Local anesthetics (LAs) are typically used for regional anesthesia but can be given systemically to mitigate postoperative pain, supplement general anesthesia, or prevent cardiac arrhythmias. However, systemic application or inadvertent intravenous injection can be associated with substantial toxicity, including seizure induction. The molecular basis for this toxic action remains unclear.
We characterized inhibition by different LAs of homomeric and heteromeric K channels containing TASK-1 (K2P3.1, KCNK3) and TASK-3 (K2P9.1, KCNK9) subunits in a mammalian expression system. In addition, we used TASK-1/TASK-3 knockout mice to test the possibility that TASK channels contribute to LA-evoked seizures.
LAs inhibited homomeric and heteromeric TASK channels in a range relevant for seizure induction; channels containing TASK-1 subunits were most sensitive and IC₅₀ values indicated a rank order potency of bupivacaine > ropivacaine > lidocaine. LAs induced tonic-clonic seizures in mice with the same rank order potency, but higher LA doses were required to evoke seizures in TASK knockout mice. For bupivacaine, which produced the longest seizure times, seizure duration was significantly shorter in TASK knockout mice; bupivacaine-induced seizures were associated with an increase in electroencephalogram power at frequencies less than 5 Hz in both wild-type and TASK knockout mice.
These data suggest that increased neuronal excitability associated with TASK channel inhibition by LAs contributes to seizure induction. Because all LAs were capable of evoking seizures in TASK channel deleted mice, albeit at higher doses, the results imply that other molecular targets must also be involved in this toxic action.
What We Already Know about This Topic
Sodium channel blockade is the major target for the therapeutic effects of local anesthetics, but the mechanisms of their pro-convulsant effects are unclear
What This Article Tells Us That Is New
Local anesthetics inhibited 2-pore domain TASK potassium channels with the same potency profile as for inducing seizures in mice, and knockout mice lacking TASK channels were less sensitive to local anesthetic-induced seizures
Neuronal hyper-excitability resulting from TASK channel inhibition contributes to local anesthetic induced seizures
LOCAL anesthetics (LAs) can be used to complement surgical anesthesia and for acute or chronic pain management.1,2Although typically used for regional anesthesia, it is also well known that LAs have beneficial effects when provided systemically in low and moderate doses. For example, perioperative intravenous lidocaine can decrease postoperative and neuropathic pain,3,4decrease postoperative morphine consumption,3,–,5and reduce requirements for general anesthetics6,–,9and counter their arrhythmogenic actions.10,11Although modern LAs are generally safe, risks persist and toxic reactions remain a problem with accidental intravascular injection, inadvertent intrathecal injection, or administration of an excessive systemic dose of these drugs.12An early expression of LA toxicity is seizure activity due to central nervous system (CNS) excitation13; subsequent manifestations are associated with CNS or cardiovascular depression.13
The potency for toxic reactions is different with different LAs.14For example, lidocaine has been in clinical use for close to 60 yr and remains one of the safest LA agents ever manufactured.15Bupivacaine, on the other hand, is a long-acting LA with a greater potential for toxic reactions15,16; ropivacaine shares many characteristics with bupivacaine but is less toxic.15,16It is clear that the major mechanism accounting for LA-induced regional anesthesia involves inhibition of voltage-gated sodium (NaV) channels,17but it is not certain that inhibition of NaV channels can account for systemic toxic effects of the drugs, including the initial CNS excitation and proconvulsive actions.18
TASK channels (TASK-1, K2P3.1; TASK-3, K2P9.1) are members of the K2P family of potassium channels that are prominently and differentially expressed throughout the brain.19,–,21TASK channels generate neuronal pH-sensitive, background (or ‘leak’) K+currents22,–,24and, of relevance here, they are inhibited by LAs18,25at toxic systemic concentrations (approximately 10–100 μM).18,26Inhibition of TASK channels causes membrane depolarization and increased neuronal excitability,23–24which could contribute to central LA toxicity. To this point, however, behavioral consequences associated with inhibition of TASK channels by LAs have not been examined.18,25
In this study, we explored the hypothesis that neuronal excitation due to TASK channel inhibition contributes to proconvulsive actions of LAs.18We show that cloned TASK channels are inhibited by LAs with a potency profile consistent with their known toxic actions (i.e. , bupivacaine > ropivacaine ≫ lidocaine). Importantly, we find that knockout mice with global deletion of TASK channels are less susceptible to LA-induced seizures. These data suggest that TASK channels are an important molecular target for central toxic effects of LAs.
Materials and Methods
Adult TASK-1−/−:TASK-3−/−double knockout and wild-type C57BL/6 mice (2–3 months) used in this study were matched for age, sex, and weight. The double knockout mice (hereafter called TASK−/−) were generated, validated, and moved onto a C57BL/6 background, as described previously.27,–,29All procedures involving animals were approved by the University of Virginia Animal Care and Use Committee.
Cell Culture and Transfections
Human embryonic kidney 293 cells stably expressing the thyrotropin-releasing hormone receptor (E2 cells) were maintained in DMEM/F-12 containing 10% fetal bovine serum, penicillin (100 U/ml), and streptomycin (100 μg/ml) and supplemented with G418 (400 μg/ml; Invitrogen Inc., Carlsbad, CA). Cells were transfected with TASK channel constructs using LipofectAMINE 2000 (Invitrogen, Charlottesville, Virginia); constructs were cotransfected (TASK-1) or tagged with enhanced green fluorescent protein (TASK-3 and TASK-1/TASK-3). Cells were plated onto poly-L-lysine (100 μg/ml)-coated glass coverslips approximately12–16 h after transfection and allowed to adhere for 30–40 min at 37°C before recording; they were visualized under infrared differential interference contrast and epifluorescent optics, and individual transfected cells with green fluorescence were selected for recording.
Whole cell recordings were obtained at room temperature using 3–5 MΩ patch pipettes and an Axopatch 200B amplifier in a bath solution consisted of (in mM): 130 NaCl, 3 KCl, 2 MgCl2, 2 CaCl2, 10 HEPES, and 10 glucose, with pH adjusted using NaOH or HCl. Different concentrations of bupivacaine (5, 25, 50, 100, and 200 μM; Sigma–Aldrich, St. Louis, MO), ropivacaine (5, 25, 50, 100, 200, and 400 μM; Naropin®; APP Pharmaceuticals LLC, Schaumburg, IL), lidocaine (4, 20, 100, 200, 400, and 800 μM; Sigma–Aldrich), and picrotoxin (50, 100, and 300 μM; Sigma–Aldrich) were added to the perfusate. Internal solution contained (in mM): 120 KCH3SO3, 4 NaCl, 1 MgCl2, 0.5 CaCl2, 10 HEPES, 10 EGTA, 3 MgATP, and 0.3 GTP-Tris, pH 7.2. Voltage commands were applied and currents recorded and analyzed with pClamp software (Molecular Devices, Sunnyvale CA). Cells were held at −60 mV and depolarizing ramps (0.2 V/s, from −130 to +20 mV) were applied at 5-s intervals. Slope conductance was determined by linear fits to currents from −130 to −60 mV. Concentration-response data for channel activity were fitted and analyzed statistically in Prism 5.0 using a Boltzmann equation of the form: Y = max/(1 + exp((IC50-X)/slope)) with three free parameters (slope, IC50, and maximum) and a fixed origin.
As a measure of CNS toxicity, we quantified effects of increasing doses of intravenous bupivacaine (2.5, 3.75, 5, 6.25, and 7.5 mg/kg), ropivacaine (2.5, 3.75, 5, 6.25, 7.5, 8.75, and 10 mg/kg) and lidocaine (5, 10, 15, and 17.5 mg/kg) on seizure induction and duration in wild-type (n = 14 for bupivacaine and ropivacaine, and 15 for lidocaine) and TASK−/−mice (n = 12 each). In addition, we determined effects of picrotoxin (0.5, 0.75, 1, and 2 mg/kg, intravenously), a γ-aminobutyric acid receptor antagonist, on seizure duration and incidence in C57BL/6 and TASK−/−mice. For each drug, the dosage was chosen based on preliminary data to span a range that encompassed doses that caused no observable seizure to those that consistently evoked seizures. Individual mice received increasing doses of a drug intravenously, with a single dose administered on any given day and at least a 1-week interval between applications. If death ensued after high-dose LA administration, data from those animals were not available for the lethal dose and any higher doses. In some cases, mice received only a single dose of the LA; data obtained with single and multiple administration protocols were not different. The incidence and duration of clonic/tonic convulsions were recorded. Convulsive signs were characterized by whole-body jumps or bursts of running motions (clonic seizure) and rigidity with forelimbs and hind limbs extended caudally (tonic seizure).30A series of quantal dose-response curves were fitted and analyzed statistically in Prism 5.0 using a logistic equation of the form: Y = max/(1 + 10((logEC50−X)*slope)) with three free parameters (slope, EC50, and maximum) and a fixed origin.
Bipolar insulated stainless steel electrodes were implanted over motor cortex under ketamine/xylazine anesthesia and secured to the skull with dental acrylic. After a 5- to 7-day recovery period, each mouse was administered a single intravenous dose of bupivacaine (3.75, 5, or 6.25 mg/kg). Electroencephalogram and video monitoring began 10 min before the LA administration and terminated when the electroencephalogram returned to normal baseline or showed irregular spikes without recurrence of seizures in a subsequent observation period of 15 min. After the experiments, animals were euthanized by deep anesthesia. Electrographically recorded seizures were obtained from three wild-type and TASK-deleted mice at each bupivacaine dose; seizure episodes of 15 s duration immediately after bupivacaine injection were subjected to power spectral analysis (Spike 2 software, v. 5.14; Cambridge Electronic Design, Cambridge, United Kingdom).
Results are presented as mean ± SEM. Data were analyzed statistically using chi-square analysis, Student t test or one-way and two-way repeated measures ANOVA, with Bonferroni correction post hoc test as appropriate. Differences were considered significant if P < 0.05 in a two-tailed analysis.
Concentration-dependent Inhibition of pH-Sensitive TASK Currents by Local Anesthetics
We used whole cell recording to test inhibition by three different local anesthetics (bupivacaine, ropivacaine, and lidocaine) of pH-sensitive homodimeric and heterodimeric TASK channels expressed in mammalian cells. The records presented in figure 1A are from an exemplar experiment in which increasing concentrations of bupivacaine were administered to a cell expressing TASK-1/TASK-3 tandem heterodimeric channels. TASK channel currents were evoked by using ramp voltage commands (0.2 V/s, from −130 to +20 mV) and peak current obtained from the corresponding I -V curves (fig. 1A, ii) was plotted as a function of time (fig. 1A, i). From an initial alkalized bath solution (pH = 8.4), currents were decreased by bath acidification (to pH 5.9) and partially restored at physiologic pH (pH = 7.3); subsequent exposure to bupivacaine caused a concentration-dependent decrease in TASK currents. Similar data were obtained for all LAs when tested with any of the TASK channel constructs, but significant differences were observed in the potency of inhibition by the different drugs on each channel type and among channel conformations for each compound. Again, this is exemplified for TASK-1/TASK-3 (fig. 1B) and for bupivacaine (fig. 1C). We derived IC50values from concentration-dependent inhibition of TASK-1/TASK-3 currents by different anesthetics that indicated a rank order of potency of bupivacaine > ropivacaine ≫ lidocaine (fig. 1B). A similar rank order potency of these anesthetics was also obtained for TASK-1 and TASK-3 homodimeric constructs (table 1); indeed, across TASK channels, bupivacaine was approximately 1.4- to 1.7-fold more potent than ropivacaine and approximately 3.1- to 4.4-fold more potent than lidocaine. As shown for bupivacaine (fig. 1C), each compound was also more potent at inhibiting homomeric TASK-1 than the other two TASK channel conformations; in this case, the rank order potency across drugs was TASK-1 ≫ TASK-1/TASK-3 > TASK-3 (table 1). Taken together, these data indicate that bupivacaine is the most potent of the LAs tested, regardless of TASK channel conformation, and channels that include TASK-1 subunits are most sensitive to LA inhibition.
TASK Knockout Mice Are Less Sensitive to Seizures Induced by Systemic Local Anesthetics
Our recordings reveal a rank order of potency for inhibition of TASK channels by LAs of bupivacaine > ropivacaine ≫ lidocaine, matching the rank order of potency reported for seizures induced by systemic administration of LAs.5,14,31,–,34To test whether TASK channels contribute to this toxic effect of LAs, we determined seizure incidence and duration induced by increasing intravenous doses of bupivacaine (2.5– 7.5 mg/kg), ropivacaine (2.5–10 mg/kg), and lidocaine (5–17.5 mg/kg) in wild-type and TASK-deleted mice.
Seizure incidence increased with increasing doses of LAs for all genotypes and, as expected, higher concentrations were required for induction of seizures by lidocaine than for ropivacaine than for bupivacaine (fig. 2A). Of particular importance, however, TASK knockout mice were markedly less likely to have seizures when injected with any of these LAs. For example, whereas 10 of 14 wild-type mice exhibited seizure activity at 3.75 mg/kg bupivacaine (approximately 71%), only 1 of 12 TASK−/−mice had seizures at this same dose (approximately 8%; P < 0.001). Likewise, at 5 mg/kg ropivacaine, seizures were observed in approximately 72% of the wild-type mice (n = 10 of 14) but in only approximately 17% of the TASK−/−mice (n = 2 of 12; P < 0.005), whereas at 6.25 mg/kg, seizures occurred in all wild-type mice (n = 14 of 14) but in only approximately 58% of TASK knockout animals (n = 7 of 12; P < 0.007). Although seizure sensitivity with lidocaine was much less than for the other drugs, similar differences in incidence were noted between wild-type and TASK knockout mice. Quantal dose-response curves describing seizure incidence for increasing doses of bupivacaine, ropivacaine, and lidocaine were well fitted by using a logistic equation; for all drugs, rightward shifts in dose-response curves were obtained with TASK channel deletion, with significantly higher EC50values for induction of seizures by comparison to wild-type mice (lidocaine: 17.3 ± 1.0 mg/kg vs. 13.5 ± 0.7 mg/kg; ropivacaine: 6.0 ± 0.2 mg/kg vs. 4.7 ± 0.1 mg/kg; bupivacaine: 4.5 ± 0.1 mg/kg vs. 3.4 ± 0.1 mg/kg; P < 0.0006, P < 0.0001, and P < 0.0003, respectively). In both wild-type and TASK-deleted mice, bupivacaine was approximately 1.4-fold more potent for induction of seizures than ropivacaine and approximately fourfold more potent than lidocaine, indicating that TASK channel deletion did not change the relative toxicity among these compounds. In addition, it should be noted that LAs remained capable of inducing seizures in most TASK channel knockout mice, even if at higher drug concentrations.
In animals that experienced seizures, we determined the duration of the seizure at each dose of LA (fig. 2B). For lidocaine and ropivacaine, the LA-induced seizures were relatively brief in duration (typically shorter than 25 s and shorter than 150 s, respectively), and little difference could be discerned between wild-type and TASK knockout mice. By contrast, seizures evoked by bupivacaine were more prolonged in general than with the other LAs, especially at higher doses in wild-type mice where they were also significantly longer in duration than those observed in TASK−/−mice (P < 0.0001). Thus, TASK knockout mice were less sensitive to seizures induced by any of the tested anesthetics and, for bupivacaine in particular, both seizure incidence and duration were reduced by TASK channel deletion.
To verify seizures electrographically, we recorded electroencephalogram activity under control conditions and immediately after an intravenous injection of bupivacaine (at 3.75, 5, and 6.25 mg/kg) in both wild-type and TASK knockout mice. As exemplified in figure 3for representative mice of each genotype that had seizures after a 3.75 mg/kg bupivacaine dose, electroencephalogram activity was characterized by a pronounced increase in peak power at frequencies ⩽5 Hz during the seizure; electroencephalogram power in this δ frequency range was enhanced during LA-induced seizures in animals from both genotypes and at all tested doses of bupivacaine (by at least sevenfold at peak seizure frequency, compared with respective baseline conditions). There was a tendency for slightly higher peak frequency at the two lower bupivacaine doses during seizures in TASK-deleted mice (e.g. , see fig. 3; 3.4 ± 0.4 Hz vs. approximately 4.4 ± 0.2 Hz in control and TASK−/−mice, n = 6), but this trend was not observed at the higher bupivacaine dose and its significance remains unclear.
Picrotoxin Does Not Inhibit TASK Channels and TASK Channel Deletion Has No Effect on Picrotoxin-induced Seizures
To rule out the possibility of nonspecific effects of TASK gene deletion on seizure susceptibility, we tested the effect of another seizure-inducing pharmacologic agent, specifically one that is silent at TASK channels. For this, we found that picrotoxin, a GABAAreceptor antagonist,35has essentially no effect on homomeric or heteromeric TASK channels, even at concentrations up to 300 μM (fig. 4A and B). As shown in figure 4A, for an exemplar cell expressing TASK-1/TASK-3 heterodimeric channels, picrotoxin had no discernible effect at 100 μM and caused only a slight decrease in peak current at 300 μM (less than 10%). Averaged data for all tested TASK channel constructs, at three different picrotoxin concentrations, revealed nonsignificant decreases in current amplitude of no greater than approximately 13% (fig. 4B). In light of this, we examined the incidence and duration of seizures induced by picrotoxin (0.5–2.0 mg/kg, intravenous) in wild-type and TASK−/−mice (fig. 4C); we found that both genotypes were equally susceptible to picrotoxin-induced seizures (EC50approximately 1.0 mg/kg), with similar durations for the two doses at which seizures were induced consistently. These data indicate that deletion of TASK channels does not nonspecifically alter sensitivity to seizures evoked by picrotoxin, a drug that does not appreciably modulate TASK channels.
In this study we characterized concentration-dependent inhibition by various LAs of TASK channels in different conformations, including a heteromeric TASK-1/TASK-3 construct. Among TASK channel subunits, TASK-1 conferred the greatest sensitivity to LAs (TASK-1 ≫ TASK-1/TASK-3 > TASK-3) with a rank order potency of channel inhibition by the LAs tested that matched the expected toxicity profile of those drugs (bupivacaine > ropivacaine ≫ lidocaine). We also showed that intravenous LA administration caused behavioral (tonic/clonic) and electrographic seizures in mice, with seizures induced by this panel of LAs following the same potency profile. In TASK-1−/−:TASK-3−/−double knockout mice, a rightward shift in the dose-response curves for seizure incidence was obtained for these three LAs, indicating that mice lacking TASK channels are less sensitive to LA-induced seizures. For bupivacaine, the longest-acting agent tested, seizure duration was significantly reduced in TASK−/−mice. We found no differences in incidence or duration of picrotoxin-evoked seizures between wild-type and TASK knockout mice, as expected given our observation that picrotoxin does not modulate TASK channels and arguing against simple, nonspecific effects of TASK gene deletion on excitability. In summary, these results suggest that increased neuronal excitability due to LA-mediated inhibition of TASK channels likely facilitates seizure induction by those drugs; because LAs retained the ability to evoke seizures in TASK−/−mice, with the same rank order potency, it appears that other targets must also be involved in this CNS toxic action of LAs.
A common concern when studying complex behavioral phenotypes in knockout mice is that a measured difference in drug action reflects some nonspecific effect on neuronal function secondary to gene deletion, rather than loss of a specific molecular target. Our results suggest that this is unlikely to be the case here. First, we performed control experiments with picrotoxin, a drug that we showed is silent at TASK channels and induced seizures equally well in wild-type and TASK−/−mice. Although this argues against some general change in seizure susceptibility in the knockout mice, the neural mechanisms underlying picrotoxin-induced and LA-induced seizures are not known to overlap, and it is therefore possible that nonspecific changes in excitability could have taken place in the circuits involved in LA-induced seizures but not in those mediating picrotoxin-induced seizures. It is noteworthy in this respect, however, that ablation of TASK channels led to a decrease in LA-induced seizure susceptibility despite the fact that loss of these neuronal background K+channels would be predicted to have an opposite nonspecific effect (i.e. , enhance excitability and be proconvulsive). Thus, together with the relatively benign unchallenged phenotypes of these and most other K2Pchannel knockout mice,24,29,36,–,41it seems most likely that TASK channel deletion led to a general homeostatic compensation that preserved neuronal excitability. In this context, the simplest explanation of our results is that TASK channels represent a behaviorally relevant target for LA toxicity; according to this idea, LA-induced TASK channel inhibition enhances neuronal excitability to promote seizure induction in wild-type mice, but loss of this drug target in TASK−/−mice renders animals less sensitive to LA-induced seizures.
It should be noted that systemic administration of LAs was effective at evoking seizures in TASK−/−mice, even if higher doses were required than with wild-type mice. In addition, there were no obvious differences in the electroencephalogram characteristics of seizures produced by higher doses of LAs in wild type and TASK−/−mice. Thus, TASK channel inhibition by LAs may play a permissive, rather than mediating, role in this toxic central action. We found that the rank order potency of LA-induced seizures was also retained in TASK knockout mice. This suggests that the rank order potency shared between LA modulation of TASK channel currents in vitro and LA induction of seizures in vivo may be coincidental rather than causative. Indeed, if the relative sensitivity of TASK channel inhibition to different LAs was responsible for differential seizure susceptibility to those same LAs, then one would have predicted that those drugs would be equipotent in TASK−/−mice. This also suggests that additional targets of LA action relevant for seizure induction may share this same rank order drug sensitivity. In this respect, NaV channel inhibition by these LAs follows the same rank order potency14,31,–,34although it is unlikely that LA-evoked seizures result from NaV channel inhibition.18,25,42Although expressed in the brain at much lower concentrations than TASK-1 or TASK-3,19,43it is possible that modulation of the distantly related (but similarly named) alkaline-activated TASK-2 channel18,25could also contribute to LA-induced seizures.
It is clear from previously published reports that both TASK-1 and TASK-3 are inhibited by various LAs18,25at concentrations relevant for toxic effects of these drugs.18,26However, in general, these data have been obtained in different laboratories with distinct expression systems, and they usually have focused on effects of a single compound in the context of one of the homomeric TASK channels; there has been no information regarding LA actions on heterodimeric TASK channels, which appear to represent a predominant native conformation.28,36,37,44,–,47Our whole cell recordings in mammalian cells confirm inhibition by three different LAs of homomeric TASK-1 and TASK-3 channels,18and indicate that TASK-1 is the more sensitive subunit. Further, we find that all three drugs inhibit heterodimeric TASK-1/TASK-3 channels (by approximately 55–65%, with IC50values of approximately 24 μM, approximately 32 μM, and approximately 82 μM for bupivacaine, ropivacaine, and lidocaine). Thus, these data indicate that all of these TASK channel conformations are sensitive to LA-mediated inhibition in a clinically relevant range.33,48,–,50
In conclusion, these results indicate a prominent contribution of TASK channel inhibition in facilitating seizures induced by LA compounds, a major toxic effect associated with systemic actions of those drugs. Thus, our data further suggest that counterscreens against TASK channel subunits could be used in development of new LA compounds with reduced CNS toxicity.
The authors thank Samuel L. Kowalski, undergraduate student laboratory aide, University of Virginia, Charlottesville, Virginia, for assistance in electroencephalogram data collection and analysis.