Differential Effects of General Anesthetics on G Protein–coupled Inwardly Rectifying and Other Potassium Channels

To construct pBKSGIRK2-087-IRK1-154-GIRK2, we prepared the following four fragments: the 0.2-kilobase (kb) pair Nco  I-Sal  I (created at amino acid residues 87–88 of GIRK2 protein with silent mutations) fragment prepared by the polymerase chain reaction (PCR) from pBKSGIRK2 25; the 0.2-kb Sal  I-Pvu  I (created at amino acid residues 142–144 of IRK1 with silent mutations) fragment prepared by PCR from pGEMIRK1 25; the 1.1-kb Pvu  I (created at amino acid residues 154–156 of GIRK2 with silent mutations)-Not  I fragment prepared by PCR from pBKSGIRK2; and the 3.0-kb Not  I-Nco  I fragment from pBKSGIRK2. These fragments were ligated to yield pBKSGIRK2-087-IRK1-154-GIRK2 (the source of the N-terminal portion, followed by the amino acid residue number in which the chimeric joint is made, and followed by the source of the C-terminal portion; residue numbering is take from the GIRK2 protein 27). To construct pBKSGIRK2-154-IRK1-204-GIRK2, the following four fragments were ligated: the 0.4-kb Nco  I-Pvu  I fragment prepared by PCR from pBKSGIRK2; the 0.2-kb Pvu  I-Spe  I (created at amino acid residues 192–194 of IRK1 with silent mutations) fragment prepared by PCR from pGEMIRK1; the 0.9-kb Spe  I (created at amino acid residues 204–206 of GIRK2 with silent mutations)-Not  I fragment prepared by PCR from pBKSGIRK2; and the 3.0-kb Not  I-Nco  I fragment from pBKSGIRK2. To construct pBKSGIRK2-087-IRK1, the following three fragments were ligated: the 0.3-kb Nde  I (created at amino acid residue 1 of GIRK2)-Sal  I fragment prepared by PCR from pBKSGIRK2; the 1.1-kb Sal  I-Not  I fragment from pGEMIRK1; and the 3.0-kb Not  I-Nde  I fragment prepared by PCR from pBKSGIRK2. To construct pBKSIRK1-061-GIRK2, the following three fragments were ligated: the 0.2-kb Nde  I (created at amino acid residue 1 of IRK1)-Sal  I fragment (composed of amino acid residues 1–49 from IRK1 followed by amino acid residues 61–87 from GIRK2) prepared by a two-step PCR 28from pGEMIRK1 and pBKSGIRK2; the 1.3-kb Sal  I-Not  I fragment prepared by PCR from pBKSGIRK2; and the 3.0-kb Not  I-Nde  I fragment from pBKSGIRK2-087-IRK1. The 0.4-kb Nde  I-Pvu  I fragment prepared by PCR from pGEMIRK1, the 1.1-kb Pvu  I-Not  I fragment from pBKSGIRK2-087-IRK1-154-GIRK2, and the 3.0-kb Not  I-Nde  I fragment from pBKSGIRK2-087-IRK1 were ligated to yield pBKSIRK1-154-GIRK2. The amino acid residues 136–143, 152, 159–168, 169–178, 179–188, and 190–197 of pBKSGIRK2 were replaced by the corresponding residues of IRK1 to yield the pBKSGIRK2-MC1, MC2, MC3, MC4, MC5, and MC6, respectively, by using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA) according to the two-stage PCR protocol. 29All constructs were confirmed by DNA sequencing.

The cDNA encoding the human ERG1 subunit 8in pSP64 vector, cDNA encoding the rat ELK1 subunit 11in pBluescript SK vector, cDNAs encoding the human KCNQ2 and rat KCNQ3 subunits 13in pBluescriptII SK vector, and cDNA encoding the rat μ-opioid receptor 30in pRc/CMV vector were used for cRNA synthesis. In vitro  transcripts were prepared using the mRNA capping kit (Stratagene, La Jolla, CA). The cRNAs for GIRK1, GIRK2, GIRK4, IRK, GIRK2-IRK1 chimeras and multiple amino acid mutants, and ROMK1 potassium channels were synthesized as described previously. 25 

The use of experimental animals (frogs) was approved by the Animal Care and Use Committees of University of Texas. Isolation of Xenopus laevis  oocytes and microinjection of the cRNA was performed as described previously. 25Isolated oocytes were placed in modified Barth saline (MBS) containing 88 mm NaCl, 1 mm KCl, 2.4 mm NaHCO³, 0.82 mm MgSO⁴, 0.91 mm CaCl², 0.33 mm Ca(NO³)², and 10 mm HEPES adjusted to pH 7.5. The cRNAs (5–30 ng/oocyte, in a 1:1 molar ratio for heteromeric channels) were injected into the cytoplasm of oocytes. The injected oocytes were singly placed in Corning cell wells (Corning Glass Works, Corning, NY) containing incubation medium (sterile MBS supplemented with 10 mg/l streptomycin, 10,000 U/l of penicillin, 50 mg/l of gentamicin, 90 mg/l theophylline, and 220 mg/l pyruvate) and incubated at 15–19°C. Two to 5 days after injection, oocytes were used in electrophysiologic recording.

Oocytes expressing the potassium channels were placed in a rectangular chamber (approximately 100-μl volume) and perfused (2 ml/min) with MBS. The animal poles of oocytes were impaled with two glass electrodes (0.5–10 MΩ) filled with 3 m KCl and voltage clamped at −80 mV, unless stated otherwise, using a Warner Instruments model OC-725A (Hamden, CT) oocyte clamp. For voltage-gated potassium channels, current responses were evoked by a 2-s voltage step to +20 mV from a holding potential of −80 mV, with 10-s interpulse intervals. During the voltage step protocol, anesthetics were applied for 1 min, since a maximal effect of anesthetics occurred within 1 min. Data collection and analysis were performed with pClamp software (Axon Instruments, Foster City, CA). For the inwardly rectifying potassium channels, oocytes were first bathed in MBS, then switched to a high potassium (hK) solution containing 2 mm NaCl, 96 mm KCl, 1.5 mm CaCl², 1 mm MgCl², and 5 mm HEPES at pH 7.5 to elicit an inward current through basally active channels. 31,32After establishing a stable response, anesthetics were applied in the hK solution for 1 min before changing back to the hK solution.

Solutions of anesthetics were prepared as described previously. 33The anesthetic concentrations in the text represent the final bath concentrations quantified by gas chromatography as described previously. 33For nitrous oxide, concentrations tested were the maximal concentrations that could be achieved in our perfusion system. Data were obtained from three to nine oocytes taken from at least two different frogs. All experiments were performed at room temperature (23°C).

Halothane was obtained from Halocarbon Laboratories (River Edge, NJ); isoflurane was obtained from Anaquest Co. (Madison, WI); enflurane was obtained from Ohio Medical Products (Madison, WI); F3 (1-chloro-1,2,2-trifluorocyclobutane) and F6 (1,2-dichlorohexafluorocyclobutane) were obtained from Lancaster Synthesis, Inc. (Windham, NH); ethanol was obtained from Aaper Alcohol and Chemical Co. (Shelbyville, KY); 2,6-diisopropylphenol (propofol) was obtained from Aldrich Chemical Co. (Milwaukee, WI); etomidate was obtained from Janssen Pharmaceutica (Beerse, Belgium); alphaxalone was obtained from Research Biochemicals International (Natick, MA); nitrous oxide, xenon, and nitrogen were obtained from Airgas-Southwest, Inc. (San Antonio, TX); and 1-hexanol, pentobarbital sodium, ketamine hydrochloride, [D-Ala², N -Me-Phe⁴,Gly⁵-ol]enkephalin (DAMGO), and other reagents were purchased from Sigma Co. (St. Louis, MO). Propofol and alphaxalone were first dissolved in dimethyl sulfoxide and then diluted in MBS or hK solution. The highest final dimethyl sulfoxide concentration (0.01%) did not affect current responses of potassium channels tested.

The results obtained were statistically analyzed by t  tests or one-way analysis of variance followed by Scheffé multiple comparison tests. P  values < 0.05 were considered significant. Data are represented as mean ± SD.

The ERG1 channels were activated by a 2-s voltage step to +20 mV from a holding potential of −80 mV. A step back to the holding potential of −80 mV elicited a large tail current (fig. 1A). Effects of anesthetics on ERG1 channels were evaluated by measuring amplitudes of the tail currents during perfusion of the anesthetics. Enflurane at 1.2 mm (twice minimum alveolar concentration [MAC]) slightly inhibited the tail current of ERG1 channels (fig. 1A). Figure 1Bshows effects of 2 MAC or twice the anesthetic EC⁵⁰of various anesthetics 2,34–36on ERG1 channels. Volatile anesthetics (halothane, isoflurane, enflurane, and F3) only slightly (4–11%) inhibited the ERG1 channels. Ethanol (380 mm) inhibited ERG1 channels by 24%, whereas hexanol (1.1 mm) inhibited them by 6%. A nonimmobilizer F6 and intravenous anesthetics (pentobarbital, propofol, etomidate, alphaxalone, and ketamine) barely affected the ERG1 channels. The current responses of KCNQ2/3 channels were evoked by the same voltage step protocol, and the effects of anesthetics on the tail currents were evaluated (fig. 1C). The KCNQ2/3 channels were also relatively insensitive to the anesthetics tested. ELK1 channels were activated by a 3-s voltage step from a holding potential of −80 mV over the range −70 to +40 mV in 10-mV increments (fig. 1D). Enflurane (1.2 mm) produced little effect on the current responses. Figure 1Eshows effects of 2 MAC of anesthetics on ELK1 channels. ELK1 channels were resistant to all anesthetics tested.

Current responses of inwardly rectifying potassium channels were obtained by perfusion of a hK solution that elicits an inward current through basally active channels. 31,32After establishing a stable response, effects of anesthetics were evaluated by application of the anesthetics in the hK solution. The currents of GIRK2 channels were markedly inhibited by 2 MAC of halothane (fig. 2A). Although currents of GIRK2 channels were inhibited immediately after application of volatile anesthetics, recovery from inhibition was slow and often incomplete. Conversely, currents of IRK1 channels were not affected by halothane. Application of anesthetics in MBS did not change the baseline currents in MBS for both GIRK2 and IRK1 channels. The effects of 2 MAC or twice the anesthetic EC⁵⁰of various anesthetics on GIRK1/2, GIRK1/4, and GIRK2 channels were examined (figs. 2B, 2C, and 2D). Halothane, isoflurane, and enflurane inhibited GIRK1/2 and GIRK2 channels, whereas they only slightly affected the GIRK1/4 channels. F3 prominently inhibited GIRK1/2, GIRK1/4, and GIRK2 channels, whereas a structurally related nonimmobilizer F6 only slightly inhibited the GIRK channels. Ethanol markedly potentiated the currents of GIRK channels as reported previously, 25,26whereas hexanol slightly inhibited GIRK channels. Not only ethanol, but also methanol and propanol, are reported to prominently potentiate the currents of GIRK channels. 25,26All GIRK channels were insensitive to intravenous anesthetics. To assure that GIRK1/2 channels were actually expressed by heteromeric assembly of GIRK1 and GIRK2 subunits, the amount of GIRK2 subunit cRNA was reduced to one tenth of the original amount, and the current amplitude was compared between the GIRK1 (∼10 ng of GIRK1 subunit cRNA/oocyte), GIRK1/2 (∼10 and ∼1 ng of GIRK1 and GIRK2 subunit cRNAs/oocyte, respectively), and GIRK2 (∼1 ng of GIRK2 subunit cRNA/oocyte) channels. During these conditions, current responses of GIRK1/2 channels were very significantly larger than those of GIRK1 and GIRK2 channels, indicating heteromeric assembly for the GIRK1/2 channels, which is consistent with previous reports. 37In contrast to GIRK channels, IRK1 and ROMK1 channels were completely resistant to volatile anesthetics, alcohols, and intravenous anesthetics (figs. 2E and 2F).

Effects of nitrous oxide on voltage-gated and inwardly rectifying potassium channels were examined (fig. 3). Nitrous oxide (0.58 atmosphere [atm]) did not affect ERG1, ELK1, and KCNQ2/3 channels. Conversely, nitrous oxide potentiated GIRK1/2, GIRK1/4, and GIRK2 channels by 8–9%. Perfusion of a hK solution bubbled with 100% nitrogen produced no measurable effects on GIRK1/4 and GIRK2 channels (% change of 1 ± 1% and 0 ± 2%, respectively; n = 4), indicating that the observed potentiation of GIRK channels by nitrous oxide is not caused by displacement of oxygen from the solutions. IRK1 and ROMK1 channels were insensitive to nitrous oxide.

To examine differences in sensitivity to volatile anesthetics among inwardly rectifying potassium channels, we evaluated the dose–inhibition relations of GIRK1/2, GIRK1/4, GIRK2, and IRK1 channels for halothane and F3. GIRK1/2 and GIRK2 channels were inhibited by halothane in a dose-dependent manner (fig. 4A). Halothane at 0.25 mm (1 MAC) inhibited GIRK1/2 and GIRK2 channels by 31 ± 3% and 39 ± 4%, respectively (n = 4–7). Conversely, GIRK1/4 and IRK1 channels were not inhibited by halothane even at 5 MAC. F3 potently inhibited GIRK1/2, GIRK1/4, and GIRK2 channels in a dose-dependent manner, and GIRK1/4 channels were less sensitive to F3 than GIRK1/2 and GIRK2 channels (fig. 4B). F3 at 0.8 mm (1 MAC) inhibited GIRK1/2 and GIRK2 channels by 78 ± 5% and 78 ± 3%, respectively (n = 4–7), whereas it inhibited GIRK1/4 channels by 41 ± 3% (n = 6). In contrast, IRK1 channels were completely resistant to F3. We next examined the voltage dependence of inhibition of GIRK2 channels by halothane. The GIRK2 currents elicited by a hK solution showed inward rectification, a characteristic of GIRK channels (fig. 4C). Halothane inhibition of GIRK2 channels was not different among membrane potential tested (fig. 4C, P > 0.92). Similarly, inhibition of GIRK2 channels by 0.08 mm F3 was not different for the membrane potentials tested (data not shown). These results demonstrate that inhibition of GIRK channels by volatile anesthetics is not voltage-dependent.

To examine whether anesthetics inhibit GIRK channels activated by G protein–coupled receptors, GIRK channels were coexpressed with μ-opioid receptors, and anesthetic effects on GIRK channel currents activated by μ-opioid receptors were evaluated. In oocytes expressing both GIRK2 channels and μ-opioid receptors, application of a selective μ-opioid receptor agonist DAMGO evoked an inward current in hK solution that is mediated by GIRK2 channels activated by μ-opioid receptors. Effects of halothane were evaluated by measuring current responses to DAMGO during incubation of halothane. Halothane inhibited the DAMGO-evoked currents of GIRK2 channels in a reversible manner (fig. 5A). Halothane (0.5 mm) inhibited DAMGO-evoked currents of GIRK2 channels by 53 ± 3% (fig. 5B). In contrast, 0.5 mm halothane did not affect DAMGO-evoked currents of GIRK1/4 channels. Propofol (2 μm) and ketamine (20 μm) barely affected DAMGO-evoked currents of GIRK2 channels, whereas nitrous oxide (0.58 atm) potentiated them by 12 ± 2% (n = 6; 100% nitrogen did not affect them [2 ± 2% inhibition, n = 5]). Therefore, effects of anesthetics on GIRK channels was similar between basally active and receptor-activated GIRK channels.

Because GIRK2 channels were markedly inhibited by volatile anesthetics but IRK1 channels were completely insensitive to them, and both GIRK2 and IRK1 subunits form homomeric channels, we constructed chimeric subunits between GIRK2 and IRK1 subunits to identify the subunit region determining the distinct anesthetic sensitivity. The GIRK2-087-IRK1 subunit was composed of the GIRK2 subunit sequence from the amino terminus to the junction site (residue 087 of GIRK2) and the IRK1 subunit sequence from the junction site to the carboxyl terminus (fig. 6). The currents of the GIRK2-087-IRK1 channels were resistant to both 0.5 mm halothane and 1.6 mm F3, indicating that anesthetic sensitivity is dependent on the carboxyl terminal side of the residue 087. The GIRK2-087-IRK1-154-GIRK2 and GIRK2-154-IRK1-204-GIRK2 channels produced little current response to an hK solution even after coexpression with GIRK1 subunits. The GIRK2-223-IRK1 channels 25coexpressed with GIRK1 subunits were markedly inhibited by halothane and F3, indicating that the responsible site resides on the amino terminal side of the residue 223. The IRK1-061-GIRK2 channels were inhibited by volatile anesthetics, suggesting the involvement of the carboxyl terminal side of the residue 061. The IRK1-154-GIRK2 channels did not produce current responses even after coexpression with GIRK1 subunits. Data obtained from these chimeras suggest that anesthetic sensitivity is determined by the region of residues 087-223 that contains the transmembrane (M1 and M2) and pore-forming (P) domains.

Because there is evidence of involvement of channel-lining residues in anesthetic inhibition of ligand-gated ion channels, 38,39and the P and M2 domains correspond to the channel-lining regions of potassium channel pore, 40we constructed multiple amino acid mutants around the P and M2 domains of GIRK2 channels and examined whether this region determines anesthetic sensitivity. The region encompassing the P and M2 domains were divided into six segments containing 8–15 amino acid residues (MC1–MC6;fig. 7A), and each segment of GIRK2 subunits was replaced by the corresponding segment of IRK1 subunits to create GIRK2-MC1, GIRK2-MC2, GIRK2-MC3, GIRK2-MC4, GIRK2-MC5, and GIRK2-MC6 mutant subunits. The GIRK2-MC1, GIRK2-MC2, GIRK2-MC3, and GIRK2-MC4 channels produced currents in response to an hK solution, but GIRK2-MC5 and GIRK2-MC6 channels produced little current even after coexpression with GIRK1 subunits. The GIRK2-MC1, GIRK2-MC2, GIRK2-MC3, and GIRK2-MC4 channels were inhibited by 0.5 mm halothane to the extents that were not significantly different from the wild-type GIRK2 channels (fig. 7B). F3 (1.6 mm) also inhibited GIRK2-MC1, GIRK2-MC2, GIRK2-MC3, and GIRK2-MC4 channels to the extents similar to the wild-type channels (fig. 7C). Thus, the regions MC1–MC4 may not be involved in determining differences in anesthetic sensitivity between GIRK and IRK channels.

In the current investigation, we showed that voltage-gated potassium channels, ERG1, ELK1, and KCNQ2/3 channels, are insensitive to clinical concentrations of volatile, intravenous, and gaseous anesthetics. Thus, these voltage-gated potassium channels are not likely to be involved in anesthetic mechanisms. Conversely, among inwardly rectifying potassium channels, GIRK channels were sensitive to volatile anesthetics, ethanol, and nitrous oxide. GIRK channels were inhibited by clinical concentrations of volatile anesthetics, whereas nitrous oxide and ethanol selectively potentiated GIRK channels.

GIRK1/2 channels were markedly inhibited by halothane, isoflurane, enflurane, and F3. In contrast, GIRK1/4 channels were resistant to halothane, isoflurane, and enflurane but were inhibited by F3. The brain GIRK channels are considered to be heteromultimers composed of GIRK1 and GIRK2 subunits, whereas the cardiac GIRK channels are heteromultimers composed of GIRK1 and GIRK4 subunits. 20–23Thus, our results suggest that brain GIRK channels, but not cardiac GIRK channels, are inhibited by volatile anesthetics used in clinical practice. In bullfrog atrial myocytes, however, it is reported that muscarinic-activated potassium current (I^K(ACh)), which is considered to be mediated by heteromultimers of GIRK1 and GIRK4 subunits, 23is affected by isoflurane and halothane. 24The activation rate and rapid desensitization of I^K(ACh)elicited by acetylcholine were reduced by isoflurane (0.8 mm) and halothane (0.9 mm), whereas the magnitude of I^K(ACh)was slowly reduced by isoflurane but rapidly augmented by halothane. 24Thus, although effects of isoflurane and halothane on steady state amplitudes of GIRK1/4 channels were small in our studies, it is possible that channel gating kinetics of GIRK1/4 channels is somewhat altered by the volatile anesthetics.

In the central nervous system, activation of GIRK channels results in the outward flow of potassium ions, which causes a decrease in neuronal excitability. 1,32Thus, inhibition of brain GIRK channels by volatile anesthetics would increase excitability and is not likely to be involved in mechanisms producing anesthesia. This suggestion is supported by the finding that equipotent concentrations (in terms of immobility) of volatile anesthetics inhibited GIRK channels by different degrees. For example, 2 MAC of isoflurane inhibited GIRK1/2 channels by only 22%, whereas 2 MAC of F3 inhibited them by 81%. Because F3 inhibited GIRK channels but structurally related nonimmobilizer F6 did not, inhibition of GIRK channels may be related to some excitatory, rather than inhibitory, behavioral actions induced by volatile anesthetics.

G protein–coupled inwardly rectifying potassium channels are activated by many G protein–coupled receptors. 1The inhibition of GIRK2 channels by volatile anesthetics may imply that a part of function of G protein–coupled receptors that is mediated by GIRK channels in the nervous system is inhibited by volatile anesthetics. In the current study, we showed that current responses of μ-opioid receptors to DAMGO, mediated by GIRK2 channels, were inhibited by halothane. This inhibition of μ-opioid receptor function by halothane is likely caused by inhibition of GIRK channels, not opioid receptors, because the extent of inhibition by halothane was similar to that of basally active GIRK2 channels, and current responses of μ-opioid receptors coupled with GIRK1/4 channels were not inhibited. Although signal transduction pathways of opioid receptors include not only activation of GIRK channels but also alteration of Ca²⁺channel function and levels of second messengers such as cyclic adenosine monophosphate, 41it is possible that some actions of opioids might be inhibited by volatile anesthetics. Interestingly, halothane antagonizes suppressive effects of morphine on the heart rate increase to noxious stimulation, but not on the purposeful movement to noxious stimulation. 42 

G protein–coupled inwardly rectifying potassium channels were inhibited by volatile anesthetics, but homologous IRK channels were completely resistant to volatile anesthetics. Although GIRK channels are activated by binding of the G protein β and γ subunits, 43inhibition of GIRK channels by volatile anesthetics may not be caused by inhibition of activity or binding of G proteins, but may be caused by inhibition of potassium channels themselves, because GIRK1/4 channels were insensitive to halothane, isoflurane, and enflurane at concentrations that markedly inhibited GIRK2 channels. Previous studies showed the involvement of cytoplasmic domains of various potassium channel subunits in actions of anesthetics 3and alcohols. 25,44,45As for ethanol actions on GIRK channels, the carboxyl terminus is identified to be critical for enhancement of GIRK channels by ethanol. 25Our data from chimeric GIRK-IRK channels, however, suggest that cytoplasmic amino and carboxyl termini are not involved in determining anesthetic sensitivity of GIRK channels. Thus, not only the direction of effects (inhibition or potentiation), but also mechanisms or sites of actions on GIRK channels may be different between volatile anesthetics and ethanol. The multiple amino acid mutations in the channel-lining region did not affect sensitivity of GIRK2 channels to halothane and F3. Because GIRK2-MC5 and GIRK2-MC6 channels did not produce measurable currents, we cannot exclude the possibility of involvement of this region in determining anesthetic sensitivity. However, our results suggest that the region around the P domain may not be important for determining differences in anesthetic sensitivity between GIRK and IRK channels.

Previous studies reported that, in hippocampal neurons, ethanol in low concentrations (22–44 mm) inhibits the M current. 17,18This current was originally identified as a time- and voltage-dependent potassium current susceptible to inhibition by muscarinic agonists in sympathetic neurons. 16Analogous currents (M-like currents) have subsequently been identified in a variety of other neuronal and nonneuronal cells, and the M-like current plays a prominent role in inhibiting the excitability of the cell. Recently, the M current was proposed to be composed of a heteromeric assembly of KCNQ2 and KCNQ3 subunits. 13Furthermore, two components of the M-like current in neuronal cells are reported to be mediated by two distinct molecular species of potassium channel: KCNQ2/3 and ERG1 channels. 15In the current study, low concentrations of ethanol (10–50 mm) did not affect KCNQ2/3 and ERG1 channels (data not shown), and even high concentrations (380 mm) of ethanol only slightly inhibited KCNQ2/3 and ERG1 channels. Therefore, ethanol inhibition of the M current in hippocampal neurons may not be caused by direct inhibition of the potassium channels, but may involve other target components.

In conclusion, we determined anesthetic sensitivity of various voltage-gated and inwardly rectifying potassium channels. Among the potassium channels tested, GIRK channels, especially those composed of GIRK2 subunits, were markedly inhibited by clinical concentrations of volatile anesthetics. These results suggest that function of G protein–coupled receptors, which is mediated by GIRK channels in the nervous system, may be inhibited by volatile anesthetics. Although inhibition of GIRK channel function by anesthetics is unlikely to account for the anesthetic actions, it may contribute to some excitatory side effects. Results from chimeras and multiple amino acid mutants between GIRK2 and IRK1 subunits suggest that mechanisms or sites of actions on GIRK channels are different between volatile anesthetics and ethanol, and that the region containing the transmembrane domains, but not the P domains, may be involved in determining differences in anesthetic sensitivity between GIRK and IRK channels.

The authors thank Drs. Jane E. Dixon, Ph.D. (Research Assistant Professor, Institute of Molecular Cardiology, State University of New York, Stony Brook, NY), and Michael C. Sanguinetti, Ph.D. (Professor, Internal Medicine Admin., Eccles Institute of Human Genetics, University of Utah, Salt Lake City, UT), for providing the potassium channel clones.