The sensitivity of G protein-activated K+ channels toward halothane is essentially determined by the C terminus.

G protein-activated K(+) channels (GIRKs or Kir3.x) are targets for the volatile anesthetic, halothane. When coexpressed with the m(2) acetylcholine (ACh) receptor in Xenopus oocytes, agonist-activated GIRK1(F137S)- and GIRK2-mediated currents are inhibited by halothane, whereas in the absence of ACh, high concentrations of halothane induce GIRK1(F137S)-mediated currents. To elucidate the molecular mechanism of halothane action on GIRK currents of different subunit compositions, we constructed deletion mutants of GIRK1(F137S) (GIRK1(Delta363*)) and GIRK2 (GIRK2(Delta356)) lacking the C-terminal ends, as well as chimeric GIRK channels. Mutated GIRK channels showed normal currents when activated by ACh but exhibited different pharmacological properties toward halothane. GIRK2(Delta356) showed no sensitivity against the inhibitory action of halothane but was activated by halothane in the absence of an agonist. GIRK1(Delta363*) was activated by halothane more efficiently. Currents mediated by chimeric channels were inhibited by anesthetic concentrations that were at least 30-fold lower than those necessary to decrease GIRK2 wild type currents. Glutathione S-transferase pulldown experiments did not show displacement of bound Gbetagamma by halothane, indicating that halothane does not interfere with Gbetagamma binding. Single channel experiments revealed an influence of halothane on the gating of the channels: The agonist-induced currents of GIRK1 and GIRK2, carried mainly by brief openings, were inhibited, whereas higher concentrations of the anesthetic promoted long openings of GIRK1 channels. Because the C terminus is crucial for these effects, an interaction of halothane with the channel seems to be involved in the mechanism of current modulation.

The effects of anesthetics that induce general anesthesia are poorly understood. On the cellular level, the synaptic transmission of nerve impulses appears to be impaired by anesthetics (1). The underlying molecular targets for this effect are still a matter of debate. Besides ion channels, which have been shown to be modulated by anesthetics (2), other elements of signal transduction such as receptors and G proteins may also be involved in general anesthesia.
The family of G protein-activated inwardly rectifying K ϩ (GIRK) 1 channels is comprised of at least five isoforms, designated GIRK1-GIRK5 (Kir3.1-3.5) (3). There is strong evidence that functional GIRK channels are homotetrameric or heterotetrameric complexes with the mammalian GIRK1-GIRK4 isoforms being differentially expressed primarily in the central nervous system and cardiac tissue (4 -6). Functionally, GIRK channels are activated by the binding of G␤␥ subunits to the intracellular portions of the channel protein (7,8). G␤␥ is released from heterotrimeric, inactive G␣␤␥ subunit complexes that have been activated by the binding of an agonist to a G protein-coupled receptor. On the cellular level, the opening of GIRK channels stabilizes the membrane potential at E K ϩ and thus counteracts membrane excitability. As a result, the activation of GIRK channels leads to a decrease in heart rate after the release of acetylcholine from the vagus nerve, and in the central nervous system GIRK channels play an important role in the mediation of opioidand ethanol-induced analgesia (9 -12).
Pharmacologically, GIRK channels are modulated by volatile anesthetics (13). GIRK1 F137S or GIRK1 containing heteromeric channels such as GIRK1/GIRK4, but not homomeric GIRK4 channels, are activated by high concentrations of halothane in the absence of an agonist (14). In contrast, low concentrations of halothane are able to inhibit the agonist-induced currents through GIRK channels when expressed in Xenopus oocytes. GIRK2 channels are not activated by halothane at the basal level but are most sensitive to the inhibitory action of halothane when activated by an agonist. These effects are independent from the receptor coexpressed and specific for the GIRK channels (14). However, the mechanism of this dualistic modulation of GIRK channels is not clear. The existence of at least two independent mechanisms of halothane modulation of GIRK channels had to be assumed. From our previous experiments (14) we concluded that halothane was able to activate the GIRK1 channel, probably by increasing its affinity to G␤␥, and that the inhibition of agonist-activated GIRK channels was because of the impairment of the G protein signaling cascade. This is consistent with the findings of various other laboratories that demonstrate the inhibition of signaling through G␣ i by halothane (15)(16)(17)(18).
Structurally, GIRK channels consist of a core region with two transmembrane domains (TM1 and TM2) with a reentrant P-loop in between (Fig. 1A, P). This core region is flanked by N-and C-terminal regions, which constitute the "cytoplasmic pore" as judged from crystal structure analysis (19,20). This structure is exposed to the cytoplasm and is believed to encom-* This work was supported by Grant 8266 from the Oesterreichische Nationalbank (to L. G. W.), Austrian Research Foundation Grants P17702-B09 (to L. G. W.), T124-B04 (to B. S.-F.), and SFB007/F708 (to W. S.), and Austrian Ministry of Science Grant UGP4 (to W. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. Tel.: 43-1-40400-4147; Fax: 43-1-40400-6422; E-mail: lukas.weigl@univie.ac.at. pass the G␤␥ binding sites (21)(22)(23)(24). These parts of the proteins show a high degree of homology between the different isoforms. The most divergent regions are the distal N and C termini of the protein. The long distal C terminus of GIRK1 is unique among the GIRK channels and shows virtually no homology to the C termini of other GIRK channels. The role of the distal C terminus is not clear, yet it is not thought to take part in constitution of the cytoplasmic pore. Rather, it protrudes into the cytoplasm where it may function as an interaction domain for other proteins or the channel itself.
To elucidate the mechanism of halothane action on GIRK channels on the molecular level and to investigate the structural requirements for the modulation by halothane, we performed single channel experiments and used deletion mutants of GIRK1 F137S (the substitution of a serine for a phenylalanine in GIRK1 is necessary to get conductive homooligomeric GIRK1 channels; Ref. 25) and GIRK2 as well as chimeric proteins between these two channels. In addition, we tested whether halothane interacts with the binding of G␤␥ to GST fusion proteins, which comprise the putative G␤␥ binding sites of the N-and C termini of GIRK1.
Preparation of cRNAs-Plasmids were isolated from bacteria and linearized using standard procedures (28). The following DNAs were used: m 2 receptor (29), rat GIRK1 F137S from atrium, and GIRK2 WT from mouse testis cloned into pBS-mxt. The truncated GIRK1 ⌬363* (aa 1-363) and GIRK2 ⌬356 (aa 1-356) channels were created by PCR via primers containing the appropriate restriction sites EcoRV/EcoRI and ClaI/SpeI, respectively, as well as a stop codon at the end of each coding region (LW5, 5Ј-CCCGATATCATGTCTGCACTCCGAAGG-3Ј; LW6, 5Ј-GCGGAATTCCTATTCTTCCTGCTCTTTCACAC-3Ј; LW1, 5Ј-CCCAT-CGATATGGACCAGGATGTGGAAAGCC-3Ј; LW2, 5Ј-GCGACTAGTC-TACTCCGCTAGCTCTTTGGCAC-3Ј). The introduced restriction sites allowed the insertion of the amplified DNA into pBSmxt, which is a cloning vector designed to enable the synthesis of cRNA for injection into X. laevis oocytes. The PCR products were cleaned by EtOH precipitation prior to digestion with the respective enzymes. The DNA was further cleaned by gel extraction using the QUIAEX kit (Quiagen). The cleaned DNA was ligated and transformed into XL-1-BLUE cells. For the displacement of the C-terminal region of GIRK2 by the homologous area of GIRK1, the same truncated GIRK2 cDNA as described above was constructed except that there was no stop codon at the 3Ј-end (LW1; LW2_1, 5Ј-GCGACTAGTCTCCGCTAGCTCTTTGGCACT-3Ј). The additional PCR product created from the cytoplasmic loop of GIRK1 was cloned into the SpeI site of the GIRK2 ⌬356 pBS-mxt plasmid. For exchange of the transmembrane domains of GIRK2 and GIRK2/1 356 by the homologous domain of GIRK1 F137S (aa 88 -202), restriction sites for AflII and EheI were introduced by site-directed mutagenesis. The corresponding sequence of GIRK1 was cloned into the restriction sites at base positions 201 and 549 of GIRK2 and GIRK2/1 356 , respectively. These chimeric channels were designated GIRK2/1/2 and GIRK2/1/2/1. The correctness of the DNA constructs was verified by automated sequencing. The DNAs were linearized with SalI, and cRNA was synthesized with the Ambion megascript kit (Ambion, Austin, TX).
Electrophysiology-For two-electrode voltage clamp recordings, the oocytes were placed in a chamber that allowed the superfusion of the cells at 19 -21°C. A virtual complete exchange of the bath solution could be reached within 4 s, as judged by the changes in offset potentials. For halothane-containing solutions a gas tight superfusion system was used that was made of glass syringes and Teflon tubes to prevent evaporation of the anesthetic. The halothane solutions were prepared from a saturated stock solution (17.8 mM) by dilution to the required working concentrations. The actual concentration of the applied halothane solutions was analyzed in mock experiments by gas chromatography using an HP 5890GC (Hewlett Packard) device. Concentrations were found to be 0.96 Ϯ 0.03 mM for 1 mM halothane (n ϭ 10), 312 Ϯ 13 M for 300 M (n ϭ 10), 97.2 Ϯ 7.8 M for 100 m halothane (n ϭ 12), and 29.1 Ϯ 7.3 M for 30 M (n ϭ 3). In figures and text these concentrations are referred to as 1 mM, 300 M, 100 M, and 30 M halothane.
The agonist-induced current (I ACh ) was evoked by superfusion with 10 M ACh. The quantification of the effect of halothane was done 1 min after the start of rinsing with the halothane-containing solution. Control values for the halothane effect on I ACh were obtained by determining the current value 1 min after the peak current was reached during rinsing with 10 M ACh without halothane. Control values reflect the current desensitization. Current traces were digitized at 50 Hz using an Axolab 1200 interface and were recorded with Axotape (Axon Instruments) on an IBM-compatible computer. Analysis of current recordings was performed with the Fetchan 6.0 software (Axon Instruments).
IC 50 values were determined by least square fitting of the relative inhibition of ACh-induced currents to the following equation h ))ϩy 0 where %I is the percentage inhibition of I ACh , [halo] the concentration of halothane, h the Hill coefficient, and y 0 the control value.
For patch clamp experiments the vitelline membrane of oocytes was removed with fine forceps after putting the cells in a hypertonic solution (PG200Ca, in mM: glutamate (180), KCl (37.5), CaCl 2 (1), MgCl 2 (1), Hepes (10), pH 7.5). Immediately after devitellinization the oocyte was transferred to the recording chamber filled with 500 l of bath solution (BS, in mM: KCl (140), NaCl (6), EGTA (1), MgCl 2 (4), ATP Na 2 (1), HEPES (10), KOH to pH 7.5). Patch pipettes were pulled from borosilicate glass 1B150 (WPI), sylgard coated, fire polished, and back filled with pipette solution (PS, in mM: KCl (144), NaCl (2), MgCl 2 (1), CaCl 2 (1), GdCl 3 (1), HEPES (10), KOH to pH 7.5). The tip was filled with the same solution without GdCl 3 . Measurements were performed in the cell-attached mode at a holding potential of Ϫ80 mV with a HEKA EPC-9 amplifier and recorded with the Pulse program (HEKA, Germany) at a sampling rate of 15 kHz and a filter frequency of 5 kHz. The basal channel activity was monitored for at least 5 min. From a saturated halothane stock solution 28 l were directly applied to the bath to obtain a concentration of 1 mM. The solution was mixed by gentle pipetting. In mock experiments the loss of halothane because of evaporation was determined to be 23% within 5 min.
For analysis, the data were digitally filtered at 1.5 kHz. The number of channels in the patch was estimated from the overlaps of channel openings during the course of an experiment. Experiments with three or fewer channels were analyzed. Event detection was based on a 50% criterion, and event tables were generated using the TAC program GST Pulldown-GST fusion proteins that contained GST and the N terminus of GIRK1 (aa 1-84, CD1) or the C terminus of GIRK1 (aa 183-501, CD2) were created using PCR. DNA fragments that encoded the indicated amino acid residues from rat GIRK1 were cloned into the vector pGEX-cs using the restriction enzymes BamHI and NcoI or XhoI, respectively. All DNA constructs were verified by nucleotide sequencing. GST⅐GIRK fusion proteins were expressed in BL21(DE3) cells and purified using glutathione affinity resin as described (30). The protein was eluted from the resin with elution buffer (in mM: Hepes (20), EDTA (1), dithiothreitol (2), KCl (5), glutathione; Ref. 15). If necessary, the protein was concentrated and dialyzed overnight in elution buffer lacking glutathione. [ 35 S]G␤ 1 ␥ 2 was synthesized in reticulocyte lysate (Promega) and diluted 1:2. The fusion proteins (4 -7 g/ml) were mixed with 5 l of the diluted lysate and incubated for 2 h under gentle shaking in 700 l of binding buffer (Hepes/Na, pH 7.4, (25), MgCl 2 (5), EGTA (5), Tween 20 (0.05%)). The mixture was then incubated with glutathione-Sepharose beads for 30 min. The beads were recovered by filtering through 0.45-M spin filters and washed three times with binding buffer. Bound proteins were eluted with glutathione buffer (in mM: glutathione (15), NaCl (120), Tween 20 (0.05%), Tris (100), pH 8.0) and analyzed on a 12% SDS-polyacrylamide gel followed by Coomassie Blue staining and autoradiography using phosphorimaging as described (31). All steps of the procedure, except the gel electrophoresis, were performed in the presence of the respective concentrations of halothane. The halothane concentration in the solutions was achieved by diluting a saturated stock solution as described above. Halothane-containing wash solutions were prepared directly before use to minimize loss caused by evaporation.
Data Presentation and Statistics-Experiments were repeated at least twice, and results are given as mean Ϯ S.E. Tests for statistical significance were performed with the two-tailed Student's t test according to Ref. 32 or performed with Sigmaplot 6.0 (SPSS Inc., Erkrath, Germany). In the figures, levels of significance are given by asterisks: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.

RESULTS
Current Characteristics of Mutated Channels-The expression of GIRK channels led to an increase in the background or basal currents when rinsed with HK solution (I HK ) at a membrane potential of Ϫ70 mV for all types tested. The application of 10 M acetylcholine evoked additional inward currents (I ACh ) when the m 2 ACh receptor was coexpressed. All GIRK constructs that we used (Fig. 1A) showed normal activation and desensitization properties when compared with GIRK wild type channels. However, the current response to HK and ACh was higher through truncated channels when compared with the respective wild type channels (Fig. 1B). The average AChinduced current in oocytes injected with 150 pg of GIRK2 ⌬356 cRNA was 2.8 Ϯ 0.4 A (n ϭ 11), whereas GIRK2 wild type-injected oocytes (1500 pg) reacted to ACh with currents of only 0.7 Ϯ 0.2 A. Even with low amounts of injected cRNAs, both the deletion mutant GIRK2 ⌬356 and the chimera GIRK2/1 356 showed, on the average, higher currents than GIRK2 wild type channels. An increased ratio of I ACh /I HK from 3.3 to about 7.8 for the deletion mutant GIRK2 ⌬356 and 8.1 for the chimera GIRK2/1 356 was observed. The difference in currents was less pronounced for GIRK1 F137S and GIRK1 ⌬363* channels; the truncated channels showed an increase in the current amplitude by a factor of about 4.5 and no change in the current ratios (Fig. 1B) when the same amount of cRNA was injected. Injection of oocytes with even high amounts of chimeric GIRK2/1/2 RNA gave only small GIRK currents and reduced the survival rate of oocytes.
Effect of Halothane on Currents through Homooligomeric Wild Type GIRK Channels-As exemplified in Fig. 2A, halothane was able to modulate the current through homooligomeric GIRK1 F137S channels in Xenopus oocytes, corroborating our previous results (14). Background currents were augmented by about 90% in the presence of 1 mM halothane, but the current induced by 10 M acetylcholine (I ACh ) was inhibited by low concentrations of halothane with maximum attenuation of about 50% at a concentration of 100 M halothane. When the halothane concentration was increased, the effect reversed, and rinsing the oocytes with 1 mM halothane led to an increase in I ACh (14).
To test whether the inhibition of ACh-induced currents by halothane is because of the disturbance of the binding of G␤␥ to FIG. 1. Constructs used for the investigation of GIRK channel sensitivity against halothane. A, the transmembrane regions TM1 and TM2 with the pore region (P) in between them are flanked by the N and C termini. GIRK1 (black) is characterized by a unique long distal C terminus. This C terminus (aa 364 -501) was transferred to a deletion mutant of GIRK2 (white) to yield GIRK2/ 1 356 . GIRK2/1/2 and GIRK2/1/2/1 are chimeric GIRK proteins with the transmembrane domains of GIRK1 F137S between aa 88 and aa 202 inserted into the core of GIRK2 and GIRK2/1 356 , respectively. B, basal currents (I HK , filled boxes) and agonist-induced currents (10 M acetylcholine, I ACh, empty boxes) mediated by different GIRK isoforms. Note the differences in amount of injected cRNAs. the channel, we performed pulldown experiments with GST⅐GIRK1 fusion proteins. These fusion proteins contained the N terminus (aa 1-84, CD1) or the C-terminal part of the channel (aa 183-501), thus including the putative G␤␥ binding sites that are believed to be responsible for the agonist-induced current activation (24). With CD1 the amount of pulled down [ 35 S]G␤␥ was not sufficient to perform a quantitative analysis. With CD2 the amount of G␤␥ that was pulled down was less than 10% of the total of labeled G␤␥ in the extract. Halothane (100 M, n ϭ 7; 300 M, n ϭ 6; and 1 mM, n ϭ 5) did not change the amount of bound G␤␥ (Fig. 2, B and C), thus giving no evidence for inhibition of G␤␥ binding to the channel by halothane.
Patch clamp experiments showed low basal activity of the channels in the absence of an agonist with a relative open probability of 4.0⅐10 Ϫ3 Ϯ 1.2⅐10 Ϫ3 (n ϭ 5). However, the open time distribution of detected events showed the existence of at least two populations of channel openings with a long and a short dwell time (Table I) (Fig. 2F). In addition the number of openings increased (Fig. 2D), which led to an increase in open probability by 852 Ϯ 368% (Fig. 2G). The single channel conductivity of 15.7 Ϯ 0.5 picosiemens as well as the inward rectification properties of the channel remained unchanged by halothane (Fig. 2, E and H).
Addition of 50 nM ACh to the pipette solution activated GIRK1 F137S channels in cell-attached patches. The relative open probability was increased more than 20-fold compared with channels in the absence of ACh (relative open channel probability 0.080 Ϯ 0.029, n ϭ 5). Again the current was mainly caused by brief openings with a duration of less than one ms (Table I and Fig. 3, A and B). Longer lasting events contributed to only 17 Ϯ 8% of the total number of events. The addition of 1 mM halothane decreased the total number of events and decreased the open probability to 39 Ϯ 23% of control (n ϭ 4). Short events were more affected, so that a shift to longer openings was observed. Longer openings contributed to 61 Ϯ 10% of the number of events in the presence of halothane  (Table I and Fig. 3, C and D), resulting in an increase in mean open time from 1.1 Ϯ 0.4 ms (n ϭ 5) to 6.6 Ϯ 2.4 ms (n ϭ 4, p Ͻ 0.05). GIRK2 WT channels have been shown to be most sensitive to the inhibitory action of halothane; activation of the basal current was never observed (13,14). Concentrations of more than 100 M halothane significantly reduced the current (Fig. 4A). In two-electrode voltage clamp experiments, the least square fitting of normalized current values to the Hill equation yielded a half-maximum effect of 143 M halothane on ACh-induced currents and a Hill coefficient of 1.3. In one of three single channel experiments, 1 mM halothane completely blocked channel openings induced by 50 nM ACh. In two more experiments the number of events was reduced to an extent that prevented meaningful statistical analysis. The open channel probability was reduced to 6.5 Ϯ 4.2% of control (n ϭ 3, Fig. 4C).
Effects of Halothane on Deletion Mutants of GIRK Channels-Because GIRK1 and GIRK2 channels reacted differently to halothane, we constructed deletion mutants that lacked the unique C terminus of GIRK1 and the corresponding C terminus of GIRK2 to investigate whether the respective pharmacological properties reside on the channel itself.
The halothane-induced activation of background currents that were mediated by truncated GIRK1 ⌬363* channels was more pronounced than with GIRK1 F137S currents (Fig. 5, A and  B). The background currents of GIRK1 ⌬363* were activated by 514 Ϯ 75% (n ϭ 21, p Ͻ 0.001) with 1 mM halothane. Lower concentrations were less effective, but still 300 M halothane  The agonist-induced currents of GIRK1 ⌬363* were similarly sensitive to inhibition by halothane when compared with GIRK1 F137S channels. The inhibition was most pronounced at 30 M halothane with a decrease in the current of 59 Ϯ 3% (n ϭ 4, p Ͻ 0.001) and less at higher concentrations of the anesthetic, where activation of the channel obviously overrides the inhibition (see Fig. 5A, application of 1 mM halothane, and 5C).
Single channel experiments of the halothane effect on GIRK1 ⌬363* yielded similar results compared with GIRK1 F137S channels. The basal current was stimulated by 1 mM halothane because of an increase in the number of channel openings and open time (data not shown). The ACh-induced current was mainly carried by brief channel openings. Only 14 Ϯ 10% of the opening events were characterized by a long time constant of 10.5 Ϯ 2.7 ms (n ϭ 7) (Fig. 5, D and E and Table I). The single channel conductivity was determined to be 14.5 Ϯ 0.18 picosiemens and, therefore, the same as for GIRK1 F137S channels. As expected from the two-electrode voltage clamp experiments, halothane at a concentration of 1 mM had no effect on the open probability of ACh-activated GIRK1 ⌬363* channels, but similar to GIRK1 F137S channels, it propagated channel openings with longer open times (Table I and Fig. 5, F and G).
Wild type GIRK2 channels never responded to halothane with activation. The truncation of the C terminus of GIRK2 at aa 356 (GIRK2 ⌬356 ) resulted in a channel protein that was activated by halothane in two-electrode voltage clamp experiments. Concentrations as low as 100 M halothane induced an increase in basal currents (Fig. 6, A and B). Higher concentrations of halothane were more effective, and 1 mM of the anesthetic augmented the background current by 178 Ϯ 39% (n ϭ 16, p Ͻ 0.001). In contrast to the wild type GIRK2 channels, I ACh mediated by truncated GIRK2 ⌬356 channels proved resist-ant against halothane inhibition. No reduction of the agonistinduced current was observed by halothane in concentrations of 10 M-1 mM when compared with control values (Fig. 6, C  and D).
As expected, patch clamp experiments showed no influence of halothane on ACh-evoked channel activity. The single channel conductivity was 30.0 Ϯ 1.0 picosiemens and therefore higher compared with the GIRK2 wild type channel with 21.7 Ϯ 3.9 picosiemens. Both channel types showed frequent openings that obviously did not Effect of Halothane on Chimeric GIRK Channels-To further investigate the role of the GIRK C terminus in halothaneinduced channel modulation, we transferred the unique GIRK1 C terminus to the GIRK2 ⌬356 channel. The sensitivity of the chimera against halothane was increased ϳ25-fold when compared with GIRK2 WT channels. I ACh through GIRK2/1 356 channels was nearly completely inhibited by 100 M halothane (Fig. 7, A and B), and the IC 50 value was found to be 5.5 M halothane with a Hill coefficient of 0.7. In contrast to GIRK1 F137S or GIRK2 ⌬356 channels, the basal current of the chimera was inhibited instead of being activated by halothane FIG. 4. GIRK2 differed in its pharmacological properties from GIRK1. A, time course of current through homooligomeric GIRK2 channels. Agonist-induced currents were more sensitive against halothane than currents through GIRK1 F137S channels. Basal currents were not activated by halothane. B, single channel activity was nearly completely suppressed by 1 mM halothane. C, relative changes in open probability due to halothane treatment. Control values are normalized to 100%.
Because of the lipophilic nature of halothane, the site of interaction with integral membrane proteins might lie within the transmembrane-spanning regions of the channel (33). We sought to test whether the halothane-induced activation of GIRK1 channels is a property that can be traced to a structure within these transmembrane domains or the pore region of the channel. Therefore, we substituted the region between aa 88 and aa 202 of the GIRK2 channel and the GIRK2/1 356 channel for the homologous region of GIRK1 F137S . This part of the channel comprises the transmembrane and the pore-forming domains of the channel. Both chimeras showed a clear increase in sensitivity against the inhibitory action of halothane (Fig. 7,  C and E). In neither case was activation of GIRK currents observed with halothane. The agonist-induced current I ACh was inhibited at an estimated IC 50 value of 10 M halothane for GIRK2/1/2 (Fig. 7D) and 300 nM halothane for GIRK2/1/2/1 (Fig. 7F). In the case of GIRK2/1/2/1-mediated currents, 10 M halothane was sufficient to completely abolish I ACh , and higher concentrations diminished the current value beneath basal values.
It has been postulated that halothane increases the affinity between G␣ i and G␤␥ (16). Thus, the inhibition of activated GIRK channels by halothane might be the result of a shift in the equilibrium distribution of G␤␥ from the effector (the activated channel) to the G protein ␣ subunit. If this is the case, the rate of current decay because of halothane inhibition should be comparable with the rate of current decay because of the deactivation by the wash out of the agonist. However, the time course of halothane-induced inhibition of I ACh was rapid compared with the decrease of the agonist-induced current due to the wash out of ACh. othane is thus about 6-fold more rapid than the deactivation of the current due to deprivation of the agonist. DISCUSSION From our previous work (14) it was not clear how halothane was able to inhibit agonist-activated GIRK currents with high potency but activate the background current with lower potency. A mechanism with two independent targets seemed most probable. An increase in G␤␥ affinity to the channel could have explained the activation by halothane, whereas disruption of the G protein-signaling cascade downstream of the receptor could have been a reason for the inhibition of agonistinduced currents. However, in light of our new findings, the existence of two such independent mechanisms for modulation of GIRK currents by halothane, i.e. one on the G protein and the other on the channel, seems unlikely. The inhibition of open channels by halothane cannot be assigned to the inhibition of the G protein, because it can be abolished by deletion of the C terminus of the channel and therefore has to be a property of the channel itself. Halothane does not physically block the pore, but it seems to interact with the gating mechanism of the channel. Obviously there are at least two gates that are affected: a fast one is inhibited; it promotes flickering like channel openings due to the activation of the G-protein, whereas another slow one is responsible for longer openings and can be activated by halothane.
In the absence of an agonist, halothane induced activity of GIRK1 F137S and GIRK1 ⌬363* channels. The number of opening events and the duration of the openings was increased (Figs. 2,  D and F, and 5F). An increase in the number of long events was also seen when halothane was applied to channels that had been activated by ACh. In addition, halothane decreased the number of short openings in the presence of the agonist (Fig.  3C), thus explaining how the anesthetic was able to reduce the agonist-induced current but activate the background current.
The fact that the sensitivity of GIRK channels to halothane was determined by the C terminus indicates that this part of the protein interacts with the G␤␥ binding sites or directly modulates the gating of the channel. Deleting this part of the channel abolished the halothane sensitivity of GIRK2 channels (Fig. 6C). On the other hand, introduction of the C terminus of GIRK1 into GIRK 2 further increased its sensitivity against halothane (Fig. 7A). The functional role of the C terminus in GIRK channels is still not clear. Although crystallography studies revealed the three-dimensional structure of inwardly rectifying K ϩ channels (19,20,34), the distal C terminus downstream residue 371 could not be incorporated into the threedimensional structure. However, it is assumed that this part of the molecule is not part of the cytoplasmic pore but protrudes into the cytoplasm and probably mediates protein-protein interactions. From functional studies we know that the C terminus of GIRK1 may physically block the channel because treatment with trypsin from the cytosolic side activates GIRK in a manner similar to ACh (35). The coexpression of the C-terminal tail of GIRK1 effectively inhibits channel activity in Xenopus oocytes (36); furthermore, a peptide of 20 amino acid residues derived from the very C-terminal end of GIRK1 is a potent and reversible blocker of GIRK channels (37). Interestingly, the deletion mutants of GIRK1 as well as GIRK2 lacking the Cterminal end that we used in this study did not show any obvious gating abnormalities. This would be expected, if this very part of the C terminus directly took part in the binding of G␤␥ and gating of the channel. The C-terminal deletion mutants were readily activated by ACh via the coexpressed m 2 receptors and deactivated rapidly after washout of the agonist. The only difference seen was a general higher level of currents in oocytes injected with the cRNA of the deletion mutants (Fig.  1B). However, in patch clamp experiments we did not observe a general higher open probability of single mutated channels.
In vitro synthesized, metabolically labeled G␤␥ has been successfully used in the past to identify specific G␤␥-binding sites of GIRK channels (24) or voltage-dependent Ca 2ϩ channels (31). Our experiments showed no displacement of bound G␤␥ to the GST⅐GIRK1 fusion protein in concentrations of up to 1 mM halothane. The fusion proteins were incubated in a solution with saturating G␤␥ concentrations (about 10-fold excess); thus, with this method it is not possible to detect a potential FIG. 7. Chimeric GIRK channels developed increased sensitivities against halothane. A, the time course of GIRK2/1 356 -mediated currents measured with the two-electrode voltage clamp technique. B, agonist-induced currents were effectively inhibited by halothane. Sensitivity against halothane was increased about 30-fold in the chimeric channel when compared with GIRK2 channels, the most sensitive of the wild type channels. C, GIRK2/1/2 channels were also inhibited by low concentrations of halothane. D, the IC 50 value was estimated to be about 10 M, 15 times less than for the GIRK2 WT channel. E, exchange of the C terminus of GIRK2/1/2 against that of GIRK1 yielded the GIRK2/ 1/2/1 channel. The current mediated by this channel showed further increased sensitivity against halothane. A concentration of 300 nM halothane was sufficient to reduce the current by about 50% as shown in panel F. increase in G␤␥ affinity to the channel protein. In contrast, a potential decrease in G␤␥ affinity would have been detected because the bound complexes were washed with halothanecontaining solutions in the absence of G␤␥. Any decrease in affinity because of halothane would have reduced the amount of bound G␤␥. Even if we take into account that there is more than one binding site for G␤␥ in the C terminus of GIRK1 that could obscure a possible halothane-induced shift of binding from one site to another, it seems unlikely that washing the complexes in the absence of free G␤␥ would not reduce the total amount of bound protein. Therefore, we conclude that halothane does not interrupt the G protein-signaling on the level of G␤␥ binding to the channel.
The C terminus of GIRK, which plays a crucial role in mediating the anesthetic effect of halothane, is rather hydrophilic and therefore thought to protrude into the cytoplasm (19). Judged from this property one would not expect strong interaction of hydrophobic substances like halothane with the C terminus of GIRK channels directly. However, it is impossible to predict binding of anesthetics to proteins from the amino acid sequence alone (38). Our findings show a strong dependence of the halothane action on the C terminus of the GIRK channel, and therefore a direct interaction of halothane with this part of the protein seems possible. Still, other mechanisms, such as so-far unknown interactions with other proteins mediated by the C terminus, may play a role in the observed halothane effects.