The γ Subunit Determines Whether Anesthetic-induced Leftward Shift Is Altered by a Mutation at α¹S270 in α¹β²γ^2LGABA^AReceptors

cDNAs for the α¹, β², and γ^2Lsubunits of human GABA^AR in pCDM8 vectors were provided by Dr. Paul J. Whiting (Merck Sharp & Dohme Research Labs, Essex, United Kingdom). The S270I mutation was introduced into wild-type α¹cDNA by high-fidelity polymerase chain reaction oligonucleotide-directed mutagenesis. The presence of the mutation and absence of stray mutations in the cDNA were confirmed by dideoxynucleotide sequencing.

mRNAs were transcribed in vitro  from linearized cDNA templates using T7 RNA polymerase (Ambion, Austin, TX). mRNAs were isolated from the transcription reaction using affinity beads (BIO-101, Vista, CA) and stored at −80°C.

Wild-type and mutant GABA^ARs consisting of α¹β²γ^2L, α¹β²(γ-less), or α¹γ^2L(β-less) subunit mixtures were expressed in Xenopus laevis  oocytes. Detailed methods of oocyte preparation and injection have been described previously. 15Animal maintenance and oocyte harvest procedures were approved by the Massachusetts General Hospital Committee on Animal Research. Selected stage IV and V oocytes were injected with mixtures of GABA^AR subunit mRNAs (total, 20–100 ng in 50–75 nl) in ratios of 1α:1β:1γ, 1α:1β, or 1α:1γ. Oocytes were incubated at 17°C and used for electrophysiologic experiments on days 1–3 after injection.

All experiments were performed at room temperature (20–22°C). Oocytes were placed in a 0.1-ml flow chamber and rested in a small depression in the recording chamber with the animal pole facing upward. They were impaled with two capillary glass electrodes with open tip resistances of 0.2–2 MΩ containing 3 m KCl and were voltage clamped (model OC-725; Warner Instrument Corp., Hamden, CT) at −50 mV. Extracellular solution containing 96 mm NaCl, 2 mm KCl, 1 mm CaCl², 0.8 mm MgCl², 10 mm HEPES, pH 7.6, was used to continuously superfuse oocytes at 2 ml/min.

Drugs were directly added to extracellular solution shortly before the experiments. GABA alone or in combination with volatile anesthetic was applied for 20–50 s to elicit a peak current response, and a 5–15-min interval was used between GABA applications for recovery of channel activity from desensitization. Anesthetics were not preapplied to oocytes before coapplication with the agonist. Currents were digitized at 50–200 Hz, recorded on a personal computer, and analyzed offline.

Effects of anesthetics on GABA-activated currents (0.03 μm to 1 mm GABA) were established in individual oocytes. Responses to low GABA concentrations were alternated with control measurements at 1 mm GABA (a maximal-activating concentration) to control for run-up or run-down of currents. GABA concentration responses in the presence of 0.75 mm isoflurane or 0.85 mm halothane were studied in the same oocyte, again using 1 mm GABA (without anesthetic) as a control. For each condition (wild-type or mutant form of the α¹β²γ^2Lor α¹β²GABA^AR), four to eight individual oocytes were studied.

For EC⁵studies, the GABA concentration eliciting a peak current of 5% of control peak at 1 mm GABA was determined for individual oocytes, and responses to the EC⁵concentration with and without anesthetic were measured in triplicate. Each measurement was paired with a control measurement using 1 mm GABA. Data were obtained from oocytes isolated from at least two different frogs.

Before use, halothane (Halocarbon Laboratories, River Edge, NJ) was passed over an activated aluminum oxide (Aldrich Chemical, Milwaukee, WI) column to remove preservatives. Saturated solutions of halothane or isoflurane (Ohmeda, Liberty Corner, NJ) were prepared by adding a surplus of the volatile anesthetic to extracellular solution and by stirring in airtight glass bottles for at least 3 h before use. The concentrations of these stocks were taken to be 15 mm for isoflurane and 17 mm for halothane, as determined by Firestone et al . 16Based on Bunsen water–gas partition coefficients of 1.08 for isoflurane and of 1.2 for halothane at 25°C, concentrations of 0.48 mm isoflurane or 0.54 mm halothane correspond to approximately 1 vol%. 16Saturated stocks were diluted immediately before use and transferred to a glass syringe reservoir. All parts of the oocyte superfusion system consist of glass or polytetrafluoroethylene, and the oocyte was positioned within 1 mm of the end of the tubing to minimize adsorptive and evaporative loss of anesthetics. Gas chromatographic measurements have shown that the loss of volatile anesthetics is less than 15% in this system. 17 

GABA-activated whole oocyte currents were normalized to the average of the preceding and following control currents. For concentration–response experiments, normalized responses from individual oocytes were fitted with logistic equations of the form: where I^maxis the maximum current, EC⁵⁰is the GABA concentration eliciting a half-maximal response, and nH is the Hill coefficient of activation. Nonlinear least-squares fits were performed using Origin software (Microcal Inc., Northampton, MA). Fitted values are reported as mean ± SD.

For leftward shift analysis, the ratios of fitted EC⁵⁰values in the absence of anesthetic to the EC⁵⁰values in the presence of anesthetic (EC⁵⁰0/EC⁵⁰Anes) were calculated for individual oocytes from logistic fits. To compare volatile anesthetic effects in wild-type versus  mutant GABA^ARs, two-tailed Student t  tests were used to compare sets of EC⁵⁰ratios (leftward shift) and nH values. Data from oocytes expressing the α¹β²γ^2Land α¹β²subunits were analyzed independently.

Among oocytes expressing α¹β²γ^2LGABA^ARs, EC⁵GABA concentrations ranged from 2 to 9 μm. This variation necessitated evaluation of anesthetic effects at EC⁵in individual oocytes. At GABA EC⁵, both isoflurane and halothane markedly and reversibly enhanced currents (figs. 2A and 2C). Enhancement ratios in the presence of isoflurane were 8.0 ± 1.3 at approximately 2 minimum alveolar concentration (MAC) and 4.6 ± 0.7 at approximately 1 MAC. Similar enhancement ratios were observed with halothane (7.8 ± 0.9 at approximately 2 MAC and 4.7 ± 0.4 at approximately 1 MAC). In the absence of GABA, no direct activation of currents was observed at these anesthetic concentrations.

GABA concentration–response relations were measured in the absence and presence of a fixed concentration of isoflurane or halothane. Concentrations near 2 MAC were chosen for each drug because a pronounced leftward shift could be observed, improving sensitivity to mutant associated changes.

In individual oocytes expressing wild-type α¹β²γ^2LGABA^ARs, currents elicited by different concentrations of GABA were fit with EC⁵⁰values ranging from approximately 40 to 70 μm, and nH values ranged from 1.3 to 1.5 (table 1). Because of the variation of EC⁵⁰values among different oocytes, effects of anesthetics on individual cells were assessed. When isoflurane or halothane was added to GABA solutions, a leftward shift of the concentration–response curves was observed (isoflurane, fig. 3A; halothane, fig. 3C). The average ratio of control EC⁵⁰without anesthetic to that in the presence of isoflurane was 3.4 ± 0.7 (n = 4;table 1) and to that in the presence of halothane was 2.9 ± 0.6 (n = 4;table 1). In the presence of halothane or isoflurane, activation nH values were reduced (to 1.0 ± 0.05 with isoflurane and to 1.1 ± 0.07 with halothane, n = 4;table 1). These values were significantly less than the control nH (P ≤ 0.01 for both anesthetics).

Both isoflurane and halothane enhanced currents at all GABA concentrations, and the enhancement decreased with increasing GABA concentration. The maximal peak current at 1 mm GABA increased as much as 15% (isoflurane, figs. 3A and 4A; halothane, figs. 3C and 4C). At 10 mm GABA, peak currents were slightly higher than those elicited at 1 mm, and volatile anesthetic enhancement was still present. After discontinuation of stimulation with high GABA in the presence of anesthetic, a brief rebound increase in current occurred, reflecting recovery of channels from anesthetic inhibition during deactivation (figs. 4A and 4C, center traces).

Incorporation of the α¹(S270I) mutation in α¹β²γ^2LGABA^ARs resulted in significant changes in the behavior of GABA-induced currents. The mutation rendered GABA^ARs more sensitive to GABA, with average EC⁵⁰values of the mutant channels approximately fourfold lower than wild-type (table 1). EC⁵⁰values varied markedly among individual oocytes (table 1). In addition, the nH values were significantly lower in α¹(S270I)β²γ^2Lreceptors (0.9 ± 0.1, n = 8) compared with those of wild-type receptors.

With EC⁵GABA (0.1–0.9 μm) activation of α¹(S270I)β²γ^2Lreceptors, both isoflurane and halothane enhanced currents, but the degree of enhancement was significantly less than that observed in wild-type α¹β²γ^2Lreceptors (P ≤ 0.01;figs. 2B and 2D). Anesthetic effects were reversible after wash, and no direct activation by anesthetics was observed in the absence of GABA.

With addition of volatile anesthetics, GABA concentration–response curves measured in α¹(S270I)β²γ^2Lreceptors were consistently shifted leftward (figs. 3B and 3D). Although the magnitude of the leftward shifts appeared smaller than that observed in wild-type GABA^ARs, the EC⁵⁰0/EC⁵⁰Anes ratios derived from logistic fits were statistically indistinguishable from wild-type values using both isoflurane (P = 0.2;table 1) and halothane (P = 0.8). nH values were not altered by the anesthetics. At high GABA concentrations, both drugs slightly reduced the peak currents up to 5% (figs. 3B, 3D, 4B, and 4D). Again, after removal of GABA and anesthetic, rebound currents were observed (figs. 4B and 4D, center traces).

Leftward shift analysis revealed previously unknown details about the impact of the α¹(S270I) mutation on anesthetic modulation in α¹β²γ^2LGABA^ARs. We extended the leftward shift analysis to both α¹β²(γ-less) and α¹γ^2L(β-less) GABA^ARs in oocytes containing the α¹(S270I) mutation to see if non-α subunits influenced our results.

Oocytes (n = 31 for wild-type and mutant channels) injected with mRNA for α¹and γ^2Lsubunits produced no measurable currents within 96 h. Connor et al.  18also reported that the GABA^AR α¹subunit requires a β²subunit for receptor assembly and functional surface expression in Xenopus  oocytes. The α¹β²combination produced functional GABA^ARs.

The gating behavior of wild-type α¹β²GABA^ARs was distinct from that of α¹β²γ^2Lreceptors in that the average EC⁵⁰was nearly threefold lower (18 vs.  52 μm;table 2), and the average nH was reduced (0.9 vs.  1.4). GABA concentration–response studies in oocytes expressing α¹β²GABA^ARs in the absence and presence of 0.75 mm isoflurane showed large leftward shifts (fig. 5A). The average EC⁵⁰0/EC⁵⁰Iso ratio (8.4 ± 1.4, n = 3;table 2) was much larger than that for wild-type α¹β²γ^2Lreceptors (3.4 ± 0.7, n = 4;table 1). The low nH was not altered by isoflurane. Currents at all GABA concentrations were enhanced by isoflurane, and the effect was more pronounced at low GABA concentrations (fig. 5A). Rebound currents were observed at the termination of anesthetic application at high GABA concentrations (fig. 5E).

Compared with wild-type α¹β²receptors, α¹(S270I)β²receptors showed a sixfold decrease in the average GABA EC⁵⁰and significantly lower nH values (table 2). However, isoflurane caused only very small leftward shifts in GABA concentration–response curves, with an average EC⁵⁰0/EC⁵⁰Iso that was not significantly different from 1.0 (1.6 ± 0.6). At GABA concentrations greater than 3 μm, the dominant effect of isoflurane on α¹(S270I)β²receptors was inhibition, whereas only weak enhancement was observed at GABA concentrations less than 1 μm. Evidence of isoflurane inhibition was present in the form of rebound currents at all GABA concentrations (figs. 5D and 5F). Anesthetic effects were fully reversible after wash.

Investigation of the α¹(S270I) mutation in α¹β²γ^2LGABA^ARs demonstrates several important findings that point to a complex role for the α¹S270 residue. Our data show that the α¹(S270I) mutation causes significant changes in GABA^AR gating behavior and sensitivity to the inhibitory effects of volatile anesthetics. In contrast to results obtained at single low GABA concentrations (EC⁵), the analysis of GABA concentration–response curves suggests that the α¹(S270I) mutation causes almost no change in sensitivity to volatile anesthetic–induced leftward shift of α¹β²γ^2LGABA^ARs. However, this finding is not simply a result of using a different analytic methodology, because the α¹(S270I) mutation profoundly affects anesthetic-induced enhancement in α¹β²GABA^ARs. Indeed, we found that GABA^AR sensitivity to anesthetic modulation and the impact of the α¹(S270I) mutation are significantly influenced by the γ subunit.

In comparing the GABA-dependent function of wild-type recombinant GABA^ARs with those containing the α¹(S270I) mutation, we observed several consistent changes. Receptors containing the α¹(S270I) mutation had lower GABA EC⁵⁰values, in agreement with previous reports. 13,14In α¹β²γ^2LGABA^ARs, the mutation reduced the average EC⁵⁰approximately fourfold (table 1), whereas in α¹β²GABA^ARs, the mutation reduced average EC⁵⁰sixfold (table 2). Because of the variability of EC⁵⁰values derived from studies on different oocytes, these two ratios are not significantly different, indicating that the mutation has about the same impact on gating whether or not a γ subunit is present in the receptors. On a molecular level, reduced EC⁵⁰can be caused by higher affinity of GABA^ARs for GABA (e.g. , by slowing GABA unbinding) 19or by stabilization of the active open-channel state by the mutation.

A second effect of the α¹(S270I) mutation on gating is to reduce the nH of GABA activation. In α¹β²γ^2LGABA^ARs, the mutation reduced the average nH from 1.4 to 0.9 (table 1), whereas in α¹β²GABA^ARs, the mutation reduced average nH from 0.9 to 0.6 (table 2). The reduction of nH in whole oocyte experiments could be caused by an increase in receptor desensitization rates, which cannot be measured using these methods, or the result of a decrease in cooperativity in GABA-dependent gating.

Third, both α¹(S270I)β²γ^2Land α¹(S270I)β²GABA^ARs are inhibited by volatile anesthetics in the presence of high GABA, whereas the corresponding wild-type receptors are not. In traces from oocytes expressing mutant receptors, rebound currents observed after discontinuation of the superfusion of GABA plus anesthetics were larger than those in wild-type currents (figs. 4 and 5). Rebound currents may represent reopening of anesthetic-blocked open channels or recovery of anesthetic-stabilized, agonist-bound closed states (e.g. , desensitized) via  the open state. These results suggest that the α¹(S270I) mutation sensitizes activated GABA^ARs to inhibition by volatile anesthetics, accounting for up to a 20% difference in anesthetic effects at maximal GABA concentrations.

We examined the effects of volatile anesthetics on GABA-induced gating enhancement in both wild-type α¹β²γ^2Land mutant α¹(S270I)β²γ^2LGABA^ARs. Results with both isoflurane and halothane at approximately equipotent concentrations were essentially the same, suggesting a common mechanism for GABA^AR enhancement by halogenated alkane and ether anesthetics. We analyzed anesthetic enhancement and the impact of the α¹(S270I) mutation using two different approaches. One approach to assess enhancement, which has been used by a number of researchers, is to measure anesthetic-induced change in currents at a single, low, equipotent GABA concentration (e.g. , EC⁵). At GABA EC⁵, enhancement of currents by anesthetics was significantly lower in oocytes expressing mutant versus  wild-type channels (fig. 2). Thus, based on EC⁵results alone, the α¹(S270I) mutation apparently reduces, but does not abolish, anesthetic enhancement in α¹β²γ^2LGABA^ARs, as reported in previous studies of α¹β¹, α²β¹, or α²β¹γ^2LGABA^ARs. 12,13 

Our second approach to quantifying anesthetic enhancement of GABA^AR function was leftward shift analysis. Compared with studies using a single “equipotent” GABA concentration, leftward shift analysis requires accumulation of significantly more data, but it is more informative in assessing both inhibiting and enhancing anesthetic actions and how they are affected by mutations. For example, the reduced anesthetic enhancement observed in α¹(S270I)β²γ^2LGABA^ARs at EC⁵might be explained by a number of changes associated with the mutation: (1) weakened anesthetic enhancement; (2) increased anesthetic inhibition; or (3) reduced cooperativity of gating. Characterization of receptors containing the α¹(S270I) mutation over a wide range of GABA concentrations showed that both GABA gating and sensitivity to anesthetic inhibition were, in fact, altered.

Unexpectedly, the EC⁵⁰0/EC⁵⁰Anes ratios were the same in oocytes expressing wild-type α¹β²γ^2Land α¹(S270I)β²γ^2Lchannels (fig. 4and table 1). On casual inspection, data in figure 4appear to show smaller anesthetic-induced leftward shifts in α¹(S270I)β²γ^2Lthan in α¹β²γ^2Lreceptors, but this appearance is actually the result of the mutant-associated change in anesthetic inhibition. After renormalization to correct for the different inhibitory effects of anesthetics in wild-type and mutant receptors at high GABA concentrations, the similarity of the isoflurane-induced leftward shifts becomes apparent (fig. 6, horizontal arrows). Thus, based on leftward shift analysis, the α¹(S270I) mutation does not alter anesthetic-induced enhancement of α¹β²γ^2LGABA^ARs.

These leftward shift results in α¹β²γ^2LGABA^Aare not in direct conflict with previous reports of α(S270I) mutations, 12–14because this is the first study using this subunit combination. Indeed, other investigators have reported that subunit composition can significantly impact anesthetic modulation of GABA^ARs. Anesthetic actions in GABA^ARs have been shown to be dependent on the β-subunit isoform 20–22as well as on receptor subunit mixture. 23We therefore experimentally addressed whether the impact of the α¹(S270I) mutation on GABA^AR modulation by anesthetics could be dependent on non-α subunits.

In GABA^ARs expressing only α¹and β²subunits, isoflurane at approximately 2 MAC caused very strong enhancement at EC⁵as well as a large leftward shift in GABA concentration–response curves. Introduction of the α¹(S270I) mutation in these receptors nearly eliminated isoflurane-induced enhancement by either measure (fig. 5and table 2). Thus, for α¹β²GABA^ARs, both EC⁵and leftward shift analyses of our data agree completely with previous studies of the α¹(S270I) mutation. 12,13 

Our results demonstrate that incorporation of the γ subunit into wild-type GABA^ARs reduces isoflurane-induced left shift at 2 MAC isoflurane by almost a factor of three (8.4-fold in α¹β²vs.  3.4-fold in α¹β²γ^2L;tables 1 and 2). Furthermore, the presence of the γ subunit nearly nullifies the effect of the α¹(S270I) mutation on anesthetic modulation, as shown by the contrasting results of our studies on the α¹(S270I) mutation in α¹β²γ^2Lversus α¹β²GABA^ARs. Other investigators previously noted that GABA^ARs containing the γ subunit show less anesthetic enhancement than those formed of only α- and β-subunit types. 24In addition, incorporation of either δ or ε subunits as replacement for γ further suppresses anesthetic modulation. 25,26 

On the other hand, several reports have suggested that the γ subunit plays a negligible role in anesthetic actions. For example, the γ subunit apparently has little impact on GABA^AR enhancement by propofol. 27,28Two studies have suggested a minor role for the γ subunit in the presence of αS270 mutations. The α²(S270H) mutation reduced isoflurane enhancement in both α²β¹and α²β¹γ^2SGABA^ARs at low GABA concentrations. 29In GABA^ARs containing α²and β¹subunits, the α²(S270I) mutation reduced ethanol enhancement at low GABA concentrations, and addition of the γ^2Lsubunit slightly reduced the effects of the mutation. Moreover, the homologous S280I mutation in the γ^2Lsubunit TM2 domain only weakly altered ethanol sensitivity in α²β¹γ^2LGABA^ARs. 13It is noteworthy that these studies drew their conclusions based on results obtained at single low GABA concentrations, because our EC⁵results in wild-type and mutant α¹β²γ^2LGABA^Aalso show a significant impact of the α¹(S270I) mutation (fig. 2). It is only by quantifying anesthetic-induced leftward shifts that we demonstrated the magnitude of the influence of the γ subunit.

From studies of GABA^AR mutants, Mihic et al.  12proposed that anesthetics bind at an intrasubunit pocket formed by residues on TM2 and TM3. Other investigators have extended this model by suggesting that the α¹S270 sidechain makes direct contact with anesthetic molecules, and that when the size of this sidechain is increased, it occupies the space where anesthetics bind, mimicking anesthetic-induced gating changes. 14Evidence for interactions between anesthetics and the α¹S270 sidechain is based on cysteine mutagenesis and modification studies. 30Notably, incorporation of a photo-activatable general anesthetic into nicotinic receptors from Torpedo , which are close structural homologs of GABA^ARs, also identify a residue at the outer end of a TM2 domain. 31Our studies partially conform to the idea that α¹S270 contributes to an anesthetic binding site, because the decreased EC⁵⁰and nH caused by the α¹(S270I) mutation are similar to those observed in the presence of anesthetics. However, a detail that emerges from our data is that changes in gating behavior associated with the α¹(S270I) mutation are unaffected by the γ subunit, whereas changes in anesthetic-induced leftward shift caused by the mutation in α¹β²receptors are nearly abolished by incorporation of the γ subunit. This dissociation of two effects of the mutation indicate that, although α¹S270 influences anesthetic modulation of GABA^ARs, it is unlikely to be in direct contact with anesthetic molecules, at least when the γ subunit is present.

Previous studies have emphasized the roles of α and β subunits in channel gating and anesthetic modulation of GABA^ARs, whereas our data show that incorporation of the γ subunit in wild-type GABA^Asignificantly influences EC⁵⁰, activation nH, and anesthetic-induced leftward shift, while nullifying the impact of the α¹(S270I) mutation on anesthetic modulation. Thus, agonist gating and anesthetic enhancement are significantly influenced by all three types of subunits in the dominant GABA^AR isoform from mammalian brain. Furthermore, changes caused by mutation in one subunit (α¹) can be modified via  interactions with another subunit (γ^2L). Our results do not rule out an intrasubunit site for anesthetics, but we speculate that volatile anesthetics interact with sites formed at GABA^AR subunit interfaces, causing their effects by altering the coordinated interactions between subunits that are necessary for agonist-activated gating. Indeed, other families of drugs that influence receptor gating, such as GABA receptor agonists and benzodiazepine compounds, bind at the interfaces between GABA^AR subunits. 32This model may explain how subunit composition and mutations influence both GABA^AR gating behavior and anesthetic modulation.

The authors thank Carol Gelb, M.A. (Senior Laboratory Technician, Department of Anesthesia and Critical Care, Massachusetts General Hospital, Boston, MA), for expert technical assistance, and Paul J. Whiting, Ph.D. (Merck Sharp & Dohme Research Labs, Essex, United Kingdom), for generously supplying the cDNAs for the GABA^AR subunits. We greatly appreciate the helpful advice and comments from Keith W. Miller, D.Phil. (Professor, Department of Anesthesia and Critical Care, Massachusetts General Hospital and Harvard Medical School, Boston, MA).