Influence of GABA^AReceptor γ²Splice Variants on Receptor Kinetics and Isoflurane Modulation

Human embryonic kidney 293 cells (American Type Culture Collection CRL 1573, Manassas, VA) were cultured in minimum essential medium with l-glutamine, supplemented with minimum essential medium amino acids solution (0.1 mm), sodium pyruvate (1 mm), 1% (volume/volume) streptomycin, and 10% fetal bovine serum (Harlan, Indianapolis, IN). Cells were grown at 37°C in a humidified 5% CO²–95% air atmosphere and subcultured twice weekly using trypsin (0.25%)–EDTA (1 mm). Twenty-four to 72 h before transfection, cells were plated on 60-mm culture dishes. Complementary DNAs (cDNAs) encoding the rat GABA^Areceptor subunits α¹(“flag”epitope tagged38), β², γ^2L, γ^2S, α⁵, and β³were inserted into the multiple cloning site of the mammalian expression vector pCEP4 (Invitrogen, Carlsbad, CA) as was the cDNA for enhanced green fluorescent protein (EGFP) from pEGFP-N1 (BD Biosciences Clontech, Palo Alto, CA). After reaching 80–90% confluence, cells were cotransfected with α¹β²(1:1), α¹β²γ^2L(1:1:10), α¹β²γ^2S(1:1:10), α⁵β³(1:1), or α⁵β³γ^2S(1:1:10) subunits using Lipofectamine 2000 (Gibco Life Technologies, Inc., Grand Island, NY). Overexpression of the γ²subunits was used to bias expression toward receptors containing all three subunits.39,40cDNA quantities were adjusted to obtain maximum peak currents of 200 pA to 8 nA. Usually, 250–500 ng α¹, β², α⁵, or β³and 2.5–5 μg γ^2Sor γ^2LcDNA was cotransfected with 180 ng EGFP cDNA. One to 3 h before performing electrophysiologic recordings, cells were replated onto 12-mm glass coverslips (Fisher Scientific, Pittsburgh, PA).

Experiments were performed at room temperature (22°–24°C) on the stage of an upright microscope (BX50WI; Olympus, Melville, NY) equipped with a long-working distance water-immersion objective (Achroplan 40×; 0.75 numerical aperture; Carl Zeiss, Thornwood, NY) and differential interference contrast (Nomarski) optics. For visual identification of individual transfected cells, we used a bead immunolabeling technique.41Dynabeads (Dynal, Oslo, Norway) were coated with sheep anti-rat antibodies, incubated with anti-flag rat antibodies directed against an N-terminus “flag” epitope fused to the α¹subunit, and added to the culture dish at a 1:500 dilution. For some experiments, cells that had been cotransfected with EGFP were also visualized using a mercury arc lamp (Mercury 100; Chiu Technical Corp., Kings Park, NY) and an EGFP filter set (Chroma Technology Corp., Rockingham, VT). Whole cell recordings were made from only the smallest cells to maximize mechanical stability and to minimize solution exchange time. After gaining stable whole cell access, negative pressure was applied, and the cell was lifted from the coverslip and positioned in front of the rapid application pipette.

Pipettes were fabricated from borosilicate glass (1.7-mm OD, 1.1-mm ID; KG-33; Garner Glass, Claremont, CA) using a two-stage puller (Flaming-Brown model P-87; Sutter Instruments, Novato, CA) and coated with Sylgard 184 (Dow Corning Company, Midland, MI) to reduce pipette capacitance. Pipette tips were fire polished and had open-tip resistances of 2–5 MΩ when filled with recording solution.

All whole cell recordings were obtained at a holding potential of −40 mV using an Axopatch 200B patch clamp amplifier (Axon Instruments, Union City, CA) and pClamp 8.0 software (Axon Instruments). Access resistances were typically 4–11 MΩ and were compensated by 75–90%. Access resistance and cell capacitance were monitored throughout the course of the experiment. Recordings were terminated if these became unstable (> 15% change). Data were low-pass filtered at 5 kHz using internal amplifier circuitry, sampled at 10 kHz (Digidata 1200; Axon Instruments), and stored on the hard disk of a Pentium-based computer.

Whole cells lifted from the coverslip were exposed to solutions using a two-barrel “theta” application pipette mounted to a piezoelectric stacked translator (model P-245.50; Physik Instrumente, Costa Mesa, CA). The application pipette was fabricated from “thin theta” glass (Sutter Instruments, Novato, CA) and connected to solution reservoirs using polyimide and Teflon tubing (Cole Parmer Instruments, Vernon Hills, IL). The voltage input to the high-voltage amplifier (model P-270; Physik Instrumente) used to drive the stacked translator was filtered at 100 Hz using an 8-pole Bessel filter (model 902LPF; Frequency devices, Haverhill, MA) to reduce oscillations arising from rapid acceleration of the pipette. Solution exchange rates were estimated by measuring open-tip junction currents with a diluted solution at the end of each experiment. The exchange speed was adjusted by changing the height of the reservoirs so that, typically, open-tip exchange times of 500 μs (10–90% rise time) were achieved. For whole cells positioned near the interface of the two flowing solutions (approximately 100 μm from the tip of the pipette), solution exchange rates were slightly slower (approximately 2 ms) than indicated by the open-tip rate.42 

γ-Aminobutyric acid–evoked currents were elicited by stepping between a “control” extracellular solution and one containing GABA (1 mm). To measure the effects of isoflurane on GABA-evoked responses, both solutions flowing from the application pipette were switched to solutions containing isoflurane using low-volume, manually controlled Teflon valves (model 1126; Omnifit Limited, Cambridge, United Kingdom). Cells were typically exposed to isoflurane for longer than 1 min before responses to GABA plus isoflurane were obtained.

The recording chamber was perfused continuously with HEPES-buffered “extracellular saline” consisting of 130 mm NaCl, 3.1 mm KCl, 10.9 mm Na-HEPES, 1.44 mm MgCl², and 2.17 mm CaCl²at a pH of 7.3. The same standard saline served as “control” solution in the rapid-application pipette. Patch pipettes were filled with 140 mm CsCl, 10 mm Na-HEPES, 10 mm EGTA, 5 mm QX-314, and 2 mm MgATP at a pH of 7.2 and 280–290 mOsm. GABA was dissolved in extracellular saline to obtain a 1 mm solution. GABA solutions were prepared daily from powder and not used for longer than 5 h. ZnCl²was prepared as a 10 mm stock solution in H²O and diluted with extracellular saline to achieve the desired concentration.

Isoflurane solutions were prepared from saturated stock solutions (15 mm) in extracellular saline and diluted to the final concentration in gas-tight Teflon bags (Chemware, Portage, WI). Concentrations in the reservoir were confirmed using gas chromatography. All tubing and connectors were made of Teflon to minimize loss of the anesthetic. Based on previous data, the minimal alveolar concentration for isoflurane in the rat corresponds to an aqueous concentration of 0.31 mm.10,43 

All chemicals were obtained from Sigma (St. Louis, MO). Cell culture reagents were from Gibco (Life Technologies, Inc.) unless indicated otherwise. Distilled water was used to prepare all solutions.

Data were analyzed using ClampFit 8.0 (Axon Instruments, Union City, CA), Origin 6.1 (MicroCal, Northampton, MA), Excel (Microsoft, Redmond, WA), and Prism 3.0 (GraphPad, San Diego, CA). Decay kinetics (deactivation and desensitization) were analyzed by fitting current traces to the exponential function y = y⁰+Σ Aⁿexp[−t/τⁿ], where y⁰is the steady state current during desensitization (0 for deactivation), and Aⁿand τⁿare the amplitude and the time constant of the nth component of a multiexponential fit. Goodness of fit was evaluated by visual inspection. Percent contribution of each component (Aⁿ%) was calculated as Aⁿ/ΣAⁿ× 100. In most cases, deactivation and desensitization were well fitted by biexponential functions. However, some of the responses were best fitted with only one or more than two time constants. To facilitate comparison of responses with differing numbers of decay constants, we calculated a weighted time constant τ^wt=Σ(Aⁿτⁿ)/ΣAⁿ. To analyze the effect of isoflurane on deactivation and desensitization kinetics, currents were normalized to the peak amplitude and calculated as the ratio τ^wt(iso)/τ^wt(ctrl) with at least three to five traces averaged for each. Isoflurane effects on net charge transfer were determined by integrating the area under the curve of the averaged raw current traces under control conditions and in the presence of isoflurane.

Results are reported as mean ± SD. Statistical comparisons of decay rates and peak amplitudes between data sets were made using the Student t  test or one-way analysis of variance followed by the Tukey post hoc  test, as appropriate. Differences were considered significant at P < 0.05.

We first examined the effect of the γ²subunit itself on macroscopic kinetics. We showed previously that when the γ^2Ssubunit is coexpressed with the α¹and β²subunits, deactivation after a brief pulse of GABA is accelerated, and desensitization, the decline in current that occurs during the continued presence of a high concentration of GABA, is reduced.39To determine whether the γ^2Ssubunit causes these same effects when it is coexpressed with different α and β subunits and whether the γ^2Land γ^2Ssubunits produce similar effects, we expressed γ^2Swith α⁵and β³as well as α¹and β², and expressed both γ^2Sand γ^2Lwith α¹and β²subunits.

Examples of responses to brief pulses (10 ms) and to prolonged application (2 s) of a high concentration of GABA (1 mm) are shown in figure 1. The results of fitting current deactivation and desensitization to exponential functions are presented in tables 1 and 2. As observed previously by other investigators, deactivation and desensitization were in most cases fitted best by biexponential functions. Similar to our previous results, coexpression of γ^2Swith α¹and β²accelerated deactivation (fig. 1Aand table 1) and reduced desensitization by causing either a loss or a significant reduction of the faster component of desensitization so that a large portion of γ^2S-containing receptors exhibited monophasic desensitization (fig. 1Band table 2). This also held true for coexpression of γ^2Swith α⁵and β³subunits. In contrast, coexpression of γ^2Lwith α¹and β²did not significantly accelerate deactivation, whether after a brief (fig. 1A) or a long (fig. 1B) pulse of GABA (table 1; P > 0.05). This was not due to lack of incorporation of the γ^2Lsubunit, because these receptors were insensitive to block by ZnCl²(100 μm)44(fig. 2). Like γ^2S, coexpression of γ^2Lwith α¹and β²subunits reduced the rate and extent of desensitization (table 2). Deactivation was significantly slower for α⁵β³than α¹β²receptors, whether GABA was applied for 10 ms (Student t  test, P < 0.001) or 2 s (P < 0.001). Therefore, the effects of γ^2Sand γ^2Ldiffered for deactivation but not desensitization, and the identity of α and β subunits did not seem to be a determining factor for the impact of γ^2Son deactivation or desensitization.

Similar to its effects on IPSCs,8–10,13isoflurane prolonged the decay and decreased the amplitude of currents evoked by brief pulses of GABA (1 mm, 10 ms). These effects were concentration dependent, were fully reversible, and were seen for all subunit combinations tested (figs. 3 and 4). However, there were striking differences in the degree of changes observed, depending on the presence or absence of a γ²subunit and on the γ²splice variant that was expressed. Effects on both deactivation and amplitude were most pronounced in receptors composed of only α and β subunits: α¹β²receptors exhibited a nearly 10-fold increase in the weighted time constant of deactivation accompanied by greater than 90% reduction in amplitude at an isoflurane concentration of 1 mm (figs. 3A, 4B, and 4D), and α⁵β³receptors were sixfold slower to deactivate and nearly 90% reduced in amplitude with 0.5 mm isoflurane (figs. 3C, 4B, and 4D). Coexpression of γ^2Lwith α¹and β²subunits substantially reduced the effect of isoflurane on amplitude (figs. 3B, 4C, and 4D) but not deactivation (figs. 4A and B). Coexpression of the γ^2Ssubunit with either α¹β²or α⁵β³receptors similarly reduced the effect of isoflurane on amplitude (figs. 3C and 4D), and it also reduced its effect on deactivation (figs. 3C and 4B). Therefore, just as γ^2Sand γ^2Ldifferently affected deactivation under control conditions, these two splice variants also differed in their influence on modulation by isoflurane.

An additional difference between the two γ²splice variants was a substantial variability in response to isoflurane with γ^2Sbut not γ^2L(compare error bars in fig. 4D, γ^2Sversus γ^2L, and fig. 5). Whereas in many cells expressing γ^2Sthe peak current amplitude was reduced by isoflurane, the degree of reduction was quite variable, and in some cells, there was even an increase in amplitude (fig. 5A) so that, on average, there was no change (fig. 4D; P > 0.05). Figure 5Bsummarizes the diversity of isoflurane effects on peak amplitude observed for α¹β²γ^2Sreceptors (n = 28). Responses ranged from 81% reduction to 52% enhancement. Cells labeled cell 1  or cell 2  at either end of the spectrum correspond to current traces labeled cell 1  or cell 2  in figure 5A. Accompanying the variable effects on amplitude were variable effects on deactivation. Cells in which peak current amplitude was most strongly reduced by isoflurane also exhibited the most pronounced effects on decay (figs. 5A and C, cell 1). Conversely, increased amplitude was associated with modest prolongation (figs. 5A and C, cell 2).

This correspondence applied not only to effects of isoflurane on amplitude and deactivation, but also to differences in baseline kinetics of different subunit combinations: Those that displayed the greatest reduction in amplitude in the presence of isoflurane (fig. 5A) were slower to deactivate in the absence of the anesthetic (fig. 5C) and displayed a greater degree of desensitization (fig. 5D). The strong correlations between amplitude reduction and decay prolongation by isoflurane and degree of desensitization under control conditions in α¹β²γ^2Sare summarized in figure 5E.

Comparison of isoflurane effects on receptors of different subunit composition revealed that the variability of isoflurane effects on α¹β²γ^2Sreceptors paralleled a general difference between αβ and αβγ receptors: Those most strongly blocked also were most prolonged by isoflurane (fig. 6A) and showed slower deactivation (fig. 6B) and faster desensitization (fig. 6C) in the absence of the anesthetic.

In previous studies, anesthetic-induced enhancement of peak currents occurred when low (subsaturating) concentrations of GABA were used but not when near-saturating concentrations were applied,33–35,45as in this study. Because slow solution exchange resulting in exposure to only subsaturating concentrations of GABA during brief (10-ms) pulses represents a possible explanation for the observed increases in peak current amplitude in α¹β²γ^2Sreceptors in the presence of isoflurane,16we therefore examined the 10–90% rise time (t^10–90) for different subunit combinations. However, in all cases, t^10–90values were 1.5–4 ms, which is substantially briefer than the pulse (duration 10 ms). Only for α¹β²receptors was there a significant decrease in rise time, and for these receptors, the effect was modest (t^10–90= 84% of control at 0.5 mm isoflurane, P < 0.05; t^10–90= 71% of control at 1 mm isoflurane, P < 0.01, n = 16). For other subunit combinations, effects were not significant (P > 0.05), and there was no correlation between the effect of isoflurane on the amplitude and rise time (r =−0.11, P > 0.05, n = 47). These findings argue against slow solution exchange as a cause for the increased amplitudes in α¹β²γ^2Sreceptors.

Although in many cases it may be useful to distinguish anesthetic effects on amplitude and deactivation, in some situations, net charge transfer may be the important parameter. The two major effects of isoflurane that we observed, reduced amplitude and slowed deactivation, have opposing influences in this regard. We assessed the net effect of isoflurane on charge transfer for different subunit combinations and for a range of isoflurane concentrations (fig. 7). Net charge transfer was significantly increased by isoflurane for all receptors containing a γ²subunit but was significantly reduced for α¹β²and α⁵β³receptors. Therefore, for αβ receptors, the slowed deactivation induced by isoflurane, although dramatic, was not sufficient to fully compensate for its profound blocking effect.

Desensitization is a prominent feature of all types of GABA^Areceptors. It has been suggested that modulation of desensitization by anesthetics may contribute to their net effect,16,45–47and desensitization of GABA receptors may have profound effects on the shape of IPSCs.48,49We tested the impact of the two γ²splice variants on modulation of desensitization by isoflurane by applying a high concentration of GABA (1 mm) for 2 s in the presence and absence of the anesthetic. For α¹β²receptors (n = 4), in addition to reducing peak amplitude, isoflurane increased significantly the rate and extent of desensitization (fig. 8). In contrast, even when peak amplitude was reduced, desensitization was not significantly altered in receptors containing either a γ^2Sor γ^2Lsubunit (P > 0.05, n = 7; figs. 8B–D).

“Coapplication” of GABA and isoflurane together resulted in a significantly smaller reduction in peak current amplitude compared with the “continuous” application of isoflurane (fig. 9). In receptors lacking a γ²subunit, coapplication also produced “rebound currents” after termination of a 2-s pulse (figs. 9A and C).

Given the extensive structural diversity that exists for GABA^Areceptors in the brain and previous reports that subunit composition influences physiologic18properties and pharmacologic sensitivity to a number of drugs that target GABA^Areceptors,25–29we considered whether modulation by isoflurane depends on the constituent subunits or splice variants. To investigate this issue, we tested the influence of the two splice variants of the γ²subunit on macroscopic kinetics and on modulation by isoflurane, comparing the influence of γ^2Sand γ^2Lsplice variants when they were coexpressed with a single αβ combination (α¹β²). Further, to determine whether the effect of the γ²subunit depends on the identity of α and β subunits with which it combines to form functional receptors, we compared effects of the γ^2Ssubunit when it was coexpressed with two different αβ combinations (α¹β²and α⁵β³).17,18We found that although there were quantitative differences between the two splice variants in the extent of their influence on baseline kinetics and on isoflurane modulation, in most regards, they imparted qualitatively similar effects: Receptors that incorporated a γ²subunit displayed less desensitization and faster deactivation under control conditions, effects of isoflurane on peak amplitude and on kinetics of deactivation and desensitization were substantially reduced in γ²-containing receptors, and net charge transfer was increased by isoflurane for αβγ but decreased for αβ receptors. We conclude that subunit composition does influence modulation of GABA^Areceptors by isoflurane and, more specifically, that the presence of the γ²subunit and the identity of its splice variants are important factors in determining physiologic properties and pharmacologic sensitivity.

Although the influence of GABA^Areceptor subunit composition on drug modulation has not been examined as extensively or systematically for volatile anesthetics as it has been for many drugs, existing studies have documented that a number of subunit combinations are modulated by a variety of volatile agents. In general, it has been found that responses to GABA are augmented, as reflected by an increase in response amplitude when low concentrations of agonists are applied, or a leftward shift when full GABA concentration–response relations are compared.30–33,35Because qualitatively similar results were observed for αβ- and αβγ-containing receptors, it was suggested that the identity of constituent subunits has little influence on susceptibility to modulation by volatile agents.30By contrast, our current results indicate that the γ²subunit strongly influences anesthetic modulation of peak amplitude and kinetic properties. How can we reconcile these apparently discordant findings? A likely explanation lies with the methods by which channels were activated in the different experiments. Many previous studies were performed using subsaturating concentrations of GABA and drug application techniques that had low temporal resolution, e.g. , bath application to Xenopus  oocytes.30,32,33,35In this case, several processes that influence net current amplitude occur simultaneously as agonist is applied. These include activation, desensitization, and possibly receptor block. Also, the rates of desensitization and deactivation are often not resolved unless techniques that permit rapid solution exchange are used. Anesthetic-induced shifts in the GABA concentration–response33,34,45may well overcome any coexisting block, thus obscuring a reduction in peak amplitude, particularly when low agonist concentrations are tested. Those studies that showed that isoflurane can reduce peak current amplitude used techniques that produced solution exchanges in the range of tens of milliseconds, approaching the rapid solution exchanges in our current study.37,50The suggestion that concentration and time course of GABA exposure to the receptor play a crucial role in the different effects of anesthetics on GABA-mediated currents may also help to explain the results of Jones et al. ,7,8who found that IPSC amplitudes were decreased by volatile anesthetics, although peak current amplitudes evoked by exogenous GABA application were increased in hippocampal neurons.

Consistent with previous results from our group39and others,51we found that incorporation of a γ²subunit had a major impact on kinetic properties of receptors. αβ-containing receptors showed significantly faster desensitization and slower deactivation than αβγ receptors. This difference was most prominent for α⁵β³compared with α⁵β³γ^2Sreceptors (fig. 1and tables 1 and 2) but was seen for α¹β²receptors and for γ^2Land γ^2Sas well. The slower deactivation of αβ receptors might be a consequence of the enhanced desensitization, because desensitized receptors are thought to reenter the “open” state before agonist unbinding occurs, thereby prolonging deactivation.48Prolonged deactivation of αβ receptors could also reflect slower agonist unbinding or channel closing,16both of which would lead to lower GABA EC⁵⁰in αβ- compared with αβγ-containing receptors.39 

As demonstrated previously for recombinant receptors37,50,52and for native receptors at synapses,8–10,13our data show that isoflurane exerts two opposing actions: reduction of amplitude and slowing of deactivation (figs. 3 and 4). Incorporation of a γ²subunit significantly affected both of these actions, reducing the effect on prolongation (figs. 3B, 3D, and 4B) and reducing the blocking effect of the anesthetic (figs. 3B, 3D, and 4D). Because we explored only a limited range of isoflurane concentrations, we cannot distinguish between differences in potency versus  efficacy in any of the drug actions. Also, overlapping ranges of concentrations that modulate, activate, and block receptors do complicate interpretation of such experiments. Differences in the relative concentrations of isoflurane versus  enflurane that slowed IPSC decay and reduced IPSC amplitude in hippocampal pyramidal cells suggested that the two effects are produced by anesthetics binding to two different sites.10Similarly, the different relations between block and prolongation for αβversus αβγ receptors (fig. 6A) may again reflect actions at two different sites, with the γ²subunit influencing the two sites differently. Although the presence of two anesthetic-sensitive sites of action seems to us to be the most likely explanation, it is also possible that differences in gating properties of αβ and αβγ receptors could lead to these differences. Until binding sites that underlie the two effects are identified, it will remain difficult to make stronger conclusions. In this regard, several amino acids near the extracellular ends of the transmembrane domains have been identified that influence the ability of isoflurane to enhance responses.53It has been proposed that these residues line an anesthetic-binding pocket.54Whether mutations of these residues also influence the susceptibility to block is not known.

It has been suggested that “open channel block” underlies the reduction of GABA-evoked peak responses by several volatile anesthetics.8,50,55,56This hypothesis is based in part on the occurrence of rebound (“hump”) currents that follow coapplication of GABA with anesthetic and that have been interpreted as the reversal of an open channel block due to faster unbinding rates of the anesthetic compared to the agonist. We also observed rebound currents after 2-s pulses of isoflurane together with GABA, but only in α¹β²and α⁵β³receptors, not in receptors that contained a γ²subunit (fig. 9). Although this could reflect a different mechanism of block, it could also reflect differences in unbinding rates. If so, a slower unbinding rate for γ²-containing receptors would have to be more than compensated by an even slower binding rate, because the degree of block is less for γ²-containing receptors. Alternatively, isoflurane might act as a negative allosteric modulator rather than an open channel blocker. This would be consistent with the observation that there is not a strong use dependence of block (fig. 4C).

Several different kinetic mechanisms might account for the effect of isoflurane on deactivation, including a decrease in agonist unbinding rate16or a direct change in gating.57Whereas peak responses of recombinant and native GABA^Areceptors are usually increased by anesthetics only when low concentrations of GABA are applied,3we found that in γ^2S-containing receptors, peak currents were at times increased even when a near-saturating agonist concentration was tested. This is most consistent with the hypothesis that isoflurane alters gating characteristics, e.g. , increases the opening rate or decreases the closing rate, so that peak open probability is increased. If this is not accompanied by block, it would lead to an increase in the peak current.

Volatile anesthetics have been reported to enhance receptor desensitization in some studies45,58,59but not others.16We show here that this occurs only for receptors that lack a γ²subunit. For these receptors, accelerated transition into a fast desensitized state could contribute to the reduction in peak current and, at the same time, prolong deactivation as receptors reenter the open state before unbinding occurs.48Considering the strong correlations between block and prolongation and between block and desensitization (fig. 6), enhanced desensitization by isoflurane represents a promising mechanism that might account for or contribute to the observed blocking and prolonging actions for αβ receptors, as proposed for neurosteroids.60However, an open channel block mechanism has been described that simultaneously prolongs agonist-induced responses and decreases response amplitude.61Also, it may be that distinct mechanisms produce to the two effects of isoflurane, as suggested for pentobarbitone.11 

Compared with α¹β², α⁵β³, and α¹β²γ^2Lreceptors, coexpression of a γ^2Ssubunit resulted in a large degree of variability of responses to isoflurane (figs. 4 and 5). Although we can only speculate on the mechanisms underlying the observed differences between γ^2Land γ^2Ssplice variants, one intriguing possibility is that the variability is related to a difference in phosphorylation. The long splice variant (γ^2L) differs from the short variant (γ^2S) by the inclusion of eight additional amino acids in the intracellular loop between TM3 and TM4. These amino acids contain an additional consensus site for phosphorylation.62This additional site in γ^2Lmight increase the possibility that at least one of the TM3–4 loop sites is phosphorylated. Although it does seem somewhat counterintuitive to suggest that more sites for phosphorylation could lead to less heterogeneity, this might happen if phosphorylation of any one site were sufficient to alter responses to isoflurane. Therefore, the more variable responses seen in γ^2Scontaining receptors might reflect a more variable fraction of phosphorylated versus  dephosphorylated γ^2Ssubunits in a given cell.

To what extent do differences in γ subunit and splice variant–specific modulation influence responses in the brain or in the whole animal? Potentially, the impact could be substantial, considering the large and even qualitatively different effects that we describe here. However, because there have been relatively few detailed studies with volatile agents and different types of inhibitory synapses and receptors in situ , particularly in preparations in which subunit compositions of receptors have been established, this question is difficult to answer. The γ²subunit plays a role in targeting and clustering of receptors to synapses,36so effects may depend on whether responses of synaptic or extrasynaptic receptors are examined and on brain region and cell type. There is evidence that receptors containing only α and β subunits do exist in the brain, and they are present not only during development but also postnatally.17If they are located primarily at extrasynaptic sites, the transmitter concentration and temporal profile may be quite different than the brief, high concentrations that we used in our current studies. Therefore, it may not be appropriate to extrapolate our current findings to this situation or, in general, to expected effects of anesthetics on “tonic” inhibitory currents.63 

Although both γ subunit splice variants are expressed throughout the brain, γ^2Sis relatively more abundant in neocortical layer VI, the hippocampus, and the olfactory bulb, whereas cerebellar Purkinje cells, the medulla, and the pons express higher levels of γ^2L.64,In vitro  studies that use expressed receptors and low concentrations of GABA have generally been consistent in demonstrating increased responses with a variety of volatile agents, but results using intact tissues and synaptic circuits have sometimes been contradictory and confusing.10,65–67Even within an individual neuron, striking differences in responses of distinct types of IPSCs to the volatile agent enflurane have been observed.9It seems likely that at least some of this variability reflects subunit-specific drug effects. Subunit-specific modulation also may account for our observation that receptors excised from the soma of CA1 pyramidal neurons, which have different kinetic properties than synaptic receptors,68are potently blocked by isoflurane at concentrations that do not alter the amplitude of synaptic responses (M. I. B. and R. A. P., unpublished data, 1997).

In summary, our current results indicate that the subunit composition of GABA^Areceptors, specifically the presence or identity of the γ²subunit, influences the response to isoflurane. This may contribute to the heterogeneity of synaptic and behavioral responses induced by volatile anesthetics.

The authors thank Cynthia Czajkowski, Ph.D. (Associate Professor), Andrew Boileau, Ph.D. (Associate Scientist), and Mathew Jones, Ph.D. (Assistant Professor) (all from the Department of Physiology, University of Wisconsin, Madison, Wisconsin), for helpful discussions and comments on the manuscript and Mark Perkins, B.S. (Research Specialist, Department of Anesthesiology, University of Wisconsin), for excellent technical assistance.