Numerous Classes of General Anesthetics Inhibit Etomidate Binding to γ-Aminobutyric Acid Type A (GABAA) Receptors*

Enhancement of γ-aminobutyric acid type A receptor (GABAAR)-mediated inhibition is a property of most general anesthetics and a candidate for a molecular mechanism of anesthesia. Intravenous anesthetics, including etomidate, propofol, barbiturates, and neuroactive steroids, as well as volatile anesthetics and long-chain alcohols, all enhance GABAAR function at anesthetic concentrations. The implied existence of a receptor site for anesthetics on the GABAAR protein was supported by identification, using photoaffinity labeling, of a binding site for etomidate within the GABAAR transmembrane domain at the β-α subunit interface; the etomidate analog [3H]azietomidate photolabeled in a pharmacologically specific manner two amino acids, α1Met-236 in the M1 helix and βMet-286 in the M3 helix (Li, G. D., Chiara, D. C., Sawyer, G. W., Husain, S. S., Olsen, R. W., and Cohen, J. B. (2006) J. Neurosci. 26, 11599–11605). Here, we use [3H]azietomidate photolabeling of bovine brain GABAARs to determine whether other structural classes of anesthetics interact with the etomidate binding site. Photolabeling was inhibited by anesthetic concentrations of propofol, barbiturates, and the volatile agent isoflurane, at low millimolar concentrations, but not by octanol or ethanol. Inhibition by barbiturates, which was pharmacologically specific and stereospecific, and by propofol was only partial, consistent with allosteric interactions, whereas isoflurane inhibition was nearly complete, apparently competitive. Protein sequencing showed that propofol inhibited to the same extent the photolabeling of α1Met-236 and βMet-286. These results indicate that several classes of general anesthetics modulate etomidate binding to the GABAAR: isoflurane binds directly to the site with millimolar affinity, whereas propofol and barbiturates inhibit binding but do not bind in a mutually exclusive manner with etomidate.

␥-Aminobutyric acid type A receptors (GABA A Rs) 3 are major mediators of brain inhibitory neurotransmission and participate in most circuits and behavioral pathways relevant to normal and pathological function. Their regulation may under-lie many psychiatric/neurological disorders. GABA A Rs are subject to modulation by endogenous neurosteroids, as well as myriad clinically important central nervous system drugs, including general anesthetics, benzodiazepines, and ethanol (1)(2)(3). The mechanism of GABA A R modulation by these different classes of drugs is of major interest, including the localization of their binding sites in the receptor.
Each GABA A R is made up of five homologous subunits that associate around a central axis that forms the ion channel. Each subunit consists of a large extracellular N-terminal domain, a transmembrane domain made up of a loose bundle of four transmembrane ␣-helices, and a cytoplasmic domain made up of the amino acids located in the primary structure between the M3 and M4 helices. In an ␣ 2 ␤ 2 ␥ GABA A R, the neurotransmitter-binding sites are located in the extracellular domain at the interfaces between the ␤ and ␣ subunits, two sites per receptor, whereas the benzodiazepine sites are at an equivalent position at the ␣-␥ subunit interface, one per receptor (4).
Most general anesthetics enhance GABA A R responses in vitro at concentrations that produce immobilization in vivo, suggesting a link between the GABA A R and anesthesia (5,6). The expression of receptors containing chimeric subunits combining sequences from subunits conferring different anesthetic sensitivities led to the identification of two residues that determine the sensitivity of the GABA A R to volatile agents and alcohols (7). These residues, located in the transmembrane domains of both ␣ and ␤ subunits, one in M2 and the other in M3, were hypothesized to contact a single water-filled intrasubunit pocket (2,8). The M2 residue in ␤ subunits, e.g. ␤3Asn-265, is also implicated in the in vivo action of intravenous anesthetics, including etomidate, propofol, and barbiturates, but not anesthetic steroids (9,10). This single residue, when mutated in a knock-in mouse, eliminated the immobilization (␤3N265M (11)) or sedative/hypnotic (␤2N265S (12)) actions of etomidate.
We synthesized a photoreactive analog of etomidate, [ 3 H]azietomidate (2-(3-methyl-3H-diaziren-3-yl)ethyl 1-(phenylethyl)-1H-imidazole-5-carboxylate), which retained anesthetic activity (13). [ 3 H]Azietomidate photolabeled purified bovine brain GABA A R in a pharmacologically specific manner, photoincorporating into two residues (␣1Met-236 and ␤Met-286) 4 (14). These amino acids are in the membrane-spanning ␣M1 and ␤M3 helices, the latter being one of the identified affinity determinants for etomidate and propofol (15). These photolabeling results suggested a GABA A R structural model based on homology with the nicotinic acetylcholine receptor (14,16) that positioned both photolabeled amino acids within an intersubunit binding pocket at the ␤-␣ interface, two copies of a single class of sites per pentamer. This structural model is supported by cysteine replacement cross-linking data defining the orientation of various GABA A R transmembrane helices within and between subunits (17,18). Neuroactive steroids enhance rather than inhibit [ 3 H]azietomidate binding (19), indicating that the steroid gating site and etomidate modulation sites do not coincide, although other positions in ␣M1 and ␤M3, as well as in ␣M4, have been identified as neurosteroid sensitivity determinants (20). In this study, we use [ 3 H]azietomidate photolabeling to determine whether etomidate binds competitively with other structural classes of general anesthetics. Our results provide evidence that isoflurane at millimolar concentrations may inhibit binding competitively, propofol and barbiturates act as allosteric inhibitors, and ethanol and octanol at anesthetic concentrations have no effect on [ 3 H]azietomidate binding. Thus, several structural classes of general anesthetics, including propofol, barbiturates, and volatile agents, but not alcohols, interact, directly or indirectly, with the intersubunit etomidate-binding site in the transmembrane domain of GABA A R proteins.

Solubilization and Purification of Bovine Brain GABA A Rs-
The GABA A R was solubilized and purified on a benzodiazepine Ro7/1986-1 affinity column as described (14). Important modifications of previous protocols that allowed an ϳ5000-fold purification of the GABA A R at high yield and with retention of positive allosteric modulation of [ 3 H]muscimol binding by etomidate, propofol, pentobarbital, isoflurane, octanol, or neuroactive steroids included (i) the use of the detergent C 12 E 9 in conjunction with CHAPS in the solubilization buffer, (ii) extensive washing of the affinity resin with 10 mM CHAPS, 0.06% asolectin, 10% sucrose, and (iii) elution with clorazepate rather than flurazepam and urea. The purified protein contains a mixture of GABA A R subtypes of varying subunit composition that bind the benzodiazepine affinity column (14). [ 3 H]Muscimol binding assays were performed as described (14).
Photoaffinity Labeling of Purified GABA A R-The GABA A R was photolabeled on an analytical scale (ϳ6 pmol of [ 3 H]muscimol-binding sites/sample) to examine the concentration dependence of anesthetic modulation of [ 3 H]azietomidate incorporation, as determined by SDS-PAGE, and photolabeling was carried out on a preparative scale (95 pmol of [ 3 H]muscimol-binding sites/sample) to determine whether propofol inhibited [ 3 H]azietomidate photolabeling of ␣1Met-236 or ␤Met-286 or caused labeling of other amino acids. Effects of propofol, which has an aqueous solubility of 0.9 mM (21), were examined at concentrations up to 0.2 mM. A stock solution of propofol was prepared at 200 mM in dimethyl sulfoxide, which was present at a final concentration of 0.1% (v/v) in each irradiated sample. The peak [ 3 H]muscimol binding fraction (5 ml) from each affinity column elution was used for labeling without further dialysis or concentration. An aliquot of GABA A R (ϳ40 nM [ 3 H]muscimol-binding sites, 2.5 ml for preparative labeling, or 0.14 ml for analytical labeling) was equilibrated with [ 3 H]azietomidate (final concentration, 0.7 M for preparative labeling or 1 M for analytical labeling) Ϯ additional drugs in the presence of 1 mM GABA and 10 mM clorazepate (a benzodiazepine agonist), incubated on ice for 1 h in the dark (4°C), and irradiated (30 min, 365 nm). After photolabeling, the total protein was precipitated with methanol/chloroform, solubilized in sample loading buffer, and fractionated by SDS-PAGE. The resulting gel lanes were cut into 3-mm slices, and 3 H incorporation was determined either directly by liquid scintillation counting (analytical labeling) or after elution from each slice into 1 ml of elution buffer (14). Aliquots of the eluted bands were assayed for 3 H and pooled for proteolytic digestion.
Enzymatic Digestion, Reversed-phase High Pressure Liquid Chromatography (HPLC), and Protein Microsequencing-Digestion of [ 3 H]azietomidate-photolabeled GABA A R subunits with endoproteinase Lys-C, reversed-phase HPLC fractionation of the digests, and peptide microsequencing were each performed as described (14).

Numerous Chemical Classes of General Anesthetics Modulate [ 3 H]Muscimol Binding to Purified Bovine Brain GABA A R-
Chemical structures of the compounds studied are shown in Fig. 1. Propofol, pentobarbital, isoflurane, and n-octanol each allosterically enhanced the equilibrium binding affinity of [ 3 H]muscimol for affinity-purified bovine brain GABA A R in detergent solution (Fig. 2), as reported previously for etomidate (14) and for barbiturates using membranes from brain homogenates (22,23).
Some General Anesthetics Inhibit GABA A R Photolabeling by [ 3 H]Azietomidate-When purified bovine brain GABA A R photolabeled with [ 3 H]azietomidate in the absence or presence of 200 M nonradioactive etomidate was fractionated by SDS-PAGE, the 3 H incorporated into GABA A R subunit polypeptides of ϳ50 -55 kDa was inhibitable by Ͼ90% in the presence of etomidate and was shown to result from labeling of ␣1Met-236 4 (or the homologous methionine in ␣2, 3, or 5) in ␣M1 and ␤Met-286 4 in ␤M3 (14). To test for the effects of other classes of anesthetics, we photolabeled the GABA A R with [ 3 H]azietomidate in the presence of the drugs at varying concentrations, always in the presence of 1 mM GABA and the benzodiazepine clorazepate.
Propofol, which has in vivo and GABA A R-enhancing actions and potency similar to etomidate (5, 6), produced a concentration-dependent inhibition of [ 3 H]azietomidate photolabeling of the purified GABA A R, as indicated by SDS-PAGE (Fig. 3, A and B). At high concentrations, propofol reduced photolabeling maximally by ϳ50%, whereas in a parallel sample, 200 M etomidate inhibited photolabeling by ϳ90%, as seen previously (14). Propofol produced a half-maximal inhibition at a concentration (IC 50 ) of ϳ10 M, the same concentration that produced half-maximal potentiation of [ 3 H]muscimol binding to the purified GABA A R ( Fig. 2A) but ϳ5-fold higher than the EC 50 for GABA A R potentiation as measured by electrophysiological recording (24). The maximal inhibition by propofol appeared to reach a plateau in four independent photolabeling experiments using [ 3 H]azietomidate at concentrations between 0.7 and 3 M (data not shown).
Isoflurane, a volatile anesthetic, is known to enhance GABA A R function (2,26). As shown in Fig. 5 (A and B), isoflurane inhibited GABA A R subunit labeling by Ͼ80% at 10 mM, with an IC 50 of 2 mM (Fig. 5B), a concentration within a factor of 2 of that enhancing [ 3 H]muscimol binding (Fig. 2C) but ϳ5-fold higher than the electrophysiological EC 50 for GABA A R potentiation (27,28). In contrast, neither ethanol at a concentration as high as 170 mM (Fig. 5C) nor n-octanol up to 1 mM (Fig. 5D) inhibited [ 3 H]azietomidate photolabeling, whereas octanol at 1 mM enhanced [ 3 H]muscimol binding to the purified receptor by ϳ80% (Fig. 2D).

DISCUSSION
In this study, we photolabeled bovine brain GABA A Rs with the photoreactive etomidate analog [ 3 H]azietomidate to determine whether other structural classes of general anesthetics bind to the same site as etomidate. Azietomidate, like etomidate, displays stereospecificity as an anesthetic in modulating GABA A R function and in binding to GABA A Rs in vitro (13).
[ 3 H]Azietomidate photoincorporates in a pharmacologically specific manner into two amino acids in bovine brain GABA A R, ␣1Met-236 4 within ␣M1 and ␤Met-286 4 within ␤M3, with photolabeling of those amino acids enhanced in the presence of GABA and inhibited by etomidate (14). These two amino acids were proposed to be located in a drug-binding pocket at the interface between the ␤ and ␣ subunits, a conclusion that has been supported by cysteine substitution cross-linking studies (18) that define the relative orientations of the ␣M1 and ␤M3 helices and by mutagenesis studies that establish that ␣1Met-236 and ␤Met-286 are important determinants of GABA A R gating and sensitivity to etomidate and propofol (26,29,30) and that cysteine substitutions at either of those positions react with a sulfhydryl-reactive analog of etomidate (31).
For GABA A Rs photolabeled with 1 M [ 3 H]azietomidate in the presence of nonradioactive etomidate, 3 H incorporation into GABA A R subunits, analyzed by SDS-PAGE, was reduced by Ͼ90% in the presence of excess nonradioactive etomidate, with an IC 50 of 20 M (14). Any other anesthetic that binds in a mutually exclusive manner with [ 3 H]azietomidate should reduce subunit labeling to the same extent as etomidate, as will drugs that act as strong negative allosteric modulators. In contrast, allosteric inhibitors that decrease [ 3 H]azietomidate binding affinity by Ͻ5-fold will, at high concentrations, produce only a partial reduction of subunit photolabeling, and drugs that enhance azietomidate binding affinity will increase subunit photolabeling and labeling of ␣1Met-236 and/or ␤Met-286, as has been seen for anesthetic steroids (19).
The results presented here provide evidence that isoflurane, which reduces subunit photolabeling by Ͼ80%, may bind competitively with etomidate, whereas propofol and anesthetic barbiturates act as allosteric inhibitors. Propofol at high concentrations reduced 3 H incorporation at the subunit level by ϳ50%, and it also reduced photolabeling of both ␣Met-236 and ␤Met-286 by ϳ60%, without causing [ 3 H]azietomidate to photoincorporate into other amino acids in ␣M1 or ␤M3. Significantly, (Ϫ)-mephobarbital, which is active as an anesthetic and GABA A R modulator, inhibited [ 3 H]azietomidate photolabeling with an IC 50 of 30 M, whereas (ϩ)-mephobarbital, the pharmacologically inactive stereoisomer, had no effect at concentrations up to 1 mM. We found no evidence that binding of ethanol or octanol modulates etomidate binding. Because octanol potentiated [ 3 H]muscimol binding to the purified GABA A R in a manner similar to propofol, pentobarbital, and isoflurane, the lack of effect of octanol on [ 3 H]azietomidate photoincorporation cannot result from an inability of octanol to bind to the purified GABA A R.
As seen for etomidate (14), the EC 50 for potentiation of [ 3 H]muscimol binding by propofol, pentobarbital, or isoflurane was within a factor of 2 of the IC 50 seen for the inhibition of [ 3 H]azietomidate photolabeling of the same receptor preparation. In each case, that concentration was 5-10-fold higher than the EC 50 typically reported for the anesthetic potentiation of GABA A responses and closer to the EC 50 reported for direct anesthetic gating of the GABA A R in the absence of GABA (24,25,27,28). However, concentrations of etomidate up to 100 M produce a progressive decrease in GABA EC 50 without any evidence of saturation (32), and the The peptide beginning at ␣1Ile-223 (Ϫpropofol, initial amount (I 0 ) ϭ 0.6 pmol, repetitive yield (R) ϭ 91%; ϩpropofol, I 0 ϭ 1.6 pmol, R ϭ 83%) was the only peptide that persisted after treatment with o-phthalaldehyde prior to cycle 11 to prevent sequencing of any peptide not containing a proline at this cycle. Release of 3 H in cycle 14 established photolabeling of ␣1Met-236 (Ϫpropofol) at 140 cpm/pmol (95-210 cpm/pmol) that was reduced by propofol to 35 cpm/pmol (22-56 cpm/pmol). These ranges were calculated using the standard errors from the fits for R and I 0 . D, sequence analysis of photolabeling in ␤M3. The peptide beginning at ␤Ala-280 was present in both samples (Ϫpropofol, I 0 ϭ 0.7 pmol, R ϭ 93%; ϩpropofol, I 0 ϭ 0.5 pmol, R ϭ 96%), and the peak of release of 3 H in cycle 7 indicates photolabeling of ␤Met-286 (Ϫpropofol) at 50 cpm/pmol (33-79 cpm/pmol) that was reduced by propofol to 19 cpm/pmol (10 -39 cpm/pmol). E, shown is the alignment of subtypes of ␣ or ␤ subunits in the regions of M1 or M3 (both in gray), respectively, illustrating the high sequence conservation in these regions. In boldface are the labeled Met residues as well as the conserved Pro residue in cycle 11 of Edman degradation (C). data were well fit by an allosteric model with each GABA A R containing two equivalent binding sites contributing to potentiation and to direct gating. Each ␣ 2 ␤ 2 ␥ GABA A R contains at the ␤-␣ interfaces two equivalent binding sites for [ 3 H]azietomidate and etomidate, and these sites were proposed to mediate etomidate's two effects, enhancement of GABA and direct channel gating efficacies (14,19).
The partial inhibition of azietomidate labeling by propofol is compelling evidence against a direct competitive interaction (mutually exclusive binding), as long as we exclude the possi-bility that there is a population of receptors photolabeled by azietomidate that are insensitive to propofol, a hypothetical situation for which there is no experimental evidence (see below). Although a mechanism of allosteric inhibition can account for the observed partial reduction of [ 3 H]azietomidate photolabeling seen in the presence of high concentrations of propofol or barbiturates, alternative explanations must also be considered. (i) Although photoincorporated [ 3 H]azietomidate is no longer in reversible equilibrium with the competing anesthetics, this cannot account for the partial reduction in photolabeling because under our labeling conditions [ 3 H]azietomidate is incorporated into only ϳ2% of receptors (14). In addition, the fact that etomidate and isoflurane at high concentrations can produce full inhibition establishes that the extent of inhibition must reflect site occupancy by reversibly bound [ 3 H]azietomidate. (ii) Because the purified GABA A R preparation from bovine brain used for photolabeling is a heterogeneous population of GABA A Rs that bind to the benzodiazepine affinity column, i.e. receptors of variable subunit composition containing a ␥ subunit, it is possible that propofol and barbiturates bind only to a subset of the GABA A R subunit combinations that bind [ 3 H]azietomidate and etomidate. This is a very unlikely explanation because in vitro studies testing anesthetic sensitivities of various subunit combinations indicate that there is no evidence for a receptor subunit population that is sensitive to etomidate but insensitive to propofol. The subunit combinations most sensitive to etomidate are also sensitive to propofol and barbiturates (25,33), and none of the receptors most widely expressed in brain are insensitive to either of those anesthetics (10).
All structural classes of general anesthetics are positive allosteric modulators of [ 3 H]muscimol binding and agonist benzodiazepine binding, as well as negative modulators of bicuculline binding and benzodiazepine inverse agonist binding and of the channel blocker ligands like t-[ 35 S]butylbicyclophosphorothionate (22, 23, 34 -36). Thus, one might expect the negative interactions between the binding of the different positive modulatory anesthetics (inhibition of etomidate binding by propofol, barbiturates, and isoflurane) to be competitive rather than allosteric. However, our photolabeling studies were carried out in the presence of GABA and benzodiazepines, and the weak energetic coupling that can account for our data may result because etomidate and propofol, for example, actually bind in close proximity within the binding pocket at the interface between the ␤ and ␣ subunits. Clearly, further studies using photoreactive analogs of propofol or other anesthetics are required to directly identify their binding sites.
Extensive mutational analyses have identified GABA A R amino acids in each of the transmembrane helices (6,20,37) of the ␣ and ␤ subunits that make important energetic contributions to the actions of anesthetics and that may potentially contribute directly to the structure of the anesthetic-binding sites. For reference purposes, those positions can be identified by using a common nomenclature that numbers the transmembrane helices from their N-terminal ends (4,16). ␤M2-15 is a major determinant of the potency and efficacy of etomidate, propofol, and pentobarbital as GABA A R modulators in vitro and as anesthetics in vivo (9, 10). ␤M2-15 was not photolabeled  by [ 3 H]azietomidate, but in our GABA A R structural model (14,19), it is located at the border of the etomidate-binding pocket, accessible from the ␤-␣ interface and from the interior of the ␤ subunit helix bundle. ␤M3-4, the position in the ␤ subunit photolabeled by [ 3 H]azietomidate, is a sensitivity determinant for propofol (26), and different-sized analogues of propofol suggested that this residue plays a role consistent with a binding site (24). Furthermore, propofol reduces the rate of reaction of a sulfhydryl-reactive reagent with cysteine-substituted ␤M3-4 but not ␤M2-15 (29). Although this result suggested that propofol may bind in close proximity to ␤M3-4, the 50% reduction of the reaction rate produced by propofol could also result from an allosteric effect, just as the rate of modification of the cysteine at ␤M3-4 was increased 3-fold in the presence of GABA compared with its absence. These studies also suggest the possibility that ␤M2-15 may contribute indirectly to anesthetic action, i.e. by allosteric conformational coupling, rather than by the presence in a binding pocket for etomidate, propofol, and barbiturates.
Extensive mutational analyses provide strong evidence that volatile anesthetics and alcohols interact with a site distinct from the etomidate site identified by photoaffinity labeling. ␣M2-15 and ␣M3-4 were the positions first identified as important sensitivity determinants for volatile anesthetics and alcohols (7), and a series of mutations at ␣M2-15 suggest that the size of the amino acid side chain affects anesthetic action, consistent with a true binding site for volatile agents and alcohols (2,8,38). In addition, a sulfhydryl-reactive alcohol analog, propyl methanethiosulfonate, produced irreversible enhancement of GABA responses in a receptor containing a cysteine at ␣M2-15, but not at ␣M3-4, and this modification occluded further potentiation by isoflurane or octanol, but not a neurosteroid (39). Positions ␣M2-15 and ␣M3-4 were proposed to contribute to an anesthetic-binding site within the ␣ subunit helix bundle (8), although in other GABA A R structural models (14,18), ␣M3-4 would be positioned equivalent to ␤M3-4 in pockets at the interfaces between ␣-␤ and ␣-␥ subunits. Although we found that octanol at high concentrations had no effect on [ 3 H]azietomidate photolabeling, isoflurane was the only anesthetic other than etomidate to fully inhibit photolabeling, consistent with a competitive interaction. Further studies will be required to determine whether isoflurane binds with lower affinity to the etomidate site than to other sites in the GABA A R and/or whether isoflurane occupancy of the etomidate site provides an equivalent or less of an energetic contribution to gating than the binding of etomidate (or the binding of isoflurane at the other sites in the GABA A R).