Antagonism of Inhalant and Volatile Anesthetic Enhancement of Glycine Receptor Function*

Recent studies suggest that alcohols, volatile anesthetics, and inhaled drugs of abuse, which enhance γ-aminobutyric acid, type A, and glycine receptor-activated ion channel function, may share common or overlapping molecular sites of action on these receptors. To investigate this possibility, these compounds were applied singly and in combination to wild-type glycine α1 receptors expressed in Xenopus laevis oocytes. Data obtained from concentration-response curves of the volatile anesthetic enflurane constructed in the presence and absence of ethanol, chloroform, or toluene were consistent with competition for a common binding pocket on these receptors. A mutant glycine receptor, insensitive to the enhancing effects of ethanol but not anesthetics or inhalants, demonstrated antagonism of anesthetic and inhalant effects on this receptor. Although ethanol (25–200 mm) had no effect on its own in this receptor, it was able to inhibit reversibly the enhancing effect of enflurane, toluene, and chloroform in a concentration-dependent manner. These data suggest the existence of overlapping molecular sites of action for ethanol, inhalants, and volatile anesthetics on glycine receptors and illustrate the feasibility of pharmacological antagonism of the effects of volatile anesthetics.

Ethanol, volatile anesthetics, and inhaled drugs of abuse are central nervous system depressants. Although it was once thought that they produced their effects in vivo by the nonspecific disordering of cell membranes, it is now generally accepted that these compounds instead exert their actions on specific proteins (1)(2)(3). Among these protein targets are the glycine receptors (Gly-Rs), 1 responsible for the majority of inhibitory neurotransmission in the brain stem and spinal cord. Gly-Rs are members of a superfamily of ligand-gated ion channels that also include the serotonin-3, ␥-aminobutyric acid, type A (GABA A ), and nicotinic acetylcholine receptors (4). These receptor complexes are composed of five protein subunits surrounding a central ion pore. Each subunit has four transmem-brane spanning domains and distinct binding sites for a variety of ligands (5). Specific single amino acid mutations alter the actions of allosteric modulators on Gly-Rs (6,7), and chemically diverse compounds, such as n-alcohols, anesthetics, and inhalants, may even share a common binding site on these ligandgated ion channels (3,(7)(8). Mutations affecting alcohol and volatile anesthetic actions on receptors may hinder the physical interactions of these compounds with their binding sites or may instead interfere with the abilities of alcohols and anesthetics to transduce their signals to modulate channel function after the compounds have bound.
A conventional investigation of binding sites on receptors would include radioligand competition studies. Unfortunately the low potencies (M to mM) of alcohols, inhalants, and volatile anesthetics, coupled with the current lack of an anesthetic site antagonist, make these studies unfeasible (9). Recently, an indirect experimental approach by Mascia et al. (7) addressed the anesthetic-binding site/transduction site issue by substituting cysteine for single amino acids believed to line the anesthetic-binding pocket on homomeric ␣ 1 Gly-Rs: the amino acids serine 267 and alanine 288. Propanethiol or propyl methanethiosulfonate covalently cross-linked to the receptor at the substituted cysteine at Ser-267 irreversibly enhanced Gly-R currents and prevented further enhancement by the anesthetics enflurane or isoflurane (7). In the present study, we demonstrate a reversible binding site interaction through the simultaneous administration of two positive allosteric modulators on wild-type and mutant ␣ 1 Gly-Rs. Specifically, we tested the hypothesis that ethanol (EtOH), the volatile anesthetics enflurane (ENF) and chloroform (CHCl 3 ), and the inhaled drugs of abuse toluene (TOL) and 1,1,1-trichloroethane (TCE) share common or overlapping binding sites on these receptors. Some of this work has been presented previously in abstract form (10).
Oocyte Isolation and cDNA Nuclear Injection-The oocyte isolation and cDNA injection procedures were performed as described previously (3). Oocytes were injected with cDNAs (1.5 ng/30 nl) of wild-type or mutated human glycine receptor ␣ 1 -subunits subcloned into either the pCIS2 or the pBKCMV vector, which had been modified by removal of the lac promoter and the lacZ ATG (6). Oocytes usually expressed receptors in 1-2 days, and electrophysiological measurements were made 1-7 days after cDNA injection.
Oocyte Electrophysiological Recording-Animal poles of oocytes were impaled with two high resistance (0.5-10 megohms) glass electrodes filled with 3 M KCl. Oocytes were voltage-clamped at Ϫ70 mV using a Warner Instruments OC-725C oocyte clamp (Hamden, CT). The concentration of agonist producing 10% of a maximal effect (EC 10 ) was * This work was supported by National Institute on Alcohol Abuse and Alcoholism Grant AA11525 (to S. J. M.). 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: Section of Neurobiology, 2500 Speedway, MBB 1.148, University of Texas, Austin, TX 78712-1095. Tel.: 512-232-7173; Fax: 512-232-2525; E-mail: beckstead@ mail.utexas.edu. 1 The abbreviations used are: Gly-R, glycine receptor; GABA A , ␥-aminobutyric acid, type A; EtOH, ethanol; ENF, enflurane; CHCl 3 , chloroform; TOL, toluene; TCE, 1,1,1-trichloroethane; EC X , the concentration of agonist (glycine) that produces X% of a maximal current. determined for each oocyte and subsequently used in experiments testing effects of allosteric modulators. Washout periods (5-18 min) were allowed to ensure complete recovery of receptors from desensitization between drug applications. During testing, oocytes were preincubated with modulator (EtOH, ENF, CHCl 3 , TCE, or TOL) for 60 s before application of the glycine/modulator solution for an additional 30 s. Oocytes were perfused with modified Barth's saline at a rate of 2 ml/min using a Masterflex USA peristalsis pump (Cole Parmer Instrument Co., Vernon Hills, IL) that connected drug-containing vials to the perfusion chamber through 18-gauge polyethylene tubing. Changes in the clamping current were recorded on a strip-chart recorder (Cole Parmer Co.), and the peak currents were measured and used in data analysis. Oocyte bath concentrations of modulators were determined by gas chromatography as described previously by Eger et al. (11).
Statistics-Statistics were performed on data obtained from oocytes, by one-way or two-way analysis of variance and appropriate post hoc tests. Within-subject designs were employed whenever feasible. Data are presented as mean Ϯ S.E.; n values refer to the number of different oocytes from which data were obtained. Statistical significance was defined as p Ͻ 0.05 on paired t tests, Tukey's or Dunnett's post hoc test. All groups of data were collected using oocytes obtained from at least two different frogs. The theoretical curves presented in Fig. 9 were created using the following biphasic sigmoidal Equation 1, where Y is the current; X is the concentration of drug; A and D are maximal possible currents; B and E are EC 50 values, and C and F are slope constants. A ϭ 100, B ϭ 1.3, C ϭ 0.6, and D ϭ 0 for the agonistonly curve. A ϭ 100, B ϭ 2.3, C ϭ 0.6, and D ϭ 0 for the addition of the competitive antagonist. The inhibitory component of the curves was created with the addition of the second term with the following parameters: D ϭ 100, E ϭ 3, and F ϭ 0.3.

RESULTS
EtOH, TCE, ENF, and CHCl 3 were applied singly and in combination to Xenopus oocytes expressing wild-type ␣ 1 glycine receptors, and the resulting currents were analyzed for possible interactions between compounds (Figs. 1 and 2). Each of four compounds enhanced the currents elicited by EC 10 concentrations of glycine applied alone. The co-application of a second modulator produced no additional effect, except when ENF and CHCl 3 were co-applied (p Ͻ 0.01 versus ENF alone, p Ͻ 0.01 versus CHCl 3 alone, Tukey's test; Fig. 2B).
ENF concentration-response curves were next constructed in the absence or presence of EtOH (Fig. 3A), TOL (Fig. 3B), or CHCl 3 (Fig. 3C). Single concentrations of EtOH, TOL, and CHCl 3 each enhanced glycine receptor function, but these enhancing effects were progressively decreased as the ENF concentration was increased. Consistent with the data shown in Fig. 2B, 1.0 mM CHCl 3 maintained a significant effect in the presence of all concentrations of ENF, whereas the enhancing effects produced by EtOH or TOL vanished when they were co-applied with even moderate concentrations of ENF. To determine if CHCl 3 could be acting at two distinct sites on glycine receptors, with only one site shared with EtOH, TOL, TCE, and ENF, we replaced an amino acid believed to compose part of the putative alcohol and anesthetic-binding site (serine 267) with the large phenylalanine residue (S267F). This substitution completely eliminated chloroform enhancement of currents elicited by an EC 10 concentration of glycine (Ϫ0.1 Ϯ 4.7% enhancement, n ϭ 6, versus 265 Ϯ 15.1%, n ϭ 10 in wild-type ␣ 1 Gly-R, data not shown), thus suggesting that a second separate site for chloroform does not exist.
Alcohols and volatile anesthetics produce leftward shifts of glycine concentration-response curves, and as a result, percent enhancement of receptor function by alcohols and anesthetics decreases when higher agonist concentrations are used. Alcohols and anesthetics have minimal effects on peak current amplitudes when maximally effective glycine concentrations are used. We tested whether the apparent competition observed between ENF and the other compounds could be due to the current produced by ENF already being maximal, thus preventing any further enhancement by a second modulator. First we identified the concentration of glycine (EC EQ ) that closely approximated the size of the current elicited by EC 10 glycine ϩ 1.74 mM ENF. We then tested the ability of 100 mM EtOH to enhance currents of that size, whether elicited by a low concentration of glycine in the presence of ENF or by a higher EC EQ concentration of glycine in the absence of ENF. Fig. 4 illustrates that 100 mM EtOH potentiated currents elicited by EC EQ glycine but not currents elicited by EC 10 glycine in the presence of 1.74 mM ENF. Similar results were obtained FIG. 1. Co-application of 1,1,1-trichloroethane minimally affects the enhancing effects of enflurane, chloroform, and ethanol on Gly-Rs. TCE (0.84 mM) had no effect in the absence of glycine but enhanced the current produced by a 10% maximal concentration of glycine. Enhancement of Gly-R function was also produced by 1.0 mM ENF (A), 1.5 mM CHCl 3 (B), and 200 mM EtOH (C). When co-applied with each of the other allosteric modulators, TCE produced little to no further enhancement compared with that produced by the other modulator alone. Representative tracings are illustrated above each graph. Compounds were pre-applied to oocytes for 60 s before being co-applied with EC 10 glycine for a further 30 s. Reversibility of all effects was determined by reapplication of EC 10 glycine, which was 95 M for these particular tracings (not shown). Data are presented as the mean Ϯ S.E. of six oocytes. with 0.56 mM TCE in the presence of 1.0 mM ENF (data not shown).
We next tested to determine if competition for a modulatorbinding site could be observed in a Gly-R mutated at a residue at or near the putative modulator-binding pocket. When the serine residue at amino acid 267 is mutated to an isoleucine (S267I), mutant Gly-Rs are created that retain their sensitivities to ENF and CHCl 3 but are insensitive to EtOH (6). We tested the hypothesis that EtOH was still binding to its putative binding site, despite having no efficacy on S267I ␣ 1 Gly-Rs. This could be demonstrated by its antagonism of the effects of other modulators that still had enhancing effects on S267I Gly-Rs (Fig. 5). EtOH (200 mM), while having essentially no effect on its own, was able to antagonize the potentiation produced by 0.42 mM TOL (p Ͻ 0.001, n ϭ 5, paired t test), 0.25 mM CHCl 3 (p Ͻ 0.001, n ϭ 6), and 0.5 mM ENF (p ϭ 0.002, n ϭ 6). EtOH antagonized the enhancing effects of 0.25 mM CHCl 3 in a concentration-dependent manner (F (4,16) ϭ 37.1, p Ͻ 0.0001], even at EtOH concentrations as low as 25 mM (Fig. 6). No significant enhancing or inhibiting effects of 25-200 mM EtOH were noted on S267I ␣ 1 Gly-R when it was co-applied with EC 10 glycine in the absence of CHCl 3 . Ethanol (200 mM) antagonism of the CHCl 3 enhancement of function on S267I ␣ 1 Gly-Rs was reversible as illustrated in the representative tracings shown in Fig. 7A. In addition to minimally affecting the peak amplitude of the S267I ␣ 1 Gly-R, ethanol also did not appear to affect glycine-induced steady-state currents (Fig. 7B).
EtOH antagonized the enhancement of receptor function produced by a range of CHCl 3 concentrations. Co-application of ethanol with CHCl 3 shifted the CHCl 3 concentration-response curve down and to the right (Fig. 8A). A similar shift was observed for the anesthetic ENF (Fig. 8B), although EtOH had no effect when co-administered with exceedingly high ENF concentrations (ϳ8 times the concentration required to anesthetize humans (12)). If EtOH was competing with CHCl 3 and ENF for common sites on the S267I ␣ 1 Gly-R, we should have expected to see rightward shifts in the CHCl 3 and ENF concentration-response curves, with no effects of EtOH at the highest chloroform and enflurane concentrations tested. That this was not seen is almost certainly due to an inhibitory action that occurs at higher concentrations of volatile anesthetics, at an anesthetic site separate from that at S267I. Evidence favoring this hypothesis was obtained using the S267F ␣ 1 Gly-R mutant. This receptor shows no enhancing effects of low concentrations of enflurane but demonstrates inhibition at higher concentrations (Fig. 8C). DISCUSSION The serine residue at position 267 of the ␣ 1 -subunit, proposed to compose part of the alcohol and anesthetic-binding pocket in Gly-Rs, is located near the extracellular terminus of the second of four membrane-spanning regions on the protein.
A series of single amino acid mutations made at Ser-267 ␣ 1 Gly-R alter the enhancing effects of ethanol (6), volatile anesthetics (8,13), and inhaled drugs of abuse (3) on GABA A and glycine receptors. An excellent linear correlation between the effects of the ether ENF and the halogenated alkane TCE across a panel of Ser267 mutant receptors is consistent with a common binding site despite the dissimilarities in their chemical structures (3). Mutations in the glycine ␣ 1 -subunit at positions 267 and 288 (in TM3) change alcohol cut off, defined as the largest chain n-alcohol capable of enhancing receptor function, which could be interpreted as a change in the size of the alcohol-binding pocket (14). GABA A ␣ 2 ␤ 1 (S265I)-receptors (the equivalent amino acid as serine 267 in the ␣ 1 glycine receptor) lose sensitivity to isoflurane but not its isomer ENF (8). Finally, irreversible enhancement of Gly-R function is observed as a consequence of cross-linking propyl methanethiosulfonate to S267C mutant Gly-Rs; this also blocks further enhancement by enflurane and isoflurane (7). A substantial body of data thus implicates this region of the glycine and GABA A receptors as being critical for mediating the functional effects of alcohols and volatile anesthetics. Molecular sites for propofol and other intravenous anesthetics are located elsewhere on the protein (15), as are the sites for benzodiazepines on the GABA A receptors (16).

FIG. 2. Effects of co-applying combinations of ethanol, enflurane, and chloroform on
Gly-R, compared with the effects produced by the applications of single compounds. EtOH (200 mM) was unable to enhance glycinergic currents when it was co-applied with 1.0 mM ENF (A) or 1.5 mM CHCl 3 (C). However, when 1.0 mM ENF and 1.5 mM CHCl 3 (B) were co-applied, there was slightly greater enhancement than that produced by either compound alone. Compounds were pre-applied to oocytes for 60 s before being co-applied with EC 10 glycine for a further 30 s. Representative current tracings are provided above each graph. Data are presented as the mean Ϯ S.E. of six oocytes.
As with any ligand-receptor interaction, the effect that an allosteric modulator has on a ligand-gated ion channel depends on both the compound and the receptor. Compounds that bind at a receptor site may be classified as agonists, antagonists, and inverse agonists, depending on their efficacies, which can be positive, zero, or negative, respectively. Modulation of receptor function can be altered by either changing the structure of the drug or by changing a relevant portion of the receptor. For example, after residue 142 on the GABA A receptor ␥ 2subunit is conservatively mutated, benzodiazepine receptor antagonists and inverse agonists function as partial agonists (17). Mutation of ␣ 1 Gly-R Ser-267 to isoleucine (S267I) results in a receptor that demonstrates responses to glycine similar to those of wild-type receptors and retains sensitivity to the enhancing effects of enflurane but not ethanol (6). By mutating ␣ 1 Gly-R Ser-267 to isoleucine, we altered modulator efficacy, changing EtOH to an antagonist capable of antagonizing the enhancing effects of ENF, TOL, and CHCl 3 . At the concentrations tested, ethanol does not significantly decrease currents evoked by glycine in S267I ␣ 1 Gly-R, strongly suggesting that EtOH is binding to a modulator site common to anesthetics and inhalants, but possesses minimal efficacy, rather than binding elsewhere to favor the stabilization of a closed conformation of the chloride channel.
The ethanol effects on ENF and CHCl 3 concentration-response curves (Fig. 8, A and B) do not initially suggest a competitive interaction between EtOH and the volatile anesthetics in the S267I mutant receptors. EtOH produces a rightward shift at low concentrations of ENF and CHCl 3 , but the curves fail to reach the same maximum, which would be expected if the interaction were competitive. Despite the shapes of these concentration-response curves, we believe competitive antagonism may be occurring at a common alcohol/anestheticbinding site. There are two possible explanations for the shapes of the observed concentration-response curves. First, EtOH could be binding to a site distinct from that of ENF and CHCl 3 and decreasing the affinities of ENF and CHCl 3 to bind. Al- . All three of these compounds enhanced Gly-R function in the absence of ENF, but this enhancement decreased as the ENF concentration was raised. Unlike EtOH and TOL, CHCl 3 produced a statistically significant enhancement of Gly-R function even in the presence of 1.74 mM ENF. TOL and EtOH had non-significant effects in the presence of the high concentrations of ENF. Compounds were preapplied to oocytes for 60 s before being co-applied with EC 10 glycine for a further 30 s. Data are presented as the mean Ϯ S.E. of 13-16 oocytes.

FIG. 4. Ethanol enhances currents produced by a high concentration of glycine but not a low concentration of glycine plus enflurane.
The concentration of glycine that produced a 10% maximal response (EC 10 ) was co-applied with 1.74 mM ENF. EtOH (100 mM) was unable to enhance these currents. However, 100 mM EtOH enhanced currents of identical size (p ϭ 0.007, paired t test) elicited by the application of a higher concentration of glycine (EC EQ ) in the absence of enflurane. Representative current tracings are provided above the graph. Data are presented as the mean Ϯ S.E. of five oocytes. though the shapes of the curves favor this possibility at low concentrations of ENF and CHCl 3 , it is difficult to reconcile the effects seen at the high end of the ENF curve. The second possibility is that there are two separate sites for ENF on these receptors, the first a pocket at Ser-267 where EtOH can compete, and the second a low affinity inhibitory site. This second site has long been postulated based on the non-sigmoidal nature of volatile anesthetic concentration-response curves, including decreasing enhancing effects at high anesthetic concentrations as well as rebound currents that are observed upon washout of drug (18). We created a mutant (S267F ␣ 1 Gly-R) that illustrates the dissociation between the enhancing effects of volatile anesthetics and their inhibiting effects (Fig. 8C). S267F ␣ 1 Gly-R show no signs of the enhancing effects of low concentrations of enflurane but do exhibit inhibition of Gly-R function at higher concentrations of ENF. A hypothetical composite ENF concentration-response curve is presented in Fig. 9, illustrating the potential contribution from a second inhibitory binding site for ENF, based on data presented here and observed by others (18). At low anesthetic concentrations the enhancing effects predominate, but when higher concentrations are achieved, inhibition predominates. In the S267I receptors, EtOH antagonism of ENF binding to the higher affinity enhancing site but not the lower affinity inhibitory site could make competitive antagonism appear to be non-competitive.
Regardless of the mechanism responsible, our observations suggest the possibility that compounds acting as ethanol and volatile anesthetic receptor antagonists may exist. Currently, no known compound competitively antagonizes anesthetics from their sites of action on wild-type ligand-gated ion channels, although saturable and displaceable binding of volatile anesthetics has been demonstrated on bovine serum albumin (19), on Ca 2ϩ -ATPase (20), and in rat brain synaptosomes (21). The ethanol antagonism of anesthetic effects we demonstrate in S267I ␣ 1 Gly-R is a promising indication that receptor antagonists might be designed for alcohol-, inhalant-, and anes-FIG. 6. Ethanol antagonizes chloroform enhancement of Gly-R currents in a concentration-dependent manner. Representative tracings illustrate that ethanol antagonism of chloroform enhancement of S267I ␣ 1 Gly-R function occurs in a concentration-dependent manner. A presents data as the mean Ϯ S.E. of five oocytes. All concentrations of ethanol tested antagonized the enhancing effects of chloroform on S267I ␣ 1 receptors. Compounds were pre-applied to oocytes for 60 s before being co-applied with EC 10 glycine for a further 30 s. and ENF (B) concentrations. Inhibition of Gly-R function is observed when higher concentrations of anesthetics are applied. We tested whether inhibition of receptor function at these higher concentrations was independent of the enhancing effects observed at lower concentrations. A Gly-R bearing a phenylalanine residue at position 267 (S267F) was resistant to the enhancing effects of ENF. This S267F Gly-R still displayed the inhibition of receptor function by higher concentrations of ENF (C). Data are presented as the mean Ϯ S.E. of 4 -8 oocytes.
thetic-binding sites on wild-type GABA A and glycine receptors. Mascia et al. (7) demonstrated that the Cys-267 ␣ 1 Gly-R residue could be irreversibly labeled with an anesthetic compound resulting in persistent enhancement of receptor function. The current study extends this work to show that compounds that are not covalently bound to this alcohol/anesthetic site can also be used to demonstrate competition for binding to this site.
In summary, ethanol, volatile anesthetics, and inhaled drugs of abuse appear to compete for a fixed number of binding sites as evidenced by sub-additivity of enhancement of ligandevoked currents of recombinant Gly-Rs expressed in X. laevis oocytes. EtOH antagonized the enhancement produced by several of these compounds on a mutated receptor previously shown to be insensitive to EtOH. Finally, our results suggest that it may be feasible to consider the possibility that com-pounds could be designed specifically to antagonize alcohol, inhalant, and volatile anesthetic effects on ligand-gated ion channels. Such discoveries could lead to improved treatments for acute alcohol, anesthetic, or inhalant overdose, or even new pharmacotherapeutic approaches for alcoholism. FIG. 9. The downward shift of the enflurane and chloroform concentration-response curve in S267I mutant Gly-Rs does not necessarily indicate non-competitive antagonism. Hypothetical sigmoidal anesthetic concentration-response curves in the absence (A) or presence (B) of EtOH as might be expected if EtOH and anesthetic were competing for binding to the same site in S267I ␣ 1 Gly-R. Dashed lines C and D illustrate the contributions of a second, independent, inhibitory site when its effects are combined with those of the enhancing site.