Functional and Structural Analysis of the GABA A Receptor (cid:1) 1 Subunit during Channel Gating and Alcohol Modulation*

The substituted cysteine accessibility method has proven useful for investigating structural changes of the (cid:2) -aminobutyric acid type A (GABA A ) receptor during channel gating and allosteric modulation. In the present study, the surface accessibility and reaction rate of pro-pyl- and hexyl-methanethiosulfonate to cysteine residues introduced into the third transmembrane segment of the GABA A receptor (cid:1) 1 subunit were examined. GABA-induced currents in Xenopus oocytes expressing wild type and cysteine mutant GABA A receptors were recorded before and after application of methanethio-sulfonate (MTS) reagents in the resting, GABA- or alco-hol-bound (ethanol or hexanol) states. Our results indicate that a water-filled cavity exists around the Ala 291 and Tyr 294 residues of the third transmembrane segment, in agreement with previous results. Furthermore, our data indicate that a conformational change produced by alcohols (200 m M ethanol or 0.5 m M hexanol) exposure induces the water cavity around the A291C and Y294C residues to extend deeper, causing the A295C and F296C residues to become accessible to the MTS reagents. In addition, exposure of the A291C, Y294C, F296C, and V297C mutants to MTS reagents in the presence data show A is a during modulation and gating. GraphPad Prism3 software (San Diego, CA). The data were normalized to the maximal current in each oocyte and fitted according to the equation of the form: I (cid:3) I max /[1 (cid:6) (EC 50 / A ) n ], where I is the current, I max is the maximal current recorded in a given oocyte, EC 50 is the GABA concentration for half-maximal current response, A is the GABA concentration, and n is the Hill coefficient. Statistical significance of the difference between each mutant and wild type were also analyzed by one-way analysis of variance with the Dunnett’s post hoc test with p (cid:2) 0.05 representing significance using this software.

The GABA A receptor is the major inhibitory neurotransmitter-gated ion channel protein in the mammalian brain. Each subunit has a large N-terminal extracellular domain that is involved in agonist binding, four transmembrane domains, a large intracellular loop between TM3 and TM4, and a short extracellular C terminus (9,10). Several families and subtypes of GABA A receptor subunits (six ␣, three ␤, three ␥, one ␦, one ⑀, one , and one subunits) have been cloned to date (11), but the stoichiometry and subtype of most GABA A receptors in the brain are believed to be two ␣, two ␤, and one ␥ subunits (12)(13)(14).
The substituted cysteine accessibility method (SCAM) has been used to investigate structure and conformational changes of ion channel domains in different functional states (15)(16)(17)(18)(19)(20)(21). To investigate structural dynamics of extracellular binding sites or TM of ion channel during gating and modulation, a single cysteine residue is introduced into the functional domain and probed with a water-soluble, sulfhydryl-specific reagent. If the substituted cysteine reacts with the reagent to cause the function of the channel to be irreversibly changed, the cysteine is assumed to be exposed at the water-accessible protein surface (15)(16)(17)(18).
Previous SCAM results suggest that channel gating and allosteric modulation by diazepam or propofol induce conformation changes in the TM3 of GABA A receptor ␣ 1 subunit and that water-accessible cavities form around the TM3 segment. This indicates that structural movement by GABA, diazepam, or propofol binding might allow sulfhydryl-specific reagents to penetrate into the interior of the TM domain (15)(16)(17), and several amino acids in the TM2 and TM3 segments are critical for potentiating glycine and GABA A receptors by volatile anesthetics and alcohols (4). A binding cavity for alcohols may exist in a crevice near the extracellular ends of the TM2 or TM3 regions of GABA A and glycine receptors (7,22,23). However, the structure and function of the TM3 of GABA A receptor during alcohol binding remains unknown. In the present study, we asked how the structure of the alcohol-bound GABA A receptor differs from the resting or GABA-bound states.
We used the SCAM to investigate the structure and conformational changes of the TM3 region of GABA A receptor ␣ 1 subunit during channel gating and alcohol modulation. We propose that a water-filled cavity exists around the extracellular side of TM3 and that the cavity deepens to the middle region of TM3 during channel gating and alcohol binding. Furthermore, based on results of functional accessibility of PMTS and HMTS, our data suggest that the extracellular side of TM3 is more flexible or dynamic compared with cytosolic side of TM3. Thus, structural rearrangement of the existing waterfilled cavities around the extracellular side of TM3 during channel gating and alcohol binding may have an important role in both forming the binding sites for alcohols and in the mech-anism of allosteric modulation of the GABA A receptor by alcohols.

EXPERIMENTAL PROCEDURES
Mutagenesis and Expression-pGEMHE plasmids encoding wild type and 16 cysteine mutants from Ala 291 to Val 307 in TM3 of the rat GABA A ␣ 1 subunit were described previously (15). cRNAs were synthesized from pCIS2 plasmid encoding human ␤2 GABA A receptor subunit and pGEMHE plasmids encoding rat ␣ 1 or ␥ 2s GABA A receptor subunits by using a T7 RNA polymerase kit (Stratagene, La Jolla, CA). Xenopus laevis oocytes were isolated and injected with the cRNAs (10 ng/50 nl) encoding wild type or mutant ␣ 1 , wild type ␤2, and ␥ 2s subunits combinations in a 1:1:1 ratio of diethylpyrocarbonate-treated water (15). We introduced one additional mutation (I290C) using site-directed mutagenesis in pGEMHE plasmid encoding the rat GABA receptor ␣ 1 subunit and a QuikChange site-directed mutagenesis kit (Stratagene). This mutation was verified by double-stranded DNA sequencing.
Electrophysiological Analysis-GABA-induced currents were recorded from oocytes 2-5 days after cRNA injection using a two-electrode voltage clamp (23 To investigate the surface accessibility of MTS reagents to GABA A receptors, GABA control currents were first determined by using two different GABA concentrations corresponding to the EC 5-10 and EC 50 values. These currents were determined to be stable after change of Ͻ5% in the GABA-induced currents on more than two consecutive applications. The EC 5-10 or EC 50 are concentrations of agonist that evoke 5-10 or 50% of the maximal current (obtained by application of 1 mM GABA). Second, oocytes expressing wild type or mutant GABA A receptors were perfused 90 s with PMTS or HMTS solutions in the resting state (0.5 mM MTS alone), GABA-bound state (0.5 mM MTS reagents in 1 mM GABA), or alcohol-bound state (0.5 mM MTS reagents in 200 mM ethanol or 0.5 mM hexanol). Following washing (10 -15 min) with MBS, the GABA control currents (EC 5-10 and EC 50 ) were redetermined. The effect of MTS reagents was calculated as follows: The percentage of change ϭ {(I after /I before ) Ϫ 1} ϫ 100, where I before and I after indicate the values of the two control currents induced by EC 5-10 or EC 50 GABA concentrations before and after the application of the MTS reagent.
To determine the percentage of potentiation on GABA (EC 5-10 ) response by alcohols (200 mM ethanol or 0.5 mM hexanol), the oocytes were perfused with alcohol for 1 min to allow for complete equilibration with alcohols before a 20-s coapplication with GABA (EC 5-10 ). The solutions were freshly prepared immediately before use.
Measurement of Reaction Rates-The MTS reaction rates with the introduced cysteines were determined by the effect of sequential brief applications of MTS in the resting, alcohol-bound, or GABA-bound states (21,24,25). After stabilization of EC 5-10 or EC 50 GABA-induced currents on more than two consecutive applications, HMTS alone (for resting state), HMTS with 1 mM GABA (for GABA-bound state), or HMTS with 200 mM ethanol or 0.5 mM hexanol (for alcohol-bound states) solutions were applied for 20 s. 0.25-0.5 mM HMTS solution was used. All of the MTS and alcohol solutions were prepared immediately before use. After washing (10 -15 min) with MBS, the EC 5-10 and EC 50 GABA-induced currents were remeasured. To see the effects of alcohols on reaction rates in alcohol-bound states, 200 mM ethanol, or 0.5 mM hexanol solutions in MBS were preincubated for 60 s like the above alcohol potentiation experiments. This procedure was repeated until the GABA-induced response no longer changed. The normalized GABAinduced currents to the initial current were fitted to one phase exponential function using GraphPad Prism3 software (San Diego, CA) to calculate the first order rate constants (21,24). To determine the second order rate constant (M Ϫ1 s Ϫ1 ), the first order rate constants obtained from the single exponential fitting were divided by the concentration of MTS used (26). The normalized current was plotted as a function of the cumulative duration of HMTS application and fitted with a single exponential function using GraphPad Prism3 software (San Diego, CA).
Data Analysis-All of the values are presented as the means Ϯ S.E. of the mean (S.E.) from four or more independent experiments. Nonlinear regression analysis was performed to determine EC 50 and Hill coefficient values from GABA concentration-response curves by using GraphPad Prism3 software (San Diego, CA). The data were normalized to the maximal current in each oocyte and fitted according to the equation of the form: where I is the current, I max is the maximal current recorded in a given oocyte, EC 50 is the GABA concentration for half-maximal current response, A is the GABA concentration, and n is the Hill coefficient.
Statistical significance of the difference between each mutant and wild type were also analyzed by one-way analysis of variance with the Dunnett's post hoc test with p Ͻ 0.05 representing significance using this software.

Expression and Functional Characterization of Cysteine Mutants-GABA
A receptors composed of only ␣ and ␤ subunits are able to bind GABA or alcohols and transduce their effects in the recombinant system (4,6). However, the ␣␤␥ subunit (particularly ␣ 1 ␤ 2 ␥ 2 ) composition of GABA A receptor are the prevalent combination in the mammalian brain (9,14). Therefore, we used ␣ 1 ␤ 2 ␥ 2S combination for wild type and mutant GABA A receptors expressed in Xenopus oocytes in this study. To assess whether cysteine mutations affected GABA A receptor function, each individual cysteine mutant ␣ 1 subunits were coexpressed with wild type ␤ 2 and ␥ 2S subunits in oocytes. Then the GABAinduced currents (I GABA ) and alcohol potentiation of I GABA were determined (Tables I and II). Most of the mutants showed GABA-induced currents similar to wild type, except for the A300C mutant, which showed little current response to 1 mM GABA (Ͻ200 nA). In general, cysteine substitutions were well tolerated within the region Ile 290 -Val 307 of TM3. The mutations produced some changes in GABA sensitivity, with the most sensitive (Y294C) and the least sensitive (L301C) having GABA EC 50 values about 4-fold different from wild type receptors. For wild type receptors, 200 mM ethanol or 0.5 mM hexanol potentiated GABA EC 5-10 -induced currents by 66 Ϯ 5 and 70 Ϯ 7%, respectively (Table II). These concentrations approximately correspond to the anesthetic concentration in vivo (27,28). Ethanol and hexanol altered GABA EC 5-10 -induced currents of all other cysteine mutants by amounts ranging from Ϫ8 Ϯ 4 to 67 Ϯ 8% and 35 Ϯ 4 to 93 Ϯ 4%, respectively (Table II).
Reaction of Introduced Cysteines with MTS Reagents in the Resting State-We next examined the surface accessibility of uncharged propyl-and hexyl-methanethiosulfonate (PMTS and HMTS) as sulfhydryl-specific reagents to covalently label cysteines introduced into the TM3 domain. The wild type or mutant ␣ 1 subunits were coexpressed with wild type ␤ 2 and ␥ 2S subunits in Xenopus oocytes. For PMTS and HMTS, we first determined GABA-induced currents in Xenopus oocytes expressing wild type GABA A receptors before and after treatment with MTS compounds. Exposure of wild type GABA A receptors to MTS compounds had no significant effects on GABA-induced currents (Figs. 1 and 2). Additionally, application (1.5 min) of MTS compounds in the presence of GABA (1 mM), ethanol (200 mM), or hexanol (0.5 mM) did not affect GABA-induced currents in wild type GABA A receptor (Figs. 1-3). Thus, endogenous cysteine residues were inaccessible for reaction with the MTS compounds, or reaction with the MTS compounds had no functional effects in wild type GABA A receptors under these conditions (resting, GABA-or alcohol-bound states). Next, we asked whether conformational changes induced by agonist (GABA) or modulators (ethanol or hexanol) in the mutant receptors alter the accessibility of MTS reagents, PMTS and HMTS, to introduced cysteine in TM3. First, GABA-induced currents were recorded before and after treatment with MTS reagents without GABA or alcohols (resting state). The GABA-induced currents from A291C and Y294C mutants receptors were significantly changed after applying MTS reagents to these mutant receptors in the resting state ( Figs. 1 and 2). Interestingly, the Y294C mutant showed opposite results from PMTS and HMTS treatments: PMTS inhibited GABA-induced currents, but HMTS potentiated GABA-induced currents at Y294C (Fig. 3). The results from A291C and Y294C mutants were consistent with previous findings that a water pocket exists around these amino acids (15,22). Application of MTS reagents in the resting states did not affect the function of any other receptors (Figs. 1  and 2). This implies that these mutant receptors in the resting states are not accessible to MTS reagents, or reaction of the MTS reagents had no significant effects on mutant receptors.
Reaction of Introduced Cysteines with MTS Reagents in the GABA-bound State-Next, we asked whether application of MTS reagents in the presence of GABA (1 mM) affects GABAinduced currents of the mutant GABA A receptors. Application of MTS reagents with 1 mM GABA significantly changed the GABA EC 5-10 -or EC 50 -induced currents of the A291C, Y294C, F296C, and V297C mutants (Figs. 1 and 2). Y294C or F296C mutants had no significant effects of PMTS in the GABA-bound state but showed significant effects of HMTS in the GABAbound state (Figs. 1 and 2). The A291C and Y294C mutants were also affected by MTS reagents in the resting state. Treatment with PMTS or HMTS in the presence of GABA significantly inhibited GABA-induced currents at A291C, F296C, and V297C mutants (Figs. 1-3). Results of the Y294C mutant were similar to results obtained in the resting state. PMTS with GABA also inhibited GABA-induced currents, but HMTS with GABA potentiated GABA-induced currents at Y294C mutant ( Figs. 1 and 2). The other mutants were not significantly affected by MTS reagents in this condition.
Reaction of Introduced Cysteines with MTS Reagents in the Alcohol-bound States-Finally, GABA-induced currents in oocytes expressing mutant GABA A receptors were determined before and after treatment of MTS reagents with ethanol or hexanol to test whether there are changes of accessibility of MTS reagents in the alcohol-bound state. GABA-induced currents of A291C, Y294C, A295C, and F296C mutant receptors were significantly changed after applying MTS reagents with alcohols ( Figs. 1 and 2). A295C or F296C mutants had no significant effects of PMTS in the ethanol-or hexanol-bound states but showed significant effects of HMTS in the ethanoland hexanol-bound states ( Figs. 1 and 2). MTS reagents applied with alcohol had no significant effect on all other mutants tested ( Figs. 1 and 2). A291C and Y294C mutants were already reactive to MTS reagents in the resting and GABA-bound states. Treatment with PMTS or HMTS in the presence of ethanol or hexanol significantly inhibited GABA-induced currents at A291C, A295C, and F296C mutants. For the Y294C mutant, PMTS with ethanol or hexanol also inhibited GABAinduced currents, and HMTS with ethanol or hexanol potentiated the currents (Figs. 1 and 2).
Reaction Rates of HMTS with Accessible Cysteine Mutants in the Resting, Alcohol-bound, or GABA-bound States-The reaction rates of MTS reagents with an introduced cysteine side chain are determined by the access pathway to the cysteine and by the local environment of sulfhydryl group of the substituted cysteine residue (18,24,25). To investigate the local environment of reactive substituted cysteines, reaction rates of the accessible mutants were determined. First, A291C and Y294C mutants are accessible to HMTS in the resting state, and we asked whether the physical environments around A291C or Y294C residues in the resting state are different from environments in the alcohol-bound or GABA-bound states. To accomplish this, we measured the rate of reactions of HMTS to A291C and Y294C mutants in the absence (Fig. 4) and presence of alcohols and GABA. Rate constants of A291C mutant were similar for resting, alcohol-bound and GABA-bound states (Table III). Y294C in the GABA-bound state reacted significantly faster with HMTS than in the resting or alcohol-bound states (Table III). This indicates that GABA-induced conformational changes around Y294C residue caused increased interaction between HMTS molecules and Y294C residue. The rate constant of F296C mutant in presence of GABA was similar to rates in the presence of ethanol or hexanol (Table III). Additionally, the A295C and V297C mutants show specific state-dependent accessibility of MTS reagents in the presence of alcohols and GABA, respectively (Table III).

Accessibility of MTS Reagents and Structural
Rearrangements-With SCAM, it is assumed that the cysteine residues that react with MTS reagents are exposed at a water-accessible protein surface because water is required to ionize cysteine residues. MTS reagents react with ionized thiolate groups

FIG. 1. Effects of PMTS on GABA-induced currents of wild type and mutants GABA A receptors in the resting state (A), ethanolbound state (B), hexanol-bound state (C), or GABA-bound state (D).
For GABA-induced currents, the GABA EC 5-10 (top bar, white) and EC 50 (lower bar, gray) were applied on wild type (WT) and mutant GABA A receptors. Black bars indicate effects that are significantly different statistically from the effect on wild type in each condition by a one-way analysis of variance, using the Dunnett's post hoc test. The percentage of change was calculated as ϭ {(I after /I before ) Ϫ 1} ϫ 100, where I before and I after indicate the values of the two GABA-induced currents before and after the application of the sulfhydryl reagent (0.5 mM applied for 90 s). All of the values are presented as the means Ϯ S.E. from three to eight oocytes. *, A300C receptor is not included because of little current response to GABA.

FIG. 2. Effects of HMTS on GABA-induced currents of wild type and mutants GABA A receptors in the resting state (A), ethanolbound state (B), hexanol-bound state (C), or GABA-bound state (D).
For GABA-induced currents, the GABA EC 5-10 (top bar, white) and EC 50 (lower bar, gray) were applied on wild type (WT) and mutant GABA A receptors. Black bars indicate effects that are significantly different statistically from the effect on wild type in each condition by a one-way analysis of variance, using the Dunnett's post hoc test. The percentage of change was calculated as ϭ {(I after /I before ) Ϫ 1} ϫ 100, where I before and I after indicate the values of the two GABA-induced currents before and after the application of the sulfhydryl reagent (0.5 mM applied for 90 s). All of the values are presented as the means Ϯ S.E. from three to eight oocytes. *, A300C receptor is not included because of little current response to GABA.
(ϪS Ϫ ) 10 9 times faster than with un-ionized thiols (ϪSH) (18,29). Previous results have indicated that the subset of cysteine substitution mutants that react with sulfhydryl-specific reagents are markers for the specific conformational states induced by different ligands (15-17, 19, 21, 25, 26, 30 -32). In particular, using SCAM it was shown that TM3 of the GABA A receptor ␣ 1 subunit undergoes conformational changes during channel gating and allosteric benzodiazepine or propofol modulation (16,17). Alcohols are also assumed to produce allosteric modulation of GABA A receptors, but nothing is known about conformational changes produced by alcohols and possible overlap with changes produced by GABA, benzodiazepines, and propofol. In this study, we used SCAM to investigate whether there are common conformational changes in TM3 of the GABA A receptor ␣ 1 subunit during alcohol binding and GABA binding. In this study, we assume that the mutant is accessible to the MTS reagent if either of GABA EC 5-10 or EC 50 -induced currents are affected after treatment by an MTS reagent. We found that the A291C and Y294C mutants in the resting state were significantly accessible to MTS reagents, consistent with these amino acids at the extracellular end of TM3 being on the water-accessible protein surface or facing a water-filled cavity as was proposed earlier (15,23). It has been suggested that this water-filled cavity is connected to the extracellular solution, at least transiently allowing water and MTS reagents to enter and that residues lining the cavity can therefore react with applied MTS reagents in the resting state (17). MTS reagents also reacted with the A291C, Y294C A295C, and F296C mutants in the presence of alcohols. The results imply that an alcohol-induced conformational change may induce the waterfilled cavity around A291C and Y294C to extend deeper, causing the A295C and F296C residues to be accessible to MTS reagents (Fig. 5). Another possibility is that alcohol binding could induce the formation of a separate water-filled cavity around A295C or F296C that is transiently connected to the extracellular solution to allow PMTS or HMTS to reach these residues. In this case the cavity around A291C/Y294C may not be connected to the one that forms in the alcohol-bound state around A295C/F296C. Given the proximity of these residues, however, this seems unlikely. Furthermore, exposure of the A291C, Y294C, F296C, and V297C mutants to MTS reagents in the presence of GABA had significant effects on their GABAinduced currents, indicating that the water-accessible surface around A291C and Y294C residues widened to include F296C and V297C because of a structural movement induced by GABA-binding (Fig. 5). It is also possible that residue Val 297 simply becomes accessible by distortion of this local environment in the presence of GABA. However, Val 297 is located  Table III. below Tyr 294 in our model (Fig. 5); therefore, it is likely that V297C becomes accessible to MTS reagents by expansion of the cavity containing water and MTS reagents in the GABA-bound state. These data also support the idea that an extracellular structural change produced by binding of GABA to the Nterminal region must be transferred to this TM3 region possibly through the TM2-3 linker (19,33). The A291C and Y294C mutants are accessible to MTS reagents in the resting state and also in the alcohol-bound and in the GABA-bound states. Introduction of cysteines deeper than Val 297 within TM3 did not reveal any effect of MTS reagents in the absence or presence of GABA, ethanol, or hexanol in our studies. This suggests that MTS reagents are not accessible to these amino acids, although we cannot rule out the alternative possibility that the modification of the amino acids by these alkyl MTS derivatives is functionally silent. It should be noted that some sites (F298C, S299C, L301C, I302C, or V307C in the different states) appeared to be reactive to MTS reagents, but statistical comparison of wild type and these mutants did not show significant effects ( Figs. 1 and 2). Taken together, this suggests that the extracellular side of TM3 might be more flexible or dynamic compared with the intracellular side in the presence of agonist or modulator. Recently, it was shown that the protein packing is loose around the extracellular half of the GABA A receptor ␤ 1 subunit TM2 in the presence of GABA (34), consistent with this study that the extracellular side of TM3 might be more flexible. This region composed of A291C, Y294C, A295C, F296C, and V297C of TM3 may be involved in structural movements that transduce the allosteric modulation of GABA A receptor.
It is of interest to compare our data with those from previous studies that used the same mutants but different thiol reagents (15,16). The mutants A291C, Y294C, F296C, F298C, A300C, L301C, and E303C were found to be accessible to para-chloromercuribenzene sulfonate in the presence of GABA. Our data showed that MTS reagents were reactive with A291C, Y294C, F296C, and V297C in GABA-bound state. It should be noted that structurally and functionally different thiol reactive agents could show different accessibilities or reactivity with the same cysteine mutation or different reagents may have very different functional effects following modification (15,26,35,36). Furthermore, to investigate the accessibility GABA EC [5][6][7][8][9][10] or EC 50 was used in this study and GABA EC 50 or nearly saturating GABA were used in the study from Williams and Akabas (15), but the more important finding is that both studies show evidence for conformational changes in the TM3 of GABA A receptor ␣ 1 subunit induced by binding of agonist or modulators.
Reaction Rates of HMTS with the Reactive Mutants in the Different Functional States-The microenvironment and access pathway to the introduced cysteines can be characterized by determining the rate constants of reaction with MTS reagents (18,24,25). The rate at which MTS reagents react with a cysteine side chain depends on the collision frequency between the MTS reagent and the ionized sulfhydryl group. The collision frequency depends on the local concentration of the MTS reagent in the vicinity of the cysteine. This is influenced by steric factors in the access pathway from bulk solution and at the site of the cysteine. The extent to which a cysteine ionizes depends on fractional time that the residue is in contact with water and bulk solution pH and on the local electrostatic potential. To investigate the local physical environment of the accessible introduced cysteines of TM3 of the GABA A receptor ␣ 1 subunit, we determined the rate constant of reaction of HMTS to each of the reactive cysteine mutants. The rate constants of the A291C mutant did not vary with treatments, suggesting that the local environment around the A291C residue in the alcohol-bound states did not change significantly from environment in the resting-or GABA-bound states. The rate constant of the Y294C mutant in the GABAbound state is significantly faster than the rate constants in the resting or alcohol-bound states. This indicates that the local structure around Y294C and/or the access pathway stabilized by GABA is different from the resting and alcoholbound states. This is consistent with previous results with Y294C (16). F296C is accessible to MTS reagents in the alcohol-and GABA-bound states. The reaction rate of the F296C mutant in the GABA-bound state is not significantly different from reaction rate in the alcohol-bound states. Therefore, the microenvironment of the F296C residue induced by alcohol binding seems to be similar to local environment induced by GABA binding.
Conclusions-Our data suggest that conformational changes induced by GABA or alcohol cause the water-accessible surface of the TM3 segment to increase from the region around A291C and Y294C to the deeper region surrounding A295C, F296C, or V297C. Based on results of functional accessibility and reaction rate of MTS reagents, our data also suggest that the extracellular side of TM3 is more flexible or dynamic during channel gating or alcohol modulation and the cytosolic side of TM3 might adopt a more tightly packed or rigid conformation.
There are several possible implications of these findings. First, if we assume that alcohols are binding in the existing water-filled cavity formed in part by TM3, then it appears that this binding promotes the expansion of this cavity or stabilizes a receptor conformational state in which the cavity is larger. This may allow binding of additional molecules of alcohol and may also mimic some of the actions of GABA, thereby promoting channel opening. In a more general extension, it is of interest to note that the alcohol-binding site in the Drosophila protein LUSH has some similarity to the putative alcoholbinding site TM2/3 region of GABA A receptors (37). This binding region in LUSH is a water-filled cavity that adopts multiple conformations when occupied by water but is stabilized with limited movement when alcohols displace water (37,38). Thus, the high flexibility observed for this region of TM3 in the present study may also allow for multiple conformations of the water-filled cavities, and perhaps, occupation of the cavity by alcohol selectively stabilizes substrates that in turn stabilize the open state of the channel.