Sites of Excitatory and Inhibitory Actions of Alcohols on Neuronal α2β4 Nicotinic Acetylcholine Receptors

The nicotinic acetylcholine receptor (nAChR) is the best characterized member of the superfamily of cysteine-loop ligand-gated ion channels (Karlin, 2002). The alcohol modulation of these pentameric receptors has been studied extensively. Alcohols enhance the function of γ-aminobutyric acid receptors type A (GABA_AR) and glycine receptors (GlyR); increasing the chain length increases the alcohol potency (Mascia et al., 1996; Wick et al., 1998). In the case of nAChRs, alcohol action is more complex. Early studies of the muscle nAChR showed that short-chain alcohols potentiate, while long-chain alcohols inhibit, their function (Bradley et al., 1984); the same phenomenon was later observed for Torpedo (Wood et al., 1991) and neuronal (Godden et al., 2001; Zuo et al., 2001) receptors. Modulation of nAChR function by ethanol is thought to be important for several behavioral effects of the drug, including ethanol self-administration (Söderpalm et al., 2000).

Studies carried out in GlyR and GABA_AR indicated the existence of specific binding sites for alcohols and volatile anesthetics located between transmembrane (TM) segments 2 and 3 (Mihic et al., 1997; Wick et al., 1998; Ye et al., 1998; Jenkins et al., 2001). Compelling evidence came from mutating the critical amino acid in TM2 to cysteine in the GlyR and labeling it with sulfhydryl-specific reagents that are alcohol analogs and anesthetics in vivo (Mascia et al., 2000; Zhang et al., 2000). When alkylthiols were covalently bound to cysteines introduced at critical positions in TM2 [GlyR α1(S267C) or GABA_AR α2(S270C)β1], there was an irreversible potentiation of the receptor function mimicking the effect of alcohols on these receptors, and applying alcohol no longer enhanced the receptor function, which is consistent with an occupied alcohol binding site (Mascia et al., 2000). In the nAChR the homologous position in α2 is L261, but we found that mutations at this position did not reduce alcohol effects (Borghese et al., 2002), indicating that the alcohol binding sites in the nAChR do not directly correspond to the one in GlyR and GABA_AR.

Evidence from studies on Torpedo and muscle nAChRs suggested that long-chain alcohols bind in the channel pore (Bradley et al., 1984; Forman et al., 1995; Forman and Zhou, 2000; Zhou et al., 2000), and the photoactivable anesthetic 3-azioctanol labeled mainly αE262 in Torpedo nAChR TM2 (Pratt et al., 2000). These results, together with the studies in GABA_A and GlyRs, and the structural similarity among these receptors, point to the likelihood of alcohols binding to the nAChR αTM2, near the extracellular side. This leads us to several key questions: 1) are there distinct alcohol binding sites for the inhibitory and excitatory effects? 2) if so, how can they be located and differentiated? and 3) do other anesthetics interact with one or both of these sites?

The purpose of the present study was to define alcohol binding sites in the nAChR using propyl methanethiosulfonate (PMTS, a sulfhydryl-specific reagent). This compound reacts covalently with cysteine residues, leaving propanethiol attached to the cysteine thiol group acting as an alcohol analog. This strategy avoids one of the problems of site-directed mutagenesis: does the mutation cause a direct change of the ligand binding site or an allosteric modification of the receptor function that is unrelated to ligand binding (Colquhoun, 1998)? In the present study, that difficulty is overcome by covalently linking an alcohol analog to a specific site. Our key predictions are that reaction of PMTS with a cysteine-forming part of a binding site will mimic the specific action of an alcohol at this site and reduce the effect of a subsequent application of alcohol.

As noted above, the heteromeric neuronal nAChRs are enhanced by short-chain alcohols and inhibited by long-chain alcohols (Godden et al., 2001; Zuo et al., 2001, 2002), and alcohols produce dual effects on a number of different nAChR subunit combinations, depending on the chain length. We used the α2β4 nAChR because we have extensive experience with this combination, including mutagenesis, and it is strongly modulated by alcohols (Cardoso et al., 1999; Godden et al., 2001; Borghese et al., 2002). Moreover, the TM2 segment is well conserved among α subunits that form heteromeric nAChRs (Fig. 1A). The most abundant heteromeric nAChR subunits in brain are α4 and β2, but the exact subunit composition and functional importance of specific neuronal nAChRs remain controversial (Picciotto et al., 2000). We consider α2β4 as a good model of heteromeric neuronal nAChRs, but additional studies would be required to extend present findings to other types of nAChRs.

Materials and Methods

Materials. Adult female Xenopus laevis frogs were obtained from Xenopus Express, Inc. (Plant City, FL) and Nasco (Fort Atkinson, WI). Acetylcholine (ACh) chloride, atropine sulfate, ethanol, 1-butanol, 1-pentanol, 1-octanol, and urethane were purchased from Sigma-Aldrich (St. Louis, MO). Isoflurane was obtained from Marsam Pharmaceuticals Inc. (Cherry Hill, NJ). MTS reagents were purchased from Toronto Research Chemicals Inc. (North York, Toronto, Canada). All other reagents were of reagent grade.

Site-Directed Mutagenesis. The cDNAs encoding the rat nAChR α2 and β4 subunits were in pSP65 and pBluescript SK plasmids, respectively, and were kindly provided by Dr. Charles W. Luetje (University of Miami). The notation we used for the mutations denotes the position in the mature protein sequence of the wild-type receptor. The desired mutations were obtained using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). Briefly, complementary mutagenesis primers were designed with the desired mutation at the appropriate place and polymerase chain reaction was performed using the cDNA encoding the wild-type α2 clone and the mutagenesis primers. Afterward, the parental DNA was digested with DpnI, Epicurian Coli XL1-Blue supercompetent cells were transformed with the polymerase chain reaction product, and the plasmids obtained through minipreps were sequenced to verify the desired mutation.

Transcription and Oocyte Injection. After linearization, the cDNA encoding the wild-type and mutant subunits was used as a template for the synthesis in vitro of 5′-capped RNA (mCAP RNA capping kit, Stratagene). X. laevis oocytes were manually isolated from a surgically removed portion of ovary. Oocytes were treated with collagenase (type IA, 0.5 mg/ml) for 10 min and then placed in sterile modified Barth's solution (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 10 mM HEPES, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, 0.91 mM CaCl₂, adjusted to pH 7.5) supplemented with 10,000 units of penicillin, 50 mg of gentamicin, 90 mg of theophylline, and 220 mg of sodium pyruvate per liter (incubation medium). Oocytes were then injected into the cytoplasm with 30 nl of diethyl pyrocarbonate-treated water containing 10 to 40 ng of cRNA encoding the α2 and β4 subunits in a 1:1 (w/w) ratio. The injected oocytes were kept at 13°C in incubation media.

Electrophysiological Recordings. Recordings were carried out 2 to 7 days after injection. The oocytes were placed in a rectangular chamber (approximately 100 μl) and continuously perfused with buffer (2 ml/min) at room temperature (23°C). Due to the endogenous Ca²⁺-dependent Cl^– current, the experiments were carried out in a nominal absence of Ca²⁺. The perfusion buffer composition was 115 mM NaCl, 1.8 mM BaCl₂, 2.5 mM KCl, 10 mM HEPES, pH 7.3). Atropine (1 μM) was included in the perfusion buffer for some oocytes to test for endogenous muscarinic responses, but none were detected; the presence of atropine did not modify the results. The whole-cell voltage clamp at –70 mV was achieved through two glass electrodes (1.5–10 MΩ) filled with 3 M KCl, using a Warner Instrument (Hamden, CT) model OC-725C oocyte clamp.

All drugs were applied by bath-perfusion. PMTS was applied at a concentration of 0.5 mM. All solutions were prepared the day of the experiment, and PMTS, octanol, and isoflurane solutions were prepared immediately before use. The modulator concentrations were selected to provide a reliable, but submaximal and nontoxic, modulation of the receptor function; they ranged from half the anesthetic EC₅₀ value for isoflurane to four times the anesthetic EC₅₀ value for urethane (Alifimoff et al., 1989; Franks and Lieb, 1994; Hara and Harris, 2002). The preapplication of alcohol provided more reliable responses in previous studies (Cardoso et al., 1999), and preapplication of anesthetics was also used.

The concentration-response curves (CRCs) were obtained with increasing concentrations of ACh, applied for 20 s at intervals ranging from 5 to 10 min. From these CRCs the concentration evoking a half-maximal response (EC₅₀) was calculated, along with the Hill coefficient (see Statistical Analysis). For the analysis of the alcohol effect at different ACh concentrations, increasing ACh concentrations were applied with and without alcohol. For each concentration ACh was applied, followed by a 5- or 10-min washout; then the alcohol was preapplied for 1 min, immediately followed by ACh plus alcohol, another washout, and ACh applied again; all the responses were calculated with respect to the maximal current in the absence of alcohol. To analyze the effect of PMTS in the ACh EC₅₀ and maximal response, increasing ACh concentrations were applied, then PMTS was applied, and the same ACh concentrations were applied again. The CRC parameters were calculated as before, assigning a value of 100% to the maximal ACh response before PMTS application.

In the experiments on the effect of PMTS on ACh responses and its modulation by alcohols or anesthetics, the ACh concentration used was the EC₅₀ value for that particular receptor; all ACh applications lasted 20 s and were followed by a 5-min washout. The protocol consisted of an ACh application followed by a second one (considered to be the initial response, assigned a value of 100%); allosteric modulator for 1 min immediately followed by ACh plus the same modulator; ACh; PMTS for 1 min (unless noted otherwise); two applications of ACh; allosteric modulator for 1 min immediately followed by ACh plus the same modulator; and the last ACh application. All ACh responses are expressed as percentages of the initial response. To calculate the ACh response value before and after PMTS application, the means of the ACh responses before and after the coapplication of the allosteric modulator were considered. The potentiation or inhibition by allosteric modulators was calculated as where ACh + Mod represents the response to ACh in the presence of a modulator, and ACh is the mean of the previous and subsequent ACh applications. All experiments shown include data obtained from oocytes taken from at least two different frogs.

Statistical Analysis. Nonlinear regression analysis was performed with Prism (GraphPad Software Inc. San Diego, CA). CRCs were fitted to the equation where I represents the current, I_max the maximal current, EC₅₀ the agonist concentration for half-maximal response, [ACh] the ACh concentration, and n_H the Hill coefficient. For periodicity analysis, the function Y_(X) = baseline + amplitude · sine (frequency · X + phase shift) was used. Results were analyzed by two-tailed paired t test where appropriate. Data are represented as mean ± S.E.

Molecular Model. We previously used ClustalW to prepare multiple sequence alignments of six ligand-gated ion channels to predict that the topology of the TM segments of all subunits of AChRs is a bundle of four antiparallel α-helices (Bertaccini and Trudell, 2002). As a template for homology modeling we selected a four-α-helical bundle from the crystal structure of bovine cytochrome c oxidase, chain C (PDB keyword, 2occ) (Yoshikawa et al., 1998). As previously described (Yamakura et al., 2001; Trudell and Bertaccini, 2002), we superimposed five copies of this bundle onto a template of a pentameric ion channel found in the crystal structure of MscL, a bacterial mechanosensitive receptor (Chang et al., 1998). The resulting pentameric structure containing 20 α-helices was then used as a template for threading the primary sequences of nAChR α2 and β4 subunits. For each nAChR subunit the predicted α-helical segments were assigned corresponding coordinates from the structure of bovine cytochrome c oxidase, whereas the interhelical loops were generated in the Homology module of Insight 2000 (Accelrys, Princeton, NJ). The entire structure was subjected to sequentially restrained molecular mechanics energy optimization with Discover_3 using the CFF91 force field (Accelrys).

Results

Our strategy for detection of alcohol binding sites requires substitution of cysteine residues in the TM2 segment and its subsequent exposure to a sulfhydryl-specific reagent. Amino acids from S256 to E266 (10′ to 20′, Fig. 1B; see Fig. 1A and Table 1 for general notation) in nAChR α2 TM2 were mutated to cysteine one at a time, and the corresponding cRNA injected into X. laevis oocytes along with the cRNA for β4. Four of the mutants failed to provide detectable currents and were not studied further [α2(S256C)β4, α2(F260C)β4, α2(I264C)β4, and α2(E266C)β4]. The ACh CRCs showed that the replacement of the wild-type amino acid by cysteine resulted in an increased affinity for ACh [α2(V259C)β4, α2(L262C)β4, and α2(L263C)β4], and a slightly decreased ACh affinity for α2(T265C)β4 (Table 1). The variation of EC₅₀ ACh with the cysteine position showed a periodicity of 3.5 ± 0.2, as expected from a α-helical structure (Fig. 1C). Nevertheless, the variation in the EC₅₀ values is relatively small, suggesting that the introduction of a cysteine at these positions did not induce marked structural alterations in the nAChR.

Our goal was to detect residues where reaction with PMTS alters both receptor function and receptor modulation by ethanol. It is assumed that a change in the receptor properties indicates that the covalent reaction has taken place and propanethiol is covalently bound to the cysteine. To react with PMTS, the cysteine thiol group has to be in the reactive thiolate form (the reaction rate with thiolate is 10⁹ faster than with thiol; Roberts et al., 1986); therefore, the cysteine residue has to be surrounded by water, either in the channel or in a water-filled cavity. Once the cysteine in the alcohol binding site is labeled, the ACh responses and alcohol effects would be modified. The absence of change in the receptor function after PMTS application can be due to the absence of reaction, or the reaction has occurred but has no functional consequences; in either case that cysteine is irrelevant to us, since a cysteine located in an alcohol binding site would be water-accessible and there are no barriers for the neutral PMTS to reach that cysteine.

Most of the cysteine mutants tested were largely unaffected by the PMTS treatment (Table 2). Changes in the ACh response were observed after the sulfhydryl-specific reagent application for α2(T258C)β4 and α2(V259C)β4, but the ethanol effect in these mutants was not altered after PMTS labeling. Conversely, ethanol modulation of ACh responses in α2(L257C)β4 was modestly increased, but the ACh responses themselves were not modified after PMTS application.

Two cysteine mutant receptors showed consistent and significant effects of PMTS both on ACh responses and ethanol action: α2(L262C)β4 and α2(L263C)β4 (16′ and 17′). The remainder of the experiments focused on these two receptors.

To evaluate possible nonspecific alterations in the ACh or alcohol effects caused by labeling with PMTS we conducted a series of control experiments. First, we studied the effect of ethanol and PMTS application in the wild-type receptor at different ACh concentrations. Ethanol (200 mM) potentiated ACh responses at all concentrations tested (Fig. 2A), and the potentiation was similar for concentrations higher than EC₂₀ (Fig. 2B). This is consistent with previous studies in human α2β4 neuronal nAChR (Cardoso et al., 1999). PMTS application did not modify the responses to different ACh concentrations in the wild-type receptor (Fig. 2C). The CRC parameters are detailed in Table 3.

Given these results for the wild-type receptor, we tested the consequences of alcohol and PMTS treatment on mutant α2(L262C)β4 and α2(L263C)β4 nAChRs. Ethanol (200 mM) potentiated ACh responses in α2(L263C)β4 nAChRs at all concentrations tested (Fig. 3A), increasing the maximal response without modifying the EC₅₀ value (Table 3); application of PMTS resulted in the same modifications of the CRC, that is, an increase of the maximal response without changing the ACh affinity (Fig. 3B, Table 3). Octanol (57 μM) inhibited ACh responses in α2(L262C)β4 nAChRs at all concentrations tested (Fig. 3C), decreasing the maximal response without modifying the EC₅₀ value (Table 3); the same changes were observed after application of PMTS (Fig. 3D, Table 3). Because the EC₅₀ value was not modified by PMTS treatment in either wild-type or mutant nAChRs, we used the same ACh concentration before and after PMTS application.

In Fig. 4 we show representative tracings of recordings in wild-type and cysteine mutant nAChRs to illustrate the application protocol and typical results. ACh responses and ethanol effects before and after PMTS application are shown in Fig. 5. It should be noted that there were small (8–9%) but significant changes in the wild-type ACh responses after the PMTS application, but this decrease was mainly due to rundown of the responses. For α2(L262C)β4 and α2(L263C)β4, the effects of PMTS on ACh responses were again clear and consistent: there was a 64% inhibition for α2(L262C)β4, while the ACh responses of α2(L263C)β4 increased about 100% (Fig. 5A). The ethanol effect on ACh responses remained unchanged for wild-type nAChRs after exposure to PMTS, but it increased significantly for α2(L262C)β4 (309% potentiation) and it was essentially abolished in α2(L263C)β4 nAChRs after exposure to PMTS (Fig. 5B). To achieve more complete labeling of the cysteine with PMTS we increased the application time for the sulfhydryl-specific compound from 1 min to 5 min (pilot studies showed maximal effect after 5 min). With longer exposure to PMTS the changes in the ACh responses were increased in the mutants [77% inhibition for α2(L262C)β4 and 424% potentiation for α2(L263C)β4], while the small change in the wild-type receptor remained very small (11% inhibition), as shown in Fig. 5C. The ethanol effect on ACh responses was not altered in wild-type receptors, while it was even more pronounced in α2(L262C)β4 receptors (420% enhancement); interestingly, 200 mM ethanol alone produced direct activation of the channel after a 5-min application of PMTS (data not shown). In α2(L263C)β4 receptors the effect of ethanol was markedly altered by the longer exposure to PMTS, as the potentiation (52%) observed before the PMTS application was changed to inhibition (–27%) (Fig. 5D). However, the prolonged application of PMTS was harmful to the oocytes and many did not survive the treatment; therefore, we used a 1-min application of PMTS in all subsequent studies, although this likely produced only partial labeling of the cysteine residues.

The effects of intermediate-chain alcohols on ACh responses before and after the treatment with PMTS are shown in Fig. 6. We tested butanol (22 mM, Fig. 6A) and pentanol (6 mM, Fig. 6B) since they represent the transition from the enhancing short-chain alcohols to the inhibitory long-chain alcohols. Butanol slightly enhanced and pentanol slightly inhibited the ACh responses in wild-type receptors; these effects were not altered by treatment of WT receptors with PMTS. Both alcohols induced a small potentiation in α2(L262C)β4 receptors, and for both compounds the potentiation was greatly increased after PMTS application. An opposite situation was observed for α2(L263C)β4 receptors: although butanol did not have any effect on α2(L263C)β4, a clear inhibition was observed after PMTS treatment; pentanol had a small inhibitory effect on this receptor, which was almost tripled after the application of PMTS. As expected, octanol (114 μM) had a clear inhibitory effect on wild-type receptors that was not altered after treatment with PMTS. The inhibitory effect of octanol on α2(L262C)β4 receptors was reduced by half after PMTS application, while it was significantly increased in α2(L263C)β4 receptors (Fig. 6C).

To extend these findings to general anesthetics with structures different from n-alcohols, we tested urethane (40 mM, Fig. 7A) and isoflurane (0.15 mM, Fig. 7B). They were selected because urethane potentiates and isoflurane inhibits ACh responses in wild-type nAChRs (Yamakura et al., 2001; Hara and Harris, 2002). In the cysteine mutants urethane behaved like a short-chain alcohol: its potentiation was enhanced in α2(L262C)β4 and abolished in α2(L263C)β4 receptors after treatment with PMTS. In contrast, isoflurane inhibition was not modified in wild-type or α2(L262C)β4 receptors after the application of PMTS, and in α2(L263C)β4 receptors only a small augmentation of isoflurane inhibition was observed.

To assist in visualizing the location of these mutations in the receptor we constructed a composite model with alternating nAChR α2 and β4 subunits arranged with pseudosymmetry about a central ion channel (Fig. 8). Three amino acid residues are shown in α2TM2, near the extracellular surface; cysteine residues are represented in these three locations. Position 261 (15′) corresponds to the amino acid that determines anesthetic and alcohol sensitivity in GABA_AR and GlyR channels, and points toward a region between the four α2TM segments. Position 262 (16′) is in an intersubunit location near α2TM3, whereas position 263 (17′) is tangential to the ion pore, near β4TM1.

Discussion

Our results provide considerable support for two independent alcohol binding sites in neuronal nAChRs, one excitatory and another inhibitory, and provide the first evidence that different residues in αTM2 participate in each of these alcohol binding sites.

Using the same approach previously applied to GlyRs and GABA_ARs (Mascia et al., 2000), we introduced cysteines in the nAChR αTM2 and covalently bound propanethiol to the cysteine thiolate upon treatment with PMTS. Thus, we identified two residues that showed altered ACh responses and modulation by alcohols after PMTS treatment. According to our hypothesis, alcohol binding to one of these sites inhibits ACh responses, whereas alcohol binding to the other enhances the receptor function. Short-chain alcohols would preferentially interact with the enhancing site and long-chain alcohols with the inhibitory site; intermediate-chain alcohols would interact with both, and their effects would vary according to the experimental conditions.

What are the predictions from this hypothesis? First, we will assume that we have replaced a key residue in the inhibitory site with cysteine. Then irreversible binding of propanethiol to this cysteine should mimic the action of alcohol at this site, i.e., it should inhibit ACh responses; if analyzed at different ACh concentrations, propanethiol covalently bound to the cysteine at the inhibitory site should decrease the maximal response to ACh without modifying the EC₅₀ value. Furthermore, the propanethiol covalently bound to the cysteine should reduce the specific alcohol action on that site, i.e., the long-chain alcohols should have less inhibitory effect and the short-chain alcohols should have greater enhancing effect. The consequences of covalent binding of propanethiol to the enhancing site would have the opposite effect, but still would not modify the ACh EC₅₀ value. It is also of interest to consider intermediate-chain alcohols, which produce little modulation of wild-type receptors. Selective covalent binding of propanethiol to the inhibitory or excitatory sites after PMTS treatment would be predicted to reveal actions of the intermediate-chain alcohols. According to our hypothesis intermediate-chain alcohols would bind at both sites, but their actions would cancel each other and result in no net effect.

The present study provides strong support for each proposal. The treatment of α2(L263C)β4 nAChRs with PMTS produced the same changes in the ACh CRC as an enhancing alcohol, ethanol, and the application of PMTS to α2(L262C)β4 nAChRs resulted in modifications in the ACh CRC similar to the ones induced by an inhibitory alcohol, octanol. More support for action on two sites is found in the following results: PMTS application inhibited the ACh responses in α2(L262C)β4, indicating covalent binding to an inhibitory alcohol site. Before PMTS application, ethanol enhanced the function of α2(L262C)β4; this action was the net effect of potentiation and inhibition (Fig. 9A), since covalent binding of propanethiol to the cysteine in this site resulted in an enhanced ethanol potentiation, consistent with the loss of inhibitory effect due to the occupation of the inhibitory site and ethanol acting mainly on the enhancing site (Fig. 9B). The opposite situation applies to α2(L263C)β4; in this case, the 5-min application of PMTS resulted in propanethiol covalently bound to enough excitatory alcohol sites to reveal the inhibitory effect of ethanol on this mutant (Fig. 9C). The intermediate-chain alcohols where the transition from potentiation to inhibition takes place behaved in a manner consistent with our hypothesis, binding to both sites with cancellation of effects (Fig. 9D); covalent binding of propanethiol to one alcohol binding site reveals their effects on the other (Fig. 9, E and F). The long-chain octanol also acts in accordance with this hypothesis (Fig. 9, G–I). Interestingly, urethane presented a pattern similar to the alcohols.

In contrast, the volatile anesthetic isoflurane induced inhibition in wild-type, α2(L262C)β4, and α2(L263C)β4 nAChRs, but its effect was not altered except for a small increase in the inhibition after PMTS treatment on α2(L263C)β4, which suggested a small enhancing effect by isoflurane, usually masked by the more pronounced inhibitory effect. Our results point to a distinct location of the isoflurane inhibitory binding site, perhaps deeper into the channel, as suggested by others (Forman et al., 1995; Wenningmann et al., 2001). Thus, even though they share many characteristics, volatile anesthetics and alcohols do not appear to act on the same inhibitory binding site in the neuronal nAChRs. The results with isoflurane also indicate that the changes we observed in inhibitory effects of long-chain alcohols are not nonspecific, because they do not occur for all inhibitors of nAChR function.

Previous mutagenesis studies have suggested an alcohol inhibitory binding site located between positions 8′ and 13′ (equivalent to α2L254 and α2V259, respectively), and long-chain alcohols are thought to bind deeply in the pore to produce channel block (Forman et al., 1995; Forman and Zhou, 2000; Zhou et al., 2000). To date there is no direct evidence to support this hypothesis, and some data appear inconsistent, such as the lack of competition between the inhibitory actions of butanol and QX-222, a derivative of the local anesthetic lidocaine that is proposed to bind at the level of α1S248 (6′, homologous to α2S252) (Dilger and Vidal, 1994). Furthermore, there is evidence for nonluminal inhibitory binding sites in the nAChR. Such a mechanism has been proposed for quinacrine and ethidium, as well as for fatty acids and steroids (reviewed in Arias, 1998). Therefore, there is no firm evidence for an alcohol inhibitory site located in the pore, and considering the present results the participation of α2L262 in a nonluminal binding site is strongly supported. Furthermore, the carbene group of a photoactivable anesthetic 3-azioctanol binds to αE262 (20′) in Torpedo nAChR (Pratt et al., 2000), which would allow the OH group to interact with a proton donor at a distance of approximately 5 Å away from αE262 (C1–C3 ∼ 2 Å, C1–O ∼ 1 Å, H-bond between O and proton donor ∼2 Å). Since α2L262 (16′) and α2L263 (17′) are one turn of the helix (∼5.5 Å) below α2E266 (20′), they would be within range for interacting with 3-azioctanol.

Where are α2L262 (16′) and α2L263 (17′) located? The structure of the Torpedo nAChR in this region has been determined: the channel widens at the extracellular side but the four TM segments maintain their α-helical structure (Miyazawa et al., 2003). Our results (Fig. 1, B and C) and cysteine accessibility in α and βTM2 in muscle nAChR (Akabas et al., 1994; Zhang and Karlin, 1998) and TM2 peptide NMR spectroscopy (Opella et al., 1999) are also consistent with a α-helical structure in this region of the nAChR α subunit.

We found that PMTS treatment did not alter the function of α2(L261C)β4 (15′, the position homologous to GlyR α1S267 and GABA_AR α2S270, the TM2 residues forming part of the alcohol binding site in these receptors): there was no alteration either in the receptor activation or in the alcohol modulation, supporting the idea that this leucine is not critical for alcohol binding in nAChRs. The residues that participate in alcohol binding sites on the nAChR, α2(L262C) (16′) and α2(L263C) (17′), are close to but are not in the homologous position of the residue-forming part of the alcohol binding site in GlyRs and GABA_ARs. This demonstrates clear differences between the nAChRs and the GABA_AR and GlyRs with respect to the structure of this region of the receptor and to mechanisms of alcohol action.

Our model for the TM2 structure at this level locates α2(L262C) (16′, inhibitory site) at an intersubunit location, and α2(L263C) (17′, enhancing site) at the edge of the pore, with the residue tangential to the channel (Fig. 8). The residue α2(L261C) (homologous to GlyR α1S267 and the GABA_AR α2S270) is similarly located facing toward the region between the four TM segments. The position of these three residues in our model coincides with the structure recently described for the Torpedo nAChR (Miyazawa et al., 2003). Reaction with methanethiosulfonates requires the presence of water (Roberts et al., 1986); thus, α2L262 may be located in a water-filled cavity between two or more TM segments, as appears to be the case for alcohol and anesthetic binding sites in GABA_A and GlyRs (Jenkins et al., 2002) and the alcohol binding site in the Drosophila odorant-binding protein LUSH (Kruse et al., 2003).

In summary, the relatively simple picture of alcohols and volatile anesthetics acting at a single protein cavity originally obtained from GABA_AR and GlyR appears to gain complexity in the closely related nAChRs, where we provided evidence for two distinct alcohol sites that mediate the inhibitory and enhancing effects of alcohol. However, the importance of the extracellular regions of TM2 in forming part of these binding sites appears as a common motif in nAChRs, GABA_ARs, and GlyRs.