Waggoner Center for Alcohol and Addiction Research, University of Texas at Austin, 2500 Speedway, MBB 1.124, Austin, Texas, TX 78712, U.S.A. E-mail: firstname.lastname@example.org
Each residue in the second transmembrane segment (TM2) of the human GABAA receptor α2 subunit was individually mutated to tryptophan. The wild-type or mutant α2 subunits were expressed with the wild-type human GABAA receptor β2 subunit in Xenopus oocytes, and the effects of these mutations were investigated using two-electrode voltage-clamp recording.
Four mutations (V257W, T262W, T265W and S270W) produced receptors which were active in the absence of agonist, and this spontaneous open channel activity was blocked by both picrotoxin and bicuculline, except in the α2(V257W)β2 mutant receptor, which was not sensitive to picrotoxin.
Six mutations (V257W, V260W, T262W, T267W, S270W and A273W) enhanced the agonist sensitivity of the receptor, by 10–100 times compared with the wild-type α2β2 receptor. Other mutations (T261W, V263W, L269W, I271W and S272W) had little or no effect on the apparent affinity of the receptor to GABA. Eight of the tryptophan mutations (R255, T256, F258, G259, L264, T265, M266 or T268) resulted in undetectable GABA-induced currents.
The S270W mutation eliminated potentiation of GABA by ethanol, whereas T261W markedly increased the action of ethanol. The T262W mutation produced direct activation (10% of maximal GABA response) by ethanol in the absence of GABA, while other mutations did not alter the action of ethanol significantly.
These results are consistent with a unique role for S270 in the action of ethanol within the TM2 region, and with models of GABAA receptor channel function, in which specific residues within TM2 are critical for the regulation of channel gating (S270, L264), while other residues (L269, I271 and S272) have little effect on these functions and may be non-critical structural residues.
GABAA receptors expressed in vitro without the γ subunit remain sensitive to general anaesthetics and n-alcohols (Levitan et al., 1988; Pritchett et al., 1989; Harrison et al., 1993; Mihic et al., 1994a,1994b). This is an advantage for molecular and pharmacological studies on this receptor since the number of receptor cDNAs can be limited to two. Previous studies demonstrated that mutation of specific amino acids in the transmembrane segments (TM) 2 and 3 of the α or β subunit of the GABAA receptor can eliminate the action of ethanol without abolishing the response to GABA (Mihic et al., 1997; Ueno et al., 1999), and that the corresponding mutations introduced into the γ subunit have less effect (Ueno et al., 1999). These findings suggest that, in the GABAA receptor, S270 in TM2 of the α subunit may be critical for the actions of ethanol, but did not rule out a more general involvement of the entire TM2 segment.
In the present study, we replaced each TM2 residue of the GABAA receptor α subunit in turn with the large amino acid tryptophan, and used this ‘tryptophan scanning mutagenesis’ to probe the importance for ethanol sensitivity of each such residue. If the actions of ethanol are transduced by the entire TM2 segment, we reasoned that the introduction of the large volumes associated with the tryptophan side chain would cause large decrements in the activity of alcohol in a large number of these mutated receptors. If the action of ethanol were not widely distributed, but dependent upon specific binding interactions, the effects of tryptophan mutagenesis might be expected to be more discrete. We expressed each of the mutant α2 subunits together with β2 subunit in Xenopus oocytes and also modelled the TM2 region of the GABA2 receptor α2 subunit as an α-helix, in order to suggest potential structural effects of mutagenesis and to predict which residues might lie physically close to S270 (Figure 1B).
Adult female Xenopus laevis were obtained from Xenopus I (Ann Arbor, MI, U.S.A.); GABA and bicuculline methiodide from Research Biochemicals International (Natick, MA, U.S.A.); collagenase type 1A, picrotoxin, penicillin-streptomycin solution and gentamicin solution were from Sigma Chemical Co. (St. Louis, MO, U.S.A.); ethanol was purchased from Aaper Alcohol and Chemical Co. (Shelbyville, KY, U.S.A.). All other chemicals used were of reagent grade.
All tryptophan mutations were made in the cDNA encoding the human GABAA receptor α2 subunit by use of the QuikChange Site-directed Mutagenesis kit (Stratagene, La Jolla, CA, U.S.A.) in the pCIS2 vector. The cDNA encoding the wild-type human GABAA receptor β2 subunit was subcloned into the pCDM8 vector (Invitrogen, Carlsbad, U.S.A.). All point mutations were verified by double-stranded DNA sequencing.
Xenopus oocytes were isolated and injected with cDNAs (1.5–3.0 ng per 30 nl), and two-electrode voltage-clamp recordings were performed as described previously (Mihic et al., 1994b; Harris et al., 1997). GABA was applied for 20–30 s and the maximum (peak) current was used as a measure of drug response. For GABA concentration-response curves, we applied 3 nM–1 mM GABA solutions to a single oocyte with an interval of 5 min at lower concentrations, or 15–25 min when desensitization was observed. We tested the ability of ethanol to enhance the effect of a concentration of GABA that produced 5% of a maximal effect (EC5). There is a dependence of alcohol enhancement on the GABA concentration, with greater potentiation by ethanol being seen at the lower GABA concentrations (Mihic et al., 1994b). The GABA EC5 was determined individually for each oocyte. We used 1 mM GABA to produce a maximal current. Oocytes were perfused with ethanol for 2 min before coapplication of GABA, to allow for complete equilibration of the oocytes with ethanol; this was immediately followed by a 20–30 s perfusion with GABA plus ethanol. In all cases, a 15–20 min washout period was allowed following application of the ethanol/GABA solutions. The solutions were freshly prepared immediately before use. All values are presented as mean±standard error of the mean (s.e.mean) from 4–94 oocytes obtained from at least two different frogs. Curve fitting and estimation of EC50 values for concentration-response curves were performed using GraphPad Prism software (San Diego, CA, U.S.A.). Statistical analyses were also carried out by two-tailed, paired t-test using this software.
The amino acid sequences of transmembrane segments 2 and 3 (TM2 and TM3) were modelled as α-helices (Cutting et al., 1991) using the Biopolymer module of Insight II (MSI, San Diego, CA, U.S.A.). TM2 was positioned such that the polar residues that line the pore of the ion channel face to the left in Figure 7. TM3 was aligned anti-parallel with TM2 such that its axis was 11Å from TM2. The backbone atoms of the two alpha helices were tethered with a force constant of 10 kcal/Å2 and side chain packing was optimized with Discover98 using the CFF91 potential energy functions. Replacing S270 with tryptophan and then re-optimizing the side chain packing while using a force constant of 10 kcal/Å2 to tether the backbone atoms modelled the effect of the S270W mutation. The molecular volumes (van der Waals) of serine, tryptophan, and ethanol were calculated with Spartan (Wavefunction Inc. San Diego, CA, U.S.A.) using the AM1 semi-empirical parameters. The structures were optimized in an extended geometry.
Nineteen different residues in the TM2 region of the GABAA receptor α2 subunit were mutated to tryptophan. Four of these mutations, R255W (n=27), L264W (n=28), M266W (n=25) and T268W (n=24) did not produce any significant GABA-induced current (up to 10 mM GABA) when they were co-expressed with the wild-type β2 subunit in Xenopus oocytes (Figure 1B). Three other mutations, T256W, F258W and G259W, resulted in receptors with weak responses to 1 mM GABA (Table 1). Because of the small or absent GABA response, these seven mutants were not investigated further in this study. The T265W mutant produced a weak response to 10 mM GABA (Table 1) and also made the receptor tonically open (see below).
Table 1. Summary of spontaneous activities, GABA EC50s, Hill coefficients and maximal responses produced by 1 mM GABA for wild-type and mutant GABAA receptors expressed in Xenopus oocytes
Introduction of tryptophan into three positions made the receptor spontaneously open; coexpression of α2(T262W), α2(T265W) and α2(T270W) with wild-type β2 subunit produced receptors in which picrotoxin induced an outward current (Figure 2B–D and Table 1). Interestingly, bicuculline, (supposedly a competitive GABAA receptor antagonist), also produced an outward current on those mutant receptors. In receptors composed of α2(T265W)β2 or α2(S270W)β2, an inward current was observed upon removal of picrotoxin. On the other hand, V257W also produced tonic activity, which was blocked only by bicuculline, but not by picrotoxin (Figure 2A). In the α2(V257W)β2 receptor, GABA-induced current, as well as spontaneous current, was insensitive to picrotoxin (data not shown).
Compared to the wild-type α2β2 receptors, the mutations of V257W, V260W, T262W, T267W, S270W and A273W made the resulting receptors 10 to 100 fold more sensitive to GABA (Figure 3 and Table 1). It should be noted that the α2(T262W)β2 receptor desensitized too rapidly to estimate the GABA EC50 and Hill coefficient from a complete concentration-response curve. Other mutations slightly altered (T261W, V263W or S272W) or did not affect (L269W or I271W) the apparent affinity of the receptor for GABA (Table 1).
Next we examined the effects of mutations on the actions of ethanol, using an EC5 concentration of GABA. Across all the potentiation experiments for wild-type and mutant receptors reported, the actual percentage of maximal GABA response (EC values) for the test concentrations used were: wild-type α2β2 (4.9±0.1%, total 100 experiments), α2(T257W)β2 (5.7±0.2%, seven experiments), α2(V260W)β2 (5.1±0.2%, 13 experiments), α2(T261W)β2 (5.2±0.3%, 12 experiments), α2(T262W)β2 (4.9±0.2%, 10 experiments), α2(V263W)β2 (5.0±0.1%, 12 experiments), α2(T267W)β2 (4.8±0.3%, 12 experiments), α2(L269W)β2 (4.9±0.3%, 10 experiments), α2(S270W)β2 (5.1±0.2%, 10 experiments), α2(I271W)β2 (5.5±0.3%, 10 experiments), α2(S272W)β2 (5.5±0.4%, eight experiments), α2(A273W)β2 (5.2±0.13%, eight experiments).
For the wild-type α2β2 receptor, ethanol (200 mM) produced 94±3% potentiation without any direct activation (i.e., no effect of preincubation with ethanol alone) (Figures 4 and 5). The S270W mutation which has been reported to eliminate the potentiating action of isoflurane (Koltchine et al., 1999), also abolished (6±5%) the ethanol action. A small degree of direct activation by ethanol (1.9±0.2% of maximal GABA response) was obtained in the α2(V275W)β2 receptor, in which there was a slight reduction (to 64±6%) in ethanol potentiation (Figure 5). The α2(V263W)β2 or α2(T267W)β2 receptors showed 48±5% or 68±5% potentiation of GABA-induced current, respectively. Interestingly, ethanol directly activated the mutant α2(T262W)β2 receptors to the extent of 10±1% of maximal GABA response, but the potentiation of GABA-induced current was reduced (to 40±7%) in comparison to the wild-type receptor.
One mutant, α2(T261W)β2, demonstrated a very large potentiation (251±21%) by 200 mM ethanol, without showing any direct effect of ethanol (Figures 5 and 6A). For this mutant we studied the potentiation by ethanol over a wide range of ethanol concentrations, and the T261W mutant showed significant potentiation at the concentration as low as 10 mM, compared to the wild-type that showed significant potentiation only at 50 mM and higher (Figure 6B). Other tryptophan-mutant receptors, such as α2(V260W)β2, α2(L269W)β2, α2(I271W)β2, α2(S272W)β2, or α2(A273W)β2, exhibited actions of ethanol that did not differ from the wild-type receptor.
Both the Ser270 in the GABAA receptor α subunit and the corresponding residue (Ser267) in the glycine receptor α subunit have been previously reported to be critical for the actions of alcohols and volatile anaesthetics (Mihic et al., 1997; Krasowski et al., 1998). At this position the volume of the amino acid residue is negatively correlated with the potentiation produced by alcohols/anaesthetics; that is, substitution with larger amino acids at that site leads to decrease enhancement (Ye et al., 1998; Koltchine et al., 1999). The importance of side chain volume at this site is further underscored by the observation that substitution of small residues in TM2/TM3 regions of the GABA ρ1 receptor increase the alcohol ‘cutoff’ toward longer chain alcohols, suggesting that alcohols may bind in a cavity formed in part by TM2 (Wick et al., 1998).
The S270W mutation, alone among the 19 mutations studied, completely eliminated the potentiation of GABA currents by ethanol. The V257W, T262W, V263W and T267W mutations reduced but did not abolish ethanol potentiation. We therefore conclude that S270 is the only amino acid residue that is absolutely critical for the action of ethanol in the TM2 region of the GABAA receptor α2 subunit. We also suggest that the slight reduction in ethanol potentiation of GABA response produced by the mutations of V263W or T267W may be because the positions of these residues are physically close to S270, as predicted from a helical wheel diagram of the TM2 region (Figure 1B). A similar reduction in the action of ethanol has been reported by the mutation of T264 in the glycine receptor α1 subunit, which corresponds to the position of T267 in the GABAA receptor α2 subunit (Mihic et al., 1997).
According to our helical wheel diagram, the position of V257 or T262 is not supposed to be close to S270, but tryptophan mutations at these positions resulted in a small decrease in ethanol action. This could be because spontaneous activities in the resultant receptors have a role in the reduced potentiation by ethanol. Recently Thompson et al. (1999) have reported that the mutation in TM2 of the GABAA receptor β subunit makes the resultant receptors tonically open, more sensitive to GABA and fail to potentiate the GABA response by positive allosteric modulators. They conclude that the positive allosteric modulation can be reduced by mutations that shift the receptor EC50 closer to the intrinsic binding affinity because no further leftward shift in an agonist dose-response relationship is possible. If our changes in ethanol actions were due only to changes in GABA sensitivity, then we would predict that there would be a relationship between GABA EC50 and the potentiation by ethanol among the mutants. However, we found no significant correlation between these two parameters for the mutants tested (r2=0.16). Thus, our results suggest that changes in ethanol action produced by tryptophan mutations are mainly due to effects of the mutations on alcohol action rather than GABA action, although it remains possible that changes in GABA sensitivity and spontaneous activity in the mutant receptors also contribute to the differences in alcohol sensitivity.
Many of the mutants tested made the receptor more sensitive to GABA. Moreover, V257W, T262W, T265W or S270W mutants made the receptors spontaneously active and tryptophan mutations of an Ala residue in the TM3 region of GABAA receptor α2 and β1, or glycine receptor α1 subunits also produce receptors with tonically open channels (Mihic et al., 1997; Ueno et al., 1999). Moreover, Lasalde et al. (1996) have reported that tryptophan substitutions into the TM4 region of the Torpedo nicotinic acetylcholine receptor stabilize the channel in the open state.
Taken together, these results suggest that the introduction of bulky residues within a transmembrane domain of ligand-gated ion channels might reduce the free energy difference between the more stable closed and less stable open states. This reduction in the free energy difference could also interpret the direct activation by ethanol, which was observed remarkably in the T262W-containing receptor (Figure 4). Direct effects of ethanol in the absence of GABA were found only with receptors that showed spontaneous activities (T265W was too difficult to express and ethanol actions were not determiend).
Picrotoxin and bicuculline appear to inhibit the spontaneous opening of all mutants, except for the V257W mutation, which made the receptor insensitive to picrotoxin. It has been reported that the corresponding position in the GABAA receptor α1 subunit (also V257) may be a binding site of picrotoxin (Xu et al., 1995), and our data would suggest that the tryptophan mutation interferes with the binding of picrotoxin at this position. A possible interpretation for the closing of the channel by bicuculline in several of the mutant receptors studied here is that bicuculline acts as an allosteric inhibitor, rather than a true competitive antagonist, so that it can stabilize the channel in the closed state, as suggested by Chang & Weiss (1999) using other spontaneously opening GABAA receptors.
The rebound currents following removal of picrotoxin were an unusual feature of some of the mutant receptors. These currents were especially noticeable in the α(T265W) and α(S270W) mutants (Figure 2C,D). A plausible explanation for this behaviour may be provided by a consideration of a simple three-state model for these receptors:
Under ‘baseline’ conditions, in the absence of agonist, the wild-type receptors are all in the closed (C) state, which is energetically highly stable relative to the open (O) and desensitized (D) states. Hence picrotoxin produces no current in wild-type receptors. Binding of agonist to the C state then promotes transition to the stabilized O and D states and allows current to flow, until agonist is removed, prompting a return to the C state. We hypothesize that the energetics of the mutant receptors are such that, even in the absence of agonist, they exist in an equilibrium between closed (C), open (O) and desensitized (D) states, so that there is significant ‘spontaneous’ inward current through receptors in the O state. The addition of picrotoxin results in what appears to be an outward current (Figure 2D), in reality reflecting a rapid block by the drug of the standing inward current. It is presumed that picrotoxin promotes a transition from the O and D states to the C state. The subsequent removal and very rapid unbinding of picrotoxin then allows the receptors to undergo relaxation from the C state to the O state, resulting in the large inward ‘rebound’ currents we observed (Figure 2C,D). The subsequent ‘fade’ of the rebound currents reflects the slower kinetics of relaxation from the O state to the D state (Figure 2D), until the original equilibrium between the O and D states is re-established.
Rebound currents are not observed with bicuculline. It is likely that the kinetics of unbinding of bicuculline are slower (as seen in, for example, the experiment shown in Figure 2A, from the V257W mutant) relative to the C-O and O-D transitions, so that these cannot be observed independently following removal of the drug. Unbinding of bicuculline is therefore rate-limiting and the return to equilibrium state appears to follow a simpler monophasic trajectory.
It is particularly interesting that the T261W mutation produces a striking increase in the potentiation by ethanol, resulting in significant effects of 10 mM ethanol, the lowest concentration tested. There is evidence that T261 faces into the channel (Xu & Akabas, 1996), and is positioned horizontally opposite to S270 according to the helical wheel diagram (Figure 1B). One possible interpretation of our results would be that the T261W mutation changes the conformation of the protein in the region bounded by S270 and thereby increases binding of ethanol, perhaps stabilizing the channel open state in the presence of the drug.
In an attempt to visualize the effect of tryptophan mutation at position S270, TM2 and TM3 domains of the GABAA receptor wild-type α2 and α2(S270W) subunits were modelled as anti-parallel α-helices (Figure 7). The positions of TM2 and TM3 were modelled based on exposure to hydrophobic photoactivatable probes in the lipid region of the acetylcholine receptor: TM1 was only slightly labelled, TM2 was not labelled, TM3 was partially labelled, and TM4 was most labelled (Blanton & Cohen, 1994). In contrast, exposure of cysteine mutants of TM1 and TM2 of the acetylcholine receptor to a water-soluble probe revealed that TM2 was labelled in a manner consistent with an α-helix whereas TM1 was labelled in an irregular manner (Akabas & Karlin, 1995). In addition, activation by GABA increases the accessibility of residues on TM3 of GABAA receptors to water-soluble probes (Williams & Akabas, 1999).
The model of the TM2-TM3 region in wild-type α2 subunit shows that the region between S270 (TM2) and TM3 contains a small cavity that may not be filled by side chains of adjoining helices. In contrast, the model of the S270W mutation shown on the right hand side of the figure demonstrates that the side chain of tryptophan completely occupies this cavity, which could eliminate occupation of the putative capacity by ethanol. Our calculations of the molecular volumes (van der Waals) of serine and tryptophan showed that tryptophan (224 Å3) is 114 Å3 larger than serine (110 Å3). If ethanol (molecular volume of 70 Å3) cannot be accommodated within the remaining resultant cavity, the volume of the putative cavity in the wild-type receptor can be estimated as 180 Å3, similar to estimates (∼200 Å3) suggested by recent work with the volatile anaesthetic ether, isoflurane (Koltchine et al., 1999).
In summary, the introduction of individual tryptophan mutations into the TM2 region changes the gating properties of GABAA receptors significantly in many of the mutant receptors, while the S270W mutant was unique in eliminating the action of ethanol. Our simple calculations provide surprising agreement between experimental data and theoretical prediction that alcohols may potentiate GABA responses by binding in the region of S270. However, it is important to note the alternative explanation that ethanol binds elsewhere on the receptor and that the TM2 region alters gating of the receptor thereby changing actions of ethanol. Further investigation is required to determine if alcohols bind to these amino acids in the TM2 and TM3 regions.
We are grateful to Dr Paul J. Whiting for providing GABAA receptor subunit cDNAs. We also thank Virginia Bleck for technical assistance. This work was supported by NIH grants AA06399 and GM7818 to R.A. Harris, AA11525 to S.J. Mihic, GM56850 to N.L. Harrison and PO1 GM47818 to J.R. Trudell, S.J. Mihic, R.A. Harris and N.L. Harrison.