Tryptophan and Cysteine Mutations in M1 Helices of α1β3γ2L γ-Aminobutyric Acid Type A Receptors Indicate Distinct Intersubunit Sites for Four Intravenous Anesthetics and One Orphan Site

ETOMIDATE, propofol, alphaxalone, and barbiturates are intravenous general anesthetics that enhance the activity of γ-aminobutyric acid type A (GABA_A) receptors, the dominant inhibitory neurotransmitter receptors in the mammalian brain and members of the pentameric ligand-gated ion channel (pLGIC) superfamily.^1–4  GABA_A receptor subunits contain structural elements common to all pLGICs, including an N-terminal extracellular domain and a transmembrane domain with four α-helices (M1 to M4). Most GABA_A receptors consist of two α, two β, and one γ subunits arranged βαβαγ counterclockwise,⁵  creating four distinct types of subunit interfaces: α⁺–β^–, α⁺–γ^–, β⁺–γ^–, and two β⁺–α^– (fig. 1).

Anesthetic photolabels form adducts with GABA_A receptor residues that are homologs of ivermectin contacts in transmembrane intersubunit pockets of GluCl pLGICs, imaged with x-ray crystallography.⁶  Azi-etomidate labeled αM236 in α-M1 and βM286 in β-M3 in GABA_A receptors from bovine brain or cells expressing α1β3γ2L.^7,8  The barbiturate R-5-allyl-1-methyl-5-(m-trifluoromethyl-diazirinylphenyl) barbituric acid (mTFD-MPAB) photolabeled residues in β3-M1 (β3M227), α1-M3, and γ2-M3.⁸ m-Azi-propofol (azi-Pm) photolabeled both α1M236 and β3M227 in α1β3 receptors.⁹  6-Azi-pregnanalone (6-aziP), a neurosteroid derivative, photolabeled β3F301 in β3 homomeric receptors.¹⁰  Propofol, but not alphaxalone, displaced both azi-etomidate and mTFD-MPAB labeling in α1β3γ2L receptors.^8,11,12  These results suggest that α1β3γ2L receptors form overlapping etomidate and propofol sites in β3⁺–α1^– interfaces adjacent to α1-M1s and overlapping mTFD-MPAB and propofol sites in homologous α1⁺–β3^– and γ2⁺–β3^– pockets abutting β3-M1s (fig. 1). However, we do not know how propofol sites overlap with versus differ from etomidate and mTFD-MPAB sites because photolabels do not adduct all residues that contact parent drugs.^13,14  Additionally, no anesthetics have photolabeled γ2-M1, and because receptors photolabeled by azi-Pm and 6-aziP lacked γ2 subunits, propofol or neurosteroid interactions with γ2 remain untested.

To both complement and supplement photolabeling data, two approaches based on voltage-clamp electrophysiologic studies of GABA_A receptors with mutations at putative anesthetic contact residues have been widely applied. Tryptophan substitutions in putative anesthetic sites have been reported to both mimic the channel-gating effects of anesthetic binding and impair anesthetic modulation.^15–19  Cysteines substituted at putative contact residues have been covalently modified by probes such as p-chloromercuricbenzenesulfonate (pCMBS) and protected from modification by anesthetics.^14,20–24  However, neither of these approaches has been rigorously compared to photolabeling.

In the current study, we created tryptophan and cysteine mutations at two photolabeled residues in M1 helices, α1M236 and β3M227, and their homologs based on sequence alignment: α1L232, β3L231, γ2I242, and γ2L246 (table 1). At each locus, we assessed tryptophan mutant functions and performed substituted cysteine modification–protection (SCAMP) experiments to probe interactions with etomidate, propofol, alphaxalone, and mTFD-MPAB. Our analysis addressed three key questions: (1) do SCAMP results and/or tryptophan mutant sensitivity results for the study drugs agree with photolabeling at α1M236 and β3M227? (2) Do the α1-M1 contacts for etomidate and propofol or the β3-M1 contacts for mTFD-MPAB and propofol differ? (3) Do alphaxalone, propofol, or other anesthetics bind in the α1⁺–γ2^– interface?

Female Xenopus frogs were used as a source of oocytes in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (Bethesda, Maryland). Approval for animal use in this study was obtained from the Massachusetts General Hospital Institutional Animal Care and Use Committee (protocol number, 2005N000051; Boston, Massachusetts). Frogs were housed and maintained in a veterinarian-supervised facility and anesthetized using tricaine before oocyte collection. All efforts were made to minimize animal suffering.

R-mTFD-MPAB was a gift from Professor Karol Bruzik, Ph.D. (Department of Medicinal Chemistry and Pharmacognosy, University of Illinois, Chicago, Illinois). It was stored as a 100 mM solution in dimethyl sulfoxide (DMSO) and diluted in electrophysiology buffer to 4, 8, or 16 μM for experiments. Alphaxalone was purchased from Tocris Bioscience (Bristol, United Kingdom), stored as a 10 mM solution in DMSO, and diluted to 1.25, 2.5, or 5 μM in electrophysiology buffer for experiments. Propofol (2,6-diisopropylphenol) was purchased from Sigma-Aldrich (St. Louis, USA), stored as a 10 mM solution in DMSO, and diluted to 2.5, 5, or 10 μM in an electrophysiology buffer for experiments. R-Etomidate was purchased from Hospira, Inc (USA) as a 2 mg/ml (approximately 8.2 mM) solution in 35% propylene glycol:water and diluted to 1.6, 3.2, or 6.4 μM in electrophysiology buffer for experiments. We have previously shown that DMSO and propylene glycol at the dilutions used here produce no effects on GABA_A receptor function.²³ Para-cloromercuribenzenesulfonic acid sodium salt (pCMBS) was purchased from Toronto Chemical Research (Canada), and fresh stock solutions in electrophysiology buffer were prepared on the day of use and kept on ice until final dilution. γ-Aminobutyric acid (GABA), picrotoxin, salts, and buffers were purchased from Sigma-Aldrich.

Oocytes were prepared for use as previously described.¹⁵  Complementary DNAs encoding human α1, β3, and γ2L GABA_A receptor subunits in pCDNA3.1 expression vectors (Thermo Fisher Scientific, USA) were used as mutagenesis templates. Tryptophan and cysteine mutations were introduced into complementary DNAs by site-directed mutagenesis using QuikChange kits (Agilent Technologies, USA). Selected clones were sequenced through the entire coding region, and a single clone for each mutant was chosen for subsequent use. Messenger RNAs were synthesized on linearized DNA templates using mMessage Machine kits (Ambion Thermo Fisher, USA), purified, mixed in a ratio of 1α:1β:3γ, and then diluted in RNAase-free water to 1 ng/nl. Oocytes were injected with approximately 12 ng total messenger RNA mix and incubated in ND96 buffer (in mM: 96 NaCl, 2 KCl, 1 CaCl₂, 0.8 MgCl₂, 1 EGTA, and 10 HEPES, pH 7.5) supplemented with ciprofloxacin (2 mg/ml) and amikacin (100 μg/ml) at 17°C for 48 to 72 h before electrophysiologic studies.

Electrophysiologic experiments were performed in ND96 buffer at room temperature (21° to 23°C). Oocytes were positioned in a low volume (30 μl) custom-built flow cell and impaled with two borosilicate glass microelectrodes filled with 3 M KCl (resistance less than 1 MΩ). Oocytes were voltage clamped at −50 mV (model OC-725C; Warner Instruments, USA). Superfusion solutions based on ND96 were selected and delivered at a rate of 2 to 3 ml/min from glass reservoir syringes via polytetrafluoroethylene tubing and micromanifold (MP-8; Warner Instruments). Solutions were selected by activating electrical valves (VC-8; Warner Instruments). Specialized software and a digital input/output interface (pClamp 8.0 and Digidata 1322; both from Molecular Devices, USA) were used to coordinate the delivery of different superfusion solutions and recordings of voltage and current signals. Currents were filtered at 1 kHz, digitized at 100 Hz, and stored on a computer disk for offline analysis.

Currents in voltage-clamped oocytes expressing GABA_A receptors were exposed to solutions containing GABA (range, 0.1 μM to 10 mM) with or without anesthetics for 10 to 20 s, followed by 5-min ND96 wash. Normalization sweeps at 1 to 10 mM GABA alone (i.e., maximum GABA for the specific receptor) were recorded every other or every third experiment. GABA concentration–responses in the presence of etomidate or mTFD-MPAB were studied using the same method with solutions of variable GABA combined with 3.2 μM etomidate or 8 μM mTFD-MPAB. Baseline responses were measured using anesthetic without GABA. Oocytes were not preexposed to anesthetics when coapplied with GABA, and maximal normalization sweeps were in either high GABA alone or high GABA plus anesthetic. At least three oocytes from two different frogs were used for each concentration–response.

Spontaneous activation of GABA_A receptors (in the absence of GABA or anesthetics) was assessed by applying 2 mM picrotoxin to voltage-clamped oocytes. Outward currents that reversed during picrotoxin washout were assumed to represent spontaneously active channels. Spontaneous activity, if detected, was normalized to maximally activated inward GABA-elicited current in the same cell (n ≥ 3 cells).

Maximal GABA efficacy for each receptor was determined by comparing peak currents elicited with high GABA to currents elicited with high GABA supplemented with 2.5 μM alphaxalone, which positively modulated all receptors in this study. GABA efficacy was calculated by normalizing GABA responses to GABA plus alphaxalone responses, which we assumed represents 100% activation, in the same cell (n ≥ 3 cells).

Voltage-clamped oocytes expressing wild-type or mutant GABA_A receptors were repetitively exposed for 20 s to GABA EC₅ (eliciting approximately 5% of maximal GABA response) separated by 5-min wash until three stable peak responses (varying by less than 5%) were sequentially recorded. The oocyte was then exposed for 30 s to anesthetic, followed by 20-s exposure to a solution containing GABA EC₅ combined with anesthetic at 2 × EC₅₀ for loss-of-righting reflexes (LoRR) in tadpoles: 2.5 μM alphaxalone,²⁵  3.2 μM etomidate,²⁶  5 μM propofol,²⁷  or 8 μM mTFD-MPAB.²⁸  For each receptor type (wild-type and 12 mutants) and four anesthetics, triplicate measurements of current response to GABA EC₅ and GABA EC₅ plus anesthetic were obtained in at least four oocytes from two different frogs. Because receptors with the γ2L246W mutation displayed substantial spontaneous activity, 2 × EC₅₀ anesthetic enhancement of GABA EC₅ responses approached maximal GABA responses for all drugs. These receptors were also studied using 1 × EC₅₀ for tadpole LoRR of each anesthetic.

Cysteine modification and protection were performed in the presence of maximally activating GABA, as previously described.²³  Maximal GABA serves to (1) enhance anesthetic binding site occupancy in protection experiments; (2) assure in most cases that the mix of receptor states is similar in both control modification and protection studies; and (3) increase the rate of sulfhydryl modification. Each substituted cysteine mutant receptor was expressed in Xenopus oocytes, voltage clamped, and exposed to both GABA EC₅ (low) and a maximally activating GABA concentration (high). After 5-min wash, oocytes were then exposed for 30 s to ND96 containing high GABA plus pCMBS, a water-soluble cysteine-modifying reagent, ranging in concentration from 1 μM to 1 mM, followed by a 5-min wash in ND96. Responses to low and high GABA were then repeated. We thus confirmed that pCMBS produced irreversible changes in receptor function (in most cases an increase in the low:high GABA response ratio) and identified pCMBS concentrations suitable for comparing initial rates of modification. For control modification rate experiments, voltage-clamped oocytes were repeatedly tested for responses to both EC₅ and maximal GABA and then washed for 5 min in ND96, to assure that the response ratio was stable (less than 5% change) before pCMBS exposure. Oocytes were then exposed for 5 to 10 s to solutions of maximal GABA plus pCMBS at concentrations estimated to produce about 10% of the maximal modification effect, followed by a 5-min wash and retesting for low and high GABA responses. At least three cycles of pCMBS exposure/wash/low:high GABA response testing were performed on each oocyte. A final modification cycle was performed using 10 × pCMBS concentration for 20 s to complete modification of receptors, and the electrophysiologic response was assessed as the maximal modification effect.

For anesthetic protection experiments, a similar experimental protocol was used, except that oocytes were exposed to anesthetic for 30 s just before exposure to a solution of pCMBS plus GABA plus anesthetic. Postmodification wash and response tests were identical to control conditions (no anesthetic present). Anesthetic concentrations used in protection studies were 2 to 4 × LoRR EC₅₀. For each cysteine mutant, at least five oocytes were studied in control modification experiments and at least four oocytes were studied in protection experiments. Group sample sizes were based on both previous experience and a power analysis performed with G*Power 3.08 software (Franz Faul, Universitat Kiel, Kiel, Germany). Based on our experience with SCAMP, SD/mean ratios for modification rates range from 0.2 to 0.5, and in protection experiments, we aimed to detect drops of more than 50% from control modification rates. We estimated sample sizes for two groups with respective means of 2 and 1, both with SD = 0.4 (an effect size of 2.5), in a one-tail Student’s t test with α = 0.013 (adjusted for four-drug comparisons to control) and β = 0.85, indicating a need for five samples per group.

Results in the text and figures are mean ± SEM unless otherwise indicated.

Digitized GABA concentration–response data were corrected for baseline leak currents and digitally filtered (10-Hz low pass, Bessel function) using Clampfit 9.0 software (Molecular Devices). Peak currents were normalized to control (maximal currents), and combined GABA data from multiple cells (n ≥ 3) were fitted with logistic equations using Prism 5.02 (GraphPad Software Inc., USA):

where EC₅₀ is the half-maximal activating GABA concentration, and nH is the Hill slope. Mean GABA EC₅₀ and 95% CI are reported. GABA EC₅₀ shifts in the presence of anesthetics were calculated from the difference in log(EC₅₀) values (control − anesthetic) with propagation of log(SD) errors.²⁹  Mean GABA EC₅₀ shift ratios and 95% CIs were calculated. Statistical comparison of GABA EC₅₀s and anesthetic shifts between mutants and wild-type was based on the calculated CIs of the differences of means.

Comparison of both spontaneous activity and GABA efficacy between mutants and wild-type was based on one-way ANOVA with post hoc Dunnett tests (using Prism 5.02). EC₅ enhancement data for the four equipotent anesthetic concentrations in wild-type and all six tryptophan mutants were tabulated and analyzed with two-way ANOVA and Bonferroni posttests for wild-type versus mutation for each anesthetic (Prism 5.02).

Apparent pCMBS modification rates were calculated from data for individual oocytes expressing cysteine mutants. Either normalized maximal GABA response (for α1β3M227Cγ2L) or normalized low:high GABA response ratio (for the other five cysteine mutants) was plotted against cumulative pCMBS exposure (M × s), and linear least-squares analyses with y-axis intercepts fixed at 1.0 were applied. For each mutant, the resulting slopes from control and anesthetic protection studies were tabulated and compared using one-way ANOVA using Prism 5.02.

We treated photolabeling results as accepted standard evidence of anesthetic contact with a residue. For each anesthetic at each residue studied, we categorized substituted tryptophan sensitivity as positive if EC₅ enhancement was significantly reduced compared to wild-type. SCAMP results were categorized as positive if addition of anesthetic significantly reduced the initial rate of modification by pCMBS. We constructed 2 × 2 contingency tables and calculated the percentage agreement between each mutation-based method and photolabeling (i.e., concordance of positive and negative results), as well as Cohen κ, a measure of agreement between methods that corrects for chance. Contingency analysis with Fisher exact test was used to calculate a conservative P value, with the implicit null hypothesis being that mutation-based results are uncorrelated with photolabeling. Statistical significance was inferred for P < 0.05.

We expressed GABA_A receptors in Xenopus oocytes and characterized their function using two-electrode voltage clamp, assessing GABA concentration–response, maximal GABA efficacy, spontaneous activation, and modulation by intravenous general anesthetics (table 2). To quantitatively compare the effects of multiple general anesthetics in GABA_A receptors, we used equipotent drug concentrations based on LoRR tests in tadpoles. The EC₅₀s for LoRR are 1.6 μM etomidate, 2.5 μM propofol, 4 μM mTFD-MPAB, and 1.25 μM alphaxalone.^25–28  Wild-type α1β3γ2L GABA responses were similarly enhanced by equipotent 2 × EC₅₀ anesthetic concentrations, measured by either the shift in GABA EC₅₀s (fig. 2 shows results for etomidate and mTFD-MPAB) or enhancement of receptor activation at EC₅ GABA (fig. 3 shows results for all four anesthetics).

Photolabel analogs of both etomidate and propofol form adducts with α1M236.^7,9  We have previously described some of the functional characteristics of α1M236Wβ2γ2L GABA_A receptors,¹⁵  which mimic the effects of bound anesthetic, including increased sensitivity to GABA, spontaneous channel activity, and reduced modulation by etomidate, quantified as the ratio of GABA EC₅₀s in the absence versus presence of anesthetic. We hypothesized that tryptophan substitutions within anesthetic binding sites may, as a general rule, mimic the effects of anesthetics and reduce modulation by drugs that occupy those sites.

The functional characteristics of α1M236Wβ3γ2L receptors were very similar to those of α1M236Wβ2γ2L (table 2), including very weak modulation by etomidate (figs. 2 and 3). mTFD-MPAB potently activated α1M236Wβ3γ2L receptors while inducing a smaller GABA EC₅₀ shift than that in wild-type receptors (fig. 2; table 2). This may be due to this receptor’s spontaneous activity. However, EC₅ enhancement by mTFD-MPAB was similar in wild-type and α1M236Wβ3γ2L receptors (fig. 3). Propofol, and unexpectedly alphaxalone, also produced significantly less EC₅ enhancement in this mutant than in wild-type (fig. 3).

The α1 homolog of β3M227 is α1L232 (table 1). A tryptophan mutation at this site was described in a previous study of volatile anesthetic modulation.³⁰  Oocyte-expressed α1L232Wβ3γ2L receptors were characterized by a low GABA EC₅₀, low GABA efficacy, and no spontaneous activation (table 2). Based on GABA EC₅₀ shifts or EC₅ enhancement metrics, modulation of α1L232Wβ3γ2L receptors by etomidate and propofol was significantly reduced, while modulation of α1L232Wβ3γ2L receptors by mTFD-MPAB and alphaxalone was similar to that of wild-type (table 2; figs. 2 and 3).

Both mTFD-MPAB and azi-Pm form photoadducts with β3M227.^8,9  The effects of mutations at β3M227 and β3L231 have not been reported previously. Oocyte-expressed α1β3M227Wγ2L receptors displayed no detectable spontaneous activity, GABA EC₅₀ about twice that of wild-type, and high GABA efficacy (table 2). Anesthetic modulation of α1β3M227Wγ2L receptors was similar to that of wild-type for etomidate, propofol, and alphaxalone but significantly reduced for mTFD-MPAB (table 2; figs. 2 and 3).

The β3 homolog of α1M236 is β3L231 (table 1). The functional characteristics of α1β3L231Wγ2L receptors included spontaneous channel gating similar to that in α1M236Wβ3γ2L, a low GABA EC₅₀, and high GABA efficacy (table 2). Unexpectedly, etomidate modulation of α1β3L231Wγ2L receptors was significantly less than that of wild-type, while modulation by mTFD-MPAB, propofol, and alphaxalone was comparable to the effects in wild-type (table 2; figs. 2 and 3).

Basal leak and both low and high GABA responses in voltage-clamped oocytes expressing wild-type α1β3γ2L GABA_A receptors were unaffected by exposure to 1 mM pCMBS for up to 60 s (n = 4 oocytes; data not shown). We have previously described the functional effects of α1M236C mutations in GABA_A receptors (table 3), the effects of pCMBS modification at α1M236C, and evidence that both etomidate and propofol protect this substituted sidechain from modification, while alphaxalone does not.^21,31  For our current experiments, we first tested α1M236Cβ3γ2L receptors for sensitivity to anesthetics, confirming that GABA EC₅ responses are enhanced similarly by all four study drugs (table 3). For modification and protection experiments, we assessed low:high GABA response ratios, which increased by 8.9 ± 0.38-fold (n = 6) after maximal pCMBS modification. Compared to previous studies, we used lower concentrations of etomidate and propofol in protection experiments and also tested protection with 8 μM mTFD-MPAB. In agreement with our previous studies, we found that the addition of 2.5 μM alphaxalone dramatically increased the pCMBS modification rate relative to that in the presence of GABA alone (fig. 4). This is explained by observations that GABA alone activates only about 24% of α1M236Cβ3γ2L receptors, while pCMBS modification of this receptor is much faster in GABA-activated versus inactive receptors.²¹  Therefore, we used pCMBS plus GABA plus alphaxalone as the control condition for protection experiments with pCMBS plus GABA plus other anesthetics. These indicated that both etomidate and propofol protect α1M236C, while mTFD-MPAB does not (fig. 4).

We and others have also previously described the functional effects of α1L232C mutations (table 3) and evidence that etomidate protects this substituted sidechain from pCMBS modification.^21,32,33  All four anesthetics similarly enhanced GABA EC₅ responses in α1L232Cβ3γ2L receptors (table 3). By systematically testing various pCMBS concentrations, we found that very low concentrations (1 μM) of pCMBS produced irreversible enhancement of gating in α1L232Cβ3γ2L receptors (fig. 5, A and B) in contrast to the inhibitory effects of 200 to 500 μM pCMBS that were previously reported.^21,33  Maximal pCMBS modification increased the low:high GABA response ratio by 7.1 ± 0.47-fold (n = 7). In comparison to pCMBS plus GABA, addition of etomidate significantly slowed modification of α1L232Cβ3γ2L receptors (fig. 5, B and C). Alphaxalone, propofol, and mTFD-MPAB did not significantly affect the rate of pCMBS modification in this mutant.

Studies of cysteine substitutions at β3M227 and β3L231 have not been reported previously. The functional properties of α1β3M227Cγ2L receptors are summarized in table 3. Modulation of EC₅ responses in α1β3M227Cγ2L receptors by etomidate and other study drugs was similar to that in wild-type. The function of oocyte-expressed α1β3M227Cγ2L receptors was irreversibly modified by pCMBS (100 μM; fig. 6, A and B), but unlike other cysteine mutants in this study, modification reduced receptor activation without altering GABA sensitivity (the ratio of responses to EC₅vs. high GABA remained constant). Maximal pCMBS modification reduced peak currents by 71 ± 5.4% (n = 6). Protection experiments showed that both mTFD-MPAB and propofol significantly slowed modification of α1β3M227Cγ2L receptors, while alphaxalone and etomidate did not (fig. 6C).

The functional properties of α1β3L231Cγ2L receptors are summarized in table 3. This mutant was modulated similarly by all four anesthetics at equipotent concentrations. Modification by pCMBS resulted in enhanced GABA sensitivity based on the ratio of electrophysiologic currents elicited with low versus high GABA (fig. 6, D and E). Maximal pCMBS modification increased this ratio 5.7 ± 0.44-fold (n = 6). Anesthetic protection experiments showed strong protection by mTFD-MPAB but not by etomidate, propofol, or alphaxalone (fig. 6F).

Concordance between photolabeling, substituted tryptophan anesthetic sensitivity, and SCAMP was assessed for the four anesthetics at the two photolabeled residues that we studied: α1M236 and β3M227. In this set of eight anesthetic–residue pairs, there were four positive photolabeling pairs and four negative pairs. Substituted tryptophan sensitivity was categorized as positive if EC₅ enhancement by anesthetic in the mutant was significantly lower than that in wild-type (fig. 3). SCAMP was categorized as positive if the anesthetic significantly slowed the initial rate of pCMBS modification at the substituted cysteine (figs. 4 and 6C). The categorized results for all three methods are summarized in table 4. The agreement between photolabeling and substituted tryptophan sensitivity was 75% (Cohen κ = 0.5; P = 0.49 by Fisher exact test), with mismatches for propofol–β3M227 (a false-negative) and alphaxalone–α1M236 (a false-positive) interactions. The concordance between photolabeling and SCAMP results was 100% and statistically significant (Cohen κ = 1.0; P = 0.029 by Fisher exact test).

Both substituted tryptophan sensitivity and SCAMP were used to determine if any of the study anesthetics bind near the γ2-M1 helix. Effects of mutations at γ2I242 and γ2L246 have not been reported previously. The functional properties of α1β3γ2I242W and α1β3γ2L246W receptors are summarized in table 2. The γ2I242W mutation did not produce spontaneous receptor activation or enhance GABA sensitivity. Instead, we observed a high GABA EC₅₀ and reduced GABA efficacy in this mutant (fig. 7A). Oocyte-expressed α1β3γ2L246W receptors were characterized by approximately 11% spontaneous channel activation, low GABA EC₅₀, and very high GABA efficacy (table 2; fig. 7B). Modulation of α1β3γ2I242W by all four intravenous general anesthetics was similar to that observed in wild-type receptors (fig. 7C). In α1β3γ2L246W receptors, all the anesthetics produced a large amount of direct channel activation (e.g., fig. 7B), and GABA EC₅ enhancement with preexposure to the anesthetic concentrations used in other mutants produced near-maximal activation, reducing our ability to discern differences between drugs. We therefore studied EC₅ enhancement with half the usual anesthetic concentrations (i.e., 1 × EC₅₀ for tadpole LoRR; fig. 7C). Under these altered conditions, the overall amount of enhancement was lower, and the results indicated that none of the four anesthetics produced significantly more or less enhancement than the others.

The functional properties of oocyte-expressed α1β3γ2I242C and α1β3γ2L246C receptors (table 3) were similar to those of wild-type (table 2). Neither of the γ2-M1 cysteine mutations altered sensitivity to modulation by any of the four study anesthetics. Application of 10 μM pCMBS to α1β3γ2I242C receptors irreversibly enhanced GABA sensitivity (fig. 8, A and B). Maximal pCMBS modification increased the low:high GABA response ratio by 5.8 ± 0.37-fold (n = 8). None of the four anesthetic drugs significantly reduced the rate of pCMBS modification, even at twice the concentration used for most other cysteine mutants in this study (fig. 8C). Modification of α1β3γ2L246C receptors required high pCMBS concentrations (500 μM) and also enhanced GABA sensitivity (fig. 8, D and E). Maximal pCMBS modification increased the low:high GABA response ratio by 5.0 ± 0.38-fold (n = 5). None of the study anesthetics significantly reduced the rate of pCMBS modification in α1β3γ2L246C receptors.

Given the negative SCAMP results at both γ2I242 and γ2L246 with the four intravenous anesthetics, we also tested whether a flexible linear alcohol might bind in the α⁺–γ^– interface. Additional SCAMP experiments showed no protection by 120 μM n-octanol in either α1β3γ2I242C or α1β3γ2L246C receptors (n = 4 each; data not shown).

Anesthetic-photolabeling data on α1β3γ2L GABA_A receptors remain limited. We aimed to extend the map of drug contacts and test the functional roles of residues using two mutant-based approaches: substituted tryptophan sensitivity and SCAMP. For two residues, α1M236 and β3M227, and four intravenous anesthetics, published photolabeling results were 100% concordant with SCAMP but only 75% concordant with substituted tryptophan sensitivity. SCAMP experiments also indicated etomidate contact at α1L232 and mTFD-MPAB contact at β3L231 but no propofol or alphaxalone contact at these loci. At γ2I242 and γ2L246, both approaches consistently indicated no anesthetic interactions.

SCAMP is based on the formation of a covalent bond between a sulfhydryl engineered into the target receptor and a chemical probe. SCAMP is thus analogous to photolabeling in requiring proximity between the probe and the site. Both photolabeling and functional studies indicate that anesthetics bind preferentially to GABA-bound open and desensitized states.^14,34  Thus, our experiments were designed to promote formation of GABA-bound receptors with high anesthetic affinity and were powered to detect large protection effects indicating site occupation. We extended our previous SCAMP studies in α1M236C and α1L232C²¹  to include additional drugs and investigated four new cysteine mutants in β3-M1 and γ2-M1. All six substituted cysteines were accessible to pCMBS, forming covalent adducts that irreversibly altered receptor function. In all but one case, cysteine modification enhanced receptor sensitivity to GABA (increased the low:high GABA response ratio; table 3), echoing the effects of tryptophan substitution at most of these positions (table 2). The exception was β3M227C where pCMBS modification reduced GABA responses. Notably, the β3M227W mutation reduced GABA sensitivity (table 2).

SCAMP results for α1M236C and β3M227C were in perfect accord with both direct and indirect (drug competition) anesthetic-photolabeling studies (table 4). Contingency analysis with the Fisher exact test indicated that SCAMP is a sensitive and specific predictor of photolabeling results for these two residues and four anesthetics (four true-positives, four true-negatives, and no false-positives or -negatives; P = 0.029). We can extend this analysis to include βM286, another residue photolabeled by both azi-etomidate and azi-Pm, and where SCAMP evidence confirms propofol and etomidate contacts, but there is no interaction with alphaxalone.^7,9,12,20,22  Adding these two true-positives and one true-negative to our Fisher exact test analysis results in P = 0.0022. These results support SCAMP as a reliable technique to test putative anesthetic contact points beyond those identified with photolabels.

SCAMP also has limitations as an approach to mapping ligand–receptor contact residues.¹⁴  Cysteine modification is frequently dependent on the receptor state. Establishing conditions where the mix of functional receptor states is similar in both control modification and protection experiments can be challenging.²¹  Protection studies may not be feasible or interpretable when cysteine substitutions induce insensitivity to the ligand (possibly impaired binding) or when exposure to reactive probes causes little or no functional change.^21,31  To infer steric interference from SCAMP results, allosteric inhibition of cysteine modification must be ruled out. We addressed this issue by studying multiple drugs that produce similar functional effects.

Although α1L232 has not been photolabeled by anesthetic derivatives, β3L231 was photolabeled by a convulsant barbiturate.³⁵  Our current SCAMP results indicate that α1L232C is protected by etomidate, confirming a previous study using higher drug concentrations.²¹  Neither alphaxalone nor mTFD-MPAB protected α1L232C from modification. Despite evidence that propofol binds adjacent to α1-M1 and displaces azi-etomidate, α1L232C was not protected by propofol. Thus, etomidate and propofol sites in the β⁺–α^– interfaces overlap but not at α1L232. Of note, azi-Pm also photolabels α1I239,⁹  but pCMBS does not affect the function of α1I239Cβ2γ2L receptors,²¹  making SCAMP studies unfeasible.

Our data indicate that β3L231C is protected by mTFD-MPAB, extending the list of this drug’s likely contacts. However, while propofol displaces mTFD-MPAB photolabeling, it does not protect β3L231C, again suggesting overlapping but noncongruent sites adjacent to β3-M1. Consistent with photolabeling results, neither etomidate nor alphaxalone protected β3L231C. Considered together, the above SCAMP results indicate that etomidate binds exclusively in the two β⁺–α^– interfaces near α1L232 and α1M236, while mTFD-MPAB binds exclusively in homologous α⁺– β^– and γ⁺–β^– sites abutting βM227 and βL231. Propofol binds to all four of these sites, adjacent to either α1M236 or βM227.

This study extended our previous analysis of α1M236W and assessed five more tryptophan mutants in M1 helices. Both α1M236W and β2M286W mutations mimic the effects of anesthetic binding on GABA_A receptors, including decreased GABA EC₅₀, increased GABA efficacy, and spontaneous channel activation, all of which indicate enhanced channel gating.¹⁵  We hypothesized that these effects were produced when tryptophan sidechains partially filled the β⁺–α^– interfacial pockets where etomidate binds, and this approach has been applied to other transmembrane residues in putative anesthetic sites.^17–19  Both β3L231W and γ2L246W mutations produced effects similar to α1M236W (table 2). Thus, tryptophan substitution at the 10th residue after the conserved glycine-tyrosine-phenylalanine (GYF) sequence in any of the M1 helices of α1β3γ2L receptors (table 1) produced gating effects that mimic anesthetic binding. Receptors with α1L232W mutations also displayed increased apparent GABA sensitivity but not spontaneous channel activation or GABA efficacy higher than wild-type. On the other hand, β3M227W and γ2I242W produced decreases in GABA sensitivity, and none of the tryptophan mutations at the sixth residue after the conserved GYF caused spontaneous receptor gating. Overall, these results indicate that the middle segments of all M1 helices of α1β3γ2 receptors are similarly linked to ion channel gating and suggest that all five adjacent intersubunit pockets, including α⁺–γ^–, are potential allosteric modulator-agonist sites.

If tryptophan substitutions in M1 helices occlude adjacent intersubunit anesthetic sites, they should reduce modulation by anesthetics occupying these pockets. In agreement with photolabeling, α1M236W reduced receptor sensitivity to both etomidate and propofol, while β3M227W reduced sensitivity to mTFD-MPAB. However, anesthetic sensitivities of α1M236Wβ3γ2L and α1β3M227Wγ2L receptors were not in full accord with photolabeling and SCAMP results at these sites and others (table 4).

Unexpectedly, α1M236W reduced sensitivity to alphaxalone (fig. 3). Interestingly, Hénin et al.³⁶  predicted cholesterol binding sites near all six of the M1 residues we studied, and Hosie et al.³⁷  implicated αT237, immediately adjacent to αM236, as a contact for neurosteroids, based on altered tetrahydrodeoxycorticosterone sensitivity in mutants. However, SCAMP indicated no alphaxalone protection at αM236, and alphaxalone enhances azi-etomidate photolabeling at αM236, implying allosteric effects.¹²  Indeed, SCAMP provided no evidence of alphaxalone binding near any of the six residues we studied. Correlation with photolabeling is limited because photolabeling with 6-aziP used β3 homomeric receptors (thus, we lack data for α or γ subunits), and mass spectroscopic analysis missed a sequence including β3M227 and β3L231.¹⁰  The combined negative SCAMP and photolabeling results together suggest that mutant effects at αM236 and αT237 are indirectly (allosterically) coupled to neurosteroid sites.

Azi-Pm photolabeled β3M227,⁹  and SCAMP indicates propofol protection at β3M227C (fig. 6), but β3M227W did not reduce sensitivity to propofol (figs. 2 and 3). To explain this result, we suggest that propofol exerts most of its gating effects through the two sites that it shares with etomidate and relatively weak effects through sites adjacent to βM227 that it shares with mTFD-MPAB. Thus, the β3M227W mutation could occlude propofol binding to α⁺–β^– and γ⁺–β^– sites while preserving propofol modulation mediated by β⁺–α^– sites. This hypothesis is supported by evidence that βN265 mutations reduce propofol binding and reduce most of its modulatory effects.^31,38,39 

For anesthetic interactions at α1L232 and β3L231, SCAMP and substituted tryptophan sensitivity agree in five of eight drug–residue pairs (table 4). Discordant results were obtained at α1L232 for propofol and at β3L231 for both etomidate and mTFD-MPAB.

The extracellular α⁺–γ⁻ interface of GABA_A receptors forms the high-affinity benzodiazepine site,^40,41  but no anesthetic photolabels get incorporated in γ2-M1 in the transmembrane portion of this interface. In our current study, neither γ2I242W nor γ2L246W mutations selectively altered EC₅ enhancement by the four intravenous drugs we studied. Moreover, SCAMP studies at both γ2I242C and γ2L246C indicated no protection by these anesthetics, even at twice the concentrations that protected residues in other intersubunit pockets. Thus, we found no evidence of intravenous anesthetic binding in the α⁺–γ^– transmembrane interfacial pocket. We tested whether the α⁺–γ^– site was accessible to a small flexible drug, n-octanol, but observed no protection. Others have suggested that ivermectin, a macrocyclic lactone that modulates GABA_A receptors, may bind in this interface.⁴² 

Our SCAMP results are in remarkable agreement with anesthetic-photolabeling studies in GABA_A receptors while identifying new drug interactions that indicate overlapping but noncongruent sites for etomidate/propofol in β⁺–α^– interfaces and mTFD-MPAB/propofol in α⁺–β^– and γ⁺–β^– interfaces. The functional characteristics of substituted tryptophan mutants appear useful for probing allosteric linkages to the channel gate but unreliable for identifying drug-binding contacts. SCAMP provided no evidence that alphaxalone binds near any of the six residues we studied. Finally, our studies in γ2-M1 indicate coupling to channel gating similar to that in the four intersubunit anesthetic sites but no anesthetic contacts. Thus, the α⁺–γ^– transmembrane interface in α1β3γ2L GABA_A receptors is a potentially unique modulator site with no established ligands, in other words, an orphan site.

The authors thank Youssef Jounaidi, Ph.D. (Instructor, Department of Anesthesia Critical Care and Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts), for his help with molecular biology, and Timothy Houle, Ph.D. (Faculty Member, Department of Anesthesia Critical Care and Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts), for expert advice on statistical analysis. Keith W. Miller, D.Phil. (Professor, Department of Anesthesia Critical Care and Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts), Douglas Raines, M.D. (Professor, Department of Anesthesia Critical Care and Pain Medicine, Massachusetts General Hospital, Boston, Massachusetts), and Jonathan B. Cohen, Ph.D. (Professor of Neurobiology, Harvard Medical School, Boston, Massachusetts), provided helpful comments on the manuscript.

Supported in part by grants (GM089745 and GM058448) from the National Institutes of Health, Bethesda, Maryland.

The authors declare no competing interests.