Abstract
The structural basis by which agonists, antagonists, and allosteric modulators exert their distinct actions on ligand-gated ion channels is poorly understood. We used the substituted cysteine accessibility method to probe the structure of the GABAA receptor in the presence of ligands that elicit different pharmacological effects. Residues in the α1 Met113-Leu132 region of the GABA binding site were individually mutated to cysteine and expressed with wild-type β2 and γ2 subunits in Xenopus laevis oocytes. Using electrophysiology, we determined the rates of reaction of N-biotinaminoethyl methaneth-iosulfonate (MTSEA-biotin) with the introduced cysteines in the resting (unliganded) state and compared them with rates determined in the presence of GABA (agonist), 4-[6-imino-3-(4-methoxyphenyl)pyridazin-1-yl]butanoic acid hydrobromide (SR-95531; antagonist), pentobarbital (allosteric modulator), and flurazepam (allosteric modulator). α1N115C, α1L117C, α1T129C, and α1R131C are predicted to line the GABA binding pocket because MTSEA-biotin modification of these residues decreased the amount of current elicited by GABA, and the rates/extents of modification were decreased both by GABA and SR-95531. Reaction rates of some substituted cysteines were different depending on the ligand, indicating that barbiturate- and GABA-induced channel gating, antagonist binding, and benzodiazepine modulation induce specific structural rearrangements. Chemical reactivity of α1E122C was decreased by either GABA or pentobarbital but was unaltered by SR-95531 binding, whereas α1L127C reactivity was decreased by agonist and antagonist binding but not affected by pentobarbital. Furthermore, α1E122C, α1L127C, and α1R131C changed accessibility in response to flurazepam, providing structural evidence that residues in and near the GABA binding site move in response to benzodiazepine modulation.
GABAA receptors are ligand-gated ion channels (LGICs) that mediate fast inhibitory synaptic neurotransmission in the brain and are allosterically modulated by a variety of compounds, including benzodiazepines (BZDs), barbiturates, anesthetics, and neuroactive steroids (Akabas, 2004). Identifying conformational movements that occur in the receptor in response to binding of these different classes of compounds is critical for understanding the molecular mechanisms underlying their therapeutic actions. According to allosteric theory, modulators bind to a site on the protein that is distinct from the agonist binding site and exert their effects by initiating an allosteric transition in the protein that indirectly modifies the conformation of the binding site (Changeux and Edelstein, 1998). In this study, we used the substituted cysteine accessibility method to monitor the accessibility of residues in the α1Met113-Leu132 region of the GABA binding site in the presence of GABAergic ligands that elicit distinct pharmacological effects. In particular, we were interested in determining whether BZD binding induces structural rearrangements in or near the GABA binding site. In addition, we were interested in investigating the mechanisms underlying how orthosteric agonist binding triggers channel opening and orthosteric antagonist binding does not.
The GABAA receptor belongs to the Cys-loop superfamily of LGICs, which includes the nicotinic acetylcholine (nACh), serotonin-type-3 (5-HT3), GABAC, and glycine receptors. Much is known about the structure of these receptors from the recent 4-Å resolution model of the nACh receptor (Unwin, 2005), the crystallographic structure of the related acetylcholine binding protein (AChBP) (Brejc et al., 2001; Celie et al., 2004) and decades of biochemical and electrophysiological studies (Akabas, 2004). For receptors in this superfamily, binding of neurotransmitter takes place at the extracellular N-terminal domain of the protein at the interface of two adjacent subunits and is coordinated by residues from at least six noncontiguous polypeptide regions, designated loops A to F, three in each subunit (Akabas, 2004).
The α1β2γ2 GABAA receptor subtype (β:α:β:α:γ viewed counterclockwise from the synaptic cleft) is the most abundant in vivo (Sieghart and Sperk, 2002) and the two GABA binding sites are located at the interfaces between the β and α subunits (β:α), with residues from each subunit contributing to the binding site (Sigel et al., 1992; Amin and Weiss, 1993; Smith and Olsen, 1994; Boileau et al., 1999; Westh-Hansen et al., 1999; Wagner and Czajkowski, 2001; Boileau et al., 2002; Holden and Czajkowski, 2002; Newell and Czajkowski, 2003; Newell et al., 2004). The BZD binding site is also located on the extracellular N-terminal domain of the receptor and is situated at the interface between the α and γ subunits (α:γ) (Sigel, 2002). Although the structures of the GABA and BZD binding sites are relatively well known, the protein movements underlying GABA-induced channel activation and BZD-mediated receptor modulation are less well understood.
In the present study, we focused on the α1Met113-Leu132 region of the GABA binding site (loop E, Fig. 1). Loop E of the GABA binding pocket is physically linked to the BZD binding site (loop A - α1His101) by a short stretch of 11 amino acid residues. We reasoned that if BZD binding triggered structural changes in the GABA binding site, then loop E was in a perfect position for potentially detecting these movements. In addition, residues in this region of the GABA binding site, besides directly contacting ligand, are likely to be important for mediating local movements in the site involved in coupling neurotransmitter binding to channel gating. Mutations of residues in the Loop E binding regions of the nACh and 5-HT3 receptors have been shown to alter channel gating (Ohno et al., 1996; Venkataraman et al., 2002; and Beene et al., 2004).
In this study, we demonstrate that flurazepam binding to the GABAA receptor induces conformational rearrangements in the GABA binding site, providing structural evidence for reciprocal allosteric interactions between the GABA and BZD binding sites. We identify four residues, α1Asn115, α1Leu117, α1Thr129, and α1Arg131, that participate in forming part of the GABA binding site, and we present evidence that binding of GABA and the competitive antagonist SR-95531 trigger different local movements within the site, which are likely correlated with their different abilities in promoting channel opening.
Materials and Methods
Site-Directed Mutagenesis. The α1 cysteine mutants were engineered using recombinant polymerase chain reaction as described previously (Boileau et al., 1999; Kucken et al., 2000). Individual cysteine substitutions were made in the rat α1 subunit at Met113, Pro114, Asn115, Lys116, Leu117, Leu118, Arg119, Ile120, Thr121, Glu122, Asp123, Gly124, Thr125, Leu126, Leu127, Tyr128, Thr129, Met130, Arg131, and Leu132 where the numbering reflects the position in the mature α1 subunit protein sequence. The cysteine mutants were subcloned into pGH19 for expression in Xenopus laevis oocytes as described previously (Boileau et al., 1998). The presence of the mutations was verified by restriction endonuclease digestion and double-strand cDNA sequencing.
Expression in Oocytes.X. laevis oocytes were prepared as described previously (Boileau et al., 1998). cRNA transcripts were generated using the mMessage T7 kit (Ambion, Austin, TX). GABAA receptor rat α1 or α1 mutants were expressed with wild-type rat β2 and γ2 subunits by injection of cRNA into oocytes (0.1 ng of α and β subunits/oocyte and 1.4 ng of γ subunit/oocyte, except for α1R119Cβ2γ2, which was injected at 6:1:6 ng of α1/β2/γ2). α1R119C was also expressed with the wild-type rat β2 subunit alone and injected at 10:1 ng of α1/β2 for some experiments, because higher receptor expression levels were achieved in the absence of the γ2 subunit. The oocytes were stored in ND96 medium (96 mM NaCl, 2 mM KCl, 1 mM MgCl2, 1.8 mM CaCl2, and 5 mM HEPES, pH 7.4) supplemented with 100 μg/ml gentamicin and 100 μg/ml bovine serum albumin for 2 to 14 days and used for electrophysiological recordings.
Voltage-Clamp Analysis. Oocytes under two-electrode voltageclamp (Vhold of -80 mV) were continuously perfused with ND96 at a rate of 5 ml/min. The bath volume was 200 μl. GABA, SR-95531 (Sigma, St. Louis, MO), pentobarbital (Sigma/RBI, Natick, MA) and flurazepam (Sigma/RBI) were dissolved in ND96. Stock solutions of MTSEA-biotin (Biotium, Hayward, CA) were prepared in dimethyl sulfoxide (final concentration, ≤2%). Standard two-electrode voltage-clamp recording was carried out using a GeneClamp 500 amplifier (Molecular Devices, Sunnyvale, CA) interfaced to a computer with a Digidata 1200 (Molecular Devices). Electrodes were filled with 3 M KCl and had resistances of 0.5 to 2.0 MΩ in ND96. Data acquisition was performed using pClamp 6 (Axon Instruments).
EC50 Analysis. Concentration-response experiments were performed as described previously (Wagner and Czajkowski, 2001). In brief, these experiments used a low concentration of GABA (EC2-EC7) immediately before the test concentration of agonist to correct for any slow drift in GABA responses that might occur during the experiment. Currents elicited by each test concentration were normalized to the corresponding low-concentration current. Concentration-response data were fit to the following equation: I = Imax/(1 + (EC50)/[A])nH), where I is the peak response to a given concentration of GABA, Imax is the maximum amplitude of current, EC50 is the concentration of GABA that produces a half-maximal response, [A] is the concentration of GABA, and nH is the Hill coefficient.
IC50 Analysis. IC50 values were measured as described previously (Wagner and Czajkowski, 2001). SR-95531 IC50 values were measured by applying a fixed concentration of GABA (EC20-EC60) immediately followed by coapplication of the same concentration of GABA and a test concentration of SR-95531. Inhibition was calculated as IGABA+ SR-95531/IGABA, where IGABA+ SR-95531 is the current elicited in the presence of GABA and SR-95531, and IGABA is the current elicited by GABA alone. Data were fit to the following equation: inhibition = 1 - 1/(1 + (IC50/[Ant])nH), where IC50 is the concentration of antagonist that blocks half of IGABA, [Ant] is the concentration of antagonist, and nH is the Hill coefficient. Apparent KI values were calculated using the Cheng-Prusoff/Chou equation (Cheng and Prusoff, 1973; Chou, 1974): KI = IC50/(1 + ([A]/EC50)), where [A] is the concentration of GABA used and EC50 is the concentration of GABA that elicits a half-maximal response.
Pulse Protocol for Measuring the Effect of MTSEA-Biotin on Introduced Cysteines. GABA-elicited currents (EC40-60) were stabilized before application of MTSEA-biotin by applying GABA at 10-min intervals until the peak GABA-activated currents (IGABA) varied by <10%. The oocyte was allowed to recover for 3 min and treated with MTSEA-biotin (2 mM) and washed again for 5 min. GABA (EC40-60) was then reapplied. The effect of MTSEA-biotin was calculated as (IGABA-post/IGABA-pre) - 1, where IGABA-post is the current elicited by GABA after MTSEA-biotin application, and IGABA-pre is the current elicited by GABA before MTSEA-biotin application.
For α1M113Cβ2γ2 and α1N115Cβ2γ2, the pulse protocol was modified to determine whether GABA or SR-95531 had any effect on the ability of MTSEA-biotin to inhibit IGABA. In these experiments, GABA (EC40-60) was applied at 10-min intervals until the response stabilized. A high concentration of GABA or SR-95531 (EC90-100) was applied (2 min), and the cell was washed again (5 min). GABA (EC40-60) was then reapplied. The cell was stabilized until GABA-activated currents varied by <10%. High concentration GABA or SR-95531 was then applied in the presence of MTSEA-biotin (2 mM, 2 min). The cell was washed for 5 min and GABA (EC40-60) was applied again. Finally, MTSEA-biotin (2 mM, 2 min) was applied alone, the cell was washed for 5 min, and the GABA (EC40-60) was applied to determine the full extent of the MTSEA-biotin reaction.
MTSEA-biotin was used because it is impermeable to the membrane, it is not charged, and its size (similar to SR-95531) is suitable for fitting within the binding site but is bulky enough to make it likely that covalently attaching it to an introduced cysteine will result in a functional effect. In addition, MTSEA-biotin was used to examine the accessibility of introduced cysteines in other regions of the GABA binding site (Boileau et al., 1999, 2002; Wagner and Czajkowski, 2001; Holden and Czajkowski, 2002; Newell and Cza-jkowski, 2003; Newell et al., 2004).
Rate of MTSEA-Biotin Modification of Introduced Cysteines. The rate of MTSEA-biotin covalent modification of accessible introduced cysteines was determined by measuring the outcome of sequential applications of low-concentration MTSEA-biotin on IGABA. The protocol was as follows: EC20-60 GABA was applied for 5 s, the cell was washed for 30 s, MTSEA-biotin was applied for 5 to 20 s, the cell was washed for 2 to 2.5 min, and the procedure was repeated until IGABA no longer changed, indicating that the reaction was complete. Before the rate of MTSEA-biotin modification was measured, GABA was applied every 3 min until IGABA varied ≤3%, demonstrating that the observed changes in IGABA after application of MTSEA-biotin were due to effects of the MTSEA-biotin.
For all rate experiments, the change in current amplitude was plotted versus cumulative time of MTSEA-biotin exposure. We assume that the concentration of MTSEA-biotin does not change significantly during the reaction; thus, we can determine a pseudofirst-order rate constant from the rate of change in IGABA. Peak current at each time point was normalized to the peak current at t = 0, and a pseudo-first-order rate constant (k1) was determined by fitting the data with a single exponential decay or growth function: y = span × e-kt + plateau. Because the data are normalized to IGABA at time 0, span = 1 - plateau. The second-order rate constant (k2) for MTSEA-biotin reaction was determined by dividing the calculated pseudo-first-order rate constant by the concentration of MTSEA-biotin used (Pascual and Karlin, 1998). To verify the accuracy of the calculated rate constants, rates were determined using at least two different concentrations of MTSEA-biotin for several mutants (data not shown).
The effect of ligand on the rate of MTSEA-biotin modification was tested by coapplying GABA (EC90-100), SR-95531 (IC50-95), pentobarbital (50 μM for low concentration, or 500-1000 μM for high concentration), and flurazepam (1-10 μM) with MTSEA-biotin. We use low concentrations of MTEA-biotin for short times, which we established does not maximally alter IGABA at the start of the experiment and we used high concentrations of ligand. Thus, we can determine how the presence of ligand alters the reaction rate, because the covalent reaction has not gone to completion. For these studies, IGABA was stabilized before the rate of MTS reaction was measured as follows: apply GABA (EC20-60) for 5 s, wash for 30 s, apply GABA, SR-95531, pentobarbital, or flurazepam at high concentration for 5 to 20 s, wash 2 to 5 min, and repeat the procedure. Wash times were adjusted to ensure that currents obtained from test pulses of GABA (EC20-60) after exposure to high concentrations of GABA, pentobarbital or flurazepam were within 3% of the previous GABA (EC20-60) current peak. This ensured complete wash-out of drugs and recovery from desensitization.
Statistical Analysis. All data are from at least three different oocytes from at least two different frogs. Data analysis was carried out using nonlinear regression analysis included in the Prism software package (GraphPad Software, San Diego, CA). Statistical analysis was conducted using a one-way analysis of variance, followed by a post hoc Dunnett's test. Log (EC50) and (KI) values were used for statistical analyses.
Structural Homology Modeling. A model of the extracellular ligand binding domain (LBD) of the GABAA receptor was built based on the structure of the AChBP (Brejc. et al., 2001). The crystal structure of the AChBP was downloaded from RCSB Protein Data Bank (code 1I9B) and loaded into Swiss Protein Data Bank Viewer (http://ca.expasy.ord/spdbv). The rat α1 mature protein sequence from Thr12-Ile227, the β2 protein sequence from Ser10-Leu218, and the γ2 protein sequence from Gly25-Arg231 were aligned with the AChBP primary amino acid sequence as depicted in Cromer et al. (2002) and threaded onto the AChBP tertiary structure using the “Interactive Magic Fit” function of Swiss Protein Data Bank Viewer. The threaded subunits were imported into SYBYL (Tripos, Inc., St. Louis, MO) where energy minimization was carried out. SYBYL minimizations terminate when the number of iterations is reached or if the gradient change reaches 0.05 kcal/Å × mol. The first 100 iterations were carried out using Simplex minimization (Press et al., 1988) followed by 1000 iterations using the Powell conjugate gradient method (Powell, 1977). A GABAA receptor LBD (β:α:β:α:γ viewed counterclockwise from the synaptic cleft) was assembled by overlaying the monomeric subunits on the AChBP scaffold, and the resulting structure was imported into SYBYL and energy-minimized. Neither water nor entropy factors were included during the minimizations. After the global energy minimization, Ramachandran plots, χ plots, side-chain positions, and cis- and trans-bonds were all examined. The program PROCHECK (Laskowski et al., 1993) was run to examine structural features against the established database of protein parameters, most importantly the Φ/ψ torsions and side-chain conformations. Problems in the structure that were revealed by these evaluations were fixed manually, and energy minimizations were run again as needed. Regions with insertions were modeled by fitting structures from a loop database. Because the sequence identify of the AChBP and the GABAA receptor LBD is only 18%, caution must be used in interpreting the absolute positions of individual side-chain residues in the model. Docking of GABA and SR-95531 was performed using the Flo+ version of QXP (Quick eXPlore; McMartin and Bohacek, 1997). The ligands were manually placed into the β/α GABA binding site interface, and the docking simulation was carried out using a Metropolis modified Monte-Carlo search procedure for 1000 cycles with nearby side chains flexible. The lowest energy poses are shown in Fig. 8, C and D, and were produced using PyMOL (DeLano Scientific, LLC, San Carlos, CA).
Results
Expression and Functional Characterization of Cysteine Mutant Receptors. Twenty residues in the Met113-Leu132 region of the GABAA receptor α1 subunit were individually mutated to cysteine (Fig. 1) and expressed with wild-type β2 and γ2 subunits in X. laevis oocytes. All mutant subunits formed GABA-activated channels. Maximal GABA current amplitudes ranged from 1 to 10 μA and did not differ significantly from maximal currents elicited from oocytes expressing wild-type α1β2γ2 receptors, except for α1R119Cβ2γ2 (see Materials and Methods).
GABA activated wild-type receptors with an EC50 value of 13.1 μM (Fig. 2, Table 1), and cysteine substitutions at α1Met113, α1Asn115, α1Lys116, α1Thr121, α1Thr125, and α1Leu126 did not alter GABA sensitivity. Fourteen cysteine mutations significantly altered GABA EC50 values compared with wild-type receptors (Table 1). The largest increases in GABA EC50 values were observed for receptors containing α1R119C, α1L117C, and α1T129C (261-, 161-, and 102-fold, respectively). Mutations of α1Arg131, α1Glu124, α1Tyr128, and α1Leu118, increased GABA EC50 values 67-, 42-, 28-, and 25-fold, respectively, and cysteine substitutions of seven other residues (α1Pro114, α1Ile120, α1Glu122, α1Asp123, α1Leu127, α1Met130, and α1Leu132) increased GABA EC50 values by less than 15-fold. Seven cysteine substitutions, α1P114C, α1L117C, α1R119C, α1T121C, α1G124C, α1Y128C, and α1R131C, significantly altered the sensitivity for the competitive antagonist SR-95531, and these changes were ≤11-fold (Fig. 2, Table 1). The cysteine mutations had no effects on pentobarbital (50 μM) or flurazepam (1-10 μM) potentiation of GABA responses and no effects on the ability of pentobarbital (500 μM-1 mM) to activate the receptor compared with wild-type receptors, suggesting that the global structure and function of the receptor was not altered by the cysteine mutations. Nonetheless, as in any mutagenesis study, one must keep in mind that the structure of the mutant receptor may not be the same as the structure of the wild-type receptor.
Modification of Cysteine Mutants by MTSEA-Biotin. Reaction of wild-type α1β2γ2 GABAA receptors with the sulfhydryl-specific reagent MTSEA-biotin (2 mM, 2 min) had no significant effect on GABA-mediated current amplitude (IGABA) (Fig. 3). MTSEA-biotin treatment significantly altered IGABA in nine of the twenty mutant receptors (Fig. 3). In receptors containing α1M113C, α1N115C, α1K116C, α1L117C, α1R119C, α1T129C, and α1R131C, MTSEA-biotin inhibited IGABA by 25, 27, 58, 81, 34, 60, and 76%, respectively. The observed decreases in IGABA after MTS modification can result from the binding sites being physically blocked by the MTS reagent or from an increase in GABA EC50 and/or a decrease in maximal GABA response (Imax) that arise from changes in GABA binding and/or channel gating. In receptors containing α1E122C and α1L127C, MTSEA-biotin potentiated IGABA by 50 and 67%, respectively. We hypothesized that these observed increases in IGABA were due to decreases in the GABA EC50 values of α1E122Cβ2γ2 and α1L127Cβ2γ2 receptors after MTSEA-bi-otin modification. To test this hypothesis, we measured GABA concentration responses in the same oocyte expressing α1E122Cβ2γ2 receptors before and after MTSEA-biotin application (Fig. 4). MTSEA-biotin modification of α1E122C-containing receptors resulted in an approximately 1.8-fold increase in GABA sensitivity, with no change in the maximal amount of current elicited by GABA. MTSEA-biotin had no effect on receptors containing α1P114C, α1L118C, α1I120C, α1T121C, α1D123C, α1G124C, α1T125C, α1L126C, α1Y128C, α1M130C, or α1L132C (Fig. 3). Thus, either these introduced cysteines were not accessible to MTSEA-biotin modification or their modification had no functional effect. The accessibility pattern of the residues is consistent with the predicted side-chain positions observed in homology models of the GABAA receptor, which depict this region of the receptor as forming β5-loop-β5′-loop-β6 (Fig. 1), and suggests that the positions of the introduced cysteines are similar to the wild-type residues.
MTSEA-Biotin Rates of Reaction. The rate at which MTSEA-biotin covalently modifies an accessible introduced cysteine depends on several factors, including the ionization of the sulfhydryl group, local steric restrictions, the electrostatic potential near the cysteine, and the permeability of the access pathway. Methanethiosulfonate reagents react 109 to 1010 times faster with the ionized thiolate (RS-) form of cysteine than they do with the protonated form, and ionization is more likely in an aqueous environment (Roberts et al., 1986; Stauffer and Karlin, 1994). Thus, a cysteine in an open, aqueous environment will have a faster rate of reaction than a cysteine in a restrictive, nonpolar environment.
We measured the second-order rate constants for MTSEA-biotin modification of accessible introduced cysteines in the absence of ligands to establish rates of modification in the resting, closed state of the receptor. The fastest rates of covalent modification occurred at α1T129C, α1L127C, α1L117C, and α1E122C (k2 ∼ 25.2, 1.968, 1.51, and 0.142 × 105 M-1s-1, respectively; Table 2), indicating that these residues are ionized and located in an open, aqueous environment. Slower rates were measured for α1R119C, α1R131C, and α1K116C (k2 ∼ 1170, 400, and 223 M -1s-1, respectively), suggesting that the thiol groups introduced at these positions are not as well ionized and/or the residues are located in a more restricted, buried environment. Although the rates measured are consistent with the predicted positions of these residues in a homology model of the extracellular ligand binding domain of the GABAA receptor, with α1Thr129, α1Leu127, α1Leu117, and α1Glu122 being located in relatively accessible regions and α1Arg131 and α1Lys116 more buried (Fig. 8), the rates also depend on the local electrostatic potentials near the sulfhydryl groups, which are probably different at each position and contribute to the range of reaction rates measured. Rates of modification of α1M113C and α1N115C were not measured because of the small effect MTSEA-biotin modification had on IGABA for these residues (<30%, Fig. 3).
Effect of Orthosteric Ligands on MTSEA-Biotin Second-Order Rate Constants. Orthosteric ligands bind to the GABAA receptor in the neurotransmitter binding site. To identify core binding site residues and to detect whether movements occur in and near the α1Met113-Leu132 region during occupation of the binding site with orthosteric ligands, we measured the rates of modification of K116C, L117C, R119C, E122C, L127C, T129C, and R131C in the presence of the agonist GABA and the antagonist SR-95531. Changes in the rates of modification during orthosteric ligand binding can be due to steric block from the ligand itself or to local structural movements triggered by ligand binding.
In the presence of GABA (∼EC90), the rates of modification of K116C, L117C, E122C, L127C, T129C, and R131C were significantly slowed by 3-, 200-, 3-, 40-, 74-, and 36-fold, respectively (Fig. 5, A and B; Table 2). In the presence of SR-95531 (∼EC90), the rates of modification of L117C, R119C, L127C, T129C, and R131C were slowed 11-, 6-, 7-, 6-, and 2-fold, respectively (Fig. 5, A and B; Table 2).
Because MTSEA-biotin elicited only a small change in IGABA for residues M113C and N115C (<30%; Fig. 3), we examined the effect of GABA and SR-95531 on the extent of inhibition of IGABA induced by MTSEA-biotin modification. For M113C, only GABA decreased MTSEA-biotin inhibition of IGABA, whereas both GABA and SR-95531 decreased MTSEA-biotin inhibition of IGABA for N115C (Fig. 5, C and D). In summary, both GABA and SR-95531 decreased modification of five introduced cysteines: N115C, L117C, L127C, T129C, and R131C. For three mutants, M113C, K116C, and E122C, only GABA slowed their modification, whereas modification of R119C was slowed only by SR-95531.
Effect of Pentobarbital on MTSEA-Biotin SecondOrder Rate Constants. To identify conformational changes that occur within or near the α1Met113-Leu132 region during channel gating, we measured the rate of modification of accessible introduced cysteine residues in the presence of pentobarbital. Pentobarbital binds to a site that is distinct from the neurotransmitter binding pocket (Amin and Weiss, 1993) and, at high concentrations, directly activates the GABAA receptor. Thus, pentobarbital is a useful tool for monitoring conformational movements that occur within or near the GABA binding site during channel gating without sterically blocking the GABA binding pocket.
The rates of modification of most of the introduced cysteines were not altered in the presence of a concentration of pentobarbital that activates the receptor (500-1000 μM) (K116C, L117C, R119C, L127C, and T129C; Fig. 6 and Table 2). However, pentobarbital (500 μM) significantly decreased the rate of modification of E122C by ∼2-fold and significantly increased modification of R131C by ∼4-fold (Fig. 6D, Table 2). It is noteworthy that concentrations of pentobarbital that do not open the channel but potentiate GABA current (∼50 μM) increased the rate of modification of R131C by ∼2-fold (Table 2).
Effect of Flurazepam on MTSEA-Biotin Second-Order Rate Constants. BZDs, used clinically for their anxiolytic and sedative actions, allosterically modulate GABAA receptor current responses. The structural mechanisms underlying how BZDs alter GABA-activation of the receptor are not well known. To identify whether local conformational movements occur within or near the GABA binding site during allosteric modulation of the GABAA receptor by a BZD, we measured the rates of modification of accessible introduced cysteines in the presence of the BZD-positive allosteric modulator flurazepam.
Flurazepam (∼EC90) increased the rates of MTSEA-biotin modification of L127C and R131C by approximately 2-fold, whereas it decreased the rate of modification of E122C by approximately 2-fold (Fig. 7, Table 2). Flurazepam had no effect on the rates of modification of K116C, L117C, and T129C (Table 2). The ability of flurazepam to alter the rates of modification of E122C, L127C, and R131C demonstrates that residues in the β:α binding site interface undergo structural rearrangements during occupation of the BZD binding site.
Discussion
Protein movements underlying LGIC activation and allosteric drug modulation are not well established. Here, we used the substituted cysteine accessibility method to examine the dynamics of the α1Met113-Leu132 region of the GABAA receptor agonist binding site (loop E) and determined that orthosteric ligand binding, channel gating, and allosteric drug modulator binding each induce distinct structural rearrangements in this region.
GABA Binding Site Residues. Residues that line the GABA binding site may directly contact ligand, be important for maintaining binding site structure, and/or mediate local conformational movements important for activation and/or desensitization. We identified four residues, α1Asn115, α1Leu117, α1Thr129, and α1Arg131, that line the GABA binding pocket. When mutated to cysteine, these residues were modified by MTSEA-biotin, which decreased IGABA. The rates or extents of modification were decreased in the presence of GABA and SR-95531. Pentobarbital-mediated activation did not slow MTSEA-biotin modification of L117C, T129C, and R131C, suggesting that GABA slowed their modification because of steric block rather than global channel gating phenomenon. When mapped onto our homology model, these residues form part of the back wall of the binding pocket and face into the site (Fig. 8A). Detailed kinetic analyses are required to quantitatively determine the mutations' effects on microscopic binding affinity and channel gating kinetics. Recently, Sedelnikova et al. (2005) determined that residues in positions aligned to α1Leu117, α1Thr129, and α1Arg131 line the GABAC receptor binding site.
In our structure, α1Leu117 lines the back wall of the binding pocket and faces into the site behind α1Thr129. Based on our dockings, α1Leu117 is unlikely to be involved in contacting GABA or SR-99531 but possibly contributes to maintaining the site's structure via a hydrophobic interaction with β2Tyr157. The Cβ atom of α1Leu117 is less than 6 Å from the Cβ atom of β2Tyr157, a residue that forms part of the binding site's aromatic box. α1Arg131, which also lines the back wall of the pocket, may maintain binding site structure via a salt-bridge interaction with β2Asp101 (Harrison and Lummis, 2006) or could be part of a trio of arginines (α1Arg66, α1Arg131, β2Arg207) proposed to stabilize GABA's negatively charged carboxylate (Holden and Czajkowski, 2002; Wagner et al., 2004).
Orthosteric Ligand-Mediated Movements. As with the GABAA receptor α1 subunit loop D region (Holden and Czajkowski, 2002), we have identified residues in the loop E region that change accessibility in response to only GABA or SR-95531. GABA decreased modification of M113C, K116C, and E122C, whereas SR-95531 decreased modification of R119C (Fig. 5D, Table 2). Decreases in modification can result from ligand physically blocking access to the cysteine or to movements induced by ligand binding. Because α1Glu122 is located far from the binding pocket (Fig. 8B), the change in E122C accessibility as a result of GABA probably reflects movements associated with GABA-induced channel gating. Consistent with this idea, activating concentrations of pentobarbital also decreased E122C accessibility, whereas SR-95531, which does not open the channel, had no affect. In addition, GABA application changed fluorescence of a label attached to E122C (and the aligned GABAC receptor residue) simultaneously with channel opening (Chang and Weiss, 2002; Muroi et al., 2006) providing evidence that α1Glu122 and/or nearby residues move in response to channel activation.
The decreases in modification of M113C and K116C by GABA and not by SR-95531 are more difficult to interpret because these residues are located near identified GABA binding site residues (Fig. 8B). Although we cannot rule out the possibility that decreases in their modification are due to physical block by GABA, this is unlikely. If the decreases were due to steric block, one would predict that both GABA and SR-95531 would decrease MTSEA-biotin modification of the residues, based on the larger size of SR-95531 (264 Å3) compared with GABA (99 Å3) and the overlapping positions of these ligands relative to α1Met113 and α1Lys116 when computationally docked into a homology model of the binding site (Fig. 8, C and D). Because SR-95531 had no effect on the rate of modification of M113C and K116C, the data suggest that these residues change accessibility due to local movements specifically triggered by GABA binding. Furthermore, in our homology model, α1Met113 and α1Lys116 are pointing away from the core of the GABA binding site.
Structural studies (Unwin et al., 2002; Celie et al., 2004) as well as molecular dynamic simulations (Law et al., 2005) in the nACh receptor and AChBP suggest that agonist binding promotes movement of loop C inward toward loop E to cap the binding site. We envision this capping motion to trap GABA within the site and block MTSEA-biotin from M113C and K116C. SR-95531, although considerably larger than GABA, is an antagonist and presumably does not promote this movement of loop C and thus does not change these residues accessibility. Furthermore, our data demonstrating that GABA decreased the rates of modification of the binding site residues L117C, T129C, and R131C (200-, 74-, and 36-fold, respectively) to a larger extent than SR-95531 (11-, 6-, and 2-fold, respectively) are also consistent with the idea that GABA binding stabilizes a conformation of the binding site that is more restrictive and less accessible than a SR-95531 bound site. Taken together, the data provide evidence that GABA binding promotes binding site closure.
After MTSEA-biotin modification of L127C, GABA-elicited currents are potentiated (Fig. 3), demonstrating that tethering ethyl-biotin onto L127C does not inhibit the ability of GABA to bind and trigger channel opening. Thus, L127C does not lie within the binding site and the decreases observed in the rate of modification of L127C by GABA and SR-95531 are a result of structural changes induced by their binding. Thus, although SR-95531 occupation of the GABA binding site does not trigger channel opening, it does cause local structural rearrangements. Movements triggered by competitive antagonist binding have been identified in glutamate (Armstrong and Gouaux, 2000), GABAA (Boileau et al., 2002; Muroi et al., 2006), and GABAC receptors (Chang and Weiss, 2002). It is noteworthy that SR-95531 decreased R119C modification, whereas GABA had no effect, suggesting that antagonist-induced movements are distinct from agonist-induced movements. Additional experiments are needed to confirm that the effects observed are due to the ligands' functional properties (i.e., agonist, antagonist) and not to differences in size or chemical property.
Pentobarbital-versus GABA-Mediated Gating Movements. Although GABA and pentobarbital binding probably stabilize comparable open-state structures because they produce similar single channel conductances (Jackson et al., 1982), some of the protein movements initially triggered by their binding must be different as they bind to distinct sites (Amin and Weiss, 1993). Thus, to examine global protein movements associated with channel gating as well as unique movements triggered by GABA and pentobarbital binding, we measured MTSEA-biotin modification rates in the presence of an activating concentration of pentobarbital (500-1000 μM) and compared these with rates measured in the presence of GABA.
For E122C, both GABA and pentobarbital decreased the rate of MTSEA-biotin modification suggesting that, at least near Glu122, movements triggered by their binding are similar and that E122C reports gating-induced global movements. For R131C, pentobarbital significantly increased whereas GABA significantly decreased MTSEA-biotin modification. Arg131 is located in the GABA binding site core; thus, the slowing of modification by GABA is probably due to steric block by GABA. Pentobarbital's increase in modification of R131C, however, demonstrates that this region of the GABA binding site, even when unoccupied by orthosteric ligand, undergoes movements in response to pentobarbital binding/channel activation. To date, we have identified 13 of 28 positions in the GABA binding site interface that change accessibility during pentobarbital binding/gating (β2T160C, β2D163C, β2G203C, β2S204C, β2R207C, β2S209C, α1D62C, α1S68C, α1E122C, α1R131C, α1V180C, α1A181C, and α1R186C; Wagner and Czajkowski 2001; Holden and Czajkowski 2002; Newell and Czajkowski, 2003; Newell et al., 2004). At several positions (K116C, L117C, L127C, and T129C) GABA decreased rates of modification whereas pentobarbital had no affect. As discussed above, the ability of GABA to alter L127C modification was probably due to GABA-induced local movements and not physical block. Thus, pentobarbital's lack of affect at this position suggests that L127C reports unique movements associated with GABA binding/activation of the receptor.
Benzodiazepine-Mediated Movements. BZD binding allosterically triggers movements in the α1Met113-α1Leu132 region of the GABA binding site. Flurazepam significantly altered the rate of E122C modification to a similar extent as GABA and pentobarbital, and increased modification of R131C similar to pentobarbital (Table 2). Thus, flurazepam binding triggered a subset of the same movements in the receptor that GABA and pentobarbital gating induced. Consistent with this hypothesis, recent experiments indicate that BZD positive modulators stabilize the receptor's open versus the closed state perturbing the receptor's gating equilibrium (Downing et al., 2005; Rusch and Forman, 2005; Campo-Soria et al., 2006).
Flurazepam binding also induces unique movements within the GABA binding site interface. Flurazepam increased the rate of L127C modification, whereas GABA decreased its rate and pentobarbital had no effect. Thus, each ligand can trigger different conformational changes in the α1Met113-α1Leu132 region. We speculate that these unique movements are the result of their binding to distinct sites and triggering different activation pathways that lead to their functional effects. Identifying these structural pathways is an important goal for future studies.
Acknowledgments
We thank Jill Smith for help in constructing the cysteine mutants, and Lisa Sharkey, Amy Kucken, Dr. Glen Newell, and James Seffinga-Clark for treating oocytes. In addition, we thank Dr. Ken Satyshur for help in constructing the models shown in this article and Dr. Glen Newell for critical reading of the manuscript.
Footnotes
-
This work was supported by a Pharmaceutical Research and Manufacturers of America Foundation Predoctoral Fellowship (to J.H.K.), National Institute of Neurological Disorders and Stroke grant NS34727 (to C.C.), and National Alliance for Research on Schizophrenia and Affective Disorder Award (to C.C.).
-
Portions of this work were previously presented at the Annual Meeting of the Society for Neuroscience 2001 in San Diego, CA and 2003 in New Orleans, LA.
-
ABBREVIATIONS: LGIC, ligand-gated ion channel; BZD, benzodiazepine; nACh, nicotinic acetylcholine; 5-HT3, serotonin-type-3; AChBP, acetylcholine binding protein; SR-95531, 4-[6-imino-3-(4-methoxyphenyl)pyridazin-1-yl]butanoic acid hydrobromide; MTSES, methanethiosulfonate ethylsulfonate; MTSEA, methanethiosulfonate ethylammonium; LBD, ligand binding domain.
-
↵1 Current affiliation: Laboratory of Membrane Biochemistry & Biophysics, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Bethesda, Maryland.
- Received July 7, 2006.
- Accepted November 15, 2006.
- The American Society for Pharmacology and Experimental Therapeutics