Protein movements underlying ligand-gated ion channel activation are poorly understood. The binding of agonist initiates a series of conformational movements that ultimately lead to the opening of the ion channel pore. Although little is known about local movements within the GABA-binding site, a recent structural model of the GABAA receptor (GABAAR) ligand-binding domain predicts that β2Glu155 is a key residue for direct interactions with the neurotransmitter (Cromer et al., 2002). To elucidate the role of the β2Ile154-Asp163 region in GABAAR activation, each residue was individually mutated to cysteine and coexpressed with wild-type α1 subunits in Xenopus laevis oocytes. Seven mutations increased the GABA EC50 value (8- to 3400-fold), whereas three mutations (E155C, S156C, and G158C) also significantly increased the 2-(3-carboxypropyl)-3-amino-6-(4-methoxyphenyl) pyridazinium (SR-95531) KI value. GABA, SR-95531, and pentobarbital slowed N-biotinylaminoethyl methanethiosulfonate modification of T160C and D163C, indicating that β2Thr160 and β2Asp163 are located in or near the GABA-binding site and that this region undergoes structural rearrangements during channel gating. Cysteine substitution of β2Glu155 resulted in spontaneously open GABAARs and differentially decreased the GABA, piperidine-4-sulfonic acid (partial agonist), and SR-95531 sensitivities, indicating that the mutation perturbs ligand binding as well as channel gating. Tethering thiol-reactive groups onto β2E155C closed the spontaneously open channels, suggesting that β2Glu155 is a control element involved in coupling ligand binding to channel gating. Structural modeling suggests that the β2 Ile154-Asp163 region is a protein hinge that forms a network of interconnections that couples binding site movements to the cascade of events leading to channel opening.
- piperidine-4-sulfonic acid
- substituted cysteine accessibility method
- Xenopus laevis oocytes
- two-electrode voltage clamp
Agonists and antagonists induce different molecular rearrangements in neurotransmitter-binding sites of ligand-gated ion channels (LGICs) (Armstrong and Gouaux, 2000; Boileau et al., 2002; Chang and Weiss, 2002). Agonists, but not antagonists, promote opening of the ion channel pore. It is likely that movements of amino acids near or within the neurotransmitter recognition site trigger the cascade of events leading to channel opening (Boileau et al., 2002; Torres and Weiss, 2002; Unwin et al., 2002; Miyazawa et al., 2003; Chakrapani et al., 2004). Here, we examined the I1e154-Asp163 region of the GABAA receptor (GABAAR) β2 subunit to identify residues that mediate local movements within the binding site that initiate channel gating and residues involved in GABA binding.
GABAARs are heteropentameric LGICs that mediate fast synaptic inhibitory neurotransmission in the brain. The α1β2γ2 GABAAR subtype is the most abundant in vivo, and heterologous expression studies suggest a β-α-β-α-γ stoichiometry and subunit arrangement (Chang et al., 1996; Tretter et al., 1997; Farrar et al., 1999; Baumann et al., 2001, 2002). Expression of α and β subunits also gives rise to functional GABAAR with putative stoichiometries of either 3α:2β (Im et al., 1995) or 3β:2α (Baumann et al., 2001; Horenstein et al., 2001) that lack sensitivity to benzodiazepines (Schofield et al., 1987; Pritchett et al., 1989), are responsive to barbiturates, and have a high apparent affinity for agonists (Boileau et al., 1999, 2002; Wagner and Czajkowski, 2001).
A recent homology model of the GABAAR agonist-binding site predicts that β2Glu155 interacts with the positively charged moiety of GABA (Cromer et al., 2002). In addition, mutagenesis studies have determined that nearby residues, β2Tyr157 and β2Thr160, are important for GABA binding (Amin and Weiss, 1993). Similarly, amino acid residues in aligned regions of the muscle-type nicotinic acetylcholine α1 (Trp148, Tyr151, and Asp152) (Dennis et al., 1988; Galzi et al., 1991; Sugiyama et al., 1996), glycine α1 (Asp148, Gly160, and Tyr161) (Vandenberg et al., 1992, 1993), and serotonin type 3 receptor subunits (Trp160) (Spier and Lummis, 2000) have been determined to be critical for agonist-antagonist binding (see Fig. 1) and define region “B”a of the ligand-binding site. The contributions of these residues in forming their respective agonist binding sites are supported by homology models that place this region (from β-strand 7 and loop 8) within the putative core of LGIC neurotransmitter-binding sites (Cromer et al., 2002; Holden and Czajkowski, 2002; LeNovère et al., 2002; Newell and Czajkowksi, 2003; Reeves et al., 2003).
Here, we demonstrate that expression of α1β2(E155C) gives rise to spontaneously open GABA channels. Mutation of β2Glu155 alters both channel-gating properties and impairs agonist binding. In addition, we provide evidence that β2Thr160 and β2Asp163 are found on an aqueous surface within or near the GABA-binding site and undergo conformational rearrangements during pentobarbital-mediated gating events. Together, the data suggest that movement in this region of the GABA-binding site is one of the initial triggers for coupling binding to gating.
Materials and Methods
Mutagenesis and expression in oocytes. Rat cDNAs encoding the GABAAR α1 and β2 subunits were used in this study. The β2 cysteine mutants were engineered using a recombinant PCR method, as described previously (Boileau et al., 1999; Kucken et al., 2000). Cysteine substitutions were made in the β2 subunit at Ile154, Glu155, Ser156, Tyr157, Gly158, Tyr159, Thr160, Thr161, Asp162, and Asp163 (see Fig. 1). Cysteine substitutions were verified by restriction endonuclease digestion and double-stranded DNA sequencing.
All wild-type and mutant cDNAs were subcloned into the vector pGH19 (Liman et al., 1992; Robertson et al., 1996) for expression in Xenopus laevis oocytes. Oocytes were isolated as described previously (Boileau et al., 1998). cRNA transcripts were prepared using the T7 mMessage machine (Ambion, Austin, TX). GABAA receptor α1 and β2 or β2 mutant subunits were coexpressed by injection of cRNA (675 pg/subunit) in a 1:1 ratio (α:β). The oocytes were maintained in modified ND96 medium [containing (in mm): 96 NaCl, 2 KCl, 1 MgCl2, 1.8 CaCl2, and 5 HEPES, pH 7.4] that had been supplemented with 100 μg/ml gentamicin and 100 μg/ml bovine serum albumin. Oocytes were used 2-7 d after injection for electrophysiological recordings.
Two-electrode voltage-clamp analysis. Oocytes under two-electrode voltage clamp were perfused continuously with ND96 recording solution [containing (in mm): 96 NaCl, 2 KCl, 1 MgCl2, 1.8 CaCl2, 5 HEPES, pH 7.4] at a rate of ∼5 ml/min. The holding potential was -80 mV. The volume of the recording chamber was 200 μl. Standard two-electrode voltage-clamp procedures were performed using a GeneClamp500 Amplifier (Axon Instruments, Foster City, CA). Borosilicate electrodes were filled with 3 m KCl and had resistances of 0.5-3.0 MΩ in ND96. Stock solutions of GABA, 2-(3-carboxypropyl)-3-amino-6-(4-methoxyphenyl) pyridazinium (SR-95531) and piperidine-4-sulfonic acid (P4S) (Sigma, St. Louis, MO) were prepared in water, whereas stock solutions of picrotoxin (PTX) (Sigma) and N-biotinylaminoethyl methanethiosulfonate (MTSEA-biotin; 100 mm; Biotium, Hayward, CA) were prepared in dimethylsulfoxide (DMSO). All compounds were diluted appropriately in ND96 such that the final concentration of DMSO was ≤2%. The vehicle did not affect GABA-activated currents.
To measure the sensitivity to GABA or P4S, the agonist (0.0001-100 mm) or partial agonist (0.00001-10 mm) was applied via gravity perfusion or by pipette application (∼5-8 sec) with a 3-15 min washout period between each application to ensure complete recovery from desensitization. Peak GABA- or P4S-activated current (IGABA or IP4S) was recorded. To correct for slow drift in the amplitude of the response as a function of time, concentration-response data were normalized to a low concentration of agonist (EC2-EC5). The apparent affinity of pentobarbital using concentrations between 0.01 and 10 mm was determined via gravity perfusion (∼5-8 sec) with a 3-5 min washout period between each application. Peak pentobarbital-activated current was recorded. Concentration-response data for pentobarbital were normalized to a previous application of pentobarbital (100 μm). Concentration-response curves for GABA, P4S, or pentobarbital were generated for each recombinant receptor, and the data were fitted by nonlinear regression analysis using Prism software (GraphPad, San Diego, CA). Data were fitted to the following equation: I = Imax/(1 + (EC50/[A])n), where I is the peak amplitude of the current for a given concentration of GABA, P4S, or pentobarbital ([A]), Imax is the maximum current, EC50 is the concentration required for half-maximal receptor activation, and n is the Hill coefficient.
To determine SR-95531 or PTX IC50 values, GABA (EC50) was applied via gravity perfusion followed by a brief washout period (20 sec) before application of GABA (EC50) and increasing concentrations of SR-95531 or PTX. The response to the application of SR-95531/PTX and GABA was normalized to the response elicited by the agonist alone. Concentration-inhibition curves were generated by nonlinear regression analysis using GraphPad Prism software. Data were fitted to the following equation: 1 - 1/(1 + (IC50/[Ant])n), where IC50 is the concentration of antagonist ([Ant]) that reduces the amplitude of the GABA-evoked current by 50%, and n is the Hill coefficient. SR-95531 KI values were calculated using the Cheng-Prussof correction: KI = IC50/(1 + ([A]/EC50)), where [A] is the concentration of GABA used in each experiment, and EC50 is the concentration of GABA that elicits a half-maximal response for each receptor (Cheng and Prussof, 1973).
Modification of cysteine residues by MTSEA-biotin. MTSEA-biotin was the cysteine-specific reagent used in this study because it is a relatively impermeant compound (Daniels and Amara, 1998), the dimensions (14.5 Å unreacted moiety; 11.2 Å reacted moiety) of which are similar to SR-95531 (13.5 Å) but longer than GABA (4.5 Å). Therefore, it is reasonable to assume that MTSEA-biotin can occupy the GABA-binding site and that this reagent will principally modify extracellular cysteine residues. We used the following criterion for stability of the response for these studies: ≤10% variation in IGABA (EC50) on two consecutive applications at regular intervals (10 min). Oocytes were then allowed to recover fully, after which a high concentration of MTSEA-biotin (2 mm) was applied (2 min). After MTSEA-biotin application, cells were washed (5 min) with ND96, after which GABA (EC50) was reapplied to determine the effect of MTSEA-biotin application on IGABA. Effects of MTSEA-biotin were calculated as the difference in the amplitude of the GABA-gated current before and after MTSEA-biotin application as follows: (IGABAPRE - IGABAPOST/IGABAPRE) × 100, where “post” refers to the amplitude of the GABA-gated current after MTSEA-biotin application, and “pre” refers to the amplitude of the stabilized GABA-gated current before covalent modification by MTSEA-biotin.
Rate of MTSEA-biotin reaction. The rate at which MTSEA-biotin modified introduced cysteine residues (E155C, T160C, and D163C) was measured using low MTSEA-biotin concentrations, as described previously (Newell and Czajkowski, 2003). The concentrations of MTSEA-biotin used were 50 nm (D163C), 100 nm (T160C), and 2 mm (E155C). The experimental protocol is described as follows: GABA (EC50) application (5 sec), ND96 washout (25 sec), MTSEA-biotin application (10-20 sec), ND96 washout (2.2-2.3 min). The sequence was repeated until IGABA no longer changed after the MTSEA-biotin treatment (i.e., the reaction had proceeded to apparent completion). The individual abilities of GABA, SR-95531, and pentobarbital to alter the rate of cysteine modification by MTSEA-biotin were determined by coapplying GABA (5 × EC50), SR-95531 (6 × KI), or pentobarbital (50 or 500 μm) during the MTSEA-biotin pulse. In all cases, the wash times were adjusted to ensure that currents obtained from test pulses of GABA (EC50) after exposure to high concentrations of GABA, SR-95531, or pentobarbital were stable. This ensured (1) a complete washout of drugs and that (2) reductions in the current amplitude resulted from the application of MTSEA-biotin.
For all rate experiments, the decrease in IGABA was plotted as a function of the cumulative time of MTSEA-biotin exposure and fit to a single-exponential decay function using GraphPad Prism software. A pseudo-first-order rate constant (k1) was determined, and the second-order rate constant (k2) was calculated by dividing k1 by the concentration of MTSEA-biotin used in the assay (Pascual and Karlin, 1998). Second-order rate constants were determined using at least two different concentrations of MTSEA-biotin to ensure accuracy of the protocol.
Statistical analysis. Log (EC50) values, log (KI) values, and log (k2) rates were analyzed using a one-way ANOVA followed by a Dunnett's post hoc test to determine levels of significance.
Structural modeling. The mature protein sequences of the rat α1 and β2 subunits were homology modeled with a subunit of the ACh binding protein (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 Bank Viewer (SPDBV). The α1 protein sequence from Thr12-Ile227 and the β2 protein sequence from Ser10-Leu218 were aligned with the AChBP primary amino acid sequence as depicted by Cromer et al. (2002) and threaded onto the AChBP tertiary structure using the “Interactive Magic Fit” function of SPDBV. The threaded subunits were imported into SYBYL (Tripos, St. Louis, MO), in which energy minimization was performed (<0.5 kcal/Å). The first 100 iterations were performed using Simplex minimization (Press et al., 1988) followed by 1000 iterations using the Powell conjugate gradient method (Powell, 1977). A β2/α1 GABA-binding site interface 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, several TRIPOS programs were run to evaluate the accuracy of the model. Ramachandran plots, χ plots, side-chain positions, and cis- and trans-bonds were all examined. Problems in the structure that were revealed by these evaluations were fixed manually, and energy minimizations were run again as needed. Our model is quite similar to models published recently for the nicotinic ACh receptor (nAChR) and GABAAR ligand-binding domains (Cromer et al., 2002; LeNovère et al., 2002). Regions with insertions were modeled by fitting structures from a loop database. Because the sequence identity of the AChBP and the GABAAR extracellular ligand-binding domain is only 18%, caution must be used in interpreting the absolute positions of individual side-chain residues in the model.
Expression and functional characterization of cysteine mutants
Cysteine substitutions were engineered at 10 individual positions (Fig. 1) in the GABAAR β2 subunit (Ile154, Glu155, Ser156, Tyr157, Gly158, Tyr159, Thr160, Thr161, Asp162, and Asp163), coexpressed with wild-type α1 subunits in X. laevis oocytes, and analyzed using two-electrode voltage clamp. Expression of most mutant β2 subunits produced functional channels (IGABA = 1-10 μA), with the exceptions Y157C and Y159C. We speculate that introduction of cysteine residues at these positions impaired assembly of mutant receptors.
For the cysteine mutants that did express, seven of eight significantly increased GABA EC50 values, demonstrating that this region is particularly sensitive to structural perturbation. Expression of receptors containing I154C, E155C, S156C, G158C, T160C, D162C, and D163C increased GABA EC50 values 8-, 3375-, 22-, 260-, 23-, 18-, and 9-fold relative to wild type (1.6 μm) (Fig. 2A, Table 1). The Hill coefficients for GABA activation of G158C- and D163C-containing receptors were significantly lower than wild type. The KI values for the competitive antagonist SR-95531 were significantly different from wild type (KI = 163 nm) for E155C, S156C, and G158C by 11-, 3-, and 18-fold, respectively (Table 1, Fig. 2B). Small currents (Imax < 90 nA) of receptors containing G158C precluded additional analysis.
Pentobarbital is a barbiturate that exerts its pharmacological effects (allosteric modulation and channel opening) via interactions with the GABAA receptor at a binding site distinct from the GABA site (Akk and Steinbach, 2000). Pentobarbital is therefore useful for assessing the consequences of mutating residues located near the GABA-binding site on overall receptor structure-function. Pentobarbital activated wild-type receptors with an EC50 of 1.1 ± 0.3 mm (n = 4) (Table 2) but failed to elicit current in receptors containing Y157C or Y159C, again suggesting that these mutant β2 subunits did not assemble into functional channels. Sensitivity to pentobarbital was increased approximately twofold for E155C- and T160C-containing receptors [EC50 values = 0.53 ± 0.1 mm (n = 4) and 0.32 ± 0.03 mm (n = 3), respectively], whereas expression of D163C (EC50 = 2.0 ± 0.3 mm; n = 3) had no significant effect on pentobarbital EC50. These data suggest that the rightward GABA EC50 shifts (Table 1) measured for E155C-, T160C-, and D163C-containing receptors are attributable to local effects at the GABA-binding site.
Although it is impossible to know whether the introduced cysteine residues occupy positions equivalent to wild-type residues, because SR-95531 and pentobarbital apparent affinities were similar or, in certain cases, more potent for some mutant receptors in which there were large rightward shifts in GABA EC50, we believe gross structural reorganizations of the GABAA receptor did not occur as a result of these mutations.
Spontaneous openings at α1β2(E155C)
Expression of α1β2(E155C) receptors gave rise to higher than normal resting conductances (Ileak), the magnitudes of which (-609 ± 76 nA; n = 9) were ∼12-fold greater than injection-matched wild-type receptors (-51 ± 9 nA; n = 9). To determine the nature of this high resting conductance, we applied the GABAAR channel blocker PTX. PTX (100 μm) reduced Ileak by 72.1 ± 4.2% (n = 3) (Fig. 3A), demonstrating that spontaneously open GABAAR channels accounted for the high resting conductance. PTX inhibited GABA-activated currents elicited from α1β2(E155C) receptors with an IC50 value of 3.1 μm, which was not significantly different from wild-type receptor values (4.7 μm) (Fig. 3B-D, Table 2).
Spontaneously active LGICs often arise as a consequence of mutations in the M2 channel-lining segment, and a characteristic of these constitutively open channels is a leftward shift in agonist concentration responses (Bertrand et al., 1992; Filatov and White, 1995; Labarca et al., 1995; Tierney et al., 1996; Chang and Weiss, 1998, 1999; Thompson et al., 1999; Findlay et al., 2000). However, this was not the case on expression of β2E155C, wherein we observed a 3375-fold decrease in GABA sensitivity. The sensitivity of the partial agonist (P4S) was also reduced (with no apparent reduction in efficacy) (Fig. 4A,B), albeit to a lesser extent than that of GABA (152-fold) (Table 2, Fig. 4C). The mutation also decreased the sensitivity of the competitive antagonist SR-95531 (11.1-fold rightward shift) (Table 2). Again, it should be noted that for this mutation, pentobarbital sensitivity was increased twofold relative to wild type (Table 2, Fig. 4D). Mutation of ρ1Y102 located in the GABAC receptor D region of the agonist binding site (i.e., β-strand 2) has also been reported to result in spontaneously open channels with similar properties (Torres and Weiss, 2002).
Alterations in EC50 values are difficult to evaluate because changes in either ligand binding and/or channel gating can alter this macroscopic constant (Colquhoun, 1998). The increase in open probability for the E155C mutant indicates that the mutation altered GABAA receptor channel gating. If the mutation altered gating exclusively, similar-fold changes in the apparent affinities of a series of ligands as well as leftward shifts in their concentration responses would be expected (Zhang et al., 1994). The apparent affinities for GABA, P4S, and SR-95531 were altered by different factors (Table 2), and rightward shifts in their concentration responses were observed. Thus, these data indicate that, besides altering gating, E155C decreased the binding of orthosteric ligands to the GABA-binding site. We can exclude the possibility that these differential effects arise from a mixed population of receptors [i.e.,α1β2(E155C) and β2(E155C)] because expression of β2E155C alone produced no currents.
In addition to PTX, covalent modification of E155C by thiol-specific reagents closed the spontaneously open channels. The leak current was reduced by MTSEA-biotin (2 mm; 57.2 ± 0.7%; n = 3), MTSEA-biotin-CAP (N-biotinylcaproylaminoethyl methanethiosulfonate) (2 mm; 63.0 ± 1.8%; n = 3), 2-aminoethyl methanethiosulfonate (MTSEA) (2 mm; 39.4 ± 1.4%; n = 3), MTSET (2-(trimethylammonium)ethyl methanethiosulfonate) (2 mm; 30.5 ± 0.9%; n = 3), and MTSES (2-sulfonatoethyl methanethiosulfonate) (2 mm; 26.8 ± 7.5%; n = 3) (Fig. 5) The observation that Ileak is reduced by tethering different chemical groups directly onto E155C suggests that this region of the binding site may play a key role in the triggering of allosteric transitions from the closed to open state.
MTSEA-biotin modification of cysteine residues
To further examine the β2 subunit Ile154-Asp163 region, we assessed the accessibility of cysteine residues introduced into this region. Wild-type and mutant receptors were exposed to MTSEA-biotin (2 mm; 2 min), a thiol-specific reagent that modifies water-accessible cysteine residues (Karlin and Akabas, 1998). MTSEA-biotin significantly reduced IGABA for receptors containing E155C (91.8 ± 1.5%; n = 6), G158C (34.6 ± 4.5%; n = 4), T160C (60.9 ± 3.0%; n = 10), and D163C (98.9 ± 3.7%; n = 8) (Fig. 6). MTSEA-biotin had no effect on wild-type receptors or those containing I154C, S156C, T161C, and D162C. Lack of effect indicates that the thiol group was not accessible to modification or that modification produced no detectable 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 envision this region of the GABAA receptor forming a loop structure (Cromer et al., 2002).
MTSEA-biotin rates of reaction
The rate of reaction of MTSEA-biotin with an introduced cysteine mainly depends on the ionization of the thiol group and the access route of the methanethiosulfonate reagent (Karlin and Akabas, 1998). Cysteine residues with ionized sulfhydryls react 108 to 109 times faster than nonionized sulfhydryls (Roberts et al., 1986). Second-order rate constants therefore provide information about the local environment of a substituted cysteine. The fast second-order rate constants (in the absence of other ligands) for MTSEA-biotin modification of D163C (604,771 m-1sec-1) and T160C (286,100 m-1sec-1) indicate that both residues are found in an open, aqueous environment. The second-order rate constant for E155C is significantly slower (27.9 m-1sec-1), suggesting that the thiol group is not well ionized and/or that the introduced cysteine residue is in a restricted-buried environment (Table 3, Fig. 7).
To determine whether a given residue lies near the neurotransmitter binding site, MTSEA-biotin reaction rates were measured in the presence of GABA and the competitive antagonist SR-95531. These ligands promote different conformational changes in the binding site (Boileau et al., 2002), and thus, if the rate at which MTSEA-biotin reacts with an introduced cysteine is slowed by both ligands, then it is likely that the ligands are sterically blocking the reaction and that the sulfhydryl side chain is facing into or near the GABA-binding site. Both GABA (at EC90 concentration) and SR-95531 (at IC90 concentration) significantly slowed modification of T160C and D163C approximately twofold (Fig. 8), suggesting that these residues are found within or near the GABA-binding site (Fig. 9). Although the data are consistent with GABA and SR-95531 causing a steric block, it is feasible that the binding of either ligand induces local allosteric changes in the receptor that leads to the slowing of MTSEA-biotin reaction rates. Neither ligand slowed modification of E155C (Table 3). Because α1E155C receptors are spontaneously open, the control rate of MTSEA-biotin modification of E155C likely reflects reaction to a “ligand-bound, active” conformation of the binding site. Thus, the result that GABA and SR-95531 had no effect on modification rate is not surprising.
Effect of pentobarbital on MTSEA-biotin second-order rate constants
To identify whether movements occur in and near the Ile154-Asp163 region of the GABA-binding site during channel gating and modulation, we measured the rates of reaction of MTSEA-biotin with T160C, D163C, and E155C in the presence pentobarbital. The ability of pentobarbital to alter the rates of modification provides an indirect measure of changes that occur within this region of the binding cleft in the transition from the resting to the active-desensitized states. As a result of the slowness of drug application using the oocyte expression system, we cannot easily determine whether the movement is associated with open or desensitized states. Nevertheless, coapplication of pentobarbital and MTSEA-biotin should capture receptor states that differ from that captured by application of MTSEA-biotin alone. Concentrations of pentobarbital (500 μm) that activate the receptor slowed the rate of MTSEA-biotin modification of T160C and D163C approximately twofold but had no effect on the second-order rate constant for E155C (Table 3). Thus, T160C and D163C act as reporters of barbiturate-mediated channel gating. Concentrations of pentobarbital that do not open the channel but potentiate GABA current (e.g., 50 μm) (Table 3) slowed modification of T160C but not D163C (Fig. 7B,C; Table 3). These data suggest that movements within the binding site associated with channel gating versus allosteric modulation are distinct.
Binding-site movements that initiate ligand-gated ion channel activation are not well established. Here, we provide evidence that movement in the β2 Ile154-Asp163 region of the GABAAR is involved in coupling GABA binding to channel gating, and we describe a role for β2Glu155 as an initial trigger for ion channel opening.
β2 Ile154-Asp163 mutations affect ligand binding and channel gating
If the β2 Ile154-Asp163 region plays a pivotal role for coupling binding to gating, one would expect that mutations within this domain would affect both processes. Seven cysteine substitutions significantly increase GABA EC50 values, which reflect changes in either microscopic binding affinity and/or channel gating (Colquhoun, 1998). Three of the seven mutations that shift the GABA EC50 (E155C, S156C, and G158C) also significantly reduce SR-95531 KI, suggesting that at least one effect of these mutations is to alter ligand binding, because SR-95531 does not gate the channel (but see Ueno et al., 1997). α1β2(G158C) and α1β2(D163C) receptors have significantly reduced Hill coefficients for GABA activation, consistent with a reduction in gating efficacy (Colquhoun, 1998). β2E155C results in spontaneously open GABAARs, clearly demonstrating that this mutation alters channel gating. In addition, expression of β2E155C differentially shifts the concentration dependencies rightward for GABA, SR-95531, and P4S, indicating that the mutation also perturbs ligand binding (Zhang et al., 1994). Furthermore, tethering thiol-reactive groups onto β2E155C closes the spontaneously open channels. Together, the data suggest that β2Glu155 occupies a key position in the activation pathway involved in coupling ligand binding site to channel gating. Although detailed kinetic analyses of these mutations are required to quantitatively tease apart the effects of each of these mutations on microscopic binding affinity and channel gating properties, the above results indicate that mutations in β2 Ile154-Asp163 region of the GABA-binding site disrupt both affinity and efficacy.
Structural rearrangements during gating transitions
The β2 subunit forms the principal side of the GABA-binding site. We conclude that β2Asp163 and β2Thr160 are found within or near the GABA-binding site, based on a slowing of the rate of MTSEA-biotin modification of T160C and D163C by both GABA and SR-95531. Protection by both ligands suggests steric hindrance of MTSEA-biotin modification, because agonists and antagonists promote different conformational changes within the binding site (Armstrong and Gouaux, 2000). In addition, mutagenesis and homology modeling studies suggest that the β2 Ile154- Asp163 region lines part of the GABA-binding site (Amin and Weiss, 1993).
Tierney et al. (1996) predict that movements within the β2 subunit are critical for channel gating. To test the hypothesis that the β2 Ile154-Asp163 region of the binding site undergoes structural rearrangements during channel activation, we measured the rate of MTSEA-biotin modification of introduced cysteines in the presence of pentobarbital (500 μm). Although the binding sites for GABA and pentobarbital differ, the final structure of the activated GABAA receptor channel is likely similar because both drugs produce similar single channel conductances (Jackson et al., 1982; Akk and Steinbach, 2000). We can therefore monitor pentobarbital-induced movements in the GABA-binding site during state transitions from resting to open-desensitized states. Coapplication of pentobarbital and MTSEA-biotin should capture this region of the receptor in a conformation that differs from that captured by application of MTSEA-biotin alone. The observation that pentobarbital significantly slows modification of T160C and D163C (approximately twofold) indicates that the environment surrounding these residues changes and that they are “conformationally sensitive” to channel activation, supporting our hypothesis that movements within this region are critical for channel gating. T160C also appears to act as a reporter for movements that occur during allosteric modulation, because the rate of MTSEA-biotin modification of T160C was slowed almost twofold in the presence of a low concentration of pentobarbital (50 μm) that potentiates but does not gate the channel.
β2 Ile154-Asp163 is a protein hinge
Local agonist-induced movements within LGIC binding sites precede a conformational wave that leads to channel gating (Chakrapani et al., 2004). Structural studies suggest that the binding of ACh induces a 15° clockwise rotation of the inner β-sheets of the N-terminal ligand binding domain of the nAChR α1 subunits. This, in turn, brings the β1-β2 loop (loop 2) into contact with the extracellular M2-M3 loop, and movement of the M2-M3 loop then causes the M2 region to rotate, which leads to opening of the channel gate (Miyazawa et al., 2003). Linear free energy analysis of the nAChR suggest that a conformational wave begins at the binding site and region A of the ACh-binding site (β-strand 4 and loop 5), followed by movements of loops 2, 7 (Cys-Cys loop) and the M2-M3 linker at the extracellular juxtapore region, and finally movement of the transmembrane domains (Grosman et al., 2000; Chakrapani et al., 2004). Studies examining the structural mechanisms of GABAAR activation have identified pairs of interacting residues within these regions that are necessary for coupling GABA binding to channel gating-desensitization. These include electrostatic interactions between negatively charged residues in loops 2 (α1Asp57) and 7 (α1Asp149) of the GABAAR and a positively charged lysine (α1Lys279) in the M2-M3 loop (Kash et al., 2003). Recently, a study using a chimeric receptor comprised of AChBP fused to the transmembrane pore domain of the 5-HT3A receptor demonstrated that only when loops 2, 7, and 9 (region F of the binding site) from AChBP were replaced with 5-HT3A receptor sequences did ACh binding trigger channel opening (Bouzat et al., 2004). This indicates that loops 2, 7, and 9 are critical elements involved in coupling the extracellular binding site domain to the transmembrane channel gating domain. During examination of homology models of the GABAAR, we noticed that the β2 Ile154-Asp163 region of the binding site (β-strand 7 and loop 8) physically links loops 2 and 9. Thus, we speculate that the β2 Ile154-Asp163 region may act as a protein hinge and that structural rearrangements within this region of the binding site could therefore be an efficient means for simultaneously propagating movements to both loops 2 and 9.
Movements in region B are also likely to be transmitted to the region C of the GABA-binding site (i.e., end of β-strand 9, loop 10 and beginning of β-strand 10). Based on homology models, possible interactions within regions B and C include the following amino acid pairs: β2Glu153 and β2Lys196, β2Glu155 and β2Arg207, as well as β2Glu165 and β2Lys197. Previously, we demonstrated that β2Arg207 stabilizes GABA binding (Wagner and Czajkowski, 2001; Wagner et al., 2004). We predict that the carboxylate side chain of Glu155 is within 2.5 Å from the guanido group of β2Arg207 (Fig. 9) and may play a role in positioning the β2Arg207 side chain. This may explain why mutation of β2Glu155 disrupts orthosteric ligand binding. However, mutation of β2Glu155 also produces spontaneously open channels and indicates that perturbation of this residue has additional long-range allosteric effects that are likely propagated to the channel gate by changes in the positions of regions B and C located in the binding site as well as loops 2 and 9 near the juxtapore region. Additional experiments are needed to test these hypotheses. Support for interactions between regions B and C of LGIC binding sites comes from studies of the nAChR. It has been reported that a hydrogen bond between a residue in region B (G152K) and in region C (P193I) of the nAChR α7 subunit is important for nAChR activation (Grutter et al., 2003) and may serve to explain how the α1G153S human polymorphism gives rise to a slow channel myasthenic syndrome (Sine et al., 1995).
Finally, the β2 Ile154-Asp163 region may also be involved in intersubunit interactions. Models of the N-terminal domains of GABAAR predict that β2Asp163 forms a salt bridge with α1Arg119 and β2Arg28. The roles of salt bridges in GABAAR function are not presently known, but it is believed that salt bridges may limit the number of conformations of a protein complex, be key participants in determining ligand binding geometry, or be important for the association of subunits in multiprotein complexes (Hendsch and Tidor, 1994). The disruption or formation of bonds among these charged residues at subunit interfaces may be important for conformational changes that occur during activation and/or desensitization. Mutation of β2Asp163 significantly decreased the Hill coefficient for GABA activation of the receptor and is consistent with this hypothesis.
The β2 Ile154-Asp163 region of the GABAAR-binding site appears to be a protein hinge that is uniquely positioned to transduce binding site movements to the cascade of events that lead to opening of the ion channel. We demonstrate that the region undergoes conformational rearrangements during pentobarbital-mediated gating events, and mutation of β2Glu155 gives rise to spontaneously open channels, suggesting that movements in this region of the GABA-binding site are one of the initial triggers for coupling binding to gating. Ultimately, precise mapping of inter-residue contacts will be required to test this activation mechanism and to define the pathway leading from the binding site to the channel gate.
↵ a Throughout this paper, the six previously identified binding site regions are named by the letters A-F (Galzi and Changeux, 1994; Lester et al., 2004). Based on the structure of AChBP (Brejc et al., 2001), these regions can be defined as follows: region A, β strand 4 and loop 5; region B, β strand 7 and loop 8; region C, β strand 9, loop 10 and β strand 10; region D, β strand 2; region E, β strands 5′ and 6; region F, loop 9. When we refer to regions that are not part of the binding site, these are named by numbers based on AChBP structure.
This work was supported by National Institute of Neurological Disorders and Stroke Grant 34727 to C.C. and a Postdoctoral Fellowship from the Natural Sciences and Engineering Research Council of Canada to J.G.N.
Correspondence should be addressed to Cynthia Czajkowski, Department of Physiology, University of Wisconsin-Madison, 601 Science Drive, Madison, WI 53711. E-mail:.
J. G. Newell's present address: Department of Physiology, University of Toronto, Room 3318, Medical Sciences Building, 1 King's College Circle, Toronto, Ontario, Canada M5S 1A8.
R. A. McDevitt's present address: Graduate Program in Neurobiology and Behavior, University of Washington, Box 357270, Room T-471, Seattle, WA 98195.
Copyright © 2004 Society for Neuroscience 0270-6474/04/2411226-10$15.00/0