Mutagenesis of the Rat α1 Subunit of the γ-Aminobutyric Acid_A Receptor Reveals the Importance of Residue 101 in Determining the Allosteric Effects of Benzodiazepine Site Ligands

Materials and Methods

Preparation of Transcripts and Oocyte Injection.

Site-directed mutagenesis of α1 subunits was performed as described previously (Davies et al., 1998) using the Altered Sites kit (Promega, Madison, WI). The α1 subunit was subcloned into the pAlter-1 vector for mutagenesis. Potential mutants were identified by the presence of a restriction site that was introduced as a silent mutation by the mutagenic oligonucleotide. The presence of the correct substitutions was verified by sequencing. The mutant α1 cDNAs were then subcloned into the pcDNA3 expression vector (InVitrogen, San Diego, CA).

cDNAs encoding the α1, β2, and γ2L subunits in the pcDNA3 vector were linearized, and cRNA transcripts were prepared by standard procedures as previously described (Hope et al., 1993). The cRNA transcripts were injected (50 nl of 1 mg/ml cRNA) into X. laevis oocytes (stage V to VI) that had been previously defolliculated by treatment with 2 mg/ml collagenase A (Boehringer-Mannheim, Indianapolis, IN) for 3 h at room temperature (Belelli et al., 1996). Injected oocytes were individually maintained at 19–20°C for up to 14 days in 96-well plates containing 200 μl of standard Barth’s solution [15 mM HEPES, pH 7.6, 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 0.5 mM Ca(NO₃)₂, 0.5 mM CaCl₂, 1 mM MgSO₄] supplemented with 0.1 mg/ml gentamycin.

Electrophysiological Recordings.

Oocytes were used for experimentation 2 to 14 days after cRNA injection. The methodology was essentially the same as described previously (Pistis et al., 1997). Briefly, electrical recordings were made from oocytes voltage-clamped at −60 mV using a GeneClamp 500 amplifier (Axon Instruments, Inc., Foster City, CA) in the two-electrode voltage-clamp mode. The oocytes were held in a chamber (0.5 ml) and continuously superfused (7–10 ml/min) with frog Ringer’s solution (5 mM HEPES, pH 7.4, 120 mM NaCl, 2 mM KCl, and 1.8 mM CaCl₂). The voltage-sensing and current passing electrodes were filled with 3 M KCl and had resistances of 0.5 to 1.5 MΩ when measured in frog Ringer’s solution. Agonist-induced responses were low pass filtered at 100 Hz, recorded onto videotape, and simultaneously displayed on a chart recorder. The peak amplitude of the agonist-evoked response was measured manually. All drugs were applied via the superfusion system.

For each oocyte, a maximal concentration of GABA (3 mM) was applied, and the resulting peak current amplitude was measured. This concentration of GABA was reapplied at approximately 20-min intervals until the current amplitude was consistent to within ±2% over three successive challenges.

To investigate the effects of the benzodiazepines, GABA concentrations eliciting a peak current approximately 10% (EC₁₀) or 50% (EC₅₀) of the maximum GABA response were used as indicated in the text. Pharmacological effects were first screened using a concentration of GABA approximating its EC₅₀ value. Subsequently, the EC₁₀ concentration of GABA was used to examine potentiating effects and, unless where noted in the text, the EC₅₀ concentration of GABA was used to further characterize inverse agonist and antagonist effects. Benzodiazepines were dissolved in dimethyl sulfoxide to give stock concentrations of 30 mM, and these stocks were stored frozen at −20°C. In all experiments, the concentration of dimethyl sulfoxide was constant at 0.02%, a concentration that produced no overt vehicle effects. In each experiment, control currents using the indicated concentrations of GABA were first recorded. This was followed by 5-min washing with frog Ringer’s solution and then by 3-min perfusion with the benzodiazepine before coapplication with the appropriate concentration of GABA. After recording of the evoked current, the oocyte was again washed with frog Ringer’s for at least 5 min before further benzodiazepine perfusion. During each experiment, control currents were recorded periodically to verify that the response remained stable throughout and that any benzodiazepine effects were fully reversible.

Data Analysis.

All data represent the mean (±S.E.) of observations made from at least three oocytes. Data were analyzed by nonlinear regression techniques using InPlot 4 (GraphPad Software, San Diego, CA). The GABA concentration dependence of the observed current was fit by the equation:where I is the measured amplitude of the evoked current, [L] is the GABA concentration, EC₅₀ is the concentration of GABA producing 50% of the maximal response (I_max), and n is the Hill coefficient. In each experiment, the observed current was normalized to theI_max (100%), and these normalized data were used to construct concentration-effect curves.

Unless otherwise indicated, the effects of the benzodiazepines were fit by the equation:where I is the amplitude of the observed current; [X] is the logarithm of the concentration of the benzodiazepine;I₀ and E_max are the currents observed in the absence of benzodiazepine and in the presence of the maximally effective concentration, respectively;C is the logarithm of the EC₅₀ for the benzodiazepine response; and n is the Hill coefficient. As shown in the figures, the concentration dependencies of many of the benzodiazepine effects deviated from a simple process, giving Hill coefficients of greater than or less than 1. Unfortunately, without detailed knowledge of the mechanisms underlying these allosteric interactions, it is not presently feasible to develop a molecular framework within which to interpret these Hill coefficients.

Curves that were obviously biphasic were fit by the equation:where E₁ andE₂ are the amplitudes of the first and second phases, C₁ and C₂are the logarithms of the corresponding EC₅₀values, and the other parameters are as described above.

In experiments to estimate antagonist affinities, the effects of Ro15-1788 or Ro15-4513 on the FNZ-induced potentiation were investigated. IC₅₀ values were calculated from the equation:where I is the observed current,I_FNZ is the current in the presence of FNZ,I₀ is the current in the presence of both FNZ and a saturating concentration of benzodiazepine antagonist, [X] is the logarithm of the antagonist concentration, C is the logarithm of the IC₅₀ concentration, andn is the Hill coefficient. In these experiments, the data were normalized by first expressing the magnitude of the FNZ-induced potentiation of the GABA response (in the absence of antagonist) as 100% and then calculating the inhibition of this current in the presence of different concentrations of antagonist.

Drugs.

GABA and FNZ were obtained from Sigma Chemical Co. (St. Louis, MO). Ro15-4513 and Ro15-1788 were generous gifts from Hoffmann-La Roche and Co. (Basel, Switzerland).

Results

At a holding potential of −60 mV, oocytes preinjected with cRNAs encoding rat α1, β2, and γ2 GABA_A receptor subunits responded to bath-applied GABA with an inward current response. Similarly, receptors carrying the various α subunit mutations displayed robust current responses to GABA application. All receptors showed some modulation by benzodiazepine-site ligands as illustrated for the wild-type and H101Y mutant receptor in Fig.1. Thus, all receptors examined displayed functional coupling between the GABA and benzodiazepine binding sites.

Analysis of the GABA concentration-response curve for the wild-type α1β2γ2 receptor gave an EC₅₀ value of 31.8 ± 3.4 μM (n = 4) and a Hill coefficient of 1.32 ± 0.16 (n = 4). None of the His101 mutations examined had a dramatic effect on GABA-induced currents (Table1), and although the EC₅₀ value was slightly increased in each case, this remained within a factor of about 2 of the wild-type receptor.

Mutations of the α1 subunit His101 to residues of altered chemical specificity did, however, have significant effects on both the potencies of the benzodiazepine-site ligands and on the manner in which they modulated GABA-evoked currents. For clarity, the qualitative changes in the pharmacological effects of FNZ, Ro15-1788, and Ro15-4513 are first summarized in Table 2.

The effects of FNZ on wild-type receptors were distinctly biphasic (Fig. 2A), suggesting a relatively potent potentiating effect (EC₅₀ = 5.8 nM) and a lower affinity inhibitory response (EC₅₀ = 178 nM). In the H101F, H101Y, and H101Q mutants, FNZ potentiated currents to a similar extent (E_max = 217–290% of control; Table 3). In contrast, FNZ had no effect on GABA-evoked currents in the H101K receptor (Fig. 2A), which is consistent with the lack of measurable FNZ binding to this receptor mutant when expressed in tsA201 cells (Davies et al., 1998). In the H101E mutant, which also displayed no measurable FNZ binding in mammalian cells, FNZ did have a potentiating effect but only at high ligand concentrations (>1 μM; Fig. 2B). Representative results for the effects of FNZ on GABA-induced currents are shown in Fig. 2, and the data are summarized in Table 3, where their measured EC₅₀ values are compared with their apparent affinities, measured in binding assays of receptors expressed in tsA201 cells. Thus, FNZ is an agonist of all receptors that recognize this ligand.

The consequences of H101 mutations on the effects of Ro15-4513 on functional responses were more complex. In the wild-type receptor, Ro15-4513 was a very limited inverse agonist, decreasing GABA-evoked currents by only 6 to 10% with an EC₅₀ of 1.8 nM (Fig. 3 and Table4). However, in the H101K and H101E mutants (and to a small extent in the H101Y mutant), this ligand became a partial agonist (see Fig. 3). Ro15-4513 had no direct effect on currents elicited by the H101Q mutant, but it did inhibit the potentiation induced by FNZ, indicating that it is a potent antagonist of this receptor. As previously reported for these receptors expressed in tsA201 cells (Davies et al., 1998), none of the mutations caused a pronounced decrease in the apparent affinities for Ro15-4513, and in some cases, notably H101F and H101Q, the potency of this ligand was increased compared with the wild-type receptor (Fig. 3 and Table 4).

Figure 4 illustrates the effects of the mutations on the ability of Ro15-1788 to modulate GABA-gated currents. In the wild-type receptor, this ligand had no direct effect on the GABA-evoked currents (Fig. 4A) but inhibited the FNZ-induced potentiation with an IC₅₀ value of 7.9 ± 1.9 nM (Fig. 4B). In the H101F mutant, Ro15-1788 also acted as an antagonist, but in the H101Y, H101K, and H101E receptors, it became a partial agonist (Fig. 4 and Table 5). Ro15-1788 had no direct effect on GABA-evoked currents in the H101E mutant receptor. Unfortunately, the lack of a robust agonist response of FNZ (Fig. 2B) or Ro15-4513 (Table 4) precluded further characterization of its possible antagonist properties on this receptor.

Discussion

A central issue concerning the benzodiazepine site of the GABA_A receptor is how its occupation can result in a spectrum of pharmacological effects ranging from inverse agonism to full agonism. We examined the effects of mutations of His101 of the rat α1 subunit on the pharmacological specificity of FNZ, Ro15-1788, and Ro15-4513 (i.e., compounds that are commonly regarded as having agonist, antagonist, and partial inverse agonist properties, respectively). Although it has recently been suggested that His102 (human numbering) is not directly involved in determining the efficacies of benzodiazepine-site ligands (McKernan et al., 1998), the above results demonstrate that the chemical nature of the residue at the homologous position of the rat α1 subunit has a significant effect on the actions of these compounds.

The mutations that were introduced had relatively minor effects on the concentration dependence of GABA-evoked currents (Table 1), suggesting that none of the substitutions had serious adverse effects on the overall structure of the receptor. Other groups have, however, reported larger shifts in the EC₅₀ values for GABA activation when the corresponding residue in the α6 subunit was modified (Im et al., 1997).

The shifts in EC₅₀ values obtained for the effects of FNZ, Ro15-1788, and Ro15-4513 on GABA-gated currents tended to parallel the shifts in binding affinities reported previously (Davies et al., 1998). In most cases, there was excellent agreement between the present functional data for receptors expressed in oocytes and previous binding data using receptors expressed in tsA201 cells. One exception is the effect of FNZ on the H101F mutant, where the apparent affinity obtained from binding experiments was 92.8 ± 9.1 nM compared with 8.6 ± 1.0 nM measured in the functional studies. In the absence of any other interpretation, it must be assumed that this is a cell type-specific phenomenon.

The binding of FNZ is particularly sensitive to the nature of the residue occupying position 101. Incorporation of the positively charged lysine resulted in receptors that were insensitive to FNZ. This is consistent with previous reports (Wieland et al., 1992; Benson et al., 1998) in which substitution of His101 in different α subunits by arginine produced receptors that were insensitive to diazepam. Similarly, the introduction of the negatively charged amino acid glutamate drastically affected FNZ binding, and in the H101E mutant, this ligand potentiated GABA-evoked currents only at high concentrations. In contrast, substitution by the uncharged glutamine restored sensitivity to FNZ (Table 3). The ability of glutamine to substitute for histidine in the recognition site has been discussed previously (Davies et al., 1998). In all mutant receptors that recognized FNZ, this drug acted as a strong agonist, potentiating GABA-induced currents to levels similar to those seen with wild-type receptors. This suggests that although the mutations reduce the affinity for FNZ, they do not produce qualitative changes in the protein conformational transitions that allosterically modulate GABA-gated chloride flux.

The efficacies of Ro15-1788 and Ro15-4513 were dramatically altered by some of the mutations, but the pattern of the changes differed for the two ligands. In the wild-type receptor, Ro15-4513 was a poor inverse agonist, decreasing GABA-evoked currents by less than 10%. Decreases of a similar magnitude have been reported previously in studies using both the human recombinant α1β2γ2 subtype (Hadingham et al., 1996) and acutely dissociated rat pyramidal neurons (Smith et al., 1998). As expected from previous studies (Sieghart, 1995), Ro15-1788 acted as an antagonist of the wild-type receptor. It has previously been shown that both Ro15-4513 and Ro15-1788 act as partial agonists in α6 receptors in which arginine is present in this position (Hadingham et al., 1996; Wafford et al., 1996). Recently, the importance of this residue was further confirmed by showing that substitution of His101 by arginine in the α1, α2, α3, and α5 subunits resulted in Ro15-4513 becoming a positive allosteric modulator (Benson et al., 1998). Another mutation in the γ2 subunit has been shown to change the efficacy of Ro15-4513 (Mihic et al., 1994), suggesting that determinants in both the α and γ subunits affect the interaction of this compound with the receptor. A recent report by Im et al. (1997)suggests that residues C terminal to position 101 may be involved in determining the efficacy of Ro15-1788 at α6-containing receptors. In the present study, we have shown that manipulation of this amino acid, at least in α1 subunits, is sufficient to bring about changes in efficacy, not only for Ro15-1788 but also for Ro15-4513.

Relatively small differences in the structure of the residue at position 101 had differential effects on receptor modulation by Ro15-1788 and Ro15-4513. This is illustrated by comparing the effects of these compounds on the H101Y and H101F receptors. In the H101Y mutant, Ro15-1788 is an agonist, whereas Ro15-4513 acts as an antagonist. In the H101F receptor, the converse is true. Thus, the pharmacological specificity of these ligands is dictated by the presence or absence of a hydroxyl group in this position. The affinities for both agents were reduced in the tyrosine mutant, indicating that changes in efficacy do not parallel changes in apparent affinity.

Because structural information on the GABA_Areceptor is lacking, any rationalization of the differential effects of the mutations on the actions of Ro15-4513 and Ro15-1788 is speculative. Molecular modeling has suggested that agonists, inverse agonists, and antagonists differentially affect the receptor by interacting with different lipophilic pockets within the binding site or sites. In the models proposed by Villar et al. (1989) and Zhang et al. (1995), the lipophilic pocket or pockets occupied by antagonists and inverse agonists are structurally related to each other, but the pocket that accommodates the 5-phenyl ring of benzodiazepine agonists lies in a much different position. Because of the similarities in the binding of antagonists and inverse agonists, small perturbations in binding site structure may be sufficient to change the particular lipophilic pocket with which these compounds interact, thus changing the efficacy of a ligand from inverse agonist to antagonist. To explain the partial agonism of Ro15-1788 and Ro15-4513 at some mutant receptors, two possibilities must be considered. First, both ligands may interact with the same regions of the binding cleft as in wild-type receptors but now act as agonists (i.e., the mutations affect events that are downstream from the initial binding interaction). Second, the structural changes induced by the mutations allow these compounds to interact with the lipophilic pocket that is normally “seen” only by agonists (i.e., the signal transduction mechanism remains the same as in wild-type receptors, but the initial receptor-ligand interaction is altered). We cannot presently distinguish between these possibilities, but it is hoped that further information will come from direct identification of binding domains for different ligands by peptide mapping of photolabeled mutant receptors.

It is generally assumed that agonists, antagonists, and inverse agonists interact with a common binding site because their binding is mutually exclusive at equilibrium. However, Buhr et al. (1997) recently showed that a single mutation in the γ2 subunit produced receptors that displayed biphasic displacement curves in which zolpidem, methyl 6,7-dimethoxy-4-ethyl-β-carboline-3-carboxylate, and Ro15-1788 were able to displace only about 50% of bound [³H]FNZ. The authors suggested that this may indicate the presence of two benzodiazepine binding sites within a single receptor molecule. In the present study, we identified one mutant (H101Q) that displays a shallow concentration dependence (Hill coefficient = 0.45 ± 0.11) for FNZ potentiation, suggesting that two benzodiazepine sites of different affinities may be present in these receptors. However, no evidence of binding site heterogeneity was obtained in direct binding studies in which FNZ was shown to completely displace [³H]Ro15-4513 (Davies et al., 1998). Also in the current study, we found that high concentrations of FNZ (>1 μM) potentiated GABA-gated ion flux in the H101E mutant, yet previous competition studies using this receptor showed that FNZ at concentrations up to 10 μM was unable to displace bound [³H]Ro15-4513. This was previously interpreted to mean that these receptors did not recognize FNZ. However, in light of the functional data for this mutant, it is possible that there is an additional low-affinity binding site for FNZ that is distinct from the high-affinity site for Ro15-4513.

The bell-shaped appearance of the curve for FNZ potentiation of the wild-type α1β2γ2 receptor (Fig. 2A) suggests that the presence of more than one benzodiazepine site is not exclusive to the mutant receptors. Evidence for more than one diazepam-responsive component in this subtype has also been reported by others (Amin et al., 1997). Im et al. (1993) showed that an additional benzodiazepine site may lie at subunit interfaces other than α/γ, and it has been suggested bySieghart (1995) that GABA_A receptors may carry both a high-affinity and a low affinity site for benzodiazepine agonists. The significance of the lower-affinity benzodiazepine site or sites is not yet known.

In conclusion, the results described above show that His102 of the α1 subunit is an important determinant of both the affinity and pharmacological efficacy of ligands that bind to the benzodiazepine site of the rat α1β2γ2 GABA_A receptor.