Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The principal therapeutic actions of drugs such as the benzodiazepines (e.g., diazepam), imidazopyridines (e.g., zolpidem), and triazolopyridazines (e.g., zaleplon) are effected through the family of GABA_A receptors (Lüddens et al., 1995; Korpi et al., 1997; Sigel and Buhr, 1997). Based on sequence homology, 17 distinct subunits belonging to six related families (_1-6, _1-3, _1-3, , , _1-2, ) have been identified as members of this group of ligand-gated ion channels (for review, see McKernan and Whiting, 1996; Sigel and Buhr, 1997; Bonnert et al., 1999). Assuming a pentameric arrangement (Nayeem et al., 1994), there is a remarkable potential for GABA_A receptor heterogeneity. Nonetheless, no more than 10 to 20 distinct GABA_A receptor isoforms have been identified in the adult rat central nervous system (Fritschy and Mohler, 1995; De Blas, 1996; McKernan and Whiting, 1996), with the majority existing as heteromers composed of -, -, and -subunits (Fritschy and Mohler, 1995; De Blas, 1996). Although the stoichiometry of native GABA_A receptors has not been definitively established, several studies have proposed a GABA_A receptor configuration as consisting of 2-, 2-, and 1-subunit (Chang et al., 1996; Tretter et al., 1997).

Studies using recombinant GABA_A receptors have demonstrated that subunit composition defines ligand pharmacology at these ligand-gated ion channels (Pritchett and Seeburg, 1990; Hadingham et al., 1993). This principle is amply illustrated by the impact of the -subunit on the affinities of a chemically diverse group of substances often termed benzodiazepine site ligands (for review, see Lüddens et al., 1995; Korpi et al., 1997). For example, prototypic 1,4- benzodiazepines such as diazepam and flunitrazepam possess high (nM) affinities for GABA_A receptors containing _1,2,3 and ₅ subunits [comprising the "diazepam-sensitive" (DS) family of GABA_A receptors], but are essentially inactive at receptors containing ₄ and ₆ subunits [the "diazepam-insensitive" family of GABA_A receptors] (Korpi et al., 1992; Wong et al., 1992; Wieland and Lüddens, 1994; Fritschy and Mohler, 1995). This remarkable effect on ligand affinity is determined in a large part, by a single histidine residue in homologous positions ₁101, ₂101, ₃126, and ₅105 of the DS -subunits and the cognate arginine in position 100 on the ₄ and ₆ receptors (Wieland et al., 1992; Benson et al., 1998).

The affinities of 1,4-benzodiazepines are very similar among both recombinant and native DS receptors (Mohler et al., 1978; Pritchett and Seeburg, 1990; Graham et al., 1996). Several nonbenzodiazepine molecules, including CL 218,872 and zolpidem, exhibit some selectivity for recombinant GABA_A receptors bearing an ₁ subunit and possess higher affinities in brain regions (e.g., cerebellum) that are relatively enriched in this species (Squires et al., 1979; Pritchett and Seeburg, 1990; Hadingham et al., 1993). Only recently have very high-affinity, selective compounds been developed for less abundant GABA_A receptor isoforms. Thus, based on the ~10-fold selectivity of Ro 15-4513 for GABA_A receptors containing an ₅ subunits (Hadingham et al., 1993; Lüddens et al., 1994), compounds such as RY-80, RY-24, and L-655,708 have been developed (Liu et al., 1995, 1996; Quirk et al., 1996). These imidazodiazepine derivatives exhibit high affinity and selectivity for wild-type and recombinant GABA_A receptors containing an ₅ subunit (Liu et al., 1995, 1996; Skolnick et al., 1997; Sur et al., 1998, 1999). Using these compounds as probes, we now identify a single amino acid residue on the ₅ subunit (Ile215) that is critical for ligand selectivity at recombinant ₅₃₂ receptors.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Transfection of Recombinant GABA_A Receptors and Membrane Preparation. cDNAs encoding rat ₁ and ₅ subunits were subcloned into a pRc/CMV vector, as described elsewhere (Skolnick et al., 1997). The ₃ and _2S cDNAs were subcloned into pcDNA3 (Gunnersen et al., 1996). Site-directed mutagenesis was performed with QuickChange mutagenesis kit (Stratagene, La Jolla, CA). Presence of the desired mutations was verified by direct sequencing. To verify the absence of new, unwanted substitutions, the complete coding regions were sequenced for each mutant. In case of ₅I215F mutant, the plasmid resulting from the QuickChange mutagenesis reaction was digested with PflMI endonuclease, and the fragment containing the desired I215F substitution was gel-purified and ligated into the similarly digested wild-type pRc/CMV/a5 vector. Human embryonic kidney 293 cells (American Type Culture Collection, Manassas, VA) were maintained at 37°C in 5% CO₂ as previously described (Gunnersen et al., 1996). Cells were transfected with equal amounts (5 µg of each DNA/90-mm dish) by calcium phosphate precipitation as described previously (Gorman et al., 1990). The cells were harvested 48 h after transfection, by washing with ice-cold phosphate-buffered saline and centrifuged at 1000g. Cells were washed three times by homogenization in ice-cold 50 mM Tris-citrate buffer, pH 7.8, and centrifuged at 20000g. These membrane suspensions were stored at 70°C until needed.

Radioligand Binding. Incubations were performed in a final volume of 600 to 1000 µl and contained resuspended cell membranes (~0.02-0.1 mg of protein), 0.2 M NaCl, [³H]Ro 15-1788, or [³H]RY-80 (87 and 55.4 Ci/mmol, respectively; DuPont-New England Nuclear, Boston, MA), and 50 mM Tris-citrate buffer, pH 7.8, to volume. In competition experiments, 50 µl of buffer was replaced by drugs. [³H]Ro 15-1788 was used at concentrations equal to its K_D values at the respective receptor subtype. Nonspecific binding was determined with Ro 15-1788 (10 µM). [³H]Muscimol binding was determined using a membrane suspension (~0.02-0.1 mg of protein), [³H]muscimol (20 Ci/mmol; DuPont-New England Nuclear), and 50 mM Tris-citrate buffer, pH 7.8, to volume. Nonspecific binding in this case was determined in presence of 1 mM GABA. Assays were incubated at 4°C for 2 h and terminated by rapid filtration (Brandel M-48R, Gaithersburg, MD) through GF/B filters followed by two 5-ml washes with ice-cold Tris-citrate buffer. The filter-retained radioactivity was determined by liquid scintillation counting. Data were analyzed with GraphPad Prism software (GraphPad Software Inc., San Diego, CA), and K_i values were calculated from the equation, K_i = IC₅₀/(1 + [radioligand]/K_D). Both K_i and K_D values were calculated from at least three independent experiments performed in duplicate. Statistical significance was determined using a one-way ANOVA followed by a Dunnett's multiple comparison post hoc test. Protein concentrations were determined using the bicinchoninic acid protein assay kit (Pierce, Rockford, IL). RY-24 and RY-80 were synthesized at the University of Wisconsin-Milwaukee, CL 218,872 was obtained from Lederle Laboratories (Mont-St-Guibert, Belgium), and zolpidem was obtained from Synthelabo (Laboratoire Experimental Recherche Synthelabo, Paris, France). Flunitrazepam was purchased from Research Biochemicals International (Natick, MA). The structures of Ro 15-1788 and related ₅-selective benzodiazepines are given in Fig. 1. All other reagents and chemicals were from standard commercial sources.

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Structures of imidazobenzodiazepines with selectivity for 5-containing GABAA receptors: Comparison with Ro 15-1788.

Expression in Xenopus laevis Oocytes. X. laevis frogs were purchased from Xenopus-1 (Dexter, MI). Collagenase B was from Boehringer Mannheim (Indianapolis, IN). All other compounds were from Sigma Chemical Co. Capped cRNA was synthesized from linearized template cDNA encoding the subunits using mMESSAGE mMACHINE kits (Ambion, Austin, TX). Oocytes were injected with cRNAs encoding the specified ₅ subunit variants along with the ₃ and ₂ subunits in a ratio of 1:1:1 as determined by gel electrophoresis. Mature X. laevis frogs were anesthetized by submersion in 0.1% 3-aminobenzoic acid ethyl ester, and oocytes were surgically removed. Follicle cells were removed by treatment with collagenase B for 2 h. Each oocyte was injected with 5 to 25 ng of cRNA in 50 nl of water and incubated at 19°C in modified Barth's saline (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 0.41 mM CaCl₂, 0.82 mM MgSO₄, 100 µg/ml gentamicin, and 15 mM HEPES, pH 7.6). Recordings were performed 1 to 7 days post injection.

Electrophysiological Recordings Oocytes were perfused at room temperature in a Warner Instruments oocyte recording chamber #RC-5/18 (Hamden, CT) with perfusion solution (115 mM NaCl, 1.8 mM CaCl₂, 2.5 mM KCl, 10 mM HEPES, pH 7.2). Perfusion solution was gravity fed continuously at a rate of 15 ml/min. GABA was dissolved in the perfusion solution. Drugs were added as a 10 mM solution in ethanol to the perfusion solution to achieve the appropriate concentration.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Amino Acid Ile215 on the ₅ Subunit Contributes to Ligand Selectivity at ₅₃₂ Receptors. In an attempt to define amino acid residues on the ₅ subunit that are important for high affinity and selectivity to compounds such as RY-80, we considered amino acid residues that are conserved among all other DS -subunits (Fig. 2). Based on this sequence comparison, four residues in the ₅ subunit N-terminal domain were selected for the initial analysis as the most likely to be involved in defining ligand selectivity. The corresponding amino acids on the other DS -subunits were substituted on the ₅ subunit, yielding ₅G24R, ₅P166T, ₅H196D, and ₅I215V variants, respectively. Wild-type and mutant ₅₃₂ receptors were transiently expressed in the human embryonic kidney 293 cells. No additional mutations were found after sequencing the complete coding regions for each mutant. The screening strategy applied to identify amino acid(s) important for ligand selectivity at ₅ subunit was based on the premise that at the concentrations approximating the K_D value of each ligand the binding of [³H]Ro 15-1788 and [³H]RY-80 to the wild-type ₅₃₂ receptor will be similar, yielding a ratio of ~1. If a particular amino acid substitution introduced in the ₅ subunit altered [³H]RY-80 binding, this ratio would change. Among the amino acid substitutions tested, only ₅I215V yielded a dramatically different ratio (17.8) than the value obtained (1.04) in wild-type ₅₃₂ receptors (data not shown). Based on this observation, the mutant receptor ₅I215V₃₂ was chosen for further analysis. Saturation analysis revealed a ~16-fold decrease in affinity for [³H]RY-80 in ₅I215V₃₂ receptor (5.6 ± 0.6 nM) compared with wild-type receptor (0.35 ± 0.02 nM) (Fig. 3A). Consistent with these data, the affinities of the ₅-selective compounds RY-80 and RY-24 in mutant receptors were reduced ~20- and ~10-fold, respectively, in competition studies using [³H]Ro 15-1788 (Fig. 3B; Table 1). In contrast, the affinity of the nonselective ligand [³H]Ro 15-1788 was not significantly affected by this mutation (Table 1; Fig. 3B).

View larger version (65K): [in this window] [in a new window]   Fig. 2.   Alignment of the rat GABAA 1, 2, 3, and 5 subunits. The sequences shown represent N-terminal regions of the corresponding -subunits up to the first putative transmembrane domain (TM1). Arrows: amino acids chosen for mutagenesis.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Reduced affinities of 5-selective ligands at 5I215V32 receptor. A, binding of [3H]RY-80 to wild-type 532 () and 5I215V32 () receptors. These are representative isotherms with KD values for [3H]RY-80 of 0.38 nM (Bmax = 1008 fmol/mg of protein) and 6.5 nM (Bmax = 2419 fmol/mg of protein) for the 532 and 5I215V32 receptors, respectively. Because of differences in receptor density and/or transfection efficiency, the quantity of bound [3H]RY-80 was normalized to the estimated Bmax values. KD and Bmax values were calculated by nonlinear least-squares fit of specifically bound [3H]RY-80 (under Materials and Methods). B, displacement of [3H]Ro 15-1788 binding with RY-24 from wild-type 532 () and 5I215V32 () receptors. [3H]Ro 15-1788 was used at concentrations equal to its KD value for each receptor. Ki values were calculated using the equation of Cheng and Prusoff (under Materials and Methods). Ki values for RY-24 were equal to 0.71 and 4.6 nM for 532 and 5I215V32 receptors, respectively. Representative curves are shown. Also shown are representative [3H]Ro 15-1788 saturation isotherms for wild-type 532 () and 5I215V32 () receptors. KD values for [3H]Ro 15-1788 were 0.34 and 0.57 nM in wild-type and mutant receptors, respectively. The data from [3H]Ro 15-1788 saturation isotherms were transformed as (1  B/Bmax) × 100% for ease of comparison.

View this table: [in this window] [in a new window]   TABLE 1 Binding properties of the wild-type and mutant 132 and 532 receptors KD values for [3H]Ro 15-1788 and [3H]RY-80 were determined in saturation experiments. Ki values were derived from the displacement of [3H]Ro 15-1788 at a concentration equivalent to its KD value at each receptor subtype. Values are mean ± S.E.M. for at least three experiments performed in duplicate.

Properties of ₁V211I₃₂ Receptor. The dramatic effect produced by substitution of a valine in position ₅215 on the affinities of the ₅-selective ligands prompted an examination of the effect of a cognate substitution (isoleucine for valine) on position ₁211 (corresponding to ₅215). Consistent with the previous reports (Liu et al., 1996) both RY-24 and RY-80 exhibited low affinities (56 ± 17 and 47 ± 8 nM, respectively) for wild-type ₁₃₂ receptors (Table 1). This exchange (₁V211I₃₂) increased the affinities of RY-24 and RY-80 by more than one order of magnitude (to 3.3 ± 0.3 and 3.7 ± 0.3 nM, respectively) (Table 1). Moreover, this mutation decreased the affinities of the ₁-selective ligands zolpidem and CL 218,872 by 3-fold without affecting the affinity of [³H]Ro 15-1788 (Table 1; Fig. 3).

Properties of Mutant ₅₃₂ Receptors with Lipophilic Amino Acid Substitutions in Position ₅215. Based on the hypothesis that interaction with a lipophilic pocket is required for ligand selectivity at the ₅-containing GABA_A receptors (Liu et al., 1996), isoleucine in position 215 of the wild-type ₅ subunit was exchanged with alanine, leucine, or phenylalanine. The ligand binding properties of the ₅I215A₃₂ receptor were similar to those of the ₅I215V receptor. The alanine substitution decreased (by >10-fold) the affinities of both RY-24 and RY-80 without altering the affinity of [³H]Ro 15-1788. In contrast to the valine substitution, introduction of alanine in position ₅215 increased the affinity of CL 218,872 by 10-fold (Table 1). Substitution of either leucine or the more lipophilic phenylalanine resulted in no significant change in the affinities of RY-24 and RY-80. Furthermore, substitution ₅I215F resulted in a slightly reduced affinity of Ro 15-1788 and a lower affinity of CL 218,872. The affinity of flunitrazepam was unchanged (compared with wild-type receptors) for both ₅I215A₃₂ and ₅I215L₃₂ receptors (K_i values of 0.6 ± 0.1 and 0.9 ± 0.2 nM, respectively), whereas the affinity for flunitrazepam at the ₅I215F₃₂ receptors was decreased ~7-fold (K_i = 6.9 ± 1.8 nM). A decrease in the affinity of ₅-selective ligands at ₅I215A₃₂ and ₅I215V₃₂ receptors prompted us to further reduce the size of the side chain of the residue ₅215. However, introduction of glycine resulted in levels of [³H]Ro 15-1788, [³H]RY-80, or [³H]muscimol binding that were barely detectable.

Properties of Mutant ₅₃₂ Receptors with Charged or Polar Amino Acid Residues in Position ₅215. Substitution of the negatively charged aspartate residue at position 215 produced a modest decrease in affinity of [³H]Ro 15-1788 (1.7 ± 0.3 nM compared with 0.36 ± 0.04 nM for the wild-type receptor), and a similar, modest decrease in the affinities of RY-80 and RY-24 binding (Table 1). Substitution of a threonine (a more hydrophilic amino acid) for isoleucine yielded a receptor with properties similar to ₅I215V₃₂ receptor. This receptor produced a 10-fold decrease in affinities of both RY-24 and RY-80 without significantly affecting [³H]Ro 15-1788 binding. However, isoleucine-to-threonine substitution resulted in a small (3-fold) increase in the affinity of flunitrazepam (Table 1). Substitution of a basic lysine residue for isoleucine produced a 5- to 6-fold decrease in the affinities of both RY-24 and RY-80 without affecting the affinity of [³H]Ro 15-1788. The affinity of flunitrazepam also was not substantially changed (Table 1). Additionally, ₅I215D₃₂, ₅I215T₃₂, and ₅I215K₃₂ receptors displayed an increase in affinity of CL 218,872 (4-fold) compared with wild-type receptors; however, none of the amino acid substitutions in position ₅215 yielded a receptor variant with any measurable affinity for zolpidem.

Efficacy of RY-24 and RY-80 at Wild-Type and Mutant ₅₃₂ Receptors. Mutation of a conserved histidine residue in the N-terminal domain of all DS -subunits (₁H101R, ₂H101R, ₃H126R, and ₅H105R) to arginine not only confers diazepam insensitivity to the respective _x2/32 receptors but also alters the efficacies of several ligands at these receptors (Benson et al., 1998). Based on these observations, the potential role of ₅215 in modulating ligand efficacy was examined. Three mutant receptors, ₅I215V₃₂, ₅I215K₃₂, and ₅I215T₃₂ were examined. Introduction of either valine, lysine, or threonine in position ₅215 did not change the potency of GABA at these receptor subtypes (EC₅₀ = ~30 µM for all receptors tested; see Table 2, legend).

View this table: [in this window] [in a new window]   TABLE 2 Mutation of residue 5215 changes efficacy of benzodiazepine-site ligands Wild-type or mutant GABAA receptors were expressed in X. laevis oocytes and electrophysiological recordings were performed as described under Materials and Methods. Responses to concentrations of GABA in presence of drugs are reported as a percentage of the response to GABA alone (% control). GABA was applied at concentration 30 µM, approximating its EC50 value at each receptor subtype tested (EC50 = 25.8 ± 7.8 µM). RY-80, RY-24, and flunitrazepam were used at a saturating concentration of 1 µM. Values are means ± S.D. for at least three oocytes.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The objective of the present study was to localize the molecular features of the ₅ subunit responsible for the high affinity and selectivity of ligands such as RY-80. Because the N-terminal extracellular domain exhibits the greatest sequence divergence among -subunits, it was hypothesized that this region was most likely to be involved in defining ligand selectivity. Four amino acid residues conserved in this region among the _1-3 subunits but different in the ₅ subunit (₅G24, ₅P166, ₅H195, and ₅I215) were considered. Substitution of each of these four residues in the ₅ subunit with the corresponding amino acids conserved among the _1-3 subunits resulted in a significant reduction in [³H]RY-80 binding only in the ₅I215V₃₂ mutant (under Results). Saturation analysis confirmed that this reduction in [³H]RY-80 binding reflects an ~16-fold increase in the K_D value of this radioligand (Table 1; Fig. 3) compared with wild-type ₅I215V₃₂ receptors. This mutation concomitantly reduced the selectivity of RY-80 for GABA_A receptors containing an ₅ subunit from ~134- to ~8.4-fold compared with cognate receptors containing an ₁ subunit. This mutation also increased the K_i of RY-24 by >6.0-fold and reduced its selectivity for ₅-containing GABA_A receptors from ~80- to ~12-fold (Table 1; Fig. 3). Because all known ₅-selective ligands are structurally related (Fig. 1), it is not known whether the affinities of other, structurally unrelated compounds exhibiting ₅-subtype selectivity would be similarly affected. However, the observation that the affinity of Ro 15-1788 was not significantly altered in the ₅I215V₃₂ mutant and that the affinity of CL 218,872 was slightly increased supports the hypothesis that this amino acid is essential for a selective interaction at ₅₃₂ receptors. We hypothesized that if Ile215 is essential for ligand selectivity at ₅₃₂ receptors, then substitution of this residue at this corresponding position in ₁₃₂ receptors (i.e., at Val211) should produce a significant increase in the affinity of compounds such as RY-80. Consistent with this hypothesis, the affinities of both RY-80 and RY-24 were increased ~20-fold in ₁V211I₃₂, whereas the affinities of other ligands were either unchanged or slightly reduced (Fig. 4).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Increased affinities of 5-selective ligands at 1V211I32 receptor. A, binding of [3H] RY-80 to wild-type 132 () and 1V211I32 () receptors. These are representative isotherms with KD values for [3H] RY-80 of 48.4 nM (Bmax = 353 fmol/mg of protein) and 2.3 nM (Bmax = 1453 fmol/mg of protein) for the 132 and 1V211I32 receptors, respectively. Because of differences in receptor density and/or transfection efficiency the quantity of bound [3H]RY-80 was normalized to the estimated Bmax values. KD and Bmax values were calculated by nonlinear least-squares fit of specifically bound [3H]RY-80 (under Materials and Methods). B, displacement of [3H]Ro 15-1788 binding with RY-24 from wild-type 132 () and 1V211I32 () receptors. [3H]Ro 15-1788 was used at concentrations equal to its KD value at each receptor. Ki values were calculated using the equation of Cheng and Prusoff (under Materials and Methods). Ki values for RY-24 were equal to 55.9 and 3.3 nM for 132 and 1V211I32 receptors, respectively. Representative curves are shown. Also shown are representative [3H]Ro 15-1788 saturation isotherms for wild-type 132 () and 1V211I32 () receptors. KD values for [3H]Ro 15-1788 were 1.2 and 0.82 nM in wild-type and mutant receptors, respectively. The data from [3H]Ro 15-1788 saturation isotherms were transformed as (1  B/Bmax) × 100% for ease of comparison.

Based on the affinities of a structurally diverse group of ligands, an inclusive pharmacophore of the ₅₂₂ receptor has been developed (Liu et al., 1996). A large lipophilic regionl (L₂) appears to contribute to the unique pharmacological properties of this GABA_A receptor isoform (Liu et al., 1996). Thus, although the L₂ descriptor is common to the pharmacophore of other GABA_A receptors (e.g., ₁₂₂ receptors), the larger volume of L₂ in ₅₂₂ receptors has been proposed to result in low affinities for ligands (e.g., zolipidem) that do not extend into this domain, and very high affinities for compounds (e.g., RY-80) capable of filling this area (Liu et al., 1996). Both the >10-fold reduction in the affinities of ₅-selective ligands produced by a subtle change in the lipophilicity of residue 215 (i.e., isoleucine-to-valine) and the corresponding increases in affinity produced by the cognate (i.e., "back") mutation in ₁₃₂ receptors prompted us to hypothesize that Ile215 constitutes a portion of this L₂ domain. If this hypothesis is correct, then increasing the lipophilicity of the residue at ₅215 should increase the affinities of ligands such as RY-80 and RY-24 without remarkably affecting the affinities of nonselective ligands. Conversely, reducing lipophilicity, either by substituting a nonpolar amino acid with a smaller side chain or introduction of polar or charged residues at ₅215 should produce a further reduction in the affinities of these compounds. Increasing the lipophilicity of this residue (e.g., leucine, phenylalanine) did not remarkably affect the affinities of RY-80 and RY-24. In contrast, 5- to 6-fold decreases in affinities of nonspecific ligands Ro 15-1788 and flunitrazepam were observed (Table 1). However, neither reducing lipophilicity by introducing an alanine residue nor substitution of polar and/or charged residues (e.g., threonine, lysine, and aspartic acid) decreased the affinities of the ₅-selective ligands beyond that produced by the original ₅I215V mutation (Table 1). These observations indicate that although contributing to the high affinity and selectivity of RY-80 and RY-24 at ₅₃₂ receptors, residue 215 may not directly participate in the formation of L₂. Thus, the original model, based on an L₂ lipophilic region exerting influence over ligand selectivity at ₅-containing GABA_A receptors merits reconsideration. Furthermore, in the absence of crystallographic studies of ligand-bound receptor (Dingledine et al., 1999), it is possible that ₅215 modulates the affinities of RY-24 and RY-80 through an allosteric mechanism rather than as an integral part of the binding pocket.

Although the present findings demonstrate that ₅Ile215 substantially contributes to ligand selectivity at ₅₃₂ receptors, it cannot be the sole determinant of the unique pharmacological profile of this receptor isoform. Thus, RY-80 and RY-24 retain modest selectivities (~8-14-fold) for ₅I215V, ₅I215A, and ₅I215T₃₂ receptors compared with ₁₃₂ receptors. Moreover, despite a ~20-fold increase in the affinities of RY-80 and RY-24 for ₁₃₂ receptors containing a back mutation (i.e., ₁V211I₃₂), these compounds remain significantly (~5-fold) more potent in wild-type ₅₃₂ receptors. Finally, the very low affinity of zolpidem at wild-type ₅-containing receptors is maintained through a range of mutations at this residue. Other likely candidates contributing to the pharmacological profile of ₅₃₂ receptors (i.e., selectivity for RY-80 and RY-24) are one or more of the other N terminus amino acid residues that differ between the ₅ and _1-3 subunits, as well as residues on the -subunit that may act in concert with ₅I215 to produce a unique pharmacology. This latter hypothesis is consistent with both the dramatic reduction in the affinity of zolpidem produced by substitution of a ₃ for a ₂ subunit in recombinant ₁-containing GABA_A receptors (Lüddens et al., 1994), and the absolute requirement for a -subunit for high-affinity binding of benzodiazepine-site ligands (Pritchett et al., 1989; Wong et al., 1992; Boileau et al., 1998).

The -subunit has been closely linked to the efficacy of benzodiazepine-site ligands (Knoflach et al., 1991; Puia et al., 1991). Nonetheless, a series of conservative mutations (₁H101R, ₂H101R, ₃H126R, and ₅H105R) that imparts diazepam insensitivity to the corresponding _x2/32 receptors (Wieland et al., 1992; Benson et al., 1998) were recently shown to affect the efficacy of several benzodiazepine-site ligands (Benson et al., 1998). This finding prompted us to examine the role of ₅215 in controlling ligand efficacy at recombinant ₅₃₂ receptors. Three mutations (₅I215V, ₅I215T, and ₅I215K) were chosen for study based on their structural divergence from the residue present in wild-type receptors and the reduced affinities of ₅-selective agents. Introduction of valine, lysine, or threonine in position ₅215 does not affect the potency of GABA compared with wild-type receptors. These mutations did not alter the ability of flunitrazepam to act as an agonist, increasing currents evoked by subsaturating concentrations of GABA (Fig. 5; Table 2). In contrast, the characteristic ability of RY-80 and RY-24 to reduce GABA-gated currents in wild-type ₅₃₂ receptors (Fig. 5; Table 2; Liu et al., 1996)) was abolished in ₅I215V₃₂ and ₅I215T₃₂ receptors, but retained in the ₅I215K₃₂ mutants. The failure to observe a change in the efficacies of RY-24 and RY-80 in the ₅I215K mutants indicates this effect on ligand efficacy is independent of changes in ligand affinity because all three mutations reduced (albeit to different degrees) the affinities of these ₅-selective ligands. These data demonstrate that in addition to a well described role in defining the affinities of benzodiazepine-site ligands, the -subunit can also impact ligand efficacy. The identification of amino acid residues contributing to ligand selectivity at GABA_A receptor isoforms may provide insights resulting in compounds with a more limited spectrum of action than traditional 1,4-benzodiazepines.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Representative voltage-clamp trace recordings from X. laevis oocytes injected with the wild-type or mutant 532 constructs 532 (left), 5I215V32 (center), and 5I215K32 (right). GABA (30 µM) was perfused over oocyte for the duration indicated by the black bar. GABA plus saturating concentrations of compound (1 µM) were perfused over oocyte for the duration indicated by the gray bar. Oocytes were voltage-clamped at 60 mV. Scale bars, 50 nA/10 s. Note that flunitrazepam potentiates currents at all receptor subtypes tested, whereas RY-80 changes its mode of action from being an inverse agonist at the wild-type 532 and mutant 5I215K32 receptors to being an antagonist at the 5I215V32 receptor.