Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

-Aminobutyric acid (GABA) receptors are the major inhibitory neurotransmitter receptors in the mammalian brain. The native receptor is likely to be a heteropentameric protein (Nayeem et al., 1994), assembled from multiple subunit subtypes: 6 , 4 , 3 , 1 , 1 , 3 , and 1  (Barnard et al., 1998). The receptor contains an integral chloride-selective channel with specific binding sites for GABA and a variety of neuroactive drugs, including benzodiazepines (BZDs), barbiturates, neurosteroids, and anesthetics (Sieghart, 1995; Smith and Olsen, 1995). BZDs, clinically used for their anxiolytic, muscle relaxant, sedative, and antiepileptic actions, exert their therapeutic effects by allosterically modulating the activation of the GABA_A receptor. Because of their clinical usefulness, a substantial effort has been made to understand the structural determinants of BZD binding in this receptor.

A variety of structurally diverse ligands bind with high affinity to the BZD binding site. These compounds include classic benzodiazepines, triazolopyridazines, imidazopyridines, cyclopyrrolones, pyrazoloquinolinones, and -carbolines (Barnard et al., 1998). Depending on the ligand and the subunit composition of the GABA_A receptor, the modulatory actions of these compounds range from full agonist (positive modulator) to inverse agonist (negative modulator). BZD positive modulators decrease the GABA concentration needed to elicit half-maximal channel activity (EC₅₀), whereas BZD negative modulators increase the GABA EC₅₀ value. BZD antagonists block the effect of both positive and negative modulators. Although all BZD binding site ligands appear to compete for a common binding site (McKernan et al., 1998), it is likely that different microdomains within the site interact with different subsets of BZD ligands (Davies et al., 1996).

The BZD binding site of the GABA_A receptor has been proposed to lie at the interface between the - and -subunits, with residues from each subunit contributing to the binding site (Smith and Olsen, 1995; Sigel and Buhr, 1997). In the 1 subunit, photoaffinity-labeling (Duncalfe et al., 1996) and mutagenesis (Wieland et al., 1992; Davies et al., 1998) experiments have identified histidine at position 101 (1 H101) as forming part of the BZD binding site. Other 1 residues implicated in BZD binding include Tyr-159, Thr-162, Gly-200, Thr-206, Tyr-209, and Val-211 (Pritchett and Seeburg, 1991; Wieland and Luddens, 1994; Amin et al., 1997; Buhr et al., 1997b). In the 2 subunit, only two residues have been identified as key determinants for BZD binding: 2 Phe-77 (Buhr et al., 1997a; Wingrove et al., 1997) and 2 Met-130 (Buhr and Sigel, 1997; Wingrove et al., 1997).

Previously, by using 2/1 chimeric subunits, we identified two domains of the 2 subunit, Lys-41-Trp-82 and Arg-114-Asp-161, that together are necessary for high-affinity BZD binding (Boileau et al., 1998). In this study, by using / chimeric subunits and 2 point mutations, we focused on identifying residues within the 2 Lys-41-Trp-82 region that contribute to BZD binding. We identify three novel residues in the 2 subunit, 2 Met-57, Tyr-58, and Ala-79, that are important determinants for high-affinity BZD binding.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chimera Nomenclature. All 2/1chimeric constructs in this study contain 2 amino acids from Arg-114 to Asp-161 because this region was found to be necessary for BZD binding (Boileau et al., 1998). For ease of reading, the chimeric constructs () are named for the 2 residue before the junction of the first 2/1 crossover in the mature rat protein sequence (Fig. 1). For example, 40 contains 2 sequence from Gln-1 to Asn-40 and from Arg-114 to Asp-161, whereas 82 contains 2 sequence from Gln-1 to Trp-82 and from Arg-114 to Asp-161. Mutations produced in chimeric backgrounds were named for the 1 residue mutated, the aligned position in the mature 2 subunit, and the 2 residue introduced. For instance, 56 V76I denotes that the 1 residue (Val) in the 56 subunit was mutated to the aligned residue (Ile) at position 76 of the mature 2 sequence. Mutations produced in the 2 subunit were named for the targeted 2 residue, the position in the mature 2 subunit, and the mutant amino acid. For example, 2 A79C denotes that the Ala at position 79 in 2 was mutated to Cys.

View larger version (25K): [in this window] [in a new window]   Fig. 1.   Constructed chimeric subunits and mutations used in the identification of 2 residues important for BZD binding. 2/1 chimeras () are named for the 2 residue before the junction of the first 2/1 crossover. All s contain additional amino acid residues from 2 Arg-114 to Asp-161. Therefore, 40 contains the 2 sequence from Gln-1 to Asn-40 and from Arg-114 to Asp-161 (see Materials and Methods). The 2 sequence is shown in gray, the 1 sequence is shown in white, and the transmembrane segments M1 through M4 are shown in black. 1282 receptors specifically bind BZDs, whereas 1240 receptors do not (Boileau et al., 1998). 1 and 2 sequence homology in the region from 2 Lys-41 to Trp-82 is shown in expanded form. Identical amino acid residues are shown in light gray. Residues indicated with an asterisk were targeted for mutation. Vertical lines indicate crossover transitions for 56 and 65. Residues that were found to influence BZD binding are boxed.

Molecular Cloning. Rat cDNA clones for the 1, 2, and 2 GABA_A receptor subunits were used for all molecular cloning. 40 and 82 were produced as previously described (Boileau et al., 1998). Point mutations were made in 40 using either the MORPH Site-Specific Plasmid DNA Mutagenesis kit (5 Prime-3 Prime, Boulder, CO) or a modified form of recombinant polymerase chain reaction (PCR). In this method, a forward mutagenic oligonucleotide was paired with a reverse template-specific oligonucleotide and amplified by PCR using an appropriate template. The resulting product was purified to remove excess oligonucleotides using the High Pure PCR Product Purification kit (Boehringer-Mannheim Biochemicals, Indianapolis, IN). Using the same template, the purified primary product, now acting as a reverse primer, was paired with an upstream vector-specific oligonucleotide and amplified. After the secondary amplification, the final product was purified as earlier and subcloned into the appropriate background.

Transient Expression in HEK 293 Cells. Cells were grown in 100-mm tissue culture dishes in minimum essential medium with Earle's salts (Life Technologies, Inc., Gaithersburg, MD) containing 10% fetal bovine serum (Sigma-Aldrich, St. Louis, MO) in a 37°C incubator under a 5% CO₂ atmosphere. Cells were cotransfected at 50 to 60% confluency with 1-pCEP4, 2-pCEP4, and either 2-pCEP4 or -pCEP4 using a standard CaHPO₄ precipitation method (Graham and van der Eb, 1973). The vector pAdVAntage (Promega, Madison, WI) was added to enhance expression levels (4 µg of each subunit and 12 µg of pAdVAntage/100-mm plate). Cells were harvested and membrane homogenates were prepared as previously described (Boileau et al., 1998).

Binding Assays. Competition binding experiments with various BZD-site ligands were performed as previously described (Boileau et al., 1998). In brief, membrane homogenates (100 µg) were incubated at room temperature with [³H]flunitrazepam (85 Ci/mmol; DuPont-New England Nuclear, Boston, MA) or [³H]Ro15-4513 (21.7 Ci/mmol; DuPont-New England Nuclear) at a concentration slightly lower than K_I and 7 to 10 concentrations of unlabeled competing ligand in a final volume of 250 µl. The unlabeled BZDs, flunitrazepam, Ro15-1788, and Ro15-4513 were generously supplied by Dr. Sepinwall (Hoffman-La Roche, Nutley, NJ). Data were fit by using the equation y = B_max/[1 + (x/IC₅₀)], where y is the specifically bound dpm, B_max is the maximal binding, and x is the concentration of displacing drug (Prism; GraphPad Software, San Diego, CA). K_I was calculated according to the Cheng-Prusoff/Chou equation (Cheng and Prusoff, 1973; Chou, 1974).

Immunofluorescence. 40, 56, 82, and 2 were tagged between the third and fourth residues of the mature subunit with the myc 9E10 epitope (EQKLISEEDL) using recombinant PCR and subcloned back into each respective template. The myc epitope tag had no effect on the function or expression of the subunits. Cells were grown and transfected in 12-well dishes on poly(D)-lysine (Sigma-Aldrich)-coated 12-mm glass coverslips. Forty-eight hours after the transfection, cells were washed and fixed in 2% paraformaldehyde. Nonspecific immunoreactivity was reduced by blocking cells with 2% BSA in PBS containing: 2.7 mM KCl, 1.5 mM KH₂PO₄, 0.5 mM MgCl₂, 137 mM NaCl, and 14 mM Na₂HPO₄, pH 7.1. Antibodies were diluted in the corresponding blocking buffer. In some cases, the cells were permeabilized using PBS plus 0.1% Triton X-100 before the antibody addition. The primary antibody was an anti-myc 9E10 antibody, generously supplied by Dr. Johannes Hell (University of Wisconsin-Madison), diluted at 1:500; the secondary antibody, biotin-SP goat anti-mouse IgG (Jackson ImmunoResearch, West Grove, PA), was diluted to 4.4 µg/ml. The final incubation was in Texas Red-conjugated streptavidin (Jackson ImmunoResearch), diluted to 4.2 µg/ml. After several washes, the coverslips were mounted onto slides and visualized. Fluorescent images of cells were acquired with a Zeiss 35 M inverted microscope (Carl Zeiss, Thornwood, NY), 63×/1.4 NA Plan-Apochromatic objective, Texas Red filter set (Chroma Technology, Brattleboro, VT) and a Princeton Instruments MicroMax cooled CCD digital camera (Princeton Instruments, Trenton, NJ). All images were acquired at full chip resolution (Kodak KAF-1400 chip, 1035 × 1317 pixels, 6.8-µm pixel size) within the dynamic range of the camera (12-bit, 4096 gray levels) using Metamorph 4.1 imaging software (Universal Imaging, West Chester, PA). Images were scaled appropriately, converted to 8-bit images, and imported into Adobe Photoshop (ADOBE Systems, Mountview, CA).

Statistical Analysis. We compared the effects of the mutations with the use of one-way ANOVA, applying Dunnett's post-test for significance of differences (Prism; GraphPad Software).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of 2 Residues Essential for Flunitrazepam Binding. The focus of the current study was to identify specific amino acid residues within the 2 Lys-41-Trp-82 region that contribute to BZD binding. Previously, by comparing a 2/1 chimera that bound BZDs with high affinity (82) with one that did not (40), we determined that the 2 Lys-41-Trp-82 region was important for BZD binding (Boileau et al., 1998). An amino acid alignment of the 2 Lys-41-Trp-82 region with the equivalent 1 region (Arg-28-Trp-69) reveals 22 identical residues, 8 conservative substitutions, and 12 nonconservative substitutions (Fig. 1). In an attempt to identify single 2 amino acid residues that contribute to BZD binding, 14 of the 20 nonidentical 1 residues in 40 were individually mutated to the aligned 2 residues (Fig. 1, asterisks). When the 40 mutants were expressed with wild-type 1 and 2 subunits in HEK 293 cells, no specific [³H]flunitrazepam binding was detected (results not shown). Because no single 1-to-2 amino acid residue substitution tested restored high-affinity binding, it is likely that more than one 2 residue in the Lys-41-to-Trp-82 region is important for [³H]flunitrazepam binding.

View larger version (12K): [in this window] [in a new window]   Fig. 2.   2 Met-57 and Tyr-58 are essential for high-affinity flunitrazepam binding. [3H]Flunitrazepam affinity was measured by homologous competition radioligand binding assays on membranes prepared from HEK 293 cells transfected with , 56, 65, 56-I57M/F58Y, and 56-I57M receptors. 56 data were fit by linear regression analysis, and the total binding of [3H]flunitrazepam was normalized to 1.0 for this receptor combination. The slope did not significantly deviate from zero. Data for the other receptors were normalized to specific binding and fit by nonlinear regression analysis. Data shown are single representative experiments; points are mean ± S.E. of triplicate determinations. KI values are summarized in Table 1.

View this table: [in this window] [in a new window]   TABLE 1 Binding affinities of BZD site ligands for wild-type, chimeric, and mutant chimeric GABAA receptors KI values were determined by displacement of [3H]flunitrazepam binding. Results shown are the mean ± S.E.; n is the number of independent experiments. Statistical differences between log KI values were determined by one-way ANOVA using Dunnett's post test. The flunitrazepam affinity for 56-I57M was not included in the statistical analysis due to n = 2, however, the trend suggests that it is significantly different from 65.

Surface Expression of ₁₂ GABA_A Receptors. Although the ability of 2 Met-57 and 2 Tyr-58 to restore high-affinity flunitrazepam binding to 1256 receptors may indicate that these residues are directly involved in flunitrazepam binding, it is also possible that these residues are required for the proper assembly and cell surface expression of 1256 receptors. To distinguish between these possibilities, we examined the assembly and cell surface expression of the 2/1 chimeric subunits through the use of immunohistochemistry. Because nonpentameric GABA_A receptors and unassembled 1, 2, and 2 subunits are not expressed on the cell surface (Connolly et al., 1996; Tretter et al., 1997), cell surface expression of the 2/1 chimeric subunits provides strong evidence that they are assembling into mature, pentameric receptors.

View larger version (48K): [in this window] [in a new window]   Fig. 3.   Immunofluorescence of GABAA receptors expressed in HEK 293 cells demonstrates that chimeric subunits reach the cell surface. Cells transfected with 1, 2, and myc-tagged 2 or myc-tagged 2/1 chimeric subunits were either surface labeled () with an anti-myc 9E10 antibody or permeabilized with Triton X-100 before labeling (+) as described in Materials and Methods.

Ro15-4513 and Ro15-1788 Binding to ₁₂ GABA_A Receptors. The apparent affinities (K_I values) of 1265, 1282, and 122 receptors for [³H]flunitrazepam (BZD agonist), Ro15-1788 (BZD antagonist), and Ro15-4513 (BZD inverse agonist) were measured and compared. The affinity of 1265 receptors for [³H]flunitrazepam was not significantly different from that of 1282 receptors and was only 2-fold lower than that of 122 receptors (Fig. 4, Table 1). In contrast, the affinities of 1265 receptors for Ro15-4513 and Ro15-1788 were significantly lower than those of 1282 receptors (approximately 3-fold) and were 25- and 10-fold lower than those of the wild-type receptors, respectively (Fig. 4, Table 1). These results suggest that amino acid residues within the 2 Asn-66-Trp-82 region preferentially influence Ro15-4513 and Ro15-1788 binding.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Displacement of [3H]flunitrazepam binding from membranes prepared from HEK 293 cells transfected with (), 82 (), and 65 (). Competition of [3H]flunitrazepam binding by flunitrazepam (top), Ro15-4513 (middle), and Ro15-1788 (bottom) is shown. Data are from single representative experiments; points are mean ± S.E. of triplicate determinations. Data were fit by nonlinear regression analysis. The KI values are summarized in Table 1.

Analysis of Point Mutations at 2 Ala-79 and Thr-81. To confirm the importance of 2 Ala-79 and 2 Thr-81 for Ro15-4513 binding, a series of point mutations were made directly in the wild-type 2 subunit. 2 Ala-79 was mutated to arginine (2A79R), cysteine (2A79C), tyrosine (2A79Y), and glutamine (2A79Q), whereas 2 Thr-81 was mutated to serine (2T81S), cysteine (2T81C), and alanine (2T81A). The mutant 2 subunits were expressed with wild-type 1 and 2 subunits in HEK 293 cells, and the binding of flunitrazepam, Ro15-4513, and Ro15-1788 was measured and compared. Individual mutations altered the affinities of the three drugs by different amounts. For example, the 2A79Y mutation had no significant effect on flunitrazepam binding but decreased Ro15-4513 and Ro15-1788 binding affinities by 52- and 21-fold, respectively, whereas the 2 T81C mutation significantly decreased flunitrazepam, Ro15-4513, and Ro15-1788 binding affinities by 4-, 4-, and 2.5-fold, respectively (Fig. 5, Table 2). In general, the mutations affected Ro15-1788 and Ro15-4513 binding more than flunitrazepam binding, and mutations at 2 Ala-79 had much larger effects than mutations at 2 Thr-81 (Fig. 6, Table 2). These results confirm and extend our chimeric data and indicate that 2 Ala-79 and, to a lesser extent, 2 Thr-81 are important for BZD binding. Moreover, the result suggests that 2 Ala-79 preferentially influences imidazobenzodiazepine binding.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Affects of 2 A79Y and 2 T81C mutations on BZD binding. The myc-tagged 2 (), 2 A79Y (), and 2 T81C () were expressed with wild-type 1 and 2 subunits in HEK 293 cells. Displacement of [3H]flunitrazepam binding by flunitrazepam (top left), Ro15-4513 (top right), and Ro15-1788 (bottom) is shown. Data are from single representative experiments; points are mean ± S.E. of triplicate determinations. Data were fit by nonlinear regression analysis. KI values are summarized in Table 2.

View this table: [in this window] [in a new window]   TABLE 2 Binding affinities of BZD site ligands for myc-tagged 2, 2 A79 mutants, and 2 T81 mutants Ro15-4513 KI values for myc and T81 mutant receptors were determined by homologous competition radioligand binding assays. All other KI values were determined by displacement of [3H]flunitrazepam binding. Results shown are the mean ± S.E.; n is the number of independent experiments. Statistical differences between log KI values were determined by one-way ANOVA using Dunnett's post-test.

View larger version (25K): [in this window] [in a new window]   Fig. 6.   Analysis of point mutations at 2 Ala-79 and 2 Thr-81. The ratios of KI mutant to KI myc for flunitrazepam (FNZM, medium gray), Ro15-4513 (dark gray), and Ro15-1788 (light gray) are shown. KI values for each mutation are summarized in Table 2.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We are interested in identifying the residues that form the BZD binding site. A binding site not only is formed by residues that directly contact agonists, antagonists, and/or inverse agonists but also includes other residues that maintain the structural integrity of the site and/or are involved in local conformational changes that occur when a ligand binds. In this study, the use of 2/1 intersubunit chimeras allows us to identify regions and residues unique to the -subunit that are required for the high-affinity binding of BZD ligands. In contrast to subunit subtype chimeras (e.g., 3/2, 1/2), which are useful for examining the subtle pharmacological differences between -subunit subtypes, 2/1 chimeric subunits can identify residues conserved between -subunit subtypes that are necessary for BZD binding.

By comparing the 2/1 crossover positions (Fig. 1) of a chimera that bound flunitrazepam with high affinity (65) with one that did not (56), we conclude that residues within the ₂ Met-57-Val-65 region are essential for high-affinity [³H]flunitrazepam binding (Fig. 2, Table 1). The substitution of 2 Met-57 in 56 increases flunitrazepam binding affinity by more than 400-fold. When 2 Met-57 and 2 Tyr-58 are simultaneously substituted into 56, flunitrazepam affinity increases by more than 1800-fold (Table 1). Thus, within the 2 Met-57-Ile-62 region, 2 Met-57 and Tyr-58 are key determinants for high-affinity flunitrazepam binding. To a lesser extent, the other unique 2 residues in this region, 2 Asn-60 and 2 Ile-62, may also influence flunitrazepam binding. Only when all four 2 residues are substituted into 56 is high-affinity flunitrazepam binding completely restored to 65 receptor levels (Table 1).

As in all mutagenesis experiments, it is difficult to establish whether these residues define a part of the BZD binding site pocket of the GABA_A receptor or participate in nonlocal allosteric actions that affect BZD binding. Several lines of evidence, however, argue that the residues identified in this study, 2 Met-57 and Tyr-58, are near the BZD binding site. The 56 subunit is expressed in cell-surface GABA_A receptors (Fig. 3). Thus, the ability of 2 Met-57 and 2 Tyr-58 to restore high-affinity flunitrazepam binding to 1256 receptors is not due to an indirect affect of these residues on the assembly and/or expression of 56 but more likely indicates that these residues are involved in flunitrazepam binding. Our conclusion that 2 Met-57 and 2 Tyr-58 are important for flunitrazepam binding is also supported by results using 3/2 chimeras that suggest 2 Met-57 influences zolpidem binding (Buhr and Sigel, 1997). Furthermore, a homologous region in the nicotinic acetylcholine receptor,  Lys-34 and  Ser-36, has been identified as being part of the acetylcholine binding site (Sine et al., 1995). Thus, we conclude 2 Met-57 and Tyr-58 are involved in BZD binding. 2 Met-57 is conserved in all 2 subunits of various species, and 2 Tyr-58 is conserved in all -subunit subtypes. We speculate that these residues interact with the aromatic groups of flunitrazepam by forming a hydrophobic subdomain in the BZD binding site.

The 2 Met-57-Ile-62 region is the most amino-terminal region so far identified to contribute to flunitrazepam binding. If the GABA binding region of the -subunit is homologous to the BZD binding region of the 2 subunit, then our finding that the Met-57 region in the 2 subunit is needed for BZD binding suggests that there may be another region in the 1 subunit, amino terminal to 1 Phe-64 (1 Ile-44-Phe-49), that is involved in the formation of a GABA binding site. Alternatively, the 2 Met-57 region may represent an area uniquely involved in the binding of BZDs.

Differences in the binding affinities of Ro15-4513 and Ro15-1788 for 1265 and 1282 receptors (Fig. 3, Table 1) indicate that an additional region of the 2 subunit, 2 Asn-66-Trp-82, influences BZD binding. By using two complementary approaches, mutagenesis of 65 to identify mutations that increase BZD binding affinity and mutagenesis of wild-type 2 to identify mutations that disrupt binding affinity, we conclude that 2 Ala-79 and, to a lesser extent, 2 Thr-81 are important for BZD binding. The inclusion of 2 Ala-79 or Thr-81 in 65 significantly increases Ro15-4513 binding affinity (Table 1). Consistent with these results, the substitution of 2 Ala-79 with a variety of amino acid residues significantly decreases the affinities of flunitrazepam, Ro15-4513, and Ro15-1788 (Fig. 6, Table 2). Depending on the amino acid that is substituted, the mutation of 2 Thr-81 also causes significant decreases in BZD binding affinities (Table 2). For this residue, the substitutions were fairly conservative (the side chains are all nearly the same size with similar hydrophilicities), which may explain the relatively small effects on affinity that were measured.

The decreases in apparent BZD binding affinity after the mutation of 2 Ala-79 and Thr-81 can be explained by one of two mechanisms. One possibility is that the mutations directly alter the BZD binding site and disrupt the free energy of ligand binding (i.e., the microscopic binding rate constants). Alternatively, the mutations may work indirectly, at a distance, to disrupt the structural integrity of the BZD binding site. Although it is experimentally difficult to distinguish between these two mechanisms (Colquhoun, 1998), several convergent lines of evidence argue that 2 Ala-79 and Thr-81 are located near the BZD binding site. Using a simple allosteric receptor mechanism, it has been shown that unequal shifts in the binding sensitivities of different competitive ligands in response to a mutation indicate that the mutation directly disrupts the binding site (Zhang et al., 1994). As seen in Fig. 6, individual mutations of 2 Ala-79 and Thr-81 alter the affinities of flunitrazepam, Ro15-4513, and Ro15-1788 by different amounts, suggesting that these residues are part of the BZD binding site. The identification of 2 Ala-79 and Thr-81 as BZD binding site residues is concordant with their proximity to 2 Phe-77, a previously identified BZD binding site residue (Buhr et al., 1997a; Wingrove et al., 1997). Covalent modification of 2 A79C with sulfhydryl-specific reagents is slowed in the presence of BZD ligands, lending further support that 2 Ala-79 is part of the BZD binding site (Teissére et al., 1999). 2 Ala-79 and Thr-81 are in homologous aligned positions as 1 Arg-66 and Ser-68, which we have recently shown to be part of the GABA binding site (Boileau et al., 1999). Because the GABA and BZD binding sites appear to be conserved structures (Sigel and Buhr, 1997), it seems probable that 2 Ala-79 and Thr-81 contribute to part of the BZD binding site.

2 Ala-79 and 2 Thr-81 are conserved in the majority of GABA_A receptor -subunit subtypes. In general, substitutions at 2 Ala-79 affect the binding of Ro15-4513 and Ro15-1788 more than flunitrazepam binding. Tyrosine substitution of 2 Ala-79 alters Ro15-4513 and Ro15-1788 binding but not flunitrazepam binding (Fig. 5). Interestingly, tyrosine substitution of 2 Phe-77, an identified binding site residue located near 2 Ala-79, affects flunitrazepam binding but not Ro15-1788 binding (Sigel et al., 1998). This region of the 2 subunit most likely plays a role in BZD ligand discrimination.

The orientation of BZD ligands relative to these amino acid side chains and other identified binding site residues is not known. The stabilization of BZD binding most likely involves electrostatic, van der Waals, and hydrophobic interactions as well as hydrogen bonding between the different BZD component groups and the side chains of binding site amino acid residues. The aromatic binding site residues (e.g., 1 His-101, 2 Phe-77) may be involved in / stacking interactions with the aromatic portions of BZD ligands. Recent evidence suggests that the pendant phenyl group of classic BZDs such as flunitrazepam may interact with 1 His-101 (McKernan et al., 1998), 2 Phe-77 (Sigel et al., 1998), and/or 2 Met-130 (Wingrove et al., 1997). Other residues, such as 1 Tyr-159, 1 Thr-206, and 1 Tyr-209, may be important for hydrogen bond interactions with the seven-member amino-lactam ring of BZDs. Ultimately, confirmation of these structural predictions is dependent on crystallization of the GABA_A receptor.

Many of the /-interface residues that have been identified as being important for BZD binding (2 Phe-77, 1 Tyr-159, 1 Thr-206, 1 Tyr-209) are homologous to the /-interface residues that are important for the binding of GABA (1 Phe-64, 2 Tyr-157, 2 Thr-202, and 2 Tyr-205). In the aligned sequences of the subunits, these residues are identical. However, because the molecular structures of GABA and BZDs are quite distinct, it is likely that nonconserved residues are required to impart pharmacological specificity to these sites. Because 2 Met-57 and 2 Ala-79 are not conserved in the aligned positions of any -subunit, we hypothesize that these particular amino acid side chains may be uniquely involved in BZD binding specificity.

In summary, our results are most simply explained by a model in which 2 Met-57, Tyr-58, and Ala-79 line part of the BZD binding site. The identified residues are clustered in two distinct domains separated by about 20 residues in the linear 2 sequence. Not all of these residues have to make contact with BZDs. Some of the residues may be important for maintaining the local physicochemical properties of the site or be involved in the local changes that occur at the binding site when BZDs bind. In the absence of a high-resolution crystal structure, identification of the amino acid residues involved in BZD binding is a first step toward building a detailed molecular model of the BZD binding site pocket.