Introduction

Top Summary Introduction Materials & Methods Results Discussion References

The GABA_A receptor, a member of the ligand-gated ion channel family, mediates synaptic inhibition through the gating of chloride ions, resulting in hyperpolarization of the cell membrane. It is the site of action of a number of pharmacological agents, including BZs, barbiturates, and anesthetics. The hetero-oligomeric receptor is formed from the coassembly of five different subunit classes [, , ,  (1, 2), and  (3, 4)] in a presumed pentameric arrangement (5, 6) to yield a family of receptor subtypes. It is the heterogeneity within these subunits that provides the molecular basis for the differences in pharmacology of receptor subtypes (7).

Classic BZ pharmacology is exhibited by receptors containing a 2 subunit in combination with an  and a  subunit (8). The affinity of BZ ligands for the receptor is dependent on the  subunit isoform, and hence compounds such as CL218,872 and zolpidem have higher affinity for 1n2 (n = 1, 2, or 3) receptors than for other  subunit-containing receptors (9, 10), and flunitrazepam and diazepam (11, 12) have very low affinity (>10 µM) for 4n2 and 6n2. Mutagenesis studies have identified two amino acids on the  subunit as contributing to the BZ binding site (13, 14). Photoaffinity-labeling of the receptor by BZ ligands [³H]flunitrazepam and [³H]Ro15-4513 also highlights the proximity of the  subunit (15, 16); His102 has been shown to be the major site of incorporation of [³H]flunitrazepam into the 1 subunit (17). It is clear, however, from the studies of both Stephenson et al. (15) and McKernan et al. (16) that the  subunit also contributes significantly to the BZ binding site. In addition, pharmacological studies have demonstrated that the type of  subunit (1, 2, or 3) coexpressed with an  and a  subunit profoundly influences the affinity and efficacy of BZs such as flumazenil and flunitrazepam (18-21). GABA_A receptors containing a 1 subunit have a >5000-fold lower affinity for the antagonist flumazenil than do those containing a 2 or 3 subunit, whereas 1- and 3-containing receptors have a 10-30-fold lower affinity for flunitrazepam than do receptors containing 2 (18-22). In this study, we used these two observations as a starting point to identify the key amino acids of the 2 subunit that contribute to the BZ site of the GABA_A receptor.

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Construction of Chimeric Subunits

Human 1, 1, 1, 2S, and 3 cDNAs have been reported previously (10, 19, 21). The 2S splice isoform is used throughout this study and is referred to simply as 2. A PCR-based method, as described previously (23), was used in construction of the 1/2 chimeric subunits.

12.1. A 2 PCR product was generated with the pCDM8 vector-specific, sense oligonucleotide 5-AGTCCGAAAGAATCTGCTCCCTGCTT-3 and the 2-specific, antisense primer 5-GACAATGAGTATGCATGGGATATAGG-3. This was digested with HindIII and NsiI and inserted into similarly cut 1 in pCDM8.

12.2. A 2 fragment was obtained using the pCDM8 sense primer and the 2-specific antisense primer 5-GTGTTCATCCATGGGAAAATTGTGCA-3. This was digested with HindIII and NcoI and inserted into similarly cut 1 in pBS. The construct was subcloned into pcDNAIamp.

12.3. Two 2-specific primers, 5-TGCACAATTTTCCCATGGATGAACA C-3 (sense) and 5-GACAATGAGTATGCATGGGATATAGG-3 (antisense), were used to amplify the 2 portion, which was cut with NcoI and NsiI and inserted into similarly cut 1 in pBS. The construct was subcloned into pcDNAIamp.

12.4. A BglII site was introduced into 1 in pcDNAIamp by site-directed mutagenesis using the primer 5-GTGGCTGATCCTAGATCTTGGAGATTAT AT-3 (1BglII). A 2 fragment generated with the 12.3 sense primer and 5-AAGCCTCCAAGATCTTGTGTCGCC-3 was digested with NcoI and BglII and inserted into similarly cut 1BglII.

12.5. A 2 fragment was obtained with the 12.3 antisense primer and 5-AAGCCTCCAAGATCTTGTGTCGCC-3, cut with BglII and NsiI, and inserted into similarly cut 1BglII.

12.6. A 2 fragment generated with the 12.4 antisense primer and 5-CACTGTCATCTTGAATTCCCTGCTGGAAG-3 was digested with EcoRI and BglII and inserted into similarly cut 1BglII.

12.7. A 1 PCR fragment was generated using the sense primer 5-ATAGATATATTTTTTGCGCAAACCT-3 and antisense 5-CTTAAAATAGGTACCATACTAGTCACATTTTA-3 and then digested with FspI and SpeI. This was inserted into 2 in pCDM8 digested with FspI and XbaI.

Site-Directed Mutagenesis

Oligonucleotide-directed mutagenesis was performed as described previously (23) using single-stranded 1, 2, or 3 cDNAs in pcDNAI-amp as template and sense-strand oligonucleotides. Mutations were verified by DNA sequencing.

Transient Expression and Radioligand Binding

The  subunit constructs were cotransfected with 1 and 1 cDNAs and the vector pAdVAntage (Promega, Madison, WI) to enhance expression levels (2 µg of each subunit DNA/plate and 6 µg of pAdVAntage). Transient transfection in human embryonic kidney 293 cells (4 × 10⁶ cells/10-cm plate) was performed through calcium phosphate precipitation (24). After 2 days, the cells were harvested by being scraped into phosphate-buffered saline and pelleted through centrifugation. The cell pellet was washed twice in 10 mM potassium phosphate, pH 7.4, with pelleting between washes before being resuspended in assay buffer (10 mM potassium phosphate, pH 7.4, 100 mM potassium chloride) and homogenization by passage through a 27-gauge needle.

Saturation binding curves were obtained by incubation of membranes with [³H]flumazenil, [³H]flunitrazepam, or [³H]Ro15-4513 (all from New England Nuclear Research Products, Boston, MA) at 0.1-30 nM in a total volume of 0.5 ml. Nonspecific binding was determined in the presence of 10 µM flunitrazepam, except for 1I79Y, for which 10 µM Ro15-1788 was used. After 90 min at 4°, the assay was harvested by filtration onto GF/B filters (Brandel, Montreal, Quebec, Canada) using a TOMTEC (Orange, CT) cell harvester. Filters were washed three times with ice-cold assay buffer and dried before filter-retained radioactivity was detected by liquid scintillation counting. Dissociation constants, K_d values, were calculated by Scatchard analysis using GraFit. Displacement of [³H]flumazenil or [³H]flunitrazepam (at a concentration equivalent to the calculated K_d value) by -CCM (Research Biochemicals, Natick, MA), CL218,872 (Lederle, Mont-St-Guibert, Belgium), clonazepam (Sigma Chemical, Poole, Dorset, UK), triazolam (Sigma), flunitrazepam (Sigma), and zolpidem (Synthelabo, Paris, France) was performed in a similar manner. The structures of the compounds are given in Fig. 1. Experimental data points were fitted to a single-site dose-response curve using GraFit, 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 and expressed as mean ± standard error.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Structures of BZ site ligands used in this study.

Electrophysiology

Adult female Xenopus laevis specimens were anesthetized by immersion in a 0.4% solution of 3-aminobenzoic acid ethylester for 30-45 min (or until unresponsive). Ovary tissue was removed via a small abdominal incision, and stage V and VI oocytes were isolated with fine forceps. After mild collagenase treatment to remove follicle cells (Type IA; 0.5 mg/ml for 8 min), the oocyte nuclei were directly injected with 10-20 nl of injection buffer (88 mM NaCl, 1 mM KCl, 15 mM HEPES, pH 7, filtered through nitrocellulose) or sterile water containing different combinations of human GABA_A subunit cDNAs (20 ng/µl) engineered into the expression vector pCDM8 or pcDNAI/Amp. 1, 1, 1, and 1 mutant subunit cDNAs were mixed in a 1:1:3 or 1:1:10 ratio to ensure preferential assembly of receptors. After incubation for 24-72 hr, oocytes were placed in a 50-µl bath and perfused at 4-6 ml/min with modified Barth's solution consisting of 88 mM NaCl, 1 mM KCl, 10 mM HEPES, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, 0.91 mM CaCl₂, and 2.4 mM NaHCO₃, at pH 7.5. Cells were impaled with two 1-3-M electrodes containing 2 M KCl and voltage-clamped between 40 and 70 mV.

In all experiments, drugs were applied in the perfusate until the peak of the response was observed. The effects of GABA_A receptor modulators were examined on control GABA responses using a concentration that elicited 20% of a maximum GABA response on each oocyte (EC₂₀) and a BZ preapplication time of 30 sec. Three minutes were allowed between each application to prevent desensitization. All values are shown as mean ± standard error.

	Results

Top Summary Introduction Materials & Methods Results Discussion References

The initial search for determinants of the 2 subunit that contribute to the BZ binding site was based on the observation that flumazenil has a ~5000-fold higher affinity for 112 and 113 receptors than 111 (18) (Table 1). A simple, single-point assay was used to identify 2 sequences contained within chimeric 1/2 subunits that conferred high affinity binding. A series of six chimeric 1/2 subunits were constructed (Fig. 2) that, when coexpressed with 1 and 1 cDNAs in human embryonic kidney 293 cells, allowed delineation of the determinants for high affinity binding to residues Asn33 to Pro159 of 2 (numbering as for mature peptide). The affinities (K_d values) of [³H]flumazenil for 1112.2 and 1112.6 receptors were 3.07 and 2.59 nM, which is very similar to the affinity at 112 (0.91 nM). A comparison of the  subunit amino acid sequences for this region (Fig. 3) reveals six positions at which the amino acid is conserved in 2 and 3 but not in 1. These residues were targeted for site-directed mutagenesis, with the 1 subunit sequence being changed to that of 2. When coexpressed with 1 and 1 subunits, only one point mutant (1I79F) conferred high affinity binding of [³H]flumazenil (Fig. 4).

View this table: [in this window] [in a new window]   TABLE 1 Affinities for selected BZ site ligands at GABAA receptors containing wild-type, chimeric, or mutant  subunits

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Localization of the BZ binding site within the  subunit. Black areas, chimeric 1/2 subunits represented with the putative signal peptide and transmembrane spanning domains (TM1-4). Shaded areas, portions contributed by the 2 subunit. Column 1, amino acid range. Column 2, subunits coexpressed with 11 and assayed for binding of 2 nM [3H]flumazenil. Column 3, displacement of this binding by 1 µM flunitrazepam and 10 nM zolpidem.

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Alignment of human 1, 2, and 3 subunit sequences. The partial amino acid sequences are shown aligned to the 2 portion encompassed by residues Asn33 and Pro159 (numbering as mature peptide). Residues in capital letters, conserved between sequences. *, Residues common to only 2 and 3. , Residues between 2Gln80 and 2Pro159 where 2 differs from 1 and 3, regardless of whether the latter two are the same (note that 2Ser142 was not mutated in this study and thus is not indicated with ). Boxes, positions of 1 Ile79/2Phe77 and 1Leu132/2Met130.

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Binding of [3H]flumazenil to recombinant receptors containing mutant 1 subunits. The 1 mutants were coexpressed with 1 and 1 and assayed for binding of [3H]flumazenil. Results are the mean values of two experiments.

The affinity of [³H]flumazenil for receptors containing 1I79F was 3.13 nM (Table 1), close to that of receptors containing wild-type 2. Receptors containing 2F77I (as in 1) had an affinity of 1.42 µM. These data confirm the critical role of 2Phe77. To assess the nature of the interaction between the ligand and receptor, a series of subsequent point mutations were made at Ile79 of the 1 subunit. This residue was changed to aspartate, glutamate, histidine, tryptophan, and tyrosine. Only introduction of a tyrosine residue conferred high affinity binding of [³H]flumazenil (Fig. 5, Table 1), demonstrating the requirement of a phenyl ring at this position. It was apparent, however, that the binding of [³H]flumazenil was not displaced by flunitrazepam. Indeed, the affinity for flunitrazepam was significantly lower for receptors containing 1I79Y. K_i values for -CCM and CL218,872 are similar for 111I79F and 112 receptors, demonstrating the importance of this residue. The affinities for triazolam and clonazepam are also significantly increased at 111I79F, although not quite to the affinities at 112, suggesting the requirement for additional determinant or determinants in the 2 subunit. In contrast, the affinity for flunitrazepam and zolpidem at receptors containing 1I79F is not increased to the affinity of receptors containing a 2 subunit (Table 1). This also suggested that additional amino acids within the 2 subunit were required for the high affinity binding of these compounds. Two of the previously constructed chimeras (12.2 and 12.6) were coexpressed with 1 and 1, and the ability of flunitrazepam and zolpidem to displace [³H]flumazenil binding was determined (Fig. 2). An additional subunit chimera, 12.7, was then constructed to further delineate the critical residue (Fig. 2). This last chimera has the phenylalanine residue necessary for [³H]flumazenil binding, but the radioligand was not displaced by flunitrazepam or zolpidem; hence, a second residue delineated by Gln80 and Pro159 of the 2 subunit was conferring higher affinity for flunitrazepam and zolpidem. Receptors containing the 2 subunit (but not those containing 1 or 3) have a high affinity for both flunitrazepam and zolpidem (Table 1); therefore, point mutants were made in the 1 subunit between Gln82 to Pro161 equivalent to positions at which 2 has a different residue from either 1 or 3, regardless of whether these latter two subunits had an identical residue (Fig. 3). Six amino acid positions satisfied this criterion, and the 1 point mutant 1I79F was also mutated at each of these positions so that they could be assayed for displacement of [³H]flumazenil binding by flunitrazepam and zolpidem on coexpression with 1 and 1.

View larger version (12K): [in this window] [in a new window]   Fig. 5.   Introduction of various amino acids at position 79 of the 1 subunit. The isoleucine at position 79 of the 1 subunit was mutated to aspartic acid, glutamic acid, histidine, tryptophan, and tyrosine. Recombinant receptors 11 were assayed for binding of [3H]flumazenil. Height of bar, total binding. White area, binding not displaced by 10 µM flunitrazepam. Values are mean of three experiments.

All of these double mutants were able to bind [³H]flumazenil with high affinity (data not shown), but only one, 1I79F/L132M, showed displacement by flunitrazepam and zolpidem at the concentrations chosen for the assay. The single-point mutant, 1L132M, was subsequently constructed and on coexpression with 1 and 1 formed receptors with high affinity for [³H]flunitrazepam (K_d = 3.72 nM; Table 1). K_i values for a number of compounds were obtained for 111L132M receptors by displacement of [³H]flunitrazepam (Table 1). Both triazolam and clonazepam had significantly increased affinities at 111L132M (16- and 56-fold, respectively, compared with 111), approaching their affinities at 112. Flumazenil, -CCM, CL218,872, and zolpidem have low affinity for 111L132M, suggesting 2Met130 is not a critical residue for the binding of these compounds. The affinity of zolpidem was further investigated at receptors containing the 1I79F/L132M double mutant and found be to 10-fold higher than for receptors containing 1 subunits with either single-point mutation but still 30-fold lower than for 112, indicating an interaction with additional determinants.

The 3 subunit is similar to 2 in having a phenylalanine residue at position 80 and similar to 1 in having a leucine residue at position 133 (Fig. 3), and receptors containing a 3 subunit have a distinct BZ pharmacology (21). To confirm the importance of the residues identified above, the 3 subunit was altered to give the mutants 3F80I and 3L133M. Binding of [³H]flumazenil was abolished to receptors containing 3F80I (n = 2; data not shown), confirming the importance of the phenylalanine residue at this position. Receptors containing 3L133M had a 4-5-fold increase in affinities for flunitrazepam, triazolam, and clonazepam (Table 1) and a 17-fold increase in affinity for zolpidem, confirming the importance of this residue at the BZ binding site. In contrast, the affinity for flumazenil was essentially unaffected by changes at this position of the 3 subunit, further demonstrating that amino acids at this position of the  subunit do not contribute to flumazenil binding.

In addition to affecting affinity,  subunits can confer differences in BZ efficacy (19, 21). To determine the contributions to BZ efficacy of the individual amino acids identified in this study, mutated  subunits were coexpressed with 11 in X. laevis oocytes, and modulation by BZs was compared with wild-type receptors (Table 2). We reported previously that 211 receptors are modulated by BZs, although with generally lower efficacy than 212 receptors (19). Here, we report that 111 receptors were not modulated by flunitrazepam, CL218,872, -CCM, or zolpidem (Table 2). The presence of the 1 subunit in the 111 receptor complex was confirmed by the higher GABA EC₅₀ value compared with 11 and the relative insensitivity to zinc compared with 11 (Table 3). 111I79F receptors, however, were potentiated by flunitrazepam, but unlike 112, 100 nM flunitrazepam did not elicit a maximum response (Fig. 6), suggesting a lower affinity for the former subunit combination, which in fact is the case (Table 1). A comparison of the data in Fig. 6 with the affinities for flunitrazepam given in Table 1 reveals that the EC₅₀ value for flunitrazepam at 112 and 111I79F is a little higher than the K_i value derived from radioligand binding. This is not unusual (25) and presumably reflects the fact that one is a direct measurement of the binding energy, whereas the other is a functional determination. 111I79F receptors were also potentiated by CL218,872 and inhibited by the inverse agonist -CCM, although with lower efficacy than receptors containing 2 (Table 2). They were not modulated by zolpidem, reflecting the low affinity of this compound for 111I79F receptors. Like 111, receptors containing 1L132M were not modulated by any of the compounds tested, including flunitrazepam, which binds with an affinity of 3 nM. Coassembly of the 1L132M subunit into the receptor complex was again confirmed by higher GABA EC₅₀ values and insensitivity to zinc (Table 3). Taken together, these data suggest a critical role for 2Phe77 in conferring BZ efficacy to the receptor. To confirm this hypothesis, the mutant 2F77I was constructed and coexpressed with 1 and 1 subunits. 112F77I receptors were not modulated by flunitrazepam (which binds with an affinity of 7.62 nM; Table 1) or any of the other BZ site ligands tested (Table 2), confirming the importance of this residue in conferring BZ efficacy to the receptor.

View this table: [in this window] [in a new window]   TABLE 3 GABA EC50, Hill coefficients, and sensitivity to zinc of GABAA receptors expressed in X. laevis oocytes EC50 values are geometric mean ( standard error, + standard error) and the slope arithmetic mean ± standard error. Inhibition by 3 µM zinc represents the percent inhibition of a GABA EC50 response for each receptor combination.

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Percent modulation of EC20GABA responses at 111, 112, 111I79F, 111L132M, and 112F77I GABAA receptors by 100 nM and 1 µM flunitrazepam. Modulation is expressed as the percent potentiation of an EC20 concentration of GABA for each oocyte. Values are mean ± standard error of at least four oocytes. For reference, the affinity (determined by radioligand binding; see Table 1) of flunitrazepam at the various subunit combinations is also shown.

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

Although some studies have been performed to identify residues responsible for the  subunit-selective binding profile of BZ site ligands (13-14), few studies have been made of the role of the  subunit. The  subunit is an essential component of the BZ binding site (8), and the BZ pharmacology is profoundly affected by the type of  subunit present in the receptor complex (18-21). For example, flumazenil has a much greater affinity for 2- and 3-containing receptors than those containing 1. This observation provided the criterion that initiated this study. The sequence homology between the  subunits made it possible to construct chimeric  subunits that would coassemble with 1 and 1 subunits. At this stage, and during the creation of subunit point mutants, the strategy was to introduce the determinants that conferred high affinity binding.

A single amino acid, 2Phe77, was found to be necessary for high affinity binding of [³H]flumazenil. The presence of this residue confers 5000-fold increase in affinity, indicating that it is one of the key constituents of the BZ binding site. It is a similarly important determinant for the binding of other structurally diverse BZ site ligands (i.e., CL218,872 and -CCM). High affinity ("2-like") binding of triazolam and clonazepam is also dependent on the presence of this phenylalanine residue. It was hoped that more could be learned of the interaction between receptor and ligand by making further amino acid substitutions at 1Ile79. Of the five substitutions made, only 1I79Y conferred a high affinity for [³H]flumazenil (Table 1 and Fig. 5). Phenylalanine and tyrosine differ only by the addition of a para-hydroxyl group, suggesting the common benzene ring is interacting with flumazenil. Interestingly, receptors containing 1I79Y had a >100-fold reduction in affinity for flunitrazepam compared with 1I79F- containing receptors, suggesting the presence of the para-hydroxy group disrupts binding.

A second residue in the  subunit, 2Met130, also contributes to the BZ binding site. This residue seems to have no significant effect on the binding affinity of flumazenil, -CCM, or CL218,872; however, it is an important determinant for the binding of flunitrazepam, triazolam, and clonazepam, as demonstrated by the increased binding affinity when a methionine residue is introduced at the equivalent position in both 1 and 3. The latter three compounds have a pendant phenyl ring (Fig. 1), allowing speculation that the interaction with the methionine residue occurs through this moiety.

The affinity for zolpidem of receptors containing either of the 1 point mutants is >5 µM. When both changes are introduced together, the affinity is increased but remains >30-fold lower than that at receptors containing 2. Similarly, the presence of both the phenylalanine and methionine residues in 3L133M increases the affinity for zolpidem to 330 nM but is still 8-fold less than that at 2-containing receptors. Additional amino acid determinants in the 2 subunit may therefore be necessary to attain the 40 nM affinity achieved at 112 receptors. The location of the two (or possibly more) determinants required for high affinity binding of zolpidem on the  subunit reveals the considerable contribution of this subunit to the binding site. However, zolpidem (a so-called BZ1-selective compound) has higher affinity for 1-containing receptors than for receptors containing other  subunits (9, 10). These data may be reconciled if the binding site for zolpidem is formed largely by determinants from the 2 subunit; the lower affinity of zolpidem for 312 receptors compared with 112 receptors is due to increased steric hindrance by the large amino acid residue in 3, which is responsible for the selectivity (3Glu225; Ref. 13) compared with the small glycine residue at the equivalent position in 1.

The functional properties observed when the various  subunit mutants where coexpressed with 1 and 1 indicate that Phe77 is also required for allosteric modulation of the receptor by the BZ. When this position is occupied by an isoleucine (as in 1, 1L132M, and 2F77I), the receptor is not modulated, despite affinities of 3-44 nM for flunitrazepam. Conversely, the receptors containing 1I79F were modulated by all the BZs tested, with the exception of zolpidem. The lack of modulation of 111 by BZs is in contrast to that observed for 211 (a combination likely to exist in vivo; Refs. 26 and 27), in which BZs are able to allosterically modulate the receptor (19). This suggests that a residue or residues in the 2 subunit can partially compensate for the effects of the phenylalanine residue and confer a degree of positive modulation to the receptor, albeit less than that conferred by Phe77; all BZ compounds tested had lower efficacy on 221 (19). One apparent contradiction is that although Phe77 is not required for the binding of flunitrazepam, it is a requirement for efficacy. For other BZ site ligands, such as -CCM and CL218,872, Phe77 is required for both binding and, presumably, modulation. These data can be reconciled if Phe77 is an absolute requirement for allosteric modulation, not necessarily by direct interaction with the ligand, and is used as a binding determinant by some classes of BZ site ligands. An alternative hypothesis is that Phe77 could be a contact point for flunitrazepam but not necessary for the compound to bind (i.e., other contact points satisfy the energy requirements for high affinity binding); in the absence of Phe77, the compound could occupy the binding site in a conformation that is incapable of initiating the allosteric changes leading to modulation of the channel.

The data reported here, in conjunction with previous studies characterizing the BZ pharmacology of 3-containing receptors, also suggest that Met132 does not influence the efficacy of BZs because despite differences in affinity, in a comparison of 2- and 3-containing receptors, flunitrazepam, dimethoxy-4-ethyl--carboline-3-carboxylate, bretazenil, zolpidem, and CL218,872 have similar degrees of efficacy (21).

An interesting insight from this study is that 2Phe77 is at a position homologous to 1Phe64. The latter is a critical residue at the GABA binding site; mutations at this position affect the affinity of GABA (28), and this residue is the site of photoincorporation of the GABA site radiolabel [³H]muscimol (29). The GABA site has contributions from both the  (28, 29) and  subunits (30), whereas as discussed, the BZ site has contributions from both the  and  subunits. One interpretation is that the BZ site is a vestigial GABA binding site that over time has mutated and lost its ability to bind GABA, but by chance synthetic molecules (i.e., BZs) are able to bind to this site and thereby modulate receptor function. Indeed, the recent observation by Amin et al. (31) that 1Tyr159 and 1Tyr 209 (both conserved in all  subunits) are components of the BZ binding site supports this hypothesis; these two residues are homologous to 2Tyr157 and 2Tyr205, previously demonstrated to be part of the GABA binding site (30).

The observation that the aromatic residues tyrosine and phenylalanine are key components of both the GABA and BZ binding sites is a recurring theme in ligand-gated ion channels. The aromatic residues phenylalanine and tyrosine are thought to also contribute to the acetylcholine binding site on the nicotinic receptor subunit (32), the glycine binding site of the strychnine-sensitive glycine receptor (33), and the glycine coagonist site of the N-methyl-D-aspartate-type glutamate receptor (34).

A recent report has also demonstrated that 2Phe77 is an important determinant for the binding of BZ site ligands (35), which is in good agreement with the current data. However, this study also reported that diazepam was able to potentiate receptors containing 2F77I; in contrast, we found that this phenylalanine is a key determinant for modulation of receptors by a number of BZ site ligands. The reason for this apparent discrepancy is unclear. Another amino acid in the 2 subunit that has also been shown to directly affect the efficacy of BZ compounds is Thr142, which when mutated to serine increased the efficacy of BZ ligands, changing flumazenil and Ro15-4513 to agonists (36). However, this mutation did not affect BZ affinity.

In conclusion, we demonstrated that at least two residues in the  subunit are key determinants of the BZ site of the GABA_A receptor. 2Phe77 is required for high affinity binding of some, but not all, BZ site ligands, but according to current results, it seems to be an absolute requirement for functional modulation by these compounds. 2Met130 is also required for high affinity binding of some but not all BZ site ligands, but it does not seem to influence allosteric modulation.