The Major Site of Photoaffinity Labeling of the -Aminobutyric Acid Type A Receptor by HFlunitrazepam Is Histidine 102 of the Subunit

The α subunit of the -aminobutyric acid type A (GABA) receptor is known to be photoaffinity labeled by the classical benzodiazepine agonist, [3H]flunitrazepam. To identify the specific site for [3H]flunitrazepam photoincorporation in the receptor subunit, we have subjected photoaffinity labeled GABA receptors from bovine cerebral cortex to specific cleavage with cyanogen bromide and purified the resulting photolabeled peptides by immunoprecipitation with an anti-flunitrazepam polyclonal serum. A major photolabeled peptide component from reversed-phase high performance liquid chromatography of the immunopurified peptides was resolved by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate. The radioactivity profile indicated that the [3H]flunitrazepam photoaffinity label is covalently associated with a 5.4-kDa peptide. This peptide is glycosylated because treatment with the enzyme, peptide-N4-(N-acetyl-β-glucosaminyl)asparagine amidase, reduced the molecular mass of the peptide to 3.2 kDa. Direct sequencing of the photolabeled peptide by automated Edman degradation showed that the radioactivity is released in the twelfth cycle. Based on the molecular mass of the peptides that can be generated by cyanogen bromide cleavage of the GABA receptor α subunit and the potential sites for asparagine-linked glycosylation, the pattern of release of radioactivity during Edman degradation of the photolabeled peptide was mapped to the known amino acid sequence of the receptor subunit. The major site of photoincorporation by [3H]flunitrazepam on the GABA receptor is shown to be α subunit residue His (numbering based on bovine α sequence).

The ␥-aminobutyric acid type A (GABA A ) 1 receptor mediates the majority of rapid inhibitory synaptic transmission through-out the mammalian central nervous system. As a member of the superfamily of ligand-gated ion channels (1), the GABA A receptor is believed to be a hetero-pentameric protein that spans the neuronal membrane to create a chloride conducting pore. The homologous subunits that assemble to form the receptor-chloride channel complex are encoded by distinct but related genes (2). Six ␣, four ␤, four ␥, one ␦, and two subunit isoforms plus splice variants for many of the genes have been identified and classified by sequence similarity. However, the precise stoichiometry of subunit isoforms that comprise native receptors remains unknown.
A multiplicity of neuroactive drugs have been shown to interact specifically with the GABA A receptor complex to modulate inhibitory neurotransmission throughout the brain (3). These include the benzodiazepines, barbiturates, some steroids and general anaesthetics, and possibly alcohol (4). Because of the clinical usefulness of the benzodiazepines as anxiolytics, hypnotics, and anticonvulsants, their interaction with the GABA A receptor has been extensively studied. The benzodiazepines are known to be allosteric modulators of GABA A receptors in that the classical agonists potentiate whereas the inverse agonists reduce GABA-mediated chloride conductances.
An area of particular interest has been the identification of protein domains involved in the interaction of benzodiazepines with the GABA A receptor. Using multidisciplinary approaches, several groups have identified structural features of subunit isoforms that are important for ligand recognition and for the modulatory effects of the benzodiazepines (reviewed in Ref. 5). To date, site-directed mutagenesis has identified the amino acids Gly 225 of the ␣ 1 subunit (6), His 101 of the ␣ 1 subunit (7), and Thr 142 of the ␥ 2 subunit (8) as residues that play a role in conferring the differential binding affinities of benzodiazepine ligands for the GABA A receptor. Biochemical approaches have shown that the site of photoaffinity labeling by the classical agonist, [ 3 H]flunitrazepam, is associated with the ␣ subunit of the GABA A receptor (9,10) within the large extracellular amino-terminal domain (11,12). In addition, partial sequences of proteolytic fragments from photoaffinity labeled receptors have indicated that the [ 3 H]flunitrazepam site occurs within amino acid residues 8 -297 of the ␣ 1 subunit (13), and using subunit specific antibodies, the site has been predicted to occur within residues 59 -158 of the ␣ 1 sequence (14). We have mapped the [ 3 H]flunitrazepam photoaffinity labeled peptides generated by hydroxylamine cleavage to known GABA A receptor sequences and have demonstrated that the site of photolabeling occurs within amino acids 1-103 of the ␣ subunit (15). These studies, considered together, limit the predicted site of labeling to within residues 59 -103 of the ␣ 1 subunit or within homologous segments of other ␣ subunit isoforms. In the present study, we have employed immunoprecipitation and HPLC techniques to purify a [ 3 H]flunitrazepam photoaffinity labeled peptide that was generated by cyanogen bromide cleavage of labeled GABA A receptors from bovine cerebral cortex. It is shown by peptide mapping and microsequence analysis that the major site of [ 3 H]flunitrazepam photoincorporation by the GABA A receptor is likely to be the amino acid His 102 of the bovine ␣ 1 subunit.
Preparation of Membranes from Bovine Cerebral Cortex and Photoaffinity Labeling with [ 3 H]Flunitrazepam-Adult bovine brain was obtained from a local slaughterhouse. The cerebral cortex was dissected out and was immediately frozen on dry ice and stored at Ϫ80°C. Brain membranes were prepared from approximately 100 g of partially thawed cortex as described previously (16). For the photoaffinity labeling reaction, membrane aliquots were thawed and diluted to a final protein concentration of 2 mg/ml in 20 mM Tris citrate pH 7.4 buffer containing 1 mM EDTA, 1 mM benzamidine, 0.5 mM dithiothreitol, 0.3 mM phenylmethylsulfonyl fluoride, 10 g/ml soybean trypsin inhibitor, 20 g/ml bacitracin, and 0.02% NaN 3 . [ 3 H]Flunitrazepam, isotopically diluted 5-fold to a specific activity of 17.2 Ci/mmol, was added to a final concentration of 10 nM. The mixture was incubated for 45 min in the dark on ice with constant shaking, before irradiating with long wavelength ultraviolet light for 45 min using a Spectroline ENF 260C lamp at a distance of 6 cm. The membranes were subjected to repeated cycles of centrifugation (150,000 ϫ g, 45 min) and resuspension until the radioactivity in the supernatant fell to close to the background, at which point the membranes were finally resuspended at a concentration of 15 mg/ml in the above buffer.
Solubilization of Photoaffinity Labeled Membranes and Protein Precipitation-Photoaffinity labeled membranes were stirred on ice at 4°C, and an equal volume of solubilization buffer containing 20 mM Tris citrate, pH 7.5, 0.5 M KCl, 3% CHAPS, 0.3% asolectin, and protease inhibitors as noted above was added dropwize. After being stirred for 60 min at 4°C, the mixture was centrifuged for 75 min at 100,000 ϫ g. To the supernatant containing the photoaffinity labeled receptor, a volume of trichloroacetic acid was added to a final concentration of 12% (w/v). Following incubation on ice for 15 min, the solution was centrifuged at 10,000 ϫ g for 15 min, and the protein pellets were washed twice with acetone.
Cyanogen Bromide Cleavage of the Photoaffinity Labeled Receptor-CNBr cleavage (17) of the photolabeled receptor was carried out by dissolving the solubilized and precipitated protein to a concentration of 10 mg/ml in 70% formic acid and adding a volume of 25% CNBr in formic acid solution to achieve a final amount of CNBr equal to that of the total protein (mg per mg basis). Following 24 h of incubation at room temperature in the dark, the reaction mixture was diluted into 15 volumes of distilled water, freeze dried twice, and resuspended to about 10 mg/protein ml in 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.02% NaN 3 buffer for subsequent immunopurification.

Immunoprecipitation of [ 3 H]Flunitrazepam Photoaffinity Labeled
CNBr Peptides-Sheep raised anti-flunitrazepam polyclonal serum (approximately 1.25 l of neat antiserum) was added per pmol of the [ 3 H]flunitrazepam photoaffinity labeled CNBr peptides, and the mixture was incubated for 2 h at 37°C. About 50 l of 50% protein G matrix in 10 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.02% NaN 3 buffer was added per pmol of labeled peptide, and the slurry was incubated for 24 h at 4°C with constant mixing. The Protein G matrix was extensively washed by repeated cycles of centrifugation (10,000 ϫ g, 15 min) and resuspension in a series of different solutions; 1) 10 mM Tris-HCl pH 7.4, 150 mM NaCl buffer, 2) 10 mM Tris-HCl pH 7.4 buffer, 3) 20% (v/v) ethanol, and 4) distilled water. To specifically recover the labeled peptides while avoiding spurious contaminants that were present in the precipitation assay, a large excess of free unlabeled flunitrazepam was used to elute the immunoprecipitated peptides from the antibody-protein G complex. A volume of flunitrazepam dissolved in 10% ethanol was added to the protein G beads equivalent to 100 nmol of free drug/pmol of labeled peptide. The slurry was incubated for 24 h at 4°C with constant mixing, before the beads were pelleted by centrifugation and the supernatant containing the immunopurified peptides was re-moved and freeze-dried. The dried product was resuspended in distilled water, and the insoluble free flunitrazepam was removed by ether extraction. After gentle mixing with diethyl ether/distilled water (3:1) to solubilize the drug, the ether phase was discarded, and the aqueous phase containing the immunopurified peptide components was again freeze-dried. The [ 3 H]flunitrazepam photoaffinity labeled CNBr peptides were finally resuspended in 0.1% trifluoroacetic acid and stored at 4°C for analysis by HPLC.

Reversed-Phase HPLC of the Immunoprecipitated [ 3 H]Flunitrazepam Photoaffinity Labeled CNBr
Peptides-Further purification of the immunoprecipitated peptides was achieved by reversed-phase HPLC using a Vydac C18, 25 ϫ 0.46-cm column. The peptide components were eluted from the column using a linear gradient established with aqueous 0.05% trifluoroacetic acid (A) and acetonitrile/0.05% trifluoroacetic acid (B) at a flow rate of 1 ml/min. After 45 min of 1% B per min, the mixture was ramped to 90% B and held for 5 min and then returned to 100% A. Fractions of 1 ml were collected and the 3 H radioactivity profile of the fractions was determined. Distinct peaks were pooled and concentrated by vacuum centrifugation for further study.
Deglycosylation with N-Glycanase-The hydrolysis of all common classes of Asn-linked oligosaccharides at the ␤-aspartyl-glycosylamine bond between the innermost GlcNAc and the asparagine residue is catalyzed by the enzyme, N-Glycanase (18,19). For deglycosylation experiments, pooled HPLC fractions were dried and resuspended in 0.5 M sodium phosphate, pH 8.0, 0.5% SDS, and 50 mM ␤-mercaptoethanol and boiled for 5 min. Prior to the addition of 0.3 units of N-Glycanase enzyme, the sample was diluted 3-fold and Nonidet P-40 (final concentration, 1.25%) and 1,10-ortho-phenanthroline (final concentration, 10 mM) were added. The mixtures were incubated for 18 h at 37°C, before the deglycosylated peptides were prepared for resolution by HPLC or SDS-PAGE.

SDS-PAGE and Scintillation
Counting-The immunoprecipitated photoaffinity labeled CNBr peptides and the components purified by reversed-phase HPLC were analyzed by Tricine SDS-PAGE for the superior resolution of species with molecular mass in the 1-100-kDa range (20). Electrophoretic separations were carried out on gels that consisted of a 16.5% T, 6% C resolving slab, a 10% T, 3% C spacer gel, and a 4% T, 3% C stacking gel (where T denotes the total percentage concentration of both monomers (acrylamide and bisacrylamide) and C denotes the percentage concentration of the cross-linker relative to the total concentration T; see Ref. 20). Immediately following electrophoresis, sample lanes were cut into 0.25-cm slices and eluted in 0.5 ml of 2% SDS at 50°C overnight, and radioactivity was determined by scintillation counting. Molecular weight (M r ) values of peptides were estimated by the migration profiles of parallel standards using low molecular weight rainbow standards.
Sequence Analysis-Amino-terminal sequence analysis was performed with an Applied Biosystems model 473A pulsed liquid/gas phase protein sequencer. Pooled HPLC samples were concentrated by vacuum centrifugation and stored at 4°C. Aliquots from the fractions were immobilized on polybrene-treated filters for sequencing by standard Edman degradation. The PTH amino acids generated by each sequencer cycle were directly transferred to a fraction collector instead of analyzing the derivatives by HPLC, and the amount of [ 3 H] in each cycle was assessed by scintillation counting. In some instances, the sequencer was allowed to operate with PITC omitted from the standard reaction protocol. This allowed estimation of the amount of nonspecific elution of radioactivity during automated sequencing of particular samples.

RESULTS
To characterize the site of photoincorporation by [ 3 H]flunitrazepam on the GABA A receptor, a photoaffinity labeled peptide component from bovine cerebral cortex was purified by immunoprecipitation with a polyclonal antiserum raised against free flunitrazepam and with the precipitating reagent, protein G-linked Sepharose. In addition to quantitatively precipitating free [ 3 H]flunitrazepam, the anti-flunitrazepam serum was shown to immunoprecipitate photoaffinity labeled peptides by specifically recognizing the [ 3 H]flunitrazepam ligand covalently associated with the GABA A receptor. After confirming (by SDS-PAGE) that the photoincorporation of [ 3 H]flunitrazepam with the GABA A receptor from bovine cerebral cortex was associated with a major 53-kDa protein, previously defined as the ␣ subunit(s) (9, 10), and that there was no significant labeling of other species, the photoaffinity labeled protein was solubilized and subjected to specific chemical cleavage at Met residues by treatment with CNBr. The concentration dependence of immunoprecipitation of [ 3 H]flunitrazepam photoaffinity labeled CNBr peptides with anti-flunitrazepam serum was investigated to assess the optimal antibody concentration for use in large scale purification. A maximum immunoprecipitation of 70 Ϯ 9% was achieved with less than 1 l of neat antiserum/pmol of [ 3 H]flunitrazepam photolabeled CNBr peptide. Nonspecific adsorption of labeled peptides to the protein G matrix was measured in the absence of antiserum and represented less than 1% of the total yield.
Batch immunoprecipitation was used to purify [ 3 H]flunitrazepam photoaffinity labeled CNBr peptides in quantities sufficient for further characterization. About 50% of the total yield of immunoprecipitated product could be specifically eluted from the antibody-protein G complex by incubation with free flunitrazepam. The radioactivity profile of the immunopurified peptides resolved by Tricine SDS-PAGE (Fig. 1A) demonstrated that CNBr cleavage of the GABA A receptor preparation generated a [ 3 H]flunitrazepam photoaffinity labeled peptide of 5.5-kDa molecular mass, with a minor component that resolved as a 2.5-kDa species. Unfortunately, the inadequate amount of purified protein precluded the use of silver stain for protein detection. The elution profile from reversed-phase HPLC of the immunoprecipitated peptides (Fig. 1B) displays three apparent peaks of radioactivity, arbitrarily marked (i), (ii), and (iii). Although the 3 H peak marked (iii) invariably represented 55-60% of the total radioactivity loaded onto the column, the relative amount of radioactivity in the other two peaks varied between peptide preparations. The percentage of 3 H that eluted with the peaks ranged from 17 to 35% for (i) and from 8 to 26% for (ii). However, the combined cpm in peaks (i) and (ii) routinely represented 40 -45% of the total radioactivity.
The pooled fractions from each of the radioactive HPLC elution peaks were resolved by Tricine SDS-PAGE for determination of the molecular mass of the [ 3 H]flunitrazepam labeled CNBr peptide components. In addition, the samples were treated with the deglycosylation enzyme, N-Glycanase, to assess whether the radiolabeled peptides contained asparaginelinked oligosaccharides. On SDS-PAGE, HPLC peak (iii) resolved as a radiolabeled 5.4-kDa peptide ( Fig. 2A), i.e. a profile similar to the unfractionated peptides (Fig. 1A). This peptide was glycosylated because after N-Glycanase treatment of peak (iii), the apparent molecular mass of the labeled peptide was reduced to 3.2 kDa (Fig. 2B). Reversed-phase HPLC of deglycosylated peak (iii) also showed a small but consistent shift to a higher percentage of acetonitrile required for the elution of this peptide from the column. In both the SDS-PAGE and HPLC profiles, some undigested peak (iii) apparently remained after N-Glycanase treatment, and this accounted for approximately 20% of the total radioactivity.
When HPLC peak (ii) was resolved by SDS-PAGE, it ran as a broad band of radioactivity with apparent molecular mass ranging from 8.7 to 12.1 kDa. After digestion with N-Gly-

FIG. 1. Immunoprecipitated [ 3 H]flunitrazepam photoaffinity labeled CNBr peptides resolved by SDS-PAGE and HPLC. A, representative radioactivity profile of immunopurified [ 3 H]flunitrazepam photolabeled peptides resolved by Tricine SDS-PAGE is shown.
Determination of the cpm per gel slice was performed as described under "Experimental Procedures." The numerals above each arrow indicate the relative position of the molecular mass standards (expressed in kDa). The major peak of radioactivity corresponds to a peptide band with approximate molecular mass of 5.5 kDa, whereas the smaller component resolves as a 2.5-kDa peptide. The data shown are typical of several preparations of immunopurified peptides. B, immunoprecipitated [ 3 H]flunitrazepam photolabeled peptides were resolved by reversed-phase HPLC as described under "Experimental Procedures." The data are shown as the representative radioactivity profile of the elution of [ 3 H] in 1-ml fractions as determined by scintillation counting of 100-l aliquots. Approximately 14,000 cpm of the immunoprecipitated peptides were loaded directly onto the column, and the recovery of [ 3 H] was greater than 93%. The peak marked as (i) eluting at fraction 21 contained 4,700 cpm, the small peak marked (ii) eluting at fraction 26 contained 1,100 cpm, and the largest peak (iii) eluting at fraction 33 contained approximately 7,250 cpm.

FIG. 2. Peak (iii) from reversed-phase HPLC of immunopurified [ 3 H]flunitrazepam CNBr peptides, resolved by SDS-PAGE before and after treatment with N-Glycanase.
The 3 H profiles of native and deglycosylated peptides of HPLC peak (iii) resolved by SDS-PAGE are shown as described in the legend to Fig. 1. A, peak (iii) from reversed-phase HPLC of immunoprecipitated peptides resolves by SDS-PAGE to an apparently single radiolabeled peptide with molecular mass of about 5.4 kDa. B, following treatment of the peptides eluted in peak (iii) with N-Glycanase as described under "Experimental Procedures," the major radiolabeled species resolves to a 3.2-kDa peptide with some residual 5.4-kDa peptide evident. canase, this peak resolved to a more distinct but still broad band of 3 H that corresponds to a radiolabeled peptide of about 9 kDa. It was not possible to resolve the radioactivity present in HPLC peak (i) to a distinct protein band by electrophoresis. In repeated attempts, the radioactivity showed a diffuse migration pattern spanning the middle portion of the gel that did not vary when the pore size of the gel was altered. The incubation of peak (i) with N-Glycanase had no effect on the apparent inability of this fraction to be resolved by gel electrophoresis.
The [ 3 H]flunitrazepam photoaffinity labeled peptides purified by HPLC (see Fig. 1B) were subjected to direct automated sequencing to measure the release of radioactivity during each cycle. The pattern of release of radioactive PTH amino acids generated by Edman degradation of the [ 3 H]flunitrazepam photolabeled peptides present in HPLC peak (iii) shows that a maximum release of 3 H occurred in sequencer cycle 12 (Fig. 3). The release of radioactivity remained elevated for about four sequencer cycles before returning to baseline levels. The 3 H elution seen in cycles 13-16 is most likely due to NH 2 -terminal cleavage from residual peptide containing the [ 3 H]flunitrazepam-associated amino acid, originally residue 12, and/or incomplete extraction of the modified amino acid. It is probable that the covalent incorporation of the ligand with a particular amino acid adversely affects the efficiency of the Edman degradation reaction. The moderate level of radioactivity seen in the first cycle likely results from desorption of peptide from the filter cartridge system of the sequencer. The pattern of radioactive release from automated sequencing of HPLC peaks (i) and (ii) were indicative of nonspecific elution during the Edman degradation cycles (Fig. 4). Therefore, duplicate aliquots of these samples were processed in the same manner as for standard sequencing, except that PITC was omitted from the reaction chemistry. The elution of 3 H during the cycles from sham sequencing of peaks (i) and (ii) was parallel to that obtained for the standard Edman reaction cycles. DISCUSSION Several structural determinants required for the allosteric modulation of GABA A receptors by the benzodiazepines have been characterized by site-directed mutagenesis of recombinant receptors (see Introduction). Using this information, the benzodiazepine binding domain has been modelled as a composite of these structural features (5). However, the identity of specific amino acid residue(s) in native GABA A receptors that are directly involved in benzodiazepine binding has remained an area of intense interest. One approach used extensively has been to specifically and irreversibly label the benzodiazepine binding site with the photoactivatable agonist, [ 3 H]flunitrazepam, as first described by Möhler et al. (21). Although strong evidence has been reported to show that the major site of [ 3 H]flunitrazepam photoaffinity labeling occurs on the GABA A receptor ␣ subunit (5), the precise position of the photolabel on the polypeptide has not previously been established.
To identify the [ 3 H]flunitrazepam photoaffinity labeling site on the GABA A receptor, a purified preparation of labeled peptide was required in sufficient quantities to allow for characterization by conventional biochemical techniques. The development of an anti-flunitrazepam immunoprecipitation assay has provided a highly specific method to purify the photoaffinity labeled peptides from a crude brain preparation. The [ 3 H]flunitrazepam photoaffinity labeled peptides generated by CNBr cleavage were precipitated with an antiserum directed against free flunitrazepam that also specifically recognized the covalently attached ligand. Whereas the major component evident in the immunopurified preparation was a 5.5-kDa photolabeled peptide when resolved by SDS-PAGE, three distinct peaks of radioactivity were resolved by reversed-phase HPLC.
The [ 3 H]flunitrazepam present in HPLC peak (iii) represented the majority (approximately 60%) of the total immunoprecipitated radioactivity and was shown by SDS-PAGE to be  HPLC peak (iii). The amount of 3 H associated with the PTH amino acids generated by each cycle of Edman degradation of the photolabeled peptides present in HPLC peak (iii) was determined as described under "Experimental Procedures." The radioactivity shown is that observed and has not been corrected for repetitive yield of sequencer cycles. Approximately 30% of the radioactivity loaded onto the sequencer was recovered in the fractions collected from each cycle; about 25% remained on the filter and cartridge seal, and the remainder was presumably lost in the washes.

FIG. 4. Release of radioactivity from HPLC peaks (i) and (ii)
subjected to automated Edman degradation. The amount of 3 H that eluted with each cycle during standard conditions of Edman degradation and when PITC was omitted from the automated reaction was investigated (see "Experimental Procedures"). Equivalent amounts of radioactivity were loaded onto the sequencer for all instances. A, automated sequencing of HPLC peak (i) in the presence (E) and absence (ϫ) of PITC. Of the 3 H recovered, 28% was released during the cycles of Edman degradation, 4% was left on the filter and cartridge seal, and 68% was presumed lost in washes. B, automated sequencing of HPLC peak (ii) in the presence (E) and absence (ϫ) of PITC; the recovery of the radioactivity loaded onto the sequencer was 17% in the normal reaction cycles, 3% remained on the filter, and 80% was lost in washes. The radioactivity released from peaks (i) and (ii) during the sham sequencing cycles (minus PITC) was slightly reduced, whereas the amount of 3 H presumed to be lost in washes was increased compared with that for the standard sequencer runs. associated with a peptide of apparent molecular mass 5.4 kDa. It was also shown that this photolabeled peptide contains asparagine-linked carbohydrate, because after N-Glycanase digestion, it migrated as a 3.2-kDa peptide. Recognizing that: 1) CNBr specifically cleaves peptide bonds on the carboxyl side of Met residues with high efficiency, except for Met-Thr and Met-Ser bonds, which are essentially resistant to CNBr and 2) asparaginyl-linked glycosylation exclusively occurs at the consensus sequence of Asn-Xaa-Thr or Asn-Xaa-Ser, the origin of the photolabeled peptide can be mapped to the known amino acid sequence of the NH 2 -terminal domain of the GABA A receptor ␣ 1 subunit (see Fig. 5). Considering the potential sites for cleavage by CNBr and for asparaginyl glycosylation, the only peptide that could be generated from the ␣ subunit to contain Asn-linked carbohydrate and to resolve by electrophoresis as described above is Ala 91 -Met 114 . This peptide has a predicted molecular mass of 3.1 kDa without consideration given for glycosylation, which is close to the estimates obtained by SDS-PAGE analysis. The other CNBr peptide that could be generated from the ␣ subunit to contain Asn-linked oligosaccharide is Gln 1 -Met 59 , which would have a molecular mass after deglycosylation of greater than 6.5 kDa. Because previous attempts to sequence the ␣ subunit indicated that the polypeptide has a blocked NH 2 terminus (1), the peptide beginning with Gln 1 would also be refractory to the Edman chemistry. Furthermore, the peptide bonds carboxyl to Met 90 and Met 114 have previously been shown to be susceptible to CNBr cleavage, and sequencing of the peptides Ala 91 -Ala 109 and Pro 115 -Thr 130 provided the information necessary for the first cloning of the GABA A receptor subunits (1). The [ 3 H]flunitrazepam photolabeled peptide has been mapped to the ␣ 1 subunit, because this isoform has been shown to occur in the vast majority of native GABA A receptors from bovine cerebral cortex (22). Although other, less abundant, isoforms of the ␣ subunit have been identified in cortical GABA A receptors, the interpretation for the origin of the photolabeled peptide is not compromised, because the isoforms possess a high degree of sequence identity throughout this domain.
The pattern of release of radioactive PTH amino acids obtained from automated Edman degradation of the major photolabeled peptide (Ala 91 -Met 114 ) indicated the [ 3 H]flunitrazepam is covalently associated with the twelfth residue, which corresponds to His in position 102 of the ␣ 1 subunit. The photoincorporation of [ 3 H]flunitrazepam with His 102 is consistent with the findings of hydroxylamine cleavage experiments that demonstrated photolabeling occurred prior to Asn 103 , as well as previous reports that predicted the site occurred within limited subunit domains (see Introduction). The involvement of a histidine residue in the interaction of benzodiazepines with the GABA A receptor was implicated in earlier studies that investigated the effects of chemical modification and the pH dependence of radioligand binding (23,24). In addition, His 102 of the ␣ 1 subunit is the residue that was shown by point mutation to be required for the high affinity binding of benzodiazepine agonists in recombinantly expressed GABA A receptors (7).
The chemical nature of the radioactivity present in HPLC peaks (i) and (ii) has not been established. HPLC peak (ii), which accounted for 10 -25% of the total radioactivity, resolved by SDS-PAGE to a molecular mass ranging from 9 to 12 kDa, with deglycosylation causing a marginal shift in the gel profile that indicated a 9-kDa photolabeled peptide. The radioactivity profile obtained from direct sequencing of peak (ii) did not suggest the [ 3 H]flunitrazepam label was associated with a particular amino acid residue. Automated sequencing in the absence of PITC demonstrated that the release of radioactivity seen during the first few cycles of Edman degradation was not due to the generation of radiolabeled PTH amino acids resulting from NH 2 -terminal cleavage of a labeled peptide. Although a reliable prediction for the origin of the photolabeled peptide is difficult, we can speculate that a 9-kDa peptide containing [ 3 H]flunitrazepam-labeled His 102 could result from incomplete cleavage at one or more of the potential CNBr sites in the FIG. 5. Partial amino acid sequence of the bovine GABA A receptor ␣ 1 subunit. The GABA A receptor sequence of the large extracellular amino-terminal domain of the ␣ 1 subunit from Gln 1 to the first putative transmembrane domain is shown (1). Met residues are in bold type with the potential cleavage sites for CNBr indicated by down arrows, the potential sites for asparagine-linked glycosylation are underlined (Asn-Xaa-Thr), the large bracket between Asn 103 and Gly 104 marks the only potential site in the subunit for specific cleavage by hydroxylamine, and the putative membrane spanning region is marked by the shaded box above the initial residue Phe 226 . NH 2 -terminal domain of the ␣ subunits. Alternatively, the peptide may have been generated from a different domain of the ␣ subunit or from another subunit subtype, thereby representing a minor fraction of photolabel incorporated with a residue other than His 102 . Recent studies have suggested that the benzodiazepine binding domain is made up of determinants from both the ␣ and ␥ subunits. Site-directed mutagenesis has found that Thr 142 of the ␥ subunit is involved in conferring the modulatory effects of benzodiazepine ligands (8), and previous photoaffinity labeling experiments have shown data that suggest some [ 3 H]flunitrazepam label may incorporate with the ␥ subunit (14). The apparent inability to resolve the radioactivity present in HPLC peak (i) by electrophoresis suggested the [ 3 H]flunitrazepam was not associated with a distinct peptide. In addition, the radioactivity profiles obtained from automated sequencing of peak (i) were not consistent with the release of labeled PTH amino acids from the Edman degradation reaction. It is possible that the peak (i) fraction of the immunopurified product may represent free [ 3 H]flunitrazepam or ligand incorporated with carbohydrate or some other nonprotein molecule.
In conclusion, we have presented evidence that photoaffinity labeling of the GABA A receptor with [ 3 H]flunitrazepam leads to the covalent association of the ligand with His 102 of the ␣ subunit. Despite our intention to characterize the photolabeled peptide components by microsequence analysis, current technical limitations and insufficient yields of the purified peptides have precluded the identification of the PTH amino acids generated during each cycle. Therefore, although His 102 of the ␣ subunit is likely to be the major site of photoincorporation, it is not possible to substantiate the possible existence of other amino acids that may be photolabeled by [ 3 H]flunitrazepam.