Probing the Structure of the Nicotinic Acetylcholine Receptor with 4-Benzoylbenzoylcholine, a Novel Photoaffinity Competitive Antagonist*

[ 3 H]4-Benzoylbenzoylcholine (Bz 2 choline) was synthe- sized as a photoaffinity probe for the Torpedo nicotinic acetylcholine receptor (nAChR). [ 3 H]Bz 2 choline acts as an nAChR competitive antagonist and binds at equilibrium with the same affinity ( K D 5 1.4 m M ) to both agonist sites. Irradiation at 320 nm of nAChR-rich membranes equilibrated with [ 3 H]Bz 2 choline results in the covalent incorporation of [ 3 H]Bz 2 choline into the nAChR g - and d -subunits that is inhibitable by agonist, with little specific incorporation in the a -subunits. To identify the sites of photoincorporation, g - and d -subunits, isolated from nAChR-rich membranes photolabeled with [ 3 H]Bz 2 choline, were digested enzymatically, and the la- beled fragments were isolated by sodium dodecyl sul-fate-polyacrylamide gel electrophoresis and/or re-versed-phase high performance liquid chromatography. For the g -subunit, Staphylococcus aureus V8 protease produced a specifically labeled peptide beginning at g Val-102, whereas for the d -subunit, endoproteinase Asp-N produced a labeled peptide beginning at d Asp-99. Amino-terminal sequence analysis identified the homologous residues g Leu-109 and d Leu-111 as the primary sites of [ 3 H]Bz 2 choline photoincorporation.

The Torpedo nicotinic acetylcholine receptor (nAChR) 1 is composed of four homologous subunits in a stoichiometry of ␣ 2 ␤␥␦ that associate pseudosymmetrically about a central axis that is the ion channel. Cryoelectron microscopic analyses of two-dimensional crystalline arrays of nAChRs provide a definition of three-dimensional structure approaching 5-Å resolution (1), including a definition of changes in structure in the transmembrane region between closed and open channel states (2). Since there is as yet no direct identification of the path of the polypeptide chains or of the structure of the acetylcholine (ACh)-binding sites, our knowledge of the structure of the AChbinding sites has been derived primarily from the results of affinity labeling and mutational analyses.
Although there is considerable information about the identity of amino acids contributing to the agonist-binding sites, little is known about the orientation of agonists or antagonists bound within the site. Exceptions include the labeling of Loop A by [ 3 H]acetylcholine mustard which suggests an interaction between ␣Tyr-93 in the desensitized nAChR and the quaternary ammonium group of ACh (20) and recent double mutant cycle analyses that have begun to identify likely pairwise interactions between amino acids within the peptide ␣-neurotoxins and amino acids of the nAChR, including residues in Loops C (21) and F (22). The structures of the reactive intermediates formed upon UV irradiation of dTC and nicotine are unknown. Therefore, these photolabeling studies, while identifying residues contributing to the agonist sites, do not define the orientation of the drugs within the binding sites.
In this report, we demonstrate that 4-benzoylbenzoylcholine (Bz 2 choline), which possesses a well defined photochemistry, can be used as a photoaffinity probe of the agonist-binding sites. As with other benzophenones, Bz 2 choline is activated by ultraviolet light at Ͼ320 nm, producing aĊ-Ȯ diradical that can insert into C-H bonds but will not react with water (reviewed in Refs. [23][24][25]. Thus, proximity alone should determine which residues are covalently labeled, independent of the inherent reactivity of the amino acid side chains, which should enable Bz 2 choline to identify amino acids in proximity to the para position of a benzoylcholine ester when it is bound in the AChbinding sites. We report here that Bz 2 choline is a competitive antagonist of the Torpedo nAChR binding with equal affinity (K D ϳ1.4 M) at the two ACh-binding sites. [ 3 H]Bz 2 choline specifically photoincorporates into the nAChR ␥and ␦-subunits with similar efficiencies, with specific incorporation within the ␣-subunit at no more than 10% the level in the ␥or ␦-subunit. The primary sites of specific photoincorporation for [ 3 H]Bz 2 choline are the homologous residues within Loop E of the non-␣-subunits, ␥Leu-109 and ␦Leu-111.
Synthesis of Bz 2 choline-Bz 2 choline was synthesized by coupling Bz 2 acid to choline with 1,1Ј-carbonyl diimidazole (CDI). To the mixture of Bz 2 acid (0.23 g, 1 mmol) and CDI (0.20 g, 1.2 mmol) was added 3 ml of anhydrous N,N-dimethylformamide. The reaction mixture was stirred at room temperature for 2 h before addition of dry choline p-toluene sulfonate (0.55 g, 2 mmol). After reacting overnight with stirring, the product was purified by preparative layer chromatography to remove unreacted Bz 2 acid (5:10:1 chloroform/methanol/acetic acid; R f values, Bz 2 choline, 0.16; Bz 2 acid, 0.81). The product was then extracted from the preparative layer chromatography plate in 15 ml of 50% methanol and concentrated to 1 ml by centrifugal evaporation. To remove excess choline, the sample was loaded on a Sephadex LH-20 column (1.5 ϫ 25 cm) and eluted with water (1 ml/min, 2 min/fraction). The eluate was monitored by absorbance at 261 nm, and for the radioactive synthesis, 3  To confirm further the purity of Bz 2 choline, a sample was hydrolyzed (100 nmol) in concentrated NH 4 OH for 8 h. The sample was then dried by centrifugal evaporation, and the residual was dissolved in water. Half of the sample was used to analyze the amount of 4-benzoylbenzoic acid by spectrophotometry, using an extinction coefficient of 20,000 cm Ϫ1 M Ϫ1 for 4-benzoylbenzoic acid at pH 7.0 (28). The other half was used to quantify the amount of choline released by base hydrolysis. This was determined by coupling the phosphorylation of choline by choline kinase to the oxidation of NADH (29). The ratio of benzoylbenzoic acid to choline was 0.95 Ϯ 0.02.
[ 3 H]Bz 2 choline was synthesized by the method described above either starting with Bz 2 acid and [methyl-3 H]choline (5 mCi, isotopically diluted to 70 Ci/mol) on a 1:20 scale or from [3, H]4-benzoylbenzoic acid (5 mCi, isotopically diluted to 700 Ci/mol) and choline p-toluene sulfonate on a 1:200 scale. The higher specific activity compound was used preferentially for the preparative photolabeling experiments.
Electrophysiology-Torpedo nAChR subunit-specific cRNAs were transcribed in vitro as described (15). Isolated, follicle-free oocytes were microinjected with 10 ng of Torpedo nAChR subunit-specific RNA in a molar stoichiometry of 2:1:1:1 ␣/␤/␥/␦. Oocytes were maintained in low-Ca 2ϩ ND96 solution containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH 7.6, and 50 g/ml gentamicin for at least 48 h prior to use. Currents elicited by ACh were measured using a standard two-electrode voltage clamp (Oocyte Clamp OC-725B, Warner Instrument Corp.) at a holding potential of Ϫ70 mV. The recording chamber (150 l volume) was continually gravity perfused with low-Ca 2ϩ ND96 solution (ϩ1 M atropine, pH 7.6). 3 M ACh with various concentrations of Bz 2 choline was applied to the oocyte in the recording chamber for 5 s through solenoid valves.
Radioligand Binding Assays-The equilibrium binding of radioactive cholinergic ligands to nAChR-rich membranes in Torpedo saline (TPS, 250 mM NaCl, 5 mM KCl, 3 mM CaCl 2 , 2 mM MgCl 2 , 5 mM sodium phosphate, pH 7.0) was measured by centrifugation in a Tomy MXT-150 microcentrifuge (30 I-␣BgTx binding, nAChR-rich membranes in TPS (5 nM ACh-binding sites) were incubated with various concentrations of Bz 2 choline or dTC for 20 min at room temperature. The binding reaction was initiated by adding 125 I-␣-BgTx to a final concentration of 10 nM, and after appropriate intervals, 125 I-␣-BgTx binding was terminated by removing aliquots and adding them to non-radioactive ␣-BgTx (5 M final concentration). The quenched reaction solutions were centrifuged at 150,000 ϫ g for 20 min in a Beckman 42.2Ti rotor. The supernatants were discarded, and the pellets were washed twice with TPS before counting. Under these assay conditions, 125 I-␣-BgTx bound to ϳ35% of the available sites in 10 min, and the nonspecific binding was ϳ2% of the total 125 I. Three time points were used to determine the initial rates of binding of 125 I-␣-BgTx for each condition.
Data Analysis-The concentration dependence of Bz 2 choline inhibition of ACh-induced currents or of radioligand equilibrium binding was fit to Equation 1, where f(x) is the initial rate of 125 I-␣-BgTx binding seen in the presence of inhibitor concentration x; f 0 is the rate of binding in the absence of inhibitor, and K H and K L are the dissociation constants for the high and low affinity binding sites, respectively. Equilibrium binding data for [ 3 H]Bz 2 choline were fit to a single site model with a linear, nonspecific binding component as shown in Equation 3,  (15 M) and additional cholinergic ligands for 20 min prior to photolysis, except for ␣-BgTx, which was incubated with the membranes for 60 min prior to addition of [ 3 H]Bz 2 choline. The samples were then irradiated on ice for 60 min using a 100-watt, focused, short-arc, mercury lamp with a Pyrex glass filter and a 310-nm cut-off filter. Polypeptides were resolved by SDS-PAGE and visualized by Coomassie stain. The 3 H incorporation into individual polypeptides was determined by scintillation counting of excised gel slices as described (12) or by fluorography as described previously (26), except that Amplify (Amersham Pharmacia Biotech) was used as the fluorographic reagent.
For labeling on a preparative scale, membrane suspensions (3 mg/ml; 3-7 ml in 2.5-cm diameter plastic Petri dishes) were equilibrated with [ 3 H]Bz 2 choline (15 M) and, where indicated, with 30 M proadifen, a desensitizing aromatic amine noncompetitive antagonist (31). To visualize the nAChR subunits resolved by preparative SDS-PAGE without staining and destaining of the gel, after photolysis with [ 3 H]Bz 2 choline, membrane suspensions were photolabeled with 1-azidopyrene, a fluorescent photoreactive hydrophobic compound (32). 1-Azidopyrene solution (10 mg/ml in Me 2 SO) was added to final concentration of 60 g/ml, and the sample was irradiated with a 365-nm hand-held lamp for 3 min. After SDS-PAGE, the unstained gel was visualized under UV light, and the nAChR subunits were excised based upon the fluorescence of incorporated 1-azidopyrene. The excised gel pieces containing ␥or ␦-subunit were diced, and the protein was passively eluted for 3 days into 30 ml of gel elution buffer (100 mM NH 4 HCO 3 , 0.1% SDS, pH 7.0). The eluate was filtered (Whatman No. 1 filter paper), and the protein was concentrated to ϳ200 l by Centriprep-30 (Amicon). Excess SDS was removed by acetone precipitation (80% final concentration, overnight at Ϫ20°C). Starting from 15 mg of nAChR-rich membranes (typically ϳ15 nmol of ACh sites), ϳ500 g of protein was recovered from the subunit bands. Based upon amino-terminal sequence analyses, the ␥and ␦-subunit preparations both contained as a major contaminant the Na ϩ /K ϩ -ATPase ␤-subunit present at 50 -100% of the level of ␥or ␦-subunit, which was not unexpected based upon previous reports (13). The ␦-subunit preparations also contained ␥-subunit at 10 -30% the level of ␦-subunit. This contamination by ␥-subunit results because ␦-subunit is the subunit most difficult to visualize based upon the fluorescence of the azidopyrene adduct.
Enzymatic Digestions and Peptide Purification-For digestion with S. aureus V8 protease, the nAChR subunits or subunit fragments were resuspended in gel elution buffer and digested for 3 days with V8 protease (20% w/w). For digestion with EndoLys-C, the nAChR subunits were resuspended in 100 mM Tris, 0.1% SDS, pH 8.6, and digestion conditions are described in the figure legends. Digests were fractionated on 1.5-mm thick 16.6% T, 6% C Tricine/SDS-PAGE gels (32,33). To identify radioactive fragments, a strip was cut from the gel, cut into 2-mm slices, and counted, whereas the remainder of the gel was stored at 4°C. This 3 H profile was used as a template to identify the regions of the preparative gel containing 3 H. Material eluted from the gel was resuspended in gel elution buffer for further digestion with V8 protease or in the same buffer supplemented with 1% Genapol C-100 for digestion with EndoAsp-N (2 g, 25°C, 7 days). After purification by SDS-PAGE, eluted material (with or without additional protease treatment) was further purified by reversed-phase HPLC with a Brownlee Aquapore C-4 column (100 ϫ 2.1 mm). Solvent A was 0.09% trifluoroacetic acid in water, and solvent B was 60% acetonitrile/40% isopropyl alcohol/0.05% trifluoroacetic acid. The gradients (in % solvent B) are indicated in the figures as dashed lines. The eluate (0.2 ml/min) was monitored by absorbance at 215 nm, by fluorescence emission at 432 nm (355 nm excitation), and by counting fraction aliquots (5-10%) for 3 H.
To determine whether the labeled ␥-subunit fragment generated by V8 protease was glycosylated, protease digestion was terminated with 1 mM diisopropyl phosphofluoridate and then an aliquot of the digest (ϳ30 g) in 50 l was treated with endoglycosidase H (10 milliunits) for 3 days. This material as well as a parallel sample not treated with glycosidase were fractionated by analytical Tricine/SDS-PAGE, with the distribution of 3 H on the gel determined by fluorography.
Sequence Analysis-Amino-terminal sequence analysis was performed on an Applied Biosystems model 477A protein sequencer using gas-phase cycles optimized (34) for sequencing on chemically modified glass-fiber peptide supports (Beckman 290111). HPLC fractions were reduced in volume to 50 l and loaded directly onto the filters. Onethird of each cycle was injected for mass determination by an on-line model 120A PTH-derivative analyzer, and two-thirds were collected for 3 H counting. Cycle mass yields for PTH-derivatives were estimated from chromatographic peak heights. Values plotted are actual cpm and mass values detected. Initial peptide mass (I 0 ) and repetitive yield (R) were calculated by nonlinear least squares regression (Sigma Plot) of the function f(n) ϭ I 0 ϫ R n where f(n) is the observed mass released in cycle n. PTH-derivatives of Ser, Cys, Arg, His, and Trp were excluded from the fits due to known problems with measuring mass yields of these residues. In some cases, sequencing of some peptides was blocked by treatment of the sample on the filter with o-phthalaldehyde (Pierce) prior to cycles containing known proline residues. o-Phthalaldehyde reacts with primary, but not secondary, amines and can be used to block Edman degradation of any peptide not containing an amino-terminal proline (35). o-Phthalaldehyde treatment was carried out as described (12).

Bz 2 choline
Is an nAChR Antagonist-We first examined interactions of Bz 2 choline with Torpedo nAChRs expressed in Xenopus oocytes. Concentrations of Bz 2 choline up to 100 M elicited no detectable currents, but it was a potent antagonist of ACh-induced currents (Fig. 1A). When co-applied for 5 s with 3 M ACh, Bz 2 choline produced a dose-dependent inhibition of ACh responses (IC 50 ϭ 1 M) that was completely reversible after removal of the Bz 2 choline from the oocyte perfusion solution. Inhibition of 125 I-␣-BgTx Binding-Initial rates of binding of 125 I-␣-BgTx were determined for nAChR-rich membranes equilibrated with Bz 2 choline or dTC (Fig. 1C). For the assay conditions used, 125 I-␣-BgTx bound to the two ACh sites at equal rates, and inhibition of binding by dTC was well fit, as expected (5)  choline, a concentration sufficient to occupy ϳ90% of the agonist sites. After 1 h of photolysis, the pattern of incorporation was assessed by SDS-PAGE followed by fluorography or by scintillation counting of bands excised from the gel. As seen in the fluorograph of an 8% acrylamide gel (Fig. 2), [ 3 H]Bz 2 choline was incorporated primarily in the nAChR ␥and ␦-subunits, with incorporation in ␣-subunit at lower efficiency. The incorporation in ␥and ␦-subunits was inhibited by Carb, whereas incorporation in ␣-subunit appeared largely nonspecific. There was also Carb-inhibitable incorporation in a band with a mobility close to that of rapsyn (43K) in a region of the gel known to contain a proteolytic fragment of ␥-subunit (36). Surprisingly, whereas incorporation in ␥-subunit was inhibited by dTC or ␣-BgTx, dTC only partially inhibited the incorporation in ␦-subunit, and in the presence of ␣-BgTx there was apparently increased incorporation in ␦-subunit. The pharmacological specificity of the photolabeling at the level of the nAChR subunits was characterized by quantifying 3 H incorporation in excised gel bands in this and additional experiments. Incorporation in ␥-subunit was reduced by 80 -90% by Carb, dTC, or ␣-BgTx, whereas for ␦-subunit, Carb reduced labeling by 70%, but dTC reduced labeling by only 30%, and ␣-BgTx actually increased labeling by 15-20%. The amount of Carbinhibitable 3 H incorporation in ␥and ␦-subunits indicated specific labeling of 6 -8% of the nAChRs. Carb-inhibitable labeling of the ␣-subunit band varied between 3 and 20% that of the ␥-band. Furthermore, when ACh sites were occupied by ␣-BgTx, [ 3 H]Bz 2 choline incorporation in ␦-subunit was inhibited 85% by tetracaine, an aromatic amine noncompetitive antagonist that binds with high affinity to the nAChR ion channel in the presence of ␣-BgTx (37, 38) (not shown). The pharmacological specificity of [ 3 H]Bz 2 choline photoincorporation in ␥-subunit indicated that ϳ90% of the incorporation at the subunit level was in amino acids within the agonist-binding site. For ␦-subunit, the inhibition by Carb along with the tetracaine-inhibitable labeling in the presence of ␣-BgTx indi- cated that there might be incorporation both within the agonist-binding site and within the ion channel domain.

Effects of Bz 2 choline on Equilibrium Binding of [ 3 H]ACh, [ 3 H]dTC, and [ 3 H]H 12 -Histrionicotoxin-Equilibrium
[ 3 H]Bz 2 choline Photoincorporation in ␥-Subunit-To provide an initial characterization of the site(s) of agonist-inhibitable labeling, ␥-subunit was digested in solution with V8 protease and fractionated by SDS-PAGE, with a part of the digest also treated with endoglycosidase H, which removes high mannose and some hybrid Asn-linked carbohydrates (39). As seen in the fluorogram of the gel (Fig. 3), the majority of [ 3 H]Bz 2 choline incorporation was contained in a band of ϳ12 kDa. This 3 Hlabeled fragment also contained the endoglycosidase H-sensitive Asn-linked carbohydrate in nAChR ␥-subunit, which is known to be attached to ␥Asn-141 (15).
To identify the site of labeling in the ␥-subunit, [ 3 H]Bz 2 choline-labeled ␥-subunit was isolated by preparative SDS-PAGE and digested with V8 protease in solution for 3 days. The digest was then fractionated by Tricine/SDS-PAGE, and a strip of the preparative gel was excised and cut into 2-mm slices for scintillation counting (Fig. 4A). There was a single major peak of 3 H at ϳ12 kDa. About 25% of the 3 H loaded on the gel was recovered when this region was cut out from the preparative gel and eluted. When this material was further purified by reversed-phase HPLC (Fig. 4B), about 45% of 3 H loaded on the column was recovered in a single peak of 3 H centered at 60% solvent B, with the 3 H in that peak reduced by 80% for the sample labeled in the presence of Carb. Material in that peak (fractions 26 -29) was pooled and concentrated for amino-terminal sequence analysis (Fig. 4C). The only nAChR subunit sequence detected began at ␥Val-102 (ϪCarb, I 0 ϭ 54 pmol; ϩCarb, I 0 ϭ 28 pmol). In addition, in each sample there were three fragments from the ␤-subunit of the Na ϩ /K ϩ -ATPase beginning at Tyr-230 (I 0 ϭ 50 pmol), Ser-142 (I 0 ϭ 20 pmol), and Gly-172 (I 0 ϭ 15 pmol). For the sample labeled in the absence of agonist, there was a prominent peak of 3 (15). Incorporation of [ 3 H]Bz 2 choline into the fragment beginning at ␥Val-102 was consistent with the observation that the 3 Hlabeled fragment also contained ␥Asn-141, since the first glutamate after ␥Val-102 is ␥Glu-155. To verify the identification of the labeled peptide and labeled amino acid, an aliquot of the material purified by HPLC from the ␥-subunit V8 digest was digested with chymotrypsin. Since digestion with chymotrypsin would cleave after ␥Tyr-105, if the ␥Val-102 peptide contained the 3   An alternative fragmentation strategy was developed to isolate the ␥-Val-102 fragment in greater purity. Digestion of the ␥-subunit in solution with endoproteinase Lys-C (EndoLys-C) produces a fragment of ϳ28 kDa extending from ␥Glu-47 to ␥Lys-192 (13) which could be isolated for subsequent digestion by V8 protease. [ 3 H]Bz 2 choline-labeled ␥-subunit was digested with EndoLys-C in solution, and the digest was then fractionated by preparative Tricine/SDS-PAGE with an aliquot of the digest separated in an adjacent analytical lane. The distribution of 3 H in the analytical lane was determined by gel slice analysis (Fig. 5A), and the resulting profile was used as a template to isolate the ϳ26-kDa band containing the majority of the 3 H label in the preparative gels. Sequence analysis of this material identified a ␥-subunit fragment beginning at ␥Glu-47, with no detectable 3 H release (Ͻ0.2 cpm/pmol) in or around cycle 9, corresponding to ␥Trp-55, the primary site of photolabeling by [ 3 H]dTC (13). Aliquots of this material were subsequently digested with V8 protease (72 h at 25°C), and the digest was fractionated by reversed-phase HPLC (Fig. 5B). About 45% of the loaded 3 H was recovered in a single peak eluting at 38% solvent B (fractions 16 -17). Sequence analysis of the material in fraction 16 (Fig. 5C) revealed the presence of a primary sequence beginning at ␥Val-102 (ϪCarb, I 0 ϭ 38 pmol; ϩCarb, I 0 ϭ 40 pmol) with prominent, Carb-inhibitable release of 3 H in cycle 8 consistent with incorporation at ␥Leu-109. Two secondary sequences at ϳ10% the level of the primary sequence were also detected, beginning at Val-1 of V8 protease and Arg-147 from the Na ϩ /K ϩ -ATPase ␤-subunit. No other nAChR peptides were present at Ͼ1 pmol mass level. The 3 H released in cycle 8 was 16 cpm/pmol of ␥Leu-109, which, based upon the radiochemical specific activity of the [ 3 H]Bz 2 choline, indicated that ϳ4% of the ␥Leu-109 residues were labeled.
[ 3 H]Bz 2 choline Photoincorporation in ␦-Subunit-For nAChRrich membranes photolabeled with 15 M [ 3 H]Bz 2 choline, incorporation in ␦-subunit was inhibited ϳ90% by Carb, but it was not inhibited at all by ␣-BgTx (Fig. 2). To provide an initial localization of the site(s) of agonist-inhibitable photoincorporation and of the site of photoincorporation in the presence of ␣-BgTx, aliquots of ␦-subunit isolated from nAChR-rich membranes labeled in three conditions (Control, ϩCarb, and ϩ␣-BgTx) were digested with En-doLys-C in solution (3 days, 25°C). The digests were fractionated by Tricine/SDS-PAGE, with the distribution of 3 H determined by gel slice analysis (Fig. 6). For the ␦-subunit isolated from membranes labeled with [ 3 H]Bz 2 choline in the absence of competing drugs (Control), digestion with EndoLys-C produced a major 3 H band of ϳ20 kDa as well as a band of ϳ30 kDa, and 3 H incorporation was not seen in either band for the digests of subunits isolated from nAChRs labeled in the presence of Carb or ␣-BgTx. For the ␦-subunit isolated from membranes labeled in the presence of ␣-BgTx, there was a major 3 H band of ϳ10 kDa that was not present in the other digests. This band had a mobility similar to a [ 3 H]tetracaine-labeled fragment beginning at the NH 2 terminus of ␦M2 that was also isolated from EndoLys-C digests (38)  increased 3 H incorporation in the 10-kDa band, likely to contain the transmembrane domain of the nAChR, can account for the apparent lack of inhibition by ␣-BgTx of [ 3 H]Bz 2 choline labeling at the level of the intact ␦-subunit. When the 20-kDa 3 H-labeled band was isolated from an EndoLys-C digest of labeled ␦-subunit and sequenced, ␦-subunit fragments beginning at ␦His-20, ␦His-26, and ␦Glu-47 were present at similar mass levels, as well a fragment from the Na ϩ /K ϩ -ATPase ␤-subunit beginning at Arg-147. There was no 3 H release above background during 15 cycles of Edman degradation, indicating that there was no detectable specific 3 H incorporation in the region of ␦Trp-57 (Ͻ0.1 cpm/pmol), the amino acid photolabeled by [ 3 H]dTC (13).
We wanted to determine whether [ 3 H]Bz 2 choline was photoincorporated in ␦Leu-111, the position homologous to ␥Leu-109. Digestion with V8 protease would not be useful for this purpose, because it would be difficult to obtain sufficient yields at that position when sequencing a fragment beginning after ␦Glu-85, the first glutamate amino-terminal to ␦Leu-109. Digestion with endoproteinase Asp-N (EndoAsp-N), which cleaves before aspartates and sometimes before glutamates (40), could provide an alternative strategy if there was efficient cleavage at ␦Asp-99. EndoLys-C digests of labeled ␦-subunit were fractionated by preparative Tricine/SDS-PAGE (Fig. 7A), and the region of the gel between 20 and 29 kDa was excised and eluted. This material was then digested with EndoAsp-N (1 week, 25°C), and the digest was separated by reversedphase HPLC (Fig. 7B). When the principal peak of 3 H (fraction 24) from the ϪCarb sample was sequenced, the primary 3 H release was in cycle 13 with lower level release in cycle 9 (Fig.  7C). 3 H release in those cycles was reduced by Ͼ90% for the sample isolated from nAChRs labeled in the presence of Carb. Two ␦-subunit peptides were detected beginning at ␦Asp-99 (ϪCarb, I 0 ϭ 39 pmol; ϩCarb, I 0 ϭ 48 pmol) and ␦Asp-76 (ϪCarb, I 0 ϭ 27 pmol; ϩCarb, I 0 ϭ 42 pmol). In addition there were fragments from nAChR ␥-subunit beginning at ␥Asp-76 and ␥Glu-101 (ϪCarb, I 0 ϭ 40 and 17 pmol, respectively) and a primary sequence from the ␤-subunit of the Na ϩ /K ϩ -ATPase beginning at Arg-147 (ϪCarb, I 0 ϭ 160 pmol). Although the polypeptide composition of the sample was complex, the 3 H release in cycle 13 was consistent with specific [ 3 H]Bz 2 choline photoincorporation in ␦Leu-111 in the fragment beginning at ␦Asp-99, and the release in cycle 9 would be consistent with labeling of ␥Leu-109 in the fragment beginning at ␥Glu-101.
Additional samples were sequenced to clarify the source of 3 H release seen in cycles 9 and 13 (Fig. 8). Sequence analysis of the adjacent fractions (fractions 22, 23, and 25) revealed that the level of 3 H release in cycle 13 was largest for fraction 24, which had the highest level of the ␦-Asp-99 fragment, whereas release in cycle 9 was greatest for fraction 23, which contained the ␥Glu-101, ␥Asp-76 and ␦Asp-76 fragments at the highest levels. Based upon the 3 H release in cycle 13 and the masses of the ␦Asp-99 fragment in the fractions, the calculated 3 H incorporation at ␦Leu-111 in fractions 23-25 was 10, 6, and 8 cpm/ pmol, respectively. Although 3 H release in cycle 9 correlated reasonably well with the mass levels of ␥Glu-101, ␦Asp-76, or ␥Asp-76 in the adjacent fractions, an additional experiment established that neither of the fragments beginning at ␦Asp-76 or ␥Asp-76 could be the source of 3 H release in cycle 9 (or 13). We took advantage of the fact that both fragments contain prolines prior to the observed 3 H release in cycle 9 and that sequence analysis can be interrupted at any desired cycle of Edman degradation to treat the material on the sequencing filter with o-phthalaldehyde, which blocks Edman degradation of all peptides without an amino-terminal proline (12,35). Thus, during sequence analysis of aliquots of fraction 23, samples were treated with o-phthalaldehyde prior to cycle 8 (␦Pro-83) or cycle 6 (␥Pro-81). Following these treatments the only sequence present was ␦Asp-76 (45 pmol) or ␥Asp-76 (65 pmol), depending upon the cycle of treatment, and neither had 3 H release in cycle 9 or 13 (data not shown). Therefore, the 3 H release in cycles 9 and 13 did not result from photolabeling of the fragments beginning at ␥Asp-76 and ␦Asp-76. EndoAsp-N can cleave on the amino-terminal side of glutamates (40), and in separate experiments with labeled ␥-subunit, we verified that EndoAsp-N does cleave at ␥Glu-101 (not shown). Based upon the 3 H release in cycle 9 and the masses of the ␥Glu-101 fragment in the fractions, the calculated 3 H incorporation at ␥Leu-109 in fractions 23-25 was 15, 13, and 5 cpm/pmol, respectively, which was similar to the level of incorporation at ␥Leu-109 seen in the ␥-subunit band analyzed from the same labeling experiment (16 cpm/pmol, Fig. 5). DISCUSSION In this paper we introduce [ 3 H]Bz 2 choline as a novel antagonist photoaffinity probe of the nAChR. Bz 2 choline binds with For the preparative lanes, the region from 20 to 29 kDa was excised and eluted (300 l; ϪCarb, 210,000; ϩCarb, 69,000 cpm). This material, resuspended in gel elution buffer supplemented with 1% Genapol C-100, was digested with 2 g of EndoAsp-N (1 week at 25°C). B, reversed-phase HPLC purification of the digests with 8% of each fraction was counted. Also shown are the absorbance at 215 nm (O) and the fluorescence (⅐⅐⅐) profiles (ϪCarb) as well as the % solvent B (---). C, 3 H and mass release from sequence analysis of HPLC fraction 24 (q, 5,200 cpm loaded, 3,900 cpm left on the filter; E, 1,600 cpm loaded, 830 cpm left). Two ␦-subunit fragments were detected beginning at ␦Asp-99 (Ⅺ, ϪCarb, I 0 ϭ 40 pmol, R ϭ 92%; ϩCarb, I 0 ϭ 51 pmol, R ϭ 91%) and at ␦Asp-76 (ϪCarb, I 0 ϭ 27 pmol, R ϭ 84%). Also present were fragments from nAChR ␥-subunit beginning at ␥Asp-76 (ϪCarb, I 0 ϭ 40 pmol, R ϭ 87%) and ␥Glu-101 (ϪCarb, I 0 ϭ 17 pmol, R ϭ 80%) as well as from the Na ϩ /K ϩ -ATPase ␤-subunit beginning at Arg-147 (I 0 ϭ 170 pmol, R ϭ 90%). The sequence of the ␦Asp-99 peptide is shown at top. high affinity (K D ϭ 1 M) to the two ACh sites, and its interaction with the ion channel domain is 1000-fold weaker, as judged by the inhibition of binding of [ 3 H]H 12 -HTX. Whereas Bz 2 choline binding to the ACh site at the ␣-␥ interface is 40-fold weaker than that of dTC (K H ϭ 25 nM), it binds with similar affinity as dTC at the ␣-␦-binding site. Bz 2 choline is thus a potent competitive antagonist that upon UV irradiation will form a reactive intermediate of known structure (the excited state triplet ketone) that reacts via hydrogen abstraction and CH bond insertion (23). For Torpedo nAChR-rich membranes equilibrated with [ 3 H]Bz 2 choline, UV irradiation results in selective photoincorporation into ␥Leu-109 and the homologous residue in ␦-subunit, ␦Leu-111, amino acids within Loop E of the ACh-binding sites. These results, which define the orientation in the ACh-binding site of the para position of this substituted benzoylcholine ester, provide a first definition of the orientation of a choline ester competitive antagonist when bound in the ACh site.
When analyzed at the level of the intact subunits, [ 3 H]Bz 2 choline was specifically photoincorporated into the nAChR ␥and ␦-subunits at similar efficiencies, with specific incorporation into the ␣-subunit at less than 10% the efficiency of labeling of the ␥or ␦-subunit. The inefficient labeling of the ␣-subunit was surprising. Whereas the sulfhydryl specificity of [ 3 H]MBTA (4-(N-maleimido)-␣-benzyltrimethylammonium) (41)) restricts its reactivity to the ␣-subunits, containing the only known cysteines in the ACh-binding sites, the alkylating antagonist [ 3 H]lophotoxin also reacted only with the ␣-subunits (42), and for two competitive antagonist photoaffinity labels, photoincorporation in ␣-subunit either predominated ([ 3 H]DDF (43)) or was at least at similar efficiency ([ 3 H]dTC (3)) as in the ␥or ␦-subunit.
Under equilibrium conditions with ACh sites fully occupied by [ 3 H]Bz 2 choline, irradiation for 1 h resulted in specific pho-tolabeling of ϳ3-5% of ␥or ␦-subunit. This efficiency of photolabeling was ϳ10-fold higher than the labeling seen for [ 3 H]dTC (3) or [ 3 H]nicotine (12), but considerably lower than the stoichiometric labeling seen for benzophenone-based peptide photoaffinity probes of myosin light chain kinase (44) or of many peptide hormone receptors (reviewed in Refs. 23 and 25).
Within the nAChR ␥-subunit, ϳ90% of 3 H photoincorporation was inhibitable by Carb as well as dTC and ␣-BgTx. Thus the primary site of photolabeling is within the agonist-binding site. For the ␦-subunit, whereas ϳ70% of the 3 H incorporation was inhibitable by Carb, it was not inhibited by ␣-BgTx. The origin of this difference was identified by analyzing the distribution of 3 H in the ␦-subunit fragments produced by En-doLys-C (Fig. 6). The 3 H incorporation seen in the presence of ␣-BgTx was in an ϳ10-kDa fragment distinct from the ϳ20-kDa fragment that contains the Carb-inhibitable 3 H incorporation in ␦-subunit. The ␦-subunit labeling seen in the presence of ␣-BgTx is likely to be within the ion channel domain, as it is inhibitable by tetracaine, and EndoLys-C is known to produce a 10-kDa fragment beginning at the amino terminus of ␦M2 (38). These data suggest that for nAChRs in the resting state, To identify directly the amino acid(s) of the nAChR ␥-subunit that were specifically photolabeled by [ 3 H]Bz 2 choline, labeled ␥-subunit was first digested with EndoLys-C which produced a 26-kDa fragment that was isolated and digested with V8 protease. This double digestion strategy was necessary to eliminate considerable contamination by the Na ϩ /K ϩ -ATPase ␤-subunit, which originates from contaminating membrane fragments present in the nAChR-rich membrane preparations and which has a mobility similar to the nAChR ␥and ␦-subunits in our SDS-PAGE conditions (13). Reversed-phase HPLC of the V8 protease digest isolated a 3 H peak which contained a ␥-subunit fragment beginning at ␥Val-102, and 3 H release in cycle 8 identified ␥Leu-109 as the primary site of [ 3 H]Bz 2 choline photoincorporation in ␥-subunit (Fig. 5). This conclusion was confirmed by radiosequence analysis of a chymotryptic digest of the labeled ␥Val-102 fragment which revealed 3 H release in cycle 4 (i.e. cleavage after ␥Tyr-105) and lack of 3 H release in cycle 8 (Fig. 4). In addition, it was consistent with the fact that the labeled ϳ12-kDa ␥-subunit fragment produced by V8 protease also contained the site of endoglycosidase H-sensitive glycosylation in ␥-subunit, ␥Asn-141 (15).
A similar approach was utilized to identify ␦Leu-111 as the primary site of agonist-inhibitable [ 3 H]Bz 2 choline photoincorporation within the ␦-subunit. Digestion with EndoLys-C produced labeled fragments of 20 -29 kDa, which were isolated. Since ␥-subunit was known to contaminate the ␦-subunit iso- lated by preparative SDS-PAGE, the material isolated from the EndoLys-C digest could also contain the labeled ␥-subunit fragment of similar mobility. The ␦-subunit fragment was digested with EndoAsp-N to take advantage of the presence of ␦Asp-99. Sequence analysis of the peak of 3 H from the reversed-phase HPLC separation of the EndoAsp-N digest revealed the presence of a fragment beginning at ␦Asp-99, and the major peak of 3 H release was in cycle 13, consistent with labeling of ␦Leu-111, the homologous residue to ␥Leu-109 (Fig. 7). However, this HPLC fraction also contained fragments beginning at ␦Asp-76, ␥Asp-76, and ␥-Glu-101 as well as a fragment from the Na ϩ / K ϩ -ATPase ␤-subunit, and there was 3 H release in cycle 9 at a lower level than in cycle 13. Sequence analysis of the HPLC fractions on either side of the peak of 3 H revealed that the mass levels of only the ␦Asp-99 fragment correlated with the level of 3 H release in cycle 13 (Fig. 8). In addition, the levels of the fragment beginning at ␥Glu-101 did account for the 3 H release seen in the ninth cycle of Edman degradation. Despite the fact that the ␦Asp-99 fragment was not isolated from the contaminating fragments, the data support the conclusion that ␦Leu-111 is the primary site of specific [ 3 H]Bz 2 choline photoincorporation in ␦-subunit.
The levels of 3 H incorporation in ␥Leu-109 (16 cpm/pmol) and ␦Leu-111 (ϳ8 cpm/pmol) in subunit fragments isolated from the same labeling experiment indicate that they are labeled at similar efficiencies and that [ 3 H]Bz 2 choline is incorporated into ϳ2-4% of the subunits at that position. The labeled 20 -29-kDa fragments of the ␥and ␦-subunits produced by EndoLys-C began at ␥Glu-47 and ␦Glu-47, and sequence analysis of those fragments allow direct estimation of the levels of [ 3 H]Bz 2 choline incorporation in ACh-binding site Loop D, defined by ␥Trp-55 and Trp-57, the primary sites of [ 3 H]dTC photoincorporation in the ␥and ␦-subunits. No 3 H release above background was detected while sequencing through ␥Asn-61 or ␦Ala-61, which indicated that any incorporation in binding site Loop D is at Ͻ2% the efficiency of labeling of ␥Leu-109/␦Leu-111.
The selective photoincorporation of [ 3 H]Bz 2 choline in ␥Leu-109 and ␦Leu-111 is striking. These are the first amino acids identified by affinity labeling within the nAChR agonist-binding sites other than Tyr, Trp, or Cys, amino acids with intrinsic side chain reactivity. Benzophenone affinity probes have been shown to react especially well with Met, but also with a wide variety of amino acids including Tyr, Ser, Gly, and Pro (24). To our knowledge this is the first example of a benzophenone derivative reacting with a leucine residue within a native protein, presumably via CH bond insertion in the leucine side chain. Thus, the covalent incorporation of [ 3 H]Bz 2 choline into ␥Leu-109/␦Leu-111 certainly results from the proximity of the bound, excited state ketone with those amino acids. It is notable that there is no photolabeling of the adjacent ␥Tyr-111/ ␦Arg-113 which in Torpedo nAChR have been identified by [ 3 H]dTC affinity labeling and mutational analyses (15) as contributors to ACh-binding site Loop E. Also there is no photoincorporation in ␥Met-116/␥Tyr-117, positions that have been identified as dTC affinity determinants in the mouse nAChR ␥-subunit (17).
In addition to the photoincorporation in ␥Leu-109/␦Leu-111, there was also agonist-inhibitable photolabeling within ␣-subunit, albeit at Ͻ10% the efficiency in ␥or ␦-subunit. Identification of the amino acids in the ␣-subunit that are also in proximity to photoactivated Bz 2 choline will further define proximity relations in the ACh-binding site.
The results of photoaffinity labeling and mutational analyses have both provided consistent evidence that the ACh-binding sites are at subunit interfaces, with amino acids from ␥-(or ␦-) subunit making important contributions to binding sites for agonists and competitive antagonists. In contrast, available data concerning the structure of the nAChR based upon cryoelectron microscopy (1) have been interpreted to suggest that ACh sites are within the ␣-subunits, with the non-␣-subunits possibly contributing to the access way to the binding sites. The highly selective labeling of ␥Leu-109/␦Leu-111, in conjunction with the lack of efficient specific labeling of any amino acids in the ␣-subunit, provides further evidence that amino acids from the non-␣-subunits are directly involved in the structure of the ACh-binding sites.