[3H]Epibatidine Photolabels Non-equivalent Amino Acids in the Agonist Binding Site of Torpedo and α4β2 Nicotinic Acetylcholine Receptors*

Nicotinic acetylcholine receptor (nAChR) agonists, such as epibatidine and its molecular derivatives, are potential therapeutic agents for a variety of neurological disorders. In order to identify determinants for subtype-selective agonist binding, it is important to determine whether an agonist binds in a common orientation in different nAChR subtypes. To compare the mode of binding of epibatidine in a muscle and a neuronal nAChR, we photolabeled Torpedo α2βγδ and expressed human α4β2 nAChRs with [3H]epibatidine and identified by Edman degradation the photolabeled amino acids. Irradiation at 254 nm resulted in photolabeling of αTyr198 in agonist binding site Segment C of the principal (+) face in both α subunits and of γLeu109 and γTyr117 in Segment E of the complementary (−) face, with no labeling detected in the δ subunit. For affinity-purified α4β2 nAChRs, [3H]epibatidine photolabeled α4Tyr195 (equivalent to Torpedo αTyr190) in Segment C as well as β2Val111 and β2Ser113 in Segment E (equivalent to Torpedo γLeu109 and γTyr111, respectively). Consideration of the location of the photolabeled amino acids in homology models of the nAChRs based upon the acetylcholine-binding protein structure and the results of ligand docking simulations suggests that epibatidine binds in a single preferred orientation within the α-γ transmitter binding site, whereas it binds in two distinct orientations in the α4β2 nAChR.

Nicotinic acetylcholine receptor (nAChR) agonists, such as epibatidine and its molecular derivatives, are potential therapeutic agents for a variety of neurological disorders. In order to identify determinants for subtype-selective agonist binding, it is important to determine whether an agonist binds in a common orientation in different nAChR subtypes. To compare the mode of binding of epibatidine in a muscle and a neuronal nAChR, we photolabeled Torpedo ␣ 2 ␤␥␦ and expressed human ␣4␤2 Nicotinic acetylcholine receptors (nAChRs) 3 are prototypical members of the Cys loop superfamily of neurotransmitter-gated ion channels that mediate the actions of the neurotransmitter acetylcholine (1). nAChRs from vertebrate skeletal muscle and the electric organs of Torpedo rays are heteropentamers of homologous subunits with a stoichiometry of 2␣:␤: ␥(⑀):␦ that are arranged pseudosymmetrically around central cation-selective ion channels (1,2). There are 12 mammalian neuronal nAChR subunit genes: nine neuronal ␣ subunits (␣2-␣10) and three neuronal ␤ subunits (␤2-␤4). The ␣4␤2 nAChR is the most abundant and widely distributed nAChR subtype expressed in the brain and is a major target for potential therapeutic agents for neurological diseases and conditions, including nicotine dependence and Alzheimer and Parkinson diseases (3,4). Although the ratio of ␣4 to ␤2 subunit in vivo is uncertain, expressed receptors containing either three ␣4 or three ␤2 subunits have distinct pharmacological properties (5,6).
The agonist binding sites (ABS) of nAChRs are located within the amino-terminal extracellular domain at the interface of adjacent subunits (␣-␥ and ␣-␦ in the Torpedo nAChR), and different nAChR subunit combinations form ABS with distinct physical and pharmacological properties (3,7). Affinity labeling studies with Torpedo nAChR and site-directed mutational analyses of muscle and neuronal nAChRs identified key amino acids delineating the ABS from three noncontiguous stretches of the ␣ subunit (Segments A-C, the principal component (ϩ face)) and three noncontiguous regions of the non-␣ subunit (Segments D-F, the complementary component (Ϫ face)) (8,9). The three-dimensional structure of the ABS in the absence and presence of nAChR agonists or competitive antagonists has been determined for snail acetylcholine-binding proteins (AChBPs) that are soluble homopentamers homologous to the extracellular (amino-terminal) domain of a nAChR (10 -12). In the AChBP, four aromatic amino acids from Segments A-C that are conserved within ␣ subunits, along with a conserved Trp in Segment D, form a core aromatic "pocket" with a dimension optimal for accommodation of a trimethylammonium group. The other amino acids in the non-␣ subunits closest to the aromatic pocket, which are generally not conserved among ␥, ␦, or neuronal ␤ subunits, are on three antiparallel ␤ strands. The AChBP structure was used to refine the structure of the Torpedo nAChR in the absence of agonist to 4 Å resolution (13). In this structure, there is a reorientation of Segments A-C, resulting in the absence of a well defined core aromatic binding pocket.
Analysis of agonist interactions with mutant nAChRs containing fluorine-substituted core aromatic residues provides evidence that cationinteractions, particularly with ␣Trp 149 in Segment B, are important determinants of agonist binding affinity (14) and for the higher affinity binding of nicotine to ␣4␤2 nAChRs compared with ␣ 2 ␤␥␦ nAChRs (15). Mutational analyses and molecular docking calculations have also provided evidence that two molecules of very similar structure may actually bind to a single receptor in very different orientations, as seen for two high affinity antagonists, D-tubocurarine and its quaternary ammonium analog metocurine, binding to the AChBP and to the muscle nAChR (16,17).
Photoaffinity labeling provides an alternative means to identify amino acids contributing to a drug binding site (18,19) and has been used to determine the orientation of drugs bound in the ABS of Torpedo nAChR (20). Epibatidine binds with very high affinity (ϳ10 pM) to heteromeric neuronal nAChRs (e.g. ␣4␤2) and with nanomolar affinity to ␣7 and muscle-type/Torpedo nAChRs (3). Utilizing a photoreactive analogue of epibatidine (azidoepibatidine; Fig. 1) and mass spectrometry, Tomizawa et al. (21) identified photolabeled amino acids in the Aplysia AChBP (Tyr 195 in Segment C and Met 116 in Segment E), establishing an orientation for bound azidoepibatidine consistent with the orientation of epibatidine in an AChBP crystal structure (12).
In this report, we use [ 3 H]epibatidine as a photoaffinity reagent to identify the amino acids photolabeled in an expressed ␣4␤2 nAChR and in the Torpedo ␣ 2 ␤␥␦ nAChR. Comparisons of the labeled amino acids seen in the Torpedo nAChR ␣-␥ binding site and in the ␣4␤2 nAChR, in conjunction with the results of docking calculations for epibatidine binding to homology models of the ␣ 2 ␤␥␦ and ␣4␤2 nAChRs, suggests that epibatidine binds in a single orientation in the ␣-␥ site but in two orientations in the ␣4␤2 ABS.
Preparation of Torpedo nAChR-Torpedo californica nAChRrich membranes for radioligand binding studies and for affinity purification were isolated from frozen electric organs (Aquatic Research Consultants, San Pedro, CA), as described previously (22). Torpedo nAChR-rich membranes at 1 mg/ml protein were solubilized in 1% sodium cholate in vesicle dialysis buffer (VDB; 100 mM NaCl, 0.1 mM EDTA, 0.02% NaN 3 , 10 mM MOPS, pH 7.5) and treated with 0.1 mM diisopropylfluorophosphate after insoluble material was pelleted by centrifugation (91,000 ϫ g for 1 h). The nAChR was affinity-purified on a bromoacetylcholine bromide-derivatized Affi-Gel 10 column and then reconstituted into lipid vesicles composed of dioleoyl phosphatidic acid/dioleoyl phosphatidylcholine/cholesterol (at a molar ratio of 3:1:1), as described (23,24). The lipid/nAChR ratio was adjusted to molar ratio of 400:1. Based upon SDS-PAGE, after purification, the nAChR comprised more than 90% of the protein in the preparation. Both the nAChR-rich membranes and purified nAChRs were stored at Ϫ80°C.
Radioligand Binding to Torpedo nAChR-rich Membranes-The effect of epibatidine and dTC on the binding of [ 3 H]epibatidine to Torpedo nAChR-rich membranes was determined using a centrifugation assay with duplicate samples. Membranes at 0.1 mg/ml protein (16 nM ACh binding sites) in Torpedo physiological saline (250 mM NaCl, 5 mM KCl, 3 mM CaCl 2 , 2 mM MgCl 2 , and 5 mM sodium phosphate, pH 7.0) were incu- Bound and free [ 3 H]epibatidine were separated by centrifugation (39,000 ϫ g for 1 h) and then quantified by liquid scintillation counting. Nonspecific binding was determined in the presence of 1 mM carbamylcholine.
Data Analysis-The concentration-dependent inhibition of [ 3 H]epibatidine binding by epibatidine and dTC were fit according to a single-site model as follows, and to a two-site model, where f (x) is the [ 3 H]epibatidine binding in the presence of competitor concentration x, T is the total specific binding, NS is the nonspecific binding determined in the presence of 1 mM carbamylcholine, and IC 50 is the concentration of competitor that inhibits 50% of the total specific binding. Sigmaplot version 11 (Systat Software) was used for non-linear least squares fit of the data, and the S.E. values of the parameter fits are indicated. . After a 2-h incubation at room temperature, the samples in glass test tubes were irradiated for 30 min with a 254-nm UV lamp (Spectroline EN-280L). The labeled membranes were then pelleted by centrifugation (39,000 ϫ g for 1 h), resuspended in electrophoresis sample buffer (12.5 mM Tris-HCl, 2% SDS, 8% sucrose, 1% glycerol, 0.01% bromphenol blue, pH 6.8), and resolved on 1-mm-thick, 8% polyacrylamide, 0.33% bisacrylamide gels (27). After staining with Coomassie Blue R-250 and destaining to visualize bands, gels were impregnated with fluor (Amplify; GE Biosciences) for 30 min, dried, and exposed to Eastman Kodak Co. X-Omat LS film at Ϫ80°C (1-4-week exposure). For some 8% gels, following staining and destaining, gels were soaked in distilled water overnight, and bands corresponding to the ␣ and ␥ subunits were excised, soaked in overlay buffer (5% sucrose, 125 mM Tris-HCl, 0.1% SDS, pH 6.8) for 30 min, transferred to the wells of a 15% acrylamide mapping gel, and digested in gel with 10 g of V8 protease (28). After electrophoresis, mapping gels were processed for fluorography (3-6-week exposure) as described above. For both 8 and 15% mapping gels, to quantify the amount of 3 H cpm incorporated into nAChR subunits or subunit proteolytic fragments, bands were excised from the gel, transferred to 5-ml scintillation vials, and soaked in 0.5 ml of 0.1% SDS for 4 days with occasional mixing. Then 3 ml of liquid scintillation mixture was added, and samples were counted for 5 min.
Reversed-phase HPLC Purification and Sequence Analysis-Prior to sequence analysis, all of the [ 3 H]epibatidine-labeled peptides were purified using reversed-phase HPLC (rpHPLC) on a Shimadzu LC-10A binary HPLC system, using a Brownlee Aquapore C 4 column (100 ϫ 2.1 mm). Solvent A was composed of 0.08% trifluoroacetic acid in water, and Solvent B contained 0.05% trifluoroacetic acid in 60% acetonitrile, 40% 2-propanol. A non-linear elution gradient at 0.2 ml/min was employed (25-100% Solvent B in 100 min, shown as dotted line in the figures), and fractions were collected every 2.5 min (40 fractions/run). The elution of peptides was monitored by the absorbance at 210 nm, and the amount of 3 H associated with each fraction was determined by liquid scintillation counting of 5% aliquots.
For sequence analysis, rpHPLC fractions containing peaks of 3 H were pooled, diluted 3-fold with 0.1% trifluoroacetic acid, and loaded onto polyvinylidene difluoride filters using Prosorb sample preparation cartridges (catalog number 401959; Applied Biosystems). The filters were then treated with Biobrene, as recommended by the manufacturer. Sequencing was performed on an Applied Biosystems PROCISE TM 492 protein sequencer configured to utilize one-sixth of each cycle of Edman degradation for amino acid quantification and to collect the other five-sixths for 3 H counting. To determine the amount of sequenced peptide, the pmol of each amino acid in a detected sequence was quantified by peak height and fit to the equation f(x) ϭ I 0 R x , where I 0 represents the initial amount of the peptide sequenced (in pmol), R is the repetitive yield, and f(x) is the pmol detected in cycle x. Ser, His, Trp, and Cys were not included in the fits due to known problems with their accurate detection/quantification. The fit was calculated in Sigma-Plot 11 using a non-linear least squares method, and figures containing 3 H release profiles include this fit as a dotted line. Some sequencing samples were treated with o-phthalaldehyde (OPA) prior to a cycle known to contain a proline (29). OPA reacts with all amino-terminal amino acids (but not with the imino acid proline) and blocks further Edman degradation (30). Thus, release of 3 H in a cycle after an OPA treatment establishes that the 3 H release originates from a peptide with a proline in the OPA-treated cycle. Quantification of 3 H incorporated into a specific residue (cpm/pmol) was calculated by Molecular Modeling-Models of the extracellular domain of the human (␣4) 2 (␤2) 3 and Torpedo nAChR were constructed from the x-ray structure of the epibatidine-bound form of the Aplysia AChBP (12) (Protein Data Bank code 2BYQ) using the Discovery Studio (Accelrys, Inc.) software package. Epibatidine (volume 156 Å 3 ) was docked into the ABS of the Aplysia AChBP crystal structure and in the models of the Torpedo ␣-␥ and human ␣4-␤2 ABS using CDOCKER (31,32), a CHARMmbased (33) molecular dynamics simulated annealing program that treats the ligand as fully flexible while maintaining a rigid receptor. In each docking experiment, 50 replicas of epibatidine (protonated form) were seeded in random orientations within the ABS defined by a binding site sphere of 12 Å radius. For each starting seed, CDOCKER was used to generate 10 ligand conformations using high temperature molecular dynamics, and then the 10 lowest energy orientations were identified using random rigid body rotations, followed by simulated annealing and a full potential final minimization step. For visualization, we show in Fig. 7 the Connolly surface representations, defined by a 1.8-Å diameter probe, of the ensemble of the 20 docking solutions with the lowest CDOCKER interaction energies.

[ 3 H]Epibatidine
Binding to Torpedo nAChR-In equilibrium binding studies with Torpedo nAChR-rich membranes, epibatidine fully inhibited the binding of 0.5 nM [ 3 H]epibatidine, with the data well fit by a single-site model with an IC 50 of 11 Ϯ 2 nM ( Fig. 2A). We used the displacement of [ 3 H]epibatidine binding by dTC to explore the agonist binding site selectivity of epibatidine. dTC binds with higher affinity (K d ϭ 35 nM) at the Torpedo ␣-␥ agonist binding site than the ␣-␦ site (K d ϭ 8 M) (34). The dTC displacement data (Fig. 2B, (Fig. 2B, open circles) with little effect on the relative occupancy by epibatidine of the ␣-␥ and ␣-␦ sites.

Photoincorporation of [ 3 H]Epibatidine into Torpedo nAChR-We
used SDS-PAGE followed by fluorography (Fig. 3A) and scintillation  counting of excised gel bands (Fig. 3B) to provide an initial characterization of the subunit and pharmacological specificity of 3 H incorporation when affinity-purified Torpedo nAChRs were photolabeled on an analytical scale (ϳ50-g samples). As seen in the fluorograph, [ 3 H]epibatidine was incorporated primarily in the nAChR ␣ and ␥ subunits (ϪEpi lane). Photolabeling was eliminated by the addition of an excess (40 M) of nonradioactive epibatidine (ϩEpi lane) and inhibited in a concentration-dependent manner by dTC. Based upon scintillation counting, the concentration dependence of dTC inhibition of 3 H photoincorporation in the ␥ subunit (Fig. 3B, open  circles) was well fit by a single-site model with an IC 50 value of 0.45 Ϯ 0.07 M, whereas inhibition of labeling in the ␣ subunit (Fig. 3B, solid circles)

Identification of Amino Acids Photolabeled by [ 3 H]Epibatidine in the Torpedo ␣ and ␥ Subunits-To identify amino acids photolabeled by [ 3 H
]epibatidine within the ␣ and ␥ subunits, the incorporation in each subunit was first mapped by in-gel digestion with V8 protease, which for the ␣ subunit produces fragments of ϳ20 kDa (␣V8-20, beginning at ␣Ser 173 and containing ACh binding site Segment C), ϳ18 kDa (␣V8-18, beginning at ␣Thr 52 and containing binding site Segments A and B), and ϳ10 kDa (␣V8-10, beginning at ␣Asn 338 and containing the M4 membrane-spanning helix (35)). The corresponding fluorograph of a mapping gel (Fig. 4) established that all detectable (and specific) labeling in the ␣ subunit was contained within a Coomassie-stained band of ϳ20 kDa (␣V8-20), whereas, based upon liquid scintillation counting, the labeling in ␣V8-18 was less than 5% that of ␣V8-20 (supplementary Fig.  S1). For the ␥ subunit, all 3 H incorporation was contained within an ϳ14 kDa band (␥V8-14).
To identify the residues photolabeled by [ 3 H]epibatidine, ␣V8-20 and ␥V8-14 were isolated from affinity-purified Torpedo nAChRs photolabeled on a preparative scale (3.5 mg). Labeled ␣V8-20 was digested with EndoLys-C, which cleaves ␣V8-20 after ␣Lys 185 (36). When the digest was fractionated by rpHPLC, ϳ90% of the recovered 3 H eluted in a single peak at ϳ87% solvent B (supplemental Fig. S1). Sequence analysis of the peak fractions (Fig. 5A) revealed a primary sequence beginning at ␣His 186 (23 pmol) and a single peak of 3 H release in cycle 13, corresponding to labeling of ␣Tyr 198 (28 cpm/pmol), one of two tyrosines in Segment C contributing to the core aromatic binding pocket for ACh. During sequencing, the sample was treated after cycle 8 (see the arrow in Fig. 5A) with OPA, which reacts with primary, but not secondary, amines and blocks Edman degradation of any peptide without an amino-terminal proline at the time of addition (29,30). The only sequence detected after cycle 8 was the primary sequence (with ␣Pro 194 in cycle 9), providing additional evidence that the 3 H release in cycle 13 resulted from [ 3 H]epibatidine incorporation into ␣Tyr 198 .

Photoincorporation of [ 3 H]Epibatidine into
␣4␤2 nAChR-Affinity-purified neuronal ␣4␤2 nAChRs were photolabeled on an analytical scale (ϳ50-g samples) with 880 nM [ 3 H]epibatidine in the absence and presence of excess nonradioactive epibatidine (40 M), and the 3 H incorporation was assessed by SDS-PAGE and fluorography (Fig. 6). [ 3 H]epibatidine photoincorporated primarily into the ␣4 subunit and in a broad band migrating with an apparent molecular mass of ϳ36 kDa, with low level labeling of the ␤2 subunit, and labeling in each band was fully inhibited by epibatidine. Based upon liquid scintillation counting of excised gel bands, 1,600, 550, and 2,400 cpm were incorporated into the ␣4 and ␤2 subunits and 36 kDa band, respectively, and photolabeling in each band was inhibited by Ͼ95% by excess nonradioactive epibatidine. Based on previous studies with purified ␣4␤2 nAChRs (26), the [ 3 H]epibatidine-labeled material migrating at ϳ36 kDa probably included proteolytic fragments of both the ␣4 and ␤2 subunits, which was confirmed by protein sequencing (supplemental Fig. S2).

Identification of Amino Acids Photolabeled by [ 3 H]Epibati
dine in the ␣4␤2 nAChR-To identify specific residues labeled by [ 3 H]epibatidine in the ␣4 and ␤2 subunits, each subunit was isolated from ␣4␤2 nAChRs labeled on a preparative scale (1.2 mg of nAChR; 800 nM [ 3 H]epibatidine). Labeled subunits were digested with V8 protease for 4 days, and the digests were fractionated by rpHPLC. When the ␣4 subunit digest was fractionated by rpHPLC (Fig. 7A), the 3 H eluted in peaks centered at fraction 25 (ϳ50% solvent B) and fraction 28 (ϳ57% solvent B) and in the column flow-through. 4 Sequence analysis of the pool of fractions 24 -26 (Fig. 7B) revealed a fragment beginning at ␣4Trp 181 (5 pmol) as well as the amino terminus of V8 protease, which was the primary sequence. The major peak of 3 H release in cycle 15 corresponded to labeling of ␣4Tyr 195 (10 cpm/ pmol), a residue that is contained within Segment C of the ABS and is equivalent to ␣Tyr 190 of the muscle-type nAChR.

DISCUSSION
In this report, we use the intrinsic photoreactivity of the nAChR agonist [ 3 H]epibatidine to compare its mode of binding in the Torpedo and neuronal ␣4␤2 nAChRs. Although the reactive intermediates formed upon photolysis of epibatidine have not been directly identified, halo pyridines, such as the chlorinated pyridine ring of epibatidine, are known to undergo photoaddition reactions initiated by the cleavage of the C-Cl bond (37)(38)(39), and it is that carbon that is probably reactive in epibatidine. Irradiation at 254 nm results in [ 3 H]epibatidine photoincorporation into the ␣4-␤2 and Torpedo ␣-␥ ABS, with little or no labeling detected in regions outside of that domain. Likely contributors to the high specificity of [ 3 H]epibatidine labeling , and the stained ␣ and ␥ subunits bands were excised and digested in gel with V8 protease on a second gel, as described under "Experimental Procedures." Material eluted from ␣V8-20 was digested with EndoLys-C and then fractionated by HPLC, whereas material eluted from ␥V8-14 was purified directly by HPLC (supplemental Fig. S1). A, 3 H (F) and PTH-derivatives (Ⅺ) released during amino acid sequence analysis of the 3 H peak (fractions 34 -36; 31,000 cpm) from the HPLC purification of the EndoLys-C digest of ␣V8-20. During sequencing, the filter was treated with OPA before cycle 9 (indicated by an arrow) to chemically isolate the primary peptide detected (␣His 186 , I 0 ϭ 23 Ϯ 9 pmol, r ϭ 96%, with ␣Pro 194 in cycle 9) by preventing further sequencing of fragments not containing a proline in this cycle (29,30). The only sequence detected after cycle 8 was the primary sequence, establishing that the 3 H release in cycle 15 resulted from [ 3 H]epibatidine incorporation into ␣Tyr 198 (28 cpm/pmol). B, 3 H (F) and PTH-derivatives (Ⅺ) released during amino acid sequence analysis of the 3 H peak (fractions 28 -30; 22,500 cpm) from the HPLC purification of ␥V8-14. The only fragment detected began at ␥Val 102 (I 0 ϭ 55 Ϯ 5 pmol, r ϭ 94%) and was present at Ͼ20-fold higher amount than any other sequences. The 3 H releases in cycles 8 and 16 correspond to photolabeling of ␥Leu 109 (23 cpm/ pmol) and ␥Tyr 117 (13 cpm/pmol). include the very high binding affinity to ␣4␤2 nAChRs (ϳ100 pM) (40) and the molecular rigidity of epibatidine. In contrast to photolabeling studies conducted with the iodo-analog of epibatidine ([ 125 I]epibatidine) (41), the amino acids in the ABS of the ␣4␤2 and Torpedo nAChRs that reacted with [ 3 H]epibatidine were readily identified by Edman degradation.
In the Torpedo nAChR, our results establish that [ 3 H]epibatidine binds to both the ␣-␥ and ␣-␦ ABS with high affinity (ϳ11 nM). Although epibatidine may bind with 3-4-fold higher affinity to one of the two sites in the Torpedo nAChR, as does ACh (42), it did not have the high selectivity between the sites seen in the mouse muscle nAChR (Ͼ170-fold higher affinity for the ␣-␥ than the ␣-␦ site in the desensitized state (43) (44).
In the ␣4␤2 nAChR, [ 3 H]epibatidine photolabeled the ␣4 subunit at ϳ3-fold higher efficiency than the ␤2 subunit, and consistent with this, [ 3 H]epibatidine photolabeled ␣4Tyr 195 (10 cpm/pmol; equivalent to Torpedo ␣Tyr 190 ) in Segment C more efficiently than the two labeled amino acids in the ␤2 subunit: ␤2Val 111 (2 cpm/pmol) and ␤2Ser 113 (ϳ0.6 cpm/pmol) in Segment E (equivalent to Torpedo ␥Leu 109 and ␥Tyr 111 , respectively). Tomizawa et al. (21) reported that [ 3 H]azidoepibatidine also photolabeled the ␣4 subunit more efficiently (no labeling of the ␤2 subunit was evident by fluorography), although the photolabeled amino acids were not identified. Although Proposed Orientations of Epibatidine in the Torpedo and ␣4␤2 nAChR Agonist Binding Sites from Photolabeling and Molecular Modeling-As one approach to identify factors that could explain the selective labeling of non-equivalent core aromatic amino acids within Segment C of the Torpedo (␣Tyr 198 ) and ␣4␤2 (␣4Tyr 195 ) ABS by what must be the same epibatidine photoreactive intermediate, we used computational methods to predict favored epibatidine docking orientations in the ABS in homology models of the Torpedo and ␣4␤2 nAChRs based on the structure of the epibatidine-bound form of Aplysia AChBP (12). When epibatidine was docked in the Torpedo ␣-␥ ABS, a single binding orientation was highly favored (Fig. 8A) that was essentially the same as epibatidine in the crystal structure of the ligand-bound form of AChBP (12) and as we found for epibatidine docked in the AChBP crystal structure (not shown) or had been found for epibatidine docked in a structural model of chick ␣7 nAChR (45). In this orientation, the azabicycloheptane ring occupies the "aromatic box" formed by ␣Tyr 93 , ␣Tyr 190 , ␣Try 198 , ␣Trp 149 , and ␥Trp 55 , and the chloropyridyl ring is oriented toward Segment E, with C6 of the pyridine ring (the most likely reactive site) positioned within 6 Å of the labeled ␣Try 198 , ␥Tyr 117 , and ␥Leu 109 but 11 Å from the unlabeled ␣Tyr 190 (Fig. 8B).
When epibatidine was docked into the ABS of the ␣4␤2 nAChR homology model, two distinct binding orientations  A and B) and ␤2 (C and D) subunits isolated from [ 3 H]epibatidine-labeled affinity-purified ␣4␤2 nAChR (1.2 mg ␣4␤2 nAChR, 800 nM [ 3 H]epibatidine). The elution of peptides during HPLC was monitored by absorbance at 210 nm (solid line), and 3 H elution was quantified by liquid scintillation counting of 5% of each fraction (F). B, 3 H (F) and PTH-derivatives (E) released during amino acid sequence analysis of 3 H peak (fractions 24 -26; 4,200 cpm) from the HPLC purification of the V8 protease digest of the ␣4 subunit. The primary peptides detected began at ␣4Trp 181 (I 0 ϭ 5 Ϯ 0.3 pmol, r ϭ 90%) and at the amino terminus of V8 protease (ϳ10 pmol), with a peak of 3 H release in cycle 15 corresponding to labeling of ␣4Tyr 195 (10 cpm/pmol). For the material in HPLC fraction 28, which coeleuted with a major peak of UV absorbance, the amino terminus of V8 protease was the dominant sequence, and no 3 H release was detected in 30 cycles of Edman degradation (not shown). D, 3 H (F) and PTH-derivatives (Ⅺ) released during amino acid sequence analysis of the 3 H peak (fractions 27-29; 3,100 cpm) from the HPLC purification of the V8 protease digest of the ␤2 subunit. The primary peptides detected began at ␤2Val 104 (I 0 ϭ 19 Ϯ 3 pmol, r ϭ 89%) and at the amino terminus of V8 protease, with 3 H release in cycles 8 and 10 corresponding to labeling of ␤2Val 111 (1.9 cpm/pmol) and ␤2Ser 113 (ϳ0.3 cpm/pmol).
were predicted with similar energy and frequency: one orientation similar to that seen in the Torpedo ␣-␥ ABS ( Fig. 8C; up orientation with reference to the position of chlorine) and a second orientation with epibatidine rotated by ϳ180°and translated by 3 Å, orienting the chloropyridyl ring toward the core aromatic side chains ( Fig. 8E; down orientation). In both orientations, the positively charged nitrogen in the azabicycloheptane ring is located 2.6 Å from the backbone carbonyl group of ␣4Trp 143 and ϳ4.6 Å from the aromatic side chain (supplemental Fig. S3), preserving equivalent potential for the hydrogen bonding and/or cationinteractions predicted to be important determinants of binding affinity (9,12,14,15). In both orientations, the positively charged nitrogen is also ϳ4 Å from ␣4Tyr 202 , which was not photolabeled by [ 3 H]epipabitidine. In the up orientation (Fig. 8C), the pyridyl C6 is positioned ϳ5 Å from the labeled ␤2Val 111 but 11 Å from the labeled ␣4Tyr 195 (Fig. 8D). In the down orientation (Fig. 8E), the pyridyl C6 is positioned ϳ5 Å from the labeled ␣4Tyr 195 but 12 Å from the labeled ␤2Val 111 (Fig. 8F). These simple proximity relations suggest that epibatidine is likely to bind in an up orientation when ␤Val 111 is labeled and in a down orientation when ␣4Tyr 195 is photolabeled.
Although only a single binding orientation has been seen for epibatidine and other agonists and antagonists in the Aplysia AChBP crystal structures (11,12), mutational analyses have provided evidence that the competitive antagonist dTC and its FIGURE 8. Molecular models of epibatidine docked in the Torpedo ␣-␥ and ␣4␤2 nAChR agonist binding sites. Epibatidine was docked into the ABS of the Torpedo ␣-␥ (A and B) and human ␣4-␤2 (C-F) homology models using CDOCKER, as described under "Experimental Procedures." A, C, and E, views of the ␣-␥ (A) and ␣4-␤2 (C and E) ABS in a flat ribbon representation (gold, ␣ and ␣4; cyan, ␥; pink, ␤2) showing Connolly surface of the ensemble of the 20 lowest energy solutions of epibatidine, which docked in a single orientation in the ␣-␥ ABS and in two orientations in the ␣4-␤2 ABS, one orientation (C) similar to that in the ␣-␥ ABS, and a second orientation rotated by ϳ180°and translated by 3 Å. In each image, the lowest energy epibatidine orientation and amino acids within the ABS are shown in stick format with the [ 3 H]epibatidine-labeled amino acids colored red and unlabeled amino acids in blue. B, D, and F, the distances in Å between the C6 of the chloropyridyl ring of epibatidine and amino acids within the ABS for A, C, and E, respectively. The colors reflect atom type: carbon (black), oxygen (red), nitrogen (blue), and chlorine (green). See supplemental Fig. S3 for an alternative view of ␣4␤2 ABS that highlights the orientations of epibatidine relative to ␣4Trp 143 . quaternary ammonium analog, metocurine, bind in distinctly different orientations in both the AChBP and the muscle nAChR (16,17). Moreover, the prediction of two distinct epibatidine binding orientations in the ␣4␤2 ABS parallels the prediction that dTC and metocurine can bind in distinct orientations in the AChBP and human ␣-⑀ ABS, respectively (16,17). Our interpretation is based upon a plausible, but unproven, assumption that cleavage of the C-Cl bond of epibatidine produces the only reactive intermediate, and it is also possible that the differences in the patterns of [ 3 H]epibatidine photolabeling between the Torpedo and ␣4␤2 nAChRs result from differences in the structures of the transmitter binding sites between the two nAChRs, including the position and orientations of the core aromatic amino acids. However, the proposal that nAChR ligands bind in a "normal" and an "inverted" orientation provides a plausible explanation for our photolabeling results, as for photolabeling studies of neonicotinoids binding to the Lymnaea AChBP (46).