Identification of Residues in the Neuronal α7Acetylcholine Receptor That Confer Selectivity for Conotoxin ImI*

To identify residues in the neuronal α7 acetylcholine subunit that confer high affinity for the neuronal-specific toxin conotoxin ImI (CTx ImI), we constructed α7-α1 chimeras containing segments of the muscle α1 subunit inserted into equivalent positions of the neuronal α7 subunit. To achieve high expression in 293 human embryonic kidney cells and formation of homo-oligomers, we joined the extracellular domains of each chimera to the M1 junction of the 5-hydroxytryptamine-3 (5HT-3) subunit. Measurements of CTx ImI binding to the chimeric receptors reveal three pairs of residues in equivalent positions of the primary sequence that confer high affinity of CTx ImI for α7/5HT-3 over α1/5HT-3 homo-oligomers. Two of these pairs, α7Trp55/α1Arg55 and α7Ser59/α1Gln59, are within one of the four loops that contribute to the traditional non-α subunit face of the muscle receptor binding site. The third pair, α7Thr77/α1Lys77, is not within previously described loops of either the α or non-α faces and may represent a new loop or an allosterically coupled loop. Exchanging these residues between α1 and α7subunits exchanges the affinities of the binding sites for CTx ImI, suggesting that the α7 and α1 subunits, despite sequence identity of only 38%, share similar protein scaffolds.

To identify residues in the neuronal ␣ 7 acetylcholine subunit that confer high affinity for the neuronal-specific toxin conotoxin ImI (CTx ImI), we constructed ␣ 7 -␣ 1 chimeras containing segments of the muscle ␣ 1 subunit inserted into equivalent positions of the neuronal ␣ 7 subunit. To achieve high expression in 293 human embryonic kidney cells and formation of homo-oligomers, we joined the extracellular domains of each chimera to the M1 junction of the 5-hydroxytryptamine-3 (5HT-3) subunit. Measurements of CTx ImI binding to the chimeric receptors reveal three pairs of residues in equivalent positions of the primary sequence that confer high affinity of CTx ImI for ␣ 7 /5HT-3 over ␣ 1 /5HT-3 homooligomers. Two of these pairs, ␣ 7 Trp 55 /␣ 1 Arg 55 and ␣ 7 Ser 59 /␣ 1 Gln 59 , are within one of the four loops that contribute to the traditional non-␣ subunit face of the muscle receptor binding site. The third pair, ␣ 7 Thr 77 / ␣ 1 Lys 77 , is not within previously described loops of either the ␣ or non-␣ faces and may represent a new loop or an allosterically coupled loop. Exchanging these residues between ␣ 1 and ␣ 7 subunits exchanges the affinities of the binding sites for CTx ImI, suggesting that the ␣ 7 and ␣ 1 subunits, despite sequence identity of only 38%, share similar protein scaffolds.
The two neurotransmitter binding sites of muscle nicotinic acetylcholine receptors (AChR) 1 are generated by apposition of pairs of nonequivalent subunits, ␣ 1 /␦, ␣ 1 /␥, and ␣ 1 /⑀. By contrast, the binding sites of ␣ 7 neuronal nicotinic receptors are generated by apposition of pairs of identical subunits, ␣ 7 /␣ 7 (1). Because only the ␣ 7 subunit contributes to both faces of the ligand binding site, one can study the traditional ␣ and non-␣ faces by mutagenesis of a single ␣ 7 cDNA.
Ligand affinities of ␣ 7 neuronal and muscle AChRs differ owing to the different subunits that form their binding site interfaces. For example, the muscle-specific ␣-conotoxins MI, GI, and SI bind with high affinity to muscle receptors, whereas they bind with low affinity to ␣ 7 neuronal receptors (2). On the other hand, ␣-conotoxin ImI (CTx ImI) binds with high affinity to ␣ 7 receptors but binds with low affinity to muscle receptors (3). As the only known ␣ 7 -specific ␣-conotoxin, CTx ImI is a valuable probe of the homo-oligomeric ␣ 7 binding site.
Understanding of the ␣ 7 binding site has been limited by low expression of ␣ 7 receptors in mammalian cell lines (4,5). Part of the problem appears due to cell type, as neuronal cell lines promote expression of ␣ 7 receptors more efficiently than nonneuronal cell lines (6). The sequence of the subunit also affects expression, as chimeras derived from ␣ 7 and 5HT-3 subunits express high levels of functional homo-oligomers in non-neuronal cells (7). Joining the ␣ 7 extracellular domain to the M1 junction of 5HT-3 permits expression in 293 HEK cells and preserves the pharmacology of the ␣ 7 binding site (7). Thus, inserting portions of 5HT-3 sequence is a powerful tool to express receptors with an intact ␣ 7 binding site.
Studies of the muscle AChR have led to a basic scaffold hypothesis to account for observations that residues in equivalent positions of the homologous ␥, ⑀, and ␦ subunits contribute similarly to ligand affinity (8 -10). The hypothesis postulates that owing to their high degree of homology, these subunits fold into similar peptide scaffolds such that residues equivalent in the linear sequence occupy equivalent positions in three-dimensional space. Thus, ligand affinity for a particular site is dictated by differences in primary structure rather than differences in secondary or tertiary structures.
The primary sequence of the ␣ 7 subunit reveals conserved residues that contribute to both the ␣ and non-␣ faces of the ligand binding site. Within the ␣ face of the binding site, ␣ 7 and ␣ 1 share conserved aromatic residues that stabilize agonists, including ␣ 7 Tyr 92 , ␣ 7 Trp 148 , ␣ 7 Tyr 187 , and ␣ 7 Tyr 195 (11)(12)(13). On the other hand, ␣ 7 and non-␣ muscle subunits (␥, ⑀, and ␦) share the conserved ␣ 7 Trp 55 , which contributes to binding of agonists and antagonists (14,15). Thus, despite only 31-38% sequence homology with muscle subunits, ␣ 7 subunits maintain conserved residues that contribute to both faces of the ligand binding site.
The experiments described herein identify residues of the ␣ 7 binding site that determine selectivity for the competitive antagonist CTx ImI and examine the question of whether ␣ 7 neuronal and ␣ 1 muscle subunits form similar protein scaffolds. By constructing chimeras composed of ␣ 7 and ␣ 1 subunits, we identified three pairs of residues in equivalent positions of the subunits that confer selectivity of CTx ImI for binding sites formed from ␣ 7 versus ␣ 1 subunits. Moreover, exchanging these three selectivity determinants between ␣ 7 and ␣ 1 subunits exchanges the affinity conferred by the subunit, indicating that ␣ 7 and ␣ 1 subunits share similar protein scaffolds.

EXPERIMENTAL PROCEDURES
Materials-125 I-Labeled ␣-bungarotoxin (␣-bgt) was purchased from NEN Life Science Products, d-tubocurarine chloride from ICN Pharmaceuticals, (ϩ)-epibatidine and methyllycaconitine from Research Biochemicals, 293 human embryonic kidney cell line (293 HEK) from the American Type Culture Collection, and unlabeled ␣-bgt from Sigma. Human ␣ 7 and rat 5HT-3 subunit cDNAs were generously provided by Drs. John Lindstrom and William Green. Sources of the human acetyl-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Receptor Biology Laboratory, Dept. of Physiology and Biophysics, Mayo Foundation, 200 First St. S.W., Rochester, Minnesota 55905. Tel.: 507-284-5612; Fax: 507-284-9420; E-mail: sine.steven@mayo.edu. 1 The abbreviations used are: AChR, acetylcholine receptor; CTx ImI, choline receptor subunit cDNAs were as described previously (16). Synthesis and Purification of Conotoxin ImI-Conotoxin ImI was synthesized by standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry on an Applied Biosystems 431A peptide synthesizer. During synthesis, cysteine (S-triphenylmethyl)-protecting groups were incorporated at cysteines 3 and 12, and acetamidomethyl-protecting groups were incorporated at cysteines 2 and 8. The linear peptide was purified by reversed phase high performance liquid chromatography using a Vydac C18 preparative column with trifluoroacetic acid/acetonitrile buffers. Two intramolecular disulfide bridges were formed as follows: the cysteine S-triphenylmethyl-protecting groups of cysteines 3 and 12 were removed during trifluoroacetic acid cleavage of the linear peptide from the support resin, and the peptide was oxidized by molecular oxygen to form the 3-12 disulfide by stirring in 50 mM ammonium bicarbonate buffer, pH 8.5, at 25°C for 24 h. The peptide was lyophilized prior to the formation of the second bridge. The acetamidomethylprotecting groups on cysteine 2 and 8 were removed oxidatively by iodine as described (17) except that the peptide/iodine reaction was allowed to progress 16 h prior to carbon tetrachloride extraction. Residual iodine was separated from the pure product by high performance liquid chromatography, and CTx ImI was characterized by mass spectrometry.
Mutagenesis and Expression in HEK Cells-Acetylcholine receptor subunit cDNAs were subcloned into the cytomegalovirus-based expression vector pRBG4 (18). Mutant cDNAs were constructed by bridging naturally occurring or mutagenically installed restriction sites with double-stranded oligonucleotides. The chimeras are named as follows: the first subunit is the amino-terminal sequence of the chimera, the number following gives the position of the chimeric junction, and the final subunit gives the subunit from which the carboxyl-terminal sequence of the extracellular domain is taken. The extracellular domains of all chimeras are joined at M1 to the rat 5HT-3 sequence. Chimera ␣ 7 /5HT-3 (␣ 7 200/5HT-3) was constructed by bridging a 58-bp synthetic oligonucleotide from a TfiI site in ␣ 7 to an EcoRV site in rat 5HT-3. Chimera ␣ 1 /5HT-3 (␣ 1 205/5HT-3) was constructed by bridging a 69-bp synthetic oligonucleotide from a DraIII site in ␣ 1 to a StuI site in rat 5HT-3. All constructs were confirmed by dideoxy sequencing. HEK cells were transfected with wild type or mutant cDNAs using calcium phosphate precipitation as described (18). Two days after transfection, intact cells were harvested by gentle agitation in phosphate-buffered saline with 5 mM EDTA.
Ligand Binding Measurements-Ligand binding to intact cells was measured by competition against the initial rate of 125 I-labeled ␣-bgt binding (18). The cells were briefly centrifuged, resuspended in potassium Ringer's solution, and divided into aliquots for ligand binding. Potassium Ringer's solution contains 140 mM KCl, 5.4 mM NaCl, 1.8 mM CaCl 2 , 1.7 mM MgCl 2 , 25 mM HEPES, and 30 mg/liter bovine serum albumin, adjusted to a pH of 7.4 with 10 mM NaOH. Specified concentrations of ligand were added 30 min prior to addition of 3.75 nM 125 I-labeled ␣-bgt, which was allowed to bind for 15 min to occupy approximately half of the surface receptors. Binding was terminated by addition of 2 ml of potassium Ringer's solution containing 600 M of d-tubocurarine chloride. All experiments were performed at 24 Ϯ 2°C. Cells were harvested by filtration through Whatman GF-B filters using a Brandel cell harvester and washed three times with 3 ml of potassium Ringer's solution. Prior to use, filters were soaked in potassium Ringer's solution containing 4% skim milk. Nonspecific binding was determined in the presence of 10 nM ␣-bgt and was typically 1% of the total number of binding sites. The total number of binding sites was determined by incubation with toxin for 120 min. The initial rate of toxin binding was calculated as described previously (19) to yield the fractional occupancy of ligand. Binding measurements were analyzed according to the Hill equation: where Y is fractional occupancy of ligand, K app is the apparent dissociation constant and n H is the Hill coefficient. Parameter estimates and standard errors were obtained using UltraFit (BIOSOFT). For multiple experiments, means of the individual fitted parameters and standard deviations are presented (Tables I and II).
To measure time courses of ␣-bgt dissociation, receptors were incubated with 3.75 nM 125 I-labeled ␣-bgt for 120 min to achieve full occupancy, free 125 I-labeled ␣-bgt was removed, and 200-l aliquots of cells were filtered at specific times. Radioactivity at each time point was normalized to that of the maximum radioactivity at the time of toxin removal. Nonspecific binding was measured by incubation with 10 nM unlabeled ␣-bgt.
Sucrose Gradient Centrifugation-Two days after transfection, 293 HEK cells expressing various receptors were harvested by gentle agitation in phosphate-buffered saline and resuspended in high potassium Ringer's solution. Following labeling with 3.75 nM 125 I-labeled ␣-bgt for 2 h, cells were washed free of unbound radioactivity. Samples for sucrose gradient centrifugation were solubilized in 1 ml of Triton X-100 buffer (0.6% Triton X-100, 150 mM NaCl, 5 mM EDTA, 50 mM Tris, 35 g/ml phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 1 g/ml pepstatin A, pH 7.5). Extracts were layered on sucrose gradients (3-30%) and centrifuged for 22 h at 40,000 rpm, and fractions were collected and counted with a ␥ counter. Radioactivity in each fraction was normalized to that of the fraction containing the maximum radioactivity in each gradient.

RESULTS
Characterization of ␣ 7 /5HT-3 and ␣ 1 /5HT-3 Chimeric Receptors-Previous studies described construction of a chimera containing the extracellular domain of chick ␣ 7 joined to the M1 junction of the rat 5HT-3 subunit (␣ 7 201/5HT-3) (7). The studies further showed that addition of 5HT-3 sequence maintained ligand recognition properties of the native ␣ 7 binding site. We constructed a similar chimera by joining the extracellular domain of human ␣ 7 to the rat 5HT-3 subunit, with the chimera junction formed at position 200 (␣ 7 200/5HT-3) (Fig. 1A). To determine whether our human ␣ 7 /5HT-3 receptor has similar ligand recognition properties to wild type ␣ 7 , we expressed the constructs in 293 HEK cells and measured binding of agonists and antagonists by competition against the initial rate of 125 I-FIG. 1. Agonists and antagonists bind to wild type ␣ 7 and ␣ 7 / 5HT-3 receptors with similar affinities. Panel A is a schematic drawing of the wild type ␣ 7 and ␣ 7 /5HT-3 subunits: ␣ 7 /5HT-3 contains ␣ 7 sequence to position 200 followed by 5HT-3 sequence to the carboxyl terminus. Shaded portions represent 5HT-3 sequence. Panels B-D, 293 HEK cells were transfected with ␣ 7 or ␣ 7 /5HT-3 subunit cDNAs, and binding of CTx ImI, methyllycaconitine (MLA), or (ϩ)-epibatidine was determined as described under "Experimental Procedures." The curves through the data are fits to the Hill equation; means and S.E. of the fitted parameters are given in Table I. Expression of wild type ␣ 7 and ␣ 7 /5HT-3 surface receptors was typically 7 fmol and 6 pmol per 10-cm plate, respectively. labeled ␣-bgt binding. Although expression of ␣ 7 /5HT-3 receptors exceeds that of wild type ␣ 7 receptors by 1000-fold (see Fig.  1 legend), the competitive antagonists CTx ImI and methyllycaconitine and the agonist (ϩ)-epibatidine bind with identical affinities to the two types of receptors ( Fig. 1, B-D; Table I). Thus, the ␣ 7 ligand binding domain is preserved in ␣ 7 /5HT-3 receptors, and expression is greatly enhanced by addition of 5HT-3 sequence.
To investigate the basis of the specificity of CTx ImI for ␣ 7 receptors, we needed a subunit homologous to ␣ 7 and with low affinity for CTx ImI to serve as a frame of reference. We therefore constructed an analogous ␣ 1 /5HT-3 chimera by joining the ␣ 1 subunit extracellular domain to the M1 junction of the 5HT-3 receptor. When transfected into 293 HEK cells, the ␣ 1 /5HT-3 cDNA leads to expression of ␣-bgt binding sites on the cell surface. Moreover, CTx ImI binds 50-fold less tightly to ␣ 1 /5HT-3 than to ␣ 7 /5HT-3 receptors ( Fig. 2A). Similarly, CTx ImI binds 50-fold less tightly to adult human muscle receptors, further demonstrating neuronal specificity of CTx ImI ( Fig.  2A). Thus, the ␣ 1 /5HT-3 receptor provides a homologous muscle-like frame of reference for investigating specificity of CTx ImI for the ␣ 7 /5HT-3 receptor.
To determine whether surface receptors generated by ␣ 7 / 5HT-3 and ␣ 1 /5HT-3 are homo-oligomers containing five subunits, we labeled them with 125 I-labeled ␣-bgt and centrifuged the solubilized receptors on sucrose density gradients. We ran a parallel gradient containing the ␣ 2 ␤⑀␦ human muscle receptor to provide a 9S pentamer standard. The ␣ 7 /5HT-3 receptor migrates with a sedimentation coefficient just greater than our 9S muscle receptor standard, as described previously (20), and the profile was noticeably broader (Fig. 2B). The increased molecular weight and broad profile suggest either increased glycosylation or an additional 40 kDa due to binding of ␣-bgt to the five potential binding sites. The ␣ 1 /5HT-3 chimera co-migrates with our 9S muscle receptor standard. Both the ␣ 1 / 5HT-3 and human muscle receptors show significant 1.3S peaks due to free 125 I-labeled ␣-bgt, indicating dissociation of some of the ␣-bgt-receptor complexes during centrifugation. These results show that both ␣ 7 /5HT-3 and ␣ 1 /5HT-3 subunits form pentameric homo-oligomers on the cell surface.
To further investigate differences in stability of the ␣-bgtreceptor complexes suggested by sedimentation analysis, we compared time courses of 125 I-labeled ␣-bgt dissociation from ␣ 1 /5HT-3, ␣ 7 /5HT-3, adult mouse muscle, and adult human muscle receptors. 125 I-labeled ␣-bgt dissociates from ␣ 7 /5HT-3 and adult mouse muscle receptors with a single slow rate constant (t1 ⁄2 ϭ 20 h), whereas adult human muscle receptors show a more rapid single rate of dissociation (t1 ⁄2 ϭ 2.5 h). However, 125 I-labeled ␣-bgt dissociates from ␣ 1 /5HT-3 recep-tors in a biphasic manner, with a rapid component having a t1 ⁄2 of 13 min and a slow component having a t1 ⁄2 of 13.7 h. The amplitudes of the two components are approximately equal, indicating similar numbers of two classes of sites in the ␣ 1 / 5HT-3 receptor. Thus, the kinetics of 125 I-labeled ␣-bgt dissociation reveal quantitative differences in binding sites of ␣ 7 / 5HT-3, ␣ 1 /5HT-3, and muscle receptors.
Dissection of the 55-77 Segment-We constructed a series of stepwise chimeras to further localize selectivity determinants within the 55-77 segment. Starting with our reference chimera ␣ 7 55␣ 1 77␣ 7 , we maintained the ␣ 7 /␣ 1 junction at position 55 but shifted the carboxyl-terminal junction from position 77 to position 57 (Fig. 4). Surpassing only one mismatched residue, the chimera ␣ 7 55␣ 1 76␣ 7 increases CTx ImI affinity toward that of ␣ 7 , suggesting that the pair ␣ 7 Thr 77 /␣ 1 Lys 77 contributes to selectivity. Shifting the junction from position 76 to 59 produces no further change in affinity, but shifting from position 59 to 57 reveals an additional increase in affinity toward that of ␣ 7 . The last chimera in this series, ␣ 7 55␣ 1 57␣ 7 , falls 3-fold short of pure ␣ 7 -like affinity, indicating that residues 55-57 contribute the remaining increment of selectivity. Thus, the stepwise chimeras reveal at least three subsets of selectivity determinants within the 55-77 segment; one is between positions 55 and 57, the second is between positions 57 and 59, and the third is the pair ␣ 7 Thr 77 /␣ 1 Lys 77 .
Point Mutants of Selectivity Determinants-We further local-ized selectivity determinants by constructing point mutations in the ␣ 7 /5HT-3 cDNA. Beginning with the 55-57 segment, the point mutation ␣ 7 W55R decreased CTx ImI affinity by the same amount observed in the stepwise chimeras. Similarly, FIG. 3. CTx ImI binding to receptors formed by ␣ 7 -␣ 1 chimeras. Panel A is a schematic drawing of the ␣ 7 55␣ 1 77␣ 7 chimera, which contains ␣ 1 sequence from residues 55 to 77. The black portion represents ␣ 1 sequence; the unshaded portion represents ␣ 7 sequence; and the shaded portion represents 5HT-3 sequence. Comparison of ␣ 1 and ␣ 7 sequence in this segment indicates candidate residues that confer CTx ImI selectivity. Panel B, CTx ImI binding to intact cells expressing ␣ 7 /5HT-3, ␣ 1 /5HT-3, and ␣ 7 55␣ 1 77␣ 7 receptors.  4. Dissection of selectivity determinants for CTx ImI starting with the base chimera ␣ 7 55␣ 1 77␣ 7 . CTx ImI affinities for ␣ 1 /5HT-3 and ␣ 7 /5HT-3 receptors are shown by the vertical dashed lines, and the error bars indicate S.D. Affinities of the chimeric receptors are expressed as the log of the ratio of the dissociation constant of the chimera divided by that of the ␣ 7 /5HT-3 receptor. At right is a schematic representation of the chimeras in this series, with ␣ 7 sequence unshaded, ␣ 1 sequence black, and 5HT-3 sequence shaded. The text below indicates amino acid sequences in the 55-77 segment with ␣ 1 sequence in boldface and underlined, and ␣ 7 sequence in plain text. Data are means Ϯ S.D. of at least three experiments. mutating within the 57-59 segment, ␣ 7 S59Q decreases affinity by the same amount observed in the stepwise chimeras. Finally, the mutation ␣ 7 T77K decreases affinity as observed in the chimeras. Thus, three residues, ␣ 7 Trp 55 , ␣ 7 Ser 59 , and ␣ 7 Thr 77 , contribute to CTx ImI selectivity for ␣ 7 (Fig. 5).
We next combined mutations of two or three determinants into one receptor to look for interactions between the determinants and to ask whether the set of three determinants fully accounts for selectivity of CTx ImI. The three possible double mutations, ␣ 7 (W55R/S59Q), ␣ 7 (W55R/T77K), and ␣ 7 (S59Q/ T77K), decreases affinity in an additive manner (Table II), indicating that these pairs of residues contribute independently. Moreover, the triple mutation ␣ 7 (W55R/S59Q/T77K) fully decreases affinity to that observed for both the ␣ 7 55␣ 1 77␣ 7 and ␣ 1 /5HT-3 chimeras (Fig. 5). Thus, ␣ 7 Trp 55 , ␣ 7 Ser 59 , and ␣ 7 Thr 77 confer CTx ImI selectivity for ␣ 7 receptors. The overall results support the basic scaffold hypothesis because exchange of the selectivity determinants from ␣ 1 to ␣ 7 exchanges the affinity for CTx ImI.
Exchange of Selectivity Determinants between the ␣ 7 and ␣ 1 Subunits-To further confirm that ␣ 7 W55R, ␣ 7 S59Q, and ␣ 7 T77K confer CTx ImI selectivity, we sought to convert ␣ 1 to ␣ 7 affinity by constructing the converse mutations in the ␣ 1 / 5HT-3 subunit (Fig. 5). Two of the three point mutants, ␣ 1 Q59S and ␣ 1 K77T, maintain good levels of expression and increase affinity for CTx ImI, as expected from the chimeras. Combining these two mutations into a single receptor with ␣ 1 (Q59S/K77T) increases CTx ImI affinity in an additive manner, again showing that these determinants contribute independently. However, the third point mutation, ␣ 1 R55W, does not express alone nor when present in the triple mutant ␣ 1 (R55W/Q59S/K77T). The residual affinity between the double point mutation ␣ 1 (Q59S/K77T) and ␣ 7 /5HT-3 equals that conferred by the point mutant ␣ 7 W55R, further suggesting that basic scaffold of the ␣ 1 subunit is similar to that of the ␣ 7 subunit. DISCUSSION We probed the binding site of the ␣ 7 receptor using the neuronal-specific toxin CTx ImI together with chimeras containing portions of ␣ 1 sequence substituted into the extracellular domain of ␣ 7 . The results reveal three pairs of residues in equivalent positions of the subunits that confer selectivity of CTx ImI for ␣ 7 over muscle-like AChRs. Because exchange of these residues between ␣ 7 /5HT-3 and ␣ 1 /5HT-3 exchanges affinity for CTx ImI, the extracellular domains of ␣ 7 and ␣ 1 subunits appear to fold into similar basic scaffolds. This basic scaffold hypothesis was originally developed to explain ligand selectivity conferred by subunits with high homology, such as ␥ and ␦ subunits (49% identity), ⑀ and ␦ subunits (47%), and ␥ and ⑀ subunits (53%) (8 -10, 18, 21). We find that the hypothesis extends to the less homologous ␣ 7 and ␣ 1 subunits, which are only 38% identical.
To investigate the basis for neuronal specificity of CTx ImI, we needed to achieve high levels of expression of receptors with ␣ 7 binding sites and a homologous, muscle-like frame of reference. As described by others (7), we find that substituting 5HT-3 sequence from the M1 domain to the carboxyl terminus markedly increases expression, yet preserves the ligand recognition properties of the ␣ 7 binding site. The increased expression was initially surprising because the extracellular domain was widely known for its importance in receptor assembly (22)(23)(24). However, studies of ␣ 7 -␣ 3 chimeras show that formation of homo-oligomers requires matching of particular residues in the M1 and M2 transmembrane domains (25). Thus, our results further confirm that the region carboxyl-terminal to M1 contributes to assembly of homo-oligomers.
Another surprise is that our ␣ 1 /5HT-3 construct forms homooligomers on the cell surface. This observation contrasts with expression of the ␣ 1 subunit alone, which remains monomeric and retained within the cell. Thus, 5HT-3 sequence between M1 and the carboxyl terminus promotes homo-oligomer formation. We found differences, however, between ␣ 1 /5HT-3 and ␣ 7 /5HT-3 homo-oligomers in their kinetics of ␣-bgt dissociation. ␣-bgt dissociates from ␣ 7 /5HT-3 homo-oligomers with a single slow rate constant, whereas the toxin dissociates from ␣ 1 / 5HT-3 homo-oligomers with one rapid and one slow rate constant. Thus, despite the presence of only one type of subunit, binding sites in ␣ 1 /5HT-3 homo-oligomers appear to be nonequivalent. The origin of nonequivalent sites is not known, but they may arise during the course of subunit folding and oligomerization that produces a fully assembled pentamer with high affinity for ␣-bgt (26). Because acquisition of the toxin binding site is associated with a protein folding event, and folding requires interaction with specific subunits, successive addition of ␣ 1 /5HT-3 subunits may produce unusual interactions, leading to nonequivalence of the binding sites.
In addition, we find that ␣ 1 /5HT-3 and ␣ 2 ␤⑀␦ muscle receptors bind CTx ImI with similar low affinities. Because the muscle receptor contains the stabilizing residues ⑀Trp 55 / ␦Trp 57 , and the ␣ 1 /5HT-3 homo-oligomer contains the destabilizing residue ␣ 1 Arg 55 , one might expect higher affinity of the muscle receptor compared with that of ␣ 1 /5HT-3. Resolution of the apparent paradox likely lies in the different contributions of the (Ϫ) face in the two types of receptors, because they contain identical (ϩ) faces. Potentially, residues flanking the three determinants identified here, but unique to the ⑀ and ␦ subunits, may decrease affinity, despite the presence of ⑀Trp 55 / ␦Trp 57 in the muscle receptor. The unique residue differences may cause small changes in the protein scaffold and prevent interaction between CTx ImI and ⑀Trp 55 /␦Trp 57 , as well as other determinants of affinity. In addition, the low affinity may result from reorientation of CTx ImI at the binding site, and the low affinity may be due to stabilization by residues common to the ␣ 1 , ⑀, and ␦ subunits.
We chose CTx ImI to probe the ␣ 7 binding site because it is a constrained two-loop structure owing to its two disulfide bridges, similar to the muscle-specific ␣-conotoxins (Fig. 6). CTx ImI contains four amino acids in its first loop and three in its second, whereas muscle-specific ␣-conotoxins contain three FIG. 5. CTx ImI binding to ␣ 7 /5HT-3 and ␣ 1 /5HT-3 receptors containing point mutations of selectivity determinants. For each mutant receptor, CTx ImI affinity is expressed as in Fig. 4. NE, no expression. To the right is a schematic of the ␣ 7 55␣ 1 77␣ 7 chimera. The text below indicates the amino acid sequences of the 55-77 segment, with mutant residues underlined, ␣ 1 sequence in boldface, and ␣ 7 sequence in plain text. and five amino acids in its first and second loops, respectively. In addition, CTx ImI contains basic residues in both loops (Arg 7 and Arg 11 ) and an acidic residue in the first loop (Asp 5 ), unlike muscle-specific ␣-conotoxins; these unique residues may further contribute to neuronal specificity of CTx ImI.
Photoaffinity labeling and mutagenesis studies establish that the ligand binding sites of the muscle AChR contain contributions of both ␣ and non-␣ subunits. Residues of the ␣ or (ϩ) face of the binding site cluster into three linearly separate regions, leading to a three-loop model of the ␣ subunit contribution to the binding site. Similarly, residues of the non-␣ or (Ϫ) face of the binding site cluster into four separate regions of the linear sequence, leading to a four-loop model of the non-␣ contribution to the site (reviewed in Refs. 27 and 28). Studies of the highly homologous ␦, ⑀, and ␥ subunits show that residues in equivalent positions of the subunit make equivalent contributions to ligand affinity (8 -10, 18, 21). Thus, the (Ϫ) face contributed by these subunits harbors virtually superimposable peptide scaffolds. Here, we extend the basic scaffold hypothesis to the less homologous ␣ 7 and ␣ 1 subunits by showing that exchanging selectivity determinants between ␣ 7 and ␣ 1 subunits exchanges affinity for CTx ImI.
Two of the three pairs of ␣ 7 selectivity determinants, ␣ 7 Trp 55 /␣ 1 Arg 55 and ␣ 7 Ser 59 /␣ 1 Gln 59 , are within one of the four loops that contribute to the (Ϫ) face of the binding site. The contribution of ␣ 7 Trp 55 was first demonstrated by photoaffinity labeling with d-[ 3 H]tubocurarine, which labeled the equivalent residues ␥Trp 55 /␦Trp 57 in Torpedo receptors (29). Subsequent mutagenesis of ␣ 7 201/5HT-3 homo-oligomers revealed contributions of the equivalent residue in chick ␣ 7 Trp 54 to agonist and antagonist affinity (15). The second pair in this segment, ␣ 7 Ser 59 /␣ 1 Gln 59 , is equivalent to the residues in non-␣ subunits in muscle receptors, ⑀Asp 59 /␦Ala 61 , which contribute to dimethyl-d-tubocurarine selectivity of the adult receptor (9). Thus, approximately half of the neuronal specificity of CTx ImI is due to stabilization by one of the four loops at the (Ϫ) face of the subunit.
The third pair of selectivity determinants, ␣ 7 Thr 77 /␣ 1 Lys 77 , is not contained within previously described loops of either the ␣ or non-␣ faces of the binding site. One member of this pair, ␣ 1 Lys 77 , is immediately carboxyl-terminal to the main immunogenic region, which extends from the tip of the extracellular lobe of the AChR (30). ␣ 1 Lys 77 may be far enough from the main immunogenic region that it can fold back to the binding site. We cannot say whether ␣ 7 Thr 77 /␣ 1 Lys 77 contributes di-rectly or allosterically to the binding site. However, the positively charged ␣ 1 Lys 77 may repel one of the arginine side chains in CTx ImI, whereas ␣ 7 Thr 77 may be neutral or stabilize CTx ImI through hydrogen bonding. We observed equal and opposite changes in free energy of binding with ␣ 7 T77K and ␣ 1 K77T, suggesting direct contributions to affinity. A long range interaction would not be expected to show such equal and opposite free energy changes because it would have to propagate through intervening residues.
The overall results reveal three pairs of equivalent residues in the ␣ 7 and ␣ 1 subunits that confer selectivity of CTx ImI and show that the extracellular domains of ␣ 7 and ␣ 1 subunits fold into similar basic scaffolds. The precise contacts between CTx ImI and ␣ 7 await experiments that mutate residues in both the toxin and the receptor.