The contributions of aspartyl residues in the acetylcholine receptor gamma and delta subunits to the binding of agonists and competitive antagonists.

The acetylcholine (ACh) receptors in muscle have the composition α2βγδ and contain two ACh binding sites. One is formed between an α subunit and the γ subunit, and the other is formed between an α subunit and the δ subunit. Among the residues in the ACh binding sites are αCys-192 and αCys-193. The negatively charged δAsp-180 is at an appropriate distance from αCys-192/193 also to be in the ACh binding site and to interact electrostatically with the positively charged ammonium group common to agonists and competitive antagonists. Mutation to Asn of either δAsp-180 or the aligned residue in the γ subunit, γAsp-174, decreased the affinities of three agonists, acetylcholine, tetramethylammonium, and succinyldicholine 170-560-fold. By contrast, these mutations decreased the affinities of three competitive antagonists, (+)-tubocurarine, hexamethonium, and dihydro-β-erythroidine, only 2-15-fold. Agonists, but not antagonists, promote the transitions of the receptor from the resting state to the higher affinity active and desensitized states, and the greater effects of the mutations of γAsp-174 and δAsp-180 on the apparent affinities of agonists could reflect the involvement of these residues in the conformational changes of the receptor corresponding to its transitions to higher affinity states. In these transitions, one possibility is that γAsp-174 and δAsp-180 move closer to bound agonist.

The binding of acetylcholine (ACh) 1 by nicotinic receptors promotes the transitions of the receptor from the resting state to the open and the desensitized states (Katz and Thesleff, 1957). Muscle-type ACh receptors contain two nonidentical ACh binding sites (Damle and Karlin, 1978;Neubig and Cohen, 1979;Dowding and Hall, 1987). Normally, these receptors are pentamers composed of four types of subunits in the stoichiometry ␣ 2 ␤␥␦ (Reynolds and Karlin, 1978;Lindstrom et al., 1979;Raftery et al., 1980). One of the ACh binding sites is formed in the interface between the first ␣ subunit and the ␥ subunit, and the other site is formed between the second ␣ subunit and the ␦ subunit (Kurosaki et al., 1987;Blount and Merlie, 1989;Pedersen and Cohen, 1990;Sine and Claudio, 1991;Czajkowski and Karlin, 1991).
Several residues have been identified as in or close to the ACh binding sites. In the ␣ subunit, disulfide-linked and four aromatic residues, were labeled by binding site-directed reagents (Kao et al., 1984;Kao and Karlin, 1986;Dennis et al., 1988;Abramson et al., 1989;Galzi et al., 1990;Cohen et al., 1991;Middleton and Cohen, 1991). Furthermore, the involvement of these six residues in ACh binding was supported by the functional consequences of site-directed mutagenesis (Mishina et al., 1985;Tomaselli et al., 1991;Galzi et al., 1991a;O'Leary and White, 1992). The adjacent cysteines and the four aromatic residues are highly conserved among all ␣-subunit sequences.
An early assumption was that the ammonium group common to all potent agonists and antagonists of the ACh receptor is bound to a negative subsite of the binding site. From the rates of reaction and functional effects of affinity labels of different lengths, it was inferred that this negative subsite is about 12 Å from the binding site disulfide in the resting state of the receptor and about 9 Å from the binding site disulfide in the open state (Karlin, 1969). Using a radiolabeled cross-linker that reacts with a sulfhydryl at one end and a carboxyl group at the other end, we showed that three negatively charged residues in the ␦ subunit of Torpedo ACh receptor, ␦Asp-165, ␦Asp-180, and ␦Glu-182 (see Fig. 1), are within approximately 9 Å of ␣Cys-192/193 (Czajkowski andKarlin, 1991, 1995). 2 Furthermore, in the complex of mouse ACh receptor ␣, ␤, and ␦ subunits, the mutation of mouse ␦Asp-180 to Asn decreased the apparent affinity of the receptor for ACh by 2 orders of magnitude (Czajkowski et al., 1993). In addition, mutation of ␦Glu-189 to Gln decreased the apparent affinity for ACh by 1 order of magnitude. Therefore, both ␦Asp-180 and ␦Glu-189 could contribute to the negative subsite of the ACh binding site formed between the ␣ and ␦ subunits. By contrast, mutations of ␦Asp-165, ␦Glu-182, or eight other negatively charged residues in ␦ had only small effects on the apparent affinity for ACh.
The cross-linking of  to ␦Asp-180 supports the location of an ACh binding site in the interface of an ␣ subunit and the ␦ subunit. The photolabeling of the aligned residues ␥Trp-55 and ␦Trp-57 by the competitive antagonist (ϩ)-tubocurarine also supported the location of ACh binding sites in the interfaces between ␣ and ␥ and between ␣ and ␦ (Pedersen and Cohen, 1990;Chiara and Cohen, 1992). Mutation of these Trp residues had modest effects on the binding of agonists and antagonists (O'Leary et al., 1994;Corringer et al., 1995). These Trp residues and some of the aromatic residues identifed in the ␣ subunit could also contribute to the negative subsite; i.e. to the binding of the ammonium group (Dougherty and Stauffer, 1990;Galzi et al., 1990). The ␥ subunit is the most similar in sequence to the ␦ subunit, and mouse-muscle ␥ and ␦ substitute for each other in forming fully functional receptor complexes with just three types of subunits, presumably with the stoichiometries ␣ 2 ␤␥ 2 or ␣ 2 ␤␦ 2 (Kullberg et al., 1990;Sine and Claudio, 1991). The residues in ␥ that align with ␦Asp-180 and ␦Glu-189 are ␥Asp-174 and ␥Glu-183 (see Fig. 1). These two residues are identically conserved among all aligned sequences of the ␥, ␦, and ⑀ subunits (Czajkowski et al., 1993). (The ⑀ subunit substitutes for ␥ in adult muscle ACh receptor (Mishina et al., 1986).) We mutated to Asn or Gln, ␥Asp-174, ␥Glu-183, and each of the other eight Asp and Glu residues in a 60-residue segment preceding the first membrane-spanning segment (see Fig. 1) and expressed each mutant ␥ subunit, together with wild-type ␣ and ␤ subunits, in Xenopus oocytes. We also expressed ␦Asp-180 mutated to Asn and ␦Glu-189 mutated to Gln together with wild-type ␣ and ␤ subunits. The salient findings were that the mutation of ␥Asp-174 to Asn, just like the mutation of ␦Asp-180 to Asn, decreased the apparent affinity for ACh by 2 orders of magnitude and that both of these mutations decreased the apparent affinities for agonists much more than the affinities for competitive antagonists.
ACh-induced currents were recorded with a two-electrode voltage clamp at a holding potential of Ϫ40mV as described (Akabas et al., 1992). The currents induced by various concentrations of ACh were fit by the Hill equation, Oocyte membranes were prepared as described (Czajkowski et al., 1993). The final membrane suspension contained about 1.5 g of protein/l. The yield was approximately 5 fmol of 125 I-␣-bungarotoxin binding sites and about 15 g of protein/oocyte.
The binding of 125 I-labeled ␣-bungarotoxin at different toxin concentrations was determined by diluting 20 l of membrane suspension to 400 l with NP50 (50 mM NaCl/10 mM sodium phosphate/1 mM EDTA, pH 7.0) containing 0.2% Triton X-100, mixing with 125 I-␣-bungarotoxin (at five concentrations from 0.05 to 5 nM), and incubating for about 24 h at 18°C. The samples were diluted with 5 ml of ice-cold wash buffer (0.2% Triton X-100/10 mM NaCl/10 mM sodium phosphate, pH 7.4), filtered through Reeves-Angel glass-fiber filters pre-soaked in 2% polyethyleneimine, and washed two times with 5 ml of wash buffer. The amount of 125 I-␣-bungarotoxin on the filters was determined by liquid scintillation counting. Specific binding was defined as that binding of 125 I-␣-bungarotoxin blocked by 1 M nonradioactive ␣-bungarotoxin. The concentration of toxin giving half-maximal binding, KЈ tox , was estimated by a nonlinear least squares fit of the specifically bound toxin, Y, to the equation, Y ϭ Y max /(1 ϩ (KЈ tox /X) n ), where X is the 125 I-␣-bungarotoxin concentration. Given the rate constants for the association and dissociation of ␣-bungarotoxin (Sine and Claudio, 1991), the binding would not be expected to reach equilibrium in 24 h, especially at low concentrations of ␣-bungarotoxin, and therefore KЈ tox is not an equilibrium dissociation constant.
The binding of 125 I-␣-bungarotoxin to intact oocytes was determined by placing 5-10 oocytes in a final volume of 400 l of 0.5 nM 125 I-␣bungarotoxin in MOR2 (82 mM NaCl/2.5 mM KCl/1 mM Na 2 HPO 4 /5 mM MgCl 2 /0.2 mM CaCl 2 /5 mM Hepes, pH 7.4), containing 0.1% bovine serum albumin, and incubating at 18°C overnight. The oocytes were placed on a single DE81 filter and washed four times with 5 ml of ice cold MOR2 (without bovine serum albumin). The amount of 125 I-␣bungarotoxin on the filter was determined by liquid scintillation counting. In the case of intact oocytes, specific binding was defined as that blocked by 100 nM nonradioactive toxin.
The binding of agonists and antagonists was determined by their inhibition of the binding of 125 I-␣-bungarotoxin to wild-type and mutant receptors to a crude membrane fraction of oocytes. Membranes (50 l), pretreated for 20 min with 200 M diisopropyl phosphofluoridate, 0.75 nM 125 I-␣-bungarotoxin (50 l), and various concentrations of agonist or antagonists (50 l), all in NP50, were combined in a final volume of 150 l. The final 125 I-␣-bungarotoxin concentration was 0.25 nM. After 5 h at room temperature, the suspension was filtered, and the bound 125 I-␣bungarotoxin was determined as above.
The dissociation constant, K ligand , for ligand was obtained by the nonlinear least squares fit of the following equation to the binding data: where Y is the specifically bound 125 I-␣-bungarotoxin (that blocked by 1 M ␣-bungarotoxin), Y 0 is the specifically bound 125 I-␣-bungarotoxin in the absence of ligand, and A is the concentration of ligand. 3 U is the specifically bound 125 I-␣-bungarotoxin not blocked by saturating concentrations of the ligand; U was estimated by the fit. Over all the ligands tested, U ranged from 13 to 42% of specifically bound 125 I-␣-bungarotoxin.
The inhibition data was also analyzed in terms of two binding sites by the following equation: where g is the fraction of blockable binding to site 1, K ligand,1 is the dissociation constant of the ligand for site 1, K ligand,2 is the dissociation constant for site 2, and the other parameters are defined above.

RESULTS
The ten acidic residues between and including ␥Glu-163 and ␥Glu-203 ( Fig. 1) were mutated one at a time, Asp to Asn and Glu to Gln. In addition, ␥Asp-174 was mutated to Glu, and ␥Glu-183 to Asp. These mutant ␥ subunits were expressed in Xenopus oocytes together with wild-type ␣ and ␤ subunits and, where w ϭ 1/(1 ϩ a/K ligand ), k 1 and k Ϫ1 are the toxin-association and toxin-dissociation rate constants, respectively, and r total is the concentration of all receptor. Furthermore, For small values of the exponents, this equation is approximated by which is also the result obtained by assuming that toxin is binding at its initial rate and that dissociation is negligible. Given the rate constants for the binding of ␣-bungarotoxin to mouse muscle ACh receptor, k 1 ϭ 5.5 ϫ 10 5 M Ϫ1 s Ϫ1 and k Ϫ1 ϭ 3.3 ϫ 10 Ϫ6 s Ϫ1 (Sine and Claudio, 1991), and the conditions used here, x ϭ 0.25 nM, t ϭ 5 h, the exact expression is 0, 6, and 12% larger than the approximation at a ϭ 0, a ϭ K ligand , and a Ͼ Ͼ K ligand , respectively. Fitting data generated by the exact expression with the approximate equation yields a value for K ligand 13% larger than that used to generate the data. The error in the ratio of K ligand,mutant /K ligand,wild-type , as used in Tables I and II, would  in some cases, together with ␣, ␤, and ␦ subunits. We determined the effects of each of these mutations on the whole cell currents elicited by ACh and on the inhibition by ACh of the binding of 125 I-␣-bungarotoxin. For three mutations of ␥, ␥D174N, ␥D174E, and ␥E183Q, and for two mutations of ␦, ␦D180N and ␦E189Q, we also determined the binding of two additional agonists and of three antagonists (Fig. 2). The complex of wild-type ␣, ␤, and ␥ (␣ 2 ␤␥ 2 receptor), similar to the complex of ␣, ␤, and ␦ (␣ 2 ␤␦ 2 receptor; Czajkowski et al., 1993), yielded an EC 50 that was close to that of the complex of wild-type ␣, ␤, ␥, and ␦ (␣ 2 ␤␥␦ receptor; Table I). No AChinduced current was obtained with just ␣ and ␤ (Table I). Also, Liu and Brehm (1993) found that complexes of ␣ and ␥ or of ␣ and ␦ yielded ACh-induced currents that were 200 or 40 times smaller than the currents yielded by complexes of ␣, ␤, and ␥ or of ␣, ␤, and ␦, respectively. Initially, we tested the ␥ mutants in complexes just with ␣ and ␤.
For 10 of the 12 ␥ mutants, the complexes with ␣ and ␤ gave maximal ACh-induced currents that were 2-50% of wild-type currents ( Table I). The largest change in the EC 50 for ACh was a 4.6-fold increase shown by ␥E176Q; ␥E183Q gave a 4-fold increase; the other EC 50 values were within a factor of 3 of the EC 50 of wild-type ␣ 2 ␤␥ 2 receptor ( Table I). The Hill coefficients of all mutants were very close to the Hill coefficient of wild-type receptor. Also, for these 10 mutants the K ACh , determined by the inhibition by ACh of 125 I-␣-bungarotoxin binding, was slightly less than that of wild-type ␣ 2 ␤␥ 2 (Table I).
Two mutants, ␥D174N and ␥E202Q, co-expressed with ␣ and ␤, gave no detectable ACh-induced current at ACh concentrations up to 2 mM. Furthermore, there was no specific binding of 125 I-␣-bungarotoxin to the surface of the intact oocytes expressing ␣ϩ␤ϩ␥D174N or ␣ϩ␤ϩ␥E202Q. 4 Therefore, we conclude that the complexes formed by these subunits were not transported to the cell surface. These subunits did form agonistbinding complexes in cytoplasmic membranes, the predominant constituents of the crude membrane fraction used in the binding experiments. Because the dissociation constant for ACh of the complex containing ␥E202Q was slightly less than that of all wild-type ␣ϩ␤ϩ␥, this mutant was not further characterized (Table I).
Unlike the complex of ␣, ␤, and ␥D174N, the complex of ␣, ␤, ␦, and ␥D174N was expressed on the cell surface. The AChinduced current was characterized by an EC 50 of 24 M, six times the EC 50 of all wild-type ␣ 2 ␤␥␦ ( Fig. 3; Table I). Because the complex of all wild-type ␣ 2 ␤␦ 2 receptor had an EC 50 of 2.7 M (Czajkowski et al., 1993) and the complexes of ␣ϩ␤ϩ␥D174N were not expressed on the cell surface, the increase in the EC 50 must have been due to the expression of the pentameric complex of ␣ 2 ␤␦␥D174N. Given that wild-type ␣ 2 ␤␦ 2 complex may also have been present, the observed EC 50 of 24 M ACh is a lower limit to the EC 50 of the complex of ␣ 2 ␤␦␥D174N.
The binding of ACh by complexes of ␣, ␤, and ␥D174N in a subcellular membrane fraction of oocytes was characterized as K ACh , derived from the inhibition by ACh of 125 I-␣-bungarotoxin binding. K ACh for ␣ϩ␤ϩ␥D174N was 170 times the K ACh for wild-type ␣ϩ␤ϩ␥ (Fig. 4A and Table I). All of the other 11 mutants tested had K ACh slightly less than that of wild-type ␣ϩ␤ϩ␥.
The mutation ␥D174N also affected the binding of ACh in the 4 Receptor complexes transported to the cell surface are pentamers (Sine and Claudio, 1991), and therefore complexes whose properties were determined by electrophysiological assay we designate with pentameric stoichiometry, e.g. ␣ 2 ␤␥ 2 . The binding properties of receptor complexes were determined in a membrane fraction that included both surface and cytoplasmic membranes. The latter may contain incompletely assembled receptor complexes, as well as pentamers (Kreienkamp et al., 1995); we designate the mixture of oligomers obtained by expressing combinations of subunits with plus signs, e.g. ␣ϩ␤ϩ␥D174N.
FIG. 1. Aligned sequences of Torpedo ␦ subunit and mouse ␥ and ␦ subunits from residue 161 to residue 224 in ␦ and from residue 158 to residue 218 in ␥. Torpedo ␦ and mouse ␦ have the same numbering. In ␦, the first membrane-spanning segment starts near Pro-225 and in ␥, near Pro-219. In Torpedo ␦, the residues shown by cross-linking to be within 9 Å of ␣Cys-192/193 are underlined (Czajkowski et al., 1995). Each Asp and Glu in the mouse ␥ and ␦ segments were mutated individually to Asn or Gln; ␥Asp-174 and ␦Asp-180, the mutations of which caused the largest changes in affinities, are in large print. context of the ␣ϩ␤ϩ␦ϩ␥D174N complexes. The binding by these complexes were characterized by two dissociation constants, K 1 and K 2 , with K 1 Ͻ K 2 . K 1 for the mutant receptor was within a factor of 2 of K 1 for wild-type ␣ϩ␤ϩ␥ϩ␦ receptor; K 2 for the mutant, however, was 27 times larger than K 2 for wild-type receptor ( Table I). The high affinity site 44 Ϯ 8% of the binding (g in the two-site equation under "Experimental Procedures"), and the low affinity site accounted for 56 Ϯ 8%. The complex(es) formed by the co-expression of ␣, ␤, ␥D174N, and ␦D180N bound ACh with a single dissociation constant of 45 Ϯ 5 M (n ϭ 2), 200 times the dissociation constant of 0.21 M for the high affinity site of wild-type receptor and eight times the dissociation constant of 5.7 M for the low affinity site (Fig. 4B).
We determined the effects of the mutations of ␥Asp-174 and FIG. 3. Current as a function of ACh concentration. The receptor complexes are all wild-type ␣ 2 ␤␥␦ (circles), ␣ 2 ␤␦␥D174N (squares), and ␣ 2 ␤␦␥E183Q (triangles). Oocytes expressing these complexes were superfused with five or six concentrations of ACh for 10 s each with a 5-min wash between, and the peak currents were recorded under twoelectrode voltage clamp at Ϫ40 mV (see "Experimental Procedures"). Each concentration was added twice. The peak current (I) is the average of the duplicates and is plotted as a fraction of the peak current (I max ) at infinite ACh concentration, calculated by fitting the Hill equation to the data (Table I)

TABLE I Effects of mutations in the gamma subunit on ACh-induced current and on ACh binding
Subunit mRNA was injected into Xenopus oocytes, and after 1-3 days ACh-induced currents were recorded, as described under "Experimental Procedures." Peak current as a function of ACh concentration was fitted by the Hill equation. The dissociation constants of ACh were determined from its retardation of the binding of toxin (see "Experimental Procedures"). Where two values are given, the data were better fitted by a two-site than a one-site fit. Means, S.E.s, and number of independent experiments are given. ␥Glu-183, and of the aligned residues, ␦Asp-180 and ␦Glu-189, on the binding of two additional agonists, tetramethylammonium, the smallest agonist of the ACh receptor, and the bisquaternary agonist, succinyldicholine (Fig. 2). The dissociation constant for tetramethylammonium, K TMA , for ␣ϩ␤ϩ␥D174N was 180 times larger than for wild-type ␣ϩ␤ϩ␥ (Table II). By contrast, K TMA values for ␣ϩ␤ϩ␥D174E and for ␣ϩ␤ϩ␥E183Q were not appreciably different than K TMA for wild-type (Table  II). The dissociation constant for succinyldicholine, K SuCh , was 300 times greater for ␣ϩ␤ϩ␥D174N than for wild-type ␣ϩ␤ϩ␥ (Table II). The dissociation constants of ␣ϩ␤ϩ␦D180N for the three agonists were also 2 orders of magnitude greater than those of wild-type ␣ϩ␤ϩ␦ (Table II). Thus, the mutations to Asn of the aligned residues ␦Asp-180 and ␥Asp-174 had similar effects, even though wild-type ␣ϩ␤ϩ␦ had an order of magnitude higher affinity for the agonists than did wild-type ␣ϩ␤ϩ␥.
None of the mutations tested had much of an effect on the binding of ␣-bungarotoxin. The concentrations giving half-maximal binding (in 24 h) were 130 Ϯ 30 pM for wild-type ␣ϩ␤ϩ␥, 160 Ϯ 30 pM for ␣ϩ␤ϩ␥D174N, 170 Ϯ 30 pM for ␣ϩ␤ϩ␥E176Q, and 290 Ϯ 110 pM for ␣ϩ␤ϩ␥E183Q (n ϭ 3 for each). Mutations of ␦Asp-180 and ␦Glu-189 also had little effect on the binding of ␣-bungarotoxin by ␣ϩ␤ϩ␦ receptor (Czajkowski et al., 1993). DISCUSSION We previously identified two residues in the ␦ subunit, ␦Asp-180 and ␦Glu-189, that could contribute to the negative subsite of the ACh binding site formed between ␣ and ␦ (Czajkowski and Karlin, 1991;Czajkowski et al., 1993;Czajkowski and Karlin, 1995). One of these, ␦Asp-180, was cross-linked via a 9-Å cross-link to one of the adjacent Cys residues, ␣Cys-192 or ␣Cys-193, that form the binding site disulfide, and the mutation of ␦Asp-180 to Asn caused a 2 orders of magnitude decrease in the apparent affinity of the ␣ϩ␤ϩ␦ receptor complex for ACh, measured both by activation and by binding. Although ␦Glu-189 was not cross-linked to ␣Cys-192/193, its mutation caused a 1 order of magnitude decrease in the apparent affinity for ACh, also measured by both methods. ACh binding was uniquely sensitive to the mutation of these two residues among the 11 negatively charged residues in a stretch of 60 residues just preceding the first membrane-spanning segment of the ␦ subunit. Because ␥ and ␦ each forms an ACh binding site with an ␣ subunit (Kurosaki et al., 1987;Blount and Merlie, 1989;Pedersen and Cohen, 1990;Sine and Claudio, 1991;Czajkowski and Karlin, 1991), we determined the effects on agonist and competitive antagonist binding of mutating to Asn or Gln each Asp or Glu residue in the aligned stretch of residues in the ␥ subunit. Among the 10 acidic residues in this stretch of ␥, only the mutation of ␥Asp-174, which aligns with ␦Asp-180, caused

effects of mutations on the binding of agonists
The dissociation constants for ligands were determined from their retardation of the binding of toxin, as described under "Experimental Procedures." Where two values are given, the data were better fitter by a two-site fit than a one site fit. S.E.s and the number if independent experiments are given.  Czajkowski et al (1993).
The effects of the mutations of ␥D174N and ␦D180N were most obvious in complexes of just three of the four types of receptor subunits. In mouse receptor, ␥ and ␦ substitute for each other to form a functional complex of three types of subunits with properties very similar to those of the complex with all four types of subunits (Kullberg et al., 1990;Sine and Claudio, 1991). On the other hand, ␣ and ␤ alone, gave no current and ␣ and ␥ or ␣ and ␦, gave much lower currents than ␣, ␤, and ␥, or ␣, ␤, and ␦ complexes (Liu and Brehm, 1993). Therefore, to the extent that we were able to characterize the mutations by their effect on agonist-induced currents, we could be certain that we were characterizing only complexes containing the mutant subunit when either the mutant ␥ or the mutant ␦ was expressed with wild-type ␣ and ␤ subunits. This approach was successful with most of the mutants tested.
␥D174N, however, was not expressed on the oocyte surface with ␣ and ␤ alone. Nevertheless, ␥D174N was incorporated into a functional complex on the oocyte surface when it was expressed together with wild-type ␣, ␤, and ␦. The EC 50 characterizing the ACh-induced current was six times the EC 50 for all wild-type ␣ 2 ␤␥␦. In the complexes formed by these four subunits in both surface and cytoplasmic membranes, the binding of ACh was characterized by two dissociation constants. The value for the low affinity binding site, presumably the ␣-␥ site (Blount and Merlie, 1989;Sine and Claudio, 1991), was 27 times higher than that for all wild-type ␣ϩ␤ϩ␥ϩ␦. When both ␥D174N and ␦D180N were co-expressed with wild-type ␣ and ␤, a single dissociation constant was obtained that was greater than both dissociation constants of wild-type ␣ϩ␤ϩ␥ϩ␦. Therefore, both ␥D174N and ␦D180N affect binding, even in complexes of all four types of subunits.
The largest effects of the mutation ␥D174N were obtained after coexpression just with wild-type ␣ and ␤. In this case, the binding of agonists and competitive antagonists were entirely to complexes in intracellular membranes. These complexes could include ␣␥ dimers, ␣␥␤ trimers, (␣␥) 2 tetramers, and (␣␥) 2 ␤ pentamers (Kreienkamp et al., 1995). We obtained little ␣-bungarotoxin binding and negligible toxin binding blocked by ACh when we coexpressed just ␣ and ␥ (data not shown); thus, for this analysis, we can ignore ␣␥-dimers and (␣␥) 2 tetramers. The mutation ␥D174N could, however, have shifted the distribution between ␣␤␥ trimers and ␣ 2 ␤␥ 2 pentamers. If these complexes had different binding properties, the effect of ␥D174N on binding could have been due to the shift in the distribution of complexes. ␥Asp-174, however, is not in a region found to effect receptor assembly (Kreienkamp et al., 1995).
Furthermore, the mutation of the aligned residue, ␦D180N, had a comparable effect on ACh binding in functional ␣ 2 ␤␦ 2 pentamers expressed on the oocyte surface (Czajkowski et al., 1993).
It is remarkable that the mutation to Asn of either ␥Asp-174 or ␦Asp-180 had a 10 -100 times greater effect on the binding of agonists than on the binding of competitive antagonists (Tables  II and III). Agonists and competitve antagonists bind to overlapping sites; some of the same residues are labeled by agonist and antagonist affinity labels (Kao et al., 1984;Galzi et al., 1990;Cohen et al., 1991;Middleton and Cohen, 1991), and the mutations of these residues affect the binding of both agonists and antagonists, albeit not equally. Furthermore, ACh receptors altered by chemical modification (Karlin and Winnik, 1968) or by mutations (Bertrand et al., 1992) can be activated by ligands that normally are competitive antagonists, consistent with the overlap of agonist and competitive antagonist sites. Thus, the difference in the effects of mutations on agonist and competitive antagonist binding was not likely due to completely separate sites for the two types of ligands. Nevertheless, the differences could have resulted from nonidentical contacts of agonists and antagonists within overlapping binding sites. In the results presented here, it is clear that the difference was not dependent on the number of ammonium groups on the ligands; the binding of all three agonists was affected much more than the binding of all three antagonists (Tables II and  III).
The binding of agonists promotes the transitions of the receptor from the resting state to the higher affinity active and desensitized states. Therefore, alteration of the kinetics of these transitions could affect both the EC 50 and K ligand . The binding of competitive antagonists does not normally promote activation, and the observed K ligand for a competitive antagonist is likely to be simply an equilibrium dissociation constant. The uniformly greater effects of the mutations on agonist binding than on antagonist binding is therefore likely due to effects on the kinetics of the agonist-induced transitions. Consistent with this interpretation, coexpression of ⑀D175N (⑀Asp-175 aligns with ␥Asp-174 and ␦Asp-180), with wild-type ␣, ␤, and ␦, caused an 8-fold increase in EC 50 , most of which could be accounted for by a decrease in the channel opening rate .
Mutations of other residues in or close to the ACh binding site also affected the kinetics of state transitions. Based on photoaffinity labeling, ␣Tyr-93 and ␣Tyr-190 are in or close to the ACh binding site (Dennis et al., 1988;Abramson et al., 1989;Galzi et al., 1990;Cohen et al., 1991). Mutations of these residues also had a much greater effect on agonist binding than on competitive antagonist binding (Sine et al., 1994). An analysis of ␣Y190F showed that the 2 orders of magnitude increases

of mutations on the binding of competitive antagonists
The dissociation constants for ligands were determined from their retardation of the binding of toxin, as described under "Experimental Procedures." Where two values are given the data were better fit by a two-site fit than a one site fit. S.E.s and the number of independent experiments are given. in the EC 50 were attributable to changes both in binding and in gating kinetics (O'Leary and White, 1992;Chen et al., 1995). The structure of the binding site changes on the binding of agonists (Karlin, 1969;Damle and Karlin, 1980) or on the transition to the desensitized state (Galzi et al., 1991b). The involvement of residues in or close to the ACh binding site in the agonist-induced transitions between states is consistent with these residues moving during the transitions. We have determined by cross-linking that ␦Asp-180 is close to the ␣-␦ ACh binding site (Czajkowski and Karlin, 1995), and by symmetry, ␥Asp-174 and ⑀Asp-175 are also likely to be close to the ␣-␥ and ␣-⑀ ACh binding sites. Nevertheless, we do not know whether or not these residues participate directly in the binding of agonists. One possible mechanism, however, that places these residues in the binding sites and incorporates their movement as an integral part of activation is that on the binding of agonist to the ␣ subunit, the side chain of the Asp on the neighboring subunit, ␥Asp-174, ⑀Asp-175, or ␦Asp-180, moves closer to the agonist ammonium group, increasing the electrostatic interaction between these oppositely charged groups and bringing other side chains into more favorable interactions with the agonist. We previously postulated a negative subsite that interacted with the ammonium group of agonists and inferred that on the binding of agonist this negative subsite moved a few Ångstroms closer to the binding site disulfide (Karlin, 1969). We now suggest that ␥Asp-174 contributes to the negative subsite of the ACh binding site formed between ␣ and ␥, and similarly ␦Asp-180 contributes to the negative subsite of the ACh binding site formed between ␣ and ␦ Karlin, 1991, 1995;Czajkowski et al., 1993). The postulated contraction of the binding site crosses the subunit interface and could trigger the sliding of neighboring subunits. This relative movement of the subunits could be a mechanism for the propagation of structural changes across the membrane, from the ACh binding sites, in the extracellular domain, to the gate, close to the intracellular end of the channel (Czajkowski et al., 1993;Akabas et al., 1994;Unwin, 1995).