Epibatidine binds with unique site and state selectivity to muscle nicotinic acetylcholine receptors.

Ligand binding sites in fetal (alpha2betagammadelta) and adult (alpha2betadeltaepsilon) muscle acetylcholine receptors are formed by alphadelta, alphagamma, or alphaepsilon subunit pairs. Each type of binding site shows unique ligand selectivity due to different contributions by the delta, gamma, or epsilon subunits. The present study compares epibatidine and carbamylcholine binding in terms of their site and state selectivities for muscle receptors expressed in human embryonic kidney 293 cells. Measurements of binding to alphagamma, alphadelta, and alphaepsilon intracellular complexes reveal opposite site selectivities between epibatidine and carbamylcholine; for epibatidine the rank order of affinities is alphaepsilon > alphagamma > alphadelta, whereas for carbamylcholine the rank order is alphadelta congruent with alphaepsilon > alphagamma. Because the relative affinities of intracellular complexes resemble those of receptors in the closed activable state, the results suggest that epibatidine binds with unique site selectivity in activating the muscle receptor. Measurements of binding at equilibrium show that both enantiomers of epibatidine bind to adult and fetal receptors with shallow but monophasic binding curves. However, when receptors are fully desensitized, epibatidine binds in a biphasic manner, with dissociation constants of the two components differing by more than 170-fold. Studies of subunit-omitted receptors (alpha2betadelta2, alpha2betagamma2, and alpha2betaepsilon2) reveal that in the desensitized state, the alphadelta interface forms the low affinity epibatidine site, whereas the alphagamma and alphaepsilon interfaces form high affinity sites. In contrast to epibatidine, carbamylcholine shows little site selectivity for desensitized fetal or adult receptors. Thus epibatidine is a potentially valuable probe of acetylcholine receptor binding site structure and of elements that confer state-dependent selectivities of the binding sites.

functionally distinct subunits with compositions ␣ 2 ␤␥␦ in fetal and ␣ 2 ␤⑀␦ in adult muscle (4). Within the pentamer are two binding sites for acetylcholine (ACh); one is formed at the ␣␦ subunit interface, whereas the other is formed at the ␣␥ interface in the fetal receptor and at the ␣⑀ interface in the adult. Each type of binding site displays distinct selectivities for agonists and competitive antagonists. For example, carbamylcholine binds 30-fold less tightly to the ␣␥ binding site than to the ␣␦ and ␣⑀ binding sites owing to different contributions of the ␥, ␦, and ⑀ subunits (5,6). Previous studies used ␥-␦ and ⑀-␦ subunit chimeras to identify residues in these subunits which confer site selectivity for curare, conotoxin M1, and carbamylcholine (6 -9). The overall findings show that four loops, well separated in the primary sequence of the non-␣ subunits, contribute to the binding site interface (10).
To investigate further the structure of the ligand binding site, the present work examines binding of epibatidine to sites of fetal and adult muscle AChRs. We reasoned that a structurally constrained ligand such as epibatidine would be less able to accommodate differences in binding site structure and thus might show greater or different selectivity compared with flexible agonists such as carbamylcholine and ACh. We show that epibatidine binds with novel site selectivity to muscle AChRs, selecting strongly for the ␣⑀ and ␣␥ binding sites over the ␣␦ site. Further, unlike carbamylcholine, epibatidine maintains strong site selectivity when the receptor is converted to the high affinity desensitized state. Owing to its unique site and state selectivity, epibatidine is a potentially valuable probe of binding site structure and of elements that confer state-dependent selectivity.
Ligand Binding Measurements-Cells expressing surface pentamers were harvested by gentle agitation in phosphate-buffered saline containing 5 mM EDTA, centrifuged at 1000 ϫ g for 1 min, and resuspended in potassium Ringer's solution (140 mM KCl, 5.4 mM NaCl, 1.8 mM CaCl 2 , 1.7 mM MgCl 2 , 25 mM HEPES, 30 mg/l bovine serum albumin, adjusted to pH 7.4 with 10 -11 mM NaOH). Cells expressing intracellular complexes were permeabilized with saponin before harvesting (6). Agonist binding was determined by competition against the initial rate of 125 I-labeled ␣-BTX binding as described previously (6). In brief, the receptor complexes were equilibrated with agonist for 40 min before addition of 5 nM 125 I-labeled ␣-BTX. The cells were then incubated for a further 20 -40 min to allow occupancy of at most 50% of the binding sites by 125 I-labeled ␣-BTX. The total number of sites was determined by incubating with 25 nM 125 I-labeled ␣-BTX for 20 -40 min and subtracting a blank determined in the presence of 10 mM carbamylcholine. The cells were harvested using a Brandell cell harvester and counted in a ␥ counter.
Data Analysis-We fit the following two equations to our data using Prism 2.0 (GraphPad Software): where [L] is the concentration of competing ligand, K is an apparent dissociation constant, n is the Hill coefficient, K 1 and K 2 are intrinsic dissociation constants, and P is the fraction of sites with dissociation constant K 1 .

RESULTS
Binding to Intracellular Complexes-Transfection of HEK-293 cells with ␣ and ␥, ␦, or ⑀ subunit results in robust expression of agonist displaceable intracellular ␣-BTX binding sites. To determine selectivity of epibatidine for these intracellular complexes, we measured binding by competition against 125 Ilabeled ␣-BTX binding. Both enantiomers of epibatidine bind with highest affinity to ␣⑀ complexes, intermediate affinity to ␣␥, and lowest affinity to ␣␦ ( Fig. 1A and Table I). Although both isomers of epibatidine bind with the same affinity to ␣⑀ complexes, the (Ϫ)-isomer binds with significantly higher affinity to ␣␥ and ␣␦ complexes compared with the (ϩ)-isomer. By contrast, carbamylcholine shows opposite site selectivity to epibatidine, binding with high affinity to ␣␦ and ␣⑀ and low affinity to ␣␥ complexes ( Fig. 1B and Table I). In this and previous studies (6) we found that agonists bind to intracellular complexes with Hill coefficients significantly less than 1. Because complexes of an ␣ and a non-␣ subunit should form only one type of binding site, the shallow binding curves we observe indicate some type of site heterogeneity. Despite this heterogeneity, selectivities of carbamylcholine and acetylcholine for intracellular complexes (5, 6) coincide with selectivities of receptors in the closed activable state determined by single channel kinetic analysis (12,13). Further, intracellular complexes show the same rank order of affinity for the competitive antagonist d-tubocurarine to that observed in native receptors, ␣␥ Х ␣⑀ Ͼ ␣␦ (14). Intracellular complexes, unlike fully assembled native receptors, do not enter a high affinity desensitized state and are presumably incapable of channel gating (6). Thus the binding site selectivity of intracellular complexes may most closely resemble that of closed activable receptors.
Binding to Cell Surface Pentamers-To assess site selectivity of epibatidine for receptors containing the full complement of subunits, we expressed adult or fetal receptors in HEK-293 cells and measured binding to intact cells under control and desensitizing conditions ( Fig. 2 and Table II). Under control conditions, both enantiomers of epibatidine displace 125 I-labeled ␣-BTX with much higher affinity than carbamylcholine. Although fetal and adult receptors bind epibatidine with similar affinities, fetal receptors distinguish between (ϩ)-and (Ϫ)-epibatidine, whereas adult receptors do not. Hill coefficients for epibatidine are only ϳ0.8 compared with values for carbamylcholine of ϳ1.3. Although a Hill coefficient less than 1 implies multiple binding sites, for agonists equilibrium binding cannot be interpreted using a simple two site model. At equilibrium, agonist binding is determined by affinities of the sites in the resting, open, and desensitized states as well as the allosteric constants governing channel opening and desensitization (see under "Discussion").
To measure binding of agonist to a single functional state of the receptor, we used the local anesthetic proadifen to promote entry into the high affinity desensitized state (15,16). When fully desensitized with proadifen, adult and fetal receptors bind carbamylcholine with high affinity and Hill coefficients close to 1, indicating binding to a single class of high affinity sites ( Fig.   FIG. 1. Epibatidine and carbamylcholine binding to intracellular complexes. A, epibatidine binds with high affinity to ␣⑀ complexes, with intermediate affinity to ␣␥ complexes, and with low affinity to ␣␦ complexes. B, carbamylcholine binds with high affinity to ␣⑀ and ␣␦ complexes and with low affinity to ␣␥ complexes. Binding of agonists to intracellular complexes was determined as described under "Experimental Procedures." In both panels the data are means of three to five experiments with the error bars representing the standard error. The curves are fits of Equation 1 (see under "Experimental Procedures") to the data with parameters given in Table I. yield intrinsic dissociation constants different by more than 170-fold (Fig. 2, A and C and Table II). Thus, whereas carbamylcholine does not distinguish between the ␣␥ and ␣␦ binding  Table II. In all cases the data are means of three to five experiments with standard errors of less than 5% of total.

TABLE II Binding of (ϩ)-and (Ϫ)-epibatidine and carbamylcholine to adult and fetal receptors
The parameters (K 1 , K 2 , dissociation constants; P, proportion of binding sites with dissociation constant K 1 , n, Hill coefficient) are derived from fits of either Equation 1 or 2 (see under "Experimental Procedures") to the data in Fig. 2  sites in the desensitized fetal receptor nor between the ␣⑀ and ␣␦ sites in the desensitized adult receptor, epibatidine selects strongly between the two binding sites of each receptor.
Comparison of epibatidine binding to desensitized fetal and the corresponding ␣ 2 ␤␥ 2 and ␣ 2 ␤␦ 2 triplet receptors shows that the high affinity component corresponds to the ␣␥ site, whereas the low affinity component corresponds to the ␣␦ site. Although expression of ␣ 2 ␤⑀ 2 pentamers was too low to accurately measure under desensitizing conditions, at equilibrium epibatidine binds to these receptors 30-fold more tightly than to the low affinity binding site of desensitized adult receptors. Thus in the desensitized adult receptor, the high affinity component corresponds to the ␣⑀ site, whereas the low affinity component corresponds to the ␣␦ site.
Effects of Proadifen: Concentration Dependence-Our preliminary experiments with (Ϯ)-epibatidine tested a range of proadifen concentrations to determine an optimum concentration for promoting desensitization. As observed previously, increasing proadifen concentrations progressively increase carbamylcholine affinity to approach a limiting value corresponding to the agonist affinity for the desensitized state (15). For the fetal receptor, the limiting carbamylcholine affinity is reached at 200 M proadifen; at all concentrations of proadifen carbamylcholine binding is well described by Hill coefficients of 1 or greater, implying binding to a single class of sites ( Fig. 4 and Table IV). For the adult receptor the limiting affinity is achieved by only 30 -60 M proadifen; at proadifen concentrations up to 60 M, carbamylcholine binding is described by Hill coefficients of 1 or greater, again indicating binding to a single class of high affinity sites ( Fig. 4 and Table IV). However, in the presence of 100 -200 M proadifen, carbamylcholine binds to adult receptors with Hill coefficients of 0.7-0.8. As carbamylcholine affinity reaches a limit at lower proadifen concentrations, the observed broadening of the binding curve at high proadifen concentrations likely results from an action distinct from enhancement of desensitization.
At both fetal and adult receptors, epibatidine binds in a distinctly biphasic manner in the presence of proadifen; biphasic binding is observed at all proadifen concentrations greater than 10 M. Furthermore, for both sites of the adult receptor, epibatidine affinity increases to reach a limiting value with proadifen concentrations of 30 M or greater, as observed for carbamylcholine (Table III). For the fetal receptor, the maximum increase in affinity is reached at 60 -100 M proadifen, similar to results with carbamylcholine (Table IV). Thus for both fetal and adult receptors, a given concentration of proadifen increases carbamylcholine and epibatidine affinities to similar extents.
We noticed for epibatidine, however, that the relative weights of the high and low affinity components depend on proadifen concentration (Fig. 4 and Table IV). For fetal receptors in the presence of 30 M proadifen, the high affinity component dominates, whereas for adult receptors the two compo-  Table III.   Table IV.

TABLE IV Dependence of desensitized state agonist binding on proadifen concentration
Adult or fetal receptors were incubated with the indicated concentration of proadifen, and agonist affinity was determined. The parameters (K 1 , K 2 , dissociation constants; n, Hill coefficient; P, fraction of sites with dissociation constant K 1 ) are derived from fits of Equation 1 or 2 (see under "Experimental Procedures") to the data in Fig. 4 and are expressed Ϯ S.E. For the adult receptor, carbamylcholine binding curves at 100 and 200 M proadifen were fit using both the Hill equation and the sum of two equally weighted sites. Receptor type Proadifen (Ϯ)-Epibatidine Carbamylcholine To examine this additional action of proadifen, we measured the initial rate and total number of 125 I-labeled BTX binding sites at varying concentrations of proadifen. The results show that increasing proadifen concentrations reduce the rate of toxin binding and the total number of sites approximately in parallel (Fig. 5). As these determinations reflect contributions of both sites in each receptor type, they do not reveal which of the ␣␦, ␣⑀, or ␣␥ sites are preferentially affected. However, our results on site selectivity of epibatidine show that proadifen decreases the rate and number of BTX sites in a site-selective manner (Fig. 4); site selectivity for proadifen follows the rank order, ␣⑀ Ͼ ␣␦ Ͼ ␣␥.

DISCUSSION
The present study examines various combinations of AChR subunits to characterize the interaction of (ϩ)-and (Ϫ)-epibatidine with the three types of binding sites of muscle AChRs. Our results reveal that epibatidine is unique among AChR agonists thus far characterized in that it binds with high affinity to ␣⑀ and ␣␥ sites and low affinity to ␣␦ sites. Further, unlike classical agonists, epibatidine selects between these binding sites in the desensitized state of the receptor.
Binding of agonists to the AChR can be described by the following conventional scheme: where R is the closed, activable state of the receptor, O is the open state, D is the desensitized state, and A is agonist. K 1 and K 2 are the dissociation constants for the activable state binding sites, and K d1 and K d2 are the dissociation constants for the desensitized state binding sites, M is an allosteric constant governing the transition between the activable and desensitized states, and ⌰ is the channel opening equilibrium constant (18,19). Because apparent affinity measured in equilibrium binding assays depends on the four intrinsic dissociation constants as well as the state equilibrium constants, such measurements do not directly reflect intrinsic agonist affinities of a particular binding site. Nevertheless, estimates of parameters in the upper limb of Scheme 1 can be derived from the kinetics of single channel currents. Single channel kinetic analysis established that K 1 and K 2 for ACh differ by 30 -100-fold in fetal mouse and Torpedo receptors (12,20) and by 5-fold in adult human receptors (21) but are indistinguishable in adult mouse receptors (13). Here, we observe a similar selectivity pattern for carbamylcholine binding to mouse intracellular complexes: ␣␦ and ␣⑀ complexes bind agonist with high affinity but ␣␥ complexes bind agonist 20 -30-fold less tightly. As in previous studies (6) we find that intracellular complexes bind agonists with Hill coefficients less than 1. Because complexes of an ␣ and a single type of non-␣ subunit should form only one class of binding site, a Hill coefficient less than 1 suggests heterogeneous binding sites. Heterogeneity of intracellular complexes could arise from 1) formation of tetrameric complexes of the form ␣x␣x, where x is ␥, ␦, or ⑀ (22); 2) access of agonist to immature or partially degraded complexes due to permeabilization by saponin; 3) a fixed proportion of the receptors arrested in the low affinity activable and high affinity open channel and desensitized states. Despite this apparent heterogeneity, carbamylcholine selectivity determined in our binding assay parallels selectivity of the activable state of the native receptor determined from single channel kinetic analysis (12,13). The overall results suggest that the ligand selectivities of ␣␥, ␣␦, and ␣⑀ complexes most closely resemble those of the corresponding closed activable binding sites in native pentamers. Thus in activating the receptor, epibatidine likely binds with high affinity to the ␣␥ and ␣⑀ sites but with low affinity to the ␣␦ site.
Unlike the closed activable state, the desensitized state can be studied in isolation by measuring agonist binding in the presence of a desensitizing agent such as proadifen (15,16). Here we observe that carbamylcholine does not distinguish between the desensitized ␣␥, ␣␦, and ␣⑀ binding interfaces. This is evidenced by the monophasic binding curves of desensitized fetal and adult receptors and the similar affinities of desensitized ␣ 2 ␤␦ 2 and ␣ 2 ␤␥ 2 triplet receptors. Similar results were obtained previously for carbamylcholine and ACh (17,23). Epibatidine, by contrast, distinguishes between the two sites in desensitized fetal and adult receptors.
Previous studies showed that ␣ 2 ␤␥ 2 and ␣ 2 ␤␦ 2 pentamers bind a range of ligands with affinities similar to those of the two sites in native fetal receptors (7,17), suggesting that binding sites of subunit-omitted receptors are close in structure to binding sites in the corresponding native receptor. For epibatidine, we observe similar dissociation constants for desensitized ␣ 2 ␤␥ 2 receptors and the high affinity component of desensitized fetal receptors and similar dissociation constants for desensitized ␣ 2 ␤␦ 2 receptors and the low affinity components of desensitized fetal and adult receptors. These results strongly suggest that the biphasic binding of epibatidine under desensitizing conditions is due to intrinsic differences in the ␣␥ and ␣␦ binding sites, i.e. the different contributions of the ␥ and ␦ subunits to the binding sites confer different epibatidine affinities. A second possibility is that biphasic binding might represent uncoupled binding sites of a single receptor type where one component corresponds to the high affinity desensitized state and the other corresponds to the low affinity activable state, i.e. proadifen might eliminate the cooperativity of the binding sites predicted by the allosteric model of Monod et al. (24). However, this hypothesis seems unlikely because the low affinity component in the presence of proadifen is shifted to the left of the overall curve in its absence. In addition, uncoupling of the binding sites by proadifen should result in a biphasic binding curve for carbamylcholine, whereas we observe essentially monophasic binding of carbamylcholine to both adult and fetal receptors. Thus the weight of evidence suggests that unlike acetylcholine and carbamylcholine, epibatidine selects between the desensitized ␣␥ and ␣␦ binding sites in the fetal receptor and between the ␣⑀ and ␣␦ binding sites in the adult receptor. Whereas previous studies with the competitive antagonist d-tubocurarine showed site selectivity for desensitized Torpedo receptors (25), to our knowledge epibatidine is the first agonist that selects between the binding sites of the desensitized receptor.
Our preliminary experiments sought to determine the optimal concentration of proadifen for desensitizing each receptor type. These studies revealed additional complexities in the interaction of proadifen with the AChR. In the absence of agonist, high concentrations of proadifen inhibit the binding of 125 I-labeled ␣-BTX; this inhibition manifests as decreases in both the initial rate and maximal number of toxin binding sites. A decrease in the maximal number of toxin binding sites suggests that proadifen shifts the receptor into a conformation that no longer binds ␣-BTX. On the other hand, slowing of the rate of ␣-BTX association could be due to either noncompetitive or competitive inhibition by proadifen. If proadifen competitively inhibits ␣-BTX association, one might expect a decrease in agonist affinity as the concentration of proadifen increases. However, we observed no decrease in agonist affinity even at the highest concentration of proadifen. Thus if the interaction between ␣-BTX and proadifen is competitive, the site of interaction does not overlap with the agonist binding site.
We observed an additional action of proadifen when we measured epibatidine binding in the presence of varying concentrations of proadifen. Our results establish that epibatidine binds in a biphasic manner in the presence of proadifen. However, the apparent ratio of high to low affinity sites varies with proadifen concentration. Whereas for both adult and fetal receptors the fraction of high affinity sites decreases as the proadifen concentration increases, the effects of proadifen differ quantitatively for the two types of receptors. At 60 M proadifen, adult receptors show approximately equally weighted high and low affinity components, whereas fetal receptors show about 3-fold more high than low affinity sites. In contrast, at 200 M proadifen, fetal receptors show equally weighted high and low affinity components, whereas adult receptors show a dominant low affinity component. The changes in the apparent ratio of sites indicate different proadifen sensitivities of the ␣␥, ␣␦, and ␣⑀ interfaces to the inhibitory effects of proadifen. In the fetal receptor, the ␣␦ site is more sensitive to proadifen than the ␣␥ site. Thus at low concentrations of proadifen, ␣-BTX binding to the ␣␦ site is inhibited more, and the apparent weight of the ␣␥ site is greater. As the proadifen concentration is increased, ␣-BTX binding to the ␣␥ also becomes inhibited while effects on the ␣␦ site reach a maximum. This yields equally weighted high and low affinity components at 200 M proadifen. Conversely, in the adult receptor the ␣⑀ site is more sensitive to proadifen than the ␣␦ site. At low concentrations of proadifen, the adult receptor shows two equally weighted components, but as the proadifen concentration is increased the weight of the high affinity ␣⑀ component decreases. Further experiments are required to determine whether changes in the ratio of high and low affinity binding sites owe to changes in the rates at which the ␣␥, ␣␦, and ␣⑀ interfaces bind ␣-BTX or whether the binding sites differ in their propensity to enter a conformational state which does not bind ␣-BTX.
Our overall results demonstrate that epibatidine binds with unique site and state selectivity to fetal and adult muscle AChRs. In contrast to carbamylcholine, epibatidine selects strongly for binding sites formed by the ␣␥ and ␣⑀ subunit interfaces compared with the ␣␦ interfaces. Further, unlike carbamylcholine, selectivity of epibatidine is maintained in the desensitized state. Structure-function studies using epibatidine should allow identification of new binding site determinants in the AChR thus refining our understanding of this receptor.