Mutagenesis, functional, labeling, and structural studies have provided detailed information about the ligand binding site in the nAChR (fig. 1A). Binding sites for agonists and competitive antagonists are located at the α–δ and α–ε (or α–γ) subunit interfaces.^** Studies have identified seven noncontinuous “loops” of amino acids that participate in agonist and competitive antagonist binding to the nAChR: loops A–C are on the α subunit (primary component of the binding site), and loops D–G are on the non-α subunit (complementary component of the binding site). The three dimensional arrangement of these loops became apparent by making analogies to the structure of AChBP.^4,6 AChBP is composed of 10 β strands arranged in an immunoglobulin-like topology. Several of the loops are located between the β strands, and the loops from the primary and complementary components come together at the interface between subunits. The binding site is centered around a conserved tryptophan in loop B that provides stability for the quaternary nitrogen moiety of agonists and antagonists via a strong π–cation interaction.⁷ Four additional aromatic amino acids are distributed around the primary and complementary components. The C loop may act as a cap that closes upon binding of agonist and greatly decreases the agonist dissociation rate.⁸ The amino acids on the complementary component are not conserved between the γ, δ, and ε subunits; this leads to subunit specificity for ligand binding.

Agonists such as acetylcholine have a higher affinity for the α–δ interface compared with the α–ε interface; the extent of the difference depends on both species (Torpedo receptors exhibit a 100-fold preference for acetylcholine binding to the α–δ interface⁹) and subtype (fetal mouse receptors exhibit a 32-fold preference,¹⁰ and adult mouse receptors exhibit very little preference¹¹). Both sites must be occupied for efficient opening of the channel.¹¹ Some competitive antagonists are known to bind differently to the two sites. Metocurine, for example, has a 70- to 170-fold higher affinity for the α–ε interface compared with the α–δ interface.^12,13 Occupation of just one of the sites by a competitive antagonist is sufficient to prevent channel opening. Most of the information about the residues that participate in antagonist binding comes from studies of either (+)-tubocurarine or metocurine.¹⁴ Even for these similar ligands (metocurine has three additional methyl groups; fig. 1), there are differences in sensitivities to mutations. Using computational chemistry to dock ligands to a homology model of the nAChR, Wang et al. ¹⁵ showed that (+)-tubocurarine and metocurine bind with different orientations within the α–ε interface and have different contacts with the amino acid residues there.

Although the clinical action of competitive nAChR antagonists, muscle paralysis, is a straightforward consequence of their molecular action, there are outstanding questions about the effects of nondepolarizing muscle relaxants. Specific issues include the mechanisms of tetanic fade¹⁶ and of muscle relaxant synergy.¹⁷ Although these phenomena are often ascribed to actions on presynaptic nAChRs, postsynaptic explanations may still be viable. As a first step to address these questions, we use a functional determination of current inhibition to examine competitive antagonism by (+)-tubocurarine, metocurine, pancuronium, vecuronium, and cisatracurium (fig. 1B) on mouse adult wild-type and five mutant nAChRs. We chose these mutations because they have been shown to affect metocurine and/or (+)-tubocurarine binding.^13,15,18,19 The αY198F mutation is in loop C, and the εD59 mutations and εD173 mutations are in loops E and G, respectively. We used the δD180K mutation to test the strength of binding to the α–δ interface. We postulated that the antagonists would show different sensitivities to mutations but that all of them would have a higher affinity for the α–ε interface compared with the α–δ interface.

Acetylcholine chloride (purity > 99%), (+)-tubocurarine chloride (purity 98%), and pancuronium dibromide (purity > 99%) were obtained from Sigma-Aldrich Corp. (St. Louis, MO). Metocurine iodide was synthesized from (+)-tubocurarine²³ at the Chemical Synthesis Center, Department of Chemistry, Stony Brook University (Stony Brook, NY). Purity of 99% was determined by ¹H-nuclear magnetic resonance. Vecuronium bromide was obtained as the clinical formulation from Bedford Laboratories (Bedford, OH), 1 mg/ml (1.8 mm) in a solution containing 2.1 mg/ml anhydrous citric acid, 1.6 mg/ml sodium phosphate, and 9.7 mg/ml mannitol. Dilutions were prepared in distilled water. The highest concentration of vecuronium used, 1 μm, contained 95 μm mannitol. Cisatracurium besylate was obtained as the clinical formulation from GlaxoSmithKline (Philadelphia, PA), 2 mg/ml (2.1 mm) in a 35% benzene sulfonic acid solution. Dilutions were prepared in distilled water. The highest concentration of cisatracurium used, 1 μm, contained 0.017% benzene sulfonic acid.

Cells were prepared for patch clamp recording by replacing the culture medium with an extracellular solution consisting of 150 mm NaCl, 5.6 mm KCl, 1.8 mm CaCl₂, 1.0 mm MgCl₂, and 10 mm HEPES, pH 7.3. Subsequently, 3−5 μl of polystyrene beads coated with a monoclonal antibody specific for the CD8 antigen (Dyna-beads; Dynal, Lake Success, NY) were added to the culture dish. Good expression of nAChR channels in excised patches was found for most cells having two or three beads attached.

Patch pipettes, filled with a solution consisting of 140 mm KCl, 5 mm EGTA, 5 mm MgCl₂, and 10 mm HEPES, pH 7.3, had resistances of 3−6 MΩ. An outside-out patch²⁴ with a seal resistance of 5 GΩ or greater was excised from a cell and moved into position at the outflow of a HSSE-2 rapid perfusion device (ALA Scientific Instruments, Westbury, NY). The perfusion system consisted of solution reservoirs, manual switching valves, a solenoid-driven pinch valve, and two tubes inserted into the culture dish and had a time resolution of less than 100 μs.²⁵ One tube contained extracellular solution without agonist (normal solution); the other contained extracellular solution with 300 μm acetylcholine (test solution). In the control protocol, the patch, initially perfused with normal solution, was exposed to a series of ten 0.25-s exposures to the test solution at 5-s intervals. Manual valves were used to connect to reservoirs containing a defined concentration of competitive antagonist with or without acetylcholine. An equilibrium (+/+) protocol was performed by exposing the patch to acetylcholine plus antagonist for 0.25 s, with a 5-s interval of antagonist alone. After switching back to antagonist-free solutions, the control protocol was repeated.²⁶ Results were accepted if the current amplitudes during the second control were 80% or more of those in the first; usually, the ratio was 95% or more. A recovery (+/−) protocol was used to determine the time needed for currents to recover after removal of antagonist. This protocol was performed with the same perfusion system by equilibrating the patch with antagonist alone for 5 s and exposing the patch to acetylcholine alone for 0.60 s as described previously.²⁷ Currents flowing during exposure of the patch to acetylcholine were measured with a patch clamp amplifier (EPC-9; HEKA Elektronik, Lambrecht, Germany), sampled at 100 μs per point and stored on a computer. Data analysis was performed off-line as described previously.^26,27 Experiments were performed at room temperature (20°− 23°C) and at a patch potential of −50 mV unless otherwise indicated.

The ratio of the peak inward current after the rapid onset phase in the presence of antagonist (I_ant, +/+ protocol) to that in the average of the two control currents, I₀, was calculated. This provided the fraction of receptors that were inhibited by an antagonist in the absence of acetylcholine. Concentration–response curves were generated from data obtained from experiments performed with six or seven different antagonist concentrations, each used on three to six different patches. Means and SDs are shown in the figures. Individual data points were fitted to the Hill equation: where [ant] is antagonist concentration, IC₅₀ is the concentration producing 50% inhibition, and n_H is the Hill coefficient. For δD180K receptors, control currents were usually less than 20 pA in absolute value. In these experiments, a single concentration point (at twice wild-type IC₅₀) was determined. The rate of recovery from antagonist inhibition (+/− protocol) was determined from 7−19 measurements of the recovery time constant for 3−5 different patches. The average and SD of the reciprocal of the (concentration-independent) time constant were calculated. This rate was interpreted to be the antagonist dissociation rate, .²⁷ The association rate, , was calculated from .

We characterized the inhibition produced by competitive antagonists (table 1). Our results for the IC₅₀ of antagonists to wild-type adult mouse nAChR are in good agreement with published values. Fletcher and Stein-bach¹² studied channels in a stably transfected fibroblast cell line. Electrophysiologic measurements resulted in IC₅₀ values of 11, 54, 129, and 139 nm for pancuronium, (+)-tubocurarine, metocurine, and atracurium, respectively (they did not study cisatracurium). Measurements on transfected frog oocytes by Paul et al.²⁹ produced IC₅₀ values of 5.5, 9.9, and 43.4 nm for pancuronium, vecuronium, and (+)-tubocurarine, respectively. It should be noted that both of these groups studied receptors in the whole cell activated with nonsaturating concentrations of acetylcholine (0.4 μM¹² or 10 μM²⁹) using relatively slow perfusion systems. The inherent assumption with this approach is that if the antagonists dissociate from the receptor, rebinding by the antagonist is more probable than binding by agonist (and subsequent activation of the channel). In our system, rapid (< 1 ms) application of a saturating concentration of acetylcholine (300 μm) allows measurement of the uninhibited current (and, from this, determination of the degree of inhibition) before the antagonists begin to dissociate. Actual measurement of dissociation rates (≤ 400/s; table 1) validates our approach.

The results of the mutation studies (table 1 and figs. 3 and ​and4)4) show that antagonists are affected differentially by mutations. The αY198F mutation in embryonic mouse nAChR was shown to have a greater effect on inhibition by (+)-tubocurarine than by pancuronium.¹⁸ Differences between (+)-tubocurarine and metocurine binding to AChBP³⁰ and human nAChR¹⁵ were identified by Sine's group. They found that although these two antagonists differ by only three methyl groups (fig. 1), they are differentially sensitive to mutations. Our results on mouse nAChR are similar. (+)-Tubocurarine is affected by mutations at αY198, but metocurine is not. Mutations at εD59 affect metocurine more than (+)-tubocurarine, and the opposite is true for εD173. In docking simulations, Sine's group made the surprising finding that these two antagonists assume docking orientations that are rotated by 180° (AChBP³⁰) or 60° (human nAChR¹⁵). Our results with pancuronium and vecuronium underscore the sensitive interactions between ligand and receptor. Although both antagonists are affected by the αY198F mutation, mutations in the ε subunit have small but significant effects on vecuronium and no effect on pancuronium. Inhibition by both of these antagonists is affected by the δ subunit mutation, whereas inhibition by (+)-tubocurarine or metocurine is not. Cisatracurium, which is a benzylisoquinolinium compound like (+)-tubocurarine and metocurine, is affected by mutations on the α and ε subunits similarly to metocurine. However, unlike metocurine, cisatracurium is sensitive to the δ subunit mutation and has distinct kinetic responses to mutations.

Binding constants for some antagonists have been obtained through measurements of the reduction of iodinated bungarotoxin binding. These experiments can reveal the affinity of both binding sites for the antagonists, but do not always make an association between the binding constant and a particular site. Although the slope of the electrophysiologically determined concentration–response curve is not a very sensitive way to characterize the lower-affinity site, the relatively high Hill slope for cisatracurium on wild-type receptors (table 1) can be explained if there are two binding sites with affinities of 62 ± 4 and 480 ± 180 nm, respectively.²⁷ Table 2 summarizes what is known about the selectivity of adult mouse nAChR for competitive antagonists.

We must be cautious about making inferences about selectivity solely from mutation experiments because the lack of effect of a mutation could mean that the mutated residue is either not involved or only weakly involved in binding. Conversely, the presence of an effect by a mutation could mean that the mutated residue is involved in binding or that the mutation caused an allosteric change to affect the affinity of the antagonist. Our mutagenesis results are consistent with previous results showing that the α–ε interface as the high-affinity site for metocurine and (+)-tubocurarine. For pancuronium, vecuronium, and cisatracurium, our finding that receptors with a mutation in the δ subunit are inhibited more than wild-type receptors suggests that that the α–δ interface is the higher-affinity site. The small and absent effects of ε subunit mutations on vecuronium and pancuronium respectively also support this idea. Experiments in which the ε subunit is replaced with a second copy of the δ subunit (fig. 4) were consistent with the idea that pancuronium, vecuronium, and cisatracurium now had two high-affinity binding sites. We could not obtain precise estimates of the interface selectivity of pancuronium or vecuronium. Because cisatracurium is potently affected by mutations in the ε subunit, this antagonist is not as selective (3- to 8-fold).

Although these α₂βδ₂ receptors were not inhibited as strongly as wild-type receptors by (+)-tubocurarine and metocurine, this effect was not in quantitative agreement with the results of toxin binding experiments that show high selectivity of these antagonists for the α–ε interface. We do not have a simple explanation for this finding. We note, however, that α₂βδ₂ receptors may perturb the structure of the receptor significantly more than a single site mutation.

Our conclusion differs from the results of a recent study of adult mouse nAChR exposed to pairs of competitive antagonists.¹⁷ The results of that study suggested that (+)-tubocurarine and pancuronium compete for the same binding site. However, those experiments were performed under conditions of relatively low receptor occupancy where synergistic effects are expected to be small. We are currently performing experiments under conditions of high receptor occupancy to clarify this issue.

Our kinetic measurements provide additional information about the binding of competitive antagonists to the nAChR. The αY198F mutation caused a relatively small, 20% decrease in the IC₅₀ of cisatracurium. However, this was the result of large changes in the kinetics of inhibition: a 3.6-fold decrease in the dissociation rate and a 2.8-fold decrease in the association rate. An equilibrium assay alone might have led to the conclusion that this amino acid plays a minor role in the interaction of cisatracurium with the receptor. One possible interpretation is that cisatracurium encounters αY198 on its journey into and out of its binding site but does not bind too closely to this residue. A similar effect was seen with the effect of this mutation on the kinetics of acetylcholine binding; both the association and dissociation rates decreased by a factor of 2, such that the overall affinity was unchanged.³¹ Most of the antagonist/mutation combinations we examined, however, suggest that the dissociation rate is the primary determinant of the IC₅₀ value. There are some notable exceptions, and we are currently conducting experiments with other analogs of (+)-tubocurarine to better understand these observations.