Chimeric Analysis of a Neuronal Nicotinic Acetylcholine Receptor Reveals Amino Acids Conferring Sensitivity to α-Bungarotoxin*

We have investigated the molecular determinants responsible for α-bungarotoxin (αBgtx) binding to nicotinic acetylcholine receptors through chimeric analysis of two homologous α subunits, one highly sensitive to αBgtx block (α1) and the other, αBgtx-insensitive (α3). By replacing rat α3 residues 184–191 with the corresponding region from the Torpedo α1 subunit, we introduced a cluster of five α1 residues (Trp-184, Trp-187, Val-188, Tyr-189, and Thr-191) into the α3 subunit. Functional activity and αBgtx sensitivity were assessed following co-expression in Xenopus oocytes of the chimeric α3 subunit (α3/α1[5]) with either rat β2 or β4 subunits. Agonist-evoked responses of α3/α1[5]-containing receptors were blocked by αBgtx with nanomolar affinity (IC50 values: 41 nm for α3/α1[5]β2 and 19 nm for α3/α1[5]β4). Furthermore, receptors containing the single point mutation α3K189Y acquire significant sensitivity to αBgtx block (IC50 values: 186 nm for α3K189Yβ2 and 179 nm for α3K189Yβ4). Another α3 chimeric subunit, α3/α7[6], similar to α3/α1[5] but incorporating the corresponding residues from the αBgtx-sensitive α7 subunit, also conferred potent αBgtx sensitivity to chimeric receptors when co-expressed with the β4 subunit (IC50 value = 31 nm). Our findings demonstrate that the residues between positions 184 and 191 of the αBgtx-sensitive subunits α1 and α7 play a critical functional role in the interaction of αBgtx with nicotinic acetylcholine receptors sensitive to this toxin.

Nicotinic acetylcholine receptors (nAChRs) 1 are multimeric ligand-gated ion channels expressed on skeletal muscle cells and on select groups of nerve cells in the peripheral and central nervous systems (1)(2)(3). Muscle nAChRs have pentameric structures made up of two ␣1 subunits and one each ␤1, ␥, and ␦(or ⑀) subunits; they are among the best characterized ion channels and serve as a model for understanding the structure and function of related ligand-gated channels responding to glycine, ␥-aminobutyric acid, and 5-hydroxytryptamine (4). Advances in the characterization of muscle nAChRs have been significantly aided by the discovery of a high affinity competitive antagonist, ␣-bungarotoxin (␣Bgtx). ␣Bgtx is used extensively in experiments on the molecular properties of nAChRs and for following expression, targeting, and clustering of these receptors on muscle during synapse formation (1)(2)(3)(4). Much less is known about nAChRs on neurons, in part because comparable antagonists are in limited quantity or are nonexistent. The purpose of this paper is to define the amino acid residues that are essential for high affinity ␣Bgtx binding through a chimeric subunit approach, by conferring ␣Bgtx sensitivity to a neuronal ␣ subunit that is normally insensitive to ␣Bgtx.
Previous work indicates that the main ␣Bgtx binding site is between residues 173 and 204 on the ␣1 subunit of the muscle nAChR. Specifically, studies of peptides derived from the Torpedo ␣1 sequence capable of binding ␣Bgtx with sub-micromolar affinity suggest that the major determinants of toxin binding are located in a region adjacent to the vicinal cysteines 192 and 193 (e.g. see Ref. 5). Recent studies of heterologously expressed muscle nAChRs have identified residues in this region of the native receptor that appear to interact with the short ␣-neurotoxin I from Naja mossambica mossambica (NmmI (6,7)) and with ␣Bgtx (8,9). Residues in this region are also involved in forming the binding sites for agonists and non-␣neurotoxin antagonists (6, 7, 10 -12). In such studies, singlesite mutations in the muscle type ␣1 subunit have not been very helpful in fully defining the ␣-neurotoxin binding site in the native nAChR, as most mutations studied fail to produce large changes in ␣Bgtx affinity (6 -9). Therefore, in this study, rather than eliminate ␣Bgtx binding, we have used site-directed mutagenesis of a neuronal nAChR to introduce a toxin binding site. As a consequence, we identified the molecular determinants responsible for ␣Bgtx binding to nAChRs.
Eleven different genes encode neuronal nAChR subunits (1-3): eight ␣ subunit genes (␣2-␣9) and three ␤ subunits (␤2-␤4). Sequence homology demonstrates that all muscle and neuronal nAChR subunits share a common structural motif; each has four hydrophobic, putative membrane-spanning domains and a long extracellular amino terminus that contains invariant cysteines at positions 128 and 142 (␣1 subunit numbering). In addition, all neuronal ␣ subunits contain the tandem cysteines at residues 192 and 193. Functional expression studies demon-strate that pairwise assembly of ␣2, ␣3, and ␣4 subunits with either ␤2 or ␤4 in the same complex is sufficient to form an ACh-gated ion channel (1,2). Although these neuronal nicotinic receptors have several physiological properties in common with nAChRs of muscle, they are not blocked by ␣Bgtx. In this study we take advantage of this key difference between muscle and neuronal nAChRs to investigate the molecular determinants responsible for ␣Bgtx binding.
By measuring ACh-evoked macroscopic currents from heterologously expressed receptors, we show that as few as five ␣1 residues can confer a high affinity ␣Bgtx block of chimeric ␣3/␣1 receptors. Using a similar chimeric ␣3 subunit in which the same stretch of residues is replaced with those of the ␣Bgtx-sensitive ␣7 subunit, we show that the ␣Bgtx binding site is effectively modular. This approach holds considerable promise for a full description of those residues required for toxin recognition in the ␣Bgtx-sensitive nAChRs and is likely to help elucidate the molecular basis for the insensitivity of various neuronal nAChRs to ␣-neurotoxins.

EXPERIMENTAL PROCEDURES
Chemicals-Unless otherwise noted, all chemicals were reagent grade from Sigma. ␣Bgtx was from Research Biochemicals Inc. (Natick, MA).
Oocyte Preparation and Injections-Oocytes were collected from mature Xenopus laevis frogs by survival surgery and were prepared for injection essentially as described by Bertrand et al. (13). Following brief collagenase treatment to remove the follicular cell layer, healthy stage V-VI oocytes were manually selected and incubated for 1 day in ND96 (in 96 mM NaCl, 2 mM MgCl 2 , 2 mM KCl, 1.2 mM CaCl 2 , 5 mM HEPES, pH 7.5) supplemented with 100 units/mL penicillin and 100 g/ml streptomycin. Oocytes were then injected with either cRNAs or cDNAs and maintained in antibiotic-supplemented ND96 for 2-3 days at 18°C before recording. cRNA transcripts were generated using the SP6 or T7 MessageMachine kit from Ambion (Austin, TX). Plasmids bearing cDNAs of the rat ␣3, ␤2, and ␤4 subunits and the ␣3/␣1 chimeric subunits were prepared using standard procedures. Following extrac-tion with organic solvents, cDNA plasmids were ethanol-precipitated and resuspended in injection buffer (in 88 mM NaCl, 1 mM KCl, 15 mM HEPES, pH 7.0). A Drummond Nanoject was used to inject ϳ23 ng (in 46 nl) for cRNAs and ϳ15 ng (in 14 nl) for cDNAs. Nucleic acids for the various subunit combinations were combined in equimolar ratios. For recordings, cells were perfused with OR2 medium (in 115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl 2 , 10 mM HEPES, 10 Ϫ3 mM atropine sulfate, pH 7.4). Concentrated stocks of acetylcholine chloride (ACh) and toxins in water were diluted directly in OR2.
Electrophysiological Recordings-ACh-evoked currents were measured from injected oocytes using the two-electrode voltage clamp method with a Warner 752B amplifier at a membrane potential of Ϫ60 mV. Data were acquired on a PC using Fetchex software (Axon Instruments, Foster City, CA). Electrodes of resistance 0.5-4.0 megaohms were filled with 3 M KCl. Recordings were performed in a custom Sylgard chamber or a Warner RC-8 chamber (each with an incubation volume of ϳ300 l) with gravity perfusion flow (ϳ5 ml/min). The flow of various drug solutions into the chamber was regulated using solenoid valves driven by a Warner VC-6 valve controller. As noted by others (14,15), neuronal nAChRs can display "run-down" with repeated agonist application, which we also observed in some batches of oocytes. For each subunit combination and each recording session, we performed rundown control determinations by measuring currents for the same dose (ϳEC 50 ) of ACh given 6 -10 times over a time period similar to that used for the experimental data acquisitions. In cases where run-down was observed, values for ACh-evoked inward currents were corrected by linear interpolation.
Co-application experiments were used to measure the onset of toxin block. After collecting initial responses evoked by a control dose of ACh, oocytes were challenged with a solution containing both the test dose of ACh and the desired ␣Bgtx concentration. Thereafter, cells were perfused between successive co-applications with a solution containing only the test concentration of toxin; in some cases, perfusion was stopped after ϳ30 s to conserve toxin. To measure the recovery from toxin block, ACh responses were obtained for each cell before and after exposing the oocyte to a blocking concentration of ␣Bgtx. After 10 min in toxin solution, the degree of block was measured, and recovery was initiated by perfusing the oocyte with toxin-free solution. The responses to test doses of ACh were measured at various times during the washout of toxin. Data measuring the concentration dependence of ␣Bgtx block (Figs. 3 and 6) were fit by nonlinear regression to the logistic dose-response function (16) included in Origin 5.0 (Microcal, Northampton, MA).

RESULTS AND DISCUSSION
We prepared four chimeric ␣3 subunit constructs, focusing on the region between residues 177 and 199. Residues from the ␣Bgtx-sensitive Torpedo ␣1 sequence were substituted for the homologous sites in the rat ␣3 subunit as shown in Fig. 1. For assessment of function, the chimeric ␣3 subunits were co-expressed with the appropriate non-␣ subunits (␤2 or ␤4) in Xenopus oocytes (13). The ␣3/␣1 [5] chimera, with the residue changes Y184W, E187W, I188V, K189Y, and N191T, and the point mutant ␣3K189Y both expressed well, producing robust ACh-evoked currents with little apparent deleterious effect on ACh sensitivity. As indicated in Table I, the EC 50 values for ACh of the various chimeric receptors co-expressed with ␤2 or ␤4 subunits, determined from dose-response titrations, are either the same as or less than those for the wild type combinations of ␣3␤2 and ␣3␤4. For ␣3 chimeric subunits co-expressed FIG. 1. Sequence comparison of ␣Bgtx-sensitive ␣1 and ␣Bgtx-insensitive ␣3 nAChR subunits. Residues changed in the chimeric subunit constructs are boxed. Asterisks indicate residues in a position to contribute to ␣Bgtx binding based on a substituted cysteine accessibility analysis (8). Underlined residues were found to give rise to intermolecular nuclear Overhauser effect signals in a complex formed between ␣Bgtx and an ␣1 subunit-derived dodecapeptide (23). Tyrosines 190 and 198 and cysteines 192 and 193 have been localized to the ligand binding site by chemical modification or cross-linking with agonist/antagonist derivatives (4).
with the ␤2 subunit, the EC 50 values ranged from 50 -80 M, in good agreement with the range 10 -150 M previously reported for wild type ␣3␤2 (17-21); for the chimeras co-expressed with ␤4, the values ranged from 70 -230 M compared with 100 -220 M reported for wild type ␣3␤4 (15,17,18,20). Increases in agonist sensitivities have also been noted by Luetje and coworkers for several nAChR chimeras (17,18). Because the EC 50 values we measured for the various chimeric subunits are in the ranges previously reported for the wild type combinations, we conclude that these substitutions cause no major structural perturbations to the receptors.
We also constructed the ␣3/␣1 [16] chimeric subunit (see Fig.  1) in which a more extensive region of Torpedo ␣1 (residues 177-199) was substituted into ␣3. No expression of this construct was observed following co-injection with either ␤2 or ␤4 subunit genes (46 oocytes from 5 frogs). High concentrations of ACh, dimethylphenylpiperazinium, and cytisine failed to evoke currents in these oocytes, suggesting that the additional substitutions in the ␣3/␣1 [16] chimera adversely affected subunit folding and/or assembly and membrane targeting. This phenomenon is not well understood but has been observed for other chimeric nAChRs in which residues of the amino-terminal extracellular domain were examined (17,22).
Five Residues from the ␣1 Subunit Confer ␣Bgtx Sensitivity to ␣3-We tested whether the five homologous residues of the Torpedo ␣1 sequence adjacent to the tandem cysteines 192-193 were sufficient when substituted into the rat ␣3 sequence to confer sensitivity to ␣Bgtx on these chimeric nAChRs. As shown in Fig. 2A, following an initial co-application of 100 M ACh and 15 nM ␣Bgtx and further incubation in toxin between successive test co-applications, the evoked currents of an oocyte expressing the ␣3/␣1 [5]␤4 combination were greatly reduced over a 15-min period. In contrast, the ACh-evoked currents from an oocyte expressing the wild type ␣3␤4 combination showed no block after a 10-min incubation with 1 M ␣Bgtx (Fig. 2B). In other experiments we confirmed that wild type ␣3␤2 and ␣3␤4 receptors show no block of ACh-evoked currents following incubations for up to 30 min in 10 M ␣Bgtx (data not shown).
Combined data from several co-application experiments using ␣3/␣1 [5]␤2 and ␣3/␣1 [5]␤4 nAChRs, shown in Fig. 2C, describe the rate of onset of ␣Bgtx block. Fractional block (1-I co-application /I control , proportional to the fraction of receptor bound with toxin) was determined as a function of the time of exposure to toxin. The onset of block can be well fit to a simple bimolecular association model for complex formation under pseudo-first-order conditions. We obtained a value of k app of 1.
. These rate constants are very similar to those measured for association of ␣Bgtx with muscle type receptors (24,25). The plateau value of ϳ0.6 for the two data sets shown in Fig. 2C represents the maximal block achievable with 15 nM ␣Bgtx. Operationally, these data also show that for receptors incorporating ␣3/␣1 [5] chimeric subunits, an incubation time of 10 min is sufficient to approach full equilibration for ␣Bgtx concentrations in excess of 15 nM.
The concentration dependence of ␣Bgtx block of chimeric ␣3/␣1[5]-containing receptors was determined following a 10min incubation with ␣Bgtx as shown in Fig. 3. The solid curves represent the best fit to the logistic equation (16), from which 100 Ϯ 10 (n ϭ 6) Ͼ100,000 NA FIG. 2. ACh-evoked currents in oocytes expressing wild type and chimeric ␣3 subunits. A, individual current traces are shown superimposed for an oocyte expressing the ␣3/␣1 chimeric subunit in combination with rat ␤4. The control responses to a 6-s pulse of 100 M ACh were recorded with the two-electrode voltage clamp method at a membrane potential of Ϫ60 mV. At time 0, a co-application of 100 M ACh ϩ15 nM ␣Bgtx was begun, and the response was recorded. Between subsequent co-applications at the indicated times, the cell was maintained in 15 nM ␣Bgtx. B, an oocyte expressing the wild type ␣3␤4 nAChR was challenged with a control 6-s pulse of 100 M ACh, and the response (left trace) was recorded. After a 10-min incubation in 1 M ␣Bgtx, the cell was again challenged with 100 M ACh; the post-toxin response (right trace) is offset on the time axis for clarity, and the arrows indicate onset of ACh application. C, data from co-application experiments as described in A above for chimeric receptors ␣3/␣1 [5]␤2 (open squares, n ϭ 6 cells) and ␣3/␣1 [5]␤4 (open circles, n ϭ 3 cells) are shown as plots of 1 Ϫ I co-application /I control (fractional block) versus time in 15 nM toxin, where I co-application is the peak response evoked by the co-application at that time point. Curves represent fits to a bimolecular association with apparent association constants of 1.4 Ϯ 0.1 ϫ 10 5 M Ϫ1 s Ϫ1 for ␣3/␣1 [5]␤2 and 1.7 Ϯ 0.3 ϫ 10 5 M Ϫ1 s Ϫ1 for ␣3/␣1 [5]␤4.
the IC 50 values of 19 Ϯ 3 nM and 41 Ϯ 9 nM for the ␣3/␣1 [5]␤4 and ␣3/␣1 [5]␤2 receptors, respectively, were determined. Significantly, these IC 50 values are the same order of magnitude as that found for the block of ␣3␤2 receptors by -bungarotoxin (Bgtx, also known as neuronal bungarotoxin), a related toxin from the same venom as ␣Bgtx (26), and only about 1 order of magnitude higher than the IC 50 of 2.4 nM ␣Bgtx measured for mouse muscle receptors expressed in oocytes (27).
For neuronal nAChRs, the ␤ subunit contributes extensively in determining the affinities for many ligands, including the sensitivity to Bgtx (17,20,26,28). For example, Bgtx blocks oocyte-expressed ␣3␤2 receptors with high affinity in a prolonged manner but blocks ␣3␤4 receptors transiently and only at high concentrations (17,20,28). This is in contrast to our finding that the chimeric subunit ␣3/␣1 [5] co-expressed with either ␤2 or ␤4 showed little difference in affinity for ␣Bgtx. Recently, Sine (10) has shown that a conserved Leu residue (position 119 in the ␥ and ⑀ subunits, 121 in ␦) was in proximity to the ␣Bgtx binding site of muscle receptors. Interestingly, in both rat ␤2 and ␤4 subunits the homologous position is also occupied by Leu. If ␣Bgtx makes contacts with the neighboring ␤ subunit in the ␣3/␣1 chimeric receptors, these contacts are likely to be similar for both ␤2 and ␤4. Harvey and Luetje (18) find that the region 54 -63 of the ␤2 subunit, in particular residue Thr-59, is the major determinant of the selectivity of Bgtx for ␣3␤2 over ␣3␤4 receptors. Together these results suggest possible differences in which residues of ␣Bgtx and Bgtx are oriented toward the ␤ subunit in the nAChR-bound state.
Importance of Position 189 in ␣Bgtx Binding-In the region between residues 184 -191, one of the most divergent sites between ␣1 and ␣3 sequences occurs at position 189; this residue is Tyr in Torpedo ␣1 (Phe in mouse ␣1) and Lys in ␣3. In fact, all known ␣Bgtx-insensitive ␣ subunits in rat, chick, and human have Lys in this position. To test the importance of residue 189 in ␣Bgtx sensitivity, we mutated Lys-189 in ␣3 to Tyr and expressed the point mutant subunit ␣3K189Y with ␤ subunits in oocytes. As shown in the dose-response curves for the point mutant combinations (Fig. 3), ␣Bgtx blocked the ACh-evoked currents of the ␣3K189Y␤2 and ␣3K189Y␤4 receptors. For each oocyte, the maximal block at each concentration of ␣Bgtx was calculated by extrapolation to t ϭ 0 of the washout/recovery period using 4 -8 peak current measurements over the course of 10 min following removal of ␣Bgtx. In other experiments using co-applications of ACh and ␣Bgtx (i.e. block measured in the presence of toxin), the fractional block obtained after 10 min of toxin incubation was identical to that obtained by extrapolating the recovery data. The IC 50 values for the ␣Bgtx block of the subunit combinations ␣3K189Y␤4 (179 Ϯ 59 nM) and ␣3K189Y␤2 (186 Ϯ 42 nM) were about an order of magnitude greater than those of the ␣3/␣1 [5] receptors and again were not significantly affected by which of the two ␤ subunits was used for co-expression. Considering that the IC 50 values of ␣Bgtx block of wild type ␣3␤2 and ␣3␤4 receptors must be greater than 100 M (see Table I), the sensitivity to block of these ␣3K189Y chimeric receptors represents an increase in affinity of approximately 3 orders of magnitude or more compared with the wild type ␣3.
The observation that the single amino acid substitution ␣3K189Y results in considerable ␣Bgtx affinity demonstrates a central role for position 189 in mediating ␣Bgtx interactions. However, as both the ␣3/␣1 [5] and ␣3K189Y chimeric receptors fall short of the affinity for ␣Bgtx characteristic of wild type Torpedo nAChR (cf. 24,27), other residues are also likely to contribute to high affinity binding in ␣1 subunit-containing nAChRs. On the basis of double mutant cycle analysis, Taylor and co-workers (7) suggest that residues Val-188, Tyr-190, Pro-197, and Asp-200 are involved in binding the short ␣-neurotoxin Nmm I. Although mutations at these four sites had less dramatic effects on ␣Bgtx binding (6), substituted cysteine mutagenesis of residues Val-188 and Tyr-190 suggest that these as well as Trp-187, Phe-189, and Pro-194 (studied in mouse ␣1) also play a role in ␣Bgtx binding (8). Of these proposed sites, the ␣3/␣1 [5] chimera includes Trp-187, Val-188, Tyr-189, Tyr-190, and Asp-200 but not Pro-194 and Pro-197. Interactions involving residues in the vicinity of Pro-194 and Pro-197 in the ␣1 subunit may therefore contribute to the 10 -20-fold higher affinity ␣Bgtx binding seen with the native muscle receptor.
Three of the other residue changes made in the ␣3/␣1 [5] chimera are fairly conservative: Y184W, I188V, N191T. In contrast, the substitution of Trp for Glu at position 187 is more dramatic. To test whether the E187W change together with the other three conservative substitutions might contribute significantly to ␣Bgtx binding, we prepared the ␣3/␣1[4] chimera ( Fig. 1) and co-expressed it with the ␤4 subunit. Although the ␣3/␣1 [4]␤4 combination led to robust ACh-evoked currents and had an EC 50 value (40 Ϯ 1 M) in the range observed for the other ␣3 chimeras, no measurable block of these receptors was evident after incubation with 1.5 M ␣Bgtx. The average fractional block using 1.5 M ␣Bgtx was 0.06 Ϯ 0.03 (4 cells; see also Fig. 6). Because no significant block was detected, there is no indication that residues 184, 187, 188, and 191 participate directly in ␣Bgtx binding in the absence of Tyr-189. These positions, when occupied by the ␣1 amino acids, may create a local conformation for position 189 that is more favorable than that found in the ␣3K189Y chimera. Nonetheless, this result provides further support for the importance of residue 189 in determining ␣Bgtx sensitivity and suggests that the role of each of the other ␣1-derived residues in ␣Bgtx recognition needs to be assessed in the context of Tyr-189.  Table  I). Oocytes were then incubated with ␣Bgtx for 10 min, and the responses to the test dose of ACh were measured in rapid succession following the removal of toxin. The fraction of the control response (I post-toxin /I pre-toxin ) is plotted versus the ␣Bgtx concentration; data points represent averages (ϮS.E.) for 3-5 different oocytes at each ␣Bgtx concentration. The following ACh concentrations were used to elicit responses: ␣3/␣1 [5]␤2 (open squares) and ␣3K189Y␤2 (open circles), 30 M; ␣3/␣1 [5]␤4 (filled squares) and ␣3K189Y␤4 (filled circles), 100 M. The IC 50 values obtained from these data are given in Table I.

Different
Affinities of ␣3/␣1 [5] and ␣3K189Y nAChRs for ␣Bgtx Are Predominantly Due to Different Dissociation Rates-As shown above, the rate constants of association of ␣Bgtx to ␣3/␣1 [5]␤2 and ␣3/␣1 [5]␤4 chimeric receptors were very similar to that for the association of ␣Bgtx with the muscle type of nAChR (cf. 24,25). This suggests that the difference in affinities for ␣Bgtx between ␣3/␣1 [5] receptors and muscle type nAChRs is due to a difference in dissociation rates. To test this, we carried out measurements of the time course of recovery from toxin block as an indicator of toxin dissociation. After a 10-min incubation with ␣Bgtx, oocytes were continuously perfused with buffer lacking toxin and periodically challenged with the test dose of ACh. Representative results are presented in Fig. 4, where the fraction of maximal block [(I control Ϫ I tϭx )/ (I control Ϫ I t ϭ 0 )] is plotted as a function of the time of washout. Single exponential fits (solid curves through data points) revealed the following half-times (t1 ⁄2 ) for recovery: 233 min for Torpedo ␣1␤␥␦, 18 min for ␣3/␣1 [5]␤4, and 1.0 min for ␣3K189Y␤4. The same ϳ20-fold difference in t1 ⁄2 values for ␣3/␣1 [5] and ␣3K189Y was observed for ␤2-containing chimeric receptors (Table I). In all cases, the differences in the IC 50 values of ␣Bgtx block demonstrated in Fig. 3 correlate well with the rates of recovery from block obtained for the chimeric receptors.
Modularity of ␣Bgtx Binding Sequences-The cysteine residues at positions 192 and 193 are invariant in all nAChR ␣ subunits, and the Gly at position 183 is also highly conserved, occurring in 63% of known sequences. On this basis we postulated that the region 183-193 is structurally conserved in nAChR ␣ subunits and that it may be modular with respect to ␣Bgtx binding. We tested this by preparing another chimeric subunit in which the residues of ␣3 in the region 183-193 were substituted with those of the ␣Bgtx-sensitive rat ␣7; this yielded the ␣3/␣7 [6] construct. Fig. 5 shows a further comparison of some ␣Bgtx-sensitive and ␣Bgtx-insensitive ␣ subunits in this region. Note that of the residues that are not invariant or highly conserved among all ␣ subunits in this region, only position 189 is well conserved (either Tyr or Phe) in ␣Bgtxsensitive subunits.
We found that the ␣3/␣7 [6] chimera, when co-expressed with the rat ␤4 subunit, was highly sensitive to block by ␣Bgtx. As shown in Fig. 6, ACh-evoked currents of oocytes expressing the ␣3/␣7 [6]␤4 combination were blocked by ␣Bgtx concentrations in the nanomolar range (IC 50 ϭ 31 Ϯ 2 nM) using a 10-min toxin incubation. This is similar to the apparent affinity of ␣Bgtx for the ␣3/␣1 [5] chimera co-expressed with either the ␤2 or ␤4 subunit. The ␣3/␣7 [6] chimera has an apparent affinity for ␣Bgtx about 1 order of magnitude lower than that of wild type ␣7 receptors (29,30). Although the residues 183-193 form a high affinity ␣Bgtx binding unit that is modular in the sense that it can be substituted into the background of insensitive subunits to confer binding, amino acids elsewhere in the sequence must also contribute to give wild type affinity. In comparing ␣1, ␣3, and ␣7 sequences in the region 183-193, position 189 stands out as being most likely to determine the ␣Bgtx sensitivity of ␣1 and ␣7 receptors. It is possible and perhaps likely that ␣Bgtx interacts with ␣1 and ␣7 nAChRs in subtly different ways, but our results suggest that the core of these interactions is mediated by residues 183-193 and that an aromatic ring at position 189 is an important feature in ␣Bgtx recognition.
Chimeric Mouse ␣1 with Val-188 and Phe-189 Replaced by Their ␣4 Counterparts-Chemical modifications of a substituted cysteine have suggested that Phe-189 of the ␣1 subunit is in the proximity of the ␣Bgtx binding site in native mouse muscle nAChRs (8). In contrast, mutation of Phe-189 to Lys leads to less than a 3-fold reduction in the apparent dissociation constant for NmmI (6). One interpretation of these results, together with those presented here, is that position 189, although playing an important role in ␣Bgtx binding, may not be as critically involved in the recognition of the short ␣-neurotoxin NmmI. Because Val-188 has been suggested to contribute to contacts with NmmI (7), we constructed a double mutant of the mouse muscle ␣1 subunit (␣1/␣4 [2], see Fig. 5) in which Val-188 is replaced with the positively charged residue Arg (as in the rat ␣4 subunit), and Phe-189 is replaced with Lys (as in the rat ␣3 and ␣4 subunits). Ackermann and Taylor (6) show that the introduction of a positive charge at position 188 (V188K) of the mouse ␣1 subunit leads to a 20-fold reduction in affinity for 1 of the 2 neurotoxin binding sites (that associated with the ␣␦ subunit interface) and a 390-fold reduction in affinity for the other (at the ␣␥ interface). NmmI has the unique characteristic among ␣-neurotoxins of being able to discriminate between the two neurotoxin binding sites in the Torpedo nAChR (31). We reasoned that the effect of a V188R  Table I).
FIG. 5. Sequence comparison of ␣Bgtx-sensitive and ␣Bgtxinsensitive nAChR subunits. Nicotinic receptors containing ␣1 or ␣7 subunits are highly sensitive to block by ␣Bgtx, whereas nAChRs containing ␣3 or ␣4 subunits are completely insensitive. The chimeric subunit ␣3/␣7 [6] has the 6 divergent residues of rat ␣3 in the region 184 -191 replaced with those of rat ␣7. The chimeric ␣1/␣4 [2] subunit is a double mutant in which Val-188 and Phe-189 of the mouse ␣1 sequence are replaced with Arg and Lys, respectively; these are the amino acids occupying the homologous positions in the rat ␣4 sequence. Residues changed in the chimeric subunit constructs are boxed. mutation should be similar to V188K, allowing us to analyze the results in terms of a direct functional comparison between the mouse ␣1 subunit and the ␣Bgtx-insensitive rat ␣4 subunit. If the two residues Val-188 and Phe-189 are indeed important for ␣Bgtx binding, the double mutation V188R and F189K in ␣1/␣4 [2] would be expected to produce a very dramatic reduction in ␣Bgtx affinity. As shown in Fig. 6, upon co-expression with the mouse ␤, ␥, and ␦ subunits, the ␣1/␣4 [2] chimera gave rise to a receptor that remained very sensitive to ␣Bgtx block. With an IC 50 value of 7 Ϯ 0.5 nM, the sensitivity to ␣Bgtx of receptors containing the ␣1/␣4 [2] chimera was reduced only about 3-fold compared with wild type mouse ␣1␤␥␦ receptors (27). This is much less than would be predicted based on the binding studies with NmmI toxin (6) if one assumes that the same residues are recognized by both ␣-neurotoxins.
First of all, our results suggest a fundamental difference in the role of positions 188 and 189 in ␣Bgtx binding and in NmmI binding. The conclusion that ␣Bgtx and NmmI differ substantially in their modes of interaction with mouse muscle nAChRs receives further support from the recent study of Osaka et al. (32). These authors report that Glu-176 of the ␥ subunit comes into close apposition with Lys-27 of NmmI, and that positions 175 and 176 of the ␥ and ␦ subunits contribute to the high affinity of the binding sites at the ␣␥ and ␣␦ interfaces. These conclusions were based on the observation that the homologous residues of the ⑀ subunit, Thr and Ala, conferred 1000-fold lower NmmI affinity to the mouse ␣1␤⑀␦ nAChR. Of most relevance to the present study, Osaka et al. (32) show that the on-rate of ␣Bgtx association to ␣1␤⑀␦ was reduced only about 4-fold compared with mouse ␣1␤␥␦ receptors (with no significant effect on off-rate), in marked contrast to the dramatic effects observed for the NmmI interaction. Although Lys-27 in NmmI appears to play a very important role in binding to muscle nAChRs (6, 7, 32), the removal of the positive charge from the corresponding residue in ␣Bgtx by the K26A substi-tution leads to only a 10-fold reduction in ␣Bgtx affinity (27). Furthermore, ␣Bgtx blocks homo-oligomeric neuronal ␣7 receptors with high affinity, whereas short ␣-neurotoxins show greatly diminished activity on these receptors (33). In combination, these results suggest significant differences in the molecular basis of binding to nAChRs for the short and long ␣-neurotoxins (34).
A second conclusion derived from the results with the ␣1/ ␣4 [2] chimera is that the introduction of a positively charged side chain and the concomitant removal of the aromatic side chain at position 189 cannot alone account for the marked ␣Bgtx insensitivity of the ␣3 and ␣4 subunits. This contrasts sharply with our demonstration here that the reciprocal mutation at position 189 in the ␣3 background (i.e. ␣3K189Y) leads to a dramatic enhancement of ␣Bgtx sensitivity of more than 2 orders of magnitude, from an IC 50 Ն100 M to ϳ0.2 M (See Fig. 3 and Table I). In the ␣1 background, it is possible that multiple alternative contacts with ␣Bgtx can accommodate and mitigate the effects of single-site mutations such as F189K, making the ␣Bgtx-nAChR interface effectively over-determined. Differences in the orientations of introduced side chains due to differences in the local environment of the neighboring sequence may also contribute to the apparent nonreciprocal nature of the amino acid substitutions studied at position 189. In any case, the use of the ␣3 subunit as a background allows for the sensitive detection of residues that contribute to ␣Bgtx binding in the ␣1 subunit. Additional substitution studies will be needed to test whether the major effect of the K189Y mutation in ␣3 is due to the introduction of an aromatic side chain allowing favorable interactions to occur or due to the removal of a positive charge that interferes with toxin-receptor association. Also, the further application of this approach to investigate the role of other residues divergent in the ␣1 and ␣3 sequences should allow a full description of such ␣1 residues that directly contribute to ␣Bgtx recognition.
General Implications for ␣Bgtx Binding to nAChRs-Our results with a homologous substitution analysis utilizing an ␣Bgtx-insensitive ␣3 subunit background clearly indicate an important role for residues 184 -191 in mediating ␣Bgtx recognition for native nAChRs. It is unlikely that the ␣3 mutations studied here cause gross structural alterations that somehow permit aberrant ␣Bgtx binding, given that the chimeric receptors all have EC 50 values for ACh activation in the same range as those previously determined for the wild type ␣3␤2 or ␣3␤4 combinations. The studies reported here also provide important support for the physiological relevance of the NMR-based structure of the complex formed between ␣Bgtx and a dodecapeptide corresponding to ␣1 residues 185-196 (23). Tyr-189 was among 5 receptor residues found to be in close contact with ␣Bgtx in this slow exchange protein-peptide complex.
The results reported here also demonstrate the benefit and desirability of carrying out reciprocal mutant and chimeric analyses in the study of ligand binding sites. For large ligands with high affinity like ␣Bgtx, the intermolecular interactions are probably over-determined, and such redundant interactions may mask the important contributions of individual residues. In addition to mutagenesis aimed at eliminating ligand binding, reciprocal mutations designed to introduce a gain of function such as ligand binding are critical to a complete understanding of ligand-receptor interactions.
␣6 subunits. In any event, further studies utilizing the chimeric approach, single-site mutagenesis, and double mutant cycle analysis will be required to fully elucidate of the basis of the remarkable affinity and selectivity of ␣Bgtx.