Biotinylation of Substituted Cysteines in the Nicotinic Acetylcholine Receptor Reveals Distinct Binding Modes for α-Bungarotoxin and Erabutoxin a*

Although previous results indicate that α-subunit residues Trp187, Val188, Phe189, Tyr190, and Pro194 of the mouse nicotinic acetylcholine receptor are solvent-accessible and are in a position to contribute to the α-bungarotoxin (α-Bgtx) binding site (Spura, A., Russin, T. S., Freedman, N. D., Grant, M., McLaughlin, J. T., and Hawrot, E. (1999) Biochemistry38, 4912–4921), little is known about the accessibility of other residues within this region. By determining second-order rate constants for the reaction of cysteine mutants at α184–α197 with the thiol-specific biotin derivative (+)-biotinyl-3-maleimidopropionamidyl-3,6-dioxaoctanediamine, we now show that only very subtle differences in reactivity (∼10-fold) are detectable, arguing that the entire region is solvent-exposed. Importantly, biotinylation in the presence of saturating concentrations of the long neurotoxin α-Bgtx is significantly retarded for positions αW187C, αF189C, and reduced wild-type receptors (αCys192 and αCys193), further emphasizing their major contribution to the α-Bgtx binding site. Interestingly, although biotinylation of position αV188C is not affected by the presence of α-Bgtx, erabutoxin a, which is a member of the short neurotoxin family, inhibits biotinylation at position αV188C, but not at αW187C or αF189C. Taken together, these results indicate that short and long neurotoxins establish interactions with distinct amino acids on the nicotinic acetylcholine receptor.

The nicotinic acetylcholine receptor (AChR) 1 is the major prototype for neurotransmitter-gated ion channels and is found at high concentrations in the postsynaptic membranes of muscle cells, where it mediates the rapid propagation of electrical signals at the neuromuscular junction. It is a pentameric protein composed of four subunit types in a molar ratio 2␣:␤:␥:␦ (see Ref. 1 for review).
An important first step in assessing the structure and function of such ion channels at a molecular level is to determine their transmembrane topology. To address this issue, a variety of techniques have been developed (see Ref. 2 for review). The most commonly used methods are the epitope protection assay (3)(4)(5), in which an epitope that is recognized by a specific antibody is fused to the protein of interest, and N-linked glycosylation tagging, wherein N-linked glycosylation sites can be engineered into the protein under investigation and glycosylation can then be evaluated (6 -8).
The above methods, however, are only useful to assess overall topology of membrane proteins. For the nAChR, there is general consensus on the overall topology, although final proof will have to await high resolution structural data; each subunit possesses a large, extracellular amino-terminal domain, which is followed by four transmembrane-spanning regions and a short extracellular carboxyl terminus (9). In contrast, the key structural issue of whether individual residues are solventexposed has not been resolved. To address this issue, Gallivan et al. (10) have recently employed the in vivo nonsense suppression technique to incorporate derivatives of the unnatural amino acid biocytin into the nAChR heterologously expressed in Xenopus oocytes. By evaluating the binding of 125 I-streptavidin to biotinylated receptors, they studied the surface exposure of individual residues comprising the main immunogenic region (spanning positions 67-76; Ref. 11) and showed that position ␣70 was highly exposed.
In the current study, we have used the substituted cysteine accessibility method (12,13) to systematically map the accessibility of individual residues between positions 184 and 197 of the ␣-subunit, the main determinant for agonist and competitive antagonist binding to the nAChR (14 -16). To achieve this, we have introduced cysteine residues into the nAChR and labeled them with thiol-reactive, water-soluble biotin derivatives. Subsequently, we precipitated biotinylated receptors with immobilized streptavidin and probed the immunoprecipitates by Western blotting. Previous studies of oocyte-expressed Cys substitution mutations of ␣-subunit residues 181-197 (17) indicate that the majority of these substitutions are well tolerated and lead to minimal perturbations in receptor function.
Here, we show that positions 184 -197 are all amenable to biotinylation, suggesting that the entire region is surface-exposed. In addition, modifications with the uncharged biotin derivative occur with similar rates for all of these residues. Moreover, we demonstrate that preincubation with the competitive antagonist ␣-Bgtx, a long-chain ␣-neurotoxin, selectively blocks biotinylation of positions 187, 189, and cysteines 192 and 193 in reduced wild-type receptors, demonstrating the importance of these residues in the binding of ␣-Bgtx and further supporting results we obtained previously (16). In contrast, preincubation with Ea, a short-chain ␣-neurotoxin, yields a different footprint, preventing only position 188 and reduced wild-type receptors from biotinylation. Thus, our findings strongly argue that long-and short-chain ␣-neurotoxins interact selectively with different positions when blocking agonist binding on the nAChR.

Reagents
MTSEA-biotin was from Toronto Research. PEO-biotin and streptavidin-agarose beads were from Pierce, mAb 35 from Research Biochemicals International, and Protein G-agarose beads from Santa Cruz Biotechnology.

Mutagenesis
We used a cytomegalovirus-based expression vector (GWI, British Biotechnology, Oxford, United Kingdom) to express the cDNAs for the ␣-, ␤-, ␥-, and ␦-subunits of the mouse muscle nicotinic AChR. Mutations were introduced using the Quikchange mutagenesis kit (Stratagene, La Jolla, CA) following the manufacturer's specifications. Mutations were confirmed by diagnostic restriction enzyme digests and bidirectional sequencing of the entire insert following a DyeDeoxy terminator protocol (Perkin-Elmer).

Transfections and Cell Lines
These have been described previously (16).
Step 1: N-Boc-2,2Ј-(ethylenedioxy)bis(ethylamine) Derivative (18)-2,2Ј-(Ethylenedioxy)bis(ethylamine) (38 g, 257 mmol) was dissolved in 40 ml of deionized water and 40 ml of dioxane. While this solution was stirring, di-tert-butyl dicarbonate (8 g, 36.7 mmol) dissolved in 80 ml of dioxane was added dropwise. Formation of the bis-Boc-protected product was discouraged by performing the dropwise addition at 0°C. The reaction was warmed to room temperature and stirred for 6 h. The solvents were removed by rotary evaporation, and the product was dissolved in ethyl acetate and washed with saturated NaHCO 3 . The organic layers were dried over sodium carbonate, gravity-filtered, and rotary-evaporated to remove ethyl acetate. Flash chromatography using a 20% MeOH, 1% NH 4 OH, CH 2 Cl 2 solvent system gave a 66% yield (6 g, 2.4 mmol) of a yellowish oil.
Step 3: NЈ-Biotinyl-2,2Ј-(ethylenedioxy)bis(ethylamine) Derivative (19)-The Boc-protected biotin diamine (0.5 g, 0.1 mmol) was dissolved in 4.9 ml of trifluoroacetic acid and stirred at room temperature for 20 min. The reaction flask was rotary-evaporated to remove trifluoroacetic acid, producing a brownish oil. This oil was dissolved in a few drops of deionized water and placed under vacuum to remove any remaining trifluoroacetic acid. Thin layer chromatography (15% MeOH/CH 2 Cl 2 ) showed the product spot very close to the base line, suggesting that it contained the extremely polar free amine.
Step 5: N-Methanethiolsulfonyl-NЈ-biotinyl-2,2Ј-(ethylenedioxy)bis-(ethylamine)-The iodobiotin diamine (0.75 g, 1.4 mmol) was taken up in 10 ml of dimethylformamide (DMF). To this solution was added sodium methanethiolsulfonate (0.37 g, 0.003 mmol), and the reaction was allowed to stir at room temperature for 2 h. The flask was placed under high vacuum to remove the DMF, and the product was flashchromatographed using a 15% MeOH/CH 2 Cl 2 solvent system. An 88% yield (0.64 g) of a yellowish oil was isolated. NMR spectra for MTSEDEbiotin and its intermediates were determined to confirm the purity of the product (data not shown) and are available upon request.
Step Step 7: Sodium Methanethiolsulfonate (21)-Sodium hydrosulfide was dried over P 2 O 5 for 3 days. This dried sodium hydrosulfide (11 g, 20 mmol) was then dissolved in 150 ml of absolute ethanol. Methanesulfonyl chloride (11.4 g, 9.9 mmol) was added dropwise to this solution as it stirred at room temperature and under a nitrogen atmosphere. After all of the methanesulfonyl chloride had been added, the reaction was allowed to stir for another 2 h, under nitrogen. As nitrogen gas was bubbled through the reaction, it was forced through a drying tube and then bubbled through 2000 ml of bleach in order to neutralize the developing hydrogen sulfide gas. After 2 h, the reaction flask was heated to 65-70°C for 1 h. The flask was allowed to cool, and then 100 ml of absolute ethanol were added before leaving the flask under nitrogen overnight. After this time, the solution was gravity-filtered to remove NaCl and then rotary-evaporated to remove ethanol. Recrystallization was performed with warm ethanol to yield a white solid with a melting point of 271.5°C. The yield was 3.8 g (2.88 mmol).
Step 8: Iodoacetic Anhydride-Iodoacetic acid (20 g, 107 mmol) was dissolved in 290 ml of ethyl acetate. Diisopropyl carbodiimide (6.8 g, 54 mmol) was added to this solution, and the reaction was stirred at room temperature for 1 h under nitrogen. IR spectroscopy showed that the reaction had gone to completion by the formation of two strong, sharp peaks at 1793.7 and 1735.6 cm Ϫ1 and the disappearance of a strong, broad peak at 3401.5 cm Ϫ1 (COOH peak). Ethyl acetate was removed by rotary evaporation, yielding a ruby red oil of iodoacetic anhydride.

Covalent Cysteine Modification with Biotin Reagents and Preincubations with ␣-Bgtx or Erabutoxin a
Two days after transfection, cells were harvested by gentle agitation in phosphate-buffered saline containing 5 mM Na 2 -EDTA (ϳ0.5-1 ϫ 10 7 cells obtained from one 75-cm 2 tissue culture flask). After a brief centrifugation at ϳ600 ϫ g, the cells were resuspended in high potassium Ringer's solution (22), pooled, divided into 300-l aliquots, and incubated for the specified times with 5-500 M MTSEDE-biotin or PEObiotin. For MTSEA-biotin, the reagent was dissolved in Me 2 SO instead of water before being added to the cells at 500 M. For each biotin reagent, we added an excess of BrACh (1.5 mM) to terminate the reaction. The unbound biotin was removed by pelleting the cells (2 min at 20°C), and resuspending the pellet in 0.5 ml of high potassium Ringer's. This wash was repeated three times in total. For preincubations with ␣-Bgtx or Ea, the cells were incubated for 2 h at room temperature with 10 M amounts of the respective toxin to allow for a saturation of binding sites. PEO-biotin was then added directly into the tubes for the times indicated. Typically, HEK-293 cells transiently transfected with wild-type AChR subunits yielded 50 -100 fmol of biotinylated surface receptor/cm 2 of confluent cells.

Precipitation with Streptavidin-Agarose Beads
After treatment with PEO-biotin, cells were lysed in 450 l of ice-cold RIPA solution (1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) with added proteinase inhibitors (10 g/ml aprotinin, 200 M phenylmethylsulfonyl fluoride, and 100 M benzamide) and incubated on ice for 15 min. The lysate was centrifuged at 14,000 ϫ g for 15 min, and the supernatant was precipitated with 50 g of streptavidin-agarose beads (Pierce) at 4°C overnight. Under these conditions, we found that the maximal amount of biotinylated receptor was precipitated, as the addition of larger volumes of beads did not lead to an increase in detectable ␣-subunit. The precipitated samples were washed three times with 500 l of RIPA solution (4,000 ϫ g, 2 min, 4°C). Typically, lysed cells containing ϳ1 mg of total protein were precipitated and the equivalent of ϳ200 fmol of toxin binding sites was retrieved and loaded onto SDS-polyacrylamide gels.

Surface Labeling with mAb 35 and Immunoprecipitation
To determine receptor surface expression, intact cells containing ϳ1 mg of total protein were incubated with 5 g of monoclonal anti-nAChR mAb 35 (11,23) in a final volume of 500 l of Ringer's solution for 90 min on ice. Subsequently, unbound antibody was removed by washing as above, cells were lysed with 450 l of RIPA solution as above, and receptors were precipitated overnight using 50 l of Protein G-agarose. The precipitated samples were washed three times with 500 l of RIPA solution (4,000 ϫ g, 2 min, 4°C). The conditions chosen are saturating, as neither the addition of larger amounts of Protein G-agarose or mAb 35 nor the prolonged exposure with mAb 35 lead to an increased precipitation of ␣-subunit.

SDS-Polyacrylamide Gel Electrophoresis and Western Blot Analysis
Biotinylated receptor-streptavidin-agarose bead complexes (or Protein G-receptor complexes in the case of mAb 35-treated samples) were brought to a final concentration of 4% SDS, 0.002% bromphenol blue, 0.12 M Tris-HCl, pH 6.8, and 10% glycerol. Prior to loading, DTT and ␤-mercaptoethanol were added to a final concentration of 500 mM each. Samples were heated to 95°C for 3 min before being loaded onto a 10% SDS-polyacrylamide gel (24). The gels were transferred onto polyvinylidene difluoride membranes and blocked with phosphate-buffered saline containing 0.1% Tween and 3% bovine serum albumin (Sigma). Primary antibody incubations were performed in the same buffer using a 1:500 dilution of antibody 43.37 (120 g/ml stock) (25). Blots were then incubated with anti-mouse IgG horseradish peroxidase-conjugated secondary antibody (working dilution 1:2500; Transduction Laboratories). Proteins were visualized using the enhanced chemiluminescence detection method (ECL Plus, Amersham Pharmacia Biotech).

Calculation of the Rates of Receptor Biotinylations
Visualized receptor bands were quantitated by densitometry using the software ImageJ (National Institutes of Health, Bethesda, MD). A calibration curve relating band intensities to femtomoles of ACh binding sites was established for mutant and wild-type receptors by running known concentrations of Torpedo membranes on a gel, followed by Western blotting, ECL exposure, and densitometry quantitation. These defined amounts of Torpedo membranes were run alongside with the biotinylated receptors and subjected to identical experimental conditions. Using these calibration curves, intensities of biotinylated receptor bands were then converted from pixel values to femtomoles of ACh binding sites for each time point. Rate constants for the biotinylation of AChR were then determined using a second-order rate equation (26). In cases where ␣-Bgtx or Ea led to a substantial inhibition (Ͼ10-fold), second-order reaction rates were estimated by comparing the maximal amounts of biotinylation in the absence or presence of the toxins for three PEO-biotin concentrations (5, 50, and 500 M).

RESULTS
In previous studies, we demonstrated that a cysteine can be substituted for all of the individual amino acids between positions 184 and 197 of the mouse muscle-type ␣-subunit without dramatic effect on receptor functionality (Ͻ6-fold changes in the EC 50 ) as measured by ACh responsiveness in Xenopus oocytes (17). Moreover, we showed that within this region, residues Trp 187 , Val 188 , Phe 189 , Tyr 190 , and Pro 194 are solventaccessible and are in a position to contribute to the ␣-Bgtx binding site. In order to explore further the topology of the bracketing region extending from 184 to 197, we have now modified the appropriate Cys-substituted mutants with thiolspecific biotin derivatives following their expression in HEK-293 cells, permitting a more detailed analysis of the accessibility of these Cys-substituted residues.
PEO-and MTSEDE-biotin Specifically Modify Cysteines 192/193 in Reduced Wild-type nAChR-Initially, we wanted to determine whether alkyl methane thiosulfonate derivatives of biotin could be used to specifically modify surface-exposed cysteines in the nAChR. Previously, we and others showed that smaller alkyl methane thiosulfonate derivatives and bromoacetylcholine, an alkylammonium compound containing an ␣-haloacyl ester moiety, react covalently and specifically with Cys 192 and Cys 193 of the nAChR following their selective reduction with 1 mM DTT (16,17,(27)(28)(29). As shown in Fig. 2, when cells expressing the wild-type nAChR were selectively reduced at positions 192 and 193 with 1 mM DTT and then treated with 500 M of MTSEDE-biotin ( Fig. 2A, lane 4), we detected a band corresponding to the ␣-subunit of the nAChR.
As these studies were under way, a similar biotin derivative (PEO-biotin) became commercially available. When we applied this reagent to reduced wild-type receptors, we obtained results comparable to those for MTSEDE-biotin (Fig. 2B, lane 4). Importantly, this band was not present when unreduced nAChR was exposed to MTSEDE-biotin ( Fig. 2A, lane 3) or PEO-biotin (Fig. 2B, lane 3).
Biotinylation was blocked by preincubation with either 1.5 mM BrACh or 1.5 mM MTSET for both MTSEDE-and PEO-biotin, confirming its specificity (Fig. 3B, lane 3; Fig. 5A). In contrast, exposure of cells to the more hydrophobic MTSEAbiotin leads to considerable nonspecific labeling, as we detected ␣-subunit labeling that was equally pronounced both before and after reduction with DTT. These results suggest that this reagent is capable of penetrating the lipid bilayer to a large degree and may have access to the internal cysteines at position ␣Cys 222 (in presumed transmembrane segment M1) and ␣Cys 418 (M4) (compare Fig. 2A, lanes 6 and 7). In addition, we observed that MTSEA-mediated biotinylation of reduced wildtype and mutant receptors cannot be blocked by 1.5 mM BrACh (data not shown), further pointing to its reactivity with an internal cysteine. For PEO-and MTSEDE-biotin, we detected nonspecific labeling of native unreduced receptors only when concentrations were raised to 1.5 mM and above (data not shown). The weak signal we obtained in the presence of DTT alone may suggest nonspecific absorption of the receptor to the streptavidin beads that are used for precipitation, although the result presented here was somewhat exceptional, and we generally did not observe a signal in the presence of DTT alone ( Fig. 2A, lane 2). Additionally, labeling of surface-expressed ␣-subunit by incubation of intact cells with the mouse monoclonal nAChR antibody, mAb 35 (11) (directed against region ␣67-76), followed by immunoprecipitation enabled us to the detect an ␣-band of an intensity comparable to that for the biotinylated wild-type receptor, indicating that at least a large fraction of the surface population of wild-type nAChRs can be biotinylated ( Fig. 2A, lane 5). However, it should be noted that a direct comparison of the signal intensities is not 100% accurate. Although we have optimized conditions such that maximal amounts of both mAb 35-labeled and biotinylated receptor are precipitated, it remains possible that the recovery of receptor eluted from the beads is different for the two methods. Nevertheless, we could use mAb 35 to test surface expression of all the cysteine mutants from position ␣184C to ␣197C, and we detected no noticeable difference in cell-surface expression levels among these mutants (Fig. 3A).
Rate of Biotinylation of Reduced Wild-type nAChR and Retardation in the Presence of ␣-Bgtx-In an effort to quantitate the reactivity of reduced wild-type and mutant nAChRs with PEO-biotin, we exposed nAChRs to 5 M PEO-biotin for various incubation times (Fig. 4A). Second-order reaction rates were  (Table I). Comparable results were obtained with MT-SEDE-biotin (data not shown). However, since the MTS compound was more light-sensitive and degraded rapidly at room temperature, we focused the remainder of our studies exclusively on the effects mediated by PEO-biotin.
When 10 M ␣-Bgtx was added to cells and allowed to equilibrate for 2 h, the reactivity of reduced wild-type receptors toward PEO-biotin was substantially slowed (Ͼ100-fold). Using 5 and 50 M PEO-biotin, we could not detect any biotinylation of the ␣-subunit, and even after incubation with 500 M PEO-biotin, we obtained only a weak signal (ϳ10% of the maximum intensity; Fig. 4B). A precise determination of the second-order rate constant in the presence of ␣-Bgtx was not possible, since PEO concentrations of 1.5 mM and above would have to be used, and under these conditions, there was considerable background stemming from the presumed modification of an internal cysteine or other unidentified sites.
Rate of the Biotinylation of Mutant ␣V188C nAChR Is Unaffected by the Presence of ␣-Bgtx-Previous results showed that mutant ␣V188K leads to a ϳ680-fold decrease in affinity for a short ␣-neurotoxin, NmmI, although the effect of this mutation was much less pronounced in affecting ␣-Bgtx binding (15,30,31). In addition, modification of ␣V188C with 1.5 mM BrACh or MTSET leads to a ϳ50% reduction in the number of ␣-Bgtx binding sites (16). On the other hand, a mutation introducing a negative charge at this position (␣V188D) merely produced a ϳ10-fold decrease in NmmI binding. To more closely examine the role of this position in ␣-Bgtx binding, we determined second-order reaction rates for ␣V188C with 50 M PEO-biotin in the presence (Fig. 5B) and absence of ␣-Bgtx (Fig. 5A). In the absence of ␣-Bgtx, we obtained a rate of 102 M Ϫ1 s Ϫ1 , whereas incubation in the presence of ␣-Bgtx resulted in only a slightly diminished rate of 77 M Ϫ1 s Ϫ1 . Therefore, we found no indication that ␣-Bgtx binding to the ␣-subunit protected this Cyssubstituted residue from biotinylation. Interestingly, a comparison between Figs. 4 and 5 shows that the signal intensity for TABLE I Second-order reaction rates of mutant ␣-subunits with PEO-biotin Wild-type or the respective mutant ␣-subunits encoding cDNAs were transfected together with wild-type ␤-, ␥-, and ␦-subunits into HEK-293 cells, which were harvested after 2 days of incubation, and treated with PEO-biotin. Streptavidin precipitation, electrophoresis, Western blotting and visualization of the ␣-subunit and calculation of second-order reaction rates were performed as described under "Materials and Methods." Second-order reaction rate constants in the absence (k) and presence of ␣-Bgtx (k Bgtx ) were determined using 50 M PEO-biotin for all of the mutants, except for mutant ␣W187C (500 M) and reduced wildtype receptors (5 M). Positions 192 and 193 form a vicinal disulfide bond in the wild-type receptor and are not listed separately, since they are indistinguishable and become modified in PEO-biotin-exposed reduced wild-type receptors. In cases where ␣-Bgtx led to a blockade of the second-order reaction rate with PEO-biotin, only estimates of the maximal rates of biotinylation in the presence of ␣-Bgtx could be calculated, due to nonspecific labeling at high concentrations of PEO-biotin. The reduction in the rate of reactivity was estimated by comparing the intensities obtained for 10-min incubations with 5, 50, and 500 M biotin in the presence and absence of the ␣-Bgtx.

FIG. 4. Reactivity of reduced wild-type receptors with PEObiotin and inhibition of biotinylation in the presence of ␣-Bgtx.
Wild-type ␣-subunit encoding cDNAs were transfected together with wild-type ␤-, ␥-, and ␦-subunits into HEK-293 cells, which were harvested after 2 days of incubation, and treated with PEO-biotin. the biotinylation of mutant ␣V188C is less than that for reduced wild-type receptors, despite the fact that their reaction rates are comparable (Table I). Most likely, two factors contribute to this observation. First, reduced wild-type receptors contain two modifiable free thiol groups (␣Cys 192/193 ), i.e. although the reaction rate is comparable to that for ␣V188C, the signal intensity will be twice as high for any given time point. Additionally, it is possible that, for ␣V188C, only a portion of the receptor population contributes to the signal, whereas the rest is either refractory to biotinylation or to the subsequent sterically constrained reaction with streptavidin beads. This view is strengthened by the fact that preincubation with ␣-Bgtx enhances the yield of biotinylated receptor approximately 2-fold (Fig. 6C).
Evidence That All Amino Acid Residues within Region ␣184 to ␣197 Are Solvent-exposed-Although positions ␣W187C, ␣V188C, ␣F189C, ␣Y190C, and ␣P194C were modifiable with BrACh and their modification resulted in a substantial blockade in ␣-Bgtx binding (16), the surface disposition of the other residues spanning region ␣184 to ␣197 is not well understood. Specifically, it was previously impossible to distinguish whether cysteine-substituted residues 184 -186, 191, and 195-197 were modified following the application of thiol-specific reagents. By systematically examining the reactivity of these residues with PEO-biotin, we now show that all of these residues are surface-exposed and accessible to biotinylation to a similar degree. Their second-order reaction rate constants fell within the range of ϳ10 to ϳ100 M Ϫ1 s Ϫ1 (␣W187C: 10.8 M Ϫ1 s Ϫ1 ; ␣V188C: 101.9 M Ϫ1 s Ϫ1 ; Table I).
Table I also lists the second-order reaction rate for positions ␣D71C and ␣W184C to ␣P197C in the presence of saturating concentrations of ␣-Bgtx. Importantly, the reaction rate for the biotinylation of ␣D71C is unaffected in the presence of ␣-Bgtx, confirming its location outside of the ␣-Bgtx binding site (11). Apart from reduced wild-type receptors, only residues ␣W187C (at least a 10-fold reduction) and F189C (at least a 100-fold reduction) exhibit a substantial reduction in the reaction rate in the presence of ␣-Bgtx. It should be noted that an accurate determination of reaction rates for these mutants in the presence of ␣-Bgtx was not possible. This would have required using concentrations of PEO-biotin in excess of 1.5 mM, a concentration that leads to the modification of other sites in addition to the engineered cysteine. Due to its reduced reactivity with PEO-biotin even in the absence of ␣-Bgtx, this is especially true for mutant ␣W187C. and ␣Phe 189 in the binding of ␣-neurotoxins. For example, using a double mutant cycle analysis of the short-chain ␣-neurotoxin NmmI with nAChRs, Ackermann et al. (31) found no indication for the involvement of ␣Trp 187 and ␣Phe 189 in toxin binding. On the other hand, our previous work suggests that all three positions (187-189) contribute to the binding of the longchain ␣-neurotoxin ␣-Bgtx. To characterize further the precise roles of these residues, we expanded our biotinylation studies to investigate the effect of a short ␣-neurotoxin. In this assay, Erabutoxin a (10 M) was added to HEK-293 cells transfected with ␣W187C, ␣V188C, and ␣F189C 2 h prior to PEO-biotin addition. Fig. 6 shows the profile of ␣-neurotoxin (␣-Bgtx or Ea) protection from biotinylation for reduced wild-type, ␣W187C, ␣V188C, and ␣F189C mutant receptors. Biotinylation of reduced wild-type receptors is blocked by approximately 90% with Ea and completely with ␣-Bgtx (Fig. 6A, lanes 2-4) when 5 M PEO-biotin is added for 10 min. In contrast, Ea does not protect positions ␣W187C and ␣F189C from biotinylation ( Fig.  6, B (lanes 2 and 4) and D (lanes 1 and 3), respectively). ␣-Bgtx, on the other hand, clearly blocks biotinylation at these positions (Fig. 6, B (lane 3) and D (lane 2)). Similarly, Ea completely abrogates biotinylation of mutant ␣V188C (Fig. 6C, lanes 2 and  4), whereas ␣-Bgtx has little effect (Fig. 6C, lane 3). It should be noted that different PEO-biotin concentrations were used to account for the different reaction rates of the various mutants. However, we repeated these experiments for reduced wild-type nAChRs with 50 and 500 M PEO-biotin and for ␣V188C and ␣F189C with 500 M and obtained comparable results (data not shown).

Effects of Erabutoxin a Co-incubation on the Reactivity of Residues ␣W184C to ␣P197C with PEO-biotin-Previous
For all other positions, the effects of Ea and ␣-Bgtx on PEO biotinylation of engineered cysteines seem to be comparable; none of the other positions shows a substantial blockade of biotinylation, as is summarized in Table II. Again, as stated above (and in the legend for Tables I and II), accurate determinations of second-order rate constants for the biotinylation in the presence of Ea were not possible for positions ␣W187C, ␣V188C, and reduced wild-type receptors. Furthermore, in those cases where Ea did not appear to protect against biotinylation, precise reaction rates were not calculated. Instead, the reduction of the rate was estimated by comparing the intensities obtained for 10-min incubations with 5, 50, and 500 M PEO-biotin in the presence and absence of Ea.

DISCUSSION
Our primary aim was to define in more detail the topology and solvent accessibility of residues 184 -197 in the ␣-subunit of the mouse muscle nicotinic acetylcholine receptor. To achieve this, we engineered individual cysteine substitutions and modified the introduced cysteine side chains with watersoluble, thiol-reactive derivatives of biotin. A similar approach has been used, for example, by Grunewald and co-workers (32) to determine the topology of the astroglial glutamate transporter GLT-1. Likewise, the solvent-accessible structure of the ␥-aminobutyric acid A receptor has been investigated with a combined cysteine mutagenesis/biotinylation procedure (33).
Thiol-reactive Biotin Derivatives That Are Hydrophilic Are More Selective in Modifying External Cysteines-The success and accuracy of cysteine modifications with thiol-reactive biotin derivatives strongly depends on the hydrophilicity of the reagent. As can be seen in Fig. 1 (A and B), the relatively hydrophobic modifier MTSEA-biotin reacts strongly with wildtype nAChR. In these receptors, there are intrinsic cysteine disulfide pairs at positions 128 and 142 as well as at positions 192 and 193, but no free thiols in the extracellular domain, and thus there should be no reactivity with thiol reagents. The observed labeling is most likely explained by MTSEA-biotin entering the lipid bilayer and reacting with the solvent inaccessible unpaired ␣Cys 222 and/or ␣Cys 418 . This is further supported by the fact that preincubation with 1.5 mM BrACh does not protect against biotinylation of reduced wild-type and mutant receptors when MTSEA-biotin is used (data not shown). In sharp contrast, the more hydrophilic, water-soluble derivatives, MTSEDE-biotin and PEO-biotin, do not react with wildtype nAChR unless the disulfide bond at positions 192/193 is selectively reduced with 1 mM DTT (27,28), indicating that these reagents are not membrane-permeable. With the hydrophilic reagents, biotinylation was blocked completely using either 1.5 mM BrACh or MTSET (see Fig. 3B for MTSEDE-biotin and Fig. 5A for PEO-biotin). As BrACh and MTSET are charged derivatives, their block of biotinylation confirms the specificity of the reaction and supports the conclusion that biotinylation with MTSEDE-biotin or PEO-biotin is restricted to the solvent-accessible surface of the nAChR in our studies using intact cells. Either biotin conjugate would have been useful in the present study, although we decided to concentrate on PEO-biotin, as MTSEDE-biotin was more light-sensitive and degraded more rapidly at room temperature. Nonetheless, MTSEDE-biotin would offer advantages in cases where the introduction of a reversible disulfide bond is desired.
All Amino Acid Residues between Positions ␣184 and ␣197 Are Solvent-exposed- Table I summarizes the second-order reaction rates for the individual Cys substitutions of amino acid residues between positions ␣184 and ␣197. Their second-order reaction rate constants fall within the range of ϳ10 to ϳ100 M Ϫ1 s Ϫ1 (␣W187C: 10.8 M Ϫ1 s Ϫ1 ; ␣V188C: 101.9 M Ϫ1 s Ϫ1 ). As even the least reactive of these positions, ␣W187C, clearly contributes to the ␣-Bgtx binding site and is modifiable by the charged reagent BrACh (16), we conclude that second-order reaction rates for modification with PEO-biotin that are as low as 10 M Ϫ1 s Ϫ1 are consistent with solvent accessibility. We further conclude, therefore, that all of the residues within the region tested are surface-exposed. Minor perturbations in reactivity occur, in general, over a 10-fold range and are likely to be due to variations in the chemical and steric environment surrounding the individual positions (34). Additionally, favorable electrostatic and steric interactions may enhance biotinylation. As an additional reference point, we included an analysis of mutant ␣D71C. It is well established that Asp 71 forms one of the main determinants in the epitope recognized by ␣-Bgtx and erabutoxin a Wild-type or the respective mutant ␣-subunits encoding cDNAs were transfected together with wild-type ␤-, ␥-, and ␦-subunits into HEK-293 cells, which were harvested after 2 days of incubation, and treated with PEO-biotin. Only region ␣187 to ␣193 is shown, since there was no difference in the observed effects for ␣-Bgtx and Ea for the other mutants examined. Results for positions ␣192 and ␣193 were obtained with reduced wild-type receptor, which contains two free thiols at these positions. The symbol "ϩ" in the table denotes a greater than 10-fold reduction in the second-order reaction rate constant k in the presence of either ␣-Bgtx or Ea. The symbol "Ϫ" is used to represent a change of less than 2-fold in the reaction rate (for ␣-Bgtx), or a Ͻ50% reduction in the maximum amount of labeling in the presence of Ea. For reasons described in the text and Table I, rates for the biotinylation of mutants ␣W187C, ␣F189C,and ␣Cys 192/193 in the presence of ␣-Bgtx are estimates of the maximum rate and could not be determined accurately from the data. This is also the case for the biotinylation of positions ␣V188C and ␣Cys 192/193 in the presence of Ea.
antibodies directed against the main immunogenic region (11) and therefore should be surface-accessible. Indeed, the secondorder reaction rate (39.5 M-1 s Ϫ1 ) is also in the range obtained with the 184 -197 series of Cys substitutions. Similarly, a comparable reaction rate constant (ϳ200 M Ϫ1 s Ϫ1 ) has been reported for the reaction of an N-ethylmaleimide derivative with ␤-mercaptoethanol in phosphate buffer (35). Further, the second-order reaction rate with PEO-biotin seems to be identical for the two ␣-subunits of the nAChR, regardless of position of the introduced cysteine mutation. Using a linear regression fitting routine, we calculated a curve with a very good fit to the data (R Ͼ 0.98 for all the mutants investigated; data not shown) assuming a single class of binding sites.
Our conclusions on the solvent accessibility of the 184 -197 region are in line with a number of recently published studies. Structural predictions derived from the NMR studies of a receptor-peptide fragment bound to ␣-Bgtx (36) suggest that residues 187-190 are likely to be surface-accessible in the native receptor. The conclusions concerning surface accessibility of ␣-subunit residues 187-190 are also consistent with studies of the Bgtx-resistant nAChRs found in cobra and mongoose muscle and of HEK-expressed mouse muscle nAChRs containing glycosylation signals found in the cobra and mongoose AChR (23,37). The surface accessibility of ␣Val 188 was demonstrated by McLaughlin et al. (17) and is also supported by recent studies of Ackermann et al. (30,31). A recent double mutant cycle analysis concludes that position ␣Pro 197 interacts with NmmI residues Arg 33 and Lys 27 and must therefore be surfaceexposed. In addition, our study demonstrates for the first time that positions corresponding to ␣Trp 184 , ␣Lys 185 , ␣Ser 191 , ␣Thr 195 , and ␣Thr 196 are on the solvent-accessible surface of the receptor.
␣-Bgtx Protects Mutants ␣W187C, ␣F189C, and Reduced Wild-type Receptors, but Not Mutant ␣V188C, from Biotinylation with PEO-biotin-In the present study, we provide evidence that positions ␣71, ␣184, ␣185, ␣191, ␣195, ␣196, and ␣197 do not form part of a stable ␣-Bgtx binding site, as the presence of bound ␣-Bgtx does not have a significant effect on the second-order reaction rate for PEO-biotin (Table I). One reservation concerning the interpretation of the Cys substitution studies is that the substitution itself may locally distort the structure. Thus, negative results of the protection experiments may not be as conclusive as positive results. Furthermore, it is possible, depending on the local geometry, that an introduced cysteine is both in some proximity to bound Bgtx and in a position where Bgtx does not occlude the site from biotinylation.
Interestingly, the biotinylation of mutants ␣V188C and ␣Y190C is not affected by the presence of ␣-Bgtx, even though the tethering of a methylammonium moiety to these residues leads to a blockade of ␣-Bgtx binding (16). These results allow us to refine our understanding of the interaction between ␣-Bgtx and positions 188 and 190, as they suggest that these positions, at least when substituted with cysteine side chains, do not interact directly and stably with ␣-Bgtx ( Fig. 3 and Table I). Nevertheless, there is good evidence that these positions are within 8 Å of the toxin binding site. The introduction of an alkylammonium moiety, through the action of either BrACh or MTSET, leads to a significant perturbation of the receptor-toxin interaction (16). With both BrACh and MTSET, a covalently attached adduct of cysteine is formed that would fit into a cylinder 8 Å long and 6 Å in diameter (38). Furthermore, the results presented here are compatible with NMR structural studies of the complex formed between the dodecapeptide ␣185-196 and ␣-Bgtx (36). The NMR analysis re-vealed intermolecular nuclear Overhauser effect cross-peaks between one ␥-methyl group of ␣Val 188 and the two ␥-methyl groups of Bgtx residue Val 39 , thus placing these latter methyl protons at a distance of ϳ4 -5 Å from that of ␣Val 188 . Although this distance constraint certainly places Val 188 in the proximity of Bgtx residue Val 39 in this complex, there was no further indication of a more intimate or extensive contact between these hydrophobic side chains. In contrast, ␣-Bgtx clearly blocks the biotinylation of residues ␣W187C and ␣F189C, and thereby provides independent support that these residues are critical for a stable receptor-toxin interaction (16,37,39). Furthermore, the biotinylation of positions 192 and 193 in reduced wild-type receptors is also inhibited by ␣-Bgtx and underlines the importance of these residues for ␣-Bgtx binding. Nevertheless, the modification of these cysteines and of ␣W187C and ␣F189C with non-methylammonium-containing modifiers does not affect ␣-Bgtx binding significantly (16), suggesting additional complexity in Bgtx binding. In addition, it should be emphasized that the saturation binding assay used in these modification studies would not have detected decreases of ␣-Bgtx binding affinity Ͻ50-fold (16). Theoretically, it is also possible that the presence of ␣-Bgtx leads to the complete abrogation of biotinylation on only one of the two ␣-subunits, whereas the other site remains unaffected. If this were the case, we would expect to see a 2-fold reduction of the maximum intensity of the biotinylation signal. Our results argue against such a scenario, since the maximum intensity of the signal is largely unaltered for all of the investigated mutants.
It is somewhat surprising that the "footprint" of ␣-Bgtx protection from biotinylation is not larger. Previous studies have suggested that the Bgtx-receptor contact site would involve multiple points and a large portion of the toxin surface (40,41). Our results suggest that the number of strong contacts may be fewer than expected. In addition, any additional contacts may be flexible enough to allow sufficient structural fluctuation to permit access to the reactive derivatives over the time course of the reaction incubation. Alternatively, some of the residues in this region may interact with Bgtx and the biotinylation reagents via non-overlapping surfaces.
Short and Long Chain Neurotoxins Establish Distinct Contact Points with the nAChR-Ackermann et al. (31) have argued that the short neurotoxin NmmI interacts with ␣Val 188 , whereas their results do not implicate positions ␣Trp 187 and ␣Phe 189 in NmmI binding. The results presented here (Fig. 6, Table II) seem to reconcile the NmmI studies with our previous results, suggesting an important role of ␣Phe 189 in ␣Bgtx binding (16,39). Second-order reaction rates for the biotinylation of positions ␣W187C and ␣F189C are inhibited at least 10-fold by bound ␣-Bgtx, whereas ␣V188C is not affected. In contrast, Ea, a short neurotoxin very similar to NmmI, leads to a Ͼ10-fold protection of position ␣V188C, but does not alter biotinylation at positions ␣W187C and ␣F189C. This suggests that ␣-Bgtx and Ea interact with distinct amino acids on the nAChR.
A Model for the Bgtx-mediated Blockade of Receptor Biotinylation-The model we propose here for the mechanism of ␣-Bgtx-mediated blockade of biotinylation is based on the mode of interaction between fasciculin and acetylcholine esterase (42)(43)(44)(45)(46)(47). In this model, ␣-Bgtx, which carries a large net positive charge (ϩ4; see Refs. 48 and 49) could bind to a site near a gorge leading to the agonist binding site and could obstruct PEO-biotin access to substituted cysteines (or ACh access to its binding site), which, in our experiments, would be reflected by a decrease in the second-order reaction rate of biotinylation. It is unlikely, however, that ␣-Bgtx blocks the putative gorge entirely, as only some of the residues implicated in ACh binding can be protected from biotinylation by concomitant ␣-Bgtx incubation (␣W187C, ␣F189C, and ␣Cys 192/193 ), whereas other residues that are thought to be crucial for the receptor-ACh interaction (e.g. ␣Y190C; Ref. 16), are not affected. Rather, our results argue that the gorge would be only partially blocked by ␣-Bgtx, and that the remaining cavity is large enough to accommodate the entry of PEO-biotin (29 Å in fully extended length and 5.6 Å in width at the biotin ring). Finally, it is interesting that, between positions ␣186 and ␣190, biotinylation of only every other residue is inhibited by ␣-Bgtx binding. Therefore, our results are consistent with NMR studies arguing that residues ␣186 -190 are in an extended ␤-sheet orientation (36). In addition, these results suggest that ␣-Bgtx may establish contact with one of the two faces of the ␤-sheet, whereas the other face remains accessible to PEO-biotin.