TABLE OF CONTENTS

Table of content 1

SUMMARY 3

List of abbreviation 4

INTRODUCTION 5

EXPERIMENTAL PROCEDURES

Peptide synthesis and complex formation 8

Dynamic light scattering measurements 9

NMR sample preparation 9

NMR measurements 9

Experimental constraints 11

Structure determination 11

RESULTS

Optimization of sample conditions 12

Mapping the N-terminus of the α-BTX-binding determinant 12

Mapping the C-terminus of the α-BTX-binding determinant 13

Sequential assignments of the aBTX/ap25 complex resonances 15

Secondary structure of the αp25/αBTX complex 15

Peptide flexibility in the aBTX/ap25 complex 18

Structure determination 19

Structure of the bound aBTX 19

Structure of the bound aAChR peptide 20

Binding interactions 21

DISCUSSION

Mapping of the aAChR determinant 22

Residue flexibility 23

Structure of the aAChR/ap25 complex 23

Conformation and structure of the bound ap25 peptide 24

Changes in aBTX conformation of upon peptide binding 24

Peptide-toxin interaction in the aBTX/ap25 complex 25

The importance of the vicinal disulfide bridge ^RC192-^RC193 25

Comparison with site directed mutagenesis 25

Species variation of the aAChR sequence 26

Agonist binding domain of the aAChR 28

Comparison to other a-neurotoxins 28

aBTX dimerization 29

Kinetical study and complex formation 30

Biomedical application 30

CONCLUSIONS 30

REFERENCES 32

List of publication resulting from this study 39

Instructions concerning the joined disk 40

SUMMARY

The a-subunit of the nicotinic acetylcholine receptor (αAChR) contains a binding site for the snake venom α-bungarotoxin (a-BTX). Previous studies have established that a segment comprising residues 173-204 of aAChR contains the major determinant interacting with the toxin, but the precise boundaries of this determinant have not been defined clearly to date. In this study, we applied NMR dynamic filtering to determine the exact sequence constituting the major aAChR determinant interacting with α-BTX. Based on previous studies, two overlapping synthetic peptides corresponding to segments 179-200 and 182-202 of αAChR were prepared and complexed with α-BTX. HOHAHA and ROESY spectra of these complexes, sensitive to proton T_1ρ relaxation time, were acquired with long mixing times. These spectra discriminate between immobile and flexible protons, and highlight the residues of the peptide that do not interact with the toxin and retain considerable mobility upon binding to α-BTX. Moreover, a novel HOHAHA experiment enabled the evaluation of T₁_rfor each residue of the aAChR peptide/aBTX complex, allowing a quantitative differentiation between the residues according to their mobility. These results, together with the dynamic filtering data, suggest the contact region is formed by residues 184-200.

The three-dimensional structure of the aAChR-peptide corresponding to residues 182-202 bound to α-BTX was determined using 2D ¹H-NMR techniques. The bound peptide adopts a β-hairpin conformation in which the segments ^RH186-^RV188 and ^RY198-^RD200 form the β-strands. The two strands are connected by a loop consisting of the ^RC192-^RC193-^RP194-Xxx-Xxx-^RP197 rigid motif that contains a disulfide bond conserved in most species. The tight binding of the peptide to the toxin is mediated by residues ^RV188-^RT191, which form an intermolecular b-sheet with the second finger of aBTX. The high affinity is also attributed to the hydrophobic interactions of residues ^RW184 and ^RW187-^RC192 with the toxin. Moreover, the NMR data indicates that peptide binding is accompanied by an increase of the b-sheet content of aBTX. Our structure correlates the observed changes in toxin binding with the naturally occuring mutations of aAChR in different species. It also provides a structural explanation for the results of previous site-directed mutagenesis studies. This study therefore enhances our understanding of the a-neurotoxins binding site upon aAChR, and sheds light on the structure of the major AChR determinant involved in both acetylcholine and toxin binding.

List of abbreviations

AChR, nicotinic acetylcholine receptor; αAChR, α subunit of AChR; α-BTX, α-bungarotoxin; NMR, nuclear magnetic resonance; 2D-NMR, two-dimensional NMR; HOHAHA, homonuclear Hartmann-Hahn spectroscopy; NOE, nuclear Overhauser effect; NOESY, nuclear Overhauser spectroscopy; ROESY, rotating-frame Overhauser spectroscopy; DQF-COSY, double-quantum filtered correlation spectroscopy; T_1ρ, spin-lattice relaxation time in the rotating frame; T₂, transverse relaxation time constant; CD, circular dichroism; DLS, dynamic light scattering; FPLC, fast protein liquid chromatography; RP-HPLC, reverse-phase high-pressure liquid chromatography; α-BTX and αAChR-peptide residues are designated by a superscript B (^BX) and R (^RX) respectively before the one-letter amino-acid code and position in sequence.

INTRODUCTION

The nicotinic acetylcholine receptor (AChR) is the prototype of the superfamily of ion channel-coupled receptors, which are gated by specific neurotransmitters. In addition to AChRs, this superfamily includes the g-aminobutyric acid receptor (GABA_AR and GABA_CR types), the glycine receptor (GlyR), and the 5-hydroxytryptamine receptor (5-HTR)[1]. The nicotinic AChR is a ligand-gated cation channel that is activated upon binding of acetylcholine. It is a 290 kDa membranal glycoprotein found in muscle and neuronal tissues consisting of five homologous subunits in the stoichiometry of a₂bdg or a₂bde (Fig. 1) [2-5]. The five subunits are arranged in the clockwise order agabd to create a cylindrical complex around the ion channel. Each subunit is composed of extra- and intracellular domains as well as transmembranal loops. Located on the postsynaptic surface of the neuromuscular junction, the AChR translates the chemical signal of acetylcholine binding into an electrical one, leading to muscle contraction. The acetylcholine binding site is formed by the ad and ag subunits [6]. The a-subunit also contains a high affinity binding site for a-neurotoxin antagonists [7-9], as well as the major immunogenic region (MIR) which is the main target of autoimmune antibodies in myasthenia gravis [10-12].

a-bungarotoxin (a-BTX) is a 74 amino-acid, 8 kDa a-neurotoxin derived from the venom of the snake Bungarus Multicinctus. It binds to the postsynaptic muscle-AChR with a K_d of 10^-11M [13], competitively inhibiting acetylcholine binding, thereby preventing the depolarizing action on postsynaptic membranes and blocking neuromuscular transmission [14]. Since no single-site mutation of aBTX appears to eliminate receptor binding, it has been suggested that receptor binding involves a multipoint attachment [15]. Although a number of amino acid residues have been implicated in receptor recognition, there has been little direct evidence to reveal the molecular mechanism underlying receptor binding and specificity. Electron image analysis of aBTX bound to AChR enriched membranes suggest that aBTX lies atop the synaptic crest of the receptor at positions corresponding to the two a-subunits and may protrude some distance into the central vestibule [16-18]. Photo-labeling experiments indicate that the a-BTX binding site is restricted mostly to the a-subunit [1, 19], and partially overlaps that of acetylcholine [20-22]. αAChR isolated from the Torpedo Californica fish is able to bind α-BTX even after denaturation, indicating that a linear segment of this subunit can form a binding site for α-BTX [19]. A major domain of the toxin-binding site was mapped to residues 173-204 of aAChR [23, 24]. A 32 residue peptide corresponding to this segment binds a-BTX with a K_D of 1.4 x 10^-7M, which is comparable to the affinity of the intact denatured aAChR to the toxin (K_D = 1.2 x 10^-8) [24, 25]. Several studies have focused upon the a-BTX-affinity to aAChR synthetic peptide analogs in an attempt to locate the a-BTX-binding domain on the receptor more precisely. These studies propose various putative binding domains within the aforementioned 32 amino acid peptide such as residues 185-199 [26], 181-200 [27], 179-196 [24, 25], 184-200 [28], 187-200 [1] and 185-196 [29]. Concurrently, site-directed mutagenesis and affinity labeling studies have pinpointed residues within this domain critical to acetylcholine and antagonist binding [30, 31]. Residues contributing to aBTX binding were determined as ^RW184 [32-34], ^RK185 [32], ^RH186 [30], ^RW187 [29, 32, 33, 35, 36], ^RY189 [30, 32, 33, 35-37], ^RY190 [30, 32, 37], ^RT191 [33], ^RC192 [26, 30, 32, 37, 38], ^RC193 [26, 30, 32, 38], ^RP194 [30, 32, 35], ^RD195 [30, 32], ^RT196 [32], ^RY198 [32] and ^RD200 [28]. In summary, it seems that the entire segment 184-200 is important for optimal binding to aBTX.

A complex of an aAChR peptide analog with a-BTX has also been the subject of structural studies. The solution structure of a-BTX in complex with a dodecapeptide corresponding to residues 185-196 (KHWVYYTCCPDT) of Torpedo aAChR was solved by Basus and co-workers using NMR spectroscopy [39]. They established interactions between a-BTX and the segment corresponding to residues 186-190 of the aAChR, yet no interactions were detected for the C-terminal half (191-196) of the peptide. Thus, the short peptide studied by Basus and co-workers represents only a fraction of the a-BTX binding domain in aAChR.

Anglister, Katchalski-Katzir and coworkers [40] recently determined the three-dimensional solution structure of a-BTX in complex with a 13-residue peptide (MRYYESSLKSYPD) selected from a phage-displayed peptide library. This complex is of special interest, since this tridecapeptide exhibits a 15-fold higher a-BTX-affinity in comparison to the dodecapeptide aAChR peptide used by Basus et al. While the peptide corresponding to residues 185-196 of aAChR adopts an extended conformation when bound to the toxin [39], the toxin-bound library peptide is nearly globular and occupies a larger surface area of the a-BTX binding site. In view of the larger number of interactions and the 15-fold higher binding constant for the library peptide, the globular conformation of this peptide seems to mimic a larger aAChR determinant, and provides a more detailed picture of the a-BTX binding-site for AChR.

Despite these recent advances in the structural understanding of aAChR-peptides in complex with a-BTX, the precise boundaries of the binding domain on the receptor are uncertain to date. A systematic serial truncation of the 32-residue peptide has not been performed, leaving the exact boundaries of the binding domain undetermined. The inhibition of a-BTX binding to AChR by the 12 residue (residues 185-196, IC₅₀=1.3 X 10^-5 M [39]) and the 18 residue peptides (residues 181-198, IC₅₀=9.3 X 10^-6 M [41]) used by Basus is lower than that by the 32 residue peptide (residues 173-204, IC₅₀=1.4 X 10^-7 M), indicating that a significant contact area is contributed by residues outside the shorter segments. Similarly, the library tridecapeptide displays lower a-BTX-affinity than the 32-residue peptide. These findings underline the need for structural studies of a-BTX complexes with longer peptide analogs of aAChR.

Several studies have attempted to predict the secondary structure of the aAChR [42-44]. However, the different models of the subunit disagree on the presence of secondary structure in the α-BTX binding domain. A complete high-resolution structure of AChR has not been solved yet due to the difficulties in crystallizing this membrane protein, and a cryoelectron microscopy derived structure at 9 Å resolution [45] is unsuitable for molecular analyses of the aAChR ligand binding domain. Nevertheless, tritium hydrogen exchange kinetics of the AChR have been analyzed, and a retardation in the exchange rate was observed in presence of α-BTX [46]. It was suggested that α-BTX shields the AChR by forming an intermolecular β-sheet, thereby decreasing solvent accessibility. Circular dichroism (CD) measurements of a peptide corresponding to residues 185-196 of the αAChR indicated an increase of β-structure upon α-BTX binding [47]. Structure determination of peptides corresponding to the ligand binding domain of aAChR when bound to aBTX will shed light at the molecular level on this important binding determinant.

In recent years, NMR spectroscopy has become a powerful tool for studying the three-dimensional structure of proteins in solution. This was made possible by the introduction of 2D-NMR techniques as well as the advent of higher spectrometer fields. The first step towards structure determination by NMR is the sequential assignment, a procedure by which each resonance is assigned to its corresponding proton. This is accomplished using two types of magnetization transfer in 2D-NMR experiments. The first type is homonuclear Hartman Hahn (HOHAHA) experiments (Bax & Davies, 1985) that allow through bond magnetization transfer, produce intra-residue cross-peaks, and are used to assign spin systems to amino acid types. The second type is nuclear Overhauser spectroscopy (NOESY) and rotating frame Overhauser spectroscopy (ROESY) experiments (Bothner-By, 1984) that allow through space magnetization transfer, produce intra- and inter-residue NOE cross-peaks, and are utilized for sequential assignment. Next, structural data are compiled, and include the interproton distances derived from the NOE cross-peak volume-integrals, the dihedral angles constraints calculated from coupling constants and hydrogen bond constraints obtained from solvent exchange data. Finally, distance geometry and simulated annealing calculations translate these data into cartesian coordinates, which after several energy minimization steps generate final three-dimensional structures [48].

Another facet of NMR is its ability to study dynamic processes governed by relaxation parameters of proteins. In this study we employ the ROESY experiment [49] and HOHAHA experiments, to discriminate between peptide protons that interact with aBTX and peptide protons that do not interact with the toxin. Upon binding to aBTX, peptide protons interacting with the toxin are immobilized and assume a T₁_ρ relaxation time comparable to that of the toxin protons. Peptide protons with no interaction with the toxin retain some mobility and have considerably longer T₁_r relaxation times. The mixing period in the HOHAHA and ROESY experiments is tuned to discriminate between the immobilized and flexible peptide protons. This method, known as dynamic filtering, enables us to accurately locate the a-BTX binding determinant upon the aAChR. Moreover, T₁_r-filtered HOHAHA experiments allows exact determination of the T₁_r value of each residue within the aBTX/ap25 complex. The T₁_r value is influenced by both the overall mobility of the molecule and the internal flexibility. In this study we use the aforementioned capabilities of NMR to map the αAChR determinant involved in the strong binding of α-BTX, determine the relative flexibility of all residues in the complex, and elucidate the three-dimensional structure of the αAChR-peptide bound to α-BTX.

EXPERIMENTAL PROCEDURES

Peptide synthesis and complex formation. Peptides αp22 and αp25 as shown in table 1, corresponding to segments of the aAChR, were synthesized on an AMS422 automated multiple peptide synthesizer (Gilson) at the WIS, and purified by reverse-phase HPLC.

Text Box: Table 1: Sequence and origin of peptides used in this study. The numbering system follows the Torpedo sequence.

Peptide Sequence Origin αAChR residues

αp22 KEARGWKHWVFYSCCPTTPYLD Mouse 179-200

αp25 EERGWKHWVYYTCCPDTPYLDITEE Torpedo 182-202

αp25 was elongated with two glutamic residues at each terminus to increase peptide solubility. Formation of the Cys192-Cys193 disulfide bond was ensured for all peptides, emulating their oxidation state in the native AChR [50], although this state has little effect on α-BTX binding [51]. In αp25 this disulfide bond was formed by air oxidation. The oxidation reaction was carried out in a dilute peptide solution (10 mg / 250 ml) to avoid oligomerization [52] and completion of the reaction was monitored by the Ellman reagent [53]. αp22 was not treated, but its mass spectrum suggested it had oxidized during purification. The oxidized peptides were lyophilized and purified by RP-HPLC with an acetonitrile gradient. The composition of the peptides was verified by amino acid analysis, and their molecular mass and oxidation state was confirmed by mass spectrometry. α-BTX was purchased from Sigma. The α-BTX/αp22 complex was prepared by mixing the peptide and α-BTX at a molar ratio of 0.75:1 respectively. The α-BTX/αp25 complex was prepared by addition of excess oxidized αp25 to α-BTX. To obtain a 1:1 complex of ap25 and aBTX, this complex was purified by gel-filtration FPLC on a Pharamcia S-75 gel filtration column, using a solution of 250 mM NH₄HCO₃ as a running buffer followed by lyophilization. The formation of the complexes was verified by polyacrylamide gel electrophoresis, transverse NMR relaxation time (T₂) measurements and dynamic light scattering. Complex formation was also monitored by the disappearance of the ^BH4(H^d) and the ^BW28(H^N) resonances of the free toxin at at 6.32 [39] and 8.6 ppm respectively.

Dynamic light scattering measurements. Dynamic light scattering measurements were performed using a Dynapro® Molecular Sizing instrument at 25 ºC. Concentration of all protein solutions was ~2 mg/ml. The molecular weight of α-BTX and the complexes was determined at several pH values and at various salt concentrations.

NMR sample preparation. The peptide, αp22 (2 mM), was dissolved in a solution of 90% H₂O, 10% D₂O and 0.05% NaN₃ and acidified with HCl to pH ~ 4. The peptide, αp25 (2 mM), was dissolved in a solution of 80% H₂O, 10% D₂O, 10% CD₃CO₂H. The α-BTX/peptide complex was dissolved at various pH and its aggregation was evaluated by measurement of the transverse relaxation time (T₂) of the amide protons using the 1,1-echo pulse sequence [54] . Only at pH 4 were the T₂values of the complexes approximately 35 msec,which is characteristic of the molecular mass of the complex (11.2 kDa). Therefore, all complexes were dissolved in 90% H₂O / 10% D₂O and 0.05% NaN₃ and acidified with HCl to pH 4. Final concentrations of α-BTX/αp22 and α-BTX/αp25 samples were 1.8 and 2 mM respectively. For experiments of α-BTX/αp25 in 99.99% D₂O, the sample was prepared by two cycles of redissolving the lyophilized complex in 99.8% D₂O, incubation at 42 ºC for 14 hours, lyophilization, and final redissolving in 99.99% D₂O. This sample was acidified to pH 4 using d₄-acetic acid.

NMR measurements. All NMR spectra were acquired on Bruker DMX-500 MHz and DRX-800 MHz spectrometers. The pulse sequence used for the 2D HOHAHA measurements combined a WALTZ [55] or DIPSI-2 [56] sequence for isotropic mixing, sensitivity enhancement [57], Echo/Antiecho-TPPI gradient selection [57] and a 3-19-9 pulse sequence with gradients for water suppression [58]. ROESY and NOESY measurements used States-TPPI for phase sensitivity and the WATERGATE (WATER suppression by GrAdient-Tailored Excitation) or 3-19-9 sequences for water suppression [58, 59]. The DQF-COSY spectrum was acquired by conventional procedures [60].

HOHAHA, ROESY and NOESY spectra of αp22 were acquired at 20 ºC using mixing times of 150, 150 and 300 msec respectively. HOHAHA spectra of αp25, were acquired at 30 ºC and 47 ºC. Sequential assignment of αp22 and αp25 was performed according to the well-known method of Wüthrich [48].

To map the binding determinant, HOHAHA and ROESY spectra of the α-BTX/ap22 and the α-BTX/ap25 complexes were acquired with increasing mixing times of 100-400 msec, with 2-4K points in the F₂ and 256-600 increments in the F₁ dimensions at 20 ºC and 30 ºC. Sequential assignment of the observed peptide spin-systems of the α-BTX/ap22 complex was performed using the HOHAHA and ROESY spectra with a mixing time of 400 msec. Sequential assignment of the observed peptide spin-systems of the α-BTX/ap25 complex was performed using the HOHAHA and ROESY spectra with a mixing time of 250 msec and 200 msec respectively.

To determine the T₁_rrelaxation time of all amide protons in the aBTX/ap25 complex, a T₁_r-filtered HOHAHA experiment was designed. In this experiment a spin-lock pulse was introduced to a common HOHAHA pulse sequence prior to the mixing time, correlating signal decay to the T₁_rrelaxation time. Six T₁_r-filtered HOHAHA spectra were acquired at 37ºC with increasing spin-lock duration from 0 to 150 msec. The amide proton intensity was determined using the AURELIA software package [61] and the T₁_rcalculated using an in-house script.

For the complete sequential assignment of α-BTX/αp25 complex, HOHAHA and NOESY spectra with 8K data points in F₂ and 800 increments in F₁ were acquired with mixing times of 70 msec and 150 ms respectively at 30 ºC and 37 ºC. Sequential assignment was performed using the previously determined chemical shifts of both toxin [40] and peptide. The relative intensities of the sequential NOEs as well as the cross-peaks volume were determined using the AURELIA software package [61]. For assignment of the aliphatic region of these spectra, HOHAHA and NOESY spectra with water presaturation were acquired for the 99.99% D₂O sample at 37 ºC with mixing times of 70 msec and 150 msec respectively.

To identify slowly exchanging amide protons, the α-BTX/αp25 complex was lyophilized from H₂O and redissolved in 99.8% D₂O at pH 4.1 (uncorrected for isotope effects). A series of six HOHAHA experiments at 37 ºC, each 2.5 hours long, was initiated 45 minutes later. Amide protons still giving rise to cross-peaks after 3 hours were considered to be in slow exchange with the solvent.

³J_HNHα-couplings of free αp22 and ap25 were measured from the HOHAHA spectrum acquired with a mixing time of 150 msec at 20 ºC with 4K points (zero-filled to 8K) in F₂ and 400 increments in F₁ dimension. ³J_HNHα-couplings of the αp25/α-BTX complex were measured from a DQF-COSY spectrum acquired at 37 ºC with 8K points (zero-filled to 16K) in F₂ and 1024 increments in F₁ dimension. The AURELIA software package [61] was used to fit the anti-phase doublets and obtain the J-couplings.

Experimental constraints. Structure determination by NMR is based on a large number of constraints on proton-proton distances that are obtained from the analysis of the NOESY spectrum [48]. The calibration curve for distance restraints in the amide and aromatic regions was based on the volume integrals of sequential H^a(i)/H^N(i+1) cross-peaks involved in b-sheet, corresponding to 2.2 Å. In the aliphatic region of the spectrum recorded in 99.99 % D₂O, volumes of the H^a/H^a interstrand cross-peaks in b-sheets, corresponding to a 2.3 Å distance, were used as a reference for NOE distances. For structure calculation, using the CNS program an input list of inter-proton distances was prepared, assigning each NOE cross-peak a lower bound distance of 1.8 Å and an upper bound of 130 % to account for internal motion and proton multiplicity [62]. Dihedral angle restraints were determined using the ³J_HNH_a couplings. The f angles of residues with a coupling constant smaller than 6 Hz were constrained to -65º ±20º, and the f angles of residues with a ³J_HNH_a larger than 8 Hz were constrained to –120º ±20º. ³J_HNH_avalues between 6 and 8 Hz were regarded as uninformative [62]. For each of the five disulfide bonds, ^BC3/^BC23, ^BC29/^BC33, ^BC48/^BC59, ^BC60/^BC65 and ^RC192/^RC193, d_S_g_-S_g was constrained between 2.01 and 2.03 Å. Two constraints were applied for each hydrogen bond; the H···O distance was constrained between 1.8 and 2.3 Å and the corresponding N···O distance between 2.5 and 3.3 Å.

Structure determination. Structure calculations were performed on a Silicon Graphics OCTANE workstation with version 1.0 of the CNS program [63]. Structures were generated with a hybrid distance geometry-dynamical simulated annealing method containing two slow cooling steps, followed by 10 steps of minimization. The structure calculations were accomplished in several steps. Initial input list included only H^N/H^aandH^a/H^adistance restraints, dihedral angle restraints and few unambiguous hydrogen-bond constraints. Structures generated from these constraints were used to complete assignment of ambiguous NOE peaks, deduce possible oxygen acceptors for slowly exchanging labile hydrogens. This cycle was repeated with gradual addition of aromatic constraints, methyl constraints and b-protons constraints. The structures were displayed for analysis using RasMol Molecular Renderer and InsightII (MSI Technologies, Inc.) programs.

RESULTS

Optimization of sample conditions. To determine the optimal pH of the α-BTX/peptide complexes for NMR experiments, a series of dynamic light scattering (DLS) measurements at increasing pH values were performed. The measurements indicated that the aBTX/ap25 complex dimerized at pH > 5, consistent with previous studies where aBTX was found to be a homodimer [64, 65]. Furthermore, salt concentrations up to 0.25 M failed to inhibit dimerization. These findings were supported by T₂ relaxation time measurements of the complexes. At pH 6, the complexes displayed T₂ values of 13-20 msec, indicating that aggregation had occurred. At pH 4, T₂values of 35-40 msec, typical of the molecular mass of the complex, were measured. Except of the dimerization, no major change in the overall structure of free α-BTX has been observed between pH 4 and 5.8 [39], and therefore all complexes were prepared at pH 4.

Mapping the N-terminus of the α-BTX-binding determinant. To map the N-terminus of the αAChR segment recognized by α-BTX, a set of HOHAHA spectra with mixing periods of 300, 350, 400, 500 and 600 msec were acquired for the α-BTX/αp22 complex. The HOHAHA and ROESY spectra measured with a mixing time of 400 msec retained the cross-peaks of the mobile part of the αAChR peptide with a good signal-to-noise ratio, while the contribution of the toxin cross-peaks was minimal (Fig. 2). Using these spectra, five residues (^RE180-^RW184) of αp22 could easily be assigned, as well as the cross-peaks arising from ^RK185(H^Nε) (not shown). Proton chemical shifts of residues ^RE180-^RG183 were practically identical to those observed for the free peptide, indicating that these residues are flexible and do not participate in α-BTX binding. The chemical shifts of ^RW184(H^α) and ^RK185(H^Nε) differ from those of the free peptide by 0.1 and 0.25 ppm respectively, and their HOHAHA cross-peaks were very weak, indicating that these residues are within the AChR determinant recognized by α-BTX. The cross-peaks of ^RH186-^RY198 were undetectable in the spectra. We therefore concluded that N-terminal residues ^RK179-^RG183 lie outside the determinant recognized by a-BTX, and that ^RW184 is at the boundary of this determinant. The C-terminal residues ^RL199 (not shown) and ^RD200 gave rise to weak cross-peaks in the HOHAHA spectra with 400 msec mixing time and their assignment was possible using the ROESY spectrum with shorter mixing time (200-300 msec), suggesting they are somewhat flexible. However, the change in their proton chemical shifts implied that they do interact with α-BTX. Determination of the C-terminal boundary of the binding determinant for α-BTX using this peptide was therefore inconclusive at this stage, because the ^RL199-^RD200 residues could be subject to terminal effects, and residues downstream of ^RD200 could participate in the binding. Cross-peaks arising from the mobile C-terminal residues of α-BTX were also detected in the long mixing time spectra. They were differentiated from peptide protons based upon the assignment of the library peptide/α-BTX complex and arise from residues ^BK70-^BG74 [40]. This assignment was later verified by the complete assignment of the aBTX/peptide complexes.

Mapping the C-terminus of the α-BTX-binding determinant. The above results suggested that a longer peptide was necessary to determine the C-terminus of the α-BTX binding determinant. A 29-residue peptide, RGWKHWVYYTCCPDTPYLDITYHFIMQRI, corresponding to residues 182-210 of the Torpedo aAChR, was insoluble in water, and no complex with a-BTX was obtainable. To increase the peptide solubility four C-terminal residues (MQRI) were omitted and two glutamic acid residues were added at each terminus. The resulting 29-residue peptide, EERGWKHWVYYTCCPDTPYLDITYHFIEE, was soluble in basic pH, but precipitated at the acidic pH required for monomeric complex formation. To increase the solubility at acidic pH, four additional C-terminal residues (YHFI) were omitted. The resulting peptide, αp25, was soluble at both acidic and basic pH, and a 1:1 complex with α-BTX could easily be obtained and purified by FPLC. A decrease in the retention volume was observed on a Pharmacia S-75 gel filtration column, α-BTX eluted at 14 ml while the complex eluted at 12.5 ml. DLS measurements of the molecular weight of α-BTX/αp25 concurred with the expected value of 11.2 kDa. Furthermore, in comparison to unbound aBTX, shorter T₂ relaxation times of 36 msec were measured for the complex, corresponding to molecular weights of 10-15 kDa. Several HOHAHA spectra were acquired with varying mixing times. The spectrum measured with a 250 msec mixing time showed a number of peptide cross-peaks with a high signal-to-noise-ratio while only a limited number of a-BTX cross-peaks were observed. The peptide cross-peaks corresponded to residues which retained a longer T_1ρ relaxation time upon complexation with α-BTX and were assigned to ^RE180-^RW184 and ^RI201-^RE204 using the ROESY spectrum measured with 200 msec mixing (Fig. 3). We therefore concluded that the determinant recognized by a-BTX comprised residues W¹⁸⁴KHWVYYTCCPDTPYLD²⁰⁰of the Torpedo αAChR

Sequential assignments of the aBTX/ap25 complex resonances. The resonances of all residues of the a-BTX/ap25 complex were successfully assigned using the common sequential assignment technique [48]. Resonances in the amide region were assigned utilizing NOESY and HOHAHA spectra recorded in H₂O with mixing times of 150 msec and 70 msec respectively. Assignment of the aliphatic region was performed in the corresponding spectra recorded in D₂O. The majority of the aBTX resonances did not change their chemical shift upon peptide binding [66], however some loop residues or residues in contact with the aAChR peptide (^BH4, ^BT6, ^BA7, ^BT8, ^BS9, ^BI11, ^BN21, ^BL22, ^BW28, ^BC29, ^BD30, ^BF32, ^BG37, ^BK38, ^BV40, ^BE41, ^BC48, ^BY54) as well as the C-terminal region (^BK70, ^BQ71) of the toxin experienced considerable changes (>0.2 ppm). Furthermore, due to a partial mutation of A31V in the commercially available aBTX [67], sequential assignment of four residues upstream in the sequence was complicated as they appeared twice. Peptide residues lying outside the binding determinant exhibited identical resonances to those of the free ap25, while residues within the binding site showed altered resonances. Interestingly, protons of ^RC192 were observed at two distinct chemical shift as the eight-membered ring of the disulfide bonded ^RC192-^RC193 adopt either cis- or trans- conformations. In this study, only the cis conformation was used for structure determination.

Secondary structure of the αp25/αBTX complex. The determination of the secondary structure of the complex was based on the characteristics of different secondary structure elements. For example, a-helices are characterized by the coexistence of strong H^N(i)/H^N(i+1) and medium H^a(i)/H^N(i+1) connectivies, as well as small (< 6 Hz) ³J_HNHαcouplings. On the other hand, b-sheets are characterized by strong H^a(i)/H^N(i+1) connectivities, by large (>8 Hz)³J_HNHαcouplings, by slowly exchanging amide protons and by large (>0.3 ppm) positive deviation of the H^a chemical shifts from their random coil values [48]. ³J_HNHα coupling constant were measured for α-BTX/αp25 residues and the deviation of the H^a chemical shifts from their random coil values were calculated as well. Figure 3 summarizes the sequential NOE, solvent exchange, ³J_HNHα coupling constants and H^a chemical shift deviation for a-BTX and ap25 in their complex.

Overall, the long range NOE interactions summarized in figure 4 indicate that αp25 forms a β-hairpin with the two b-strands ^RH186-^RV188 and ^RY198-^RD200 interacting with each other while the segment ^RV188-^RT191 forms an intermolecular b-sheet with residues ^BK38-^BE41 of the second finger of a-BTX. The (Cys)₂-Pro-Xxx-Xxx-Pro motif typical of αAChR forms the loop at the tip of the b-hairpin. The high ³J_HNHα coupling constant measured for ^RD195 and ^RT196 (9.7 and 9.6 Hz respectively) indicates that these two residues also assume an extended conformation. The slow solvent exchange rate of the amide protons of residues ^RY190, ^RC192 and ^RC193 suggests that they are shielded by intra-peptide or peptide-toxin interactions.

The secondary structure of unbound aBTX is composed of five b-strands [68] (fig. 4), consisting of residues ^BV2-^BT5, ^BS12-^BT15, L22-^BW28, ^BV40-^BA45 , and ^BE56-^BS60 [66]. Long range NOE contacts summarized in figure 4 as well as solvent exchange data indicate that the latter three strands are further elongated upon peptide binding to include residues ^BC29, ^BD30, ^BG37, ^BK38, ^BV39 and ^BE55 (fig. 4). The b-sheet interaction with the peptide accounts for the slow exchange rate of the amide protons of ^BK38 and ^BV40.

Peptide flexibility in the aBTX/ap25 complex During the spin-lock period in the T₁_r-filtered HOHAHA experiments the magnetization decays in the y-axis according to T₁_r, and with it the peak intensity according to the equation: I = I_o e^-t/T¹^r, where I_o is the initial peak intensity and I is that after t msec. As the spin-lock duration (t) in the edited HOHAHA spectra increases, peak intensity (I) decreases. Using six T₁_r-filtered HOHAHA experiments with increasing pulse duration, the T₁_r value of each amide protons in the complex was calculated. Small molecules with fast tumbling rates have long T₁_r values. T₁_r decreases as the molecular weight increases. Regions with internal flexibility will have increased T₁_r in comparison to well structured regions.The calculated values are shown in figure 5.

The overall data indicates that in aBTX the T₁_r values for amide protons are larger (40-60 sec) in loops where flexibility is high, and lower (20-40 msec) in ordered b-structure, or in residues that interact with ap25. In ap25, residues 184-200 (with the exception of ^RD195) attain T₁_r values of 20-40 msec, corresponding to the values of aBTX amide protons in the ordered secondary structure. This indicates that the binding domain is indeed the segment 184-200 and that ^RD195 does not interact with the toxin in agreement with the mapping of the binding determinant using the dynamic filtering experiments.

Structure determination. For the structure determination, we used a total of 1,002 distance constraints, of which 273 were long-range. Long-range NOE cross-peaks (connecting protons more than five residues apart) included 62 peptide/toxin, 20 intrapeptide and 191 intratoxin interactions. The f angle restraints included 46 angles for the toxin and 12 for the bound peptide. Twenty final structures that satisfy the experimental restraints without NOE violations larger than 0.5 Å and torsional violations exceeding 5º were obtained. Figure 6 shows the backbone ribbon diagrams of ten superimposed structures. Figure 7 displays a cartoon diagram of the minimum energy structure of the complex.

Structure of the bound aBTX. The structure of the bound aBTX is similar to that of aBTX in complex with a library peptide, previously determined in our laboratory [40]. The overall structure of aBTX consists of three long fingers and a tail, all stabilized by four disulfide bridges (C3/C23, C16/C44, C48/C59, and C60/C65) (Fig. 7). Two b-hairpins are formed by fingers I and II of aBTX. The first consists of two antiparallel b-strands, namely finger residues ^BV2-^BT5 and ^BS12-^BT15, which are connected by a 6 residue loop. The second involve the two antiparallel b-strands, ^BL22-^BD30 and ^BG37-^BA45 also connected by a six-residue loop. A third b-strand corresponding to residues ^BE55-^BC60 of finger III forms a triple-stranded antiparallel b-sheet with the two b-strands of the second finger. These motifs are present in all a-neurotoxins, thus creating the central core of the macromolecule. The structure reveals that the long C-terminus of aBTX is parallel to the b-hairpin of finger I. This observation together with the slow amide proton solvent exchange observed for residues ^BK64, ^BN66 and ^BH4 suggest that the segment ^BK64-^BN66 of the C-terminus is close to creating a parallel b-sheet with the segment ^BH4-^BT5 of finger I. Finally, a pair of hydrogen bonds, between ^BA7 and ^BL42, which was repetitively displayed in the calculated structures, stabilizes the interaction between the first and second finger of aBTX.

Structure of the bound aAChR peptide. As described earlier, the peptide adopts a b-hairpin conformation involving four hydrogen bonds (residues ^RV188 with ^BY198, and ^RH186 with ^BD200) (fig .7). Sidechains of 188, 190 and 198 are in close contact, further stabilizing this conformation by aromatic ring stacking. The disulfide bridged vicinal cysteines 192 and 193 form an eight-member ring in the preferred cis conformation.

Binding interactions. The binding between the aAChR peptide and aBTX is mediated by a network of hydrophobic interactions as well as hydrogen bonds. Surrounded by the toxin, the peptide fits snugly into the aBTX binding site. Sixteen residues of the toxin interact with seven peptide residues. The sidechains of ^RW184, ^RW187 and ^RY189 interact through hydrophobic interaction with residues of the first finger of aBTX (fig.8). Residues 188-191 interact with the second finger via an intermolecular b-sheet involving four hydrogen bonds (fig. 9). Finally, residues 189-192 interact with the C-terminus of the toxin (fig.10). Residue ^RY189 is the most prominent residue in the binding and forms the largest number of interactions with the toxin. The protons involved in the intermolecular b-sheet exchange very slowly with the solvent, and their cross-peaks were observable for over 24 hours in spectra recorded in D₂O. These findings suggest the binding is very tight. The NMR - derived intermolecular interactions in the complex are summarized in figure 11. (For 3-D representation of the interactions, please consult the disk hereby joined.)

DISCUSSION

In this study, we applied the NMR dynamic filtering technique [69, 70] to map the aAChR determinant recognized by α-BTX and to determine the relative residue immobilization. The three dimensional structure of the major AChR determinant involved in aBTX and acetylcholine binding in complex with the toxin was determined by 2D-NMR. Optimization of measurement conditions together with the resolution and sensitivity enhancement of the 800 MHz made the structure determination of this 11.2 KDa complex possible.

Mapping of the aAChR determinant. In a previous study conducted in our laboratory, a combination of HOHAHA and ROESY spectra with long mixing periods was used to map the gp120 epitope recognized by the HIV-1 neutralizing antibody 0.5β [71]. The three-dimensional structure of the complex between the gp120 peptide and the Fv fragment of the antibody was later solved by multi-dimensional NMR [72, 73] and was in excellent agreement with the dynamic filtering results. This attests to the potential of the dynamic filtering method in precisely mapping peptide segments interacting with a much larger protein. The current study pinpoints the major αAChR determinant interacting with α-BTX to residues ^RW184-^RD200. This segment was shown to bind α-BTX with a K_Dof 2.5 x 10^-7 M, similar to that of the denatured α-subunit to α-BTX. The shorter 184-196 αAChR segment exhibits a K_D of 4.7 x 10^-6 M, indicating that the sequence 197-200 contributes to α-BTX binding [28]. The mapping of the determinant interacting with aBTX was subsequently corroborated by the three dimensional structure of the complex. The higher affinity of intact pentameric AChR to α-BTX (K_D= 10^-11M) can be attributed to distal segments of the α-subunit as well as to residues in other subunits [1, 74].

Residue flexibility. In a more quantitative approach we also determined the T₁_r values of all amide protons in the aBTX/ap25 comlplex. Flexible peptide residues that do not contribute to aBTX binding attain large T₁_r values and vice-versa. The observation that residues ^RR182, ^RG183, ^RI201, ^RT202 and the flanking glutamates in ap25 have relatively large T₁_r values indicates they are highly flexible in comparison with the majority of a-BTX residues and do not contribute to the toxin binding. The finding that ^RD195 possesses a high T₁_r value of ~ 56 msec, suggests it is highly flexible and does not contribute to aBTX binding. This result correlates excellently with the structure of the bound peptide which was solved subsequently, where D195 is solvent exposed, and therefore does not contribute to the binding.

Structure of the aAChR/ap25 complex. The Ramachandran plot of the lowest energy structure is shown in figure 12. The f and y angles predominantly occupy allowed regions. Only three residues reside outside the allowed region, namely ^RC192, ^RE181 and ^RT202. The f and y angles of ^RC192 are distorted due to the vicinal disulfide bridge with ^RC193. Such disulfides are rare in proteins due to their unusual constraint, suggesting they must play an important role in agonist binding [1]. Residues of both peptide termini such as ^RE181 and ^RT202, adopt unusual dihedral angles due to the low number of NOE constraint in these random coil domain. The calculated rmsd of the backbone and sidechain atoms of twenty structures is 1.8 Å and ~3.5 Å respectively. The structures presented here include only a partial list of NOE constraints. The structure will be better defined after we include NOEs involving H^gand H^d of the sidechains, a process currently underway.

Conformation and structure of the complexed ap25 peptide and implications for aAchR conformation. Several studies have attempted to predict the secondary structure of αAChR. Finer-Moore and Stroud applied amphipathic analysis to αAChR and predicted a β-strand conformation for the ^RW184-^RD200 segment [42] with a turn formed by ^RC192 and ^RC193. Chaturvedi et al. [30] proposed a b-hairpin conformation in which ^RC192, ^RC193 and ^RP194 form a loop and all the other residues form b-strands. On the other hand, very recently Changeux and co-workers predicted no secondary structure for the ^RR182-^RD200 segment and a β-strand formed by residues ^RI201-^RM207 only [44]. In another recent modeling study, Ortells et al. used six different prediction methods and 118 aligned sequences to extend this β-strand to include residues ^RY198-^RI210, with residues ^RG183-^RP197 remaining unstructured [43]. This controversy between the earlier and more recent secondary structure predictions emphasizes the importance and relevance of the experimental structural data presented in our study.

The NMR data provides evidence for a b-hairpin conformation in which residues ^RH186-^RV188 and ^RY198-^RD200 form the two interacting b-strands, and residues ^RC192-^RP197 form the turn. As mentioned before, the ^RV188-^RT191 segment within the b-hairpin forms a b-sheet with the second finger of a-BTX. Basus and co-workers observed an extended conformation for the five N-terminal residues of the aAChR peptide ^RK185-^RT196 bound to a-BTX [39], but not the b-hairpin conformation, since ^RY198-^RD200 was absent in their peptide. The intermolecular b-sheet formation has not been observed as well, although this type of interaction was postulated by Hawrot and co-workers [41].

The β-hairpin structure of αp25 bound to a-BTX emulates that of the corresponding αAChR segment in complex with α-BTX. Designed by nature to bind to the corresponding segment of αAChR, the α-BTX binding site serves as a template and forces the flexible peptide to fold into a conformation similar to that of native aAChR. This enables the peptide to fit into the considerably more rigid binding site of a-BTX. The fact that the peptide must maintain a correct secondary structure in order to bind the toxin optimally is further supported from a study by Conti-Tronconi and coworkers. They show that a peptide corresponding to residues 182-201 of mouse aAChR bound aBTX weakly when attached to a solid support consisting of nitrocellulose, but was effective in competing for ¹²⁵I-aBTX in solution [51]. It was suggested that the solid phase assay might constrain a peptide to a limited set of conformation, which is not representative of its conformational repertoire in solution, and therefore the AChR competition assay is a more reliable assay than solid phase assays.

Changes in aBTX conformation upon peptide binding. Our findings demonstrate that aBTX undergoes conformational change upon binding ap25. The b-sheet of the second finger of aBTX is elongated to accommodate the peptide binding. Hawrot and coworkers have described that residues at the edge of b-sheet of the second finger, ^BW28 and ^BV39, zip together upon a binding of a dodecapeptide 185-196 [39]. However, in our structure with the longer peptide ap25, additional residues ^BC29-^BD30 and ^BG37-^BK38 extend the b-sheet, illustrating the importance of ^RP197-^RD200 in stabilizing the complex. The concerted elongation that occurs upon peptide binding and complex stabilization accounts for the changes in chemical shift of residues ^BW28-^BD30, ^BG37-^BE41 at the second fingertip.

Peptide-toxin interactions in the aBTX/ap25 complex. Our results show that hydrogen bonds as well as interactions of aromatic residues are important for aBTX binding of the aAChR peptide. This is in agreement with models for aBTX binding, which propose that the contacts between aBTX and the aAChR involve primarily hydrophobic and hydrogen bonding, with only a few electrostatic interactions [75]. Upon formation of the aBTX/ap25 complex, residues ^BT6-^BI11, ^BD30-^BV40 and ^BH68-^BR72 experience a chemical shift change in comparison with the unbound state. This is in agreement with our structure that indicates the aforementioned residues are important in binding of ap25. The structure of a library peptide bound to aBTX, solved previously in our lab, showed that the peptide interacts with residues of the first (^BT6, ^BA7, ^BI11) and second (^BD30, ^BR36, ^BK38, ^BV39, ^BV40) finger of aBTX as well as residues of the C-terminus (^BH68, ^BK70, ^BQ71, ^BR72) [40]. The high similarity of the toxin residues involved in either the library peptide or the aAChR peptide binding suggests that the library peptide imitates the binding of the natural aAChR.

Importance of the vicinal disulfide bridge ^RC192-^RC193. The significance of the vicinal disulfide bond ^RC192 and ^RC193 in the native receptor and its importance in aBTX binding has been the subject of much discussion in the literature. Reduction of the disulfide bond did not decreased significantly the peptide affinity to aBTX, however methylation of the free thiols decreased the affinity considerably [50, 51]. This latter finding coincide with our structure that indicates that ^RC192 is directly involved in aBTX binding, whereas ^RC193 and its disulfide-bridge are not. Nevertheless, it should be stated that the disulfide-bridge stabilizes the loop and passively contributes to ligand binding.

Comparison with site directed mutagenesis. In an attempt to elucidate residues required for aBTX binding to aAChR, Conti-Tronconi et al. substituted residues of a Torpedo 181-200 peptide with glycine and measured competition by unmodified peptides [32]. They showed that ^RV188, ^RY189, ^RY190, ^RC192, ^RC193 and ^RP194 were necessary for aBTX binding. In a similar study, systematic cysteine mutagenesis followed by covalent modification of the introduced cysteines with thiol specific reagent, showed that residues ^RW187, ^RV188, ^RF189, ^RY190 and ^RP194 contribute to aBTX binding [76]. Concurrently, other studies showed that mutations of residues ^RH186, ^RY189, ^RY190, ^RC192, ^RC193 and ^RP194 greatly reduced or abolished α-BTX binding, and mutation of ^RY198 and ^RD200 decreased the binding to a lesser extent [28, 30]. Finally, it was shown that α-BTX-affinity was conferred upon an α-BTX-insensitive α-subunit by introducing a cluster of five residues from the Torpedo sequence (^RW184, ^RW187, ^RV188, ^RY189 and ^RT191), indicating that these residues are important for toxin binding [33]. Thus, a large number of residues within the 184-200 sequence including ^RW184 and ^RD200 have been implicated by various experimental methods as important for α-BTX binding and coincide with our findings. However all studies were unable to determine whether the mutations decreasing the toxin binding were caused by covalent modification or by conformational change of the receptor. Our structure indicates that the residues ^RW184, ^RW187, ^RY189, ^RT191 and ^RC192 contribute to aBTX binding through their side chains and therefore the chemical modification leads to steric hindrance and a decrease in toxin affinity. Residues ^RH186, ^RV188, ^RY190, ^RC193, ^RP194, and ^RP197-^RD200 contribute to aBTX binding passively by maintaining the secondary structure and therefore the mutations destabilize the b-hairpin conformation and lead indirectly to a decreased affinity.

Species variation of the aAChR sequence. During evolution, natural selection resulted in mutations of the aAChR and subsequently nature provided us with a variety of aAChR sequences that differ from one species to another [77]. The natural preys of Bungarus Multicinctus are frogs and chicks, and it is therefore not surprising that aBTX binds lethally and with the highest affinity to their aAChR. Torpedo California and zebrafish possess similar sequences to those of frogs, and therefore present similar affinities [51]. On the other hand, snakes themselves and their natural predators such as the mongoose are naturally resistant to snake-venom in general, and aBTX in particular. Other species such as humans and hedgehogs, the latter being closely related to the mongoose, are subject to aBTX poisoning although they possess partial resistance [78]. In table 2, sequences of the aAChR of various species are presented together with their relative binding affinity. Determination of the influence of a mutation on the actual binding is a powerful tool in relating aAChR structure to its function.

180 190 200 aBTX affinity

Torpedo: EWVMKDYRG WKHWVYYTCCPDTPYLD ITYHF +++

Zebrafish: --------S -------A--------- ----- +++

Frog: --M-----C ------------K---- ----- +++

Chicken: --------- -------A--------- ----- +++

Calf: ---I-ES-- -----F-A---S----- ----- ++

Cat: ---I-ES-- -----F-A---T----- ----- ++

Shrew: ---I-EA-- -----F-A---T----- ----- ++

Mouse/Rat: ---I-EA-- -----F-S---T----- ----- ++

Hedgehog: ---I-EA-- ---R-I-A---S----- ----- +

Human: ---I-ES-- ---S-T-S--------- ----- +

Mongoose: ---I-EA-- ---N-T-A--LT-H--- ----- -

Cobra/Boa: --TL----- FW-S-N-S--L------ ----- -

Drosophila: A VRNEKF---- EE---- -VFN +

Locust: ER-EK--P--AE --P- +

-loop-

Table2: Sequences of aAChR of different species involved in aBTX binding and their affinity. Affinity values are transcribed from references [51, 77] and [79]. Residues contributing to aBTX binding are in bold. Residues interacting with aBTX are marked with colored arrows and residues involved in the b-hairpin are marked with black arrows. Most conserved residues within the binding determinant are colored red, and mutation leading to a decrease in aBTX affinity in yellow.

Overall, the table indicates that mutations of residues directly interacting with aBTX such as ^RW184, ^RW187 and ^RY189, lead to a decrease or loss of binding capability. Residues in the binding determinant that possess high flexibility, and do not contribute to aBTX binding, will not affect aBTX affinity, i.e. mutations of the solvent exposed residues ^RD195S or ^RD195T and ^RT196K in respectively mammals and frogs do not alter the binding affinity [51]. Our studies indicate that residues ^RH186, ^RW187, ^RV188, ^RY198, ^RL199 and ^RD200 are involved in b-hairpin formation in all aAChR species, because they are all conserved. Some of these residues may be involved in acetylcholine binding. In such a b-hairpin, every second residue points away from aBTX and therefore no natural selection leading to a mutation of this position occurred. In contrast, every other residue facing the toxin, such as ^RW187 and ^RY189, have been mutated in resistant species i.e. snakes, where ^RW187 is replaced by a serine and ^RY189 by a asparagine, or in the mongoose, where ^RW187 is replaced with an N-glycosylated asparagine [78], and ^RY189 by a threonine. The latter modifications hinder toxin contact and abolish the binding. Mutations of residues passively involved in aBTX binding by maintaining the rigid loop such as ^RP194L and ^RP197H in the mongoose, also decrease the binding affinity, suggesting they are important for aBTX binding. The ability of the human and hedgehog aAChR to bind aBTX, despite the nonconservative substitution of ^RW187 and ^RY189 [51], may result from the aptitude of adopting the necessary secondary structure and from the presence of W184, residing in the binding determinant. In contrary, the inability of neuronal aAChRs, such as rat a2, a3 and a4 (sequences not shown) to bind the toxin [29] results from the substitutions of ^RW184, ^RW187, ^RY189 that contribute direct contacts with aBTX and to the ^RP197I and ^RL199P mutation that may result in a loss of the b-hairpin structure, while maintaining acetylcholine affinity.

Agonist binding domain of the aAChR. It is commonly accepted that aBTX competitively inhibits channel opening by physically blocking the acetylcholine binding site. aAChR Structural comparisons of nicotinic agonists and antagonists have long predicted a negative subsite on the receptor that interacts with the positively charged alkyl-ammonium moiety common to nearly all nicotinic agents. It has been pointed out that aromatic residues could form an electronegative subsite [80-82], i.e., through the formation of a tyrosinate anion or the p electrons of electron-rich aromatic systems. The importance of aromatic rings in the binding of acetylcholine is supported by the determination of the three-dimensional structure of acetylcholinesterase, that reveals the active site to be lined by 14 aromatic residues and only a small amount of negatively charged residues [83]. Intrinsic fluorescence spectroscopic analyses together with binding studies of selectively modified peptide fragments of the AChR suggest that one or two invariant tyrosine residues at positions 190 and 198 on the a-subunit provide the critical negative subsite required for ligand binding [84]. Interestingly, the sidechains of ^RY190 and ^RY198 are adjacent (< 5 Å) in the bound peptide, possibly emulating their conformation in native aAChR. If so, the binding of aBTX and acetylcholine to the receptor would be on the two opposite faces of the b-hairpin. This hypothesis together with our findings that the aAChR peptide is immobilized upon aBTX binding, suggest that aBTX also acts as an antagonist by “freezing” the aAChR binding determinant so that acetylcholine (if still binding) could induce no conformational changes required for channel opening. However this still remains to be clarified.

Comparison to other a-neurotoxins. NMR structural data [66] indicate that the solution structure of aBTX, while consistent with the x-ray structures of a-cobratoxin and erabutoxin [85-87], is different in several key respects from the x-ray structure of aBTX, i.e. the amount of b-sheet structure [68]. Several other members of the a-neurotoxin family have been studied by NMR [15, 64, 65, 88-91]. A repeating structural motif, consisting of three fingers and a flexible C-terminus, is observable in the a-neurotoxin family, and their sequence

Bungarotoxin	IVCHTTATS P ISAVTCPPGE NLCYRKMWCD AFCSSRGKVV ELGCAATCPS KKPYEEVTCC STDKCNPHPK QRPG
Cobratoxin	IRCFI T PDITSKDCPNG HVCYTKTWCD AFCSIRGKRV DLGCAATCPT VKSGVDINCC STDNCNPFPT RKRP
Erabutoxin	RICFNHNSSNPN TTKTCSPGN SSCYHKNWSD F RGTII GRGCG CPT VKPGIKLSCC ESEVCNN
Acanthophis antarticus	VICYRGY NNP Q TCPPGE NVCFTRTWCD AFCSSRGKVV ELGCAATCPI VKSYNEVKCC STDKCNPFPV RPRRPP
King cobra	TKCYKTGDR I ISE ACPPGQ DLCYMKTWCD VF