Identifying Key Amino Acid Residues That Affect α-Conotoxin AuIB Inhibition of α3β4 Nicotinic Acetylcholine Receptors*

Background: α-Conotoxin AuIB interacts with α3β4 nAChRs and GABAB receptors, but structural determinants of these interactions are unknown. Results: Using alanine scanning mutagenesis and molecular dynamics, we identified residues crucial for AuIB·α3β4 nAChR interaction. Conclusion: We identified the key residues that mediate AuIB·α3β4 nAChR interaction. Significance: Ability to direct α-conotoxin binding to nAChRs or GABAB receptors will improve analgesic conopeptides. α-Conotoxin AuIB is a selective α3β4 nicotinic acetylcholine receptor (nAChR) subtype inhibitor. Its analgesic properties are believed to result from it activating GABAB receptors and subsequently inhibiting CaV2.2 voltage-gated calcium channels. The structural determinants that mediate diverging AuIB activity at these targets are unknown. We performed alanine scanning mutagenesis of AuIB and α3β4 nAChR, homology modeling, and molecular dynamics simulations to identify the structural determinants of the AuIB·α3β4 nAChR interaction. Two alanine-substituted AuIB analogues, [P6A]AuIB and [F9A]AuIB, did not inhibit the α3β4 nAChR. NMR and CD spectroscopy studies demonstrated that [F9A]AuIB retains its native globular structure, so its activity loss is probably due to loss of specific toxin-receptor residue pairwise contacts. Compared with AuIB, the concentration-response curve for inhibition of α3β4 by [F9A]AuIB shifted rightward more than 10-fold, and its subtype selectivity profile changed. Homology modeling and molecular dynamics simulations suggest that Phe-9 of AuIB interacts with a two-residue binding pocket on the β4 nAChR subunit. This hypothesis was confirmed by site-directed mutagenesis of the β4-Trp-59 and β4-Lys-61 residues of loop D, which form a putative binding pocket. AuIB analogues with Phe-9 substitutions corroborated the finding of a binding pocket on the β4 subunit and gave further insight into how AuIB Phe-9 interacts with the β4 subunit. In summary, we identified critical residues that mediate interactions between AuIB and its cognate nAChR subtype. These findings might help improve the design of analgesic conopeptides that selectively “avoid” nAChR receptors while targeting receptors involved with nociception.

␣-Conotoxin AuIB is a selective ␣3␤4 nicotinic acetylcholine receptor (nAChR) subtype inhibitor. Its analgesic properties are believed to result from it activating GABA B receptors and subsequently inhibiting Ca V 2.2 voltage-gated calcium channels. The structural determinants that mediate diverging AuIB activity at these targets are unknown. We performed alanine scanning mutagenesis of AuIB and ␣3␤4 nAChR, homology modeling, and molecular dynamics simulations to identify the structural determinants of the AuIB⅐␣3␤4 nAChR interaction. Two alanine-substituted AuIB analogues, [P6A]AuIB and [F9A]AuIB, did not inhibit the ␣3␤4 nAChR. NMR and CD spectroscopy studies demonstrated that [F9A]AuIB retains its native globular structure, so its activity loss is probably due to loss of specific toxin-receptor residue pairwise contacts. Compared with AuIB, the concentration-response curve for inhibition of ␣3␤4 by [F9A]AuIB shifted rightward more than 10-fold, and its subtype selectivity profile changed. Homology modeling and molecular dynamics simulations suggest that Phe-9 of AuIB interacts with a two-residue binding pocket on the ␤4 nAChR subunit. This hypothesis was confirmed by site-directed mutagenesis of the ␤4-Trp-59 and ␤4-Lys-61 residues of loop D, which form a putative binding pocket. AuIB analogues with Phe-9 substitutions corroborated the finding of a binding pocket on the ␤4 subunit and gave further insight into how AuIB Phe-9 interacts with the ␤4 subunit. In summary, we identified critical residues that mediate interactions between AuIB and its cognate nAChR subtype. These findings might help improve the design of analgesic conopeptides that selec-tively "avoid" nAChR receptors while targeting receptors involved with nociception.
Peptides isolated from the venom of cone snails belonging to the genus Conus are valuable pharmacological tools, and some are also promising drug leads (1)(2)(3)(4)(5). ␣-Conotoxins are a subfamily of these peptides and consist of 12-19 amino acid residues, including four cysteines with a characteristic CC-C-C arrangement (type I cysteine framework) (6,7). These four cysteines can yield three possible disulfide connectivities: globular (I-III, II-IV), ribbon (I-IV, II-III), and beads (I-II, III-IV). However, naturally occurring ␣-conotoxins typically exhibit the "globular" conformation (7). The number of amino acids in each of the two loops between the framework cysteine residues is used to divide ␣-conotoxins into subclasses. For example those with four amino acids in loop 1 and six in loop 2 are referred to as 4/6-␣-conotoxins.
Nicotinic acetylcholine receptors (nAChRs) 6 are transmembrane proteins that form cationic ligand-gated channels that mediate fast excitatory cholinergic neurotransmission in the central nervous system (CNS). They also have an important regulatory role in the body, modulating the release of several neurotransmitters. The importance of nAChRs is emphasized by their involvement in various CNS disorders, including Alzheimer disease, schizophrenia, pain, nicotine addiction, and cancer (8 -11).
The first disulfide bond in each peptide was formed by incubating peptides in 0.1 M NH 4 HCO 3 (pH 8.2, 0.3 mg/ml) overnight at 22°C, then purifying them by RP-HPLC. The second disulfide bond was formed by incubating peptides with iodine in acidic conditions. Peptides were dissolved in 50% aqueous acetic acid (0.5 mg/ml). To this solution 100 l of 1 M HCl/mg of peptide was added, then a solution of 0.1 M I 2 in 50% acetic acid was slowly added until the solution became pale yellow. The reaction mixture was stirred for 12 h at 22°C in the dark. The reaction was quenched by adding 1 M ascorbic acid until the mixture became colorless. The peptide was purified by RP-HPLC, and fractions were combined after analytical RP-HPLC confirmed purity. Electrospray mass spectrometry was used to confirm the peptide identity.
Nuclear Magnetic Resonance (NMR) and Circular Dichroism (CD)-NMR spectra for all AuIB analogues were recorded on samples dissolved in 90% H 2 O and 10% D 2 O at a pH of ϳ4. Bruker Avance 500 and 600 MHz NMR spectrometers were used to acquire spectra, including 1 H, total correlation spectroscopy (TOCSY) and nuclear Overhauser effect spectroscopy (NOESY) data, as described previously (18,31), and processed using Topspin (Bruker). All spectra were recorded at 290 K, and chemical shifts were referenced to the residual water signal at 4.85 ppm. Processed spectra were analyzed and assigned within the program Sparky (32).
For CD spectroscopy experiments, 70 M concentrations of each peptide were dissolved in 20 mM sodium phosphate buffer at pH 7. To examine the helical propensity of each isomer, CD data were also obtained for each product after 30% tetrafluoroethene was added to the solution. Spectra were acquired on a Jasco J-810 spectropolarimeter, which was routinely calibrated using 0.6% (w/v) ammonium-D-camphor-10-sulfonate. All experiments were conducted at room temperature (21-23°C) under a nitrogen atmosphere (15 ml/min). The experimental parameters were set to a scanning speed of 50 nm/min, response time of 1 s, sensitivity range of 100 millidegrees, and a step resolution of 1 nm. Absorbance was measured in the far UV region (185-260 nm) using a 1-mm path length quartz cuvette. Each recording was an accumulation of four scans. To eliminate any possible interference from the solvent, cuvette, and spectropolarimeter optics, we subtracted CD spectra of the pure solvents from each sample.
Protein Sequence Alignment-The National Institute of Health's online Constraint-based Multiple Alignment Tool, COBALT (www.ncbi.nlm.nih.gov), was used to align protein sequence, and conservative domains were taken into account. Residues of nAChR subunits were numbered according to the sequences of the mature proteins that lacked the signal peptide at the start of their sequences.
cRNA Preparation-Plasmid DNAs encoding rat ␣3, ␣4, ␣9, ␣10, ␤2, and ␤4 nAChR subunits and human ␣7 nAChR subunits were obtained from Dr. J. Patrick, Baylor College of Medicine, Houston TX, Dr. J. Lindstrom University of Pennsylvania, Philadelphia PA, and OriGene Technologies Inc., Rockville MD. The plasmids were linearized with appropriate restriction enzymes, and cRNA was synthesized in vitro using a SP6 or T7 in vitro transcription kit (mMessage mMachine; Ambion, Foster City, CA). RNA for different nAChR subunits to be co-injected into the same oocytes was synthesized in parallel on the same day using identical procedures to maximize the consistency of concentration and purity between subunits.
Electrophysiological Recordings and Data Analysis-Twoelectrode voltage clamp recordings from oocytes were carried out at room temperature using a GeneClamp 500B amplifier (Molecular Devices Corp., Sunnyvale, CA) at a holding potential of Ϫ80 mV. The voltage-recording and current-injecting electrodes were pulled from borosilicate glass (GC150T-15, Harvard Apparatus Ltd.) and had resistances of 0.3-1.5 megaohms when filled with 3 M KCl. Oocytes were continuously per-fused in a recording chamber with a volume of ϳ50 l, with ND96 solution at 2 ml/min, applied by a gravity-fed perfusion system. nAChR-mediated currents were evoked by pipetting 100 l of acetylcholine (ACh) into the bath when the perfusion was temporarily halted. ACh concentration was 50 M unless specified otherwise. Oocytes were preincubated with the peptide for ϳ5 min, then ACh and the peptide were co-applied. Peak ACh-evoked current amplitude was recorded before and after peptide incubation using pClamp 9 software (Molecular Devices). The effects of native AuIB and its peptide analogues on ACh-evoked nAChR-mediated currents were defined as peak current amplitudes relative to the average peak current amplitude of 3-5 control ACh applications, recorded before preincubation with the peptides. Concentration-response curves for AuIB and [F9A]AuIB were fitted by unweighted nonlinear regression to the logistic equation, where E x is the response, X is the peptide concentration, E max is the maximal response, nH is the slope factor (Hill slope), and IC 50 is the peptide concentration that gives 50% inhibition of the maximal response. All electrophysiological data were pooled (n ϭ 3-6 for each data point) and represent the arithmetic means Ϯ S.E. of the mean. One-way analysis of variance followed by Bonferroni's post hoc test was used to compare current amplitudes affected by AuIB analogues with those of the native peptide. Data were statistically analyzed using SigmaPlot Version 11.0 (Systat Software Inc., San Jose, CA) or Prism5 (GraphPad Software Inc., La Jolla, CA). Molecular Modeling and Docking Simulation-Homology models of the extracellular ligand binding domain of the rat (␣3) 2 (␤4) 3 nAChR bound to AuIB, [F9A]AuIB, or [F9Y]AuIB were constructed using the crystallographic coordinates of Aplysia californica acetylcholine-binding protein (AChBP) co-crystallized with the double mutant ␣-conotoxin PnIA[A10L,D14K] (Protein Data Bank accession code 2BR8) as a template. This template was chosen to provide a suitable 4/7 ␣-conotoxin-bound conformation of the receptor for subsequent molecular dynamics (MD) simulations and analyses.
AuIB and mutant peptides were modeled bound to the two ␣3(ϩ)␤4(Ϫ) receptor binding sites using the geometry of the PnIA mutant in the AChBP crystal structure as a template. Rat ␣3 and ␤4 sequences were obtained from the Swiss-Prot database (codes P04757 and P12392, respectively) and aligned with the template sequence using the ClustalW server. BLOSUM was used as the scoring matrix. With the multiple alignment as input, 10 AuIB⅐␣3␤4, [F9A]AuIB⅐␣3␤4, and [F9Y]AuIB⅐␣3␤4 complex models were generated using Modeler9v6. The topranking models were selected and validated using PROCHECK. MD simulations were carried out on the top models of both receptor complexes.
MD Simulations-Before MD simulations, the energies of the complexes were minimized using the steepest descent algorithm and an energy gradient convergence criterion of 0.01 kcal/ mol/Å. Each receptor complex was placed in icosahedral simulation boxes with edge lengths of 100 ϫ 100 ϫ 100 Å and solvated with 34,000 TIP3P water molecules. To neutralize charge and pro-␣-Conotoxin AuIB Interaction with ␣3␤4 nAChRs vide a salt concentration of ϳ150 mM, 94 Na ϩ and 70 Cl Ϫ ions were added to the solvent. All simulations were performed using GROMACS Version 4.5 (75) with the CHARMM27 forcefield (with cmap) (76,77). All subsequent simulations were performed using a constant particle number, pressure, and temperature ensemble. Temperature was maintained at 300 K using the Nose-Hoover temperature coupling algorithm, and pressure was maintained at 1 atm using Berendsen's pressure coupling algorithm. Time steps of 2 fs were used to integrate all simulations. Solvent equilibration simulations of 100 ps lengths were performed. The non-hydrogen atoms of the receptor and peptides were positionally restrained so the solvent and ions could relax from an initially semi-crystalline structure. "Data collection" simulations of both complexes were then conducted, and all atoms of the system were free to undergo dynamics. Each complex was simulated for 100 ns. All analyses were performed on the final 20 ns of the trajectories to reduce bias from initial homology model conformations. Molecular graphics were produced using VMD Version 1.9.2 (78). All analyses were done using a combination of VMD, GROMACS analysis software suite, and in-house scripts.

Alanine Scanning Mutagenesis Identifies Residues in the AuIB Sequence That Are Critical for Its Interaction with the
␣3␤4 nAChR Subtype-To find residues in the AuIB sequence that contribute most to ␣3␤4 nAChR inhibition, we performed alanine scanning mutagenesis of the peptide. We systematically substituted each of the original residues to alanine, except for the four cysteines essential for maintaining the peptide globular structure and the native alanine at position 10 (see Fig. 1, A and B). All peptides were successfully synthesized using solid-phase peptide synthesis and a regioselective approach to form disulfide bonds to produce the globular disulfide framework. For future reference, we refer to globular AuIB as native AuIB. Each alanine mutant structure was analyzed using NMR and CD spectroscopy. The NMR spectral data for each peptide except [P6A]AuIB were successfully assigned whereby ␣H i -NH iϩ1 NOE connectivities observed in NOESY spectra were used in the sequential assignment of the individual spin systems determined from TOCSY spectra.
The [P6A]AuIB mutant exhibited broadened signals and multiple conformations in the TOCSY and NOESY spectra, so could not be assigned. The loss of secondary structure in the [P6A]AuIB mutant was confirmed by CD spectroscopy (Fig.  1C). Pro-6 is the only highly conserved amino acid residue in ␣-conotoxins apart from the cysteines and is responsible for helix initiation by inducing the 3 10 helix turn in the peptide backbone (33,34). Mutation of Pro-6 to alanine has also been shown to disrupt the ␣-conotoxin Vc1.1 structure (35).
Secondary ␣H chemical shifts represent the difference between an observed ␣H chemical shift and that of the corresponding residue in a random coil peptide. They are strong indicators of the presence of a secondary structure (36 (Fig. 1D). These nega-tive ␣H secondary shifts indicate a helical region, which is consistent with the previously reported three-dimensional structure of native AuIB (21) and other ␣-conotoxin structures reported to date (37). Interestingly, NMR secondary shift data for [Y5A]AuIB and [N12A]AuIB were more consistent with the ribbon isomer of AuIB (21) despite using a regioselective disulfide bond formation strategy to form the native (globular) isomer (Fig. 1D). This was confirmed by CD spectroscopy (Fig.  1C).
The effect of point modifications in the AuIB analogues on ␣3␤4 nAChR-mediated current inhibition was examined by two-electrode voltage clamp recordings in oocytes expressing the ␣3␤4 nAChR subtype. The relative amount of inhibition AuIB alanine-substituted analogues produced was compared with that of non-modified peptides at a fixed concentration (3 M; ϳ IC 50 AuIB (14)) ( Fig. 2, A and B).
Only three residues in the AuIB sequence significantly reduced inhibition of relative peak ACh-evoked current amplitude when alanine was substituted for them. These were Gly-1 (0.82 Ϯ 0.11, n ϭ 3; p Ͻ 0.05), Pro-6 (1.02 Ϯ 0.07, n ϭ 3; p Ͻ 0.001), and Phe-9 (1.07 Ϯ 0.15, n ϭ 3; p Ͻ 0.001) compared with a relative peak current amplitude of 0.51 Ϯ 0.07 (n ϭ 6) obtained in the presence of native AuIB (Fig. 2B). Interestingly, Gly1 replacement by Ala reduced ␣3␤4 nAChR inhibition, although Gly1 does not belong to either loop 1 or loop 2 of AuIB, which are thought to be the primary mediators of ␣-conotoxin interaction with nAChRs (6, 7). Unlike the [G1A]AuIB mutation, which retained minor inhibitory activity, P6A and F9A analogues exhibited a complete loss of inhibitory activity. As the NMR and CD data showed that secondary structure of [P6A]AuIB is irregular (Fig. 1, C and D), we concluded that this peptide's loss of inhibitory activity was due to disruption in the three-dimensional structure. This is not surprising, as Pro-6 is believed to be responsible for inducing the 3 10 helix turn in the backbone of ␣-conotoxins (33,34). In contrast, [F9A]AuIB NMR and CD spectrum data were consistent with the native AuIB structure (Fig. 1, C and D). This led us to conclude that the F9A substitution is specific and relevant to the peptide-receptor interaction rather than a general disruption of the peptide structure.
Characterization of [F9A]AuIB Inhibition of the ␣3␤4 nAChR Subtype-Our next step was to probe the degree to which [F9A]AuIB inhibitory activity is impaired by modifying the Ala to Phe substitution. Because the [F9A]AuIB analog was not active at ϳIC 50 concentration of the native AuIB (3 M) ( Fig.  2A), we tested this analog at higher concentrations, up to the highest practical concentration available (30 M). The [F9A]AuIB analog applied at 10 M mildly inhibited AChevoked currents, with the amplitude reduced to 0.77 Ϯ 0.04 (n ϭ 4) of normalized control currents (Fig. 3, A and B). This indicates that Phe to Ala substitution at position 9 of AuIB may strongly reduce the affinity of the modified peptide for ␣3␤4 nAChRs but does not disrupt peptide-receptor interaction altogether. However, native AuIB completely blocked the current at this concentration (0.05 Ϯ 0.01, n ϭ 6) (Fig. 3, A and B). [F9A]AuIB interaction with ␣3␤4 nAChRs is markedly reduced (Fig. 3B).

Homology Modeling and MD Simulations Suggest Loss of Interactions between AuIB and Key Receptor Residues When
Ala Is Substituted for Phe-9-Having established that position 9 Phe in AuIB is crucial for the peptide inhibiting ␣3␤4, we used atomistic simulations of native AuIB and the [F9A]AuIB mutant bound to ␣3␤4 to provide molecular-level explanations of why the mutation so markedly reduces inhibition. In particular, whereas experimental alanine scanning mutagenesis on AuIB identified peptide residues crucial for its efficacy, homol-ogy modeling and MD simulations identified receptor residues that might play important roles in AuIB inhibition of ␣3␤4 nAChR. Representative homology models of AuIB-bound ␣3␤4 are shown in Fig. 5, A and B.
To elucidate which receptor residues probably need contact with the peptide for inhibition to occur, we first calculated the average number of interatomic contacts between native AuIB or [F9A]AuIB and each receptor ectodomain residue. We then subtracted the number of contacts between each receptor residue and [F9A]AuIB from the number of contacts between each receptor residue and native AuIB. This measured the loss (or gain) of peptide contacts at each receptor residue caused by the mutation at AuIB position 9. Receptor residues with a negative contact difference might be important for AuIB inhibition.   ␣-Conotoxin AuIB Interaction with ␣3␤4 nAChRs NOVEMBER 29, 2013 • VOLUME 288 • NUMBER 48

JOURNAL OF BIOLOGICAL CHEMISTRY 34433
5C shows the ␣3␤4 receptor residues and mutation-induced changes in terms of the number of contacts with the peptide.
At the ␣3(ϩ) face, loss of peptide contact at several residues (e.g. Tyr-93, Cys-192, and Cys-193) is offset by increased contact at others (e.g. Ile-188, . Therefore, on average, peptide contact at the (ϩ) face slightly increased. In contrast, at the ␤4(Ϫ) face, the F9A mutation caused widespread loss of contact across many residues (with notable exceptions at Arg-115). Therefore, removing the bulky phenyl side chain at position 9 makes AuIB detach from the ␤4(Ϫ) face. This slightly increased the peptide's number of interatomic contacts at the opposing ␣3(ϩ) face.
Based on this analysis, we propose several specific receptor residues that are especially important for AuIB inhibition of ␣3␤4. First, because we demonstrated that the F9A mutation substantially reduces AuIB inhibitory efficacy, we examined the receptor residues that bind Phe-9, because these are probably crucial for AuIB inhibition of ␣3␤4. The homology model and MD simulation trajectory of wild-type AuIB/␣3␤4 suggested that Phe-9 makes contact exclusively at the ␤4(Ϫ) face, with its phenyl ring sandwiched between the Trp-59 indole ring and hydrocarbon segment of the Lys-61 side chain (Fig. 6B). As expected, both of these residues lose contact with the peptide when position 9 of AuIB is substituted (Fig. 5C).
The geometry of contacts between AuIB-Phe-9 with ␤4-Trp-59 and ␤4-Lys-61 suggests that cation-andinteractions may be important in AuIB inhibition of ␣3␤4. However, we caution that the CHARMM27 force field does not explicitly account for interactions involving electrons and acknowledge that methods which explicitly model cation-interactions are needed to produce quantitatively accurate geometries involving charged and aromatic side chains. Nonetheless, the force field we used partially mimicked cation-interactions and predicted Lys-Phe and Phe-Trp interaction geometries that are qualitatively in agreement with similar interactions observed in the PDB. In particular, our simulation suggests that contacts are made between ␤4-Lys-61 and the AuIB-Phe-9 phenyl ring primarily through the methylene carbons adjacent to the amine. In the PDB, Lys is known to engage with systems more commonly via carbon (38,39). Furthermore, our simulation suggests an offset-stacked interaction between ␤4-Trp-59 and AuIB-Phe-9. This geometry was identified as a common structural motif in Trp-Phe interactions in the PDB (40).
In addition, at the ␣3(ϩ) face, there is modest but statistically significant loss of contact between AuIB and both C-loop cysteines when position 9 of AuIB is substituted (Fig. 5C). This is due to reduced contact between the AuIB-Cys2/8 cysteines and ␣3-Cys-192/193. A snapshot from the MD trajectory of the native AuIB (Fig. 6C) illustrates close contact between the Cys-2-8 sulfurs of the peptide and the Cys-192/193 sulfurs of ␣3. In contrast, a similar snapshot for the [F9A]AuIB trajectory (Fig.  6D) illustrates loss of close contacts between the peptide and receptor disulfides. This result is especially intriguing, given that direct contact between peptide and C-loop cysteines is proposed to be essential for the peptide competitive antagonism of nAChRs, based on results from dicarba-conotoxin variants (41). Fewer close interatomic contacts between AuIB and ␣3 C-loop cysteines, indirectly caused by substituting Ala for Phe at position 9 (which lies on the opposite side of the peptide), may contribute to loss of inhibitory efficacy of this mutant. Therefore, we propose that Phe-9 has two roles in the AuIB inhibition of ␣3␤4: 1) direct contact with ␤4-Trp-59 and Lys-61 and 2) indirect facilitation of close contact between AuIB Cys-2/8 and ␣3-Cys-192/193.

JOURNAL OF BIOLOGICAL CHEMISTRY 34435
was present when ␤4-Trp-59 was mutated alone ([W59A]␤4). Therefore, we used a high agonist concentration (2.5 mM ACh) to evoke currents in all groups to create a meaningful comparison. The high agonist concentration should not interfere with AuIB inhibition, because AuIB inhibits the ␣3␤4 nAChR noncompetitively irrespective of ACh concentration (14). We assessed the effect of ␤4 mutations using a high concentration of native AuIB (10 M), which efficiently inhibits wild-type ␣3␤4 nAChRs (14).
How Side Chain Size and Hydrophobicity at Position 9 of AuIB Affect Inhibition at the ␣3␤4 nAChR-Given the critical interaction between AuIB-Phe-9 and ␤4-Trp-59 and the proposed importance of cation-andinteractions for the inhibitory activity of AuIB, we assessed how side chain size, aromaticity, and hydrophobicity of the amino acid residue at position 9 in the AuIB sequence affect nAChR inhibition. In particular, we sought to find position 9 substituents that may better fit the Trp-59 -Lys-61 pocket on the ␤4 nAChR subunit than Phe. We probed the inhibitory action of AuIB analogues substituted at position 9 by large aromatic (Tyr, Trp, 3-(2naphthyl)-L-alanine (NAL)) and small (Gly) side chains on wildtype ␣3␤4 nAChRs. CD analysis confirmed that the globular peptide structure was retained in all of these analogues (Fig. 8A). Substituting Phe-9 to glycine ([F9G]AuIB, 10 M) dramatically reduced inhibition (0.90 Ϯ 0.04, n ϭ 3; p Ͻ 0.001) compared with native AuIB (0.047 Ϯ 0.01, n ϭ 6) (Fig. 8B). This is consistent with our proposed model in which AuIB interaction with the binding pocket on the ␤4 subunit is crucial. This interaction is presumably absent when Phe-9 is replaced by the side chainfree Gly.
Another AuIB analog in which Trp was substituted for Phe-9 (F9W) substantially reduced inhibition (0.80 Ϯ 0.07, n ϭ 3; p Ͻ FIGURE 6. Homology modeling and MD simulation of AuIB⅐␣3␤4 binding. A, main interacting residues between AuIB and ␣3 subunit. Receptor backbone is in transparent blue ribbon form, with receptor residues of interest in Corey-Pauling-Koltun representation and indicated with arrows. Peptide backbone is in opaque red ribbon form, with residues of interest shown as van der Waals spheres and indicated with arrows. B, main interacting residues between AuIB and ␤4 subunit. Receptor backbone is in transparent yellow ribbon form. At the ␣3(ϩ) ␤4(Ϫ) interface, the only "unique ␣3" residue making direct contact with AuIB is Gln-198. However, this occurs via backbone only. On the ␤4 subunit, Phe-9 of AuIB is stabilized by Lys-61 and Trp-59 of the receptor. Phe-9 substitution to alanine results in loss of interactions with both these residues. Trp-59 is common to all neuronal nAChRs and so does not determine AuIB selectivity. Lys-61 is unique to the ␤4 subunit, so determines AuIB selectivity for ␣3␤4. C, snapshot illustrating close contact between AuIB-Cys2 (van der Waals spheres) and ␣-Conotoxin AuIB Interaction with ␣3␤4 nAChRs 0.001) presumably due to steric clashes. Indeed, automated blind docking of [F9W]AuIB to ␣3␤4 using Autodock (42) with all peptide side chain torsions free to rotate showed that none of the docking solutions involved binding of analog [F9W]AuIB to the canonical C-loop pocket. In contrast, docking of both native and [F9Y]AuIB predicted binding conformations very close to that of the double mutant PnIA-AChBP co-crystal structure (2BR8) (data not shown). An analog with an even larger unnatural amino acid, NAL, to substitute Phe-9 ([F9NAL]AuIB) also showed dramatically reduced inhibition (0.82 Ϯ 0.07, n ϭ 3; p Ͻ 0.001) (Fig. 8B). The NAL was attached to the 2-position of the Ala residue. Unlike the Trp residue, which has an N-heterocyclic five ring, NAL has a homocarbocyclic six ring.
Together, these results demonstrate that side chain size, aromaticity, and hydrophobicity at position 9 of AuIB are important for interaction between the peptide and ␤4 sub-unit of the ␣3␤4 pentamer. In contrast with the above mentioned analogues with substitutions at position 9, inhibition in a control AuIB analog with a histidine substitution at position 12 (N12H) was not significantly impaired (0.069 Ϯ 0.003, n ϭ 3) (Fig. 8B).

DISCUSSION
Using alanine scanning mutagenesis, we identified three residues in the AuIB sequence (Gly-1, Pro-6, and Phe-9) that affect inhibition of ␣3␤4 nAChRs. The Gly to Ala substitution only moderately reduced inhibition. Homology modeling of the AuIB⅐␣3␤4 complex suggests that the N terminus NH 3 ϩ of AuIB forms a salt bridge with the ␤4-Asp-172 side chain. The G1A mutation introduces a non-polar CH 3 side chain that may weaken the favorable interaction between the peptide N ter- it has an auxiliary role in forming the binding pocket for the hydrophobic interaction with Trp-59, which is key for inhibition. Note that only high ACh concentration elicited measurable currents in the W59AϩK61A mutant. Therefore, 2.5 mM ACh was used as an agonist to compare inhibition in the mutants and wild-type receptors. Data represent the means Ϯ S.E., n ϭ 3-6. ** indicates p Ͻ 0.005; *** indicates p Ͻ 0.001 for relative reduction of current inhibition versus wild-type ␣3␤4 nAChRs.
␣-Conotoxin AuIB Interaction with ␣3␤4 nAChRs NOVEMBER 29, 2013 • VOLUME 288 • NUMBER 48 minus and ␤4-Asp-172 side chain. Additionally, in a recent study investigating N-terminal post-translational modification of ␣-conotoxin MI, truncation of the N-terminal residue decreased the affinity for the muscle-type nAChR, a direct pharmacological effect indicating the N-terminal residues are important for bioactivity (43). ␣-Conotoxin LsIA has also shown reduced potency at ␣7 and ␣3␤2 nAChRs when N-terminally truncated (44), which corroborates the potential importance of the N-terminal amino acid in ␣-conotoxin potency and selectivity.
The other two substitutions, P6A and F9A, caused an even greater decrease in peptide activity on ␣3␤4 nAChR than G1A. Our NMR and CD data revealed that the [P6A]AuIB analog structure is disrupted, but that of [F9A]AuIB is not. This is not surprising considering the structural effect Pro has on protein structures in general and ␣-conotoxin structures in particular (34). Therefore, we reasoned that the ability of Phe-9, but not Pro-6, to reduce inhibition is due to Phe-9 specifically interacting with the ␣3␤4 nAChR. We found that the number of mutations that reduce AuIB inhibition of the ␣3␤4 nAChR is small compared with other ␣-conotoxins. For instance, alanine substitutions of almost all intercysteine residues in another ␣-conotoxin, Vc1.1, affected ␣9␣10 nAChR inhibition (35). We also found no alanine mutants that inhibited the ␣3␤4 nAChR significantly more than native AuIB.
As well as substantially reducing ␣3␤4 inhibition, [F9A]AuIB selectivity for nAChR subtypes also shifted. This indicates that Phe-9 has a role in the peptide specific interaction with the FIGURE 8. The effects of various substitutions at position 9 of AuIB on its inhibition of the ␣3␤4 nAChR subtype. A, CD spectra of AuIB analogues in which Phe-9 was substituted by various other residues. These spectra were similar to the spectrum of native globular AuIB, confirming that all of these analogues retained their structure. B, bar graph showing inhibition of ␣3␤4 nAChR ACh-evoked current amplitude by native AuIB and its F9-substituted analogues (10 M). Note that inhibition is abolished if the position 9 residue is too small or too large. Still present, but reduced inhibition by the Tyr analog suggests that hydrophobic interaction with the binding pocket at the ␤4 subunit is reduced because of steric hindrance. A random control mutation at position 12 of AuIB ([N12H]AuIB) does not attenuate maximal inhibition by AuIB. Data represent the means Ϯ S.E., n ϭ 3-6. *** indicates p Ͻ 0.001 for relative reduction of current inhibition versus native AuIB.

␣-Conotoxin AuIB Interaction with ␣3␤4 nAChRs
␣3␤4 nAChR and is needed to maintain selectivity for this particular subtype. Our docking simulation and MD calculations showed that the (Ϫ) side of the ␤4 subunit is the site where AuIB makes the most intermolecular contacts. Modeling also suggested that Phe-9 is sandwiched by Trp-59 and Lys-61, which may form a binding pocket on the ␤4 subunit. This model's validity was corroborated by mutational analysis of the ␤4 subunit and position 9 of AuIB.
The Role of the ␤4 nAChR Loop D and Its Conserved Tryptophan in Binding to AuIB and the Pharmacology of Various Ligands-Analysis of the ␤4 sequence (see also Fig. 9) indicates that the AuIB binding pocket partially overlaps with the ACh binding site. Residues 59 -61 are situated on the ␤2 strand of the ␤4 subunit and form the main functional part of loop D on the complementary ACh-binding site component. The tryptophan of loop D (TrpD; Trp-59 in ␤4 numbering) is highly conserved across different nAChR subunits and AChBPs and helps form the aromatic box/cage of hydrophobic residues essential for ACh binding and nAChR activation (45,46). The residue in position ϩ2 relative to TrpD (Lys-61 in ␤4 numbering) is also an important part of the ACh binding site (e.g. it corresponds with Gln-57 in the ␣7 subunit).
Studies involving receptor mutations and co-crystallization with AChBPs have shown that ␣-conotoxins bind in the extracellular part of nAChR subunit interface close to the ligand binding site (47,48). Usually ␣-conotoxins do not rely on a single critical residue/contact, and their binding determinants can be found on both principal and complementary sides. For instance, ImI has been proposed to bind to AChBP via a "network of interactions" (49). Similar findings were obtained in an earlier ImI-AChBP co-crystallization study (47) and using docking simulations for ImI, GIC, MII, PnIA, and GID (50 -52). A study measuring ImI binding to mutated ␣7 nAChRs reported a single dominant interaction between Arg-7 of ImI and Tyr-195 of ␣7 as well as multiple weak interactions between several other residues (53).
Using ␣3␤2 receptor chimeras, determinants for MII conotoxin specificity were found on ␣ and ␤ subunits (58). Interestingly, one of the determinants was mapped to loop D of the ␤2 subunit and identified as Thr-59 (homologous to ␤4-Lys-61). Therefore, the Lys to Thr substitution in the ␤2 subunit was one factor that defined ␤2/␤4 selectivity for MII. BuIA conotoxin has a different wash-off rate for ␣3␤2 than ␣3␤4 nAChRs. The same difference in the sequence (Lys/Thr-59) in ␤4 and ␤2 was also a determinant for the different wash-off rates for the various nAChR subtypes (59).
Consistent with the studies mentioned above, our modeling and experimental data show that the ␤4 single-point mutation [K61A]␤4 only reduces AuIB inhibition, but [W59A]␤4 (TrpD) completely abolishes it. Therefore, TrpD is essential for AuIB to inhibit ␣3␤4 nAChRs. TrpD involvement in nAChR pharmacology is well documented. Mutations of TrpD in AChBP have been shown to affect its conformation as well as alter its potency, desensitization, efficacy, and selectivity for various ligands (60 -64). However, its effects depend on the subunit and ligand in question and are more ambiguous than mutations of the principal subunit's aromatic residues. For instance, the effects of the TrpD mutation on ACh affinity for muscle nAChRs differed between ␦ and ␥ subunits (20,000 versus 7,000-fold reduction) (61). TrpD mutation caused opposite shifts in the potency of 4OH-GTS21 agonists in ␣7 and ␣4␤2 nAChRs (62).
When expressed in Xenopus oocytes and tested by two-voltage clamp recordings, the [W59A]␤4 mutation dramatically reduced the ACh-induced current without altering its kinetics. To induce measurable currents, we had to use the saturating ACh concentration of 2.5 mM instead of 50 M. In a previous study we showed that high ACh concentration has practically no effect on AuIB inhibition of ␣3␤4, as this inhibition is noncompetitive (14).
There are examples where the binding/action of pharmacological agents is critically dependent on interaction with TrpD. Apolipoprotein E inhibits ␣7 nAChRs via hydrophobic interactions with ␣7 TrpD (Trp55 in ␣7 numbering) (64). Varenicline, a smoking cessation drug, has been shown to interact with ␤2-TrpD in the ␣4␤2 nAChR. TrpD substitution to Ala converted varenicline from a partial to full agonist and abolished varenicline-induced desensitization at high concentrations (63).
Nature of AuIB-Phe-9⅐␤4-Trp-59 Interaction and Its Implications for AuIB Pharmacology-The sequence alignment of several nAChR subunits (Fig. 9) centered at the AuIB binding pocket shows that only TrpD (Trp-59 in ␤4 numbering) is absolutely conserved across the different subunits. However, substitutions in the WLK pocket are often homologous. For example, FIGURE 9. Protein sequence alignment of nAChR subunits. Sequence alignments showing the (Ϫ)(complementary) side of the main neuronal nAChR subunits centered around loop D and its conserved tryptophan residue TrpD, the main ␤4 subunit residue that interacts with Phe-9 of AuIB (␤4-Trp-59). Absolutely conserved residues in the sequences are emphasized in bold. Boxed residues are homologous to the WLK binding pocket on ␤4(Ϫ) subunit. Residue numbering according to the ␤4 subunit sequence is shown. Note the ubiquitous conservation of tryptophan.

JOURNAL OF BIOLOGICAL CHEMISTRY 34439
instead of ␤4-Lys at the TrpD ϩ2 position, ␣9 and ␣10 subunits have a similar positively charged Arg. There is a single different residue, Thr instead of Lys, between ␤2 and ␤4, and this can cause differences in pharmacology. It is unlikely that Lys/Thr substitution underlies AuIB selectivity for the ␣3␤4 subtype, as the [K61A]␤4 mutation only reduces inhibition, and AuIB is not active at the ␣9␣10 subtype when Lys is changed to an essentially homologous Arg. Interestingly, loop D of the ␣3 subunit has the same WLK pocket as the ␤4 subunit.
AuIB could also bind at the non-canonical binding site (ϩ)␤/ (Ϫ)␣, as suggested for galanthamine and cocaine binding to AChBP (65), anti-helminthic compound morantel binding to ␣3␤2 nAChRs (66), and recently for ␣-conotoxin Vc1.1 binding to ␣9␣10 nAChRs (67). However, this is unlikely because the docking model does not support binding at a non-canonical site and the [W59A]␤4 mutant loses inhibition despite identical WLK pockets being available on the ␣3 subunit (Ϫ) side. Another possibility could be that AuIB anchoring at the WLK pocket is controlled by its interactions with the ␣ subunit.
Our homology and MD-simulated model suggested that Tyr-5 of AuIB interacts with the ␣3 subunit Gln-198. However, [Q198A]␣3 mutation did not affect AuIB inhibition, and mutation of its putative partner, Tyr-5-AuIB, did not reduce inhibition on the wild-type receptor. Mounting evidence suggests that differences in three-dimensional structure across subunits probably underlies distinct bond patterns for the same ligand despite the presence of identical residues, as demonstrated for nicotinic ligands (68,69). This is certainly possible for larger molecules, such as AuIB.
There is also the question of the mechanism behind AuIB inhibition of nAChRs and how it can act non-competitively despite overlapping with the ACh-binding site. At least two larger peptides, apolipoprotein E and A␤ 1-42 amyloid, block nAChRs non-competitively as well. In addition, apolipoprotein E binds to loop D by interacting with TrpD (64,70). AuIB may be able to act non-competitively because it can bind to ␣3␤4 in the presence of ACh. Because TrpD may be involved in desensitizing or transducing the gating signal (54,63,71), AuIB may work by stabilizing the desensitized state or blocking the gating signal. Another possibility is steric hindrance to the movement of loop C, which is thought to be associated with agonist activation of nAChRs (47,72).
What could be the nature of the AuIB-Phe-9 interaction with TrpD? It is well established that residues around ligand binding sites and those that form the aromatic box in Cys-loop family receptors often make cationbonds with their cognate ligands (73). The reverse situation, when aromatic moieties of ligands engage in a cationbond formation with positively charged residues of the receptor, is also possible (74). AuIB does not have positively charged residues so cannot make a cationbond with ␤4-Trp-59 (TrpD). A situation in which a positively charged ␤4-Lys-61 forms a cation-bond with Phe-9 of AuIB is possible. However, our experiment with single-point mutations of the WLK pocket showed that removing Lys-61 does not abolish inhibition but that Trp-59 mutation does. This suggests that Trp-59 is indispensable for AuIB inhibitory effect and Lys-61 plays an auxiliary role in it.
Our experiments with second-generation AuIB mutants at position 9 demonstrated that size and hydrophobicity/aromaticity of the residue in this position are important for the peptide to inhibit ␣3␤4 nAChRs. Although simple hydrophobic interaction of AuIB-Phe-9 and ␤4-Trp-59 is possible, it is more likely that Phe-9 and Trp-59 interact viastacking due to the deep insertion of Phe-9 in the WLK pocket. Because removing positively charged Lys-61 reduces inhibition, it will likely interact with Phe-9 of AuIB and/or stabilize AuIB-Phe-9 interaction with ␤4-Trp-59. In conclusion, we identified determinants of AuIB binding/action on the ␣3␤4 nAChR.