Alanine Scan of α-Conotoxin RegIIA Reveals a Selective α3β4 Nicotinic Acetylcholine Receptor Antagonist*

Background: The molecular mechanism by which α-conotoxin RegIIA inhibits α3β4, α3β2, and α7 nAChRs is unknown. Results: Alanine scanning mutagenesis and molecular dynamic simulations of RegIIA revealed Asn11 and Asn12 confer improved selectivity at α3β4 nAChR. Conclusion: We synthesized the [N11A,N12A]RegIIA analog that selectively inhibits α3β4. Significance: These findings could be used to develop α3β4-selective drugs to treat lung cancer. Activation of the α3β4 nicotinic acetylcholine receptor (nAChR) subtype has recently been implicated in the pathophysiology of various conditions, including development and progression of lung cancer and in nicotine addiction. As selective α3β4 nAChR antagonists, α-conotoxins are valuable tools to evaluate the functional roles of this receptor subtype. We previously reported the discovery of a new α4/7-conotoxin, RegIIA. RegIIA was isolated from Conus regius and inhibits acetylcholine (ACh)-evoked currents mediated by α3β4, α3β2, and α7 nAChR subtypes. The current study used alanine scanning mutagenesis to understand the selectivity profile of RegIIA at the α3β4 nAChR subtype. [N11A] and [N12A] RegIIA analogs exhibited 3-fold more selectivity for the α3β4 than the α3β2 nAChR subtype. We also report synthesis of [N11A,N12A]RegIIA, a selective α3β4 nAChR antagonist (IC50 of 370 nm) that could potentially be used in the treatment of lung cancer and nicotine addiction. Molecular dynamics simulations of RegIIA and [N11A,N12A]RegIIA bound to α3β4 and α3β2 suggest that destabilization of toxin contacts with residues at the principal and complementary faces of α3β2 (α3-Tyr92, Ser149, Tyr189, Cys192, and Tyr196; β2-Trp57, Arg81, and Phe119) may form the molecular basis for the selectivity shift.

nAChRs have been implicated in the pathophysiology of a number of health conditions, including Alzheimer disease, schizophrenia, tobacco addiction, and lung cancer (5). Our understanding of isoform distribution and neurophysiological roles of individual receptor subtypes in these conditions is limited by a lack of adequate isoform-specific probes (6).
An initial report suggesting that nAChRs may regulate cancer cell growth (7) was followed by a number of studies investigating the role of nAChRs in cancer development and progression (reviewed in Ref. 8). Two studies identified tobacco-specific nitrosamines as potent nAChR agonists and inhibitors of cancer cell apoptosis (9). In addition, genome-wide studies identified associations between lung cancer and several single nucleotide polymorphisms within the gene cluster encoding the ␣3, ␣5, and ␤4 nAChR subunits (10). Additionally, activation of the ␣3␤4 nAChR, a predominant subtype expressed in sympathetic and parasympathetic neurons of mammalian autonomic ganglia (11)(12)(13), is known to be associated with nicotine addiction and drug abuse (14,15). However, our understanding of the physiological relationship is limited.
Conotoxins are bioactive peptides isolated from the venom of cone snails of the genus Conus (16). ␣-Conotoxins are a specific class of short, disulfide-constrained peptides with a conserved Cys-framework, CCX n CX m C. X n and X m represent the number of amino acids and are used to subclassify peptides such as ␣4/3, ␣4/4, and ␣4/7, which specifically target various nAChR isoforms (17). Thus they represent excellent molecular probes to elucidate the physiological roles of nAChR subtypes in normal and disease states (18). The structural and functional properties of a number of ␣-conotoxins have been characterized (16 -19). Although ␣-conotoxins such as ImII and RgIA exhibit selective inhibitory activity at ␣7 (20) and ␣9␣10 (21) nAChR subtypes, respectively, most known ␣-conotoxins target multiple nAChR subtypes (16,17). ␣-Conotoxin AuIB is the only known peptide that selectively targets the ␣3␤4 nAChR subtype albeit with low potency (IC 50 ϭ 2.5 M) (22,23).
Mutagenesis experiments have become an important tool to improve selectivity and potency of receptor inhibitors (see for example, Ref. 24). Previously we reported the discovery and isolation from Conus regius venom of the ␣4/7-conotoxin RegIIA (25). RegIIA potently inhibits ACh-evoked currents of ␣3␤4, ␣3␤2, and ␣7 nAChR isoforms. Given the pathophysiological association of the ␣3␤4 nAChR subtype with various disorders such as lung cancer and nicotine addiction, we have now employed mutagenesis with the aim of improving the selectivity profile of RegIIA. Using alanine scanning mutagenesis and modeling studies, we also identified critical ␣-conotoxin RegIIA residues that interact with ␣3␤2, ␣3␤4, and ␣7 nAChR ACh-binding sites.

EXPERIMENTAL PROCEDURES
Peptide Synthesis-All of the peptide analogs were assembled on rink amide methylbenzhydrylamine resin (Novabiochem; 0.7 mmol g Ϫ1 ) using o-benzotriazole-N,N,NЈ,NЈ-tetramethyluronium hexafluorophosphate (HBTU)-mediated manual solidphase peptide synthesis, with an in situ neutralization procedure for N-(9-fluorenyl)methoxycarbonyl (Fmoc) chemistry. Each cycle consisted of Fmoc deprotection with 20% piperidine in dimethylformamide, followed by Fmoc amino acid coupling using HBTU and N,N-diisopropylethylamine in dimethylformamide. A 2-fold excess of Fmoc amino acids was used in the coupling reactions. All peptides were synthesized in globular conformation (I-III and II-IV disulfide connectivity) through incorporation of Fmoc-Cys acetamidomethyl (Acm)-OH at positions 2 and 8 of the amino acid sequence (I-III disulfide bond). The efficiency of the coupling reactions were checked using the Kaiser Ninhydrin test.
Peptides were cleaved from the dried resin (0.4 g) by treatment with 100 ml of TFA, triisopropylsilane, and water as scavengers (95:2.5:2.5, TFA:triisopropylsilane:water, v/v/v). The reaction was allowed to proceed at room temperature (20 -23°C) for 2.5 h. The TFA was then evaporated, and the peptide was precipitated with ice-cold ether, filtered, dissolved in 50% buffer A/B (buffer A: H 2 O, 0.05% TFA; buffer B: 90% CH 3 CN, 10% H 2 O, 0.045% TFA) and lyophilized. Crude peptides were purified by reversed phase-high performance liquid chromatography (RP-HPLC) on a Phenomenex C18 column using a gradient of 0 -80% buffer B for 80 min and the elutant was monitored at 215/280 nm. Unless otherwise stated, the same conditions were used in subsequent purification steps. Electrospray-mass spectroscopy (ESI-MS) confirmed the molecular mass of the fractions collected.
Fractions displaying the correct molecular mass for linear peptide were pooled and lyophilized for oxidation. Linear peptides were oxidized in two steps. First, they were dissolved in 0.1 M NH 4 HCO 3 (pH 8.2) at a concentration of 0.3 mg/ml. Stirring overnight at room temperature formed the II-IV disulfide bond. The acetamidomethyl (Acm) protecting groups were stable under these conditions. Second, iodine (0.01-0.1 M) was added for 5 min at 37°C, and excess iodine was destroyed by adding sodium ascorbate. This formed the I-III disulfide bond. The oxidized peptides were purified by RP-HPLC using a gradient of 0 -80% buffer B over 160 min. Analytical RP-HPLC and ESI-MS confirmed the purity and molecular mass of the synthesized peptides.
NMR Spectroscopy-Nuclear magnetic resonance (NMR) data for all peptides were recorded on Bruker Avance 500-and 600-MHz spectrometers, with samples dissolved in 90% H 2 O, 10% D 2 O. Two-dimensional NMR experiments included TOCSY (total correlation spectroscopy) and NOESY (nuclear Overhauser effect spectroscopy) recorded at 280 K. Spectra were analyzed using Topspin 1.3 (Bruker) and Sparky software. Unless specified, spectra were recorded at pH 3.5.
Two-electrode Voltage Clamp Electrophysiological Recordings of nAChRs Expressed in Xenopus Oocytes-Stage V-VI oocytes were harvested from mature female Xenopus laevis anesthetized with 0.1% tricaine, following protocols approved by the RMIT Animal Ethics Committee. RNA and Xenopus oocytes were prepared, then nAChR subtypes were expressed in oocytes as described previously (26). Briefly, cDNAs encoding the rat ␣3, ␤2, and ␤4 subunits, and human ␣7 subunit were subcloned into oocyte expression vector pT7TS and the ␣6/␣3 chimera plasmid was kindly provided by Dr. Michael McIntosh (University of Utah). mRNA corresponding to each subunit was prepared using the mMESSAGE mMACHINE Kit (Ambion, Invitrogen). Oocytes were injected with 5 ng of cRNA for each subtype in a 1:1 ratio. However, for the expression of ␣6/␣3 receptors, 15-20 ng of cRNA for each subunit was injected. The injected oocytes were then incubated for 2-5 days at 18°C in ND96 buffer (96 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM HEPES, pH 7.4) supplemented with 50 mg/liter of gentamicin and 100 g/units/ml of penicillin-streptomycin, before recording. Membrane currents from Xenopus oocytes were recorded at room temperature (20 -23°C), using a bath solution of ND96 as described above. A two-electrode voltage clamp (virtual ground circuit) with either a GeneClamp 500B amplifier (Molecular Devices, Sunnyvale, CA) or an automated work station with eight channels in parallel, including drug delivery and online analysis (OpusXpress TM 6000A, Axon Instruments Inc.) was used.
All recordings were made using voltage recording and current-injecting electrodes were pulled from borosilicate glass (GC150T-7.5, Harvard Apparatus Ltd., Holliston, MA) and had resistances of 0.3-1.5 M⍀ when filled with 3 M KCl. Oocytes were voltage clamped at a holding potential of Ϫ80 mV. During recordings, oocytes were perfused continuously at a rate of 2 ml/min, with ACh (200 M for ␣7 and 50 M for all other nAChR subtypes) applied for 2 s at 2 ml/min. A 180-to 240-s interval washout period was used between every ACh application. The inhibitory effect of each peptide at the respective concentration represents the ratio of ACh-evoked peak amplitude evoked before and following 300-s incubation with peptide. Data were filtered at 10 Hz and sampled at 500 Hz (27).
Data Analysis-Concentration-response curves for antagonists were fitted by unweighted nonlinear regression to the logistic equation, where E x is the response, X is the antagonist concentration, E max is the maximal response, n H is the slope factor, and IC 50 is Structure-Activity Relationship of ␣-Conotoxin RegIIA the antagonist concentration giving 50% inhibition of the maximal response. All electrophysiological data were pooled (n ϭ 4 -8 for each data point) and represent arithmetic mean Ϯ S.E. of the fit. Computation was done using GraphPad Prism 6.03 (GraphPad Software, Inc., La Jolla, CA).
Before undergoing MD simulations, the complexes were energy minimized using the steepest descent algorithm and an energy gradient convergence criterion of 0.01 kcal/mol/Å. All MD simulations were performed using a constant particle number, pressure, and temperature ensemble, with temperature maintained at 300 K using the v-rescale temperature coupling algorithm (34), and pressure maintained at 1 bar using the Parrinello-Rahman pressure coupling algorithm (35). Time steps of 2 fs were used to integrate all simulations.
Solvent equilibration simulations of 100 ps lengths were performed. In these simulations, the non-hydrogen atoms of the receptor and toxins were positionally restrained so the solvent and ions could undergo motions to reach equilibrium from an initially energetically unfavorable state, without disturbing the protein. Subsequent simulations of all four complexes were then performed with all atoms free of the system to undergo dynamics.
To improve conformational sampling, we performed 10 independent simulations for each complex, using different random seeds to assign initial particle velocities. Each simulation was performed for 20 ns (i.e. 200 ns of trajectory per complex). To reduce bias from initial homology model conformations, all analyses were performed on the final 10 ns of the trajectories.
Unless indicated, all data were taken as an average over the 10 independent simulations. Molecular graphics were produced using VMD version 1.9.2 (36). All analyses were performed using a combination of VMD, GROMACS analysis software suite, and in-house scripts.
The interatomic contact difference plot (⌬N) was calculated by determining the total number of toxin contacts within 4.5 Å of each receptor residue for wild-type and mutant RegIIA, averaged over 10 independent simulations and both ␣(ϩ)␤(Ϫ) interfaces (20 data points per receptor residue). To obtain ⌬N values, the wild-type contact number was subtracted from that of the double mutant.

RESULTS
Synthesis of RegIIA Analogs-We used an alanine scan mutagenesis approach to elucidate the molecular mechanism underlying inhibition of the ␣3␤4 nAChR subtype by RegIIA. This technique is well established, and in conjunction with atomistic MD simulations, has enabled significant advances in molecular pharmacology (see for example Refs. 37 and 38). ␣-Conotoxin RegIIA belongs to the ␣4/7 subclass having the conserved Cys-framework CCX n CX m C. Due to its I-III and II-IV disulfide connectivity, native RegIIA exhibits a classical helical, globular structure (25). This globular conformation balances the shape, charge, and polarity of the peptide. As ␣-conotoxins contain four cysteines there are two additional disulfide isomers that can form during oxidative folding or disulfide reshuffling, the ribbon (I-IV and II-III disulfide bonds) or bead (I-II and III-IV disulfide bonds) isomers. Changes in the disulfide connectivity are reflected in the peptide conformation and can have a significant impact on ␣-conotoxin potency and specificity at nAChRs (39,40).
We used regioselective disulfide bond formation with Acmprotected cysteine residues incorporated at positions 1 and 3, and a two-step oxidation procedure, to produce the alanine mutant peptides in a globular conformation (I-III and II-IV disulfide bonds). The two-step oxidation process was confirmed by RP-HPLC, ESI-MS (Fig. 1), and two-dimensional NMR. Fig. 2 show the negative 1 H shift values between amino acid positions 3 and 7 indicating the presence of an ␣-helix secondary structure.
Because the interactions between nAChR and RegIIA-Asn 11 / Asn 12 are similar for both ␣3␤2 and ␣3␤4, it is difficult to rationalize the substantial selectivity change after double mutation by employing "static" homology models alone. We therefore employed MD simulations, which take into account solvent, temperature, and protein dynamics, to compare changes in atomic contacts at the wild-type and mutated receptors. In particular, we examined the effects of the [N11A,N12A] mutation on contacts at sites distant from these positions (including the ␤(Ϫ) face, which differs substantially between ␤2 and ␤4), which may help explain the basis of RegIIA selectivity. Fig. 7 shows the change in number of toxin atoms (y axis) residing within 4.5 Å of the molecular surface of each receptor residue (x axis) at the ␣(ϩ)␤(Ϫ) interfaces ("⌬N profiles"). This is a measure of loss (negative values on the y axis) or gain (positive values) of toxin contacts as a result of the [N11A,N12A] double mutation. The ⌬N profiles for ␣3␤2 and ␣3␤4 reveal that the double mutation generally reduces toxin-receptor contacts at ␣(ϩ) face residues. However, the effect at the ␤(Ϫ) face is mixed, with reduced or increased number of contacts for  different residues. As expected, mutation of Asn 11 and Asn 12 leads to decreased contact with receptor residues in the immediate proximity of these (RegIIA) positions. Contact between the Asn 11 position with ␣3-Tyr 196 /␤2-Arg 81 (␤4-Arg 79 ), and the Asn 12 position with ␣3-Cys 192 , are reduced compared with that of wild-type RegIIA (Fig. 6B).
The [N11A,N12A] mutation also markedly reduces toxinreceptor contacts at regions distant from the mutation sites. Of particular interest and to explain the selectivity of the mutant for ␣3␤4, are receptor residues with substantially fewer toxin contacts for ␣3␤2 than ␣3␤4 nAChR subtypes. Some of the pairwise interactions involving these residues are illustrated in Fig. 6C and are also marked with asterisks in Fig. 7. The most prominent residues include ␣3-Tyr 92 , Ser 149 , Tyr 189 , Cys 192 , and Tyr 196 at the principal face, and ␤2-Trp 57 (␤4-Trp 55 ) and ␤2-Phe 119 (␤4-Gln 117 ) at the complementary face. These residues may be essential for RegIIA inhibition of ␣3␤2. In particular, the marked loss of contact at ␣3-Cys 192 in both ␣3␤2 and ␣3␤4 nAChR subtypes is consistent with similar observations from previous studies of ␣-conotoxin analogs and modifications. van Lierop et al. (41) showed dicarba modification of the C2-C8 disulfide bond in Vc1.1 resulted in loss of activity at ␣9␣10. Their MD simulations showed reduced contact between the modified toxin and Cys-loop disulfide atoms. Grishin et al. (38) found [F9A]AuIB lost its activity at ␣3␤4 with the MD simulations showing reduced contact between the toxin and Cys-loop sulfur atoms due to this mutation. Our current data supports the crucial role of Cys-loop sulfur atoms in conotoxin inhibition of neuronal nAChRs.
Double mutation also caused a loss of contact between Asn 9 / Pro 6 and ␤2-Trp 57 , whereas slightly increasing the contact between Asn 9 /Pro 6 and the homologous position at ␤4-Trp 59 (Fig. 7). This, in addition to functional data for [N9A]RegIIA showing complete loss in activity at ␣3␤4 (Fig. 3), suggests interaction between Asn 9 and/␤4-Trp 59 might be important for inhibition of the ␣3␤4 subtype. This loss of toxin contact at ␣3␤2, but not ␣3␤4, may also contribute to the marked selectivity change of the double mutant [N11A,N12A]RegIIA toward the ␣3␤4 nAChR subtype.

DISCUSSION
Since the discovery, in the worm-hunting cone snail Conus imperialis, of ␣-conotoxin ImI and its action on neuronal nAChRs, numerous ␣-conotoxins have been identified and functionally characterized (16 -19). A number of peptides have been identified from the venom of C. regius, a Western Atlantic worm-hunting cone snail species. These peptides belong to various superfamilies with eight being part of the A-superfamily that is distinguished by cysteine framework I (42). The conotoxin composition of C. regius venom is clinically relevant because it contains the ␣-conotoxin RegIe and RegIIA both of which have been identified as potential therapeutics. RgIA (free  carboxyl C terminus form of [O6P]RegIe) selectively targets pain transmission by modulation of the ␣9␣10 nAChR subtype and high voltage-activated N-type calcium channel currents, via GABA B receptor activation (43). ␣-Conotoxin RegIIA potently inhibits activity of the ␣3␤4 nAChR subtype (25) that has been implicated in the pathophysiology of lung cancer.
RegIIA exhibits homology with a number of peptides (Table  2) having a conserved SHPA sequence in loop 1 and a NNP motif in loop 2. The Ser and Pro residues of the SHPA motif are highly conserved in peptides of different subclasses that inhibit various nAChR subtypes ( Table 2). Sequence variations in loop 2 contribute to the unique nAChR subtype selectivity of ␣-conotoxins (44). However, the NNP motif is conserved in peptides such as ␣-conotoxins OmIA, EpI, PnIA, TxIA, and ArIB that specifically inhibit ␣3␤2 and ␣7 nAChR subtypes (16). This observation is consistent with our finding that alanine mutation of the NNP motif of RegIIA significantly affected inhibition at ␣3␤2 and ␣7 nAChR subtypes (Fig. 3A). Alanine mutations at other positions provided further information about the structure-function relationship between ␣-conotoxins and neuronal nAChRs. Asparagine (RegIIA position nine) to alanine mutation completely abolished inhibition of the ␣7 and ␣3␤4 nAChRs.
The marked shifts in selectivity observed by the analogs for either ␣3␤2 or ␣3␤4 subtype is intriguing given that both receptors share a common principal subunit face at the presumed conotoxin binding site (Fig. 6A). Previous x-ray studies of ␣-conotoxin-acetylcholine-binding protein complexes, with support from synthetic analogs, have improved our understanding of ligand-receptor interactions (45)(46)(47). The extracellular N-terminal domains of the ␤2 and ␤4 nAChR subunits bind ␣-conotoxins and exhibit 70% sequence homology. Recent MD simulation studies also reveal well preserved structural topology of ␣3␤2 and ␣3␤4 nAChRs. However, the ACh-binding pocket interface between the ␣ and ␤ subunits was larger in ␣3␤2 than ␣3␤4 nAChR subtypes (48). Even though, both Asn 11 and Asn 12 primarily interact with the ␣3(ϩ) interface, this difference could explain the shift in selectivity of [N11A] and [N12A]RegIIA for ␣3␤4 over ␣3␤2.
We report the successful synthesis of [N11A,N12A]RegIIA, an ␣3␤4 nAChR subtype-selective antagonist. [N11A,N12A]-RegIIA is ϳ4-fold more potent than AuIB and could be used to decipher the physiological role of ␣3␤4 nAChR in pathological states. MD simulations reveal a direct loss in pairwise contacts between positions 11 and 12 and the principal face of the receptor, including destabilization in other toxin-receptor contacts at the complementary face. This is qualitatively consistent with the 1000-fold decrease in [N11A,N12A]RegIIA inhibition of the ␣3␤2 nAChR subtype. Our homology models and MD simulations also suggest that at the ␤2 subunit, Asn 11 is close to two basic residues (␤2-Arg 81 and Lys 79 ); whereas at the ␤4 subunit, Asn 11 is close to one basic (␤4-Arg 79 ) and one non-polar residue (␤4-Ile 77 ). This has implications for rational toxin modification. Based on the receptor environment surrounding Asn 11 , it is possible that a N11K mutant would enhance selectivity for ␣3␤4, because a Lys at position 11 would introduce electrostatic repulsion with ␤2-Arg 81 and ␤2-Lys 79 . There would likely also be a reduction in affinity at ␣3␤4. However, the repulsion between N11K and ␤4-Arg 79 might be partially offset by favorable contacts between the CH 2 groups of N11K and ␤4-Ile 77 . Further efforts to examine and optimize the selectivity of RegIIA are presently ongoing.
Our study increases understanding of the interactions of RegIIA with various nAChR subtypes. It also identifies key residues such as Asn 9 , Asn 11 , and Asn 12 involved in toxin-receptor interaction. This information will be valuable in the design and development of potent, ␣3␤4-selective drugs to treat lung cancer and nicotine addiction.