Scanning Mutagenesis of α-Conotoxin Vc1.1 Reveals Residues Crucial for Activity at the α9α10 Nicotinic Acetylcholine Receptor*

Vc1.1 is a disulfide-rich peptide inhibitor of nicotinic acetylcholine receptors that has stimulated considerable interest in these receptors as potential therapeutic targets for the treatment of neuropathic pain. Here we present an extensive series of mutational studies in which all residues except the conserved cysteines were mutated separately to Ala, Asp, or Lys. The effect on acetylcholine (ACh)-evoked membrane currents at the α9α10 nicotinic acetylcholine receptor (nAChR), which has been implicated as a target in the alleviation of neuropathic pain, was then observed. The analogs were characterized by NMR spectroscopy to determine the effects of mutations on structure. The structural fold was found to be preserved in all peptides except where Pro was substituted. Electrophysiological studies showed that the key residues for functional activity are Asp5–Arg7 and Asp11–Ile15, because changes at these positions resulted in the loss of activity at the α9α10 nAChR. Interestingly, the S4K and N9A analogs were more potent than Vc1.1 itself. A second generation of mutants was synthesized, namely N9G, N9I, N9L, S4R, and S4K+N9A, all of which were more potent than Vc1.1 at both the rat α9α10 and the human α9/rat α10 hybrid receptor, providing a mechanistic insight into the key residues involved in eliciting the biological function of Vc1.1. The most potent analogs were also tested at the α3β2, α3β4, and α7 nAChR subtypes to determine their selectivity. All mutants tested were most selective for the α9α10 nAChR. These findings provide valuable insight into the interaction of Vc1.1 with the α9α10 nAChR subtype and will help in the further development of analogs of Vc1.1 as analgesic drugs.

Marine snails belonging to the Conus genus produce a variety of neurotoxic peptides in their venom glands that they use for the capture of prey (1)(2)(3). Within this repertoire of conopeptides, those that are disulfide-rich are referred to as conotoxins. Conotoxins typically range in size from 12 to 30 amino acids, contain 4 or more Cys residues, and exhibit high potency and selectivity toward a variety of membrane receptors and ion channels (4,5). The ␣-conotoxin subfamily members typically range in size from 12 to 19 amino acids, contain 2 disulfide bonds in a Cys I -Cys III and Cys II -Cys IV connectivity, and have an amidated C terminus, as depicted in Fig. 1. They interact with nicotinic acetylcholine receptors (nAChRs), 4 of both the muscle and the neuronal type, which have been implicated in a range of neurological disorders varying from Alzheimer disease to addiction (6 -8).
The nAChRs are ligand-gated ion channels that respond to ACh, nicotine, and other competitive agonists/antagonists. They are composed of five subunits, with differing nAChR subunit composition according to the site of expression. The muscle-type nAChRs are composed of two ␣ subunits, a ␤ and ␦ subunit, and either an ⑀ or a ␥ subunit (9 -12). The neuronal forms exist either as homomeric channels composed of ␣ subunits alone or ␣␤ heteromeric channels. The wide variety of possible subunit combinations has led to unique subtypes with distinct pharmacological properties. This makes ␣-conotoxins valuable neuropharmacological tools and drug leads, because they have the ability to distinguish between different nAChR subtypes. Effectively, they are small rigid scaffolds that display amino acids on their surface to selectively target their receptors (13).
Of particular interest in this study is the ␣-conotoxin Vc1.1, a synthetic derivative of a naturally occurring peptide from the venom of the marine cone snail, Conus victoriae. It was discovered using PCR screening of cDNA extracted from the snail venom duct (14). Fig. 1 depicts the sequences of selected ␣-conotoxins, including Vc1.1, which is 16 amino acids in length and displays the classic disulfide bond connectivity observed for ␣-conotoxins, together with a short helical segment as depicted in Fig. 1b. The conserved Cys framework of ␣-conotoxins defines two backbone loops, which vary in size and residue composition, and are classified by an n/m nomenclature to define subclasses of ␣-conotoxins. For example, Vc1.1 is a 4/7 subclass ␣-conotoxin, because it contains four residues in loop 1 and seven in loop 2. RgIA (15)(16)(17) is another conotoxin of interest in this study, because it is also selective for the ␣9␣10 nAChR subtype, and has a 4/3 framework. Vc1.1 contains an amidated C terminus, a post-translational modification common to most ␣-conotoxins, but it is not present in RgIA. Vc1.1 lacks the post-translationally modified hydroxyproline and ␥-carboxyglutamate residues present in the native peptide, vc1a, isolated from the venom duct of C. victoriae (18).
Vc1.1 has been under development as a drug lead for neuropathic pain (19). When tested in rat models of neuropathic pain, Vc1.1 induced analgesia when injected intramuscularly near the site of injury (20). Initially, it was thought that ␣3-containing subtypes of nAChRs may be the target for Vc1.1 (21); however, it was then reported that Vc1.1 has a 100-fold higher affinity at the ␣9␣10 nAChR subtype (22,23). The ␣9␣10 nAChR mediates synaptic transmission between efferent olivocochlear fibers and cochlear hair cells (24 -26). The mRNA of these receptor subtypes is expressed in many different tissue types from the inner ear, dorsal root ganglion (27), skin keratinocytes (28), and lymphocytes (29) to the pituitary (26). The ␣10 subunit has to be expressed with the ␣9 subunit to form a functional receptor. In the auditory system, the ␣9␣10 nAChR plays an important role in hair cell development, but its role in other tissues is yet to be characterized (22,26,30,31).
Owing to the promising antinociceptive effects of Vc1.1 in animals, its analogs are of interest as leads for the treatment of neuropathic pain (14,20). To date, studies have predominantly focused on the ␣9␣10 nAChR, but the very recent finding that Vc1.1 also targets the ␥-aminobutyric acid, type B receptor (32) has raised interest in the molecular mode of action of Vc1.1 in analgesia. Hence there is a need to define structure-activity relationships of this peptide at several targets, including human and rat forms of the ␣9␣10 nAChR. In particular, we were interested in analogs that maintain potency at the rat ␣9␣10 nAChR but also show significant improvement in potency at human forms of the receptor, while maintaining selectivity over other nAChR subtypes.
In this study we determined such structure-activity relationships for Vc1.1 at the ␣9␣10 nAChR by successively mutating each non-Cys residue of Vc1.1 to either an "inert" residue (Ala), a negatively charged residue (Asp), or a positively charged residue (Lys) and observing the impact on the structure and functional activity of Vc1.1. Once the key residues had been identified, a second generation of analogs with new substitutions was synthesized and tested at the rat ␣9␣10 nAChR. The analogs were also analyzed at the human ␣9/rat ␣10 (h␣9r␣10) hybrid clone, because a recent report 5 suggested differences in the activity of Vc1.1 at the human and rat clones of the ␣9␣10 nAChR. We also examined the effect of pH change on the structure of Vc1.1 using NMR ␣H chemical shift analysis. The results from this study provide valuable insight into the key residues involved in the interaction of Vc1.1 with the ␣9␣10 nAChR subtype and have the potential to assist in the development of conotoxin analogs as drug leads for the treatment of neuropathic pain (4,33).

EXPERIMENTAL PROCEDURES
Synthesis and Cleavage of Mutants-All of the peptide mutants were assembled on rink amide methylbenzhydrylamine resin (Novabiochem) using manual solid-phase peptide synthesis with an in situ neutralization/2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate activation procedure for Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry. Cleavage of peptides from the resin was achieved by treatment with trifluoroacetic acid and triisopropylsilane and water as scavengers (9:0.5:0.5 trifluoroacetic acid: triisopropylsilane:water). The reaction was allowed to proceed at room temperature (20 -23°C) for 2.5 h. The trifluoroacetic acid 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% trifluoroacetic acid; Buffer B: 90% CH 3 CN/ 10%H 2 O/0.045% trifluoroacetic acid), and lyophilized. Crude peptides were purified by reversed phase-HPLC on a Phenomenex C18 column using a gradient of 0 -80% B in 80 min, with the eluant monitored at 215/280 nm. These conditions were used in subsequent purification steps unless stated otherwise. Electrospray-mass spectroscopy confirmed the molecular mass of the fractions collected, and those displaying the correct FIGURE 1. ␣-Conotoxin sequences and structure of Vc1.1. a, the sequences of selected ␣-conotoxins relevant to this study are shown by one-letter amino acid codes. The asterisk indicates an amidated C terminus, which is a common post-translational modification found in ␣-conotoxins. The conserved cysteine residues are highlighted in yellow, and the Cys I -Cys III and Cys II -Cys IV disulfide connectivity is indicated by the connecting lines under the sequence. The number of residues between the cysteines define two backbone "loops," which are used to classify ␣-conotoxins into subclasses. For example, RgIA has four residues in loop 1 and three residues in loop 2, making this a 4/3 loop subclass ␣-conotoxin. b, structural representation of Vc1.1 (PDB 2H8S), with disulfide bonds depicted in yellow. The cysteines, the loops, and the termini are labeled. molecular mass of linear peptide were pooled and lyophilized for oxidation. The linear peptides were oxidized by being dissolved in 0.1 M NH 4 HCO 3 (pH 8.2) at a concentration of 0.3 mg/ml with stirring overnight at room temperature. The oxidized peptides were then purified by reversed phase-HPLC using a gradient of 0 -80% buffer B over 160 min. Analytical reversed phase-HPLC and electrospray-mass spectroscopy confirmed the purity and molecular mass of the synthesized peptides.
NMR Spectroscopy-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 and NOESY recorded at 280 K. Spectra were analyzed using Topspin 1.3 (Bruker) and Sparky software. Most spectra were recorded at pH 3.5, although a pH titration over the range 3-7 was done for Vc1.1. This involved adjusting the pH with NaOH and recording a series of TOCSY spectra. Most of the ␣H and amide NMR chemical shifts were unaffected by a change in pH from pH 3.03 to 5.77, suggesting that there are no pH-dependent conformational changes. Protonation effects were found for residues near His 12 . Although TOCSY spectra were collected up to pH 7, the spectra beyond pH 5.77 could not be fully assigned as they deteriorated in quality due to rapid amide exchange. Nevertheless, analysis of the data showed that the changes in ␣H and amide shifts reached a plateau near pH 4.41 and are likely to be unchanged at pH values higher than 5.77.
Electrophysiological Recordings from nAChRs Expressed in Xenopus Oocytes-RNA preparation, oocyte preparation, and expression of rat ␣9␣10, human ␣9/rat ␣10 hybrid, rat ␣3␤2, rat ␣3␤4, and rat ␣7 nAChR subunits in Xenopus oocytes were performed as described previously (21,34). We used the human ␣9/rat ␣10 hybrid as a surrogate for the human receptor because in our hands the human ␣9␣10 nAChR expressed poorly. Azam and McIntosh 5 earlier showed that the human ␣9/rat ␣10 hybrid receptor displayed the same sensitivity to the ␣-conotoxin RgIA, as human ␣9␣10 and the sequence difference resides in the N-terminal binding region of ␣9 subunits. All oocytes were injected with 5 ng of RNA and kept at 18°C in ND96 buffer (96 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM HEPES, at pH 7.4) supplemented with 50 mg/liter gentamicin and 5 mM pyruvic acid 2-5 days before recording. Membrane currents were recorded from Xenopus oocytes using a GeneClamp 500B amplifier (Molecular Devices), and an automated workstation with eight channels in parallel, including drug delivery and on-line analysis (OpusXpress TM 6000A workstation, Molecular Devices, Sunnyvale, CA). Both the voltage-recording and the current-injecting electrodes were pulled from borosilicate glass (Harvard Apparatus Ltd., Edenbridge, UK) and had resistances of 0.3-1.5 M⍀ when filled with 3 M KCl. All recordings were conducted at room temperature using a bath solution of ND96, as described above. During recordings, the oocytes were perfused continuously at a rate of 1.5 ml/min, with 300-s incubation times for the conotoxin. Acetylcholine (10 M unless otherwise specified) was applied for 2 s at 5 ml/min, with 180 -300 s washout periods between applications. Vc1.1 and mutants were bath-applied and co-applied with the agonist. Oocytes were voltage clamped at a holding potential of Ϫ80 mV. Peak current amplitude was measured before and after incubation of the peptide (23).

RESULTS
Peptide Synthesis and Oxidative Folding-Alanine scanning mutagenesis is a powerful technique for identifying residues important for the structure and activity of peptides. Here we combined it with Lys and Asp scanning to probe the role of neutral, positive, and negative charge substitutions at all non-Cys positions of Vc1.1. The Cys positions were not mutated because they are essential for the disulfide bonds that define the native structure. The suite of mutant peptides was assembled using solid-phase peptide synthesis, oxidized in ammonium bicarbonate buffer, purified using HPLC, and characterized by mass spectrometry and NMR. A sufficient quantity (Ͼ1 mg) of natively folded peptide was obtained for each of the 34 single point mutants for structural and electrophysiological studies. The native Cys I -Cys III and Cys II -Cys IV disulfide connectivity is referred to as the globular isomer (35) and was the predominant form observed after oxidation of most of the peptides. For some peptides a Cys I -Cys IV and Cys II -Cys III disulfide connectivity, referred to as the ribbon isomer, was also formed. In these cases the isomers were separated by HPLC, and the globular form was used for further analysis. The sequences, molecular masses, disulfide isomers, and yields of the synthesized peptides are summarized in Table 1. The overall yield of globular folded peptide ranged from 30 to 70%, based on the quantity of crude peptide obtained after cleavage from the resin.
Although our primary focus was on the globular isomer, it is of interest to note that in some cases the mutations caused changes in the relative distribution of globular versus ribbon isomer. For example, the replacement of Arg 7 with an Ala or a Lys led to exclusive formation of the globular isomer, whereas substituting it with an Asp led to a small proportion of the ribbon isomer. In general, replacement of Pro residues led to a lower proportion of globular isomer, as did substitutions of Asp 5 or His 12 . A sample analytical HPLC trace for mutant I15K is shown in Fig. 2. This mutant formed both the globular and the ribbon isomers, which were separated by preparative HPLC.
NMR Analysis-NMR spectroscopy was used to confirm the native globular fold of the purified peptides. This analysis required assignment of the spectra, and for this task ␣H i Ϫ NH iϩ1 NOE connectivities observed in NOESY spectra were used in the sequential assignment of the individual spin systems determined from TOCSY spectra. For most of the mutants, sequential ␣H i Ϫ NH iϩ1 NOEs were observed for the entire peptide chain except, as expected, at Pro 6 and Pro 13 because these residues lack amide protons.
In general, ␣H NMR chemical shifts can be used to indicate the nature of secondary structural elements present in peptides. Specifically, secondary ␣H chemical shifts represent the difference between an observed ␣H chemical shift and that for the corresponding residue in a random coil peptide and are strong indicators of the presence of secondary structure (36). Analysis of the measured chemical shift data indicated that, for most of the mutants, residues Cys 2 -Ser 4 , Pro 6 -Asp 11 , and Pro 13 -Cys 16 have negative ␣H secondary shifts (Fig. 3). As reported previ-ously (21), the negative ␣H secondary shifts between Pro 6 -Asp 11 indicate the conservation of the ␣-helix as seen in the parent peptide. This predominance of helical secondary struc-ture is characteristic of other ␣-conotoxin structures reported to date (7). The trends in shifts for most of the mutants are almost identical to those for Vc1.1, illustrating the high degree of structural similarity between the mutants. In a few cases the change in residue type causes a structural change. This is particularly so with the mutations of the Pro residues. The P6K mutant folded to yield two isomers, both of which displayed similar secondary ␣H shifts. Fig. 3 shows the shifts of what is believed to be the globular isomer and clearly illustrates how they are different from the shifts of the other mutants, suggesting a change in structure. Likewise, P13K, P13D, and P13A have significant changes in ␣H secondary shifts of loop 2 relative to Vc1.1, suggesting a perturbation in the core of the structure. Small localized differences were seen in the secondary ␣H shifts for N9I, N9G, S4R, and S4KϩN9A, as shown in the lower panel of Fig. 3.
Aside from the exceptions noted above, most mutations did not lead to perturbations of chemical shifts. ␣H shifts are sensitive to the backbone conformation and, with most shifts superimposing well, any differences in the biological activities between the mutant peptides can therefore be attributed to changes in the nature of the side-chain interactions with the receptor rather than structural perturbations of the conotoxin ligands.
pH Titration Experiment-It has been reported that Vc1.1 has increased potency at lower pH values, presumably as a result of His 12 protonation (21). Because His 12 is unlikely to be protonated at physiological pH, we conducted a titration to determine the effect of pH on the structure of Vc1.1. No global change in the structure was apparent with changes in pH, as judged from the lack of change in the chemical shifts for most residues. There were some localized changes in the ␣H and amide shifts of Tyr 10 , Asp 11 , and His 12 with changes in pH from 5.2 to 3.0, with His 12 ␣H shifts being the most affected; however, these reflect local charge-state effects rather than a change in the overall fold of the peptide. Fig. 4 shows the backbone and side-chain orientation of Asp 11 and His 12 in a structure determined at pH 3.5, and highlights the proximity of His 12 to Asp 11 . The largest observed shift with pH was for the amide of Glu 14 . Fig. 4 shows that the backbone amide proton of Glu 14 and its carboxyl side chain form a hydrogen bond. Because Glu 14 is a surface-exposed residue, it is not surprising that it is susceptible to changes in pH, but again the effect is only local. Overall, we saw no change in the structure of Vc1.1 at the pH at which the NMR data were collected and at the pH used for the electrophysiological experiments to functionally characterize the mutants.
Potency of Vc1.1 and Analogs at the Rat ␣9␣10 nAChR-The functional activity of Vc1.1 and analogs was investigated via its effects on ACh-evoked membrane currents in Xenopus oocytes. The EC 50 for activation of rat ␣9␣10 nAChR by ACh was determined to be ϳ10 M, similar to that reported previously (37). The IC 50 for inhibition of ACh-evoked current by Vc1.1 was 109 nM. All Vc1.1 analogs were initially screened at a single concentration of 100 nM upon activation of ␣9␣10 nAChR with 10 M ACh. Representative traces of ACh-evoked currents inhibited by an analog more potent, less potent, and equally potent to Vc1.1 are shown in Fig. 5a. The pooled data for all of  the mutants at 100 nM are summarized in Table 2. These results were statistically analyzed using a two-sample t test at p ϭ 0.05 (assuming equal variances). From the single dose screen, it is clear that any mutation to residues Asp 5 -Arg 7 and Asp 11 -Ile 15 significantly reduced the inhibition of ACh-evoked current by Vc1.1. G1A, G1D, G1K, N9D, Y10A, and Y10K had comparable inhibitory activities to Vc1.1, whereas S4K and N9A were both significantly more potent, with 71 and 93% inhibition of ACh-evoked current amplitude, respectively, relative to 47% inhibition by Vc1.1 at 100 nM concentration (Fig. 5b). Concentration-response curves are shown in Fig. 5c. The second generation analogs, namely N9I, N9L, N9G, S4R, and S4KϩN9A, were significantly more potent than Vc1.1, with 82%, 82%, 89%, 88%, and 96% inhibition, respectively, of ACh-evoked current amplitude at 100 nM (see Fig.  5b). The concentration-response curves are shown in Fig. 5c. The IC 50 values from the pooled data for all of these mutants are summarized in Table 3. For the most potent mutants, N9G, N9I, and N9L, a ϳ12-fold decrease in their IC 50 values or leftward shift in the concentration-response curves, relative to Vc1.1 was observed.
Potency of Vc1.1 and Analogs at the Human ␣9/rat ␣10 Hybrid (h␣9r␣10 Hybrid)-Given that Vc1.1 has been reported 5 to be ϳ100-fold less potent at the human compared with the rat ␣9␣10 nAChR, Vc1.1 and the potent analogs at the rat ␣9␣10 nAChR were also tested at the h␣9r␣10 hybrid nAChR. The EC 50 for ACh activation of the h␣9r␣10 hybrid was ϳ50 M, and the IC 50 for inhibition of ACh-evoked currents by Vc1.1 was 549 nM. The Vc1.1 analogs N9A, N9I, N9L, N9G, S4R, S4K, and S4KϩN9A were tested at the h␣9r␣10 hybrid where, at 100 nM, they inhibited the ACh-evoked current amplitude by 83%, 82%, 74%, 86%, 93%, 53%, and 84%, respectively, relative to Vc1.1, which exhibits 38% inhibition at this concentration. The concentration-response curves for all of these analogs at the h␣9r␣10 hybrid are shown in Fig. 5d. Although, in general, saturation was observed at a higher concentration than that observed with the rat ␣9␣10, which indicated a generic increase in the IC 50 values of everything tested on the h␣9r␣10 hybrid nAChR, there was still a leftward shift in the concentration- The similarity in secondary ␣H shifts of the mutants to Vc1.1 indicates that the secondary structures are very similar. The mutants indicated by the arrows show significant differences in their secondary ␣H shifts, suggesting disturbances in the overall fold of these peptides. In the last panel the local changes around position 9 are highlighted. NOTE: The spectra for D5A, D5K, H12K and H12D were not fully assignable. response curves of the analogs relative to Vc1.1, suggesting they are still more potent. The IC 50 values from the pooled data are summarized in Table 3.

DISCUSSION
In this study we synthesized an extensive suite of single point mutants of Vc1.1 and determined their potency for inhibiting ACh-evoked currents at the rat ␣9␣10 nAChR and a hybrid h␣9r␣10 receptor. These studies identified the key residues important for biological function and the derived structureactivity relationships allowed us to design a second generation of novel mutants that elicited substantially increased inhibition of ACh-evoked current over the parent peptide. To determine the selectivity of the more potent mutants for particular nAChR subtypes, they were also screened at the ␣3␤2, ␣3␤4, and ␣7 nAChR subtypes. All showed clear selectivity for the ␣9␣10 nAChR. Overall the study provides important information that helps define the binding interactions of Vc1.1 in atomic detail and may assist in the development of analogs with applications as drugs for the treatment of neuropathic pain.
Mutations of residues 5-7 and 11-15 to Ala, Lys, or Asp led to a significant decrease or abolishment of functional activity at  . Membrane potential, Ϫ80 mV. b, inhibitory activity of Vc1.1 and mutants at 100 nM at the rat ␣9␣10 and human ␣9/rat ␣10 hybrid nAChRs. All the mutants shown are significantly more potent than Vc1.1 (p ϭ 0.05). c, concentration-response curves at the rat ␣9␣10 nAChR (n ϭ 3-6). d, concentration response at the human ␣9/rat ␣10 hybrid nAChR. The key for c and d is the same (n ϭ 3-6). Relative inhibitory activity was calculated by dividing the % response of each analog at 100 nM with the % response of Vc1.1 at 100 nM. 100 nM of each was used, because it is the ϳIC 50 value of Vc1.1. the nAChR. In principle, the loss of biological function observed for these analogs may be a consequence of the mutation either causing a disruption in the native structure that renders the peptide unable to bind to the receptor, or alternatively affecting the specific amino acid interactions required to induce a biological response; we refer to these as effects on "fold" or "function." NMR ␣H secondary shifts were used to assess the effect of the mutations on the fold of the individual peptides. Because ␣H shifts are sensitive to backbone conformation, similar ␣H secondary shifts for different peptides indicate similar backbone conformations. The ␣H secondary shifts for synthetic Vc1.1 measured here were identical to those originally reported by Clark et al. (21), confirming correct folding of the parent peptide in the current study. Most of the mutants also had similar patterns of ␣H shifts, unequivocally showing that there were no major changes in the secondary structure of the mutants. In general, therefore, the observed changes in activity can be interpreted in terms of function rather than fold.

Residue Ala Asp Lys
However, there are two minor exceptions. Variations in ␣H secondary shifts associated with mutation of Pro 13 suggest that mutations at this site cause disturbances in the structure of loop 2 of Vc1.1. This effect is apparent for all three Pro 13 mutants, namely P13A, P13K, and P13D, suggesting that Pro 13 has an important structural role to play in defining the native conformation (fold) of the peptides. Disturbances in the ␣H shifts associated with Pro 6 mutation are also apparent, again suggesting that the conformation of the peptide is affected by this mutation. Pro 6 has been directly implicated in receptor binding of ␣-conotoxins through a hydrophobic interaction with the ␤ subunit of the nAChR (38), and so for this residue the mutations are likely to affect both fold and function. The high conservation of Pro at this position in a number of different ␣-conotoxins is consistent with the fact that it is likely to play an important role defining both the ligand conformation and receptor binding activity.
Having established that the vast majority of the mutations do not affect the global fold of the peptides, we then interpreted the mutagenesis data in terms of binding and functional effects at the nAChR. Fig. 7 summarizes the effects of the various mutants of Vc1.1 on ACh-evoked currents at the wide range of nAChRs studied. For comparison, literature data (17) on RgIA are also shown.
Strikingly, the residues for which any mutation reduces inhibitory activity (i.e. residues 5-7 and 11-15) are clustered in two focal but distinct regions of Vc1.1, as illustrated in Fig. 8. The bifunctional nature of the bioactive surface of Vc1.1 is consistent with proposals for other ␣-conotoxins that suggest a subset of residues in loop 1 is primarily responsible for binding and that a set of residues in loop 2 is responsible for subtype selectivity. This is consistent with the fact that there is strong conservation of residues in loop 1 for a range of ␣-conotoxins, yet they have differential preferences for different nAChR subtypes (17,21,39,40).
In contrast to the loss of activity seen with substitutions at residues 5-7 and 11-15, substitutions at two other positions, Ser 4 and Asn 9 , improve the functional activity of Vc1.1 at the ␣9␣10 nAChR. Replacing Ser 4 with a positive residue, a Lys or an Arg, significantly increased the inhibitory activity of Vc1.1, as did the introduction of a hydrophobic Ala, Leu, or Ile at position 9. Before interpreting these functional effects it was important to check that they were not associated with a change in the structure of the peptide resulting from the mutations. The bottom panel of Fig. 3 shows that the secondary shifts of all of the second generation mutants superimpose precisely with those of Vc1.1, apart from some highly localized perturbations at residue 9. These perturbations mainly reflect local charge effects associated with the substitution at residue 9, although substitution at residue 4 also has a minor effect at residue 9. The surface representation in Fig. 8 shows that these two residues are adjacent to one another, so some crosstalk is not surprising, but overall the potency-enhanced mutants have the same folds as Vc1.1 itself. Given the lack of change in the global fold of the second generation mutants, it is clear that the increased potencies of residue 4 and residue 9 mutants are a result of either a stronger binding interaction or a more favorable functional interaction with the receptor. Their adjacent location on the surface of the molecule is consistent with either possibility. Ser 4 is highly conserved among many conotoxins, including ImI, MII, AuIB, EPI, PeIA, Vc1.1, RgIA, PnIA, PnIB, and others.
Here we show that a positive residue, instead of Ser 4 , is more favorable for potency at the ␣9␣10 nAChR. Mutations to either an Ala or an Asp at position 4 reduce the activity of Vc1.1. RgIA, another analgesic ␣-conotoxin, showed no change in its activity toward the ␣9␣10 nAChR when its Ser 4 was replaced by Ala (17), suggesting that the role of Ser 4 in different conotoxins may not be the same.
Replacing the polar residue Asn 9 with a hydrophobic residue such as Ala, Leu, or Ile significantly increased the potency of Vc1.1 at both the rat ␣9␣10 and the h␣9r␣10 hybrid nAChR. The 4/7 ␣-conotoxins usually contain a hydrophobic patch on the surface, which is thought to play a role in receptor binding and subtype specificity. Fig. 8 shows that, when Asn 9 is replaced with a hydrophobic residue, then two smaller hydrophobic regions are joined, thus forming a more distinct hydrophobic patch.
Interestingly, RgIA shows the opposite trend. Ellison et al. (17) found that substitution of Arg 9 to an Ala in RgIA led to a loss of activity at the ␣9␣10 nAChR. The difference in behavior of Vc1.1 and RgIA reflects the fact that one is a 4/7 conotoxin and the other a 4/3 conotoxin. An overlay of the two structures (data not shown) shows that, whereas loop 1 is completely superimposable, the conformations of loop 2 are quite different.
The structural similarities (loop 1) and differences (loop 2) between 4/7 and 4/3 ␣-conotoxins help to explain some of the differential effects of mutagenesis in the two classes of ␣-conotoxins. For example, both RgIA and Vc1.1 have a Tyr at position 10. In this study we found that mutating Tyr 10 to an Ala or a Lys had no effect on the activity of Vc1.1. However, mutating it to an Asp reduces the activity, suggesting that a polar, hydrophobic, or positively charged residue, but not a negatively charged residue, is acceptable at this position. The substitution of Tyr 10 in RgIA to a Trp had no effect on its activity, similar to the findings of the current study. Furthermore, the importance of the Asp 5 , Pro 6 , and Arg 7 triad for functional activity has been highlighted for RgIA (17) and ImI (39). Because ImI is selective FIGURE 7. A summary of the substitutions that lead to an increase, decrease, or no change in the activity of Vc1.1 and RgIA at different nAChR subtypes. The blue circles represent substitutions that lead to loss of activity, the yellow circles represent mutations with comparable activity to the native peptide, and the red circles represent substitutions that lead to an increase in potency. The disulfide bond connectivity is shown by the lines connecting the cysteines. The mutational study data summarized here for RgIA was reported by Ellison et al. (17). ith respect to one another. Asp 5 and Asp 11 are highlighted in a darker shade of blue as these residues were not mutated with an Asp because the parent peptide already has an Asp in these positions. Additionally, they were the first residues in each cluster (Asp 5 -Arg 7 and Asp 11 -Ile 15 ) leading to the loss of functional activity. c, shows a surface representation, highlighting a hydrophobic patch (light green) that is formed when residue 9 (dark red) is mutated to a hydrophobic residue, leading to increased potency. Ser 4 is highlighted in a lighter shade of red to show its proximity to position 9. d, ribbon representation highlighting the positions of the Ser 4 and Asn 9 side chains. The structures are based on PDB deposition 2H8S.
for the ␣7 nAChR, this implies that the triad is important in a generic role for ␣-conotoxins, irrespective of the nAChR for which they are most selective. Therefore, in the case of Vc1.1 we suggest that residues 5-7 are crucial for binding, whereas residues 11-15 are required for selectivity.
After establishing the key positions important for the potency at the ␣9␣10 nAChR, we conducted selectivity studies on the more potent mutants at other nAChR subtypes, namely ␣3␤2, ␣3␤4, and ␣7, known to be targeted by other conotoxins. No mutants found to be potent at the ␣9␣10 nAChR led to an increase in potency at the ␣3␤4 nAChR subtype. However, there was a 30-fold increase in the potency of N9A at the ␣3␤2 nAChR subtype, indicating the importance of an extended hydrophobic patch for binding at this nAChR subtype as well as at ␣9␣10. Interestingly, as the hydrophobic residue extended further, the potency of the mutants decreased from 30-fold, 20-fold, and 2-fold for N9A, N9I, and N9L, respectively. This is not unexpected, because the larger hydrophobic residues are more likely to disrupt the binding if the site at which residue 9 interacts is spatially limited. With the ␣9␣10 nAChR, we see that the N9G mutant is equally potent to the hydrophobic mutants N9A, N9I, and N9L, suggesting that a small non-hydrophobic residue can also be used to increase the potency of Vc1.1. However, it is clear that the increased potency of N9A, N9I, and N9L at the ␣3␤2 subtype is more closely related to the chemical nature of the residue than the size of the residue, because there is no increase in the potency of the N9G mutant at this subtype. Because Gly is a small and "inert" amino acid, it should essentially confer no real chemical change and instead act as a space filler. Consequently, the N9G mutant shows that the hydrophobicity and the size of a residue are both important for conferring potency, although the emphasis on each characteristic may vary at different nAChR subtypes.
In conclusion, we have shown that mutations at positions 4 and 9 are important in increasing the potency of Vc1.1 at the ␣9␣10 nAChR. All second generation mutants were found to be more selective for the ␣9␣10 nAChR over the other subtypes, with four mutants, N9G, S4K, S4KϩN9A, and S4R, being exclusively selective for the ␣9␣10 nAChR. Therefore, we have elucidated the key residues involved in the interaction of Vc1.1 with the ␣9␣10 nAChR subtype. The detailed understanding of the structure-function relationships of Vc1.1 that this study provides has the potential to assist in the design and development of conotoxin analogs for the treatment of neuropathic pain. Further optimization of potent analogs as stable peptide drugs might be achieved through backbone cyclization. This approach has been shown to be successful for increasing the stability of two classes of conotoxins while maintaining their biological activity (41,42).