A New (cid:97) -Conotoxin Which Targets (cid:97) 3 (cid:98) 2 Nicotinic Acetylcholine Receptors*

We have isolated a 16-amino acid peptide from the venom of the marine snail Conus magus which potently blocks nicotinic acetylcholine receptors (nAChRs) com- posed of (cid:97) 3 (cid:98) 2 subunits. This peptide, named (cid:97) -cono-toxin MII, was identified by electrophysiologically screening venom fractions against cloned nicotinic receptors expressed in Xenopus oocytes. The peptide’s structure, which has been confirmed by mass spectrometry and total chemical synthesis, differs significantly from those of all previously isolated (cid:97) -conotoxins. Disulfide bridging, however, is conserved. The toxin blocks the response to acetylcholine in oocytes expressing (cid:97) 3 (cid:98) 2 nAChRs with an IC 50 of 0.5 n M and is 2–4 orders of magnitude less potent on other nAChR subunit combinations. We have recently reported the isolation and characterization of (cid:97) -conotoxin ImI, which selectively targets homomeric (cid:97) 7 neuronal nAChRs. Yet other (cid:97) -conotoxins selectively block the muscle subtype of nAChR. Thus, it is increasingly apparent that (cid:97) -conotox-ins represent a significant resource for ligands with which to probe structure-function relationships of various nAChR subtypes.

The muscle subtype of nicotinic acetylcholine receptor (nAChR) 1 is one of the best understood ligand-gated channels due in part to the availability of a large number of protein and small molecule ligands which serve as specific probes for this channel. The nAChR is a heteropentameric ion channel complex and is a member of a superfamily that includes glycine, GABA A , and 5-HT 3 receptors (1). The mammalian nAChR has the subunit composition (␣1) 2 ␤1␥␦ in developing muscle, and the ␥ subunit is replaced by an ⑀ subunit in mature muscle. In mammalian neurons the situation is much more complex with at least seven ␣ subunits, designated ␣2-␣7 and ␣9 (in chick there is also an ␣8 subunit), and three ␤ subunits, ␤2-␤4. The ␣2, ␣3, and ␣4 subunits can each combine with ␤2 or ␤4 subunits to form functional channels when expressed in Xenopus oocytes, e.g. ␣2␤2, ␣3␤2, ␣2␤4, etc. In addition, ␣7 and ␣9 subunits can be expressed as functional homooligomers in this system. Studies employing either nucleotide probes or antibodies indicate that each of these ␣ and ␤ subunits have a unique pattern of anatomical expression in the central nervous system (2). However, the precise structural composition and functional role of the different neuronal subtypes of nicotinic receptors are less well understood. The development of subtype-specific ligands will greatly aid progress in this area.
Although a number of valuable nicotinic antagonists have been described, few are highly subtype-selective, particularly in the case of neuronal nAChRs. d-Tubocurarine, an alkaloid from the Chondrodendron tomentosum bush, used for centuries as an arrow poison to kill wild game, blocks both muscle and neuronal nAChRs (3). In addition it binds to all neuronal nicotinic receptors with more or less similar affinities (4). Likewise, dihydro-␤-erythroidine, the hydrogenated derivative of erythroidine, isolated from trees and shrubs of the genus Erythrina is a competitive antagonist at both muscle and neuronal nAChRs (4). Lophotoxin, a small cyclic diterpene, is used by the soft shell coral Lophogorgia rigida to discourage its consumption by fish (5). This toxin forms a covalent bond with Tyr 190 of the ␣-subunit of Torpedo nAChRs, irreversibly blocking the binding of ACh to the receptor (6,7). Studies with nAChRs expressed in Xenopus oocytes reveal that this toxin blocks muscle nAChRs as well as ␣2␤2, ␣3␤2, and ␣4␤2 neuronal nAChRs (8). Neosurugatoxin, a glycoside from the gastropod Babylonia japonica (9), potently but nonselectively blocks ␣x␤2 nAChRs expressed in oocytes, where x is 2, 3, or 4 (8). The synthetically derived small molecules trimethaphan and mecamylamine discriminate between ganglionic and neuromuscular nAChRs and are used clinically as ganglionic blocking agents (3).
Numerous protein toxins which act at muscle nAChRs have been isolated from a variety of snake venoms and proven highly useful for studying nAChRs. Two toxins from the Taiwanese banded krait, Bungarus multicinctus, have been particularly well characterized. The major nicotinic antagonist in this venom, ␣-bungarotoxin, in addition to blocking the muscle receptor, potently blocks ␣7 subunit-containing neuronal nAChRs (10). Methyllycaconitine, an alkaloid toxin from the seeds of Delphinium brownii has markedly greater affinity for the 125 I-␣-bungarotoxin binding site in brain versus that in muscle demonstrating that these receptor subtypes can be pharmacologically distinguished (11). A minor component of Bungarus venom, -bungarotoxin (also known as neuronal bungarotoxin, toxin F, or Bgt 3.1) preferentially blocks ␣3␤2 receptors (8), although the presence of venom purification contaminants has led to inconsistent findings (8,12). Unfortunately, due to limited availability of venom, this potent toxin is commercially unavailable at the present time.
A growing number of nicotinic antagonists have been isolated from the venom of the carnivorous marine snail Conus and are known as ␣-conotoxins. In contrast to snake ␣-toxins (ϳ60 -80 amino acids), ␣-conotoxins are much smaller (ϳ12-25 amino acids), a feature which has allowed them to be readily chemically synthesized (13). ␣-Conotoxins, which target the muscle nAChR are enjoying increasing use due to their recently discovered ability to differentiate between the two acetylcholine binding sites on the receptor. In the mouse muscle-derived BC 3 H-1 cell line ␣-conotoxins MI, GI, and SIA (respectively from Conus magus, Conus geographus, and Conus striatus) selectively bind to the ACh binding site at the ␣/␦ interface with more than 10 4 -fold greater affinity than the site at the ␣/␥ interface (14,15). With Torpedo nAChR, the situation is reversed. ␣-Conotoxins MI and GI bind at the ␣/␥ interface with approximately 2 orders of magnitude greater affinity than the ␣/␦ interface (15,16). Like ␣-conotoxins MI, GI, and SIA, ␣-conotoxin EI, from Conus ermineus, prefers the ␣/␦ to the ␣/␥ interface of receptors in BC 3 H-1 cells, but with only 30-fold difference. In contrast to these other ␣-conotoxins, with Torpedo receptors, ␣-conotoxin EI preferentially binds the ␣/␦ versus the ␣/␥ interface by a 400-fold difference in affinity and is the only ligand known to possess this selectivity (17). Thus, these ␣-conotoxins can serve as specific probes to investigate structure-function relationships of nAChRs (18).
There are approximately 500 species of Conus. Each of these predatory gastropods hunt prey from one of five different phyla, and all of these prey have cholinergic synapses (19). Thus, there is a potentially very wide diversity of nAChRs for conotoxins to target, and it is likely therefore that there are a comparably wide spectrum of cholinergically active peptides in the venom of Conus. We are seeking to exploit this situation to develop a bank of peptides which act on specific subtypes of neuronal nicotinic receptors. By use of a bioassay involving intracranial injections into mice to guide purification, we previously isolated ␣-conotoxin ImI which, unlike other ␣-conotoxins, selectively targets ␣7, and to a lesser degree ␣9, nAChRs (20,21). In the present study we used a much more specific screening assay to purify a novel nicotinic antagonist from C. magus venom. Voltage-clamped Xenopus oocytes expressing ␣3␤2 nAChRs were used in the assay to isolate ␣-conotoxin MII. We report the structural characterization and nAChR subtype selectivity of this peptide.

Peptide Isolation and Sequencing
Venom Extraction-Crude venom from dissected ducts of C. magus was collected in the Philippines, lyophilized, and stored at Ϫ70°C until used. All reagents were precooled to, and extraction procedures were conducted at, 4°C. Fifteen ml of 0.1% trifluoroacetic acid was added to 500 mg of lyophilized venom, and the mixture was vortexed for 20 min. This mixture was centrifuged at 17,000 ϫ g for 20 min. The supernatant was transferred to a separate tube, and another 15 ml of 0.1% trifluoroacetic acid was added to the pellet which was then sonicated with a Sonifier (Branson Instruments) at setting #4, vortexed for 10 min, and centrifuged as above. The supernatants were combined and filtered through a Whatman GF/C filter (Whatman, Ltd, Maidstone, UK), and then placed in two Centriprep 30 microconcentrators (Amicon, Beverly, MA) which have a 30,000 molecular weight cut-off. The Centripreps were centrifuged at 1500 ϫ g until the retentate in each was reduced to 5 ml (ϳ45 min). The filtrate was removed and 10 ml of 0.1% trifluoroacetic acid was added to the retentate of each Centriprep. The Centripreps were again centrifuged at 1500 ϫ g for 120 min. The addition of trifluoroacetic acid reduced the viscosity of the retentate and improved the recovery of filtrate. Filtrates were combined and used for further purification of ␣-conotoxin MII.
Pyridylethylation and Purification of Modified Peptide-Peptide from the final purification was stored in the RPLC buffer in which it eluted. A 287-l solution of this purified peptide (ϳ250 pmol) was combined with 14.4 l (20:1 v/v) of 0.5 M Tris base which raised the pH to a value between 7 and 8 as measured with pH paper. Seventy-five l of 50 mM dithiothreitol was added (final concentration 10 mM); the reaction vessel was flushed with argon, and the reaction incubated at 65°C for 15 min. The solution was allowed to cool; 15 l of 20% 4-vinyl pyridine in ethanol was added, and the solution was reacted for a further 25 min at room temperature in the dark. The solution was diluted 3-fold with 0.1% trifluoroacetic acid, and the alkylated peptide was loaded on the Brownlee column. After washing the column with 20% buffer B to allow the baseline to return to 10% of the initial reading, the peptide was eluted with the gradient described in Fig. 1, Sequence Analysis-Sequencing was performed with Edman chemistry on an Applied Biosystems 477A Protein Sequencer at the Protein/ DNA Core Facility at the University of Utah Cancer Center. Mass spectrometry was performed as described previously (17).

Peptide Synthesis
Linear Peptide-All amino acid derivatives were purchased from Bachem (Torrance, CA). The linear peptide chain was built on Rink amide resin by Fmoc (N-(9-fluorenyl)methoxycarbonyl) procedures with 2-(1H-benzotriole-1-yl)-1,1,3,3-tetramethyluronium tetrafluoroborate coupling, using an ABI model 430A peptide synthesizer. Side chain protection of non-Cys residues was in the form of t-butyl (Glu, Ser) and trityl (Asn, His). Orthogonal protection was used on cysteines: Cys 3 and Cys 16 were protected as the stable Cys(S-acetamidomethyl), while Cys 2 and Cys 8 were protected as the acid-labile Cys(S-trityl). After assembly of the resin-bound peptide, the terminal Fmoc group was removed in situ by treatment with 20% piperidine in N-methylpyrrolidone. Linear peptide amide was cleaved from 93 mg of resin by treatment with 1 ml of trifluoroacetic acid/H 2 O/ethanedithiol/phenol/thioanisole (90/5/2.5/ 7.5/5 by volume) for 1.5 h at 20°C. Released peptide was precipitated by filtering the reaction mixture into methyl-t-butyl ether which had been cooled to Ϫ10°C. This procedure simultaneously cleaved peptide from the resin and deprotected the Cys(S-trityl) and the non-Cys residue side chains, but not the Cys(S-acetamidomethyl) residues. The cleavage reaction vessel was rinsed with 100% trifluoroacetic acid, and this rinse was also filtered into the methyl-t-butyl ether solution. The precipitate was washed two additional times with chilled methyl-t-butyl ether, and the supernatants were discarded. Pelleted peptide was dissolved by the addition of approximately 10 ml of 0.1% trifluoroacetic acid in 60% acetonitrile, with gentle swirling (to avoid foaming). The linear peptide solution was diluted with 190 ml of 0.1% trifluoroacetic acid and purified by RPLC on the preparative C18 Vydac column with a 10 -60% buffer B gradient over 50 min. Flow rate was 20 ml/min. This gradient was also used for all subsequent preparative RPLC purifications of the synthetic peptide.
Peptide Cyclization-To form a disulfide bridge between Cys 2 and Cys 8 (i.e. the first and third cysteines), the major peptide fraction from the preparative RPLC (see above) was diluted to 1 liter with H 2 O and solid Tris base was added to increase the pH to 7.6. The solution was placed in a 4-liter flask and gently swirled at room temperature for 38 h at which time the reaction was judged to be complete by analytical RPLC. The pH of the solution was decreased to a value of 2-3 (measured with pH paper) by the addition of 4 ml of trifluoroacetic acid. The monocyclic peptide was then purified by RPLC and collected in a volume of 45 ml. Removal of the S-acetamidomethyl groups and closure of the second disulfide bridge (Cys 3 -Cys 16 , i.e. the second and fourth cysteines) was carried out simultaneously by iodine oxidation. The 45 ml of RPLC eluent containing the monocyclic peptide was dripped into a rapidly stirred 50-ml solution of 20 mM iodine in H 2 0/trifluoroacetic acid/acetonitrile/MeOH (50:20:20:10 by volume) over 5.5 min at room temperature. The reaction was allowed to proceed for another 15 min and terminated by the addition of ascorbic acid. The solution was diluted to 1 liter and the bicyclic peptide purified by RPLC. cRNA Injection-cRNA was injected with a Drummond 10-l microdispenser (Drummond Scientific, Broomall, PA) essentially as described by Goldin (29). It was fitted with micropipettes pulled from glass capillaries provided for the microdispenser. The pipette tips were broken to an OD of 22-25 m and back-filled with paraffin before mounting on the microdispenser. cRNA was drawn into the micropipette and 50 nl, containing 5 ng of cRNA of each subunit, was injected into each oocyte. In the case of muscle subunits, 0.5-2.5 ng of each subunit was injected.
Voltage-clamp Recording-An injected oocyte was placed in a ϳ30-l recording chamber consisting of a cylindrical well (ϳ4 mm diameter ϫ 2 mm deep) fabricated from Sylgard, and gravity-perfused with either ND96 or ND96 containing 1 M atropine (ND96A) at a rate of ϳ1 ml/min. All toxin solutions also contained 0.1 mg/ml bovine serum albumin to reduce nonspecific adsorption of toxin. The perfusion medium could be switched to one containing toxin or acetylcholine (ACh) by use of a distributor valve (SmartValve, Cavro Scientific Instruments, Sunnyvale, CA) and a series of three-way solenoid valves (model 161TO31, Neptune Research, Northboro, MA). ACh-gated currents were obtained with a two-electrode voltage-clamp amplifier (model OC-725B, Warner Instrument Corp., Hamden, CT) set for "fast" clamp and with clamp gain at maximum (ϫ 2000). Glass microelectrodes, pulled from fiber-filled borosilicate capillaries (1 mm outer diameter ϫ 0.75 mm inner diameter, WPI Inc., Sarasota, FL) and filled with 3 M KCl, served as voltage and current electrodes. Resistances were 0.5-5 megohm for voltage, and 0.5-2 megohm for current electrodes. The membrane potential was clamped at Ϫ70 mV, and the current signal, recorded through virtual ground, was low-pass filtered (5 Hz cut-off) and digitized at a sampling frequency of 20 Hz. The solenoid perfusion 2 J. Boulter, unpublished data. The gradient was 20 -50% buffer B/60 min at a flow rate of 1 ml/min. Panel C, the right half of a center cut of the absorbance peak indicated by the arrow in Panel B was diluted with 2 volumes of 0.1% trifluoroacetic acid and re-chromatographed as described in panel B to obtain the final purified product. A 5-ml sample loading loop was used in all chromatography. Buffer A, 0.1% trifluoroacetic acid; and buffer B, 0.1% trifluoroacetic acid, 60% acetonitrile. Absorbance was monitored at 220 nm. valves were controlled with solid state relays (model ODC5 in a PB16HC digital I/O backplane, Opto 22, Temecula, CA). A/D conversion and digital control of solenoid valves were performed with a Lab-LC or Lab-NB board (National Instruments, Austin, TX) in a Macintosh (Quadra 630 or IIcx) computer. The computer communicated with the distributor valve via its serial port. Data acquisition and activities of the distributor and solenoid valves were automatically controlled by a home-made virtual instrument constructed with the graphical programming language LabVIEW (National Instruments, Austin, TX). To apply a pulse of ACh to the oocyte, the perfusion fluid was switched to one containing ACh for 1 s. This was automatically done every 5 min. The concentration of ACh was 1 M for oocytes expressing ␣1␤1␥␦, 1 mM for ␣7, and 300 M for all others. The ACh was diluted in ND96A for all except ␣7, in which case the diluent was ND96. For control responses, the ACh pulse was preceded by perfusion with ND96 (for ␣7) or ND96A (all others). No atropine was used with oocytes expressing ␣7, since it has been demonstrated to be an antagonist of these receptors (30). For responses in toxin (test responses), the oocyte was perfused with toxin solution until equilibrated (generally 5 min, but up to 25 min at lower toxin concentrations) before the ACh pulse was applied. All ACh pulses contained no toxin, for it was assumed that little, if any, bound toxin would have washed away in the brief time (Ͻ2 s) it takes for the responses to peak (see Fig. 3). The peak amplitudes of the ACh-gated current responses were measured by the virtual instrument. The average of three control responses just preceding a test response was used to normalize the test response to obtain "% response." Each data point of a dose-response curve represents the average Ϯ S.E. of at least three oocytes. Dose-response curves were fit, with Prism software (GraphPad Software Inc., San Diego, CA), to the equation: % response ϭ 100/{1 ϩ ([toxin]/IC 50 ) nH }, where n H is the Hill coefficient. All recordings were done at room temperature (ϳ22°C).
Bioassay-Intraperitoneal injections of toxin into Swiss Webster mice and intramuscular injections into goldfish were performed as described previously (17,20).

RESULTS
Purification and Characterization of ␣-Conotoxin MII-Serial dilutions of a 50 mg/ml ND96 buffer extract of crude C. magus venom were tested for their ability to block the AChinduced current in Xenopus oocytes expressing ␣3␤2 nAChRs. Dose-dependent block was observed; 82% block was produced with 0.071 mg/ml crude venom extract solution (data not shown). C. magus venom was purified by RPLC as described under "Materials and Methods." For the initial RPLC fractionation (Fig. 1), 5-ml fractions were collected. Aliquots of 0.2% of each fraction were pooled in groups of three, lyophilized, and 30% of each pool was tested on oocytes expressing ␣3␤2 subunits (see "Materials and Methods"). Individual fractions of the active pool were then tested within the oocyte system and the active fraction purified to homogeneity via RPLC.
The purified peptide was reduced, alkylated, and sequenced as described under "Materials and Methods." The sequence is: GCCSNPVCHLEHSNLC. Liquid secondary ion mass spectrometry indicated that Cys residues are present as disulfides and that the COOH-terminal ␣-carboxyl group is amidated (monoisotopic MH ϩ : calculated 1710.65, observed 1710.6). The sequence was further verified by total chemical synthesis (see below). The sequence resembles previously isolated ␣-conotoxins in its spacing of Cys residues, yet differs substantially in other amino acids. As will be shown below, the peptide potently targets the nAChR, and we have therefore named the peptide ␣-conotoxin MII in accordance with the nomenclature previously proposed for conotoxins (31).
Chemical Synthesis-Solid phase chemical synthesis of ␣-conotoxin MII was undertaken to provide an abundant supply of peptide. It was assumed that the disulfide bridging of ␣-conotoxin MII would be analogous to all previously characterized ␣-conotoxins, i.e. 1st Cys-3rd Cys, 2nd Cys-4th Cys. Cys groups were protected in pairs to direct disulfide formation. The acid-labile trityl protecting groups were removed from Cys 2 and Cys 8 (i.e. the first and third cysteines) during the cleavage reaction which released linear peptide from the resin. Closure of the disulfide bridge was accomplished by air oxida- tion, and the monocyclic peptide was purified by RPLC. The acid-stable acetamidomethyl group was next removed from Cys 3 and Cys 16 (i.e. the second and fourth cysteines), and the disulfide bridge closed by rapid iodine oxidation and the bicyclic peptide purified by RPLC. Peptide yield was 41.7 nmol/mg peptide resin.
The order of RPLC elution of the synthetic peptides is of note: linear first, followed by monocyclic and bicyclic last. With the formation of each disulfide bridge, ␣-conotoxin MII becomes increasingly hydrophobic. This is exactly the opposite behavior from ␣-conotoxin EI, where the formation of each disulfide bond results in decreased retention time on RPLC (17), and may indicate that the disulfide bridges in ␣-conotoxin MII force hydrophobic residues to face outward. Synthetic peptide comigrated with native on RPLC (Fig. 2). Liquid secondary ionization mass spectrometry of synthetic ␣-conotoxin MII was consistent with the amidated sequence (monoisotopic MH ϩ : calculated 1710.65, observed 1710.8). Lyophilization or bath application of small (fmol Ϫ pmol) amounts of ␣-conotoxin MII resulted in apparent loss of peptide (data not shown). The hydrophobic nature of the peptide may account for this problem which was minimized by the use of carrier protein (bovine serum albumin) in all solutions and continuous perfusion rather than bath application of peptide.
Electrophysiology-␣-Conotoxin MII purification by RPLC was guided by an assay which used Xenopus oocytes expressing ␣3␤2 receptors. RPLC fractions were tested for their ability to block the ACh-induced response in this assay. We also examined the effect of the toxin on other nAChR subunit combinations expressed in oocytes. Both native and synthetic toxin blocked the ␣3␤2 nAChRs with equal potency (data not shown). Due to very limited availability of native toxin, synthetic toxin was used for all subsequent experiments. Synthetic ␣-conotoxin MII showed dose-dependent block of ␣3␤2 receptors at low nanomolar concentrations (Fig. 3). This block slowly reversed with washing (Fig. 4). ␣-Conotoxin MII blocks ␣3␤2 nAChRs with an IC 50 of 0.5 nM, with an apparent Hill coefficient, n H , of 0.8 (Fig. 5). ␣-Conotoxin MII was also tested on other nAChR subunit combinations. Results are shown in Fig.  5 and indicate that ␣-conotoxin MII is 2-4 orders of magnitude more potent at ␣3␤2 nAChRs than at other nAChR subtypes.
We have previously shown that ␣-conotoxin MI (also from C. magus venom), ␣-conotoxin GI, and ␣-conotoxin ImI have no effect on ␣3␤2 receptors at up to 5 M concentration (21). Thus, ␣-conotoxin MII is the only conotoxin known to potently block this neuronal receptor subtype.
In Vivo Activity-Intraperitoneal injections of 10 nmol of ␣-conotoxin MII into 8 -10-g mice did not result in any signs of paralysis (n ϭ 3). This is in contrast to ␣-conotoxin MI, 0.67 nmol of which kills a 20-g mouse in 20 min (32). Intramuscular injection of 5 nmol of ␣-conotoxin MII into fish did not result in any signs of paralysis (n ϭ 3). This is in contrast to ␣-conotoxin MI where 0.5 nmol is paralytic. DISCUSSION nAChR Selectivity-Xenopus oocytes expressing mammalian neuronal ␣3␤2 nAChRs were used in an assay which successfully guided the isolation of the novel 16-residue peptide, ␣-conotoxin MII. This is significant in that it is the first ␣-conotoxin known to target ␣3␤2 receptors. Most previously reported ␣-conotoxins target the muscle nAChR. Exceptions are ␣-conotoxin ImI, which selectively blocks homomeric ␣7 and ␣9 receptors (20,21), and ␣-conotoxins PnIA/B, which block molluscan neuronal nAChRs (33). We have shown elsewhere that ␣-conotoxins MI and GI potently target muscle nAChRs expressed in Xenopus oocytes, but are inactive at all neuronal nAChRs tested, including ␣3␤2 receptors (21). As demonstrated in this report, ␣3␤2 receptors are blocked by ␣-conotoxin MII with an IC 50 of 0.5 nM. The effectiveness of the toxin on the other nAChR subunit combinations tested is 2-4 orders of magnitude less. Thus, ␣-conotoxin MII has an entirely unique activity profile among the ␣-conotoxins (Table I), and represents a potent and selective new probe for studying nAChRs. Notably, the small size of ␣-conotoxin MII has allowed it to be chemically synthesized and thus readily available.
Structural Relationships among ␣-Conotoxins-Reported ␣-Conotoxin MII Targets ␣3␤2 nAChRs ␣-conotoxins can be classified into two main groups based on the spacing of the cysteine residues. One group has a "3,5 spacing" where the numerals indicate the number of amino acids between the second and third Cys and the third and fourth Cys, respectively (see Table II). The other group, which includes ␣-conotoxin MII, has a "4,7 spacing." Individual toxins from both groups can potently block the muscle nAChR, suggesting that it is not the Cys spacing which is responsible for ␣-conotoxin MII's selectivity. Aside from the Cys residues, the only completely conserved residue in all reported ␣-conotoxins is a proline between the second and third Cys. The other non-Cys residues in ␣-conotoxin MII are strongly divergent from all other ␣-conotoxins. Of the non-Cys residues in the 4,7 group, ␣-conotoxin MII from C. magus shares only 4 of 12 residues with ␣-conotoxin PnIB from C. pennaceus and only 1 of 12 residues with ␣-conotoxin EI from C. ermineus. Furthermore, except for the proline, ␣-conotoxin MII shares little if any homology with ␣-conotoxin MI (which has a 3,5 spacing) although both are from the same Conus species. However, despite the difference in Cys spacing and strong divergence in other amino acids, the disulfide bridges are exactly analogous between the two toxins, 1st Cys to 3rd Cys, 2nd Cys to 4th Cys, and this arrangement is also conserved in all other ␣-conotoxins where the disulfide linkages have been studied (Table II). Recently, a new family of Conus peptides targeted to nAChRs was reported. These ␣A-conotoxins have a distinctly different structure including three disulfide bridges instead of two (34).
Multiple nAChR-targeted toxins have previously been isolated from two other Conus species, C. geographus and C. striatus (see Table II). In both of these cases, however, the ␣-conotoxins have considerable structural homology and all target the muscle nAChR in contrast to the MI and MII peptides isolated from C. magus that differ substantially in both structure and function. It will be of interest to determine which residues in MII confer ␣3␤2 nAChR selectivity and which in MI confer ␣1␤1␥␦ selectivity.
Biological Role-Injection of venom by Conus snails results in prey immobilization and capture. Nevertheless, the majority of the 100 -200 peptides present in the venom of fish-hunting Conus do not induce paralysis when injected into fish. 3 The functional targets and roles of these non-paralytic peptides are under investigation. However, all previously reported ␣-conotoxins from fish-hunting Conus do cause rapid paralysis when injected into fish. ␣-Conotoxin MII is unique in not causing paralysis in this assay. It has recently been shown that only as few as three amino acid substitutions in the ␥ or ␦ subunit of mouse nAChR can result in a 10 4 -fold change in the affinity of this receptor for ␣-conotoxin MI (18). It has also been shown that a single amino acid substitution in ␣-conotoxin SI increases its affinity for mouse muscle nAChRs by 2 orders of magnitude (35). It is possible, therefore, that ␣-conotoxin MII does potently target the muscle nAChR of its natural tropical fish prey, and that a few amino acid substitutions in the goldfish nAChR used in our assay may be responsible for the observed substantial differences in toxin potency. Poor dispersal of the more hydrophobic ␣-conotoxin MII might also lead to an apparent lack of activity in our assay. Since C. magus already has a toxin which potently blocks muscle nAChRs in the form of ␣-conotoxin MI, another possibility is that C. magus uses ␣-conotoxin MII to selectively target ganglionic or adrenal nAChRs in fish to lessen the sympathetically mediated fight or flight response. In frog, ␣-conotoxin MII blocks ganglionic neurotransmission. 4 A related example may be neosurugatoxin. This glycoside, isolated from mid-gut gland of the Japanese ivory shell, appears to be responsible for human poisonings following ingestion of this carnivorous gastropod (9). Poisoning symptoms are consistent with blockade of autonomic ganglia. Like ␣-conotoxin MII, neosurugatoxin preferentially targets neuronal versus muscle nAChRs. In contrast, however, neosurugatoxin is non-selective among ␣x␤2 nAChRs (4,8).
The discovery of ␣-conotoxin MII, which differs substantially in both structure and function from other ␣-conotoxins, provides further evidence of the enormous diversity of nAChRtargeted toxins present in Conus. This report demonstrates the feasibility of using specific nAChR subunit combinations expressed in oocytes as a functional screen to initially detect and ultimately guide the purification of these peptides.