α-Conotoxins ImI and ImII SIMILAR α7 NICOTINIC RECEPTOR ANTAGONISTS ACT AT DIFFERENT SITES

A novel conotoxin, α-conotoxin ImII (α-CTx ImII), identified from Conus imperialis venom ducts, was chemically synthesized. A previously characterized C. imperialis conotoxin, α-conotoxin ImI (α-CTx ImI), is closely related; 9 of 12 amino acids are identical. Both α-CTx ImII and α-CTx ImI functionally inhibit heterologously expressed rat α7 nAChRs with similar IC50 values. Furthermore, the biological activities of intracranially applied α-CTx ImI and α-CTx ImII are similar over the same dosage range, and are consistent with α7 nAChR inhibition. However, unlike α-CTx ImI, α-CTx ImII was not able to block the binding of α-bungarotoxin to α7 nAChRs. α-Conotoxin ImI and α-bungarotoxin-binding sites have been well characterized as overlapping and located at the cleft between adjacent nAChR subunits. Because α-CTx ImI and α-CTx ImII share extensive sequence homology, the inability of α-CTx ImII to compete with α-BgTx is surprising. Furthermore, functional studies in oocytes indicate that there is no overlap between functional binding sites of α-CTx ImI and α-CTx ImII. Like α-CTx ImI, the block by α-CTx ImII is voltage-independent. Thus, α-CTx ImII represents a probe for a novel antagonist binding site, or microsite, on the α7 nAChR.

Marine snails in the genus Conus have venoms that contain a remarkable number of small peptide neurotoxins. Many of these peptides, the conotoxins, are rich in cysteine residues and are highly disulfide-bonded. Known conotoxins may be divided into families based on shared features (reviewed in Refs. 1 and 2). Members of a given conotoxin family have a characteristic number and spacing of cysteines, a conserved disulfide connectivity, and similar receptor targets. However, the toxins in a given family show great variability in their intercysteine sequence, and this accounts for the high degree of receptor subtype specificity within a toxin family. For example, the ␣-conotoxins are inhibitors of nicotinic acetylcholine receptors (nAChRs), 1 but individual ␣-conotoxins show a high degree of selectivity for different nAChR subtypes including the neuromuscular subtype and various neuronal subtypes (1). Minor changes in the sequence of the non-Cys residues of conotoxins can profoundly change their receptor subtype specificity. For example, the conotoxin ␣-CTx PnIA preferentially targets the ␣7 nAChR and ␣-CTx PnIB preferentially targets the ␣3␤2 nAChR despite the fact that the toxins only vary in two of 16 amino acids (3).
In this report, we describe the discovery of a novel ␣-conotoxin, ␣-conotoxin ImII (␣-CTx ImII) from the worm-hunting snail, Conus imperialis. This molecule is very similar to the previously characterized C. imperialis toxin ␣-conotoxin ImI (␣-CTx ImI) (it is identical in 9 of 12 amino acids). Like ␣-CTx ImI, ␣-CTx ImII inhibits the ␣7 nAChR, and both toxins display very similar potencies against this receptor. Unlike ␣-CTx ImI, however, ␣-CTx ImII does not compete with ␣-bungarotoxin (␣-BgTx), a classical competitive inhibitor of the ␣7 nAChR. Additionally, we show that ␣-CTx ImI and ␣-CTx ImII share little, if any, overlap in their functional binding sites on the receptor. The discovery of ␣-CTx ImII thus illustrates that not only can small changes in intercysteine amino acids alter subtype specificity, but they can also result in toxins that target the same receptor subtype at different sites.
Discovery of ␣-CTx ImII-The sequence of ␣-CTx ImII was obtained as part of a systematic analysis of ␣-conotoxin sequences, using PCR amplification of both cDNA and genomic DNA (5)(6)(7). The specimen of C. imperialis analyzed was collected in the Philippines, and hepatopancreas and venom duct tissue was isolated and stored at Ϫ70°C. The cDNA was prepared from venom duct as described previously (8), and genomic DNA was extracted from hepatopancreas using Puregene reagents and the marine invertebrates protocol provided by the manufacturer (Gentra).
Peptide Synthesis and Folding-Linear ␣-CTx ImI was synthesized and oxidized to form disulfide bridges (folded) as described previously (9). Linear ␣-CTx ImII was synthesized by standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry, using an ABI model 430A peptide synthesizer at the University of Utah core facility. The peptide was folded to give the correct disulfide connectivity (first Cys to third Cys and second Cys to fourth Cys) using orthogonal Cys protection. The first and third Cys residues had stable Cys(S-acetomidomethyl) protection, whereas the second and fourth Cys residues had acid-labile Cys(Strityl) protection. A previously described folding scheme (3) that se-* This work was supported in part by National Institutes of Health Grants GM48677 and MH53631. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper for portions of the ␣-CTx ImI and ␣-CTx ImII genes that encode the mature toxins has been submitted to the GenBank TM 1 The abbreviations used are: ACh, acetylcholine; ␣-BgTx, ␣-bungarotoxin; ␣-CTx, ␣-conotoxin; MLA, methyllycaconitine; PR, potassium quentially closed the second Cys to fourth Cys bridge and then the first Cys to third Cys bridge was used to generate toxin. The analogs [P6R]␣-CTx ImI and [R6P]␣-CTx ImII were generated in the same way as ␣-CTx ImII.
Bioassays-Biological activity of synthetic ␣-conotoxins was tested by intracranial injection into young mice as described previously (10).
Electrophysiology-Complementary RNA encoding rat ␣7 nAChR was prepared and injected into Xenopus laevis oocytes as described previously (11). The RNA was generated by in vitro transcription using a plasmid that was a gift from Dr. J. Boulter. The plasmid carries a rat ␣7 nAChR cDNA clone (accession number M85273) inserted into the EcoRI site of pBS SK(Ϫ). RNA was transcribed from the T7 promoter of SmaI linearized plasmid. Oocytes were injected 1-2 days after harvesting and used for voltage clamping 1-7 days after injection.
Voltage clamping was done essentially as has been described elsewhere (11). Briefly, oocytes were clamped at a holding potential of Ϫ70 mV with a two-electrode system and were perfused in a 30-l bath with ND96 (96 mM NaCl, 2.0 mM KCl, 1.8 mM CaCl 2 , 1.0 mM MgCl 2 , 5 mM HEPES, pH 7.1-7.5). Currents were elicited with 1-s pulses of 200 M ACh in ND96 every 1 min. Only oocytes that yielded stable responses to successive ACh pulses were used. To determine the concentration dependence of inhibition of rat ␣7 nAChRs, toxin was applied using a static bath method. That is, the ACh pulses and ND96 flow were halted, and conotoxin was applied to the bath. The bath was allowed to equilibrate for 5 min before the ND96 flow was resumed at the same time an the ACh pulse was applied. ACh pulses and ND96 flow continued until stable ACh-evoked currents were re-established. To determine the inhibition at different conotoxin concentrations, the peak current elicited by the first ACh pulse following toxin exposure was normalized to the peak current elicited following controls where ND96 alone, instead of toxin, were applied.
To show that the 5-min exposure to toxin was sufficient for toxin/ receptor binding to reach equilibrium, 5-and 10-min exposures of 1, 0.33, and 0.033 M ␣-CTx ImI or 1, 0.33, and 0.1 M ␣-CTx ImII were carried out, and no difference in percentage inhibition was seen at the two times.
To investigate the voltage dependence of ␣-CTx ImII inhibition, the block caused by 1 M ␣-CTx ImII was measured as described above, but at a range of holding potentials randomly altered between Ϫ110, Ϫ90, Ϫ70, Ϫ50, Ϫ30, and Ϫ10 mV. To determine the current-voltage relationship in the absence of toxin, the ratio of current amplitudes at two successive potentials was determined from the average of at least five currents before and after the voltage change.
Preparation of Cells Expressing Rat ␣7-5-HT 3 Chimera-The plasmid pZeoSV2-␣7 (V201) /5-HT 3 was a gift from Dr. N. S. Millar (4). It encodes a chimeric receptor that has the N-terminal ACh-binding domain of the rat ␣7 nAChR and the C-terminal of the homologous 5-HT 3 receptor. The chimera was expressed in HEK293 cells, which are null for endogenous ␣-BgTx binding. HEK293 cells were grown in Dulbecco's modified Eagle's medium containing 10% heat-inactivated fetal calf serum, 100 g/ml streptomycin, and 100 units/ml penicillin. Cells were transfected with pZeoSV2-␣7 (V201) /5-HT 3 using LipofectAMINE (Invitrogen) according to the manufacturer's instructions for HEK293 cells. After 48 h the cells were washed with ice-cold potassium Ringer's (PR) solution (140 mM KCl, 5.4 mM NaCl, 1.7 mM MgCl 2 , 25 mM HEPES, and 30 g/ml bovine serum albumin, adjusted to a pH of 7.4 with 10 mM NaOH). The cells were detached into fresh, ice-cold PR (3 ml/10-cm cell culture plate) with a cell scraper and were spun down (10,000 ϫ g). The cells were washed twice by resuspension in ice-cold PR followed by recentrifugation. The suspension was passed 10 times through an 18gauge needle, divided into 1-ml aliquots, snap-frozen in liquid nitrogen, and stored at Ϫ70°C until use.
Preparation of Crude Rat Brain Membranes-Crude rat brain membranes were prepared as described previously (3) except that membranes were frozen and stored in PR.
Competition Binding Assays-Each assay (300 l total volume) consisted of the following in PR: 200 l of thawed cells or crude rat brain membranes, a final concentration of 4 nM 3-125 I-␣-BgTx (Amersham Biosciences) and various concentrations of ␣-CTx ImI, ␣-CTx ImII, or 100 M MLA (to determine nonsaturatable binding). ␣-CTx ImI, ␣-CTx ImII, or MLA were preincubated with cells for 30 min prior to the addition of 3-125 I-␣-BgTx (applied in a volume of 4 l). The radioligand was allowed to bind for 15 min during which its association with receptor was linear with time (data not shown). The assays were quenched with 500 l of ice-cold d-tubocurarine (2400 M). The cells were harvested (using a Brandell cell harvester) through Whatman GF-B filters pretreated with 4% nonfat dry milk. The filters were washed three times with about 800 l of PR and were counted using a COBRAII ␥-counter (Packard). Nonsaturatable binding determined in assays containing 100 M MLA was subtracted from all readings and the resulting specific 3-125 I-␣-BgTx binding was normalized as a percentage of specific binding in the absence of toxin. Assays were done at 25 Ϯ 2°C.

RESULTS
PCR-based Discovery of ␣-CTx ImII-Members of a conotoxin family, both from a given Conus species as well as from different species, share conserved sequence elements in their gene structure (12,13). Thus, PCR strategies can amplify fragments of conotoxin genes that include sequence encoding the mature toxin. PCR was used to amplify ␣-conotoxin gene fragments from C. imperialis genomic DNA and cDNA. The heterogeneous pools of PCR product were cloned and independent clones were sequenced; sequences encoding two closely related peptides, ␣-CTx ImI and ␣-CTx ImII, were found ( Fig. 1). The ␣-CTx ImI peptide had previously been purified from C. imperialis venom (9) and is a potent and specific competitive inhibitor of rat ␣7 nAChRs (14,15). Based on the predicted sequence from the clone, ␣-CTx ImII was chemically synthesized and folded to form disulfide bonds (see ''Experimental Procedures''), and the synthetic peptide was then used to evaluate potential interactions with ␣7 nAChRs.
The Biological Activity of ␣-CTx ImII Is Similar to That Seen for the ␣7 nAChR-targeting Toxins ␣-CTx ImI and ␣-BgTx-␣-CTx ImI and ␣-BgTx have been shown to cause complex seizures when introduced intracranially into rats (14). This behavior is believed to be because of inhibition of ␣7 nAChRs. To see if ␣-CTx ImII caused similar effects, intracranial injections of ␣-CTx ImI and ␣-CTx ImII were made in young mice. As can be seen in Table I, the effects of both toxins were generally similar and are consistent with both toxins acting on the neuronal ␣7 subtype of the nAChR.
Block by ␣-CTx ImII, Like That by ␣-CTx ImI, Is Voltageindependent-The functional inhibition of oocyte-expressed rat ␣7 nAChRs by 1 M ␣-CTx ImII was measured at different holding potentials. As can be seen from Fig. 2C, the percent block was independent of holding potential indicating that the activity of ␣-CTx ImII is not voltage-dependent. Block by ␣-CTx ImI is also voltage-independent (15).
The 5-HT 3 receptor is highly homologous to the ␣7 nAChR, and the N-terminal ACh-binding domain of the ␣7 nAChR has been used to replace the N-terminal 5-HT-binding domain from the 5-HT 3 receptor (4,16,17). The resulting chimera can be expressed in HEK293 cells such that ␣-BgTx-binding sites are produced at a level ϳ1000-fold higher than when the native ␣7 receptor is used (17). In addition, the chimera retains the pharmacology of the wild-type receptor with respect to many cholinergic agonists and antagonists (16,17). The ability of ␣-CTx ImII and ␣-CTx ImI to inhibit 3-125 I-␣-BgTx binding to rat ␣7-5-HT 3 chimera was tested as described under "Experimental Procedures." As shown in Fig. 3B, the same pattern of inhibition was seen with the chimera as with the native ␣7 receptor. Again, ␣-CTx ImII was unable to significantly block 3-125 I-␣-BgTx binding, but ␣-CTx ImI inhibited all specific binding of the radiolabel.
Competition for Functional ␣-BgTx-binding Sites-It was previously shown using rat hippocampal neurons (15) that preincubation of ␣7 nAChRs with ␣-CTx ImI prevents the very slowly reversible functional block by ␣-BgTx. We have used a similar approach to investigate the functional binding sites of ␣-CTx ImII and ␣-CTx ImI on oocyte-expressed rat ␣7 nAChRs. It was found that a 5-min bath application of 100 nM ␣-BgTx is sufficient to block about 95% of ACh-gated current in oocytes expressing rat ␣7 nAChRs. Because of the very slow off-rate of ␣-BgTx, no significant recovery was observed after washing toxin from the oocyte bath (Fig. 4A).
However, when oocytes were pretreated for 5 min with 100 M ␣-CTx ImI and then subjected to a 5-min co-application of ␣-BgTx and ␣-CTx ImI, very rapid and essentially full recovery was observed after washing out the toxins. This result is consistent with ␣-CTx ImI binding preventing the slowly reversible block by ␣-BgTx, i.e. that the two toxins compete for the same functional site. However, a much more limited ability to protect against block by ␣-BgTx (Fig. 4C) is achieved by a similar preincubation with ␣-CTx ImII. Note that 5 min of bath application of ␣-CTx ImI and ␣-CTx ImII is sufficient for both to reach equilibrium with receptor (see ''Experimental Procedures'').
Preincubation with a High Concentration of ␣-CTx ImI Does Not Inhibit Binding of ␣-CTx ImII to Oocyte-expressed Rat ␣7 nAChRs-The ability of ␣-CTx ImII to bind to oocyte-expressed receptor was tested with and without pre-equilibration of oocytes with a high concentration of ␣-CTx ImI. As can be seen in Fig. 5A, a 5-min bath application of ␣-CTx ImI (100 M) or ␣-CTx ImII (10 M) is sufficient to completely inhibit AChgated ion currents in oocyte-expressed rat ␣7 nAChRs. Subsequent washout results in full recovery for both toxins; however, ␣-CTx ImII has a noticeably slower off-rate than ␣-CTx ImI. Although the differences are subtle, they are highly reproducible and a diagnostic functional difference between the toxins.
When 100 M ␣-CTx ImI was bath-applied to oocytes expressing rat ␣7 nAChRs for 10 min, the characteristic fast off-rate of ␣-CTx ImI was observed (Fig. 5C). However, when 100 M ␣-CTx ImI was bath-applied for 5 min and 10 M ␣-CTx ImII was then added, giving 5 min of co-application of ␣-CTx ImI and ␣-CTx ImII, the characteristic slow off-rate for ␣-CTx ImII was observed, and the result was not detectably different from that of the control experiment in Fig. 5B (no toxin was applied for 5 min, 10 M ␣-CTx ImII was then added for 5 min).
This suggests that 100 M ␣-CTx ImI does not inhibit ␣-CTx ImII binding to rat ␣7 nAChRs despite this concentration being  1. A, fragments of the ␣-CTx ImII gene were PCR amplified from genomic DNA and cDNA prepared from C. imperialis tissue collected in the Philippines. The nucleotide sequence in the vicinity of the region encoding mature toxin is shown with the predicted translation product. The putative mature peptide sequence is in bold letters. The N-terminal of the mature toxin is deduced by the presence of an Arg (R) in the larger precursor molecule that can act as a cleavage site for the release of the mature toxin. The C terminus is deduced from the presence of a stop codon; for the mature peptide we assume that the C-terminal Gly (G) is post-translationally removed, leaving Cys-12 amidated (amidation is represented by #). Fragments of the ␣-CTx ImI gene were also found in the pools of PCR products. A fragment of the ␣-CTx ImI gene in the vicinity of the region encoding mature toxin is shown, as is the known amino acid sequence of mature ␣-CTx ImI (9). B, the mature toxin sequences of ␣-CTx ImI and ␣-CTx ImII. The residues of ␣-CTx ImII that differ from ␣-CTx ImI are underlined.
about 520 times greater than the functional IC 50 . These, as well as the previous data, are consistent with the conclusion that ␣-CTx ImI and ␣-CTx ImII have little if any overlap in their high affinity binding sites on the ␣7 receptor. Nevertheless, occupancy of the two different sites by each ␣-conotoxin leads to functional block of the receptor.
Analogs of ␣-Conotoxins ImI and ImII-The peptides ␣-CTx ImI and ␣-CTx ImII are identical in 9 of 12 amino acids. Because they appear to target different sites on the ␣7 nAChR, we performed a structure/function study to identify which amino acids were critical for the difference in targeting. Of the three differences, those at positions 6 (Pro versus Arg) and 9 (Ala versus Arg) seem the most striking. At position 1 (Gly versus Ala), the two residues differ only by a methyl group. Additionally, the absence of a first loop Pro is very unusual in ␣-conotoxins (see Table II). The two analogs shown in Table II were thus synthesized (see "Experimental Procedures") and characterized.
Both analogs are significantly less functionally potent than the corresponding native peptides as determined by electrophysiological characterization of ␣7 nAChR inhibition (data not shown). Nevertheless, what is clearly indicated by the data is that the presence of a proline residue at position 6 is the major determinant of whether a peptide will compete with radiolabeled ␣-bungarotoxin for binding to the ␣7 receptor (Fig. 6). Thus, R6P ␣-CTx ImII is better at displacing ␣-bungarotoxin Pulses of ACh (1 s at 1-min intervals) gate currents in oocytes expressing rat ␣7 nAChRs. ␣-CTx ImII applied in a static bath (see "Experimental Procedures") results in a concentration-dependent reduction of the peak height of ACh-gated currents obtained simultaneously with the resumption of buffer flow. C, filled circles, the amplitudes of ACh-gated currents through oocyte-expressed rat ␣7 nAChR at different holding potentials are normalized such that the average amplitude of responses gated at Ϫ70 mV is Ϫ1. Open circles, the amplitudes of ACh-gated currents at different holding potentials following 5-min applications of 1 M ␣-CTx ImII are normalized such that the Ϫ70 mV response after toxin application is equal to the Ϫ70 mV response in the absence of toxin (i.e. Ϫ1). Data points and error bars, mean Ϯ S.E. for 3 to 6 measurements. than native ␣-CTx ImII. In contrast, replacement of the Pro-6 residue in ␣-CTx ImI with Arg results in failure to displace ␣-bungarotoxin even at a concentration of 100 M (compared with an EC 50 for native ␣-CTx ImI of 407 nM). Thus, the presence or absence of proline at position 6 determines whether or not these peptides preferentially bind to a site that overlaps with the ␣-bungarotoxin-binding site or another site. DISCUSSION We report the discovery and characterization of ␣-CTx ImII that has high sequence identity (9 of 12 amino acids) to ␣-CTx ImI (Fig. 1); both peptides are from the venom ducts of C. imperialis (9). ␣-CTx ImI is a specific competitive inhibitor of the ␣7 nAChR subtype (14,15). Given the close sequence similarity of ␣-CTx ImI and ␣-CTx ImII, it was not surprising that ␣-CTx ImII was also found to inhibit the ␣7 nAChR. However, most unexpectedly, the two closely related peptides appear to cause their similar functional effects by binding to different sites on the ␣7 nAChR.
␣-CTx ImII was found to be similar to ␣-CTx ImI in the behavioral effects observed when injected intracranially into mice; both peptides elicited complex seizures, weakness, tremors and, at higher doses, death. Similar behavior was also observed following intracerebral-ventricular injection of ␣-BgTx, another ␣7 nAChR inhibitor, into rats (14). In view of its homology to ␣-CTx ImI and the characteristic symptoms observed when it was injected into the central nervous system, ␣-CTx ImII was tested for its ability to inhibit ACh-gated currents in Xenopus oocytes expressing rat ␣7 nAChRs.

FIG. 3. Inhibition of 3-125 I-␣-BgTx binding to crude rat brain membranes (A) and ␣7-5-HT 3 chimera (B). ␣-CTx ImI (squares) and
␣-CTx ImII (circles) were added to compete with radiolabeled ␣-bungarotoxin as described under ''Experimental Procedures.'' The specific binding of 3-125 I-␣-BgTx at each conotoxin concentration is normalized to that obtained in the absence of conotoxins. For ␣-CTx ImI, the EC 50 is 1,560 nM (n H ϭ 0.59) on crude membranes and 407 nM (n H ϭ 0.71) on ␣7-5-HT 3 chimera. The ␣-CTx ImI data were fit to a curve and the ␣-CTx ImII data were fit to a straight line as described under "Experimental Procedures." Data points and error bars, mean Ϯ S.E. for 3 to 6 measurements.

FIG. 4. Functional competition between ␣-BgTx and ␣-conotoxins.
Xenopus oocytes expressing rat ␣7 nAChRs were voltageclamped as described under ''Experimental Procedures'' and their responses to 1-s ACh pulses at 1-min intervals were recorded. The peak heights are all normalized to the average of 5 peaks recorded prior to toxin application. A, bath perfusion was paused for 11 min. After 1 min, 1 l of ND96 was added; after 6 min, ␣-BgTx was added (black bar) in 1 l of ND96 to a final concentration of 100 nM; after 11 min, ND96 flow and ACh pulses were resumed. B, bath perfusion was paused for 11 min. After 1 min, ␣-CTx ImI in 1 l of ND96 was added (striped bar) to a final concentration of 100 M; after 6 min, ␣-BgTx was added (black bar) in 1 l of ND96 to a final concentration of 100 nM; after 11 min, ND96 flow and ACh pulses were resumed. C, the same protocol as in B was used except that ␣-CTx ImII was added (white bar) instead of ␣-CTx ImI. Data points and error bars, mean Ϯ S.D. for four repetitions.
␣-CTx ImII was found to inhibit the receptor with an IC 50 similar to that of ␣-CTx ImI (Fig. 2) when the toxins were tested using identical protocols.
The first surprising result was obtained when ␣-CTx ImII was tested in a competition assay with 3-125 I-␣-BgTx. As had been previously demonstrated by others (18), we found that ␣-CTx ImI competed with ␣-BgTx for binding to the receptor. In contrast, ␣-CTx ImII did not appreciably displace ␣-BgTx binding in the concentration range tested. These results for ␣-CTx ImI and ␣-CTx ImII were obtained with both rat brain ␣7 FIG. 5. Functional competition between ␣-CTx ImI and ␣-CTx ImII. Xenopus oocytes expressing rat ␣7 nAChRs were voltage-clamped as described under ''Experimental Procedures'' and their response to brief ACh pulses at 1-min intervals were recorded. The peak heights are all normalized to the average of 5 peaks recorded prior to toxin application. A1, bath perfusion was paused for 6 min. After 1 min, ␣-CTx ImI in 1 l of ND96 was added to a final concentration of 100 M (striped bar); after 6 min, ND96 flow and ACh pulses were resumed. A2, the same protocol as in A1 was applied except that ␣-CTx ImII was added to a final concentration of 10 M (white bar). B, bath perfusion was paused for 11 min. After 1 min, 1 l of ND96 was added; after 6 min, ␣-CTx ImII was added in 1 l of ND96 to a final concentration of 10 M (white bar); after 11 min, ND96 flow and ACh pulses were resumed. C, bath perfusion was paused for 11 min. After 1 min, ␣-CTx ImI was added in 1 l of ND96 to a final concentration of 100 M (striped bar); after 6 min, 1 l of ND96 was added; after 11 min, ND96 flow and ACh pulses were resumed. D, bath perfusion was paused for 11 min. After 1 min, ␣-CTx ImI was added in 1 l of ND96 to a final concentration of 100 M; after 6 min, ␣-CTx ImII was added in 1 l of ND96 to a final concentration of 10 M; after 11 min, ND96 flow and ACh pulses were resumed. Data points and error bars, mean Ϯ S.D. for four repetitions for B, C, and D. A1 and A2 are representative traces. nAChRs (Fig. 3A) as well as rat ␣7-5-HT 3 chimeras (Fig. 3B). Furthermore, experiments using rat ␣7 nAChRs expressed in oocytes (Fig. 4) demonstrated that preincubation with ␣-CTx ImI prevents ␣-BgTx from binding to its functionally relevant site, a result consistent with competitive antagonism, and previously shown by others (15). On the other hand, ␣-CTx ImII had only a very weak effect on ␣-BgTx inhibition of oocyteexpressed receptor, consistent with a different binding site.
The binding site for competitive antagonists of nAChRs is located at the interfaces between subunits that make up the receptor (reviewed in Refs. 19 and 20). The site includes contacts in three conserved loops from one subunit (loops A, B, and C) that make up the "ϩ" face, and four loops from an adjacent subunit (loops I to IV) that make up the "Ϫ" face. The binding of ␣-CTx ImI to the ␣7 nAChR is affected by mutations in or near loops A, B, and C, and II and III (17,18). The ␣-BgTx site on the ␣7 nAChR has also been mapped to the A, B, and C loops (21), and a loop II mutation causes a minor reduction in ␣-BgTx affinity (22). This and other evidence are consistent with ␣-CTx ImI-binding sites in ␣7 nAChRs overlapping with ACh and ␣-BgTx-binding sites, and being at subunit interfaces.
The data in Fig. 5 suggest that ␣-CTx ImI and ␣-CTx ImII do not bind to the same site at a subunit interface. Assuming a potential five identical subunit interfaces in the ␣7 nAChR pentamer, and that occupation of even one site by ␣-CTx ImI results in inhibition of the receptor, then the concentration of ␣-CTx ImI that occupies half the potential sites, K d , is related to the functional IC 50 by IC 50 /K d ϭ 0.15 (3), i.e. K d ϭ IC 50 ϫ 6.67. The IC 50 of ␣-CTx ImI on the ␣7 nAChR is 191 nM (Fig. 2). Therefore, 100 M ␣-CTx ImI (see Fig. 5) would clearly occupy most subunit interface-binding sites on the ␣7 nAChR (assuming these are identical) and should significantly reduce binding of ␣-CTx ImII if ␣-CTx ImI and ␣-CTx ImII share a binding site. The ␣-CTx ImII-binding site awaits definitive characterization; however, several possibilities are outlined below.
Because the primary structures of ␣-CTx ImI and ␣-CTx ImII are so similar, and because they share the characteristic ␣-conotoxin disulfide framework, it seems possible that ␣-CTx ImII also binds to the interface between ␣7 subunits. In this case, the inability of ␣-CTx ImII to compete with ␣-BgTx or ␣-CTx ImI might be explained by the following models.
One potential explanation for the results is that ␣-CTx ImI and ␣-CTx ImII can simultaneously bind at a single subunit interface by positioning differently within the cleft at different microsites. In fact, ␣-BgTx appears to make more contacts with the ϩ face than with the Ϫ face at ␣7 nAChR subunit interfaces (21,22). It is possible, for example, that ␣-CTx ImII binds predominantly to the Ϫ face and is thus unable to displace ␣-BgTx, whereas ␣-CTx ImI, because of many contacts in the ϩ face, disrupts many ␣-BgTx-receptor interactions, and is thus able to compete with this toxin.
An alternative explanation is based on the work of Green and co-workers (23), who have shown that despite amino acid sequence identity, the subunits of a functional ␣7 nAChR receptor are not identical. Evidence was presented that the functional ␣7 nAChR complex requires a mixture of ␣7 subunits that are in at least two states that differ in their N-terminal domain conformation and the oxidation state of Cys residues (23). A direct consequence of this nonidentity is that putative ligand-binding sites located between subunits become distinguishable. One possibility is that one type of interface between ␣7 subunits is the ␣-CTx ImI and ␣-BgTx-binding site, whereas another type of subunit interface does not bind ␣-BgTx, but is the ␣-CTx ImII target site. Bertrand and co-workers (24) have shown that for the competitive ␣7 nAChR antagonist MLA there are five identical binding sites. This is not necessarily incompatible with a heterogeneous interface model. MLA may recognize structural elements at interfaces that are unaffected by the state of flanking subunits. However, other ligands might FIG. 6. Competition binding of ␣-conotoxin analogs. A, inhibition of 3-125 I-␣-BgTx binding to ␣7-5-HT 3 chimera by ␣-CTx ImI (squares) and P6R ␣-ImI (circles). The specific binding of 3-125 I-␣-BgTx at each conotoxin or mutant toxin concentration is normalized to specific 3-125 I-␣-BgTx binding in the absence of peptide. The ␣-CTx ImI data was fit to a curve and the P6R ␣-CTx ImI data was fit to a straight line as described under "Experimental Procedures." For ␣-CTx ImI, the EC 50 is 407 nM and the n H is 0.71. B, inhibition of 3-125 I-␣-BgTx binding to rat ␣7-5-HT 3 chimeras by ␣-CTx ImII (circles) and R6P ␣-CTx ImII (squares) was determined as described under "Experimental Procedures." The specific binding of 3-125 I␣-BgTx at each conotoxin or mutant toxin concentration is normalized to specific 3-125 I-␣-BgTx binding in the absence of peptide. The R6P ␣-CTx ImII data was fit to a curve and the ␣-CTx ImII data was fit to a straight line as described under ''Experimental Procedures.'' For R6P ␣-CTx ImII, the EC 50 is 19.3 M and the n H is 0.78. Data points and error bars, mean Ϯ S.E. for 3 to 6 measurements. Although ␣-CTx ImI and ␣-CTx ImII show extensive sequence homology, it is possible that ␣-CTx ImII binds to a nonsubunit-interface site on the receptor. For example, it might bind extracellular regions of the receptor that are not in the N-terminal ACh-binding domain, i.e. the extracellular loop that occurs between two transmembrane helices of the ␣7 nAChR or the C-terminal extracellular region. ␣-CTx ImII could also potentially bind to nonsubunit interface regions on the N-terminal ACh-binding domain or the channel pore; however, because ␣-CTx ImII block is not voltage dependent, this supports the model that it is not an open channel blocker.
The experiments with analogues suggest that although ␣-CTx ImI and ␣-CTx ImII have very similar sequences, the amino acid residue at position 6 (Pro in ␣-CTx ImI, Arg in ␣-CTx ImII) is critical in determining where they bind on the ␣7 nAChR. Relative to wild-type ␣-CTx ImI, P6R ␣-CTx ImI is a very poor competitor of ␣-BgTx binding to ␣7-5-HT 3 chimera. In contrast, R6P ␣-CTx ImII has an enhanced ability to compete with ␣-BgTx compared with wild-type ␣-CTx ImII. Because the two native toxins apparently target different sites, a key determinant for selectivity is which amino acid is present at position 6.
Additional information about interactions of ␣-CTx ImI and ␣-CTx ImII with their distinct binding sites can be derived from the analog toxin data if one assumes the initial Gly and Ala residues in the two toxins are functionally equivalent. In this case, the P6R ␣-CTx ImI analog is equivalent to R9A ␣-CTx ImII and the R6P␣-CTx-ImII analog is equivalent to A9R ␣-CTx-ImI. Because P6R ␣-CTx ImI does not compete with ␣-BgTx for binding to the ␣7 nAChR, this strongly suggests that R9A ␣-CTx ImII would be like ␣-CTx ImII and also not compete with ␣-BgTx. Because R6P ␣-CTx ImII has some ability to compete ␣-BgTx but is not as potent as ␣-CTx ImI, this strongly suggests that A9R ␣-CTx ImI would compete with ␣-BgTx for binding to the ␣7 nAChR but would be a less potent competitor than ␣-CTx ImI. Taken together, these observations imply that the residues at position 9 in ␣-CTx ImI and ␣-CTx ImII are not critical in determining whether the ␣-CTx ImI or ␣-CTx ImII site is targeted, but are important for ensuring optimal affinity of ␣-CTx ImI and ␣-CTx ImII for their respective sites.
The discovery of ␣-CTx ImII reveals that C. imperialis has two toxins that inhibit the rat ␣7 nAChR, and that these act at different sites. Although caution must be applied when extrapolating this observation to the native prey, it suggests that C. imperialis may target marine worms with both ␣-CTx ImI and ␣-CTx ImII, which may bind to different sites on an ''␣7-like'' receptor in native prey. This would represent a second example of cone snail venom containing two distinct antagonists of the same nAChR. It was previously demonstrated that Conus purpurascens produces two structurally unrelated nAChR antagonists, a competitive ␣A-conotoxin and a noncom-petitive -conotoxin (reviewed in Ref. 1). The present case is different, however, in that the toxins are both ␣-conotoxins that are very closely related to each other in sequence. A caveat that must be applied to this model is that natural, venom-derived ␣-CTx ImII may possess post-translational modifications that were not incorporated in the synthetic peptide used in this study. The native toxin may thus differ from the synthetic molecule in its functional properties, i.e. it may not target an ␣7-like receptor at all. On the other hand, post-translational modification in ␣-conotoxins isolated from venom have so far been limited to C-terminal amidation and tyrosine sulfation (␣-CTx ImI and ␣-CTx ImII lack tyrosine residues).
Previously, it has been shown that very minor changes in the intercysteine amino acid sequences of conotoxins can drastically affect their specificity. The toxins ␣-CTx PnIA and ␣-CTx PnIB from Conus pennaceus are different in only 2 of 16 amino acids, but preferentially block ␣3␤2 and ␣7 nAChRs, respectively (3). The discovery of ␣-CTx ImII illustrates that in C. imperialis, minor differences between two toxins result in molecules that target, not distinct receptor subtypes, but distinct sites on a single nAChR subtype.