Neuronally Selective μ-Conotoxins from Conus striatus Utilize an α-Helical Motif to Target Mammalian Sodium Channels*

μ-Conotoxins are small peptide inhibitors of muscle and neuronal tetrodotoxin (TTX)-sensitive voltage-gated sodium channels (VGSCs). Here we report the isolation of μ-conotoxins SIIIA and SIIIB by 125I-TIIIA-guided fractionation of milked Conus striatus venom. SIIIA and SIIIB potently displaced 125I-TIIIA from native rat brain Nav1.2 (IC50 values 10 and 5 nm, respectively) and muscle Nav1.4 (IC50 values 60 and 3 nm, respectively) VGSCs, and both inhibited current through Xenopus oocyte-expressed Nav1.2 and Nav1.4. An alanine scan of SIIIA-(2–20), a pyroglutamate-truncated analogue with enhanced neuronal activity, revealed residues important for affinity and selectivity. Alanine replacement of the solvent-exposed Trp-12, Arg-14, His-16, Arg-18 resulted in large reductions in SIIIA-(2–20) affinity, with His-16 replacement affecting structure. In contrast, [D15A]SIIIA-(2–20) had significantly enhanced neuronal affinity (IC50 0.65 nm), while the double mutant [D15A/H16R]SIIIA-(2–20) showed greatest Nav1.2 versus 1.4 selectivity (136-fold). 1H NMR studies revealed that SIIIA adopted a single conformation in solution comprising a series of turns and anα-helical motif across residues 11–16 that is not found in larger μ-conotoxins. The structure of SIIIA provides a new structural template for the development of neuronally selective inhibitors of TTX-sensitive VGSCs based on the smaller μ-conotoxin pharmacophore.

1.9 are implicated in neuropathic pain states (4,5), the TTX-S Na v 1.7 is implicated in allodynia and neuropathic pain (6 -10), and the TTX-S Na v 1.3 is up-regulated in inflammatory pain (5). To learn more about the different roles played by each of the VGSC subtypes in normal and disease states requires new inhibitors with improved VGSC subtype selectivity.
Inhibitors of TTX-S VGSCs include tetrodotoxin (TTX) from puffer fish, the saxitoxins (STXs) from marine dinoflagellates, and the -conotoxins from the venom of cone snails. These toxins act competitively at Site 1 in the P-loop region of the ion-conducting pore of the ␣-subunit of the VGSC (3,11). While TTX and STX act with nanomolar potency across the six TTX-S VGSCs (Na v 1.1-1.4, 1.6, and 1.7), -conotoxins are more discriminating. Currently 12 -conotoxins have been identified from nine species of fish hunting cone snails (Table  1). -Conotoxins GIIIA, GIIIB, and GIIIC from Conus geographus selectively target the skeletal muscle Na v 1.4, PIIIA from Conus purpurasence (12)(13)(14) and TIIIA from Conus tulipa (15) target Na v 1.2 and 1.4, and -conotoxins SmIIIA from Conus stercusmuscsarum (16,17), SIIIA from Conus striatus (18,19), and KIIIA from Conus kinoshitai (18) target amphibian TTX-R sodium channels. More recently, KIIIA has also been shown to inhibit TTX-S sodium channels in mouse DRG and rat TTX-S sodium channels expressed in Xenopus oocytes (20). Lastly, -conotoxins CnIIIA, CnIIIB from Conus consors, CIIIA from Conus catus, and MIIIA from Conus magus were found to inhibit TTX-R and TTX-S sodium channels in frog DRG but not rat or mouse DRG neurons (21). Most reported -conotoxins have a conserved arginine in loop 2, which is essential for high affinity interactions at TTX-S sodium channels (15,(22)(23)(24)(25). In contrast, SIIIA and KIIIA have a much shorter loop 2 and a lysine instead of an arginine in the corresponding position. Recent structure-activity studies on TIIIA (15) and KIIIA (20) have confirmed that the arginine in loop 2 is crucial for TIIIA affinity at both neuronal and skeletal muscle sodium channels, whereas the structurally equivalent lysine was only important for affinity at the skeletal muscle VGSC. Three-dimensional structures of -conotoxins GIIIA, GIIIB, PIIIA, TIIIA, and SmIIIA have been determined by NMR (14,15,17,26,27). These structures reveal a similar fold for the -conotoxins despite high sequence divergence. At present, structures of -conotoxins SIIIA and KIIIA are modeled from SmIIIA (18,20,21).
This study reports the isolation and structure-activity of -conotoxins SIIIA and SIIIB from the milked venom of C. striatus. To define aspects of their structure that contribute to binding to mammalian VGSCs, we determined the NMR structure of SIIIA. Distinct from other -conotoxins, SIIIA adopts a single conformation in solution that includes an ␣-helical motif not previously seen in other members of this class. This structural change was associated with a significant pharmacophore shift from the dominant arginine residue in loop 2 of the larger -conotoxins to a series of five similarly important residues, including four associated with the ␣-helical region, in the smaller SIIIA. Interestingly, a number of SIIIA analogues had significantly enhanced neuronal selectivity, opening the way to the development of subtype-selective neuronal VGSC inhibitors.

EXPERIMENTAL PROCEDURES
Isolation of -Conotoxins SIIIA and SIIIB-Crude C. striatus milked venom was separated into 80 1-min fractions by semipreparative RP-HPLC using a Vydac C18 column (250 ϫ 10 mm, 5 ) eluted at 3 ml/min with a 1% gradient from 100% A to 80% B over 80 min (solvent A 0.05% trifluoroacetic acid, solvent B 90% ACN ϩ 0.045% trifluoroacetic acid). The eluant was monitored at 230 nm and 125 I-TIIIA-guided fractionation (15) of the crude milked venom RP-HPLC fractions of C. striatus used to isolate SIIIA and SIIIB to homogeneity. The sequences of SIIIA and SIIIB (Table 1) were determined by Edman sequencing only after pyroglutamate aminopeptidase digestion, indicating that the N-terminal residue of both these conotoxins was a pyroglutamate. The observed masses for native SIIIA and SIIIB (2206.8 and 2120.7 Da, respectively) were consistent with the Edman sequencing results for peptides with an amidated C terminus. Co-injected synthetic and native SIIIA and SIIIB (2:1 ratio) co-eluted on a Vydac C18 reversed-phase HPLC column (250 ϫ 4.6 mm, 5 ) eluted with a 1% gradient from 100% A to 50% B over 50 min, confirming the identity of the native peptides.
Peptide Synthesis--Conotoxins SIIIA and SIIIB and analogues were prepared by Boc chemistry (28) using methods described previously (14). Side-chain protecting groups chosen were Arg(Tos), Asp(OcHex), Lys(CIZ), Ser(Bzl), Asn(Xan), His(DNP), and Cys(p-MeBzl). The crude reduced peptides were purified by preparative RP-HPLC using a 1% gradient (100% A to 80% B over 80 min) and UV monitoring (230 nm), and oxidized at 0.02 mM in either aqueous 0.33 M NH 4 OAc/0.5 M GnHCl/2 M NH 4 OH, or 100 mM ammonium bicarbonate in the presence of oxidized and reduced glutathione (29). Oxidized peptides were purified by preparative RP-HPLC. Peptides were quantified initially by triplicate amino acid analysis to create an external reference standard for HPLC quantitation of each peptide (30). Mass spectra were acquired on a PE-Sciex API III triple quadrupole electrospray mass spectrometer in positive ion mode (m/z 500 -2000 at 0.1-0.2 Da steps, declustering potentials of 10 -90 V, and dwell times of 0.4 -1.0 s). Data were deconvoluted using MacSpec 3.2 (Sciex, Canada) to obtain the molecular weight from the multiply charged ion species. Mass spectrometry was used to confirm purity and to monitor peptide oxidation (data not shown).
Radioligand Binding Studies-125 I-TIIIA (prepared using IODOGEN) radioligand binding studies were performed using rat brain and rat skeletal muscle preparations to measure Na v 1.2 and Na v 1.4 affinity, as previously described (15). Nonlinear regressions (Hillslope-1) were fitted to triplicate data obtained for each experiment using Prism software (GraphPad, San Diego, CA).
Recording of Depolarization-activated Sodium Ion Currents in Rat DRG Neurons-Neonatal rat DRG neurons were prepared as described previously (31). Whole cell TTX-S and TTX-R sodium ion currents were recorded in using an Axopatch 200A patch clamp amplifier (Molecular Devices Corp.). Series resistance was routinely compensated by 70 -80%. Capacitive and leakage currents were digitally subtracted using a ϪP/6 pulse protocol. Internal pipette solution contained (in mM): NaCl 10, CsF 130, CsCl 10, EGTA 10, HEPES-CsOH 10, pH 7.2, 290 -300 mOsM (with sucrose). The bath solution contained (in mM): NaCl 50, KCl 5, MgCl 2 1, CaCl 2 1, glucose 10, tetraethylammonium (TEA)-Cl 90, HEPES-TEA-OH 10, pH 7.35. TTX-resistant sodium ion currents were recorded from small DRG neurons (Յ25 m diameter) in the presence of 300 nM TTX. Large neurons (Ն40 m diameter), in which the sodium ion current exhibited fast activation and inactivation kinetics and was mostly TTX-sensitive (Ͼ90%), were used to study toxin effects on TTX-S sodium ion current after complete washout of TTX block. Toxins were diluted in extracellular solution and added via a gravity-fed perfusion system.
Xenopus Oocyte Experiments-Depolarization-activated sodium ion currents were recorded from Xenopus oocytes expressing various VGSC ␣-subunit subtypes using a two-electrode (virtual ground circuit) voltage clamp at room temperature (20 -23°C) with a GeneClamp 500B amplifier and pCLAMP 8 software (Axon Instruments Inc, Union City, CA). Bath solution contained (in mM) 100 NaCl, 2 KCl, 1 MgCl 2 , 0.3 CaCl 2 , and 20 HEPES-NaOH, pH 7.5. Data were low pass-filtered at 2 kHz, digitized at 10 kHz, and leak-subtracted on-line using a ϪP/6 protocol and analyzed off-line. Data were analyzed using Clampfit 8 software (Axon Instruments Inc., Union City, CA). Rat Na v 1.2 and rat Na v 1.3 was a gift from A. Goldin (University of California), human Na v 1.5 a gift from R. Kass (Columbia University), and human Na v 1.7 a gift from N. Klugbauer (Teknischen Universität, Munich). Toxins were added by direct bath application (static bath) followed by extensive mixing to reach the final concentrations stated. The steady-state block was reached after ϳ8 min of toxin exposure, and washout of SIIIA was followed for 20 -30 min.

H Nuclear Magnetic Resonance (NMR) Spectroscopy-All
NMR experiments were recorded on a Bruker ARX 500 spectrometer equipped with a z-gradient unit, a Bruker DMX 750, or a Bruker 600 Avance spectrometer equipped with a x,y,zgradient unit. Peptide concentrations were ϳ2 mM. SIIIA and SIIIB were examined in 95% H 2 O/5% D 2 O (pH 3.0, 275-298 K) and in 100% D 2 O (260 -293 K). SIIIA analogues were examined in 90% H 2 O/10% D 2 O, pH 3.0. For SIIIA, 1 H NMR experiments recorded included NOESY, (mixing times 120, 300, and 400 ms), TOCSY (mixing time 80 ms), double-quantum filtered (DQF)-COSY, and E-COSY in 100% D 2 O. Slowly exchanging amide protons were detected by acquiring one-dimensional NMR spectra immediately after dissolving the peptide in 100% D 2 O at pH 3.5. For all other analogues, TOCSY experiments with a mixing time of 80 ms were acquired. Additional NOESY experiments were only run if ambiguities arose during assignment. All spectra were run over 6024 Hz (500 MHz), 7184 Hz (600 MHz), or 8192 Hz (750 MHz) with 4K data points, 400 -512 FIDs, 8 -64 scans, and a recycle delay of 1-2 s. The solvent was suppressed using the WATERGATE sequence (32). Spec-tra were processed using UXNMR as described previously (29) and using Aurelia, subtraction of background was used to minimize T 1 -noise. Chemical shift values of SIIIA and analogues were obtained in 90% H 2 O, 10% D 2 O and referenced internally to DSS at 0.00 ppm. Secondary H␣ shifts were measured and compared with values for random coil shifts (33).
SIIIA was assigned in sparky (34), and distance information obtained by integration of a 300 ms mixing time NOESY spectrum. Backbone dihedral restraints were derived from 3 J HN-H␣ coupling constants from a DQF-COSY spectrum or a one-dimensional 1 H NMR spectrum. The dihedral angle restraints included were restrained to 120 Ϯ 30°for 3 J HN-H␣ Ͼ8 Hz and Ϫ60 Ϯ 30°for 3 J HN-H␣ Ͻ5.8 Hz. 1 dihedral angels were determined from intraresidue nuclear Overhauser effect (NOE) and 3 J H␣-H␤ coupling patterns for an E-COSY spectrum. Initial structures were generated using dyana (35), and final structures were calculated in explicit water with CNS (36) as previously described (37). An initial 50 structures were calculated, and the 17 lowest overall energy structures selected to represent the solution structure of SIIIA.

RESULTS
Isolation of -Conotoxins SIIIA and SIIIB-To identify -conotoxins selective for neuronal VGSCs, fractionated milked C. striatus venom was assayed for the ability to displace 125 I-TIIIA from rat brain sodium channels (15). Three active fractions with retention times of 20, 21, and 24 min (Fig. 1) were purified to homogeneity, digested using pyroglutaminase, and Edman-sequenced to reveal three peptides with a CC-C-C-CC motif characteristic of -conotoxins (38). Fraction 24 contained the sequence of TIIIA, previously identified in C. tulipa by PCR (15), fraction 20 the sequence for SIIIA, while fraction 21 contained a sequence for a new -conotoxin we named SIIIB (Table 1). SIIIA and SIIIB each comprised 20 residues with a net charge of ϩ2 and ϩ3, respectively. Mass spectrometry and coelution studies of the native and synthetic peptides confirmed the sequences and C-terminal amidation of both peptides (data not shown).
Radioligand Binding Studies-Synthetic -conotoxin SIIIA and SIIIB each potently displaced 125 I-TIIIA from VGSCs in rat brain and rat skeletal muscle (Fig. 2). The pIC 50 values for SIIIA and SIIIB and selected additional -conotoxins are summarized in Table 2. Similar to TTX and STX, SIIIA had a small but   Effects of SIIIA on Whole Cell Sodium Ion Currents-The effects of -conotoxin SIIIA was investigated on both TTX-S and TTX-R sodium ion current in neonatal rat DRG (31). SIIIA at 3 M produced no detectable inhibition of either TTX-S or TTX-R sodium ion currents in rat DRG neurons (data not shown).
Effects of -Conotoxins on Oocyte-expressed VGSC Subtypes-SIIIA was examined for inhibition of sodium ion current through VGSC subtypes expressed in Xenopus oocytes. SIIIA at 3 M caused 95% inhibition of Na v 1.2 current, 45% inhibition of Na v 1.3 current (Fig. 3A), but no detectable inhibition of sodium ion current through Na v 1.5, Na v 1.7, or Na v 1.8 (data not shown). The IC 50 values for SIIIA was 790 nM (95% CI 630 -990 nM) at Na v 1.2, 330 nM (95% CI 300 -360 nM) at Na v 1.4, and 5.3 M (95% CI 3.6 -7.6 M) at Na v 1.3 (Fig. 3B). SIIIB at 3 M caused 93% inhibition of Na v 1.2 current and 10% inhibition of Na v 1.7 current (data not shown) but was not tested on other VGSC subtypes. The inhibitory effect of SIIIA was not reversed upon washout (Fig. 3C), in contrast to TIIIA which was fully reversible upon washout at Na v 1.2 (15).
SIIIA was examined by NMR spectroscopy under aqueous conditions over a range of different temperatures to confirm proton assignments (data not shown). NMR solution structures were then calculated based on a total of 288 distance restraints derived from 102 intraresidue, 93 sequential, 89 mediumand long-range NOEs, 4 hydrogen bond restraints defining 2 hydrogen bonds, and 12 and one 1 dihedral angle restraints. A total of 46 of 50 structures converged to a consensus structure. These structures con-   2 NOE violations greater than 0.2 Å and no dihedral violations greater than 3°. The local and medium range NMR data that provide information on the secondary structure of SIIIA are given in supplemental Fig. S1A. The presence of several H␣-NH iϩ2 , NH-NH iϩ2 , H␣-NH iϩ3 , and H␣-NH iϩ3 NOEs suggested the presence of several turns, and identified that a helical secondary structure existed across residues 11-16. This ␣-helical motif was supported by a number of i-iϩ3 and i-iϩ4 NOEs and the presence of small 3 J HN-H␣ dihedral angles and negative secondary H␣-chemical shifts in this region of the peptide (Fig. 5).
The 17 lowest energy structures of SIIIA are superimposed globally in Fig. 6A, and a representative ribbon structure highlighting secondary structural motifs is shown in Fig.  6B. Overall, the structures are welldefined with a backbone root meansquare deviation (RMSD) of 0.21 across residues 3-20 and 0.35 across residue 1-20. The angular order parameters for and angles are also well ordered and Ͼ0.92 across residues 6 -20. The slightly disordered N-terminal residues do not appear to arise from conformational exchange or structural flexibility, because no line-broadening was observed for these residues, and are more likely due to a lack of NOEs to loop 1, which contains two Gly residues. Unlike the longer loop 1 in other -conotoxins, which protrudes outward, loop 1 of SIIIA comprises a tightly folded ␤-turnlike motif that is positioned to interact with loop 3 (see Fig. 7). Analysis of the surface exposure of each residue illustrates this structural shift, with loop 1 and the first two residues in loop 2 of SIIIA being relatively buried compared with the equivalent region of TIIIA (15), whereas Lys-11, Trp-12, Arg-14 within the ␣-helical region, together with Arg-18, are relatively exposed to solvent (supplemental Fig. S1B). As expected, all Cys residues are buried in the core of the molecule. Comparing the NMR structures of SIIIA and TIIIA reveals that the key binding determinants of SIIIA are distributed across the ␣-helical motif (Lys-11, Trp-12, Arg-14), and Arg-18 with each contributing similarly to affinity, whereas a single arginine (Arg-14) is the key determinant of TIIIA affinity (Fig. 7). In supplemental Fig. S2, these comparisons are extended to PIIIA, GIIIA, and SmIIIA.

DISCUSSION
The venoms of cone snails provide a rich source of -conotoxin inhibitors of VGSCs, including several recently isolated -conotoxins with a smaller loop 2 (18,19). In this study we used assay-directed fractionation of milked C. striatus venom to identify a new -conotoxin SIIIB, together with the recently reported SIIIA (18) and TIIIA (15). Interestingly, the smaller -conotoxins from C. striatus exploit a new pharmacophore  that allows the development of peptides with selectivity for Na v 1.2 and potentially other TTX-S subtypes.
SIIIA and SIIIB both displaced 125 I-TIIIA and [ 3 H]STX binding to rat brain (Na v 1.2) and rat skeletal muscle membrane (Na v 1.4), indicating they act at site 1 of the VGSC like the larger -conotoxins PIIIA (24), TIIIA (15), GIIIA, and GIIIB (38). SIIIA also inhibited current through oocyte-expressed Na v 1.2 and Na v 1.4 at nanomolar potencies, confirming it acts as a pore blocker of mammalian VGSCs. Like KIIIA (20), inhibition of Na v 1.2 was not reversed by toxin washout. SIIIA also inhibited current through Na v 1.3 at higher concentrations, showing that these smaller -conotoxins have the potential to target additional neuronal subtypes. -Conotoxins SIIIA, KIIIA, SmIIIA, CnIIIA, and CIIIA all potently inhibit frog TTX-R sodium ion current (18,21). However, we were unable to detect any inhibition of TTX-R or TTX-S sodium ion current in rat DRG neurons by SIIIA, in contrast to an earlier study showing inhibition of the TTX-R current in adult rat DRG by SIIIA (19). The inability of SIIIA to inhibit TTX-S sodium ion current in neonatal rat DRG indicates that Na v 1.2 and 1.3 are not expressed at significant levels in these neurons.
Replacing His-16 with alanine also had a dramatic effect on SIIIA and KIIIA (20) affinity. However, H␣-chemical shift changes indicated that this altered affinity arises, at least in part, from changes in structure of the C-terminal half of the peptide that potentially disrupt the ␣-helix and thus the orientations of the major binding determinants required for high affinity interactions at VGSCs. Interestingly, the potent and Na v 1.2-selective [D15A/H16R]SIIIA-(2-20) did not show a change in secondary H␣-chemical shifts, apart from minor local shifts at residue 15, indicating that Arg-16 might play a similar structural role to His-16.
Comparing the NMR solution structure of SIIIA with TIIIA it is apparent they adopt distinct conformations in solution (Fig.  7). Superimposition of loop 2 residues 8 -13 of SIIIA with the corresponding residues 11-16 of TIIIA and PIIIA and gave high RMSDs of 1.08 1.53, respectively. This affect is even more dramatic when SIIIA and SmIIIA are compared across loops 2 and 3 (RMSD of 2.58), despite having almost identical sequences. In contrast, comparing the same residues between PIIIA and TIIIA gave a lower RMSD of 0.64. Examining the secondary structures of the different -conotoxins provides an explanation. GIIIA, GIIIB, PIIIA, SmIIIA, and TIIIA each comprise a series of turns and loops (14,15,17,25,27,42), with a ␤-hairpin seen in the N-terminal part of GIIIA and GIIIB and the minor conformation of PIIIA (14). These turns in GIIIB, SmIIIA, PIIIA, and TIIIA produce a 3 10 -helix across residues 13-22, 13-20, 13-17, and 12-15, respectively (numbering specific to each conotoxin) (14,15,17,25,27,42) compared with the ␣-helix found in SIIIA. Previously described structures of SIIIA and KIIIA (18,20,21), as well as CnIIIA, CnIIIB, CIIIA, and MIIIA (21), were modeled from the NMR structure of SmIIIA (17). However, the NMR solution structures of SIIIA reveals that loop 1 of SIIIA is not extended as seen in TIIIA (Fig. 7) and other larger -conotoxins (supplemental Fig. S2) but instead folds in toward the ␣-helical region. This altered conformation, supported by a series of long-range NOEs between loops 1 and 3, within loops 2 and 3, and between loops 2 and 3, permits a more extended C-terminal region not seen in the larger -conotoxins. Interestingly, these structural shifts allow Lys-11, Arg-14, and Arg-18 of SIIIA and Arg-14, Arg-17, and Lys-6 of TIIIA to adopt similar relative positions (Fig. 7) that might allow binding to a negatively charged crevice in or near the mouth of the VGSC. A comparison of the sequences of KIIIA, SIIIA, and SIIIB reveals they are similar across loops 2 and 3, and include all important binding determinants, suggesting they likely have similar binding modes and similar conformations in solution.
The present study of SIIIA and SIIIB extends our knowledge of the structure-activity relationship of smaller -conotoxins and defines key residues contributing to Na v 1.2 selectivity. SIIIA contains an ␣-helix across residues 11-16 unlike previously determined -conotoxin NMR structures. The reduction in size of loop 1 from five residues in the larger -conotoxins to three in smaller -conotoxins like SIIIA appears to underlie a shift from a 3 10 -helical motif to the ␣-helix. SIIIA targets Na v 1.2, 1.3, and 1.4, whereas the even smaller KIIIA additionally targets Na v 1.1, 1.6, and 1.7 (20), indicating that the size of loop 1 can influence not only the structure but also the selectivity of -conotoxins. The range of VGSC subtypes inhibited by these smaller -conotoxins opens the door to the development of peptides with novel subtype specificity and analgesic potential.