Structures of (cid:1) O-conotoxins from Conus marmoreus INHIBITORS OF TETRODOTOXIN (TTX)-SENSITIVE AND TTX-RESISTANT SODIUM CHANNELS IN MAMMALIAN SENSORY NEURONS*

, The (cid:1) O-conotoxins are an intriguing class of conotoxins targeting various voltage-dependent sodium channels and molluscan calcium channels. In the current study, we have shown MrVIA and MrVIB to be the first known peptidic inhibitors of the transient tetrodotoxin-resistant (TTX-R) Na (cid:2) current in rat dorsal root ganglion neurons, in addition to inhibiting tetrodotoxin-sensitive Na (cid:2) currents. Human TTX-R sodium channels are a therapeutic target for indications such as pain, highlighting the importance of the (cid:1) O-conotoxins as potential leads for drug development. Furthermore, we have used NMR spectroscopy to provide the first structural information on this class of conotoxins. MrVIA and MrVIB are hydrophobic peptides that aggregate in aqueous solution but were solubilized in 50% acetoni-trile/water. The three-dimensional structure of MrVIB consists of a small (cid:3) -sheet and a cystine knot arrange-ment of the three-disulfide bonds. It contains four backbone “loops” between successive cysteine residues that are exposed to the solvent to varying degrees. The largest of these, loop 2, is the most disordered part of the molecule, most likely due to flexibility in solution.

Conotoxins are of great interest as ligands in neuroscience and are valuable leads in drug design based on their specificity and potency for therapeutically relevant receptors and ion channels (1,2). Several subtypes of one such ion channel, the sodium channel, have been implicated in clinical states, including pain (3)(4)(5)(6), stroke (7,8), and epilepsy (9,10). Consequently, conotoxins targeting voltage-gated sodium channels are not only important for further characterization of the channel but may themselves be useful as therapeutics.
The O-conotoxins, MrVIA and MrVIB from the venom of Conus marmoreus (11), are a particularly interesting family of conotoxins, because they block voltage-sensitive sodium channels (VSSC) 1 but also interact with the calcium channel in molluscan neurons. MrVIA and MrVIB have been shown to block tetrodotoxin (TTX)-resistant (TTX-R) Na ϩ current in molluscan neurons (11), the TTX-sensitive (TTX-S) Na ϩ current through Na v 1.2 expressed in Xenopus oocytes, and the TTX-S Na ϩ current in hippocampal neurons (12). Despite the high sequence conservation between MrVIA and MrVIB (only three residues differ) the peptides behave differently at the calcium channel in molluscan neurons. At low micromolar concentrations, MrVIA acts as a calcium channel agonist, whereas MrVIB acts as an antagonist. However, at higher doses both toxins block the fast-inactivating calcium channel. This crosschannel activity observed for O-conotoxins has been suggested to represent an "intermediate" variant of conotoxin in the diversification of one conotoxin structural family (13).
Comparison of the O-conotoxins with other conotoxins targeting either sodium or calcium channels highlights some striking differences as well as similarities. Although the -conotoxins target VSSCs and contain three disulfide bonds, the similarities to the O-conotoxins appear to end there. Apart from the cysteine residues there is no sequence homology, and the spacings between the cysteine residues are significantly different, as shown in Fig. 1. Furthermore, MrVIA inhibits mammalian VSSCs through a novel mechanism distinct from site 1 targeted by the -conotoxins, saxitoxin, and tetrodotoxin (12).
The O-conotoxins show greater similarities to both theand ␦-conotoxins than to the -conotoxins. These similarities involve the spacings between the cysteine residues ( Fig. 1) and similar precursor sequences, and consequently all three families have been placed in the O superfamily (11). Although the sequence homology in the mature peptides is restricted to the cysteine residues for these three families, the O-conotoxins are active at both the sodium and calcium channels similar to the ␦and -conotoxins, respectively. It should be noted that the action of ␦-conotoxins at the VSSC is distinct from the O-conotoxins, because they slow the rate of inactivation of sodium channels resulting in channel activation.
The O-conotoxins are the first group of conotoxins reported to inhibit TTX-R Na ϩ currents in molluscan neurons. To determine if O-conotoxins are able to inhibit both TTX-S and TTX-R Na ϩ currents in vertebrate neurons, we assessed the ability of MrVIA and MrVIB to inhibit TTX-S and TTX-R Na ϩ currents in isolated rat dorsal root ganglion (DRG) neurons. DRG neurons, which express a variety of TTX-S and TTX-R VSSCs, constitute the first connection between the peripheral afferents and the central nervous system and contain the cell bodies of small, unmyelinated C-fibers important for sensory transmission, including pain.
To date, there is no structural information available on the O-conotoxin family despite their sequences being discovered nearly a decade ago (11). This is likely the case because only limited amounts of native peptides may be isolated, and synthetic derivatives of the peptides are difficult to oxidize to the correct disulfide connectivity and appear to aggregate in aqueous solution. In the present study, we have shown that titration of the hydrophobic solvent acetonitrile into an aqueous solution of MrVIB ameliorates the aggregation and allows high quality NMR spectra to be recorded. We here report the three-dimensional structure of native MrVIB and its effects on TTX-S and TTX-R Na ϩ currents in rat DRG neurons.

EXPERIMENTAL PROCEDURES
Isolation of O-conotoxin MrVIA/MrVIB-Crude venom from the venom duct of C. marmoreus (440 mg by protein estimation at A 280 nm ) was fractionated by HPLC on a Vydac C18 semipreparative column eluted at 3 ml/min with a gradient of 0% to 80% solvent B over 80 min (solvent A, 0.1% trifluoroacetic acid) in H 2 O; solvent B, 0.09% trifluoroacetic acid in 90% acetonitrile and 10% H 2 O). The eluant was monitored at 230 nM, and 1-min fractions were collected and analyzed by mass spectrometry on an LCT, electrospray time-of-flight mass spectrometer (MicroMass, Manchester, UK). Fractions exhibiting the desired mass for MrVIA and MrVIB were further analyzed by analytical HPLC on a Vydac C4 analytical column eluted at 1 ml/min with a gradient of 0% to 80% solvent B over a 40-min period (eluant monitored at 214 nm). Further purification was obtained on a Phenomenex C4 semipreparative column eluted at 3 ml/min with a gradient of 0% to 80% solvent B over 80 min (monitored at 230 nm). Fractions were further analyzed for purity as before on the Vydac analytical C4 column, and their masses were confirmed by ESTOF MS. Yields for MrVIA and MrVIB were each ϳ1% of the crude venom.

Three-dimensional Structural Analysis
NMR Spectroscopy-Samples for 1 H NMR measurements contained ϳ1 mM peptide in various solvent conditions. Deuterated solvents were obtained from Cambridge Isotope Laboratories, Woburn, MA. Spectra were recorded between 290 and 308 K on a Bruker DMX 750 spectrometer equipped with a shielded gradient unit. Two-dimensional NMR spectra were recorded in phase-sensitive mode using time-proportional phase incrementation for quadrature detection in the t 1 dimension. The two-dimensional experiments consisted of a TOCSY spectrum using an MLEV-17 spin lock sequence with a mixing time of 80 ms, DQF-COSY, ECOSY, and NOESY spectra with mixing times of 100 -200 ms. Solvent suppression was achieved using a modified water suppression by gradient-tailored excitation (WATERGATE) sequence (14). Spectra were acquired over 6024 Hz with 4096 complex data points in F 2 and 512 increments in the F 1 dimension. 3 J NH-H␣ coupling constants were measured from a one-dimensional spectrum or from the DQF-COSY spectrum.
Spectra were processed on a Silicon Graphics Indigo workstation using XWINNMR (Bruker) software. The t 1 dimension was zero-filled to 2048 real data points, and 90°phase-shifted sine bell window functions were applied prior to Fourier transformation. Chemical shifts were referenced to the water resonance at 290 K.
Structure Calculations-Preliminary structures of MrVIB were calculated using a torsion angle-simulated annealing protocol within the program DYANA as described previously (15). Final structures were calculated using simulated annealing and energy minimization protocols within CNS version 1.1 (16). The starting structures were generated using random and dihedral angles and energy-minimized to produce structures with the correct local geometry. A set of 50 structures was generated by a torsion angle-simulated annealing protocol (17,18). This protocol involves a high temperature phase comprising 4000 steps of 0.015 ps of torsion angle dynamics, a cooling phase with 4000 steps of 0.015 ps of torsion angle dynamics during which the temperature is lowered to 0 K, and finally an energy minimization phase comprising 500 steps of Powell minimization. Structures consistent with restraints were subjected to further molecular dynamics and energy minimization in a water shell, as described previously (19). The refinement in explicit water involves the following steps. First, heating to 500 K via steps of 100 K, each comprising 50 steps of 0.005 ps of Cartesian dynamics. Second, 2500 steps of 0.005 ps of Cartesian dynamics at 500 K before a cooling phase where the temperature is lowered in steps of 100 K, each comprising 2500 steps of 0.005 ps of Cartesian dynamics. Finally, the structures were minimized with 2000 steps of Powell minimization. Structures were analyzed using PROMO-TIF (20) and PROCHECK-NMR (21).

Electrophysiological Experiments
Preparation of DRG Neurons-DRGs were dissected from along the length of the spinal cord of 7-to 12-day-old rats and treated with 1 mg/ml collagenase (type II, Worthington Biochemical Corp., Freehold, NJ) in sterile physiological salt solution for 35 min at 37°C. Ganglia were subsequently triturated with a fine-bore glass pipette and resuspended in Dulbecco's modified Eagle's media supplemented with 10% fetal calf serum and 1% penicillin/streptomycin, plated onto plastic coverslips coated with poly-L-lysine, and stored at 37°C in 95% O 2 :5% CO 2 . Following 1-2 days in culture, neurons with a capacitance of 2-35 picofarads that were relatively free of neurites were selected for experiments.
Electrophysiological Recordings from DRG Cells-Whole cell Na ϩ currents were recorded from isolated DRG neurons using the whole cell patch clamp recording technique. Patch pipettes (PG150T-7.5; Harvard Apparatus Ltd., Edenbridge, Kent, UK) were pulled, fire-polished, and had resistances of 0.8 -1.5 M⍀ when filled with intracellular solution. Membrane currents were recorded using either an EPC-7 (List Medical, Darmstadt, Germany) or an Axopatch 200A patch clamp amplifier (Axon Instruments Inc., Union City, CA), and voltage steps were generated from a PC Pentium computer using pCLAMP7 software and a Digidata 1200 A interface (Axon Instruments Inc., Union City, CA). Series resistance was routinely compensated by 70 -80%. Capacitive and leakage currents were digitally subtracted using a ϪP/6 pulse protocol.
Solutions and Toxins-Depolarization-activated Na ϩ currents in DRG neurons were recorded using patch pipettes filled with an internal solution containing (in mM): 10 NaCl, 130 CsF, 10 CsCl, 10 EGTA, 10 HEPES-CsOH, pH 7. Data Analysis-All data were analyzed using Clampfit 8 software (Axon Instruments Inc., Union city, CA). Tetrodotoxin (TTX)-resistant Na ϩ currents were recorded in the presence of Ն300 nM TTX, at least 5 min following bath perfusion. Cells that contained Ն90% current that was abolished by application of 300 nM TTX were used to study toxin effects on TTX-S Na ϩ current. In this case, TTX was applied at the beginning of the experiment to classify the Na ϩ current and then washed out. Data are expressed as means Ϯ S.E. with n indicating the number of neurons used. Concentration-response curves were analyzed using GraphPad Prism 2.0 (San Diego, CA) and fitted with a one-site binding equation, Y ϭ 1/(1 ϩ 10 (xϪlog EC50) ). K D values for kinetic experiments were calculated using the following formulas: K D ϭ K off /K on (M), where *K off ϭ 1/ off s Ϫ1 and *K on ϭ 1/( on -Ϫ K off /[toxin]) M Ϫ1 .s Ϫ1 . K off and K on were calculated from single exponential functions fitted to the time-course curves using GraphPad Prism 2.0.

RESULTS
A preliminary analysis of MrVIA and MrVIB using NMR spectroscopy revealed broad signals in aqueous solutions, indicative of aggregation. A range of pH values and temperatures did not improve the quality of the NMR spectra, but titration with acetonitrile had a beneficial effect, as illustrated in Fig. 2. The titration was carried out from 0 to 90% aqueous acetonitrile, but the highest quality spectra were observed at 50%, and the peptides were insoluble in 100% acetonitrile. Methanol titrations were also performed but did not enhance the quality of the spectra. Both MrVIA and MrVIB behaved in a similar manner, and further structural studies were performed only on MrVIB.
Assignments of the NMR spectra of MrVIB were made using two-dimensional homonuclear methods (22), and the 1 H chemical shifts are supplied as supplementary material. Chemical shifts in the amide region are generally well dispersed, and the relatively large number of resolved cross-peaks in the NOESY spectrum allowed determination of a well defined structure for most regions of the molecule. Interestingly, the amide proton of the N-terminal alanine is visible in the TOCSY spectrum, suggesting that it is involved in long-lived hydrogen bonds or is otherwise protected from exchanging with the solvent. Peaks from terminal amino groups are not normally observed in NMR spectra of peptides.
There are four backbone loops between successive cysteine residues of the O-conotoxins, as shown in Fig. 1. The residues in loop 2, the largest of the four loops, display significantly fewer NOEs than the other parts of the molecule. Furthermore, one residue in this loop, Leu 14 , and the amide proton of Cys 20 were not visible in the spectra and therefore could not be assigned. Several residues gave stronger cross-peaks in the TOCSY spectra recorded at 308 K compared with those recorded at 290 K, including Cys 9 , Phe 16 , and Tyr 18 . However, in general the quality of the NOESY spectra was better at lower temperatures, and thus structures were calculated from the data recorded at 290 K.
Secondary Structure of MrVIB-Trends in the observed chemical shifts provided a useful first insight into the secondary structure of MrVIB. Chemical shift indices (CSI (23)) for the backbone ␣-protons were calculated using the measured chemical shifts, and random coil values (24) and are illustrated in Fig. 3. Several of the residues have chemical shifts that differ from the random coil values by more than 0.1 ppm and hence have CSI values of Ϯ1, but there are no significant stretches of either positive or negative CSI values, suggesting that the structure is not dominated by either helices or ␤-sheets.
From a consideration of the experimental NOE, slow exchange and coupling patterns illustrated in Fig. 3, together with some longer range NOEs between backbone protons, an indication of the likely secondary structure was derived and is schematically illustrated in Fig. 4. The major secondary structural element is a ␤-sheet involving residues 24 -25 and 30 -31. The presence of ␣ H(iϪ1) -␦ Hi NOE signals and the absence of ␣ H(iϪ1) -␣ Hi NOE signals in the NOESY spectrum for all three proline residues (Pro 12 , Pro 21 , and Pro 27 ) confirms that their X-Pro peptide bonds are in the trans conformation. It is not uncommon that proline residues have cis peptide bonds when adjacent to aromatic residues (25). This does not appear to be the case for MrVIB, because, although Phe 28 follows Pro 27 , the peptide bond between Gly 26 and Pro 27 is in the trans conformation. However, Phe 28 adopts a positive phi angle that is presumably necessary to stabilize this region of the molecule.
After completion of the assignments for the 50% acetonitrile/ water data, spectra at lower acetonitrile concentrations were re-examined. These spectra showed fewer peaks than in 50% acetonitrile, due to the broadening/disappearance of certain residues. For example, in the spectra recorded at 25% acetonitrile/water several residues from loop 2 are not visible. The absence of signals from residues in loop 2 suggests that they are broadened, possibly as a result of intermediate exchange between multiple conformations. These conformers may reflect intermolecular interactions associated with the aggregation observed in aqueous solution.
Structure Determination-Solution structures for MrVIB were determined using simulated annealing, including experimental distance restraints and dihedral angle restraints. Of the final 50 calculated structures, the 20 lowest energy structures consistent with experimental data were chosen to represent the family of MrVIB solution structures as shown in Fig.  5. The disulfide bond connectivities used in the structure calculations were those reported by McIntosh et al. (11) (Cys 2 -Cys 20 , Cys 9 -Cys 25 , and Cys 19 -Cys 30 ). They were determined using chemical synthesis with a selective protection strategy for two of the cysteine residues.
A summary of the structural statistics for this family is given in Table I based on the broadened peaks observed for several residues in this region. Fig. 6 shows a region of the NOESY spectrum highlighting the broader peaks observed for Val 11 and Ile 13 compared with other peaks in the spectrum.
An analysis of the family of structures using the program PROMOTIF (20) shows that the main element of secondary structure of MrVIB is an anti-parallel ␤-sheet, consistent with the prediction from the NMR chemical shift, coupling, and NOE data. The strands of the ␤-sheet consist of residues 24 -25 and 30 -31 with an inverse ␥ turn between residues 26 and 28.
Preliminary structures were calculated without any hydrogen bond restraints, but these structures were consistent with hydrogen bonds across the ␤-sheet between 24 HN and 31 CO, 26 HN and 29 CO, and 31 HN and 24 CO. The latter hydrogen bond was not detected as a slow exchange amide, but this is most likely due to the low intensity of the signal due to broadening. Additional hydrogen bonds are present in several ␤-turns between the CO of residue i and HN of residue iϩ3. In particular, type I ␤-turns are present between residues 4 and 7 and between 12 and 15, and there is a type IV ␤-turn between residues 20 and 23. The type I turn for residues 12-15 is present in a region of the molecule having very few NOEs and is consequently poorly defined. The structures also provide an explanation for the observation of a signal from the N terminus in TOCSY and NOESY spectra: 70% of the structures indicate a hydrogen bond between the amino group of Ala 1 and the hydroxyl of Tyr 18 .
The disulfide bonds between Cys 2 and Cys 20 and between Cys 19 and Cys 30 are classified as right-handed hooks in a majority of the 20 final structures. The disulfide bond between residues 9 and 25 does not have a single preferred geometry throughout the structures. These structural findings are based on the assumption that the disulfide connectivity reported pre-viously (11) is correct, as seems likely, but to provide further support for this suggestion additional calculations were done with alternative connectivities. In previous studies, MrVIB was synthesized with Cys 9 and Cys 25 selectively protected, and the four remaining cysteine residues were allowed to form disulfide bonds in a nonselective fashion (11). These four cysteine residues can theoretically form three connectivities, and structures were calculated with each of these connectivities, keeping the Cys 9 -Cys 25 bond fixed. From these studies the Cys 2 -Cys 20 , Cys 9 -Cys 25 , and Cys 19 -Cys 30 disulfide set was found to have significantly better fits to the experimental restraints. For instance, the Cys 2 -Cys 19 , Cys 9 -Cys 25 , and Cys 20 -Cys 30 connectivity had several large NOE violations, including an average violation of 0.71 Å for the NOE between Cys 19 HB3 and Leu 23 HN. The 1 angle restraint on Cys 19 also showed significant violations for this connectivity (average violation of 15°). In the Cys 2 -Cys 30 , Cys 9 -Cys 25 , and Cys 19 -Cys 20 connectivity, a large number of relatively small NOE violations were observed, and again, the Cys 19 1 angle restraint violated significantly (average violation of 31°) in addition to the 1 angle restraints of Cys 2 and Cys 30 (average violations of 6.3°and 17.1°, respectively). The connectivity pattern of Cys I -Cys IV , Cys II -Cys V , and Cys III -Cys VI confirmed in the current study to be present in MrVIB is the same as is seen in a range of cystine knot structures (26). Analysis of the three-dimensional structure of MrVIB suggests that the topology is that of an inhibitor cystine knot (ICK) peptide (27).
Effects of MrVIA and MrVIB on Neuronal Whole Cell Na ϩ Currents-The effects of O-conotoxins MrVIA and MrVIB on peak TTX-R and TTX-S Na ϩ currents in rat DRG neurons were investigated. Large diameter DRG neurons (capacitance 20 -35 picofarads) exhibited mostly TTX-S Na ϩ currents with fast activation and inactivation kinetics, whereas small (2-10 picofarads) and medium sized (10 -20 picofarads) neurons exhibited both fast TTX-S and slow TTX-R Na ϩ currents, similar to that described previously (28,29). Bath application of MrVIA and MrVIB (1 M) reversibly reduced the peak amplitude of both TTX-S (Ն90% of current sensitive to 300 nM TTX) and TTX-R Na ϩ currents in rat DRG neurons (Fig. 7). Concentration-response curves indicated that MrVIA and MrVIB exhibit a similar potency, however, given that MrVIA was more readily available than MrVIB, MrVIA was used for further experi-  ments. MrVIA was found to have a 10-fold higher affinity for inhibition of TTX-R than TTX-S Na ϩ currents in rat DRG neurons (Fig. 8, A and B). The half-maximal inhibitory concentration (IC 50 ) for reduction of peak TTX-R Na ϩ current was 82.8 nM (n ϭ 7) and 1.02 M (n ϭ 7) for the TTX-S Na ϩ current (Fig. 8B). The kinetics of onset and recovery from block by MrVIA are shown for TTX-R and TTX-S Na ϩ currents in Fig. 9. From this experiment, the K d for the TTX-R Na ϩ current was 82 and 647 nM for the TTX-S Na ϩ current. The off-rate was relatively similar for TTX-R versus TTX-S Na ϩ current, and K d was determined largely by the ϳ10-fold difference in on-rate. In addition, MrVIA did not change the current-voltage relationship for either TTX-R or TTX-S Na ϩ current (Fig. 10). In a series of experiments to examine the effect of MrVIA on voltagedependent Ca 2ϩ channels, MrVIA at concentrations of up to 10 M had no effect on depolarization-activated Ba 2ϩ current amplitude in rat DRG neurons (n ϭ 6) as shown in Fig. 11.

DISCUSSION
In this study, we have determined for the first time the three-dimensional structure adopted by O-conotoxins and established that members of this class act at a mammalian TTX-R sodium channel. The O-conotoxins MrVIA and MrVIB were found to inhibit both TTX-S and TTX-R Na ϩ currents in rat DRG neurons, with IC 50 values of ϳ1 M and ϳ80 nM, respectively. The IC 50 value for TTX-R Na ϩ currents is very similar to that reported previously for inhibition of Na ϩ currents in molluscan neurons (11,13), whereas the IC 50 value for TTX-S Na ϩ currents is ϳ10 times higher. The binding site for MrVIA/B on the VSSC is currently unknown but is distinct from site 1 where -conotoxins bind (12).
The transient TTX-R Na ϩ current in DRG neurons appears to be mediated by the Na v 1.8/PN3/SNS VSSC subtype (31,32). Na v 1.8 is expressed almost exclusively in small and medium sized DRG neurons, most of which are nociceptors (33,34), and is up-regulated in inflammatory pain states (35,36). The role of Na v 1.8 in pain was confirmed by studies of Na v 1.8 null mutant mice, which exhibited a decreased response to noxious mechanical stimuli and reduced inflammatory-induced hyperalgesia (3). These observations suggest that selective inhibition of Na v 1.8 could produce analgesia without side-effects (37)(38)(39)(40).
The O-conotoxins, MrVIA and MrVIB, are the first known peptide inhibitors of the mammalian TTX-R Na ϩ current in DRG neurons, and it is possible that engineering of these toxins could produce the first selective inhibitors for TTX-R Na ϩ channels.
Analysis of the structures of the O-conotoxins has revealed some interesting features that may have functional significance. Both MrVIA and MrVIB aggregate in aqueous solution at millimolar concentrations, based on the very broad signals observed in the NMR spectra. This finding is consistent with the late elution time on HPLC of these peptides and their high content of hydrophobic residues. Although the peptides visually appear to be soluble in aqueous solution, the broad NMR peaks suggest that they are highly aggregated. Certainly, dimers or small oligomers would not be expected to result in the degree of broadening observed. The presence of organic solvent, in particular acetonitrile, substantially improved the quality of the spectra, presumably by stabilizing a monomeric conformation.
The main element of secondary structure in MrVIB is a small region of ␤-sheet between residues 24 and 25 and between 30 and 31. The ␤-sheet is stabilized by hydrogen bonds between 24 HN and 31 CO, 26 HN and 29 CO, and 31 HN and 24 CO. Additional hydrogen bonds are also present to stabilize several ␤-turns in MrVIB, as well as to stabilize the N terminus. Unusually, a signal from the N-terminal amino group is present in the TOCSY and NOESY spectra. The unusually slow exchange of the N-terminal amine appears to be the result of a hydrogen bond with the hydroxyl of Tyr 18 .
Both strands involved in the ␤-sheet of MrVIB contain cysteine residues that form disulfide bonds and connect the strands to other regions, forming a compact core in the molecule. The disulfide connectivity was previously established using synthetic methods to be Cys 2 -Cys 20 , Cys 9 -Cys 25 , and Cys 19 -Cys 30 (11). The NMR data in the current study have confirmed this connectivity but also show that the disulfide bonds are topologically arranged to form an ICK motif (27). This motif involves the disulfide bonds between Cys 2 and Cys 20 and between Cys 9 and Cys 25 and their connecting backbone segments forming a ring through which the Cys 19 to Cys 30 disulfide bond threads. The knotted structure is likely to play a significant role in stabilizing the overall fold.
The ICK motif is now known to be present in a variety of proteins, many of which are toxins (26,27). In particular, both theand ␦-conotoxins contain a cystine knot motif (41,42), and a comparison of the structures of members of these families with MrVIB is shown in Fig. 9. The overall fold is maintained in all three families despite the sequence variation, suggesting a crucial role for the disulfide bonds in maintaining the structure. Other studies on cystine knot proteins have also high- FIG. 8. Effects of MrVIA on TTX-sensitive and TTX-resistant Na ؉ currents in rat DRG neurons. In A: i, superimposed traces of Na ϩ currents recorded from a DRG neuron exhibiting only fast, TTX-S Na ϩ current. Depolarization-activated Na ϩ currents were obtained in the absence (control) and presence of 100 nM MrVIA and were completely blocked by 300 nM TTX. ii, superimposed Na ϩ currents recorded from a different DRG neuron in the presence of 300 nM TTX (TTXresistant Na ϩ current) in the absence (control) and presence of 100 nM MrVIA. B, neurons exhibiting Na ϩ current that was abolished by 300 nM TTX (Ն90%) were used for comparison of toxin affinity on TTX-R versus TTX-S Na ϩ current. The half-maximal inhibitory concentration (IC 50 ) for the TTX-R Na ϩ current was 82.8 nM (n ϭ 7) and 1.02 M (n ϭ 7) for the TTX-S Na ϩ current.
lighted the importance of the disulfide bonds. For example, the structures of disulfide isomers of -conotoxin GVIA differ significantly from the native structure and have no activity at the N-type calcium channel (43). Variations in activity have also been seen for disulfide bond isomers of ␣-conotoxins (44,45).
Despite the structural similarities observed for the O-, ␦-, and -conotoxins, they clearly act in distinct ways that must be related to structural differences. An analysis of the surface features of the peptides shown in Fig. 12 reveals such differences. For instance, most of the surface of MrVIB is hydrophobic. It has previously been noted that the ␦-conotoxin TxVIA contains an unusual surface-exposed hydrophobic patch (Fig.  12) (41), but it is apparent from the present study that the extent of hydrophobicity is quite different from O-MrVIB. This is reflected in the differing aggregation states of the peptides. The three-dimensional structure of TxVIA was determined in aqueous solution at a quite high concentration (ϳ5 mM (41)), whereas MrVIA and MrVIB appear to aggregate in aqueous solution at much lower concentrations (ϳ1 mM). Both TxVIA and MVIIA have a larger number of hydrophilic and charged residues than MrVIB, and this is highlighted in Fig.  12. The hydrophobic patch of TxVIA has been implicated in playing an important role in binding to sodium channels. MrVIB has a hydrophobic patch in a similar location but, as FIG. 9. Time course of onset and recovery from block of depolarizationactivated TTX-sensitive and TTX-resistant Na ؉ currents by MrVIA. A, the relative peak Na ϩ current amplitude plotted as a function of time during bath application of 100 and 300 nM MrVIA and upon washout. TTX-R Na ϩ current was recorded in the presence of 300 nM TTX (n ϭ 4). Representative TTX-R Na ϩ currents corresponding to those elicited before and during the onset and recovery from block are shown below. B, the relative peak TTX-sensitive Na ϩ current amplitude plotted as a function of time during bath application of 1 M MrVIA and upon washout (n ϭ 3). Error bars (Ϯ S.E.) are obscured by the data points. Representative TTX-S Na ϩ currents recorded at different times before and during the onset and recovery from block are shown below. From estimates of on-and off-rates the K d for inhibition of the TTX-R Na ϩ current was 82 nM and for the TTX-S Na ϩ current was 647 nM. TTX-R Na ϩ currents recorded in the presence of 300 nM TTX were inhibited by 100 nM MrVIA, whereas TTX-S Na ϩ currents were inhibited by 1 M MrVIA.
FIG. 11. Depolarization-activated Ca 2؉ channel currents in DRG neurons are insensitive to MrVIA. Superimposed Ba 2ϩ currents in response to step depolarization from Ϫ120 mV to Ϫ20 mV obtained in the absence (control) and presence of 1 M and 10 M MrVIA. Depolarization-activated Ba 2ϩ currents were recorded in a Na ϩ -free extracellular solution containing 4 mM Ba 2ϩ . mentioned above, also has a large number of other surfaceexposed hydrophobic residues, and therefore it is difficult to predict if this particular region plays a role in binding to the receptor. Mutagenesis studies will be necessary to elucidate the importance of various surface residues.
The most striking difference between MrVIB and theand ␦-conotoxins is the disorder observed in loop 2 of MrVIB. Relaxation experiments will be required to determine if this disorder is a result of structural flexibility or if indeed the aggregation observed in aqueous solution that appears to be modulated by acetonitrile is responsible for the disorder. Such a disordered loop is unprecedented in the O superfamily conotoxins, although a limited degree of flexibility has been reported for loop 2 of the -conotoxin MVIIA. In that case, it is known that loop 2 is critical for high affinity binding to voltagegated calcium channels. Broadening of peaks from residues 11-13 suggested slow conformational exchange (42), and this was also supported by NMR relaxation measurements (46). It has been suggested that this conformational exchange may have functional significance and may allow MVIIA to bind to two different receptor subtypes in the molluscs that it preys upon (46). Flexibility in proteins is increasingly being taken into account in drug design applications, because disordered regions can allow variability in protein recognition motifs, permitting a given sequence to adaptively recognize multiple binding partners (47). The apparent flexibility observed in the structure of MrVIB reported here may have implications for its cross-channel activity at the mollusc neuronal sodium and calcium channels in addition to activity at both TTX-R and TTX-S mammalian sodium channels.
The unusual cross channel activity at both sodium and calcium channels in molluscan neurons has been observed in at least three other peptide toxins, namely the tarantula peptides ProTx-I and ProTx-II (48) and the ␣-scorpion-toxin-like peptide, kurtoxin (49). The only sequence conservation between these peptides and MrVIA/MrVIB is the high content of cysteine residues, with the ProTx peptides containing six cysteine residues and kurtoxin containing eight. The disulfide connectivity has been elucidated for the ProTx peptides and shown to be Cys I -Cys IV , Cys II -Cys V , and Cys III -Cys VI (48). It has been suggested that they contain an inhibitor cystine knot motif based on this connectivity. Because the three-dimensional structures of these toxins have not been reported, it remains to be seen if structural flexibility is also a distinguishing feature of these toxins. Structural analysis may assist in determining key features of these toxins that interact with both sodium and calcium channels. Other peptide toxins containing a cystine knot motif that have been implicated in sodium channel activity include GS (50) and TVIIA (30).
In summary, we have shown MrVIA and MrVIB to be the first peptidic inhibitors of the transient TTX-R Na ϩ current in rat DRG neurons. The equivalent ion channels in humans are therapeutic targets for the treatment of pain. Furthermore, we have used NMR spectroscopy to provide the first structural information on this class of conotoxins. A disordered loop 2 in the structure of MrVIB is likely to result from structural flexibility and may allow binding to calcium and TTX-R and TTX-S sodium channels. However, structure activity studies are required to provide information on the importance of particular residues in binding to these different ion channels. Ribbon representations are shown with the ␤-strands as arrows and the disulfide bonds in a ball-and-stick format. Surface representations of each molecule are shown under the ribbon diagrams in two views (the lower view rotated 180°about the z-axis relative to the other view). Hydrophobic residues are shown in green, positively charged in blue, and negatively charged in red. The remaining residues are shown in gray. Selected charged residues are labeled with the single-letter amino acid codes. The hydrophobic patch in ␦-TxVIA implicated in binding to sodium channels is on the left side of this upper surface diagram.