A delta-conotoxin from Conus ermineus venom inhibits inactivation in vertebrate neuronal Na+ channels but not in skeletal and cardiac muscles.

We have isolated delta-conotoxin EVIA (delta-EVIA), a conopeptide in Conus ermineus venom that contains 32 amino acid residues and a six-cysteine/four-loop framework similar to that of previously described omega-, delta-, microO-, and kappa-conotoxins. However, it displays low sequence homology with the latter conotoxins. delta-EVIA inhibits Na+ channel inactivation with unique tissue specificity upon binding to receptor site 6 of neuronal Na+ channels. Using amphibian myelinated axons and spinal neurons, we showed that delta-EVIA increases the duration of action potentials by inhibiting Na+ channel inactivation. delta-EVIA considerably enhanced nerve terminal excitability and synaptic efficacy at the frog neuromuscular junction but did not affect directly elicited muscle action potentials. The neuronally selective property of delta-EVIA was confirmed by showing that a fluorescent derivative of delta-EVIA labeled motor nerve endings but not skeletal muscle fibers. In a heterologous expression system, delta-EVIA inhibited inactivation of rat neuronal Na+ channel subtypes (rNaV1.2a, rNaV1.3, and rNaV1.6) but did not affect rat skeletal (rNaV1.4) and human cardiac muscle (hNaV1.5) Na+ channel subtypes. delta-EVIA, in the range of concentrations used, is the first conotoxin found to affect neuronal Na+ channels without acting on Na+ channels of skeletal and cardiac muscle. Therefore, it is a unique tool for discriminating voltage-sensitive Na+ channel subtypes and for studying the distribution and modulation mechanisms of neuronal Na+ channels, and it may serve as a lead to design new drugs adapted to treat diseases characterized by defective nerve conduction.

Defective nerve conduction is frequently associated with neurological diseases, particularly with those causing axonal demyelinization (1,2). Nerve conduction may be facilitated by blocking voltage-gated K ϩ channels and by inhibiting inactivation of voltage-gated Na ϩ channels. Ligands that act on such channels are plentiful, but most of them lack selectivity, i.e. they affect channels present in various tissues, including nerve tissue and skeletal, smooth, and cardiac muscles.
Voltage-dependent Na ϩ channels play a fundamental role in cell membrane excitability, and they are targets for a large number of animal and plant toxins (3,4). The ␣-subunit of each channel is a large membrane-spanning glycoprotein containing four homologous domains associated with smaller accessory ␤-subunits (3). Mammals possess at least 10 Na ϩ channel genes, each encoding a distinct channel subtype (5), and various Na ϩ channel repertoires endow various types of neurons and muscles with distinct transduction and encoding properties.
During recent decades, many peptides isolated from cone snail venoms have been reported to target specific subtypes of voltage-sensitive Na ϩ channels (6 -10). Thus, the -conotoxins GIIIA and GIIIB selectively block tetrodotoxin-sensitive Na ϩ channels in skeletal muscles (6,10,11). The -conotoxin PIIIA also differentiates Na ϩ channel subtypes, i.e. toxin concentrations that block rNa V 1.2 (rat brain type II/IIA) and rNa V 1.4 (rat skeletal muscle) channels only slightly inhibit rNa V 1.7 (rat peripheral nerve PN1) channels (12,13). The ␦-conotoxins represent a novel category of Na ϩ channel probes because their receptor/binding site differs from those of other neurotoxins affecting Na ϩ channel inactivation (7, 14 -18). The goal of our studies was to discover ␦-conopeptides that exclusively inhibit inactivation of Na ϩ channel subtypes in the nervous systems of vertebrates.
We report here the isolation and characterization of ␦-conotoxin EVIA (␦-EVIA), 1 a conopeptide in Conus ermineus venom. The purified and synthesized toxin inhibits Na ϩ channel inactivation in neuronal membranes from amphibians and mammals (subtypes rNa v 1.2a, rNa v 1.3, and rNa v 1.6), without affecting rat skeletal muscle (subtype rNa v 1.4) and human cardiac muscle (subtype hNa v 1.5) Na ϩ channels. Thus, ␦-EVIA may be a valuable new tool for discriminating Na ϩ channel subtypes, for studying the mechanisms involved in neuronal Na ϩ channel modulation, and for designing new drugs to treat neurological diseases characterized by defective nerve conduction.

EXPERIMENTAL PROCEDURES
Reagents-Specimens of C. ermineus were collected from the Atlantic Ocean along the Senegalese coast of West Africa and maintained in an aquarium. Venom stripped from freshly dissected ducts, was stored at Ϫ80°C after lyophilization. ␦-Conotoxin TxVIA was isolated from the venom of Conus textile collected in New Caledonia. Tubocurarine chloride and tetrodotoxin were from Sigma-Aldrich, acetonitrile (UV grade) was from Merckeurolab (Fontenay sous Bois, France), fluorescein isothiocyanate-conjugated ␣-bungarotoxin (␣-BTX) was from Molecular Probes Europe BV (Leiden, The Netherlands). Other solvents and chemicals were purchased from commercial sources and were of the highest purity commercially available.
Amino Acid Sequencing-Aliquots of ␦-EVIA were incubated for 1 h at room temperature in a solution containing 6 M guanidine hydrochloride, 20 mM dithiothreitol, 2 mM EDTA, and 0.5 M Tris-HCl (pH 7.5). They were then treated for 3 h at room temperature with 1.5 mM 4-vinylpyridine. The derivative was purified by reverse phase HPLC with a C18 Vydac column (4.6 mm x 25 cm; 5-m particle size), and the amino acid sequence was determined by Edman degradation using an Applied Biosystems 477A microsequencer.
Bioassays-The bioassays were performed with mosquito fish (Gambusia affinis, 1.0 -1.5 g) and Swiss-Webster mice (10 -15 g). The fish were injected i.m., and the mice were injected intracerebroventricularly with a stereotaxic system (Harvard/ASI Apparatus, Kent, UK). The ED 50 was defined as the dose producing hyperactivity in 50% of the tested mice within 2 h postinjection. Isolated frog (Rana esculenta) nerve muscle preparations were used to detect excitoxin(s) in venom fractions (excitotoxins produce spontaneous muscle contractions) and to characterize the activity of purified ␦-EVIA.
Electrophysiological Studies-Frogs (R. esculenta, weighing 20 -25 g) were killed by double pithing. Cutaneous pectoris nerve muscle preparations were dissected and placed in a standard solution containing 115 mM NaCl, 2.1 mM KCl, 1.8 mM CaCl 2 , and 5 mM HEPES (pH 7.25). In most experiments, excitation-contraction was uncoupled with 2 M formamide, as previously described (19). Membrane potential and synaptic potentials were recorded at 20 -22°C, with intracellular microelectrodes filled with 3 M KCl (8 -12 M⍀ resistance), using conventional techniques (19). In some experiments, a dual microelectrode currentclamp technique, with the aid of an Axoclamp 2B system (Axon Instruments, Union City, CA), was used to trigger and record action potentials at a membrane potential of Ϫ100 mV.
Electrophysiological experiments with nodes of Ranvier were performed at 13-16°C with single myelinated axons isolated from the frog sciatic nerve. Action potentials and Na ϩ currents were recorded using current-and voltage-clamp techniques, as previously described (20). Na ϩ current recordings were performed after suppressing K ϩ currents with internal 110 mM CsCl and 10 mM external tetraethylammonium chloride. ␦-EVIA was applied to the nodal compartment of the chamber (for details, see Ref. 20).
Na ϩ Channel Expression and Two-electrode Voltage-Clamp Assay Using Xenopus Oocytes-cRNAs encoding the ␣-subunits of rat skeletal muscle (rNa V 1.4), human cardiac muscle (hNa V 1.5), and rat brain IIa (rNa V 1.2a), III (rNa V 1.3), and Na6 (rNa V 1.6) Na ϩ channels and the human auxiliary ␤1-subunit were transcribed in vitro, injected into Xenopus laevis oocytes, and analyzed by a two-electrode voltage-clamp protocol, as described (24). The bath solution contained 96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , and 5 mM HEPES (pH 7.85). A stock solution of ␦-EVIA (2 mM) was prepared in 50% acetonitrile, 0.05% trifluoroacetic acid, and was diluted 200-fold with bath solution containing bovine serum albumin (1 mg/ml). A BPS-8 perfusion system (ALA Scientific Instruments Inc., Westbury, NY) was used to wash the oocytes with bath solution (0.7 ml/min, flow rate; 4 p.s.i., positive pressure). Approximately 1 ml of toxin-containing solution was perfused over the oocytes mounted in a 200-l chamber at room temperature. Exposure to the vehicle (0.25% acetonitrile) did not affect the oocytes.
Staining of Nerve Terminals and Confocal Microscopy-Isolated frog cutaneous pectoris nerve muscle preparations were stained (1 h, 22°C) with 2 M ␦-R EVIA dissolved in standard solution and double-labeled (15 min, 22°C) with fluorescein isothiocyanate-conjugated ␣-BTX (diluted 1:100). After being washed several times with standard medium, the preparations were examined with a Leica TCS scanning laser confocal system (Leica Microsystems, Mannheim, Germany) mounted on an upright microscope. The images were collected in the sequential acquisition mode, using a water immersion lens (ϫ40, N.A. 0.8).
Data Analysis-All of the data are expressed as the means Ϯ S.E. Statistical comparisons were performed using Student's t test (p values Ͻ 0.05 were considered to be statistically significant).

RESULTS
Purification and Amino Acid Sequencing of ␦-EVIA-Chromatography of C. ermineus venom on Sephadex G-50 leads to the profile shown in Fig. 1. Only one fraction caused spontaneous muscle contractions when applied to isolated frog neuromuscular preparations (not shown). The fraction was purified to homogeneity by successive reverse phase HPLC with C18 Vydac (Fig. 1, inset) and C18 Merck analytical columns. Gas phase automated Edman degradation and analysis of the pyridylethylated peptide gave the following unambiguous sequence: (H)-DDCIKOYGFCSLPILKNGLCCSGACVGVCADL-(NH 2 ). Electrospray ionization mass spectrometry analyses (observed average MH ϩ ϭ 3,288.3; calculated average MH ϩ ϭ 3,288.9) were consistent with an amidated C terminus and three disulfide bridges. The pairing of the disulfide linkages unambiguously showed the arrangement C1-4, C2-5, and C3-6 (not shown). Therefore, ␦-EVIA consists of 32 amino acid residues and three disulfide linkages forming a four-loop cysteine scaffold similar to that of other conotoxins from various families (Table I). On the basis of cDNA analysis, a 29-amino acid-long C. ermineus conotoxin (collected in the western part of Atlantic Ocean), with a deduced amino acid sequence similar to that of ␦-PVIA and ␦-NgVIA (15,16) has previously been reported (18) and designated according to the current nomenclature of Conus peptides as E6a (encoded by clone E6.4). Interestingly, there is low homology between the amino acid sequence of this E6a peptide of unknown mechanism of action and that of our described ␦-EVIA (Table I). In contrast to other ␦-conotoxins, ␦-EVIA possesses an aspartic acid at positions 2 and 31, a tyrosine at position 7, a glycine at position 8, a small side chain (alanine) at position 24, and a longer loop composed of 9 amino acid residues, which, however, is also observed in conotoxins with unrelated functions; i.e. -conotoxin PnVIA (28) and O-conotoxins (29,30).
In Vivo Activity of Venom and ␦-EVIA-2-5 min after intracerebroventricular injection, low doses of C. ermineus venom induced rapid running and jumping of mice (ED 50 ϭ 0.25 g/g of body mass), whereas higher doses (Ͼ0.4 g/g of body mass) caused convulsions and death within a few minutes. Intramuscular injection of venom (0.5 g/g of body mass) into mosquito fish produced an immediate extension of the dorsal and caudal fins, a loss of balance control, and violent convulsions just prior to death. Native and synthetic ␦-EVIA exhibited similar excitatory activity in mice (ED 50 ϭ 1.8 and 1.9 pmol/g of body mass, respectively). ␦-EVIA (40 pmol) produced hyperactivity, rapid running, and seizures in mice 1-3 min after intracerebroventricular injection. Higher doses (1 nmol) induced convulsions and death in about 1 min.
Effects of ␦-EVIA on Frog Neuromuscular Preparations-Under control conditions, a single nerve stimulus elicited a single muscle action potential recorded at the junctional region of the muscle fiber ( Fig. 2A, top trace). However, the addition of ␦-EVIA (0.2-1 M) to the medium induced trains of repetitive action potentials after a single nerve stimulus ( Fig. 2A, middle  trace), and continued nerve stimulation (0.5 Hz) revealed that the repetitive action potentials were triggered by repetitive end plate potentials (EPPs) (Fig. 2A, bottom trace). When the membrane was preset to Ϫ45 mV (to inactivate most voltage-dependent Na ϩ channels), a single nerve stimulus evoked a single EPP (Fig. 2B, upper trace) in the absence of the toxin. However, in the presence of ␦-EVIA, a single nerve stimulus evoked trains of EPPs (Fig. 2B, lower trace) at high frequency. Similar effects and in the same range of ␦-EVIA concentrations were obtained in isolated mouse phrenic-hemidiaphragm preparations (data not shown). Thus, the main effect of ␦-EVIA was the enhancement of nerve terminal excitability, so that a single nerve stimulus elicited trains of repetitive EPPs, which in turn triggered trains of repetitive muscle action potentials.
We then examined whether ␦-EVIA affects directly elicited muscle action potentials using a dual microelectrode currentclamp technique and nerve muscle preparations equilibrated in standard medium containing 20 M tubocurarine to block muscle nicotinic acetylcholine receptors and neutralize the presynaptic effects of ␦-EVIA. Under these conditions, ␦-EVIA (0.2-2 M) did not affect muscle excitability or modify the amplitude, overshoot, or duration of action potentials (n ϭ 6) recorded at Ϫ100 mV (Fig. 2C). Furthermore, the membrane potential of muscle fibers was not significantly affected by the range of ␦-EVIA concentrations used (Ϫ88.8 Ϯ 3.6 mV for controls and Ϫ90.7 Ϯ 4.7 mV after toxin treatment; n ϭ 20, four different muscles). Clearly, ␦-EVIA does not act directly on the frog skeletal muscle membrane.

Effects of ␦-EVIA on Action Potentials and Na ϩ Currents in Frog Myelinated Axons and Xenopus Spinal Neurons-␦-EVIA
(0.2 M) added to the medium bathing current-clamped single myelinated axons consistently prolonged action potentials without modifying the resting membrane potential or the peak amplitude (Fig. 2D). Within 1 min after toxin exposure, the duration of the axonal action potentials, measured at 50% membrane repolarization, increased from 1.2 Ϯ 0.1 to 11.4 Ϯ 1.3 ms (n ϭ 3). The effect was not reversible within 30 min of superfusion with standard toxin-free solution. A similar action of ␦-EVIA was observed with current-clamped spinal neurons (n ϭ 6; data not shown).
We also characterized the effects of ␦-EVIA (0.2 M) on the time dependence of tetrodotoxin-sensitive Na ϩ currents in myelinated axons (Fig. 3A) and spinal neurons (Fig. 3B). Under control conditions, all of the Na ϩ channels inactivated as a function of time and thus were closed at the end of depolarizing steps. ␦-EVIA significantly suppressed time-dependent Na ϩ channel inactivation, which led to maintenance of an inward Na ϩ current. The sustained current, measured at the end of depolarizing steps, was 20.7 Ϯ 1.7% (n ϭ 6) of the peak current, and the time corresponding to 50% inactivation of the inactivatable peak current increased 3.6 Ϯ 1.6-fold (n ϭ 6) when compared with controls. ␦-EVIA exposure did not significantly alter the amplitude of the peak current and the kinetics of current activation.
␦-EVIA also altered the voltage dependence of steady-state Na ϩ current inactivation (Fig. 3C), as determined using a twopulse protocol. In the presence of 0.2 M ␦-EVIA, a fraction of the Na ϩ current (corresponding to the sustained current described above) did not inactivate and remained constant when prehyperpolarizations were more positive than Ϫ45 mV. In addition, the toxin shifted the prepulse potential corresponding to 50% steady-state inactivation by Ϫ14.9 Ϯ 1.9 mV (n ϭ 4). In contrast, ␦-EVIA (0.2 M) did not affect the voltage dependence FIG. 1. Reverse phase HPLC of ␦-EVIA in C. ermineus venom. The excitotoxic fraction obtained by molecular exclusion chromatography (Sephadex G-50) of the venom was applied to and eluted from (flow rate, ϳ3 ml/min) a C18 Vydac semi-preparative column (10 mm ϫ 25 cm; particle size, 10 m). The arrow indicates the excitotoxic fraction. Inset, the excitotoxic fraction shown in the main panel was applied to a C18 Vydac analytical column (4.6 mm ϫ 25 cm; particle size, 5 m), and ␦-EVIA was eluted (flow rate, ϳ1 ml/min) with an acetonitrile (Buffer B) gradient, as indicated by the dashed line.
of Na ϩ channel activation, i.e. we did not observe a significant change in peak conductance-voltage relationships (Fig. 3C).
␦-EVIA Affects Xenopus Oocyte-Expressed Mammalian Neuronal Na ϩ Channels but Not Skeletal and Cardiac Muscle Na ϩ Channels-The selectivity of ␦-EVIA for vertebrate neuronal Na ϩ channels was confirmed by the results of its effects on several heterologously expressed rat neuronal Na ϩ channels (subtypes rNa V 1.2a, rNa V 1.3, and rNa V 1.6) that are typical of the central nervous system. Under control conditions, Na ϩ currents fully inactivated, i.e. the ratio of the sustained current at the end of depolarizing steps (I s ) to the peak current (I p ) was Ͻ1%. The addition of 10 M ␦-EVIA inhibited Na ϩ current inactivation in the three neuronal channel subtypes (Fig. 4) without affecting the amplitude of the peak current. In the presence of ␦-EVIA, the I s /I p ratios for rNa V 1.2a, rNa V 1.3, and rNa V 1.6 were 11.1 Ϯ 2.1% (n ϭ 3), 3.7 Ϯ 0.23% (n ϭ 3), and 5.6 Ϯ 0.14% (n ϭ 3), respectively. As shown in Fig. 4, ␦-EVIA did not affect the rat skeletal muscle (rNa V 1.4) and the human cardiac muscle (hNa V 1.5) Na ϩ channel subtypes. ␦-EVIA (5 M) was also tested on rNa V 1.4 channels expressed in HEK 293 cells as previously described (31), yielding no effect in the time course of inactivation. Thus, we conclude that in the range of concentrations used, ␦-EVIA has a predominant effect on neuronal Na ϩ channels in vertebrates.
Binding Studies with Rat Brain Synaptosomes-Toxins that inhibit Na ϩ channel inactivation bind to either receptor site 3 or 6 of Na ϩ channels (4). Therefore, competition experiments were performed with [ 125 I]Lqh-2, a scorpion ␣-toxin that binds to receptor site 3 (25), and with [ 125 I]␦-TxVIA, a conotoxin that   A and B, Na ϩ currents recorded in a myelinated axon (A) and in a spinal neuron (B) during a depolarization to Ϫ20 mV, preceded by a 50-ms hyperpolarization to Ϫ140 mV applied from a holding potential of Ϫ60 mV, before and after adding 0.2 M ␦-EVIA to the external medium. Na ϩ currents were characterized by their sensitivity to tetrodotoxin (1 M). C, voltage dependence of Na ϩ current activation (circles) and inactivation (triangles). The peak current was measured in a spinal neuron during depolarizations of various amplitudes (circles) or to Ϫ20 mV (triangles), preceded by 50 ms hyperpolarizations to Ϫ140 mV (circles) or of various amplitudes (triangles) applied from a holding potential of Ϫ60 mV, before (closed symbols) and after (open symbols) treatment with 0.2 M ␦-EVIA. To determine steady-state inactivation-voltage relationships, the peak current was normalized to its maximum value at large prehyperpolarizations. To determine activation-voltage relationships, the conductance (g) was calculated from the peak current (I), according to: where V is the test pulse potential, and E Na is the current reversal potential, and was normalized to its maximum value at large depolarizations. specifically binds to receptor site 6 (7,17). ␦-EVIA did not compete [ 125 I]Lqh-2, even at high concentrations (1-10 M, data not shown), whereas, like unlabeled ␦-TxVIA, it fully inhibited binding of [ 125 I]␦-TxVIA (Fig. 5). Clearly, ␦-EVIA bound to receptor site 6, in a dose-dependent manner. The K i values deduced from these data were equal to 475 Ϯ 75 nM for ␦-EVIA (n ϭ 3) and 1.9 Ϯ 0.4 nM for ␦-TxVIA (n ϭ 3).
Staining of Frog Motor Nerve Terminals in Situ with ␦-R EVIA-The observed neuronal specificity of ␦-EVIA prompted us to determine whether ␦-R EVIA could be used to distinguish between Na ϩ channels in nerve terminals and skeletal muscle. Indeed, staining of motor nerve terminals of neuromuscular junctions was observed after exposure to 2 M ␦-R EVIA (Fig.  6A), but staining was not detected in muscle fibers. Subsequent exposure to fluorescein isothiocyanate-conjugated ␣-BTX (Fig.  6B) (which labels muscle nicotinic ACh receptors) revealed that the region of ␦-R EVIA staining was distinct from that of ␣-BTX staining (Fig. 6C), which confirmed the different, specific interactions of the toxin with presynaptic and postsynaptic elements of the neuromuscular junction. DISCUSSION We have shown that C. ermineus venom contains the toxin ␦-EVIA, which consists of 32 amino acid residues and has the same six-cysteine/four-loop framework as previously characterized ␦-, ⌷-, -, and -conotoxins (Table I). ␦-EVIA possesses an unusual specificity for neuronal Na ϩ channels. This selectivity was first observed using amphibian preparations. ␦-EVIA inhibits neuronal Na ϩ channel inactivation, which increases the duration of action potentials in myelinated axons and spinal neurons, and it considerably enhances nerve terminal excitability and synaptic efficacy at the neuromuscular junction. However, it does not directly enhance muscle excitability or induce detectable changes in directly elicited muscle action potentials in the range of concentrations used. The neuronalselective property of ␦-EVIA detected in amphibian tissues was confirmed by our observation that ␦-R EVIA (a fully active, rhodaminated derivative of ␦-EVIA) labels frog motor nerve terminals but not skeletal muscle fibers. Furthermore, the results of our subsequent studies with rat and human Na ϩ channel subtypes (heterologously expressed in Xenopus oocytes and HEK 293 cells) indicated that the toxin also differentiates between neuronal and muscle Na ϩ channels in mammals. Thus, ␦-EVIA inhibited inactivation in the three rat neuronal Na ϩ channel subtypes (rNa V 1.2a, rNa V 1.3, and rNa V 1.6) tested, but it did not discernibly affect rat skeletal muscle and human cardiac muscle subtypes (rNa V 1.4 and hNa V 1.5, respectively). Therefore, we conclude that ␦-EVIA has a predominant action on neuronal Na ϩ channels in vertebrates. The different potency of ␦-EVIA in native Na ϩ channels from amphibian spinal neurons, nodes of Ranvier, and motor nerve terminals versus heterologously expressed mammalian Na ϩ channels in oocytes may be due to: (i) differences in lipid environment and/or (ii) differences in Na ϩ channels regulatory proteins present in native versus expressed cells.
At least seven ligand-binding sites have been identified on the ␣-subunits of voltage-sensitive Na ϩ channels. Two of them, sites 3 and 6, are known to be binding sites for toxins that inhibit Na ϩ channel inactivation (4). Our binding competition studies revealed that ␦-EVIA binds only to receptor site 6, which suggests that there are structural differences in receptor site 6 of neuronal and muscle Na ϩ channels.
What is the role of ␦-EVIA in the venom of C. ermineus, a fish-hunting cone snail? ␦-conotoxins are thought to be essential for the rapid immobilization of quickly moving prey (8,18,39,40). Thus, our observation that ␦-EVIA selectively acts on neuronal voltage-dependent Na ϩ channels suggests that it contributes to the excitotoxic shock and tetanic paralysis produced in fish by C. ermineus venom. Our findings also support the idea that ␦-EVIA may be responsible for part, if not all, of the excitatory effects of the venom (i.e. spontaneous muscle contractions) observed in isolated frog neuromuscular preparations (40).
Future studies with ␦-EVIA may be of interest in various domains. First, it is a useful pharmacological probe to study channel distribution. Indeed, the results obtained with the fluorescent derivative of the toxin (␦-R EVIA) support the current notion that Na ϩ channels are absent in the distal part of frog motor nerve endings, which is consistent with data showing active conduction only in the proximal and median parts of the terminals (41). Second, the recently determined three-dimensional solution structure of ␦-EVIA (44) may provide a basis for understanding its molecular mechanism of action and its capacity to modulate the activity of neuronal voltage-sensitive Na ϩ channels. Finally, because of the specificity of the toxin for neuronal Na ϩ channels in mammals, it may be useful in designing drugs to treat diseases characterized by defective nerve conduction.