Jingzhaotoxin-III, a Novel Spider Toxin Inhibiting Activation of Voltage-gated Sodium Channel in Rat Cardiac Myocytes*

We have isolated a cardiotoxin, denoted jingzhaotoxin-III (JZTX-III), from the venom of the Chinese spider Chilobrachys jingzhao. The toxin contains 36 residues stabilized by three intracellular disulfide bridges (I-IV, II-V, and III-VI), assigned by a chemical strategy of partial reduction and sequence analysis. Cloned and sequenced using 3′-rapid amplification of cDNA ends and 5′-rapid amplification of cDNA ends, the full-length cDNA encoded a 63-residue precursor of JZTX-III. Different from other spider peptides, it contains an uncommon endoproteolytic site (-X-Ser-) anterior to mature protein and the intervening regions of 5 residues, which is the smallest in spider toxin cDNAs identified to date. Under whole cell recording, JZTX-III showed no effects on voltage-gated sodium channels (VGSCs) or calcium channels in dorsal root ganglion neurons, whereas it significantly inhibited tetrodotoxin-resistant VGSCs with an IC50 value of 0.38 μm in rat cardiac myocytes. Different from scorpion β-toxins, it caused a 10-mV depolarizing shift in the channel activation threshold. The binding site for JZTX-III on VGSCs is further suggested to be site 4 with a simple competitive assay, which at 10 μm eliminated the slowing currents induced by Buthus martensi Karsch I (BMK-I, scorpion α-like toxin) completely. JZTX-III shows higher selectivity for VGSC isoforms than other spider toxins affecting VGSCs, and the toxin hopefully represents an important ligand for discriminating cardiac VGSC subtype.

ing human, more than 10 mammalian (Na v 1.1-Na v 1.9 and Na v x) subtypes have been identified, cloned, functionally expressed, and characterized (3). Most of them can express in dorsal root ganglion (DRG) neurons, except for Na v 1.4 in skeletal muscles and Na v 1.5 in cardiac myocytes (4). These subtypes have been highly conserved during evolution (5,6). With more than 75% sequence identity among one another, they exhibit relatively similar pharmacological properties in different expression systems. However, it is the divergent residues among the sequences of these VGSC isoforms that determine their response to distinct ligands. For instance, after tyrosine 371 is substituted by serine in rNav1.6 and rNav1.3 (wild types), which are TTX-S phenotypes, the mutants become resistant to TTX (7,8).
As the major contributors to the initiation and propagation of action potentials, VGSCs become the main targets attacked by many spider toxins. With specific pharmacological properties and higher affinity with VGSCs, spider peptides have attracted the interests of many scientists. Until now, more than 30 sodium channel toxins have been purified and well characterized from venoms of various species. NMR and homology modeling techniques indicate that, irrespective of the different composition of amino acids, most of them adopt a typical inhibitor cystine knot (ICK) fold distinct from the ␣/␤ scaffold emerging in scorpion toxins (9,10). Most residues in their primary structure are believed to support the peptide framework, whereas only a few charged residues situated at the loop domains of ICK motifs are critical to interact with sodium channels. Recently, scorpion toxin determinants demonstrate that some conserved aromatic residues (Phe, Tyr, and Trp) also play an important role in modifying the sodium channel activities (11). Spider toxins exhibit limited sequence identity in the sodium channel toxins from other origins, such as marine animals, scorpions, and snakes, revealing that there is a perspective for searching for new valuable ligands to dissect variant VGSCs. Furthermore, spider toxins have been shown to lead to new insecticides and pharmaceuticals. On the functional ␣ subunit of VGSCs, more than six sites (sites 1-6) have been disclosed to bind certain ligands (12). Spider toxins mainly interact with three of them, corresponding to blocking channel pore (site 1, hainantoxin-IV (HNTX-IV) and Huwentoxin-IV (HWTX-IV)), slowing channel inactivation (site 3, -agatoxins and ␦-atracotoxins), and inhibiting channel activation (site 4, Magi 5 and ProTx I-II), respectively (13)(14)(15)(16)(17)(18). Most of them are found to have high affinity with the subtypes of VGSCs localizing on sensory neurons, but a few affect the cardiac isoform.
In this study, we report the isolation, cDNA sequence clone, and functional characterization of a novel cardiotoxin from the spider Chilobrachys jingzhao, which was identified as a new species recently (19). The crude venom is lethal to mice with an intrapertoneal LD 50 of 4.4 mg/kg. The spider toxin, denoted jingzhaotoxin-III (JZTX-III), is composed of 36 amino acid residues including 6 cysteines cross-linked in a pattern of I-IV, II-V, and III-VI. The toxin shows no effect on voltage-gated potassium channels (K v 1.1-1.3) expressed in Xenopus laevis oocytes or VGSCs and voltage-gated calcium channels (VGCCs) distributed in DRG neurons. However, it can selectively inhibit activation of TTX-R VGSCs in cardiac myocytes followed by shifting activated voltage in a depolarizing direction. We further assume that JZTX-III binds to site 4 on sodium channel proteins, which is formed by amino acid residues in the extracellular linker between domain II-S3 and domain II-S4.

MATERIALS AND METHODS
Toxin Purification and Sequencing-The venom from female C. jingzhao spiders was collected as described earlier (16). Lyophilized venom (1 mg in 0.2 ml in distilled water) was applied to a Vydac C18 analytical reverse-phase (RP) HPLC column (218TP54, 4.6 ϫ 250 mm) and eluted at a flow rate of 1 ml/min by a linear gradient of 0 -40% of buffer B (acetonitrile containing 0.1% v/v trifluoroacetic acid) over 50 min after an equilibrium period of 3 min with buffer A (distilled water containing 0.1% v/v trifluoroacetic acid). The fraction containing JZTX-III was purified further on the same RP-HPLC column by a slower linear gradient of 30 -35% buffer B over 20 min. Once purified to Ͼ99% homogeneity assessed by RP-HPLC and mass spectrometry, the peptide sample was lyophilized and stored at Ϫ20°C until use. The molecular mass was determined by MALDI-TOF mass spectrometry on a Voyager-DE TM STR Biospectrometry TM work station. The entire amino acid sequence was obtained from a single sequencing run on an Applied Biosystems/ PerkinElmer Life Sciences Procise 491-A protein sequencer.
Identification of JZTX-III cDNA-The characterization of JZTX-III cDNA was performed using 3Ј-and 5Ј-RACE methods described previously (20). First, according to the manufacturer's instructions, the total RNA was extracted from 0.1 g of fresh venom glands of female spiders using the TRIzol reagent kit. 5 g of RNA was taken to convert mRNA into cDNA using the Superscript II reverse transcriptase with an universal oligo(dT)-containing adapter primer (5Ј-GGCCACGCGTCGACTAG-TAC (dT) 17 -3Ј). The cDNA was then used as a template for PCR amplification in 3Ј-RACE. A degenerate primer 1 (5Ј-GG(A/T/C/G)CA(G/A)TT(T/ C)TGGTGGAA(A/G)TG(T/C)-3Ј) was designed corresponding to the N-terminal residues ( 5 GQFWWKC 11 ) of mature JZTX-III. The cDNA of mature toxin was amplified using primer 1 and an abridged universal adapter primer containing an additional HindIII restriction site (5Ј-CGAAGCTTGGCCACGCGTCGACTAGTAC-3Ј). Second, based on the partial cDNA sequence of JZTX-III determined by 3Ј-RACE, the antisense primers were designed and synthesized for 5Ј-RACE as follows: the genespecific primer 2 (5Ј-GCAGGCATACCCTTTGCAGCA-3Ј) corresponding to the C-terminal residues ( 18 CCKGYAC 24 ). With the strategy described by the RACE kit supplier, the 5Ј-end cDNA of JZTX-III was amplified using its gene-specific primer 2. Amplified products in both 3Ј-and 5Ј-RACE were precipitated and cloned into the pGEM-T easy vector for sequencing. DNA sequencing was performed by Bioasia Inc. Nucleic acid sequences were analyzed using the software DNAclub (by Xiongfong Chen) (www.imtech.res.in/pub/nsa/dnaclub/dos/) and DNAman (by Nynnon Biosoft) (www.Lynnon.com).
Assignment of the Disulfide Bonds of JZTX-III-To determine the disulfide connections of JZTX-III, partial reduction by Tris (2-carboxyethyl) phosphine (TCEP) at low pH was employed (16,21). 0.1 mg of JZTX-III, dissolved in 10 l of 0.1 M citrate buffer (pH 3) containing 6 M guanidine-HCl, was partially reduced by adding 10 l of 0.1 M TCEP at 40°C for 8 min at pH 3. The intermediates were isolated by RP-HPLC, and their masses were measured by MALDI-TOF mass spectrometry. Appropriate intermediates containing free thiols were dried and then alkylated by adding 100 l of 0.5 M iodoacetamide (pH 8.3). The alkylated peptide was desalted by RP-HPLC and then submitted to an Applied Biosystems 491 protein sequencer.
Preparation of Cardiac Myocytes-Single ventricular cardiomyocytes were enzymatically dissociated from adult rats by a previously described method (22) with minor modifications. Briefly, Sprague-Dawley rats (about 250 g) of either sex were killed by decapitation without anesthetization, and the heart was rapidly removed and rinsed in ice-cold Tyrode's solution containing (in mM): 143.0 NaCl, 5.4 KCl, 0.3 NaH 2 PO4, 0.5 MgCl 2 , 10.0 glucose, 5.0 HEPES, 1.8 CaCl 2 at pH 7.2. Then the heart was mounted on a Langendorff apparatus for retrograde perfusion via the aorta with non-recirculating Ca 2ϩ -free Tyrode's solution bubbled at 37°C by 95% O 2 and 5% CO 2 . After 10 min, perfusate was switched to a Ca 2ϩ -free Tyrode's solution supplemented with 0.3% collagenase IA and 0.7% bovine serum albumin, and the hearts were perfused in a recirculated mode for 5 min. After the enzymatic solution was replaced by KB buffer containing (in mM): 70.0 L-glutanic, 25.0 KCl, 20.0 taurine, 10.0 KH 2 PO 4 , 3 MgCl 2 , 0.5 EGTA, 10.0 glucose, 10.0 HEPES at pH 7.4, the partially digested hearts were cut, minced, and gently triturated with a pipette in the KB buffer at 37°C for 10 min. The single cells were obtained after undigested tissues filtered through 200-m nylon mesh. All cells were used within 8 h of isolation.
Electrophysiological Studies-Whole cell cardiac sodium currents were recorded from rod-shaped cells with clear cross-striations at room temperature (20 -25°C). Recording pipettes (2-3-m diameter) were made from borosilicate glass capillary tubing, and their resistances were 1-2 megaohms when filled with internal solution containing (in mM): 135.0 CsF, 10.0 NaCl, 5.0 HEPES at pH 7.0. External bath composition was (in mM): 30 NaCl, 5 CsCl, 25 D-glucose, 1 MgCl 2 , 1.8 CaCl 2 , 5 HEPES, 20 triethanolamine-chlorine, 70 tetramethylammonium chloride at pH 7.4. Ionic currents were filtered at 10 kHz and sampled at 3 kHz on EPC-9 patch clamp amplifier (HEKA Electronics). Linear capacitive and leakage currents were subtracted by using a P/4 protocol. Experimental data were acquired and analyzed by the program pulseϩpulsefit8.0 (HEKA Electronics). The needed concentrations of toxin dissolved in external solution were applied onto the surface of experimental cells by low pressure injection with a microinjector (IM-5B, Narishige).
Data Analysis-Data analysis was performed using Pulsefit (HEKA Electronics) and Sigmaplot (Sigma). All data are presented as means Ϯ S.E., and n is the number of independent experiments. The fitted curves of concentration-dependent inhibition were obtained by using the following form of the Boltzmann equation: Inhibition% ϭ 100/[1ϩ exp(C Ϫ IC 50 )/k], in which IC 50 is the concentration of toxin at half-maximal inhibition, k is the slope factor, and C is the toxin concentration. Fig. 1A, in which more than 20 fractions eluted were monitored at 280 nm. The fraction with the retention time of 38 min, containing JZTX-III, was further purified by a repeated RP-HPLC (Fig. 1B). Two purifications yielded about 0.05 mg of JZTX-III/mg of crude venom with a purity over 99%. Its molecular mass was determined to be 3919.4 Da by MALDI-TOF mass spectrometry. The complete amino acid sequence of the toxin was obtained by Edman degradation and found to contain 36 residues including 6 cysteines (see Fig. 4A). After being reduced by dithiothreitol and then alkylated with iodoacetamide, the molecular mass of JZTX-III increased 348 Da (58 Da ϫ 6), implying that all 6 cysteines were involved in forming three disulfide bridges. Since the primary structure had a mass of 3919.52 Da, consistent with the measured mass, the Cterminal residue could not be amidated. JZTX-III is a basic peptide sharing less than 50% sequence identity with any know peptides, although its 6 cysteines were highly conserved at corresponding positions in many toxins, such as HWTX-IV, ProTx-I, and ProTx-II (16,17).

Purification and Sequence Analysis of JZTX-III-A typical RP-HPLC chromatogram of the female spider venom was shown in
Determination of the Disulfide Bridges in JZTX-III-Because vicinal cysteines ( 18 CC 19 ) emerge in the primary structure of JZTX-III, a chemical strategy composed of partial reduction and sequence analysis was introduced instead of a traditional enzymatic method. As shown in Fig. 2, four main peaks were obtained from the RP-HPLC separation of the partial reduced mixture of JZTX-III by TCEP. MALDI-TOF mass spectrometry analysis points out that the two intermediates yielded were resolved to contain one or two disulfide bridges in peak II and I, respectively, whereas peaks R and N represent completely reduced peptide and intact peptide, respectively. Peaks I-II were collected and alkylated rapidly with iodoacetamide followed by further purification using analytical RP-HPLC. Molecular mass determination and sequencing indicated that the free thiols of these two peptides had been alkylated.
In Fig. 3A, Pth-CM-Cys signals were observed at the 4th and 19th cycles in the sequencing chromatograms of alkylated peak I, whereas no signals emerged at other cysteine cycles. The result indicates that the only reduced disulfide bond is Cys 4 -Cys 19 . When sequencing alkylated peak II, Pth-CM-Cys signals were observed at the 4th, 11th, 19th, and 24th cycles in the profiles of cysteine cycles (Fig. 3B), indicating that Cys 18 was still linked to Cys 31 by a disulfide bond. The above results indicate that two of three disulfide bridges in JZTX-III were determined to be Cys 4 -Cys 19 and Cys 18 -Cys 31 . Accordingly, the third one is cross-linked between Cys 11 and Cys 24 . Thus, JZTX-III has a conserved disulfide connectivity emerging among ICK motifs where 6 cysteines were linked in a pattern of I-IV, II-V, and III-VI (9).
Cloning and Sequencing of JZTX-III cDNA-The full-length cDNA sequence of JZTX-III was completed by overlapping two fragments resulting from 3Ј-and 5Ј-RACE. As shown in Fig.  4B, the oligonucleotide sequence of the cDNA was a 373-bp bond in which the first ATG was assumed to serve as the translation start codon. The open reading frame, ending before the first stop codon TGA at 3Ј-terminal position, encoded 63 residues corresponding to the JZTX-III precursor. It comprised a signal peptide of 21 residues, a pro-peptide of 5 residues, and a mature peptide of 36 residues. The deduced mature peptide sequence was consistent with that of native JZTX-III determined by Edman degradation. Unlike huwentoxin-IV, JZTX-III had no extra Gly or Gly ϩ Arg/Lys residues at the C terminus, which are known to allow "post-modification" ␣-amidation at the C-terminal residue (20). The prepro-regions common to all spider toxins are a hydrophobic peptide and can be processed at a common signal site -X-Arg-before mature peptide sequences, which is recognized by special endoprotelytic enzymes. In general, this region is composed of over 40 residues. Interestingly, further analysis indicated that JZTX-III had a very small prepro-region that exhibits no similarity to those of other spider toxins from diverse species including the Chinese bird spider Selenocosmia huwena Wang (also known as Ornithoctonas huwena Wang) (20,21). Furthermore, it is worth noting that the signal site anterior to mature JZTX-III was an uncommon one (-X-Ser-) (20,(23)(24)(25). A polyadenylation signal, AATAAA, was found in the 3Ј-untranslated region at position 16 upstream of the poly(A).
Effects of JZTX-III on VGSCs-Using whole cell patch clamp technique, the actions of JZTX-III were characterized on VG-SCs in rat DRG neurons and ventricular myocytes, in which both TTX-S and TTX-R types are co-expressed. TTX (200 nM) was added to the external bath solution to separate TTX-R type FIG. 2. Analytical RP-HPLC chromatogram of partial reduced JZTX-III by TCEP. Four chromatographic peaks contained intact peptide and partially reduced intermediates, respectively. As determined by MALDI-TOF mass spectrometry, their molecular masses had 2 (peak I), 4 (peak II), 6 (peak R), or 0 Da (peak N) more than that of native JZTX-III, respectively, suggesting that the main peak (peak N) represented intact toxin and that peak R represented completely reduced peptide, whereas two intermediates contained only one (peak II) or two (peak I) disulfide bonds.

FIG. 3. RP-HPLC profile of sequencing partially reduced intermediates modified with iodoacetamide.
Cys residues occur at cycles 2, 9, 16, 17, 22, and 29. The elution position of Pth-CM-Cys is marked with down arrows. A, the Cys residue cycles of alkylated peak I in Fig. 2. B, the Cys residue cycles of alkylated peak II in Fig. 2. from mixture currents. Although Maier et al. (26) suggested that some brain TTX-S subtypes were situated in transverse tubules of ventricular myocytes, in our experiments, the induced sodium currents were not changed in the absence or presence of TTX at 0.2 M (data not shown, n ϭ 4). Therefore, the effects of JZTX-III on cardiac myocytes were assayed in bath solution without TTX.
After establishing whole cell configuration, the experimental cells were held at Ϫ80 mV for over 4 min to allow adequate equilibration between the micropipette solution and the cell interior, and then the current traces were evoked using a 50-ms step depolarization to Ϫ10 mV every second. As shown in Fig.  5, A and B, 1 M JZTX-III showed no evident effects on the normal activities of both TTX-S and TTX-R VGSCs in DRG neurons (n ϭ 3). However, cardiac TTX-R currents were sensitive to the novel toxin. 1 M JZTX-III reduced the control peak amplitude to a maximum effect by 64.7 Ϯ 4.7% (Fig. 5C, n ϭ 8). JZTX-III up to 10 M completely eliminated the remaining currents within less than 1 min (Fig. 5, D and E, n ϭ 4). The rapid inhibition was dose-dependent with an IC 50 value of 0.38 Ϯ 0.04 M (Fig. 5F). It was observed that similar to ProTx I-II (15), JZTX-III failed to alternate channel inactivation, although most spider toxins (e.g. ␦and -toxins) identified to date share a common mode of slowing channel inactivation similar to scorpion ␣-toxins (7). ProTx I-II have been suggested to bind to VGSC site 3 or site 4. To further determine the detailed site for the toxin of interest, a simple competitive assay was introduced between JZTX-III and site 3 toxins. Buthus martensi Karsch I (BMK-I), acting on site 3, is a typical ␣-like scorpion toxin isolated from the Asian scorpion, B. martensi Karsch. It can slow the inactivation of VGSCs expressing in both mammalian sensory neurons and ventricular myocytes without significantly affecting the peak amplitudes (10,27). Exposed to 10 M JZTX-III, the slowing currents induced by 10 M BMK-I were eliminated completely, suggesting that the spider toxin modulated cardiac VGSCs through a mechanism distinct from site 3 toxins. Fig. 6 shows the current-voltage (I-V) curve of cardiac TTX-R VGSCs, yielding that initial activated voltage and reversal potential are Ϫ50 mV and ϩ25 mV, respectively (n ϭ 4). After 1 M JZTX-III treatment for 1 min, the inhibition of currents could be observed at tested potential from Ϫ40 mV to ϩ20 mV. JZTX-III shifted the threshold of initial activation more than ϩ10 mV in a depolarizing direction, but no change was observed significantly in the membrane reversal potential, implying that it did not change the ion selectivity of channels.
Effects of JZTX-III on VGCCs and Voltage-gated Potassium Channels (VGPCs)-There are two main categories of VGCCs distributed in rat DRG neurons: high voltage-activated channels and low voltage-activated channels, which can be discriminated by their voltage dependence and kinetics. JZTX-III (1 M) was not found to affect VGCCs (Fig. 7, n ϭ 3). Three different VGPC isoforms (K v 1.1, K v 1.2, and K v 1.3) were expressed in Xenopus Laevis oocytes and checked for toxins using the two-electrode voltage clamp technique as described previously (28). No effects were detected with JZTX-III at 1 M (Fig.  8, n ϭ 4).

DISCUSSION
In this work, we have isolated and characterized a 3.9-kDa toxin named JZTX-III from the Chinese spider C. jingzhao (19). The full sequence of the toxin was performed by Edman degradation and found to contain 36 residues including 6 cysteines. No amidation at its C-terminal residue is detected by MALDI-TOF mass spectrometry and its cDNA sequence analysis. Although it exhibits less than 50% sequence identity to any known peptides, it contains a conserved disulfide connectivity frequently emerging in ICK peptide toxins from diverse species, such as spiders and marine snails, cross-linked in a pattern of I-IV, II-V, and III-VI. Based on the analysis of precursor organization and gene structure combined with a three-dimensional fold, Zhu et al. (29) suggested that these ICK peptides from animals shared a common evolutionary origin. The molecular scaffold is highly stabilized by the three disulfide bridges, especially the third (III-VI) (9). Huwentoxin-II, from the Chinese bird spider S. huwena, adopts a scaffold distinct from ICK motif for having a unique disulfide connectivity of I-V, II-III, and IV-VI (30). The residue numbers between 2 cysteines in JZTX-III also conform exactly to the ICK definition described as a consensus sequence C I X 3-7 C II X 4 -6 C III C IV X 1-4 C V X 4 -13 C VI (where X is any residue, with the number indicated by the range) (9).
The amino acid sequence of JZTX-III is verified further by its cDNA, which produces a precursor comprising a signal peptide, an intervening pro-peptide, and a mature peptide. Concerning the structural organization, JZTX-III should be matured through a post-translational cleavage during the course of se- cretion. Many works have demonstrated that there is a common endoprotelytic site (-X-Arg-) between the sequences of prepro and mature peptide (20,(23)(24)(25). Different from known spider toxins, JZTX-III precursor contains an uncommon site (-X-Ser-), suggesting that the processing to endroproteolysis prepro-peptide should be accordingly different from that of them. Another intriguing finding in this study is that the intervening pro-peptide region of JZTX-III is the smallest one identified to date in the field of spider toxins. The region, generally rich in glutamate residues, emerges in cDNA sequences of most animal toxins from diverse sources, but it is missing in some scorpion toxins. Until now, its action in forming toxins is not yet well defined, although Diao et al. (20) inferred that it might contribute to stabilizing the toxin precursor and prevent the mature toxins from interacting with other molecules in the cytoplasm. Furthermore, the analysis of prepro-regions can provide new proof for interpreting the evolutionary relationship in animal toxins. Around 50,000 conotoxins, although targeting different receptors, can be grouped into seven superfamilies (24). However, no similar description about spider toxins demonstrated that their prepro-peptides share higher sequence identity with one another, until in our recent work, seven distinct cDNAs from the gland of S. huwena were classified into two superfamilies (20). Having two distinct characterizations, an uncommon endoproteolytic site (-X-Arg-) and a very small pre-region exhibiting limited sequence identity to others, JZTX-III defines a novel superfamily distinct from the previously reported two superfamilies.
To date, more than 30 spider toxins from BLAST databases are found to target neuronal VGSCs, but few are found to target the cardiac subtype. According to their distinct pharmacological characterization, these toxins can be classified into two groups: excitatory toxins and depressant toxins (21). ProTx-I and ProTx-II are the only agents reported in the both groups to inhibit Na v 1.5, a TTX-R subtype expressing especially in cardiac myocytes (17). A similar inhibition of channel activation is observed after the application of JZTX-III, and it belongs to the depressant toxins. It seems that its selectivity for sodium channel isoforms is even higher than that of ProTxs, which inhibit some neuronal VGSC subtypes (Na v 1.2, Na v 1.8 -1.9) with IC 50 values of less than 0.1 M. Moreover, ProTxs target outward delayed-rectifier VGPCs and T-type VGCCs (17). We also checked the effects of JZTX-III on neuronal VG-SCs isoforms and VGCCs as well as VGPCs (K v 1.1-1.3) expressed in Xenopus laevis oocytes, but no evident effects were observed. It is very likely that Na v 1.4 is not the target for JZTX-III because the peptide did not affect the normal contractions of mouse diaphragm induced by direct electrical stimulus. The properties of JZTX-III in Na v 1.5 are similar to those of scorpion ␤-toxins. They inhibit channel activation without affecting the inactivation kinetics or the ion selectivity of Na ϩ (10). This mechanism is different from that of excitatory spider toxins, such as ␦-actracotoxin-Ar1, in which they, binding to the extracellular S3-S4 loop of domain IV, modify the conformation of channel peptides and cause an uncoupling of channel activation and inactivation in a similar manner to scorpion ␣-toxins or sea anemone toxins (14). In our experiments, JZTX-III inhibited the slowing currents induced by site 3 toxin (BMK-I, a scorpion ␣-like toxin) completely, suggesting that the binding site for the spider toxin is not site 3. The mechanism of JZTX-III is also different from that of other depressant toxins, such as HNTX-IV. They block neuronal TTX-S VGSCs with no shift in the I-V curve and are assumed to be site 1-like toxins (16,18). According to the distinct effects on the VGSCs when toxins selectively bind to six sites of the channels (12), JZTX-III can be reasonably inferred to interact with site 4 located at the extracellular S3-S4 loop of domain II of the channel molecules. Furthermore, it is worth noting that although both ␤-scorpion toxin and JZTX-III inhibit channel activation, they cause a shift of the voltage dependence in different directions, implying that these toxins do not overlap the same active residues at site 4 of the VGSC protein. Thus, JZTX-III hopefully represents a useful probe for discriminating rat cardiac TTX-R VGSC isoform, although it has a lower affinity (IC 50 Ͻ 0.4 M).
Naturally occurring toxin determinants are helpful for insight into the underlying mechanism of peptides responding to distinct receptors. NMR structures of hainantoxin-I (HNTX-I) and ProTxs reveal that a hydrophobic patch formed by Phe, Tyr, Trp, and Val act as an ion channel binding site anchor and charged residues can be responsible for their pharmacological specificity (17,21). Sequence alignment in Fig. 4A indicates that JZTX-III shows limited sequence identities with other sodium channel toxins (e.g. HNTX-I and ProTxs). However, interestingly, several hydrophobic residues (Phe 7 , Tyr 22 , Trp 30 , and Val 33 ) in JZTX-III are strictly conserved at the corresponding positions in other sodium channel toxins. HNTX-IV is a potent blocker of neuronal TTX-S VGSC in DRG neurons with an IC 50 value of 44.6 nM (18). Substitutions of Lys 27 or Arg 29 with Ala reduce HNTX-IV sensitivity of TTX-S VGSC in DRG neurons by over 10-fold. 2 The 2 positive residues are also conserved in HWTX-IV, ProTxs, and HNTX-I, which are proved to inhibit Na v 1.2, whereas they are missing in JZTX-III. It is likely that the 2 residues may be responsible for binding Na v 1.2 but not Na v 1.5. JZTX-III has 8 charged residues, and most of them, except for Asp 1 and Arg 13 , can be found at corresponding positions in neurotoxins. From the listed sequences, it is still difficult to infer the crucial residues responsible for Nav1.5, but we can assume that Asp 1 and Arg 14 may result in the subtle difference in pharmacological characterization between JZTX-III and other toxins.