Unique Bell-shaped Voltage-dependent Modulation of Na+ Channel Gating by Novel Insect-selective Toxins from the Spider Agelena orientalis*

Spider venoms provide a highly valuable source of peptide toxins that act on a wide diversity of membrane-bound receptors and ion channels. In this work, we report isolation, biochemical analysis, and pharmacological characterization of a novel family of spider peptide toxins, designated β/δ-agatoxins. These toxins consist of 36–38 amino acid residues and originate from the venom of the agelenid funnel-web spider Agelena orientalis. The presented toxins show considerable amino acid sequence similarity to other known toxins such as μ-agatoxins, curtatoxins, and δ-palutoxins-IT from the related spiders Agelenopsis aperta, Hololena curta, and Paracoelotes luctuosus. β/δ-Agatoxins modulate the insect NaV channel (DmNaV1/tipE) in a unique manner, with both the activation and inactivation processes being affected. The voltage dependence of activation is shifted toward more hyperpolarized potentials (analogous to site 4 toxins) and a non-inactivating persistent Na+ current is induced (site 3-like action). Interestingly, both effects take place in a voltage-dependent manner, producing a bell-shaped curve between −80 and 0 mV, and they are absent in mammalian NaV channels. To the best of our knowledge, this is the first detailed report of peptide toxins with such a peculiar pharmacological behavior, clearly indicating that traditional classification of toxins according to their binding sites may not be as exclusive as previously assumed.

Spider venom is a rich mixture with a complex biochemical content, including proteins, peptides, and low molecular mass molecules (1)(2)(3)(4). To attain high efficiency of the venom to paralyze or kill prey, many of the bioactive components specifically act on targets in the nervous system of the recipient, such as ion channels, and voltage-gated sodium (Na V ) 4 channels in particular. Na V channels are the trademark of electro-excitable cells, carrying the fast transient inward Na ϩ current during the depolarization phase of an action potential. They consist of a ϳ260-kDa pore-forming ␣-subunit associated with auxiliary ␤-subunits of ϳ30 kDa (␤ 1 -␤ 4 for mammalian and tipE for insect Na V channels) (5)(6)(7). The ␣-subunit contains four homologous but non-identical repeats (DI-DIV), enclosing the ion conduction pore in a clockwise orientation. Each of these four repeats consists of six membrane-spanning segments (S1-S6). The first four segments (S1-S4) of each repeat comprise the voltage sensor domain with highly conserved positively charged amino acid residues (Arg or Lys) in the S4 segment serving as gating charges (5). The four positively charged S4 segments are thought to move outward through the membrane in response to depolarization, changing the channel conformation and thereby opening the ion conduction pathway to generate a transient Na ϩ current. Still, the exact mechanism underlying the gating process is poorly understood (8,9). Segments S5 and S6 from all four repeats surround the pore of the channel, with the extracellular linker between S5 and S6 dipping back into the membrane to form the ion selectivity filter (5). The Na V channel is a common target for numerous xenobiotics, including animal toxins, therapeutic drugs, and insecticides (10,11). These molecular probes can bind to at least seven different binding sites (so-called receptor sites) located on the ␣-subunit and thereby strongly alter normal channel functioning. Peptide toxins from spiders and other venomous animals can bind to at least four of these receptor sites (12,13). Site 1 is located on the extracellular mouth of the channel and is targeted by a few peptide toxins, including -conotoxins and probably some spider toxins such as hainantoxins I and III-V and huwentoxin IV (14 -17). Like the milestone tetrodotoxin and saxitoxin, these peptides are thought to act as a plug, physically blocking the pore of the Na V channel. Receptor sites 3, 4, and 6 are targeted by peptide toxins that modulate channel gating rather than block the pore of the channel. Scorpion ␣-toxins typically target site 3, together with some sea anemone and spider toxins, including Magi 4 and ␦-atracotoxins (18 -22). These site 3 toxins slow or inhibit the inactivation process of Na V channels. Scorpion ␤-toxins are known as the classic site 4 toxins and cause a hyperpolarizing shift of the activation of the Na V channel (46). Some spider toxins like Magi 5, ␦-palutoxins-IT, -agatoxins, and curtatoxins are also reported to bind to site 4, although this is only speculative for the latter two groups (21,(23)(24)(25). Finally, site 6 is targeted by ␦-conotoxins that slow the inactivation of Na V channels upon binding similar to site 3 toxins (26,27). Site 6 has not been studied in detail and probably overlaps at least partially with site 3. Moreover, some peptide toxins, such as O-conotoxins, may target receptor sites distinct from those described (28).
Because most spiders prey on insects, it does not seem to be very surprising that their venom contains toxins specifically designed to target insect Na V channels. To date, a number of spider toxins have been reported to modulate the functioning of insect Na V channels with a high selectivity (13,29,30). The insect Na V channels share a high primary structure similarity with their mammalian counterparts (31). Yet, insect Na V channels are much more conserved. Until today, only one gene (para) encoding Na V channels has been identified in insects and successfully expressed in a heterologous system, in contrast to mammalian Na V channels where nine distinct genes were found to encode for functionally distinct isoforms Na V 1.1-1.9 (isoforms 1.1-1.8 successfully expressed heterologously) (32). In insect Na V channels, however, functional diversity is thought to be achieved mainly by alternative splicing and RNA editing (33,34).
Agelena orientalis is an "Old World" spider from the family Agelenidae. These araneomorph spiders are sometimes referred to as "funnel-web" spiders but should not be confused with the Australian mygalomorph funnel-web spiders from the Hexathelidae family, which includes the infamous Atrax and Hadronyche genera (22). Certain functionally diverse groups of toxins have been isolated and characterized from the venom of agelenid spiders. Thus, agelenin isolated from Agelena opulenta is presumed to act on insect Ca 2ϩ channels (35). Curtatoxins are insecticidal toxins from Hololena curta (25,36). From Agelenopsis aperta, three groups of toxins were isolated and characterized: ␣-, -, and -agatoxins (24). The ␣and -agatoxins are known to target glutamate receptors and voltage-gated Ca 2ϩ channels, respectively, whereas the -agatoxins act on insect Na V channels. This latter group of toxins is known to shift the voltage dependence of sodium channel activation toward more hyperpolarized potentials (24,(37)(38)(39)(40). Still, very few electrophysiological data showing the effects of -agatoxins on Na V channels are available today. In this study, we present isolation, biochemical analysis, and pharmacological characterization of ␤/␦-agatoxins, a novel group of toxins isolated from the venom of the agelenid spider A. orientalis. They are unique in their narrow voltage-dependent effects on both activation and inactivation of insect Na V channels.

EXPERIMENTAL PROCEDURES
Our generalized scheme for spider toxin purification, sequencing, and analysis may be found elsewhere (41).
Toxin Purification-Crude venom of the spider A. orientalis (Agelenidae) was purchased from Fauna Laboratories, Ltd. (Republic of Kazakhstan). Lyophilized venom was dissolved in distilled water (10 mg in 1 ml) and separated by reversed-phase high-performance liquid chromatography (RP-HPLC) on a Jupiter C 5 column (4.6 ϫ 150 mm, Phenomenex) using a 45-min linear gradient of acetonitrile concentration (0 -20% in 5 min, 20 -80% in 40 min) in 0.1% triethylamine (titrated with acetic acid to pH 10) at a flow rate of 1 ml/min. All fractions obtained were exposed to a second round RP-HPLC on the same column under acidic conditions using a 60-min linear gradient of acetonitrile concentration (0 -60%) in 0.1% aqueous trifluoroacetic acid at a flow rate of 1 ml/min. Eluate absorbance was always monitored at 280 nm.
Mass Spectrometry-Molecular mass measurements were carried out using the method of matrix-assisted laser desorption-ionization (MALDI) on a M@LDI LR mass spectrometer (Micromass, UK) with identification of positive ions in linear mode. The 2,5-dihydroxy benzoic acid (10 mg/ml in 50% acetonitrile, 0.1% trifluoroacetic acid) matrix was used. Samples were prepared using the dried droplet method: equal volumes (0.5 l) of sample and matrix were mixed on the target plate and left to dry in the air. Calibration was performed using the Pro-teoMass calibration kit with a molecular mass range of 700 -66000 Da (Sigma).
Reduction of Disulfide Bonds and Modification of Thiol Groups-Dried samples were dissolved in 40 l of solution containing 6 M guanidine hydrochloride, 3 mM EDTA, 0.5 M Tris-HCl (pH 8.5). Then 2 l of 1.4 M 1,4-dithiothreitol was added, and the sample was incubated overnight at 30°C. Four microliters of 50% 4-vinylpyridine in isopropanol was added to the samples, followed by incubation for 15-20 min at room temperature in the dark. Modified polypeptides were separated by RP-HPLC on a Jupiter C 5 column (2 ϫ 150 mm, Phenomenex) using a 60-min linear gradient of acetonitrile concentration (15-60%) in 0.1% trifluoroacetic acid at a flow rate of 0.3 ml/min. Protein Sequencing-The N-terminal amino acid sequences of the alkylated peptides were determined on a Procise Model 492 protein/peptide sequencer (Applied Biosystems) according to the manufacturer's protocol.
Insect Toxicity Assay-Chromatographic fractions as well as purified toxins were assayed by injection into flesh fly Sarcophaga carnaria maggots (weighing ϳ50 -60 mg). Samples were dissolved in physiological saline: 140 mM NaCl, 5 mM KCl, 5 mM CaCl 2 , 1 mM MgCl 2 , 4 mM NaHCO 3 , and 5 mM HEPES (pH 7.2), and a volume of 2 l was injected into the fourth segment of fly larva. Groups of 3-5 individuals were used for every tested concentration. Controls received pure saline. Paralytic and lethal effects were monitored for up to 24 h after the injection. Sample toxicities were expressed as median lethal doses (LD 50 ) causing 50% mortality.
Stage V-VI oocytes were harvested from the ovarian lobes of anesthetized female X. laevis frogs as described previously (44). The oocytes were injected with 50 nl of cRNA at a concentration of 1 ng/nl using a micro-injector (Drummond). The injected oocytes were incubated in a solution containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 2 mM MgCl 2 , and 5 mM HEPES (pH 7.4), supplemented with 50 mg/liter gentamycin sulfate.
Electrophysiological Experiments-Two-electrode voltage clamp recordings were performed at room temperature (18 -22°C) using a GeneClamp 500 amplifier (Molecular Devices) controlled by a pClamp data acquisition system (Molecular Devices). Whole cell currents from X. laevis oocytes were recorded 2-5 days after injection. Voltage and current electrodes were filled with 3 M KCl. Resistance of both electrodes was kept as low as possible (Ͻ1 M⍀). The bath solution composition was: 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 2 mM MgCl 2 , and 5 mM HEPES (pH 7.4). The evoked currents were sampled at 20 kHz and filtered at 1 kHz using a four-pole lowpass Bessel filter. Linear leak currents were corrected with a P/4 protocol. To avoid overestimation of a potential toxin-induced shift in the current-voltage relationships as a result of inadequate voltage control when measuring large sodium currents in oocytes, only data obtained from cells exhibiting currents with peak amplitude below 2 A were considered for analysis.
For the electrophysiological characterization of the toxins, a number of voltage protocols were applied from a holding potential of Ϫ90 mV with a start-to-start pulse frequency of 0.2 Hz. (i) Current traces were evoked by 100-ms depolarizations to V max (i.e. the voltage corresponding to maximal I Na in control).
(ii) The current-voltage relationships were determined by 50-ms step depolarizations between Ϫ90 and 60 mV, using 5-mV increments. The sodium conductance (g Na ) was calculated from the currents according to Ohm's law (Equation 1): g Na ϭ I Na /(V Ϫ V rev ), where I Na represents the Na ϩ current peak amplitude at a given test potential V, and V rev is the reversal potential. The values of g Na were plotted as a function of voltage and fitted using the Boltzmann equation (Equation 2): Ϫ1 , where g max represents maximal g Na , V g is the voltage corresponding to half-maximal conductance, and k is the slope factor. (iii) To assess the toxininduced effects on the steady-state inactivation, a standard twostep protocol was applied. In this protocol, 100-ms conditioning, 5-mV step prepulses ranging from Ϫ90 to 60 mV were followed by a 50-ms test pulse to Ϫ30 or Ϫ10 mV. Data were normalized to the maximal Na ϩ current amplitude, plotted against prepulse potential, and fitted using the Boltzmann equation (Equation 3): where I max is the maximal I Na , V h is the voltage corresponding to half-maximal inactivation, V is the test voltage, k is the slope factor, and C is a constant representing a non-inactivating persistent fraction (close to zero in control). (iv) The recovery from inactivation was assayed with a double-pulse protocol, where a 100-ms conditioning pulse to Ϫ30 or Ϫ10 mV was followed by a 50-ms test pulse to the same voltage. Both pulses were interspersed by a repolarization to Ϫ90 mV for a gradually increasing time interval (1-40 ms). The I Na obtained in the test pulse was normalized to the I Na obtained in the conditioning pulse, plotted against the corresponding time interval, and fitted with the following exponential equation (Equation 4): f(t) ϭ Ae Ϫt/ ϩ C, where t represents the time, A is the amplitude of the current, is the time constant for the fast inactivation, and C is a constant representing a non-inactivating persistent fraction (close to zero in control). The time constants () of the Na V channel fast inactivation were measured directly from the decay phase of the recorded Na ϩ current using a single exponential fit (Equation 4). To assess the doseresponse relationships, data were fitted with the Hill equation Data were analyzed with pClamp Clampfit 8.0 (Molecular Devices) and Origin 6.1 software (OriginLab). Comparison of two sample means was made using a paired Student's t test (p Ͻ 0.05). All data are presented as means Ϯ S.E. of at least three independent experiments (n Ն 3).
Molecular Modeling-The model of the ␤/␦-agatoxin-1 spatial structure was built using the program MODELLER 9v7 (available on-line) on the basis of the known structure of the homologous ␦-palutoxin-IT2 (PDB code 1V91). Structure visualization was performed using the program PyMOL (available on-line).

RESULTS
Toxin Isolation and Purification-Venom of the spider A. orientalis showed high insect toxicity. In flesh fly larvae, for instance, its LD 50 was ϳ3 g/g. At 0.1 mg/ml, the crude venom showed a pronounced effect on the insect DmNa V 1/tipE channel (activation facilitation and inactivation inhibition) and virtually no activity on mammalian Na V 1.2, 1.5, or 1.7 channels in an X. laevis oocyte expression system (45). This venom was therefore utilized to identify novel selective molecules affecting insect Na V channels.
The new toxins were isolated using a bioassay-guided twostep RP-HPLC procedure. Crude A. orientalis venom was first separated on an RP column under basic conditions (Fig. 1), and the fractions active on DmNa V 1/tipE channels were then again separated on the same column under acidic conditions (see "Experimental Procedures"). For proteomic analysis of the venom, see Ref. 60. In total, seven novel polypeptides were purified and named ␤/␦-agatoxins (see Table 1 for sequence data). Their homogeneity was verified by analytical HPLC and matrix-assisted laser desorption ionization mass spectrometry (the measured molecular masses are listed in Table 1). The major venom constituent corresponding to at least 30% of total polypeptide content (fraction 5 in Fig. 1) retained much of the crude venom toxicity (LD 50 ϭ 7 g/g in flesh fly larvae) and was named ␤/␦-agatoxin-1 (␤/␦-Aga-1). Other ␤/␦-agatoxins were numbered in accordance with the degree of their similarity to the major peptide (see sequence data below).
Primary Structure-Alkylation of non-reduced molecules with 4-vinylpyridine revealed no free thiol groups in the native toxins. Full amino acid sequences of the new sodium channel modulators were established by automated Edman degrada-tion. Direct N-terminal sequencing of reduced and alkylated peptides provided 36 amino acid residues for ␤/␦-Aga-5 and -6, 37 residues for ␤/␦-Aga-3, -4, and -7, and 38 residues for ␤/␦-Aga-1 and -2 (see Table 1 for sequence data). Each peptide contains eight cysteine residues forming four intramolecular disulfide bridges. Because of the 1-Da difference between the measured molecular masses and the corresponding calculated values based on the sequences, it was concluded that all seven peptides are C-terminally amidated.
Electrophysiological Study-Because the spider A. orientalis is known to prey on small insects, we firstly investigated the effects of ␤/␦-Aga-1, the most abundant compound in the venom, on the cloned insect Na V channel DmNa V 1/tipE from the fruit fly Drosophila melanogaster, expressed in X. laevis oocytes. On currents elicited by depolarizations to Ϫ10 mV (i.e. V max , the test potential corresponding to maximal I Na ), addition of 1 M ␤/␦-Aga-1 resulted in three significant effects ( Fig.  2A): (i) the peak of the transient sodium current (I Na ) increased by 35.2 Ϯ 4.8% (n ϭ 6), (ii) the time-to-peak of I Na decreased from 1.07 Ϯ 0.12 to 0.86 Ϯ 0.11 ms (n ϭ 6), and (iii) a persistent current (I NaP ) of 5.5 Ϯ 2.2% (n ϭ 6) of the peak was induced. Consequently, the toxin-induced effects on the current-voltage (I-V) relationships were assayed at 1 M (Fig. 2B) and analyzed for peak currents (Fig. 2C, I Na ) and persistent currents (Fig. 2C, I NaP ). Clearly, ␤/␦-Aga-1 caused a profound hyperpolarizing shift in the I Na -V relationship, without changing the reversal potential indicating that the ion selectivity of the DmNa V 1/tipE channels was not altered. The toxin-induced persistent currents (I NaP ) could be observed only in the voltage range between Ϫ80 and 0 mV, following a bell-shaped I NaP -V relationship with

TABLE 1 Full-length sequences, average molecular masses, and accession numbers of ␤/␦-agatoxins
All peptides are C-terminally amidated (-NH 2 ). Breaks are introduced to allow sequence comparison, cysteine residues are printed in bold, and residues that differ from those in ␤/␦-Aga-1 are marked gray.
To further investigate the effects of ␤/␦-Aga-1 on the activation of the insect DmNa V 1/tipE channel, the Na ϩ conductance (g Na ) was calculated from I Na peak amplitude according to Ohm's law (Equation 1), normalized, plotted against voltage, and fitted with the Boltzmann equation (Equation 2), yielding the steady-state activation curve (Fig. 3A, top panels). At a concentration of 1 M, the toxin induced a large hyperpolarizing shift marked by a prominent change in the slope of the curve from 6.3 Ϯ 0.3 to 11.3 Ϯ 0.6 ms (n ϭ 6). The threshold potential for activation was shifted approximately from Ϫ50 to Ϫ80 mV, whereas the potential corresponding to maximal activation was not changed. The potential corresponding to half-maximal activation (V g ) was shifted from Ϫ20.8 Ϯ 0.5 to Ϫ30.4 Ϯ 1.0 mV (n ϭ 6). To display the voltage range wherein the action of toxin on DmNa V 1/tipE channel activation takes place, the normalized g Na values in control were subtracted from the normalized g Na values in the presence of toxin. The resulting ⌬g Na was then plotted against voltage (see Fig. 3A, bottom panels). The voltage dependence thus found revealed a bell-shaped relationship in the voltage range between Ϫ80 and 0 mV reaching a maximal amplitude around Ϫ35 mV. When measured at different toxin concentrations, a concentration-and voltage-dependent bellshaped relationship was observed (see compilation, Fig. 3B). To assess the dose-response relationship (Fig. 3C), the ⌬g Na (sub-tracted conductances) measured at Ϫ35 mV were plotted against toxin concentration and fitted with the Hill equation (Equation 5) yielding an EC 50 value of 288.2 Ϯ 116.4 nM (n ϭ 3-6, Hill coefficient ϭ 1.5 Ϯ 0.7).
The toxin-induced persistent currents (I NaP ) were analyzed in an analogous manner as the effect of toxin on the activation of DmNa V 1/ tipE. The I NaP was quantified by measuring the current amplitude after 50 ms and normalizing to the maximal current amplitude in control (not necessarily at the same voltage). The thus obtained normalized I NaP values were then plotted against voltage (Fig. 4A, top panels).
To display the voltage range wherein the toxin-induced persistent currents take place, control values for normalized I NaP were subtracted from normalized I NaP in the presence of the toxin (see Fig. 4B, bottom panels). The resulting ⌬I NaP followed a bell-shaped voltage dependence, reaching a maximum around Ϫ45 mV. When tested at different toxin concentrations, a concentration-and voltage-dependent bell-shaped relationship was observed (see compilation, Fig. 4B). To obtain the dose-response relationship (Fig. 4C), the ⌬I NaP (subtracted persistent currents) measured at Ϫ45 mV were plotted against toxin concentration and fitted with the Hill equation (Equation 5) yielding an EC 50 value of 220.2 Ϯ 32.9 nM (n ϭ 3-6, Hill coefficient ϭ 1.5 Ϯ 0.4).
For a better understanding of the mechanism behind the persistent currents induced by ␤/␦-Aga-1, we continued by investigating the toxin effects on the DmNa V 1/tipE inactivation process into more detail. Firstly, the steady-state inactivation was assayed with a typical double-pulse protocol (Fig. 5A). When the test pulse voltage was set at Ϫ10 mV (i.e. V max , the voltage corresponding to maximal current in control), 1 M ␤/␦-Aga-1 did not change V h (i.e. the midpoint of the steadystate inactivation curve). Yet, a small but significant non-inactivating component (5.3 Ϯ 0.6%; n ϭ 6) was induced at prepulse voltages more depolarized than Ϫ40 mV. Subsequently, the test potential was chosen more centrally in the voltage range where the persistent currents occur. At a test potential of Ϫ30 mV, V h was again not significantly shifted (from Ϫ46.7 Ϯ 0.1 to Ϫ50.0 Ϯ 0.3 mV), but the non-inactivating component was now increased up to 15.8 Ϯ 0.6% (n ϭ 3). The effects of ␤/␦-Aga-1 on the recovery from inactivation of DmNa V 1/tipE were examined next (see Fig. 5B). At a test potential of Ϫ10 mV, ␤/␦-Aga-1 decreased the time needed to recover from inactivation. The time constant for recovery from inactivation significantly decreased from 2.56 Ϯ 0.11 ms in control to 1.65 Ϯ 0.13  JUNE 11, 2010 • VOLUME 285 • NUMBER 24 ms in the presence of 1 M ␤/␦-Aga-1 (n ϭ 4). When the test potential was set at Ϫ30 mV, the time constant for recovery from inactivation decreased significantly from 2.86 Ϯ 0.08 to 1.77 Ϯ 0.06 ms (n ϭ 4). Interestingly, the channels not only recovered faster from inactivation in the presence of the toxin, they also yielded significantly higher currents (up to 114.2 Ϯ 1.0%) in the test pulse compared with the conditioning pulse when short interpulse intervals (4 -28 ms) were applied. These unusually high normalized currents (Ͼ1) decreased with increasing interpulse time interval, to return to unity when interpulse intervals were 28 ms or longer. To test whether the toxin altered the real-time kinetics of fast inactivation of DmNa V 1/tipE, the time constants of inactivation were measured directly from the decay phase of currents with a monoexponential fit as described by Equation 4. The use of a multiexponential equation did not yield beneficial fitting results over the mono-exponential equation. At voltages more depolarized than Ϫ25 mV, 1 M ␤/␦-Aga-1 did not significantly change the time constant of inactivation (see Fig. 5, n ϭ 4). At more negative potentials than Ϫ25 mV, however, the time constants of inactivation significantly decreased in the presence of 1 M ␤/␦- Aga-1 (n ϭ 4).

DISCUSSION
In this work, we describe ␤/␦-agatoxins, a new group of peptides found in A. orientalis spider venom, specifically targeting insect Na V channels. The electrophysiological profile of ␤/␦agatoxins was assayed on cloned insect and mammalian Na V channels expressed in X. laevis oocytes, revealing, to the best of our knowledge, a novel and unique behavior. The studied toxins cause a 3-fold effect on the currents mediated by the insect DmNa V 1/tipE channel: (i) an increase of peak sodium current, (ii) a decrease in time-to-peak, and (iii) generation of a noninactivating persistent current. In contrast, the mammalian rNa V 1.2/␤ 1 channel was not affected by ␤/␦-Aga-1, which was somewhat expected, as the natural prey of A. orientalis does not include mammals.
␤/␦-Agatoxins shift the current-voltage relationships of insect DmNa V 1/tipE channels toward more hyperpolarizing potentials, comparable to typical site 4 toxins such as scorpion ␤-toxins, which could explain the increase in peak current amplitude at low depolarizations. However, ␤/␦-agatoxins differ in their actions from scorpion ␤-toxins in that the latter generally cause a small current inhibition (Ͻ15%) at the voltage of maximal activation (46 -49). Moreover, the shift in the voltage dependence of activation observed with ␤/␦-agatoxins is marked by a significant increase in the slope factor of the activation curve, indicating that the cooperativity between the four S4 segments is changed, which is not the case with the classic ␤-toxins.
Importantly, the foremost prominent difference with typical site 4 toxins is the persistent current generated by ␤/␦-agatoxins. In fact, the increase in peak current amplitude and the appearance of persistent currents macroscopically resemble the effects of typical site 3 toxins, such as scorpion ␣-toxins, which are known to slow the fast inactivation of Na V channels. However, the observed actions of ␤/␦-agatoxins differ thoroughly from the actions of typical site 3 toxins. The ␤/␦-agatoxin-induced persistent currents only took place in the voltage range between Ϫ80 and 0 mV, following a bell-shaped voltage dependence with a maximal amplitude around Ϫ45 mV. In addition, ␤/␦-Aga-1 did not change the time constants of inactivation () at voltages more positive than Ϫ25 mV and even caused a decrease in at potentials below Ϫ25 mV rather than an increase, which would have been expected if the toxin actually slowed the channel inactivation. The interpretation of this unexpected observation is difficult because of the dual effect of the toxin. Besides the apparent slowing of the inactivation that could result in the persistent currents, ␤/␦-Aga-1 also inflicts a hyperpolarizing shift in the voltage dependence of activation, causing the channels to open at lower depolarizations compared with control conditions. Therefore, currents in the presence of ␤/␦-Aga-1 between Ϫ80 and Ϫ25 mV have an increased peak amplitude and a faster inactivation than currents elicited at the same voltage in control situation (see current traces in Fig. 5C). ␤/␦-Aga-1 did not shift V h nor alter the slope of the steady-state inactivation curve. Yet, a small but significant component of non-inactivating channels was observed at prepulse potentials more depolarized than Ϫ40 mV, a phenomenon previously observed with site 3 toxins (20,50,51). This small non-inactivating component was observed using a classic steadystate inactivation voltage protocol with the test potential set at V max (Ϫ10 mV). When the test potential was chosen at Ϫ30 mV, a potential more centrally located in the voltage range where the toxininduced effects take place, the non-inactivating component was increased significantly. If one assumes a bimodal gating scheme in the modulated receptor mechanism as proposed by Hondeghem and Katzung (52), it could be deduced that the toxin promotes the inactivated-to-open state transition (as indicated by the non-inactivating component in the steadystate inactivation; see Fig. 5A) and the closed-to-open state transition (as observed in real-time by the decreased time-topeak and in steady-state by the hyperpolarizing shift in activation; see Fig. 2, A-C). In addition, ␤/␦-Aga-1 increased the rate of recovery from inactivation of DmNa V 1/tipE (see Fig. 5B), which may indicate that the toxin promotes the transition from the inactivated to the closed state of the toxin-bound channel. A remarkable observation from the recovery protocol is that the normalized currents in the presence of ␤/␦-Aga-1 exceed unity with short interpulse intervals (4 -28 ms). This "additional" normalized current (I test Ͼ I cond ) gradually decreases toward unity as interpulse intervals increase. Apparently, when using such a double-pulse protocol, the duration of repolarization to Ϫ90 mV between the two pulses is crucial for unbinding of the toxin from the channel. The "additional" normalized current (I test Ͼ I cond ) observed in the recovery protocol with short interpulse intervals could be explained by the promotion of the inactivated-to-open state transition by the toxin, which leads to a higher fraction of channels that return to the open state in the test pulse compared with the conditioning pulse. As the repolarization interval increases, the fraction of unbound channels increases, and this effect decreases. All together, these observations show that binding of ␤/␦-Aga-1 relatively favors the open state of the DmNa V 1/tipE channel, giving rise to larger sodium influx in the narrow bell-shaped voltage range, with important consequences in vivo.
For ␤/␦-agatoxins, the corresponding mRNA sequences from venom glands were established, and the precursor prepropeptide structures were thus deduced (56). The sequences are organized in the conventional manner, with N-terminal signal peptides, intervening profragments and C-terminal mature peptides (57,58). Limited proteolysis Analysis of the toxin-induced effects on activation (⌬g Na ) and persistent currents (⌬I NaP ) were done in the same manner as in Figs. 3 and 4, respectively. B, bar diagrams showing the relative amplitudes of the effects on the activation (represented by ⌬g Na measured at Ϫ35 mV) and the persistent currents (represented by ⌬I NaP measured at Ϫ45 mV). Amplitudes were normalized by setting the ␤/␦-agatoxin homologue with the highest effect as 100%.
FIGURE 7. Comparison of the amino acid sequences of ␤/␦-agatoxins and similar spider toxins. Sequence identities are shown relative to ␤/␦-Aga-1, ␤/␦-Aga-6, and ␤/␦-Aga-7. Amino acid residues identical with ␤/␦-Aga-1 are shown on a black background, conservative substitutions are shown on a gray background, and cysteine residues are printed in bold. Breaks (dashes) have been introduced to maximize alignment. All the aligned toxins except for ␦-Palu-IT3 are C-terminally amidated, as indicated by asterisks. occurs at the processing quadruplet motifs: XXXR, where any X ϭ E. All precursors bear C-terminal extensions -G or -GK serving signals of C-terminal amidation and confirming our protein sequencing results.
Historically the first discovered -agatoxins from A. aperta venom (37,40) were later shown to act specifically on insect Na V channels (24,38,39). Most notably, the toxins -Aga-1 and -4 studied in detail were shown to shift the voltage-activation curve to more negative potentials similar to scorpion ␤-toxins. However, they also slowed Na V channel inactivation and produced a non-inactivating persistent current. These effects have not been explored in detail but at least for the best studied -Aga-1 and -4 qualitatively resemble those described by us for the ␤/␦-agatoxins. Thus, we believe that the "-" prefix is misleading, because it has a more widespread pore-blocking mechanism connotation and propose that these toxins be renamed accordingly. Curtatoxins from H. curta venom (25) have not been studied in detail but are believed to act similarly to -agatoxins, ␦-palutoxins-IT (␦-Palu-IT), and the newly discovered ␤/␦-agatoxins based on their sequence similarity. To date, ␦-palutoxins-IT from P. luctuosus venom (59) are probably the best studied toxins from the family. Unlike the homologous -agatoxin-1 and -4 and ␤/␦-agatoxins, ␦-palutoxins-IT1 and -2 do not affect Na V channel activation but instead induce effects resembling scorpion ␣-toxins, i.e. antagonize channel inactivation. However, profound differences between ␦-Palu-IT1 and -2 and scorpion ␣-toxins were noted, the former competing with scorpion ␤-toxins for binding to receptor site 4 and not influencing the kinetics of channel inactivation (23,53). It was therefore suggested that ␦-Palu-IT1 and -2 only macroscopically resemble typical scorpion ␣-toxins but, unlike them, influence the inactivated-to-closed state transition process (53).
Although numerous exceptions exist, high polypeptide sequence similarity generally suggests common spatial structure organization and mechanism of action. We may therefore speculate that, albeit producing diverse effects on their target, ␤/␦and -agatoxins, curtatoxins, and ␦-Palu-IT all share a common fold and bind to the same receptor site 4 of Na V channels, as shown for ␦-Palu-IT2 (23). In Fig. 8, a spatial structure model for ␤/␦-Aga-1 is shown, based on sequence similarity to ␦-Palu-IT2 (62.2% identical residues). Residues critical for channel binding have been determined in ␦-Palu-IT2 by alanine scanning mutagenesis: Arg-8, Trp-12, Tyr-22, Ser-24, Arg-26, Met-28, Tyr-30, Arg-32, and Arg-34 (colored red in Fig. 8 at the proper positions) (23). Of these, Trp-12 and Tyr-22 are almost 100% conserved across all family members (see Fig. 7) with the only W12A mutation in curtatoxin-1 (a neighboring Tyr-13 in this toxin may take over the functions) and Y22W substitution in ␦-Palu-IT3; a hydroxy amino acid residue (Ser or Thr) is found at position equivalent to Ser-24 with Asn only in curtatoxin-1; Arg-26 is substituted for Met in ␤/␦-Aga-6, Ser in curtatoxin-1, and Pro in ␦-Palu-IT3; a hydrophobic residue is present at position corresponding to Met-28; either positively charged or bulky hydrophobic residues are found at the positions of Arg-32 and Arg-34; and only Arg-8 and Tyr-30 are quite variable, still with prefer-ence for Arg/Lys, His, or Gln at position 8 and Lys or Tyr at position 30 (see Fig. 7). Therefore, the general features of the ␦-Palu-IT2 pharmacophore are well conserved among all members of the family.
In fact, ␤/␦and -agatoxins (the best studied -Aga-1 and -4) may be considered as having an "additional" scorpion ␤-toxin-like activity as compared with the ␣-toxin-like ␦-palutoxins-IT (␦-Palu-IT1 and -2). Following the same reasoning as above, residues that account for this additional feature may be located. In ␤/␦-Aga-1, these are Ser-7, Tyr-28, Phe-29, and Ile-33 (cyan in Fig. 8), the corresponding residues in ␦-Palu-IT2 being Gly-6, Ser-27, Met-28, and Arg-32. All other substitutions in ␦-Palu-IT2 as compared with ␤/␦-Aga-1 can also be found in other toxins exhibiting the ␤-toxin-like activity and are therefore considered neutral. Interestingly, all of the presumed residues assuring the ␤-toxin-like activity are located close in space (see Fig. 8) with two of the four coinciding with the ␦-Palu-IT2 pharmacophore. The envisaged future mutagenesis studies will shed more light onto these issues.
In conclusion, we argue that the "classic" concept of receptor sites 3 and 4 of Na V channels binding toxins that exhibit the ␣and ␤-effects (site 3-and site 4-like activities), respectively, becomes rather "blurred" as new data are being accumulated. Indeed, it seems that no strict correspondence exists between the toxin binding site and its effect on the channel function. Presumed functionally important residues are indicated, the corresponding residues of ␦-Palu-IT2 are given in parenthesis; the putative pharmacophore of ␦-Palu-IT2 (23) is colored red; and residues believed to conform the scorpion ␤-toxin-like activity to ␤/␦-Aga-1 are marked cyan.