Introduction

Top Abstract Introduction Materials and Methods Results Discussion Summary References

Natives of the Choco rain forest of Colombia have traditionally used secretions from the skin of frogs of the genus Phyllobates to make poisoned darts (Albuquerque et al., 1971). Batrachotoxin (BTX), the main active ingredient from these skin secretions, is an extraordinarily potent neurotoxin. Its toxic effects are caused by sustained, irreversible opening of voltage-gated sodium channels of nerve and muscle (for reviews, see Brown, 1988; Cestèle and Catterall, 2000). BTX profoundly alters various aspects of sodium channel behavior. Voltage-dependent activation is shifted to more negative potentials, inactivation is disabled, and pore conductance and selectivity are altered (Khodorov and Revenko, 1979; Khodorov et al., 1981; Quandt and Narahashi, 1982; Tanguy and Yeh, 1991; Wang and Wang, 1994). Modification of sodium channels by BTX requires channel activation. In whole-cell voltage-clamp experiments, little channel modification occurs at hyperpolarized holding potentials. However, repeated channel opening by trains of depolarizing stimulus pulses causes a progressive buildup of modified channels. These characteristics have been explained by an allosteric model in which BTX binds preferentially to the open channel conformation, thus shifting the conformational equilibrium between closed open and inactivated channel states strongly toward the open state (Catterall, 1977).

The main structural component of sodium channels is the 260-kDa  subunit. The  subunit consists of four domains (I-IV), each containing six -helical transmembrane segments (S1-S6) (Catterall, 2000). The regions between S5 and S6 in each of the four domains form pore loops that dip into the membrane to create a narrow selectivity filter at the external end of the ion-conducting pore. The remainder of the pore is formed by the four S6 segments (Doyle et al., 1998; Lipkind and Fozzard, 2000). Recent site-directed mutagenesis studies have shown that specific amino acid residues within each of the four S6 segments are important determinants of BTX action (Linford et al., 1998; Wang and Wang, 1998, 1999; Wang et al., 2000, 2001). Interestingly, some of these residues are also critical for the action of local anesthetics and other related sodium channel inhibitors. For example, substitution of alanine for a phenylalanine residue located approximately halfway through transmembrane segments IVS6 profoundly reduces sodium channel sensitivity to both BTX (Linford et al., 1998) and local anesthetics (Ragsdale et al., 1994). Based on these results, Linford et al. (1998) proposed that this site represents a point of overlap between the receptors for BTX and local anesthetics.

In a previous study, we examined how the critical phenylalanine in IVS6 stabilizes local anesthetic binding by making a series of amino acid substitutions that systematically altered the size, hydrophobicity, and aromaticity of the side chain at this site (Li et al., 1999). These data showed that low affinity binding of local anesthetics to resting sodium channels depends on hydrophobicity at this site, whereas higher affinity interaction with the inactivated channel state requires an aromatic residue at this position. In the study presented here, we investigated how these same mutations affect modification of sodium channels by BTX. We found that mutant channels with nonpolar substitutions at this site were virtually insensitive to BTX, whereas channels with polar or aromatic substitutions exhibited moderate to strong modification by BTX. We then took advantage of a mutant channel that showed intermediate sensitivity to BTX to examine the biophysical properties of toxin interaction with its receptor. For the mutant channel, we found that BTX can rapidly leave the receptor when the channel is open, but not when it is closed, as if bound toxin were trapped by the closed activation gate, and that toxin dissociation from open channels is faster at negative potentials than at positive potentials, perhaps because the binding reaction is intrinsically voltage-dependent. In these respects, BTX binding to the sodium channel resembles the action of quaternary amine local anesthetics like QX314. Thus, the notion of a hydrophilic access pathway through the cytoplasmic end of the open pore, originally proposed for QX compounds (Hille, 1977), may apply to BTX as well.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion Summary References

Site-directed mutations were introduced into the rat Na_v1.3 subtype of the sodium channel  subunit (Kayano et al., 1988) in vector pSP64t using either the Altered States mutagenesis kit (Promega, Madison WI) or polymerase chain reaction-based mutagenesis, as described in detail elsewhere (Li et al., 1999). RNA was transcribed from wild-type and mutant Na_v1.3 constructs using the mMessage mMachine RNA synthesis kit (Ambion, Austin TX), and resuspended in RNAase free 0.1 mM EDTA and 5 mM HEPES, pH 7.5

Oocytes were isolated from female Xenopus laevis frogs (Boreal, St. Catherine, Ontario) anesthetized with 3-aminobenzoic acid ethyl ester, as described previously (Li et al., 1999). On the day after isolation, oocytes were microinjected with 50 nl of wild-type or mutant Na_v1.3 RNA. Sodium currents were examined 2 to 5 days later by two-electrode voltage clamp (Li et al., 1999). All recordings were performed at room temperature in a bath chamber with a volume of 100 µl. BTX (a generous gift from Dr. John Daly, National Institutes of Health, Bethesda, MD) was dissolved in ethanol to make a 0.5 mM stock solution. This stock solution was pipetted directly into the bath to give the appropriate concentration (10 µM in most experiments) and then mixed thoroughly into the bath solution with a pipetter. In some experiments, BTX was subsequently washed out by superfusion.

	Results

Top Abstract Introduction Materials and Methods Results Discussion Summary References

BTX Alters Gating of Wild-Type Na_v1.3 Sodium Channels Expressed in X. laevis Oocytes. To investigate the molecular mechanisms of BTX action, we expressed wild-type or mutant Na_v1.3 sodium channels in X. laevis oocytes and examined their responses to BTX applied to the bath. The effects of BTX on wild-type Na_v1.3 channels expressed in oocytes are summarized in Fig. 1. Activation of wild-type channels in control conditions was just detectable at 35 mV (Fig. 1A, middle control trace), whereas depolarization to 0 mV resulted in a rapidly activating whole-cell current that decayed to baseline by the end of the 70-ms test pulse (Fig. 1A, bottom control trace). BTX had little affect on closed Na_v1.3 channels at the hyperpolarized holding potential (90 mV in this and subsequent experiments). However, repeated opening of sodium channels by application of 2400 pulses to 0 mV at a frequency of 2 Hz resulted in a number of profound changes in channel behavior. First, channel activation was shifted negative compared with control. Thus, after BTX modification, whole-cell currents were clearly present at 50 mV, a potential at which no current was present in control (Fig. 1A, top traces), and were maximal at 35 mV, a potential at which current was barely detectable in control (Fig. 1A, middle traces). Overall, activation of BTX modified channels was shifted approximately 30 mV negative compared with control (Fig. 1B). A second effect of BTX was to strongly disrupt inactivation, resulting in a large sustained current at the end of a 70-ms depolarization to 0 mV (Fig. 1A, bottom BTX trace). Mean sustained currents (scaled with respect to peak currents) for control and BTX are shown in Fig. 1C. Finally, after BTX modification, a large tail current was observed when the membrane potential was returned to 90 mV at the end of the depolarizing test pulse (Fig. 1A, bottom BTX trace). The tail currents were caused by increased driving force through BTX modified channels that were open at the end of the test pulse and closed slowly when the membrane was returned to 90 mV. Mean tail/peak currents are shown in Fig. 1D.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Batrachotoxin modifies the properties of cloned Nav1.3 sodium channels expressed in X. laevis oocytes. A, typical whole-cell sodium currents elicited in control conditions (left traces) and after 2400 35-ms pulses to 0 mV in the presence of BTX (right traces). The example traces show currents evoked by 70-ms depolarizations to 50, 35, or 0 mV. In this and subsequent figures, the holding potential was 90 mV and the concentration of BTX was 10 µM. B, complete current-voltage relationships for the experiment shown in A in control () and with BTX (). The data points show peak amplitudes of currents elicited by depolarizations to test potentials from 80 to +10 mV, plotted as a function of test potential. C, the ratios of the amplitudes of sustained currents/peak currents elicited by depolarization to 0 mV, in control and with BTX. Sustained currents were measured near the end of the 70-ms test pulse. Error bars indicate S.E.M. D, ratios of the amplitudes of the tail currents/peak currents evoked by depolarization to 0 mV in control and with BTX.

Modification of Sodium Channel Function by BTX Depends on the Properties of the Amino Acid at Position 1710. It was previously shown that substitution of alanine for a critical phenylalanine residue located approximately halfway through segment IVS6 of the sodium channel  subunit (F1710 in the Na_v1.3 channel) virtually eliminates channel sensitivity to BTX (Linford et al., 1998). Phenylananine is a hydrophobic aromatic residue. Thus, substitution of alanine could in principle disrupt BTX binding to the sodium channel by reducing hydrophobicity or aromaticity at this position. To investigate how this residue influences BTX binding, we examined the affects of various amino acid substitutions at this site. Figure 2A shows typical currents, elicited by 70-ms test pulses to 0 mV in control and after 2400 pulses in the presence of BTX, for oocytes expressing wild-type or mutant channels. Figure 2, B and C, summarize the effects of these mutations on sustained currents and tail currents, respectively. As described previously (Linford et al., 1998), alanine substitution at position 1710 (mutant F1710A) virtually eliminated sodium channel modification by BTX. Similarly, isoleucine (F1710I), a large hydrophobic residue, strongly disrupted BTX action. The properties of the F1710I mutant indicate that hydrophobicity alone was not sufficient to preserve high-affinity toxin binding. In contrast, mutant channels with polar (cysteine, F1710C), or aromatic (tryptophan, F1710W; tyrosine, F1710Y) substitutions remained sensitive to BTX, with only moderate reductions in sensitivity to toxin, compared with wild-type. As discussed in more detail below, both polar and aromatic residues can interact electrostatically with ligands. Thus, one possible interpretation of these findings is that BTX binding to the sodium channel normally involves an electrostatic interaction with the phenylalanine at position 1710, an interaction that is partially preserved by polar or aromatic substitutions at this site.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Site-directed mutations at position 1710 alter sodium channel sensitivity to BTX. A, typical sodium currents in control and with 10 µM BTX recorded in oocytes expressing wild-type or mutant Nav1.3 sodium channels. Currents were evoked by 70-ms depolarizations to 0 mV. In each case, the left-hand trace was in control and the right-hand trace was with BTX. The vertical scale bars indicate 1 µA. B, mean sustained current/peak current ratios for oocytes expressing wild-type or mutant channels. For simplicity, the sustained/peak currents in control have been subtracted off. Thus, the bars indicate the increase in sustained current caused by BTX. For wild-type, F1710I, F1710C, and F1710W channels, sustained current near the end of the test pulse was negligible in control, whereas the F1710A and F1710W mutations exhibited small sustained currents even in control conditions (see example traces). C, mean tail current/peak current ratios for wild-type or mutant channels. As in B, tail/peak currents in control have been subtracted off. Error bars indicate S.E.M.

BTX Dissociates Rapidly from F1710C Channels, but Only When They Are Open. BTX binding to wild-type sodium channels is virtually irreversible over the time course of a typical voltage clamp experiment. This property places limitations on the types of questions one can ask about the biophysical mechanisms of the toxin-receptor reaction. In our initial screen of mutations at position 1710, we found that several mutations significantly reduced but did not eliminate sodium channel modification by BTX. We wondered whether it might be possible to use these mutant channels to examine the dynamics of toxin binding and unbinding. Mutant F1710C expressed well in oocytes and showed the appropriate intermediate sensitivity to BTX. Therefore, we examined this mutation in more detail.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   BTX modification of whole-cell currents approaches steady state more rapidly in oocytes expressing mutant F1710C than in oocytes expressing wild-type channels. A, typical wild-type () and F1710C () experiments showing the progressive increase in tail current amplitudes caused by BTX modification of channels during 2 Hz trains of 35-ms depolarizations to 0 mV. In this and subsequent figures, we measured the level of BTX modification by amplitudes of tail currents. We found this to be a convenient and reliable way to assess changes in the levels of channel modification throughout the course of rapid pulse trains. Unless otherwise indicated, we scaled tail amplitudes with respect to peak currents, so that we could compare one experiment with another. BTX was present in the bath for the time indicated by the bar. B, the data from A normalized with respect to the maximum tail currents in each experiment.

View larger version (34K): [in this window] [in a new window]   Fig. 4.   The rate of BTX dissociation from F1710C channels depends on stimulus frequency. A, BTX did not dissociate from wild-type channels. In this experiment, we first induced BTX modification with a 2-Hz pulse (). We then washed BTX out of the bath and monitored toxin dissociation by applying depolarizations to 0 mV, once every 30 s (). B, BTX dissociation from F1710C mutant channels was accelerated by frequent channel activation. After building up BTX modification of mutant channels by a 2-Hz pulse train (), we washed toxin out of the bath and then monitored the rate of toxin dissociation by applying depolarizing test pulses to 0 mV, initially at a rate of once very 30 sec (). At the arrow, we increased the stimulus frequency to 2 Hz (), which caused the toxin to dissociate much more rapidly from the channels. C, the rate of toxin dissociation from F1710C channels is a function of stimulus frequency. BTX was repeatedly pulsed on with 2 Hz pulse trains to 0 mV () and then washed out of the bath and pulsed off the channels with trains of 2 Hz (), 0.33 Hz (), or 0.033 Hz ().

View larger version (21K): [in this window] [in a new window]   Fig. 5.   BTX dissociation from wild-type channels is accelerated by rapid trains of depolarizing test pulses. BTX was first pulsed on using a 2 Hz pulse train (), then washed out of the bath and pulsed off using a 2 Hz train ().

BTX Dissociation from F1710C Channels Is Voltage-Dependent. If, as predicted, BTX dissociation requires the open channel conformation, then dissociation should be rapid at test potentials where most channels are open and slower at test potentials where the probability of channel opening is low. To test this prediction, we examined the rate of toxin dissociation from F1710C channels over a range of test potentials. The results of these experiments are summarized in Fig. 6. As expected, the dissociation rate drops off rapidly between 30 and 60 mV (Fig. 6, A and B), a potential range over which the proportion of activated BTX-modified channels goes from ~0.5 to near 0 (Fig. 6C). Interestingly, however, the rate of BTX dissociation showed the reverse voltage-dependence over a more positive range of test potentials. In other words, dissociation was faster at 30 mV than at 0 mV, and faster at 0 mV than at +60 mV (Fig. 6, A and B). The clear voltage-dependence over a range of test potentials (0 mV to +60 mV) in which channel activation is maximal suggests that this voltage dependence is not related to channel activation but instead reflects voltage-dependent dissociation of toxin from open sodium channels. Possible mechanisms for this voltage-dependent dissociation are discussed below.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   The rate of BTX dissociation from F1710C channels is voltage-dependent. A, typical experiments showing normalized amplitudes of tail currents elicited by toxin washout pulses applied at different test potentials. In each case, pulses were 35 ms and were applied at a frequency of 1 Hz from a holding voltage of 90 mV. At 60 mV and 45 mV (see B), currents evoked at the test potential were too small to accurately measure. Therefore, at these test potentials, dissociation was assessed by the amplitude of the tail evoked by a single pulse to 0 mV, applied once every 100 pulses. The smooth lines are exponential fits of the data. B, rate constants for toxin dissociation, plotted as a function of test potential. Each symbol shows mean ± S.E.M. for two (60 mV, 45 mV), three (30 mV, +30 mV), or four (0 mV, +60 mV) individual experiments. For each experiment, the time course of dissociation, as assessed from the amplitudes of tail currents, was fit by single exponential functions, as in A. The time constants from these fits were multiplied by 0.035, based on the assumption that dissociation occurred only during each 35-ms depolarization. The inverse of this adjusted time constant gave the dissociation rate constant koff. C, the voltage-dependence of activation for BTX modified F1710C channels. Activation was determined, for the experiments in B, from the normalized tail currents evoked, after BTX modification, by 70 msec-long depolarizations to a range of test potentials. Error bars indicate S.E.M.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion Summary References

In this article, we report three novel findings concerning modification of voltage-gated sodium channels by BTX: 1) nonpolar substitutions for a critical phenylalanine residue in transmembrane segment IVS6 strongly reduce sodium channel sensitivity to BTX modification, whereas polar or aromatic substitutions at the same site disrupt toxin action only partially; 2) BTX dissociates rapidly from F1710C mutant channels, but only when the channels are open, not when they are closed, as if the toxin were trapped in the closed channel by the activation gate; 3) toxin dissociation from open F1710C channels is voltage-dependent, with dissociation at negative potentials faster than at positive potentials. In the following sections, we discuss each of these observations.

Sodium Channel Sensitivity to BTX Depends on the Properties of the Residue at Position 1710 in Segment IVS6. The native phenylalanine residue at position 1710 contains a hydrophobic, aromatic side chain that in principle is capable of binding ligands through hydrophobic or electrostatic interactions. At least two types of electrostatic interactions involving aromatic groups are physiologically important: cation- interactions between a positively charged moiety on the ligand and the negatively charged  orbitals of the aromatic ring (Dougherty, 1996) and aromatic-aromatic interactions between the  face of one aromatic ring and the partially positively charged hydrogen atoms on the edge of another ring (Burley and Petsko, 1985). The striking loss of BTX-mediated modification with substitution of isoleucine, a hydrophobic nonaromatic residue, suggests that hydrophobic interactions at this site are not important for stabilizing BTX binding. Furthermore, the observation that channel modification is preserved with substitution of cysteine, a polar nonaromatic residue, indicates that aromatic-aromatic interactions are not involved. This leaves the possibility that cation- (for the native Phe or Tyr and Trp substitutions) or ion-dipole (for Cys substitution) interactions are important for stabilizing BTX binding.

View larger version (46K): [in this window] [in a new window]   Fig. 7.   A model for gated access to the BTX receptor within the pore of the voltage-gated sodium channel. A, the structure of BTX. B, BTX passes through the membrane in its uncharged form, but at physiological pH it is predominantly positively charged within the cell cytoplasm. The toxin reaches its receptor only from the cytoplasmic side of the membrane. The closed activation gate blocks BTX access to the receptor and traps bound toxin within the pore.

Dissociation of BTX from F1710C Sodium Channels Requires the Open Channel Conformation. BTX binding to wild-type sodium channels is virtually irreversible over the time course of a typical electrophysiology experiment; however, BTX rapidly dissociated from F1710C mutant channels. Perhaps the most surprising finding of this study is that this rapid dissociation only occurred during depolarizing stimulus pulses, when BTX-modified channels were open. This observation was unexpected, because the open channel state is thought to put the BTX receptor in a high affinity conformation, whereas resting channels are thought to have a much lower affinity for BTX. Based on the modulated receptor model for BTX action (Catterall, 1977), one might have predicted that toxin dissociation should be faster at hyperpolarized potentials, where most channels are in low-affinity resting states, and slowed by depolarizing pulse trains, which repeatedly put channels in the high-affinity open state. Yet we observed just the opposite effects.

BTX Dissociation Is Voltage Dependent. BTX dissociation showed a surprising voltage-dependence over test potentials ranging from 30 mV to +60 mV, with progressively slower dissociation at progressively more depolarized test potentials. The fact that this voltage dependence was observed over a range of potentials at which channel activation was maximal suggests that it was not caused by the closed channel conformation's having a lower affinity for toxin than the open channel conformation. (Indeed, the data indicate that dissociation from closed channels at hyperpolarized potentials is extremely slow.) However, other state-dependent mechanisms are conceivable. For example, one possibility is that there are at least two open states (for example, see Correa et al., 1992; Correa and Bezanilla, 1994), one which predominates at moderately depolarized test potentials and has a relatively low affinity for BTX and a second that predominates at strongly depolarized test potentials and has a higher affinity for BTX. An alternative possibility is that the toxin binding reaction is intrinsically voltage-dependent, because the positive charge on the BTX molecule moves through part of the membrane electric field as the toxin enters its binding site (Woodhull, 1973). This idea is consistent with the location of a critical binding determinant approximately halfway through transmembrane segment IVS6 and with the hypothesis that the toxin binding site is close to the receptor for quaternary local anesthetics, which also exhibit voltage-dependent binding (Strichartz, 1973; Cahalan, 1978; Gingrich et al., 1993; Zamponi and French, 1994). In addition, if the toxin binds at least partially within the ion-conducting pore, it could be "knocked off" by electrostatic interactions with sodium ions entering the channel from the extracellular end of the pore. This knock off effect would be greater at hyperpolarized potentials, where single channel ion flux is high, than at more depolarized potentials, where single channel current is lower.

	Summary

Top Abstract Introduction Materials and Methods Results Discussion Summary References

In summary, we propose that the pathway to and from the BTX receptor is through the cytoplasmic end of the ion-conducting pore. This pathway is available when the channel is open but is occluded by the closed activation gate. Toxin dissociation from open channels is voltage dependent, due either to state-dependent or to electrostatic mechanisms. Finally, we suggest that the phenylalanine residue at position 1710 within transmembrane segment IVS6 stabilizes toxin binding by an electrostatic interaction, perhaps with the positively charged tertiary amine of the toxin. As long as you are not a sodium channel pharmacologist, your chances of being poisoned by the secretions of Phyllobates frogs are exceedingly low. Nevertheless, the actions of batrachotoxin are fascinating, and a clearer picture of how this toxin modifies sodium channel function will probably give new insights into the molecular mechanism of channel gating.