Evidence That Tetrodotoxin and Saxitoxin Act at a Metal Cation Binding Site in the Sodium Channels of Nerve Membrane (Solubilized Membrane/Receptors/Surface Charge) R

Proc. Nat. Acad. Sci. USA 71 Vol. 71, No. 10, pp. 3936-3940, October 1974

Evidence That Tetrodotoxin and Saxitoxin Act at a Metal Cation Binding Site in the Sodium Channels of Nerve Membrane (solubilized membrane/receptors/surface charge) R. HENDERSON*, J. M. RITCHIE, AND G. R. STRICHARTZt Departments of Molecular Biophysics and Biochemistry, and of Pharmacology, Yale University School of Medicine, New Haven, Connecticut 06510 Communicated by Frederic M. Richards, June 17, 1974

ABSTRACT The effects of monovalent, divalent, and STX. These experiments suggest how the toxins exert their trivalent cations on the binding of tetrodotoxin and saxitoxin to intact nerves and to a preparation of solubilized action, and provide a unifying explanation of how several nerve membranes have been examined. All eight divalent cations affect nerve membrane permeability. and trivalent cations tested, and the monovalent ions MATERIALS AND LiW, Tl+, and H+ appear to compete reversibly with the METHODS toxins for their binding site. The ability of lithium to Tritium-labeled TTX and STX were prepared and purified reduce toxin binding is paralleled by its ability to reduce (3, 5). Olfactory nerves from garfish, obtained from the Gulf tetrodotoxin-sensitive ion fluxes through the nerve membrane. We conclude that the toxins act at a metal cation Specimen Co., Florida, were dissected by the method of binding site in the sodium channel and suggest that this Easton (12). A detergent-solubilized extract of the nerves site is the principal coordination site for cations (normally was prepared by the method of Henderson and Wang (4). Na+ ions) as they pass through the membrane during an Binding experiments were also carried out on intact garfish action potential. The dissociation constant for Li+ is olfactory nerves and rabbit vagus nerves (3, 5). Where pos- 0.1-0.2 M and for Na+ > 0.6 M, reflecting the weak binding necessary for rapid passage of sodium ions through the sible, means ± standard errors of the mean are given. channel. Binding Assays. Most of the experiments described below consisted of the measurement of tritium-labeled toxin bound Tetrodotoxin (TTX) specifically blocks the early inward at equilibrium to nerves in the presence of different cations. sodium current that underlies the excitability of nerve and The method depended on whether intact nerve or the deter- muscle plasma membranes (1). Since the determination of its gent-solubilized extract was being used; the intact nerve was structure in 1964 (2), its relatively small size (molecular soaked in a solution containing the toxin, and the radioactivity weight 319), single positively charged guanidinium group, taken up by the nerve was determined after tissue solubiliza- and numerous hydroxyl groups have provoked much specula- tion (3), whereas the detergent extract was assayed for bind- tion about how the toxin acts. ing activity by a gel filtration equilibration technique (4). Recent studies have shown that tritium-labeled tetrodo- Corrections for nonsaturable uptake (determined largely toxin (3, 4, 6) and saxitoxin (STX) (5), a compound with some from measurements at high toxin concentrations) and for similar structural features and identical physiological action toxin present in the extracellular volume (determined with (1), can be used both to estimate the number of binding sites ['4C]mannitol) were always applied in determining the inhibi- present in nerve and to characterize the properties of the tion of saturable binding in intact nerves. The experiments toxin binding sites. The equilibrium dissociation constants with intact nerves were limited by the fact that a 6-hr soak and the kinetic rate constants for the toxin-receptor inter- was required for full equilibration of the toxin with nerve (3). action so obtained (3-7) agree well with those measured The binding to the detergent extract, by contrast, took less electrophysiologically (8-11). The labeled toxins thus promise than 5 min (4), but was limited to the pH range 6.5-8.5 by to be powerful tools in the purification of the receptor, which the lack of stability of the binding activity outside this range. must form at least part of the sodium channel (4, 7). In the detergent extract there is no detectable nonsaturable Here, we report evidence that TTX and STX act at a uptake at physiological ionic strength. Most results were specific metal cation coordination site. All the divalent and checked on both preparations. All assays were carried out at trivalent cations tested, and the three monovalent cations, 200. Li+, Tl+, and H+ reversibly block the binding of TTX and Flux Measurements were carried out on intact garfish olfactory nerves. The method used was to measure the efflux of tracer amounts of 22Na in the presence of both ouabain Abbreviations: TTX, tetrodotoxin; STX, saxitoxin; TMA, (0.2 mM), to abolish the pumped efflux of 22Na, and veratrine tetramethyl ammonium. (5.0 Ag/ml), to keep the channels "open." The specific flux of * Present address: MRC Laboratory of Molecular Biology, Hills 22Na through the sodium channels was measured by comparing Road, Cambridge CB2 2QH, England. of efflux in the and absence TTX. The t Present address: Department of Physiology and Biophysics, the rate presence of State University of New York, Stony Brook, L. I., N.Y. 11790. efflux in the absence of TTX was normally 50-100% Reprints should be requested from: Prof. J. M. Ritchie, De- higher than with TTX present. Measurements in hypertonic partment of Pharmacology, Yale University School of Medicine, solutions of both NaCl and LiCl could be made easily since the New Haven, Conn. 06510. fibers had a mean diameter of only 0.25 Am (13), allowing a 3936 Downloaded by guest on October 1, 2021 Proc. Nat. Acad. Sci. USA 71 (1974) Cationic Binding Site in Sodium Channels of Nerve 3937

TABLE 1. Binding of the toxins to solubilized membrane 1.41 in the presence of added divalent and trivalent cations 1.2 Choline Toxin bound -6 (fraction of K.pp -S. Cation (mM) Toxin control) (mM) c 1.0 0 0 La' ++ nitrate, 25 STX -0.03 <1 C .2 0.8 Cs 5 STX 0.02 <1 TMA 0 0.5 STX 0.56 <1 K Sm3+ acetate, 1 STX 0.16 <1 la 0.6 1 STX -0.10 <1 C 0 2.5 STX -0.03 <1 .0c Er'+ bromide, 1 STX 0.33 <1 04 x Be2+ chloride, 50 STX 0.02 <1 10 TTX 0.02 <1 ).2 100 STX 0.25 49 [- Mg2+ chloride, .'1 Tl b 20 STX 0.53 15 4 Ca2+ chloride, 100 STX 0.19 15 0.1 0.2 0.3 0.4 0.5 100 STX -0.10 (<1) 0 Molority of added cation 50 STX 0.14 5 50 STX 0.34 17 FIG. 1. Binding of TTX (2 nM) in the presence of monovalent 50 TTX 0.28 13 cations to solubilized membrane extract. All solutions contained 25 STX 0.28 6 10 mM Tris buffer (pH 7.2) in addition to the indicated concen- 10 STX 0.41 5 tration of the added cation. Chloride was the anion in all cases 8 STX 0.73 14 except with Tl+, where the nitrate salt was used. Furthermore, 8 TTX 0.67 11 0.15 M NaCl was also present in the experiments with Tl+. Sr2+ chloride, 160 STX 0.08 6 Each point is the mean of two or three experiments. 20 STX 0.36 7 Ba2+ chloride, 100 STX 0.29 27 20 STX 0.35 7 range 5-50 mM. The corresponding values for the trivalent ions were less than 1 mM. The inhibition was reversed on Toxin concentration was 2 nM. The binding in presence of removal of the ions. It also greatly exceeded that due to the added cation is expressed as fraction (5) of corresponding binding increased ionic strength alone; addition of 0.5 M choline in the absence of added cation. All binding measurements were chloride had only a minimal effect on toxin binding (Fig. 1). carried out on the solubilized nerve preparation by addition of The possible contribution of surface potential changes to the the appropriate concentration of ions to normal bathing solution observed inhibition is considered below. [0.15 M NaCI, 10 mM Tris-HCl (pH 7.2)]. K... for each cation, of TTX binding experiments with C, was calculated from the relation: a = {2 + KXTX(1 + [Na]/ Fig. 1 shows the results KN)}/{2 + KXTX(1 + [Na]/KN. + [C]/Kapp)I. Krrx was seven monovalent cations; similar experiments with STX gave taken as 9 nM (4, 5), KSTX as 7 nM (4, 5), KNa as 0.6 M (Fig. 1). similar results. With Tl+ and Li+ there was an appreciable inhibition of toxin binding that was clearly greater than the rapid equilibration of the internal ionic composition of the effects of any of the other monovalent cations; this inhibition nerve after the presumed initial shrinkage (see refs. 11 and was reversible. The inhibition can again be explained by a 14 for a more complete description of the methods). competitive inhibition with dissociation constants of about 10 mM for Tl+ (9.9 i 0.6 mM for the observations in Fig. 1) and RESULTS 0.13 i 0.01 M (four determinations) for Li+. The inhibition inhibition by Cations of Toxin Binding to Solubilized Mlem- by the other monovalent ions was smaller, and the differences branes. Table 1 gives the results of toxin binding assays in the between them were only slightly greater than the standard presence of eight different divalent and trivalent cations, the errors. However, the sequence Li+ > Na+ > K+ > Cs+, tetra- cation being added in the appropriate concentration to a methyl ammonium (TMA+), choline+ was clearly discerned. If solution containing 0.15 M NaCl, 10 mM Tris-HCl (pH 7.2). the effect of Na + is due to a direct competitive binding, the dis- In all cases there was an inhibition of binding to the solubilized sociation constant would have to be 0.6 i 0.1 M (four deter- membranes that was reversible and comparable in degree to minations), if the combined data for Cs+, TMA+, and choline+ that in whole nerve. If we assume that the reduction in toxin are taken to give the baseline below which the specific Na + binding is due to competitive inhibition, we can calculate an inhibition is measured. The cause of the relatively small effect equilibrium dissociation constant for the cation binding. Since of Cs+, TMA+, and choline+ themselves on TTX binding is competitive inhibition is only the simplest of many possible unknown, but the effect is clearly an order of magnitude mechanisms that might cause inhibition of toxin binding, the smaller than the effects of the other cations discussed above. results of such calculations yield only an "apparent" dissocia- Inhibition of Toxin Binding to Intact Nerve by Cations. Fig. 2 tion constant (Kapp) for the cation. The agreement obtained shows the dependence on pH of the binding of TTX to intact between values from experiments at different cation concen- garfish nerve. The inhibition at low pH is reversed when the trations (see ref. 5 and Table 1) seems to justify this proce- nerves are returned to pH 7. The curve, which is a. theoretical dure. titration curve, indicates that the binding site has a pKa of Except for Be2+, for which Kapp was less than 1 mM, all the about 5.5, in reasonable agreement with a pKa of 5.9 obtained divalent cations had apparent dissociation constants in the with STX on the rabbit vagus nerve (5). It is the toxin Downloaded by guest on October 1, 2021 3938 Physiology: Henderson et al. Proc. Nat. Acad. Sci. USA 71 (1974)

100 TABLE 2. Saturable binding of the toxins to whole nerve in the presence of varius cations

cNj 80 Apparent I disso- Toxin bound ciation a 60 Cation Toxin (test/control) constant c (mM) (nM) gar olfactory rabbit vagus (mM) 1O- co 40 A La'+, 1 TTX, 4 0.60: 0.06 (3) 1 5 TTX,4 0.48±0.11(3) 2 * 20 Ca'+, 88 TTX, 4 0.4.5 4 0.02 (3) 46 88 STX, 4 0.23 At 0.02 (2) 16 113 STX, 4 0.31 4 0.06+ (4) 29 0 Tl', 121 STX, 4 0.18 4 0,05 + (4) 16 4 5 6 7 8 100 TTX, 4 0.25±- 0.08 (3) 21 pH Lit, 154 TTX, 4 0.77 i 0.03 (3) 344 154 STX, 3 0.46 i 0.73 80 FIG. 2. pH dependence of tetrodotoxin binding to garfish 1.54 STX, 3 0.53 ± 0.06 (3) 113 olfactory nerve. 2 nM TTX, 6-hr soak. 120 TTX, 4 1.07 ± 0.10 (9) * K+, 154 STX, 4.5 0.45i:0.15 (3) 75 120 TTX, 4 1.14 0.18 (4) * receptor that is being titrated by H+, since the pK. of TTX is Cho- 154 TTX, 4 1.06 ± 0.02 (3) * 8.5 and the toxin is quite stable over the range of pH studied. line, 154 STX, 2 1. 36 ± 0.226 (3) * Table 2 shows the effect of monovalent, divalent, and tri- B valent cations on toxin binding to garfish olfactory and rabbit Li+, 850 TTX, 4 0.39 ± 0.06 (4) 3o90 vagus nerves. In general, the results agree with those from Na+, 850 TTX, 4 0.68±4 0.06 (4) 1300 soltubilized membranes (Table 1); the trivalent cation (La3+) Toxin bound (±i standard error with number of experiments in shows a dissociation constant near 1 mM; the divalent cation. parentheses) is expressed as fraction of that in normal bathing (Ca2+) has an average Kca of about 30 mM. Inhibition of medium in A, and of that in Na-free gar R1inger's solution con- toxin binding by monovalent cations is consistent with the taining 0.85 M choline chloride in B. Part A: the cations replaced results in solubilized membranes except for two results. equiosmotic amounts of Na+ in the bathing medium (except for Replacement of Na+ by K+, which had relatively little effect in La3+ experiments, where no Na + was removed). In calculation of the solubilized preparation (Fig. 1), halved the binding of apparent dissociation constant for cationsKSTx was taken as 7 STX to the intact rabbit nerve; and replacement of Na+ by nM for both gar olfactory nerve (see ref. 4 for similar value in Li+ at 120 mM, which reduced binding in the solubilized solubilized gar membrane) and intact rabbit nerve (5); KTrx was preparation, did not inhibit TTX binding to garfish nerve taken as 3.0 nM for rabbit nerve and 10 nM1 for garfish nerve (3); at 0.85 M total cation concentration the results with KNa was taken as 1300 mM. Part B: as in A except Li + and Na+ (although were added to Na+-free garfish Ringer's solution. All solutions Li+ were more consistent with those from solubilized mem- contained 0.01 M morpholinopropane sulfonate, pH 7.2. branes). * High values not calculable. Contributions from the Membrane Surface Potential. One explanation of the Inhibition by divalent and trivalent ions invokes changes of the surface potential at the nerve mem- If it is assumed that changes in the solution concentration of brane. If fixed negative charges on the membrane exist near multivalent metal cations affect TTX and STX binding the toxin binding site, they will create an electric potential at equally, except through the modification of surface charge, the binding site, 4s, that differs from the bulk solution po- then the following calculation can be applied to determine the tential, taken as zero. The multivalent ions could alter the surface electric potential. surface potential, and, hence, the amount of toxin bound; By dividing Eq. [2] for STX by Eq. [21 for TTX and taking either by binding directly to the fixed charges or by screening the logarithm, one derives: in the fixed charges by accumulating the double layer adjacent In Kapp(STX)/Kttpp(TTX) = In Ko(STX)/Ko(TTX) + to the membrane (15). This surface potential will influence the toxin concentration at the binding site [Tox]. according to qs F/RT [31 Boltzmann statistics (15) and modify the apparent binding constant of the toxin cation, Kapp. From the Boltzmann since STX is charged +2 and TTX is charged +1 at pH 7.0. equation: Table 3 shows how the surface potential, calculated on the basis of Eq. [3], is decreased by increasing external calcium [Tox]8 = [Toxlbulk exp - t'8ZTF/RT [11 concentration, particularly in the absence of other external and, therefore, cations. This decrease must contribute to the decrease in the apparent affinity of the membrane for the toxin Kapp = Ko exp - 4&8ZTF/RT [2] produced by calcium (and other cations) particularly at low where ZT is the charge of the toxin molecule, [Toxlbulk is its ionic strength. It cannot, however, be the only factor involved. bulk solution concentration, Ko the intrinsic equilibrium dis. Thus, the variation of Kapp with changing Ca++ obtained in sociation constant. F, R, and T have their usual meanings. the experiment of Table 3 was always greater than the varia- Therefore, the apparent affinity of the site for STX (ZT = 2) tion calculated from the surface potential change only (ob- should be affected more than that for TTX (ZT = 1) by tained from Eq. [3] and Table 3). factors that alter As (such as changing external [Ca2+]). The change in surface potential with calcium at the toxin- Indeed, Table 3 shows that a given increase in [Ca2+] reduces binding site in 0.12 M NaCl is significantly less than the 20- the specific uptake, U, of STX much more than that of TTX. mV shift per 10-fold increase in [Ca2+] observed for the Downloaded by guest on October 1, 2021 Proc. Nat. -tcad. Sci. USA 71 (1974) Cationic Binding Site in Sodium Channels of Nerve 3939

TABLE 3. The effect of calcium on the membrane

surface potential of the toxin binding site C E 1.00 [Ca2+] [Tox] U* Kapp Al,* 0 z .02 (mM) (nm) (fmol/mg) (nM) (mV) N r_ 0 A. Intact gar nerves in 0.12 AI NaCi (high ionic strength) C-i0 0.75 0 0 3.5 TTX, 4 179 9 0 c STX, 4 93 21 0 26 TTX, 4 134 13 5.8 0.50 a STX, 4 56 38 x .01 88 TTX, 4 100 19 6.7 c - STX, 4 38 57 0 I- 0.25 m

B. Intact gar nerves in 0.01 Al morpholinopropane sulfonate x (low ionic strength) 0 1 TTX, 4 163 10.3 -8.5 0 0 STX, 4 245 5.5 0 01 02 03 0.4 0.5 3.5 TTX, 4 157 10.9 0 Concentration of Lithium (M) STX, 4 194 8.1 10 TTX, 4 114 16.5 1.3 FIG. 3. Comparison of the effect of replacing [Na ] with STX, 4 138 12.9 [Li+] on: TTX- sensitive efflux of 22Na from intact garfish nerves 20.2 TTX, 4 107 17.8 13.4 (0) and the binding of TTX (0) or STX (0), concentration 2 STX, 4 88 22.5 nM, to the solubilized nerve membrane extract. 43.5 TTX, 4 88 22.5 19.2 STX, 4 5)9 35.6 inhibited at Li+ concentration of about 0.3 M, and is reversed 87 TTX, 4 71 28.9 21.0 when the nerves are returned to a lithium-free solution. If KNa STX, 4 44 49.1 is taken to be 1.3 M (see Table 3), KLM can be calculated to be C. Solubilized gar nerve in 0.15 Al NaCI 0.12 M. A similar measurement of radioactive toxin binding to 0 TTX, 4 0.275 6.0 0 the solubilized membrane preparation, again replacing 0.5 M STX, 2 0.256 6.6 NaCl with LiCl (Fig. 3), also shows a 50% reduction when 8 TTX, 2 0.184 10 -2.9 0.3-0.4 M Li+ replaces Na+. The six points in Fig. 3, where the STX, 2 0.187 9.8 toxin bound is reduced by more than 30%, when a sufficiently 50 TTX, 2 0.077 27 -6.5 accurate calculation can be made, give a value for KLM STX, 2 0.087 23 (calculated as in Table 1) of 0.20 ± 0.03 M, in reasonably good agreement with the value obtained from the efflux measure- The standard error of the mean was, on average, about 8% of ments. the mean value. Each value has been corrected for the amount The experiment on 22Na efflux suffers from a number of dis- on space and for In rabbit the extracellular nonspecific binding. advantages, mainly due to the method used. For example, the nerve full replacement of Na by Ca2+ (113 mM) caused the extracellular space to increase by 26 i 7% (13 observations): effect of the veratrine on sodium channel conductance may be in two experiments on garfish nerve there was a decrease of 13% altered by lithium. Several authors (see ref. 18) have shown in 88 mM Ca2+. The nonspecific binding was assumed to remain that nerves depolarized by veratrine are repolarized when constant. Since the combined amount in the extracellular sodium is replaced by lithium; these potential changes per se space and in the nonspecific compartment is so small relative to must affect the sodium fluxes. Nevertheless, we are unable to the amount specifically bound at the ion concentration of toxin devise any other method of demonstrating the effect of used, any error in estimating these amounts would be corre- lithium on fluxes through the sodium channels in these very spondingly diminished during the correction procedure and small, nonmyelinated fibers. The effect observed, however, is would not greatly affect the calculations. (Three tests at high an inhibition of 22Na efflux by lithium that roughly parallels toxin concentration in rabbit nerve in 113 mM Ca2+ gave an increase in total uptake of 15 11% above that in normal Locke the ability of lithium to displace the radioactive toxins from solution.) their binding site. At least in the range of lithium concentra- * U is the uptake of toxin per mg of dry nerve. Maximum tion below 0.3 M, the agreement is fairly good. A similar uptake is 584 fmol/mg (3, 5). Each value of U is the mean of inhibition of sodium currents by lithium also occurs in intact three or four separate estimates. nerve fibers untreated with drugs such as ouabain and vera- * AV,& is change in An, calculated as in text (Eq. [3]) from arbi- trine. Hille (19), working with frog nodes of Ranvier, and Cole trarily chosen zero potential. (20), with squid giant axons, have reported a 20-40% decrease of the peak transient currents through the sodium channels in sodium permeability function under similar conditions (16, experiments where the sodium in the normal Ringer's solution 17); increasing [Ca2+] more than 20-fold (from 3.5 to 88 mM) was completely replaced by lithium. changed bs by less than 7 mV (Table 3). Apparently, the toxin-binding site is not located at the same region of mem- DISCUSSION brane enclosing the voltage sensor that modulates the sodium These results demonstrate certain properties of the site at permeability. which TTX and STX act. The site is able to bond at least Inhibition of 22Na Efflux by Lithium. Fig. 3 shows the effect eight divalent and trivalent cations, and the monovalent of Li+ on the TTX-sensitive efflux of 22Na. Lithium progres- cations Li+, Tl+, and H+ (pKa, 5.5-5.9). All previously sively replaces sodium in the hypertonic (0.85 M cation) observed blockage of sodium currents by Ca2+ (21), H+ (21), medium bathing intact garfish nerves. The efflux is half Tl+ (19), and Li+ (19, 20) show concentration dependences Downloaded by guest on October 1, 2021 3940 Physiology: Henderson et al. Proc. Nat. Acad. Sci. USA 71 (1974) that are consistent with cation binding at this site. The sodium ion transit but with weaker binding, so that no satura- dissociation constants we observe for the binding of these tion of the sites can be detected at nearly physiological ion cations to the toxin site are within a factor of two of those concentrations. The dissociation constant for Na+ would have required to explain the blockage phenomena in the electro- to be greater than or equal to 0.6 M. physiological experiments. Kao and Nishiyama (26) have suggested that both TTX Since the two toxins are cations, the binding site may well and STX act by "plugging up" the sodium channels in the be negatively charged, encompassing perhaps the same acidic membrane, thus preventing the passage of ions. The guani- group postulated by Hille (16), so it might not seem remark- dinium group-a structural feature of both the toxins-was able that metal cation also bind. However, the charge at a envisaged as entering the mouth of the channel, as a sodium site, and the ability of the latter to bind ions, are in fact ion might, but staying there, unable to proceed further. separable factors. One can cite two examples of enzymes that Further developing this idea, Hille (24) has suggested that the bind small positively charged effectors or substrates of a size toxins mav act at a narrow, oxygen-lined part of the channel similar to TTX, but that do not bind metal cations. Acetyl- designed to account for the channel's ionic selectivity prop- cholinesterase and the proteolytic enzyme, trypsin, are both erties. The present work further elucidates the nature of the proteins with negatively charged binding sites, in one case a channel, revealing as one of its important components a site for the choline moiety and in the other case, for the side cation-binding site that is probably involved in the passage of chain of the lysine substrate. For trypsin, x-ray studies have ions across the membrane and to which the toxins TTX and shown that the binding site contains an aspartic acid side STX bind to exert their physiological action. chain (23). In neither case (22, 23) do high concentrations of calcium, for example, have any inhibitory effect on binding, This work was supported by grants from the USPHS, NS- thus demonstrating that the binding of positively charged 08304, and GM-04483. 11.H. held a fellowship from the Helen ligands can easily occur without any strong affinity for some Hay Whitney Foundation. metal cations. None of 1. Kao, C. Y. (1966) Pharmacol. Rev. 18, 997-1049. the calculated metal cation dissociation constants is 2. Woodward, it. B. (1964) Pure Appl. Chem. 9, 49-74. very low; indeed, the binding of lithium and sodium is quite 3. Colquhoun, D., Henderson, R. & Ritchie, J. M. (1972) J. weak. But if the toxin-binding site is actually on the pathway Physiol. (London) 227, 95-126. normally followed by sodium ions as they traverse the mem- 4. Henderson, R. & Wang, J. H. (1972) Biochemistry 11, 4565- brane, a tight binding site for sodium would be contra- 4569. 5. Henderson, R., Ritchie, J. M. & Strichartz, G. R. (1973) indicated; any tight binding would not allow the very high J. Physiol. (London) 235, 783-804. flux rates that are observed, for the flux rate would then be 6. Benzer, T. I. & Raftery, M. A. (1972) Proc. Nat. Acad. Sci. limited by the speed at which the ion could dissociate from the USA 69, 3634-3637. site. 7. Benzer, T. I. & Raftery, M. A. (1973) Biochem. Biophys. The existence of a negative in the sodium Res. Commun. 51, 939-944. charge channel, 8. Schwartz, J. R., Ulbricht, W. & Wagner, H. H. (1973) J. with an apparent pKa of between 5 and 6, seems now well Physiol. (London) 233, 167-194. established (16, 21). Hille (24) has provided a hypothetical 9. Hille, B. (1968) J. Gen. Physiol. 51, 199-219. description of the environs of this acidic group and the pos- 10. Colquhoun, 1). & Ritchie, J. M. (1972) J. Physiol. (London) sibility that it might bind TTX and STX. Support for this 221, 533-553. 11. Henderson, R. & Strichartz, G. (1974) J. Physiol. (London) model has until now rested on voltage-clamp experiments on 238,329-342. frog nodes (16, 21, 25). The present experiments provide 12. Easton, D. M. (1965) Cold Spring Harbor Symp. Quant. direct evidence for this hypothesis. In the light of the con- Biol. 30, 15-28. siderations in the previous paragraphs, we suggest that the site 13. Easton, D. M. (1971) Science 172, 952-955. is 14. Caterall, W. A. & Nirenberg, M. (1973) Proc. Nat. Acad. of toxin binding is a single metal cation binding site that the Sci. USA 70,3759-3763. main coordination site for sodium ions as they pass through 15. McLaughlin, S. G. A., Szabo, G. & Eisenman, G. (1971) the membrane. We again emphasize the agreement between J. Gen. Physiol. 58, 667-687. our observations of the effect of Ca2+ and H+ ions on toxin 16. Hille, B. (1968) J. Gen. Physiol. 51, 221-236. binding and those of Woodhull who observed a mem- 17. Frankenhaeuser, B. & Hodgkin, A. L. (1957) J. Physiol. (21), (London) 137, 218-244. brane potential-dependent block of sodium currents in frog 18. Ubricht, W. (1969) Ergeb. Physiol. Biol. Chem. Exp. Phar- nodes of Ranvier by both Ca2+ and H+ ions. Wagner and makol.61, 18-71. Ulbricht (25), confirming Woodhull's observations, have 19. Hille, B. (1972) J. Gen. Physiol. 59, 637-658. suggested, in addition, that H+ and TTX compete for the 20. Cole, K. S. (1968) in Membranes, Ions and Impulses (Univ. would our measurements with Calif. Press, Berkeley, Calif.), pp. 452-453. same site. We also emphasize 21. Woodhull, A. M. (1973) J. Gen. Physiol. 61, 687-708. Li+ (with the exception of those on garfish nerve in 0.12 M 22. Dudai, Y. & Silman, I. (1973) FEBS Lett. 30,49-59. Li+); the effect of lithium in reducing the binding of toxin to 23. Stroud, R. M., Kay, L. M. & Dickerson, R. E. (1971) Cold intact nerve (Fig. 3 and Table 2) agrees fairly well with the Spring Harbor Symp. Quant. Biol. 36, 125-140. effect of lithium in reducing the toxin-sensitive ion fluxes. This 24. Hille, B. (1971) J. Gen. Physiol. 58, 599-619. 25. Wagner, H.-H. & Ulbricht, W. (1974) Eur. J. Physiol. 347, implies that for lithium ions the toxin-binding site is the R34. principal site of coordination as they pass through the mem- 26. Kao, C. Y. & Nishiyama, A. (1965) J. Physiol. (London) brane. By analogy, a similar situation must occur during 180, 50-66. Downloaded by guest on October 1, 2021