Neurotoxin Binding to Receptor Sites Associated with Voltagesensitive Sodium Channels in Intact, Lysed, and Detergent-solubilized Brain Membranes*

Transcription

1 Vol. 254, No. 22, Issue of November 25, pp , 1979 Printed in U.S. A Neurotoxin Binding to Receptor Sites Associated with Voltagesensitive Sodium Channels in Intact, Lysed, and Detergent-solubilized Brain Membranes* (Received for publication, June 5, 1979) William A. Catterall, Cynthia S. Morrow, and Robert P. Hartshorne From the Department of Pharmacology, School of Medicine, University of Washington, Seattle, Washington [3H]Saxitoxin binds to a single class of receptor sites in rat brain synaptosomes and in broken membrane fractions derived from rat brain with a KD of approximately 2 11~ at 36 C. Batrachotoxin and scorpion toxin, which act at different receptor sites to activate sodium channels, have no effect on saxitoxin binding. Osmotic lysis and depolarization also have no effect on saxitoxin binding. These results show that saxitoxin binds equally well to resting, active, and inactivated sodium channels. Saxitoxin binding sites were concentrated in the crude mitochondrial fraction in subcellular fractionation experiments. Further separation of this fraction on discontinuous sucrose gradients showed that the receptor sites were enriched in fractions containing broken neuronal membranes, synaptosomes, and synaptic plasma membrane. Scorpion mono[ 251]iodotoxin also binds to a single class of receptor sites in synaptosomes. Comparison of saxitoxin and scorpion toxin binding to synaptosomes indicates that there are 3.7 saxitoxin receptor sites for each scorpion toxin receptor site. Lysis or depolarization causes a marked inhibition of scorpion toxin binding. Batrachotoxin, veratridine, and aconitine enhance scorpion toxin binding in intact synaptosomes and partially restore binding in lysed membrane fractions. Competitive interactions among these toxins in enhancing scorpion toxin binding in lysed membrane fractions indicate that they act at a common binding site. These results with lysed membrane fractions show that both the allosteric interactions between scorpion toxin and the lipid-soluble toxins and the competitive interactions among the three lipid-soluble toxins are retained in the absence of membrane potential and ion gradients. Dissolution of brain membranes with 1% Triton X- 100 results in solubilization of up to 40% of the saxitoxin receptor sites with no change in KD for saxitoxin. The soluble binding activity can be stabilized by addition of phosphatidylcholine and Ca +. Analysis of the solubilized saxitoxin receptor by sucrose density gradient velocity sedimentation indicates a single peak of binding activity with a sedimentation coefficient of 10 S. Under similar conditions, scorpion toxin binding activity is lost irreversibly. Binding of scorpion toxin may depend upon the integrity of the bilayer membrane. Voltage-sensitive sodium channels in mouse neuroblastoma * This work was supported by Grant HL from the National Institutes of Health and a grant from the University of Washington Alcohol and Drug Abuse Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. cells and in rat brain synaptosomes have three separate receptor sites for neurotoxins (1, 2). One receptor site binds the inhibitors saxitoxin and tetrodotoxin (3,4). A second receptor site binds the lipid-soluble toxins batrachotoxin, veratridine, aconitine, and grayanotoxin (1, 5, 6) which alter the voltage dependence of sodium channel activation and inactivation (7-9) and cause persistent activation of sodium channels by an allosteric mechanism (1, 6). A third receptor site binds the polypeptides scorpion toxin and sea anemone toxin II (1, lo- 12) which block sodium channel inactivation (13-20) and interact cooperatively with the lipid-soluble toxins to cause persistent activation of sodium channels (1,6). The binding of scorpion toxin to its receptor site in neuroblastoma cells, synaptosomes, and frog skeletal muscle is voltage-dependent (10, 21, 22). The voltage dependence of binding is closely correlated with the voltage dependence of activation of the sodium channel (22). These toxins provide valuable probes of the structure and function of sodium channels and can potentially be used to detect and purify sodium channel components solubilized from excitable membranes. The tetrodotoxin receptor has been successfully solubilized from membrane preparations of garfish olfactory nerve and eel electroplax using nonionic detergents (23-25) and the receptor from electric eel has been partially purified (25). In this report, we describe the binding of scorpion toxin and saxitoxin to their receptor sites in intact, osmotically lysed, and detergent-solubilized membrane preparations from rat brain. A preliminary report of some aspects of these results has been presented (26). EXPERIMENTAL PROCEDURES Materials-Chemicals were obtained from the following sources: scorpion venom (Leiurus quinquestriatus) from Sigma, tetrodotoxin from Calbiochem, and Triton X-100 from New England Nuclear. Batrachotoxin was generously provided by Drs. J. Daly and B. Witkop (National Institute of Arthritis, Metabolism and Digestive Diseases, National Institutes of Health). Saxitoxin was generously provided by Dr. E. Schantz, University of Wisconsin. Scorpion toxin was purified and radiolabeled by lactoperoxidase-catalyzed iodination as described previously (10). Preparation of Membrane Fractions-The methods used are modifications of the procedures of Gray and Whittaker (27) for preparation of synaptosomes and Cotman and Matthews (28) for preparation of synaptic plasma membrane. The brain was removed from a male Sprague-Dawley rat and homogenized in ice-cold 0.32 M sucrose, 5 mm sodium phosphate, ph 7.4, 0.1 mm phenylmethanesulfonyl fluoride, at 10% (w/v) with 10 strokes of a motor-driven Teflon/glass homogenizer. The resulting homogenate was sedimented at 1000 x g for 10 min. The supernatant was saved and the sediment was resuspended in 10 ml of 0.32 M sucrose and again sedimented at 1000 X g for 10 min. The pellet, designated Pl, was discarded except when the distribution of receptor sites throughout the fractionation scheme was to be studied. The two supernatants were combined to give Fraction Sl + S2 and sedimented at 17,000 x g for 60 min. The supernatant, designated S3, was discarded except in distribution studies. The pellet P3, containing a mixture of synaptosomes, broken membranes, and 11379

2 11380 Sodium Channels in Brain mitochondria, was used for binding assay directly, for preparation of synaptosomes, or for preparation of the lysed P3 fraction. To prepare lysed P3, the P3 preparation was homogenized in 3.3 ml/g of original brain wet weight of 5 mm Nap,, ph 7.4, 0.1 mm phenylmethanesulfonyl fluoride. After a 30-min incubation at O C, an equal volume of 260 mm choline chloride, 100 mm Hepes (adjusted to ph 7.4 with Tris base), 11 mm glucose, 1.6 mm MgS04, 10.8 mm KC1 was added. The fraction was then used in binding experiments. To prepare synaptosomes, the P3 fraction was layered onto a stepwise sucrose gradient consisting of lo-ml layers of 1.2, 1.0,0.8,0.6, and 0.4 M sucrose and sedimented at 100,000 X g for 105 min. As described previously (2), the heavy synaptosome fractions in the 1.0 M sucrose layer and the 1.0 M-1.2 M sucrose interface were collected, sedimented, and used in experiments. To prepare synaptic membranes according to Cotman and Matthews (28), the heavy synaptosome fraction from six rat brains was suspended in 10 ml of 10% (w/v) sucrose. The suspension was then diluted with 40 ml of ice cold 6 mm Tris-HCl, ph 8.1, and incubated at 0 C for 1 h. The membranes were collected by sedimentation at 75,000 x g for 30 min and resuspended in 18 ml of 10% sucrose per original rat brain at 0 C. The suspension was then layered on two discontinuous sucrose gradients consisting of lo-ml layers of 3870, 35% 32.5%, and 25% sucrose. The samples were sedimented at 100,006 x g for 90 min. The sucrose layers were collected and the 25% layer containing the synaptic plasma membrane was saved for assay. Solubilization of the Saxitosin Receptor-The P3 fraction was prepared and lysed in 5 mm sodium phosphate, ph 7.4, 0.1 mm phenylmethanesulfonyl fluoride as described above. The lysed membrane fraction was diluted with an equal volume of 2% Triton X-100, 200 mm choline chloride, 40 mrvr Hepes (adjusted to ph 7.4 with Tris base), 0.1 mm phenylmethanesulfonyl fluoride and incubated for 30 min at 0 C. The detergent solution was sedimented at 164,000 x g for 30 min and the resulting supernatant was removed and used in experiments. Preparation and Characterization of (3H/Saxitoxin-[ H]Saxitoxin was prepared by the specific [ H]HzO exchange technique of Ritchie et al. (29). Dry saxitoxin (1.8 mg) was dissolved in [ H]HLO (25 Ci) and incubated at 50 C for 3 h. Labile H was removed in U~CUO with Hz0 rinses at 0 C. These operations were performed at New England Nuclear Corp. Aliquots of this material were then purified by paper electrophoresis as described by Ritchie et al. (29). Fractions from the electrophoretogram were tested for saxitoxin biological activity and for H counts per min. The concentration of saxitoxin in each fraction was determined by comparing the inhibition of sodium channel ion flux in neuroblastoma cells by [ Hlsaxitoxin fractions and by standard saxitoxin solutions as described previously (4). A standard saxitoxin solution obtained from the U. S. Food and Drug Administration was used. The radiochemical purity of the [ Hlsaxitoxin was estimated by determining the fraction of H counts per min which can be specifically bound to brain membrane preparations as described previously (4). After purification by electrophoresis, several samples of r H]saxitoxin ranged from 50 to 65% pure. Careful analysis of radioactivity and biological activity of fractions cut from the peak of radioactivity migrating with a charge of +2 in the paper electrophoretogram revealed that it contained two components: saxitoxin and radioactive material migrating slightly faster than saxitoxin and having no biological activity. These two compounds could be separated by thin layer chromatography on Silica Gel F in ethanol:water:acetic acid (65: 70:25) at 4 C. In this solvent system, saxitoxin migrated with an RF of 0.57 while the radioactive impurity had an RF of [ HlSaxitoxin purified by this method was 86% to 90% radiochemically pure. The concentration of saxitoxin in the [ Hlsaxitoxin preparation was estimated by two methods. The inhibition by [ Hlsaxitoxin of Na+ influx due to activation of sodium channels in neuroblastoma cells was measured and compared to inhibition by standard saxitoxin samples from the U. S. Food and Drug Administration as described previously (4). In addition, the concentration was estimated by an isotopic dilution assay essentially as described by Reed and Raftery (30) for tetrodotoxin. The P3 membrane fraction was prepared and specific binding of [ Hlsaxitoxin was measured at concentrations from 0.1 nm to 5 nm saxitoxin as described below. These measurements were carried out in the presence and absence of 3 nm standard saxitoxin and the specific radioactivity and concentration of [ HIsaxitoxin were calculated as described (30). These two methods gave The abbreviation used is: Hepes, 4-(2.hydroxyethyl)-l-piperazineethanesulfonic acid. similar results. The preparations of [ Hlsaxitoxin used ranged from 9 to 22 Ci/mmol as estimated by these methods. Experiments with toxin samples of 86% to 90% radiochemical purity and those of 50% radiochemical purity gave comparable results when appropriate corrections for purity were made in the calculation of specific radioactivity. Measurement of I-labeled Scorpion Toxin Binding-Pure scorpion mono[ I]iodotoxin was used (10). The specific radioactivity is calculated from the specific radioactivity of the I- in the labeling reaction and is known with good precision. Scorpion toxin binding was measured using a rapid filtration assay as described previously (2). Brain membrane fractions were incubated with Z I-labeled scorpion toxin in standard binding medium consisting of 130 mm choline chloride, 50 mm Hepes (adjusted to ph 7.4 with Tris base), 5.5 mm glucose, 0.8 mm MgS04, 5.4 mm KCl, and 1 mg/ml of bovine serum albumin. Binding reactions were initiated by addition of brain membranes suspended in 25 ~1 of standard binding medium minus bovine serum albumin to a reaction mixture containing scorpion mono- [ Iliodotoxin and other effecters as noted in the figure legends ir 175 ~1 of standard binding medium. The samples were mixed and incubated for 30 min at 36 C. The reactions were then stopped by addition of 3.0 ml of wash medium at 36 C consisting of 163 rn~ choline chloride, 5 mm Hepes (adjusted to ph 7.4 with Tris basil, 1.8 mm CaCIZ, 0.8 mm MgSO,, and 1 mg/ml of bovine serum albumin. The membranes were immediately collected on glass fiber filters (Whatman GF/C) under vacuum and washed three times with wash medium. The filters were then placed in counting vials and radioactivity was measured. Nonspecific binding measured in the presence of 200 nm unlabeled scorpion toxin was subtracted from the results (2). All data are presented as a function of the free scorpion toxin concentration. Measurement of ( H]Saxitoxin Binding to Brain Membranes by Rapid Filtration-Binding of [ Hlsaxitoxin to brain membranes was measured routinely by a rapid filtration procedure as for scorpion toxin binding. Brain membranes suspended in 25 ~1 of standard binding medium minus bovine serum albumin were added to a reaction mixture containing [: H]saxitoxin and other effecters as noted in the figure legends in 175 ~1 of standard binding medium. The samples were mixed and incubated 20 min at 36 C. The reactions were then stopped by addition of 3.0 ml of wash medium without bovine serum albumin at 0 C and the membranes were collected by filtration and washed twice in 10 s. The filters were then placed in counting vials and radioactivity was determined. Nonspecific binding was measured in the presence of a saturating concentration (1 pm) of tetrodotoxin and was subtracted from the results (4). All data are presented as a function of the free saxitoxin concentration. Preliminary kinetic studies showed that the binding reaction was complete in 10 min at 36 C at the lowest concentration studied and that dissociation of the toxin. receptor complex had a half-time of 2 min at 0 C. The rapid filtration binding measurements therefore are adequate to measure the saxitoxin. receptor complex at equilibrium. This conclusion was verified by comparing the rapid filtration method with equilibrium dialysis carried out as described below. Measurement of ( HJSaxitoxin Binding by Equilibrium Dialysis-Binding of [ Hlsaxitoxin was also measured by equilibrium dialysis. This method was used for studies of the solubilized saxitoxin receptor and to verify that the rapid fntration assay measured true equilibrium binding for membrane fractions. Samples (100 ~1) containing either brain membranes in standard binding medium minus bovine serum albumin or soluble extracts in 100 mm choline chloride, 20 mm Hepes (adjusted to ph 7.4 with Tris base) were added to one side of a microdialysis assembly and [ Hlsaxitoxin and other effecters in the same solution were added to the other side. The dialysis assembly was stirred by rotation for 3 h at 0 C to allow complete equilibration. Samples were then removed from both sides of the dialysis membrane using a microliter syringe and radioactivity was measured. Bound [: H]saxitoxin was calculated from the difference in radioactivity between the compartments. Nonspecific binding was estimated from measurements in the presence of 1 FM tetrodotoxin and the small amount of nonspecific binding observed was subtracted from the results. [ H]Saxitoxin purified by thin layer chromatography (86% to 90% radiochemical purity) was used in all equilibrium dialysis studies so that the background of unbound radioactive impurity did not interfere with measurements at low concentrations of saxitoxin where most of the toxin is bound. Measurement of (3H/Saxitoxin Binding to Solubilized Membrane Preparations by Rapid Gel Filtration-The solubilized saxitoxin receptor site is unstable. Studies of the solubilized receptor were

3 Sodium Channels in Brain facilitated by use of a rapid gel fiitration assay for measurement of tim&ng as prel;lously aescl?bea \25,31j. Columns (1.3 my) of Sepha?lex G-50 equilibrated with 100 rnm choline chloride, 20 mm Hepes (adjusted to ph 7.4 with Tris base), 0.1% Triton X-100, 0.02% (w/v) phosphatidylcholine (from chicken egg) in 3-ml plastic syringes were centrifuged at 1000 x g for 1 min. Binding reactions were initiated by mixing membrane extract and [ Hlsaxitoxin in a final volume of 250 ~1 of 100 mm choline chloride, 20 mm Hepes (adjusted to ph 7.4 with Tris base), 0.2% Triton X-100, 0.04% egg phosphatidylcholine. After incubation for 20 min at O C, 200 ~1 of the reaction mixture was layered onto a Sephadex column prepared as above and centrifuged at 1000 x g for 1 min. The eluate containing the void volume of the column was collected. The saxitoxin.receptor complex is large and elutes in the void volume. Less than 5% of the free saxitoxin elutes under these conditions. The concentration of the [3H]saxitoxin. receptor complex was determined by measuring the radioactivity in the eluate. These measurements were verified by comparison with equilibrium dialysis. In companion experiments using identical solubilized extracts and [ Hlsaxitoxin preparations, the two methods gave identical values for KU, but the rapid gel filtration method underestimated the value of B,,,, by approximately 15%. This seemed to result from trapping of a fraction of the receptor. toxin complex in the column. In view of this small underestimation of binding capacity, the rapid gel filtration method was not used when precise estimation of the binding capacity was required. Measurement of 2 I-labeled Scorpion Toxin Binding to Solubilized Membrane Preparations by Rapid Gel Filtration-Scorpion toxin binding could not be measured by equilibrium dialysis because nonspecific binding to the dialysis membrane was high and the equilibration rate was extremely slow. Therefore, attempts were made to measure solubilized scorpion toxin binding activity using a rapid gel filtration assay that was modified from the saxitoxin binding assay. Bio-Gel P30 was used to reduce nonspecific binding of scorpion toxin to the column matrix. Batrachotoxin (1 pm) was added to enhance scorpion toxin binding. Although no soluble scorpion toxin binding activity was detected in detergent extracts of brain membranes (see Results ), the assay procedure provided an accurate measure of binding of I-labeled scorpion toxin by antisera against the toxin. The assay is therefore adequate to detect soluble receptor sites of high affinity if they are present. Other Methods-Protein was measured by the method of Peterson (32). In this method, membrane proteins are solubilized by detergent and thus are detected with the same efficiency as soluble proteins. The protein is precipitated prior to measurement to remove interfering solutes. RESULTS Binding of [3HjSaxitoxin to Receptor Sites in Synaptosomes-Previous studies (reviewed by Ritchie and Rogart (3)) have shown that saxitoxin and tetrodotoxin bind to a common receptor site associated with sodium channels in many excitable membranes. The binding of [ Hlsaxitoxin to receptor sites in synaptosomes measured using the membrane filtration method is illustrated in Fig. 1. In the left panel, the dependence of total saxitoxin binding on concentration of free toxin is illustrated (0). Nonspecific binding in the presence of 1 PM tetrodotoxin depends linearly on concentration (0). Specific binding, determined from the difference between total binding and the fitted line for nonspecific binding, is illustrated as a Scatchard plot in the right panel of Fig. 1. The plot is linear indicating a single class of binding sites with a Klj of 1.7 nm and a binding capacity of 4.9 pmol/mg of protein. The specificity of the saxitoxin receptor sites is examined in the experiment illustrated in Fig. 2. Binding of [ Hlsaxitoxin is inhibited effectively by unlabeled saxitoxin and by tetrodotoxin, but is unaffected by concentrations of scorpion toxin (A) or batrachotoxin (M) which saturate their receptor sites. These results show that batrachotoxin and scorpion toxin act at different receptor sites than tetrodotoxin and saxitoxin, consistent with our previous conclusion (1, 2) that sodium channels have three separate receptor sites for neurotoxins: one specific for tetrodotoxin and saxitoxin, a second specific for the lipid-soluble toxins like batrachotoxin, and a third FIG. 1. Binding of [3H]saxitoxin to synaptosomes. Synaptosomes were incubated with increasing concentrations of [ Hlsaxitoxin (STX) from 0.2 to 30 nm for 20 min at 36 C and bound saxitoxin was measured using the rapid membrane filtration assay as described under Experimental Procedures. Left, total binding (0) and nonspecific binding measured in the presence of 1 pm tetrodotoxin (0) are plotted versus the measured free saxitoxin concentration. Right, specific binding, calculated from the difference between the experimental values for total binding and the fitted curve for nonspecific binding, is presented as a Scatchard plot. FIG. 2. Effect of neurotoxins on [3H]saxitoxin binding. Synaptosomes were incubated with 2 nm [ Hlsaxitoxin and the indicated concentrations of unlabeled saxitoxin (O), tetrodotoxin (O), scorpion toxin (a), or batrachotoxin (U) for 20 min at 36 C and bound saxitoxin was measured using the rapid membrane filtration assay as described under Experimental Procedures. Nonspecific binding measured in the presence of 1 PM tetrodotoxin has been subtracted from the results. specific for the polypeptides scorpion toxin and sea anemone toxin. The binding of [ Hlsaxitoxin to its receptor site in synaptosomes is also inhibited by certain cations and by protonation of an acid group with a pk, of 5.9 (33, 34). In order to compare the binding affinity and binding capacity of the saxitoxin and scorpion toxin receptor sites in synaptosomes, parallel measurements of binding of both toxins were made on ten separate synaptosome preparations using five preparations of labeled toxins. The two toxins have similar high affinity for their receptor sites with KD values of 1.9 nm for scorpion toxin and 2.3 nm for saxitoxin in this series of experiments. The binding capacity for saxitoxin is higher with B max values of 4.8 pmol/mg for saxitoxin and 1.3 pmol/mg for scorpion toxin. This gives a binding site ratio of 3.7. These results are similar to those described earlier for neuroblastoma cells in which there are 2.8 saxitoxin sites per scorpion toxin site (4). The interpretation of this result is more complicated in the synaptosome system, however. Possible interpretations are considered under Discussion.

4 11382 Sodium Channels in Brain Distribution of [3H]Saxitoxin Receptor Sites in Brain Membrane Fractions-Since the binding of saxitoxin is unaffected by depolarization and lysis of excitable membrane fractions (see below), saxitoxin binding can be used to determine the distribution of sodium channels in subcellular fractions from the brain. In these experiments, a subcellular fractionation scheme similar to that of Michaelson and Whittaker (35) was used as described under Experimental Procedures. The compositions of the different fractions referred to below are taken from their work (35). All the detectable saxitoxin receptor sites are present in the low speed supernatant, Sl + S2. (Table I) and 79% of the saxitoxin receptor sites are present in the 17,000 x g sediment, Fraction P3, containing mitochondria, synaptosomes, and broken membrane fragments. The purification of receptor sites is less than 2-fold, however. Further fractionation of the P3 fraction by discontinuous density gradient sedimentation yielded substantial [3H]saxitoxin binding in each of the gradient fractions. The light fractions in 0.32 M sucrose and 0.4 M sucrose are enriched in myelin and have relatively few [ Hlsaxitoxin receptor sites. The 0.6 M sucrose fraction containing mainly broken membrane fragments has a specific binding capacity lower than the P3 fraction. The 0.8 M sucrose fraction containing a mixture of broken membranes and synaptosomes is enriched in [3H]saxitoxin binding sites relative to the P3 fraction as is the 1.0 M sucrose fraction containing mainly synaptosomes. The 1.2 M sucrose fraction containing synaptosomes and mitochondria has a lower specific binding capacity. Thus, we find a modest enrichment of [3H]saxitoxin binding sites in the 0.8 M sucrose and 1.0 M sucrose fractions containing synaptosomes and mixed broken membrane fragments. The enrichment of sites in this gradient fractionation is small, however, and the sites are distributed through several fractions. This result is expected since the sodium channels in brain are located in several different membrane compartments: cell bodies, axonal membranes of myelinated and unmyelinated nerve, and nerve endings. The synaptosomal plasma membrane can be isolated from synaptosomes by further subcellular fractionation (28). Synaptosomes from the 1.0 M sucrose fraction were lysed and their membranes were fractionated by discontinuous density gradient centrifugation as described by Cotman and Matthews (28). The results are presented in Table II. As observed in tetrodotoxin binding experiments (36), the saxitoxin receptor TABLE I Distribution of saxitoxin receptor sites duringpreparation of synaptosomes Synaptosomes were prepared as described under Experimental Procedures. Samples of each fraction were saved and protein and specific [ Hlsaxitoxin binding at 3 nrvr [3H]saxitoxin were measured. Binding capacities were calculated assuming Ko = 1.8 nrvr. The results are means of three experiments. The specific binding capacity for synaptosomes (1.0 M sucrose fraction) is approximately 20% higher than the mean for ah exueriments (4.4 pmol/mg). Fraction Total sites Homogenates s1+ s P Gradient fractions 0.32 M (11%) 0.4 M (14%) 0.6 M (21%) 0.8 M (27%) 1.0 M (34%) 1.2 M (41%) Total Specific activity PmoUw sites were enriched 2-fold in the 25% sucrose fraction which was identified as the synaptic plasma membrane on the basis of analysis of marker enzymes (28). This result is consistent with the conclusion that most of the saxitoxin receptor sites in the 1.0 M sucrose fraction are associated with synaptosomes and sediment to the lighter density characteristic of synaptic plasma membrane after osmotic lysis. Binding of Saxitoxin and Scorpion Toxin to Receptor Sites in Lysed Membrane Fractions-In our previous studies (2), we found that 251-labeled scorpion toxin binding to receptor sites in synaptosomes was markedly inhibited by depolarization or lysis of the synaptosomes. The effect of depolarization and lysis on binding of [3H]saxitoxin and 251-labeled scorpion toxin is compared in Table III. As reported previously (a), depolarization of synaptosomes with 135 mm K+ inhibits scorpion toxin binding more than 90%. This inhibition is due to depolarization rather than K+ per se (2). Osmotic lysis also inhibits scorpion toxin binding greater than 90% since lysis causes complete depolarization. In contrast to the results with scorpion toxin, incubation with 135 mm K inhibits [3H]saxitoxin binding only 30% (Table III). Moreover, osmotic lysis has no effect. Since osmotic lysis causes complete depolarization, we conclude that saxitoxin binding is unaffected by depolarization. The small effect of 135 mm K+ on saxitoxin binding probably reflects the competitive interactions between K+ and saxitoxin described previously (3, 34). The receptor sites for scorpion toxin and the lipid-soluble toxins interact allosterically (1, 2, 6). Scorpion toxin binding to receptor sites in synaptosomes is greatly enhanced by batrachotoxin and the effect of depolarization on scorpion TABLE II Distribution of saxitoxin receptor sites duringpreparation of synaptic plasma membrane Synaptic plasma membranes were prepared from synaptosomes as described under Experimental Procedures. Samples of each fraction were saved and protein and specific [ Hlsaxitoxin binding at 3 nm [3H]saxitoxin were measured. Binding capacities were calculated assuming KU = 1.8 nm. The results are means of three experiments. Fraction Total sites Total protein Synaptosomes Gradient fractions 10% 25% 32.5% 35% 38% Specific activity % PmoUW TABLE Binding of saxitoxin and scorpion toxin to intact and lysed synaptosomes Synaptosomes were prepared as described under Experimental Procedures. Half of the preparation was osmotically lysed by resuspension in HP0 and incubation for 30 min at 0 C. Specific binding of scorpion toxin (2 nm + 1 pm tetrodotoxin) and [3H]saxitoxin (3 nm) was then measured as described under Experimental Procedures. Preparation fmol Intact synap- 5mMK tosomes 135 mm K mMK++l~MBTX mm K + 1 PM BTX Lysed synap- 5mMK tosomes 5mMK++ 1pMBTX III

5 Sodium Channels in Brain toxin binding is much reduced in the presence of batrachotoxin (2). In the experiment in Table III, depolarization reduces scorpion toxin binding from 34 fmol to 3 fmol and addition of batrachotoxin restores it to 16 fmol. Similarly, batrachotoxin increases scorpion toxin binding from 3 fmol to 15 fmol after osmotic lysis. These results show that osmotic lysis under these conditions gives an inhibition of scorpion toxin binding that is equivalent to depolarization of the intact synaptosome. Rupture of the synaptosomal membrane per se does not have a marked effect on either scorpion toxin binding or its enhancement by batrachotoxin. In contrast to scorpion toxin, saxitoxin binding is unaffected by batrachotoxin (Table III). Taken together, the results of Fig. 2 and Table III show that neither batrachotoxin, scorpion toxin, nor depolarization have any effect on saxitoxin binding. Since, under these conditions, batrachotoxin causes persistent activation of sodium channels, scorpion toxin blocks inactivation of sodium channels, and prolonged depolarization causes inactivation of sodium channels, we conclude that the saxitoxin receptor site is unaltered by voltage-dependent activation and inactivation of sodium channels and that saxitoxin binds equally well to its receptor site when sodium channels are in the resting, active, or inactivated states. Scatchard plots of saxitoxin binding to the lysed P3 fraction and to purified synaptic plasma membranes are illustrated in Fig. 3. In each case, a single class of binding sites is observed with a KI, of 1.8 to 2.6 nm. Since we observe a single class of saxitoxin receptor sites with a Ko in this range in intact synaptosomes, lysed synaptosomes, synaptic plasma membrane, and lysed P3, it is likely that all saxitoxin receptor sites in the brain have similar binding characteristics. Binding of scorpion toxin in lysed membrane preparations is highly dependent on the presence of batrachotoxin (Ref. 2 and Table III). The concentration dependence of batrachotoxin enhancement of scorpion toxin binding is illustrated in Fig. 4 for synaptic plasma membrane (left) and lysed P3 (right). In both these lysed membrane preparations, batrachotoxin is as effective in enhancing scorpion toxin binding as observed previously in intact synaptosomes (2). Thus, neither membrane integrity nor membrane potential are required for batrachotoxin action. The affinity of the mixed membranes in the lysed P3 fraction and the purified synaptic plasma membranes for batrachotoxin are similar (Fig. 4). In both these membrane preparations, a single class of scorpion toxin recep- tor sites is observed as illustrated for the lysed P3 fraction in Fig. 5. We conclude from these experiments that both the scorpion toxin and batrachotoxin receptors remain functional in lysed and depolarized membranes but that, as previously shown (2), the affinity of the scorpion toxin receptor is markedly reduced. In previous work with intact synaptosomes (2), we showed that veratridine and aconitine also enhance scorpion toxin binding but are less effective than batrachotoxin, even when saturating concentrations were used. These toxins reduce the enhancing effect of batrachotoxin when added together with the more effective toxin suggesting competitive interactions among the three lipid-soluble toxins (2). It remained possible that membrane potential or ionic rearrangements influenced these results in intact synaptosomes, however. In order to rule out a role of membrane potential and ion gradients in these interactions, we tested the effect of saturating concentrations of veratridine, aconitine, and batrachotoxin on scorpion toxin binding in lysed P3 preparations, alone and in combination (Table IV). We find, as in intact synaptosomes, that batrachotoxin causes a large enhancement of scorpion toxin binding, veratridine (500 pm) is less effective, and aconitine (100 pm) is least effective. Thus, the allosteric interaction between scorpion toxin and each of the three lipid-soluble toxins is retained in the absence of membrane potential and ion gra- [Botrachotoxin] FIG. 4. Enhancement of scorpion toxin binding by batrachotoxin in lysed P3 and purified synaptic plasma membrane. Scorpion toxin binding to synaptic plasma membrane (left) or lysed P3 (right) was measured in the presence of 1 nm scorpion mono[ 251]iodotoxin and the indicated concentrations of batrachotoxin as described under Experimental Procedures. Nonspecific binding in the presence of 200 nm unlabeled scorpion toxin has been subtracted from the results.,,um j. I [3H]-Saxitoxin bound (pmol/mg) FIG. 3. Comparison of saxitoxin binding to lysed P3 and purified synaptic plasma membrane. Binding of saxitoxin to lysed membrane fractions was measured as described in Fig. 1 for lysed P3 membranes (0) and synaptic plasma membrane (0). : Scorpion toxin bound (pmollmg) FIG. 5. Binding of scorpion toxin to lysed P3 membranes. Lysed P3 membranes were incubated with increasing concentrations of scorpion mono[ * I]iodotoxin from 0.2 to 10 nm in the presence of 1 PM batrachotoxin and bound scorpion toxin was measured as described under Experimental Procedures. Specific binding is presented as a Scatchard plot.

6 11384 Sodium Channels in Brain dients. Addition of veratridine or aconitine in the presence of batrachotoxin causes a marked reduction of scorpion toxin binding in lysed P3 to the value characteristic of incubation with veratridine or aconitine alone (Table IV). The competitive interactions among lipid-soluble toxins are unchanged in lysed P3 membranes and therefore neither membrane potential nor ion gradients are required for these interactions. In our previous work, K+ inhibited scorpion toxin binding while saxitoxin and tetrodotoxin enhanced scorpion toxin binding (2). We concluded that the effect of Kt was due to depolarization and the effects of tetrodotoxin and saxitoxin were due to prevention of depolarization. These conclusions can be tested directly using lysed P3 membranes since agents acting t,hrough membrane potential alterations should have no effect in the lysed membrane fraction. The results in Table IV show that K, saxitoxin, and tetrodotoxin have no effect on scorpion toxin binding in lysed P3 confirming that their effects are mediated through alterations of membrane potential or ion gradients. Solubilization of the Saxitoxin Receptor-Experiments with garfish olfactory nerve and electric eel electroplax have shown that the saxitoxin receptor sites in those tissues can be solubilized with nonionic detergents (23-25). Saxitoxin receptor sites in the rat brain P3 fraction can be solubilized with Triton X-100 (Fig. 6). As the Triton X-100 concentration is increased above O.l%, binding activity is lost from the membrane sediment and appears in the supernatant. At 1% Triton X-100, 35 to 40% of the binding activity is recovered in the soluble fraction while 65% remains in the sedimented membranes. At higher Triton X-100 concentrations, more activity is lost from the membranes, but the activity recovered in the supernatant is also reduced indicating substantial denaturation of the receptor sites. A large number of other nonionic detergents tested at a concentration of 1% are less effective at solubilizing the saxitoxin receptor in an active form. The solubilized saxitoxin receptor was unstable at 36 C losing all its binding activity in 5 min. Therefore, the binding activity of the solubilized receptor was measured at 0 C. A comparison of the binding of saxitoxin to P3 membranes and to soluble extract as measured by equilibrium dialysis is presented in Fig. 7. The affinity of the membrane-bound receptor for saxitoxin is markedly increased at 0 C. The KU is reduced from 1.8 nm (Fig. 3) to 0.17 nm (Fig. 7, 0). A similar effect of temperature on KIj for saxitoxin has been described previously (33). Binding to the solubilized receptor is charac- TABLE Interaction of neurotoxins with receptor sites in lysed P3 membranes The lysed P3 membrane fraction was prepared and binding of scorpion toxin was measured as described under Experimental Procedures with the additions indicated. Addition Specific ~;;;;pion toxin fmolhg None PM veratridine PM aconitine PM batrachotoxin pm batrachotoxin FM veratridine 1.1 PM batrachotoxin FM aconitine 1.1 PM batrachotoxin InM KC1 (replacing choline chloride) PM batrachotoxin 1.0 FM tetrodotoxin PM batrachotoxin 1.0 FM saxitoxin IV FIG. 6. Solubilization of the saxitoxin receptor. Lysed P3 membranes were incubated for 30 min at 0 C with the indicated concentrations of Triton X-100 in 100 mm choline chloride, 20 mm Hepes (adjusted to ph 7.4 with Tris base), 0.1 mm phenylmethanesulfonyl fluoride, and sedimented at 165,000 x g for 30 min. Binding of [YH]saxitoxin by the resuspended sediment fractions was measured at a concentration of 1.2 nm using the membrane filtration method described under Experimental Procedures (0). Binding to the supernatant fractions was measured at a concentration of 1.2 nm [ Hlsaxitoxin using the equilibrium dialysis method described under Experimental Procedures (0) $ 45 loo 50 0 [Jo] Saxitoxin bound (pmol/ml) ( I 1 IO L3~] Saxitoxin bound (pmol/ml) (0 ) FIG. 7. Comparison of [3H]saxitoxin binding by the solubilized and membrane-bound saxitoxin receptor. Binding of [ Hlsaxitoxin to lysed P3 membranes and to the saxitoxin receptor solubilized in 1% Triton X-100 was measured at 0 C using the equilibrium dialysis method described under Experimental Procedures. The results are presented as a Scatchard plot. Note the different scales for the solubilized and membrane-bound receptors. terized by a linear Scatchard plot indicating a single class of receptor site with a KI, of 0.22 nm. The nearly identical values of KD for the soluble and membrane-bound receptors indicate that the saxitoxin receptor is solubilized without alteration in the conformation at the saxitoxin receptor site. The values of B max for the soluble extract and the starting P3 membranes confirm that 40% of the sites are solubilized (compare abscissa intercepts in Fig. 7). The saxitoxin binding activity in the soluble extract is stable for several hours at 0 C but is rapidly lost on dilution with the extraction buffer or incubation at 22 C or 36 C. Agnew et al. (25) have shown that the soluble saxitoxin receptor from electric eel is stabilized by addition of phospholipids to the nonionic detergent solution. The effect of phosphatidylcholine J

7 Sodium Channels in Brain on stability of the saxitoxin receptor from rat brain at 22 C is illustrated in Fig. 8. In extraction buffer, binding activity was completely lost in 15 min (0). Phosphatidylcholine (0.02% (w/ v) in 0.1% Triton X-100) stabilized somewhat (0) as did addition of 10 IIIM CaClz (A). Addition of these two compounds together gave marked stabilization, however (0). Thus, the saxitoxin receptor from rat brain requires both Ca + and phosphatidylcholine for stabilization. Agnew et al. (25) showed that the ratio of lipid to detergent was the critical factor in stabilization of the eel receptor rather than the concentration of lipid per se. Examination of the effect of different lipid concentrations and lipid/detergent ratios has confirmed this conclusion for the saxitoxin receptor from mammalian brain (data not shown). Thus, the saxitoxin receptor from mammalian brain, like the receptor from electric eel, requires a mixed detergent/phospholipid micelle of defined composition for stability in solution. Presumably, the receptor is stable in the initial soluble extract because of the presence of endogenous lipid from the extracted membranes. In addition to phospholipid, the receptor from mammalian brain requires Ca +. The role of Ca2+ could be to produce a detergent/lipid micelle with appropriate properties or to bind to a specific site on the solubilized receptor. The solubilized saxitoxin receptor was analyzed by sucrose gradient sedimentation in the presence of Ca + and phosphatidylcholine in order to estimate the size of the receptor. lipoprotein complex. The gradient profiles (Fig. 9) showed a single major peak of binding activity. The sedimentation coefficient was estimated as 10 f 1 S by comparison with the sedimentation positions of catalase and lactoperoxidase in the same gradients (37). This sedimentation behavior is consistent with that of a globular protein with a molecular weight of 200,000 to 300,000. This is at best a very rough estimate however, because the effects of asymmetry and of detergent and phospholipid binding have not been taken into account. Extensive studies of hydrodynamic properties will be required before accurate estimates of molecular weight can be made. These experiments do show that the solubilized receptor is of fairly homogeneous size and is not a mixture of aggregates. I \ I Time (mi$5 FIG. 8. Stabilization of the solubilized saxitoxin receptor by Ca2+ and phosphatidylcholine. Soluble extract containing the saxitoxin receptor in 1% Triton X-100 was prepared as described under Experimental Procedures. At zero time, the extract was diluted 4- fold into 100 rnm choline chloride, 20 mm Hepes (adjusted to ph 7.4 with Tris base), 1% Triton X-100 with no further additions (O), or a final concentration of 10 mm CaCll (A), 0.2% (w/v) phosphatidylcholine (O), or both 10 mm CaCla and 0.2% phosphatidylcholine (0) at 22 C. At the indicated times, aliquots were withdrawn and cooled to O C, and [: H]saxitoxin binding was measured using the rapid gel Fitration assay at 0 C described under Experimental Procedures. 0 BOTTOM 5 IO Fraction TOP FIG. 9. Analysis of the solubilized saxitoxin receptor by sucrose density gradient centrifugation. Soluble extract containing the saxitoxin receptor in 1% Triton X-100 was prepared as described under Experimental Procedures. Aliquots (0.5 ml) were layered on linear 7% to 22% sucrose gradients prepared in 100 mm choline chloride, 20 IIIM Hepes (adjusted to ph 7.4 with Tris base), 0.1% Triton X-100,0.02% (w/v) phosphatidylcholine, and 10 mm CaCL and sedimented at 302,000 X g for 6 h at 4 C. Gradients were fractionated from the bottom and aliquots of each fraction were assayed for [ Hlsaxitoxin binding activity using the rapid gel filtration assay. The migration positions of lactoperoxidase (5.4 S) and catalase (11.2 S) sedimented in the same gradients are indicated by arrows. The solubilized and stabilized receptor should be amenable to purification and further analysis by conventional and affinity chromatographic methods. Solubilization of the Scorpion Toxin Receptor-Although scorpion toxin binding is membrane potential-dependent, we have developed conditions (Fig. 5) under which scorpion toxin binding can be measured in lysed membrane fractions having no membrane potential. Under these conditions, it is possible in principle to treat the lysed membranes with detergent and detect the solubilization of the receptor sites by measuring loss of sites from the membranes and appearance in the supernatant as for saxitoxin. In these experiments, a rapid gel filtration assay similar to that used for saxitoxin binding was employed to measure solubilized scorpion toxin receptor (see Experimental Procedures ). Lysed P3 membranes were incubated with various concentrations of Triton X-100 or other detergents for 30 min at 0 C and then sedimented at 165,000 X g for 30 min. Binding activity in the supernatant and t,he sediment were then measured. At 0.1% Triton X-100, all of the?-labeled scorpion toxin binding activity was lost from the membrane fraction. This concentration of Triton X-100 did not solubilize the saxitoxin receptor (Fig. 6) and had little effect on membrane proteins in general. Several other nonionic detergents also caused loss of scorpion toxin binding at much lower concentrations than those required for solubilization of the saxitoxin receptor. Thus, the scorpion toxin receptor is either denatured or solubilized at detergent concentrations that have no effect on the saxitoxin receptor. The detergent supernatants were analyzed for scorpion toxin binding using the rapid gel filtration technique. This method was sensitive enough to detect solubilization of 10% of the scorpion toxin receptor. No binding activity was detected in any of the detergent supernatants and no binding activity was recovered in either membrane sediments or supernatants on removal of the detergent by dilution or dialysis. These data show that the soluble extract containing 40% of the saxitoxin receptor sites has no scorpion toxin binding activity. Either the scorpion toxin receptor is denatured by low concentrations of detergent or it is a small protein that is not excluded from Bio-Gel P30 columns in the rapid gel filtration assay. We have also been unable to detect soluble scorpion toxin receptor using ion exchange column assays to

8 11386 Sodium Channels in Brain detect the toxin. receptor complex. Equilibrium dialysis methods were found unsuitable because of large nonspecific binding values. Our working hypothesis at present is that the high affinity binding activity of the scorpion toxin receptor is dependent upon the integrity of the lipid bilayer membrane and is irreversibly lost when the membrane is solubilized by detergents. Since scorpion toxin binds to a site involved in voltage-dependent activation of sodium channels, it seems likely that the saxitoxin receptor as solubilized here and in other studies will be incapable of voltage- or toxin-dependent activation of sodium transport upon ultimate isolation and reconstitution into a bilayer membrane. DISCUSSION The experiments on interactions of neurotoxins with receptor sites in brain membrane fractions described in this report confirm and extend conclusions on the mechanism of toxin action reached in earlier studies of sodium channels in neuroblastoma cells and brain synaptosomes (1, 2). The toxins studied act, at three separate receptor sites. Saxitoxin and tetrodotoxin bind to a common receptor site which we have designated Receptor Site 1 (11). Neither the polypeptide toxins nor the lipid-soluble toxins have any effect on [3H]- saxitoxin binding in brain membranes confirming the independence of the receptor sites for these three groups of toxins. In addition, neither membrane lysis nor depolarization alters saxitoxin binding. Since batrachotoxin and scorpion toxin activate sodium channels and prolonged depolarization inactivates sodium channels, our results show that saxitoxin binds equally well to the resting, active, and inactivated states of the sodium channel and suggest that Receptor Site 1 is not located on a component of the sodium channel which undergoes a voltage-dependent change of state during activation or inactivation. It has been suggested that this receptor site is located within the ion channel at a specific ion coordination site (38, 39), but experimental evidence on this point is contradictory at present (reviewed in Ref. 40). Lipid-soluble neurotoxins acting at Receptor Site 2 and polypeptide toxins acting at Receptor Site 3 interact allosterically (1, 2, 6). The results presented here show that this allosteric interaction is retained in lysed membrane fractions and therefore does not require either membrane potential or ion gradients. As in neuroblastoma cells and intact synaptosomes (2, 6), veratridine and aconitine are partial agonists in enhancing scorpion toxin binding while batrachotoxin may be a full agonist. Competitive interactions are observed among the three lipid-soluble toxins in lysed membrane preparations as in intact synaptosomes and neuroblastoma cells (1, 2, 6). These results support the conclusion that these toxins act at a common receptor site and show that the competitive interactions observed are not mediated by alterations in membrane potential or ion gradients. The binding of scorpion toxin to Receptor Site 3 is highly dependent upon membrane potential (2, 10, 21). In those experiments, membrane potential was varied by changing the extracellular K+ concentration or by increasing sodium permeability with gramicidin A. Since these two different methods of depolarization gave comparable results, it was concluded that membrane potential was the important variable and that K+ had no direct effect. Experiments with lysed membranes confirm that conclusion by showing that K+ has no effect on binding in the absence of a membrane potential. By taking advantage of the enhancement of scorpion toxin binding by batrachotoxin, we have been able to measure binding in the absence of a membrane potential accurately for the first time. We find that the lysed P3 membrane fraction containing 79% of the sodium channels in brain has a single class of binding sites for scorpion toxin with a KD similar to that for lysed or depolarized synaptosomes or purified synaptic plasma membrane. In addition, the sodium channels in this fraction showed similar allosteric interactions to those observed in synaptosomes and purified synaptic plasma membrane. The results indicate that these are general properties of sodium channels in the brain. In addition, the results show that, while scorpion toxin binding is highly dependent on membrane potential, the receptor remains functional in lysed and depolarized membranes, but exhibits much reduced affinity for scorpion toxin. Comparison of the binding capacity of synaptosomes for scorpion toxin and saxitoxin shows a ratio of 3.7 saxitoxin receptor sites for each scorpion toxin receptor site. In neuroblastoma cells, a ratio of 2.8 was observed (4). Two interpre- tations of the results with neuroblastoma cells were considered (4): either each sodium channel complex contains three saxitoxin receptor sites and only one scorpion toxin receptor site or neuroblastoma cells contain two classes of saxitoxin receptor sites, one associated with a scorpion toxin receptor site and one not. These two interpretations may also be applied to our results with synaptosomes. In addition, other sources of excess saxitoxin receptor sites are possible in synaptosome preparations. Binding to the saxitoxin receptor is unaffected by changes in membrane potential, lysis of membranes, and even solubilization with detergents. In contrast, scorpion toxin binding is sensitive to all these conditions. It is possible, therefore, that a fraction of the scorpion toxin receptor sites in the synaptosome fraction has been destroyed during membrane isolation. Since scorpion toxin binding is voltage-dependent, whereas saxitoxin binding is not, membrane fragments contaminating the synaptosome fraction might bind saxitoxin but not scorpion toxin. We have obtained experimental evidence against this possibility. In previous work, we showed that at least 85% of the scorpion toxin receptor sites in synaptosomal preparations are voltage-sensitive and therefore likely to be associated with synaptosomes (2). In conjunction with the experiments presented in the text, we also carried out companion experiments on depolarized synaptosomes. Under these conditions, receptor sites in broken membranes should have affinities comparable to those in synaptosomes. A similar ratio of saxitoxin sites per scorpion toxin site was observed under these conditions. Finally, similar experiments were carried out with the lysed P3 fraction. These resulted in a larger ratio of 4.7 saxitoxin receptor sites to one scorpion toxin receptor site. Since we observe a single class of scorpion toxin receptor sites in each case, these data show that membrane potential variation cannot be the source of the stoichiometry observed. Our data comparing scorpion toxin and saxitoxin binding in brain membranes support the conclusion that excitable membranes contain approximately three saxitoxin receptor sites for each scorpion toxin receptor site (4). The precise interpretation of this result must await purification and analysis of the toxin receptors. A major step toward purification of a membrane component is solubilization with retention of biological activity. In this report, we have described procedures which allow solubilization of the saxitoxin receptor site in good yield without change in the binding of saxitoxin. The soluble receptor can be stabilized by addition of phosphatidylcholine and Ca +. It appears homogeneous in size with a sedimentation coefficient of 10 S. This sedimentation behavior is consistent with a globular protein of M, = 200,000 to 300,000. This estimate is approximate at best because asymmetry and detergent and lipid binding have not been considered. Nevertheless, the sedimentation coefficient is similar to that reported for soluble