Purification and Chemical and Biological Characterizations of Seven Toxins from the Mexican Scorpion, Centruroides suffusus suffusus”

Seven polypeptides highly toxic to mice were isolated from the venom of the scorpion, Centruroides suffusus suffusus (Css), and their chemical and toxic properties were characterized. It was shown that the most active toxins by intracerebroventricular injection are less ac- tive when injected subcutaneously. The complete amino acid sequence (66 residues) of toxin I1 (Css 11) has been determined. The C-terminal end is amidated as found for most other scorpion toxins. Css I1 is a B- type toxin, previously used to define the binding site for activation of the sodium channel. Using rat brain synaptosomes, we demonstrated that all Css toxins compete with 12sI-Css I1 to bind to site 4 and should be considered as &scorpion toxins. Specific binding parameters for Css VI, one of the most active toxins, were determined: KD = 100 PM; capacity in binding sites, 2.2 pmol of toxin/mg of synaptosomal protein. Css VI was shown to inhibit y-aminobutyric acid uptake by synaptosomes: K 0.5 = 100 PM, which agrees with its KD. Competition experiments between the seven Css toxins and 1261-Css I1 for antiserum

lethal to mice which are present in the venom of Centruroides suffusus suffusus, a scorpion living in the area of Durango in Mexico. Seven Css' toxins have been purified and characterized. One of them, Css 11, has been entirely sequenced here. It was the first 8-toxin to be used to demonstrate the existence of the fourth receptor site on the sodium channel (2). Specific binding parameters of Css VI, one of the most active toxins in the venom, were defined on rat brain synaptosomal fractions. The antigenic properties of these polypeptides were studied extensively.

MATERIALS AND METHOD$ RESULTS
Purification of C. suffusus suffusus Toxins-The different venom fractions obtained during the purification process are described in the flow chart given in Fig. 1. The elution patterns of the chromatographic steps can be found in the Miniprint Section (Figs. 2-5). Css toxins were numbered according to retention time on cationic exchangers. Fig. 6 shows polyacrylamide gel electrophoresis of major toxins compared with the venom's water extract. Table I gives quantitative data related to purification. When scorpion telsons were used instead of venom as starting material, the results were slightly different: only Css I1 was recovered from fraction R1, but with a lower yield (60% of that obtained with the venom), and Css I could only be detected. In this case, fraction R2 was not further purified. Toxicity tests were performed on mice both by subcutaneous and intracerebroventricular injections. The results show that the toxins are much more active by intracerebroventricular injection: about 100 times more active with Css 1-111 and VII; about 1000-1500 times with Css IV-VI. One protein, P1, inactive in mice, was also obtained in a pure form. Its amino acid composition is given in Table I1 together with those of the Css toxins. Css I and I1 have the same amino acid composition, and P1 is closely related to toxins. Isoelectric focusing (not shown) indicates that both Css I and I1 have a PI value approaching 9, Css I being slightly more acidic than Css 11. The most likely expla- nation for this difference is the presence of one (or more) amidated residue(s) in Css I1 replacing one (or more) acidic residue(s) in Css I. For all other toxins, PI is >9.0. The presence of 8 half-cystine residues represents a common feature of all scorpion toxins.

Characterizations
Amino Acid Sequence of Css II-The amino acid sequence for Css If, the major toxin of Css venom, and the NH2terminal sequence for Css I were determined. Experimental procedures used to establish these sequences are given in the Miniprint Section. Fig. 8 summarizes the different steps of sequence determination.

Effects of Css Toxins on Binding of '251-Css 11 to Anti-Css II
IgGs- Fig. 9 gives dose-response curves obtained when adding different homologous Css toxins to an equimolar mixture of "51-Css I1 and anti-css I1 IgGs. Concentrations of Css toxins which allow the half-maximum effect to be attained are given in Table 11. It appears that Css I1 has high affinity for its specific antibodies = 0.17 nM). The shape of the displacement curve is noticeable when Css I is used: at first, it follows the standard curve and then deviates, but complete competition is still possible. It would be interesting to determine the amino acid sequence of Css I to check for deamidation which could affect some but not all antigenic regions of this protein.
The common antigenic behavior of these @-toxins may also reflect a high degree of sequence homology.
Char~terization of Css Toxin Binding to Rut Brain Synap-Prn a Mexican Scorpion Venom 4453 tosomal Fraction-On the basis of their binding properties to rat brain synaptosomes and the different mechanisms which act on the sodium channels of excitable membranes, the existence of two types of scorpion toxins, a-and @-toxins, has been demonstrated (2,3). These two types of scorpion toxins are represented by Androctonus australis Hector (AaH) I1 (a) and Css I1 (@), respectively. Css toxins have been tested on rat brain synaptosomal fractions in two sets of displacement experiments using either radioiodinated AaH I1 or Css 11. Css toxins are unable to compete with lZ5I-AaH I1 for site 3 on the sodium channel, even with concentrations up to 10 p~ (data not shown). On the other hand, all Css toxins behave like &toxins because they cause '251-Css I1 to be completely displaced from site 4 (Fig. 10). The KD values calculated from these displacement experiments are given in Table 11: several toxins (i.e. Css IV-VII) were more efficient than Css I1 in displacing '251-Css I1 from its site. Finally, the venom extract was used in competition experiments with Iz5I-AaH for site 3, and concentrations up to 10 mg/ml were ineffective. In addition, the binding properties of Css VI were studied on the rat brain synaptosomal fraction: the Scatchard plot is linear and indicates that '251-Css VI binds to a single class of noninteractive binding sites (data not shown). Four independent experiments gave an average KD of 100 k 10 pM and a site capacity of 2.2 k 0.2 pmol/mg of protein. This specific binding has been further correlated with the inhibitory action of Css VI on GABA uptake by rat brain synaptosomal fractions. The dose-response curve of this effect is given in Fig. 11: the value of (100 pM) corresponds well to the binding characteristics of Css Vi. As aiready shown with Css I1 (3, IO), this inhibition of GABA uptake is probably due to a change in the potentialdependent sodium channel activity induced by @-toxin binding at site 4.

DISCUSSION
Using 15 g of c. suffusus suffusus venom, seven toxins which are highly active in mice were obtained in a pure state as evidenced by ion-exchange chromato~aphy (Miniprint Section, Figs. 4 and 5) and polyacrylamide gel electrophoresis (Fig. 6). The toxicity of the seven Css toxins accounts roughly for 40% of the venom toxicity regardless of the injection method. We believe this nonquantitative yield is the consequence of cumulative losses occurring during the numerous purification procedures. Lethality was tested on mice throughout; and consequently, only the so-called ''mammal toxins," i.e. those proteins toxic in mice, were detected and purified. Many different toxins have also been found in other Centruroides venoms, i.e. Centruroides sculpturatus Ewing ( 6 ) and Centruroides noxius (7). One may ask whether this can be attributed to the high number of animals milked or to a single animal's capacity to synthesize several toxins, or to a combination of these two factors. This represents a general problem with pooled scorpion venoms (1 1).
The amino acid sequence of Css I1 as well as preliminary results3 on Css I and other Css toxins indicate that, as with other scorpion toxins, Css toxins are single chain miniproteins with 60-66 residues cross-linked by four disulfide bridges (Table 11). This is evident in the amino acid sequence of Css I1 given in this paper and also by experiments in progress in our laboratory3 on the sequences of other Css toxins. Amino acid compositions (Table 11) show a high content in aromatic and basic amino acids, which is common to all scorpion toxins. Css toxins have several characteristics: one or, more generally, two amino acids are lacking, and the aspartic acid/glutamic ~ Sampieri, F., Beehis, G., and Rochat, H., experiments in progress. Quantitative data are from left to right: weight in toxins from crude venom, toxicities, and yields of toxins from Css venom as determined by subcutaneous injection (sc) and by intracerebroventricular injection (icv); results of radioimmunoassay between '*61-Css I1 and Css toxins and P1 for 0.2 nM anti-Css I1 IgGs (the K0.5 values are the concentrations giving 50% inhibition of '%Css I1 binding); and KU values for Css toxins binding to synaptosomes (these values are calculated from the values, i.e. the concentrations of native toxins which give the halfmaximum inhibition of the specific binding of 'p61-Css I1 at 37 "C on a rat brain synaptosomal fraction). acid ratio is approximately 1:l whereas it is much higher for toxins of the Old World scorpions.

C~r~t e r i~~i o~ of Toxins from a Mexican Scorpion Venom
The complete amino acid sequence of Css I1 was determined using standard procedures, but with two objectives: to use automated Edman degradation extensively and to avoid where possible peptide purification procedures. Css 11, like many other scorpion toxins (l), was found to be amidated at the Cterminal end. The sequence of Css I will have to be determined in order to ascertain whether Css I and I1 differ in one (or more) amidated groups. The primary structures of Css I1 and other scorpion toxins are compared in Fig. 12. The sequences are arranged such that the cysteine residues are aligned and deletions are introduced to maximize homology. These amino acid deletions occur in positions which are different in a-and ,&toxins. a-Toxins are characterized by two deletions, the first (at least of 3 amino acid residues) begins at position 20, whereas the second (generally of 2 or 3 amino acid residues) is located at position 69. The beginning of a deletion at position 46 (4-5 amino acid residues) is the main feature common to all the @toxin sequences. These differences might result in localized conformational changes, which could be related to the ability of a-and &toxins to bind to two different sites on the voltagedependent sodium channel (2,3). Css I1 relates to the @-toxin group (2, 3). Its primary structure clearly belongs to this group, although there are some differences. A tyrosine fposition 59) replaces the proline found in all other a-and 8toxins. Glutamine (position 32) and histidine (position 57) replace lysine and glycine residues present in other @-toxins, respectively. Despite these modifications, the amino acid sequence for Css I1 seems to comply to the general sequence proposed for scorpion toxins (24).     and stopped by a dilution of 1 0 1 followed by filtration using a Whatman GF/C filter. The filters were solubilized in Beckman tissue solubiliter and then put in 2-ml scintillation fluid. The radioactivity was measured using a Packard Tri-Carb 460C liquid scintillation system. roides scorpions. Although there is no evidence that each of the 120,000 animals used was able to synthesize the seven toxins, the fact that complete cross-reaction has been obtained ( Fig. 9) seems to indicate that, regardless of the Css venom batch used, an efficient antiserum will be obtained. The situation is quite different with a-toxins whose polymorphism is well documented ( i e . Androctonus, Buthus, and Leiurus genera (11, 25)). These a-toxins have been classified into four groups according to amino acid sequence homologies and antigenic group determinations (1, [26][27][28]. Moreover, this polymorphism, which exists with toxins present in individual venoms from A. australis Hector (29), poses a problem for the development of an efficient and general serotherapy. In other respects, the possibility of generating antibodies which neutralize AaH I1 (an a-toxin) using synthetic peptides which correspond to definite sequence portions of AaH I1 (30) has recently been demonstrated. This technique will be extended to Css toxins in the near future.

Asp-Tyr-Cys-Leu-Arg-Glu-Cys-Lys-Gln-Gln-Tyr-Gly-Ly~-Ser-Ser-Gl~~-Gly-Tyr-Cys-Tyr-Ala-?h~-
Unlike scorpion a-toxins which bind to site 3 on the sodium channel, Css I1 has been shown to bind to site 4 (2) and thus is considered as the first toxin of the 6-toxin group. This binding has been further correlated with a pharmacological effect on the voltage-dependent sodium channel activation (3). We attempted to define the (a-or p-) type for each of the Characterizations of Toxins from a Mexican Scorpion Venom

p-toxins
seven purified Css toxins. This was accomplished using competition experiments with labeled a-and @-toxins, i.e. 1251-AaH I1 or 1251-Css I1 (Fig, 10). The results clearly indicate that the seven Css toxins are @-type toxins. Furthermore, we found no competition when Css venom extract was used with 1261-AaH (a-toxin). These data strongly support the idea that no a-type toxin is present in the Css venom. The characteristics of Css venom are therefore different from those of Tityus serrulatus venom where both a-and @-toxins are present (31) and also perhaps from the characteristics of C. s c~~~u r~t u s Ewing, although contradictory results have been published concerning its toxins (32-34). Css I is less efficient than Css I1 in displacing '251-Css 11, whereas the other toxins, Css 111-VII, are relatively more active. In fact, the most efficient toxins in displacing '251-Css I1 from rat brain synaptosomes are those which are the most active in lethality tests performed by intracerebroventricular injection. This is significant because it indicates that when one is looking for toxins which are highly active in the central nervous system, it is best to perform intracerebroventricular injection in order to monitor the purification process. For toxins active in the peripheral nervous system, the subcutaneous method is perhaps better suited. This affinity difference may be caused by differences in sodium channels present in the central and peripheral nervous systems (35-37). Css VI1 does not fit this scheme. Indeed, its calculated KO is similar to KD values for Css IV and VI, whereas its LD, (intracerebroventricular) is close to the values obtained with Css 1-111 (Table 11). This discrepancy has not yet been clearly explained. However, one may suppose that the affinity of the toxin for its binding site results from chemical interactions between a region of the toxin and a site of its target; whereas the pharmacological effect related to toxicity may result from an interaction, not restricted to the binding site, between toxin and the sodium channel.
The number of sites for Css VI (2.2 pmol/mg of synaptosomal protein) is about twice that found for Css I1 on the same starting material (2). In the case of Css 11, this difference can be explained by the fact that the number of sites depends on pH: it is twice as high at pH 6 as at pH 7.2 (38). We did not find such a pH dependence with Css VI.
Css I1 is known to inhibit the rate of GABA uptake and to increase the rate of GABA released by synaptosomes (3). This is also true for Css VI whose inhibition corresponds to the binding affinity of this toxin with the sodium channel. These data confirm the previously presented hypothesis (10) that this effect on GABA can be used for testing @-toxin activity on synaptosomes.
In addition to the seven Css toxins, one protein (Pl) was purified; and its amino acid composition was found to be closely related to those of toxins I-VI1 (Table 11). This protein was tested for toxicity and its ability to compete with 1251-Css I1 in binding experiments, either with anti-Css I1 IgGs or synaptosomes (Figs. 9 and 10 and Table 11). This protein can be considered either as a toxin with low activity or as a nonactive analogue contaminated by traces (less than %moth) of a very active toxin. In either case, the determination of its primary structure will be of great interest for a better understanding of structure-activity relationships in @-scorpion toxins.