Importance of the Conserved Aromatic Residues in the Scorpion α-Like Toxin BmK M1

About one-third of the amino acid residues conserved in all scorpion long chain Na+ channel toxins are aromatic residues, some of which constitute the so-called “conserved hydrophobic surface.” At present, in-depth structure-function studies of these aromatic residues using site-directed mutagenesis are still rare. In this study, an effective yeast expression system was used to study the role of seven conserved aromatic residues (Tyr5, Tyr14, Tyr21, Tyr35, Trp38, Tyr42, and Trp47) from the scorpion toxin BmK M1. Using site-directed mutagenesis, all of these aromatic residues were individually substituted with Gly in association with a more conservative substitution of Phe for Tyr5, Tyr14, Tyr35, or Trp47. The mutants, which were expressed in Saccharomyces cerevisiae S-78 cells, were then subjected to a bioassay in mice, electrophysiological characterization on cloned Na+ channels (Nav1.5), and CD analysis. Our results show an eye-catching correlation between the LD50 values in mice and the EC50 values on Nav1.5 channels in oocytes, indicating large mutant-dependent differences that emphasize important specific roles for the conserved aromatic residues in BmK M1. The aromatic side chains of the Tyr5, Tyr35, and Trp47 cluster protruding from the three-stranded β-sheet seem to be essential for the structure and function of the toxin. Trp38 and Tyr42 (located in the β2-sheet and in the loop between the β2- and β3-sheets, respectively) are most likely involved in the pharmacological function of the toxin.

Scorpion neurotoxins targeting voltage-gated sodium channels are single chain polypeptides composed of 60 -70 amino acids cross-linked by four disulfide bridges. They have been divided into two major classes, ␣and ␤-toxins. Scorpion ␣-toxins, the most extensively studied group, can prolong the action potential by slowing the inactivation of Na ϩ currents with no direct effect on activation (1)(2)(3).
According to their different pharmacological properties, the ␣-toxins can be further divided into three subgroups, classical ␣-, ␣-like, and insect ␣-toxins (4,5). The classical ␣-toxins (e.g. AaH II and Lqh II) are highly toxic to mammals, whereas the insect ␣-toxins (e.g. Lqh ␣ insect toxin) are highly toxic to insects. The more recently characterized ␣-like toxins (e.g. Lqh III and BmK M1) act on both mammals and insects, but are unique in their inability to bind to rat synaptosomes despite a high toxicity by intravenous injection. Although three-dimensional structures for the classical ␣-toxins (6, 7), ␣-like toxins (8,9), and insect ␣-toxins (10) have been elucidated, in-depth structure-function studies of these long chain toxins using sitedirected mutagenesis are still rare, mainly because of folding problems; and the focus has often been on the charged residues in the toxins (11,12). Here, we report the importance of the conserved aromatic residues in ␣-toxins identified by mutagenesis analysis using the ␣-like toxin BmK M1 as template.
BmK M1 is a toxin from the venom of the scorpion Buthus martensii Karsch, which resides in eastern Asia, and is composed of 64 amino acids cross-linked by four disulfide bridges (3,7). BmK M1 has been the subject of different studies: its three-dimensional structure was determined by x-ray crystallography at 1.7-Å resolution (8); the pharmacological properties of Na ϩ channels have recently been investigated (13); and gene cloning and expression of wild-type BmK M1 have also been carried out (14,15).
Alignment of the amino acid sequences of several ␣-toxins shows that seven aromatic residues, including Tyr 5 , Tyr 14 , Tyr 21 , Tyr 35 , Tyr 42 , Trp 38 /Tyr 38 , and Trp 47 /Tyr 47 , are notably conserved ( Fig. 1A) (5). In a previously performed structural analysis, a conserved hydrophobic surface (CHS) 1 was identified (7). The CHS is assumed to be part of the functional site of scorpion toxins targeting sodium channels (16,17). Tyr 5 , Tyr 35 , and Trp 47 are located on the so-called Face A surface of the toxin (CHS). Tyr 14 and Tyr 21 are situated on the surface opposite to Face A, called Face B. Trp 38 and Tyr 42 are located in the ␤ 2 -sheet and in the loop between the ␤ 2 -and ␤ 3 -sheets, respectively (Fig. 1B).
Seven of the 15 residues conserved in scorpion toxins (5) are aromatic residues and have been studied in this work. Based on an efficient yeast expression system (15), the importance of the above-mentioned aromatic residues in the scorpion toxin BmK M1 was analyzed by site-directed mutagenesis. The results from mutagenesis and expression, characterization, bioassays, and electrophysiological analysis of the mutants are reported here. Based on these findings, the important role of these conserved aromatic residues is discussed. Y5F were created by one-step PCR. Other mutants (Y14G, Y14F, Y21G, Y35G, Y35F, W38G, Y42G, W47G, and W47F) were obtained by threestep PCR. A pair of mutagenic primers was applied in the first or second PCR with primer 1 or 2, respectively, to create two intermediate products, which shared an identical sequence. After treatment with Klenow fragment, two intermediate products acted as primers for each other, and extension of this overlap by DNA polymerase created the fulllength mutant, which had the mutation at the desired position. All PCR products were purified by gel excision.
Expression and Purification of Mutants-After digestion with XbaI and HindIII, the mutated cDNA gene was inserted into plasmid pVT102U/␣ and transformed into E. coli TG1 competent cells. The recombinant plasmid pVT102U/␣-mutant was extracted, sequenced, and transformed into S. cerevisiae S-78 using the LiCl method (18). The expression of the mutants was carried out using a described previously procedure (15). After fermentation, the supernatant of the culture was adjusted to pH 4.2 with acetic acid. The sample was directly applied to a CM32-cellulose cation-exchange column (2.8 ϫ 14 cm), which was equilibrated with 0.1 M sodium acetate at a flow rate of 1 ml/min. Upon reaching a steady base line, the column was washed by stepwise elution with 0.2, 0.3, and 0.5 N NaCl equilibration buffer. The 0.5 N NaCl fraction was directly applied to a Sephasil ® peptide C 18 reversed-phase column. Buffer A contained 0.1% trifluoroacetic acid in water; buffer B contained 0.1% trifluoroacetic acid in acetonitrile. The C 18 column was eluted with a linear gradient of 0 -80% buffer B for 15 column volumes. Reversed-phase chromatography was carried out using an Ä KTA purifier chromatography system (Amersham Biosciences AB).
Molecular Mass Determination-The molecular masses of the purified mutants were obtained using a Finnigan LCQ ion-trap mass spectrometer (ThermoQuest, San Jose, CA) equipped with an electrospray ionization source. The spray voltage was 4.50 kV. Calculations were performed using the program provided by the manufacturer.
Bioassay-Using 0.9% NaCl as a negative control and rBmK M1 as a positive control, the toxicity of the mutants was determined in mice (male, specified pathogen free level, 18 -20 g of body weight). Each group consisted of 10 mice. Various doses of toxin mutants were dissolved in 0.9% NaCl and injected into the mice through the tail vein. Survival times (times between injection and death), reaction, and doses were recorded. Evaluation of toxicity was based on the determination of LD 50 (the dose capable of statistically killing 50% of the mice) according to the method of Meier and Theakston (19).
Expression in Xenopus Oocytes, Electrophysiological Recordings, and Analysis-The human Na v 1.5 gene was subcloned into pSP64T (20). For in vitro transcription, pSP64T/Na v 1.5 was first linearized by XbaI. Using the large-scale SP6 mMESSAGE mMACHINE transcription kit (Ambion Inc.), capped cRNAs were synthesized from the linearized plasmids. The in vitro synthesis of cRNA encoding histone H1 and the isolation of Xenopus oocytes were done as described previously (21). Oocytes were injected with 50 nl of Na v 1.5 cRNA solution at a concentration of 1 ng/nl using a Drummond microinjector.
Whole cell currents from oocytes were recorded using the two-microelectrode voltage clamp technique between 1 and 3 days after injection. Voltage and current electrodes were filled with 3 M KCl. Resistances of both electrodes were kept as low as possible (ϳ0.1-0.2 megaohms). Experiments were performed using a GeneClamp 500 amplifier (Axon Instruments, Inc.) controlled by a pClamp data acquisition system (Axon Instruments, Inc.). Currents were sampled at 10 kHz and filtered at 5 kHz using a four-pole low-pass Bessel filter. Digital leak subtraction of the current records was carried out using a P/2 protocol. The bath solution composition was 96 mmol/liter NaCl, 2 mmol/liter KCl, 1.8 mmol/liter CaCl 2 , 2 mmol/liter MgCl 2 , and 5 mmol/liter HEPES (pH 7.4). This solution was supplemented with 50 mg/liter gentamycin sulfate for incubation of the oocytes. All experiments were performed at room temperature (20 -22°C).
Circular Dichroism Measurements-Samples used for analyses were dissolved in 20 mM phosphate buffer (pH 7) at a concentration of 1.0 mg/ml. Circular dichroism spectra were recorded on a Jasco J-720 spectropolarimeter. Spectra were run at 25°C from 250 to 200 nm using a quartz cell 0.5 mm in length. Data were collected at 0.5-nm intervals with a scan rate of 50 nm/min. All CD spectra resulted from averaging four scans. The final spectrum was corrected by subtracting the corresponding base-line spectrum obtained under identical conditions. Spectra were smoothed by the instrument's software. The secondary structure content was estimated by standard Jasco CD analysis.

FIG. 2. As an example, the entire purification process of Y42G is shown.
A, Tricine/SDS-PAGE before and after purification. The protein bands were stained with Coomassie Brilliant Blue R-250. Lane 1, molecular mass markers; lane 2, fermentation supernatant of Y42G, lane 3, purified Y42G. B, purification of Y42G by reversed-phase chromatography. C, Finnigan LCQ ion-trap mass spectra of Y42G. The mass spectra of the other mutants are not shown. mAU, milliabsorbance units. and W47F were created by three-step PCR. Mutants Y5G and Y5F were produced by one-step PCR. The target gene was expressed using the pVT102U/␣ vector. Tricine/SDS-PAGE analyses of yeast cultures demonstrated that mutants Y5G, Y5F, Y14G, Y14F, Y21G, Y35G, Y35F, W38G, Y42G, and W47F were expressed and secreted into the medium. Mutant Y5G could not be used in the following characterization because its expression level was in a trace amount. Mutant W47G could not be expressed at all. The expression levels of the five glycine mutants Y14G, Y21G, Y35G, W38G, and Y42G were ϳ1-2 mg/liter of culture medium. For three of the mutants with the conservative phenylalanine substitution (Y5F, Y14F, and Y35F), the expression levels were ϳ3 mg/liter, comparable to that of unmodified rBmK M1 (ϳ3 mg/liter). Remarkably, the amount of W47F in the culture medium was 9 -10 mg/liter, which is about three times the value of rBmK M1 (Table I). Apparently, the conservative phenylalanine mutations of Tyr 5 and Trp 47 changed the expression levels of mutants Y5G and W47G from a trace or nothing at all to the normal level or to even a high level.
Expressed mutants were purified by a simple and efficient protocol. One liter of culture (10 g/liter yeast extract, 20 g/liter bacteriological peptone, 20 g/liter glucose, pH after sterilization of 6.5) was harvested and initially purified by chromatography on a CM32 cation-exchange column. The next step of purification was carried out on a C 18 column. The elution peaks corresponding to target mutants were pooled and lyophilized. The Tricine/SDS-polyacrylamide gels and the mass spectra showed a high purity of the final products. As an example, the entire purification process of Y42G is shown at the following stages: Tricine/SDS-PAGE before and after purification ( Fig. 2A), reversed-phase chromatography (Fig. 2B), mass spectrometry (Fig. 2C).
Conformational Analysis-The CD spectra of rBmK M1 and its mutants in the UV range of 250 -200 nm are shown in Fig.  3. Compared with native BmK M1, the CD spectra of Y14G and Y35G dramatically changed (Fig. 3A), indicating that there are apparent changes in the secondary structures of these two mutants. The secondary structure estimation (J-700 for Windows Secondary Structure Estimation, Version 1.10.00) indicates that mutation Y14G interrupts both the ␣-helix and ␤-sheet, whereas mutation Y35G interrupts only the ␤-sheet. In both cases, the estimated random coils show a significant increase.
For Y21G, W38G, and Y42G, the CD spectra show that the secondary structures have almost not changed compared with the native toxin (Fig. 3A). It seems that the loss of the aromatic side chains in these mutants does not alter the general struc- ture of the toxin. Regarding mutants with the conservative phenylalanine substitution, the CD spectra show small alterations for Y5F, Y14F, and Y35F, but large changes for W47F compared with wild-type BmK M1 (Fig. 3B).
Bioassay-The mice showed typical symptoms of envenomation after injection with rBmK M1. The LD 50 determined by the method of Meier and Theakston (19) was ϳ0.53 mg/kg, which is consistent with that of native BmK M1 (16). Excluding W47G and Y5G, which were not expressed and expressed only in trace amounts, respectively, the other nine mutants (Y5F, Y14G, Y14F, Y21G, Y35G, Y35F, W38G, Y42G, and W47F) were used for bioassays. Each purified mutant was injected into the mice through the tail vein at different doses to determine the LD 50 value ( Fig. 4 and Table I).
Mutants Y14G, W38G, and Y42G showed no detectable toxicity even at a dose of 25 mg/kg, which is 47 times the LD 50 of rBmK M1 (Table I). Mutant Y35G lost most of its toxicity (LD 50 ϭ 16.04 mg/kg, which is 30 times the LD 50 of rBmK M1). In contrast, the LD 50 of Y21G was only twice that of rBmK M1 ( Fig. 4 and Table I).
For the phenylalanine mutants, the toxicities of Y14F (71%) and Y35F (55%) were in the same order as that of unmodified rBmK M1. The toxicity of W47F displayed a certain decrease (23%). The toxicity of Y5F was dramatically reduced to 4% in comparison with unmodified rBmK M1 (Fig. 4 and Table I).

Effect of rBmK M1 and Its Aromatic Amino Acid Mutants on
Voltage-gated Na ϩ Channels- Fig. 5 displays the effects of rBmK M1 and mutants Y14G, Y35G, W38G, Y42G, Y21G, Y14F, Y35F, Y5F, and W47F on Na v 1.5 Na ϩ channels expressed in Xenopus laevis oocytes. The currents displayed were evoked by a depolarization step to Ϫ20 mV from a holding potential of Ϫ90 mV. The current traces recorded after the addition of the toxin reveal that rBmK M1 induced a slowing of the inactivation process of Na ϩ currents. This effect appeared a few seconds after the addition of the toxin and continued until reaching a steady state after 4 -5 min. Under control conditions, the inactivation kinetics of Na v 1.5 currents were rapid, and almost no remaining currents were visible at the end of the traces, after 25 ms. The toxin-induced slowing of inactivation was evaluated by a single exponential fit (pClamp Version 8) of the current decay after the peak. The time window for each fit was manually set from the peak current to the end of the trace inactivation of the channel. The time constants of inactivation were 2.1 Ϯ 0.3 ms (n ϭ 3) and 1.7 Ϯ 0.3 ms (n ϭ 4) after the addition of 30 M Y14G and 50 M Y42G, respectively, corresponding to 150 and 121% of the time constants under control conditions, respectively. The addition of 30 M W38G increased the time constant of inactivation to 2.6 Ϯ 0.4 ms (n ϭ 3), corresponding to 186% of the control value. Y35G was the second most effective mutant, increasing the value to 4.2 Ϯ 1.3 ms (n ϭ 3) at a concentration of 30 M, corresponding to 300% of the control value. Y21G was the most effective mutant, increasing the value to 4.7 Ϯ 0.4 ms (n ϭ 3) at a concentration of 5 M, corresponding to 335% of the control value.
As shown in Fig. 5, all of the phenylalanine mutants except Y5F had about the same efficacy in Na v 1.5 as rBmK M1. The time constants of inactivation were 5.4 Ϯ 0.6 ms (n ϭ 4), 7.1 Ϯ 0.4 ms (n ϭ 4), and 5.8 Ϯ 0.6 ms (n ϭ 3) after the addition of 10 M W47F, 7.5 M Y35F, and 10 M Y14F, respectively, corresponding to 385, 507, and 414% of the time constants under control conditions, respectively. Y5F was the least effective in slowing the inactivation of the Na ϩ channel. The time constant of inactivation was only 2.3 Ϯ 0.6 ms (n ϭ 3) after the addition of 100 M, corresponding to 165% of the control value. The effects of rBmK M1 and some of its mutants on the peak Na ϩ current and time to peak were somewhat variable (oocyte-dependent) and not further analyzed.
The slowing of inactivation induced by rBmK M1 and its mutants was concentration-dependent (Fig. 6). The EC 50 values of rBmK M1, Y21G, Y35G, Y35F, Y14F, and W47F were determined by a sigmoidal fit of the -V relationship as displayed in Fig. 6. The EC 50 values determined for rBmK M1, Y21G, Y35G, Y35F, Y14F, and W47F were 0.50 Ϯ 0.03, 1.02 Ϯ 0.15, 5.05 Ϯ 0.36, 0.71 Ϯ 0.09, 3.36 Ϯ 0.48, and 3.1 Ϯ 0.3 M, respectively. The EC 50 value determined in this study for rBmK M1 is slightly higher than the EC 50 value determined for the native BmK M1 toxin (0.2 M) in one of our previous studies (13). Y21G was comparable to native rBmK M1. Y35G was ϳ30% less efficient in slowing the inactivation kinetics of Na v 1.5 channels compared with wild-type rBmK M1. The EC 50 value of Y35G was at least 10 times higher than that of rBmK M1. The EC 50 values of the phenylalanine mutants were comparable to that of rBmK M1, except Y5F. As shown in Fig. 6, the EC 50 values of Y14G, Y42G, W38G, and Y5F could not be determined because the highest concentrations used did not reach a maximal effect in the doseresponse curve. DISCUSSION When the three-dimensional structures of the scorpion toxins CsE V3 and AaH II were elucidated ϳ20 years ago, the CHS, mainly including Tyr 5 , Tyr 35 , and Trp 47 /Tyr 47 , was proposed to be responsible for the pharmacological effect of these toxins (17,22,23). Although the CHS is found in all scorpion toxin structures known today, this assumption required experimental identification. The individual residues in this cluster (e.g. Trp 45 in AaH II and Tyr 49 in Lqh ␣ insect toxin) have been assessed by chemical modification (24) and mutagenesis analysis (11,25) and shown to play an important role in bioactivity. In this study, seven aromatic residues, including three amino acids of this cluster, were analyzed by site-directed mutagenesis. Correlating the high impact substitution of glycine to the more conservative mutation of phenylalanine, our results clearly indicate that these conserved aromatic residues are specifically involved in either or both pharmacological function and structural stability.
Aromatic Residues Possibly Involved in Pharmacological Function-The bioassay showed that the toxicity of W38G and Y42G was dramatically reduced (Table I). In concordance, elec-trophysiological analysis showed that W38G and Y42G were the least effective in slowing the inactivation of the sodium channel. The EC 50 could not be determined because the highest concentration available could not induce a maximal effect in the dose-response curves (Figs. 5 and 6). Simultaneously, the CD spectra of these two mutants show the least alteration in comparison with that of unmodified BmK M1 (Fig. 3A). These results can be used to speculate that the conserved aromatic residues Trp 38 and Tyr 42 are involved in the functional performance of the toxin.
Tyr 42 is located in the loop between the ␤ 2 -and ␤ 3 -sheets (Fig. 7). Considering that this loop is remarkably different in sequence and structure between ␣and ␤-toxins (7,17), this aromatic residue may be related to the preference for the distinct target site of ␣-toxins.
Aromatic Residues Possibly Involved in Structural Stability-The non-expression of W47G and the extremely unstable expression of Y5G indicate that these two residues are essential for the general structure of the toxin. The polypeptide chain of the toxin cannot be folded correctly without these aromatic side chains. Y35G could be expressed in an amount comparable to the wild type, but its toxicity was reduced dramatically (Table I). In agreement with the obtained LD 50 value in mice, the corresponding EC 50 value of Y35G in Na v 1.5 was 10 times higher than that of rBmK M1. The CD spectrum also changed dramatically (Fig. 3A), indicating an alteration of the secondary structure of Y35G. Interestingly, the conservative phenylalanine substitution mutants were expressed very well. Y5F and Y35F were present in the culture medium in an amount of ϳ3 mg/liter, comparable to rBmK M1. The expression level of W47F was high (9 -10 mg/liter) in comparison with rBmK M1 (Table I). Compared with rBmK M1, alterations in the CD spectra were milder for Y5F and Y35F, but severe for W47F (Fig. 3B). In addition, the bioassay, in concordance with the electrophysiological characterization, showed that the bioactivities of all of the phenylalanine mutants (although in different degrees) were significantly higher than those of the mutants that lost their aromatic side chains by substitution with glycine (Table I). Hence, by constructing and thoroughly comparing glycine and phenylalanine mutants of conserved aromatic residues in BmK M1, we have shown that the aromatic side chains of W47F and Y35F are indispensable for maintaining the structure and pharmacological function of the toxin. Interestingly, these residues do not have to be Trp 47 or Tyr 35 because the aromatic side chain is the primordial component. Also indicated by this study is the fact that residue 5 has to be a tyrosine because the phenylalanine mutant displayed a very low bioactivity.
The three-dimensional structures of BmK M1 and other ␣-toxins reveal that Tyr 5 , Tyr 35 , and Trp 47 are located on the three-stranded ␤-sheet: ␤ 1 -, ␤ 2 -, and ␤ 3 -strands, respectively. The aromatic rings of these residues are positioned orthogonally one to the other in a so-called "herringbone" arrangement ( Fig. 7), which was identified as the lowest energy configuration of relatively solvent-exposed aromatic rings (26). In this way, this aromatic cluster plays an important role in the stabilization of the three-stranded ␤-sheet. It is plausible to infer that, due to the loss of the interactions between these aromatic rings, the ␤-sheet will be interrupted and maybe even disintegrated. Trp 47 is situated at the center of the cluster (Fig. 7), and its aromatic ring resides in the vicinity of the side chains of both Tyr 5 and Tyr 35 (distances of 3.5-4 Å). It can be hypothesized that the disruption of the herringbone arrangement due to the loss of the aromatic side chains in W47G and Y5G makes the mutants unable to express. This conclusion is supported by the fact that all of the phenylalanine mutants were very well expressed. W47F, Y35F, and Y14F also displayed the essential bioactivity (Table I). Instead of the non-expression for W47G, mutant W47F was expressed at a high level with a relative toxicity of 23%. However, the CD spectrum reveals a severe conformational change (Fig. 3B). In short, the glycine and phenylalanine mutants clearly indicate that, with the exception of Tyr 5 , the presence of the aromatic side chain is sufficient and important for the unique structure and the pharmacological function. Tyr 5 seems to be a unique and irreplaceable amino acid. The CD spectrum shows a milder alteration for this mutant. Inspecting the three-dimensional structure, there is a contact through a hydrogen bond between the functional -OH group of Tyr 5 and the side chain of Lys 58 /Arg 58 , which can be found in all long chain scorpion toxins (6,8). By consequence, an aromatic ring without that -OH group cannot maintain the general structure as shown by the CD spectrum and can interrupt the subtle tertiary arrangement between the N-terminal part and the C-terminal segment, which in turn affects the pharmacological function. Therefore, the importance of Tyr 5 is not only in the aromatic ring, but also in the functional -OH group; and as a consequence, Tyr 5 is highly conserved among ␣-toxins.
Aromatic Residues in Other Sites-Tyr 14 and Tyr 21 are located on Face B (Fig. 1B), which is roughly opposite to Face A, where the CHS is situated. Y14G was almost nontoxic compared with rBmK M1 (Table I). In agreement, its effect on voltage-gated sodium channels was also the least (Figs. 5 and 6). The CD spectrum was seriously altered compared with that of unmodified BmK M1 (Fig. 3A), indicating a conformational change due to the loss of the aromatic side chain in this position. The results show that Tyr 14 is essential for stabilizing the unique conformation of the toxin and is involved in the interaction with the voltage-gated sodium channel. To obtain a more thorough insight in this matter, a more conservative phenylalanine substitution for Tyr 14 was constructed. Y14F was expressed in an amount comparable to that of Y14G. However, in contrast to Y14G (no detectable toxicity and EC 50 Ͼ 100 M), Y14F possesses a high bioactivity (71% relative toxicity and EC 50 ϭ 3.36 Ϯ 0.48 M) (Table I). These data reveal that it is the aromatic side chain of Tyr 14 that mainly contributes to the proper conformation and in turn affects the pharmacological function of the toxin. This residue protrudes from the loop between the ␤ 1 -sheet and the ␣-helix (Fig. 7). The structure shows that its aromatic ring interacts with the side chain of Ile 6 , which is also a crucial conserved residue in ␣-toxins. The hydrophobic interactions between these two residues on the surface may play an important role in stabilizing the unique conformation of this loop so as to influence the ␤-sheet and the ␣-helix of the toxin. The effect of mutation Y21G was milder on bioactivity, on the EC 50 in Na v 1.5, and on the CD spectrum ( Fig. 3A and Table I), indicating that the aromatic residue Tyr 21 is putatively not a crucial determinant for the structure and pharmacological function of the toxin.