Molecular Basis of the High Insecticidal Potency of Scorpion (cid:1) -Toxins*

Scorpion (cid:1) -toxins are similar in their mode of action and three-dimensional structure but differ considerably in affinity for various voltage-gated sodium channels (NaChs). To clarify the molecular basis of the high potency of the (cid:1) -toxin Lqh (cid:1) IT (from Leiurus quinquestriatus hebraeus ) for insect NaChs, we identified by mutagenesis the key residues important for activity. We have found that the functional surface is composed of two distinct domains: a conserved “Core-domain” formed by residues of the loops connecting the secondary structure elements of the molecule core and a variable “NC-domain” formed by a five-residue turn (resi-dues 8–12) and a C-terminal segment (residues 56–64). We further analyzed the role of these domains in toxin activity on insects by their stepwise construction onto the scaffold of the anti-mammalian (cid:1) -toxin, Aah2 (from Androctonus australis

domains (D1-D4), each containing six transmembrane segments (S1-S6) and a membrane-associated re-entrant segment (SS1-SS2), connected by internal and external loops. A key feature in the function of NaChs is their gating behavior, namely the ability to rapidly activate and inactivate upon cell membrane depolarization, leading to transient increase in Na ϩ conductance (1). Due to their key role in excitability, these channels are targeted by a variety of toxins.
Long-chain scorpion toxins are 61-to 76-residue-long polypeptides that share a similar core composed of an ␣-helix packed against a three-stranded ␤-sheet and stabilized by four disulfide bonds. These toxins bind to various receptor sites on the extracellular face of NaChs and alter their gating. Traditionally, they are divided into two major classes, ␣and ␤-toxins, according to their mode of action and binding properties to distinct receptor sites on NaChs (2,3).
Scorpion ␣-toxins prolong the action potential by slowing channel inactivation, possibly through interference with the outward movement of the D4S4 segment necessary for the fast inactivation process (4). The scorpion ␣-toxin binding site, termed neurotoxin receptor site-3, has been shown to involve the extracellular regions of D1 and D4 and in particular the S3-S4 loop in D4 (5,6). Several structurally unrelated peptide toxins from sea anemones and spiders are able to compete with scorpion ␣-toxins on binding to receptor site-3 of vertebrate and invertebrate NaChs, suggesting similarity of this receptor in phylogenetically distant organisms (5,(7)(8)(9)(10)(11)(12). However, dramatic differences in the potency of various scorpion ␣-toxins toward insects and mammals were observed in vivo by toxin administration, and in vitro by competition binding assays to insect and rat-brain neuronal preparations. Based on their preferential activity in mammals and insects, scorpion ␣-toxins were divided into three pharmacological subgroups (see Fig. 1): 1) classic anti-mammalian toxins, such as Aah2 (from Androctonus australis hector) and Lqh2 (from Leiurus quinquestriatus hebraeus), which bind with high affinity to rat brain NaChs (0.2-0.3 nM) and are practically non-toxic to insects; 2) ␣-toxins highly active on insects (e.g. Lqh␣IT), which bind to insect NaChs with high affinity (ϳ0.07 nM) and are over three orders of magnitude less potent in mammalian brain compared with Aah2; and 3) ␣-like toxins that are toxic in both mammalian brain and insects (2,9,13). Despite these differences, ␣-toxins of the three pharmacological subgroups are similarly toxic to mice by subcutaneous injection and were shown to affect with similar affinity rat skeletal muscle NaChs (14). Therefore, the variations in effects of scorpion ␣-toxins on different NaChs imply heterogeneity in toxin functional surfaces as well as in their receptor sites (9,15,16). Elucidation of the inherent ability of various scorpion ␣-toxins to differentiate between insect and mammalian NaChs, or among NaCh subtypes in mammalian excitable tissues, is of potential value for future design of selective drugs and insecticides. However, although partial information on the putative functional surfaces of Lqh␣IT (17,18) and the ␣-like toxin BmK-M1 (19) has been reported, the molecular basis of their ability to distinguish among NaCh subtypes is still unclear. To address this question, we used Lqh␣IT and Aah2, which differ markedly in potency to insect and rat brain NaChs. Molecular dissection of Lqh␣IT revealed that residues important for function are clustered in two distinct domains. Construction of these residues onto the anti-mammalian ␣-toxin, Aah2 (Fig. 1B), yielded a functional chimera with anti-insect activity. Determination of the x-ray structure of this chimera highlighted structural features underlying the preference of scorpion ␣-toxins for insect NaChs.

EXPERIMENTAL PROCEDURES
Bacterial and Animal Strains-Escherichia coli DH5␣ cells were used for plasmid constructions. E. coli BL21 cells were used for the expression of the recombinant toxins using the pET-11c vector (20) as was described (21). Sarcophaga falculata blowfly larvae were bred in the laboratory.
Cloning the Gene Encoding Lqh2-Two degenerate oligonucleotide primers, spanning the first eight and last six amino acid residues of Lqh2 (  (22) as template. The deduced amino acid sequence of the PCR product was identical to that of the mature Lqh2 polypeptide (23). The gene encoding Lqh2 was engineered with NdeI and BamHI restriction sites at the 5Ј-and the 3Ј-termini, respectively, and cloned into the expression vector pET-11cK (21). Surprisingly, no recombinant Lqh2 accumulated within inclusion bodies, although its engineered derivatives were expressed successfully.
Site-directed Mutagenesis, Functional Expression, and Purification of Toxins-Mutations in the cDNAs encoding Lqh␣IT and Lqh2 were generated by PCR with pBluescript bearing the appropriate cDNA clone as template (22). Sequences of the mutated cDNAs were verified prior to expression by automatic sequencing. Production of recombinant polypeptides in a non-soluble form, renaturation, and high-performance liquid chromatography purification of the active toxins were performed as was previously described (17,21). All toxin mutants were eluted from the C 18 high-performance liquid chromatography column as major peaks at 24 -26% acetonitrile, similarly to the wild type toxin (17). Quantification of purified recombinant toxin variants was performed by amino acid analysis using an ABI system 420A/130A (Applied Biosystems Inc., Foster City, CA) after hydrolysis by 6 M HCl under vacuum (18 h at 110°C).
CD Spectroscopy-CD spectra were recorded at 25°C using a model 202 circular dichroism spectrometer (Aviv Instruments, Lakewood, NJ). Toxins (140 M) were dissolved in 5 mM sodium phosphate buffer, pH 7.0, and their spectrum was measured using a quartz cell of 0.1-mm light path. Blank spectrum of the buffer was run under identical conditions and subtracted from each of the toxin spectra.
Toxicity Assays-Four-day-old blowfly larvae (S. falculata; 150 Ϯ 20 mg of body weight) were injected inter-segmentally. A positive result was scored when a characteristic paralysis (immobilization and contraction) was observed up to 5 min after injection. Five concentrations of each toxin were injected to larvae (nine larvae in each group) in three independent experiments. Effective dose 50% (ED 50 ) values were calculated according to the sampling and estimation method of Reed and Muench (24).
Competition Binding Experiments-Radioiodinated Lqh␣IT was prepared as was previously described (9,25). Insect neuronal membranes were prepared from heads of adult cockroaches (Periplaneta americana (26)). Equilibrium competition assays were performed using increasing concentrations of the unlabeled toxin in the presence of a constant low concentration of the labeled toxin (30 -50 pM) (9). The median inhibitory concentration (IC 50 ) values for inhibition of toxin binding were determined by a non-linear regression analysis using the Hill equation, employing a Hill coefficient of 1. Mathematical curve fitting for IC 50 determination was accomplished using Kaleidagraph (version 3.08, Synergy Software). K i values were calculated by the equation K i ϭ IC 50 /1 ϩ [L*/K d ], in which L* is the concentration of the radioactive ligand and K d is its dissociation constant (27).
Crystallization-Crystals of Aah2 Lqh␣IT(face) were grown using the vapor-diffusion (hanging-drop) technique. A 2-l reservoir solution containing 100 mM trisodium citrate dihydrate (pH 5.6), 1 M monoammonium dihydrogen phosphate (Hampton Research Crystal Screen 1, condition No. 11) was mixed with an equal amount of the protein dissolved in water (10 mg/ml). Tetragonal crystals of ϳ0.15 mm ϫ 0.15 mm ϫ 0.03 mm were grown overnight at room temperature. The crystals were packed into thin-wall glass capillaries and analyzed using the synchrotron at the European Synchrotron Radiation Facility (Grenoble, France).
Data Collection and Processing-Single crystal x-ray data were collected at 100 K using an ADSC Q4 charge-couple device detector at ID14-EH4 at the European Synchrotron Radiation Facility. Prior to the data collection, the crystals were harvested for a short time (fraction of a minute) in a solution mimicking the reservoir solution supplemented with 25% ethylene glycol. The oscillation range per image was 0.25°with exposure of 0.5 s/image. In addition to the initial data set, two additional data sets were collected with direct beam intensity reduced by a factor of 12 and 36 to occlude overexposed reflections. Data were integrated with DENZO and scaled with SCALEPACK (28). The crystals belong to space group I422 or I4 1 22 with unit cell dimensions a ϭ b ϭ 74.52 Å, c ϭ 56.97 Å. Data statistics are summarized in Table I.
Molecular Replacement-The structure of Aah2 Lqh␣IT(face) was solved by molecular replacement using the program MOLREP (29). The search model was based on the structure of Aah2 from the scorpion A. australis hector (30) (Protein Data Bank entry 1AHO). The rotation and translation searches were tried in various resolutions and a recognizable solution was obtained in the resolution range of 30 -2.5 Å (r ϭ 0.51, CC ϭ 0.370). The space group ambiguity was resolved on the level of a translation search favoring space group I422.
Model Building and Refinement-The structure of Aah2 Lqh␣IT(face) was rebuilt from the model using the program O (30,31). Refinement was performed using REFMAC5 (32) with 5% of the total reflection set for cross-validation (33). Positional and anisotropic temperature factor refinement was performed. An anisotropic overall B factor correction and bulk-solvent correction were applied. No low resolution or sigma cutoff was used to truncate data. Water molecules were included using ARP/WARP (34) automatic protocol. Refinement statistics for the final model are shown in Table I. The final model of Aah2 Lqh␣IT(face) included 518 protein atoms and 132 water molecules. In addition, sulfate and nitrate ions with obvious electron density and reasonable stereochemistry were included. The final model was validated using PROCHECK (35) and fell within the limits of all quality criteria. For Aah2 Lqh␣IT(face) Ramachandran plot (36), 94.3% of the dihedral angles of the peptide backbone were in acceptable regions of the Phi-Psi diagram. The PDB TABLE I Data-collection and refinement statistics of Aah2 LqhaIT(face) structure R merge ϭ ͓⌺ h ⌺ i ͉I h Ϫ I hi ͉/⌺ h ⌺ i I hi ͔ ϫ 100, in which I h is the mean of I hi observations of reflection h. R factor or R free ϭ ⌺ʈF obs ͉ Ϫ ͉F calc ʈ/⌺͉F obs ͉ ϫ 100 for 95% of recorded data (R factor) or 5% of data (R free ). The numbers in parentheses represent statistics in the highest resolution shell. Data

RESULTS
The scorpion toxin pair, Lqh␣IT and Aah2, represents two extremes in potency for insect versus mammalian brain NaChs (3), and their three-dimensional structures are available (37,38). We have utilized the cDNA clones of Lqh␣IT and Lqh2 to analyze the molecular basis of ␣-toxin specificity for insect NaChs. Because Lqh2 differs from Aah2 only in the first and last residues (Fig. 1A), we constructed an Aah2 clone by mutagenesis of both termini.
Mutagenic Analysis of Lqh␣IT-We further analyzed by mutagenesis the external surface that has previously been implicated in Lqh␣IT function (17,18). Each mutant was produced in a recombinant form, and its binding properties were examined on cockroach neuronal membranes. These assays were accompanied by CD spectroscopy to discern between effects resulting from structural perturbation and those reflecting putative interaction with the channel receptor. In general, the CD spectra of most mutants were indistinguishable from those of the unmodified toxin. Residues, whose substitution decreased substantially the binding affinity (K i mutant/K i wild-type Ͼ 20) with no alteration of the CD spectrum were considered significant for activity and assigned to the putative functional surface. Out of 21 residues analyzed, substitution of 10 (Tyr-14, Glu-15, Asp-19, Tyr-21, Glu-24, Leu-25, Lys-28, Ala-39, Asn-54, and Pro-56) had little effect on activity (Table II) (37)) and Aah2 (red (38)) backbones. The toxins are structurally aligned, and the orientation is similar to that in C.
toxin activity with no apparent structural perturbation, as implied from the unchanged CD spectra (Table II and Fig. 2). Substitution of Trp-38 and Asn-44 by Ala resulted in altered CD spectra accompanied by a marked decrease in activity (Table II and Fig. 2). However, the activity of mutants F17W and W38Y was similar to that of the unmodified toxin (Table  II), suggesting that aromatic side chains at these positions are important for activity. Residues Phe-17, Arg-18, Trp-38, and Asn-44 form a packed cluster on the surface of Lqh␣IT held together by a network of intra-molecular interactions (e.g. aromatic and hydrophobic interactions between Phe-17, Trp-38, and Asn-44, as well as hydrophobic interactions between the phenyl ring of Phe-17 and the hydrophobic moiety of Arg-18 (37)). Therefore, substitution of any of these residues may affect the structural organization of this cluster, and thereby the activity.
Mutagenesis of most residues at the C-tail region (I57A/T, R58K, V59A/G, R58K/V59A, K62A/L/R, and R64N) reduced the activity (Table II), but the most detrimental effect was obtained upon substitution of Arg-58. Neutralization of the charge of Arg-58 (R58N) destroyed toxin function (Table II) and was accompanied by a substantial change in the CD spectrum, indicating structural perturbation (Fig. 2). In contrast, R58K (conserved substitution), with a CD spectrum similar to that of the unmodified toxin (Fig. 2), reduced the activity 77-fold (Table II). This result suggests that the charged amine groups and the aliphatic moiety at position 58 are both important for toxin function.
Thus, residues whose substitution affected the function appear on the molecule exterior in two domains (Fig. 3). One domain consists of Phe-17, Arg-18, Trp-38, and Asn-44, of the loops connecting the secondary structure motifs of the molecule ␣/␤-core ("Core-domain"), whereas the second domain ("NCdomain") is constituted by the five-residue-turn (residues 8 -12) and the C-terminal region (residues 56 -64; refs. 18, 37). To clarify the role of these domains in the unique preference of Lqh␣IT for insect NaChs, we constructed the Core-domain and/or the NC-domain of Lqh␣IT onto the scaffold of the antimammalian scorpion ␣-toxin, Aah2, and analyzed the activity of the resulting chimeras.

FIG. 2. CD spectra of recombinant Lqh␣IT and representative derivatives with altered or unchanged spectra.
as Lqh␣IT (ED 50 ϭ 37 ng/100 mg body weight), its binding affinity (K i ϭ 0.26 Ϯ 0.04 nM) for cockroach neuronal membranes was only 3.9-fold lower than that of Lqh␣IT, and it lost the high affinity for rat brain synaptosomes characterizing Aah2 (K i Ͼ 1000 nM versus 0.2 nM of Aah2). In addition, the toxicity of this chimera to mice injected subcutaneously was similar to that of Lqh␣IT. Therefore, this chimera was termed Aah2 Lqh␣IT(face) . To elucidate the fine structural details conferring preference for insect NaChs, we crystallized Aah2 Lqh␣IT(face) and determined its structure by x-ray diffraction.
Three-dimensional Structure of Aah2 Lqh␣IT(face) -The threedimensional structure of Aah2 Lqh␣IT(face) was determined by x-ray diffraction at 1.3-Å resolution. The core of Aah2 Lqh␣IT(face) is made of an ␣-helix (residues 19 -28) packed against a threestranded anti-parallel ␤-sheet (residues 2-4, 32-37, and 45-51) and is very similar to that of Aah2 (Fig. 4A) (38). The root mean square deviation (r.m.s.d.) calculated for the chimeric peptide backbone atoms encompassing residues 13-55 was 0.15 Å. In contrast, substantial deviations from the Aah2 structure were observed for the backbone atoms of residues 8 -10 and 56 -64 grafted from Lqh␣IT (r.m.s.d. ϭ 1.2 Å and 1.5 Å, respectively). The key feature underlying these deviations is a nonproline cis peptide bond observed during structure refinement between residues Asn-9 and Tyr-10 (Fig. 4B). This bond, which is rare in proteins and introduces a steric strain, is stabilized by side chain-side chain interactions between Asn-9 and Tyr-10 and a pair of main-chain hydrogen bonds between Tyr-10 and His-64 (10O…64N, 2.89 Å and 10N…64OXT, 2.97 Å; Fig. 4C). Similar backbone-backbone interactions were deduced from the NOESY NMR spectra of Lqh␣IT (37), but were not observed in the crystal structure of Aah2 (38). Rather, the peptide bond between Asp-9 and Val-10 in Aah2 assumes a trans conformation, and the distances between the backbone atoms of residues 10 and 64 (10O…64N, 10.78 Å and 10N…64O, 7.44 Å) preclude the formation of hydrogen bonds as was observed in Aah2 Lqh␣IT(face) . Instead, the indole ring of His-64 forms a hydrogen bond with the backbone hydroxyl of Tyr-42. This stabilizing interaction is possible, because the C-terminal stretch of Aah2 bends onto the toxin core (Fig. 4C). Additional differences between the structures of Aah2 and Aah2 Lqh␣IT(face) were ob- The five-residue turn and C-tail regions, which differ mostly among the toxins, are indicated. B, superimposed five-residue turn (residues 8 -12) main-chain atoms of Aah2 (carbon atoms in cyan) and Aah2 Lqh␣IT(face) (carbon atoms in yellow). Nitrogen atoms are in blue, and oxygen atoms are in red. The peptide bond between residues 9 and 10, which adopts a cis conformation in Aah2 Lqh␣IT(face) versus trans conformation in Aah2, is thickened. C, hydrogen bonds formed by the terminal histidines of Aah2 (magenta) and Aah2 Lqh␣IT(face) (green). Residues 1-56 are colored in gray; the C-terminal segments (56 -64) are in green (Aah2 Lqh␣IT(face) ) and magenta (Aah2). D, hydrogen bonds formed between the amide group of Asn11 and the C-terminal segment backbone carbonyls of Lqh␣IT (blue), Aah2 (red), and Aah2 Lqh␣IT(face) (green).

FIG. 3. The two domains that constitute the functional surface of
Lqh␣IT. The toxin backbone is shown in ribbon representation. Residues, whose substitution affected the function (see Table II) are space-filled and colored according to their chemical nature (aliphatic, green; aromatic, magenta; polar, yellow; and positive, blue). Counterpart residues in Aah2 are illustrated for comparison. served in the hydrogen bond network that connects the side chain of Asn11 with the backbone hydroxyls of residues 58 -62 (Fig. 4D).
An intriguing outcome of the above structural differences is depicted in Fig. 5, in which the molecular contact surface of Aah2 Lqh␣IT(face) is compared with those of the parental toxins, Aah2 and Lqh␣IT. Although the five-residue turn and the C-terminal segment of Lqh␣IT form a structural entity that protrudes ϳ10 Å into the solvent, the homologous region in Aah2 has a flat topology and is tightened onto the toxin core. Examination of the Aah2 Lqh␣IT(face) surface contour reveals that it is highly similar to that of Aah2 in the core area but differs substantially around the C-tail, where the flat NCdomain of Aah2 is replaced by an extended protrusion resembling the NC-domain of Lqh␣IT.
In Aah2 Lqh␣IT(face) , the side chain of Arg-58 is fixed in its position by hydrogen bonds between its N⑀ and NH1 atoms to the backbone oxygen atoms of Asn-11, Val-59, and Gly-61 (Fig.  6A). The positively charged moiety of NH 2 is situated at the base of a crevice formed by the side chains of Thr-13, Tyr-42, and the backbone atoms of residues 61-63. During structure refinement, a sulfate ion that co-crystallized with the chimeric peptide was found to occupy this positive niche. This ion is coordinated by N⑀ and NH 2 of Arg-58 and the main-chain nitrogen of Thr-13 (Fig. 6A). The intrusion of a sulfate ion at this position indicates that a functional amine group of Arg-58 is accessible to ligands, as was previously suggested for the ␣-like toxins Bmk-M1 and Bmk-M4 (39).

DISCUSSION
The structural similarity of scorpion ␣-toxins, on the one hand, but their diverse potencies for NaChs of insects and mammals, on the other hand, is an intriguing contrast. To clarify the molecular basis of this puzzle, we combined a mutagenic approach with structure determination using the most potent anti-insect scorpion ␣-toxin, Lqh␣IT, alongside the strongest anti-mammalian toxin, Aah2.
The Functional Surface of Scorpion ␣-Toxins Is Composed of Two Domains-Molecular dissection of Lqh␣IT and com-parison to Aah2 elucidates for the first time that the functional surface is composed of two distinct domains, Coredomain and NC-domain.
The Core-domain is composed of Phe-17, Arg-18, Trp-38, and Asn-44. Although Phe-17 is not conserved, an Arg residue appears at position 18 in most scorpion ␣-toxins (Fig. 1A). In addition to our mutagenesis results, the role of Arg-18 in activity may also be inferred from the lower potency to insects of the toxin variant Bmk-M4 from the scorpion Buthus martensii Karsch, which in contrast to Bmk-M1 and Bmk-M2 has a Gly residue at this position (Fig. 1A (19)). Trp-38 appears on the toxin surface, and an aromatic residue at this position is conserved in all ␣-toxins (Fig. 1A). The change in CD spectrum upon its substitution to Ala may suggest a structural role. Still, its sulfenylation in the toxin Aah2 decreased toxicity to mammals (40), and a W38G mutation impaired the toxicity of the ␣-like scorpion toxin Bmk-M1 (19). Therefore, Trp-38 may also have a role in toxin function. Asn-44 is conserved in most scorpion ␣-toxins (Fig. 1A), although the extent of its exposure to the solvent varies considerably depending on the nature and conformation of neighboring amino acids (residues 18,37,39,44). In the crystal structure of Aah2 Lqh␣IT(face) , Asn-44 is essentially buried, and its side chain forms a barrier between the aromatic rings of Phe-17 and Trp-38, preventing their collapse into the hydrophobic core of the molecule. Therefore, Asn-44 may have a structural role in stabilizing the hydrophobic Core-domain.
The NC-domain, composed of the five-residue turn and the C-terminal segment, varies in amino acid composition and spatial arrangement among ␣-toxin subgroups, suggesting a role in toxin specificity (Fig. 1) (19,39). Its sensitivity to alterations (Table II) (17)(18)(19)39) and the tight interaction between the five-residue turn and the C-tail suggest that the unique conformation of this domain is important for toxin function. The importance of a positively charged residue at position 58 for activity of scorpion ␣-toxins was previously inferred from chemical modifications (41,42) and mutagenesis (18,19). Residue 58 is highly conserved (Arg or Lys) in most scorpion ␣-toxins (Fig.  1), and it is involved in several hydrogen bonds that stabilize the conformation of the C-terminal segment. However, the solvent accessibility of its side chain varies markedly among different toxins (39). On this basis, both structural and functional roles were proposed for residue 58 to explain the dramatic decrease in toxicity upon substitution (19,39,42,43). We show that a conserved substitution, R58K, decreased the activity of Lqh␣IT 77-fold (Table II), whereas its CD spectrum remained similar to that of Lqh␣IT (Fig. 2). The lack of structural perturbation is not surprising, because Lys is capable of sustaining the important hydrogen bonds with the backbone carbonyls of Asn-11, Val-59, and Gly-61 (Fig. 6A). Therefore, the decrease in R58K toxicity could result from the elimination of a charged amine, which suggests a functional role for this charge. This suggestion is reinforced by the interaction of a sulfate ion with Arg-58 observed in the crystal structure of Aah2 Lqh␣IT(face) (Fig. 6A). The accessibility of Arg-58 for charged ions provides for the first time experimental evidence that it may interact electrostatically with a negatively charged residue of the receptor (modeled in Fig. 6B). Such an interaction is expected to be highly favorable in terms of ionic bond distance and low dielectric constant of the surrounding hydrophobic environment.
Structural Basis of Lqh␣IT Preference for Insect Sodium Channels-The close resemblance between the pharmacological properties of Aah2 Lqh␣IT(face) and Lqh␣IT suggests that their functional surfaces are similarly oriented. The crystal structure of Aah2 Lqh␣IT(face) clearly indicates that the gain of anti-insect toxicity is coupled to the arrangement of the NCdomain in a shape that protrudes to the solvent as found in Lqh␣IT (Fig. 5). Notably, this shape is common to all scorpion ␣-toxins active on insects in contrast to the flat NC-domain in anti-mammalian ␣-toxins (Fig. 5). From this observation we conclude that, although residues of the Core-domain are important for activity, it is mainly the NC-domain that confers the specificity for insects. This conclusion is corroborated by the study of Wang et al. (19), in which two mutations in the NCdomain of the ␣-like toxin BmK-M1 affected differently the activity to insects or mammals. On this basis they proposed that the specific orientation of the C-tail mediated by the fiveresidue turn could be relevant to the preference of various ␣-toxin subgroups for phylogenetically distinct NaCh receptor sites. Our results obtained for the Aah2 Lqh␣IT(face) chimera provide experimental evidence for this proposition.
The difference in geometry of the NC-domain is linked to variations found in the backbone conformations of toxins from different subgroups. Whereas a non-proline cis peptide bond was observed between residues 9 and 10 of all scorpion ␣-toxins active on insects (Lqh3, PDB ID 1FH3; BmK-M1, PDB ID 1SN1; BmK-M4, PDB ID 1SN4; Lqh␣IT, unpublished), the same bond assumes a trans conformation in the classic anti-mammalian toxins Aah2 and BmK-M8 (43,44). The occurrence of a non-proline cis peptide bond in proteins is quite rare due to a steric strain as a result of repulsion of the neighboring C ␣ atoms and is compensated by local stabilizing interactions. In a recent survey of 153,209 peptide bonds derived from a nonredundant set of 571 PDB entries (45,46), only 0.03% of the bonds were found in the non-proline cis conformation. Interestingly, most of these bonds appear in functionally important regions, such as those near the active sites of proteins.
A Putative Mechanism of ␣-Toxin Binding to the NaCh Receptor-Our results suggest that binding of scorpion ␣-toxins to the voltage-gated NaCh involves two distinct interaction sites. One interaction is between the conserved Core-domain and a channel region that may be conserved among insects and mammals. This interaction may enable recognition of receptor site-3 by all toxins of the ␣-group and could explain the ability of anti-mammalian ␣-toxins to displace at high concentrations ␣-insect toxins from their binding site on the insect channel (9,47,48). The second interaction is between the variable NCdomain and another channel region. The latter interaction supports high binding affinity and is responsible for the specificity toward NaCh subtypes. This supposition is in concert with the emerging motif from numerous studies of toxin-receptor interactions (49,50) that toxins with a similar mode of action, but diverse specificities for various receptor subtypes, usually share a minimal critical core facilitating the recognition of the receptor by providing an anchoring interaction. The subtype specificity is achieved through additional supporting interactions, which may vary from one toxin to another.
The two-site binding model suggested for Lqh␣IT fits the experimental observations reported for the ␣-mammal toxin, Lqh2, in which receptor site-3 on rat brain synaptosomes was found to occupy two conformations providing high and low affinity binding sites (25). These receptor conformations differed substantially in their membrane potential dependence and in their apparent contribution to the association and dissociation rates (k on and k off ) of the toxin-receptor interaction. Whereas depolarization of synaptosome membrane potential (from Ϫ55 to 0 mV) decreased dramatically the accessibility of the toxin (k on ) to its receptor, it had only a minor effect on its off-rate (25). These findings suggest that the ␣-toxin receptor site is made of two distinct regions, which differently govern toxin association and dissociation. These two regions may correspond to the two proposed toxin-channel interaction sites. Because the Core-domain and NC-domain involved in this interaction are ϳ15-20 Å apart, their counterpart channel regions may be contributed by two adjacent domains (51,52) and hence be differently affected by membrane depolarization. This suggestion is further substantiated by electrophysiological analysis of the interaction of Lqh2, Lqh3, and Lqh␣IT with FIG. 6. The binding pocket of Arg-58. A, interaction with the sulfate ion that co-crystallized with the chimeric toxin. B, putative interaction with a Glu residue of the receptor binding site (dashed yellow lines) modeled according to the coordinates of the sulfate ion. cloned skeletal muscle NaChs (14), in which voltage-dependent dissociation yielded similar slope factors, whereas their association was voltage-independent in the range of Ϫ140 to Ϫ80 mV. A similar scenario, described for the specific interaction of ␣4/7 conotoxins with their ␣3/␤2 neuronal nicotinic acetylcholine receptor, was termed the "Janus ligand" paradigm (50). These toxins appear to possess two distinct domains for the recognition of the nicotinic receptor ␣and ␤-subunits, respectively. Whereas alteration of the toxin at the ␤-subunit recognition site decreased the k on and had no apparent effect on the k off , alteration at the ␣-subunit recognition site dramatically reduced the k off (50). Although a Janus ligand mechanism of binding for scorpion ␣-toxins remains to be demonstrated, our findings strongly suggest a similar mode, in which two distinct regions constitute receptor site-3 on various NaChs. Notably, a recent molecular dissection of the ␤-toxin, Bj-xtrIT (from the scorpion Buthotus judaicus) has revealed two distinct domains on the molecule exterior that constitute its functional surface. One domain includes a hot spot conserved in ␤-toxins, whereas the other domain is unique to the toxin and determines its specificity for insects (53,54). Thus, it seems that the Janus ligand paradigm might be a common mechanism, by which "long-chain" scorpion toxins differentiate among various voltage-gated NaChs.