Maurotoxin Versus Pi1/HsTx1 Scorpion Toxins

Maurotoxin (MTX) is a scorpion toxin acting on several K+ channel subtypes. It is a 34-residue peptide cross-linked by four disulfide bridges that are in an “uncommon” arrangement of the type C1-C5, C2-C6, C3-C4, and C7-C8 (versus C1-C5, C2-C6, C3-C7, and C4-C8 for Pi1 or HsTx1, two MTX-related scorpion toxins). We report here that a single mutation in MTX, in either position 15 or 33, resulted in a shift from the MTX toward the Pi1/HsTx1 disulfide bridge pattern. This shift is accompanied by structural and pharmacological changes of the peptide without altering the general α/β scaffold of scorpion toxins.

Maurotoxin (MTX), 1 a toxin from the venom of the Tunisian chactidae scorpion, Scorpio maurus palmatus, is a basic 34residue peptide cross-linked by four disulfide bridges (1,2). MTX exhibits an uncommonly wide range of pharmacological activities, since it binds onto apamin-sensitive small conductance Ca 2ϩ -activated K ϩ (SK) channels and also blocks voltagegated Kv channels (Shaker B, Kv1.2 and Kv1.3) in the nanomolar concentration range (1)(2)(3)(4). The three-dimensional structure of MTX in solution has been solved by 1 H nuclear magnetic resonance (NMR) (5). It consists in a bent ␣-helix (residues 6 -17) connected by a loop to a two-stranded antiparallel ␤-sheet (residues 22-25 and 28 -31). Therefore, MTX adopts an ␣/␤ scaffold (6) that is similar to those of other scorpion toxins. Such a scaffold is thought to occur independently of the toxin chain length and ion channel pharmacological activity (7,8). In three disulfide-bridged toxins acting on K ϩ channels, this scaffold is associated with the presence of a consensus sequence (6)  . ] is observed instead, which differs from the consensus sequence by the insertion of an extra half-cystine residue within the central part of the sequence (in bold). Also, the remaining half-cystine residue required for the fourth disulfide bridge formation is located at the C-terminal end of the motif as in the case of Na ϩ channel-specific toxins.
To maintain the integrity of the ␣/␤ scaffold, the ␣-helix is connected to the anti-parallel ␤-sheet by two disulfide bridges. The bridging pattern that is generally observed for three disulfide-bridged toxins is of the type C1-C4, C2-C5, and C3-C6. A similar half-cystine arrangement (of the type C1-C5, C2-C6, C3-C7, and C4-C8) is also observed for Na ϩ channel-acting toxins cross-linked by four disulfide bridges and containing the consensus sequence. A similar pattern of half-cystine pairings is again observed in two recently characterized four-disulfidebridged scorpion toxins, Pi1 (from Pandinus imperator (9)) and HsTx1 (from Heterometrus spinnifer (10)), that contain the variant instead of the consensus sequence. These two toxins share high sequence identities (53-68%) with MTX and belong therefore to the same structural family, also referred to as the ␣-KTx6 subfamily (11). Although MTX is structurally related to Pi1 and HsTx1, it differs in its half-cystine pairings, which were unexpectedly "non-conventional" with disulfide bridges connecting C1-C5, C2-C6, C3-C4, and C7-C8. The two first disulfides are "conventional," whereas the two others are rearranged to form short cyclic domains, one between Cys 13 and Cys 19 (C3-C4), and another between Cys 31 and C-terminal amidated Cys 34 (C7-C8). At the structural level, these non-conventional pairings result in a significant difference, with the ␣-helix connected by two disulfide bridges (C2-C6 and C3-C4) to two different strands of the ␤-sheet instead of connecting the ␣-helix to the same strand as in Pi1 and HsTx1 (inferred from their three-dimensional structures (12,13)). Possibly, this unique half-cystine pairing pattern of MTX may contribute to a conformation that could be slightly different from the one that would be exhibited with pairings of the Pi1/HsTx1 type. In * This work was supported financially by CNRS and INSERM. 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.
§ Both authors contributed equally to this work. imperator; HPLC, high pressure liquid chromatography; NOE, nuclear Overhauser effect; NOESY, NOE spectroscopy; TOCSY, total correlation spectroscopy; Fmoc, N ␣ -(9-fluorenyl)methyloxycarbonyl; SK, small conductance Ca 2ϩ -activated K ϩ channels; Kv, voltage-gated K ϩ channels; CNS, crystallography and NMR system (a software suite for macromolecular structure determination); DIANA, distance geometry algorithm for NMR; HN, amide proton. addition, the positioning of key residues in toxin required for the interaction with the ion channel receptor site(s), and therefore crucial for toxin selectivity and/or affinity, could also be affected by the pattern of the disulfide bridges. It is worth noting that MTX exhibits a different pharmacological selectivity than Pi1 and HsTx1, although differences in pharmacology basically rely more on amino acid sequence variation than on disulfide bridge arrangement. For instance, Pi1 is inactive on rat Kv1.3 channels (14), whereas HsTx1 is inactive on rat brain apamin-sensitive SK channels (10), contrary to MTX (1,2).
Here, we have investigated the structural basis of the nonconventional disulfide bridge pattern of MTX. The aim of our study was to introduce, by solid-phase peptide synthesis, targeted point mutations in the MTX amino acid sequence to obtain toxin analogs with Pi1/HsTx1-like disulfide bridging (C1-C5, C2-C6, C3-C7, and C4-C8 instead of C1-C5, C2-C6, C3-C4, and C7-C8). Our rationale in MTX mutagenesis was to introduce by substitution (i) Pi1-specific amino acid residue(s) between C3 and C4 aimed at mimicking Pi1 folding ([Q14]-MTX and [Gln 15 ]MTX), and (ii) an amino acid residue that would produce steric hindrance and/or geometric constraints in order to prevent the connection between C7 and C8 ([Ala 33 ]MTX). The corresponding folded MTX analogues were obtained by solid-phase chemical synthesis and were characterized for half-cystine pairings by enzyme-based cleavage. The consequences of a shift in disulfide bridging onto peptide structure and pharmacology were assessed using 1 H NMR and electrophysiology on Kv channels expressed in Xenopus oocytes.

EXPERIMENTAL PROCEDURES
Materials-N-␣-Fmoc-L-amino acids, Fmoc-amide resin, and reagents used for peptide synthesis were obtained from PerkinElmer Life Sciences. Solvents were analytical grade products from SDS. Enzymes (trypsin and chymotrypsin) were obtained from Roche Molecular Biochemicals.
Chemical  33 ]MTX analogues were obtained by the solid-phase technique (15) using a peptide synthesizer (Model 433A, Applied Biosystems Inc.). Peptide chains were assembled stepwise on 0.25 millimolar equivalent of Fmoc-amide resin (0.65 millimolar equivalent of amino group/g) using 1 mmol of Fmoc-amino acid derivatives. The side chain-protecting groups used for trifunctional residues were: trityl for Cys, Asn, and Gln; tert-butyl for Ser, Thr, Tyr, and Asp; pentamethylchroman for Arg; and tert-butyloxycarbonyl for Lys. The Fmoc-amino acid derivatives were coupled (20 min) as their hydroxybenzotriazole active esters in N-methylpyrrolidinone (4-fold excess). The peptide resins (2.2-2.5 g) were treated for 2.5 h at room temperature with a mixture of trifluoroacetic acid/H 2 O/thioanisole/ethanedithiol (88:5:5:2, v/v) in the presence of crystalline phenol (2.25 g). After filtration of the mixture, the peptide was precipitated and washed by adding cold diethyl ether. The crude peptide was pelleted by centrifugation (3000 ϫ g, 10 min), and the supernatant was discarded. The reduced peptides were then dissolved at 2 mM in 0.  33 ]MTX by Enzyme-based Cleavage and Edman Sequencing Analysis-The two MTX analogues (800 g) were each incubated with a mixture of trypsin and chymotrypsin at 10% (w/w) in 0.2 M Tris-HCl, pH 7.4, for 12 h at 37°C. The peptide fragments were then purified by reversed-phase HPLC (Vydac, C 18 5 m, 4 ϫ 150 mm) with a 60-min linear gradient of 0.08% (v/v) trifluoroacetic acid, 0 -60% acetonitrile in 0.1% (v/v) trifluoroacetic acid, H 2 O at a flow rate of 1 ml/min ( ϭ 230 nm), and freeze dried prior to their analyses. The peptide fragments were hydrolyzed by acidolysis (6 N HCl/phenol), and their amino acid content was analyzed (Beckman, System 6300 amino acid analyzer). The peptides were further characterized by mass spectrometry analysis (RP-DE Voyager, Perseptive Biosystems) and Edman sequencing using a gas-phase microsequencer (Applied Biosystems 470A). In standard HPLC conditions for analyzing phenylthiohydantoin amino acid derivatives, diphenylthiohydantoin-cystine elutes at a retention time of 9.8 min.  1 H NMR spectra were recorded on a Bruker DRX 500 spectrometer equipped with a proton/carbon/azote probe and selfshielded triple axis gradients. The experiments were performed at 300 K. Two-dimensional spectra were acquired using the states-time proportional phase increment method to achieve F1 quadrature detection (16). Water suppression was achieved using presaturation during the relaxation delay (1.5 s), as well as during the mixing time in the case of NOESY experiments. NOESY spectra were acquired using mixing times of 80 ms. Clean-TOCSY was performed with a spin locking field strength of 8 kHz and a spin lock time of 80 ms.
For spectral analysis, identification of amino acid residue spin systems and the sequential assignments were done using the standard strategy described by Wü thrich (17) applied with the graphical software XEASY (18). The TOCSY spectra recorded in water gave the spin system signatures of the protein. The spin systems were then sequentially connected using the NOESY spectra.
For experimental restraints, NOE data were integrated by measuring the peak volumes. On the basis of known distances in regular secondary structures (d H␣-H␣ ϭ 0.23 nm and d HN-HN ϭ 0.33 nm between two strands of an antiparallel ␤-sheet), these volumes were translated into upper limit distances by the CALIBA (a supporting program converting the measured nuclear Overhauser effects into upper distance limits) routine (19) of DIANA (20) software. The lower limit was systematically set at 0.18 nm.
For structure calculations, distance geometry calculations were performed with the variable target function program DIANA 2.8. A pre- liminary set of 1000 structures was initiated including only intraresidual and sequential upper limit distances. From these, the 500 best were kept for a second round, including medium range distances, and the resulting 250 best for a third round, with the whole set of upper limit restraints, and some additional distance restraints, used to define the disulfide bridges (i.e. d S␥, S␥ 0.21 nm, d C␤, S␥ and d S␥,C␤ 0.31 nm). Starting from the 50 best structures, a REDAC (redundant dihedral angle constraints) strategy (20) was finally used to include the additional distance restraints coming from hydrogen bonds proposed by DIANA. Final energy refinement was achieved by CNS (21). The visual analysis was done using the TURBO software (22) and the geometric quality of the structures obtained was assessed by PROCHECK 3.3 and PROCHECK-NMR softwares (23).
Molecular Mechanics Calculations-Steric energy calculations were used to determine which type of half-cystine pairing pattern (Pi1/HsTx1 type versus MTX type) was the most energetically favored in [Gln 15 ]MTX and [Ala 33 ]MTX. These calculations were based on the three-dimensional structure of MTX obtained from the Protein Data Bank (5). Lys 15 and Gly 33 were substituted by Gln and Ala residues, respectively, and full minimization was performed for each of the two types of pairing patterns. Minimization was achieved using the molecular modeling program Insight II (Molecular Simulations Inc.), the Discover-based minimization, and the consistent valence force field. The mathematical method used for minimization was the gradient conjugate.
Neurotoxic Activity of MTX Analogues in Mice-The peptides were tested in vivo for toxicity by determining the LD 50 after intracerebroventricular injections into 20-g C57/BL6 mice. Groups of six mice/dose were injected with 5 l of peptide solution containing 0.1% (w/v) bovine serum albumin and 0.9% (w/v) sodium chloride.
Binding Assay of 125 I-Apamin and Competition by [Abu 19 ,Abu 34 ]MTX onto Rat Brain Synaptosomes-Rat brain synaptosomes were prepared as described by Gray and Whittaker (24). Aliquots of 50 l of 0.1 nM 125 I-apamin were added to 400 l of synaptosome suspension (0.4 mg protein/ml). Samples were incubated for 1 h at 4°C with 50 l of one of a series of concentrations of MTX analogues (10 Ϫ5 -0 Ϫ13 M) in a 500-l final volume. The incubation buffer was 25 mM Tris-HCl, 10 mM KCl, pH 7.2. The samples were centrifuged, and the resulting pellets were washed three times in 1 ml of the same buffer. Bound radioactivity was determined by ␥-counting (Packard Crystal II). The values expressed are the means of triplicate experiments Ϯ S.D. Nonspecific binding, less than 10% of the total binding, was determined in the presence of an excess (10 nM) of unlabeled apamin.
Oocyte Preparation and Electrophysiological Recordings-Stage V and VI Xenopus laevis oocytes were prepared for cRNA injection and electrophysiological recordings. The follicular cell layer was removed by enzymatic treatment with 2 mg/ml collagenase IA (Sigma) in classical Barth's medium. The cDNA plasmids were linearized with SmaI (Shaker B), NotI (rat Kv1.1), XbaI (rat Kv1.2), and EcoRI (rat Kv1.3) and transcribed with either T7 or SP6 RNA polymerase (mMessage mMachine kit, Ambion). The cells were micro-injected 2 days later with 50 nl of cRNA (0.2 g/l Shaker B, rat Kv1.1, rat Kv1.2, or rat Kv1.3 channels). To favor channel expression, cells were incubated at 16°C into a defined nutrient oocyte medium (25) 2-6 days before current recordings. Oocyte currents were then recorded at 18 -23°C by standard two-microelectrode techniques using a voltage clamp amplifier (GeneClamp 500, Axon Instruments, CA) interfaced with a 16-bit AD/DA converter (Digidata 1200A, Axon Instruments). Electrodes filled with 140 mM KCl had an electric resistance of 0.5-1 megaohms. Voltage pulses were delivered every 15 s from a holding potential of Ϫ80 mV. Current records were sampled at 10 kHz, low pass-filtered at 2 kHz using an eight-pole Bessel filter, and stored on computer for subsequent analysis. The extracellular recording solution contained (in mM): 88 NaCl, 10 KCl, 2 MgCl 2 , 0.5 CaCl 2 , 0.5 niflumic acid, 5 HEPES, 0.1% bovine serum albumin, pH 7.4 (NaOH). Leak and capacitive currents were subtracted on-line by a P/4 protocol. Residual capacitive artifacts were blanked for display purposes. Toxin solutions were superfused in the recording chamber at a flow rate of 2 ml/min using a ValveBank4 apparatus (Automate Scientific Inc.). The results are presented as the mean Ϯ S.E. Fig. 1a illustrates the amino acid sequences of MTX, Pi1, and HsTx1 scorpion toxins. MTX shares 53 and 68% sequence identity with HsTx1 and Pi1, respectively. However, the half-cystine pairing pattern of MTX differs from that of Pi1 and HsTx1. It is of the non-conventional type, C1-C5, C2-C6, C3-C4, and C7-C8, instead of C1-C5, C2-C6, C3-C7, and C4-C8 for Pi1/HsTx1 (Fig. 1b). To address the structural basis of the distinct disulfide bridging that is observed between members of the same ␣-KTx6 subfamily, we attempted to alter the nonconventional MTX pairings by single amino acid residue substitution. The aim of these selective mutations was to obtain a Pi1/HsTx1-like pattern of the disulfide bridges for the resulting MTX analogues. For HsTx1, the lack of a glycine or proline at position 32 or 33 (between C7 and C8 of the toxin) is thought not to favor the C7-C8 connection (26 -29) (and therefore the occurrence of the C3-C4 connection), thereby excluding an  MTX-like disulfide bridge arrangement. In contrast, Pi1 contains the C-terminal sequence C 7 YGC 8 , which is identical to that of MTX. Therefore, we focused our study on the variable domains between MTX and Pi1 that are likely to be involved in their distinct half-cystine pairing patterns. First, we targeted the basic Lys 15 of MTX (replaced by an uncharged Gln residue present in Pi1 at an homologous position) and synthesized a [Gln 15 ]MTX analogue. Second, we targeted the MTX C terminus (residues 31-34) between the C7 and C8 connection. A prerequisite to the formation of such a cyclic structure (also referred to as the "14-member disulfide ring" (26 -29)) is the presence of a glycine or proline within the ring (in position 32 or 33), according to geometric constraints. Such a glycine is indeed found at position 33 of MTX. Therefore, an MTX analogue without such a glycine, [Ala 33 ]MTX, has been synthesized with the aim of preventing the C7-C8 half-cystine pairing. In addition, Lys 15 and Gly 33 are convenient for mutagenesis, first because they are located outside the ␤-sheet structure, which is reported to be involved in Kv channel recognition; and second, because they do not form part of the equivalent charybdotoxin key residues required for Shaker pore recognition (30). 33 ]MTX analogues was achieved by means of Fmoc/tert-butyl chemistry (15) using a double coupling strategy with Fmoc-amino acid hydroxybenzotriazole active esters. The yield of assembly ranged between 80 and 90%. The profiles of elution by C 18 reversed-phase HPLC of both peptides at different steps of their syntheses (crude reduced, crude oxidized, and purified fractions) are shown in Fig. 2, a and 33 ]MTX were proteolyzed by a mixture of trypsin and chymotrypsin (Fig. 2c). The results demonstrate that, contrary to MTX, both analogues exhibit half-cystine pairings between Cys 3 -Cys 24 , Cys 9 -Cys 29 , Cys 13 -Cys 31 , and Cys 19 -Cys 34 (type C1-C5, C2-C6, C3-C7, and C4-C8; Fig. 1b). Thus, [Gln 15 ]MTX and [Ala 33 ]MTX differ from MTX by the two last disulfide bridges, a conventional pattern that is identical to that of both Pi1 and HsTx1. Interestingly, a single mutation of basic Arg 14 (by a Gln residue), or dual mutations of both Arg 14 and Lys 15 (by two uncharged Gln residues), also resulted in MTX analogues with Pi1/HsTx1-like half-cystine pairing patterns (data not shown). In contrast, single mutations targeted toward another domain of MTX (substitution by alanine at positions 4, 6, 7, and 10) did not result in a disulfide bridge rearrangement of the peptides (data not shown). These data strongly suggest that the pattern of halfcystine pairings is tightly dictated by the nature of some key residues within MTX amino acid sequence.

Most Favored Disulfide Bridging (MTX versus Pi1/HsTx1 Pattern) of MTX Analogues by Molecular Modeling and Steric
Energy Calculations-We investigated by computer-assisted molecular modeling and steric energy calculations which disulfide bridge pattern, MTX versus Pi1/HsTx1 type, was the most energetically favored for a number of MTX analogues. As shown in Table I, the Pi1/HsTx1 type of half-cystine pairings is associated with the lowest steric energy (most stable conformation) for [Gln 15 ]MTX and [Ala 33 ]MTX, in full agreement with the half-cystine pairings observed experimentally (Fig. 2c). This observation also extends to other analogues such as [Q14]-MTX and [Q14,15]-MTX. Conversely, steric energies were found to converge minimally toward the MTX pattern for analogues that exhibited an experimental MTX disulfide bridging (analogues with a single mutation at position 4, 6, 7, or 10). 1   different domains of MTX, either in the ␣-helix or at the C terminus between C3-C4 or C7-C8, respectively. Fig. 3 illustrates the different steps in the determination of the three-dimensional solution structures of [Gln 15 ]MTX and [Ala 33 ]MTX. The qualitative analysis of sequential NOE intensities for secondary structure determination, together with the pattern of medium range constraints, allowed us to predict a helical conformation between Ser 6 and Thr 17 for both analogues. There are some strong sequential HN(i)-HN(iϩ1), weak, or no H␣(i)-HN(iϩ1), and a stretch of medium range NOE in this region for both toxins. On the basis of the qualitative analyses of these analogues, we detected two extended fragments, one from Ala 22 to Ile 25 and another from Ser 28 to Cys 31 (strong H␣(i)-HN(iϩ1) together with weak HN(i)-HN(iϩ1) sequential NOE). These two strands are connected by a tight turn centered on Asn 26 and Lys 27 . The N-terminal fragment is in an extended conformation from Val 1 to Thr 4 . For structure calculations, sequential assignments of [Gln 15 ]MTX and [Ala 33 ]MTX were achieved first by the standard method of Wü thrich (17) (Fig. 3), and second, most protons were identified and the resonance frequencies determined for both analogues. The structures of the analogues were determined using (i) 348 NOE-based distance restraints (151 intra-residue, 114 sequential, 29 33 ]MTX) constraints per residue. The structures were calculated using a hybrid distance geometry simulated annealing protocol (the distance geometry algorithm for NMR (20) (DIANA) and crystallography and NMR system (21) (CNS)). Table II summarizes the structural statistics. All of the solution structures have a good nonbonded contact and covalent geometry. The best fits are shown in Fig. 3, c and d. The analysis of the structures, using TURBO-FRODO (22) and PROCHECK-NMR (23), indicates the presence of an ␣-helix running from Ser 6 to Thr 17 and a right-hand twisted ␤-sheet from Ala 22 to Ile 25 and Ser 28 to Cys 31 for both MTX peptides. The two strands of the ␤-sheet are connected by a type 1 ␤-turn (31) formed by the residues Asn 26 and Lys 27 (Fig. 3, c and d).
The analysis of the three-dimensional solution structure of [Gln 15 ]MTX reveals that the angle between the axis of the ␣-helix and the ␤-sheet is similar to that observed in MTX (45°v ersus 50°; Fig. 4a). We also found a root-mean square deviation value of 2.1 between [Gln 15 ]MTX and MTX. The comparison of the HN and H␣ chemical shifts of [Gln 15 ]MTX and MTX showed no significant differences except for residues at positions 4, 5, 15, 17, and 33. These differences can be attributed to distinct disulfide bridging, in particular for the C terminus and the region from residues Gln 15 to Thr 17 (Fig. 4b). Of note, these two regions are located between the half-cystine residues Cys 13 -Cys 19 (C3-C4) and Cys 31 -Cys 34 (C7-C8) that are implicated in the disulfide bridge rearrangement. Also, the chemical shift observed for Thr 4 and Gly 5 may be due to the difference in geometry of the Cys 3 -Cys 24 reticulation (C1-C5). Our data strongly suggest that the disulfide bridge organization in [Gln 15 ]MTX does not control the angle between the axis of the ␣-helix and the ␤-sheet of the toxin. To explain the change in the disulfide bridge pattern (MTX toward Pi1/HsTx1 type) in the absence of striking structural modifications, we suggest that the substitution of basic Lys 15 by uncharged Gln is likely to modify the local electrostatic fields of the Cys 13 and Cys 19 thiol groups. In MTX, the basic side chain of Lys 15 is in the close vicinity of the thiol groups of Cys 13 (11.27 Å) and Cys 19 (13.14 Å), and in [Gln 15 ]MTX, the uncharged side chain of Gln 15 is at 11.99 and 14.68 Å from the thiols of Cys 13 and Cys 19 , respectively. In a previous report, local variations in the electrostatic field were reported to affect the thiol pK a values and thereby thiol reactivities (32) and presumably half-cystine pairings. Besides, mutation-induced conformational changes in [Gln 15 ]MTX possibly modify the accessibility of either reactive thiol groups (Cys 13 /Cys 19 ) in the aqueous environment leading to differential thiol reactivities and a Pi1/HsTx1-like pattern of the disulfide bridges (33). Indeed, calculations of solvent accessibility surfaces (MOLMOL) indicate a 2.3-18.3-fold increase in the accessibility of the ␥-sulfurs of Cys 13 , Cys 19 , Cys 31 , and Cys 34 , which are involved in disulfide bridge rearrangement (data not shown).
The analysis of the three-dimensional solution structure of [Ala 33 ]MTX reveals a significant difference in the chemical shifts between [Ala 33 ]MTX and its natural counterpart (Fig.  5a). The main differences are located in region 8 -23 and in the extreme C terminus. The change in the C-terminal region is obviously because of the substitution of Gly 33 . The structural modifications in the 8 -23 region can be attributed to a marked reorientation of the ␣-helix; the latter being parallel to the ␤-sheet axis. This rotation is likely to be induced by the Gly 33 mutation and not by the novel reticulation, because there is no significant reorientation of the ␣-helix as compared with the ␤-sheet in the [Gln 15 ]MTX analogue, although it also adopts the Pi1/HsTx1 disulfide bridge pattern. In MTX, the two ␣-protons of Gly 33 are in close vicinity to the ␦-protons of Pro 20 . In [Ala 33 ]MTX, the Ala methyl group causes a steric hindrance that alters the Pro 20 positioning, inducing backbone movement in the 8 -23 region and thereby a novel ␣-helix orientation. One possible key determinant for this rotation may be the integrity of the 14-member disulfide ring itself. A similar observation is made for the [Abu 19 ,Abu 34 ]MTX analogue, which also displays the same angle rotation (4) (Fig. 5b). This peptide is a three disulfide-bridged MTX analogue, which was designed to restore the entire consensus motif of scorpion toxins by substituting the two half-cystine residues in positions 19 and 34 (corresponding to C4 and C8) by two isosteric ␣-amino butyrate derivatives. The three-dimensional structure of [Abu 19 ,Abu 34 ]MTX in solution shows that it adopts the ␣/␤ scaffold with conventional half-cystine pairings connecting C1-C5, C2-C6, and C3-C7. This novel bridging results in a reorientation of the ␣-helix regarding the ␤-sheet structure and is associated with changes in peptide pharmacology (4). It should be noted that in the case of [Ala 33 ]MTX, the structural elements underlying the shift in disulfide bridging are totally unrelated to those implicated in the [Gln 15 ]MTX shift. These elements remain to be investigated. However, we conclude with two points: (i) the formation of an MTX-like pattern is impossible in the case of [Ala 33 ]MTX, and (ii) the Pi1/HsTx1 pattern appears to be more stable than any other potential combination in half-cystine pairings. There are no significant chemical shift differences between [Ala 33 ]MTX and [Abu 19 ,Abu 34 ]MTX, with the exception of the C-terminal region, which can be attributed to the Gly/Ala 33 and/or Cys/Abu 34 mutation(s) (Fig. 5a). Also, the difference observed in the 17-22 region can be explained by  (i) the Abu 19 mutation and/or (ii) the lack of a fourth disulfide bridge. Interestingly, the additional difference observed in the 2-4 region may be attributed to a distinct C1-C5 disulfide bridge geometry, as found in the case of [  34 ]MTX, both analogues remain active but with some changes in affinity and/or specificity toward K ϩ channels.
To address the pharmacological activity of both peptides, we tested, first, their ability to compete with 125 I-apamin for binding onto SK channels of rat brain synaptosomes and, second, the effects on currents resulting from Kv-type channel expression in Xenopus oocytes. The binding experiments illustrate that there is an 8-and 33-fold decrease in affinity for apaminsensitive SK channels for [Gln 15 ]MTX and [Ala 33 ]MTX, respectively (Fig. 6) (Fig. 7).   Fig. 7d). These observations suggest that the integrity of the C-terminal domain of MTX is required for rat Kv1.3 recognition, in agreement with a previous report studying the influence of the enantiomerization of this domain (34). The loss of activity of both [Gln 15 ]MTX and [Ala 33 ]MTX on rat Kv1.3 is clearly independent of (i) the relative orientation between the ␣-helix and the ␤-sheet, because these peptides display very different angles (one MTXlike and the other [Abu 19 ,Abu 34 ]MTX-like; both Kv1.3 active), and (ii) the presence (or not) of a cyclic C-terminal domain, because [Abu 19 ,Abu 34 ]MTX and HsTx1, which lack C-terminal cyclic domains, are nevertheless active on Kv1.3 channels.
Our data demonstrate that two single mutations within the MTX amino acid sequence can result in a shift in the disulfide bridge pattern, which is accompanied by structural and pharmacological changes. The disulfide bridge organization contributes to the spatial distribution of key residues that are implicated in ion channel recognition. In the case of MTX, the shift in half-cystine pairings results in an increased selectivity of the toxin for rat Kv1.2 channels. Our data illustrate that the affinity of the MTX analogues for Kv1.2 is maintained, whereas there is a loss and a reduction of activity on rat Kv1.3 and SK channels, respectively. Additional mutagenesis on these analogues aimed at further decreasing the affinity on SK channels may render the toxin Kv1.2 selective.