Purification, characterization and biosynthesis of ... - Wiley Online Library

Eur. J. Biochem. 269, 1854–1865 (2002) Ó FEBS 2002

doi:10.1046/j.1432-1033.2002.02833.x

Purification, characterization and biosynthesis of parabutoxin 3, a component of Parabuthus transvaalicus venom Isabelle Huys1, Karin Dyason2, Etienne Waelkens3, Fons Verdonck4, Johann van Zyl5, Johan du Plessis2, Gert J. Mu¨ller5, Jurg van der Walt2, Elke Clynen6, Liliane Schoofs6 and Jan Tytgat1 1

Laboratory of Toxicology, University of Leuven, Leuven, Belgium; 2Department of Physiology, University of Potchefstroom, Potchefstroom, South Africa; 3Laboratory of Biochemistry, University of Leuven, Leuven, Belgium; 4Interdisciplinary Research Centre, University of Leuven Campus Kortrijk, Kortrijk, Belgium; 5Department of Pharmacology, University of Stellenbosch, Tygerberg, South Africa; 6Laboratory for Developmental Physiology and Molecular Biology, University of Leuven, Belgium

A novel peptidyl inhibitor of voltage-gated K+ channels, named parabutoxin 3 (PBTx3), has been purified to homogeneity from the venom of Parabuthus transvaalicus. This scorpion toxin contains 37 residues, has a mass of 4274 Da and displays 41% identity with charybdotoxin (ChTx, also called Ôa-KTx1.1Õ). PBTx3 is the tenth member (called Ôa-KTx1.10Õ) of subfamily 1 of K+ channel-blocking peptides known thus far. Electrophysiological experiments using Xenopus laevis oocytes indicate that PBTx3 is an inhibitor of Kv1 channels (Kv1.1, Kv1.2, Kv1.3), but has no detectable effects on Kir-type and ERG-type channels. The dissociation constants (Kd) for Kv1.1, Kv1.2 and Kv1.3 channels are, respectively, 79 lM, 547 nM and 492 nM. A synthetic gene encoding a PBTx3 homologue was designed and expressed as a fusion protein with the maltose-binding protein (MBP) in Escherichia coli. The recombinant protein was

The southern African scorpion Parabuthus transvaalicus Purcell, 1899, is one of the largest scorpions belonging to the Buthidae family [1], subphylum Chelicerata, order Scorpionis. Severe envenomation with P. transvaalicus causes primarily neuromuscular effects with involvement of the heart and parasympathetic nervous system [2], illustrating that this scorpion can be potentially lethal, especially for children. P. granulatus scorpionism has been described by Mu¨ller [3]. P. transvaalicus scorpionism is clinically similar, but appears to produce slightly more motor and fewer sensory symptoms [4]. Crude, diluted venom of P. transvaalicus was already tested on isolated cardiomyocytes and induced an increase in the sodium current and a retardation of the time course of inactivation, implicating the presence of an a-toxin [5]. Verdonck et al. [6] reported the occurrence of pore-forming activity in the venom of P. transvaalicus,

Correspondence to J. Tytgat, Laboratory of Toxicology, University of Leuven, E. Van Evenstraat 4, 3000 Leuven, Belgium. Fax: + 32 16 32 34 05, Tel.: + 32 16 32 34 03, E-mail: [email protected] Abbreviations: PBTx3, toxin from the venom of the scorpion Parabuthus transvaalicus; AgTx2, toxin from the venom of the scorpion Leiurus quinquestriatus var. Hebraeus; MBP, maltose-binding protein; fXa, factor Xa. Note: a website is available at http://www.toxicology.be (Received 31 December 2001, accepted 12 February 2002)

purified from the bacterial periplasm compartment using an amylose affinity resin column, followed by a gel filtration purification step and cleavage by factor Xa (fXa) to release the recombinant toxin peptide (rPBTx3). After final purification and refolding, rPBTx3 was shown to be identical to the native PBTx3 with respect to HPLC retention time, mass spectrometric analysis and functional properties. The threedimensional structure of PBTx3 is proposed by homology modelling to contain a double-stranded antiparallel b sheet and a single a-helix, connected by three disulfide bridges. The scaffold of PBTx3 is homologous to most other a-KTx scorpion toxins. Keywords: Parabuthus; purification; synthesis; scorpion; toxin.

but the variability was rather high and in some specimens this activity was absent. A study was undertaken to find compounds or toxins in the venom of P. transvaalicus that modulate physiological processes at the cellular level; this was done for the following reasons: (a) very little is known about the bioactive substances present in the venom of this scorpion [7,8]; (b) the discovery of new toxins can be the key to gain insight into the molecular mechanisms of scorpionism; (c) selective toxins can be used for purifying channels from native tissue, determining their subunit composition [9] and for elucidating the pharmacology and physiological roles of voltagedependent Na+, Ca2+ and K+ channels [10–12] in target tissues. Voltage-dependent K+ channels in particular serve important functions in many signal-transduction pathways in the nervous system: they are involved in neuron excitability; they influence the resting membrane potential, the waveforms and frequencies of action potentials; and they determine the thresholds of excitation [13]. Moreover, they are the putative target sites in the design of therapeutic drugs [14]. In our work, a new short-chain toxin acting on Kv1 channels, called parabutoxin 3 (PBTx3), has been purified to homogeneity from the venom of P. transvaalicus and its specific function on different channels has been analysed electrophysiologically. Using a recombinant expression system, the toxin was produced in high quantity to confirm our data and to facilitate the screening of the active peptide

Ó FEBS 2002

Novel K+ channel blocker parabutoxin 3 (Eur. J. Biochem. 269) 1855

PBTx3. In this way, a study of the structure–function relationship of PBTx3 to different ion channels and receptors could be performed and a structural model for this novel toxin has been proposed.

MATERIALS AND METHODS Venom collection and purification P. transvaalicus scorpions were captured in South Africa. Venoms were collected by electrical stimulation and lyophilized after dilution in a saline buffer or distilled water. The lyophilized venom was dissolved in 100 mM ammonium acetate, pH 7 (Merck, Germany). After vortexing, the sample was clarified by centrifugation at 12 000 g for 15 min and its supernatant was submitted to gel filtration (Fig. 1A) using a Superdex 30 prep grade HiLoad 16/60 FPLC column (Pharmacia LKB Biotech, Sweden) equilibrated with 100 mM ammonium acetate, pH 7. The mate-

rial was eluted with the same buffer at a flow rate of 0.2 mLÆmin)1. Absorbance of the eluate was monitored at 280 nm and 4-mL fractions were collected automatically. The fraction containing the toxin was recovered, lyophilized and applied on a PepRPC HR 5/5 C2/C18 reversed-phase FPLC column (Pharmacia, Sweden) equilibrated with 0.1% trifluoroacetic acid (TFA, Merck Eurolab, Belgium) in distilled water (Fig. 1B). Separation was performed by using a linear gradient of 0–50% UV-grade acetonitrile (LiChroSolvÒ gradient grade, Merck Eurolab), supplemented with 0.1% TFA, for 30 min. The flow rate was 0.5 mLÆmin)1 and the absorbance was measured at 214 nm. Fractions between 17 and 23 min with potential short-chain toxins were recovered, dried (Speed VacÒ Plus, Savant, USA), and applied to a monomeric 238TP54 C18 reversed-phase HPLC column (Vydac, USA) equilibrated with 0.1% trifluoroacetic acid in distilled water (Fig. 1C). Separation was performed as follows: after 4 min a linear gradient to 30% acetonitrile, for 2 min, followed by a linear gradient to 42% for the final 8 min (total run, 14 min). The flow rate was 0.75 mLÆmin)1 and the absorbance was measured simultaneously at 214, 254 and 280 nm. The toxin-containing fraction (see Fig. 1C) was recovered and dried (Speed VacÒ Plus). Sequence determination The first 36 residues of the primary structure of the peptide were resolved by direct sequencing (Edman degradation) (Fig. 2A). A glass fibre disk was coated with Biobrene (Applied Biosystems) and precycled for four cycles. Subsequently, the sample (18 pmol) was loaded onto the glass fibre disk and subjected to N-terminal amino-acid sequencing on a Perkin Elmer/Applied Biosystems Procise 492 microsequencer (PE Biosystems) running in pulsed liquid mode. To identify the last C-terminal residue, a sample of peptide was also cleaved by cyanogen bromide. By this reaction, three fragments were produced (E1–M4, R5–M28 and N29–R37), separated by HPLC by using the same C18 analytical column as described above, and then sequenced. The last amino acid (arginine) was elucidated. Construction of the recombinant genes

Fig. 1. Purification of native PBTx3 from the venom of P. transvaalicus. (A) Crude venom was first fractionated by FPLC gel filtration, yielding four peaks. The labelled fraction (*) was recovered and lyophilized. Based on a constructed gel filtration calibration curve, the molecular mass of the material in this fraction ranged from 3 to 6 kDa. (B) The second purification step was carried out using a FPLC C2/C18 reversed-phase column. Fractions eluting at 17–23 min (*) contain ÔpotentialÕ short-chain toxins and were recovered and dried. (C) The third step involved a HPLC C18 reversed-phase purification.

A cDNA fragment encoding a 36 amino-acid peptide, corresponding to PBTx3 without the C-terminal arginine, was designed as follows (Fig. 3A). Two overlapping oligonucleotide pairs 5¢-GAGGTCGACATGCGCTGCA AGTCGTCGAAGGAGTGCCTGGTCAAGTGCAAG CAG-3¢, 3¢-CTCCAGCTGTACGCGACGTTCAGCAG CTTCCTCACGGACCAGTTCACGTTCGTCCGCTG CCCGGCC-5¢, and 5¢-GCGACGGGCCGGCCGAACG GCAAGTGCATGAACCGGAAGTGCAAGTGCTAC CCGTGAG-3¢, 3¢-GGCTTGCCGTTCACGTACTTGGC CTTCACGTTCACGATGGGCACTCCTAG-5¢, respectively, ranging in length from 49 to 66 base pairs, were synthesized chemically on an Applied Biosystem device (Amersham Pharmacia Biotech, The Netherlands), purified by PAGE and phosphorylated at the 5¢ end. The complementary oligomers (100 pmol of each) were annealed to generate two duplexes that were ligated using T4 DNA ligase (NEB). The synthetic PBTx3 gene was inserted into the vector pMAL-p2X (NEB) downstream from the malE gene of Escherichia coli and also directly downstream of a fXa site

1856 I. Huys et al. (Eur. J. Biochem. 269)

Fig. 2. Sequence determination of native PBTx3. (A) The first 36 amino acid residues of PBTx3 were identified by direct sequencinga. Sequencing the last fragment, produced after cyanogen bromide cleavage, identified the C-terminal residue arginineb. (B) Alignment of the amino acid sequences of the members of subfamily 1 of short-chain a-KTx toxins isolated from scorpion venom. Dashes represent gaps that were introduced to improve the alignment. Identical amino acids are indicated with a black background. Homologous residues are indicated with a grey background. The percentage identity with ChTx is shown. ChTx (charybdotoxin [24]), charybdotoxin-Lq-2 [10], Lqh 15–1 [25] and AgTx2 (agitoxin 2 [15]), were purified from Leiurus quinquestriatus var. Hebraeus; BmTx1–2 [26] was purified from Buthus martensi Karsch; HgTx2 (hongotoxin 2 [27]), and LbTx (limbatotoxin [34]), were purified from Centruroides limbatus; IbTx (iberiotoxin [28]), and TmTx (tamulotoxin [56]), were purified from Buthus tamulus; PBTx3 (parabutoxin 3, this study) was purified from Parabuthus transvaalicus.

into a XmnI site. The gene possessed an overhang at the 3¢ end (BamHI) to direct the orientation of the insert into pMAL-p2X. The transformants containing the correctly constructed DNA fragments for PBTx3 were analysed by digestion with two different restriction enzymes NaeI and XmnI (NEB). Because insertion of the synthetic gene disrupts the XmnI recognition site, this enzyme cannot cleave the recombinant plasmid. To cleave the gene in the second part of its sequence, NaeI was used as a double control of the original duplexes. In both cases, E. coli JM109 (Promega, The Netherlands) was used for plasmid propagation. A translation termination codon was inserted at the end of the PBTx3 coding sequence. The vector possesses malE translation initiation signals to direct the toxin-fusion proteins to the periplasm, thus allowing folding and disulfide bond formation to take place in E. coli [15,16]. The method for the expression of our toxins used the strong Ptac promoter, which gave a high-level expression of the cloned sequences encoding the fusion. For comparison with PBTx3, the high affinity K+ channel blocker AgTx2 [17], which is structurally related to PBTx3, was produced by a similar strategy. Expression, purification and cleavage of fusion proteins Rich Luria–Bertani medium containing bactotryptone (Sigma, Belgium), yeast (Remel, BioTrading, Belgium),

Ó FEBS 2002

Fig. 3. Schematic diagram of the pMAL-p2X vector containing the synthetic gene for the PBTx3 homologue. (A) Two ligations were performed using a 6706-bp pMAL-p XmnI/BamHI fragment and a 111-bp fragment encoding the PBTx3 homologue, immediately downstream of the fXa cleavage site in the vector. AmpR, ampicillin resistance gene; ori, origin. (B–D) Chromatographic profiles after purification of the fusion protein (B) and recombinant toxin (C,D) rPBTx3. Fractions containing the MBP-fusion proteins were collected and prepared for cleavage with fXa. The restriction digests were applied on the same HPLC C18 column as in Fig. 1 and material eluting between 8 and 15 min was purified further on a HPLC C2/C18 column and tested on Kv1 channels expressed in Xenopus oocytes.

NaCl (Merck Eurolab, Belgium), glucose (Merck Eurolab) and ampicillin (1 lgÆmL)1) was inoculated with an overnight culture of E. coli DH5a cells, carrying the gene fusions encoding either rAgTx2 or rPBTx3, in a culture shaker incubator (Innova 4000, New Brunswick Scientific). In both situations, the cells were grown at 37 °C and when the cell density had reached A600 ¼ 0.5, expression of the fusion proteins was induced by adding isopropyl thio-b-D-galactoside (Sigma) to a final concentration of 0.2 mM. Cells were harvested by centrifugation at 2660 g at 4 °C for 20 min and subjected to osmotic shock according to the following procedures. The cells were resuspended in 400 mL 30 mM Tris/HCl (Sigma) with 20% sucrose (Sigma) pH 8.0 at 25 °C. The suspension was treated with Na2EDTA (Sigma) to give a concentration of 1 mM and incubated at room temperature with shaking. After 10 min, the mixture was centrifuged for 20 min at 2660 g at 4 °C. The supernatant was removed and the well drained pellet was resuspended in 400 mL ice-cold 5 mM MgSO4 (Sigma) in an ice bath for 10 min and centrifuged at 2660 g at 4 °C. The supernatant is the cold osmotic shock fluid which contains the periplasmic extracts. The periplasmic extracts (400 mL) were loaded

Ó FEBS 2002


to an amylose affinity resin (1.5 · 23 cm column, Biolabs, NEB) at a flow rate of 1 mLÆmin)1 in column buffer containing 20 mM Tris/HCl, 200 mM NaCl (Merck Eurolabs, Belgium), and 1 mM Na2EDTA buffer, pH 7.4. After washing of the unbound proteins, the bound maltosebinding protein (MBP)-fusion products were eluted from the amylose resin using the same column buffer containing 10 mM maltose (Merck Eurolabs). Twenty 3-mL fractions were collected and the fusion protein was easily detected by the UV absorbance spectrophotometer (UV/VIS Spectrophotometer lambda 16, PerkinElmer) at 280 nm. The protein-containing fractions were pooled and purified further using a Superdex Peptide gel filtration column on the SMART System (Pharmacia Biotech). The elution was performed with a buffer containing 20 mM Tris/HCl and 100 mM NaCl, pH 8.0 (Fig. 3B). Controls were performed with cells containing no vector or cells containing the vector without insert. The synthetic gene encoding the PBTx3 homologue was designed such that an fXa cleavage site (Ile-Glu-Gly-Arg-) immediately preceded the N-terminal Glu of the toxin (Fig. 3A). The enzymatic cleavage of the pooled fusion proteins was carried out at various conditions by fXa (different sources: Boehringer, Sigma, NEB). Optimal cleavage could be performed in the following conditions: 72 h incubation at room temperature and a concentration of 0.5 UÆlg)1 fusion protein in a buffer containing 20 mM Tris/HCl, 100 mM NaCl and 2 mM CaCl2, pH 8.0. After cleavage with this enzyme, the recombinant toxin was generated without vector-related fragments. In a parallel experiment with AgTx2, chromatographic profiles of rAgTx2 and commercially available rAgTx2 (Alomone Laboratories) under the same conditions were compared and were identical. HPLC Separations of the recombinant proteins were first performed with a 218TP104 C18 reversed-phase HPLC column (Vydac) and equilibrated with 0.1% trifluoroacetic acid (Sigma) at 25 °C (Fig. 3C). After 4 min an immediate step to 5% acetonitrile (with 0.1% trifluoroacetic acid) was followed by a linear gradient to 30% acetonitrile for 5 min and then by a linear gradient to 60% for the last 12 min. The flow rate was 0.75 mLÆmin)1 and the absorbance was measured simultaneously at 214, 254 and 280 nm. The fraction containing the recombinant toxin (arrow) was recovered and applied to a lRPC C2/C18 SC 2.1/10 reversed-phase HPLC column (Vydac). A linear gradient, starting after 6 min and ranging from 0% to 30% up to 100 min with a flow rate of 200 lLÆmin)1 (Fig. 3D), was applied and the toxin was collected, dried (Speed VacÒ Plus) and prepared for functional analysis. Mass spectroscopy For examination of mass, 1 pmol of the venom was dried and redissolved in acetonitrile (+ 0.1% trifluoroacetic acid). The molecular mass of the compounds in the venom and the masses of rAgTx2 (used as a control toxin) and rPBTx3 were determined with MALDI-TOF MS on a VG Tofspec (Micromass, UK) operating in the linear and in the reflectron mode.

Electrophysiological recording Oocyte expression – Kv1.1. For in vitro transcription, plasmids were first linearized with PstI (New England Biolabs) 3¢ to the 3¢ nontranslated b-globin sequence in our custom-made high expression vector for oocytes, pGEMHE [18–20] and then transcribed using T7 RNA polymerase and a cap analogue diguanosine triphosphate (Promega). Kv1.2. The cDNA encoding Kv1.2 (originally termed RCK5) in its original vector, pAKS2, was first subcloned into pGEMHE [19]. The insert was released by a double restriction digest with BglII and EcoRI. Next, the cDNA was loaded onto an agarose gel, fragments of interest were cut out, gene cleaned (QIAGEN) and ligated into the BamHI and EcoRI sites of pGEM-HE. For in vitro transcription, the cDNA was linearized with SphI and transcribed using the largescale T7 mMESSAGE mMACHINE transcription kit (Ambion). Kv1.3. Plasmid pCI.neo containing the gene for Kv1.3 was linearized with NotI (New England Biolabs) and transcribed as for Kv1.2 [21]. Stage V–VI Xenopus laevis oocytes were isolated by partial ovariectomy under anaesthesia (tricaine, 1 gÆL)1). Anaesthetized animals were kept on ice during dissection. The oocytes were defolliculated by treatment with 2 mgÆmL)1 collagenase (Sigma) in zero calcium ND-96 solution (see below). Between 2 and 24 h after defolliculation, oocytes were injected with 50 nL of 1– 100 ngÆlL)1 cRNA. The oocytes were then incubated in ND-96 solution at 18 °C for 1–4 days. The animals were handled in conformity with the ‘Guide for the Care and Use of Laboratory Animals’, published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). Electrophysiology. Whole-cell currents from oocytes were recorded using the two-microelectrode voltage clamp technique. Voltage and current electrodes (0.4–2 megaohms) were filled with 3 M KCl. Current records were sampled at 0.5-ms intervals after low pass filtering at 0.1 kHz. Off-line analysis was performed on a Pentium(r) III processor computer. Linear components of capacity and leak currents were not subtracted. All experiments were performed at room temperature (19–23 °C). Fitted Kd values were obtained after calculating the fraction current left over after application of several toxin concentrations in different oocyte experiments (mean ± SD, n). Solutions. The ND-96 solution (pH 7.5) contained 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 5 mM Hepes, supplemented with 50 mgÆL)1 gentamycin sulphate (only for incubation). Modeling A model was generated by an automated homology modelling server (Expert Protein Analysis System proteomics server using SWISS-MODEL-ProModII) running at the Swiss Institute of Bioinformatics (Geneva). Target (PBTx3) and template (hongotoxin 2) sequences were automatically aligned by Multiple Sequence Alignment Software (CLUSTALW), which subsequently generated the coordinates of both models. Energy minimization (GROMOS96) and simulated annealing cycles were run. SWISS-MODEL computes a confidence factor for each atom


Ó FEBS 2002

in the model structure, taking into account the deviation of the model from the template structure and the distance trap value used for framework building.

RESULTS Kv1 K+ channels were expressed in X. laevis oocytes and studied using a two-microelectrode voltage clamp. Crude venom of P. transvaalicus (340 lg) produces a reversible inhibition of the Kv1.1 K+ current elicited by depolarization up to 0 mV (data not shown). In our quest to find novel short-chain scorpion toxins in the venom of P. transvaalicus, acting on voltage-dependent K+ channels, we fractionated the crude venom of this scorpion as detailed in Materials and methods (Fig. 1). As described by Debont et al. [22], gel filtration shows three typical groups of components (Fig. 1A), the largest of which (group I) was shown to block Kv1.1 channels. Based on a constructed calibration curve (see Debont et al. 1998), the active fraction corresponded to a molecular mass between 3 and 6 kDa, which probably represents the family of short-chain scorpion toxins. After HPLC purification of this active fraction (Fig. 1C), that representing native PBTx3 (85 lM) caused an inhibition of Kv1.1 channels of 50%, whereas 550 nM native PBTx3 produces 54% and 51% block of the Kv1.2 and Kv1.3 channels, respectively (Fig. 4A–C). We have undertaken the recombinant synthesis of this toxin in order to facilitate the characterization of its biological properties. The yields of affinity-purified proteins were 40–60 mgÆL)1 culture, estimated by absorbance at 280 nm, which after cleavage resulted in the production of 2–4 mg of recombinant toxin per litre culture. The recombinant synthesis resulted in the production of a recombinant toxin with an expected molecular mass of 4118 Da, with respect to the three disulfide bridges present in the secondary structure of the PBTx3 homologue. The mass of rAgTx2 (ÔcontrolÕ toxin for comparison) was also consistent with the theoretical mass. Functional effects of recombinant toxins on Kv.1 channels were investigated by electrophysiological experiments. No block was obtained when MBP-rPBTx3 was applied to expressed K+ channels in Xenopus oocytes (n ¼ 3) (data not shown). Recombinant PBTx3 inhibits both Kv1.2 and Kv1.3 channels with weak affinities and similar potencies, whereas it is a very weak inhibitor of Kv1.1 channels: application of 550 nM rPBTx3 produced no blocking effect on Kv1.1 channels (Fig. 4D), whereas the Kv1.2 and Kv1.3 currents were reversibly blocked to 52% and 49%, respectively (Fig. 4E,F). As part of a control, rAgTx2 was applied to the same oocytes expressing Kv1.1 channels. Addition of 1 nM rAgTx2 blocked the K+ current almost completely (Fig. 4G) and this effect was reversible upon washout. After equilibration of the channels and application of the same concentration of commercially available rAgTx2, quantitatively the same effect was observed as with our laboratory prepared rAgTx2. This observation, together with the fact that co-injection of equimolar amounts of both AgTx2 on reverse-phase HPLC resulted in a single peak (data not shown), demonstrates that our rAgTx2 behaved similarly to the commercially available recombinant toxin. Blockage of the Kv1 channels induced by rAgTx2 or rPBTx3 (tested at different concentrations) was shown not to be voltage-dependent, as the degree of block was not

Fig. 4. Effects of native (A–C) and recombinant (D–F) PBTx3 on Kv1.1, Kv1.2 and Kv1.3 channels. Whole-cell K+ currents through Kv1.1, Kv1.2 and Kv1.3 channels, respectively, expressed in Xenopus oocytes, are evoked by depolarizing the oocyte from a holding potential of )90 mV to 0 mV. The oocytes were clamped back to )90 mV (A), or to )50 mV (B–G). Application of 85 lM native PBTx3 (active fraction in Fig. 1C indicated by *) on Kv1.1 channels or 550 nM on Kv1.2 and Kv1.3 channels, produced 50%, 54% and 51% inhibition, respectively, of the Kv1.1, Kv1.2 and Kv1.3 currents. (D–F) Current through Kv1.1, Kv1.2 and Kv1.3 channels, respectively, in control conditions (s) and in the presence (d) of 550 nM rPBTx3. (G) Inhibition of Kv1.1 current, produced by 1 nM of rAgTx2.

different in the range of test potentials from )30 to +20 mV. Recombinant PBTx3 (500 nM) blocked the Kv1.2 and Kv1.3 peak currents by 54% and 53% at )30 mV (n ¼ 3), and by 53% and 54% (n ¼ 3) at 20 mV. In the presence of 70 lM rPBTx3, the Kv1.1 peak current was blocked by 45% (n ¼ 3) at )30 mV and by 42% at 20 mV (n ¼ 3). Blocking of the Kv1 channels by rPBTx3 is reversible and has no influence on the gating characteristics of the channels. Therefore, the time constants for relaxation to equilibrium block of the different Kv1 channels in the presence of the toxin reflect only the progress of the binding reaction. To determine the time constants son and soff for blockade and recovery, current traces were repeated every 2 s before, during and after rPBTx3 application. The timecourses of blockade and recovery were fitted to monoexponential curves, in agreement with the results obtained for other scorpion toxins [23]. In the presence of 10 lM rPBTx3 on Kv1.1 and 3.3 lM rPBTx3 on Kv1.2 and Kv1.3 channels, blockade occurred with a mean time constant son of 8.3 ms, 2.1 ms and 1.7 ms, respectively, for Kv1.1, Kv1.2 and Kv1.3 channels. The recovery from blockade

Ó FEBS 2002


occurred with a mean soff of 9.2 ms, 23.9 ms and 10.8 ms. Corresponding kon values were therefore 1.3 · 103 M)1Æs)1, 12.7 · 104 M)1Æs)1 and 1.4 · 105 M)1Æs)1 and koff values were 0.108 s)1, 0.041 Æs)1 and 0.092 Æs)1, respectively, for Kv1.1, Kv1.2 and Kv1.3. The Kd calculated from the ratio koff/kon was in all cases in good agreement with the value obtained in the dose–response experiments (see further): 80 lM for Kv1.1 (Kd ¼ 79 lM), 322 nM for Kv1.2 (Kd ¼ 547 nM) and 657 nM for Kv1.3 (Kd ¼ 492 nM). The fraction of unblocked current at equilibrium (fu) is readily measured and is related to the rate constants according to fu ¼ koff/(kon[rPBTx3] + koff). The Hill coefficients were not significantly different from 1. From the constructed current/voltage relationship (Itest/Vtest), it can be seen that 70 pM rAgTx2 produced a marked inhibition (45%) of the K+ current of Kv1.1 channels at all Vtest (Fig. 5, 1b) as measured at the end of each 100 ms test pulse. Recombinant PBTx3 produced almost the same effect by applying 550 nM toxin on Kv1.2 (Fig. 5, 1c) and 500 nM toxin on Kv1.3 (Fig. 5, 1d) channels, whereas the same degree of inhibition was observed with 70 lM toxin on Kv1.1 channels (Fig. 5, 1a). This was in close agreement with the inhibition seen with native PBTx3 (Fig. 4A). The reversal potential for Kv1.1 currents was evaluated from the kinetics of the tail currents upon repolarization. A tail current/voltage curve (Itail/Vtest) was constructed by fitting the data with a single Boltzmann distribution function of the form Itail ¼ Itail,max/{1 + exp[(V1/2–V)/s]} where Itail is the tail current, Imax is the maximal tail current, and s the slope factor of the voltage dependence. The peak amplitudes of the tails were measured at )50 mV and plotted as a function of the preceding Vtest (Fig. 5, 2a,b). This resulted in a typical fraction open channels/membrane voltage relationship. In the study with rAgTx2 (Fig. 5, 2a), the function in the control situation (n ¼ 4) was characterized by a half-maximal potential (V1/2) and slope (s) of )19.7 ± 0.7 mV and 10.0 ± 0.7 mV, respectively. With 10 pM rAgTx2 (n ¼ 4), V1/2 was )20.3 ± 1.3 mV and s was 10.5 ± 1.3 mV, demonstrating that there was no significant shift of V1/2 and of the s-value, showing no effect on the channel gating. For the control situation (n ¼ 4) in the experiment with rPBTx3 (Fig. 5, 2b), the V1/ 2 and s were )19.5 ± 2.7 mV and 10.9 ± 2.7 mV, respectively. In the presence of rPBTx3 (n ¼ 4), V1/2 and s were )19.9 ± 1.7 mV and 9.3 ± 1.6 mV, respectively, suggesting that this new toxin did not change the midpoint of the open channel/voltage curve of Kv1.1 channels. Steady-state Kv1.2 and Kv1.3 currents were converted to conductances using a reversal potential of )80 mV and fitted to single, first-order Boltzmann distributions. Conductances were normalized to the maximum estimated from the Boltzmann fit. In control, the function was characterized by a halfmaximal potential (V1/2) of )19.6 ± 3.3 mV and –22.0 ± 6.0 mV (n ¼ 4) with a slope factor of 7.5 ± 0.3 mV and 9.3 ± 4.7 mV, for Kv1.2 and Kv1.3 channels, respectively. With 500 nM rPBTx3, there was no shift: V1/2, )19.5 ± 1.7 mV and –21.6 ± 4.3 mV and s 9.6 ± 3.3 mV and 8.5 ± 6.3 mV (n ¼ 4) for Kv1.2 and Kv1.3 channels, respectively (Fig. 5, 2c,d). The induced inhibition by rPBTx3 was concentrationdependent. Fig. 6A and B show the dose–response curves of Kv1 channels to the recombinant toxins. The half-maximal effect on Kv1.2 and Kv1.3 channels was obtained with

Fig. 5. (1a–d) The current/voltage (Itest/Vtest) relationship in control (s) and in the presence (d) of different concentrations of rPBTx3 (a, c, d) on Kv1.1, Kv1.2 and Kv1.3, respectively, and rAgTx2 (B) on Kv1.1 channels expressed in Xenopus oocytes. Currents were measured at the end of each 500 ms test pulse. In all cases, the effect was reversible. (2a, b) Corresponding fraction open channels/membrane voltage curve (Itail/ Vtest) relationship, fitted with a Boltzmann function (n ¼ 4). (2a) In the absence of toxin (s), the midpoint (V1/2) and slope factor for Kv1.1 channels were )19.7 ± 0.7 mV and 10.0 ± 0.7 mV, respectively. In the presence of rAgTx2 (d), V1/2 and s were )20.3 ± 1.3 mV and 10.5 ± 1.3 mV (2b). In the control experiment (s) for rPBTx3 on Kv1.1 channels, V1/2 and s were )19.5 ± 2.7 and 10.9 ± 2.7, respectively, whereas after addition of rPBTx3 (d), they were )19.9 ± 1.7 mV and 9.3 ± 1.6 mV, respectively. The residual in maximal fraction open channels induced by application of 10 pM rAgTx2 was 75 ± 1.47% and by application of 70 lM rPBTx3 it was 53.6 ± 9.4%. (2c,d) Maximal membrane conductances (Gmax) were calculated. The steady-state activation curves for the control (s) and in the presence of 500 nM PBTx3 (d) were obtained after fitting with a Boltzmann function I ¼ Ic/[1 + exp(–Vtest–V1/2)/s])1. In both cases, for Kv1.2 and Kv1.3, V1/2 is not shifted by rPBTx3 as illustrated by the dashed lines. Slope values (s) for the control and the toxin curves are, respectively, 7.5 and 9.6 for Kv1.2, and 9.3 and 8.5 for Kv1.3 channels. In all cases, there was no significant shift of V1/2.

547 nM and 492 nM, respectively. However, the affinity of rPBTx3 for Kv1.1 channels was very low, with Kd ¼ 79 lM, showing that rAgTx2 (Kd ¼ 59 pM) has a 1 · 106 times higher affinity toward these channels. The

Ó FEBS 2002


Other channels Finally, we investigated the effect of our new toxin on different cloned channels, included in the screening process, in order to study its selectivity profile. Recombinant PBTx3 has no effect on Kir2.1 channels, hERG-type channels, hH1 Na+ channels (plant) KAT channels, cardiac two-pore background K+ channels (cTBAK) and the calcium channel p2X expressed in Xenopus oocytes (data not shown).

DISCUSSION

Fig. 6. Dose–response curves of rAgTx2 (A) and rPBTx3 (B) with a Kd value for rAgTx2 of 59 pM (Hill coefficient of 0.9). Each point represents the mean ± SD from four oocytes. The expected Kd values for rPBTx3 on Kv1.1, Kv1.2 and Kv1.3 are, respectively, 79 lM, 547 nM and 492 nM (Hill coefficients 0.89, 1.41 and 1.16, respectively). (C) Bimolecular kinetics of PBTx3 interaction. Rate constants of blocking [kon(rPBTx3), d] and dissociation (koff, s) were measured from voltage-clamp records as a function of external rPBTx3 concentration. Each point represents the mean ± SD of three individual determinations. (D) Effect of rPBTx3 on activation and inactivation kinetics of Kv1.3 channels. After depolarizing up to 0 mV from a Vhold of )90 mV for 500 ms, the activation and inactivation process in the presence of rPBTx3 is not changed. Both current traces, control and in the presence of toxin, have been superimposed after scaling of the trace in presence of rPBTx3.

obtained Kd of rAgTx2 for Kv1.1 was in accordance with the value reported by Garcia et al. [15]. Block was reversible upon washing-out. As the toxin binding was reversible and did not alter channel gating, we investigated rPBTx3 binding to the channel. As has been explained earlier, blockade is assumed to occur by a simple bimolecular reaction. If the toxin binding to the channel indeed reflects a bimolecular reaction scheme, the apparent first-order association rate increases linearly with toxin concentration and the first-order dissociation rate remains constant. This was indeed the case as shown in Fig. 6C, where the effects of increasing rPBTx3 concentrations on the kinetics of block on Kv1.2 are illustrated. The time course of activation was fitted using a Hodgkin–Huxley type model with a 4th power function of the form: It ¼ A {1–exp[+ (t/s)]4 + C}, with It the macroscopic and time-dependent current, A the current predicted at steady-state, s the time constant, and C a constant. For a depolarizing pulse from )90 to 0 mV, the activation kinetics of Kv1.3 could be fitted with a time constant of 11.2 ± 0.7 ms and 10.94 ± 1.1 ms in the control and in the presence of 500 nM rPBTx3, respectively (Fig. 6D). Recombinant rPBTx3 did not alter the activation or inactivation time constants of Kv1.3 channels expressed in oocytes.

The number of peptides isolated from distinct phyla, like scorpions [23], sea anemones [24,25], marine cone snails [26] and snakes has increased considerably. They have a threedimensional structure with some conserved motifs [27] but their affinity and specificity towards different targets may vary. Those targets include ion channels, present in different tissues. In order to increase our knowledge of the structure– function relationship between toxins and ion channels, it is necessary to isolate peptides in scorpion venoms and characterize them as much as possible. In this study, we present the purification, primary structure and functional characterization of PBTx3, a novel peptide inhibitor from the venom of the P. transvaalicus scorpion. PBTx3 was isolated from the venom on the basis of its ability to inhibit the K+ current through cloned voltage-dependent K+ channels (Kv1) expressed in Xenopus oocytes. Separation procedures leading to the identification of this novel neurotoxin were performed by gel filtration and reversedphase HPLC, by using different types of columns, as described previously [28]. The new toxin PBTx3 has a peptidic chain of 37 amino acids and shows similarities with members of the first subfamily of a-K+ scorpion toxins [8], with a fully conserved stretch of residues G25-K26-C27-M28-N29 residing in one of the b sheets, like ChTx. The sequence Lys-Cys-XXX-Lys-Cys (X being any amino acid), with the Lys–Cys in antiparallel b sheets and XXX being a tight turn, is also conserved, as in all other small scorpion toxins that are active on K+ channels. In order to find structurally significant features in the sequence of PBTx3 (Fig. 2B), sequence alignments were performed using the program CLUSTAL 1.8 (http://searchlauncher.bcm.tmc.edu:9331/multialign/multialign.html). PBTx3 shows similarities with ChTx (41%) [29], Lqh 15-1 (44%) [30] and ChTx-Lq-2 (38%) [11] from Leiurus quinquestriatus var. Hebraeus, BmTx 1 (55%) and 2 (41%) [31] from Buthus martensi Karsch, HgTx 2 (55%) [32] and LbTx (50%) [33] from Centruroides limbatus, IbTx (47%) [34] and TmTx (52%) [35] from Buthus tamulus. Alignment of the cysteine residues (C6–C27, C12–C32, C16–C34) showed that it was a novel toxin and that the cysteine motif was highly conserved. This cysteine pattern was also found in long-chain scorpion toxins [36] and other defence proteins such as the antibacterial insect defensin A [37], as well as in plant thionins [38] and potent antifungal plant defensins [39]. Disulfide bridges are important in stabilizing the three-dimensional structure of the toxin, as demonstrated by NMR studies of ChTx [40], iberiotoxin [41] and Lq2 [42]. Definitive assignment of the disulfide linkages in PBTx3 is currently unknown but is assumed to mimic that of ChTx and other a-KTx. Specific

Ó FEBS 2002


residues in ChTx, responsible for specific properties, are also present in PBTx3. For example K26 (using PBTx3 numbering), the crucial residue in the interaction with the pore of voltage-gated K+ channels [43], is located in the centre of the molecule. Furthermore G26 (corresponding to G25 in PBTx3) has been suggested to be important for appropriate formation of the disulfide pairing [44] and is also conserved throughout these sequences of all the members of subfamily 1 of a-KTx, including PBTx3. Because of these similarities and conservation of the consensus sequence, proposed for a-KTx subfamily 1, this new toxin is supposed to be the tenth member of the a-K toxin 1 subfamily. Although this novel toxin maintains a number of expected features, present also in ChTx and known to be important for the activity, it is unique in some aspects. In contrast with other members of the ChTxsubfamily, PBTx3 lacks F2 and W14. The latter plays a role in the interaction of the other members of this subfamily with residue G380 in the outer vestibule of the Kv1.3 channel [45]. The mutation W13L could well be responsible for the lower affinity of PBTx3 for Kv1 channels, as a similar decrease in affinity was demonstrated previously for the W14A mutant of ChTx [45]. However, those two residues are seen only in the toxins known to block BK channels (large-conductance Ca2+-activated K+ channels). PBTx3 conserves also a higher content of proline residues (two), but the importance of this is not really clear. PBTx3 possesses no N-terminal pyroglutamate, a residue classified as influential in the functional map of ChTx [46]. These structural differences in PBTx3 together with differences in the sequence at crucial or influential places (one N-terminal residue fewer, R22, P23, N24, R30, K33 and P36 versus the N-terminus, T23, S24, R25, K31, R34 and S37 in ChTx) may explain why the affinity of rPBTx3 is much lower for Kv1 channels. The toxin is composed of 37 amino-acid residues, with 11 positively charged groups and three negatively charged residues, dispersed all over the molecular surface. Groups of strong hydrophobicity (M4, M28, Y35) and H-binding capacity (S8, S9 N24 and N29) would suggest that the specific block of the toxin relies upon hydrophobic as well as polar interactions. The threedimensional structure of PBTx3 is also related to the a-KTx1 subfamily. Its three-dimensional conformation is determined by homology modelling (Fig. 7) with hongotoxin 2 (a-KTx 1.9) as a template for modelling because this latter toxin shares 55% homology with PBTx3. The a/b scaffold consists of a short a helix (residues S9–A19) and a b sheet, which is not triple- but doublestranded in PBTx3. Rather than forming a third b strand as found for other a-K toxins, the N terminal region of PBTx3, based in our model, adopts an extended conformation. This can be explained by the presence of the N-terminal end of PBTx3, which is one residue shorter than that of ChTx. It has been shown that toxins acting on SK channels mostly contain a two-stranded antiparallel b sheet (leiurotoxin I and PO5), whereas toxins active on Kv channels mostly have a triple-stranded b sheet. Whether PBTx3 blocks SK channels remains to be investigated. The key feature of ChTx block of the Kv1 channels, a 1 : 1 stochiometry for toxin block of the channel, is also observed with PBTx3. Although those two toxins could share a common mechanism for blocking, there are some quantitative differences in the blocking kinetics. For instance, the

Fig. 7. A three-dimensional model for PBTx3, constructed by homology modelling. The backbone of the molecule is shown in ribbon. Residues forming the functional diad (K26 and Y35) are in yellow.

on rate of rPBTx3 binding to Kv1 channels is 10–100 times slower than that of ChTx for which, depending on the conditions, channels are blocked with an on rate of 0.2– 20 · 107 M)1Æs)1. This is not very surprising as ChTx and PBTx3 share only 41% sequence homology. Only three positively charged residues are conserved between the two toxins, and two arginine residues and lysine residues are exchanged between the two toxins, located in the a-sheet. Of the three residues in ChTx (R25, K27 and R34) crucial to toxin binding and blockade [46], only the K27 is conserved. The R34 is mutated to a lysine residue. Because of the difference in the length of their side chains, lysine and arginine could have a different effect as also described for other toxins [47]. However, structural similarities in this part of the toxins may underlie the functional similarities observed for the toxins. ChTx is a highly basic toxin, with a net charge of +5 at neutral pH, whereas PBTx3 (still more basic), carries a net charge of +9 (pH range 5–9). For PBTx3, an additional negatively charged D3 is present, and could be an explanation for some of the differences in the association rate constants of the two toxins. Several binding sites of K+ channel blocking peptides have been characterized and most of these blockers possess at least a common diad composed of two functionally important residues, separated by 6.6 ± 1.0 A˚: a positively charged residue and a hydrophobic residue [48,49]. Residues in AgTx2 and ChTx at positions equivalent to Y36 and K27 of PBTx3 have been shown to be critical for channel blocking [50,51]. These two residues are also found in anemone K+ channel toxins, despite the fact that the threedimensional folding of scorpion and anemone toxins are quite different [48]. Regarding this hypothesis and in correspondence with the diad in ChTx, K26 and the Y35 in PBTx3 are most probably involved in this diad. The distance that separates the Ca of the lysine from the centre of the benzene ring of the tyrosine is 6.805 ± 0.406 A˚. We can imagine that the toxin interacts with the channel like a moon lander system and that those two residues play an important role in the interaction with the pore of the channel. Our control toxin in the recombinant expression, AgTx2,


represents a very potent blocker of Kv1 type channels. Mutagenesis studies on AgTx2 identified a set of residues as functionally important for blocking the Shaker K+ channels (N30, K27, R24, S11, F25, T36, M29 and less important R31) [52]. Three residues are mutated in PBTx3, namely R27P, F25N and T36Y (AgTx2 numbering). The T36Y mutation is unlikely to affect drastically the affinity toward Kv1.3 channels, as it also occurs in other members of the first group. The effect of the R27P and F25N mutations could be more important, considering the diad hypothesis. Most of the other mutations are located far from the interaction surface, upstream from the a helix or within this helix. The sequence of rPBTx3 includes some similarities with subfamily three, seven and eight of the a-KTx toxins. These toxins all end with a positively charged residue at the C-terminus, preceded by a proline. Functionally, the recombinant toxin, lacking the arginine, demonstrates the same properties as the native toxin (with an additional arginine), illustrating that this residue is not important for function. In the first b sheet, PBTx3 represents a fully conserved stretch referred to as the kaliotoxin group: C18-K19-A21-G22. Comparing the S5-P-S6 regions of the three channels, we can look for specific residues in the pore-forming region that are different between Kv1.1 channels and Kv1.2 or Kv1.3, that can possibly explain the selectivity toward these latter channels. The only residue, present in both Kv1.2 and Kv1.3 and mutated in Kv1.1 is D372 (Kv1.3 numbering). This residue is probably involved directly in the intimate interaction with the toxin right at the binding site. MacKinnon et al. (1989) observed a substantial reduction on the binding affinity when the structure of this site was altered by shortening the side chain (E–D) [13]. However, some studies have shown that the same mutations in highly homologous K+ channels can produce different effects. Therefore the extrapolation of the structural and functional importance of residues should be done with caution, even with ion channels belonging to the same family [53]. It is well known that long-chain scorpion neurotoxic polypeptides from the Buthidae family generally account for about 10–50% of the crude venom and that short-chain scorpion peptides appear only in very low quantities in the venom [54]. During the past decade, a number of approaches have been developed to produce toxins. For example PBTx3 is assessed to be only about 0.06% of the venom. Expression of scorpion toxins in Cos-7 cells [55], in insect cells by means of the Baculovirus system [56], in plants [57], in NIH/3T3 mouse cells [58] and in yeast [57] led to rather low yields. The first recombinant toxin was described about 10 years ago [59] and different toxins followed. We produced rPBTx3 in order to verify that this peptide was indeed the inhibitory component in the scorpion venom, excluding the possibility of the contamination with a peptide of higher affinity to K+ channels. The system chosen to express PBTx3 in E. coli had previously been shown to be suitable for the production of soluble, correctly folded spider [60] or scorpion toxins [61]. Following the procedures described in this study, it is feasible to produce 2–4 mg of homogeneous and biologically active toxin from 1 L E. coli culture. The production of fully active rAgTx2 and rPBTx3 requires some in vitro post-translational modifications that are difficult to control: proteolytic release of the toxin from the fusion protein and correct forming of the three disulfide bonds by the six cysteines. Based on the chromatographic

Ó FEBS 2002

profile of a mixture of rAgTx2 and native AgTx2 (AlomoneÒ), which resulted in a single elution peak without additional components, and based on the identical functional activity on Kv1, we can assume that the folding process in rAgTx2 was correctly performed. In the case of rPBTx3, the elution time of the native and the recombinant toxins were identical and the effect on Kv1 channels was also comparable. Therefore, we could also conclude that PBTx3 is not amidated, because peptides of this size with a free or with an amidated residue in the C-terminal position exhibit different retention times on HPLC [62]. The reduced peptide could be air oxidized in a concentration-independent manner. This was observed previously for other short scorpion toxins acting on Ca2+-activated K+ channels (e.g. leiurotoxin I and PO5) [63,64]. The lack of activity of the fusion proteins is not unexpected as the 44 kDa additional mass could affect significantly the folding and accessibility of the toxin portion. As mentioned before, just a few studies were performed based on the native venom of P. transvaalicus. Crude venom of P. transvaalicus has been shown to modulate the ChTx binding to aortic sarcolemmal vesicles, in a way that it was able to inhibit ChTx binding in the preparation [34]. Inhibitors from scorpions, snakes and bees appear to target primarily either the Shaker-related subfamily of Kv channels or the Ca2+-activated K+ channels [15,65,66]. In our study, we used a heterologous expression in oocytes of cloned Kv channel proteins. To determine which type of voltage-gated K+ channel could be sensitive to recombinant PBTx3, electrophysiological experiments were performed on Kv1.1, Kv1.2 and Kv1.3 channels expressed in Xenopus oocytes. Kv1.3 channels have been found in several types of cells, in neurons, and in T lymphocytes and have proven to be highly sensitive to scorpion toxins [67]. Analysis of the effects of rPBTx3 on Kv1 channels showed that rPBTx3 mimicked the effects of ChTx. ChTx blocks Kv1.2 and Kv1.3 with dissociation constants in the nanomolar range, but does not block Kv1.1, even at 1 lM [68]. In parallel, rPBTx3 blocks Kv1.2 and Kv1.3 channels, but with lower channel affinities than those of ChTx. The half-maximal blockage of Kv1.2 and Kv1.3 occurred at 547 nM and 592 nM, compared with 6 nM and 1 nM for ChTx [68]. Although there is a considerable amount of sequence identity between PBTx3 and other members of subfamily 1 of the a-KTx, the values for the association rates and dissociation rate constants differed from those determined previously for AgTx 2 and ChTx [46,69]. We examined the inhibitory effects of rPBTx3 at different membrane voltages. Block induced by rPBTx3 was voltage-independent over the range )30 to +20 mV, indicating that this toxin is not very sensitive to the gating state of the channel. Channel block by AgTx2 is performed by physical occlusion of the conduction pore [23]. The overall channel conductance, measured from the slope of the current–voltage relationship, is not changed in all cases in the presence of toxin. Fig. 5 shows activation curves obtained in the absence and presence of extracellular rPBTx3 on Kv1, channels. Recombinant PBTx3 does not shift the voltage at which the channels open. Also, as demonstrated for Kv1.3, there was no shift in the activation or inactivation kinetics of those three channels, as demonstrated for Kv1.3 (Fig. 6D). For Kv1.1, both the onset and recovery from inhibition were slow. Because the toxin does not alter channel Kv1.2 gating and the binding to this

Ó FEBS 2002


channel is reversible, the time constants for relaxation to equilibrium block upon toxin exposure reflect only the process of the binding reaction. Therefore, the kinetics of rPBTx3-induced inhibition were consistent with a bimolecular reaction between PBTx3 and Kv1.2. The forward rate constant for onset of inhibition varied linearly with PBTx3 concentration, while backward rate constant for recovery from inhibition was independent of PBTx3 concentration (Fig. 6C). The dissociation constants (koff) decreased from Kv1.1 to Kv1.2 and Kv1.3, in an order that correlates with the increase of the affinity of rPBTx3 for these channels. We screened a variety of other channels to investigate the selectivity of rPBTx3, but no modulation was observed. In the future, additional functional characterization of rPBTx3 on other types of channels is planned where, for example, Ca2+-activated K+ channels and other Kv1 channels (e.g. Kv1.6) are good candidates.

ACKNOWLEDGEMENTS We thank O. Pongs for providing the cDNA for the Kv1.2 channel and C. Ulens for the subcloning of the gene encoding the Kv1.2 channel. The Kv1.3 clone was kindly provided by M. L. Garcia. We are grateful to H. Sentenac to provide the KAT1 clone. The hK1 clone was kindly provided by R. G. Kallen. We also thank E. Toth Zsamboki for providing the P2X clone and Y. Kurachi for providing the TBAK clone. I. H. and E. C. are Research Assistants of the Flemish Fund for Scientific Research (F.W.O.-Vlaanderen). This work was supported by a bilateral collaboration between Flanders and South Africa (BIL00/36).

REFERENCES 1. Newlands, G. (1974) The venom-squirting ability of Parabuthus scorpions (arachnida: buthidae). S. Afr. J. Med. Sci. 39, 175–178. 2. Bergman, N.J. (1997) Clinical description of Parabuthus transvaalicus scorpionism in Zimbabwe. Toxicon 35, 759–771. 3. Muller, G.J. (1993) Scorpionism in South Africa. A report of 42 serious scorpion envenomations. S. Afr. Med. J. 83, 405–411. 4. Bergman, N.J. (1997) Scorpion sting in Zimbabwe. S. Afr. Med. J. 87, 163–167. 5. Van Aarde, M.N., Van Rooyen, J.M. & Van Aswegen, G. (1996) Screening of the venoms of two Parabuthus scorpion species on isolated cardiomyocytes. Onderstepoort J. Vet. Res. 63, 127–130. 6. Verdonck, F., Bosteels, S., Desmet, J., Moerman, L., Noppe, W., Willems, J., Tytgat, J. & van der Walt, J. (2000) A novel class of pore-forming peptides in the venom of Parabuthus schlechteri Purcell. Cimbebasia 16, 247–260. 7. Tytgat, J., Debont, T., Rostoll, K., Muller, G.J., Verdonck, F., Daenens, P., van der Walt, J.J. & Possani, L.D. (1998) Purification and partial characterization of a ÔshortÕ insectotoxin- like peptide from the venom of the scorpion Parabuthus schlechteri. FEBS Lett. 441, 387–391. 8. Tytgat, J., Chandy, K.G., Garcia, M.L., Gutman, G.A., MartinEauclaire, M.F., van der Walt, J.J. & Possani, L.D. (1999) A unified nomenclature for short-chain peptides isolated from scorpion venoms: alpha-KTx molecular subfamilies. Trends Pharmacol. Sci. 20, 444–447. 9. Garcia-Calvo, M., Knaus, H.G., McManus, O.B., Giangiacomo, K.M., Kaczorowski, G.J. & Garcia, M.L. (1994) Purification and reconstitution of the high-conductance, calcium-activated potassium channel from tracheal smooth muscle. J. Biol. Chem. 269, 676–682. 10. Wu, C.H. & Narahashi, T. (1988) Mechanism of action of novel marine neurotoxins on ion channels. Annu. Rev. Pharmacol. Toxicol. 28, 141–161.

11. Lucchesi, K., Ravindran, A., Young, H. & Moczydlowski, E. (1989) Analysis of the blocking activity of charybdotoxin homologs and iodinated derivatives against Ca2+-activated K+ channels. J. Membr. Biol. 109, 269–281. 12. Castle, N.A., Haylett, D.G. & Jenkinson, D.H. (1989) Toxins in the characterization of potassium channels. Trends Neurosci. 12, 59–65. 13. MacKinnon, R. & Miller, C. (1989) Mutant potassium channels with altered binding of charybdotoxin, a pore-blocking peptide inhibitor. Science 245, 1382–1385. 14. Ackerman, M.J. & Clapham, D.E. (1997) Mechanisms of disease: ion channels – basic science and clinical disease. New Engl. J. Med. 336, 1575–1586. 15. Missiakas, D. & Raina, S. (1997) Protein folding in the bacterial periplasm. J. Bacteriol. 179, 2465–2471. 16. Bessette, P.H., Cotto, J.J., Gilbert, H.F. & Georgiou, G. (1999) In vivo and in vitro function of the Escherichia coli periplasmic cysteine oxidoreductase DsbG. J. Biol. Chem. 274, 7784–7792. 17. Garcia, M.L., Garcia-Calvo, M., Hidalgo, P., Lee, A. & MacKinnon, R. (1994) Purification and characterization of three inhibitors of voltage- dependent K+ channels from Leiurus quinquestriatus var. hebraeus venom. Biochemistry 33, 6834–6839. 18. Baumann, A., Grupe, A., Ackermann, A. & Pongs, O. (1988) Structure of the voltage-dependent potassium channel is highly conserved from Drosophila to vertebrate central nervous systems. EMBO J. 7, 2457–2463. 19. Stuhmer, W., Stocker, M., Sakmann, B., Seeburg, P., Baumann, A., Grupe, A. & Pongs, O. (1988) Potassium channels expressed from rat brain cDNA have delayed rectifier properties. FEBS Lett. 242, 199–206. 20. Liman, E.R., Tytgat, J. & Hess, P. (1992) Subunit stoichiometry of a mammalian K+ channel determined by construction of multimeric cDNAs. Neuron 9, 861–871. 21. Swanson, R., Marshall, J., Smith, J.S., Williams, J.B., Boyle, M.B., Folander, K., Luneau, C.J., Antanavage, J., Oliva, C., Buhrow, S.A. et al. (1990) Cloning and expression of cDNA and genomic clones encoding three delayed rectifier potassium channels in rat brain. Neuron 4, 929–939. 22. Debont, T., Swerts, A., Van der Walt, J.J., Muller, G.J., Verdonck, F., Daenens, P. & Tytgat, J. (1998) Comparison and characterization of the venoms of three Parabuthus scorpion species occurring in southern Africa. Toxicon 36, 341–352. 23. Miller, C. (1995) The charybdotoxin family of K+ channelblocking peptides. Neuron 15, 5–10. 24. Aneiros, A., Garcia, I., Martinez, J.R., Harvey, A.L., Anderson, A.J., Marshall, D.L., Engstrom, A., Hellman, U. & Karlsson, E. (1993) A potassium channel toxin from the secretion of the sea anemone Bunodosoma granulifera. Isolation, amino acid sequence and biological activity. Biochim. Biophys. Acta. 1157, 86–92. 25. Castaneda, O., Sotolongo, V., Amor, A.M., Stocklin, R., Anderson, A.J., Harvey, A.L., Engstrom, A., Werstedt, C. & Karlsson, E. (1995) Characterization of a potassium channel toxin from the Caribbean Sea anemone Stichodactyla Helianthus. Toxicon 33, 603–613. 26. Terlau, H., Stocker, M., Shon, K.J., McIntosh, J.M. & Olivera, B.M. (1996) MicroO-conotoxin MrVIA inhibits mammalian sodium channels, but not through site I. J. Neurophysiol. 76, 1423–1429. 27. Kobayashi, Y., Takashima, H., Tamaoki, H., Kyogoku, Y., Lambert, P., Kuroda, H., Chino, N., Watanabe, T.X., Kimura, T., Sakakibara, S. et al. (1991) The cystine-stabilized alpha-helix: a common structural motif of ion-channel blocking neurotoxic peptides. Biopolymers 31, 1213–1220. 28. Debont, T., Daenens, P. & Tytgat, J. (1996) An improved fractionation and fast screening method for the identification of new and selective neurotoxins. Neurosci. Res. 24, 201–206. 29. Gimenez-Gallego, G., Navia, M.A., Reuben, J.P., Katz, G.M., Kaczorowski, G.J. & Garcia, M.L. (1988) Purification, sequence,


30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43. 44.

and model structure of charybdotoxin, a potent selective inhibitor of calcium-activated potassium channels. Proc. Natl Acad. Sci. USA 85, 3329–3333. Marshall, D.L., Vatanpour, H., Harvey, A.L., Boyot, P., Pinkasfeld, S., Doljansky, Y., Bouet, F. & Menez, A. (1994) Neuromuscular effects of some potassium channel blocking toxins from the venom of the scorpion Leiurus quinquestriatus hebreus. Toxicon 32, 1433–1443. Romi-Lebrun, R., Lebrun, B., Martin-Eauclaire, M.F., Ishiguro, M., Escoubas, P., Wu, F.Q., Hisada, M., Pongs, O. & Nakajima, T. (1997) Purification, characterization, and synthesis of three novel toxins from the Chinese scorpion Buthus martensi, which act on K+ channels. Biochemistry 36, 13473–13482. Koschak, A., Bugianesi, R.M., Mitterdorfer, J., Kaczorowski, G.J., Garcia, M.L. & Knaus, H.G. (1998) Subunit composition of brain voltage-gated potassium channels determined by hongotoxin-1, a novel peptide derived from Centruroides limbatus venom. J. Biol. Chem. 273, 2639–2644. Novick, J., Leonard, R.J., King, V.F., Schmalhofer, W., Kaczorowski, G.J. & Garcia, M.L. (1991) Purification and characterization of two novel peptidyl toxins directed against K+ channels from venom of new world scorpions. Biophys. J. 59, 78a. Galvez, A., Gimenez-Gallego, G., Reuben, J.P., Roy-Contancin, L., Feigenbaum, P., Kaczorowski, G.J. & Garcia, M.L. (1990) Purification and characterization of a unique, potent, peptidyl probe for the high conductance calcium-activated potassium channel from venom of the scorpion Buthus tamulus. J. Biol. Chem. 265, 11083–11090. Escoubas, P., Romi-Lebrun, R., Lebrun, B., Herrmann, R., Moskowitz, H., Rajendra, W., Hammock, B. & Nakajima, T. (1997) Tamulotoxin, a novel member of the potassium channel active short toxins from the venom of the Indian red scorpion Buthus tamulus. Toxicon 35, 806. Dauplais, M., Gilquin, B., Possani, L.D., Gurrola-Briones, G., Roumestand, C. & Menez, A. (1995) Determination of the three-dimensional solution structure of noxiustoxin: analysis of structural differences with related short-chain scorpion toxins. Biochemistry 34, 16563–16573. Cornet, B., Bonmatin, J.M., Hetru, C., Hoffmann, J.A., Ptak, M. & Vovelle, F. (1995) Refined three-dimensional solution structure of insect defensin A. Structure 3, 435–448. Bruix, M., Jimenez, M.A., Santoro, J., Gonzalez, C., Colilla, F.J., Mendez, E. & Rico, M. (1993) Solution structure of gamma 1-H and gamma 1-P thionins from barley and wheat endosperm determined by 1H-NMR: a structural motif common to toxic arthropod proteins. Biochemistry 32, 715–724. Terras, F.R., Torrekens, S., Van Leuven, F., Osborn, R.W., Vanderleyden, J., Cammue, B.P. & Broekaert, W.F. (1993) A new family of basic cysteine-rich plant antifungal proteins from Brassicaceae species. FEBS Lett. 316, 233–240. Bontems, F., Gilquin, B., Roumestand, C., Menez, A. & Toma, F. (1992) Analysis of side-chain organization on a refined model of charybdotoxin: structural and functional implications. Biochemistry 31, 7756–7764. Johnson, B.A. & Sugg, E.E. (1992) Determination of the threedimensional structure of iberiotoxin in solution by 1H nuclear magnetic resonance spectroscopy. Biochemistry 31, 8151–8159. Martins, J.C., Zhang, W.G., Tartar, A., Lazdunski, M. & Borremans, F.A. (1990) Solution conformation of leiurotoxin I (scyllatoxin) by 1H nuclear magnetic resonance. Resonance assignment and secondary structure. FEBS Lett. 260, 249–253. Park, C.S. & Miller, C. (1992) Interaction of charybdotoxin with permeant ions inside the pore of a K+ channel. Neuron 9, 307–313. Menez, A., Bontems, F., Roumestand, C., Gilquin, G. & Toma, F. (1992) Analysis of side-chain organization on a refined model of charybdotoxin: structural and functional implications. Proc. R. Soc. Edinburgh 99B, 83–103.

Ó FEBS 2002 45. Aiyar, J., Withka, J.M., Rizzi, J.P., Singleton, D.H., Andrews, G.C., Lin, W., Boyd, J., Hanson, D.C., Simon, M., Dethlefs, B. et al. (1995) Topology of the pore-region of a K+ channel revealed by the NMR-derived structures of scorpion toxins. Neuron 15, 1169–1181. 46. Goldstein, S.A., Pheasant, D.J. & Miller, C. (1994) The charybdotoxin receptor of a Shaker K+ channel: peptide and channel residues mediating molecular recognition. Neuron 12, 1377–1388. 47. Gairi, M., Romi, R., Fernandez, I., Rochat, H., Martin-Eauclaire, M.F., Van Rietschoten, J., Pons, M. & Giralt, E. (1997) 3D structure of kaliotoxin: is residue 34 a key for channel selectivity? J. Pept. Sci. 3, 314–319. 48. Dauplais, M., Lecoq, A., Song, J., Cotton, J., Jamin, N., Gilquin, B., Roumestand, C., Vita, C., de Medeiros, C.L., Rowan, E.G., Harvey, A.L. & Menez, A. (1997) On the convergent evolution of animal toxins. Conservation of a diad of functional residues in potassium channel-blocking toxins with unrelated structures. J. Biol. Chem. 272, 4302–4309. 49. Gasparini, S., Danse, J.M., Lecoq, A., Pinkasfeld, S., Zinn-Justin, S., Young, L.C., de Medeiros, C.C., Rowan, E.G., Harvey, A.L. & Menez, A. (1998) Delineation of the functional site of alphadendrotoxin. The functional topographies of dendrotoxins are different but share a conserved core with those of other Kv1 potassium channel-blocking toxins. J. Biol. Chem. 273, 25393–25403. 50. Hidalgo, P. & MacKinnon, R. (1995) Revealing the architecture of a K+ channel pore through mutant cycles with a peptide inhibitor. Science 268, 307–310. 51. Krezel, A.M., Kasibhatla, C., Hidalgo, P., MacKinnon, R. & Wagner, G. (1995) Solution structure of the potassium channel inhibitor agitoxin 2: caliper for probing channel geometry. Protein Sci. 4, 1478–1489. 52. Ranganathan, R., Lewis, J.H. & MacKinnon, R. (1996) Spatial localization of the K+ channel selectivity filter by mutant cyclebased structure analysis. Neuron 16, 131–139. 53. Tytgat, J. (1994) Mutations in the P-region of a mammalian potassium channel (RCK1): a comparison with the Shaker potassium channel. Biochem. Biophys. Res. Commun. 203, 513–518. 54. Gomez-Lagunas, F., Olamendi-Portugal, T., Zamudio, F.Z. & Possani, L.D. (1996) Two novel toxins from the venom of the scorpion Pandinus imperator show that the N-terminal amino acid sequence is important for their affinities towards Shaker B K+ channels. J. Membr. Biol. 152, 49–56. 55. Bougis, P.E., Rochat, H. & Smith, L.A. (1989) Precursors of Androctonus australis scorpion neurotoxins. Structures of precursors, processing outcomes, and expression of a functional recombinant toxin II. J. Biol. Chem. 264, 19259–19265. 56. Carbonell, L.F., Hodge, M.R., Tomalski, M.D. & Miller, L.K. (1988) Synthesis of a gene coding for an insect-specific scorpion neurotoxin and attempts to express it using baculovirus vectors. Gene 73, 409–418. 57. Pang, S.Z., Oberhaus, S.M., Rasmussen, J.L., Knipple, D.C., Bloomquist, J.R., Dean, D.H., Bowman, K.D. & Sanford, J.C. (1992) Expression of a gene encoding a scorpion insectotoxin peptide in yeast, bacteria and plants. Gene 116, 165–172. 58. Dee, A., Belagaje, R.M., Ward, K., Chio, E. & Lai, M.H. (1990) Expression and secretion of a functional scorpion insecticidal toxin in cultured mouse cells. Biotechnology 8, 339–342. 59. Ducancel, F., Boulain, J.C., Tremeau, O. & Menez, A. (1989) Direct expression in E. coli of a functionally active protein A-snake toxin fusion protein. Protein Eng. 3, 139–143. 60. Swartz, K.J. & MacKinnon, R. (1995) An inhibitor of the Kv2.1 potassium channel isolated from the venom of a Chilean tarantula. Neuron 15, 941–949.

Ó FEBS 2002


61. Legros, C., Feyfant, E., Sampieri, F., Rochat, H., Bougis, P.E. & Martin-Eauclaire, M.F. (1997) Influence of a NH2-terminal extension on the activity of KTX2, a K+ channel blocker purified from Androctonus australis scorpion venom. FEBS Lett. 417, 123–129. 62. Romi, R., Crest, M., Gola, M., Sampieri, F., Jacquet, G., Zerrouk, H., Mansuelle, P., Sorokine, O., Van Dorsselaer, A., & Rochat, H. et al. (1993) Synthesis and characterization of kaliotoxin. Is the 26–32 sequence essential for potassium channel recognition? J. Biol. Chem. 268, 26302–26309. 63. Sabatier, J.M., Zerrouk, H., Darbon, H., Mabrouk, K., Benslimane, A., Rochat, H., Martin-Eauclaire, M.F. & Van Rietschoten, J. (1993) P05, a new leiurotoxin I-like scorpion toxin: synthesis and structure–activity relationships of the alphaamidated analog, a ligand of Ca2+-activated K+ channels with increased affinity. Biochemistry 32, 2763–2770. 64. Sabatier, J.M., Fremont, V., Mabrouk, K., Crest, M., Darbon, H., Rochat, H., Van Rietschoten, J. & Martin-Eauclaire, M.F. (1994) Leiurotoxin I, a scorpion toxin specific for Ca2+-activated K+ channels. Structure-activity analysis using synthetic analogs. Int. J. Pept. Protein Res. 43, 486–495.

65. Blatz, A.L. & Magleby, K.L. (1986) Single apamin-blocked Ca2+-activated K+ channels of small conductance in cultured rat skeletal muscle. Nature 323, 718–720. 66. Grissmer, S., Nguyen, A.N., Aiyar, J., Hanson, D.C., Mather, R.J., Gutman, G.A., Karmilowicz, M.J., Auperin, D.D. & Chandy, K.G. (1994) Pharmacological characterization of five cloned voltage-gated K+ channels, types Kv1.1, 1.2, 1.3, 1.5, and 3.1, stably expressed in mammalian cell lines. Mol. Pharmacol. 45, 1227–1234. 67. Leonard, R.J., Garcia, M.L., Slaughter, R.S. & Reuben, J.P. (1992) Selective blockers of voltage-gated K+ channels depolarize human T lymphocytes: mechanism of the antiproliferative effect of charybdotoxin. Proc. Natl Acad. Sci. USA 89, 10094–10098. 68. Pongs, O. (1992) Molecular biology of voltage-dependent potassium channels. Physiol. Rev. 72, S69–S88. 69. Laraba-Djebari, F., Legros, C., Crest, M., Ceard, B., Romi, R., Mansuelle, P., Jacquet, G., van Rietschoten, J., Gola, M., & Rochat, H. et al. (1994) The kaliotoxin family enlarged. Purification, characterization, and precursor nucleotide sequence of KTX2 from Androctonus australis venom. J. Biol. Chem. 269, 32835–32843.