This work is dedicated to Professor José C. de Freitas, an outstanding scientist who inspired generations of biologists in Brazil
Sea anemone venoms have become a rich source of peptide toxins which are invaluable tools for studying the structure and functions of ion channels. In this work, BcsTx3, a toxin found in the venom of a Bunodosoma caissarum (population captured at the Saint Peter and Saint Paul Archipelago, Brazil) was purified and biochemically and pharmacologically characterized. The pharmacological effects were studied on 12 different subtypes of voltage-gated potassium channels (KV1.1–KV1.6; KV2.1; KV3.1; KV4.2; KV4.3; hERG and Shaker IR) and three cloned voltage-gated sodium channel isoforms (NaV1.2, NaV1.4 and BgNaV1.1) expressed in Xenopus laevis oocytes. BcsTx3 shows a high affinity for Drosophila Shaker IR channels over rKv1.2, hKv1.3 and rKv1.6, and is not active on NaV channels. Biochemical characterization reveals that BcsTx3 is a 50 amino acid peptide crosslinked by four disulfide bridges, and sequence comparison allowed BcsTx3 to be classified as a novel type of sea anemone toxin acting on KV channels. Moreover, putative toxins homologous to BcsTx3 from two additional actiniarian species suggest an ancient origin of this newly discovered toxin family.
Potassium channels are by far the largest and most diverse family of ion channels that through evolution have come to play essential roles in cellular signaling processes in both excitable and non-excitable cells. There are several types of K+ channels, including voltage-gated potassium channels (KV), that are involved in the control of cardiac and neuronal excitability, release of neurotransmitters, muscle contractility and hormone secretion . Throughout evolution venomous animals, such as snakes, cone snails, scorpions, spiders and sea anemones, have developed highly specialized toxin peptides that specifically target KV channels. These toxins have been employed as invaluable probes for exploring channel structures and functions, as well as for developing novel therapeutic approaches and for the design of powerful insecticides [2, 3].
Sea anemones are members of the phylum Cnidaria (class Anthozoa, order Actiniaria) that is typified by intracellular stinging organelles, known as nematocysts. They contain a variety of toxic proteins and peptides, including potent toxins affecting the transient receptor protein vanilloid (TRPV1), acid-sensing ion channels, voltage-gated Na+ (NaV) and K+ channels, as well as secondary metabolites . Recently, the ectodermal gland cells were also reported as a toxin production compartment in sea anemones . Characterized peptide toxins from sea anemones can affect ion channels either by blocking the current or modifying the gating mechanism. Peptide toxins that delay NaV channel inactivation during the depolarization process by binding to receptor site 3 are the best studied group, with already more than 100 known toxins . In contrast, no more than 23 potassium channel toxins have been structurally and functionally characterized. They can be divided into four types according to their differences in amino acid sequences, disulfide bridge patterns and activity profiles . Type 1 toxins have 35–38 amino acid residues and three disulfide bridges. These toxins block currents through channels of Shaker (KV1.) and Shaw (KV3.) subfamilies and also intermediate conductance calcium-activated K+ channels . Potassium channel toxins of this type have been characterized from the venom of sea anemones of different families (Actiniidae, Hormathiidae, Stichodactylidae and Thalassianthidae) and can be subdivided into subtype 1a and subtype 1b, having four and eight amino acid residues respectively between the second and third cysteine residues [8, 9]. Kalicludines peptides (AsKC 1–3) from Anemonia viridis and APEKTx1 from Anthopleura elegantissima belong to type 2, with 58–65 amino acid residues and three disulfide bridges. These toxins share a structural homology with the basic pancreatic trypsin inhibitor (BPTI), a very potent Kunitz-type protease inhibitor, and dendrotoxins (from mamba snakes) which are highly potent blockers of KV channels [10, 11]. The third type is composed of toxins containing 41 or 42 amino acid residues and three disulfide bridges: BDS I and II toxins (A. viridis) inhibit KV3.1, KV3.2 and KV3.4 currents by acting as channel-gating modulators and APETx1 (A. elegantissima) is a gating modifier toxin of KV10.1 and KV11.1 [12, 13]. The type 4 family is represented by two putative homologous KV channel blocking peptides, SHTX I and II, from Stichodactyla haddoni .
In this work we present the isolation, electrophysiological characterization, as well as biochemical and evolutionary analysis of the first representative member of a novel type of KV toxin from the venom of the sea anemone B. caissarum (Corrêa, 1964)  population from Saint Peter and Saint Paul Archipelago, Brazil.
As previously described, the fractionation of B. caissarum venom was achieved by gel filtration chromatography and the components were collected in five main fractions . Fraction III, also named ‘neurotoxic fraction’ because of its high neurotoxicity (strong contractile paralysis) when tested on swimming crabs (Callinectes danae), was further subjected to reverse-phase HPLC. The first purification step was performed using a linear gradient from 10% to 50% solution B for 35 min at a flow rate of 2.5 mL·min−1 (Fig. 1A), and the second step was performed using an isocratic condition of 16% of solution B (Fig. 1B). After application of the neurotoxic fraction under these conditions, the homogeneity of BcsTx3 toxin was confirmed by MALDI-TOF measurements, which generated an m/z of 5710.970 Da. This experimental mass corresponds well with the theoretically expected molecular weight of 5710.52 Da (Fig. 1C).
Native BcsTx3 was subjected to Edman degradation. Since only the first 47 N-terminal amino acids could be reliably sequenced, the toxin was subjected to de novo sequencing by mass spectrometry using the bottom-up strategy [17, 18]. In order to predict the theoretical mass of the digested fragments, the sequence obtained by Edman degradation was used as a template. The spectrum containing BcsTx3 peptide fragments is illustrated in Fig. 2A. The digestion of BcsTx3 generated peptides that were detected as single charged ions at m/z 717.311, 1347.603, 1538.657 and 1880.637 Da. This spectrum not only allows the results obtained by Edman to be confirmed but also highlights the unknown C-terminal (fragment ion of m/z 1347.603). The fragmentation spectrum of this C-terminal peptide contains many informative ions that allow the determination of the full sequence of the peptide (Fig. 2B). However, an ambiguity must be noticed in position 50 of the sequence, due to the very close masses of lysine (K, 146.106 Da) and glutamine and amidated glutamic acid (Q/E*, 146.069 Da). In order to unravel this ambiguity, a comparison of the theoretical and experimental y and b ions  of each possibility was carried out (Table S1). The error in parts per million corresponds to the mass difference between the theoretical and experimental ions expressed in parts per million. Considering that b-type ions do not contain the C-terminus extremity (i.e. the ambiguous site), they cannot be used to solve the ambiguity. In contrast, all the y ion types contain the C-terminus extremity and allow the sequence to be completed. The average error in parts per million for the y ions that include a K is 86.09 p.p.m. compared with −1.41 p.p.m. when considering a Q or an E* (Table S1). This result eliminates the possibility for the peptide to have a K at the C-terminal extremity. In fact, the error on the mass accuracy for y1 is 251.5 p.p.m. when a K is simulated at the C-terminal (clearly out of the spectrometer capabilities) and 2.7 p.p.m. for a Q or a E*, which is much more relevant. However, this approach cannot solve the ambiguity between a glutamine and an amidated glutamic acid (Fig. 2B). The protein sequence data reported herein will appear in the UniProt Knowledgebase (UniProtKB) under accession number C0HJC4.
BcsTx3 activity was investigated on 12 cloned voltage-gated potassium channels (KV1.1–KV1.6, Shaker IR, KV2.1, KV3.1, KV4.2, KV4.3 and hERG) and three voltage-gated sodium channels (NaV1.2, NaV1.4 and BgNaV1.1). BcsTx3 (3 μm) blocks KV channels rKV1.1 (54% ± 2.75%), rKV1.2 (90% ± 2.45%), hKV1.3 (81.5% ± 2.10%), rKV1.6 (64.17% ± 2.97%) and Shaker IR (97% ± 2.44%), while the same concentration has no effect upon the other isoforms tested (Figs 3 and S1). Concentration–response curves were constructed in order to determine the concentrations at which half of the channels were blocked. The IC50 values yielded 94.25 ± 16.31 nm for Shaker IR, 172.59 ± 51.09 nm for rKv1.2, 1006.48 ± 22.51 nm for hKv1.3 and 2245.93 ± 18.97 nm for rKv1.6 (Fig. 4A). Since BcsTx3 had the highest affinity for Drosophila Shaker IR, this channel isoform was used to further characterize the effect of the toxin. To investigate whether the observed current inhibition is due to pore blockage or to alteration of channel gating upon toxin binding, current–voltage (I–V) curves in ND96 solution (n =5) and HK-ND96 solution (n =4) were constructed. Application of BcsTx3 (100 nm) caused 51.74% ± 1.79% and 53.81% ± 4.26% inhibition of the potassium current in ND96 and HK-ND96, respectively (Fig. 4B,C). In ND96, the I–V curve in the control experiments and in the presence of toxin was characterized by a voltage for half-maximal activation (V1/2) of 1.79 ± 1.53 mV and 6.27 ± 2.44 mV, respectively. This shows that the inhibition of Shaker IR channels in the presence of BcsTx3 is not associated with a change in the shape of the I versus V relationship. The toxin induced inhibition of Shaker IR channels was not voltage dependent as in a range of test potentials from −30 mV to +50 mV no difference in the degree of block could be observed (n =5) (Fig. 4D), and binding was reversible since the current was completely recovered upon washout (Fig. 4E). Competitive binding experiments using tetraethylammonium (TEA) were performed in order to test whether BcsTx3 is an external pore blocker. First, 5.75 mm (IC50) TEA was applied to Shaker IR and resulted in a reduction of the K+ current to 49% ± 1.53% (Fig. 4F). Next, a mixture of TEA (5.75 mm) and BcsTx3 (100 nm) was applied. The observed additional block of only 12% ± 0.99% indicates a competition of the two ligands (Fig. 4F). All these findings together suggest that BcsTx3 inhibits the K+ current through a pore blocking mechanism.
Sequence analysis of sea anemone KV channel toxins
The BcsTx3 amino acid sequence was used to carry out a similarities search using blast. Only one hit, a putative protein (GenBank EDO47375.1) from Nematostella vectensis (NvePTx1), was returned. In addition we found a transcript sequence from the anemone species Metridium senile that encodes another putative homolog, referred to here as MsePTx1 (GenBank FC839755.1). Since no similarity was found with any other protein sequence available in public databases and considering the biochemical properties of BcsTx3 as well as its selective activity toward KV channels, we propose to include these three peptides in a novel family of potassium channel blockers from sea anemones named sea anemone type 5 KV toxins (Fig. 5).
Genomic organization of NvePTx1
Transcript and genomic organization were determined for the novel identified putative Nematostella BcsTx3 homolog (Fig. 6). The NvePTx1 sequence seems to encode for a signal peptide and a propart at the N-terminus of the polypeptide product. The propart end includes a highly conserved Lys-Arg tandem, suspected to be the cleavage site of a sea anemone specific protease for toxin maturation . The mature portion is a 50 amino acid peptide, having eight cysteine residues which are probably crosslinked to each other (Fig. 6A). NvePTx1 gene consists of two exons of 102 and 400 bp disrupted by one intron of 387 bp. The intron sequence covers the 5′ untranslated region of the transcript and does not disrupt the signal peptide or the sequence coding the propart and mature toxin, which are encoded at the 5′ region of the second exon (Fig. 6B).
Localization of NvePTx1 expression
As a rising cnidarian model in the field of evolutionary developmental biology, N. vectensis provides relatively well established molecular tools such as in situ hybridization (ISH), which are not available for other sea anemone species such as B. caissarum . We took advantage of this fact in order to study the expression pattern of the NvePTx1 transcript. By ISH we localized the transcript in late gastrulae (30 h post fertilization) to ectodermal cells spanning most of the outer wall aside of the oral and aboral regions (Fig. 7A). In early planulae (48 h post fertilization) the pattern remained similar, with the staining noticeable in long and thin ectodermal cells, possibly nematocytes (Fig. 7B). In the late planula and primary polyp stages (96 and 144 h post fertilization, respectively) the stained cell population became smaller and more dispersed throughout the animals (Fig. 7C,D) but the cells had a similar morphology to those stained in the early planula.
A large number of voltage-gated potassium channel neurotoxins have been purified from the venom of worms, cone snails, scorpions, spiders and snakes . To date, only around 23 sea anemone KV toxins have been isolated and functionally characterized. BcsTx3 was purified from the venom of the B. caissarum population from Saint Peter and Saint Paul Archipelago by gel filtration and reverse-phase chromatography (Fig. 1), and its structure was elucidated by a combination of protein sequencing methods. BcsTx3 is a single-chain peptide containing 50 amino acids crosslinked by four disulfide bridges. However, an ambiguity between a glutamine and an amidated glutamic acid in position 50 of the sequence could not be experimentally solved (Fig. 2 and Table S1). By comparing BcsTx3 amino acid sequence with the putative homologs from Nematostella and Metridium that have a Q at the same position (Fig. 5), we suggest that BcsTx3 also ends with glutamine. Characterized sea anemone KV toxins contain from 28 to 63 amino acid residues and are mostly crosslinked by three disulfide bridges, except for SHTX I and II that have only two. These peptide toxins can be classified into four different types (types 1–4), according to their differences in amino acid sequences, disulfide bridge patterns and activity profiles . Our results clearly suggest that this classification needs to be extended to a new type 5, composed of BcsTx3 and the putative proteins from N. vectensis and M. senile, whose mature portions share 65.3% and 63.3% identity to BcsTx3 and the positions of their cysteine residues are identical (Fig. 5). The BcsTx3 disulfide bridge pattern has not been determined yet; however, its primary structure and pharmacological properties shown herein clearly support our proposed classification. Moreover, this observation strongly suggests that peptides capable of blocking or modulating KV channels evolved at least five times independently in sea anemones, a remarkable finding that highlights the strong selection that shaped different protein families again and again to fulfill similar toxic functions in different ways.
The pharmacological characterization showed that BcsTx3 is active against members of the KV1.x potassium channel subfamily (i.e. Shaker IR, KV1.2, KV1.3, Kv1.6 and KV1.1) when tested on a wide range of 15 different ion channels, including voltage-gated sodium channels. This broad screening not only reveals a potent activity against Shaker IR, but also shows an impressive selectivity for these channels over the other isoforms tested (Figs 3 and S1). This pharmacological profile might explain better the coexistence of BcsTx3 with BcsTx1 and BcsTx2, two type 1 toxins we reported previously, as their pharmacological preferences among KV channels are complementary . This complementarity might enable the animal to produce venom that is highly potent and can affect a wide array of KV channels. Moreover, BcsTx3 does not modulate the voltage dependence of gating of Shaker IR channels (Fig. 4B), and its blockage activity is not voltage dependent (Fig. 4D). Shaker IR blockage is independent of the direction of the potassium current flux and is not influenced by high extracellular concentration of K+ ions (Fig. 4C). Since the inhibition of current was reversible upon washout, it can be assumed that the site of toxin binding is located at the extracellular side (Fig. 4E). Thus, a competitive interaction between TEA (applied externally) and BcsTx3 was tested based on the rationale that, if both compounds act on the same site or by a similar mechanism, their co-application will not induce a supplementary block of the remaining current. However, if they possess different, non-overlapping binding sites, a potentiation of the blockage effect will be observed. Interestingly, an additional blockage effect was observed after co-application of TEA and BcsTx3 suggesting that their binding sites are partially overlapping (Fig. 4F). TEA inhibits KV channel function by binding within the ion conduction pathway and obstructing potassium flow. From mutagenesis studies, it has been shown that external TEA blockage is dependent upon the presence of amino acids with an aromatic side chain located near the extracellular entrance to the pore . TEA ethyl groups interact simultaneously with the tyrosine side chains of the selectivity filter on all four subunits and make favorable electrostatic interactions and van der Waals contacts with the edge of the aromatic ring . Similar to TEA, pore blocker toxins possess a ‘functional dyad’ composed of a lysine and tyrosine residue (6–7 Å distant to each other) that contributes to the blockage of the ion flux . The lysine residue positions its side chain into the channel pore, in the center of a ring composed of four aspartic acid residues of the selectivity filter, and the tyrosine residue interacts through both electrostatic forces and hydrogen bonding with a cluster of tyrosine residues of the selectivity filter, forming a physical barrier . Although our results do not confirm the existence of a functional dyad in BcsTx3, its primary structure contains two putative dyads (comprising Arg5-Tyr6 and Arg39-Tyr40) (Fig. 5). Thus, our electrophysiological analysis taken together with this evidence allow us to conclude that BcsTx3, just like TEA, inhibits Shaker IR currents by binding to the outer mouth of the pore and occluding the movement of K+ ions through it.
Nevertheless, it is worth noting that the observed potentiation of the blockage effect might be due to a multipoint interaction between BcsTx3 and the Shaker IR channel. It has been demonstrated that besides the functional dyad other determinants are involved in the interaction between pore blocker toxins and KV channels . For instance, despite the existence of a dyad (Lys22-Tyr23) in ShK, other residues (Ile7, Lys9, Arg11, Ser20 and Phe27) that are clustered around it interact with residues located in the vicinity of the selectivity filter and turret region of the KV channels [28, 29]. The amino acids Arg3, Trp5, Phe6, Lys7, His13, Ser23, Lys25, Tyr26 and Arg27 play a similar binding role in BgK [25, 30, 31]. Another example is Pi1, a scorpion toxin that has been purified from the venom of Pandinus imperator . Docking and electrophysiological experiments with Pi1 analogs have demonstrated that a ring of basic residues (Arg5, Arg12, Arg28 and Lys31) interacts with acidic residues of the KV channel turrets located in the most extracellular part of the channel outer vestibule. Moreover, three-dimensional models of toxin–channel interactions generated with scorpion toxins Pi4 (from Pandinus imperator), cobatoxin 1 (Centruroides noxius), iberiotoxin (Mesobuthus tamulus), maurotoxin (Scorpio maurus palmatus) and Lq2, agitoxin2 and charybdotoxin from Leiurus quinquestriatus hebraeus have all suggested that other amino acid residues surrounding the functional dyad make important contacts with specific residues at the turret regions of the KcsA (from Streptomyces lividans), KV1.x and calcium-activated potassium (KCa1.1 and KCa3.1) channels [33-38]. Therefore these findings support our hypothesis that BcsTx3 binds to Shaker IR channels through a multipoint interaction, and thus other amino acids than those of the selective filter might be involved in its binding and contribute to the observed potentiated blockage.
Different from many other venomous animals such as cone snails, scorpions and snakes, which have a broad characterization of their venoms based on genomic and transcriptomic approaches, knowledge on sea anemone venoms is scarce. However, the sequencing of the entire genome of N. vectensis has provided a unique opportunity to analyze evolutionary traces of the neurotoxin gene family [39, 40]. NvePTx1 seems to encode for a signal peptide and a propart at the N-terminus of the polypeptide product. The propart is probably composed of 12 amino acids characterized by ending with a pair of basic residues (Lys and Arg), a cleavage site for subtilisin-like proteases (Fig. 6). Similar structural features are observed in the precursors of many sea anemone neurotoxins [8, 9, 20], including MsePTx1 (Fig. S2). The localization of NvePTx1 in early life stages of N. vectensis to ectodermal cells that resemble nematocysts in their shape and size supports the notion that NvePTx1 encodes a neurotoxin and it may serve for defense purposes (Fig. 7). The shift in expression to more oral domains such as the tentacle buds in later stages suggests it might also have a later role in predation (Fig. 7D).
Considering the fact that no homologous sequences were found in non-actiniarian species and the phylogenetic distribution of the three sea anemone species where BcsTx3 and its homologues are found , it is unlikely that multiple evolutionary recruitment events occurred during the evolution of each species [42-44]. Thus, we suggest that the recruitment of the type 5 KV toxin family was an ancient event and that at least part of its adaptive evolution took place in the common ancestor of all three species, and possibly even in the common ancestor of all extant sea anemones . Moreover, as the split time between Metridium and Nematostella is estimated to be 250–450 million years ago  we propose that this is the minimal age of the toxin family we describe here for the first time.
In this study, we have demonstrated the biochemical and pharmacological characterization of a novel type of sea anemone KV toxin. BcsTx3 is a peptidyl toxin with a unique sequence that acts potently upon Shaker IR channels through a pore blocking mechanism. Moreover, the distribution of this newly discovered type of KV toxin among different actiniarian species suggests an ancient origin.
Materials and methods
Venom collection and neurotoxin purification
Specimens of the sea anemone B. caissarum were collected at Saint Peter and Saint Paul Archipelago (N00°55′, W29°20′) in Brazil. The sea anemones were kept in an aquarium for 24 h. The venom was obtained by electrical stimulation and fractionated as previously described . The fraction containing the neurotoxic peptides was submitted to reverse-phase chromatography on HPLC ÄKTA Purifier system (GE Healthcare, Uppsala, Sweden), using a semi-preparative CAPCELL PAK C-18 column (1 × 25 cm; Shiseido Corp., Kyoto, Japan). Protein elution was achieved using a linear gradient from 10% to 50% of acetonitrile containing 0.1% trifluoroacetic acid (TFA) (solution B) at a flow rate of 2.5 mL·min−1. BcsTx3 elution was performed using an isocratic condition of 16% of solution B at a flow rate of 1 mL·min−1 on the same column. BcsTx3 protein content was estimated by the bicinchoninic assay (Pierce, Rockford, IL, USA) and its homogeneity was verified by mass spectrometry analysis performed on an Ultraflex II TOF/TOF MALDI controlled by the flexcontrol 3.0 software (Bruker Daltonics, Bremen, Germany), under linear mode. External calibration was performed using Peptide Calibration Mix 3 [PepMix3; LaserBio Labs (Sophia-Antipolis, Valbonne, France)]. The dried droplet method was used to spot the toxin: 1 μL of BcsTx3 was spotted to a MALDI plate (384 positions; Bruker Daltonics) and mixed with 1 μL of 2,5-dihydroxybenzoic acid (20 mg·mL−1 in 50/50 0.2% acetonitrile/TFA, v/v) and α-cyano-4-hydroxycinnamic acid (20 mg·mL−1 in 50/50 0.2% acetonitrile/TFA, v/v).
Amino acid sequence determination
A native BcsTx3 sample (200 pmol) was partially sequenced by Edman degradation using an automated PPSQ-33A protein sequencer (Shimadzu, Kyoto, Japan) coupled to reverse-phase separation of PTH-amino acids on a WAKOSIL-PTH (4.6 × 250 mm) column (Wako, Osaka, Japan), according to the manufacturer's instructions. Thereafter, the toxin was subjected to mass spectrometry sequencing to verify and complete the sequence. Before mass spectrometry analysis, the BcsTx3 sample was reduced using 150 mm of Tris/(2-carboxyethyl)phosphine at 65 °C for 1 h and digested with endopeptidase LysC (Sigma-Aldrich, St. Louis, MO, USA) at 37 °C for 2 h, in an enzyme : toxin ratio of 1 : 20 in Tris/HCl (150 mm, pH 8.5). Mass spectra were acquired under linear and reflectron modes, using a Bruker Ultraflex II MALDI-TOF/TOF controlled by flexcontrol 3.0 software (Bruker Daltonics). External calibration was performed using Peptide Standard Calibration II (Bruker Daltonics) and Peptide Calibration Mix 3 (PepMix3; New England Biolabs, Ipswich, MA, USA) for reflection and linear modes, respectively. BcsTx3 was spotted as described before. MS/MS analyses were carried out by employing the LIFT™ cell for post-acceleration of the metastable fragments . Spectra were analyzed using flexanalysis 3.0 and biotools 3.1 (Bruker Daltonics Software). Additional MS/MS experiments were also performed on a Waters Synapt-G2 HDMS (Waters Corp., Manchester, UK) mass spectrometer equipped with a nano-electrospray source. The capillary and the sampling cone voltages were set at 1.2 kV and 25 V, respectively. The MS/MS spectrum was analyzed by masslynx 4.1 software (Waters Corp.) and interpreted manually.
Expression of voltage-gated ion channels in Xenopus laevis oocytes
The pharmacological effect of BcsTx3 was analyzed by heterologous expression of 12 voltage-gated potassium channels isoforms (rKV1.1, Genbank NM173095; rKV1.2, Genbank NM012970; hKV1.3, Genbank L23499; rKV1.4, Genbank NM012971; rKV1.5, Genbank NM012972; rKV1.6, Genbank NM575671; Shaker IR from Drosophila melanogaster, Genbank CG12348; rKV2.1, Genbank NM013186; hKV3.1, Genbank NM004976; rKV4.2, Genbank NM031730; rKV4.3, Genbank NM031739; hERG, Genbank NM000238) and three voltage-gated sodium channel subtypes (rNaV1.2, Genbank NM012647; rNaV1.4, Genbank M26643; BgNaV1.1 from Blattella germanica) (r, rat; h, human) in oocytes from Xenopus laevis. The linearized plasmids were transcribed using a T7 or SP6 mMESSAGE-mMACHINE transcription kit (Life Technologies, Austin, TX, USA). Oocytes were injected with 50 nL of cRNA at a concentration of 1 ng·nL−1 using a micro-injector (Drummond Scientific, Broomall, PA, USA). The oocytes were maintained in an ND96 solution (in mm: 96 NaCl, 2 KCl, 1.8 CaCl2, 2 MgCl2 and 5 HEPES; pH 7.4), supplemented with 50 μg·mL−1 gentamicin sulfate.
Two-electrode voltage-clamp recordings were performed at room temperature (18–22 °C) using a Geneclamp 500 amplifier (Molecular Devices, Sunnyvale, CA, USA) controlled by a pClamp data acquisition system (Axon Instruments, Union City, CA, USA). Whole cell currents from oocytes were recorded 1–3 days after injection. The bath solution composition was ND96 or HK-ND96 (in mm: 2 NaCl, 96 KCl, 1.8 CaCl2, 2 MgCl2 and 5 HEPES; pH 7.4). Voltage and current electrodes were filled with KCl (3 m). The resistance of both electrodes was kept between 0.8 and 1.0 MΩ. The elicited currents were filtered at 0.5 or 2 kHz using a four-pole low-pass Bessel filter. Leak subtraction was performed using a −P/4 protocol. KV1.1−KV1.6 and Shaker IR currents were evoked by 500 ms depolarization to 0 mV followed by a 500 ms pulse to −50 mV, from a holding potential of −90 mV. KV2.1, KV3.1, KV4.2 and KV4.3 currents were elicited by 500 ms pulses to +20 mV from a holding potential of −90 mV. Current traces of hERG channels were elicited by applying a + 40 mV pulse for 2.5 s followed by a step to −120 mV for 2.5 s. Sodium current traces were evoked by 100 ms depolarization steps from a holding potential of −90 mV to 0 mV. To assess the BcsTx3 concentration–response relationships, data were fitted with the Hill equation y =100/[1 + (IC50/[toxin])h], where y is the amplitude of the toxin-induced effect, IC50 is the toxin concentration at half-maximal efficacy, [toxin] is the toxin concentration and h is the Hill coefficient. In order to investigate the current–voltage (I–V) relationship, current traces were evoked by 10 mV depolarization steps from a holding potential of −90 mV. The values of IK were plotted as function of voltage and fitted using the Boltzmann equation IK/Imax = [1 + exp(Vg− V)/k]−1, where Imax represents maximal IK, Vg is the voltage corresponding to half-maximal current and k is the slope factor. To assess the concentration dependence of the BcsTx3 induced inhibitory effects, a concentration–response curve was constructed in which the percentage of current inhibition was plotted as a function of toxin concentration. Data were fitted with the Hill equation. Competitive binding experiments were performed using TEA obtained from Sigma-Aldrich. All data represent at least three independent experiments (n ≥3) and are presented as mean ± standard error. Comparison of two sample means was made using a paired Student's t test (P < 0.05). All data were analyzed using clampfit 10.3 (Molecular Devices) and origin 7.5 software (Origin Lab., Northampton, MA, USA).
Comparative sequence alignment
A multiple sequence alignment of sea anemone voltage-gated potassium channel toxins was done using clustalw2 software . The analyzed sequences were BcsTx1 (Swiss-Prot C0HJC2), BcsTx2 (Swiss-Prot C0HJC3) and BcsTx3 from the venom of the B. caissarum population from Saint Peter and Saint Paul Archipelago; Aek (Swiss-Prot P81897) from Actinia equina; AETX-K (Swiss-Prot Q0EAE5) from Anemonia erythraea; AsKs (Swiss-Prot Q9TWG1) from A. viridis; Bgk (Swiss-Prot P29186) from Bunodosoma granulifera; HmK (Swiss-Prot O16846) from Radianthus magnifica; κ1.3-SHTX-Sha1a (Genbank AB595205) from S. haddoni; κ1.3-TLTX-Ca1a (Genbank AB595207) (Cryptodendrum adhaesivum), κ1.3-TLTX-Hh1a (Genbank AB5952-08) (Heterodactyla hemprichi), κ1.3-SHTX-Sg1a (Genbank AB595204) (S. gigantea), κ1.3-SHTX-Sm1a (Genbank AB5-95206) (Stichodactyla mertensii) and κ1.3-TLTX-Ta1a (Genbank AB595209) (Thalassianthus aster); ShK (Swiss-Prot P29187) from Stichodactylahelianthus; BDS-1 (Swiss-Prot P11494), BDS-2 (Swiss-Prot P59084), AsKs (Swiss-Prot Q9TWG1), AsKC1 (Swiss-Prot Q9TWG0), AsKC2 (Swiss-Prot Q9TWF9) and AsKC3 (Swiss-Prot Q9TWF8) from A. viridis; APETx1 (Swiss-Prot P61541) and APEKTx1 (Swiss-Prot P86862) from A. elegantissima; SHTX-I/II (Swiss-Prot P0C7W7) from S. haddoni and the mature portions of the putative proteins (GenBank EDO47375.1) from N. vectensis, named Nve putative toxin 1 (NvePTx1), and from M. senile (MsePTx1) (GenBank FC839755.1).
Nucleotide and peptide similarity search
tblastn searches  using BcsTx3 amino acid sequence as the query sequence were performed against the nucleotide collection (nr/nt) and expressed sequence tags data sets of GenBank. Genomic analyses were performed via the Joint Genome Institute website (http://genome.jgi-psf.org/Nemve1/Nemve1.home.html) . Further transcriptomic and genomic analyses for a better annotated gene model based on high throughput sequencing of multiple Nematostella developmental stages was performed on a locally running version of the UCSC genome browser  (D. Fredman and U. Technau, University of Vienna, personal communication). Signal peptide and propeptide predictions were made respectively with signalp 4.0 and prop 1.0 servers.
Nematostella handling and in situ hybridization
The N. vectensis culture was kept and handled as previously described . RNA was extracted from a mixture of developmental stages with Trizol reagent (Life Technologies, Carlsbad, CA, USA). cDNA was generated with the SuperScript III enzyme (Life Technologies) and a random hexamers primer mixture (New England Biolabs), according to the manufacturers’ instructions. A transcript encoding NvePTx1 was amplified from cDNA using the primers 5′-CCGAAGACACACAAGCTGACAAGC-3′ and 5′-CATTACCAAGTGAAACCTTTTAGGT-3′. The 442 bp product was then ligated into pGEM-T Easy (Promega, Madison, WI, USA) and digoxigenin-labeled RNA probe was generated using the MEGAscript SP6 kit (Life Technologies). Probe generation and ISH were carried out as previously described .
We are very grateful to D. Fredman and U. Technau of the University of Vienna for sharing their unpublished N. vectensis high throughput sequencing data with us. We are grateful to the Interministerial Commission for Sea Resources (SECIRM). The research has been largely supported by CNPq (563874/2005-8, grant to J.C.F), FAPESP (2009/07128-7 and 2011/21031-6, Masters’ degree fellowship to D.J.B.O.) and PROAP-CAPES 2012 (grant to D.J.B.O.). J.T. was supported by the following grants: G.0433.12, G.A071.10N and G.0257.08 (F.W.O. Vlaanderen), EU-FP7-MAREX, IUAP 7/10 (Inter-University Attraction Poles Program, Belgian State, Belgian Science Policy) and OT/12/081 (KU Leuven). The FNRS and the FEDER (Walloon region) are acknowledged for the funding of the mass spectrometry platform. The authors declare that they have no conflict of interest.