Scorpions are arachnids of wide geographic distribution, with around 2200 species described in families recognized worldwide [1
]. The Buthidae
family has the most studied venoms due to its great toxicity to mammals and hence to humans [2
]. In Colombia, five genera compose this family: Anantheris
(13 species), Centruroides
(4 species), Microtityus
(2 species), Rhopalurus
(1 species), and Tityus
(13 species). The four species of the genus Centruroides
in Colombia are: C. eduardsii
, C. gracilis
, C. marx
, and C. margaritatus
]. C. margaritatus
is distributed in two geographically isolated areas: the upper and middle basin of the Cauca River (Valle del Cauca, Colombia) and the Patía river Valley (Cauca, Colombia) [4
Scorpion venom is a mixture of proteins, peptides and enzymes, carbohydrates, free amines, nucleotides, lipids, and other low molecular weight components with unknown function. Peptides that act as ion channel modulators are the main agents responsible for the venom toxicity and they have been classified according to their targets into: sodium scorpion toxins (NaScTx), with molecular masses between 6–8 kDa [6
], potassium scorpion toxins (KScTx), with molecular masses between 3–5 kDa [7
], and calcium scorpion toxins (CaScTx) that comprise peptides acting on voltage gated calcium channels and that specifically modulate ryanodine receptors [9
]. In the last decades, many details of the toxin-channel interaction have been clarified and models of different mechanisms of toxin binding have been described [11
]. Based on their structural and functional characteristics, KScTx have been classified into seven subfamilies: α-Ktx, β-Ktx, ɣ-KTx, δ-KTx, Ɛ-Ktx, κ-KTx, and ʎ-KTx (kalium database) [14
]. The ɣ-KTx family comprises toxins that selectively bind to ERG (Ether-à-go-go-Related
Gene) potassium channels. These channels are expressed in many tissues and they are especially important for the repolarization of the cardiac action potential. Mutations in the erg1 gene are responsible for congenital long QT syndrome, a disorder of cardiac repolarization, which is characterized by prolongation of the QT interval on the surface electrocardiogram, abnormal T waves, and risk of sudden cardiac death due to ventricular arrhythmias [15
]. The first member of the ɣ-KTxs was isolated from the venom of the scorpion Centruroides noxius
and named CnERG1 (ErgTx1, ɣ-KTx1.1) [16
]. Thereafter, many ɣ-KTxs sequences were identified from scorpions of the genus Centruroides
], and the toxin-channel interactions were characterized for some of these peptides [20
]. A common feature of these toxins is that, despite their concentration, ERG channel blocking effect is always partial (about 90% for CnERG1) [23
] and that ɣ-KTxs accelerate the closure kinetics due to their preference for the channel closed state [22
]. These facts have been partially explained by proposing that the ɣ-KTxs-ERG channel interaction is of the “turret” type, where toxins interact with the zone of the extracellular loop between the transmembrane segments S5 and S6, also called the “turret” of the channel [24
is a markedly synanthropic species [5
] that produces a venom of low toxicity with LD50 of 59.9 mg/kg [5
]; however, in scorpion stings by C. margaritatus
, there have been reports of clinical symptoms associated with cardiovascular disorders, leading even to scorpionism with moderate and severe systemic manifestations [4
]. Previous studies using rats as biological models showed that intravenous administration of a chromatographic fraction (peptides between 2.5 and 6.0 kDa) of the C. margaritatus
venom caused important cardiovascular alterations that included hemodynamic failure. In addition, the histological analysis showed a high density of interfibrillar hemorrhage in cardiomyocytes exposed to the venom fraction [28
]. All these alterations induced by the C. margaritatus
venom may be associated with toxins that interact directly in the heart or smooth muscle, with sodium or potassium channels and specifically with the ERG potassium channels.
Until now, little is known about the C. margaritatus
venom composition. Margatoxin 1, an α-KTx of 39 amino acids and three disulfide bonds, was earlier identified as a potent inhibitor of the Kv1.3 channel in human peripheral T lymphocytes [30
] and later found to inhibit also Kv1.1 and Kv1.2 channels with similar affinity [31
]. Thereafter, in a proteomics study of C. margaritatus
venom, two other peptides were isolated and characterized: a peptide with 24 amino acids and 3 disulfide bridges (MW = 2609.15 Da) and a peptide with 30 amino acids and 3 disulfide bridges (MW = 3376 Da), both classified as αKTxs [32
]; however, no function was tested for these peptides.
In this work, we describe the characterization of C. margaritatus venom in order to determine its activity on voltage-gated sodium and potassium channels. In addition, we present venom separation aimed to identify potassium channel toxins able to block the hERG1 potassium channel. A new γKTx toxin (CmERG1 or γKTx1.10) is described and characterized. CmERG1 was sequenced and the differences in structure and functional features with CnERG1, as well as toxin-channel interaction models, are discussed.
The venom of Buthidae scorpions is rich in neurotoxins acting on voltage-dependent ion channels that are responsible for symptomatologic signs in autonomic hyperactivity (tachycardia, hypertension, cardiac arrhythmias, mydriasis, excessive salivation and tearing, bradycardia, hypotension, and others) [4
]. When these accidents occur, they might cause death, as a consequence of cardiovascular defects attributed to the massive release of catecholamines from the adrenal glands and noradrenergic nerve terminals, together with complications associated with pulmonary edema and respiratory failure in mammals [4
Despite the fact that the C. margaritatus
scorpion belongs to the Buthidae
genus, its venom has been considered of low toxicity [1
], but as mentioned above, in Colombia this species has caused a few cases of moderate and severe scorpionism [4
]. In Colombia, two populations of this species are differentiated according to the geographical distribution: the first in the Patía Valley with a LD50 of 42.83 mg/Kg calculated in mice by intraperitoneal administration [29
] and that is the object of this study; the second, in the Cauca Valley with an LD50 of 59.9 mg/kg calculated in mice by intraperitoneal administration [5
]. Until now, scarce information was available for C. margaritatus
venom; the only venom component characterized by electrophysiology assays was Margatoxin (α-KTx2.2), isolated from C. margaritatus
venom and first identified as a potent and specific blocker of Kv1.3, but later discovered to block also Kv1.1 and Kv 1.2 [30
]. Moreover, studies in rats showed that administration of fractions of the C. margaritatus
venom caused cardiovascular alterations [28
], but which particular component is responsible for these effects is unknown.
Here we analyzed the composition of C. margaritatus
venom looking for peptides active on hERG1 potassium channels, that could explain the cardiovascular alterations reported as consequences of C. margaritatus
stings and also reproduced in rats as biological models [28
]. Potassium hERG1 channels are involved in cardiac physiology, where they contribute to the repolarization of cardiac action potentials; therefore, alteration in ERG currents is associated to arrhythmias and cardiac failure [42
]. We found in the C. margaritatus
venom a new peptide acting on the hERG1 channel. It is a 42 amino acid peptide with four disulfide bonds, which was given the trivial name of CmERG1. According to the international classification [43
], CmERG1 corresponds to the systematic code of ɣ-KTx 1.10, and it appears in the UniProt Knowledgebase under the accession number C0HLM3.
Unlike the other ɣKTxs, the new toxin CmERG1 has the ability to block 100% of the hERG1 current. The nature of the CmERG1 full blockage activity is unknown but certainly resides in its sequence. The mechanism of action of γKTxs has been proposed to be through interactions with the channel “turret” [22
], where the S5-pore loop conforms an aliphatic α-helix [44
] that makes contacts with a cationic and hydrophobic patch in the γ-toxins surface. In CnERG1, the hydrophobic patch is situated at the N-terminal end of the α-helix and the β sheet side of the toxin [36
]; furthermore, site-directed mutagenesis to alanine has confirmed that amino acids K13, Y14, Y17, Q18, M35, and F37 are important for the CnERG1-hERG1 interaction [20
Two models were proposed to explain the uncommon feature of the γKTxs that comprises a partial current block, despite the saturation concentration, and the acceleration of the closing process due to the toxin preference for the channel closed state. In one kinetic model proposed by Hill and collaborators [23
], first the toxin binds to the channel in a toxin channel encounter complex permissive to ion flow (TC*) and then passes to the blocked toxin-channel complex (TC). The incomplete blockage is explained by the relatively rapid rate of dissociation of the complex (TC) compared to the conversion rate of the toxin channel meeting complex (TC*) to TC [16
]. In addition, Tseng’s group proposed for BeKm-1, a “state-dependent” channel-toxin interaction, in which toxin binds preferentially to the closed channels while inactivation promotes toxin unbinding [22
]. This explains the acceleration of deactivation process first observed by Tseng’s and collaborators and the γKTxs proclivity to unbinding at depolarized potential. Actually, these two models are not mutually exclusive and may coexist as we have previously seen with CeErg4 [47
CmErg1 eliminates the current by 100%, suggesting a higher affinity for the pore-blocking configuration over the turret-binding one. We propose qualitative binding models of the pore-blocking configurations of CmERG1 and CnERG1 for comparison. Following Hill and collaborators, in the mindset of an equilibrium between the turret-binding mode and the pore-binding mode, the detailed analysis of the interactions of the toxins with the channels point to small differences in the interactions of F37 (a conserved residue, adjacent to the entrance of the selectivity filter), Q21 (a residue that is a T in CmERG1), and possibly Y17 (a residue that is a F in CmERG1) as the leading candidates to explain the higher efficacy of CmERG1 in blocking the channel: CnERG1 engages in more interactions with the turrets, facing the entry to the selectivity filter at an angle, weakening the interaction with the selectivity filter. The better interactions overall of CmERG1 with the pore, compared to those of CnERG1, bias the equilibrium to the blocking configuration. This leads to a concrete hypothesis testable by mutagenesis: mutating Q21 to T and Y17 to F in CnERG1 should make it a better blocker, and mutating T21 to Q and F17 to Y in CmERG1 should impair its blocking capability.
5. Materials and Methods
5.1. Venom Source, Chemical and Reagents
Fifty scorpions of the species Centruroides margaritatus were collected in the Patía Valley, municipality of Balboa, Department of Cauca, south west Colombia (2°01′21″ N, 77°10′43″ O, 800 masl), with official permission from National Environmental Licensing Authority (ANLA), Colombia (R. 0152 from 15 February 2015). Venom was obtained by electrical stimulation of the telson, dissolved in water and then centrifuged at 15,000 RPM and 4 °C for 15 min. The supernatant was lyophilized and kept at −20 °C until use. Venom concentration was estimated by absorbance measured at 280 nm with spectrophotometer Nanodrop (TermoFisher Scientific, Waltham, MA, USA). All chemicals and reagents were analytical grade substances and Mili Q water was used through all the procedures.
5.2. Peptide Purification
The venom was fractionated by means of FPLC high performance liquid chromatography model NGC chromatography System, Chrom lab Model software, and Bio Frac automatic fraction collector (Bio Rad, Hercules, CA, USA). RP-FPLC was coupled to a reversed phase C18 column, at wavelength λ = 280 nm, through a 60% mobile phase gradient: Solution A (water + 0.10% TFA) and Solution B (acetonitrile + TFA, 0, 10%), with flow rate of 1 mL/min, for 60 min. Peaks were manually collected obtaining 41 fractions (Figure 1
A). Each peak or fraction was subjected to protein quantification by Nano Drop Spectrophotometer ND 1000 equipment and then lyophilized using a Savant SC210A Speed vac Concentrator dryer, (Savant Instruments, Inc. now Thermo Fisher Scientific, Waltham, MA, USA). Peaks with activity against potassium channels were further separated by reversed phase at a high-performance liquid chromatography equipment (RP-HPLC), using an analytical C18 reverse-phase column (Vydac, Hysperia, CA, USA). The pure peptides were obtained by a linear gradient from 100% of solution A (0.12 (v
) trifluoroacetic acid (TFA) in water) to 60% of solution B (0.10 (v
) TFA in acetonitrile) in 60 min at 1 mL/min flow rate. The detection was monitored by absorbance at λ = 230 nm and components were manually collected, dried using a Savant SpeedVac dryer and storage at −20 °C until used for chemical and functional characterization.
From the first separation, we obtained a small amount of toxin CmERG1 that was not enough (in amount and purity) for sequence determination and the electrophysiological characterization. To obtain more quantity of the toxin, we performed a second venom purification with a three-step strategy. This three-step purification allowed us to process more amount of venom and achieve greater yield and purity by using not only reversed-phase chromatography as purification principle but also size exclusion and ion-exchange chromatography. For this, we first applied the soluble venom for the gel filtration on a Sephadex G-50 column (60 cm × 26 mm, L × I.D), in 20 mM ammonium acetate buffer, pH 4.7 at a flow 2 mL/min, and we obtained three fractions. From our experience, the main toxic components are in fraction FII, so it was applied to ion-exchange purification on a carboxy-methyl-cellulose (CMC) column (5 cm × 15 mm, L × I.D.) equilibrated with the same buffer. Chromatography was conducted at a flow rate of 2 mL/min with a linear gradient (0–100%; in 200 min) of 500 mM ammonium acetate buffer pH 7.4. Ten fractions were obtained and dried by lyophilization. All fractions from CMC chromatography were further separated by RP-HPLC, under the same conditions described above. We used an analytical C18 reverse-phase column (Vydac, Hysperia, CA, USA). The pure peptides were obtained by a linear gradient from 100% of solution A (0.12 (v/v) trifluoroacetic acid (TFA) in water) to 60% of solution B (0.10 (v/v) TFA in acetonitrile) in 60 min at 1 mL/min flow rate. The detection was monitored by absorbance at λ = 230 nm and components were manually collected, then dried using a Savant SpeedVac (Savant Instruments, Inc. now Thermo Fisher Scientific, Waltham, MA, USA) dryer and storage at −20 °C until used for chemical and functional characterization.
5.3. Mass Spectrometry and Sequence by Edman Degradation
Single peaks from RP-HPLC were analyzed by mass-spectrometry (MS) with an electrospray ionization (ESI) equipment LCQ FLEET from Thermo Fisher Scientific Inc. (San Jose, CA, USA). Automatic amino acid sequencing of CmERG1 was performed by Edman degradation using a Biotech PPSQ-31A Protein Sequencer equipment from Shimadzu Scientific Instruments, Inc. (Columbia, MD, USA). A sample of native peptide was directly loaded for sequencing. Additionally, a reduced and alkylated sample of the same peptide was sequenced for identification of the cysteine residues.
5.4. Reduction and Alkylation
For reduction, the pure peptide was dissolved in 200 mM TRIS-HCl buffer, pH 8.6 containing 1 mg/mL EDTA and 6 M guanidinium chloride with 2 mg of dithiotreitol (DDT). Nitrogen was bubbled to the solution for 5 min and incubated 45 min at 55 °C. Immediately after 2.5 mg of iodoacetamide was added to the reacting vial, placed in the dark at room temperature for 30 min. Reduced an alkylated peptide was recovered by RP-HPLC (similar conditions as described above).
5.5. Amino Acid Sequence Comparison of Peptide
The CmERG1 sequence data reported in this paper appears in the UniProt Knowledgebase under the accession number C0HLM3.
Basic Local Alignment Search Tool (BLAST) by National Center for Biotechnology Information (NCBI) was used to generate the protein sequence alignment in Figure 2
5.6.1. Cells and Solutions
CHO cells stably expressing hERG1 potassium channels were used for the electrophysiological experiments (hERG1 accession number: NP_000229). We previously prepared this cell line stably transfecting CHO cells with plasmid pcDNA3.1-hERG1 (a kind gift from Enzo Wanke from University of Milano-Bicocca, Italy). Briefly: CHO cells at 80% confluence in a 35 mm culture plate were transfected with 2 µg of pcDNA3.1-hERG1 mixed with 7 uL of Lipofectamine (Invitrogen), accordingly to the manufacturer instructions. After 3 days, cells were selected by adding to the culture medium 2 mg/mL of G418 (SIGMA). After 10 days of selection, cells were cloned by limiting dilution and the resulting clones were probed for their current expression by electrophysiological recordings. High glucose DMEM (Dulbecco’s Modified Eagle Medium, SIGMA, Naucalpan de Juarez, Edo de Mexico, Mexico), supplemented with 10% Fetal Bovine Serum (Biotecfron, Emiliano Zapata, Morelos, Mexico) and with 500 µg/mL of antibiotic G418 (SIGMA) was used as growth medium. Cells were routinely maintained at 37 °C with 5% of CO2 in humidified atmosphere.
Intracellular solution contained in mM: 130 K-Aspartate, 10 NaCl, 2 MgCl2, 10 HEPES, 10 EGTA, pH 7.3 adjusted with NaOH. Extracellular solution contained in mM: 95 NaCl, 40 KCl, 2 CaCl2, 2 MgCl2, 10 HEPES, 5 glucose, pH 7.3 adjusted with NaOH. A concentrated 100–1000× stock was prepared dissolving lyophilized toxin in distilled water and stored at −20 °C until the used (no more than 3 weeks). Solutions were delivered to the cell under patch by means of an active perfusion system connected to a variable speed syringe pump (model A-99 from Razel Scientific Instruments (Saint Albans VT, USA)). The perfusion rate was approximately 1µL/s.
5.6.2. Patch-Clamp Recordings and Data Analysis
Patch pipettes were manufactured from capillary borosilicate glass tubing (Warner Instruments, Hamden, CT, USA) by means of a vertical puller model P-30 (Sutter Instrument, Novato, CA, USA). When pipettes were filled with internal solution, pipette resistance was between 1.5–3 MOhm.
During the depolarizing steps, hERG1 currents are usually small (Figure S1
). This is a consequence of the slow activation kinetics (in seconds) and the very fast inactivation kinetics (in milliseconds) of this channel that can be considered an inward rectifier [48
]. Therefore, it is a common practice to record the tail currents during the hyperpolarization, where the inactivation is quickly removed and the deactivation occurs slowly. For the dose response curve (Figure 3
), tail currents of hERG1 were recorded at −120 mV for 500 ms, after a preconditioning pulse at 60 mV for 500 ms every 5 s. Peak currents recorded in absence and in presence of the toxin were normalized to the maximal peak current in control condition and plotted versus the logarithm of toxin concentration. Data were fitted by a logistic equation. For the voltage dependence of activation study, hERG1 tail currents were recorded during a step at −120 mV for 500 ms, preceded by a 5 s depolarization steps in the range between 30 to −80 mV. To determine the voltage dependence of inactivation, currents were recorded after a depolarization at 40 mV for 1 s, during depolarization steps in the range between 40 to −170 mV. Currents recorded at potentials lower than −90 mV were corrected for deactivation: the deactivation process was fitted by a single exponential extrapolated to the zero point of each step. Data from both activation and inactivation protocols, in absence and in presence of the toxin, were normalized, plotted versus membrane potential, and fitted by a Boltzmann function.
Currents were acquired by using MultiClamp 700A amplifier and DigiData 1440a (Molecular Devices, San Jose, CA, USA). Off-line analysis and graphs were performed by using Clampfit 10 (Molecular Devices) and Origin 8 (OriginLab, Northampton MA, USA).
5.6.3. Statistical Analysis
Where it is not otherwise indicated, electrophysiological data represent the mean of 3–6 cells ± standard error (S.E.). Each cell was recorded in absence and in presence of the toxin and the difference between these two conditions was analyzed by means of the paired sample t-test at 0.05 level.
5.7. Modeling of the Toxin-Channel Encounter Complexes
The CmERG1 structure was modeled by homology using the CnERG1 structure available in the Protein Data Bank [49
] (PDB-ID 1NE5, chain A) as template in Swiss-Model [50
], to ensure correct formation of the four characteristic disulfide bonds of the toxin (between residues 5–23, 11–34, 20–39, and 24–41). The sequence identity between the two toxins is 90%, resulting in a good quality model with GMQE = 0.99 and QMEAN = −1.1. Both CmERG1 and CnERG1 structures were submitted to Charmm-GUI [51
] to add hydrogen atoms and generate CHARMM45 [52
] inputs for energy minimization, which consisted in 200 steepest descent steps for hydrogen atoms only, followed by 400 steepest descent and 400 adopted-basis Newton–Raphson steps for all the atoms.
The structure of the channel, solved by cryoelectron microscopy [54
] for the open state, displays the classical tetrameric arrangement of subunits for potassium channels, but without domain swapping; it also lacks coordinates for some residues in the turrets and the extracellular loops. A recent model for the channel, including critical residues in the P-loop (N598-L602 and H578-R582) but lacking still the extracellular loops in the voltage sensor, was proposed to study both blockers and activators of the channel that bind to the intracellular vestibule [55
]. In addition, work by Noskov’s group has generated refined models for hERG1 starting from the cryo-EM structure of the channel [54
], looking at specific details of the voltage sensors, inactivation mechanisms, and binding sites for drugs [56
]. These are open channels, and no specific attention has been paid to the conformation of the turrets. An independent channel model was kindly donated by Dr. Tseng [22
]; this model was based on voltage-dependent potassium channel KvAP and has domain swapping. The structure shows perfect rotational symmetry, so it corresponds to the channel before interacting with BeKm1. Similar, domain swapped models, have been used recently to study the interaction with other toxins [18
]. In order to avoid the issue of the position of the extracellular loops of the channel, for which there is no experimental data available, we carried out our docking assays over residues 545 to 668, covering S5, the P-loop and, the S6 helix. The Tseng model was used without further modification. This model superimposes correctly onto structure 5VA1 [54
], for the available residues in the turret region.
Docking was carried out in the HADDOCK2.4 server [33
], using as active residues 582, 583, 585, 588, 592, and 628 in the channel model, and residues 13, 14, 17, 18, 21, 26, 27, 35, and 37 in the toxins. The best structures of each cluster (as indicated by HADDOCK) were inspected in the CASTp3.0 server [59
] to select those that interrupted passage through the selectivity filter, using probes of 0.9 and 1 Å radii. These structures were then inspected in VMD [60
] to determine the mode of interaction, and to find those that provided the best plugs for the channel pore. These structures were prepared in Charmm-GUI and energy-minimized in CHARMM45 with 300 steepest descent steps for hydrogen atoms only, 500 steepest descent steps for all atoms, and 500 adopted-basis Newton-Raphson steps for all the atoms. Over these structures, the contacts between toxin and channel were calculated with a 4.5 Å cutoff and the hydrogen bonds were calculated with default values (2.4 Å distance, no angle restriction). The list of contacts is rendered in Figure 6
and Figure 7