Ringhalexin from Hemachatus haemachatus: A novel inhibitor of extrinsic tenase complex

Anticoagulant therapy is used for the prevention and treatment of thromboembolic disorders. Blood coagulation is initiated by the interaction of factor VIIa (FVIIa) with membrane-bound tissue factor (TF) to form the extrinsic tenase complex which activates FX to FXa. Thus, it is an important target for the development of novel anticoagulants. Here, we report the isolation and characterization of a novel anticoagulant ringhalexin from the venom of Hemachatus haemachatus (African Ringhals Cobra). Amino acid sequence of the protein indicates that it belongs to the three-finger toxin family and exhibits 94% identity to an uncharacterized Neurotoxin-like protein NTL2 from Naja atra. Ringhalexin inhibited FX activation by extrinsic tenase complex with an IC50 of 123.8 ± 9.54 nM. It is a mixed-type inhibitor with the kinetic constants, Ki and Ki’ of 84.25 ± 3.53 nM and 152.5 ± 11.32 nM, respectively. Ringhalexin also exhibits a weak, irreversible neurotoxicity on chick biventer cervicis muscle preparations. Subsequently, the three-dimensional structure of ringhalexin was determined at 2.95 Å resolution. This study for the first time reports the structure of an anticoagulant three-finger toxin. Thus, ringhalexin is a potent inhibitor of the FX activation by extrinsic tenase complex and a weak, irreversible neurotoxin.

Scientific RepoRts | 6:25935 | DOI: 10.1038/srep25935 specifically inhibits the prothrombinase complex (CY Koh, RM Kini, unpublished observations). We also determined the mechanism of action of a novel anticoagulant protein complex, hemextin from the venom of Ringhals cobra (Hemachatus haemachatus). The tetrameric hemextin AB complex non-competitively inhibits factor VIIa (FVIIa) with nanomolar affinity 11 . On the other hand exactin isolated from the same venom inhibited the activation of factor X (FX) specifically by extrinsic tenase complex. Interestingly, exactin showed structural similarity to short-chain neurotoxins and exhibited a weak neurotoxicity (VM Girish, RM Kini, unpublished observations).
Here we report the identification, purification and characterization of a novel anticoagulant ringhalexin (Ringhals extrinsic tenase complex inhibitor) from the venom of H. haemachatus. Ringhalexin exhibited a mixed-type inhibition to FX activation by the extrinsic tenase complex and also exhibited a weak, irreversible, neurotoxicity on chick biventer cervicis muscle (CBCM) preparations. Further we determined the three-dimensional structure of ringhalexin which revealed that it has a 3FTx fold maintained by four highly conserved disulfide bonds.

Results
Purification of the anticoagulant protein ringhalexin. H. haemachatus venom was size-fractionated by Superdex 30 column (Fig. 1A). Three major peaks were obtained and the proteins eluted in peak 3 contained mostly 3FTxs. With the interest of isolating the anticoagulant proteins from 3FTx family, peak 3 was further fractionated on a C 18 RP-HPLC column. Individual fractions were lyophilized and their inhibitory activities on FX activation by the extrinsic tenase complex were examined. The estimated percent inhibition of each fraction and elution profile (Fig. 1B) indicates the presence of several extrinsic tenase complex inhibitors. Many peaks contained a mixture of different proteins and further purification by various chromatographic techniques resulted in very low yield or showed no inhibition which made further characterization difficult. In this study, we focused on the purification of ringhalexin (black arrow) as it was found to be a potent inhibitor of FX activation by the extrinsic tenase complex. Ringhalexin was further purified using a shallow gradient (Fig. 1C). The molecular mass and homogeneity of the purified ringhalexin was determined by electrospray ionization mass spectrometry (ESI-MS). ESI-MS showed 4 peaks of mass/charge (m/z) ratio ranging from +4 to +7 charges (Fig. 1D). The mass was determined to be 7437.25 ± 0.53 Da. The total yield of ringhalexin was 1-1.5 mg/g of venom. haemachatus was sub-fractionated by size-exclusion chromatography (SEC) and the proteins were eluted using 50 mM Tris-HCl (pH 7.4). Peak 3 (horizontal bar) corresponds to non-enzymatic 3FTxs. (B) The peak 3 of SEC was subjected to RP-HPLC on a Jupiter C 18 column (10 × 250 mm). A linear gradient of 28-50% of solvent B was used for the elution of proteins. The inhibitory activities of the individual fractions on FX activation by extrinsic tenase complex were determined (dotted line). The peak indicated by the black arrow contains ringhalexin. (C) The fractions containing ringhalexin were pooled and re-chromatographed using a shallow gradient of 32-38% on a Jupiter C 18 column (4.6 × 250 mm). The peak containing pure ringhalexin is indicated by the arrow. (D) ESI-MS of ringhalexin showing four peaks of mass/charge (m/z) ratio ranging from +4 to +7 charges. The mass was determined to be 7437. 25  Amino acid sequence of ringhalexin. The complete amino acid sequence of the ringhalexin was determined by Edman degradation. The first 48 residues were determined by direct sequencing of the native protein whereas the remaining residues were determined by sequencing the overlapping C-terminal fragment of o-iodosobenzoic acid cleaved S-pyridylethylated ringhalexin (Fig. S1). Sequence alignment and the position of cysteine residues showed that ringhalexin belongs to the 3FTx family. It showed 94% identity to Neurotoxin-like protein NTL2 isolated from Naja atra venom ( Fig. 2A). However, NTL2 has not been structurally or functionally characterized. It also showed significant identity (82%) to a hypothetical protein L345_15308 of Ophiophagus hannah (king cobra). Interestingly, ringhalexin showed low identity to classical short-chain neurotoxins (Fig. 2B) and cytotoxins/cardiotoxins with anticoagulant properties (Fig. 2C).
β-sheet structure of ringhalexin. The secondary structure of ringhalexin was evaluated by far-UV CD spectroscopy (Fig. S2). The CD spectrum shows a minimum at 217 nm and a maximum at 196 nm. The CD spectrum is comparable to that of haditoxin from O. hannah venom with a minimum at 215 nm and maximum at 198-200 nm 20 . However, it differed significantly from that of β -cardiotoxin, a β -blocker from the same venom 21 . Thus, ringhalexin was found to be composed of β -sheet structure similar to all other 3FTxs 4 .

Ringhalexin inhibits extrinsic tenase complex.
We determined the effect of ringhalexin on various clotting times. Ringhalexin significantly prolonged the prothrombin time in a dose-dependent manner. It also prolonged APTT and Stypven time slightly at higher concentration but had no effect on thrombin time (Fig. S3).
To determine the potency of ringhalexin, we studied its effect on the reconstituted extrinsic tenase complex. Ringhalexin inhibited FX activation by extrinsic tenase with an IC 50 of 123.8 ± 9.54 nM (Fig. 3A). However, it does not inhibit FVIIa or FXa amidolytic activity at 10 μ M (data not shown). To further understand the interactions, we examined the inhibition kinetics of ringhalexin. The ringhalexin protein showed decrease in V max and increase in K m with the increase in its concentration which is a characteristic of mixed-type inhibition. Thus, ringhalexin exhibits mixed-type inhibition of FX activation by extrinsic tenase complex (Fig. 3B). The kinetic constants, Ki and Ki' derived from the secondary plot were determined to be 84.25 ± 3.53 nM and 152.5 ± 11.32 nM for FX activation by extrinsic tenase complex (Fig. 3C,D) indicating that the affinity of ringhalexin towards the [E] complex (FVIIa/TF PCPS ) was nearly two times higher than that towards the [ES] complex (FVIIa/TF PCPS /FX).
Neurotoxic activity of ringhalexin. To observe the biological effects of ringhalexin, the mice were injected with 10 and 100 mg/kg of the protein. No effect was seen at 10 mg/kg dose. At 100 mg/kg, the mice showed typical symptoms of peripheral neurotoxicity such as hind limb paralysis and labored breathing 22,23 . The average time of death was recorded to be 96 min. Postmortem examinations showed no internal bleeding or hemorrhage. The effect of purified ringhalexin toxin (1-10 μ M) on neuromuscular transmission was studied in the CBCM. Ringhalexin toxin produced time-and concentration-dependent blockade of nerve-evoked twitch responses in indirectly stimulated CBCM. At 10 μ M, ringhalexin toxin produced 75% inhibition of the nerve-evoked twitch responses in the avian neuromuscular junction after 30 min exposure to the toxin (Fig. 4A). The KCl-induced contraction was unaffected, indicating the absence of myotoxicity produced by the toxin. The reversibility of the neuromuscular blockade produced by ringhalexin toxin was evaluated through intermittent washing of the muscle with fresh Krebs solution. No recovery of the neuromuscular blockade was observed following washing of the muscle for 30 min, hence the results indicate that the neuromuscular blockade produced by ringhalexin was irreversible. We used α -bungarotoxin, a well-characterized long-chain neurotoxin, as a positive control (Fig. 4B).
Structural analysis. The structure of ringhalexin was determined by the molecular replacement method using the Balbes program 24 using Bungarus candidus toxin Bucain coordinates (PDB code 2H8U) as model. There were three protein molecules in an asymmetric unit with each molecule consisting of residues from Arg1 to Ala65 (Fig. 5A). All the three monomers are well defined in the electron density map (Fig. 5B). The model was refined to a final R value of 0.22 (R free = 0.27) ( Table 1). The stereo-chemical parameters of the model were analysed by PROCHECK and all residues are in the allowed regions of the Ramachandran plot. Each monomer of the asymmetric unit consists of 6 anti-parallel β -strands (β 2↓ β 1↑ β 4↓ β 3↑ β 6↓ β 5↑ ) that form two β -sheets (Fig. 5A). The first β -sheet consists of two anti-parallel β -strands, β 1 (Leu2-Tyr7) and β 2 (Ser11-Ile16), while the second contains four anti-parallel strands, β 3 (Tyr23-Pro29), β 4 (Ile39-Ala43), β 5 (Cys46-Ala51) and β 6 (Val53-Cys58). The fold of ringhalexin is maintained by four disulfide bonds, and these cysteines are strictly conserved among the 3FTxs. The three fingers of ringhalexin consist of the secondary structures β 1Ω β 2, β 3Ω β 4 and β 5Ω β 6 ( Fig. 5). The electrostatic surface representation shows that the molecule is predominantly positively charged with few negative patches in the surface (Fig. 5C,D). The sequence alignment revealed the conserved residues of ringhalexin as well as its identity to cardiotoxins/cytotoxins (Figs 2 and 6A). Also, ringhalexin shared the common three-finger fold and molecular shape when compared to its structural homologues (Fig. 6B).
A search for topologically similar proteins within the Protein Data Bank (www.pdb.org) with the program DALI 25 revealed significant structural homology between ringhalexin and other 3FTxs ( Table 2). The closest homologs were bucain, cytotoxin and erabutoxin. Interestingly, none of the closest DALI homologs had anticoagulant properties.
Phylogenetic Analysis. A phylogenetic analysis was performed for ringhalexin to understand the evolutionary relationship among various 3FTxs. Our phylogenetic analysis shows that 3FTxs can be broadly divided into five branches. It can be deduced from the branch lengths of 3FTxs in their respective phylogenetic trees that these sequences have undergone significant evolutionary remodeling (Fig. S4). Ringhalexin appears to be evolutionarily closer to neurotoxin-like protein NTL2 from Naja atra and an uncharacterized protein from Ophiophagus hannah. Other two 3FTxs which share the same node are muscarinic toxin 38 from Ophiophagus hannah and an uncharacterized protein from Pantherophis guttatus.

Discussion
Haemostasis is a subtle, highly regulated system and the precise control of blood coagulation is important for the life of humans as any imbalance in its regulation can lead to excessive bleeding or unwanted clot [26][27][28] . Coronary heart diseases and cerebrovascular diseases are the major cause of mortality, resulting in most number of deaths than all other causes together in the western world 29 .
Anticoagulants are used for the prevention and treatment of thromboembolic disorders. Although coumarins, such as warfarin, and heparin are widely used anticoagulants, both have their own limitations, such as variable dose response and narrow therapeutic window 30 . Therefore, there is a great need to develop new anticoagulants targeting specific coagulation enzymes or steps in the coagulation cascade 31 . Blood coagulation cascade is initiated by the extrinsic tenase complex which makes it an important target for the novel anticoagulants development 32,33 . In the past, several inhibitors directed against extrinsic tenase complex, which is thought to be initiator of the blood coagulation cascade, have been studied as it might achieve a better anticoagulation efficacy 34 . But these inhibitors must be engineered to exert their effects only at the required site without affecting physiological haemostasis. Endogenous protein, tissue factor pathway inhibitor (TFPI), has three Kunitz-type proteinase inhibitor domains. It interacts with FXa via P1 residue (Arg107) in the second Kunitz-type domain followed by inhibiting FVIIa/TF by binding to FVIIa active site. FFR-FVIIa, generated by incorporating a tripeptide in the active site of FVIIa, limits the formation of functional FVIIa/TF complex 35 . In addition, two classes of peptide exosite inhibitors and several synthetic compounds targeting FVIIa active site have been designed but they have major limitations such as non-specific inhibition, insufficient oral bioavailability or incomplete inhibition even at saturating concentrations [36][37][38][39][40][41] . A soluble TF mutant with alanine substituted for Lys165 and Lys166 (TFAA) was developed as an anticoagulant 42 . Antibodies against TF have been shown to inhibit the proteolytic activation of FX. One type of antibodies interferes with FVIIa/TF association whereas the other type interferes only with macromolecular substrate docking 43,44 .
Several natural extrinsic tenase complex inhibitors have also been identified and characterized. Nematode anticoagulant protein c2 (NAPc2), a serine protease inhibitor from canine hookworms, inhibits the catalytic complex of FVIIa/TF by first binding to FXa 45 . In contrast to TFPI, NAPc2 binds at an exosite of FX/FXa. Ixolaris, a two-Kunitz TFPI from Ixodes scapularis, interacts with FX/FXa exosite with its second domain followed by the docking of its first domain into FVIIa/TF active site 46,47 . Although various snake venom proteins have been characterized for their anticoagulant properties, the role of 3FTxs as anticoagulants remains to be studied extensively 48,49 .
Here we report the isolation, purification and characterization of a novel protein ringhalexin which was identified by activity-based screening of the H. haemachatus crude venom. It exhibited low identity to the well characterized short-chain α -neurotoxins and cytotoxins/cardiotoxins with anticoagulant properties (Fig. 2). Ringhalexin inhibits extrinsic tenase complex with an IC 50 of 123.8 ± 9.54 nM which is comparable to that of hemextin. However, ringhalexin protein shows a mixed-type inhibition in contrast to the non-competitive inhibition exhibited by hemextin 11 . Hemextin AB complex inhibits FVIIa amidolytic and proteolytic activity non-competitively with a Ki of 50 nM. Ringhalexin does not affect the amidolytic activities of FVIIa or FXa.
Kinetic data of exactin, a mixed-type inhibitor of extrinsic tenase complex from the same venom, indicates that its affinity towards [ES] complex (FVIIa/TF PCPS /FX) is 5 times higher than that towards [E] complex (FVIIa/TF PCPS ) (VM Girish, RM Kini, unpublished observations). In contrast, our kinetic data indicates that ringhalexin binds to [E] complex (FVIIa/TF PCPS ) better than [ES] complex (FVIIa/TF PCPS /FX). On the other hand, naniproin from Naja nigricollis venom competitively inhibits prothrombin activation by prothrombinase complex. Kinetic assays ascertain that naniproin interferes with FXa-FVa interaction by competing with FVa for FXa binding with a Ki of 1.28 μ M (CY Koh, RM Kini, unpublished observations). As expected with their functional studies, ringhalexin shows low sequence identity with exactin, hemextin and naniproin (Fig. 2C).
Ringhalexin showed high sequence identity to an uncharacterized Neurotoxin-like protein NTL2 isolated from Naja atra. Upon investigation of neurotoxic effects of ringhalexin, it was found to be irreversible weak neurotoxin. At 10 μ M, ringhalexin produced 50% inhibition of the nerve-evoked twitch responses in the avian neuromuscular junction after a 15-min exposure to the toxin. However, the EC 50 values for short-chain neurotoxin erabutoxin b and long-chain neurotoxin α -bungarotoxin are 80 nM and 25 nM, respectively (data not shown). Thus, ringhalexin is quite less potent in neuromuscular blockage when compared to erabutoxin b and α -bungarotoxin. The functional invariant residues in short-chain and long-chain neurotoxins towards the muscle type receptor (α β γ δ ) have been identified previously 50,51 . The most important residues involved in binding to nicotinic acetylcholine receptors (nAChRs) are Lys27, Trp29, Asp31, Phe32, Arg33, and Lys47. Other residues involved in the recognition are His6, Gln7, Ser8, Ser9, and Gln10 in loop I; and Tyr25, Gly34, Ile36, and Glu-38 in loop II of short-chain neurotoxins 13 . Ringhalexin lacks most of these functional invariant residues explaining its low neurotoxicity.
Ringhalexin, like other 3FTxs, has three β -stranded loops extending from a central core containing four conserved disulfide bonds which resembles the three outstretched fingers of a hand (Fig. 6B). Ringhalexin showed highest structural similarity to bucain, muscarinic toxin and various cardiotoxins ( Table 2). The closest structural homolog in the neurotoxin family was Neurotoxin-1 from Naja naja oxiana venom. The DALI search did not return any 3FTx with anticoagulant activity from the PDB database and so far this is the first structure of a 3FTx with anticoagulant properties. As shown in the case of other 3FTxs, the loop II of ringhalexin was very flexible and some of the residues showed very high B values. This flexibility suggests the possible role of loop II residues in the function of ringhalexin. Since this is the first report of 3D structure of 3FTx anticoagulant, we are determining the structure of other anticoagulants for structural comparison. In addition, we plan to examine the structure-function relationships of ringhalexin.
In summary, we have structurally and functionally characterized a novel protein named as ringhalexin from H. haemachatus. It is quite possible that the protein has different sites for its anticoagulant and neurotoxic activity. This warrants further investigation and in future we would like to understand the detailed mechanism of its action.

Methods
Animals. Swiss albino mice were acquired from the National University of Singapore Laboratory Animal Center and acclimatized to the Animal Holding Unit for at least 3 days before the experiments. The animals were kept under standard conditions with food and water available ad libitum in a light-controlled room (12 h light/ dark cycle, light on 07:00 h) at 23 °C and 60% relative humidity. Domestic chicks were purchased from Chew's Agricultural Farm, Singapore and delivered on the day of experimentation.

Ramachandran statistics
Most favored and allowed regions (%) 100 Disallowed regions (%) 0 of the reconstituted fractions were examined on FX activation by the extrinsic tenase complex (described below).
The peak corresponding to ringhalexin was pooled and re-chromatographed using a shallow gradient of 32-38% on a Jupiter C 18 column (4.6 × 250 mm). The molecular weight of the protein sample was determined by electrospray ionization-mass spectrometry (ESI-MS) using API-300 LC/MS CD spectroscopy. Far-UV CD spectra (260-190 nm) were recorded using a Jasco J-810 spectropolarimeter (Jasco Corporation, Tokyo, Japan). The protein samples (20-50 μ M) were dissolved in 1 mM phosphate buffer and the measurements were carried out at room temperature using a 0.1 cm path length cuvette. The instrument optics was flushed with 30 l/min of nitrogen gas. The spectra were recorded using a scan speed of 50 nm/min, resolution of 0.1 nm and bandwidth of 1 nm. An average of three scans was taken to increase the signal to noise ratio and baseline was subtracted.  Effect of ringhalexin on plasma clotting times. All experimental protocols were approved by Institutional Review Board (NUS-IRB reference code: 08-322E) and the experiments were conducted in accordance with the approved guidelines. Following written informed consent from the healthy volunteers, citrated human blood was obtained through Tissue Repository (National University Hospital, Singapore). Fresh plasma was obtained by centrifugation at 2600 g, 4 °C for 15 min. The effect of ringhalexin (0.3 μ M to 100 μ M) in 50 mM Tris-HCl buffer, pH 7.4 were studied on Prothrombin time, Stypven time, Thrombin time and APTT of human plasma (described below). All the experiments were done at 37 °C and the fibrin clot formation was monitored using a 96-well microplate reader for 10 min at 650 nm.  Prothrombin time. Briefly, 100 μ l of plasma, 25 μ l of 50 mM Tris-HCl, pH 7.4 and 50 μ l of ringhalexin were incubated for 5 min which was followed by the addition of 25 μ l of pre-warmed thromboplastin with calcium reagent to initiate clotting. The fibrin clot formation was monitored by microplate reader.
Stypven time. Briefly, 50 μ l of plasma was incubated with 50 μ l of ringhalexin for 3 min followed by addition of pre-warmed RVV-X (50 μ l, 10 ng/ml) and incubated for another 2 min. The clotting was initiated by the addition of 50 μ l of 25 mM pre-warmed CaCl 2 and the fibrin clot formation was monitored. In vivo toxicity study. Ringhalexin protein (200 μ l in 0.9% saline) was injected intraperitoneally (i.p.) into male Swiss albino mice at doses of 10 and 100 mg/kg (n = 2) and the symptoms were observed. The control group was injected with 200 μ l of 0.9% saline (n = 2).
Ex vivo organ bath study. Isolated tissue experiments were conducted as described previously 6,52 using a conventional organ bath (6 ml) containing physiological Krebs-Henseleit buffer of the composition (in mM): 118 NaCl, 4.8 KCl, 1.2 KH 2 PO 4 , 2.5 CaCl 2 , 2.4 MgSO 4 , 25 NaHCO 3 and 11 D-(+ ) glucose), pH 7.4. Organ bath chambers were continuously aerated with carbogen (5% carbon dioxide in oxygen) and maintained at 37 °C throughout the experiment. The resting tension of the isolated tissues was maintained between 1-2 g tension and the tissues were allowed to equilibrate for 30-45 min before the start of an experiment. Electrical field stimulation (EFS) was carried out through platinum ring electrodes using a Grass stimulator S88 (Grass instruments, West Warwick, RI, USA). The magnitude of the contractile responses was measured in gram tension. Data were continuously recorded on PowerLab LabChart 6 data acquisition system using a force displacement transducer (Model MLT0201) (ADInstruments, Bella Vista, New South Wales, Australia).
Chick biventer cervicis muscle (CBCM) preparation. The CBCM nerve-skeletal muscle preparation 53 was isolated from 3-to 5-day old chicks and mounted in the organ bath chamber under similar experimental conditions as described above. Motor responses of the muscle were evoked by stimulating the motor nerve supramaximally by EFS (7-10 V, 0.1 ms, 0. Crystallization and structure determination. Crystallization screens were performed with the hanging drop vapor diffusion method using Hampton Research screens. The protein was at a concentration of 35 mg/ml, and 1:1 crystallization drops were set up with the reservoir solution. The diffraction quality crystals of ringhalexin were obtained from a reservoir solution containing 29% MPD + 0.1 M HEPES pH 7 + 0.3 M sodium citrate. Crystals were grown up to 10 days and were cryo-protected with 20% (w/v) glycerol supplemented (the mother liquor concentration was maintained by exchanging water with glycerol) with the crystallization condition. Ringhalexin crystal diffracted up to 2.95 Å resolution and belongs to P4 1 2 1 2 space group. A complete data set was collected using a Saturn944 CCD detector mounted on Rigaku X-ray generator. The data set was processed and scaled using Mosflm 54 and Aimless 55 . The structure of ringhalexin was determined by the molecular replacement method using the online program Balbes 24 . Bucain, a cardiotoxin from the Malayan Krait Bungarus candidus (PDB code 2H8U; sequence identity 45%) was used as a search model. There were three ringhalexin molecules located in the asymmetric unit. The resultant electron density map was of good quality. Several cycles of model building/refitting using the program Coot 56 , and alternated with refinement using the program Phenix-refine 57 , lead to the convergence of R-values (Table 1). Non-crystallographic symmetry (NCS) restraints were used throughout the refinement process.
Sequence Alignment and Phylogenetic Analysis. Representative 3FTxs homologs were selected from a BLAST search with ringhalexin sequence and used for phylogenetic analysis and tree building. It was performed using the Phylogeny.fr software platform using the "advanced" mode 58 . The sequence alignment was done using MUSCLE 59 , curation using G-blocks 60 , phylogeny using PhyML 61 , and tree building using TreeDyn 62 . The bootstrapping value in the phylogeny mode was set to 100 iterations.