Research paper
P-III hemorrhagic metalloproteinases from Russell’s viper venom:
Cloning, characterization, phylogenetic and functional site analyses
*
Hong-Sen Chen
a, Hsin-Yu Tsai
b, Ying-Ming Wang
b, Inn-Ho Tsai
a,b,*
a
Graduate Institute of Biochemical Sciences, National Taiwan University, Taipei, Taiwan
b
Institute of Biological Chemistry, Academia Sinica, Taipei, Taiwan Received 25 December 2007; accepted 14 May 2008
Available online 23 May 2008
Abstract
Two homologous P-III hemorrhagic metalloproteinases were purified from Russell’s viper venoms from Myanmar and Kolkata (eastern India), and designated as daborhagin-M and daborhagin-K, respectively. They induced severe dermal hemorrhage in mice at a minimum hem-orrhagic dose of 0.8e0.9 mg. Daborhagin-M specifically hydrolyzed an Aa-chain of fibrinogen, fibronectin, and type IV collagenin vitro. Anal-yses of its cleavage sites on insulin chain B and kinetic specificities toward oligopeptides suggested that daborhagin-M prefers hydrophobic residues at the P1, P10, and P20 positions on the substrates. Of the eight Daboia geographic venom samples analyzed by Western blotting,
only those from Myanmar and eastern India showed a strong positive band at 65 kDa, which correlated with the high risk of systemic hemor-rhagic symptoms elicited byDaboia envenoming in both regions. The full sequence of daborhagin-K was determined by cDNA cloning and sequencing, and then confirmed by peptide mass fingerprinting. Furthermore, molecular phylogenetic analyses based on 27 P-IIIs revealed the co-evolution of two major P-III classes with distinct hemorrhagic potencies, and daborhagin-K belongs to the most hemorrhagic subclass. By comparing the absolute complexity profiles between these two classes, we identified four structural motifs probably responsible for the phylogenetic subtyping and hemorrhagic potencies of P-III SVMPs.
Ó 2008 Elsevier Masson SAS. All rights reserved.
Keywords: metalloproteinase; Substrate specificity; Geographic variation; Phylogenetic analysis; Daboia russelii; Daboia siamensis
1. Introduction
Local and systemic hemorrhages are prominent symptoms of envenoming by snakes of the Viperidae family[1]. Symp-toms observed are caused mainly by the actions of snake venom metalloproteinases (SVMPs), which belong to the reprolysin subfamily of metzincins[2]. SVMP-induced hem-orrhage may result from fibrinogenolysis that impairs coagula-tion, and the degradation of basement membrane proteins,
which damages blood vessels [3,4]. SVMPs are categorized into P-I, P-II, and P-III groups according to the extension of several structural domains [2]. P-I enzymes comprise only the metalloproteinase domain, P-II enzymes contain a disinte-grin domain after the metalloproteinase domain, and P-III enzymes (50e70 kDa) are usually glycoproteins that contain an additional Cys-rich C-terminal domain. In general, the P-III SVMPs are more hemorrhagic than the P-I SVMPs, which indicate that the additional domains of P-IIIs might contribute to their hemorrhagic potencies.
In addition to hemorrhage, other versatile functions have also been reported for P-III SVMPs, e.g. activating prothrom-bin[5,6], inducing endothelial cell apoptosis[7], cleaving in-tegrins[3], and inhibiting platelet functions[8], which implies that more structural diversities exist between P-III SVMPs. Based on the position of the seventh cysteinyl residue in the metalloproteinase domain, three new P-III subclasses were
Abbreviations: MHD, minimum hemorrhagic dose; MMP, matrix metallo-proteinase; PMF, peptide mass fingerprinting; SVMPs, snake venom metalloproteinases.
*
The nucleotide sequence of daborhagin-K was deposited in GenBank with the accession number DQ137798.
* Corresponding author. Institute of Biological Chemistry, Academia Sinica, P.O. Box 23-106, Taipei, Taiwan. Tel.:þ886 2 2362 0264; fax: þ886 2 2363 5038.
E-mail address:bc201@gate.sinica.edu.tw(I.-H. Tsai).
0300-9084/$ - see front matterÓ 2008 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.biochi.2008.05.012
Biochimie 90 (2008) 1486e1498
revealed[2,9]: P-IIIa, which undergoes autoproteolysis to re-lease a w30-kDa fragment with disintegrin- and Cys-rich do-mains; P-IIIb, which forms a dimeric structure; and P-IIIc, which contains the seventh cysteinyl residue at position 100 in its metalloproteinase domain. However, the structural ele-ments attributed to their functional variations and hemorrhagic potencies remain elusive[2].
Two species of Russell’s vipers,Daboia siamensis and Da-boia russelii, are medically the most important Viperinae in South and Southeast Asia [10], and distributed in eastern and western ranges of their habitat, respectively [11]. Con-sumptive coagulopathy resulting in spontaneous and systemic bleeding are the main causes of fatal envenoming by the viper
[10]. The venom’s phospholipases A2, hemorrhagins, and
pro-coagulant enzymes, including Factors X and V protease acti-vators, are presumed to be responsible for these clinical manifestations [10,12]. However, striking variations in the clinical symptoms of envenomed victims [12,13] appeared to reflect some geographical differences in venom compo-nents[13].
Relative to other regions,Daboia snakebites in Myanmar and eastern India are known to cause severe internal bleeding and higher mortality[14,15]. Although the hemorrhagic prote-ase VRR-73 was isolated from D. russelii, its properties and contribution to hemorrhage have not been clarified [16]. In the present study, we purified and characterized the corre-sponding hemorrhagins (namely daborhagin) of Daboia venoms from both regions and solved its full sequence. We also examined the occurrences of P-IIIs in the venoms from various true-viper species or subspecies, and did phylogenetic analyses to trace the evolution of daborhagin and other P-III SVMPs.
2. Materials and methods 2.1. Venom and reagents
Daboia venoms from Myanmar and eastern India were gifts from Professors Yu-Yen Shu and Antony Gomes[17]. Three individualD. siamensis venom samples from southern Myan-mar were kindly given by Professor R. David G. Theakston. Venoms of Daboia russelii pulchella (Sri Lanka), D. russelii (Pakistan),D. siamensis (Taiwan), Echis leucogaster, Echis so-chureki, Echis pyramidium, Vipera lebetina turanica, Vipera lebetina mauritanica, and Vipera ammodytes montandoni were purchased from Latoxan (Rosans, France). Venom of D. siamensis (Thailand) was from the Thailand Red-Cross, Bangkok, and venom of D. siamensis (Indonesia) was from Venom Supplies (Adelaide, Australia). Vipera ammodytes andVipera berus venoms were from SigmaeAldrich Co. (St Louis, MO, USA). Western IndianD. russelii venom was ob-tained from the Haffkine Institute (Mumbai, India). Other venoms used were from the Miami Serpentarium Laboratory (Punta Gorda, FL, USA).
Peptide:N glycosidase F (PNGase F) and Endo H were from New England Biolabs (Beverly, MA, USA). Modification en-zymes, restriction enen-zymes, broad-range protein markers,
sequencing-grade modified trypsin, and pGEM-T vector were purchased from Promega Corp. (Madison, WI, USA). Protein substrates, insulin chain B, and other chemicals were from either SigmaeAldrich or Merck (Darmstadt, Germany). Three fluorogenic substrates, (7-Methoxycoumarin-4-yl)acetyl-Pro-Leu-Gly-Leu-(2,4-dinitrophenyl)-Ala-Arg-NH2 (FS-1),
(7-Methoxycoumarin-4-yl)acetyl-Arg-Pro-Lys- Pro-Tyr-Ala-Nva-Trp-Met-Lys-{Nb-3-(2,4-dinitrophenyl)-L
-2,3-diamino-propionyl}-NH2(NFF-2), and
(7-Methoxycoumarin-4-yl)ace-tyl-Arg-Pro-Lys-Pro-Val-Glu-Nva-Trp-Arg-Lys-{Nb -3-(2,4-di-nitrophenyl)-L-2,3-diaminopropionyl}-NH2 (NFF-3), were
from the Peptide Institute Inc. (Osaka, Japan). 2.2. Purification of P-III SVMPs
Approximately 20 mg of the crude venoms ofD. siamensis (Myanmar) orD. russelii (Kolkata) were dissolved in 200 ml of 0.1 M ammonium acetate (pH 6.7) and then centrifuged at 12,000 g for 5 min. The supernatant was then loaded to a Superdex G-75 column (10/300 GL; Pharmacia, Uppsala, Sweden) on an FPLC apparatus. The column was eluted at a flow rate of 1.0 ml/min, and fractions of 0.5 ml were lected. Fractions showing caseinolytic activities based on a col-orimetrical assay [18] were collected. Pooled sample was lyophilized, redissolved, injected into a Mono Q column (5/ 50 GL; Pharmacia) pre-equilibrated with 50 mM TriseHCl buffer (pH 8.0), and then eluted with a 0e0.6 M NaCl gradient in the same buffer. Finally, the sample was purified using a Mono S column (HR 16/10; Pharmacia) pre-equilibrated with 10 mM sodium phosphate (pH 6.7) and eluted with a 0e0.3 M NaCl gradient. Likewise, another two other P-IIIs were purified from C. vipera and E. leucogaster venoms, and designated as CVHRa (99 kDa) and ECLV-DM (97 kDa), respectively. The other three P-IIIs, Acurhagin, BHRa, and TSV-DM, were also purified according to pub-lished methods [18e20]. Their purity and molecular mass were confirmed by SDSePAGE.
2.3. Protein quantification
Soluble crude venom and purified proteins were quantified using a BCA protein assay kit (Pierce Chemical Co., Rock-ford, IL, USA) with bovine serum albumin as a standard. 2.4. Hemorrhage test
The hemorrhagic activity of samples was measured by the published methods [21] with minor modifications. ICR mice (28e30 g) were shaved on their dorsal regions and then subcu-taneously injected with samples in 100 ml of PBS buffer. After 24 h, the mice were sacrificed with an overdose of CO2, and
the hemorrhagic spots were measured from the inside of the moved skins. The minimal hemorrhagic dose (MHD), defined as the amount of toxins causing a 10 mm skin-lesion in 24 h, was determined from a plot of lesion diameters against the doses of toxins injected. Experiments involving mice and rab-bits were done under the approval and followed the guidelines
of the Animal Experiment Review Committee of the Acade-mia Sinica, Taiwan.
2.5. Mass spectrometry
The precise molecular weight of daborhagin-M was ana-lyzed in linear mode using the MALDI-TOF mass spectrome-ter (4700 Proteomics Analyzer; Applied Biosystems, Fosspectrome-ter City, CA, USA) equipped with an Nd:YAG laser (355-nm wavelength and 200-Hz repetition rate). One thousand shots were accumulated in positive ion mode. The sample was dis-solved in 50% acetonitrile with 0.1% formic acid and pre-mixed with a 5 mg/ml matrix solution of sinapic acid in 70% acetonitrile with 0.1% formic acid for spotting onto the target plate.
2.6. Caseinolytic activity assay and inhibitor study
Caseinolytic activity was measured colorimetrically as pre-viously described[18]. The 100 ml reaction mixture, contain-ing 0.1 mg of purified daborhagin, 0.5 mg of azocasein in 50 mM TriseHCl buffer (pH 8.0), was incubated at 37C for 90 min. Reactions were quenched by adding 200 ml of 5% trichloroacetic acid at room temperature. After spinning at 1000 g for 5 min, 150 ml of the supernatant was mixed with an equal volume of 0.5 M NaOH to halt the reaction, and absorbance was determined at 450 nm. One unit of proteo-lytic activity was defined as the amount of enzyme required to cause a 1.0 increase in absorbance at 450 nm per minute. Spe-cific activity was expressed as units/mg protein. For the inhib-itor study, 5 mM of metal ions, chelators, or protein inhibinhib-itors were added to each reaction, separately.
2.7. Hydrolysis of native proteins and oligopeptide substrates
Fibrinogen and basement membrane proteins (collagen, laminin, and fibronectin) were incubated with daborhagin-M in 100 mM TriseHCl buffer (pH 8.0) at 37C. The reactions were terminated by adding reducing buffer, and boiled at 95C for 5 min. The hydrolyzed products were then subjected to SDSePAGE. The gels were stained with Coomassie Bril-liant blue G-250 (Gelcode Blue Stain; Pierce Chemical).
Oxidized insulin chain-B (715 mM) was incubated with 1.0 mM daborhagin-M at 37C in 50 mM Tris buffer (pH 8.0). At various times, the reaction was halted by adding 25 mM EDTA. Peptide products were separated using reverse phase HPLC on a Vydac C18-column, and monitored with the
absorbance at 214 nm. The peak fractions were collected and dried in a vacuum-centrifuge device (Labconco Corp., Kansas City, MO, USA). Each sample was analyzed using MALDI-TOF or electrospray ionization (ESI) mass spectrometry.
Stock solutions of synthetic fluorogenic substrates (FS-1, NFF-2, NFF-3)[22,23]were prepared in dimethyl sulfoxide. The P-III enzymes used for the kinetic study were freshly pu-rified and kept active; their concentrations were determined based on their molecular weights. The assay was done by
incubating a 1-ml mixture containing 10 nM enzymes and 1 mM substrate in 0.1 M TriseHCl (pH 7.5) with 0.1 M NaCl, 10 mM CaCl2, and 0.05% (w/v) Brij35. The increase
in fluorescence intensity, in relative fluorescence units (RFU), was measured on a fluorescence spectrophotometer (F-3010; Hitachi Koki Co. Ltd., Tokyo, Japan), with excitation at 325 nm and emission at 393 nm. The first-order rate con-stant obtained in each experiment was divided by the enzyme concentration to calculate its specificity index:kcat/Km.
The individual kinetic parameters,Kmandkcat, for the
NFF-2 substrate were also determined by analyzing experimental data with LineweavereBurk plots. The assay mixture con-tained 10e20 nM P-III enzymes and the concentration of the substrate was 0.2e5.0 mM.
2.8. Antiserum and immunoblotting
Approximately 250 mg of purified daborhagin-M in phos-phate-buffered saline was thoroughly mixed with an equal vol-ume of Freund’s complete adjuvant for the first injection, or Freund’s incomplete adjuvant for the second and third injec-tions. It was subcutaneously injected every two weeks into the back of a male rabbit biweekly. Ten days after the third in-jection, blood was taken from the rabbit’s ear vein. The serum was harvested using centrifugation after it had clotted and been stored at 4C. For Western blotting, 100 ng of crude venoms was separated using 8% SDSePAGE under non-re-ducing conditions. After the samples had been blotted onto a PVDF membrane, they were probed using anti-daborhagin-M antiserum (1:1000 dilution) and horseradish peroxidase-conju-gated second antibody (1:2000 dilution). Immunoreactive bands were detected using the NiCl2enhancement method[24].
2.9. N-terminal sequencing
To determine theirN-terminal sequences, purified daborha-gins (10e20 mg/well) were electrophoresed on a 1.0 mm thick 8% SDSePAGE under reducing condition. The protein bands were blotted to a PVDF membrane. After the samples had been stained with Amido Black (0.2% in 7% acetic acid), the bands were cut out and sequenced using a gas-phase amino acid sequencer (Procise 492; Applied Biosystems).
2.10. Cloning and sequencing
RNA and cDNA were prepared from a pair of venom glands obtained from aD. russelii specimen from eastern India[17]. PCR amplification of the daborhagin-K cDNA was done using specific primers: a sense degenerate 17-mer complementing the N-terminal residues 7e12 (NRYFNP), and an antisense 18-mer designed according to the conserved sequences of far 30-UTR of SVMPs [25]. To clone the 50 upstream region, one primer based on the 50-end conserved sequences and the other based on the amino acid sequence ‘‘AIDLNGL’’ of the protease domain were used. PCR conditions were as follows: initial denaturation for 2 min at 94C, followed by 35 exten-sion cycles (denaturation for 1 min at 94C, annealing for
1 min at 52C, and elongation for 1 min at 72C), and a termi-nal extension step of 72C for 10 min. The products were cloned into a pGEM-T vector, and the plasmids were trans-formed toEscherichia coli strain JM 109. White transformants were selected and positive clones were subjected to DNA se-quencing using Taq-Dye-Deoxy terminator cycle sese-quencing kit (Applied Biosystems).
2.11. Peptide mass fingerprinting
For peptide mass fingerprinting (PMF), gel bands of native and deglycosylated daborhagins were excised separately and cut into pieces. Each sample was dehydrated with acetonitrile for 10 min, dried, and then dissolved in 25 mM NH4HCO3(pH
8.5) containing 100 mM dithioerythritol at 37C for 1 h. Its Cys-residues were alkylated with 65 mM iodoacetamide at 27C for 1 h in the dark. The protein was washed twice with 50% acetonitrile, dried, and hydrolyzed with 25 ng of modified trypsin (Promega) in 25 mM NH4HCO3(pH 8.5) at
37C for 16 h. The digest was twice extracted with 50% ace-tonitrile (containing 5% formic acid) for 15 min each and then dried. The resultant peptides were analyzed using MALDI-TOF/TOF with a detecting mass range of 800e4000 Da. 2.12. Phylogenetic analysis
Amino acid sequences of 26 P-III SVMPs were retrieved using BlastP search. Their sequences were aligned using the Vector NTI program (Invitrogen Corp., Carlsbad, CA, USA). A phylogenetic tree was generated using the neighbor-joining methodology of the PHYLIP program, with an RVV-X heavy chain as an out-group. The degree of confidence was deter-mined using bootstrap analyses of 1000 replicates[26]. 2.13. Absolute complexity plot and homologous
modeling
The absolute complexity plot shows the average of pairwise alignment scores of each residue using the substitution matrix blosum62mt2. Individual plots were generated using the align-ment of the members from each class with the AlignX module of the Vector NTI program.
The 3D-model of daborhagin was built using the Modeller pro-gram (http://salilab.org/modeller), with the crystal structure of ca-trocollastatin/VAP2B (PDB code 2DW0) from Crotalus atrox venom[27]as a template. Model geometry was analyzed using the PROCHECK program. The ribbon diagrams were generated using the PyMOL program (http://pymol.sourceforge.net/).
3. Results
3.1. Purification and characterization of daborhagin
The crude venom ofD. siamensis (Myanmar) was separated into several fractions using a Superdex G-75 column (Fig. 1A). The first fraction, which showed the strongest caseinolytic
activity, was collected. After it had been desalted and concen-trated, it was further partitioned using ion-exchange chroma-tography on Mono Q (Fig. 1A) and Mono S (data not shown) columns. The purified active component induced hemorrhage with a minimum hemorrhagic dose (MHD) of 0.86 mg when subcutaneously injected into mice. Based on the species origin (Dabo-), hemorrhagic activity (-rhagin), and geographic region (Myanmar), we designated it daborhagin-M. Using the same procedures, we purified another hemorrhagin from the venom of D. russelii (Kolkata, eastern India) and designated it dabo-rhagin-K. Its hemorrhagic potency with a MHD of 0.82 mg
Fig. 1. Purification and characterization of daborhagins. (A) MyanmarDaboia siamensis venom was separated using a Superdex G-75 column equilibrated with 0.1 M ammonium acetate (pH 6.7), and then a Mono Q column with a 0e0.6 M NaCl gradient in 50 mM TriseHCl (pH 8.0). The active fractions are indicated by bars. The inset indicates the SDSePAGE patterns of purified daborhagin-M and -K under reducing (R) and non-reducing (NR) conditions. (B) The precise molecular weight of native daborhagin-M was analyzed using MALDI-TOF. (C) Daborhagin-M (4 mg) was incubated with buffer only (lane 1) or with 1 unit of PNGase F (lane 2) or 5 units of Endo H (lane 3) at 37C
for 3 h. Samples were then subjected to SDSePAGE under reducing condi-tions. Arrowheads indicate deglycosylation enzymes.
was similar to that of daborhagin-M. The amounts of daborha-gin-M and daborhagin-K in the crude venoms were estimated to be 5.5% and 0.8% (w/w), and their specific activities toward azocasein were 7.1 and 7.3 units/mg, respectively.
SDSePAGE analysis of both daborhagins revealed a single band with an apparent mass of 62 and 66 kDa under non-re-ducing and renon-re-ducing conditions, respectively (Fig. 1A, inset). We further determined the mass of daborhagin-M using MALDI-TOF mass spectrometry. We identified a single-charged monomer with an average molecular mass of 65,065 Da and a double-charged monomer (m/z 31,627)
(Fig. 1B). PNGase F treatment reduced the mass to 43 kDa,
but Endo H treatment did not affect it (Fig. 1C), which sug-gested that daborhagin-M contains 4w5 complex type N-linked glycans. To examine its general proteolytic activities, we used various divalent metal ions and proteinase inhibitors in regular azocasein assays (data not shown). Adding 5 mM of Ca2þor Mg2þ increased the caseinolytic activities of the purified daborhagin-M by 5e23%, and adding 5 mM of EDTA, EGTA, or 1,10-phenanthroline strongly inhibited its activities by 84e94%. However, serine protease inhibitor, e.g. PMSF, was not inhibitory. These results suggested that da-borhagin-M is a high molecular weight metalloproteinase. 3.2. Proteolytic activity toward plasma and basement membrane proteins
The specificities of daborhagin-M were studied using poten-tial plasma and matrix proteins as substrates. At a low enzyme concentration of 75 nM, daborhagin-M cleaved the Aa-chain of human fibrinogen specifically within minutes, but it had no apparent effect on Bb- and g-chains (Fig. 2A). Thus, dabo-rhagin-M is a new a-fibrinogenase. Notably, it completely de-graded high molecular weight subunits (>200 kDa) of type IV
collagen in 2 h, and partially degraded a 250 kDa main chain of fibronectin into five fragments of 83e225 kDa. By contrast, we barely detected laminin hydrolysis after 24 h (data not shown). 3.3. Cleavage sites on insulin chain-B
We also used oxidized insulin chain-B to examine the pro-teolytic specificity of daborhagin (Fig. 2B). After oxidized in-sulin chain-B had been hydrolyzed by 1.0 mM daborhagin-M at 37C, its products at various incubation times (10 mine 24 h) were isolated using reverse-phase HPLC, and then each purified oligopeptide was analyzed using MALDI-MS spectrometry (data not shown). We found that daborhagin-M had cleaved chain-B at four sites (e.g. X-Leu and X-Phe), sim-ilar to other SVMPs[28]. The fast cleavages were at Ala14e Leu15 and Tyr16eLeu17, followed by those at His10eLeu11, and the slow cleavages were at Phe24ePhe25(Fig. 2B). 3.4. Kinetic study using fluorogenic substrates
The fluorogenic peptide substrates FS-1 (cleaving at Gly-Leu), NFF-2 (cleaving at AlaeNva), and NFF-3 (cleaving at GlueNva) were originally developed to measure matrix metal-loproteinase (MMP) activity[22,23]. To study the relationship between substrate specificity and hemorrhagic potency, we compared the kinetic specificities of daborhagin-M and the other five P-IIIs toward these substrates. The specificity index (kcat/Km) of daborhagin-M toward NFF-2 was about 3e4 times
higher than those toward FS-1 or NFF-3 (Table 1). Other strong venom hemorrhaginsdBHRa ofBitis arietans venom
[19], CVHRa ofC. vipera venom, and acurhagin of Deinagkis-trodon acutus venom[18]dalso showed higherkcat/Kmvalues
toward NFF-2 or NFF-3 than the weak hemorrhagins, e.g. TSV-DM of Trimeresurus stejnegeri venom[20] and
ECLV-Fig. 2. Substrate specificities of daborhagin-M. (A) Fibrinogen and basement membrane proteins (1 mg/ml) in 100 mM TriseHCl (pH 8.0) were hydrolyzed by 75 nM daborhagin-M (for fibrinogen) or 750 nM daborhagin-M (for other substrates) at 37C. Reaction time is shown above the gel lanes; the daborhagin-M band is marked with an asterisk. (B) Hydrolysis of oxidized insulin chain B using 1.0 mM daborhagin-M at 37C. Fast, moderate, and slow cleavages are marked by thick arrows, thin solid-line arrow and thin dashed-line arrow, respectively.
DM ofE. leucogaster venom (Table 1). By contrast, RVV-X barely hydrolyzed these synthetic substrates.
We also determined theKmandkcatvalues of all the above
P-IIIs for substrate NFF-2. The data suggested that lowerKm
and much higherkcatvalues accounted for the faster hydrolysis
of NFF-2 by the strong hemorrhagic P-IIIs than the weak or non-hemorrhagic P-IIIs (Table 1).
3.5. Occurrence of daborhagin-like enzymes in Daboia and other viperid venoms
Anti-daborhagin antiserum was prepared by immunizing one rabbit with native daborhagin-M; the antiserum reacted similarly to both daborhagins and easily recognized them be-low 1.0 ng in Western blot analysis (Fig. 3A). To avoid false-positive results, only 100 ng of each venom sample was loaded into the gel. Daborhagins were easily detected in Myanmar and eastern India Daboia venoms, with estimated levels of 8.0% and 1.5% (w/w of total soluble venom proteins) (Fig. 3B), respectively. A faint 60-kDa band was also detected in western IndiaD. russelii venom. However, no antigen was detected in the Daboia venoms collected from Pakistan, Sri Lanka, Thailand, Indonesia, and Taiwan. Thus, daborhagin is found in Daboia venoms only in particular geographical re-gions. Western blotting also revealed that three of theDaboia venom samples collected from southern Myanmar contained abundant daborhagins (data not shown).
To further explore the distribution of similar P-III enzymes in other viperid venoms, ten available venom samples under the generaVipera, Cerastes, and Echis were also tested. The results suggested that various P-III enzymes are simultaneously present in most viperid venoms (Fig. 3C). Notably,C. vipera venom is especially rich in P-III enzymes, whereas European viper venoms such asV. ammodytes and V. berus had low P-III levels. 3.6. cDNA cloning and the predicted sequence of
daborhagin-K
The cDNA of daborhagin-K was cloned and sequenced. The open reading frame of the daborhagin-K precursor
encoded 615 amino acid residues, including a highly con-served 18-residue signal peptide and a 171-residue proenzyme domain (data not shown). N-terminal sequences of predicted mature enzyme nicely matched that of purified daborhagin-K, which were determined as VATSERNRYFNPYSYV by au-tomatic sequencer. The deduced daborhagin-K contained 426 residues (with a calculated molecular mass of 48,041 Da) in-cluding three classic structural domains of P-IIIs (Fig. 5). Ad-ditionally, four potential N-glycosylation sites at Asn 74, 80, 189, and 339 were found, which accorded with our observa-tion of a 17-kDa mass increase due to glycosylaobserva-tion (Fig. 1C).
Table 1
Kinetic specificities of P-III SVMPs toward three synthetic oligopeptides in 0.1 M TriseHCl (pH 7.5) at 25C
Class P-III enzymes kcat/Km 103, M1s1 kcat
(s1) Km (mM1) FS-1 NFF-3 NFF-2 NFF-2 NFF-2 HH Daborhagin-M 30.6 41.2 125.0 0.88 7.0 BHRa 3.2 128.7 23.7 0.13 5.5 CVHRa 11.0 38.4 96.4 1.26 13.0 Acurhagin 18.2 41.9 34.2 0.32 9.4 NH ECLV-DM 0.4 3.8 3.2 0.05 16.1 TSV-DM 1.9 46.1 1.2 0.03 25.0 Other RVV-X 0.2 5.7 0.2 ea e
Substrate structures are shown in Section2.
a e, not determined.
Fig. 3. Detection of daborhagin-like or P-III SVMPs inDaboia and other vi-perid venoms by Western blot analysis. (A) Serological cross-reactivity of both daborhagins with anti-daborhagin-M polyclonal antibodies. (B) Analysis of eightDaboia geographical samples. W. India and E. India denote the venoms from Mumbai and Kolkata, respectively. (C) Analysis of ten venom samples from other Viperinae species. In (B) and (C), each sample containing 100 ng of soluble crude venom proteins was separated using SDSePAGE (8%) under non-reducing conditions. After blocking, the samples on blotting membranes were probed with anti-daborhagin-M polyclonal antibodies.
To verify the sequence, native and deglycosylated daborha-gin-K were digested in gel with trypsin, and the resulting pep-tides were analyzed using MALDI-TOF/TOF. The resultant 23 peptide fragments matched those predicted, and they represented more than 70% coverage of the entire sequence (Table 2). Similarly, 11 peptide fragments of daborhagin-M also matched those predicted from the digestion of daborhagin-K. Thus, daborhagin-K and -M shared a high degree of identity in their primary structures.
3.7. Molecular phylogeny and classification of P-IIIs
A phylogenetic tree was constructed based on 27 amino acid sequences of P-III SVMPs; some of them were predicted from cDNA clones and their functions have not been charac-terized. The topology of the tree reveals three major clusters of P-III enzymes (Fig. 4). Based on their MHD values (shown in parentheses), one cluster containing highly hemorrhagic P-III members is designated as the HH class, and another cluster containing mainly weakly or non-hemorrhagic enzymes is des-ignated as the NH class. The other small cluster is comprised of prothrombin activators derived fromEchis venom. Notably, daborhagin-K is associated with other strong hemorrhagins, e.g. HR1a (from Protobothrops flavoviridis) and HF3 (from Bothrops jararaca) within the HH class, but not with the P-IIIa and P-IIIc members[2,9].
3.8. Sequence alignment and comparison between the HH and NH classes
Protein sequences of daborhagin-K and the other eight P-III enzymes selected from each lineage of the phylogeny tree were aligned (Fig. 5). Their pI values, potential N-glycosyla-tion sites, numbers of Cys-residues, and classificaN-glycosyla-tion subtypes
[2,9]were also listed for comparison. Like other P-IIIs, dabo-rhagin contains 17 intramolecular disulfide bonds; its zinc-chelating motif, Met-turn (CIM), and three Ca2þbinding sites
[29]are highly conserved. In addition, its disintegrin domain contains a DECD sequence instead of the RGD or KGD found in P-II members [2]. Two hydrophobic ridges (HR) at posi-tions 332e333 and 356e357, which possibly create a novel interaction surface with the hyper-variable region (HVR) at 383e410[29], were also present.
To elucidate the molecular features responsible for distin-guishing the HH and NH classes, we further constructed their individual absolute complexity profiles (Fig. 6A). By examin-ing the superimposed topographies of both profiles, we found that most regions were similar. However, four structural ele-ments (designated as M1, M2, C1, and C2) were found to bear significantly higher absolute complexity in HH than in NH members. The consensus sequences of M1, M2, C1, and C2 in HH members were137YSPINLV143,176PVISxxPSKF186 (x represents a less-conserved residue), 353KGNY356, and
Table 2
Match between the calculated (MSc) and the experimental (MSe) masses of tryptic fragments of daborhagins
Tryptic peptides of daborhagin-K Position MSca MSe
Daborhagin-K Daborhagin-M VATSERNR 1e8 932.49 932.47 eb NRYFNPYSYVELIITVDHSMVTK 7e29 2805.39 2805.33 e YFNPYSYVELIITVDHSMVTK 9e29 2535.25 2535.22 e YKNDLTAIR 30e38 1093.60 1093.59 1093.60 IHDNSQLLTAIDLNGLTIGMAYVSTMCQSK 99e128 3326.60 3326.43 e YSVGVVQDHSKINLR 129e143 1714.92 1714.90 e YFSNCSYNQYR 185e195 1501.62 1501.75 e YFSNCSYNQYRR 185e196 1657.72 1657.79 e RFLTEHNPECIINPPLR 196e212 2106.09 2106.07 2106.06 FLTEHNPECIINPPLR 197e212 1949.99 1949.98 1949.99 TDIVSPPACGNELLER 213e228 1770.87 1770.86 1770.84 LHSWVECESGKCCNQCR 252e268 2209.90 2209.92 2209.92 RAGTECRPARDECDKAEQCTGR 271e292 2623.15 2623.17 e AGTECRPARDECDKAEQCTGR 272e292 2467.05 2467.05 2467.06 SANCPVDEFHENGRPCLHNFGYCYNGK 293e319 3242.36 3242.23 e CPIMYHQCHALFGQNVTGVQDSCFQYNR 320e347 3446.49 3446.32 e LGVYYAYCR 348e356 1164.55 1164.54 e LGVYYAYCRK 348e357 1292.65 1292.63 e ENGRKIPCAPKDEK 358e371 1584.82 1584.67 1584.68 LYCSYKSPGNQIPCLPYYIPSDENK 375e399 3006.40 e 3006.36 CGRLYCSYK 372e380 1206.54 1206.58 e SPGNQIPCLPYYIPSDENK 381e399 2192.03 2192.00 2192.00 SPGNQIPCLPYYIPSDENKGMVDHGT 381e407 3033.40 3033.27 3033.38 CGDGKVCSNGQCVDLNIAY 408e426 2129.90 2129.98 2129.93
All Cys residues have been transformed toS-carbamidomethylated. The MSccalculation was based on monoisotopic masses of amino acids, assuming the peptide
mass as [Mþ H]þ. a
The calculated masses (MSc) of tryptic peptides are based on predicted daborhagin-K sequence. b
401KYFY404 (Fig. 6B). The corresponding locations of these
four regions are shown in the 3D model of daborhagin (Fig. 6C).
4. Discussion
To address the problem of highly hemorrhagic symptoms elicited by Myanmar and eastern India Daboia envenoming
[13e15], we isolated the active P-III protease, daborhagin,
from bothDaboia venoms (Fig. 1). When subcutaneously in-jected into dorsal skin of mice, daborhagins caused severe hemorrhage with a MHD of 0.8e0.9 mg. Thus, daborhagins are highly hemorrhagic (MHD < 1 mg) toxins [30]. We also showed that daborhagin is a potent a-fibrinogenase, similar to the VaH1 hemorrhagic P-III from V. ammodytes venom
[31], but it has no effect on collagen- or ADP-induced platelet aggregation (data not shown). Daborhagin thus is able to inter-fere with the homeostatic system by degrading plasma fibrin-ogen. Furthermore, it has highly proteolytic activities toward type IV collagen and fibronectin, corroborating that the enzy-matic hydrolysis of matrix components is a key event of SVMP-induced microvessel disruption [3,30,32]. A recent study [4] reported that jararhagin (P-III class) was more po-tently hemorrhagic than BaP1 (P-I class) because it selectively cleaved key peptide bonds in mouse nidogen. Therefore,
daborhagin appears to be another good model for the biochem-ical study of hemorrhagic mechanisms induced by SVMPs.
Western blot analyses showed that only theDaboia venoms from Myanmar and eastern India have high levels of daborha-gin (Fig. 3), and, indeed, that the venom of Daboia from Myanmar contains 6e7 times more daborhagin than that from eastern India. Such results are consistent with the fact thatDaboia snakebites in both regions are particularly lethal and hemorrhagic, and that Myanmar’s viper envenoming fre-quently causes gastrointestinal or respiratory tract bleeding as well as pituitary infarction [13e15]. Thus, daborhagins are most likely critical in these syndromes. Recently, we found that the PLA2isoforms isolated from Daboia venom of both
regions share 97e100% sequence identities [17]. All these venom similarities strongly support that these two Daboia populations are closely related and represent a special lineage of this genus, which may explain why Myanmar antivenom for Russell’s viper was not effective against Thailand and Sri Lanka Russell’s viper venoms[33,34]. This is useful informa-tion for Daboia antivenom production and the clinical man-agement of Daboia envenomed victims.
Daborhagin-M and -K are orthologous genes in twoDaboia species, but they have some microheterogeneities imbedded in their sequences. First, both enzymes were isolated from Da-boia venoms using similar steps and had comparable
Fig. 4. Phylogenetic tree of P-III SVMPs based on amino acid sequences. The RVV-X heavy chain (AAB22477) was the out-group. Bootstrap values are shown at each node and MHD values are shown in parentheses. Abbreviations: FII act., prothrombin activators; HH, highly hemorrhagic class; NH, non-hemorrhagic class. The accession numbers and species are: Daborhagin-K (DQ137798),Daboia russelii; ACLD (AAC18911), Agkistrodon contortrix laticinctus; Acurhagin[18]
(AAS57937) and Acutolysin e (AAD27891),Deinagkistrodon acutus; MP-2 (AAX86634), Bitis arietans; Bothropasin (AAC61986), HF3[49](AAG48931), and Jararhagin[50](CAA48323),Bothrops jararaca; Berytractivase[6](AAL47169),Bothrops erythromelas; Catrocollastatin/VAP2B[27](AAC59672) and VAP1 [29] (BAB18307), Crotalus atrox; Ecarin [5] (Q90495), Echis carinatus; EoMP06 (AAP92424), Echis ocellatus; EcH-I (CAA55565) and EcH-II (CAA55566),Echis pyramidum leakeyi; Ohagin (ABM87941), Ophiophagus hannah; Kaouthiagin[51](P82942), Naja kaouthia; Cobrin (AAF00693), Naja naja; Mocarhagin (AAM51550), Naja mossambica; HR1a (BAB92013), HR1b[52](BAB92014), and HV1 (BAB60682),Protobothrops flavoviridis; TSV-DM
[20](ABC73079), Stejnihagin-A[9] (ABA40760), and Stejnihagin-B [9](ABA40759),Trimeresurus stejnegeri; VLFXA HC (AAQ17467), VLAIP-A[53]
hemorrhagic potencies. Second, their masses were almost identical according to SDSePAGE analysis. Third, anti-dabo-rhagin-M antiserum easily detected daborhagin-K in Western blot analysis, which indicated that they are closely related and share common epitopes. Furthermore, tryptic PMF analy-sis of daborhagin-M and -K confirmed that they both had 11 and 23 unique peptides, which nicely matched those predicted from daborhagin-K cDNA (Table 2). Moreover, theN-terminal sequence (1e25) of daborhagin-M is identical to that of dabo-rhagin-K, except for the replacement of the Pro6 with Arg6.
Upon obtaining the full primary sequence of daborhagin-M, one may clarify how many heterogeneities exist between them.
The previously reported VRR-73 (MHD¼ 0.5 mg) [16]
showed comparable masses and hemorrhagic potencies with daborhagin-K, and both were derived from eastern India D. russelii venom. Like other hemorrhagic SVMPs, daborhagin-K contains one typical zinc-chelating motif and three Ca2þ binding sites. Indeed, adding 5 mM Mg2þor Ca2þto daborha-gin-K increased the casein hydrolysis rate by 13e23%.
Fig. 5. Alignment of amino acid sequences of daborhagin-K with those of representative HH and NH members. The HH sequences are above the dashed lines, and the NH sequences are below. Residues identical to those of daborhagin-K are denoted by dots, and the gaps are marked with hyphens. PotentialN-glycosylation sites of daborhagin-K are underlined. Ca2þ-binding sites and non-conserved Cys-residues are shaded in gray and black, respectively. Conserved zinc-binding sites, Met-turns, and ECD motifs are boxed. Two hydrophobic ridges (HR) and hyper-variable regions (HVR) are also marked.
Presumably, the binding of Mg2þand Ca2þions is important for the stability and activity of daborhagin[29,35]. However, the only metal ion detected in VRR-73 was Mg2þ in a mol per mol ratio. Additionally, VRR-73 has high arginine estero-lytic activity and is strongly inhibited by PMSF[16], which suggests that some esterase might have been included in their purified VRR-73 sample.
Insulin chain B has been a model substrate for studying the cleavage specificities and active site microenvironments of the SVMPs [28]. Daborhagin hydrolyzed insulin chain B at four peptide bonds (His10eLeu11, Ala14eLeu15, Tyr16eLeu17, and Phe24ePhe25), all of which had a Leu or an aromatic res-idue at the P10 position. Moreover, the most rapid cleavage
bonds contained an Ala or Tyr at the P1 position, different Fig. 6. Analysis of the structural elements responsible for classification of HH and NH enzymes. (A) Absolute complexity profiles of HH and NH classes are represented as solid and dotted lines, respectively. Four structural motifs (M1, M2, C1, and C2) are indicated by arrows. (B) Sequence alignments of the four motifs (boxed) in both classes. TheN-glycosylation sites at Asn189 are indicated by arrows. (C) Locations of the four motifs in the 3D model of daborhagin. (Left) The backbones of three structural domains of daborhagin are marked and shown in different colors. HVR loop and zinc-binding residues in the active site are labeled in blue and gold, respectively. Catalytic zinc ion (Zn) is represented as a gold sphere. (Right) Their corresponding locations are highlighted in red in another orientation.
from those of weakly hemorrhagic Ht-c and Ht-d (P-I class), which prefer a small residue at the P1position[36]. In support
of this idea is the observation that daborhagin hydrolyzed NFF-2 (cleaving at AlaeNva) more efficiently than it did FS-1 (cleaving at GlyeLeu) or NFF-3 (cleaving at Glue Nva). Notably, daborhagin cleaved the bond between His10 and Leu11Val12, but not the bond between His5and Leu6Cys7 (-SO3H) in insulin chain-B, which suggested that negatively
charged residues at the P20 position were not favored. Taking
all these findings together, daborhagin has a substrate specific-ity toward hydrophobic and less polar amino acid residues at the P1, P10, and P20positions.
Molecular phylogeny has been a powerful tool in classify-ing venom protein families and identifyclassify-ing distinct functional subtypes. Our previous phylogenetic analyses of the P-I (and processed P-II metalloproteinases) clearly revealed three exist-ing subtypes[25]. Recently, two cladograms showed that P-III SVMPs could be divided into P-IIIa, P-IIIb, and P-IIIc sub-classes [2,9]. With more P-III sequences, our new phyloge-netic tree revealed two major subtypes (designated HH and NH classes). Consistent with its functional assays, daborhagin belongs to the HH class and is associated with several of the strongest hemorrhagins. Remarkably, there is no subclass-spe-cific seventh cysteinyl residue in the metalloproteinase do-mains of daborhagin, EcH-II, HR1a, or HF3 (Figs. 4 and 5). This led us to hypothesize that they constitute another new hemorrhagic group among the HH members in addition to P-IIIa and P-IIIc. It is noteworthy that these four P-III en-zymes and P-IIIc are more hemorrhagic than P-IIIa, and that they contain an average of four N-glycosylation sites while P-IIIa contains only one (Fig. 5). Further investigations of whether these glycosylation sites are important to hemorrhagic activity are needed. By contrast, the NH group is more versa-tile and further divided into three clusters, possibly with differ-ent functions, e.g. inducing endothelial cell apoptosis[7].
Our phylogenetic tree also suggests the co-evolution of pa-ralogous HH and NH enzymes in many venom species (Fig. 4), e.g.P. flavoviridis (HR1a, HR1b, HV1) and Crotalus atrox (Catrocollastatin/VAP2B, VAP1). Recent studies
[18,31,37,38]have also reported that at least two types of P-IIIs with different biological functions are expressed inD. acu-tus (Acurhagin, AAV1) and V. ammodytes (VaH1, VaH2, Ammodytase) venoms. These paralogous P-IIIs co-expressed in snake venoms might synergistically affect different targets and adapt to the ecology of their prey. Moreover, both P-III paralogs might have evolved before the subfamilies Viperinae and Crotalinae split.
Each member of the P-III SVMPs elicits varying degrees of hemorrhagic potency, and the responsible structural elements are still puzzling[2,39]. Since the tree topology clearly classi-fied the P-III SVMPs into two major classes with distinct hem-orrhagic potencies, the motifs or residues that distinguish these two classes in the phylogenetic analysis might also be critical in P-III hemorrhagic potencies. Based on this consideration, we identified four conserved motifs in the HH class, relative to the NH class. Of these, the M1 and M2 motifs are in the metalloproteinase domain and separated by the zinc-binding
region and the Met-turn. Both motifs form loop structures around the catalytic site (Fig. 6C). The M1 motif (positions 137e143) serves as a dimer-interface in the P-IIIb subclass
[29], and the formation of dimer probably hinders the entrance of substrates to the active site and thus restricts the enzyme specificity. The M2 motif in the HH class contains a conserved proline bracket [40], 176PVISxxP183, which has been sug-gested to form one wall of the extended substrate-binding site of SVMPs [41]. The other two motifs, C1 and C2, are in the Cys-rich domain. 3D homologous modeling showed that the C1 motif forms the hydrophobic ridge and that the C2 motif is the central part of the hyper-variable region (HVR). Both the C1 and C2 motifs have been suggested to constitute a potential proteineprotein interaction interface
[29]. By contrast, the disintegrin domains of both the HH and NH classes are similar in their absolute complexity pro-files, which agree with findings that the disintegrin domain might be merely a linker for the other two domains[29]. Col-lectively, these four structural motifs are probably important determinants for substrate interaction and the P-III hemor-rhagic potencies.
The pI values of HH-class members (average 5.5) were lower than those of NH-class members (average 7.4). By com-paring the surface electro-potential of 3D models between more than 10 representative P-III SVMPs (data not shown), we found that most HH enzymes bore more surface negative charges than NH enzymes did. The importance of this differ-ence remains to be determined. Remarkably, the N-glycosyla-tion site at Asn189 (Fig. 6B), on the C-terminal side of the M2 motif, is important for the hemorrhagic potencies of jararhagin
[42], and this site is strictly conserved in HH- but not in NH-class members.
Fluorogenic peptide substrates have been successfully used for comparing the specificities of SVMPs [36,43]. The syn-thetic tetrapeptide Abz-Ala-Gly-Leu-Ala-Nba (cleaving at GlyeLeu) has been examined as a good substrate for weakly hemorrhagic Ht-c and Ht-d toxins[36]. Additionally, a 38-kDa non-hemorrhagic metalloproteinase fromRhabdophis tigrinus tigrinus venom has high proteolytic specificities towards FS-1 (cleaving at GlyeLeu) [43]. Here, we found that the highly hemorrhagic P-III enzymes were more active toward three flu-orogenic substrates, especially NFF-2 and NFF-3 (Table 1). Further kinetic analyses suggested that the differences be-tween the HH and NH enzymes for NFF-2 are in both kcat
and Km. Three regionsdresidues 140e143, 176e181 and
182e185dwere found to be related to the active site domains of SVMPs[41,44,45]and located in M1 and M1 motifs. Re-markably, these regions within the HH members are more hy-drophobic (including Ile140, Asn141, Leu142, Val143, Val177, and Ile178) and rigid (Pro139, 176, and 183) (Fig. 6B). Presumably, by favorable interactions with hydro-phobic and bulky residues in the substrates (e.g. the P1and
P10 subsites), HH members might trigger a better induced-fit
effect and subsequent transition state stabilization to increase theirkcat. Notably, the specificity of the HH enzymes for these
fluorogenic substrates appears to be similar to that of the ver-tebrate hemorrhagic MMP3 (Stromelysin 1) [22,23], which
may activate several MMP zymogens[46]. Whether daborha-gin or other HH-class members activate endogenous MMPs or inflammatory cytokines, which may contribute to local tissue damage, requires further investigation[47,48].
In summary, we purified and characterized daborhagin, a highly hemorrhagic P-III metalloproteinase of Russell’s vi-per’s venom, and solved its sequence. Daborhagin has the high proteolytic activity toward fibrinogen, collagen, and fi-bronectin; this activity appears to be related to the severe bleeding symptoms of its envenomed victims. Moreover, a phylogenetic analysis unraveled the co-evolution of two pa-ralogous P-III classes with different hemorrhagic potencies in the venoms of Viperidae, and daborhagin apparently belongs to the new, highly hemorrhagic subclass. We identified four re-gions related to the classifications of the P-III SVMPs; detailed studies of their importance to hemorrhagic activities and func-tional diversities can now begin. Our findings not only unravel the important toxicology of hemorrhagin in specific popula-tions ofDaboia, but also provide a good basis for designing rational mutagenesis experiments to study the structure-func-tion relastructure-func-tionships of the P-III SVMPs.
Acknowledgments
We thank Professors Antony Gomes (University of Cal-cutta, India), Yu-Yen Shu (Kuangxi Medical University, China), and R. David G. Theakston (Liverpool School of Trop-ical Medicine, UK) for their gifts of venom samples, and Mr. S. Lin for proofreading the manuscript. This work was sup-ported by grants from the National Science Council and Aca-demia Sinica of Taiwan.
References
[1] H. Takeya, S. Iwanaga, Proteases that induce hemorrhage, in: G.S. Bailey (Ed.), Enzymes from Snake Venom, Alaken, Colorado, 1998, pp. 11e38. [2] J.W. Fox, S.M. Serrano, Structural considerations of the snake venom metalloproteinases, key members of the M12 reprolysin family of metal-loproteinases, Toxicon 45 (2005) 969e985.
[3] J.M. Gutierrez, A. Rucavado, T. Escalante, C. Diaz, Hemorrhage induced by snake venom metalloproteinases: biochemical and biophysical mech-anisms involved in microvessel damage, Toxicon 45 (2005) 997e1011. [4] T. Escalante, J. Shannon, A.M. Moura-da-Silva, J.M. Gutierrez, J.W. Fox, Novel insights into capillary vessel basement membrane damage by snake venom hemorrhagic metalloproteinases: a biochemical and immu-nohistochemical study, Arch. Biochem. Biophys. 455 (2006) 144e153. [5] S. Nishida, T. Fujita, N. Kohno, H. Atoda, T. Morita, H. Takeya, I. Kido,
M.J. Paine, S. Kawabata, S. Iwanaga, cDNA cloning and deduced amino acid sequence of prothrombin activator (ecarin) from KenyanEchis car-inatus venom, Biochemistry 34 (1995) 1771e1778.
[6] M.B. Silva, M. Schattner, C.R. Ramos, I.L. Junqueira-de-Azevedo, M.C. Guarnieri, M.A. Lazzari, C.A. Sampaio, R.G. Pozner, J.S. Ventura, P.L. Ho, A.M. Chudzinski-Tavassi, A prothrombin activator fromBothrops erythromelas (jararaca-da-seca) snake venom: character-ization and molecular cloning, Biochem. J. 369 (2003) 129e139. [7] I. Tanjoni, R. Weinlich, M.S. Della-Casa, P.B. Clissa, R.F.
Saldanha-Gama, M.S. de Freitas, C. Barja-Fidalgo, G.P. Amarante-Mendes, A.M. Moura-da-Silva, Jararhagin, a snake venom metalloproteinase, in-duces a specialized form of apoptosis (anoikis) selective to endothelial cells, Apoptosis 10 (2005) 851e861.
[8] L.C. Wijeyewickrema, M.C. Berndt, R.K. Andrews, Snake venom probes of platelet adhesion receptors and their ligands, Toxicon 45 (2005) 1051e1061.
[9] S.G. Wan, Y. Jin, W.H. Lee, Y. Zhang, Cloning of two novel P-III class metalloproteinases fromTrimeresurus stejnegeri venom gland, Toxicon 47 (2006) 465e472.
[10] D.A. Warrell, Snake venoms in science and clinical medicine. 1. Rus-sell’s viper: biology, venom and treatment of bites, Trans. R. Soc. Trop. Med. Hyg. 83 (1989) 732e740.
[11] W. Wuster, S. Otsuka, A. Malhotra, R.S. Thorpe, Population systematics of Russell’s viper: a multivariate study, Biol. J. Linn. Soc. 47 (1992) 97e113.
[12] R.E. Phillips, R.D.G. Theakston, D.A. Warrell, Y. Galigedara, D.T. Abeysekera, P. Dissanayaka, R.A. Hutton, D.J. Aloysius, Paralysis, rhabdomyolysis and haemolysis caused by bites of Russell’s viper (Vi-pera russelli pulchella) in Sri Lanka: failure of Indian (Haffkine) antive-nom, Q. J. Med. 68 (1988) 691e715.
[13] D.A. Warrell, Geographic and intraspecies variation in the clinical man-ifestations of envenoming by snakes, in: R.S. Thorpe, W. Wuster, A. Malhotra (Eds.), Venomous Snakes, Clarendon Press, Oxford, UK, 1997, pp. 189e204.
[14] Than-Than, R.A. Hutton, Myint-Lwin, Khin-Ei-Han, Soe-Soe, Tin-Nu-Swe, R.E. Phillips, D.A. Warrell, Haemostatic disturbances in patients bitten by Russell’s viper (Vipera russelli siamensis) in Burma, Br. J. Hae-matol. 69 (1988) 513e520.
[15] N.B. Prasad, B. Uma, S.K. Bhatt, V.T. Gowda, Comparative characterisa-tion of Russell’s viper (Daboia/Vipera russelii) venoms from different re-gions of the Indian peninsula, Biochim. Biophys. Acta 1428 (1999) 121e136.
[16] D. Chakrabarty, K. Datta, A. Gomes, D. Bhattacharyya, Haemorrhagic protein of Russell’s viper venom with fibrinolytic and esterase activities, Toxicon 38 (2000) 1475e1490.
[17] I.H. Tsai, H.Y. Tsai, Y.M. Wang, Tun-Pe, D.A. Warrell, Venom phospho-lipases of Russell’s vipers from Myanmar and eastern IndiadCloning, characterization and phylogeographic analysis, Biochim. Biophys. Acta 1774 (2007) 1020e1028.
[18] W.J. Wang, T.F. Huang, Purification and characterization of a novel met-alloproteinase, acurhagin, from Agkistrodon acutus venom, Thromb. Haemost. 87 (2002) 641e650.
[19] Y. Yamakawa, T. Omori-Satoh, D. Mebs, Hemorrhagic principles in the venom ofBitis arietans, a viperous snake. II. Enzymatic properties with special reference to substrate specificity, Biochim. Biophys. Acta 1247 (1995) 17e23.
[20] S.G. Wan, Y. Jin, W.H. Lee, Y. Zhang, A snake venom metalloproteinase that inhibited cell proliferation and induced morphological changes of ECV304 cells, Toxicon 47 (2006) 480e489.
[21] H. Kondo, S. Kondo, H. Ikezawa, R. Murata, Studies on the quantitative method for determination of hemorrhagic activity of Habu snake venom, Jpn. J. Med. Sci. Biol. 13 (1960) 43e52.
[22] C.G. Knight, F. Willenbrock, G. Murphy, A novel coumarin-labelled pep-tide for sensitive continuous assays of the matrix metalloproteinases, FEBS Lett. 296 (1992) 263e266.
[23] H. Nagase, C.G. Fields, G.B. Fields, Design and characterization of a flu-orogenic substrate selectively hydrolyzed by stromelysin 1 (matrix met-alloproteinase-3), J. Biol. Chem. 269 (1994) 20952e20957.
[24] E. Harlow, D. Lane, Immunoblotting, in: E. Harlow, D. Lane (Eds.), Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 1988, pp. 471e510.
[25] I.H. Tsai, Y.M. Wang, T.Y. Chiang, Y.L. Chen, R.J. Huang, Purification, cloning and sequence analyses for pro-metalloprotease-disintegrin vari-ants from Deinagkistrodon acutus venom and subclassification of the small venom metalloproteases, Eur. J. Biochem. 267 (2000) 1359e1367. [26] J. Felsenstein, Confidence limits on phylogenies: an approach using the
bootstrap, Evolution 39 (1985) 783e791.
[27] T. Igarashi, S. Araki, H. Mori, S. Takeda, Crystal structures of catro-collastatin/VAP2B reveal a dynamic, modular architecture of ADA-M/adamalysin/reprolysin family proteins, FEBS Lett. 581 (2007) 2416e2422.
[28] J.B. Bjarnason, J.W. Fox, Snake venom metalloendopeptidases: reproly-sins, Methods Enzymol. 248 (1995) 345e368.
[29] S. Takeda, T. Igarashi, H. Mori, S. Araki, Crystal structures of VAP1 re-veal ADAMs’ MDC domain architecture and its unique C-shaped scaf-fold, EMBO J. 25 (2006) 2388e2396.
[30] J.B. Bjarnason, J.W. Fox, Hemorrhagic metalloproteinases from snake venoms, Pharmacol. Ther. 62 (1994) 325e372.
[31] A. Leonardi, F. Gubenek, I. Krizaj, Purification and characterisation of two hemorrhagic metalloproteinases from the venom of the long-nosed viper,Vipera ammodytes ammodytes, Toxicon 40 (2002) 55e62. [32] E.N. Baramova, J.D. Shannon, J.B. Bjarnason, J.W. Fox, Degradation of
extracellular matrix proteins by hemorrhagic metalloproteinases, Arch. Biochem. Biophys. 275 (1989) 63e71.
[33] Win-Aung, N. Pakmanee, O. Khow, C. Nambut, V. Sitprija, Cross neutral-ization of the lethal activities of Myanmar and Thai Russell’s viper venoms by Thai and Myanmar antivenoms, J. Nat. Toxins 10 (2001) 335e342. [34] B.J. Woodhams, S.E. Wilson, B.C. Xin, R.A. Hutton, Differences
be-tween the venoms of two sub-species of Russell’s viper:Vipera russelli pulchella and Vipera russelli siamensis, Toxicon 28 (1990) 427e433. [35] F.X. Gomis-Ruth, L.F. Kress, W. Bode, First structure of a snake venom
metalloproteinase: a prototype for matrix metalloproteinases/collage-nases, EMBO J. 12 (1993) 4151e4157.
[36] J.W. Fox, R. Campbell, L. Beggerly, J.B. Bjarnason, Substrate specific-ities and inhibition of two hemorrhagic zinc proteases c and Ht-d fromCrotalus atrox venom, Eur. J. Biochem. 156 (1986) 65e72. [37] W.J. Wang, Purification and functional characterization of AAV1, a novel
P-III metalloproteinase, from FormosanAgkistrodon acutus venom, Bio-chimie 89 (2007) 105e115.
[38] A. Leonardi, J.W. Fox, A. Trampus-Bakijac, I. Krizaj, Ammodytase, a metalloprotease fromVipera ammodytes ammodytes venom, possesses strong fibrinolytic activity, Toxicon 49 (2007) 833e842.
[39] O.H. Ramos, H.S. Selistre-de-Araujo, Comparative analysis of the cata-lytic domain of hemorrhagic and non-hemorrhagic snake venom metallo-peptidases using bioinformatic tools, Toxicon 44 (2004) 529e538. [40] R.M. Kini, H.J. Evans, A novel approach to the design of potent bioactive
peptides by incorporation of proline brackets: antiplatelet effects of Arg-Gly-Asp peptides, FEBS Lett. 375 (1995) 15e17.
[41] D. Zhang, I. Botos, F.X. Gomis-Ruth, R. Doll, C. Blood, F.G. Njoroge, J.W. Fox, W. Bode, E.F. Meyer, Structural interaction of natural and syn-thetic inhibitors with the venom metalloproteinase, atrolysin C (form d), Proc. Natl. Acad. Sci. USA 91 (1994) 8447e8851.
[42] L.T. Garcia, L.T. Parreiras e Silva, O.H.P. Ramos, A.K. Carmona, P.A. Bersanetti, H.S. Selistre-de-Araujo, The effect of post-translational
modifications on the hemorrhagic activity of snake venom metalloprotei-nase, Comp. Biochem. Physiol. Part C 138 (2004) 23e32.
[43] K. Komori, M. Konishi, Y. Maruta, M. Toriba, A. Sakai, A. Matsuda, T. Hori, M. Nakatani, N. Minamino, T. Akizawa, Characterization of a novel metalloproteinase in Duvernoy’s gland ofRhabdophis tigrinus ti-grinus, J. Toxicol. Sci. 31 (2006) 157e168.
[44] K.F. Huang, S.H. Chiou, T.P. Ko, A.H.J. Wang, Determinants of the hibition of a Taiwan habu venom metalloproteinase by its endogenous in-hibitors revealed by X-ray crystallography and synthetic inhibitor analogues, Eur. J. Biochem. 269 (2002) 3047e3056.
[45] W. Gong, X. Zhu, S. Liu, M. Teng, L. Niu, Crystal structures of acuto-lysin A, a three-disulfide hemorrhagic zinc metalloproteinase from the snake venom ofAgkistrodon acutus, J. Mol. Biol. 283 (1998) 657e668. [46] R. Visse, H. Nagase, Matrix metalloproteinases and tissue inhibitors of metalloproteinases: structure, function, and biochemistry, Circ. Res. 92 (2003) 827e839.
[47] A.M. Moura-da-Silva, G.D. Laing, M.J. Paine, J.M. Dennison, V. Politi, J.M. Crampton, R.D.G. Theakston, Processing of pro-tumor necrosis fac-tor-alpha by venom metalloproteinases: a hypothesis explaining local tis-sue damage following snake bite, Eur. J. Immunol. 26 (1996) 2000e2005. [48] A. Rucavado, J. Nu´~nez, J.M. Gutie´rrez, Blister formation and skin dam-age induced by BaP1, a haemorrhagic metalloproteinase from the venom of the snakeBothrops asper, Int. J. Exp. Pathol. 79 (1998) 245e254. [49] M.T. Assakura, A.P. Reichl, F.R. Mandelbaum, Comparison of
immuno-logical, biochemical and biophysical properties of three hemorrhagic fac-tors isolated from the venom of Bothrops jararaca (JARARACA), Toxicon 24 (1986) 943e946.
[50] M.J. Paine, H.P. Desmond, R.D. Theakston, J.M. Crampton, Purification, cloning, and molecular characterization of a high molecular weight hem-orrhagic metalloprotease, jararhagin, fromBothrops jararaca venom, J. Biol. Chem. 267 (1992) 22869e22876.
[51] J. Hamako, T. Matsui, S. Nishida, S. Nomura, Y. Fujimura, M. Ito, Y. Ozeki, K. Titani, Purification and characterization of kaouthiagin, a von Willebrand factor-binding and -cleaving metalloproteinase from Naha kaouthia cobra venom, Thromb. Haemost. 80 (1998) 499e505. [52] H. Takeya, K. Oda, T. Miyata, T. Omori-Satoh, S. Iwanaga, The complete
amino acid sequence of the high molecular mass hemorrhagic protein HR1B isolated from the venom of Trimeresurus flavoviridis, J. Biol. Chem. 265 (1990) 16068e16073.
[53] K. Trummal, K. T~onisma¨gi, E. Siigur, A. Aasp~ollu, A. Lopp, T. Sillat, R. Saat, L. Kasak, I. Tammiste, P. Kogerman, N. Kalkkinen, J. Siigur, A novel metalloprotease fromVipera lebetina venom induces human en-dothelial cell apoptosis, Toxicon 46 (2005) 46e61.
