ARTICLE IN PRESS

0041-0101/$ - see

doi:10.1016/j.tox

�CorrespondiColombo 2265,

E-mail addre

Toxicon 51 (2008) 723–735

www.elsevier.com/locate/toxicon

Crystal structure of a novel myotoxic Arg49 phospholipaseA2 homolog (zhaoermiatoxin) from Zhaoermia mangshanensissnake venom: Insights into Arg49 coordination and the role

of Lys122 in the polarization of the C-terminus

Mario T. Murakamia, Ulrich Kuchb, Christian Betzelc,Dietrich Mebsb, Raghuvir K. Arnia,d,�

aCenter for Structural & Molecular Biology, Department of Physics, IBILCE/UNESP, R. Cristovao Colombo 2265,

Sao Jose do Rio Preto, Sao Paulo CEP 15054-000, BrazilbZentrum der Rechtsmedizin, Klinikum der Johann Wolfgang Goethe-Universitat, Kennedyallee 104, D-60596 Frankfurt am Main, GermanycInstitute of Biochemistry and Molecular Biology, University of Hamburg, Notkestrasse 85, c/o DESY, Build 22a, 22603, Hamburg, Germany

dCenter for Applied Toxinology, CAT-CEPID, Sao Paulo, SP, Brazil

Received 11 September 2007; received in revised form 19 November 2007; accepted 19 November 2007

Available online 4 March 2008

Abstract

The venom of Zhaoermia mangshanensis, encountered solely in Mt Mang in China’s Hunan Province, exhibits

coagulant, phosphodiesterase, L-amino acid oxidase, kallikrein, phospholipase A2 and myotoxic activities. The

catalytically inactive PLA2 homolog referred to as zhaoermiatoxin is highly myotoxic and displays high myonecrotic

and edema activities. Zhaoermiatoxin possesses a molecular weight of 13,972Da, consists of 121 amino-acid residues cross-

linked by seven disulfide bridges and shares high sequence homology with Lys49-PLA2s from the distantly related Asian

pitvipers. However, zhaoermiatoxin possesses an arginine residue at position 49 instead of a lysine, thereby suggesting a

secondary Lys49-Arg substitution which results in a catalytically inactive protein. We have determined the first crystal

structure of zhaoermiatoxin, an Arg49-PLA2, from Zhaoermia mangshanensis venom at 2.05 A resolution, which represents

a novel member of phospholipase A2 family. In this structure, unlike the Lys49 PLA2s, the C-terminus is well ordered and

an unexpected non-polarized state of the putative calcium-binding loop due to the flip of Lys122 towards the bulk solvent

is observed. The orientation of the Arg-49 side chain results in a similar binding mode to that observed in the Lys49 PLA2s;

however, the guadinidium group is tri-coordinated by carbonyl oxygen atoms of the putative calcium-binding loop,

whereas the Nz atom of lysine is tetra-coordinated as a result of the different conformation adopted by the putative

calcium-binding loop.

r 2008 Elsevier Ltd. All rights reserved.

Keywords: Zhaoermia mangshanensis; Snake venom; Arg49-phospholipase A2; Lys49-phospholipase A2; Myotoxin; Crystal structure

front matter r 2008 Elsevier Ltd. All rights reserved.

icon.2007.11.018

ng author at: Center for Structural & Molecular Biology, Department of Physics, IBILCE/UNESP, R. Cristovao

Sao Jose do Rio Preto, Sao Paulo CEP 15054-000, Brazil. Tel.: +5517 32212460; fax: +55 17 32212247.

ss: arni@ibilce.unesp.br (R.K. Arni).

ARTICLE IN PRESSM.T. Murakami et al. / Toxicon 51 (2008) 723–735724

1. Introduction

Envenoming from snakebites is a major neglectedproblem of the 21st century. Besides numerousfatalities, it results in a significantly large number ofvictims who survive with permanent physical andpsychological sequelae mostly due to the local tissue-damaging effects of snake venoms (Gutierrez et al.,2006). Among the three distinct groups of snakevenom proteins that are responsible for the directinduction of myonecrosis, the most important onesfrom pitviper venoms are group II phospholipase A2

(PLA2, EC 3.1.1.4) myotoxins. These have beendivided into two main subgroups based primarily onthe presence of a conserved aspartic acid residue atposition 49 in the catalytically active Asp49-PLA2

enzymes, or a lysine in homologues (Lys49-PLA2s;reviewed in Arni and Ward, 1996; Ownby et al., 1999;Murakami and Arni, 2003) that are considered tobe catalytically inactive due to their inability tobind the cofactor Ca2+, which in turn prevents thestabilization of the tetrahedral intermediate observedin the Ca2+-dependent catalytic reaction promotedby Asp49-PLA2s (Van den Berg et al., 1989).

Although catalytically inactive, Lys49-PLA2s aretruly multi-functional venom proteins that are highlyexpressed in pitviper venoms. They display inflam-matory properties and induce edema, an activity forwhich alternative mechanisms have been proposed inthe absence of phospholipid hydrolysis or mobiliza-tion of arachidonic acid (Lomonte et al., 1993,1994a, b; Teixeira et al., 2003; Zuliani et al., 2005),and exhibit broad antibacterial activities (Paramoet al., 1998; Santamarıa et al., 2005). Recently, Lys49-PLA2s have been implicated in the inhibition of thevascular endothelial growth factor and its receptorsystem, which plays a central role in angiogenesis(Yamazaki et al., 2005). However, from a clinicalpoint of view, the highly potent induction ofmyonecrosis by Lys49-PLA2s is most significant.This important phenomenon, its mechanisms andtheir possible inhibition have received consider-able attention. For example, strategies to elucidatethe structural determinants for the myotoxicity ofLys49-PLA2s have included chemical modification(Andriao-Escarso et al., 2000), sequence comparison(Selistre de Araujo et al., 1996; Ward et al., 1998),charge distribution (Kini and Iwanaga, 1986; Kiniand Evans, 1989), hydrophobicity profile (Kini andIwanaga, 1986), synthetic peptide (Lomonte et al.,1994b; Nunez et al., 2001) and site-directed mutagen-esis (Ward et al., 1995, 2002) studies, and compounds

such as suramin, heparin, heparin-like glycosamino-glycans, related polyanions and polyethylene glycolderivatives have been used to effectively inhibit theactivity of myotoxic Lys49-PLA2s both in vitro andin vivo (Murakami et al., 2005, 2007).

Apart from the large group of Lys49-PLA2s frompitvipers, a few other variants replacing Asp49 atthis generally highly conserved position in snakevenom group II PLA2s have been reported, e.g.,Ser49-PLA2s from the venoms of two true vipers(Krizaj et al., 1991; Polgar et al., 1996) and Asn49-PLA2s from two Asian pitvipers (Pan et al., 1998;Tsai et al., 2004) that probably originated from atleast two independent Asp49-Asn substitutionswithin the subgroup of basic Arg6-Asp49-PLA2s(Mebs et al., 2006). Recently, a novel Arg49 PLA2

analog (zhaoermiatoxin, Mebs et al., 2006) wasreported from the venom of the Mt. Mang Viper,Zhaoermia mangshanensis (Zhao and Chen, 1990), arare pitviper known only from a single mountainrange in China’s Hunan Province. This Arg49-PLA2

is the major toxin of Z. mangshanensis (formerlyTrimeresurus mangshanensis, Ermia mangshanensis;see Gumprecht and Tillack, 2004) and inducesedema and myonecrosis in mice in the absence ofcatalytic activity upon phospholipids. Sequencecomparison of zhaoermiatoxin reveals that thenovel toxin possesses most of the residues of theLys49-PLA2 subgroup, displays more than 80%identity to two Lys49-PLA2s from two distantlyrelated Asian pitvipers, and is rooted within acomprehensive sample of Lys49-PLA2s in phyloge-netic analyses (Mebs et al., 2006). In this context,zhaoermiatoxin as a member of the Lys49-PLA2

subgroup, and its Arg49 substitution as a secondaryLys49-Arg substitution does not alter its catalyticinactivity.

We present the first crystal structure of anArg49 phospholipase A2 homolog isolated fromZ. mangshanensis venom at 2.05 A resolution, whichpossesses a well-ordered C-terminus and an unex-pected non-polarized state of the putative calcium-binding loop tetra-coordinating by guanidiniumgroup of Arg49 and with Lys122 oriented towardsthe bulk solvent.

2. Materials and methods

2.1. Crystallization and X-ray data collection

The purification of zhaoermiatoxin was carriedout using gel-filtration chromatography followed by

ARTICLE IN PRESSM.T. Murakami et al. / Toxicon 51 (2008) 723–735 725

cation-exchange chromatography following thepublished protocol (Mebs et al., 2006). The purityof samples was confirmed by SDS–PAGE (Laemm-li, 1970) and protein concentration was determinedby the Bradford method (Bradford, 1976).

Prior to the crystallization trials, the sample wassubmitted to dynamic light scattering experimentsusing a DynaPro 810 (Protein Solutions) apparatusequipped with a temperature stabilizer, whichdemonstrated a monomodal distribution indicatingstructural homogeneity considered adequate forcrystallization experiments.

Crystallization was performed by the hangingdrop vapor-diffusion method as described pre-viously (Murakami et al., 2007). Typically, 1 ml ofthe protein solution was mixed with an equalvolume of screening solution and equilibrated over0.6ml of the latter in the reservoir solution. Largesingle crystals were obtained when a 1.5 ml proteindroplet was mixed with an equal volume of reservoirsolution consisting of 0.2M ammonium sulfate,0.1M sodium cacodylate trihydrate (pH 6.5) and30% polyethylene glycol 8000. X-ray diffractiondata were collected at the DB03-MX1 Beamline atthe Laboratorio Nacional de Luz Sıncrotron(LNLS, Campinas, Brazil). The crystal was trans-ferred to a cryoprotectant solution containing 10%glycerol and flash-frozen in a gaseous nitrogenstream at 100K. The wavelength of the radiationsource was set to 1.427 A and a MARCCD detectorwas used to record the X-ray diffraction data. Thedata were indexed, integrated and scaled using theDENZO and SCALEPACK programs, (Otwinowskiand Minor, 1997).

2.2. Structure determination, refinement and analysis

The crystal structure of the zhaoermiatoxin wassolved by molecular replacement techniques withthe AMoRe package (Navaza, 1994) using theatomic coordinates of the Basp-II (PDB code1Y4L) (Murakami et al., 2005) as the search model.Positional and individual isotropic thermal factorrefinements were carried out using REFMAC5(Murshudov et al., 1997) as incorporated in theCCP4 suite. The 2Fo–Fc and Fo–Fc electron densitymaps were examined; the protein model wasmanually adjusted after each refinement cycle usingTURBO FRODO (Jones, 1985) and COOT (Emsleyand Cowtan, 2004). The overall stereochemistry ofthe final model was analyzed using PROCHECK(Laskowski et al., 1993). The pictures were generated

using PyMOL (DeLano, 2002). The atomic coordi-nates and structure factors of the zhaoermiatoxinhave been deposited with the Protein Data Bank(accession code 2PH4).

2.3. Molecular modeling and quality analysis

The atomic coordinates of zhaoermiatoxin (PDBaccession code: 2PH4, this work) was used as theinitial model for obtaining a structural model ofanother Arg49 PLA2 from Protobothrops (formerlyTrimeresurus) elegans venom (PeBP-II) based onrestraint-based modeling as implemented in theMODELLER program (Fiser and Sali, 2003a).The overall model was improved enforcing theproper stereochemistry using spatial restraints andCHARMM energy terms, followed by conjugategradient simulation based on the variable targetfunction method (Fiser and Sali, 2003a). The loopswere optimized using ModLoop (Fiser and Sali,2003b) based on the satisfaction of spatial re-straints, without relying on a database of knownprotein structures. The final model was submittedunder explicit solvent molecular dynamics (MD)simulations using the GROMACS 3.2 softwarepackage (Lindahl et al., 2001) and the GROMOS-96 (43a1) force field (van Gunsteren et al., 1996) tocheck the stability and consistency of the finalmodel. The MD simulations were performed in anisothermal–isobaric ensemble, using the ‘‘leapfrog’’algorithm (Hockney and Goel, 1974) to integratethe equations of motion with a time step of 2.0 fsover a total time of 4.0 ns. The LINCS algorithm(Hess et al., 1997) was used to constrain proteincovalent bonds involving hydrogen atoms, andthe SETTLE algorithm (Miyamoto and Kollman,1992) was used to constrain water molecules. Theoverall and local quality analyses of the final modelwere assessed by VERIFY3D (Eisenberg et al.,1997), PROSA (Wiederstein and Sippl, 2007) andVADAR (Willard et al., 2003). Three-dimensionalstructures were displayed, analyzed and comparedusing the program COOT (Emsley and Cowtan,2004).

3. Results and discussion

3.1. Structure of zhaoermiatoxin

The crystal structure of zhaoermiatoxin deter-mined at 2.05 A resolution contains a dimer inthe asymmetric unit. The two molecules in the

ARTICLE IN PRESS

Table 1

Data collection and structure determination statistics

Beamline setup

Radiation source Synchrotron light

Beamline CPr, LNLS-Brazil

Wavelength (A) 1.437

Temperature 100

Detector MARCCD 165mm

Crystal preparation

Cryoprotectant solution Mother liquor+20%

glycerol

Soaking time (s) 25

Data collection statistics

Space group P64

Unit-cell parameters (A) a ¼ b ¼ 72.93,

c ¼ 93.94

Resolution range (A) 23.9–2.05 (2.12–2.05)

Matthews coefficient (A3/Da) 2.32

Corresponding solvent (%) 46.60

Number of molecules in asymmetric

unit

2

Observed reflections 305,151

Unique reflections 17,562

Average redundancy 19.5 (18.0)

Completeness (%) 99.9 (99.8)

I/s/IS 33.1 (6.9)

Rmerge 0.071 (0.332)

Refinement statistics

Resolution range (A) 23.9–2.05

Number of unique reflections used in

refinement

15771

Number of test reflections 1791

Rfactor for all reflections 0.230

Rfactor without free-reflections 0.227

Rfree 0.291

Number of atoms

Protein 1984

Water 235

R.m.s. bond-length deviation (A) 0.023

R.m.s. bond-angle deviation (%o) 2.218

Mean B factor (A2) 29.07

Ramachandran plot analysis

Most favored regions (%) 92

Allowed regions (%) 8

Disallowed regions (%) 0

Values in parentheses are for the highest resolution shell.

M.T. Murakami et al. / Toxicon 51 (2008) 723–735726

asymmetric unit are related by a two-fold axis ofrotation, and the hydrophobic surfaces surroundingthe entrance to the PLA2-active site-like regionsform the dimer interface, resulting in a large buriedarea. During the refinement residual electrondensity peaks above 2s in the Fourier differencemaps were observed in the hydrophobic channelsthat lead to the PLA2-active site-like region andhave been assigned to belong to a fragment ofpolyethylene glycol present in the crystallizingsolution. Other Lys49-PLA2 homolog crystal struc-tures have been solved with a fragment of poly-ethyelene glycol interacting with His48Nd1(Watanabe et al., 2005; Murakami et al., 2007).Additional residual electron density with tetrahedralgeometry around the putative calcium-binding loopwas identified as sulfate ions bound to the arginineresidues located in this region of the moleculeincluding Arg49.

The refinement converged to a crystallographicresidual of 20.3% (Rfree ¼ 28.4%) for all databetween 23.90 and 2.05 A (Table 1) and the finalmodel consists of the 242 amino-acid residues(homodimer, each protein molecule of 121 residues),235 solvent molecules, two fragments of PEG 4000and three sulfate ions. The model displays goodoverall stereochemistry with RMSD values of0.022 A and 2.2181 for bond lengths and bondangles, respectively. The average temperature factor(B-value) for all atoms is 29.1 A2 (Table 1). TheRamachandran plot demonstrates that 92.0% of thedihedral angles are situated in the most favoredregions, while the remaining 8.0% of residues are inthe additionally permitted regions and no residuesare found in the generously allowed and disallowedregions (Table 1).

The molecular topology of zhaoermiatoxin con-serves all the main features of Lys49-PLA2s contain-ing an N-terminal a-helix (H1) (residues 2–12) andthe two long anti-parallel disulfide linked a-helices(H2 from residues 40 to 55 and H3 from residues 86to 103) with a mean distance of 10 A between thehelical axes and two short helical turns (residues19–22; SH4 and 108–110; SH5) (Arni and Ward,1996) (Fig. 1). The b-wing region (residues 74–84) isstructurally conserved and a disulfide bridge pre-serves its relative orientation. The positions of theamino-acid residues that form the putative catalyticapparatus (His48, Tyr52, Tyr73 and Asp99, includ-ing the catalytic water molecule) are conservedexcept for Asp49, which is substituted, in this caseby arginine.

Typically the C-terminal region comprising theresidues Lys115–Lys132 are inherently flexible asobserved in several Lys49-PLA2 homolog crystalstructures (Arni and Ward, 1996; Murakami andArni, 2003); however, the zhaoermiatoxin structuredisplays a very well-ordered C-terminus with sidechains well defined in the electron density mapprobably due to the novel conformation adopted by

ARTICLE IN PRESSM.T. Murakami et al. / Toxicon 51 (2008) 723–735 727

the Lys122 side chain oriented towards the bulksolvent, which has hitherto not been observed(Figs. 1 and 2). The putative calcium-binding loopalso displays a novel architecture due to the lack ofhyperpolarization induced by Lys122, the presenceof large side chains from the residues, whichcomprise these loops and the tri-coordinated stateof Arg49 by the main-chain atoms of the putativecalcium-binding loop.

3.2. Molecular modeling of PeBP(R)-II

PeBP-II, another Arg49-PLA2 analogue fromP. elegans venom, displays high sequence homologywith the Lys49-PLA2 subgroup and identity up to90% with zhaoermiatoxin (Chijiwa et al., 2006). Theatomic coordinates of zhaoermiatoxin (PDB code2PH4, this work) solved to a resolution of 2.05 Awere used as starting model for modeling PeBP-II.The overall and local quality of the final model ofPeBP-II was assessed by analyzing its stereochemi-cal properties using VERIFY3D (Eisenberg et al.,1997), PROSA (Wiederstein and Sippl, 2007) andVADAR (Willard et al., 2003) and by comparison

Fig. 1. Results of the super positioning of the crystal structures of zhaoe

in green (zhaoermiatoxin, Arg49 PLA2) and blue (Basp-II, Lys49 PLA

with other crystallographic structures. The analysisof the Ramachandran diagram f–c plots generatedby VADAR of the PeBR-II modeled structureindicates 94% of the residues in the most favorableregion and 6% in a additional allowed region with acalculated accessible surface area of 6748.4 A2 and76% of the residues form hydrogen bonds. Overallmodel quality assessed by PROSA resulted in aZ-score of �5.99, which is in agreement with theZ-scores of all experimentally determined proteinchains currently deposited with the Protein DataBank (Fig. 3a). Local quality analysis assessed byplotting energies as a function of amino-acid sequenceposition using PROSA and VERIFY3D resulted inno positive values for all residues in the energy plot,which indicates no problematic or erroneous partsof the structure (Fig. 3b). These careful analysesstrongly indicate that the molecular models presentgood overall stereochemical quality and are suitablefor structural analysis and comparisons. The modeledstructure conserves all structural features presented inthe zhaoermiatoxin structure with no significantstructural changes. The modeled structure has beenused for surface charge distribution analysis.

rmiatoxin and Basp-II. The regions with altered conformation are

2).

ARTICLE IN PRESS

Fig. 2. Stereo view of the hydrogen bond network in the putative calcium-binding site. The main-chain atoms of zhaoermiatoxin and

Basp-II are in green and gray, respectively. The side chains of residues discussed in the text and non-conserved residues are included.

Fig. 3. (a) Graphic representation of the overall model quality parameters (Z-score) of all the structures deposited with the Protein Data

Bank (diffuse blue region) calculated by PROSA. The Z-score of the PeBR-II model is represented by a black dot. (b) Local model

quality—energy as a function of residue number of the PeBR-II model calculated by PROSA.

M.T. Murakami et al. / Toxicon 51 (2008) 723–735728

3.2.1. Arg/Lys/Asp49 coordination and nominal

active sites

Despite the fact that apparent sequence andconformational dissimilarities exist between the

D49 PLA2 and their respective subgroup Lys49and Arg49-PLA2 homolog, three functionally im-portant residues remain absolutely conserved withrespect to type and relative positions. The side

ARTICLE IN PRESSM.T. Murakami et al. / Toxicon 51 (2008) 723–735 729

chains of His48, Tyr73 and Asp99 in Lys/Arg49PLA2 homolog exhibit a hydrogen bond networkidentical to those of pancreatic and venom enzymes(Dijkstra et al., 1981; Keith et al., 1981; Brunieet al., 1985). The precise preservation of this clusterof amino acids in Lys/Arg49 PLA2s and their abilityof binding fatty acids (Lee et al., 1999; Ambrosioet al., 2005), stearic acid (Watanabe et al., 2005) andPEG (Murakami et al., 2005, 2007) at the nominalcatalytic site due to the constellation of conservedhydrophobic residues that serve to aid substratebinding seem to support the importance of theseresidues in the function of this protein.

In the case of the catalytically active Asp49PLA2s, His48 and Asp99 together with a structu-rally conserved water molecule (hydrogen bondedto His48 Nd1) participate in the nucleophilic attackat the sn-2 position of the phospholipid substrate(Scott et al., 1990). The tetrahedral transition-stateintermediate is stabilized by a calcium ion and iscoordinated by Asp49 and the main-chain atomslocated in the calcium-binding loop (White et al.,1990).

In the subgroup of Lys49 PLA2 analog, substitu-tion of Asp49 by Lys results in the Nz atom ofLys49 occupying the position of the calcium ion inAsp49 PLA2s, tetra-coordinated by carbonyl oxygenatoms from residues located at positions 28, 30, 32and 33 (Arni et al., 1995; Murakami et al., 2005)(Fig. 2). Moreover, the controversy as to whetherthese enzymes possess catalytic activity has beenclarified, since no substrate hydrolysis was detectedwith the wild-type recombinant protein (Ward et al.,2002).

Particularly in the subgroup of Arg49 PLA2

analog, the secondary substitution of Lys49 byArg results in a novel geometrical architecture of theputative calcium-binding loop where the guadini-dium group is tri-coordinated by just three carbonyloxygen atoms from residues located at positions 28,30 and 33 (Fig. 2). Specifically, the NZ1 atom is di-coordinated by carbonyl oxygens from Asn28and Arg33 backbones, whereas the NZ2 atom ismono-coordinated by carbonyl oxygen from Gly30(Fig. 2). In the radius of 3.5 A around the side chainof Arg49 no water molecules were encountered.

3.2.2. Polarization of the putative calcium-binding

loop and action mechanism

A generic mechanism for Lys49-PLA2s based onthe stabilization of a canonical structure within theC-terminal region of the molecule that is favored by

the presence of a long chain ligand within the activesite/hydrophobic channel, referred to as the ‘‘hydro-phobic knuckle’’, has been proposed (Ambrosioet al., 2005). In this model, the hyperpolarizationof the peptide bond between residues Cys29 andGly30 by the presence of Lys122 that is fullyconserved among myotoxins, including the Arg49PLA2 subgroup, causes local rearrangement of theC-terminal residues around Lys122 that becomelocked into the canonical conformation, leading tothe exposure of the hydrophobic knuckle, Phe121

and Phe124 (conserved in Arg49 PLA2s analogues),and the associated positive charge at its base(Lys115, Lys116, Lys128 and Lys129, also conservedin Arg49 PLA2 analogues). This conformation isconsidered ideal for membrane insertion, where theknuckle would interact with the hydrocarbon chainsof phospholipids within the outer leaflet and thebasic residues with the acidic polar head groups.This hyperpolarized state has been considered to beinduced by ligand binding.

However, in the crystal structure of zhaoermia-toxin neither is the peptide bond Cys29-Gly30hyperpolarized, even in the presence of a ligandbound at the nominal active site (fragment of a PEGmolecule) nor are the two hydrophobic residues,Phe121 and Phe124, exposed to the solvent (Figs. 1and 2). These observations indicate that the hydro-phobic knuckle (Ambrosio et al., 2005) does notplay a role in the myotoxicity of Arg49-PLA2

homolog and can be extended to encompass theclosely related Lys49-PLA2 subgroup.

3.2.3. PEG-binding site

During structure refinement, electron densityabove the 2s level in the 2Fo–Fc electron densitymap was observed in both the monomers at thehydrophobic channel that leads to the PLA2-activesite-like region. Considering its location, size,geometry and the compounds used for crystal-lization, this residual electron density was modeledas a fragment of a PEG-400 molecule. In bothmonomers, the PEG-400 molecules interact withthe His48Nd1 atom via a water molecule and thehydrophobic tail, and the oxy-terminus interactswith Thr23Og1 and Pro13O atoms (Fig. 4).

Similar interactions were observed in the crystalstructure of piratoxin II, a Lys49 PLA2 fromBothrops pirajai, which contains a fatty acid boundat the entrance to the active-site cavity (Lee et al.,2001) and in the crystal structure of Bothrops asper

myotoxin-II complexed with stearic acid (Watanabe

ARTICLE IN PRESS

Fig. 4. PEG-binding site. Hydrogen bonds formed between the fragment of the PEG molecule and putative calcium-binding loop and

His48 residue. The blue cage represents the electron density contoured at 2.0sU

M.T. Murakami et al. / Toxicon 51 (2008) 723–735730

et al., 2005). The crystal structure of Basp-IIcomplexed with suramin also contains a PEGmolecule bound at the active-site cavity concomi-tantly with suramin (Murakami et al., 2005) boundat the interfacial recognition face. Recently, thecrystal structure of bothropstoxin-I, a basic myo-toxin from Bothrops jararacussu venom with a PEGmolecule also bound at the PLA2-active site-likeregions (Murakami et al., 2007), has been reported.Previously, fatty acid-like molecules encountered inthe hydrophobic channels of Lys49-PLA2s analoghave been interpreted as a transient fatty acylintermediate in the catalytic cycle that leads to therelease of fatty acid in the next step, which has failed(Lee et al., 2001). However, new crystal structuresindicate that this characteristic affinity for fatty acidis the result of the conserved motif at the entrance ofthe PLA2-active site-like region in Lys49-PLA2sanalog.

3.2.4. Anion-binding site

Two electron density peaks above the 2s levelexhibiting tetrahedral geometry around the PLA2-

active site-like region, which is surrounded by basicresidues in both the 2Fo–Fc and Fo–Fc maps wereobserved. Since the crystallization experiments con-tained ammonium sulfate, these electron densities havebeen attributed to bound sulfate ions. The sulfate ionsare located close to the putative calcium-binding loopand to the residues considered important for expres-sion of its pharmacological properties mimicking theanionic surface of biological membranes.

One sulfate is bound to Arg34 and Arg49 formingfour interactions between Arg34NZ2 atom withsulfate O2 atom (distance 3.16 A), Arg34NZ1 atomwith sulfate O4 atom (distance 3.41 A), Arg49NZ2atom with sulfate O1 atom (distance 3.10 A) andArg49Ne atom with sulfate O2 atom (distance2.77 A) (Fig. 5). The other is bound to Arg34and Lys53 forming three hydrogen bonds betweenArg34Ne atom with sulfate O2 atom (distance2.64 A), Arg34Nh2 atom with sulfate O3 atom(distance 3.24 A) and LysNz atom with sulfate O3atom (distance 2.93 A) (Fig. 5).

This anion-binding site is well characterized andseveral crystal structures of Lys49-PLA2s have been

ARTICLE IN PRESS

Fig. 5. Hydrogen bond and ionic interactions in the anion-binding site. The interactions involving Arg34 and Arg49 are in red and black,

respectively. The sulfate ions are presented in orange and red. The blue cages around the sulfate ions represent the electron density

contoured at 2.0sU

M.T. Murakami et al. / Toxicon 51 (2008) 723–735 731

reported in the presence of either sulfate ions oranionic compounds bound to Arg34 and adjacentregions (Ambrosio et al., 2005; Murakami et al.,2005). This motif for binding of anionic compoundsis strictly conserved in all Lys49-PLA2s analog andalso in the Arg49-PLA2 subgroup and is likely toplay a crucial role in the binding of negativelycharged substrates or inhibitors.

Superposition of the structure of an Asp49-PLA2

from bee venom complexed with L-1-O-octyl-2-heptylphosphonyl-sn-glycero-3-phosphoethanolamine,a transition-state analogue (PDB code 1POB (Whiteet al., 1990), indicates that the position occupiedby the sulfate ions located in zhaoermiatoxin isequivalent to that occupied by the phosphate groupof the phospholipid. This would imply that thezhaoermiatoxin structure with bound sulfate corre-sponds to states that are compatible with thebinding of the head group of phospholipids via

specific interactions. This site is also involved insuramin recognition, a polysulfonated compoundused in the treatment of onchocerciasis and

trypanosomiasis, (Murakami et al., 2005), andprobably also plays a role in heparin and heparinoidbinding. Interestingly, the C-terminus that has beenpotentially considered a site where membranephospholipid head groups might bind is notinvolved in recognition or binding of anioniccompounds as expected by spectroscopic andbiochemical analyses for Lys49-PLA2s analog.

3.2.5. Charge distribution analysis

The interfacial face or i-face, which is involved inthe binding on anionic surfaces principally, encom-passes the N-terminal helix, the short helix, theputative calcium-binding loop, the elapid loop,the b-wing region, the C-terminal extension andthe area surrounding the active site cleft. Surfacecharge distribution around the i-face has beenassigned a primary role in the expression ofmyotoxicity of Lys49-PLA2s. In order to correlateelectrostatic profiles with the function of myotoxins,we have analyzed the diversity of PLA2 crystalstructures recently solved.

ARTICLE IN PRESSM.T. Murakami et al. / Toxicon 51 (2008) 723–735732

The PLA2s and its subgroups were divided intomyotoxic and non-myotoxic proteins and thequalitative analysis of surface charge distributionwas carried out. The non-myotoxic proteins clearlydisplay a conserved area around the active site,which forms a deep negatively charged channel(Fig. 6). In addition, a large part of the surface isuncharged and scattered positive regions are ob-served at the tip of the b-wing and at the end of theC-terminus. These surface charge distribution fea-tures were observed in all crystal structures of non-myotoxic PLA2s deposited with the Protein DataBank. On the other hand, the myotoxic proteins arestrikingly different when compared with the non-myotoxic PLA2s with charge inversion around theactive-site region (Fig. 6). The positive chargepattern in myotoxic proteins is high conservedaround the active-site region and C-terminus witha large neutral or slightly negatively charged spreadaround the i-face (Figs. 6 and 7). Calculationsperformed with many Lys49/Arg49 PLA2s analog

Fig. 6. Surface charge distribution of several non- and myotoxic PLA

the surface of myotoxic PLA2s (these motifs are also indicated in the

charged core in non-myotoxic PLA2s. Zhaoermiatoxin, an Arg49-PLA2

Arg49-PLA2 from Protobothrops elegans (modeled); Basp-II, a Lys49-

Asp49-PLA2 from Agkistrodon halys pallas (PDB entry: 1JIA); CA-PL

BthA-I; acid Asp49-PLA2 from Bothrops jararacussu (PDB entry: 1UM

result in a similar pattern of charge distribution,which indicates the role of charge around theC-terminus and active-site region for expression ofmyotoxicity.

4. Concluding remarks

This work represents the first crystal structure ofan Arg49-PLA2 subgroup determined at 2.05 Aresolution, with an arginine residue at position 49instead of a lysine, suggesting a secondary Lys49-Arg substitution, which does not alter the catalyticinactivity of the molecule. The structural analysisreveals a rare well-ordered C-terminus and unex-pected non-polarized state of the putative calcium-binding loop due to the flip of Lys122 towards tothe bulk solvent. The rotamers around the sidechain of Arg49 permit a manner of binding similarto the lysine residue; however, the guadinidiumgroup is tri-coordinated by carbonyl oxygen atomsof the putative calcium-binding loop, whereas the

2s. The green circles indicate conserved positive charges around

sequence alignment). The yellow circles indicate the negatively

from Zhaoermia mangshanensis (PDB entry: 2PH4); PeBR-II, an

PLA2 from Bothrops asper (PDB entry: 1Y4L); AHP-PLA2, an

A2, an Asp49-PLA2 from Crotalus atrox (PDB entry:1PP2) and

V).

ARTICLE IN PRESS

Fig. 7. Multiple sequence alignment of the zhaoermiatoxin with other Asp/Lys49 PLA2s. The two conserved motifs that significantly

contribute to the overall positively charged surfaces of myotoxic PLA2s are boxed. Zhaoermiatoxin, an Arg49-PLA2 from Zhaoermia

mangshanensis (PDB entry: 2PH4); PeBR-II, an Arg49-PLA2 from Protobothrops elegans (modeled); Basp-II, a Lys49-PLA2 from

Bothrops asper (PDB entry: 1Y4L) and BthA-I; acid Asp49-PLA2 from Bothrops jararacussu (PDB entry: 1UMV).

M.T. Murakami et al. / Toxicon 51 (2008) 723–735 733

Nz atom of lysine is tetra-coordinated using a dif-ferent architecture of the putative calcium-bindingloop. Despite the differences between Lys49 andArg49-PLA2 analog, no structural evidences wereobserved to warrant the suggestion of a differentmechanism involved in the expression of theirpharmacological effects.

Ethical statement: On behalf of, and havingobtained permission from all the authors, R.K.Arni declares that (a) the material has not beenpublished in whole or in part elsewhere; (b) thepaper is not currently being considered for publica-tion elsewhere; (c) all authors have been personallyand actively involved in substantive work leading tothe report, and will hold themselves jointly andindividually responsible for its content and (d) allrelevant ethical safeguards have been met in relationto patient or subject protection, or animal experi-mentation. There are no connections with anytobacco, alcohol, pharmaceutical or gaming indus-try. There are no contractural constraints onpublishing that existed with regard to the researchbeing reported.

Sources of funding: This project was supported bygrants from FAPESP, SMOLBNet, CAPES/DAAD/PROBRAL and CNPq to R.K.A. M.T.M. is therecipient of a FAPESP postdoctoral fellowship.

Acknowledgments

This project was supported by grants fromFAPESP, SMOLBNet, CAPES/DAAD/PROBRALand CNPq to R.K.A and CNPq (471192/2007-4)to M.T.M.

References

Ambrosio, A.L., Nonato, M.C., Selistre-de-Araujo, H.S., Arni,

R.K., Ward, R.J., Ownby, C.L., et al., 2005. A molecular

mechanism for Lys49-phospholipase A2 activity based on

ligand-induced conformational change. J. Biol. Chem. 280,

7326–7335.

Andriao-Escarso, S.H., Soares, A.M., Rodrigues, V.M., Angulo,

Y., Diaz, C., Lomonte, B., et al., 2000. Myotoxic phospho-

lipases A(2) in bothrops snake venoms: effect of chemical

modifications on the enzymatic and pharmacological proper-

ties of bothropstoxins from Bothrops jararacussu. Biochimie

82, 755–763.

Arni, R.K., Ward, R.J., 1996. Phospholipase A2—a structural

review. Toxicon 34, 827–841.

Arni, R.K., Ward, R.J., Gutierrez, J.M., Tulinsky, A., 1995.

Structure of a calcium-independent phospholipase-like myo-

toxic protein from Bothrops asper venom. Acta Crystallogr. D

51, 311–317.

Bradford, M.M., 1976. A rapid and sensitive method for the

quantitation of microgram quantities of protein utilizing the

principle of protein dye binding. Anal. Biochem. 72, 248–254.

Brunie, S., Bolin, J., Gewirth, D., Sigler, P.B., 1985. The refined

crystal structure of dimeric phospholipase A2 at 2.5 A. Access

to a shielded catalytic center. J. Biol. Chem. 260, 9742–9749.

Chijiwa, T., Tokunaga, E., Ikeda, R., Terada, K., Ogawa, T.,

Oda-Ueda, N., Hattori, S., Nozaki, M., Ohno, M., 2006.

Discovery of novel [Arg49]phospholipase A2 isozymes from

Protobothrops elegans venom and regional evolution of

Crotalinae snake venom phospholipase A2 isozymes in the

southwestern islands of Japan and Taiwan. Toxicon 48,

672–682.

DeLano, W.L., 2002. The PyMOL Molecular Graphics System.

DeLano Scientific, Palo Alto, CA, USA.

Dijkstra, B.W., Kalk, K.H., Hol, W.G.J., Drenth, J., 1981.

Structure of bovine pancreatic phospholipase A2 at 1.7 A

resolution. J. Mol. Biol. 147, 97–123.

Eisenberg, D., Luthy, R., Bowie, J.U., 1997. VERIFY3D:

assessment of protein models with three-dimensional profiles.

Methods Enzymol. 277, 396–404.

ARTICLE IN PRESSM.T. Murakami et al. / Toxicon 51 (2008) 723–735734

Emsley, P., Cowtan, K., 2004. Coot: model-building tools for

molecular graphics. Acta Crystallogr. D 60, 2126–2132.

Fiser, A., Sali, A., 2003a. ModLoop: automated modeling of

loops in protein structures. Bioinformatics 19, 2500–2501.

Fiser, A., Sali, A., 2003b. Modeller: generation and refinement of

homology-based protein structure models. Methods Enzymol.

374, 461–491.

Gumprecht, A., Tillack, F., 2004. Proposal for a replacement name

of the snake genus Ermia Zhang, 1993. Russ. J. Herpetol. 11,

73–76.

Gutierrez, J.M., Theakston, R.D.G., Warrell, D.A., 2006.

Confronting the neglected problem of snakebite envenoming:

the need for a global partnership. PLoS Med. 3, 727–731.

Hess, B., Becker, H., Berendsen, H.J., Fraaije, J.G.E.M., 1997.

Lincs: a linear constraint solver for molecular simulations.

J. Comp. Chem. 18, 1463–1472.

Hockney, R.W., Goel, S.P., 1974. Quiet high-resolution compu-

ter models of a plasma. J. Comp. Phys. 14, 148–158.

Jones, T.A., 1985. Interactive computer graphics: FRODO.

Methods Enzymol. 115, 157–171.

Keith, C., Feldman, D.S., Deaanello, S., Glick, J., Ward, K.B.,

Jones, E.O., Sigler, P.B., 1981. The 2.5 A crystal structure of a

dimeric phospholipase A2 from the venom of Crotalus atrox.

J. Biol. Chem. 256, 8602–8607.

Kini, R.M., Evans, H.J., 1989. Role of cationic residues in

cytolytic activity: modification of lysine residues in the

cardiotoxin from Naja nigricollis venom and correlation

between cytolytic and antiplatelet activity. Biochemistry 28,

9209–9216.

Kini, R.M., Iwanaga, S., 1986. Structure–function relationships

of phospholipases. II: Charge density distribution and the

myotoxicity of presynaptically neurotoxic phospholipases.

Toxicon 24, 895–905.

Krizaj, I., Bieber, A.L., Ritonja, A., Gubensek, F., 1991. The

primary structure of ammodytin L, a myotoxic phospholipase A2

homologue from Vipera ammodytes venom. Eur. J. Biochem.

202, 1165–1168.

Laemmli, U.K., 1970. Cleavage of structural proteins during the

assembly of head of bacteriophage T4. Nature 227, 680–685.

Laskowski, R.A., Moss, D.S., Thornton, J.M., 1993. Main-chain

bond lengths and bond angles in protein structures. J. Mol.

Biol. 231, 1049–1067.

Lee, W.H., Toyama, M.H., Soares, A.M., Giglio, J.R., Mar-

angoni, S., Polikarpov, I., 1999. Crystallization and pre-

liminary X-ray diffraction studies of piratoxin III, a D-49

phospholipase A2 from the venom of Bothrops pirajai. Acta

Crystallogr. D 55, 1229–1230.

Lee, W.H., da Silva Giotto, M.T., Marangoni, S., Toyama,

M.H., Polikarpov, I., Garratt, R.C., 2001. Structural basis for

low catalytic activity in Lys49 phospholipases A2—a hypoth-

esis: the crystal structure of piratoxin II complexed to fatty

acid. Biochemistry 40, 28–36.

Lindahl, E., Hess, B., van der Spoe, D.L., 2001. Gromacs 3.0: a

package for molecular simulation and trajectory analysis.

J. Mol. Mod. 7, 306–317.

Lomonte, B., Tarkowski, A., Hanson, L.A., 1993. Host response

to Bothrops asper snake venom: analysis of edema formation,

inflammatory cells, and cytokine release in a mouse model.

Inflammation 17, 93–105.

Lomonte, B., Lundgren, J., Johansson, B., Bagge, U., 1994a. The

dynamics of local tissue damage induced by Bothrops asper

snake venom and myotoxin II on the mouse cremaster

muscle: an intravital and electron microscopic study. Toxicon

32, 41–55.

Lomonte, B., Moreno, E., Tarkowski, A., Hanson, L.A.,

Maccarana, M., 1994b. Neutralizing interaction between

heparins and myotoxin II, a Lys-49 phospholipase A2 from

Bothrops asper snake venom. Identification of a heparin-binding

and cytolytic toxin region by the use of synthetic peptides and

molecular modeling. J. Biol. Chem. 269, 29867–29873.

Mebs, D., Kuch, U., Coronas, F.I., Batista, C.V., Gumprecht,

A., Possani, L.D., 2006. Biochemical and biological activities

of the venom of the Chinese pitviper Zhaoermia mangsha-

nensis, with the complete amino acid sequence and phyloge-

netic analysis of a novel Arg49 phospholipase A2 myotoxin.

Toxicon 47, 797–811.

Miyamoto, S., Kollman, P.A., 1992. SETTLE: An analytical

version of the SHAKE and RATTLE algorithm for rigid

water models. J. Comp. Chem. 13, 952–962.

Murakami, M.T., Arni, R.K., 2003. A structure based model for

liposome disruption and the role of catalytic activity in

myotoxic phospholipase A2s. Toxicon 42, 903–913.

Murakami, M.T., Arruda, E.Z., Melo, P.A., Martinez, A.B.,

Calil-Elias, S., Tomaz, M.A., Lomonte, B., Gutierrez, J.M.,

Arni, R.K., 2005. Inhibition of myotoxic activity of Bothrops

asper myotoxin II by the anti-trypanosomal drug suramin.

J. Mol. Biol. 350, 416–426.

Murakami, M.T., Vic-oti, M.M., Abrego, J.R., Lourenzoni,

M.R., Cintra, A.C., Arruda, E.Z., et al., 2007. Interfacial

surface charge and free accessibility to the PLA2-active site-

like region are essential requirements for the activity of Lys49

PLA2 homologues. Toxicon 49, 378–387.

Murshudov, G.N., Vagin, A.A., Dodson, E.J., 1997. Refinement

of macromolecular structures by the maximum-likelihood

method. Acta Crystallogr. D 53, 240–255.

Navaza, J., 1994. AMoRe: an automated package for molecular

replacement. Acta Crystallogr. A 50, 157–163.

Nunez, C.E., Angulo, Y., Lomonte, B., 2001. Identification of the

myotoxic site of the Lys49 phospholipase A(2) from Agkis-

trodon piscivorus piscivorus snake venom: synthetic C-terminal

peptides from Lys49, but not from Asp49 myotoxins, exert

membrane-damaging activities. Toxicon 39, 1587–1594.

Otwinowski, Z., Minor, W., 1997. Processing of X-ray diffraction

data collected in oscillation mode. Methods Enzymol. 276,

307–326.

Ownby, C.L., Selistre de Araujo, H.S., White, S.P., Fletcher, J.E.,

1999. Lysine 49 phospholipase A2 proteins. Toxicon 37,

441–445.

Pan, H., Liu, X.L., Ou-Yang, L.L., Yang, G.Z., Zhou, Y.C., Li,

Z.P., Wu, X.F., 1998. Diversity of cDNAs encoding

phospholipase A2 from Agkistrodon halys Pallas venom, and

its expression in E. coli. Toxicon 36, 1155–1163.

Paramo, L., Lomonte, B., Pizarro-Cerda, J., Bengoechea, J.A.,

Gorvel, J.P., Moreno, E., 1998. Bactericidal activity of Lys49

and Asp49 myotoxic phospholipases A2 from Bothrops asper

snake venom: synthetic Lys49 myotoxin II-(115–129)-peptide

identifies its bactericidal region. Eur. J. Biochem. 253, 452–461.

Polgar, J., Magnenat, E.M., Peitsch, M.C., Wells, T.N.C.,

Clemetson, K.J., 1996. Asp-49 is not an absolute prerequisite

for the enzymic activity of low-M(r) phospholipases A2:

purification, characterization and computer modelling of an

enzymically active Ser-49 phospholipase A2, ecarpholin S,

from the venom of Echis carinatus sochureki (saw-scaled

viper). Biochem. J. 319, 961–968.

ARTICLE IN PRESSM.T. Murakami et al. / Toxicon 51 (2008) 723–735 735

Santamarıa, C., Larios, S., Angulo, Y., Pizarro-Cerda, J., Gorvel,

J.P., Moreno, E., Lomonte, B., 2005. Antimicrobial activity

of myotoxic phospholipases A2 from crotalid snake venoms

and synthetic peptide variants derived from their C-terminal

region. Toxicon 45, 807–815.

Scott, D.L., White, S.P., Otwinowski, Z., Yuan, W., Gelb, M.H.,

Sigler, P.B., 1990. Interfacial catalysis: the mechanism of

phospholipase A2. Science 250, 1541–1546.

Selistre de Araujo, H.S., White, S.P., Ownby, C.L., 1996. cDNA

cloning and sequence analysis of a lysine-49 phospholipase A2

myotoxin from Agkistrodon contortrix laticinctus snake

venom. Arch. Biochem. Biophys. 326, 21–30.

Teixeira, C.F.P., Landucci, E.C.T., Antunes, E., Chacur, M.,

Cury, Y., 2003. Inflammatory effects of snake venom

myotoxic phospholipases A2. Toxicon 42, 947–962.

Tsai, I.H., Wang, Y.M., Chen, Y.H., Tsai, T.S., Tu, M.C., 2004.

Venom phospholipases A2 of bamboo viper (Trimeresurus

stejnegeri): molecular characterization, geographic variations

and evidence of multiple ancestries. Biochem. J. 377, 215–223.

Van den Berg, C.J., Slotboom, A.J., Verheij, H.M., de Haas,

G.H., 1989. The role of Asp-49 and other conserved amino

acids in phospholipases A2 and their importance for

enzymatic activity. J. Cell Biochem. 39, 379–390.

van Gunsteren, W.F., Billeter, S.R., Eising, A.A., Hunenberger,

P.H., Kruger, P., Mark, A.E., Scott, W.R.P., Tironi, I.G.,

1996. In: Groningen Molecular Simulation (GROMOS)

Library Manual. BIOMOS, Zurich, Switzerland.

Yamazaki, Y., Matsunaga, Y., Nakano, Y., Morita, T., 2005.

Identification of vascular endothelial growth factor receptor-

binding protein in the venom of eastern cottonmouth. A new

role of snake venom myotoxic Lys49-phospholipase A2.

J. Biol. Chem. 280, 29989–29992.

Ward, R.J., Monesi, N., Arni, R.K., Larson, R.E., Pac-o-Larson,

M.L., 1995. Sequence of a cDNA encoding bothropstoxin I, a

myotoxin from the venom of Bothrops jararacussu. Gene 156,

305–306.

Ward, R.J., Rodrigues Alves, A., Rugiero Neto, J., Arni, R.K.,

Casari, J., 1998. A SequenceSpace analysis of Lys49

phospholipases A2: clues towards identification of residues

involved in a novel mechanism of membrane damage and in

myotoxicity. Prot. Eng. 11, 285–294.

Ward, R.J., Chioato, L., de Oliveira, A.H.C., Ruller, R., Sa,

J.M., 2002. Active site mutagenesis of a Lys49-phospholipases

A2: biological and membrane-disrupting activities in the

absence of catalysis. Biochem. J. 362, 89–96.

Watanabe, L., Soares, A.M., Ward, R.J., Fontes, M.R., Arni,

R.K., 2005. Structural insights for fatty acid binding in a

Lys49-phospholipase A2: crystal structure of myotoxin II

from Bothrops moojeni complexed with stearic acid. Biochimie

87, 161–167.

White, S.P., Scott, D.L., Otwinowski, Z., Gelb, M.H., Sigler,

P.B., 1990. Crystal structure of cobra-venom phospholipase

A2 in a complex with a transition-state analogue. Science 250,

1560–1563.

Wiederstein, M., Sippl, M.J., 2007. ProSA-web: interactive web

service for the recognition of errors in three-dimensional

structures of proteins. Nucleic Acids Res. 35, 407–410.

Willard, L., Ranjan, A., Zhang, H., Monzavi, H., Robert, F.B.,

Brian, D.S., David, S.W., 2003. VADAR: a web server for

quantitative evaluation of protein structure quality. Nucleic

Acids Res. 31, 3316–3319.

Zhao, E.M., Chen, Y.H., 1990. Description of a new species of

the genus Trimeresurus. Sichuan J. Zool. 9, 11–12.

Zuliani, J.P., Fernandes, C.M., Zamuner, S.R., Gutierrez, J.M.,

Teixeira, C.F.P., 2005. Inflammatory events induced by Lys-

49 and Asp-49 phospholipases A2 isolated from Bothrops

asper snake venom: role of catalytic activity. Toxicon 45,

335–346.