Mapping the structural determinants of presynaptic neurotoxicity of snake venom phospholipases A2

ARTICLE IN PRESS

Contents lists available at ScienceDirect

Toxicon

Toxicon 51 (2008) 1520– 1529

0041-01

doi:10.1

$ Eth

out in

experim� Cor

E-m

journal homepage: www.elsevier.com/locate/toxicon

Mapping the structural determinants of presynaptic neurotoxicity ofsnake venom phospholipases A2

$

Petra Prijatelj a, Zala Jenko Praznikar b, Toni Petan b, Igor Krizaj b, Joze Pungercar b,�

a Department of Chemistry and Biochemistry, Faculty of Chemistry and Chemical Technology, University of Ljubljana, SI-1000 Ljubljana, Sloveniab Department of Molecular and Biomedical Sciences, Jozef Stefan Institute, SI-1000 Ljubljana, Slovenia

a r t i c l e i n f o

Article history:

Received 11 January 2008

Received in revised form

28 March 2008

Accepted 28 March 2008Available online 4 April 2008

Keywords:

Phospholipase A2

Snake venom

Vipera ammodytes ammodytes

Daboia russelli russelli

Neurotoxicity

Enzymatic activity

Interfacial binding surface

01/$ - see front matter & 2008 Elsevier Ltd.

016/j.toxicon.2008.03.031

ical statement: All experimental procedures

accordance with the EC Council Directiv

entation.

responding author. Tel.: +386 1477 3713; fax

ail address: [email protected] (J. Pungerc

a b s t r a c t

The structural features of presynaptically neurotoxic secretory phospholipases A2 (sPLA2s)

that are responsible for their potent and specific action are still a matter of debate. To

identify the residues that distinguish a highly neurotoxic sPLA2, ammodytoxin A (AtxA),

from a structurally similar but more than two orders of magnitude less toxic Russell’s

viper sPLA2, VIIIa, we prepared a range of mutants and compared their properties. The

results show that the structural features that confer high neurotoxicity to AtxA extend

from its C-terminal part, with a central role of the residues Y115, I116, R118, N119 (the

YIRN cluster) and F124, across the interfacial binding surface (IBS) in the vicinity of F24, to

the N-terminal helix whose residues M7 and G11 are located on the edges of the IBS.

Competition binding studies indicate that the surface of interaction with the neuronal

M-type sPLA2 receptor R180 extends over a similar region of the molecule. In addition, the

YIRN cluster of AtxA is crucial for the high-affinity interaction with two intracellular

binding proteins, calmodulin and R25. The concept of a single ‘‘presynaptic neurotoxic

site’’ on the surface of snake venom sPLA2s is not consistent with these results which

suggest that different parts of the toxin molecule are involved in distinct steps of

presynaptic neurotoxicity.

& 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Phospholipases A2 (PLA2s) catalyze the hydrolysis ofthe sn-2 ester bond of glycerophospholipids, generatingfree fatty acids and lysophospholipids (Berg et al., 2001).PLA2s have recently been classified into 15 groups on thebasis of their functional and structural features, and mostconsist of several subgroups (Schaloske and Dennis,2006). They are widespread in nature as extracellularand intracellular enzymes in organisms ranging fromviruses to humans (Valentin and Lambeau, 2000; Scha-loske and Dennis, 2006). Secreted PLA2s (sPLA2s) are

All rights reserved.

on mice were carried

e regarding animal

: +386 1477 3984.

ar).

relatively small (13–18 kDa), disulfide-rich and Ca2+-dependent enzymes and are some of the most toxiccomponents of snake venoms (Kini, 2003). The number ofstructurally related mammalian and non-toxic sPLA2s hasrecently risen to 11 differentially expressed isoforms, withmostly unknown (patho)physiological roles (Kudo andMurakami, 2002). The study of the action of exogenoussnake venom sPLA2s, which produce a variety of pharma-cological effects by interfering with normal physiologicalprocesses in mammalian preys, could provide new clues tothe action of the endogenous forms as well.

The most potent toxins isolated from snake venoms arepresynaptically acting sPLA2 neurotoxins, whose molecu-lar mechanism of action is still not completely understood(Rigoni et al., 2005; Pungercar and Krizaj, 2007). Thesevenom sPLA2s induce a complete failure of neuromusculartransmission leading to death of the prey due to paralysisof respiratory muscles (Chang, 1985). Although there is no

ARTICLE IN PRESS

P. Prijatelj et al. / Toxicon 51 (2008) 1520–1529 1521

clear correlation between enzymatic activity and lethalpotency of these toxins (Kini, 2003), their PLA2 activity atthe neuromuscular junction appears to be essential forneurotoxicity (Chang and Su, 1982). However, not all theobserved effects of sPLA2 presynaptic neurotoxicity can beexplained by their non-selective enzymatic action (Petanet al., 2005), strongly implying that some of the knownhigh-affinity extra- and intracellular binding proteins areindispensable for full exertion of presynaptic neurotoxi-city (Krizaj and Gubensek, 2000; Kini, 2003; Pungercarand Krizaj, 2007).

Ammodytoxins (Atxs) A, B and C are monomeric groupIIA sPLA2s with presynaptic neurotoxicity, present invenom of the long-nosed viper, Vipera ammodytes ammo-

dytes (Thouin et al., 1982; Lee et al., 1984). Soluble andmembrane-bound specific binding proteins for Atxs havebeen identified in porcine cerebral cortex and theirinvolvement in presynaptic neurotoxicity of sPLA2s iscurrently still under investigation (Pungercar and Krizaj,2007; Jenko Praznikar et al., 2008). Atxs bind to R25,which is presumably a mitochondrial protein, with highspecificity and affinity (Vucemilo et al., 1998; Sribar et al.,2003a); however, a non-toxic chimeric mutant containingthe C-terminal part of ammodytoxin A (AtxA) also bindswell to R25 (Prijatelj et al., 2002). R180, which wasidentified as a neuronal M-type sPLA2 receptor, binds bothtoxic and non-toxic sPLA2s of groups I and II (Copic et al.,1999; Vardjan et al., 2001) and binding of Atxs to R180was shown not to be obligatory for their presynapticneurotoxicity (Prijatelj et al., 2006). Acceptors for Atxsalso include the highly conserved cytoplasmic proteinscalmodulin (CaM), and 14-3-3 protein g and e isoforms, allimportant regulatory proteins (Sribar et al., 2001, 2003b;Prijatelj et al., 2003).

A weakly neurotoxic sPLA2, VIIIa (referred to here asDPLA2), has been isolated from venom of the Russell’sviper, Daboia russelli russelli (Kasturi and Gowda, 1989).Although DPLA2 and AtxA share a high level of amino acididentity (82%), the difference in their lethal potency isvery great (Prijatelj et al., 2003). AtxA is almost 150-foldmore toxic in mice than DPLA2, which displays an LD50

value of 3.1 mg/kg (Prijatelj et al., 2003). Our previousstudies suggest an important role in neurotoxicity of theYIRN cluster (residues Y115, I116, R118 and N119) in theC-terminal region of AtxA (Ivanovski et al., 2000).However, introduction of this cluster in DPLA2 resulted,surprisingly, in a significant decrease in its lethal potency(Prijatelj et al., 2003), reflecting the complex structure–function relationship of presynaptically neurotoxic sPLA2s(Kini, 2003).

In this study, we present the preparation and struc-ture–function analysis of a range of AtxA/DPLA2 chimericmutants that have allowed us to specify the structuralparameters of the two closely related toxins responsiblefor the large difference in their neurotoxic potency.Additionally, by analyzing the affinities of the mutantsfor their neuronal acceptors and their enzymatic activities,we further expand the definition of the structuraldeterminants required for binding to high-affinity proteintargets and productive interaction with the membranesurface. The results of our study reveal that the site(s)

important for neurotoxicity in each sPLA2 are formed by asubtle three-dimensional network of a number of residuespresent on a larger surface of this multifunctionalmolecule.

2. Materials and methods

2.1. Materials

AtxC was isolated from V. a. ammodytes venom asdescribed (Gubensek et al., 1980). Recombinant AtxA,DPLA2 and DPLA2

YIRN were produced in Escherichia coli andpurified as described previously (Pungercar et al., 1999;Prijatelj et al., 2003). The expression plasmid encoding ratliver fatty acid-binding protein (FABP) was from Dr. DavidC. Wilton (University of Southampton, UK) and therecombinant protein was prepared as described (Worrallet al., 1991). Restriction enzymes were obtained from MBIFermentas and New England BioLabs (USA). Vent DNApolymerase, T4 polynucleotide kinase and Taq DNA ligasewere from New England BioLabs. T4 DNA ligase was fromBoehringer Mannheim (Germany) and oligonucleotidesfrom MWG-Biotech (Germany). Hog brain CaM was fromRoche Molecular Biochemicals, radioisotopes from Perkin–Elmer Life Sciences, and disuccinimidyl suberate fromPierce. POPG was from Avanti Polar Lipids (USA), Hanks’balanced salt solution (HBSS) and 11-dansylundecanoicacid from Invitrogen (USA). All other chemicals were of atleast analytical grade, and were from Sigma-Aldrich (USA)or Serva (Germany).

2.2. Preparation of recombinant mutant sPLA2s

Expression plasmids for DPLA2 and DPLA2YIRN were

constructed as reported (Prijatelj et al., 2003). Thenucleotide sequences encoding the mutant proteins wereprepared by site-directed mutagenesis using PCR withinternal degenerated and phosphorylated oligonucleo-tides. Plasmids coding for the chimeric mutants, AtxA/DPLA2 and AtxA/DPLA2

YIRN, were constructed by introdu-cing an internal PstI restriction site in the DPLA2 andDPLA2

YIRN nucleotide sequences using the internal oligo-nucleotide primer 50-CCTGACTGCAgCCCCAAATCG-30. Inthe latter, the nucleotide shown in lower case introducedthe restriction site (underlined) and simultaneouslymutated the residue N67 in DPLA2 to a serine, which isalso present in AtxA. The outer sense oligonucleotideprimer 50-TAATACGACTCACTATAG-30 and the antisenseoligonucleotide primer 50-GTTTACTCATATATACTTTAG-30

were complementary to the expression vector pT7-7(Tabor, 1990). Previously optimized PCR cycling conditionswere used (Prijatelj et al., 2003) and the annealingtemperature was set to 49 1C. The resulting fragments of558 bp were cut at their BamHI and NotI sites and ligatedin the correspondingly linearized vectors for DPLA2 andDPLA2

YIRN. After sequencing, the PstI/PstI fragments (in thecorrect orientation) of both constructs (Prijatelj et al.,2003) were used to replace the PstI/PstI fragment in theexpression plasmid based on pT7-7 and encoding AtxA

ARTICLE IN PRESS

P. Prijatelj et al. / Toxicon 51 (2008) 1520–15291522

fused with a short peptide at its N-terminus (Pungercaret al., 1999).

The internal oligonucleotides used in the PCR muta-genesis reaction to introduce the mutations M7K/G11E,V31W and M7K/G11E/V31W on the plasmid coding forAtxA/DPLA2

YIRN were 50-CGGGAaGATGATCCTGGaGGAGAC-30 and 50-GCTACTGCGGCtgGGGGGGCAAAGGC-30, in whichthe positions of the mutations are shown in lower case.The outer amplification primers were the same as above.The PCR mutagenesis reaction was performed as described(Prijatelj et al., 2003) and the annealing temperature wasset to 50 1C in this case. The sequences of all constructswere verified using an ABI Prism 310 Genetic Analyser(Perkin-Elmer Applied Biosystems, USA).

The expression vectors were transformed into theE. coli strain BL21(DE3) (Novagen, USA) for recombinantprotein expression. Proteins in the isolated inclusionbodies were S-sulfonated, refolded, activated by trypsinand purified as described for DPLA2 and DPLA2

YIRN (Prijateljet al., 2003).

2.3. Analytical methods

Protein samples were analyzed by SDS-PAGE andreverse-phase HPLC on a C4 column, using an HP1100system (Hewlett-Packard, USA). The N-terminal sequenceswere determined on an Applied Biosystems Procise 492Aprotein sequencing system. Electrospray ionization massspectrometry (ESI-MS) was performed using a high-resolution magnetic-sector AutospecQ mass spectrometer(Micromass, UK). Protein concentrations were determinedby the method of Perkins (1986). Circular dichroismspectra were recorded from 250 to 200 nm at 25 1C onan Aviv 62A DS CD spectrometer. Samples were scanned ina cell with 1 mm path length, with a bandwidth of 2 nm, astep size of 1 nm and an averaging time of 2 s. Proteinconcentrations in 20 mM Tris–HCl, pH 7.0, were: 18.1mMfor AtxA/DPLA2, 11.9 mM for AtxA/DPLA2

YIRN, 17.1 mM forAtxAKEW/DPLA2

YIRN, 17.9mM for AtxAKE/DPLA2YIRN and

7.5 mM for AtxAW/DPLA2YIRN.

2.4. Titrimetric method for measuring PLA2 activity

Hydrolysis of egg-yolk phosphatidylcholine was mea-sured in a reaction mixture (8 ml; one egg yolk suspendedin 200 ml) containing 1% (v/v) Triton X-100 and 15 mMCaCl2 at 40 1C. The pH was set at 8.0 and the fatty acidsreleased titrated with 10 mM NaOH using a 718 STATTitrino pH-stat (Metrohm, Switzerland).

2.5. Enzymatic activity on phospholipid vesicles

Phospholipid stock solutions in chloroform were storedat �80 1C under argon atmosphere. The organic solventwas removed by evaporation under vacuum with aSpeedVac Plus centrifuge (Savant Instrument, USA), andthe dried phospholipid film was hydrated with HBSS(Invitrogen-Gibco, USA) containing 1.26 mM Ca2+ and0.9 mm Mg2+, by vigorous agitation above the phase-transition temperature. The lipids were then subjected to

8 freeze/thaw cycles and large unilamellar vesicles wereprepared by extrusion through a 100 nm polycarbonatemembrane using a Lipofast Basic extruder (Avestin,Canada). The size, structural integrity and polydispersityof the vesicle suspensions were checked routinely bydynamic light scattering measurements using a PDDLS/BatchPlus System (Precision Detectors, USA) as described(Petan et al., 2007).

The initial rate of hydrolysis of phospholipid vesicles bysPLA2s was measured by monitoring the displacement of afluorescent fatty acid analog from FABP (Wilton, 1990;Bezzine et al., 2000). Assay solutions with a final volume of1.3 ml were prepared in HBSS with 1.26 mm Ca2+ and0.9 mm Mg2+ by adding 13ml of vesicle suspension contain-ing 3 mM phospholipid, 11-dansylundecanoic acid to a finalconcentration of 1mM and approximately 10mg of recom-binant FABP. Assays were performed in acrylic fluorometriccuvettes in a thermostated chamber at 37 1C with magneticstirring, using a Perkin-Elmer LS50B fluorometer (USA).Excitation was set at 350 nm and emission at 500 nm, with10 nm slit widths. All dilutions of sPLA2s were prepared inassay buffer containing 1 mg/ml fatty-acid-free bovineserum albumin (Sigma, USA) to prevent loss of enzymedue to adsorption to the walls of the tube. Reactions werestarted by adding small volumes of sPLA2 (1–5ml) andphospholipid hydrolysis was followed in real time as adecrease in fluorescence intensity. The quantity of sPLA2

added was chosen so that the slopes of the reaction curveswere in the optimal range. Assays were calibrated bysuccessive additions of 200 pmol of oleic acid in methanolto a reaction mixture containing all components exceptsPLA2. In both assays, one enzyme unit (U) corresponds to1mmol of hydrolyzed phospholipid per minute.

2.6. Toxicity

Lethality was determined by intraperitoneal injectionof 0.5 ml of each recombinant toxin (concentrationsranging from 1 to 500 mg/ml) in 0.9% (w/v) NaCl intoBALB/c albino mice. At least seven dose levels and tenmice (in the body weight range from 20.4 to 29.2 g) perdose were used for each toxin. Neurotoxic effects onexperimental animals were observed within 24 h, andLD50 was determined using a standard method (Reed andMuench, 1938). All experimental procedures on mice wereperformed in accordance with the EC Council Directiveregarding animal experimentation.

2.7. Competition binding studies

AtxC was used in protein binding studies instead of themore toxic AtxA due to a higher recovery rate afterlabelling. AtxC was radioiodinated (125I-AtxC) (Krizaj etal., 1994) and membranes were extracted from porcinecerebral cortex as described (Sribar et al., 2001). Themembrane extract or CaM solution (Prijatelj et al., 2003), afixed concentration of 125I-AtxC (10 nM) and increasingconcentrations of unlabelled recombinant sPLA2

(2 nM–10mM) were incubated at room temperature for30 min with occasional vortexing. Cross-linking of the

ARTICLE IN PRESS

P. Prijatelj et al. / Toxicon 51 (2008) 1520–1529 1523

toxin-binding protein complex was achieved by addingdisuccinimidyl suberate (100mM), dissolved in DMSO justbefore use. After vigorous mixing for 5 min at roomtemperature, the cross-linking reaction was stopped byadding SDS-PAGE sample buffer containing dithiothreitol.Following electrophoresis, drying of the gel and autoradio-graphy, the intensities of the specific adducts on auto-radiographs were quantified by QuantiScan (Biosoft, UK)and the data analyzed usingthe nonlinear curve fitting program Grafit, Version 3.0(Erithacus Software, UK).

3. Results and discussion

3.1. Structural integrity and homogeneity of the

recombinant sPLA2s

Protein homogeneity was demonstrated by SDS-PAGE,analytical reverse-phase HPLC, ESI-MS and N-terminalsequencing. Complete removal of the peptide fused to theN-terminus of recombinant sPLA2s and the absence ofinternal cleavages after activation by trypsin were con-firmed by N-terminal sequencing. The correct primarystructure and formation of all seven disulfide bridges inthe purified recombinant sPLA2s were confirmed by ESI-MS (Fig. 1; Table 1). The CD spectra of the recombinantsPLA2s show that no obvious secondary structure pertur-bations occurred due to the mutations introduced (Fig. 2).Further evidence for proper folding and structural integ-rity of the mutants was obtained from their catalyticefficiency, evident from their high and similar rates ofhydrolysis of POPG vesicles to which Atxs bind with veryhigh affinity (Petan et al., 2005, 2007) (Table 2).

Fig. 1. Amino acid sequence alignment of DPLA2, AtxA and their mutants. Ami

Residues whose substitution resulted in increased neurotoxicity of DPLA2 are

Renetseder et al. (1985) is used.

3.2. Structural determinants of sPLA2 presynaptic

neurotoxicity

Primary structure comparison reveals that there areonly 23 residue changes and one residue deletion betweenDPLA2 and AtxA. The aim of this study was to selectivelymutate parts of the DPLA2 molecule in order to map theresidues that are crucial to a highly potent presynapticneurotoxin. According to our previous structure–functionanalyses, the C-terminal residues of AtxA, particularlythe YIRN (Y115, I116, R118 and N119) cluster, areimportant for its neurotoxicity (Ivanovski et al., 2000).The lethal potency of the AtxA-Y115K/I116K/R118M/N119L(AtxAKKML) mutant, which contains the KKML clusterpresent in DPLA2, was 290-fold lower than that of AtxA(Ivanovski et al., 2000 and Table 1). However, introducingthe YIRN cluster into DPLA2 did not increase its toxicity;the DPLA2

YIRN mutant was more than five times less toxicthan DPLA2 (Prijatelj et al., 2003). Thus, we reasonedthat a particular combination of both C-terminal andN-terminal AtxA residues must be present in the moleculefor high neurotoxic potency. In order to test this hypoth-esis and to define the N-terminal residues that supple-ment the role of the YIRN cluster in the high neurotoxicityof AtxA, we selectively mutated some of the remainingresidues that differentiate DPLA2 from AtxA. First, weprepared a chimeric AtxA/DPLA2 protein by substitutingthe N-terminal half of DPLA2 by that of AtxA. The lethalpotency of the chimera was relatively weak, in the rangeof those of DPLA2 and the AtxAKKML mutant (Table 1 andIvanovski et al., 2000), suggesting that the negativeinfluence of the C-terminal DPLA2 residues on toxicity isprimarily a consequence of the presence of the KKMLcluster. We therefore substituted the latter with the YIRNcluster of AtxA, and the resulting AtxA/DPLA2

YIRN mutantdisplayed a significant 58-fold increase in lethal potencyover that of AtxA/DPLA2, reaching a level of toxicitysimilar to that of the highly neurotoxic AtxA. Additionally,

no acid residues identical to those present in DPLA2 are shown by dots.

presented in bold type. Standard amino acid numbering according to

ARTICLE IN PRESS

Table 1Relative molecular mass, lethal potency and affinity for binding to specific proteins determined for AtxA, DPLA2 and their mutants

sPLA2 Mr (measured) Mr (theoretical) LD50 (mg/kg) IC50 (nM)

CaM R25 R180

DPLA2 13,595.7 13,596.7 3100a 300736a 200737� 300745

DPLA2YIRN 13,641.8 13,642.6 E17,000a 43714a 200730 120721

AtxA/DPLA2 13,505.5 13,504.6 2600 110710 2578� 280719

AtxAKEW/DPLA2YIRN 13,707.0 13,706.7 910 43712 110715 2274

AtxAKE/DPLA2YIRN 13,619.4 13,619.6 790 1474 8779 2474

AtxAW/DPLA2YIRN 13,636.5 13,637.6 107 2877 78712 1975

AtxA/DPLA2YIRN 13,550.4 13,550.6 45 2775 180725 100716

AtxAKKML 13,728b 13,728b E6000b 5079a 86b 257b

AtxA-V31W 13,861.0c 13,861.8c 135c n.d. n.d. n.d.

AtxA-F24S 13,715d 13,715d 380d 7.470.2d 1471d 1672d

AtxA 13,774.6 13,774.8 21e 672a 1073 1673

� Only partial inhibition of 125I-AtxC binding to the binding protein was obtained with up to 10mM concentration of the competitor sPLA2s (see text

for details). The IC50 values are means7SD of at least 3 independent measurements. The LD50 values are accurate to within 710%.a Prijatelj et al. (2003).b Ivanovski et al. (2000).c Petan et al. (2005).d Petan et al. (2002).e Thouin et al. (1982).

Fig. 2. Circular dichroism spectra of the DPLA2 mutants prepared in this

study. The spectra are shown for DPLA2 (thick solid line), AtxA/DPLA2

(solid line), AtxA/DPLA2YIRN (long-dashed line), AtxAKEW/DPLA2

YIRN (short-

dashed line), AtxAKE/DPLA2YIRN (long- and short-dashed line) and AtxAW/

DPLA2YIRN (long- and double-short-dashed line).

Table 2Rates of phospholipid hydrolysis by sPLA2s determined on mixed egg-

yolk phosphatidylcholine/Triton X-100 micelles and large unilamellar

phospholipid vesicles

sPLA2 Rate of hydrolysis (U/mg)

Mixed micelles POPC POPG

DPLA2 760730a 1.170.1 16007130

DPLA2YIRN 2072a 0.0770.02 11007120

AtxA/DPLA2 11007100 3.1470.05 13007140

AtxAKEW/DPLA2YIRN 1520750 2.270.5 12007170

AtxAKE/DPLA2YIRN 170722 0.1470.01 830780

AtxAW/DPLA2YIRN 73007500 2673 19007210

AtxA/DPLA2YIRN 460760 0.5370.06 10007120

AtxAKKML 840b 1971c 11487126c

AtxA-V31W n.d. 10277c 2102788c

AtxA-F24S 7575d 0.7170.06c 400773c

AtxA 280710e 1.270.3 14007110

The rates of hydrolysis are means7SD of at least 3 independent

measurements.a Prijatelj et al. (2003).b Ivanovski et al. (2000).c Petan et al. (2005).d Petan et al. (2002).e Thouin et al. (1982).

P. Prijatelj et al. / Toxicon 51 (2008) 1520–15291524

the fact that the presence of the YIRN cluster in theC-terminal part of AtxA and AtxA/DPLA2

YIRN was enough toconfer a neurotoxicity that was high relative to thatdisplayed by AtxAKKML and AtxA/DPLA2, allowed us toexclude the importance of the additional 11 C-terminalresidues present in AtxA and absent in DPLA2. Theseresidues, also in accordance with the findings of ourprevious studies (Pungercar et al., 1999; Prijatelj et al.,2000; Ivanovski et al., 2004), are: T70, H76, E78, G85,R100, N114, S130, E131, and K108, K111 and K128 (Fig. 3).It is clear therefore that the particular combination ofC-terminal and N-terminal residues present in AtxA isnecessary for its potent neurotoxicity.

Our next goal was to determine which of the nineN-terminal residues present in AtxA but not in DPLA2 (M7,

G11, N17, P18, L19, T20, F24, V31 and S67) are importantfor complementing the YIRN cluster’s central role in thehigh neurotoxic potency of AtxA. The aromatic F24 hasalready been identified as very important for the neuro-toxic potency of AtxA (Petan et al., 2002), since substitu-tion of F24 with a Ser, which is present in DPLA2 at thesame position, led to a significant 18-fold decrease inlethal potency. To assess the importance of the rest of theN-terminal residues, we first substituted M7, G11 and V31on the highly toxic mutant AtxA/DPLA2

YIRN with residuespresent in DPLA2—K, E and W, respectively. The resultingmolecule AtxAKEW/DPLA2

YIRN was 20-fold less potentthan the AtxA/DPLA2

YIRN mutant, suggesting a significant

ARTICLE IN PRESS

Fig. 3. Structural differences between DPLA2 and AtxA, and residues important for presynaptic neurotoxicity. The three-dimensional structure of DPLA2

(Protein Data Bank ID code 1FB2) is presented in two orientations. On the left, the putative IBS and the N-terminal residues are facing the viewer, while

the C-terminal region is located in the upper-left corner of the molecule and the b-structure in its lower-right corner (‘‘front’’ view). On the right, the

molecule is rotated by 1801 around its vertical axis (‘‘back’’ view). Amino acids that are present in both AtxA and DPLA2 are shown in white, while the 23

residues present in DPLA2, but not in AtxA, are shaded gray and black. DPLA2 residues presented in gray are not significantly involved in presynaptic

neurotoxicity, whereas those in black have an important role in the process. The N- and C-terminal residues (S1 and C133) are denoted by arrows.

P. Prijatelj et al. / Toxicon 51 (2008) 1520–1529 1525

involvement of the M7/G11/V31 residues in the neuro-toxicity of AtxA. The lethality of the partial mutantsAtxAKE/DPLA2

YIRN and AtxAW/DPLA2YIRN (Table 1) led to the

conclusion that the contribution of the pair of residues inthe N-terminal helix, M7 and G11, to the neurotoxicpotency of AtxA is substantially greater than that of V31.These results are in accordance with previous studieswhich have found residues 6–8 to be important for theneurotoxicity of trimucrotoxin, a neurotoxic sPLA2 fromTrimeresurus muscrosquamatus (Tsai and Wang, 1998), andare consistent with our previous finding that the V31Wmutation did not cause a significant change in lethality ofAtxA, despite its significant effect on enzymatic activity(Petan et al., 2005). The fact that the bulky Trp at position31 of the AtxA-V31W and AtxAW/DPLA2

YIRN mutants,despite being spatially relatively close to both F24 andthe YIRN cluster, did not have a significant impact onneurotoxicity, supports the idea that the YIRN cluster andthe spatially distant M7/G11 pair play complementary, butseparate, roles in the neurotoxicity of Atxs. If we consideronly the N-terminal half of AtxA, DPLA2 and the range oftheir mutants, the collective contribution of M7, G11 andF24 to their neurotoxicity is very significant and similar tothat of the YIRN cluster in the C-terminus. For example,the introduction of S24 instead of Phe in AtxA (Petan et al.,2002) caused a decrease in neurotoxicity (�18-fold) to anextent similar to that following the introduction of all ofthe N-terminal AtxA residues, apart from M7, G11 and V31,to DPLA2

YIRN, resulting in the AtxAKEW/DPLA2YIRN mutant.

This indicates that the remaining N-terminal residues thatdiffer between AtxA and DPLA2, the N17/P18/L19/T20cluster and S67, are not greatly involved in the process ofneurotoxicity.

The results of this study provide further support for thesuggestion that different parts of the toxin molecule areinvolved in distinct steps of the complex sequence ofevents (Pungercar and Krizaj, 2007), collectively contri-buting to the final outcome of presynaptic neurotoxicity.The identification of the most important structuralfeatures and individual residues responsible for the highneurotoxic potency of Atxs has contributed significantly toour understanding of the overall structure–functionrelationship. These features are the part of the moleculeconcentrated around the C-terminal YIRN cluster, includ-ing the aromatic residues F24 and F124, and theN-terminal helix region including the M7/G11 pair ofresidues (Fig. 3).

3.3. Sites involved in binding of sPLA2s to high-affinity

protein targets

The binding site on AtxA for CaM has been shown toinvolve the region 107–125 in the C-terminal part, withthe YIRN cluster playing a central role in the interaction(Prijatelj et al., 2003). Introducing the YIRN cluster intoDPLA2 led to a 7-fold increase in its affinity for CaM (Table 1).The introduction of the N-terminal half of AtxA toDPLA2 and DPLA2

YIRN increased their binding affinity forCaM by 2–3-fold, suggesting that the N-terminal part ofthe molecule has only a minor supporting role in thebinding process and is probably located on the peripheryof the CaM binding site. This is in accordance with ourprevious results with a range of F24 mutations, whichdid not alter the binding affinity of AtxA for CaM (Petanet al., 2002). Furthermore, the binding affinities of

ARTICLE IN PRESS

P. Prijatelj et al. / Toxicon 51 (2008) 1520–15291526

AtxA/DPLA2YIRN and AtxAW/DPLA2

YIRN for CaM were practi-cally identical, proving that even the bulky Trp does notinterfere with the binding to CaM and that the residue atposition 31, despite being spatially located relativelyclose to the C-terminal CaM binding site, is not on theCaM/AtxA interaction surface. Introducing two chargedresidues to the AtxA/DPLA2

YIRN chimera derived from theDPLA2 sequence, K7 and E11, which are markedly distantfrom the proposed binding surface, resulted in a slightincrease of affinity to CaM. This is however most probablya consequence of a slight alteration in the distribution ofsurface charge which influences the orientation of thesPLA2 molecule in the complex through long-rangeelectrostatic interactions. These results provide strongsupport for the CaM-binding site being located in theC-terminal region of AtxA and confirm the important roleof the YIRN cluster in the interaction (Prijatelj et al., 2003).

Based on our study with chimeric mutants of the non-toxic AtnI2 and AtxA, the C-terminal region of AtxA wassuggested to be involved in binding to the R25 neuronalbinding protein (Prijatelj et al., 2002). A series ofsubstitutions of F24 in the N-terminal part of AtxA had anegligible effect on its affinity for R25, apart from a subtleperturbation of the binding interaction by Trp, indicatingthat F24 is in the vicinity of the toxin-binding proteinbinding surface, located most probably in the C-terminalpart of AtxA (Petan et al., 2002). In this study, furtherevidence of the involvement of the C-terminal part of AtxAin binding to R25 has been provided. DPLA2 and AtxA/DPLA2, which have an identical C-terminal half, bothfailed to completely inhibit the binding of radioactivelylabelled AtxC to R25. The maximal inhibition achievedwith 10 mM concentrations of DPLA2 and AtxA/DPLA2 was

Fig. 4. Competitive binding experiments between 125I-AtxC and the

DPLA2 mutants and the R25 binding protein. Porcine cerebral membrane

extracts containing R25 were incubated with labelled AtxC in the

presence of increasing concentrations of the indicated competitor DPLA2

mutants, after which cross-linking and analysis of the adducts were

performed as described in Section 2. The values shown are means7SD of

at least three independent measurements. The data are displayed as

specific binding (Bsp) relative to the maximum specific binding (Bsp0)

obtained in the absence of competitor sPLA2s. DPLA2 and the AtxA/DPLA2

chimera did not completely inhibit the binding of 10 nM 125I-AtxC even

at concentrations of 10 mM.

75% and 45%, respectively (Fig. 4), suggesting theexistence of different subtypes of R25. Complete inhibi-tion could only be achieved when the KKML cluster inthese two toxins was substituted by the YIRN cluster(Table 1), emphasizing the importance of the specificcombination and position of hydrophilic and hydrophobicresidues at positions 115–119 for effective binding toapparently multiple isoforms of R25. The fact that thebinding affinities of DPLA2

YIRN and AtxA/DPLA2YIRN for R25

were very similar suggests that the N-terminal part ofAtxA does not have a crucial role in the interaction withR25. However, the slight increase in binding affinity uponintroduction of K7/E11 and/or W31 to the AtxA/DPLA2

YIRN

chimera indicates that the interaction surface extendsbeyond the C-terminal region. Additionally, although theAtxAKKML mutant had a significantly lower binding affinityfor R25 than AtxA, it was able to completely inhibit thebinding of radiolabelled AtxC to R25, in contrast to DPLA2

and AtxA/DPLA2, which also contain the KKML cluster. Thecompetitive binding studies with R25 therefore suggest acentral role for the C-terminal residues including, but notconfined to, the YIRN cluster of AtxA.

Although the interaction of AtxA with the neuronal M-type sPLA2 receptor, R180, may not be essential forpresynaptic neurotoxicity, the pathophysiological impli-cations of the interaction between the M-type sPLA2

receptors and sPLA2s in general are still unknown andintriguing (Rouault et al., 2007; Pungercar and Krizaj,2007). Site-directed mutagenesis studies of porcinepancreatic sPLA2 (group IB) have indicated that residueswithin or close to the Ca2+-binding loop are crucial for theinteraction with the muscle M-type sPLA2 receptor(Lambeau et al., 1995). Further, modification of theN-terminus of AtxA by attaching fusion peptides of fiveor more residues prevents its binding to R180 (Prijateljet al., 2006). This is in keeping with the fact that theproenzymes of the homologous group IB and group Xmammalian sPLA2s do not bind to the M-type sPLA2

receptor or that their binding is substantially reduced(Lambeau et al., 1995; Yokota et al., 2000; Hanasaki andArita, 2002), suggesting an involvement of the N-terminalinterfacial binding surface (IBS) residues in the interac-tion. Additionally, a recent study on the homologousmammalian sPLA2s showed that high-affinity binding ofmouse sPLA2 isoforms to the soluble form of the mouseM-type sPLA2 receptor leads to inhibition of their enzy-matic activity (Rouault et al., 2007). The affinity of DPLA2

for R180 was 19-fold lower than that of AtxA (Table 1).Introduction of the N-terminal half of AtxA to DPLA2 didnot affect the interaction; however, the presence of theYIRN cluster in DPLA2

YIRN or AtxA/DPLA2YIRN increased their

binding affinities for R180 considerably, although still notreaching the low nanomolar affinity of AtxA. This result isin accordance with the decreased affinity for R180 of thereverse AtxAKKML mutant (Ivanovski et al., 2000) (Table 1).Therefore, the YIRN cluster is an important part of theAtxA/R180 interaction surface, but other parts of themolecule also contribute considerably to the bindingaffinity of AtxA. Interestingly, introduction of the N-terminal DPLA2 residues (K, E and/or W) to AtxA/DPLA2

YIRN

had a significant positive effect on the binding affinity for

ARTICLE IN PRESS

P. Prijatelj et al. / Toxicon 51 (2008) 1520–1529 1527

R180, confirming that residues in the N-terminal part ofthe molecule do play a role in binding to R180. Therefore,the results of this and previous studies collectivelyindicate that the interaction surface of AtxA and itsneuronal M-type receptor, R180, extend from theC-terminal part of the molecule, across the area close tothe Ca2+-binding loop and the IBS, to the N-terminal helixon the edges of the IBS.

3.4. Interfacial binding surface properties and enzymatic

activity of DPLA2 and Atxs

Atxs are very efficient sPLA2 enzymes (Petan et al.,2005), binding strongly to and rapidly hydrolyzing bothanionic and zwitterionic phospholipid bilayers, includingPC-rich mammalian plasma membranes. Additionally,their products of hydrolysis can cause extensive mem-brane and cell damage (Petan et al., 2007), which is incontrast to their potent and specific presynapticallyneurotoxic action. Clearly, no simple correlation betweenthe enzymatic activity of neurotoxic sPLA2s and theirlethal potency can be expected (Rosenberg, 1997; Punger-car and Krizaj, 2007). The role of enzymatic activity in theprocess is obscured by the numerous factors affectingboth sPLA2 activity and neurotoxicity, especially the, as yetunknown, (sub)cellular location, accessibility, composi-tion and physical properties of the target membrane.

Productive interfacial binding is a prerequisite for thesuccessful extraction and hydrolysis of membrane phos-pholipids by sPLA2s (Berg et al., 2001; Singer et al., 2002),including Atxs (Petan et al., 2005). AtxA binds to themembrane surface through its putative IBS residues, L2,L3, L19, T20, F24, V31, S67, K69, T70, R72, R118, N119 andF124 (Petan et al., 2005), while the equivalent IBS ofDPLA2 contains several substitutions, I19, P20, S24, W31,N67, S70, M118 and L119. The present study providesfurther understanding of the importance of certain surfaceresidues for the interfacial binding and activity of Atxs.The range of mutants analyzed enables us to pinpoint theIBS residues that are crucial for the previously observeddifferences in interfacial binding and activity of AtxA andDPLA2 (Petan et al., 2005). The fact that all the recombi-nant proteins used in this study had high and similaractivities on negatively charged POPG vesicles (Table 2) isindicative of high-affinity binding. The strong electrostaticforces between negatively charged phospholipid head-groups and the positively charged protein dominate theinteraction and mask the relatively small differences inthe IBS of the mutants (Singer et al., 2002; Petan et al.,2005). However, their enzymatic activities on zwitterionicPOPC vesicles were lower by two–three orders ofmagnitude. This is a consequence of the much loweraffinity of these sPLA2s for such surfaces (Petan et al.,2005), which enables even small differences in the IBS tobe reflected in significant variations in the measured ratesof hydrolysis (Berg et al., 2001). The variations in the ratesof hydrolysis observed on large unilamellar POPC vesiclescorrelate very well with those determined on mixedmicelles of egg-yolk PC and Triton X-100 (Table 2).

The YIRN cluster in AtxA includes the hydrophilic R118and N119, which are on the putative IBS, and thehydrophobic Y115 and I116, located close to, but notforming a part of, the presumed IBS (Fig. 3). Theintroduction of the YIRN cluster in DPLA2, or in thevarious AtxA/DPLA2 variants, resulted in a substantialdecrease of the initial rates of hydrolysis of charge-neutralPC vesicles (Table 2). This is in accordance with theexpected negative impact of lowering the hydrophobic/aromatic character of the IBS surface of Atxs on interfacialbinding and, consequently, enzymatic activity (Petanet al., 2005). Thus, the presence of the hydrophobic IBSresidues M118 and L119 in DPLA2 and the AtxAKKML

mutant, as well as of M118 and the aromatic Y119 in AtxB,in contrast to the hydrophilic R118 and N119 in AtxA, havebeen shown to significantly enhance interfacial binding toanionic and, especially, zwitterionic phospholipid surfaces(Ivanovski et al., 2000; Petan et al., 2005). Similarly,substitution of the hydrophobic M7 and nonpolar G11,which are on the edges of the presumed IBS surface ofAtxs, with the charged K7 and E11 present in DPLA2,resulted in reduced enzymatic activity. Furthermore,while the introduction of W31 to the AtxA/DPLA2

YIRN

chimera caused a very significant 50-fold increase in therate of hydrolysis of PC vesicles, the impact of the sameresidue, when introduced to the AtxAKE/DPLA2

YIRN mutant,was only 16-fold and the resulting activity was only 4-foldhigher than that of AtxA/DPLA2

YIRN. This is clear confirma-tion of the earlier suggestion (Petan et al., 2005) that thepresence of K7 and E11 on the IBS of DPLA2 is for the mostpart responsible for the relatively low enzymatic activityof DPLA2, despite the presence of W31, relative to AtxAand AtxA-V31W on PC-rich membranes (Petan et al.,2005). The results presented in Table 2 confirm that thewell-known positive impact of W31 on interfacial bindingand enzymatic activity of mammalian and venom sPLA2s(Han et al., 1999; Beers et al., 2003; Petan et al., 2005,2007) is in a large part counterbalanced by the presence ofK7 and E11 in the case of DPLA2. The electrostaticinteractions between K7 and E11 and the zwitterionicmembrane surface most probably result in an alteredorientation of DPLA2 on the membrane, preventing aproductive interaction of Trp-31 with the interface, and isunfavorable for efficient catalysis. Additionally, the nega-tive effect of the charged K7 and E11 is supplemented inpart by the presence of the polar S24 in DPLA2, which isevident from its negative impact on both interfacialbinding and enzymatic activity in AtxA-F24S (Petanet al., 2005). The most significant differences in interfacialbinding and enzyme activity of AtxA and DPLA2 aretherefore a consequence of the natural substitutions thathave led to significant changes in the hydrophobic/aromatic properties of residues on or near the IBS: M7K,G11E, F24S, V31W, R118M and N119L.

4. Conclusion

The results of this study demonstrate that thestructural determinants of snake venom sPLA2 presynap-tic neurotoxicity are not confined to a single ‘‘toxic’’ site

ARTICLE IN PRESS

P. Prijatelj et al. / Toxicon 51 (2008) 1520–15291528

on the surface of the molecule. Specific combinations ofC- and N-terminal residues collectively contribute to thephenomenon of presynaptic neurotoxicity, reflecting thecomplex nature of its molecular mechanism. The struc-tural features that confer high neurotoxic potency to AtxAextend from the C-terminal part of the molecule,across the IBS, to the N-terminal helix region. Residuesin the C-terminal part of AtxA are crucial for its high-affinity binding to the intracellular binding proteins, CaMand R25, while its interaction surface with the neuronalM-type sPLA2 receptor R180 in a large part coincides withthe distribution of residues important for the neurotoxiclethal effects. Our results also underline the fact that theinterfacial binding properties, and consequently theenzymatic activity of sPLA2s, depend on the jointcontributions of a number of residues and are fine tunedby the specific arrangement of hydrophobic and hydro-philic residues on or near the IBS.

Acknowledgments

We sincerely thank Dr. David C. Wilton for theexpression plasmid encoding rat FABP and Dr. Roger H.Pain for critical reading of the manuscript. This work wassupported by Grant P1-0207 from the Slovenian ResearchAgency.

References

Beers, S.A., Buckland, A.G., Giles, N., Gelb, M.H., Wilton, D.C., 2003. Effectof tryptophan insertions on the properties of the human group IIAphospholipase A2: mutagenesis produces an enzyme with character-istics similar to those of the human group V phospholipase A2.Biochemistry 42, 7326–7338.

Berg, O.G., Gelb, M.H., Tsai, M.D., Jain, M.K., 2001. Interfacial enzymology:the secreted phospholipase A2-paradigm. Chem. Rev. 101,2613–2654.

Bezzine, S., Koduri, R.S., Valentin, E., Murakami, M., Kudo, I., Ghomashchi,F., Sadilek, M., Lambeau, G., Gelb, M.H., 2000. Exogenously addedhuman group X secreted phospholipase A2 but not the group IB, IIA,and V enzymes efficiently release arachidonic acid from adherentmammalian cells. J. Biol. Chem. 275, 3179–3191.

Chang, C.C., 1985. Neurotoxins with phospholipase A2 activity in snakevenoms. Proc. Natl. Sci. Council. B ROC 9, 126–142.

Chang, C.C., Su, M.J., 1982. Presynaptic toxicity of the histidine-modified,phospholipase A2-inactive, b-bungarotoxin, crotoxin and notexin.Toxicon 20, 895–905.

Copic, A., Vucemilo, N., Gubensek, F., Krizaj, I., 1999. Identification andpurification of a novel receptor for secretory phospholipase A2 inporcine cerebral cortex. J. Biol. Chem. 274, 26315–26320.

Gubensek, F., Ritonja, A., Zupan, J., Turk, V., 1980. Basic proteins of Viperaammodytes venom. Studies of structure and function. Period. Biol. 82,443–447.

Han, S.K., Kim, K.P., Koduri, R., Bittova, L., Munoz, N.M., Leff, A.R., Wilton,D.C., Gelb, M.H., Cho, W., 1999. Roles of Trp31 in high membranebinding and proinflammatory activity of human group V phospho-lipase A2. J. Biol. Chem. 274, 11881–11888.

Hanasaki, K., Arita, H., 2002. Phospholipase A2 receptor: a regulator ofbiological functions of secretory phospholipase A2. ProstaglandinsOther Lipid. Mediat. 68–69, 71–82.

Ivanovski, G., Copic, A., Krizaj, I., Gubensek, F., Pungercar, J., 2000.The amino acid region 115–119 of ammodytoxins plays an importantrole in neurotoxicity. Biochem. Biophys. Res. Commun. 276,1229–1234.

Ivanovski, G., Petan, T., Krizaj, I., Gelb, M.H., Gubensek, F., Pungercar, J.,2004. Basic amino acid residues in the b-structure region contribute,but not critically, to presynaptic neurotoxicity of ammodytoxin A.Biochim. Biophys. Acta 1702, 217–225.

Jenko Praznikar, Z., Kovacic, L., Rowan, E.G., Romih, R., Rusmini, P., Poletti,A., Krizaj, I., Pungercar, J., 2008. A presynaptically toxic secretedphospholipase A2 is internalized into motoneuron-like cells where itis rapidly translocated into the cytosol. Biochim. Biophys. Acta—Mol.Cell Res., in press, doi:10.1016/j.bbamcr.2008.01.011.

Kasturi, S., Gowda, T.V., 1989. Purification and characterization of a majorphospholipase A2 from Russell’s viper (Vipera russelli) venom.Toxicon 27, 229–237.

Kini, R.M., 2003. Excitement ahead: structure, function and mechanismof snake venom phospholipase A2 enzymes. Toxicon 42, 827–840.

Krizaj, I., Gubensek, F., 2000. Neuronal receptors for phospholipases A2

and b-neurotoxicity. Biochimie 82, 807–814.Krizaj, I., Dolly, J.O., Gubensek, F., 1994. Identification of the neuronal

acceptor in bovine cortex for ammodytoxin C, a presynapticallyneurotoxic phospholipase A2. Biochemistry 33, 13938–13945.

Kudo, I., Murakami, M., 2002. Phospholipase A2 enzymes. ProstaglandinsOther Lipid Mediat. 68–69, 3–58.

Lambeau, G., Ancian, P., Nicolas, J.P., Beiboer, S.H., Moinier, D., Verheij, H.,Lazdunski, M., 1995. Structural elements of secretory phospholipasesA2 involved in the binding to M-type receptors. J. Biol. Chem. 270,5534–5540.

Lee, C.Y., Tsai, M.C., Chen, Y.M., Ritonja, A., Gubensek, F., 1984. Mode ofneuromuscular blocking action of toxic phospholipases A2 fromVipera ammodytes venom. Arch. Int. Pharmacodyn. Ther. 268,313–324.

Perkins, S.J., 1986. Protein volumes and hydration effects. The calcula-tions of partial specific volumes, neutron scattering matchpoints and280-nm absorption coefficients for proteins and glycoproteins fromamino acid sequences. Eur. J. Biochem. 157, 169–180.

Petan, T., Krizaj, I., Gubensek, F., Pungercar, J., 2002. Phenylalanine-24 inthe N-terminal region of ammodytoxins is important for bothenzymic activity and presynaptic toxicity. Biochem. J. 363, 353–358.

Petan, T., Krizaj, I., Gelb, M.H., Pungercar, J., 2005. Ammodytoxins, potentpresynaptic neurotoxins, are also highly efficient phospholipase A2

enzymes. Biochemistry 44, 12535–12545.Petan, T., Krizaj, I., Pungercar, J., 2007. Restoration of enzymatic activity in

a Ser-49 phospholipase A2 homologue decreases its Ca2+-indepen-dent membrane-damaging activity and increases its toxicity. Bio-chemistry 46, 12795–12809.

Prijatelj, P., Copic, A., Krizaj, I., Gubensek, F., Pungercar, J., 2000. Chargereversal of ammodytoxin A, a phospholipase A2-toxin, does notabolish its neurotoxicity. Biochem. J. 352, 251–255.

Prijatelj, P., Krizaj, I., Kralj, B., Gubensek, F., Pungercar, J., 2002. TheC-terminal region of ammodytoxins is important but not sufficientfor neurotoxicity. Eur. J. Biochem. 269, 5759–5764.

Prijatelj, P., Sribar, J., Ivanovski, G., Krizaj, I., Gubensek, F., Pungercar, J.,2003. Identification of a novel binding site for calmodulin inammodytoxin A, a neurotoxic group IIA phospholipase A2. Eur. J.Biochem. 270, 3018–3025.

Prijatelj, P., Vardjan, N., Rowan, E.G., Krizaj, I., Pungercar, J., 2006. Bindingto the high-affinity M-type receptor for secreted phospholipases A2

is not obligatory for the presynaptic neurotoxicity of ammodytoxinA. Biochimie 88, 1425–1433.

Pungercar, J., Krizaj, I., 2007. Understanding the molecular mechanismunderlying the presynaptic toxicity of secreted phospholipases A2.Toxicon 50, 871–892.

Pungercar, J., Krizaj, I., Liang, N.-S., Gubensek, F., 1999. An aromatic, butnot a basic, residue is involved in the toxicity of group-IIphospholipase A2 neurotoxins. Biochem. J. 341, 139–145.

Reed, L.J., Muench, H., 1938. A simple method of estimating fifty per centendpoints. Am. J. Hygiene 27, 493–497.

Renetseder, R., Brunie, S., Dijkstra, B.W., Drenth, J., Sigler, P.B., 1985. Acomparison of the crystal structures of phospholipase A2 frombovine pancreas and Crotalus atrox venom. J. Biol. Chem. 260,11627–11634.

Rigoni, M., Caccin, P., Gschmeissner, S., Koster, G., Postle, A.D., Rossetto,O., Schiavo, G., Montecucco, C., 2005. Equivalent effects of snake PLA2

neurotoxins and lysophospholipid-fatty acid mixtures. Science 310,1678–1680.

Rosenberg, P., 1997. Pitfalls to avoid in the study of correlations betweenenzymatic activity and pharmacological properties of phospholipaseA2 enzymes. In: Kini, R.M. (Ed.), Venom Phospholipase A2 Enzymes:Structure, Function and Mechanism. Wiley, Chichester, pp. 155–184.

Rouault, M., Le Calvez, C., Boilard, E., Surrel, F., Singer, A., Ghomashchi, F.,Bezzine, S., Scarzello, S., Bollinger, J., Gelb, M.H., Lambeau, G., 2007.Recombinant production and properties of binding of the full set ofmouse secreted phospholipases A2 to the mouse M-type receptor.Biochemistry 46, 1647–1662.

ARTICLE IN PRESS

P. Prijatelj et al. / Toxicon 51 (2008) 1520–1529 1529

Schaloske, R.H., Dennis, E.A., 2006. The phospholipase A2 superfamilyand its group numbering system. Biochim. Biophys. Acta 1761,1246–1259.

Singer, A.G., Ghomashchi, F., Le Calvez, C., Bollinger, J., Bezzine, S.,Rouault, M., Sadilek, M., Nguyen, E., Lazdunski, M., Lambeau, G., Gelb,M.H., 2002. Interfacial kinetic and binding properties of the completeset of human and mouse groups I, II, V, X, and XII secretedphospholipases A2. J. Biol. Chem. 277, 48535–48549.

Sribar, J., Copic, A., Paris, A., Sherman, N.E., Gubensek, F., Fox, J.W., Krizaj,I., 2001. A high affinity acceptor for phospholipase A2 withneurotoxic activity is a calmodulin. J. Biol. Chem. 276, 12493–12496.

Sribar, J., Copic, A., Poljsak-Prijatelj, M., Kuret, J., Logonder, U., Gubensek,F., Krizaj, I., 2003a. R25 is an intracellular membrane receptor for asnake venom secretory phospholipase A2. FEBS Lett. 553, 309–314.

Sribar, J., Sherman, N.E., Prijatelj, P., Faure, G., Gubensek, F., Fox, J.W.,Aitken, A., Pungercar, J., Krizaj, I., 2003b. The neurotoxic phospho-lipase A2 associates, through a non-phosphorylated binding motif,with 14-3-3 protein g and e isoforms. Biochem. Biophys. Res.Commun. 302, 691–696.

Tabor, S., 1990. Expression using the T7 RNA polymerase/promotersystem. In: Ausubel, F.A., Brent, R., Kingston, R.E., Moore, D.D.,Seidman, J.G., Smith, J.A., Struhl, K. (Eds.), Current Protocols inMolecular Biology. Greene Publishing & Wiley-Interscience, NewYork, pp. 16.2.1–16.2.11.

Thouin Jr., L.G., Ritonja, A., Gubensek, F., Russell, F.E., 1982. Neuromus-cular and lethal effects of phospholipase A2 from Vipera ammodytesvenom. Toxicon 20, 1051–1058.

Tsai, I.H., Wang, Y.M., 1998. Effect of site directed mutagenesis on theactivity of recombinant trimucrotoxin, a neurotoxic phospholipasefrom Trimeresurus mucrosquamatus venom. Toxicon 36, 1591–1597.

Valentin, E., Lambeau, G., 2000. Increasing molecular diversity ofsecreted phospholipases A2 and their receptors and binding proteins.Biochim. Biophys. Acta 1488, 59–70.

Vardjan, N., Sherman, N.E., Pungercar, J., Fox, J.W., Gubensek, F., Krizaj, I.,2001. High-molecular-mass receptors for ammodytoxin in pig aretissue-specific isoforms of M-type phospholipase A2 receptor.Biochem. Biophys. Res. Commun. 289, 143–149.

Vucemilo, N., Copic, A., Gubensek, F., Krizaj, I., 1998. Identification of anew high-affinity binding protein for neurotoxic phospholipases A2.Biochem. Biophys. Res. Commun. 251, 209–212.

Wilton, D.C., 1990. A continuous fluorescence displacement assay for themeasurement of phospholipase A2 and other lipases that releaselong-chain fatty acids. Biochem. J. 266, 435–439.

Worrall, A.F., Evans, C., Wilton, D.C., 1991. Synthesis of a gene for rat liverfatty-acid-binding protein and its expression in Escherichia coli.Biochem. J. 278, 365–368.

Yokota, Y., Higashino, K., Nakano, K., Arita, H., Hanasaki, K., 2000.Identification of group X secretory phospholipase A2 as a naturalligand for mouse phospholipase A2 receptor. FEBS Lett. 478, 187–191.