Eur. J. Biochem. 209, 635-641 (1992) 0FEBS 1992

Characterization and molecular cloning of neurotoxic phospholipases A2 from Taiwan viper ( Vipera russelli formosensis) Ying-Ming WANG, Pei-Jung LU, Chewn-Lang HO and Inn-Ho TSAI Institute of Biological Chemistry, Academia Sinica, Taipei, Taiwan, Republic of China (Rcccived June 2/July 30, 1992) - EJB 92 0779

Two phospholipases A2 (PLA2s), designated as RV-4 and RV-7 were purified from venom of the Taiwan Russell’s viper ( V$era russelli fnrmnsensis) by gel-filtration and reverse-phase HPLC. Their primary structures were solved by both protein sequencing and cDNA cloning and sequencing. The cDNA synthesized was amplified by the polymerase-chain reaction using a pair of synthetic oligonucleotide primers corresponding to the N- and the C-terminal flanking regions of the enzymes. The deduced amino acid sequences of RV-4 and RV-7 were 92% identical to those of the vipoxin and vipoxin inhibitor, respectively, from the Bulgarian Vipera a. ammodytes. RV-4 itself was neurotoxic, whereas RV-7 had much lower enzymatic activity and was not toxic. The low enzymatic activity of RV-7 may be attributed to five acidic residues at positions 7, 17, 59, 114 and 119, which presumably impair its binding to aggregated lipid substrates. Based on the sequence comparison among all the known group I1 PLA2s, residues 6, 12, 76-81, and 119-125 were identified as important for the neurotoxicity. RV-4 and RV-7 exist in the crude venom as heterodimers, which were again formed by mixing together the HPLC-purified RV-4 and RV-7. Moreover, RV-7 inhibited the enzymatic activity of RV-4 in vitro but potentiated its lethal potency and neurotoxicity. It is suggested that RV-7 may facilitate the specific binding of RV-4 to its presynaptic binding sites, probably by preventing its non-specific adsorption. The extracellular phospholipases A2 (PI,A2) constitute a large family of homologous 14-kDa proteins which are major components of snake venoms [l- 31. The venom PLA2s, in the presence of Ca2+, catalyze the hydrolysis of the 2-acyl ester bond of 1,2-diacyl-sn-3-phosphoglycerol lipids of the biological membranes or aggregated phospholipid. Many of the enzymes have been shown to possess different pharmacological actions : neurotoxic [4 61, cardiotoxic, hemolytic, myonccrotic, anticoagulant [7], convulsant, hypotensive [8, 91 or edema-inducing action [lo]. From a structural-biology perspective, such a functional diversity within this group of structurally similar proteins raises questions of the relationship between their structure and their effect; for example. one may wonder what structural features are responsible for the choice between lipid interface and specific acceptor protein as the first target [5, 11 - 131 or for the capability of forming a complex with another PLA2 leading to an increase of toxicity [4,141. The neurotoxic PLA2s may be used as probes to study neurotransmitter release, the role of membrane phospholipids in synaptic transmission, or irreversible neural damage [4, 51. ~

Corrrspondmce to I.-H. Tsai, Institute of Biological Chemistry, Academia Sinica, P. 0. Box 23-106, Taipei, Taiwan 10798. Republic of China Ahhreviation.y. PCR, polymerase chain reaction; PLA2. phospholipase Az; Hxo2GroPCho, a-L-dihexanoylglycerophosphocholine; Pam2GroPCho, a-L-dipalmitoylglycerophosphocholine;RP-HPLC, reverse-phase high-performance liquid chromatography ; RV-4/RV-7 (1 :l), one-to-one complex of' RV-4 and RV-7. Enzyme. Phospholipase Az (EC 3.1.1.4). Note. The novel nucleotide sequence data published herc havc been deposited with the EMBL sequcnce data bank.

The neurotoxicity appears to arise from specific binding and hydrolysis at sites on the presynaptic membrane that are critical for transmitter release, and is probably mediated by both PLA,-independent and PLA2-dependent steps [4, 5, 15, 161. It has been demonstrated that some of the purified PLA2s from the venom of Russell’s viper are neurotoxic [8, 17, 181. However, it was not possible to analyze their toxic sites because of the lack of sequence data. In the present work, we determined the amino acid sequences of the neurotoxic PLA2s from Taiwan Russell’s viper venom efficiently by cDNA cloning and sequencing, and characterized their toxic domains. We found the toxic PLA2 is a complex containing both a basic, weakly toxic, and an acidic, nontoxic, PLAz.

MATERIALS AND METHODS Materials The venom powder and fresh glands of Vipera russelli jormosensis (Taiwan Russell’s viper) were provided by Dr M. Y. Liao (Institute of Preventive Medicine, Taipei, Taiwan). The [ c ~ - ~ ~ P ] ~was A Tpurchased P from Amersham (Les Ulis, France). Sequencing kits were from United States Biochemical Corp. (Cleveland, USA). Modification and restriction enzymes were from Boehringer Mannheim (Mannheim, FRG). Phospholipids were purchased from Avanti Polar lipids (Alabaster, AL, USA). Other chemicals were from Merck (Darmstadt, FRG), Aldrich (Milwaukee, USA) or Sigma Chemical Co. (St Louis, USA).

636 Purification of phospholipase A2 About 90 mg of the crude venom of Taiwan Russell’s viper was fractionated on a Sephadex G-100 column (1.5 x 120 cm; Pharmacia, Sweden) in 0.1 M ammonium acetate pH 6.8 at 4°C. The PLA, fractions was further purified by reversephase(RP) HPLC using a column of silica gel (ODS-H, 5 pm, 3 x 25 cm) (Chemcosorb Scientific Co., Osaka) in 0.07% aqueous trifluoroacetic acid, eluting with a 25 - 50% linear gradient of CH3CN with 0.07% trifluoroacetic acid. Protein quantitation and PLAz assay The dye-staining method of Bradford was used for quantitation of PLA, [19]. Bovine serum albumin (1 mg/ml, Azso = 0.56) was used to establish the standard curve. PLA, activity was measured by the pH-stat titration method [lo]. Dipalmitoylglycerophosphocholine(3 mM) was mixed with 3 mM sodium deoxycholate or 6 mM Triton X100 and 100 mM NaCl in a glass/Teflon tissue homogenizer. CaCI, was added just before addition of the enzyme. The liberated fatty acid was titrated with 8 mM NaOH at 37°C and pH 7.4 using a pH-stat apparatus (Radiometer, RTS 822, Denmark). The reaction rate was corrected for non-enzymatic spontaneous rate which was usually below 20% of the enzymic rate. SDS/PAGE analysis and isoelectric focusing The molecular masses of the PLAzs were analyzed by SDS/ PAGE according to Laemmli [20]. The gel was stained with Coomassie blue and the mobilities of proteins were plotted against the logarithm of the molecular masses of some standard proteins (phosphorylase b, 94 kDa; bovine serum albumin, 67 kDa; ovalbumin, 43 kDa; carbonic anhydrase, 30 kDa; soybean trypsin inhibitor, 20.1 kDa; a-lactalbumin, 14.4 kDa). Isoelectric focusing was performed according to the method in the manual of Hoefer Scientific Instruments Co. using Ampholyte of pH 3.5 - 10.

prepared using an oligo(dT)-cellulose column (Pharmacia). Complementary DNA (cDNA) was synthesized from 1 pg poly(A)-rich RNA by using a cDNA synthesis system plus kit (Amersham, England).

cDNA amplification, cloning and sequencing In order to amplify the PLA, cDNA,the polymerase chain reaction (PCR) [23] was conducted by using the cDNA of the total mRNA of the venom gland as template. Two mixed-base oligonucleotide primers (see below) were designed based on the highly conserved cDNA regions for some group 11 PLA2s [24] from Viperu snake venoms 125- 271. Both primers were synthesized using an Applied Biosystems model 381A DNA synthesizer. Primer 1 (2 1 residues) :

5’ ...... TCTGGATT(z)AGGAGGATGAGG .....3’ Met Arg .... (noncoding region) (signal peptide .....) Primer 2 (18 residues) :

5’. ..... GCCTGCAGAGACTTAGCA.. ....3’ (noncoding region) (stop cys133 .....) codon The PCR procedures were performed with Vent@DNA polymerase (Biolabs, New England, USA). As shown by 1YO agarose gel electrophoresis, a 0.45-kb fragment was specifically amplified. After being treated with Klenow fragment and polynucleotide kinase [22], the amplified DNA was inserted into the HincTI site of the pGEM-4Z plasmid (Promega Biotec., Wisconsin, USA) and transformed into Escherichiu coli strain JMlOl. White transformants were picked-up and both strands of the cDNA were sequenced by the dideoxynucleotide method using a sequencing kit. RESULTS Purification of PLA,s from Viperu r. formosensis venom

Assays of neurotoxicity and other pharmacological effects Neurotoxicity was assessed using the biventer cervicis nerve-muscle preparation of 3 - 10-day-old chicks. The contraction was stimulated indirectly with a stimulator and recorded isometrically by means of a force displacement transducer coupled with a polygraph (model 79D, Grass Instrument Co., USA) [4,21]. The platelet aggregation experiments were performed as described previously ”7, 101. The lethal potencies of RV-4 and RV-7, alone or together, was assessed on ICR mice (1820 g) by intravenous injection of the appropriate dose in 50 p1 0.15 M NaC1. The animals injected were constantly observed for symptoms of neurotoxicity. The number of dead mice was scored after 24 h. Amino acid sequence determination N-terminal amino acid sequences of the purified PLA,s were determined by automated gas-phase sequencer (model 477A, Applied Biosystems) with an on-line phenylthiohydantoin amino acid analyzer using the Normal-I program. Construction of venom gland cDNA Total RNA was isolated by the guanidinium thiocyanate and cesium chloride method [22] and poly(A)-rich RNA was

After being fractionated by the Sephadex G-100 gel permeation chromatography, the V . r. formosensis venom was separated into five major protein fractions (Fig. 1). The PLA, activity was found mainly in fraction I11 and was relatively weak in fraction 11. Fraction I11 was further separated by RPHPLC into seven peaks (Fig. 2). The PLA, activities were associated with two major peaks (RV-4 and RV-7), which together contribute more than 90% of the protein absorbance of fraction 111. Fraction I1 could similarly be fractionated into two relatively small PLA, peaks at the same elution times as RV-4 and RV-7.

Physicochemical properties and the N-terminal sequences Based on the results of SDS/PAGE, RV-4 and RV-7 appeared, after HPLC purification, to be homogeneous; the subunit masses were about 14 kDa (Fig. 2, inset). The isoelectric points of RV-4 and RV-7 determined by isoelectric focusing were found to be 10 and 4.3, respectively. When the native molecular masses and their mutual interaction were studied with a HPLC gel-filtration column (TSK G2000SW, 0.75 x 30 cm) at pH 6.8, it was found that RV-4 eluted alone at a retention time of 22.4 min (corresponding to 12.5 kDa; Fig. 3A) and RV-7 formed homodimers which eluted at 17.1 min (corresponding to 30 kDa; Fig. 3B); a 1: 1 complex

631

0

0

0

10

20

30

I 1 11

(0

N

a

50

30

10

70

90

Fraction No. Fig. 1. Gel permeation chromatography of Vipera r. formosensis venom on Sephadex 6-100. The column was equilibrated and eluted with 0.3 M ammonium acetate pH 6.8; 2-ml fractions were collected and the flow rate was adjusted to 18 ml/h. Fractions I1 and TI1 had PLA2 activities

-

0 0

RV7 RV4M

RV-4

Retention time (min)

Fig. 3. Study of complex formation between RV-4 and RV-7. A HPLC gel-filtration column (TSK GZOOOSW) was equilibrated and eluted with 0.15 M CH3COONH4 pH 6.8 at 0.5 ml/min and room temperature. The column was first calibrated with molecular mass standards: (I), bovine serum albumin (67 kDa), (2) ovamucoid trypsin inhibitor (45 kDa), (3) a-chymotrypsinogen (25 kDa) and (4) cytochrome c (13.5 kDa). The arrows denote their retention times. In separate experiments, the crude or HPLC-purified RV-4 and RV-7 in the elution bufferwereapplied.(A)0.15 mgRV-4;(B)0.15 mgRV-7;(C)0.15 mg RV-4 plus 0.15 mg RV-7; (D) 0.3 mg fraction 111 (Fig. 1).

1 0

,

0

R

J L

10

20

30

Fig. 2. RP-HPLC of fraction 111from Sephadex G-100. A Chemcosorb RP column (ODS-H, 5 pm, 1 x 25 cm) was used to isolate PLA2 from fraction 111 (Fig. 1). Solvent A was 0.07% trifluoroacetic acid in H 2 0 , solvent B was 0.07% trifluoroacetic acid in CH,CN, and the elution was by a linear gradient (25 - 50%) over 36 min at a flow rate of 2 ml/ min. Fractions denoted RV-4 and RV-7 have PLA2 activities. Inset: SDSjPAGE showing the purity and molecular mass of RV-4 and RV7. Lane M, molecular mass markers, from top to bottom: 67,43, 30, 20.1 and 14 kDa.

formed upon mixing of RV-4 and RV-7 (Fig. 3C) eluted at 18.1 min (corresponding to 27 kDa). This complex, denoted RV-4/RV-7(1: 1), was the major form of PLA2 in fraction 111 of Fig. 1 (Fig. 3D). The dimeric complex obtained in Fig. 3D gave the same twin peaks of RV-4 and RV-7 as Fig. 2 upon RP-HPLC separation. The absorption coeffcents at 280 nm of the 1% solution of RV-4 and RV-7 in NH4HC03 pH 7.6

were 15.6 and 15.0, respectively. The ratio of the absorbances of RV-4 and RV-7 peaks in Fig. 2 is about 15.6: 15.0, suggesting that they exist in equal amount and form a complex in the crude venom. The N-terminal 50 amino acid residues of pure RV-4 and RV-7 were analysed by automated sequencer successfully. Enzymatic and pharmacological activities of RV-4 and RV-7 The enzymatic hydrolysis of the micellar substrates (Pam,GroPCho plus detergents) and the monomeric substrate (2.0 mM Hxo2GroPCho) by RV-4 or RV-7 were studied by the pH-stat titration method; the results are shown in Table 1. RV-4 hydrolyzed the mixed micelles of Pam2GroPCho plus Triton X-100 faster than those of Pam2GroPCho plus deoxycholate. The enzyme activity of RV-7 was about 100-fold lower than that of RV-4 but it cannot be attributed to minor contamination by other active PLAz since additional purification steps(e.g. HPLC with TSK

638 Table 1. Enzymatic activities of RV-4, RV-7 and RV-4/RV-7 (1:l). The initial rate was measured in a titrimeric assay at pH 7.5 and 37 ' C in the presence of 1.0 mM Ca2+ and 0.1 M NaCI, using 2 mM La-dihexanoylglycerophosphocholine(Hxo,GroPCho), or detergentmixed micellcs of 3 mM synthetic 1.-a-dipalmitoylglycerophosphocholine (Pam2GroPCho) and 6 mM Triton X-100 or 3 mM deoxycholate. The concentrations of PLAzs used were 1 - 10 nM. Enzymc

Spccific activity on PamzGroPCho Pam2GroPCho HxozGroPCho deoxycholate triton X-100 (monomeric)

+

+

twice the amount of RV-7 caused a marked shortening of the blocking time to 53 & 6 min ( n = 4). A mixture containing non-toxic doses of both RV-4 (0.35 pg/ml) and RV-7 (1.0 pg/ ml) caused neuromuscular blockade in the chick muscle. RV-7 was also found to increase the lethal potency of RV4 in mice. The LD,, (intravenous) of RV-4 and its RV-7 complex were estimated to be 0.32 and 0.15 pg/g mice, respectively. RV-7 alone did not kill any mouse at a dose as high as 10 pg/g. The treated mice showed neurotoxic symptons such as hind-limb paralysis, convulsion and respiratory distress. cDNA cloning and sequencing

1000 1.5 460

RV-4

304 2.4 RV-4/RV-7( 1:1 ) 192

RV-7

21 0.5 9

RV-7

Ach d V - 7

2bms