Sequences and structural organization of phospholipase A 2 genes from Vipera aspis aspis , V. aspis zinnikeri and Vipera berus berus venom Identification of the origin of a new viper population based on ammodytin I1 heterogeneity Isabelle Guillemin*, Christiane Bouchier†, Thomas Garrigues‡, Anne Wisner§ and Vale ´ rie Choumet* Unite ´ des Venins, Institut Pasteur, Paris, France We used a PCR-based method to determine the genomic DNA sequences encoding phospholipases A 2 (PLA2s) from the venoms of Vipera aspis aspis (V. a. aspis), Vipera aspis zinnikeri (V. a. zinnikeri), Vipera berus berus (V. b. berus) and a neurotoxic V. a. aspis snake (neurotoxic V. a. aspis) from a population responsible for unusual neurotoxic envenomations in south-east France. We sequenced five groups of genes, each corresponding to a different PLA2. The genes encoding the A and B chains of vaspin from the neurotoxic V. a. aspis,PLA2-IfromV. a. zinnikeri,andthe anticoagulant PLA2 from V. b. berus are described here. Single nucleotide differences leading to amino-acid substi- tutions were observed both between genes encoding the same PLA2 and between genes encoding different PLA2s. These differences were clustered in exons 3 and 5, potentially altering the biological activities of PLA2. The distribution and characteristics of the PLA2 genes differed according to the species or subspecies. We characterized for the first time genes encoding neurotoxins from the V. a. aspis and V. b. berus snakes of central France. Genes encoding ammodytins I1 and I2, described previously in Vipera ammodytes ammodytes (V. am. ammodytes), were also pre- sent in V. a. aspis and V. b. berus. Three different ammo- dytin I1 gene sequences were characterized: one from V. b. berus, the second from V. a. aspis, V. a. zinnikeri and the neurotoxic V. a. aspis, and the third from the neurotoxic V. a. aspis. This third sequence was identical with the reported sequence of the V. am. ammodytes ammodytin I1 gene. Genes encoding monomeric neurotoxins of V. am. ammodytes venom, ammodytoxins A, B and C, and the Bov-B LINE retroposon, a phylogenetic marker found in V. am. ammodytes genome, were identified in the genome of the neurotoxic V. a. aspis. These results suggest that the population of neurotoxic V. a. aspis snakes from south-east France may have resulted from interbreeding between V. a. aspis and V. am. ammodytes. Keywords: ammodytin; neurotoxic; phospholipase A 2 ; vaspin; viper. Phospholipases A 2 (PLA2s) are major components of snake venoms. They catalyze the Ca 2+ -dependent hydrolysis of the 2-acyl ester bond of 1,2-diacyl-3-sn-phosphoglycerides releasing fatty acids and lysophospholipids. These enzymes can be separated into 11 groups. Those belonging to group II have six to eight disulfide bonds and a C-terminal extension not present in group I venom PLA2s [1]. They are found in the venoms of Crotalinae and Viperinae snakes and in human platelets, liver and spleen [2,3]. Snake venoms contain a large number of PLA2 isoenzymes which differ in neurotoxicity, myotoxicity, cardiotoxicity, anticoagulation and edema-inducing properties [2]. To date, the structures of only six Viperinae PLA2 genes have been studied: ammodytin I1 (DDBJ/EMBL/GenBank accession no. AF253048), ammodytin I2 (DDBJ/EMBL/ GenBank accession no. X84018), ammodytoxin C (DDBJ/ EMBL/GenBank accession no. X76731) and ammodytin L (DDBJ/EMBL/GenBank accession no. X84017) from Vipera ammodytes ammodytes (V. am. ammodytes)andtwo genes encoding an acidic inhibitor (VP7) (DDBJ/EMBL/ GenBank AC AF373342) and a basic PLA2 protein (VP8) (DDBJ/EMBL/GenBank AC AF373342) from Vipera palaestinae venom [4–6]. All these genes are composed of five exons and four introns, like genes encoding human Correspondence to V. Choumet, Unite ´ de Biochimie et de Biologie Mole ´ culaire des Insectes, 25, Rue du Docteur Roux, 75724 Paris Cedex 15, France. Fax: + 33 1 40 61 34 71, Tel.: + 33 1 45 68 86 30, E-mail: vchoumet@pasteur.fr Abbreviation: PLA2, phospholipase A 2 . *Present address:Unite ´ de Biochimie et Biologie Mole ´ culaire des Insectes, Institut Pasteur, 25, Rue du Dr Roux, 75724 Paris Cedex 15, France. Present address: Genopole, Institut Pasteur, 28, Rue du Docteur Roux, 75724 Paris Cedex 15, France. àPresent address:Unite ´ d’Ecologie des Syste ` mes Vectoriels, Institut Pasteur, 25, Rue du Docteur Roux, 75724 Paris Cedex 15, France. §Present address: Laboratoire de Recherche et de De ´ veloppement: Pharmacologie des Re ´ gulations Neuroendocriniennes, Paris, France. Note: The nucleotide sequences reported in this paper have been deposited in the DDBJ/EMBL/GenBank nucleotide sequence data- bases under accession numbers AY158634, AY158635, AY158636, AY158637, AY158638, AY158639, AF548351, AY152843, AY159807, AY159808, AY159809, AY159810, AY159811, AY243574, AY243575, AY243576, AY243577. (Received 2 December 2002, revised 4 April 2003, accepted 22 April 2003) Eur. J. Biochem. 270, 2697–2706 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03629.x group II PLA2s. In contrast, PLA2 genes from Crotalinae snakes such as Trimeresurus flavoviridis, Trimeresurus gramineus, Trimeresurus (Ovophis) okinavensis and Crotalus scutulatus scutulatus are organized into four exons and three introns, like group I PLA2s [7–10]. Despite this difference, nucleotide sequence analyses have shown that, unusually, introns are more conserved than exons in both Viperinae and Crotalinae. Moreover, mutations leading to amino-acid changes are common in the protein-coding regions (but not in the signal peptide exon), but are limited to the third exon in V. palaestinae [6,11]. Thus, the genes encoding the group II PLA2 of snake venoms evolved by gene duplication, followed by divergence from a common ancestral gene by accelerated Darwinian selection, probably as a means of acquiring new functions [9,11]. In this paper, we extend the study of PLA2 genes to the French vipers Vipera aspis aspis (V. a. aspis), Vipera aspis zinnikeri (V. a. zinnikeri), and Vipera berus berus (V. b. berus). We were prompted to carry out this study by the recent identification of a distinct, unusually neurotoxic population of V. a. aspis in the south-east of France [12]. Several cases of envenomation resulting in symptoms of neurotoxicity have been reported in recent years in two French de ´ partements (Alpes-Maritimes and Alpes-de- Haute-Provence). Such symptoms have never been observed after envenomation by V. a. aspis snakes in other regions of France. The venom of these snakes was detected by ELISA in the plasma of the patients; it contained PLA2s that cross- reacted with antibodies to ammodytoxin [12]. We investi- gated gene expression in the venom gland of one of the snakes captured after one case of human envenomation. We showed, by RT-PCR, that the venom of this snake contained two neurotoxins. One was monomeric and identical with ammodytoxin B, and the other (vaspin) was heterodimeric and similar to vipoxin, the toxic complex of Vipera ammodytes meridionalis snake venom [13]. These two toxins were responsible for the symptoms of neurotoxicity observed after envenomation [13]. We then investigated the genes encoding the PLA2s present in the venoms of all the venomous snake species of France. On the basis of its venom PLA2 characteristics and, particularly, sequence analysis of the ammodytin I1 gene, we suggest a possible origin for the neurotoxic snake population. We also report high levels of polymorphism, both for individual PLA2 genes and between genes encoding different PLA2s. These polymorphisms may have implications for the structure and/or function of the enzyme. Experimental Procedures Snakes were captured in various regions of France: V. a. aspis and V. b. berus in the Puy-de-Doˆ me, V. a. zinnikeri in the Gironde, and neurotoxic V. a. aspis in the Alpes-Maritimes. The V. a. aspis snake captured in the Alpes-Maritimes was responsible for one case of neurotoxic envenomation [12]. We studied one individual per snake species. Genomic DNA was extracted from snake livers as previously described [14]. DNA was amplified with a set of primers (Genset Oligos, Paris, France) targeting conserved regions of the PLA2 genes for which sequences were available in databases. The primer-binding sites were located upstream from the 5¢-UTR (PLA5G) and downstream from the 3¢-UTR (PLA3G). Amplification reactions were carried out in a final volume of 50 lL containing 2.5 lLPLA5Gand 2.5 lLPLA3G(10l M each), 1 lLdNTPs(dATP,dCTP, dTTP and dGTP, 10 m M each), 5 lL Taq buffer supplied with the enzyme, 0.25–0.5 lg genomic DNA and 1.5 U rTaq polymerase (Amersham Biosciences, Orsay, France). The DNA was denatured by heating at 95 °Cfor7min.It was then subjected to 30 amplification cycles as follows: denaturation at 95 °C for 1 min and annealing coupled with extension at 69 °C for 6 min. A final extension step was carried out, at 72 °C for 10 min. The reaction product was analyzed by agarose gel electrophoresis in 1 · Tris/borate buffer. DNA fragments of the expected size (2.1 kb) were purified from the gel with the QIAquick Gel Extraction Kit (Qiagen S.A., Courtabœuf, France), and inserted into the pCRÒ2.1-TOPOÒ vector of the TOPO TA CloningÒ Kit (Invitrogen SARL, Cergy Pontoise, France). Plasmid DNA was purified with the Montage Plasmid Miniprep 96 Kit (Millipore, Saint-Quentin-en-Yvelines, France). Sequen- cing reactions were performed from both ends of the DNA plasmid, using the ABI PRISM BigDye Terminator Cycle Sequencing Ready-Reaction Kit, and a 3700 Genetic Analyzer (Applied Biosystems). The trace files were base-called with Phred [15]. Sequences not meeting our production quality criteria (at least 100 bases with a quality over 20) and insert-less vector sequences (detected by cross-matching; [15]) were discarded. Complete nuc- leotide sequences were determined on both strands, with a set of nine primers, designed from database sequences (Table 1). Results and Discussion Identification of PLA2 genes We obtained 96 clones of PLA2 genes per snake species. Complete nucleotide sequences were further analyzed for 81 clones of V. a. aspis, 80 clones of V. b. berus,65of V. a. zinnikeri, and 59 of the neurotoxic V. a. aspis.Nuc- leotide and amino-acid sequences were compared with sequences in gene and protein databases, using BLASTN and BLASTP , respectively. We subsequently amplified the corres- ponding genomic DNA fragments from each snake with primers (Table 1) specific for the PLA2s previously charac- terized in Vipera venoms. We also designed primers specific for the Bov-B LINE retroposon, a phylogenetic marker previously identified in some PLA2 genes of Viperidae snakes including V. am. ammodytes [4,5]. Five groups of snake venom PLA2 genes were sequenced. Nucleotide polymorphism was identified in each group. Two groups of genes were most similar, in terms of nucleotide sequences, to cDNAs encoding chains A and B of vaspin (DDBJ/EMBL/GenBank accession no.s AJ459806 and AJ459807, respectively) [13], chains A and B of vipoxin (Gi numbers: 16974941 and 16974940) from V. am. meridionalis [16] and the two subunits of PLA2-I (Gi numbers: 1709547 and 1709548) from V. a. zinnikeri [17]. They also showed a high level of nucleotide identity with cDNAs encoding the presynaptic neurotoxic complex RV4/ RV7 from Daboia russelli formosensis (DDBJ/EMBL/ 2698 I. Guillemin et al.(Eur. J. Biochem. 270) Ó FEBS 2003 GenBank accession no.s X68385 and X68386, respectively [18]). Two other groups were most similar, in terms of nucleotide sequence, to the ammodytin I1 (DDBJ/EMBL/ GenBank accession no. AF253048) and ammodytin I2 (DDBJ/EMBL/GenBank accession no. X84017) genes from V. am. ammodytes, respectively [19]. The last group of genes was most similar to the V. b. berus PLA2 protein (Gi number: 423975) [20]. A list of all the venom PLA2 genes identified in each captured snake is presented in Table 2. Unexpectedly, genes encoding the A and B chains of vaspin, a heterodimeric neurotoxin, were identified in V. a. aspis and V. b. berus snakes collected in central France (Table 2). These genes are probably either expressed at a very low level or not expressed at all in the venom of these snakes because no neurotoxic envenomation has ever been reported in the Clermont-Ferrand region (Gabriel Montpied Hospital, personal communication). The expres- sion of these neurotoxin subunits in the venom of the neurotoxic V. a. aspis may be related to the diet of the snake, but this hypothesis remains to be proven [21]. More surprisingly, genes encoding ammodytoxins A, B and C were identified only in the neurotoxic V. a. aspis snake (Table 2). However, only the ammodytoxin B mRNA was detected in the venom gland of this snake [13]. Thus, not all the ammodytoxin genes were expressed. Finally, partial sequencing of the PCR product revealed the presence of the Bov-B LINE retrotransposon in intron D of the ammodytoxin C gene of the neurotoxic V. a. aspis (Table 2). This feature was also specific to this population of neurotoxic snakes. Table 1. Sequences of primers used for PCR amplification and sequencing of PLA2 genes. F, Forward; R, reverse. Y ¼ CorT; W ¼ AorT. Primer name Sequences 5¢)3¢ M13Reverse (F) CCCTATAGTGAGTCGTATTA T7 Promoter (R) CAGGAAACAGCTATGAC PLA5G (F) CGGAATTCTGAAGGTGGCCCGCC AGGTGACAG PLA3G (R) CGCGGATCCAATCTTGATGGGGC AGCCGGAGAGG PLA5G1 (F) AGGAYTCTCTGGATAGTGG PLA3G1 (R) CTCACCACAGACGATWTCC PLA5G2 (F) CGGTAAGCCCATAACGCCCA PLA3G2 (R) CAGGCCAGGATTTGCAGCC PLA3G4 (R) CATAAACAYGAGCCAGTTGCC ARTF a (F) GAGTGGATGCACAGTCGTTG ARTR a (R) GAAACGGAGGTAGTGACACAT AtxBF b (F) GCCTGCTCGAATTCGGGATG AtxBrc b (R) CTCCTTCTTGCACAAAAAGTG AtxACF c (F) CTGCTCGAATTCGGGATG AtxACrc c (R) GTCYGGGTAATTCCTATATA AmlF d (F) GTGATCGAATTTGGGAAGATGATCCA Amlrc d (R) CCCTTGCATTTAAACCTCAGGTACAC a Specific primers used for amplification of the Bov-B LINE retroposon; b specific primers used for amplification of the ammo- dytoxin B gene; c specific primers used for amplification of the ammodytoxin A and C genes; d specific primers used for amplifica- tion of the ammodytin L gene. Table 2. Characteristics of the venom PLA2 genes of V. a. as pi s, V. a. zinnikeri,neurotoxicV. a. aspis and V. b. berus. The PLA2 content of V. am. ammodytes venom is as previously reported [2,4]. Snake species PLA2 genes a Length of intron D in ammodytin I1 (bp) AmI1(form) AmI2 Vb VaspA VaspB V. a. aspis + (Ia) +–+ c + c 133 V. a. zinnikeri + (Ia) – – + + 133 Neurotoxic V. a. aspis + + – + + 133/259 1st group of clones + (Ia) 133 2nd group of clones + (In) 259 V. b. berus + (Ib) + + + c + c 259 V. am. ammodytes + (In) + – – – 259 Snake species Ammodytin I1 protein sequence b AtxA AtxB AtxC AmL Retroposon V. a. aspis L70, S71, E78, L12 – – – – – V. a. zinnikeri L70, S71, E78, L12 – – – – – Neurotoxic V. a. aspis + c + c + c –+ c (AtxC) 1st group of clones L70, S71, E78, L123 2nd group of clones M70, G71, Q78, F123 V. b. berus T3 (peptide signal) – – – + c + c (AmL) N1 K56 V. am. ammodytes M70, G71, Q78, F123 + + + + + (AtxC, AmL) a AmI, ammodytin I1; AmI2, ammodytin I2; VaspA, vaspin chain A; VaspB, vaspin chain B; AtxA, ammodytoxin A; AtxB, ammodytoxin B; AtxC, ammodytoxin C; AmL: ammodytin L; b Only amino acids differing between ammodytin I1 molecules are represented. The isoform of ammodytoxin I is indicated in parentheses, as shown in Fig. 3. c The genes were identified by PCR and partially sequenced. Ó FEBS 2003 Genomic analysis of phospholipases A 2 from French viper venoms (Eur. J. Biochem. 270) 2699 Structural organization of PLA2 genes We report here the first genomic sequences for the genes encoding vaspin, a heterodimeric neurotoxin from Vipera snakes, and PLA2 from V. b. berus. The nucleotide sequences of the genes encoding chains A and B of vaspin and V. b. berus PLA2 span about 1.9 kb, as do the ammodytin I1 and I2 genes. They show a similar organiza- tion to Viperinae PLA2s, with five exons separated by four introns (Table 3). Exons 3, 4 and the 5¢ part of exon 5 encode the mature protein; exon 1 encodes the 5¢-UTR, and exon 2 and part of exon 3 encode the signal peptide. The 110 nucleotides at the 3¢ end of exon 5 encode the 3¢-UTR. The 5¢ donor and 3¢ acceptor splice sites conformed with the GT/AG rule (Table 3). These features are common to Viperinae PLA2 genes [1,4]. Regardless of the snake from which the PLA2 gene was obtained, the lengths of the exons and the 5¢-UTR were identical, except for exon 5 in the ammodytin I2 gene, which was three nucleotides shorter than the equivalent exon in the other PLA2 genes (Table 3). The ammodytoxin C, ammo- dytins I1 and I2 and ammodytin L genes of V. am. ammo- dytes, the VP7 and VP8 genes of V. palaestinae,andthe PLA2 genes of Crotalinae snakes (Trimeresurus flavoviridis) were also of similar length [4–6,8]. Interestingly, a 476 bp insertion was observed in the 3¢-UTR of the ammodytin I2 gene from V. b. berus. This fragment was similar to a region located upstream from the TATA-box-binding protein gene of T. gramineus and T. flavoviridis [22], suggesting a probable common ancestry of V. b. berus and Trimeresurus species. ThelengthofintronDintheammodytinI1gene depended on the species. It was 133 bp long in V. a. zinnikeri and V. a. aspis whereas it was 259 bp long in V. b. berus and V. am. ammodytes (DDBJ/EMBL/ GenBank accession no. AF253048). Interestingly, introns of both lengths were found in the genome of the neurotoxic V. a. aspis, with six of the 10 sequenced clones having the 126 bp deletion as in V. a. aspis, V. a. zinnikeri,andthe remaining four clones having an intron D similar to that of the V. am. ammodytes ammodytin I1 gene. All PLA2 genes contained a TAA stop codon, an AATAAA polyadenylation site 80 bp downstream from the stop codon, and a TATA-like box (CATAAAA) 270 bp upstream from the ATG translation initiation codon, as found in other Viperinae and Crotalinae genes [2,22]. Table 3. Structural organization of V. a. aspis, V. a. zinnikeri, V. b. berus and neurotoxic V. a. aspis PLA2 genes. PLA2 gene (no. of clones) a Exon Exon length (bp) Intron Intron length (bp) Splice sites 5¢donor/3¢acceptor Ammodytin I1 1 66 A 163 CAGCTgtaag/tccagGTCTG (88) 2 56 B 241–244 AGGCGgtgag/caaagCTGAA 3 133 C 671–680 GACCGgtaag/tccagCTGCT 4 101 D 133–259 CTGTGgtgag/tgcagGAGGC 5(3¢-UTR) 140 (110) Ammodytin I2 1 66 A 163 CAGCTgtaag/tccagGTCTG (59) 2 56 B 243 AGGCGgtgag/tccagTTGAA 3 133 C 662–670 GACCGgtaag/tccagCTGCT 4 101 D 133 CTGTGgtgag/tgcagGAGGC 5(3¢-UTR) 137 (110) Vaspin A 1 66 A 163 CAGCTgtaag/tccagGTCTG (29) 2 56 B 201 AGGCGgtgag/tccagTTGAA 3 133 C 678 GACCGgtaag/tccagCTGCT 4 101 D 259 GTGCGgtgag/tgtagGAGAC 5(3¢-UTR) 140 (110) Vaspin B 1 66 A 163 CAGCTgtaag/tccagGTCTG (54) 2 56 B 241–243 AGGCGgtgag/tttagTTGAG 3 133 C 667–674 GACCGgtaag/tccagCTGCT 4 101 D 239 CTGCGgtgag/tgcagGAAAA 5(3¢-UTR) 140 (110) V. berus PLA2 1 66 A 163 CAGCTgtaag/tccagGTCTG (52) 2 56 B 243 GGGCGgtgag/tccagTTGAA 3 133 C 671 GACCGgtaag/tccagCTGCT 4 101 D 261 CTGTGgtgag/tgcagGAAAC 5(3¢-UTR) 140 (110) a Clones harboring complete sequences of PLA2s are presented. Fig. 1. Alignment of some of the variant genes encoding chain B of vaspin from V. a. aspis (neurotoxic) and V. a. zinnikeri. Two vaspin chain B gene variants are shown for V. a. zinnikeri (vp0016B10VAZ and vp0016C06VAZ, DDBJ/EMBL/GenBank accession no.s AY243574 and AY243577, respectively) and V. a. aspis (vp0015F11VAN and vp0015C10VAN, DDBJ/EMBL/GenBank accesion no.s AY243575 and AY243576, respectively). The nucleotides forming the introns are shown in italics, and those constituting the exons are underlined. Stars below the sequence indicate nucleotides conserved in all sequences. Dashes correspond to deleted nucleotides. Putative transcription factors are boxed. 2700 I. Guillemin et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Several putative regulatory sequences were identified with the TRANSFAC 4.0 databases: binding sites for the transcription factors Sp1 (CCCGCCA), NF-IL6 (TGGG GAA), NF-jB (GGGGAAGTCCC) and AP-2 (CCCTG CC) were identified in PLA2 genes (Figs 1 and 2) [22]. These trans-acting factors may act as stress-response Ó FEBS 2003 Genomic analysis of phospholipases A 2 from French viper venoms (Eur. J. Biochem. 270) 2701 elements or may be responsible for tissue-specific regula- tion [23]. Conservation of the nucleotide sequence We identified variant genes for the same PLA2 in the genomes of all of the snakes. Such variants were identified for all the sequenced PLA2s. Thorough analysis of the nucleotide sequences showed that, although the organiza- tion of introns and exons and the sequences of these PLA2 variants were well conserved (with ‡ 84% identity over 1900 bp), single-nucleotide polymorphisms were present throughout the gene sequence. An example is given for some of the vaspin chain B gene variants identified in V. a. zinnikeri and the neurotoxic V. a. aspis (Fig. 1). Some of the nucleotide mutations were found in both subspecies whereas others were subspecies-specific. These genes probably resulted from complete duplication events. However, some of the nucleotide polymorphisms observed may be also accounted for by the error rate of the Taq polymerase used for DNA amplification, which is estima- ted by the manufacturers to be  10 )4 . There were slightly more nucleotide polymorphisms in introns than in exons. This was particularly true for intron C, which contained several short deletions and insertions (Fig. 1). Two micro- satellite regions of tandem CA and TCCC repeats were particularly prone to insertion and deletion (Fig. 1), as reported in previous studies on T. flavoviridis and O. okinavensis PLA2 genes [8,10]. These features revealed intragenomic hypervariability within snake PLA2 genes. Exon nucleotide polymorphisms leading to amino-acid substitutions might result in the creation of PLA2s with different functions. Indeed, PLA2 isoforms, differing by a few amino acids and in lethal potency or enzymatic activities, have been isolated from the venoms of individual Crotalinae snakes [24]. We then identified a consensus nucleotide sequence for each of the five PLA2 genes (Fig. 2). For the genes encoding chains A and B of vaspin and PLA2 from V. b. berus,a single consensus sequence was obtained, regardless of the snake species. For the ammodytin I1 and I2 genes, however, two to three consensus sequences were obtained, according to the snake species or subspecies. Only one of the consensus sequences for the ammodytin I1 and I2 genes is presented in Fig. 2. In contrast with that observed in comparisons of gene variants encoding the same PLA2 (Fig. 1), the alignment of these consensus sequences showed that nuc- leotide variations were more common in exons than in introns (Fig. 2). Exons 3 and 5 were the most divergent, whereas the signal peptide, the 5¢-UTR and 3¢-UTR and the promoter region were the most highly conserved (Fig. 2). The nucleotide substitutions mostly involved transitions rather than transversions, in contrast with that observed in conopeptide genes, thus excluding the involvement of DNA polymerase V in genomic hypervariability [25]. These observations are not consistent with the neutral evolution theory, which states that the strong conservation of exons serves to maintain the function of the mature protein [26]. The protein-coding regions of the PLA2 genes of French vipers most probably evolved in an accelerated Darwinian manner, as reported for the PLA2 genes expressed in Crotalinae and V. am. ammodytes venom [2,3,5,7,8]. Deduced amino-acid sequence analysis We aligned the amino-acid sequences deduced from the consensus nucleotide sequence of each PLA2 (Fig. 3). The peptides encoded by exons 3 and 5 were the least well conserved for all the PLA2 genes sequenced, with only 45% and 49% identity, respectively. In contrast, the peptide encoded by exon 2 was the most highly conserved, with 76% identity between all PLA2 sequences. Similar obser- vations have been reported for the V. palaestinae and Trimeresurus PLA2s [6,9]. The mature PLA2 proteins displayed a mean of 51% identity in terms of their amino- acid sequences. The ammodytin I1 and ammodytin I2 proteins were the most similar, displaying 78% amino-acid sequence identity. The B chain of vaspin was the most divergent, displaying 67% identity with the vaspin A chain, and 70% identity with the V. b. berus PLA2. The signal peptides of all the proteins were 16 amino acids long. The mature proteins consisted of 122 amino acids for ammodytin I1, the A and B chains of vaspin and V. b. berus PLA2, and 121 amino acids for ammo- dytin I2. The amino-acid sequence of the vaspin A chain was identical in all snake species and was 100% identical with that of the acidic subunit of PLA2-I from V. a. zinnikeri [17]. The deduced amino-acid sequence of the V. b. berus anticoagulant PLA2 protein was identical with that of the PLA2 purified from V. b. berus venom [20]. No difference was observed between the ammodytin I2 sequences from the neurotoxic V. a. aspis snake, V. b. berus,andV. am. ammodytes [19]. However, for V. a. aspis, one group of genes (23 of 42) contained a sequence identical with that found in V. am. ammodytes, whereas another group (19 of 42) had one amino-acid difference (Asn111Ser), corresponding to a mutation in the fifth exon (Fig. 3). The sequence of the vaspin B chain was identical in V. a. zinnikeri and the neurotoxic V. a. aspis. However, it differed by one residue from the sequence of the B chain of vipoxin from V. am. merid- ionalis [19], and by three residues from the published Fig. 2. Alignment of the consensus sequences of ammodytin I1 (AmtI1), ammodytin I2 (AmtI2), vaspin chains A and B and V. berus PLA2 genes isolated from French vipers. The ammodytin I1 consensus sequence was defined from the sequences of V. a. aspis, V. a. zinnikeri,neurotoxic V. a. aspis isoforms 1 and 2 and V. b. berus PLA2 (DDBJ/EMBL/ GenBank accession no.s AY159807, AY159810, AY159808, AY159809 and AY159811, respectively). The ammodytin I2 consensus sequence was defined from the sequences of V. a. aspis,neurotoxic V. a. aspis and V. b. berus (DDBJ/EMBL/GenBank accession no.s AY158637, AY158638 and AY158639, respectively). The vaspin chain A consensus sequence was defined from the sequences of V. a. zin- nikeri and neurotoxic V. a. aspis (DDBJ/EMBL/GenBank accession no.s AY152843 and AF548351) and that of vaspin chain B, from the sequences of V. a. zinnikeri and neurotoxic V. a. aspis (DDBJ/EMBL/ GenBank accession no.s AY158635 and AY158634). Dots indicate identity with the ammodytin I1 sequence. Asterisks indicate nucleo- tides conserved within PLA2s. Dashes correspond to deleted nucleo- tides if the ammodytin I1 sequence is taken as the reference sequence. Italics indicate DNA tandem repeats. Putative transcription factor- binding sites are boxed. 2702 I. Guillemin et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Ó FEBS 2003 Genomic analysis of phospholipases A 2 from French viper venoms (Eur. J. Biochem. 270) 2703 sequence of the basic subunit of PLA2-I from V. a. zin- nikeri [17]. Finally, we obtained three different ammody- tin I1 sequences (In, Ia and Ib) as shown in Table 2 and Fig. 3. The amino-acid sequence of ammodytin In was identical with that reported for the protein from V. am. ammodytes (DDBJ/EMBL/GenBank AC AF253048). This sequence was identified in four ammo- dytin I1 clones from the neurotoxic V. a. aspis. Ammo- dytin Ia accounted for the remaining six clones, and all the genes of V. a. aspis and V. a. zinnikeri. Its sequence was 97% identical (four amino-acid differences: Met70Leu, Gly71Ser, Gln78Glu and Phe123Leu) with that of the ammodytin I1 of V. am. ammodytes (Fig. 3). The six clones of the neurotoxic V. a. aspis harboring the ammody- tin Ia sequence were those for which a 126-bp deletion in the ammodytin I1 gene had been identified (Table 3). Ammo- dytin Ib was found only in V. b. berus, and differed from ammodytin Ia by three amino-acid residues (98% identity: Ile/Thr, in the signal peptide, and His1Asn, Asn56Lys). In fact, ammodytin In is a hybrid molecule derived from the N-terminus of ammodytin Ia and the C-terminus of ammo- dytin Ib. Hybrid PLA2s have been reported in several venomous species of pit vipers [27]. However, this hybrid could not have been produced by recombination between ammodytin Ia and Ib genes in the neurotoxic V. a. aspis snakes, because there was no gene for ammodytin Ib in the genome of this snake. These findings provide clues to the evolutionary position of the neurotoxic V. a. aspis population with respect to the other snakes studied. The genome of one of these neurotoxic snakes displayed features characteristic of V. am. ammo- dytes (monomeric ammodytins A, B and C, and ammodytin I1n with a 259 bp intron D, the Bov-B LINE retroposon) and of V. a. aspis (vaspin A and B chains and ammodytin I1a with a 133 bp intron D; Table 2). This suggests possible interbreeding between these two species, leading to a hybrid V. a. aspis with a higher level of polymorphism in venom PLA2 genes in this snake (Table 2). The identification of natural hybrids between V. a. aspis and V. am. ammodytes in Italy is consistent with the hypothesis of horizontal transfer [28]. Moreover, immunological analysis of albumin proteins also suggests that V. aspis and V. ammodytes are closely related species [29]. The unusually strong conserva- tion of introns in PLA2 genes may facilitate homologous recombination events between PLA2 genes from different species. Amino-acid substitutions: implications for PLA2 structure and/or function The amino-acid substitutions due to the variant PLA2 genes are indicated below the protein sequence alignment in Fig. 3. Frameshifts were observed in exons 4 and 5 of the ammodytin I1 gene, and in exon 3 of the ammodytin I2 gene Fig. 3. Alignment of V. a. aspis, V. a. zinnike ri, V. b. berus and neurotoxic V. a. aspis PLA2 protein sequences. Dots indicate amino acid residues identical with those of the ammodytin I2 protein. Dashes indicate gaps introduced to optimize the alignment, using Renetseder’s numbering system [39]. AmI2 (blue) corresponds to ammodytin I2, VaspB (red) corresponds to the vaspin chain B protein and VaspA (green) corresponds to vaspin chain A. AmI1n corresponds to ammodytin from the neurotoxic V. a. aspis. AmI1a corresponds to the ammodytin I1 of V. a. aspis, V. a. zinnikeri and the neurotoxic V. a. aspis. AmI1b corresponds to ammodytin I1 from V. b. berus. VB (black) corresponds to V. b. berus PLA2. The cysteine residues involved in disulfide bridges are indicated in yellow. . indicates residue Asp49. Amino acid substitutions resulting from nucleotide polymorphisms are indicated below the alignment, in the color used for the PLA2. 2704 I. Guillemin et al.(Eur. J. Biochem. 270) Ó FEBS 2003 (data not shown). There was also a single nonsense mutation leading to insertion of the TAA stop codon in exon 3 of one of the vaspin A chain genes of V. a. zinnikeri (data not shown). These four clones are inactive. The existence of pseudogenes has also been reported in the genome of O. okinavensis [10]. As previously shown by Kini & Chan [11], the highest mutation frequencies corresponded to the exposed regions of the molecule, particularly those involved in the pharmacological activities of PLA2. These mutations may affect the enzymatic activity, stability and toxicity of the PLA2s. Enzymatic activity and stability of PLA2s Key residues involved in or required for PLA2 structure and catalysis were well conserved (Fig. 3). Most of the amino acids forming the hydrophobic channel and its opening are located in the N-terminal helix and helix a3 [14,30,31]. These amino acids were conserved in the five PLA2s, with only rare substitutions observed (Tyr22His; Ala102Val; Ala103- Val; Fig. 3). The amino-acid residues that form part of the calcium-binding loop, including Tyr28, Gly30, Gly32 and Asp49, were very well conserved among PLA2s. Neverthe- less, the Tyr28His, Gly30Asp, and Asp49Asn substitutions were found in some genes (Fig. 3), and these substitutions may affect Ca 2+ binding, which is necessary for the catalytic activity of PLA2 enzymes. In group II PLA2s, 12–16 cysteine residues are involved in the formation of six to eight disulfide bridges that stabilize the structure of the molecule. Six of these residues were substituted (Fig. 3), probably decreasing the stability of the molecule by preventing the formation of disulfide bonds. Toxicity and heterocomplex formation Little is known about the toxicity of ammodytin I1. Komori et al. [32] purified three PLA2s from V. aspis venom (PLA2-I, PLA2-II and PLA2-III) and determined their biological activities. The N-terminal sequence of PLA2-III is identical with that deduced here from the nucleotide sequence of ammodytin I1. Thus, if PLA2-III corresponds to ammodytin I1, it is not lethal. Ammodytin I2 is a nontoxic PLA2 [33]. Vaspin is a neurotoxin [13,17] and the PLA2 of V. b. berus has potent anticoagulant activity [20]. Mutations leading to amino-acid substitutions were most common in exon 4. They were clustered in an exposed region defined as the b-wing and in a short segment defined as the anticoagulant region [34]. Such mutations were also found in exons 3 and 5, specifically in the first 16 residues of exon 3 and at the 3¢ extremity of exon 5 (Fig. 3). It has been suggested that the b-wing region and the region between amino acids 106 and 128 in exon 5 are involved in PLA2 toxicity [30,35,36]. If this is indeed the case, then substitu- tions occurring in these areas may affect the neurotoxicity or anticoagulant effect of the PLA2. The substitutions in the A and B chains of vaspin, in the inhibitor and PLA2 subunits, respectively, should be considered together, as these proteins associate in a het- erodimeric complex to exert their neurotoxic effects. The formation of this complex involves intermolecular inter- actions [17,37,38]. The inhibitor subunit stabilizes the unstable PLA2 subunit through hydrophobic, ionic and electrostatic interactions, and hydrogen bonds [17,37,38]. Substitution of the residues involved in these interactions would therefore be expected to impair the formation or stabilization of the complex, or both, thereby affecting the toxicity of the protein. This is probably the case for changes in the residues involved in the interaction between the two subunits: Asn1Ser, Phe3Leu in the PLA2 subunit and Gln34His in the inhibitor subunit (Fig. 3) [17,37]. Conclusion Several groups of genes encoding Vipera venom PLA2s (chains A and B of vaspin, ammodytin I1, ammodytin I2 and an anticoagulant PLA2 from V. b. berus venom) were sequenced in this study. Nonsynonymous mutations were observed in these genes, demonstrating a high level of genetic variability in Viperinae PLA2s. Some of these mutations led to amino-acid changes, most commonly in the sequences encoded by the third and fifth exons, which are involved in the biological functions of PLA2. Some genes were pseudogenes, inactivated by frameshifts or by mutations leading to the presence of a stop codon in the sequence. The level of expression of the functional genes is probably controlled by stress-responsive promoters. Analysis of venom gland PLA2 cDNAs is currently underway to determine which of these genes are expressed. The presence in the neurotoxic snake of two ammodytin I1 isoforms (In and Ia) and of several neurotoxin-encoding genes, some specific to V. am. ammodytes venom, and of the Bov-B LINE retroposon, which was isolated from the V. am. ammodytes genome but not from the V. a. aspis genome, leads us to conclude that the new population of neurotoxic V. a. aspis is of ÔhybridÕ origin. Phylogenetic and evolutionary analyses are underway to confirm this hypothesis. Acknowledgements I. G. holds a postdoctoral fellowship from the Direction des Programmes Transversaux de Recherche (PTR) of the Pasteur Institute. This work was funded by the Direction des PTR of the Pasteur Institute. We also thank Stephane Ferris and Eliana Ochoa from the genomics platform Genopole-IP for technical assistance. We are grateful to Y. Doljanski, O. Grosselet and A. Teynie ´ for capturing the snakes and for carrying out the herpetological survey. References 1. Danse, J.M., Gasparini, S. & Me ´ nez, A. (1997) Molecular biology of snake venom phospholipases A2. In Venom Phospholipase A2 Enzymes: Structure, Function and Mechanism (Kini, R.M., ed.), pp. 29–71. Wiley & Sons Ltd, Chichester, UK. 2. Kini, R.M. (1997) Phospholipase A2: a complex multifunctional protein puzzle. In Venom Phospholipase A2 Enzymes: Structure, Function and Mechanism (Kini, R.M., ed.), pp. 1–28. Wiley & Sons, Chichester, UK. 3. Yoshikawa, T., Naruse, S., Kitagawa, M., Ishiguro, H., Naga- hama, M., Yasuda, E., Semba, R., Tanaka, M., Nomura, K. & Hayakawa, T. (2001) Cellular localization of group IIA phos- pholipase A2 in rats. J. Histochem. Cytochem. 49, 777–782. 4. Kordis, D. & Gubensek, F. (1996) Ammodytoxin C gene helps to elucidate the irregular structure of Crotalinae group II phospho- lipase A2 genes. Eur. J. Biochem. 240, 83–90. Ó FEBS 2003 Genomic analysis of phospholipases A 2 from French viper venoms (Eur. J. Biochem. 270) 2705 5. Kordis, D. & Gubensek, F. (1997) Bov-B long interspersed repeated DNA (LINE) sequences are present in Vipera ammodytes phospholipase A2 genes and in genomes of Viperidae snakes. Eur. J. Biochem. 246, 772–779. 6. Kordis, D., Bdolah, A. & Gubensek, F. (1998) Positive Darwinian selection in Vipera palaestinae phospholipase A2 genes is unexpectedly limited to the third exon. Biochem. Biophys. Res. Commun. 251, 613–619. 7. John, T.R., Smith, L.A. & Kaiser, I.I. (1994) Genomic sequences encoding the acidic and basic subunits of Mojave toxin: unusually high sequence identity of non-coding regions. Gene 139, 229–234. 8. Nakashima, K., Nobuhisa, I., Deshimaru, M., Nakai, M., Ogawa, T.,Shimohigashi,Y.,Fukumaki,Y.,Hattori,M.,Sakaki,Y.& Hattori, S. (1995) Accelerated evolution in the protein-coding regions is universal in Crotalinae snake venom gland phospho- lipase A2 isozyme genes. Proc. Natl. Acad. Sci. USA 92, 5605–5609. 9. Nakashima, K., Ogawa, T., Oda, N., Hattori, M., Sakaki, Y., Kihara,H.&Ohno,M.(1993)AcceleratedevolutionofTrimer- esurus flavoviridis venom gland phospholipase A2 isozymes. Proc. Natl. Acad. Sci. USA 90, 5964–5968. 10. Nobuhisa, I., Nakashima, K., Deshimaru, M., Ogawa, T., Shi- mohigashi, Y., Fukumaki, Y., Sakaki, Y., Hattori, S., Kihara, H. & Ohno, M. (1996) Accelerated evolution of Trimeresurus okina- vensis venom gland phospholipase A2 isozyme-encoding genes. Gene 172, 267–272. 11. Kini, R.M. & Chan, Y.M. (1999) Accelerated evolution and molecular surface of venom phospholipase A2 enzymes. J. Mol. Evol. 48, 125–132. 12. De Haro, L., Robbe-Vincent, A., Saliou, B., Valli, M., Bon, C. & Choumet, V. (2002) Unusual neurotoxic envenomations by Vipera aspis aspis snakes in France. Hum. Exp. Toxicol. 21, 137–145. 13. Jan, V., Maroun, R.C., Robbe-Vincent, A., De Haro, L. & Choumet, V. (2002) Toxicity evolution of Vipera aspis aspis venom: identification and molecular modeling of a novel phos- pholipase A2 heterodimer neurotoxin. FEBS Lett. 527, 263–268. 14. Winnepenninckx, B., Backeljau, T. & De Wachter, R. (1993) Extraction of high molecular weight DNA from molluscs. Trends Genet. 9, 407. 15. Ewing, B., Hillier, L., Wendl, M.C. & Green, P. (1998) Base- calling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res. 8, 175–185. 16. Perbandt, M., Wilson, J.C., Eschenburg, S., Mancheva, I., Alek- siev, B., Genov, N., Willingmann, P., Weber, W., Singh, T.P. & Betzel, C. (1997) Crystal structure of vipoxin at 2.0 A ˚ : an example of regulation of a toxic function generated by molecular evolution. FEBS Lett. 412, 573–577. 17. Komori, Y., Masuda, K., Nikai, T. & Sugihara, H. (1996) Com- plete primary structure of the subunits of heterodimeric phos- pholipase A2 from Vipera a. zinnikeri venom. Arch. Biochem. Biophys. 327, 303–307. 18. Wang,Y M.,Lu,P.J.,Ho,C.L.&Tsai,I.H.(1992)Character- ization and molecular cloning of neurotoxic phospholipase A2 from Taiwan viper (Vipera russelli formosensis). Eur. J. Biochem. 209, 635–641. 19. Krizaj, I., Liang, N S., Pungercar, J., Strukelj, B., Ritonja, A. & Gubensek, F. (1992) Amino acid and cDNA sequences of a neu- tral phospholipase A2 from the long-nosed viper (Vipera ammo- dytes ammodytes)venom.Eur. J. Biochem. 204, 1057–1062. 20. Krizaj, I., Siigur, J., Samel, M., Cotic, V. & Gubensek, F. (1993) Isolation, partial characterization and complete amino acid sequence of the toxic phospholipase A2 from the venom of the common viper, Vipera berus berus. Biochim. Biophys. Acta 1157, 81–85. 21. Daltry, J.C., Wu ¨ ster,W.&Thorpe,R.S.(1996)Dietandsnake venom evolution. Nature (London) 379, 537–540. 22. Nakashima, K., Nobuhisa, I., Deshimaru, M., Ogawa, T., Shimohigashi, Y., Fukumaki, Y., Hattori, M., Sakaki, Y., Hattori, S. & Ohno, M. (1995) Structures of genes encoding TATA box-binding proteins from Trimeresurus gramineus and T. flavoviridis snakes. Gene 152, 209–213. 23. Lavrovsky, Y., Schwartzman, M.L., Levere, R.D., Kappas, A. & Abraham, N.G. (1994) Identification of binding sites for tran- scription factors NF-kappa B and AP-2 in the promoter region of the human heme oxygenase 1 gene. Proc. Natl Acad. Sci. USA 91, 5987–5991. 24. Faure, G. & Bon, C. (1988) Crotoxin, a phospholipase A2 neurotoxin from the South American rattlesnake Crotalus durissus terrificus: purification of several isoforms and comparison of their molecular structure and of their biological activities. Biochemistry 27, 730–738. 25. Conticello, S.G., Gilad, Y., Avidan, N., Ben-Asher, E., Levy, Z. & Fainzilber, M. (2001) Mechanisms for evolving hypervariability: the case of conopeptides. Mol. Biol. Evol. 18, 120–131. 26. Kimura, M. (1983) The Neutral Theory of Molecular Evolution. Cambridge University Press, Cambridge, UK. 27. 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. 28. Naulleau, G. (1997) La Vipe ` re Aspis, pp. 42–43. Eveil Nature, Angoule ` me, France. 29. Herrmann, H.W. & Joger, U. (1997) Evolution of viperine snakes. In Venomous Snakes: Ecology, Evolution and Snakebite (Thorpe, R.S., Wu ¨ ster, W. & Malhotra, A., eds), pp. 43–61. The Zoological Society of London and Clarendon Press, Oxford, UK. 30. Dupureur, C.M., Yu, B.Z., Jain, M.K., Noel, J.P., Deng, T., Li, Y., Byeon, I.J. & Tsai, M.D. (1992) Phospholipase A2 engineering. Structural and functional roles of highly conserved active site re- sidues tyrosine-52 and tyrosine-73. Biochemistry 31, 6402–6413. 31. 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. 32. Komori, Y., Nikai, T. & Sugihara, H. (1990) Comparative study of three phospholipase A2s from the venom of Vipera aspis. Comp. Biochem. Physiol. 97, 507–514. 33. Sribar, J., Copic, A., Paris, A., Sherman, N.E., Gubensek, F., Fox, J. & Krizaj, I. (2001) A high affinity acceptor for phospholipase A2 with neurotoxic activity is a calmodulin. Eur. J. Biochem. 276, 13493–12496. 34. Kini, R.M. & Evans, H.J. (1989) A model to explain the phar- macological effects of snake venom phospholipases A2. Toxicon 27, 613–635. 35. Curin-Serbec,V.,Novak,D.,Babnik,J.,Turk,D.&Gubensek,F. (1991) Immunological studies of the toxic site in ammodytoxin A. FEBS Lett. 280, 175–178. 36. Ivanovski, G., Copic, A., Krizaj, I., Gubensek, F. & Pungercar, J. (2000) The amino acid region 115–119 of ammodytoxins plays an important role in neurotoxicity. Biochem. Biophys. Res. Commun. 276, 1229–1234. 37. Banumathi, S., Rajashankar, K.R., Notzel, C., Aleksiev, B., Singh, T.P., Genov, N. & Betzel, C. (2001) Structure of the neu- rotoxic complex vipoxin at 1.4 A ˚ resolution. Acta Crystallogr. D 57, 1552–1559. 38. Betzel, C., Genov, N., Rajashankar, K.R. & Singh, T.P. (1999) Modulation of phospholipase A2 activity generated by molecular evolution. Cell. Mol. Life Sci. 56, 384–397. 39. Renetseder, R., Dijkstra, B.W., Huizinga, K., Kalk, K.H., Drenth, J., Krizaj, I. & Gubensek, F. (1988) Crystal structure of bovine pancreatic phospholipase A2 covalently inhibited by p-bromo- phenacyl-bromide. J. Mol. Biol. 200, 181–188. 2706 I. Guillemin et al.(Eur. J. Biochem. 270) Ó FEBS 2003 . Sequences and structural organization of phospholipase A 2 genes from Vipera aspis aspis , V. aspis zinnikeri and Vipera berus berus venom Identification. encoding phospholipases A 2 (PLA2s) from the venoms of Vipera aspis aspis (V. a. aspis) , Vipera aspis zinnikeri (V. a. zinnikeri) , Vipera berus berus (V. b. berus) and