THE J O U R N A L OF BIOLOGICAL CHEMISTRY Vol. 264, No. 32, Issue of November 15, pp. 19259-19265,1989 Printed in U.S.A.

Precursors of Androctonus australis Scorpion Neurotoxins STRUCTURES OF PRECURSORS, PROCESSING OUTCOMES, AND EXPRESSION OF A FUNCTIONAL RECOMBINANT TOXIN 11*

(Received for publication, May 11, 1989)

Pierre E. BougisS, Herve Rochate, and Leonard A. Smith From the Toxinology Department, Pathology Diuision, United States Army Medical Research Institute of Infectious Diseases, Fort Detrick, Frederick, Maryland 21701-5011 and the SUniuersite d’Aix-Marseille II, Faculte de Medecine Secteur-Nord, Biochimie, Centre National de la Recherche Scientifique URAlI79-Institut National de la Sante et de la Recherche Medicale U172, Bd. P. Dramard, 13326 Marseille Cddex 15, France

From a cDNA library made from telsons of scorpions of the species Androctonus australis, full-length cDNAs of about 370 nucleotides encoding precursors of toxins active on mammals or on insects have been isolated using oligonucleotide probes. Sequence analy- sis of the cDNAs revealed that the precursors contain signal peptides of about 20 amino acid residues. In addition, precursors of toxins active on mammals have extensions at their COOH-terminal ends: Arg or Gly- Arg. The processing steps required to generate toxins from their respective precursors are thus not identical for all of them. Southern blot analysis performed at the genomic level with a cDNA encoding toxin I1 sug- gested a single copy gene having a minimum size of 2800 base pairs. Finally, in an attempt to successfully express an animal toxin, monkey kidney COS-7 cells transfected with a plasmid harboring a cDNA encoding toxin I1 transiently expressed a recombinant toxin hav- ing the immunological and biological properties of toxin 11.

The Buthid scorpion venoms are known to contain single- chain basic proteins of 60-70 amino acid residues (1,2) tightly reticulated by four disulfide bridges (3). The acute toxicity of these proteins is due to their high binding affinity for voltage- sensitive sodium channels, thus imparing the initial rapid depolarization phase of the action potential in nerve and muscle. The pharmacology of these neurotoxins has been reviewed elsewhere (4). Interestingly, they may be species- specific, and a venom may contain toxins selectively active on mammals and others lethal to either insects (5) or crusta- ceans (6).

* The research described was in compliance with National Insti- tutes of Health guidelines involving recombinant DNA molecules. The investigators adhered to the “Guide for the Care and Use of Laboratory Animals,” as promulgated by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animals Resources, National Research Council. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequencefs) reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession numberfs) 505102 and J05103.

$ Recipient of a National Research Council/United States Army Medical Research Institute of Infectious Diseases Research Associ- ateship from the National Academy of Sciences during this work; permanent address for correspondence: Universiti. d’Aix-Marseille 11, Facult6 de Mkdecine Secteur-Nord, Biochimie, CNRS URA1179- INSERM U172, Bd. P. Dramard, 13326 Marseille Ci.dex 15, France.

During the last 30 years, many studies have focused on the purification and characterization of the structural and im- munochemical relationships of numerous scorpion neurotox- ins. Five structural groups of toxins active on mammals have been defined according to both sequence and antigenic ho- mologies (7, 8). From their binding properties and pharma- cology, they have also been classified into a- or P-type toxins (8, 9). Chemical modification has allowed the determination of critical amino acid residues involved in the toxic activity (10). Five main regions appear responsible for antigenicity and have been mapped by using homologous toxins ( l l ) , chemically modified toxins (121, and/or synthetic peptides (reviewed in Ref. 13). Thus, while scorpion neurotoxins are polymorphic proteins, conserved domains exist in which pu- tative toxic sites have been proposed to be located (14). The first three-dimensional structure of a “variant” protein with mild toxicity was solved by x-ray crystallography analysis a few years ago (15), and very recently that of the most potent among the a-toxins has been determined at high resolution (16).

For improved serotherapy of Buthid scorpion stings as well as the development of a possible vaccine, we wished to pursue a study of structure-activity relationships in scorpion neuro- toxins by carrying out specific amino acid substitutions in toxin molecules using recombinant DNA techniques. Consid- ering the lack of knowledge of the molecular biology of scor- pion neurotoxins, as a first step we cloned and sequenced cDNAs encoding precursors of Androctonus australis neuro- toxins. The venom of this scorpion species living in North Africa has been extensively studied. Toxins comprising less than 2% of venom dry weight were purified from animals collected in Algeria (area of Chellala) and sequenced AaH I (17), AaH I’ (17), AaH I” (17), AaH I1 (18), AaH I11 (19), toxins active on mammals, and AaH IT (20), a toxin active on insects. The venom of animals collected in Tunisia (area of Tozeur) differed by the absence of AaH I” and the presence of a second toxin active on insects: AaH IT2; AaH IT1 corresponded to AaH IT on the basis of amino acid composi- tion (21). The processing steps required to generate toxins from precursors were analyzed by comparing their amino acid sequences. Southern blot analysis was performed at the ge- nomic level. Furthermore, the present paper describes exper- iments in which COS-7 cells were transfected with a construct containing an A. australis toxin I1 cDNA to investigate the ability of COS-7 cells to express a biologically active recom- binant toxin.

EXPERIMENTAL PROCEDURES

Materials-The following reagents were obtained from the indi- cated sources: 35S and 32P-labeled nucleotides, Du Pont-New England

19259

19260 Scorpion Neurotoxin Precursors Nuclear; 1261-iodide, carrier-free, Amersham Corp.; restriction en- zymes, Moloney murine leukemia virus reverse transcriptase, ter- minal dideoxynucleotidyltransferase, Bethesda Research Laborato- ries; commonly used molecular biology chemicals, Sigma; oligo(dT)- tailed pcDV-I-derived plasmid and oligo(dG)-tailed pL1 Hind111 linker for cDNA cloning and expression, ~ s c ~ r i c h ~ coli, K12 MC1061 strain, Pharmacia LKB Biotechnology Inc.; Sequenase, United States Biochemical Corp.; X-AR and X-RP autoradiography films, Kodak.

Purification and Characterization of Telson mRNA-Scorpions of the species A. australis were collected in the area of Beni-Khedache (Tunisia) for the Pasteur Institute of Tunis. Animals were trans- ported alive to the United States Army Medical Research Institute of Infectious Diseases where they were killed 2 days after manual extraction of their venom to allow the toxin-producing cells of the venom glands to enter the secretory phase. The origin of the mRNA was the telson, the last segment of the tail containing the two venom glands. Total RNA was extracted from homogenized telsons using the guanidinium hot phenol method (22). Vanadyl-ribonucleoside complex was used as an effective ribonuclease inhibitor. Poly(A)* mRNA was further selected using oligo(dT)-cellulose chromatogra- phy (23). The intactness of mRNA and its ability to serve as a template for full-length cDNA transcription was measured using 1 pg of mRNA, 0.4 pg of oligo(pdT)lz-ls as primer, and 200 units of Moloney murine leukemia virus reverse transcriptase. 32P-Labeled primary transcripts were submitted to electrophoresis on a 1.5% alkaline agarose gel, blotted onto Whatman DE-81 paper, and exposed to X- AR film with intensifying screens for 48 h at -70 "C.

Construction of a cDNA Library-The Okayama-Berg cloning and expression system (24) was used to generate the library. From 9 pg of telson mRNA and 2 pg of oligo(dT)-tailed pcDV-1 plasmid primer,

about 8 X IO5 E. coli K12 MC1061 transformants were generated, which constituted the cDNA library used in this work.

Screening of the cDNA Librwy-For each screening experiment, 400,000 clones from the cDNA library were analyzed by oligonucleo- tide probes having sequence homology to regions coding for toxins. Probes were synthesized on a Biosearch 8700 DNA synthesizer by @- cyanoethyi phosphoramidite chemistry and were 32P end-labeled by T4 polynucleotide kinase. High and low density screenings of bacte- rial colonies for recombinant plasmids were performed on nitrocel- lulose filters (25). Filters were prehybridized for 2 h at 37 "C in 6 X SSC, pH 7.0, containing 1 X Denhardt's, 0.5% SDS, 100Mg/ml sheared and denatured salmon sperm DNA, and 0.05% sodium pyrophosphate. Filters were then h y b r i ~ d in 6 X SSC, pH 7.0, containing 1 X Denhardt's, 20 pg/ml yeast tRNA, 0.05% sodium pyrophosphate, and the 32P end-labeled oligonucleotide probe for 16 h at 37 "C. Successive washes were performed in 6 X SSC, pH 7.0, containing 0.05% sodium pyrophosphate at 37 "C for 1 h and once at 47 "C for 5 min before autoradiography using X-AR film with intensifying screens for 36 h at -70 "C.

DNA Sequence Analysis-Inserts within BamHf and PstI restric- tion sites adjacent to the poly(A/T) and poly(G/C) tracts of the pcD vector were subcloned into both M13mp18 and M13mp19. Nucleotide sequence was determined by the dideoxy sequencing method (26). Sequenase, [35S]deoxyadenosine 5'-(a-thio)triphosphate, and the 17- base universal M13 primer were used. Because of difficulties in sequencing through the poly(A/T) tract (except in the case of pcD- 401) and because of limited enzyme restriction sites within cDNA sequences, the following primers were synthesized and used to prime MI3 clones: clones pcD-633, -634, -635, -639 subcloned into M13mp18, 5'GCGACGGTTTATGTAAG3'; clones pcD-644, -645, -648 subcloned into M13mpl8, S'CAAAGGATATTGCTGCT3';

TABLE I S y n t ~ t i c o l i g o n ~ l e o t ~ e s used for h y b r i ~ i z a t ~ o ~ probes

Shown are portions of the amino acid sequences of AaH I (17), AaH I' (l?), AaH I11 (19), AaH I1 (181, and AaH IT (20) including their reverse translation in all possible nucleotide sequences. I is inosine.

AaH I 8 19 T y r P r o A s n A s n C y s V a l T y r H i s C y s V a l P r o P r o

5' TAC C C I AAC AAC TGT GTI TAC CAC TGT G T I C C I CC 3' T T T T T

21 30 39 AaH I1 T y r C y s A s n G l u G l u C y s T h r Lys L e u L y s G l y G l u Ser G l y T y r C y s G l n T r p A l a

5' TAC TGC AAC GAA GAA TGC AC 3' T T T G G T

24 30 35 AaH I T A s n G l n C y s Thr L y s V a l His T y r A l a A s p L y s Gly

5' GGA TAC TGC CAA TGG GC 3 8 C T T G G T

5 ' AAC CAA TGC ACI AAA GTI CAC TAT GCI GAC AAA GG 3' T G T T T

Scorpion ~ e u r o t o x i ~ Precursors 19261

FIG. 1. Nucleotide sequence of cDNAs encoding toxins active on mammals and predicted amino acid sequence of toxin precursors. The nucleotide sequences ~ g ~ n n i n g with the 5' end of the cDNA inserts are repre- sented in the 5' to 3' direction and num- bered on top. Sequences are aligned for maximum homology with that of clone pcD-633, and differing nucleotides in each sequence are indicated by symbols placed on top of the nucleotide sequence. The predicted protein sequences are given below the n u c l ~ t i d e sequences and are numbered starting at the NH2- terminal amino acid residue of the toxin. Signal peptide sequences are underlined. A potential polyadenylation signal of AATAAA is ~ ~ a ~ ~ c ~ z e d ,

AACAATCTAT 10

AACAATTCAT ..

+tt ACTCAT

AACAATTCAT * *

90 100 110 1 2 0 130 1 4 0 150 160 CTTCTCCTCA TGATAGGTGT GGAGAGTAAA CGTGACGGTT ATATTGTCTA TCCCAATAAC TGTGTATACC ATTGTG"" L L L N I G V E S K R D G Y I V Y P N N C V Y H C V -

CTTCTCCTCA TGATAGGTGT GGAGAGTAAA CGTGATGGTT ATATTGTCTA TCCCAATAAC TGTGTGTACC ATTGTA---- L L L M I G V E S K R D G Y I V Y P N N C V Y H C I -

CTTCTCCTCA TGACAGGTGT GGAGAGTGTA CGTGATGGTT ATATTGTCGA TTCCAAGAAC TGTGTATACC ATTGTG"" L L L M T G V E S V R D G Y I V D S K N C V Y H C Y - CTTCTCTTCG TGACAGGTGT AGAGAGTGTA AAAGPCGGTT ATATTGTCGA CGATGTAAAC TGCACATACT TTTGTGGTAG

* I e.. . **ttt*. I * , t f ".*. L L F V T G V F . S V K D G Y I V D D V N C T Y F C G R

t + + t + + t

* * . 41 t10

1 70 180 190 200 210 220 230 240 "TTCCACCA TGCGACGGTT TATGTAAGAA AAACGGTGGT TCGAGTGGCT CTTGCTCTTT TTTAGTTCCA TCCGGACTTG - P P C D G L C K K N G G S S G S C S F L V P S G L A

"TTCCACCA TGCGACGGTT TATGTAAGAA AAACGGTGGT TCAAGTGGCT CTTGCTCTTT TTTAGTTCCA TCCGGACTTG - P P C D G L C K K N G G S S G S C S F L V P S G L A

+ +tt + t4

"TTCCACCA TGCGACGGTT TATGTAAGAA AAACGGTGCT AAAAGTGGCT CTTGCGGTTT TTTAATTCCA TCCGGACTTG - P P C D G L C K K N G A K S G S C G F L I P S G L A

AAATGCATAC TGCAACGAGG AATGTACCAA G T T M G G T GAGAGTGGTT ATTGCCAATG GGCAAGTCCA TATGGAAACG e * . t *,* * *** * tt *.***e* ** f *.* " * * * *. I. I f .

N A Y C N E E C T K L K G E S G Y C Q W A S P Y G N A t20 +30 t40

250 2 60 270 280 290 300 310 320 CCTGCTGGTG CAAAGACTTG CCCGATAACG TACCGATTAA AGATACATCA CGGAAATGCA CCCGCTGATA AACCTGTAGA C W C X D L P D N V P I K D T S R K C T R E n d

CCTGCTGGTG CAAAGACTTG CCCGATAACG TACCGATTAA AGATACATCA CGGAAATGCA CCCGCTGATA AACCTGTAGA C W C K D L P D N V P Z K D T S R K C T R E n d

t + + + t t i 1 t+t t

CCTGCTGGTG CGTAGCCTTG CCCGATAACG TACCGATTAA AGATCCATCA TATAAATGCC ATAGCCGATA AACCTGTAAR C W C V A L P D N V P I K D P S Y K C H S R E n d

1 f t t *

CCTGTTATTG CTATAAATTG CCCGATCATG TACGTACTAA AG---GACCA GGAAGATGCC ATGGCCGATA A"------- * * .* I **.* * I . * t t * L *.+*.. t.t

C Y C Y K L P D H V R T K G - P G R C H G R E n d +so t60

330 340 350 360 370 380 GTAAAATCAG AARGAATGTA T C C T W T AACTGGTAAA TAAACATAAG TAA-----TA AAAAAA?lA7> pcD-633

(AaH I1

GTAARRTCAG AAAGAATGTA GCCTAAAAAT AACTGGTAAA TAAACATAAG TAA-----TA AAAAAAAA>> pcD-639 (AaH 1')

i f + t GCAARR-CAC AAAGAATGTA TCCTGAAAAT AACTGGTAAA TAAACATAAG TM-----TR AAAAAAAA>> pcD-634

(AaH 111) ** ti * ** .. * * * "ATTATAAG ATGGAATGTA TCCTAAGTAT CAATGTTAAA TRAATATAAT CAAAA?lATTA AAAAAAAA7, pcD-402

f . *tit* clones pcD-402, -403 subcloned into M13mp18, 5'AGTTGAAAG- GTGAGAGTB'; clone pcD-402 subcloned into M13rnp19,S'TACG- TACATGATCGGGGC3'; clones pcD-633, -634, -639 subcloned into M13mp19, S'TCGGTACGTTATCGGGC3'; clones pcD-644, -645, -648 subcloned into M13mp19, S'GTCATT"AGACCGAAGC3'.

Sequence Analysis-Programs from Intelligenetics including SEQ, PEP, QUEST, and IFIND, were used on a VAX from Digital Equip- ment.

DNA ~ra~~~ect~on-Transfection of cesium chloride-purified plas- mid was performed by the DEAE-dextran method using chloroquinine as described by Luthman and Magnusson (27). COS-7 (SV40-trans- formed African green monkey kidney) cells obtained from the Amer- ican Type Culture Collection (CRL 1651, passage number: 9) were maintained at 37 "C under 5% CO, in Dulhecco's modified Eagle's medium supplemented with 10% fetal calf serum. Cells were plated onto 75-cmZ plates and were subconfluent at the time of transfection, Post-transfection, cells were fed each day with 5 ml of Dulbecco's modified Eagle's medium Supplemented with 10% fetal calf serum.

I ~ m ~ ~ p u r i f ~ ~ t ~ o n and Characterization of the Recombimnt ~O~~n-CNBr-activated Sepharose 4B from Pharmacia LKB Bio- technology Inc. was used to prepare an anti-AaH II immunoabsorbent column of 1-mi capacity with 10 mg of antibodies purified from rabbit

(AaH 11)

AaH 11-immune serum (28) using an ImmunoPure column from Pierce. After elution with 0.1 N formic acid containing 0.15 M NaC1, the immunoabsorbed recombinant toxin was desalted on Sephadex G-10 in the presence of 0.1 N acetic acid containing 0.1% bovine serum albumin to minimize loss and then lyophilized. AaH 11 was purified as previously described (2) and was iodinated with Iz5I by the lactoperoxidase method and recovered by immunoprecipitation (29). AaH 11-specific immunoassay was essentially as described elsewhere (301, except that 40 X M of AaH 11-specific antibodies and 6 X lo"* M of '"I-AaH I1 were used to increase the sensitivity of the assay. 'T-AaH 11-receptor assay was performed on rat brain synap- tosomes, as previously described (31).

In Vivo Assay-Toxicity on male C57/Bt6 mice from Evic-Ceba, 33 Blanquefort, France, weighing 20 rt 3 g was assayed by intracere- broventricular injection as described by Haley and McCormick (32).

Sou~hern Blot Analysis-High molecular weight genomic DNA was prepared from A. australis scorpion muscle tissue. After digestion with appropriate restriction endonucleases of 10 yg of DNA, samples were submitted to electrophoresis in 0.8% agarose gel, blotted onto nitrocellulose membrane, and probed with the BamHIIPstI insert of clone pcD-402 encoding AaH 11. The probe was 32P-laheled using the nick translation kit from Boehringer Mannheim. The filter was

19262 Scorpion Neurotoxin Precursors

10 20 30 40 50 60 70 80 CTCAAAAATA AAGCCAACTA CTGAAGTATC TCGATAAAAA GAAACGAAAA TGAAATTTCT CCTATTGTTT CTCGTAGTCC

M K F L L L F L V V L

AAGCCAACTA CTGAAGTATC TCGATAAAAA GAAACGAAAA TGAAATTTCT CCTATTGTTT CTCGTAGTCC M K F L L L F L V V L

AAAAAATA AAG-CAACTA CTGAAGTATC TCGATAAAAA GAAACGAAAA TGAAATTTCT CCTATTGTTT CTCGTAGTCC M K F L L L F L V V L -18 -10

FIG. 2. Nucleotide sequence of cDNAs encoding toxins active on in- sects and predicted amino acid se- quence of toxin precursors. The nu- cleotide sequences beginning with the 5’ end of the cDNA inserts are represented in the 5’ to 3‘ direction and numbered on top. Sequences are aligned for maxi- mum homology with that of clone pcD- 644, and differing nucleotides in each sequence are indicated by symbols placed on top of the nucleotide sequence. The predicted protein sequences are given be- low the nucleotide sequences and are numbered starting at the NH2-terminal amino acid residue of the toxin. Signal peptide sequences are underlined. A po- tential polyadenylation signal of AA- TAAA is italicized.

90 100 110 120 130 140 150 160 TTCCAATAAT GGGGGTGCTT GGCAAGAAGA ATGGATATGC CGTCGATAGT AGTGGTAAAG CTCCTGAATG TCTTTTGAGC

P I M G V L G K K N G Y A V D S S G K A P E C L L S

TTCCAATAAT GGGGGTGTTT GGCAAGAAGA ATGGATATGC CGTCGATAGT AGTGGTAAAG CTCCTGAATG TCTTTTGAGC -GKKN G Y A V D S S G K A P E C L L S

+

TTCCAATAAT GGGGGTGCTT GGCAAGAAGA ATGGATATGC CGTCGATAGT AGTGGTAAAG CTCCTGAATG TCTTTTGAGC P I M G V L G K K N G Y A V D S S G K A P E C L L S

+1 tl0

170 180 190 200 210 220 230 240 AATTACTGTA ACAACGAATG CACAAAAGTA CATTATGCTG ACAAAGGATA TTGCTGCTTA CTTTCATGTT ATTGCTTCGG N Y C N N E C T K V H Y A D K G Y C C L L S C Y C F G

AATTACTGTA ACAACGAATG CACAAAAGTA CATTATGCTG ACAAAGGATA TTGCTGCTTA CTTTCATGTT ATTGCTTCGG N Y C N N E C T K V H Y A D K G Y C C L L S C Y C F G

AATTACTGTT ACAACGAATG CACAAAAGTA CATTATGCTG ACAAAGGATA TTGCTGCTTA CTTTCATGTT ATTGCTTCGG N Y C Y N E C T K V H Y A D K G Y C C L L S C Y C F G t20 t30 t40

250 260 270 280 2 90 300 310 320 TCTAAATGAC GATAAAAAAG TTTTGGAGAT TTCGGACACA AGGAAAAGTT ATTGTGACAC CACAATAATT AATTAATTGT L N D D K K V L E I S D T R K S Y C D T T I 1 N E n d

TCTAAATGAC GATAAAAAAG TTTTGGAGAT TTCGGACACA AGGAAAAGTT ATTGTGACAC CACAATAATT AATTAATTGT L N D D K K V L E I S D T R K S Y C D T T I 1 N E n d

TCTAAATGAC GATAAAAAAG TTTTGGAGAT TTCGGACACA AGGAAAAGTT ATTGTGACAC CCCAATAATT AATTAATTGT L N D D K K V L E I S D T R K S Y C D T P I 1 N E n d

t5O + 60 t70

330 340 350 360 AATAATTATG AAGTATTGAA TTGATCTAAA ATAAAATGCA CATATATTAA T->>

370 pcD-644 (AaH IT1)

AATAATTATG AAGTATTGAA TTGATCTAlVl ATAAAATGCA CATATATTAA T U > > pcD-648 (AaH IT1)

AATAATTATG AAGAATTGAA TTGATCTAAA ATAAAATGCA CATATATTAA TAAAAAAA>> DCD-645

washed extensively in 0.1 X SSC containing 0.1% sodium dodecyl sulfate.

RESULTS

Isolation and Sequencing of cDNAs Encoding Toxins- Starting with 30 fresh telsons giving 2 g of tissue, 42 pg of poly(A)+ mRNA were obtained. To both characterize and assay the mRNAs, primary transcripts were synthesized by oligo(pdT)-primed reverse transcriptase. After electrophore- sis, two classes of transcripts were chiefly observed corre- sponding in size to about 360 and 1100 bases, the former being in larger amount. Although the size of mRNAs encoding toxins of about 65 amino acid residues should be close to 360 bases, we did not select telson mRNAs by size to construct a large cDNA library using the procedure of Okayama and Berg (24).

The first toxin to be cloned successfully was AaH 11. Two other screening experiments were performed to clone AaH I, AaH 1’, AaH 111, and two iso-AaH IT. For each experiment, 400,000 clones were plated and hybridized to oligonucleotide probe mixtures representing all possible complementary nu- cleotide sequences corresponding to selected amino acid se- quence regions of AaH 11, AaH I, and AaH IT (Table I). Probes for AaH I1 were designed to maximize the amino acid sequence differences between AaH I and AaH 11, belonging to the first and second structural groups, respectively. By contrast, the probe for AaH I was designed to maximize sequence homologies between all toxins belonging to the first

lAaH IT21

structural group (AaH I, AaH 1’, and AaH 111). Using dupli- cate filters to screen the cDNA library, we obtained about 100 positive clones each time. During the cloning for AaH I, clones showing low intensity hybridization were pooled separately from those of high intensity. Clones encoding AaH I’ and AaH I11 were also obtained. Clones chosen at random were colony-purified and analyzed for insert content. The size of inserts within the BarnHI and PstI restriction sites suggested that the percentage of full-length cDNAs was high. The nucleotide sequences of inserts of a total of 11 pcD clones were subsequently determined in full (see “Experimental Pro- cedures”).

Fig. 1 shows the nucleotide sequences of representative cDNAs encoding toxins active on mammals. cDNA sequences were compared and aligned for maximum homology with that of clone pcD-633. Clones pcD-633 (AaH I), pcD-639 (AaH I’), and pcD-402 (AaH 11) contained complete cDNAs beginning with the same sequence: AACAA. In the case of pcD-634 (AaH 111) the first four bases of the insert were most probably missing. Clones pcD-635 and pcD-636 were identical to pcD- 633. Clone pcD-403 had an insert 5’ truncated by 24 bases and differed from that of pcD-402 by two substitutions: C/T at position 245 corresponding in the open reading frame to the third base of Cys codons; T/C at position 371 in the 3’- noncoding region. The insert of pcD-401 was also 5‘ truncated starting at position 145 and was identical to that of pcD-403.

Fig. 2 shows the nucleotide sequences of cDNAs encoding

Scorpion Neurotoxin Precursors 19263

I Toxins active on mammals I pcD-401. -402, -403 ( AaH II )

I \ 1 9 a a A

'GR'

pcD-633. -635. -636 ( AaH I ) pcD-639 ( AaH II ) pcD-634 ( AaH 111 )

i 9 a a A R

A

NH2 COOH

1 Toxins active on insects I pcD-644, -648 ( AaH IT1 ) pcD-645 ( AaH IT2 )

18a.a A

NH2 C W H

I I I I I I I I I 0 100 200 300 bp 400

FIG. 3. Schematic representation of the structure of toxin precursors deduced from the cDNAs and processing outcomes. From top to bottom are shown, respectively, the sequences of cDNA, toxin precursor, and toxin. The poly(A)+ is not represented. The open reading frame is boxed, and the peptide signal domain is dashed. The COOH-terminal extra domain is in black. A, represents a site of proteolytic cleavage.

toxins active on insects. Inserts of pcD-644 (AaH IT1) and pcD-648 (AaH IT1) differed once at position 98 by a substi- tution: C/T corresponding in the open reading frame to the first base of Leu and Phe codons. The third clone, pcD-645 (AaH ITZ), had an insert differing from that of pcD-644 by four substitutions: two in the 5"noncoding region; T/A at position 170 corresponding to the first base of Asn and Tyr codons; A/C at position 302 corresponding to the first base of Thr and Pro codons.

All cDNA sequences displayed one major open reading frame of about 310 nucleotides. At the 5' end, multiple poten- tial initiation codons were observed. In each case, the 5' proximal ATG codon should have been responsible for the initiation of translation (33). At the 3' end, a putative poly- adenylation signal (AATAAA) was found 10-16 nucleotides upstream of the poly(A)+ tail (34). It appeared that the 5'- and 3"nontranslated regions were approximately of the same length.

Primary Structures of Toxin Precursors-For all cDNAs, the open reading frame demonstrated a coding capacity for toxin precursors because they were larger in size than toxins that have been previously purified and characterized from the venom (Fig. 3).

Precursors of toxins active on mammals possessed se- quences extended at both NH2 and COOH termini (Fig. 1). The proposed initiating Met was the first amino acid residue of a series of highly hydrophobic amino acid residues, sug- gesting the presence of a signal peptide of 19 residues. Ho- mology among signal peptide sequences was very high. Signal peptide cleavage occurred for each precursor at Ser. Thus, the predicted NHP-terminal residue of each toxin was identical to the one chemically determined on toxins isolated from the venom. By contrast, COOH-terminal amino acid residues predicted from the cDNA nucleotide sequences did not cor- relate with the COOH-terminal residues chemically deter- mined on toxins isolated from the venom. The toxin precur- sors ended with additional amino acid residues as compared with toxins: Gly-Arg for AaH I1 and Arg for AaH I, AaH 1', and AaH I11 (Fig. 3). One discrepancy was observed between the sequence encoded by clone pcD-634 and the AaH I11 sequence previously determined by peptide sequencing (19): Asp at position 8 instead of Asn.

Three different precursors of toxins active on insects were found (Fig. 2). For each of them, the signal peptide was 18 amino acid residues long ending with Gly. These precursors had no additional amino acid residue at their COOH-terminal ends (Fig. 3) when compared to the amino acid sequence of AaH IT previously determined by peptide sequencing (20). The two precursors encoded by clones pcD-644 and pcD-648 had peptide signals substituted at position -2, but each gave the same toxin containing one difference with AaH I T Glu at position 25 instead of Gln. Clone pcD-645 contained a cDNA coding for an isotoxin differing from AaH IT by two additional substitutions: Tyr at position 23 instead of Asn; Pro at position 67 instead of Thr.

Expression of a Recombinant Toxin in COS-7 Cells-The cloning vector was a transient eukaryotic expression system containing an SV40 origin of replication (24). We used the system to further prove that clone pcD-403 encoding AaH I1 directed the expression in COS-? cells of a biologically active recombinant toxin. The culture medium was collected be- tween 24 and 168 h post-transfection and was assayed for the appearance of AaH 11-related antigen by a specific radio- immunoassay (Fig. 4a). The optimum quantity of plasmid used to transfect COS-7 cells was 5 pg, and expression was maximum (0.2 pg/106 cells) at 120 h. Culture medium from COS-7 cells transfected in the absence of DNA or transfected with the 5' truncated clone pcD-401 had background values. These results demonstrated that clone pcD-403 was capable of directing the expression of AaH I1 immunoreactive material in COS-7 cells.

Immunological and Biological Properties of the Recombinant AaH ZI-To purify the recombinant AaH I1 secreted by COS- 7 cells, the 120-h cell culture medium was chromatographed on an immunoaffinity column. To test whether or not the recombinant AaH I1 was as active in uivo as AaH 11, the affinity-purified product was injected intracerebroventricu- larly in mice. Mice died with symptoms identical to those observed with mice dying from AaH 11. From the results (Table 11), the concentration of the recombinant AaH I1 was estimated to be 6.9 X lo-' M, with the assumption that its activity was the same as that determined for AaH I1 (LD50 = 0.5 ng/mouse). The immunopurified recombinant AaH I1 was further characterized on the basis of its binding properties to

Scorpion Neurotoxin Precursors -

19264

0.25

0.20

v) - - 5 0.15

9 . cn I - - v

$ O.1°

0.05

w Time (h)

-13 -12 -11 -10 -9 -8 -7

Log AaH II (M)

1 .oo m m'

0.75

0.50

0.25

0.00

-13 -12 -11 -10 -9 -8 -7

Log AaH II (M)

FIG. 4. Expression in COS-7 cell and immunological and biological characterization of a recombinant AaH 11. a, monkey kidney COS-7 cells were transfected with: 0, 2 pg; 0, 5 pg; or A, 10 pg of cesium chloride-purified plasmid from clone pcD-403 or 0, in the absence of DNA as well as with plasmid from clone pcD-401 whose insert was 5' truncated. The expression of the recombinant toxin was followed on a daily basis by AaH 11-specific immunoassays performed on cell culture medium. After being purified by immu- noaffinity chromatography, the recombinant toxin was further char- acterized by AaH 11-specific ( b ) immunoassay and (c) receptor assay. For both assays, 0 depicts the standard curve. Ten-fold serial dilu- tions of the sample of recombinant AaH 11, at an initial concentration of 6.9 X lo-' M as estimated from the in vivo experiment of Table 11, were tested in both assays, and X represents the results of such experiments.

TABLE I1 In vivo activity of recombinant AoH II

Five pl of appropriate dilutions of the recombinant AaH I1 sample were injected intracerebroventricularly to mice. According to a LDW of 0.5 ng/mouse for AaH 11, the initial concentration of the recom- binant toxin was estimated to be 6.9 X lo-' M.

Dilution factor of the recombinant AaH I1 samole

Deadrinjected

None 212

1/10 013 115 315

AaH I1 Deadlinjected

w 0.70 616 0.35 016

AaH 11-specific antibodies (Fig. 4b) and rat brain voltage- sensitive sodium channels (Fig. 4c). The same serial dilutions of the sample were tested in parallel. Results established that the recombinant AaH I1 had the same ability as AaH I1 to compete in both assays with lZ5I-AaH 11. Thus, it was most probable that the antigenic and toxic sites of the recombinant AaH I1 were closely related or identical to those of AaH 11.

Genomic Distribution of AaH 11 Sequence-To determine the size and copy number of the AaH I1 gene, a Southern blot analysis was performed using high molecular weight DNA

-

. 12216

,. 6 108 - 5 090

4072

3054

- 2036

- 1636

1018

FIG. 5. Southern blot analysis of A. australis DNA. Ten- microgram samples of muscle DNA were digested with EcoRI, BgZII, SphI, SmaI, X h I , StuI, and Nrul and electrophoresed on an 0.8% agarose gel. The digests were then blotted onto nitrocellulose filter and probed with the BamHIIPstI insert of clone pcD-402 encoding AaH 11. The size markers, given in base pairs, consist of a kilobase DNA ladder from Bethesda Research Laboratories.

from scorpion muscle and a 32P-labeled insert of clone pcD- 402. The results of this experiment are shown in Fig. 5. A single band was observed using the following restriction en- zymes: EcoRI, BgZII, SphI, SmaI, XhoI, StuI, and NruI; the cDNA probe contained no restriction sites for these enzymes. Thus, the results indicated a unique copy gene for AaH I1 having a minimum size of 2800 base pairs.

DISCUSSION

Size analysis of the pool of mRNAs obtained from the scorpion telson indicates that two main mRNA populations exist and that our initial assumption that 360 bases should be the expected size for monocistronic mRNAs encoding scor- pion neurotoxin precursors has been experimentally verified. Indeed, by using synthetic oligonucleotides as screening probes, we isolated numerous full-length cDNAs of about 370 base pairs encoding precursors of A. australis scorpion neu- rotoxins.

The comparison of the amino acid sequences of both toxin precursors (as predicted by cDNAs) and toxins (as determined from peptide sequencing) gives an insight into the structure of precursors and the processing they require in order to generate toxins (Fig. 3). There exist an almost complete homology among signal peptide sequences within toxins active on mammals as well as within toxins active on insects. By contrast, the signal peptide sequences are very different when comparing precursors of toxins from both sets. These findings reinforce the classification of scorpion neurotoxins based on amino acid compositions and sequences and CD spectra in which the toxins active on insects are placed apart from all the toxins active on mammals (35). For precursors of toxins active on insects, cleavage of the signal peptide by a signal protease is enough to generate toxins. For precursors of toxins active on mammals, two possibilities exist. For AaH 11, in

Scorpion Neurotoxin Precursors 19265

addition to the cleavage of the signal peptide, the removal of the COOH-terminal dipeptide Gly-Arg is required together with the a-amidation of the His (18). Precursors of secretory peptides synthesized as part of large and inactive precursor proteins contain sites for proteolysis and a-amidation. These are often marked by the sequence -X-Gly-B-B; X is the COOH-terminal amino acid residue in the mature peptide that is a-amidated, and B is either Lys or Arg (36). Thus, the additional dipeptide Gly-Arg is in agreement with these ob- servations although it has one basic residue instead of the usual two or none a t all as in the case of melittin from bee venom (37). The toxins active on mammals that are not amidated (AaH I, AaH 1‘, and AaH 111) require the removal of the signal peptide but also the cleavage of the COOH- terminal Arg residue most probably by a specific exopeptidase. We have no explanation for the fact that different processing steps seem to exist for scorpion neurotoxins. It is worth noting that the removal of the AaH I signal peptide leads to the amino acid sequence of AaH I”, an isotoxin of AaH I differing by an additional Arg residue at the COOH-terminal end (17). However, as stated previously, AaH I” has only been found in the venom of animals of Chellala origin. We may suppose that the cDNA encoding AaH I” compared with that encoding AaH I has been mutated at position 306, as is the case for AaH 111: the stop codon TGA being replaced by the Arg codon CGA preceding the second stop codon TAA. For both AaH I11 and AaH IT, discrepancies exist between the sequences derived from the cDNA and those previously determined by peptide sequencing. In the case of AaH 111, we have sequenced two samples of toxin purified from the venom of animals of both Chellala and Tozeur origins and found in both cases Asp a t position 8 as predicted by the cDNA sequence. The amino acid sequence of AaH I11 published previously (19) has to be corrected accordingly. In the case of AaH IT, the discrepancies observed have to be confirmed by peptide sequencing of AaH IT1 and AaH IT2. From the amino acid compositions of AaH IT1 and AaH IT2 (21) it seems likely that the sequences encoded by clones pcD-644 and pcD-645 correspond to those of AaH IT1 and AaH IT2, respectively. All the preceding observations may be of concern if we consider, as has been previously suggested, that subspecies of A. australis may exist as A. australis Hector in Algeria and A. australis garzonii in Tunisia (38).

The expression of a recombinant AaH I1 was achieved by transfection of COS-7 monkey kidney cells with an expression plasmid harboring a cDNA encoding AaH 11. As far as we know, this is the first successful expression of a recombinant animal toxin. The rationale for choosing a mammalian expres- sion system was that the biological activity of AaH 11 is, stricto sensu, dependent on the correct formation of four disulfide bridges and the proper folding of the protein (39). The recombinant AaH I1 expressed by the COS-7 cells was characterized in three different ways: by in vivo assay, im- munoassay, and receptor assay. The results obtained support the conclusion that COS-7 cells expressed a recombinant protein that behaves as AaH I1 in each of the three assays. Therefore, we assume that the recombinant toxin that mon- key kidney cells transiently expressed upon their transfection is the mature form of AaH 11. The fact that the recombinant toxin was secreted from the COS-7 cells in a biologically active form implies that the signal peptide had been removed. Many bioactive peptides have carboxyl-terminal a-amide res- idues and, in general, the presence of an a-amide is critical

for biological activity. We do not know what influence the a- amidation of the COOH-terminal His of AaH I1 has upon the biological activity of this toxin. Only a biochemical analysis of the recombinant toxin can certify its identity to AaH I1 and address the accuracy of the double processing of the toxin precursor. The amount of recombinant toxin obtained did not permit such an analysis. We are, a t present, attempting to increase the levels of expression of the recombinant toxin in other host systems to answer these questions.

Acknowledgments-Dr. Mohamed El Ayeb is recognized for provid- ing the scorpions. We thank Drs. Linda Lee and Catherine Loridan for their kind help, Dr. Marie-France Martin for assistance with lethality assays, and Pascal Mansuelle for peptide sequencing of AaH 111. We are grateful to Drs. Ariane Monneron and Catherine Ronin for their comments on the manuscript.

REFERENCES 1. Miranda, F., and Lissitzky, S. (1961) Nature 190 , 443 2. Miranda, F.! Kopeyan, C., Rochat, H., Rochat, C., and Lissitzky, S. (1970)

3. Kopeyan, C., Martinez, G., Lissitzky, S., Miranda, F., and Rochat, H.

4. Catterall, W. A. (1986) Annu. Rev. Bwchem. 56, 953-985 5. Zlotkin, E . , Rochat, H., Kopeyan, C., Miranda, F., and Lissitzky, S. (1971)

Biochamie 53,1073-1078 6. Zlotkin, E., Martinez, G., Rochat, H., and Miranda, F. (1975) Inset Biochem.

7. Rochat, H., Bernard, P., and Couraud, F. (1979) in Neurotoxins: Tools in 5 , 243-250

Neurobiology (Ceccarelli, B., and Clementi, F., eds) Vol. 3, pp. 325-334, Raven Press, New York

8. Martin, M. F., Garcia y Perez, L. G.. El Ayeb, M., Kopeyan, C., Bechis, G., Jover, E., and Rochat, H. (1987) J. Blol. Chem. 262,4452-4459

9. Couraud, F., Jover, E., Dubois, J. M., and Rochat, H. (1982) Toxicon 20 ,

10. Darbon. H.. Jover. E.. Couraud. F.. and Rochat. H. (1983) Int. J. Peot. 9-16

Eur. J. Blochem. 16,514-523

(1974) Eur. J. Biochem. 47,483-489

11. El Aveb. M.. MI

13. Granier; C., Novotny, J., Fonticilla-Camps, J. C., Fourquet, P., El Ayeb, M., and Bahraoui, E. M. (1989) Mol. Immunol. 26,503-513

14. Fontecilla-Camps, J. C., Almassy, R. J., Ealkick, S. E., Suddath, F. C., Watt, D. D., Feldman, R. J., and Bugg, C. E. (1981) Trends Biochem.

15. Fontecilla-Camps, J. C., Almassy, R. J., Suddath, F. L., Watt, D. D., and Sci. 6, 291-296

16. Fontecilla-Camps, J. C., Habersetzer-Rochat, C., and Rochat, H. (1988) Bugg, C. E. (1980) Proc. Natl. Acad. Sci. U. S. A. 77,6496-6500

17. Martin, M. F., and Rochat, H. (1984) Toxicon 22,695-703 Proc. Natl. Acad. Sci. U. S . A. 85, 7443-7447

18. Rochat, H., ,Rochat, C., Sampieri, F., Miranda, F., and Lissitzky, S. (1972)

19. Kopeyan, C., Martinez, G., and Rochat, H. (1979) Eur. J . Biochem. 9 4 , Em. J. Blochem. 28,381-388

f inwlh 20.

21.

22.

23.

Darbon, H., Zlotkin, E., Kopeyan, C., Van Rietschoten, J., and Rochat, H.

De Dianous, S., Kopeyan, C., Bahraoui, E. M., and Rochat, H. (1987) (1982) Int. J. Pept. Protein Res. 20,320-330

Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloni A Toxicon 2 5 , 731-741

Laboratory Manuul, pp. 194-195, Cold Spring Harbor Laboratory,?!old Spring Harbor, NY

Aviv, H., and Leder, P. (1972) Proc. NatL Acad. Sci. U. S. A. 69, 1408-

”_ ”_

l d l ? 24. OkaFaia, H., and Berg, P. (1983) Mol. Cell. Biol. 3, 280-289 25. Hanahan, D., and Meselson, M. (1980) Gene (Amst.) 10,63-67 26. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U. 27. Luthman, H., and Magnusson, G. (1983) Nucleic Acids Res. 11,1295-1308 28. Delori, P., Van Rietschoten, J., and Rochat, H. (1981) Toxicon 1 9 , 393-

29. Rochat, H., Tessier, M., Miranda, F., and Lissitzky, S. (1977) Anal.

30. El Ayeb, M., Delori, P., and Rochat, H. (1983) Toxicon 2 1 , 709-716 31. Jover, E., Martin-Moutot, N., Couraud, F., and Rochat, H. (1978) Biochern.

32. Haley, T. J., and McCormick, W. G. (1957) Br. J . Pharmol . 12 , 12-15 33. Kozak, M. (1984) Nucleic Acids Res. 12,857-872 34. Proudfoot, N. J., and Brownlee, G. G. (1976) Nature 2 6 3 , 211-214 35. Dufton, M. J., Drake, A. F., and Rochat, H. (1986) Biochim. Biophys. Acta

36. Mains, R. E., Eipper, B. A., Glembotski, C. C., and Dores, R. M. (1983)

37. Kreil, G. Mollay, C Kaschnitz R., Haiml, L., and Vilas, V. (1980) Ann.

S. A. 74,5463-5467

407

Biochem. 82,532-548

Bwphys. Res. Commun. 85,377-382

869,16-22

Trends Neurosci. 6, 229-235

N. Y. .&. sci. 323.338-345‘ 38. Martin, M. F., and Rochat,~H. (1986) Toxicon 24,1131-1139 39. Sabatier J. M., Darbon H., Fourquet P. Rochat, H., and Van Rietschoten,

J. (1987) Int. J . Pepi Protein Res. h0: 125-134