BIOCttlMIE. 1985, 67, 1231-1239

I I I - Isolation and characterization of the and subunits of the platelet-activating glycoprotein from the venom of Crotalus durissus cascavella.

Guy MARLAS.

Unitd des Venins et UnRd associde Institut Pasteur, I N S E R M N ° 285, Ddpartement de physio- pathologie Expdrimentale, Institut Pasteur, 25-28, rue du Dr. Roux, 75015 Paris, France.

(Refu le 22-3-7985, acceptd aprEs revision le 17-7-1985).

R~sum~ - - Nous conchdons lors d'une prdcddente publication [13] que la glycoprotdine de haut poids moldculaire, isolde du venin de Crotalus durissus cascavella [11-12] et capable d'activer tes plaquettes sanguines, est un hexamdre de type cts fls constitud de deux sous-unitds distinctes. Les donndes expdrimentales rapportdes ici ddmontrent que ces deux sous-unit~s peuvent Etre sdpardes l'une de l'autre par chromatographie dchangeuse d'ions en conditions ddnaturantes, qu'elles possEdent des masses moldculaires voisines (a--12 540 et f l=13 770) et qu'elles existent selon un rapport stoechiomdtrique de 1 / I au sein de la moldcule native. L'analyse de la partie glucidique a indiqud que les deux sous-unitds sont glycosyldes par de faibles contenus glycaniques dont les compositions sont similaires. I1 est, de plus, montrd que ces deux sous-unitds ont des compositions en acides aminds ldgdrement diffdrentes, qu'elles partagent la m#nle sdquence N-terminale au moins jusqu'au cinquiOme rdsidu et qu'elles ont des coefficients d'extinction molaires similaires, h 280 nm. La ddtermination des masses moldculaires des sous-unitds a permis d'attribuer une masse moldculaire calculde de 78 930 g~ la moldcule native. Enfin, il est ddmontrd que la convuixine du venin de Crotalus durissus terrificus [3] est la mEme substance capable d'activer les plaquettes sanguines que la glycoprotdine (PAG) prdsentement ddcrite et isolde du venin de Crotalus durissus cascavella [11-13].

Mots-cl~s : venin de serpent / glycnprot~ine / plaquettes sanguines.

Summary - - It was concluded in a previous paper [13] that the high Mr platelet-activating glycoprotein isolated earlier from the venom of Crotalus durissus cascavella [11-12] has an hexameric stntcture o f the ct3fls type involving two distinct subunits. Data reported here demonstrate that these two subunits are separable from each other by ion exchange chromato- graphy under denaturathlg conditions, have similar Mrs (Ct--12,540 et fl=13,770) and exist in a one to one ratio within the native molecule. Carbohydrate analysis indicated that they are both similarly glycosylated to a small extent. They have slightly different amino-acid compositions, a common N-terminal sequence up to the fifth residue and similar extinction coefficients at 280 nm. The native molecule has a calculated Mr o f 78,930. Additional data demonstrated that convulxin

from the venom ofCrotalus durissus terrificus [3] is the same platelet-activating agent as the presently described platelet-activating glycoprotein (PAG) from the venom of Crotalus durissus cascavella [11-13].

Key-words : snake venom / glycnprntein / blond platelets.

Abbreviations : SDS-PAGE : sodium dodeo'l sulphate-polyaco'lanlide electro-

phoresis PAG : platelet-activating glyeoprotein DTT : oL-dithiothreitol DNP-alanine : dinitrophenol-alanine.

1232 G. Marlas

Introduction

The venom o f Crotahts durissus terrificus has been reported to contain a noncoagula t ive factor capable o f activating rabbit b lood platelets hi vitro. In vivo, the venom component , when injec- ted into rabbits intravenously, induced a fall o f the arterial b lood pressure and th rombocytope- nia, a marked reduct ion o f the number o f circu- lating platelets [1,2]. Later, this putative factor purified from the venom o f Crotahts durisstts terrificus [3-6] and also present in the venom of Crotalus durissus cascavella (Prado-Franceschi , J., personal communica t ion) was shown to generate, when injected intravenously, transient tachypnea and apnea associated with a convulsive crisis and death in mice. Hence, this componen t was named convulxin by the latter authors. Our laboratory showed that convulxhl from the subspecies Cro- talus durissus cascavella activated guinea-pig [7] rabbit [8], rat [9] and human b lood platelets [10] and that it was not convulsant nor lethal for mice [I1, 12]. Convulxht, also purified from the subs- pecies Crotalus durisstts terrificus activates rabbit b lood platelets (Marlas, G., unpublished). Our results suggest that the term convulxhl has been

w r o n g l y applied to the platelet-activating glyco- protein under study. Seventeen years ago the presence o f a platelet-stimulating factor and o f convulxin in the venom o f Crotalus durissus terrifi- cus was reported. Recently, a potent platelet- activating componen t has been purified from the venom o f Crotahts durissus cascavella [11-13]. This factor was shown to be a high molecular weight glycoprotein consisting o f two types o f subunits with similar molecular weights [13]. The purpose o f the present work was to isolate and further characterize these two subunits for a better unders tanding o f their respective involvement in b lood platelet-activation.

Materials and Methods

Crotalus durissus cascavella venom was purchased from Enzyfarma Ltda, Brazil. Crotalus durissus terrificus venom was a gift of Prof. S. Ferreira, University of Sao Paulo, R. Prato, Brazil. All other reagents were of analytical grade.

Preparation of PAG and of convulxin : These pro- teins were purified from the venoms of C.d. cascavella and of C.d. terrificus as previously described [12].

Isolation o f the platelet-activathlg glycoprotein subunits

The separation of the a and 13-subunits of the platelet-activating glycoprotein (PAG) was achieved on a column (2.5 x 7 cm) of DE-52 microganular cellulose (Whatman) or DEAE-Trisacryl M (LKB) equilibrated with 0.1 M Tris-HCI buffer, pH 8.5, 8 M urea and 0.1 M I~-mercaptoethanol (Sigma). Elution was performed at a flow rate of 33 ml/cm:/h and fractions of 9.5 ml were collected. The acidic ct-subunit was eluted with a gradient of sodium chloride (from 0 to 0.5 M, 2 x 100 ml) dissolved in the above buffer. Proteins were monitored at 280 nm in a 1 cm path cuvette with a Zeiss PMQ II spectrophometer. The isolated ct and 13-subu- nits were dialyzed against distilled water and freeze- dried. The two fractions were then dissolved in 0.2 M Tris-HCI buffer, pH 8.6, 6 M guanidine. HCI, reduced with 0.05 M DTI', and voided through a Sephadex G-50 column (2 x 77 cm, fractions 1.5 ml) equilibrated with the same buffer devoid of D'IT. This was done to get rid of expected mixed disulfide bridges between the ct and [I-subunits and 13-mercaptoethanol; flow speed was 0.8 ml/cm2/h. The two subunits were then dialyzed against distilled water and freeze-dried.

Molecular weight determination of the reduced, alkylated ct, 13-subunits and native PAG was performed by molecular sieve chromatography on a column (I x 85 cm, 1 ml fractions) of Sepharose CL-6B equilibrated with 0.1 M phosphate buffer, pH 7, 0.1 M NaCI and 8 M urea; flow rate was of 12.7 ml/cm"/h. The native PAG or standard proteins (1 mg) were reduced for 30 min with 150 I11 of 0.1 M Tris-HCl buffer, pH 8.5, 8 M urea containing 0.05 M D-IT and 0.2 % blue dextran 2 000 under constant agitation at room temperature; 150 lal of the above buffer contai- ning only 0.15 M iodoacetamide and 0.2 % blue dextran 2 000 were then added to the reaction mixture for another 30 min. DNP-Ala was incorporated to the 300 I~1 samples which were layered sequentially on top of the column. Concerning the ct and 13-subunits, each protein (i mg) was reduced and alkylated by the same procedure but in the presence of 0.2 M Tris-HCl, pH 8.3, 5 mM EDTA and 6 M Gua.HCI. Monitoring was done at 230 nm and 280 nm for proteins and 340 nm for DNP-AIa. Standard proteins used in this experiment were : bovine serum albumin, M, = 66,296 [141; pepsin, M, = 34,700 [15]; soya bean trypsin inhibi- tor, M, = 20,095 [16]; horse heart cytochrome C, M , = 12,384 [17]; bovine lung trypsin inhibitor, Mr = 6,500 [181 and glucagon, Mr = 3483 [16].

Homogeneity of the reduced, alkylated tt and I~- subunits was checked by SDS-PAGE, according to Segrest and Jackson [13]. Duplicate 7 % polyacrylamide gels were treated with the periodic acid Schifrs reagent for carbohydrates.

N-terminal amino-acid determination of the intact molecule was done according to Zanetta et aL [19].

Platelet-activating glycoprotein front snake venom 1233

N-terminal amino-acid sequences of the tt and l~- subunits were determined by the micro-sequence analy- sis described by Chang et aL [20].

Carbohydrate analysis : carbohydrates of the ct and l~-subunits were determined by gas-liquid chromato- graphy, as described by Chambers and Clamp [21], on a 5750 G Hewlett Packard chromatograph.

Amino-acid compositions were determined on a Beckman Multichrom B with a ICAP I0 integrator calculator, using a single column procedure according to Hummel [22]. The samples were hydrolyzed in vacuo in 6 N HC1 for 15, 48 and 72 hours at II0°C according to Spackman et aL [23]. The tryptophan content was determined spectrophotometrically with a Zeiss PMQ II spectrophotometer according to Beaven and Holiday [24]. Methionine was determined as methionine sulfone after performic oxydation and half cystine was esti- mated as cysteic acid after performic oxydation, accor- ding to Hirs [25].

Determh!ation o f the molecular absorption coefficients at 280 nm of the intact molecule and of the a and 13-subunits was performed by calculation according to Edelhoch [26]; no corrections were made for absorp- tions of half cystines or reduced disulfide bonds.

Antibodies against the intact molecule or against the ct and 13-subunits were raised in white Buscat rabbits. The animals were injected intradermally in the back with one of the three proteins (about 100 to 200 lag) dissolved in acidified saline and emulsified with com- plete Freund's adjuvant (CFA), (protein/CFA = v/v). Antibody production was boosted three weeks after the first injection by injecting either the native intact molecule or the et and I]-subunits. Blood samples were collected at intervals during the sequential stimulation and assayed against the three antigens.

Double diffusion analysis was done according to Ouchterlony [27], staining and destaining according to Ouchterlony and Nilsson [28].

Toxicity of convulxin from the venom of Crotalus durissus terrificus was determined as previously descri- bed for PAG from the venom of Crotalus durissus cascavella [12].

Comparative SDS-PAGE of convulxin from the ve- nom of Crotalus durissus terrificus with PAG from the venom of Crotalus durissus cascavella was done accor- ding to Laemmli [29] in a 20 % polyacrylamide slab gel with a I0 % stacking gel.

Results

Before the two subunits were separated, the dissociat ion o f native PAG into a monodisperse state (in the presence o f the denaturat ing buffer devoid o f the reducing agent) was confirmed as

1,5

! - -

0 . 5 - -

--J ~AA

1 0

I 20 30 40

FRACTIONS

FIG. I. -- Separation of the two subunits, a and fl, presenl in the platelet-activating glycoprotein MtTtcture.

The two subunits were isolated by ion exchange chromato- graphy on a DEAE Trisacryl M column with 8 M urea, 0.1 M Tris-HCi buffer, pH 8.5 and 0.1 M 13-mercaptoethanol. The dashed line represents the NaCI gradient from 0 to 0.5 M,

50

previously described [13]. Its molecular weight was 72,000 [13]. The reduct ion o f native PAG with 0.1 M J3-mercaptoethanol in the presence o f the denaturat ing buffer p romoted the elution o f a first componen t believed to be basic since it was not retained by the anionic exchanger. This componen t is subsequently referred to as fL A second componen t which was retained by the gel, was eluted with a sodium chloride gradient (from 0 to 0.5 M). It was considered to be acidic and is subsequently referred to as ~ (Fig. 1). Results indicated that the total optical densities at 280 nm were roughly similar for both subunits (co = 23.2 and 13 = 22.4). This observat ion was conf i rmed after gel filtration o f the two DTT-re-reduced and 13-subunits through a Sephadex G-50 gel, equilibrated with 6 M guanid ine-HCl at basic pH (~ - 19.2 and ~ = 21,4, Fig. 2). After extensive dialysis against distilled water and freeze-drying, about 13 mg of each subunit were obtained (c~-- 12.6 mg and ~ - - 12.55 mg) which gives a yield o f a round 72 % for 34.75 mg of native PAG.

1234 G. Marlas

2

~o lOO

FRACTX~NS

2

1

m

5o too

FRACTIONS

L

FIG. 2. -- Gel filtration o f the two isolated a and fl subunits through a Sephade.v G-50 cohtmn equilibrated with 6 M guanidine. HCI and buffered with 0.2 31 Tris-HCl at pH 8.6.

The total optical densities at 280 nm for each subunits were tt = 19.2 and 13 = 20.7. Before gel filtration, the two subunits were re-reduced with 0.05 M D-IT and then voided through the column.

Homogeneity o f the a and fl-subunits

In spite of severe reduction and subsequent alkylation with iodoacetamide, electrophoresis of the ct and ~-subunits in SDS-PAGE (7 % polyacry- lamide gels, 8 M urea) indicated marginal reoxy- dation or aggregation, as shown by the presence of discrete polymers up to pentamers (Fig. 3). The 13-subunit appeared to be more resistant to re- duction than the ct-subunit since a slightly aggre- gated material was stained by coomassie blue on top of the gels. However, the relationship log Mr-log N (polymers of number N) yielded straight lines extrapolating to the a and l~-mono- mers (not shown). The M,s of the monomers were determined by molecular sieve chromatography under denaturating conditions (see next section). These results allowed us to consider the two purified subunits as chemically pure and not cross-contaminated. On the other hand, neither the ~ nor the 13-subunits were positive t6 the periodic acid Schiff's reagent after 8 M urea, SDS-PAGE. This negative result was obtained twice.

e

i L_

FIe3. 3. - - Homogenei O" o f the two subunits, a and fl, present in the platelet-activating gl)'eoprotein structure.

The fully reduced and alkylated subunits were electro- phoresed in 8 M urea, 0.1 per cent SDS and 0.1 M phosphate buffer, pH 7. The concentration of polyacrylamide was 7 %.

Platelet-activating glycoprotein from snake venom 1235

Determination of the apparent molecular weights of the a and fl-subunits

The apparent M, of the reduced and alkylated oligomer was first determined. The Kay obtained for the natural mixture of the ct and I~-subunits and interpolated into the Kav-log Mr plot o f the standard proteins used yielded an apparent M, of

12,540 (Fig. 5, 2 determinations). Figure 4d shows the symmetrical elution profile of the ct, 13 natural mixture. Gel filtration of the ct-subunit also gave an apparent Mr of 12,540 (Fig. 5, 2 determina- tions); its elution profile was also fairly symmetri- cal (Fig. 4a). On the contrary, the I~-subunit appeared slightly larger with an apparent Mr of 13,770 (Fig. 5, 2 determinations) and had a

I I 0.5

| m a

0.75

0.5 - -

Vo a v, ~'~ Vo

J I Io.,, i

l I I .I I / / t I i ~ I 30 40 50 60 70 80 20 30 40 50 150 70

FRACTIONS FRACTIONS

b

v,

t

! 8O

o o

0.5

Vo !

C

a.13

I_ '1 ~1 40 50 60 70

FRACTIONS

vo 1

o o <~<~ I I O5 --

, I '20

d

,11 I I I ! 20 30 80 30 80 40 50 60 70

F R A C T I O N S

FIG. 4. - - Elution profiles o f the full)' reduced attd alkylated a- al~d fl-subunits present in the platelet-activating glycoprotein.

(a) and (b) : the ¢t- and 13-subunits were separately gel filtered, respectively. (c) : the two subunits were artificially mixed in a one to one ratio and gel filtered together. (d) gel filtration of the fully reduced and alkylated a3~3 otigomer. The samples were gel filtered under the conditions of Figure 5. Vo and Vt represent the void and the total volumes of the gel for blue dextran 2 000 and DNP-alanine, respectively.

1236 G. Marlas

10

g m

0.5 --

O 0 ',/; I 21o 3

C( f 1 2 . ~ _ ~03,77o1

! ! I I II 1111 ! tl I 1 ! t I l l

M r

FIG. 5. -- Apparent M, of the (t and I~ subunits and of their natural or artificial mixtures, after full reduction and alkyla- tion, by gel filtration with 8 M urea, 0.1 M phosphate buffer, pH 7.6, in the presence of 0.1 M NaCI. A Sepharose CL-6B gel column was used in this experiment. Standard proteins were : glucagon, bovine lung trypsin inhibitor, horse heart cyto- chrome C, soya bean trypsin inhibitor, pepsin and bovine serum albumin.

shoulder on the left side of the protein peak Suggesting a reoxydated or aggregated [~2 dimer (Fig. 4b). In confirmation with the value obtained for the natural subunit mixture, the apparent M~ of t h e reconstituted ¢t, I[I mixture was still of 12,540 (Fig. 5, 2 determinations). The symmetrical elution profile had a shoulder on the left side corresponding to a 13~ dimer (Fig. 4c). All the values were calculated by linear regression and no statistics were available.

N-terminal amino-acid determination

The analysis of the N-terminal amino-acid of the intact ct3133 oligomer yielded a unique spot corresponding to a residue of glycine (not shown).

N-termhlal amino-acid sequence of the a and fl-subunits

Manual degradation of the N-terminal amino- acids of the ct and 13-subunits gave single spots for each step indicating a high degree of purity for the two purified proteins. Up to five amino-acids were determined without any ambiguity. They were identical in both subunits and were as follows :

Gly-Phe-Arg-Pro-Asp-X-

Carbohydrate analysis

Carbohydrate analysis of PAG confirmed that it contained a small amount of sugar, as 'pre- viously described [12]. All sugars were recovered, but a 4 hours hydrolysis in 3 N HCI revealed the presence of N-acetylgalactosamine which was not

quoted in our previous determinations. Carbohy- drate analysis o f the two subunits indicated that they were similarly glycosylated to a small extent. All sugars present in PAG were recovered in each subunit, except for fucose and N-acetylneurami- nic acid which were probably destroyed in the course of experiments. Values could not be quantified accurately by the techniques employed and therefore are not (given) in this paper.

Amhlo-acid compositions

Table I shows the amino-acid compositions for the ct- and 13-subunits. Amino-acids were calcu- lated for subunit masses of ct = 12,540 and 13 = 13,770. Results showed that proline, methio- nine and tryptophan residues occurred in equal numbers in both subunits. The ~- and ~-subunits would thus contain 108 and 119 residues with Mrs of 12,496 and 13,673, respectively. The total a3 I~s molecule would have 681 residues for a molecular mass of 78,507. Table I shows that the I~-subunit

TABLE I Amino-acid compositions obtained for the a- and fl- subunits present in the platelet-actirating gl)'coprotein structure. Threonine and serine were corrected for zero time hydrolysis. Nanomoles amounts o f amino-acids were normalized to alanine and expressed as moles per mole o f subunit, assuming M. o f 12,540 and 13,770for a and fl, respectively. Tryptophan was determined spectrophoto- metrically for the two subunits and for the intact ctsfls

o/igomer. Starred amino-acids are non polar and hydrophobic.

a ~ 3a+3 0

ASX 10.26 10 11.04 11 63

THR 7.O3 7 5,37 5 36

5ER 11.97 12 11.11 11 6 9

GLX 12.94 13 15.05 15 84

PRO 'k 2.64 3 3.37 3 18

GLY 7,37 7 8.10 8 45

A L A * 4 .88 5 7.17 7 36

VAL t 6 .45 6 7.45 7 39

½CYS 3.61 4 5.1 6 5 27

MET lit 0 .88 1 1,36 1 6

ILEU "A" 2.88 3 4.94 5 24

L E U * 4 .25 4 7.67 8 36

IYR 4.49 4 5.8B 6 3 0

TRP "~ 4 .84 5 4.66 5 3 0

PHE* 9.77 10 6.45 6 4 8

HIS 4 .49 4 3.08 3 21

LYS 7.72 8 10.03 10 54

.~,RG 1.95 2 2,51 3 15

108 119 681

Mrs 121496 13j673 78~07

Platelet-activating glycoprotein from snake venom 1237

contained 5 half-cystines, a result which will be discussed hereafter. Spectroscopic determinations of the tryptophan contents with MTyr /MTrp ratios of ~t = 0.93 and 13 = 1.26 yielded 5 residues

o f tryptophan for each subunit. Non polar, hydrophobic residues represented 34,3 % of the ct-subunit, 35,3 % of the 13-subunit and 34,8 % of the total ot3 133 molecule. Calculation of the extinc- tion coefficients at 280 nm, according to the amino-acid compositions, yielded :

et-subunit : 33,570 = 2.67 12,540

13-subunit : 36,130 = 2.62 13,770

Total et3 133 molecule : 209,100 = 2.64 78,930

Double diffusion analysis

Immune rabbit antiserum raised against native PAG precipitated with PAG and with convttlxht from the venom of Crotalus durissus terrifictts, no spur were seen in the gels between the two proteins (Fig. 6). Rabbit antisera raised separately against the ct and the 13-subunits neither precipita- ted with PAG nor with the separated subunits which were insoluble in our experimental system.

Comparative SDS-PAGE of PAG and convu lx in

Comparative SDS-PAGE of the two reduced proteins showed (Fig. 7) that convulxin from the venom of Crotalus durissus terrificus also posses-

(X..~ "~"'(X

. . . . . . . . . . . . . . . i

FIG. 7. - - Comparat ive SDS-PAGE of the platelet-activa- ting glycoprotein (PAG) and of the platelet-activating convn- txin isolated from the venoms of Crotalus dltrisszts cascarella (C.d.c.) and Crotalus durissus terrificus (C.d.t.). The separated a- and ~-subunits of the PAG from C.d.c. were included in the run. All proteins were incubated in 0.5 N NaOH and 0.5 % SDS and reduced with 0.05 M DTT overnight. The run was performed at basic pH in a 20 % polyacrylamide gel with 0.1% SDS; a 10% stacking gel was used in this experiment.

sed two subunits with Mr~ identical to those of the PAG rx and 13-subunits, i.e., ct = 12,330 and I 3 = 13,980 in the present determination.

Toxicity o f convu lx in

Convulxin f r o m the venom of Crotahts durissus terrificus, dissolved in acidified saline, (pH 3.5), and injected into mice intravenously (0.57-4.57 mg/kg) produced only transient apnea and appa- rent discomfort, as did PAG from the venom of Crotahts durissus cascavella [12]. This protein like PAG was not lethal for mice and did not induce convulsions.

C.d.c.

0" ~,,x:l.t. en.l cxlx:.O B C.d .

en.g, e Cxl.e.

Schematic model o f PAG

Figure 8 shows one simplified representation of the molecule with its six subunits held by their disulfide bonds. The subunits were drawn in an alternative mode to conserve supposed heterolo- gous associations between them. Spikes represent the carbohydrate moiety.

Discussion FIG. 6. - - Double diffusion analysis of PAG and com'uLrin

from the venoms ofCrotahts durissus cascavella (C.d.c.) and Crotahts durissus terrificus (C.d.t.), 5 lag per well. The central well contained 5 Izl of anti-PAG rabbit anti-serum.

In a previous report [13], it was concluded that the acid-soluble platelet-activating glycoprotein of Mr 72,000, purified from the venom of Crotalus

1238 G. Marlas

Ot31~:3 FiG. 8. -- A proposed and simplified model of the

platelet-activating glycoprotein isolated from the venom of Crotahts durissus cascarella. The subunits volumes were represented according to the apparent MR obtained by gel filtration, and disulfide bridges were drawn from results of amino-acid compositions. Spikes on each subunit represent the carbohydrate moiety.

durissus cascavella [12] has an hexameric structure and is an cq [13 complex containing two types of subunits.

Additional data reported here confirms this hypothesis and shows that these two subunits exist in a one to one ratio within the native molecule, as shown by their equivalent amounts obtained after their isolation through ion ex- change chromatography under denaturating con- ditions. Homogeneity of the two subunits was demonstrated by 8 M urea, SDS-PAGE and by analysis of their respective N-terminal amino-acid sequences which showed a high degree of purity. The determination of their apparent M, by molecular sieve chromatography in the presence of 8 M urea was quite reliable since the apparent Mr of the reduced and alkylated ct3 [~3 oligomer was 12,540, a value in very good agreement with the one previously determined in the presence of 6 M guanidine-HCl, which was 12,555 [13]. These figures are consistent with the mean value of 13,000 previously obtained by 8 M urea, SDS-PAGE with the reduced and alkylated ft31~3 oligomer [13] Furthermore, gel filtration data showed that the 13-subunit had a slightly larger apparent Mr : 13,770 as compared with 12,540 for the ct-subunit. These figures are in excellent agreement with the SDS-PAGE, previously repor- ted values o f f t = 12,600 and [~ = 13,580 [13]. On the other hand, the natural or reconstituted mixtures (w/w) of the reduced and alkylated fz and ~-subunits showed an apparent Mr of 12,540 instead of 13,155 (mean calculated Mr). This deviation from the mean calculated M: could he

due to experimental uncertainty. However, addi- tion of the values of the subunit masses yielded a total calculated Mr of 78,930 for the native molecule, a value very close to that of 78,400 previously determined by 8 M urea, SDS-PAGE [13]. Also, addition of the subunit masses obtai- ned by SDS-PAGE [131 furnished a total calcu- lated Mr of 78,540 which is close to 78,400 obtained for the native molecule.

Carbohydrate analysis confirmed that the native molecule contains a small amount of sugar in agreement with the low value of 5.3 % pre- viously reported [12] and showed that both su- bunits are glycosylated. They contain mannose , galactose, glucose, N-acetylgalactosamine and N-acetylglucosamine. Glucose should not elute from Sephadex since the desalting of subunits was done in the presence of 6 M guanidine-HCl. On the other hand, fucose and N-acetylneurami- nic acid, indeed present in the native molecule [12] and this work, were not recovered in the two subunits. This could be due to destruction during experiments.

Although, the data from carbohydrate analysis of the two subunits are preliminary, it is supposed that the subunits, on an average, have similar compositions and percentages of sugar. A more elaborate technology would be necessary to have a definite insight into the carbohydrate composi- tions and sequences of the two subunits. Howe- ver, a 4-5 % carbohydrate moiety for the native molecule would mean 2 to 3 moles of sugar per mole of subunit since they have similar Mrs. This would a priori constitute a negligible hindrance with regard to the hydrodynamic behaviour of the reduced and alkylated subunits, even if these 2-3 residues were located in only one place. Thus, the hydrodynamic Mrs values obtained for the two subunits should be significant.

The amino-acid compositions of the two subu- nits appeared not to be identical, except for the proline, me th ion ine and tryptophan residues which seemed to occur in equal numbers. A value of 5 half-cystines was calculated for the l~-subu- nit : indeed, the figure should be of 4 or 6 as the native molecule was shown to contain no free -SH group [11]. Results indicated that the hydrophobi- cities of the two subunits were quite similar (c~ = 34.3% and 1] = 35.3%) as well as their respective extinction coefficients at 280 nm (ft = 2.67 U.O.D. /cm/mg/ml and i] --" 2.62 U.O.D. /cm/mg/ml) , as hypothesized in the course of their isolation. The previous reported value of 2.5 U.O.D. /cm/mg/ml [12] for the

Platelet-activathlg glj'coprotehl from snake venont 1239

extinction coefficient of the native molecule at 280 nm fits very well with the present calculated figure of 2.64 U.O.D. /cm/mg/ml . The presence of a single glycine as the N-terminal residue for the native molecule was explained by the sharing of a common N-terminal sequence between the two subunits, determinable up to the fifth residue. This would suggest sequence homologies encoded by similar genes. Non homologous regions of the sequences of the two subunits would then account for their different behaviours during ion exchange chromatography. Rabbit immune serum raised against native PAG from the venom of Crotahts durissus cascavella displayed cross-reactivity by double-diffusion analysis with the platelet-activa- ting convulxin from the venom of Crotahls durissus terrificus. The latter technique indicated total identity between PAG and convulxhl. This identity was confirmed by the comparative SDS-PAGE of the two proteins which revealed two subunits in the structure of convulxin, the Mrs of which were identical to those of the PAG subunits. This, with the display~ of identical platelet-stimulating activi- ties on washed human platelets (not shown) and similar effects in mice (absence of convulsions and of lethality in spite of a fast and severe thrombocytopenia) demonstrates that the two platelet-activating agents PAG and convuL,:in, pre- sent in two zoologically different subspecies of Crotalus durissus, are the same multi-subunit glycoprotein.

Acknowledgments We thank Prof. Dr M.E. Goldberg, Dr B.B. Vargaftig

and Prof. Dr C.L. Ownby for revision of the manuscript. This work was supported by funds from the Institut Pasteur. We are grateful to Miss C. Goulu from the Unitd de Biologie et Pathologie Mol~culaire des Glycoprotdines, Facultd de Mddecine LariboisiOre-Saint-Louis, Paris (Prof. Y. Goussault), for the determination of the carbo- hydrate content of PAG and of its subunits. We thank A.M. Gilles from the Un#d des Protdines, lnstitut Pasteur, Paris (Prof. B. Keil), for the determination of the amino- acid compositions of the PAG subunits. This work is part of the doctoral thesis of Guy Marlas.

REFERENCES

1. Markwardt, F., Barthel, W., GIusa, E. & Hoffman, A. (1966) Naunyn-Schmiedeb. Arch. exp. Pathol. und Pharmacol. 252, 297-304.

2. Markwardt, F. (1966) Mere. Inst. Butantan Simp. Internac., 33, (3), 851-854.

3. Vital Brazil, O., Prado-Franceschi, J. & Waisbich, E. (1967) Gi~ncia cultura, 19, 658-665.

4. Prado-Franceschi, J. & Vital Brazil, O. (1969) Convulxina, uma nova neurotoxina da pe~onha

da Crotalus durissus terrificus. Presented at the IX Congr. Latino-Americano, Ci~nc. Fisiolog., Brazil, 91-92.

5. Prado-Franceschi, J., Tavares, D.Q., Hertel, R. & Lobo De Arafijo, A. (1981) Toxieon, 19, No 5, 661-666.

6. Prado-Franceschi, J. & Vital Brazil, O. (1981) Toxicon, 19, No 6, 875-887.

7. Vargaftig, B.B., Prado-Franceschi, J., Chignard, M., Lefort, J. & Marlas, G. (1980) Eur. J. Parmacol., 68, 451-464.

8. Vargaftig, B.B., Joseph, D., Wal, F., Marlas, G., Chignard, M. & Chevance, L.G. (1983) Fur. J. Pharmacol., 92, 57-68.

8. Randon, J., Lecompte, T., Chignard, M., Siess, W., Marlas, G., Dray, F. & Vargaftig, B.B. (1981) Nature, 293, No 5834, 660-662.

10. Vargaftig, B.B. (1982) Toxicon, 20, No l, 279-287. I I. Marlas,G. (1982) Toxicon, 20, No l, 289-290. 12. Marlas, G., Joseph, D. & Huet, C. (1983) Biochimie,

65, No 7, 405-416. 13. Marlas, G., Joseph, D. & Huet, C. (1983) Biochimie,

65, No ll-12, 619-628. 14. Berhens, P.Q., Spiekerman, A.M. & Brown, J.R.

(1975) Fed. Proc., 34, 591. 15. Muravek, L. & Kostka, V. (1974) Febs Lett., 43,

207-21 I. 16. Dayhoff, M.O. (1972-1976) in : Atlas of Protein

Sequence and Structure (National Biomedical Research Foundation, Silver Spring, Ed.) Vol. 5.

17. Schetjer, A., Glauser, S.C., George, P. & Margo- liash, E. (1963) Biochim. Biophys. Acta, 73, 641-643.

18. Trautschold, E., Zickgraf-Riidel, W. & Zickgraf- RQdel, G. (1967) Biochem. Pharmacol., 16, 59-72.

19. Zanetta, J.P., Vineendon, G., Mandel, P. & Gom- bos, G. (1970) J. Chromatog., 51, 441-458.

20. Chang, J.Y., Brauer, D. & Wittmann-Liebold, B. 0978) Febs Lett., 93, No 2, 205-214.

21. Chambers, R.E. & Clamp, J.R. (1971) Biochem. J., 125, I009-1018.

22. Hummel, B.C.W. (1959) Can. J. Biochem. Physiol., 37, 1393-1399.

23. Spackmann, D.H., Stein, W.H. & Moore, S. 0958) Anal. Biochem., 30, 1190-1206.

24. Beaven, G.H. & Holiday, E.R. (1952) in : Advance in Protein Chemistry. (Anson, M.L., Bailey, K. & Edsall, J.T., Ed.) vol. 7, pp. 319-387, Academic Press, New York.

25. Hirs, C.H.W. 0967) in : Methods in Enzymology. (Hirs, C.H.W., Ed.) vol. I l, pp. 59-62, Academic Press, New York and London.

26. Edelhoch, H. (1967) Biochemistry, 6, No 7, 1948-1954.

27. Outchterlony, O. (1953) Acta Path. Microb. scand., 32, 231-240.

28. Oucllterlony, O. & Nilsson, L.-A. (1978) in : Hand- book of Experimental Immunology. (Weir, D.M. Ed.) vol. l, pp. 19.1-19.44.

29. Laemmli, U.K. 0970) Nature, London, 227, 680-685.