(Received for publication, January 9, 1997)
From the Theodor Kocher Institute, University of
Berne, Freiestrasse 1, CH-3012 Berne, Switzerland,
§ Experiment, Hämostaseforschung, Innere Medizin A,
Universitätskliniken Münster,
Domagkstrasse 3, D-48149 Münster, Federal Republic of Germany,
and ¶ Geneva Biomedical Research Institute, Glaxo Wellcome
Research and Development S.A.,
14, Chemin des Aulx, CH-1228 Plan-les-Ouates, Geneva, Switzerland
Convulxin, a powerful platelet activator, was
isolated from Crotalus durissus terrificus venom, and 20 amino acid N-terminal sequences of both subunits were determined. These
indicated that convulxin belongs to the heterodimeric C-type lectin
family. Neither antibodies against GPIb nor echicetin had any effect on
convulxin-induced platelet aggregation showing that, in contrast to
other venom C-type lectins acting on platelets, GPIb is not involved in
convulxin-induced platelet activation. In addition, partially
reduced/denatured convulxin only affects collagen-induced platelet
aggregation. The mechanism of convulxin-induced platelet activation was
examined by platelet aggregation, detection of
time-dependent tyrosine phosphorylation of platelet
proteins, and binding studies with 125I-convulxin.
Convulxin induces signal transduction in part like collagen, involving
the time-dependent tyrosine phosphorylation of Fc receptor
chain, phospholipase C
2, p72SYK, c-Cbl, and p36-38.
However, unlike collagen, pp125FAK and some other bands are
not tyrosine-phosphorylated. Convulxin binds to a glycosylated 62-kDa
membrane component in platelet lysate and to p62/GPVI
immunoprecipitated by human anti-p62/GPVI antibodies. Convulxin
subunits inhibit both aggregation and tyrosine phosphorylation in
response to collagen. Piceatannol, a tyrosine kinase inhibitor with
some specificity for p72SYK, showed differential effects on
collagen and convulxin-stimulated signaling. These results suggest that
convulxin uses the p62/GPVI but not the
2
1 part of the collagen signaling
pathways to activate platelets. Occupation and clustering of p62/GPVI
may activate Src family kinases phosphorylating Fc receptor
chain
and, by a mechanism previously described in T- and B-cells, activate
p72SYK that is critical for downstream activation of
platelets.
A large number of C-type lectins from snake venoms have been described over the last few years with effects on hemostasis. While most of these inhibit the function of the coagulation factors and platelet components that they bind to, a few activate platelets by direct or indirect effects. So far all of these have been shown to affect the von Willebrand factor (vWf)1-platelet GPIb-V-IX axis. They include botrocetin (1) and bitiscetin (2) that bind to and change the conformation of vWf so that it can bind to GPIb and thus activate platelets and alboaggregin B (3) that activates platelets directly by binding to, and presumably clustering, GPIb. A further snake peptide from the venom of some Crotalus species has subunits with a molecular mass similar to the C-type lectins, is a strong activator of platelet phospholipase C, and has been termed convulxin (4-8). We have isolated a similar, possibly identical, molecule from Crotalus durissus terrificus venom and show that it belongs to the heterodimeric, C-type lectin family. It activates platelets not via GPIb but through the p62/GPVI component of the platelet collagen receptor, probably by a clustering effect, and induces signals similar to a set of those induced by collagen in platelets.
Several platelet membrane glycoproteins have been implicated as
collagen receptors, in particular GPIa-IIa (the
2
1 integrin) (9-12), CD36 (13), and
p62/GPVI (14-16). Since patients platelets lacking
2
1 do not adhere to or aggregate to
collagen (9, 10), and antibodies against
2
1 block platelet adhesion to immobilized
collagen (17-19), there is general agreement about the pivotal role of
2
1 in the adhesion of platelets to
collagen fibers. Furthermore, it was suggested that platelet adhesion
to collagen through the
2
1 integrin
induces rapid tyrosine phosphorylation of pp125FAK; thus,
the
2
1 integrin plays a more direct role
in early events of collagen-induced platelet activation, other than
mediating the initial adhesion to collagen (20, 21). Many of the
platelet responses, induced by collagen, can be induced by synthetic,
triple helical, collagen-like peptides based upon typical collagen
sequences incorporating multiple glycine-proline-hydroxyproline repeats (22-24). These responses are independent of the
2
1 receptor, which shows that receptor(s)
other than
2
1 are also involved in the
collagen-induced events. A platelet-activating antibody, anti-p62 IgG,
found in a patient with autoimmune thrombocytopenia with a selective
deficiency in collagen-induced platelet aggregation (25),
immunoprecipitated a 62-kDa protein from normal platelets that was
later described as GPVI (14). All four p62/GPVI-deficient patients,
reported so far (14-16, 25), showed defective platelet responses only
to collagen despite the normal expression of
2
1. Although CD36, which has been found
to be associated with Src family kinases (26), also may play some role,
the fact that patients lacking CD36 adhere normally and aggregate to
collagen (27-30) shows that CD36 is less critical in collagen-induced
platelet activation than
2
1 and p62/GPVI.
A characteristic of platelet activation by collagen compared via
G-protein-linked receptors, such as that for thrombin, is the early
engagement of a wide range of tyrosine kinases, phosphatases, and their
substrates. Collagen stimulation of platelets results in the
phosphorylation of numerous proteins (reviewed in Ref. 31). Recent
reports include collagen-induced tyrosine phosphorylation of
phospholipase C
2 (PLC
2) providing a mechanism for further
platelet activation by increasing cytosolic Ca2+ and
activating protein kinase C (32, 33) and of the Fc receptor
chain
(Fc
) providing a mechanism for activation of the major tyrosine
kinase p72SYK (24). Because there are at least two collagen
receptors on platelets, but the relative contribution of each to
collagen-induced platelet activation has not yet been determined, it is
of interest to find platelet agonists that mimic the effect of collagen
by acting through only one of the collagen receptors.
Here we show that convulxin, a powerful platelet agonist purified from C. durissus terrificus venom, belongs to the C-type lectin family, induces platelet activation/aggregation predominantly through one of the collagen receptors, p62/GPVI, and identify some of the signaling molecules involved.
Lyophilized C. durissus terrificus
venom, Protein A-Sepharose, peroxidase-conjugated goat anti-mouse and
anti-rabbit antibodies, bovine serum albumin, wheat germ agglutinin
(WGA), and Triton X-114 were from Sigma. Sepharose 4B was from
Pharmacia Fine Chemicals (Uppsala, Sweden). Iloprost was a kind gift
from Schering AG (Zürich, Switzerland). Na125I was
from Amersham Corp. (Zürich, Switzerland), IODO-GEN and the
SuperSignal chemiluminescence detection systems were from Pierce.
Autoradiography (Fuji RX) films were from Fujifilm (Dielsdorf AG,
Switzerland). Methylated type I calf skin collagen was a kind gift from
Dr. J. Rauterberg. Piceatannol was from Boehringer Mannheim (Germany).
Anti-phosphotyrosine mAb (4G10) was from Upstate Biotechnology Inc.
(Lake Placid, NY); anti-p72SYK (4D10) mAb, anti-c-Cbl
(C-15), and anti-pp125FAK (A-17) polyclonal antibodies and
Grb2-(54-164)-AC were from Santa Cruz Biotechnology, Inc. (Santa Cruz,
CA); anti-FcRII (IV.3) mAb was from Medarex Inc. (Annandale, NJ);
AP-1 anti-GPIb mAb was a kind gift from Dr. T. J. Kunicki; SZ-2
anti-GPIb mAb was a kind gift from Dr. C. Ruan; Ib-23 anti-GPIb mAb and
Ro44-9883, a GPIIb-IIIa inhibitor, were kind gifts from Dr. B. Steiner; PTA-1 mAb was a kind gift from Prof. G. Burns; human anti-GPVI
polyclonal antibody was a kind gift from Prof. M. Okuma; anti-GP60 (a
putative laminin receptor) mAb was a kind gift from Dr. S. Sentoso; mAb 6F1 against
2
1 was a kind gift from Prof.
B. Coller; polyclonal antibody against Fc
RI
was a kind gift from
Prof. J.-P. Kinet. Echicetin was purified from lyophilized Echis
carinatus sochureki venom (Latoxan, Rosans, France) as described
previously (34). PVDF membranes were PolyScreen from DuPont NEN.
Octyl-N-methylglucamide was from Oxyl Chemie (Bobingen,
Germany).
Lyophilized C. durissus terrificus venom was dissolved at 250 mg/8 ml in 300 mM NaCl, 100 mM ammonium formate, pH 3.5 (buffer A). Insoluble components were removed by centrifugation, and the supernatant was loaded on a Fractogel EMD BioSEC 650 (S) column (16 × 1200 mm; Merck, Darmstadt, Germany) equilibrated with buffer A. At 0.5 ml/min flow rate, 5-ml fractions were collected. Fractions were analyzed by SDS-PAGE/silver staining under nonreducing and reducing conditions and assayed for their ability to induce platelet aggregation. The pH of 100-µl aliquots was adjusted to 7.4 by adding 10 µl of 1 M Tris, pH 8.5, before the aggregation studies. Fractions that induced platelet aggregation and showed strong 85-kDa (nonreduced) and 14- and 16-kDa (reduced) bands on analysis by SDS-PAGE/silver staining were pooled (15 ml), dialyzed against diluted HCl, pH 3.5, dissolved in 3 ml of buffer A, and rechromatographed under the same conditions as above. Fractions corresponding to the main peak, high purity convulxin, were concentrated to about 20% of their original volume on a Speed-Vac. If a sample became opalescent during this step, a small volume of 0.1% trifluoroacetic acid was added to redissolve the protein. After dialysis against first, 150 mM NaCl, 50 mM phosphate, pH 5.0, second, the same buffer at pH 6.5, and finally, the same buffer or 150 mM NaCl, 50 mM Tris/HCl, pH 7.4, the convulxin was stored at 4 °C until used.
Purification and Sequence Analysis of Convulxin SubunitsTo
1 volume of 300 µg/ml convulxin in 6 M guanidine HCl, 0.1 M Tris, pH 8.0, was added to 1/20 volume 440 mM
DTT followed by incubation at 45 °C for 30 min. 1/20 volume of 1 M 4-vinylpyridine was added, and after 1 h incubation
at room temperature the sample was diluted with 2 volumes of 0.2%
trifluoroacetic acid. Modified convulxin subunits were isolated by
reverse phase HPLC on a wide pore C4 column (4.6 × 250 mm; J. T. Baker Inc.) using an acetonitrile gradient (0.1% trifluoroacetic
acid). N-terminal sequencing of S-pyridylethylated and
subunits of convulxin was performed on an Applied Biosystem model
477 A pulsed-liquid-phase protein sequencer with model 120 A on-line
phenylthiohydantoin amino acid analyzer.
Partially reduced/denatured convulxin was prepared in
the same way as S-pyridylethylated and
subunits of
convulxin, omitting 4-vinylpyridine so that the free
SH groups were
not chemically modified in these samples. The protein was isolated,
free from denaturing and reducing agents, by reverse phase HPLC,
freeze-dried, and dissolved in 0.05% trifluoroacetic acid. This
treatment resulted in three main fractions corresponding to the two
subunits and a fraction that contained both subunits. The ratio of
these three fractions varied between preparations.
Convulxin was labeled by the IODO-GEN procedure (35). Convulxin, 80 µg, and 800 µCi of Na125I were added to a glass tube that had been coated with 150 µg of IODO-GEN. The reaction mixture was incubated for 30 min at room temperature with gentle agitation. The reaction was stopped by adding 30 mg/ml (final) KI and labeled convulxin was separated from free Na125I on a Bio Gel P-6 DG column (0.9 × 16 cm, Bio-Rad) using 5 mg/ml bovine serum albumin, 300 mM NaCl, 50 mM Tris/HCl, pH 7.4. The specific activity of labeled convulxin was between 0.5 and 1.2 × 109 cpm/mg.
Computer Analysis of the Sequence DataComputer analysis of the sequence data was performed on a VAX/VMS system using the suite of programs from the University of Wisconsin Genetics Computer Group.
Protein DeterminationProtein determination was performed by the BCA protein assay (Pierce) with bovine serum albumin as a standard.
SDS-PAGE/Silver StainingSDS-PAGE was performed according to Laemmli (36), and the gels were silver-stained by the method of Morrissey (37).
Preparation of Washed Platelets, Platelet Aggregation, and ImmunoprecipitationsHuman platelets were isolated from buffy coats, less than 20 h after blood collection, obtained from the Central Laboratory of the Swiss Red Cross Blood Transfusion Service. To one buffy coat was added 30 ml of 100 mM citrate, pH 6.5. Platelet-rich plasma and the platelet pellet was isolated by successive centrifugation steps. Platelets were resuspended in 113 mM NaCl, 4.3 mM K2HPO4, 4.3 mM Na2HPO4, 24.4 mM NaH2PO4, 5.5 mM glucose, pH 6.5 (buffer B) and centrifuged at 250 × g for 5 min. The platelet-rich supernatant was centrifuged at 1000 × g for 10 min, and platelets were washed with buffer B once more. Washed platelets were resuspended in 20 mM Hepes, 140 mM NaCl, 4 mM KCl, 5.5 mM glucose, pH 7.4 (buffer C), and the platelet count was adjusted to 5 × 108 platelets/ml by dilution with buffer C. Samples were kept at room temperature until used for aggregation studies. Platelet aggregation was monitored by light transmission in an aggregometer (Lumitec, France) with continuous stirring at 1100 rpm at 37 °C. Platelets were preincubated in buffer containing 2 mM CaCl2 at 37 °C for 2 min before starting the measurement by adding the samples for analysis. For immunoprecipitation, aliquots (700 µl, 5 × 108 platelets/ml) of control, resting as well as activated platelets were solubilized in phosphate-buffered saline containing 1.2% Triton X-100 with 1 mM phenylmethylsulfonyl fluoride, 5 mM EDTA, 2 mM N-ethylmaleimide (NEM), 2 mM benzamidine, and 2 mM sodium orthovanadate. After centrifugation, platelet lysates, precleared with Protein A-Sepharose, were stirred for 2 h with specific antibodies before adding 20 µl of Protein A-Sepharose followed by 6-8 h incubation.
Triton X-114 Phase Separation and Wheat Germ Agglutinin Affinity ChromatographyHuman platelets were isolated from 10 buffy coats as described previously (38). The platelet count was adjusted to 1 × 109 platelets/ml with 20 mM Tris/HCl, pH 7.4, 10 mM EDTA, 154 mM NaCl (TENA buffer), and Triton X-114 phase separation was performed as described previously (39). The Triton phase was diluted 10 times with 2 mM sodium orthovanadate, 2 mM NEM, 4 mM EDTA, 20 mM NaCl, 10 mM Tris/HCl, pH 7.4 (buffer D), and centrifuged at 10,000 × g for 30 min at 4 °C. The supernatant was applied to a column of WGA-Sepharose 4B (2.6 × 15 cm, 1 mg of WGA/ml of Sepharose 4B) equilibrated with buffer D. The column was washed intensively with buffer D containing 0.1% octyl-N-methylglucamide. The bound material was eluted with 2.5% N-acetylglucosamine in buffer D containing 0.5% octyl-N-methylglucamide. Fractions containing glycoproteins were pooled, concentrated 10-fold by ultrafiltration, dialyzed against water, and freeze-dried. The sample was dissolved in 4 ml of phosphate-buffered saline, pH 7.2, containing 2 mM sodium orthovanadate, 2 mM NEM, 4 mM EDTA. 750-µl aliquots of this preparation were used for immunoprecipitations.
Identification of 125I-Labeled Convulxin-binding Proteins in Immunoprecipitated Platelet Proteins by an Affinity Blotting MethodAliquots (750 µl) of platelet glycoprotein samples (see above) were incubated at 4 °C for 3 h with 5 µl of anti-PTA1, anti-p62/GPVI, and anti-GP60 antibodies followed by incubation with 5 mg of Protein A-Sepharose at 4 °C overnight. The Protein A-Sepharose samples were washed four times by centrifugation in ice-cold phosphate-buffered saline, pH 7.2, containing 2 mM sodium orthovanadate, 2 mM NEM, and 4 mM EDTA. The samples were boiled in 40 µl of 2% SDS, 50 mM Tris/HCl, pH 8.0, for 5 min and centrifuged, and the supernatants were electrophoresed on a 7-17% gradient SDS-polyacrylamide gel. Proteins were transferred electrophoretically to PVDF membrane, and convulxin-binding proteins were detected using 125I-labeled convulxin. The membranes were incubated with 1 µg/ml labeled convulxin in 2% bovine serum albumin, 300 mM NaCl, 50 mM Tris/HCl, pH 7.4, for 3 h followed by washing 7 times with the same buffer for 30 min. The membranes were rinsed with water, dried, and radioactive labeling detected by autoradiography using a PhosphorImager (Molecular Dynamics).
Convulxin was
purified from lyophilized C. durissus terrificus venom by a
two-step gel filtration method (Fig. 1). The final product showed an 85-kDa broad band under nonreducing conditions and
two bands, 14 and 16-kDa, under reducing conditions when analyzed by
SDS-PAGE/silver staining (see inset in Fig. 1B).
Treatment with DTT alone under nondenaturing conditions was not
sufficient to separate the subunits by reverse phase HPLC. However,
S-pyridylethylation allowed a separation of the subunits by
reverse phase HPLC. Fractionation of reduced
S-pyridylethylated convulxin subunits is shown in Fig. 2. Peaks in the 3-10-min interval represent non-protein
peaks originated from the reagents used for reduction and
S-pyridylethylation. Fractions 1 and 2 contained homogenous
single chain proteins with apparent Mr of 16 and
14, respectively, on analysis by SDS-PAGE/silver staining (see
inset of Fig. 2). Following the nomenclature of Marlas (7),
we called the 14-kDa protein (fraction 2) the subunit and the
16-kDa protein (fraction 1) the
subunit. N-terminal sequence
analysis gave the 20 amino acid sequence
GFCCPSHWSXYDRYCYKVFK for the
subunit and
GLHCPSDWYYYDQHCYRIFN for the
subunit. Computer analysis of the
sequences showed that convulxin is a heterodimeric C-type lectin with
strong sequence similarity to other venom C-type lectins (Fig.
3).
Convulxin Does Not Act through GPIb or
Convulxin is a powerful platelet
agonist and, at concentrations as low as 3-10 ng/ml, induced maximal
aggregation of isolated human platelets (data not shown). The strong
sequence similarity to C-type lectins that bind to GPIb suggested that
GPIb might be involved in platelet activation induced by convulxin.
However, neither antibodies against GPIb (Ib-23, SZ-2, and AP-2) nor
echicetin, a snake venom C-type lectin that binds to GPIb and inhibits
platelet agonists acting through GPIb (34), had any effect on
convulxin-induced platelet aggregation (data not shown). In addition,
partially reduced/denatured convulxin had no effect on
ristocetin-vWf-induced platelet agglutination (see below). These data
show that GPIb is not involved in convulxin-induced platelet
activation. A mAb against 2
1 (6F1), which
inhibits collagen-induced platelet activation, had little effect on
convulxin-induced platelet aggregation (data not shown).
Neither of the reduced,
S-pyridylethylated convulxin subunits activated platelets
(data not shown), suggesting that the platelet activation effect of
convulxin requires its native conformation rather than a linear
sequence in its subunits. Neither incubation of convulxin with up to 50 mM DTT at 37 °C for 30 min nor freezing and thawing the
convulxin sample in the presence of 50 mM DTT decreased the
ability of convulxin to aggregate platelets. After reduction of
convulxin under denaturing conditions, in the presence of 6 M guanidine HCl, reverse phase HPLC gave three main peaks corresponding to the two subunits and a fraction that contained both
subunits. The ratio of these peaks varied between preparations (n = 10, data not shown). None of these partially
reduced/denatured convulxin subunits induced platelet aggregation or
signal transduction. Seven out of ten preparations, containing either
,
, or both subunits, had no effect on platelet aggregation
induced by various agonists. However, the others inhibited
convulxin-induced aggregation as well as collagen-induced platelet
aggregation in a concentration-dependent manner, whereas
they had no effect on platelet aggregation induced by thrombin or
ristocetin-vWf. These data show that convulxin and its two subunits
bind to a common receptor that is involved in collagen activation of
platelets.
The fact that the N-terminal
sequences of convulxin are rich in tyrosine suggested that it might be
readily labeled by 125I without affecting its biological
activity, and this proved to be the case. The platelet membrane protein
binding 125I-labeled convulxin was identified. It is a
62-kDa hydrophobic, platelet membrane glycoprotein (Fig.
4, lane 1) present in the Triton phase after
Triton X-114 phase separation of platelet membrane proteins that binds
to WGA (Fig. 4, lane 2). 125I-Labeled convulxin
binds strongly only to the 62-kDa protein immunoprecipitated with
anti-p62/GPVI antiserum (Fig. 4, lane 6) and not to other
platelet membrane glycoproteins with similar molecular masses and
physicochemical properties immunoprecipitated by specific antibodies
(Fig. 4, lane 4, anti-PTA-1 mAb and lane 5,
anti-GP60 laminin receptor mAb). This argues convincingly that the
p62/GPVI collagen receptor is the binding site for convulxin on
platelets.
Tyrosine Phosphorylation Signal Transduction by Convulxin Compared with Collagen
Fig. 5 shows a time range of
tyrosine phosphorylation for platelets activated by 5 mg/ml collagen
compared with that with 30 ng/ml convulxin. These relative amounts were
chosen because they gave comparable rates of platelet aggregation as
estimated by the slope of the aggregometer curve. Such aggregation
slopes are often used as the basis for comparison of activation rates. With convulxin several proteins are phosphorylated more rapidly but
also more intensely than with collagen. These include Fc, p36-38,
p72SYK, PI3K, c-Cbl, and PLC
2. These tyrosine
phosphorylations induced by convulxin were not inhibited by raising the
adenylate cyclase level by adding Iloprost (final concentration 0.15 nM) to the platelet suspension 5 min before the agonist
(data not shown). These components were identified by
immunoprecipitation with specific antibodies or, in the case of
p36-38, by precipitation with a GST-Grb2 SH2 fusion protein. Some of
these data are shown in Fig. 6. In Fig. 6A
the first three lanes show the tyrosine phosphorylation pattern at 30 s of resting platelets, platelets activated by
collagen, and platelets activated by convulxin, respectively, and the
following three lanes show the phosphotyrosine proteins
binding to GST-Grb2 SH2 domain from these platelet preparations. This
clearly identifies the tyrosine-phosphorylated p36-38 band present in
the platelets as a major Grb2 SH2-binding protein that is much more
tyrosine-phosphorylated in convulxin-activated platelets and is
distinct from the immunoprecipitated Fc
RIIA receptor that has a
higher molecular mass as shown in the five lanes on the
right. Thus p36-38 is perhaps the human equivalent of, or at
least related to, the Lnk adaptor protein (40). Fc
RIIA also shows
raised levels of tyrosine phosphorylation after platelet activation by
collagen or convulxin but much less than p72SYK, Fc
, and
c-Cbl. Similarly, in Fig. 6B the upper bands show
the level of tyrosine phosphorylation in immunoprecipitates from
resting, collagen-, and convulxin-activated platelets, with
p72SYK, Fc
, and c-Cbl antibodies, respectively, and the
lower bands show the same blots after stripping and
restaining with the corresponding antibodies to show that equal amounts
of each component were immunoprecipitated from each platelet
preparation. p72SYK, Fc
, and c-Cbl are all
tyrosine-phosphorylated after platelet activation by collagen compared
with virtually nil or very low levels in resting platelets but are even
more dramatically tyrosine-phosphorylated after platelet activation by
convulxin. Fig. 6C shows that whereas convulxin is a
powerful activator of pp125FAK, this was delayed compared
with collagen and was completely inhibited in the presence of 1 µM Ro44-9883, a specific, efficient inhibitor of
fibrinogen binding to GPIIb-IIIa (41), indicating that the activation
of pp125FAK by convulxin is via activation of GPIIb-IIIa
and release and binding of fibrinogen rather than via
2
1 that is an early event in platelet
activation by collagen and is not prevented by GPIIb-IIIa inhibitors.
In some cases, however, the collagen induced a more rapid and stronger
earlier phosphorylation than convulxin. This was the case with a band
at 125 kDa (Fig. 5), identified as p125FAK in Fig.
6C, and with so far unidentified bands at about 32 and 28 kDa (Fig. 5). As well as p36-38, GST-Grb2 SH2 fusion protein precipitated an additional band at about 30 kDa from a lysate of
collagen-activated platelets but not from convulxin-activated platelets
(Fig. 6A).
Partially Reduced/Denatured Convulxin Subunits Inhibit Aggregation and Tyrosine Phosphorylation Signal Transduction by Collagen
Partially reduced/denatured convulxin subunits inhibit
collagen-induced platelet aggregation in a dose-dependent
way. Fig. 7A shows the aggregation response
to collagen in the presence and absence of an intermediate dose of a
typical preparation of subunit(s). In the presence of the subunit(s)
the aggregation is rapid at first but quickly reaches a maximum and
reverses, leading to platelet disaggregation. Fig. 7B shows
the time range tyrosine phosphorylation profile. Both subunits gave
similar effects. Compared with the collagen control, in the presence of
the subunits the tyrosine phosphorylation is slower, much less intense,
of shorter duration, and those bands that do show clear but brief increases included those corresponding to pp125FAK and the
32- and 28-kDa groups. Tyrosine phosphorylation of bands corresponding
to p36-38 and to Fc were strongly inhibited and did not increase
during the aggregation phase.
Effects of Piceatannol on Signaling through Tyrosine Phosphorylation by Convulxin Compared with Collagen
As previously
reported (42) piceatannol inhibits both platelet aggregation and
tyrosine phosphorylation induced by collagen. In the case of convulxin,
however, the aggregation was not completely inhibited at doses of
convulxin giving an equivalent aggregation rate, whereas that of
collagen was completely inhibited. In this experiment lower amounts of
the agonists were used to be able to inhibit convulxin-induced
aggregation at relevant doses of piceatannol. With convulxin a brief
reversible aggregation similar to, but weaker than, that shown in Fig.
7A for the convulxin subunits was observed. In addition,
unlike with collagen, the tyrosine phosphorylation in general was not
inhibited and, compared with control, convulxin- activated platelets
lasted longer (Fig. 8). Those proteins with prolonged
tyrosine phosphorylation included p36-38, Fc, and several
unidentified bands including two at about 80 and 180 kDa. On the other
hand, tyrosine phosphorylation of some bands was inhibited by
piceatannol in convulxin-activated platelets (Fig. 8B).
These included unidentified bands at 85 and 120 kDa. Tyrosine
phosphorylation of p72SYK lasted longer in collagen than in
convulxin-activated platelets, peaking in the latter between 10 and
30 s. Tyrosine phosphorylation of p72SYK was clearly
inhibited by piceatannol in collagen-activated platelets. In
convulxin-activated platelets it appeared inhibited at 10 s but
less so at 30 s, otherwise the time course was not affected.
The venom from several Crotalus species contains a
heterodimeric protein that is a powerful platelet activator. We
isolated, separated, and determined the N-terminal sequence, 20 amino
acids, of the and
subunits of this protein purified from
C. durissus terrificus venom and show that this protein
belongs to the C-type lectin family, with sequences most similar to
alboaggregin B. The previously published very short N-terminal
sequences of the protein identified and defined as convulxin in the
venom of Crotalus durissus cascavella, GFRPD for both
subunits (7), are not identical to the sequences we found. Thus, while
it is likely to be the same protein and would appear to have similar
properties, at the moment we cannot exclude that it is a closely
related protein since many snake venoms do contain such families of
proteins, and the relative amounts may vary depending on the place
where the snake comes from and the major prey in that area (43). In the
absence of information to the contrary we shall use the name convulxin
here. Because so far the only C-type lectins that activate platelets do
so via the vWf-GPIb axis, we first of all checked if this was true for
convulxin, but the platelet activation was inhibited neither by
specific anti-GPIb antibodies that inhibit platelet-activation by
vWf-ristocetin nor by echicetin, another C-type lectin from E. carinatus venom demonstrated to bind to GPIb and to inhibit
platelet activation (34).
Earlier results showing that platelet activation by both collagen and
convulxin was antagonized by methylation inhibitors that did not
interfere with other platelet agonists, such as thrombin, ADP, or
Ca2+ ionophore A 23187 (44, 45), and specific crossed
desensitization between convulxin- and collagen-activated platelets
(5), suggested that convulxin and collagen might share common
receptors. The mechanism of platelet activation by collagen is still
controversial. At least three molecules have been implicated as
receptors, GPIa-IIa (the integrin 2
1)
(9-12), CD36 (13), and p62/GPVI (14-16). Accumulated evidence points
to a major role of
2
1 in adhesion, whereas p62/GPVI may be more important in activation, shape change, and
secretion. To find out whether convulxin receptors might be involved,
we prepared purified subunits of convulxin under conditions where we
expected at least some retention of activity and examined their effect
on collagen-induced platelet activation. There has been some
controversy about the functionality of reduced, separated subunits of
snake venom C-type lectins. For example, among those recognizing GPIb,
echicetin
was shown to retain activity and to inhibit thrombin and
vWf activation of platelets (46), whereas the isolated subunits of
GPIb-binding protein were inactive (47). An explanation of these
differences might be that the active preparations were probably only
partly reduced and could, at least partly, refold to an active
conformation, whereas the inactive preparations had been efficiently
reduced and alkylated, so that refolding and disulfide bridge formation
were effectively prevented. Thus it is essential not to block the free
SH groups to allow reformation of disulfide bridges after separation
of the subunits. Partially reduced/denatured samples containing
,
, or both subunits of convulxin did not induce platelet aggregation
or tyrosine phosphorylation (Fig. 7), but they did inhibit platelet
aggregation induced by collagen while having no effect on platelet
aggregation induced by thrombin or vWf-ristocetin, strongly implicating
the collagen receptor(s) in convulxin-induced platelet activation.
However, a mAb to
2
1 that blocks
collagen-induced platelet aggregation had only very weak effects on
convulxin-induced aggregation. A remaining candidate for the convulxin
receptor is p62/GPVI. 125I-Labeled convulxin bound to a
62-kDa membrane glycoprotein and to a similar sized molecule in the
material immunoprecipitated by anti-p62/GPVI antibodies known to
activate platelets by a collagen-like mechanism (48) but not to
material immunoprecipitated by antibodies recognizing other platelet
membrane glycoproteins of similar molecular mass and physicochemical
properties (anti-PTA-1 or anti-GP60 putative laminin receptor
antibodies). Thus, the evidence supports a model where both subunits of
convulxin bind to the p62/GPVI collagen receptor. This heterodimeric
binding to the same protein together with the fact that convulxin
exists as disulfide-linked hexamers,
3
3,
which are able to form even larger structural units up to 150 Å in
diameter (6, 7), immediately suggests an activation mechanism based on
cross-linking and clustering p62/GPVI. It has been suggested many times
that collagen with its repetitive structure and many available binding
sites may also work by such a mechanism. In addition, signal
transduction by convulxin shows many similarities to that induced by
collagen and, as recently shown, by collagen-like peptides (23).
Important molecules identified in collagen-induced signaling include
the
subunit of Fc
RI, p72SYK, PLC
2, and a p36-38
molecule (24). In platelets activated by convulxin these are all
rapidly tyrosine-phosphorylated as well suggesting that, as with
collagen itself, it is the clustering of p62/GPVI that has a major role
in activating these pathways. Studies with collagen, collagen-like
peptides, and WGA have all shown that Fc
and PLC
2 are rapidly
phosphorylated/activated, and it has been suggested that p62/GPVI may
be tightly associated with Fc
and activates Src family kinases by a
mechanism previously described in T- and B-cells (reviewed in Ref. 49)
leading to activation of p72SYK that is critical for
downstream engagement of the enzymes involved in further platelet
stimulation (24). The results described above clearly indicate that
convulxin interacts with platelets predominantly via the p62/GPVI
receptor, whereas collagen uses additional receptors including GPIa/IIa
(
2
1) and possibly CD36. It might
therefore be expected that the signal transduction resulting from
engagement of these receptors would show related differences. Overall
the pattern of tyrosine phosphorylation in platelets activated by
convulxin compared with collagen did show considerable similarity indicating that they do share common pathways. However, there were also
marked differences in the strength and timing of the signals. A problem
in comparing such signals is choosing an appropriate concentration of
reagent to use in each case. In this case we decided to use amounts
that give comparable rates of aggregation as reflected by the slope of
their aggregation curves. The rate of aggregation reflects the final
confluence of signaling pathways in activation of platelets. In the
convulxin-activated platelets Fc
, p36-38, p72SYK, PI3K,
c-Cbl, and PLC
2 show a more rapid and more intense phosphorylation compared with platelets activated with collagen. Logically these components should then be linked to the p62/GPVI receptor pathway. On
the other hand, bands at 125, 32, and 28 kDa were less
tyrosine-phosphorylated than those in the corresponding controls
activated by collagen and should therefore be related to other
receptors such as
2
1. In
collagen-activated platelets the 125-kDa band was shown to be
pp125FAK, and it was also found to be strongly
tyrosine-phosphorylated later in convulxin-activated platelets (Fig.
6). However, since this could be prevented by a specific GPIIb-IIIa
inhibitor, it is clearly related to release of fibrinogen and binding
to activated GPIIb-IIIa in convulxin-activated platelets and not to
binding to
2
1 as with collagen. It was
also of interest to note that when platelet aggregation to collagen was
blocked by convulxin subunits the tyrosine phosphorylation of the
pp125FAK, 32-, and 28-kDa bands was markedly less affected
suggesting that, in this case, the bulk of the signal transduction
occurs via
2
1 but that it was not
adequate to maintain the platelets in an activated state and they
therefore disaggregated. This distinctive use by convulxin of the
p62/GPVI pathway but not the
2
1 is also supported by the lack of effect of
anti-
2
1 antibodies on convulxin-induced platelet aggregation as well as the strong parallels with the effects
of the collagen-like peptides and the anti-p62/GPVI antibodies on
platelets. Activation of platelets with collagen-like peptides that are
also thought not to use the
2
1 receptor
gave a very similar tyrosine phosphorylation pattern to that obtained
with convulxin (23, 24). The collagen-like peptides also gave enhanced and rapid tyrosine phosphorylation of bands in the positions of p36-38, p72SYK, PI3K, c-Cbl, and PLC
2, whereas bands at
125, 32, and 28 kDa were also less tyrosine-phosphorylated than those
in the corresponding controls activated by collagen (23).
Recently, it was demonstrated that the tyrosine kinase inhibitor,
piceatannol, which shows specificity for p72SYK (50), was
an efficient inhibitor of collagen-induced platelet aggregation and,
indeed, inhibited tyrosine phosphorylation of signal transduction
components (42). Surprisingly, with convulxin as agonist it showed some
dramatically different effects causing rather a prolonged
phosphorylation of several components including p36-38. This would
indicate that in convulxin-activated platelets p72SYK has a
major role in activation of tyrosine phosphatases, whereas in
collagen-activated platelets there are alternative pathways for
activation of these phosphatases via other receptors (51) or,
alternatively, that piceatannol can also inhibit directly a phosphatase
activated by the p62/GPVI pathway, whereas the other pathways activate
additional non-inhibitable phosphatases. It might also indicate that
p72SYK is not involved in phosphorylating p36-38 in
convulxin-activated platelets. It was also recently shown that
anti-2
1 antibodies block tyrosine
phosphorylation of both p72SYK and PLC
2 by collagen
(42). On the other hand, cross-linking anti-
2
1 antibodies activated platelets
and increased phosphorylation of both p72SYK and PLC
2.
However, this involved the Fc
RIIA receptor, and F(ab)2 fragments were not stimulatory. There is thus good evidence linking p62/GPVI with Fc
and p72SYK in a common pathway that
resembles that of the T- and B-cell receptors. Both p36-38-like
molecules and c-Cbl have been closely implicated in these latter
receptor mechanisms (52, 53).
The fact that agents such as convulxin or the collagen-like peptides
are able to activate platelets efficiently and in a very similar way to
collagen raises the question why 2
1 and
CD36 are necessary in physiological situations. The main evidence for a
role for these other receptors comes from inhibitory studies with
specific antibodies. The CD36 negative platelets respond relatively
normally to collagen (27-30), and it was never possible to check if
the
2
1-deficient platelets (9, 10) had
normal levels of p62/GPVI nor was signal transduction investigated.
However, since collagen contains p62/GPVI binding sites, supported by
the poor response to collagen of p62/GPVI-deficient platelets, as well
as the inhibitory effect of the convulxin subunits on collagen-induced activation and the characteristics of the platelet response to collagen-like peptides, why are the
2
1
receptors necessary? There are several plausible answers. The
2
1 receptors are certainly necessary for
platelet adhesion (19). In addition, the p62/GPVI-binding sites on
collagen may be cryptic and require the binding of the
2
1 receptor to expose them correctly to
the p62/GPVI receptor. This is supported by evidence that the
2
1 receptor has less critical
requirements for binding collagen than does p62/GPVI, which needs
intact tertiary and quaternary structures (22). Alternatively, since
both receptors appear capable of mediating signals, the activating
potential may depend upon an adequate matrix of platelet binding sites
of both types being present on collagen to ensure that the receptors
are clustered and brought into sufficient proximity for activation
mechanisms to occur. If either receptor (or binding site) is blocked
then an adequate clustering is not attained and activation is
prevented. This would explain why ligands, containing one type of
binding site presented in multimeric form in a more closely packed
arrangement, such as in convulxin or the collagen-like peptides, are
such potent activators of platelets. It will be of considerable
interest to see if collagen-like peptides containing multimeric forms
of the
2
1-binding motif from collagen can
also activate platelets directly or whether this motif is only
effective in conjunction with p62/GPVI- or Fc
RIIA-binding motifs. It
was also recently shown, in one of the patients with p62/GPVI-deficient
platelets, that while collagen did not activate p72SYK it
did activate c-Src kinase normally (54), implying that this occurs via
other receptors, presumably
2
1. The
importance of
2
1 may also be associated
with its primary role in adhesion and with the necessity to modulate
the extremely powerful responses induced via p62/GPVI alone. Damaged
subendothelial tissue probably exposes sites on collagen for both types
of receptor. It is important that platelet activation after binding to
collagen should be limited in extent and should be responsive to
feedback mechanisms, both positive and negative; hence, under
physiological conditions a two-receptor mechanism provides the required
flexibility.
The finding that convulxin binds to and acts predominantly, if not exclusively, via p62/GPVI adds it to anti-p62/GPVI antibodies and collagen-like peptides as p62/GPVI-specific reagents. The unique structure and properties of convulxin offer considerable opportunities for the isolation and characterization, as well as for further exploration of the signaling responses of this important receptor.
We thank Corinne Birbaum for technical assistance. We are grateful to the Central Laboratory of the Swiss Red Cross Blood Transfusion Service, Berne, for the supply of buffy coats.