(Received for publication, April 1, 1997)
From the Department of Biochemistry and Molecular Biophysics, Washington University School of Medicine, St. Louis, Missouri 63110
Integrin-associated protein (IAP or CD47) is a
receptor for the cell/platelet-binding domain (CBD) of thrombospondin-1
(TS1), the most abundant protein of platelet granules. Although it associates with
IIb
3, IAP has no known function in platelets. TS1, the CBD, and an IAP agonist peptide (4N1K) from the CBD of TS1
activate the platelet integrin
IIb
3, resulting in platelet spreading on immobilized fibrinogen, stimulation of platelet
aggregation, and enhanced tyrosine phosphorylation of focal adhesion
kinase. Furthermore, 4N1K peptide selectively stimulates the
phosphorylation of LYN and SYK and their association with FAK. The
phosphorylation of SYK is blocked by pertussis toxin, implicating a
Gi-like heterotrimeric G protein. IAP solublized from
membranes of unstimulated platelets binds specifically to an affinity
column of 4N1K peptide. Both
IIb and
3 integrin subunits and
c-Src bind along with IAP. This complex of proteins is also detected
with immunoprecipitation. Activation of platelets with the agonist
peptide 4N1K results in the association of FAK with the IAP-
IIb
3
complex. Thus an important function of TS1 in platelets is that of a
secreted costimulator of
IIb
3 whose unique properties result in
its localization to the platelet surface and the fibrin clot.
Integrin-associated protein (IAP)1 is
important in host defense where it is required for
integrin-dependent functions of polymorphonuclear leukocytes (1). IAP also appears to be important in modulating integrin
function in other cells (2) and in signal transduction upon ligand
binding by certain integrins with which it associates (3, 4). We have
recently discovered that IAP is a receptor for the COOH-terminal cell
binding domain (CBD) of the thrombospondins (TSs) including TS1 (5),
the most abundant protein of platelet granules (6). A peptide from
the CBD, kRFYVVMWKk (4N1K) has been identified as an IAP agonist (2,
5). TS1 is thought to have a role in augmenting platelet aggregation
(7, 8). A mAb (C6.7) against the IAP-binding domain of TS1 can block
secretion-dependent platelet aggregation (8), but the
mechanism of this effect has remained obscure (9). IAP is present on
platelets (10), but it was initially reported to have no functional
role in platelet activation or aggregation (11). However, Dorahy
et al. (12) have recently reported that the 4N1K agonist
peptide, which we had identified in the CBD of TS1, can activate washed
platelets causing their aggregation. In nucleated cells, IAP associates with
v
3 and modulates its function (3, 4). For example, the CBD
of TS1 and the 4N1K peptide stimulate the chemotaxis of endothelial
cells on RGD-containing substrata, and this effect is blocked
specifically by mAbs against IAP and
v
3 (5). TS1, its CBD, and
4N1K peptide all stimulate the rapid spreading of C32 melanoma and
NIH3T3 cells on sparse vitronectin substrata, which support only weak,
slow spreading of these cells in the absence of TS1 (2). This
stimulation of
v
3-dependent spreading is specifically
inhibited by pertussis toxin, indicating the participation of a
heterotrimeric Gi-like protein in a pathway linking IAP to a common cellular pathway resulting in protein kinase C activation, which leads to cell spreading (2) and motility (5). This sort of
stimulation of an integrin-dependent function via G
protein-dependent pathways is reminiscent of the
costimulation of
IIb
3 function in platelets by agents that act
via heptahelical or seven transmembrane spanning receptors such as ADP,
epinephrine, and thrombin (13, 14). Here we have examined the
hypothesis that IAP ligation by TS1 has a role in modulating
IIb
3
function. We find that IAP stimulation by its agonist 4N1K activates
IIb
3 as judged by enhanced binding of the conformationally
sensitive mAb PAC-1 (13-15) resulting in spreading of platelets on
fibrinogen-coated surfaces, aggregation of stirred platelets (12), and
assembly of a signaling complex containing IAP, the integrin,
c-Src, FAK, and SYK. All of these actions of IAP are blocked by
pertussis toxin, indicating the essential participation of a
Gi-like heterotrimeric G protein (2). In unstirred, washed
platelets where the integrin is not engaged, 4N1K stimulates the
tyrosine phosphorylation of SYK, an early event in platelet activation
by many agonists (17). This activation of SYK is blocked by pertussis
toxin but not by inhibitors of downstream signaling events. These
results provide a novel explanation for the role of TS1 in platelet
function and establish IAP as signaling coreceptor for platelet
stimulation.
Apyrase, bovine serum albumin, cytochalasin D, human fibrinogen,
indomethacin, and prostaglandin E1 were obtained from Sigma. Epinephrine,
D-phenylalanyl-L-prolyl-L-arginine
chloromethyl ketone, calphostin C, herbimycin A, and wortmannin were
from Calbiochem. Pertussis toxin was from Life Technologies, Inc. BAPTA
was from Molecular Probes (Eugene, OR). Monoclonal antibody PAC1 was a gift from Dr. Sanford J. Shattil. Anti-human IAP mAbs, B6H12, and 2D3
were kindly provided by Dr. Eric J Brown. mAb against c-Src was a
generous gift of Dr. D. Lublin. mAbs against FAK (clone 77 for Western
blotting) and phosphotyrosine (PY20) were products of Transduction
Laboratory (Lexington, KY). mAbs against FAK (2A7 for
immunoprecipitation), and phosphotyrosine (4G10) and anti-SYK polyclonal IgG were from Upstate Biotechnology Inc. Anti-LYN polyclonal IgG was from Santa Cruz Biotechnology, Inc (Santa Cruz, CA). Rabbit anti-human integrin 3 polyclonal antibody and mouse anti-integrin
IIb mAb were purchased from Novous Molecular, Inc. (San Diego, CA).
All peptides used were synthesized and purified by the Protein and
Nucleic Acid Chemistry Laboratory of Washington University as described
previously (2). TS1 and its cell binding domain were prepared as
described (2).
Blood was obtained from healthy human donors in 0.15 volume of acid-citrate/dextrose solution supplemented with 1 µM prostaglandin E1 and 1 unit/ml apyrase (14). Platelet-rich plasma was separated from whole blood at 1000 rpm in a Beckman GP table top centrifuge for 30 min and further centrifuged at 2000 rpm for 30 min to pellet the platelets. Pelleted platelets were washed once and resuspended in incubation buffer (137 mM NaCl, 2.7 mM KCl, 1 mM MgCl2, 5.6 mM glucose, 1 mg/ml bovine serum albumin, 3.3 mM NaH2PO4, and 20 mM Hepes, pH 7.4) at concentration of 2 × 108 platelets/ml. Apyrase was then added to a final concentration of 10 units/ml to remove ADP. All other inhibitors were preincubated for 15 min except for pertussis toxin (preincubated for 2 h). Platelet adhesion to fibrinogen was studied on Lab-Tek 8 chamber slides (VWR Scientific), which were precoated with human fibrinogen (100 µg/ml, Sigma) overnight at 4 °C and blocked with 10 mg/ml bovine serum albumin. After 60 min at room temperature, adherent platelets were fixed with paraformaldehyde, permeabilized with 0.1% Triton X-100, and stained with rhodamine-phalloidin to determine the degree of platelet spreading on fibrinogen.
To detect phosphorylation events at various time points, adherent
platelets were lysed in RIPA buffer and cell lysates were subjected to
immunoprecipitation and immunoblot analysis as described (2). The
extent of activation of IIb
3 was determined with PAC-1 mAb
binding using a FACStar flow cytometer as described (14). Aggregation
of freshly collected, washed platelets were performed in a Chrono-Log
aggregometer (14).
For affinity chromatography of IAP protein complexes, Ni-NTA-agarose beads (0.25 ml, Qiagen) were charged with His-tagged 4N1K (His6-KRFYVVMWKK) or His-tagged 4NGG (His6-KRFYGGMWKK) peptide (1 µmol) in Hepes-buffered saline (HBS), and washed extensively in HBS. Washed platelets were lysed with 1% Triton X-100 in HBS containing protease and phosphatase inhibitors (2). Lysates were centrifuged at 13,000 rpm in a refrigerated microcentrifuge for 30 min. The soluble fraction was mixed with the peptide-charged agarose beads for 3 h at 4 °C. The beads were rinsed into a Quik Snap column (0.6 × 6 cm, MidWest Scientific, St. Louis, MO) and washed extensively with 1% Triton X-100 in HBS (20 ml), followed by 20 mM imidazole (20 ml) and 100 mM imidazole (20 ml) in the same buffer. Proteins bound via the His tag were eluted with buffer containing 1 M imidazole. 100-µl fractions were collected, 5 × SDS sample buffer containing 2-mercaptoethanol was added, and fractions were immediately boiled and subjected to SDS gel electrophoresis and immunoblot analysis with the indicated antibodies.
We have employed a system used by Shattil and co-workers (14, 15)
to assess IIb
3 activation in which washed platelets are allowed
to adhere to a fibrinogen-coated surface in the presence of apyrase, an
ADP scavenger. In the absence of a costimulator, the platelets adhere
but cannot spread (Fig. 1A). Epinephrine (Fig. 1C), thrombin receptor peptide (Fig. 1D),
and ADP itself bind to heptahelical receptors that are coupled via G
proteins to the generation of lipid intermediates including
arachidonate, prostaglandins, leukotrienes, and diacylglycerols (13).
These ultimately activate protein kinase C, which leads in an as yet unknown way to activation of the integrin resulting in the spreading of
the platelets on fibrinogen (Fig. 1, C and D)
(13, 14). With this assay, we find that TS1, the recombinant CBD, and
active IAP binding peptides such as 4N1K (Fig. 1B)
specifically stimulate the spreading of platelets adherent to
fibrinogen. At its maximum, the stimulation by 4N1K peptide is equal to
or greater than that achieved with either epinephrine or thrombin
receptor peptide (Fig. 2A). The IAP
dependence of the stimulation by 4N1K is demonstrated by the inhibition
of the effect by the function blocking anti-IAP mAb B6H12 but not by
mAb 2D3, which binds IAP but fails to block function (16). The same
results were obtained with F(ab)
2 fragments of the mAbs
(not shown). That the stimulation by 4N1K might be due to generation of
active thrombin was ruled out by the finding that sufficient
D-phenylalanyl-L-prolyl-L-arginine
chloromethyl ketone to inhibit activation by 1 unit/ml active thrombin
had no effect on the action of 4N1K (not shown).
A, specificity of the stimulation of
spreading by 4N1K. Platelet spreading on fibrinogen was performed as
described in the legend to Fig. 1. After fixing and staining with
rhodamine-phalloidin, five randomly selected fields were counted for
total platelets (more than 100) and spread platelets. 4N1K was present
as indicated at 50 µM and the antibodies B6H12, 2D3, and
mouse IgG at 100 µg/ml. Epi, 10 µM
epinephrine; TRP, 5 µM thrombin receptor
peptide, SFLLRN. Apyrase was present in all samples at 10 U/ml.
B, effect of inhibitors on platelet spreading stimulated by
4N1K. Stimulation with 4N1K in all cases as in A with
apyrase as a scavenger (, apyrase alone). Treatments were pertussis
toxin for 2 h at room temperature (PT, 100 ng/ml, the
B oligomer caused no inhibition under these conditions, not shown);
herbimycin A (HA, 27 µM); indomethacin
(Indo, 10 µM); NDGA (20 µM);
cytochalasin D (cyto D, 10 µM); wortmannin (10 nM); calphostin C (Calp. C, 100 nM);
BAPTA (10 µM). Details of incubations as in Ref. 2.
C, effect of platelet costimulators on
IIb
3 activation
determined by PAC-1 binding.
IIb
3 activation was measured by mAb
PAC-1. Bound PAC-1 was determined by FACScan analysis using a
fluorescent secondary antibody as in Ref. 14. The mean fluorescence
intensity (arbitrary units) of the brightest 85% of the counted
platelets is graphed here. (
), 10 units/ml apyrase (in all samples)
present alone; Epi, 10 µM epinephrine;
TRP, 5 µM thrombin receptor peptide. 4N1K
(KRFYVVMWKK), 4N7G (KRFYVVMGKK), and 4NGG (KRFYGGMWKK) were all at
50 µM.
Fig. 2B summarizes the effects on 4N1K-stimulated platelet
spreading of a number of inhibitors previously shown to inhibit IIb
3-dependent spreading of platelets and/or
IIb
3 activation (13, 14). As we had found for the TS and 4N1K
stimulation of C32 cells on low density VN surfaces (2), pertussis
toxin, but not the carbohydrate binding B oligomer of the toxin, blocks more than 65% of the 4N1K-stimulated platelet spreading on fibrinogen with only a 90-min preincubation. As noted by others (17), the tyrosine
kinase inhibitor herbimycin A effectively blocks spreading, reflecting
the several steps at which tyrosine kinase activation has been
implicated (13, 18). Interestingly, indomethacin (a cyclooxygenase
inhibitor) and nordihydroguaretic acid (NGDA, a lipoxygenase inhibitor)
each have a partial effect alone (Figs. 1, E and
F, and 2B), but together totally obliterate
spreading (Figs. 1G and 2B). This indicates that
both pathways are activated by 4N1K and that each contributes to the
effect. This sort of bifurcation can presumably add redundancy to the
response. As noted by others (14), cytochalasin D effectively blocks
spreading (Figs. 1G and 2B) by preventing actin
filament polymerization. Inhibition of PI 3-kinase by wortmannin
partially blocks the response to 4N1K, whereas the protein kinase C
inhibitor calphostin c and the calcium chelator BAPTA are quite
effective as noted by Shattil et al. (14) for epinephrine
costimulation. Thus the properties of the CBD peptide stimulated
spreading on fibrinogen mirror those of classical costimulators of
IIb
3 function such as ADP, epinephrine and thrombin (13).
Comparable data were obtained for the recombinant CBD of TS1 (not
shown).
As an independent index of IIb
3 activation, we tested the ability
of 4N1K to promote the binding of mAb PAC1 to its
activation-dependent epitope in the ligand binding site of
IIb
3 (14). As seen in Fig. 2C, 4N1K, at maximum, was
able to stimulate a level of PAC1 binding greater than that of either
epinephrine or thrombin receptor peptide. The specificity of the effect
is demonstrated by the partial activity of peptide 4N7G (KRFYVVMGKK)
and the complete lack of activity of peptide 4NGG (KRFYGGMWKK). In
other IAP-dependent assays such as chemotaxis and cell
spreading, 4N7G is partially active, whereas 4NGG is completely
inactive (2). Further, the development of PAC1 binding in the presence
of 4N1K was blocked by either calphostin C or indomethacin (not shown),
indicating that it resulted from stimulation of intracellular signaling
pathways and was not mediated by an extracellular effect of 4N1K. The
classical costimulators of
IIb
3 also initiate platelet
aggregation, and we have previously shown that mAb C6.7 directed
against the TS1 CBD inhibits secretion-dependent platelet
aggregation (8). Therefore, we tested 4N1K for its ability to cause
aggregation of washed platelets in an aggregometer assay. Fig.
3 (A and B) show that the maximal
response of the platelets to 4N1K was equal to that for thrombin
receptor peptide and greater than that for epinephrine, whereas the
inactive control peptide 4NGG was without effect. Walz et
al. (9) and Tuszynski et al. (19) have previously reported that whole TS1 can augment the aggregation of gel-filtered platelets. The inhibition of platelet aggregation by antibodies against
the NH2-terminal heparin binding domain of TS1 (20) suggests that binding of TS1 to negatively charged glycans helps to
tether the protein to the platelet surface.
We next examined an intracellular event related to the activation of
platelet spreading and aggregation: phosphorylation of proteins on
tyrosine residues (15, 21). Preliminary experiments with platelets
spreading on fibrinogen-coated surfaces revealed that 4N1K could
greatly enhance the rate and extent of tyrosine phosphorylation of
several proteins previously identified as early targets of tyrosine
kinases in activated platelets (15, 18, 21). Fig.
4A shows the time course of tyrosine
phosphorylation of FAK in platelets attached to fibrinogen in the
presence of apyrase with (left) and without
(right, mirror image) 4N1K. At the earliest times of 1 to 2 min of 4N1K treatment, FAK contains significantly elevated levels of
phosphotyrosine that increase throughout the time course. In these
experiments we noted the early appearance of a tyrosine phosphorylated
protein of molecular mass of approximately 70 kDa whose phosphorylation
and association with FAK was much more prominent in 4N1K-treated
platelets than in control or epinephrine-stimulated platelets. Fig.
4B shows that this protein is the 72-kDa tandem SH2 domain
kinase SYK, whose phosphorylation on tyrosine has been identified as a
very early event in platelet activation (13, 17). 4N1K treatment results in a significant stimulation of SYK association with FAK compared with epinephrine (lane 4). As seen in Fig.
4B, the association of SYK with FAK is selectively augmented
by 4N1K compared with epinephrine. The association of LYN with FAK
(Fig. 4C) is also selectively stimulated with a time course
similar to that of SYK. In B cells, LYN has been found to phosphorylate
sites in ITAM-containing cytoplasmic domains of receptors thus creating
docking sites for SYK (22). It is interesting that FAK phosphorylation
increases until 60 min, whereas SYK and LYN association with FAK is
highest at 30 min and decreases by 60 min of spreading.
To determine the events connecting 4N1K-IAP binding with SYK
phosphorylation, suspended, unstirred washed platelets were stimulated with 4N1K. These conditions prevent platelet aggregation and hence fibrinogen binding by IIb
3, which leads to outside in signaling and the initiation of a cascade of downstream events including a
massive wave of tyrosine phosphorylation and cytoskeletal
reorganization (13). In these unstirred platelets the 4N1K-stimulated
tyrosine phosphorylation of SYK is maximal at 3-5 min and declines
thereafter (Fig. 4D). At later times, when slow platelet
aggregation begins to occur despite the lack of stirring, SYK
phosphorylation again increases as previously reported (Ref. 17 and
data not shown). Again, the control peptide 4NGG is inactive. These
results indicate that platelet aggregation and signaling downstream of
IIb
3 ligation are not required for SYK activation. Using this
same protocol we tested the effects of a number of inhibitors on the
ability of 4N1K to stimulate tyrosine phosphorylation of SYK (Fig.
4E). Interestingly, only pertussis toxin strongly inhibited
SYK phosphorylation at 5 min in response to 4N1K. Cytochalasin D and
prostaglandin E1, which elevates cAMP, gave partial and
variable inhibition, whereas other inhibitors did not reduce SYK
phosphorylation (three experiments gave comparable results).
Significantly, indomethacin and NDGA had no effect, nor did calphostin
c or BAPTA, protein kinase C inhibitors. Because BAPTA is a
Ca+2 chelator, it seems unlikely that the effect of 4N1K on
SYK phosphorylation is mediated by Ca+2 flux, a known
action of IAP in other cell types (23, 24). These data indicate that
SYK phosphorylation occurs quite proximally to IAP activation and
requires a heterotrimeric Gi-like protein but not other
downstream effectors of platelet activation. Similar results in terms
of SYK phosphorylation have been obtained upon stimulation of unstirred
platelets via activation of thrombin, collagen, and ADP receptors
(25-27). The robust stimulation of SYK phosphorylation perhaps
indicates a special role for TS1 in activating SYK kinase. The
mechanism of this effect is under investigation.
To gain insight on the proteins that might initiate IAP signaling, we
have used an affinity column strategy to ask which proteins, if any,
are in a preformed complex with IAP in resting platelets. To do this,
unstimulated platelets were dissolved in a buffer containing Triton
X-100 (1% v/v), and the soluble extract was applied to Ni-NTA agarose
beads, which had been charged with 4N1K peptide containing a
hexahistidine tag at its amino terminus. The control was His-tagged
4NGG peptide charged beads. After washing with 0.2 M
imidazole, the IAP and other specifically bound proteins were eluted
with 1 M imidazole. Because only hexahistidine-tagged proteins or peptides remain bound to Ni-NTA matrix after washing with
0.2 M imidazole, it is likely that proteins that elute with 1 M imidazole are bound via interactions with IAP that
binds 4N1K. As seen in Fig. 5A, no IAP is
eluted from the 4NGG column (fractions 1 through 7 are shown), whereas
IAP is readily detectable in fractions eluted from the 4N1k column.
Both integrin subunits, IIb and
3 coeluted with IAP, confirming
that under these rather stringent conditions in the presence of Triton
X-100, IAP is complexed with the integrin heterodimer. Western blots of
these fractions were probed with antibodies against tyrosine kinases
involved in platelet signaling. Interestingly, only a mAb specific for
c-Src versus other Src-type kinases gave a strong signal (Fig.
5A) while FYN, LYN, and YES were not detected in the eluate
(not shown). FAK was marginally detectable (Fig. 5A) and SYK
was not detected in the column eluate even on prolonged exposures (not
shown). These results suggest that the integrin and IAP are in a
complex with c-Src, and perhaps FAK but not SYK in resting
platelets. However, the data in Fig. 4A indicate that FAK
becomes phosphorylated in response to 4N1K stimulation. To determine if
FAK could become associated with the integrin-IAP complex after 4N1K
stimulation, suspended, unstirred platelets were stimulated with 4N1K
as above, and at various times samples were lysed in Triton X-100
buffer and immunoprecipitated with an anti-
3 mAb, run on
SDS-polyacrylamide gel electrophoresis, and blotted. In agreement with
the results in Fig. 5A, lysates from platelets treated with
the control peptide 4NGG (at 5 min) displayed relatively little
association of FAK with the integrin-IAP complex (Fig. 5B).
However, as early as 1 min after the addition of 4N1K, a significant
increase in the amount of FAK associated with the complex was detected
(Fig. 5B). IAP and c-Src were also detected in the
3
immunoprecipitate when platelets were treated with either 4NGG or 4N1K
(not shown) in further agreement with the results in Fig.
5A. The same increase in FAK association with the
integrin-IAP complex was also detected in anti-IAP immunoprecipitates
(not shown).
These data provide a new paradigm for the action of TS1 at the
site of its greatest physiological concentration, the platelet. Upon
platelet activation, TS1 is secreted and binds to the platelet surface
via a number of receptors (9, 28-31). These include sulfatides,
which bind to the NH2-terminal heparin binding domain (32),
platelet gpIV, or CD36, which binds to a region within the type 1 repeats in the central stalk or "core" of the TS1 subunit, and
v
3 and
IIb
3, both of which have been shown to
bind TS1 in an RGD-dependent manner (29, 33). The binding
of TS1 to resting platelets does not require Ca2+ ions,
whereas binding to the receptors exposed on activated platelets is
blocked by EDTA (9, 12, 20, 30). We have reported that binding of cells
to the CBD and 4N1K peptide and affinity labeling of IAP by 4N1K are
not sensitive to EDTA (2, 5), suggesting, as noted by Dorahy et
al. (12), that IAP is the primary TS1 receptor on resting
platelets. Our present data indicate that the CBD of TS1 can bind to
platelet IAP and that the peptide 4N1K is a specific agonist for IAP
signaling. Because the RGD sequence of TS1 is just amino-terminal to
the CBD, the binding of TS1 to IAP may facilitate the exposure of the
RGD site and its binding to either
v
3 or
IIb
3
associated with IAP. This could have consequences for modulation of
fibrinogen binding by
IIb
3 and may allow TS1 to participate
directly in platelet-platelet cross-linking. This is likely to be the
case in washed platelets where TS1 is much more abundant than
fibrinogen after
-granule discharge. Whether or not the same TS1
molecule engages IAP and associated
IIb
3, the surface bound TS1
would appear to serve the function of a tethered costimulator of
IIb
3 activation and hence of ligand binding and platelet
cross-linking. The previous model for TS1 action in platelets
emphasized the role of TS1 as a cross-linker of fibrinogen bridges
between platelets (7, 34). Our data indicate that TS1 can strengthen
ligand binding to
IIb
3 and hence platelet cross-linking, by a
direct effect on the classical intraplatelet costimulatory signaling
cascade resulting in an enhancement of the affinity/avidity of
IIb
3 (13).
The experiments reported here are in fact the first report of a
detergent-stable complex containing the integrin IIb
3 and IAP
obtained from platelets. An early report showed coimmunoprecipitation of IAP with an anti-
3 mAb (3), but because platelets contain
v
3, this does not prove association with
IIb
3. Dorahy
et al. (12) used "disrupted platelet membranes" obtained
by sonicating platelets that were then passed over a 4N1K affinity
column. IAP and other proteins including
3 subunits were eluted with
RFYVVM peptide. Because no detergent was employed, these data cannot argue for complex formation. We have shown by both affinity
chromatography and immunoprecipitation of detergent solubilized
platelet membranes that
IIb
3 and IAP do indeed exist in a
complex. Because there are approximately equal numbers of
IIb
3
and IAP molecules in the platelet membrane (3), each IAP may be
associated with an
IIb
3 molecule. However, because IAP appears to
activate
IIb
3 via a signaling pathway, direct physical
association of each
IIb
3 with an IAP may not be required. Further
we find that c-Src is also resident in this complex with
IIb
3 and IAP. At least five members of the Src-type kinase
family are expressed in megakaryocytes and platelets (36), but only
c-Src could be detected in the
IIb
3-IAP complex. Even though
c-Src is present in platelets greater abundance than any other
family member, a clear functional role for c-Src in platelets has
not been identified. Thus we suggest that at least one role of
c-Src in platelets is to initiate the IAP-dependent
tyrosine phosphorylation of SYK. 4N1K stimulates the tyrosine
phosphorylation of SYK in unstirred platelets (Fig. 4D), and
this is inhibited by pertussis toxin and to a lesser extent by
cytochalasin D but not by other inhibitors of platelet activation (Fig.
4E). This is precisely the inhibitor profile found for
thrombin stimulation of SYK phosphorylation by Clark et al.
(26), who reported that c-Src and SYK are phosphorylated in a
first wave of tyrosine phosphorylation that does not require integrin
engagement or platelet aggregation. There is ample precedent for the
stimulation of Src-like tyrosine kinases by receptors that act
through heterotrimeric G proteins including Gi (37-40); however, the precise mechanism(s) linking G protein subunits to activation of Src-type kinases is unknown (41). For example, it
has recently been reported that Gi derived G
activates c-Src-dependent tyrosine phosphoryation of
the epidermal growth factor receptor (41). These authors propose that
PI 3-kinase may mediate the activation of c-Src, and in fact,
G
activation of PI 3-kinase in platelets can occur (41, 42).
However, the IAP-dependent stimulation of SYK
phosphorylation that we observe upon treatment of platelets with 4N1K
is not sensitive to inhibition by wortmannin (Fig. 4E), a
property of all three PI 3-kinase isoforms (43). The mechanism of
c-Src activation by IAP is currently under investigation.
As seen in Fig. 5 (A and B), relatively small
amounts of FAK are associated with the integrin-IAP complex extracted
from resting platelets, but FAK association with this complex is
rapidly increased upon agonist activation of IAP. Furthermore, SYK
associates with FAK with equal rapidity (Fig.
4B),2 and this association is
blocked by pertussis toxin (not shown). Taken together our data suggest
that ligation of IAP with the agonist peptide 4N1K activates a
heterotrimeric pertussis toxin-sensitive Gi protein.
Through an as yet unknown mechanism, which does not appear to involve
PI 3-kinase activation, the G protein may activate c-Src, which in
turn could phosphorylate SYK and FAK. The SYK homolog ZAP-70 functions
in T cell activation where it is initially activated by binding to ITAM
motifs that are phosphorylated on tandem tyrosines by the Src-type
kinase, FYN (44). The resultant clustering of ZAP-70 molecules
presumably allows intermolecular tyrosine phosporylation to occur. The
phosphorylated and activated ZAP-70 can then phosphorylate a number of
downstream targets leading to T cell proliferation and other responses.
A similar ITAM-dependent mechanism has been proposed for
SYK activation in B cells (45). SYK activation in platelets has
remained unexplained. Recent reports (46, 47) indicate that the
cytoplasmic domain of the integrin 3 subunit can be tyrosine
phosphorylated, but it is not clear that the two tyrosines in that
sequence can function as an ITAM motif. Although the spacing between
the tyrosines is precisely that of a cannonical ITAM, the residues in
the Y+3 positions are not well conserved (44). It has been reported
that cross-linking of platelet Fc receptors, Fc
RII, leads to
phosphorylation of SYK, even in the absence of
IIb
3 (48).
Tyrosine residues in the cytoplasmic domains of Fc
RII chains occur
in sequence contexts (YXXL) reminiscent of ITAMs but the
spacing of the tyrosines is somewhat different (44). The possibility
remains that a tyrosine kinase binds via a single SH2 domain to one of
these phosphorylated sites in the integrin or the Fc receptor tail,
which could then phosphorylate an ITAM-like motif in another protein,
perhaps FAK itself.
In summary, we have identified TS1 as the natural ligand for IAP and
4N1K as a peptide agonist (2, 5). It is now possible to isolate the IAP
costimulatory pathway and study its function. The role for TS1
suggested by our experiments is that of an autocrine or juxtacrine
costimulator of IIb
3 function whose ability to bind to the
platelet surface (9, 28, 30) and fibrin clot (35) differentiates TS1
from small molecule and other protein costimulators. The robust effect
of 4N1K on LYN and SYK phosphorylation and FAK association (Fig. 4)
suggests that this may be a unique consequence of TS1-IAP mediated
signaling. Finally the modulatory effect of TS1/IAP on
IIb
3
function presents a novel and potentially efficacious point at which to
"tune down" the function of
IIb
3 without totally obliterating
it, thus perhaps controlling acute thrombotic events without
precipitating dangerous bleeding episodes.
We thank Drs. Fred Lindberg and Eric Brown for many insightful discussions and anti-IAP mAbs; Dr. Sam Santoro for help with aggregometry; Dr. Sandy Shattil for mAb PAC1 and helpful advice; Tom Broekelmann for help with fluorescence microscopy; and the Protein and Nucleic Acid Chemistry Laboratory of Washington University for peptide synthesis, purification, and characterization. Julie Dimitry prepared recombinant CBD as in Ref. 2, and Mary Beth Finn obtained platelets and prepared platelet thrombospondin as described previously (8). We thank Anna Goffinet for preparation of the manuscript.