Content uploaded by Siegfried Ussar
Author content
All content in this area was uploaded by Siegfried Ussar on Nov 22, 2014
Content may be subject to copyright.
Kindlin-3 is essential for integrin activation and
platelet aggregation
Markus Moser
1
, Bernhard Nieswandt
2,3
, Siegfried Ussar
1
, Miroslava Pozgajova
2
& Reinhard Fa
¨ssler
1
Integrin-mediated platelet adhesion and aggregation are
essential for sealing injured blood vessels and preventing
blood loss, and excessive platelet aggregation can initiate
arterial thrombosis, causing heart attacks and stroke1.To
ensure that platelets aggregate only at injury sites, integrins
on circulating platelets exist in a low-affinity state and shift
to a high-affinity state (in a process known as integrin
activation or priming) after contacting a wounded vessel2.
The shift is mediated through binding of the cytoskeletal
protein Talin to the bsubunit cytoplasmic tail3–5.Herewe
show that platelets lacking the adhesion plaque protein
Kindlin-3 cannot activate integrins despite normal
Talin expression. As a direct consequence, Kindlin-3 deficiency
results in severe bleeding and resistance to arterial thrombosis.
Mechanistically, Kindlin-3 can directly bind to regions of
b-integrin tails distinct from those of Talin and trigger integrin
activation. We have therefore identified Kindlin-3 as a novel
and essential element for platelet integrin activation in
hemostasis and thrombosis.
Cellular control of integrin activation is essential for virtually all cells,
including platelets, which seal injured blood vessels and stop bleeding.
At sites of injury, the platelet receptors GPIb and GPVI bind to von
Willebrand factor (vWF) and collagen, respectively1–3, which together
with locally produced thrombin trigger activation of integrin a
IIb
b
3
and the release of soluble platelet agonists including ADP and
thromboxane A2 (TxA2). Activated a
IIb
b
3
integrins bind fibrinogen,
vWF and fibronectin, thus allowing firm platelet adhesion and platelet
aggregation. The central role of integrin activation in platelet adhesion
and aggregation sparked the search for critical integrin tail-binding
proteins that control integrin affinity for ligands. Irrespective of the
platelet-activating stimulus and signaling pathways, Talin binding to
the b-integrin tails was shown to be the final common step in a
IIb
b
3
integrin activation and ligand binding4,6,7. Talin, a major cytoskeletal
protein at integrin adhesion sites, consists of a large C-terminal rod-
like domain and an N-terminal FERM (protein 4.1, ezrin, radixin,
moesin) domain with three subdomains: F1, F2 and F3 (ref. 8). The
phosphotyrosine-binding (PTB) subdomain in the F3 domain sequen-
tially binds to two distinct regions in the bcytoplasmic tails and is
sufficient for integrin activation in vitro8,9.
In addition to Talin, other FERM domain–containing proteins,
including the Kindlins, interact with integrin btails10. The Kindlin
protein family consist of three members (Kindlin-1, Kindlin-2 and
Kindlin-3), all of which localize to integrin adhesion sites11–13.In
contrast to the widely expressed Kindlin-1 and Kindlin-2, Kindlin-3 is
restricted to hematopoietic cells and is particularly abundant in
megakaryocytes and platelets12. The structural hallmark of Kindlins
is a FERM domain whose F2 subdomain is split by a pleckstrin
homology (PH) domain. In a comparison of FERM-domain proteins,
the F3 subdomains of Kindlins have been found to share highest
homology with the F3 domain of Talin10.
To address the function of Kindlin-3 in vivo, we disrupted the Kind3
(also called Fermt3)geneinmice(Supplementary Fig. 1 online). Mice
heterozygous for the Kindlin-3–null mutation (Kind3
+/–
)werenor-
mal, whereas mice lacking Kindlin-3 (Kind3
/
;Fig. 1a) died within a
week of birth and showed a pronounced osteopetrosis (unpublished
data) and severe hemorrhages in the gastrointestinal tract, skin, brain
and bladder, which were already apparent during development
(Fig. 1b and data not shown). To test whether the severe bleeding
of Kindlin-3–deficient mice was due to impaired platelet production
and/or function, we generated fetal liver cell chimeras by transferring
either Kind3
/
or wild-type fetal liver cells into lethally irradiated
wild-type recipient mice. Tail-bleed assays revealed that Kind3
/
chimeras suffer from a pronounced hemostatic defect like that of
Kindlin-3–deficient mice. After the tail-tip cut, bleeding in control
mice arrested within 10 min (mean of 5.4 ± 4.3 min), whereas
Kind3
/
chimeras bled for longer than 15 min, suggesting severe
platelet dysfunction (Fig. 1c).
Kind3
/
chimeras showed platelet counts similar to those of wild-
type chimeras (Fig. 1d), ruling out an essential role for Kindlin-3 in
platelet formation. Analysis of glycoprotein abundance on platelets
revealed elevated levels of the vWF receptor complex GPIb-IX in
Kindlin-3–deficient as compared to wild-type platelets, whereas levels
of other glycoproteins, including GPVI, CD9, and b
1
and b
3
integrins,
were reduced (Supplementary Table 1 online). Thus Kindlin-3 has an
apparent yet undefined role in the expression of several glycoproteins.
However, the reduced expression of integrin a
IIb
b
3
does not account
for the hemostasis defect, as mice carrying a heterozygous null
mutation in the b
3
integrin express even less a
IIb
b
3
integrin on their
platelets (50% of wild-type) without developing a bleeding defect14.
Received 11 October 2007; accepted 4 January 2008; published online 17 February 2008; doi:10.1038/nm1722
1
Department of Molecular Medicine, Max Planck Institute of Biochemistry, Am Klopferspitz 18, 82152 Martinsried, Germany.
2
University of Wu
¨rzburg, Rudolf Virchow
Center, Deutsche Forschungsgemeinschaft Research Center for Experimental Biomedicine, Zinklesweg 10, 97080 Wu
¨rzburg, Germany.
3
Institute of Clinical
Biochemistry and Pathobiochemisty, Josef-Schneider-Str. 2, 97078 Wu
¨rzburg, Germany. Correspondence should be addressed to R.F. (faessler@biochem.mpg.de).
NATURE MEDICINE VOLUME 14
[
NUMBER 3
[
MARCH 2008 325
LETTERS
©2008 Nature Publishing Group http://www.nature.com/naturemedicine
To determine the mechanism of the platelet defect, we performed
ex vivo platelet aggregation studies. Wild-type platelets aggregated in
response to ADP, the TxA2 analog U46619, thrombin, collagen and
the GPVI-activating collagen-related peptide (CRP), whereas none of
the agonists induced aggregation of Kind3
/
platelets (Fig. 2a).
Notably, all agonists induced a comparable activation-dependent
change from discoid to spherical shape in control and Kind3
/
platelets, which can be seen in aggregometry as a short decrease in
light transmission following the addition of agonists. This suggests a
selective defect in a
IIb
b
3
-dependent aggregation rather than a general
impairment of signaling pathways in Kind3
/
platelets.
To test whether activation of a
IIb
b
3
integrin is indeed abrogated in
Kind3
/
platelets, we determined their ability to bind fibrinogen
using flow cytometry. Wild-type platelets showed robust fibrinogen
binding in response to ADP, ADP plus U46619 and CRP, whereas
Kind3
/
platelets failed to bind fibrinogen upon agonist treatment
(Fig. 2b). The antibody JON/A-PE, which specifically detects the
activated form of mouse a
IIb
b
3
integrin14, likewise did not bind
stimulated Kind3
/
platelets (Fig. 2c). When cellular activation of
a
IIb
b
3
integrin was bypassed by the addition of MnCl
2
, comparable
fibrinogen binding to Kind3
/
and wild-type platelets occurred
(Fig. 2b). Together, these findings indicate that loss of Kindlin-3
expression prevents energy-dependent conformational rearrangements
required for integrin-a
IIb
b
3
activation.
Resting platelets store P-selectin in a-granules, which fuse with the
plasma membrane during agonist-induced platelet activation1. Both
CRP and thrombin induced P-selectin translocation on wild-type and
Kind3
/
platelets, although a significant reduction was consistently
observed at intermediate concentrations of thrombin (Po0.001;
Fig. 2d). As expected, the weak agonist ADP did not induce P-selectin
surface expression in wild-type and Kind3
/
platelets. Thus, although
loss of Kindlin-3 specifically disables integrin activation, it still permits
agonist-induced P-selectin translocation.
Integrin-a
IIb
b
3
is also involved in adhesion to immobilized ligands,
including collagen-bound vWF, where it acts in concert with the
collagen-binding a
2
b
1
integrin2,15. We analyzed the ability of Kind3
/
platelets to interact with fibrous collagen in a whole-blood perfusion
Spleen
Kindlin-3
Kind3+/+
Kind3+/–
Kind3–/–
Kind3+/+
Kind3–/–
GAPDH
Platelets
cd
ab
> 900
900
600
Kind3+/+ Kind3 –/–
Kind3+/+
Kind3–/–
Tail bleeding times (s)
300
0
1,000
Platelets/µl (10–6)
800
600
400
200
0
ADP
5 µM
Light transmission
U46619
1 µM
Thrombin
1 U/ml
CRP
5 µg/ml
Collagen
3 µg/ml
Time
Kind3+/+
Kind3+/+
Kind3–/–
Kind3–/–
Kind3+/+
Kind3+/+
Kind3–/–
Kind3–/–
10
Rest. ADP
(µM)
CRP
(µg/ml)
Thrombin
(U/ml)
9 min 25 min
10 0.1 0.01 0.001
1 min
Kind3+/+
Kind3–/–
Kind3+/+
Kind3–/–
1,000
800
600
400
200
0
MFI
(Alexa-488–fibrinogen)
Rest. ADP ADP
+
U46619
CRP Mn2+
600
400
MFI
200
0
10
Rest. ADP
(µM)
CRP
(µg/ml)
Thrombin
(U/ml)
10 0.1 0.01 0.001
Kind3+/+
Kind3–/–
150
100
MFI
50
0
** 60
40
20
0
Surface coverage (%)
0
10
20
30
40
Occlusion time
(min)
>40
abc
def
Figure 2 Impaired platelet function in Kind3
/
mice. (a) Platelet aggregation assay reveals impaired aggregation of Kind3
/
platelets (gray lines) in
response to ADP, U46619, thrombin, CRP and collagen when compared with control platelets (black lines). Arrows denote addition of agonist. (b) Wild-type
(Kind
+/+
) platelets (black bars), but not Kind3
/
platelets (gray bars), bind fibrinogen in response to ADP (10 mM), ADP (10 mM) plus U46619 (3 mM) or
CRP (10 mg/ml). Treatment with MnCl
2
(Mn
2+
; 3 mM) triggers comparable binding. Resting (rest.) platelets were used as a control. (c,d)Kind3
/
platelets
(gray bars) show a complete block in activation of integrin-a
IIb
b
3
after stimulation with ADP (10 mM), CRP (10 mg/ml) and different concentrations (0.001–0.1
U/ml) of thrombin (c), whereas platelet degranulation measured by the surface expression of P-selectin is largely intact after the same treatments (d). Wild-type
platelets (black bars) were used as a control. At the intermediate thrombin concentration, moderately but significantly reduced degranulation was observed with
mutant platelets (** Po0.01). MFI, mean fluorescence intensity. (e)Kind3
/
platelets in whole blood failed to form thrombi when perfused over a
collagen-coated (0.25 mg/ml) surface at a wall shear rate of 1,000 s
–1
. Scale bar, 30 mm. (f) Mesenteric arterioles were injured with FeCl
3
, and adhesion
and thrombus formation of fluorescently labeled platelets were monitored by in vivo video microscopy. Representative images (left) and time to vessel
occlusion (right) are shown. Each symbol represents one individual. Scale bar, 30 mm.
Figure 1 Kindlin-3–deficient animals show severe hemorrhages. (a) Western
blot analyses from spleen and platelet lysates of wild-type (Kind3
+/+
),
heterozygous (Kind3
+/
) and Kindlin-3–deficient (Kind3
/
)mice.
(b) E15.5 embryos reveal severe bleeding. Postnatally, Kind3
/
mice
show skin (arrowhead) and intestinal (arrows) bleeding. All scale bars,
1mm.(c) Tail-bleeding times in wild-type and Kind3
/
mice.
(d) Peripheral platelet counts in wild-type and Kind3
/
chimeras.
LETTERS
326 VOLUME 14
[
NUMBER 3
[
MARCH 2008 NATURE MEDICINE
©2008 Nature Publishing Group http://www.nature.com/naturemedicine
assay. Under high- (1,000 s
–1
,Fig. 2e) or low-shear (150 s
–1
,datanot
shown) flow conditions, wild-type platelets adhered to the collagen
fibers and rapidly built stable three-dimensional aggregates (Fig. 2e).
In contrast, Kind3
/
platelets never established stable adhesions and
either detached immediately or translocated along the fibers for a few
seconds, resulting in virtually no platelets attaching to the collagen-
coated surface at the end of the 4-min perfusion time (Fig. 2e).
Furthermore, agonist-stimulated Kind3
/
platelets failed to bind the
monoclonal anti-b
1
integrin antibody 9EG7, which specifically recog-
nizes the activated form of b
1
integrins16, and also failed to adhere to
soluble, pepsin-digested collagen type I under static conditions (Sup-
plementary Fig. 2 online), a process known to be mediated exclusively
by a
2
b
1
integrins17. Together these findings indicate that Kind3
/
platelets can establish initial contacts with vWF/collagen via GPIb and
probably GPVI, but are unable to adhere firmly as a result of defective
activation of a
IIb
b
3
and a
2
b
1
integrins.
As platelet aggregation may lead to pathological occlusive thrombus
formation, we examined whether lack of Kindlin-3 is protective
against ischemia and infarction after mesenteric arteriole injury.
This injury was induced by ferric chloride and assessed by in vivo
fluorescence microscopy. Five minutes after injury, numerous platelets
adhered firmly to the denuded vessel wall in control chimeras (5,380 ±
2,465/mm
2
); after approximately 10 min, the first thrombi
were observed; and after 18–39 min, the vessels were occluded
(mean occlusion time: 27.6 ± 8.1 min; Fig. 2f). In contrast, a few
Kind3
/
platelets transiently (o5 s) attached to the injured vessel
wall, but virtually none adhered firmly throughout the 40-min
observation period (50 ± 24 / mm
2
). Furthermore, no thrombi
formed in the injured vessels of Kind3
/
chimeras, and blood flow
was maintained in all vessels tested (Fig. 2f). These results confirm the
ex vivo results and underscore the pivotal function of Kindlin-3 in
integrin-mediated platelet adhesion to injured vessels in vivo.
The requirement for Kindlin-3 to trigger agonist-induced integrin
activation on platelets, integrin-mediated platelet adhesion and
thrombus formation suggests that it may be a downstream target of
cellular signaling pathways that activate integrins. To test whether
Kindlin-3 is able to activate integrins, we overexpressed Kindlin-3 in
integrin-a
IIb
b
3
–overexpressing CHO cells. In these cells Kindlin-3 was
unable to trigger integrin activation, likely because the hematopoietic
Kindlin-3 is not recruited to integrin containing focal adhesions13.
Kindlin-3 overexpression in the mouse macrophage cell line RAW
264.7 (RAW), however, yielded a 2.2-fold increase in binding of the
Cy5-labeled fibronectin repeat 7-10 (FN7-10), which harbors the
integrin-binding RGD motif (Fig. 3a). Enhanced green fluorescent
protein (EGFP)-transfected RAW cells showed virtually no increase in
FN7-10 binding as compared to untransfected cells, whereas Talin
overexpression and treatment with Mn
2+
induced a 2.5-fold increase.
Notably, overexpression of a Kindlin-3 variant with a point mutation
in the PTB-like domain (Q597A), which in Talin (R358) reduces
binding to btails4, did not trigger FN7-10 binding. These data
indicate that Kindlin-3 is capable of activating FN binding integrins
and that this activity requires an intact PTB-like domain.
How does Kindlin-3 activate integrins? As reduced Talin expression
did not account for the defect (Fig. 3b), we investigated whether the
FERM domain of Kindlin-3, which is similar to that of Talin, might
play a direct role in integrin activation. As previously shown, recom-
binant Talin bound wild-type b
1
and b
3
integrin tails, and this binding
was significantly reduced when alanine mutations were introduced
into the membrane-proximal tryptophan residue or NPxY motif
(b
3
W739A, b
3
Y747A, b
1
W780A and b
1
Y788A)18. Kindlin-3 was also
able to interact with the wild-type b
1
and b
3
integrin tails (Fig. 3c), in
thepresenceandabsenceofTalin1(Supplementary Fig. 3 online),
and the F3 subdomain of Kindlin-3 was sufficient for this interaction
and this interaction occurred in a direct manner (Fig. 3d). However,
specific point mutations within the b
1
and b
3
integrin cytoplasmic
tails revealed that the binding properties of Kindlin-3 were different
from those of Talin, as the former still bound to b
3
W739A, b
3
Y747A,
b
1
W780A and b
1
Y788A tails, although less efficiently than to wild-
type tails (Fig. 3e,f). Mutations to the membrane distal NxxY motif of
the b
1
and b
3
tails (b
1
Y800A, b
3
Y759A) and the b
1
TT793/794AA and
3
2
1
0
EGFP
Fold FN
binding
Kindlin-3
Kindlin-3
Q597A
Talin
EGFP
+ Mn2+
P = 0.045
P =
0.019
Kindlin-3
Kindlin-3
Coomassie 10 kDa
GST-K3F3
αllb
β3
Coomassie
Tal i n
5% input
αllb
β1
β3
+/+
–/–
Tal i n
GAPDH
Kindlin-3
Coomassie
5% input
GST
β3β3S752P
a
g
bc d
Tal i n
Kindlin-3
Coomassie
5% input
GST
β1β1Y788A
β1Y800A
β1W780A
β1TT793/794AA
αllb
e
Tal i n
Kindlin-3
Coomassie
5% input
GST
β3β3Y747A
β3Y759A
β3W739A
β3ST752/753AA
αllb
f
Figure 3 Biochemical analyses of Kind3
/
platelets. (a) Binding of Cy-5–
labeled fibronectin III 7-10 fragments (FNIII7-10; FN) to RAW 264.7 cells
transfected with EGFP, Talin, EGFP-Kindlin-3 or EGFP-Kindlin-3(Q597A).
Binding of Cy-5–labeled fibronectin III 7-10 lacking the RGD binding loop
(FNIII7-10DRGD) was used to estimate nonspecific binding, and binding in the
presence of 5 mM Mn
2+
served as a positive control. Data shown are mean ±
s.e.m. of five independent experiments, normalized to FNIII7-10 binding to
EGFP-transfected cells and with the background binding of FNIII7-10DRGD
subtracted. Pvalues were obtained by unpaired t-test. (b) Western blot analysis
of lysates from wild-type (+/+) control and Kind3
/
(–/–) platelets.
(c) Pull-down experiment with His-tagged aIIb, b
1
and b
3
integrin tails incubated with 100 mg platelet lysate. (d) Protein-protein interaction assay of
recombinant GST–Kindlin-3 F3 domain (GST-K3F3) incubated with His-tagged aIIb and b
3
integrin tails reveals direct binding between the F3 domain of
Kindlin-3 and the b
3
tail. GST pull-down experiments were performed with recombinant GST-b
1
A (wild-type), GST-b
1
Y788A, GST-b
1
Y800A, GST-b
1
W780A
and GST-b
1
TT793/794AA (e); with GST-b
3
(wild-type), GST-b
3
Y747A, GST-b
3
Y759A, GST-b
3
W739A and GST-b
3
ST752/753AA (f); and with GST-b
3
(wild-
type) and GST-b
3
S752A (g) integrin cytoplasmic domains after incubation with 100 mg platelet lysate. GST protein and GST-aIIb were used as controls. 5%
of the platelet lysate used for the pull-down experiment is shown as input control. Bound Talin and Kindlin-3 proteins were detected by western blotting.
Coomassie blue staining showed that equal amount of GST fusion proteins were used. Shown are results from pull-down assays representative of a minimum
of seven experiments.
LETTERS
NATURE MEDICINE VOLUME 14
[
NUMBER 3
[
MARCH 2008 327
©2008 Nature Publishing Group http://www.nature.com/naturemedicine
b
3
ST752/753AA, however, abolished Kindlin-3 but not Talin binding
(Fig. 3e,f). Moreover, the b
3
S752P mutation found in a subset of
individuals with Glanzmann’s thrombasthenia also abolished Kindlin-
3binding
19 (Fig. 3g). These data indicate that Talin primarily requires
membrane-proximal residues for binding, whereas Kindlin-3 requires
membrane-distal residues for binding to the b
1
and b
3
tails.
Ligand-occupied integrins transduce signals that lead to the activa-
tion of Src family kinases, resulting in cell spreading (outside-in
signaling). We tested the role of Kindlin-3 in outside-in signaling by
analyzing the adhesion of washed control and Kind3
/
platelets to
fibrinogen under static conditions. As mouse platelets, in contrast to
human platelets, do not spread well on immobilized fibrinogen
without cellular activation20, we performed the experiments in the
presence of 0.01 U/ml thrombin, with and without added Mn
2+
.
Comparable adhesion of control and Kind3
/
platelets to the
fibrinogen matrix occurred, confirming a previous observation that
integrin-a
IIb
b
3
activation is not required for static adhesion of platelets
to fibrinogen21. However, whereas control platelets readily formed
lamellipodia and spread within 10–15 min, Kind3
/
platelets only
formed filopodia, with occasional transient small lamellipodia, and
completely failed to spread for up to 45 min (Fig. 4a,b). We obtained
similar results when we carried out adhesion in the presence of
Mn
2+
(Fig. 4c). Thus, Kindlin-3 is also required for integrin a
IIb
b
3
-
dependent outside-in signaling.
Our study demonstrates that Kindlin-3 is essential for platelet
integrin activation and subsequent integrin outside-in signaling.
Furthermore, we found that Kindlin-3 regulates activation of both
b
3
(a
IIb
b
3
)andb
1
(a
2
b
1
) integrins, suggesting that Kindlin-3, like
Talin, is a general regulator of integrin activation. We propose that this
regulatory mechanism is mediated through a direct interaction
between the PTB site of the F3 domain in Kindlin-3 and the integrin
b
3
and b
1
tails, including their distal NxxY motifs. Because Talin
binding requires an intact proximal NPxY motif, our findings raise
questions regarding the roles of Kindlin-3 and Talin in integrin
activation and the hierarchy of their binding to the integrin btails.
Future studies on the structure of the Kindlin-3–integrin complex are
required to examine the relative roles of Kindlin-3 and Talin inter-
actions with integrin tails so as to fully understand how these receptors
become activated. Finally, we show that elimination of Kindlin-3
prevents the formation of pathological thrombi. As Kindlin-3 is
selectively expressed in cells of hematopoietic origin, it may serve as
a potential target for the design of therapeutics aimed at specifically
disrupting integrin activation in platelets.
METHODS
Inactivation of the Kindlin-3 gene. A BAC clone containing the Kind3
(Fermt3) gene was isolated and used to generate the targeting construct
containing an IRES-lacZ-neo cassette between exons 3 and 6. The targeting
vector was electroporated into R1 ES cells, and targeted ES-cell clones were
identified by southern blotting and injected into host blastocysts to generate
germline chimeras.
Generation of fetal liver cell chimeras. Fetal liver cells from E15 wild-type and
Kind3
/
embryos were obtained by pushing the liver through a cell strainer
(Falcon). 4 10
6
cells were injected into the tail vein of lethally irradiated
(10 Gy) recipient C57BL/6 mice. At 3–4 weeks after transfer, platelets were
isolated from whole blood collected from the retro-orbital plexus.
Western blot analysis. Platelet lysates were subjected to a 5–15% gradient
SDS-PAGE. After blotting, PVDF membranes were probed with anti–Kindlin-3
(ref. 12), anti-Talin (Sigma) and anti-GAPDH (Chemicon).
GST fusion protein pull-down assays. The b
1
Aandb
3
integrin cyto-
plasmic domains and their mutant forms (GST-b
1
Y788A, GST-b
1
Y800A,
GST-b
1
W780A and GST-b
1
TT793/794AA; GST-b
3
Y747A, GST-b
3
Y759A,
GST-b
3
W739A, GST-b
3
ST752/753AA and GST-b
3
S752A) were expressed as
GST-fusion proteins in BL21 cells upon induction with 1 mM IPTG. Bacteria
were washed and lysed in buffer A (150 mM NaCl, 1 mM EDTA, 10 mM
Tris, pH 8) containing 100 mg/ml lysozyme for 15 min on ice and then
sonicated. After dialysis against buffer B (100 mM NaCl, 50 mM Tris
pH 7.5, 1% NP-40, 10% glycerol, 2 mM MgCl
2
), GST-fusion proteins were
bound to glutathione-Sepharose beads (Novagen), eluted in 50 mM Tris,
pH 8, 20 mM glutathione, and dialyzed against buffer C (20 mM Tris
pH 7.5, 20% glycerol, 100 mM KCl, 0.2 mM EDTA, 2 mM DTT,
10 mM b-glycerophosphate).
For GST pull-down experiments, GST fusion proteins were bound to
glutathione-Sepharose beads for 1 h at room temperature (B20 1C) in buffer
A. Fresh platelet lysates were incubated with GST or GST-integrin cytoplasmic-
domain fusion proteins for 4 h or overnight at 4 1C. The glutathione-Sepharose
beads were washed four times with buffer A containing 1% Triton X-100 and
10 mM EDTA. Bound proteins were eluted from the beads by boiling in
Laemmli buffer (2% SDS, 10% glycerol, 5% 2-mercaptoethanol, 0.002%
bromphenol blue, 62.5 mM Tris-HCl, pH 6.8) after separation by a SDS-PAGE
and western blotting.
For direct protein-protein interaction assay, recombinant Kindlin-3 F3
domain, spanning amino acids 550–665 (GST-K3F3), was expressed in BL21
cells as described above. Histidine (His)-tagged integrin cytoplasmic tails
were expressed in BL21 bacteria upon induction with 1 mM IPTG and
purified under denaturing conditions. Ten micrograms of GST-K3F3 was
incubated with His-tagged aIIb, and b
3
integrin cytoplasmic tails bound to
Ni
2+
-coated beads for 2 h in buffer D (50 mM NaCl, 10 mM PIPES, 150 mM
sucrose, 0.1% Triton X-100, pH 6.8) containing phosphatase and protease
inhibitor cocktails (Sigma, Roche). After being washed in buffer D, bound
proteins were analyzed by SDS-PAGE and western blotting. Loading of
Ni
2+
-coated beads with the recombinant integrin tails was assessed by
Coomassie Blue staining.
a
b
c
0 min 5 min 15 min
100
80
60
Percentage
of lamellipodia
forming platelets
40
20
0Kind3+/+ Kind3 –/–
Figure 4 Defective adhesion and spreading of Kind3
/
platelets.
(a) Washed wild-type and Kind3
/
platelets were stimulated with
0.01 U/ml thrombin and then allowed to adhere to immobilized fibrinogen
for 15 min. Scale bar, 5 mm. (b) Scanning electron microscopy of wild-type
and Kind3
/
platelets after thrombin stimulation and adhesion to
fibrinogen for 30 min. Scale bars, 1 mm. (c) Washed platelets were allowed
to adhere to immobilized fibrinogen in the presence of 3 mM Mn
2+
for
30 min. Left, representative DIC images. Scale bar, 5 mm. Right, number
of lamellipodia-forming platelets (% of adherent platelets; mean ± s.d.
of four experiments per group).
LETTERS
328 VOLUME 14
[
NUMBER 3
[
MARCH 2008 NATURE MEDICINE
©2008 Nature Publishing Group http://www.nature.com/naturemedicine
Fibronectin binding assay. RAW 264.7 cells were electroporated with 4 mgof
the indicated DNAs using a Macrophage Kit from the Amaxa system. Twenty-
four hours after transfection, cells were trypsinized, washed in FACS Tris-buffer
(24 mM Tris-HCl, pH 7.4, 137 mM NaCl and 2.7 mM KCl) and incubated for
40 min with 0.3 mM Cy5-labeled recombinant fibronectin III 7-10 fragment or
FNIII 7-10DRGD fragment22. As a positive control, EGFP-transfected cells were
incubated with 5 mM Mn
2+
. After washing, the amount of cell-bound
fibronectin fragment was measured with a FACSCalibur (Becton Dickinson).
Dead cells were excluded from FACS analysis by addition of 2.5 mg/ml
propidium iodide and gating for living (propidium iodide negative) and
EGFP-positive cells.
Chemicals. The anesthetic drugs xylazine (Rompun) and ketamine (Imalgene
1000) were purchased from Bayer and Me
´rial, respectively. High-molecular-
weight heparin and human fibrinogen and ADP were from Sigma and
collagen was from Kollagenreagent Horm, Nycomed. Monoclonal antibodies
conjugated to either fluorescein isothiocyanate (FITC) or phycoerythrin (PE)
were from Emfret Analytics. Alexa-Fluor-488–labeled fibrinogen was from
Molecular Probes.
Aggregometry. To determine platelet aggregation, light transmission was
measured using washed platelets (2 10
8
/ml) in the presence of 70 mg/ml
human fibrinogen. Transmission was recorded on a Fibrintimer 4 channel
aggregometer (APACT Laborgera
¨te und Analysensysteme) over 10 min and was
expressed as arbitrary units with transmission through buffer defined as
100% transmission.
Flow cytometry. Heparinized whole blood was diluted 1:20 with modified
Tyrode’s-HEPES buffer (134 mM NaCl, 0.34 mM Na
2
HPO
4
,2.9mMKCl,
12 mM NaHCO
3
, 20 mM HEPES, pH 7.0) containing 5 mM glucose, 0.35%
bovine serum albumin (BSA) and 1 mM CaCl
2
. For assessment of glycoprotein
expression and platelet count, blood samples were incubated with appropriate
fluorophore-conjugated monoclonal antibodies for 15 min at room tempera-
ture and directly analyzed on a FACSCalibur (Becton Dickinson). Activation
studies were performed with blood samples washed twice with modified
Tyrode’s-HEPES buffer, which then were activated with the indicated agonists
or 3 mM MnCl
2
for 15 min, stained with fluorophore-labeled antibodies for
15 min at room temperature and directly analyzed.
Adhesion under flow conditions. Rectangular coverslips (24 60 mm) were
coated with 0.25 mg/ml fibrillar type I collagen (Nycomed) for 1 h at 37 1Cand
blocked with 1% BSA. Perfusion of heparinized whole blood was performed as
described15. Briefly, transparent flow chambers with a slit depth of 50 mm,
equipped with the collagen-coated coverslips, were rinsed with HEPES buffer,
pH 7.45, and connected to a syringe filled with the anticoagulated blood.
Perfusion was carried out at room temperature using a pulse-free pump at low
(150 s
–1
) and high shear stress (1,000 s
–1
). During perfusion, microscopic
phase-contrast images were recorded in real time. Thereafter, the chambers
were rinsed by a 10-min perfusion with HEPES buffer, pH 7.45, at the same
shear stress, and phase-contrast pictures were recorded from at least five
different microscopic fields (63objectives). Image analysis was performed
off-line using Metamorph software (Visitron). Thrombus formation results are
expressed as the mean percentage of total area covered by thrombi.
Analysis of bleeding time. Mice were anesthetized and a 3-mm segment of the
tail tip was cut off with a scalpel. Tail bleeding was monitored by gently
absorbing the bead of blood with a filter paper without contacting the wound
site. When no blood was observed on the paper after 15-s intervals, bleeding
was determined to have ceased. The experiment was stopped after 15 min.
Intravital microscopy of thrombus formation in FeCl
3
injured mesenteric
arterioles. Mice 4–5 weeks old were anesthetized, and the mesentery was gently
exteriorized through a midline abdominal incision. Arterioles (35–60-mm
diameter) were visualized with a Zeiss Axiovert 200 inverted microscope
(10) equipped with a 100-W HBO fluorescent lamp source and a
CoolSNAP-EZ camera (Visitron). Digital images were recorded and analyzed
off-line using Metavue software (Visitron). Injury was induced by topical
application of a 3-mm
2
filter paper tip saturated with FeCl
3
(20%) for 10 s.
Adhesion and aggregation of fluorescently labeled platelets in arterioles were
monitored for 40 min or until complete occlusion occurred (blood flow
stopped for 41min).
Platelet spreading. Cover slips were coated overnight with 1 mg/ml human
fibrinogen and then blocked for 1 h with 1% BSA in PBS. Washed platelets of
wild-type or Kind3
/
mice were resuspended at a concentration of 0.5 10
6
platelets/ml and then further diluted 1:10 in Tyrode’s-HEPES buffer. Shortly
before platelets were seeded on the fibrinogen-coated coverslip, they were acti-
vatedwith0.01U/mlthrombin.Plateletswereallowedtospreadfor30minand
analyzed by differential interference contrast (DIC) microscopy. In parallel,
platelets were fixed in 2.5% glutaraldehyde in Tyrode’s-HEPES buffer and
processed for scanning electron microscopy as previously described23.Inanother
set of experiments, washed platelets were allowed to adhere to fibrinogen in the
presence of 3 mM Mn
2+
without thrombin stimulation and analyzed as above.
Note: Supplementary information is available on the Nature Medicine website.
ACKNOWLEDGMENTS
We thank D. Calderwood (Yale University) and I. Campbell (Oxford University)
for recombinant integrin tails and integrin tail expression vectors and help with
pull-down assays, G. Wanner for imaging of platelets by scanning electron
microscopy, M. Sixt and M. Boesl for mouse manipulation experiments, and
R. Zent, A. Pozzi, M. Humphries and M. Schwartz for critical reading of the
manuscript. This work was supported by the Deutsche Forschungsgemeinschaft
and the Max Planck Society.
AUTHOR CONTRIBUTIONS
M.M. and R.F. designed and supervised research. M.M., B.N. and R.F. wrote the
manuscript. M.M., B.N., S.U. and M.P. performed experiments. All authors
discussed the results and commented on the manuscript.
Published o nline at http: //www.nature.com/n aturemedicine
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions
1. Ruggeri, Z.M. Platelets in atherothrombosis. Nat. Med. 8, 1227–1234 (2002).
2. Savage, B., Almus-Jacobs, F. & Ruggeri, Z.M. Specific synergy of multiple substrate-
receptor interactions in platelet thrombus formation under flow. Cell 94, 657–666
(1998).
3. Nieswandt, B. & Watson, S.P. Platelet-collagen interaction: is GPVI the central
receptor? Blood 102, 449–461 (2003).
4. Tadokoro, S. et al. Talin binding to integrin btails: a final common step in integrin
activation. Science 302, 103–106 (2003).
5. Critchley, D.R. Cytoskeletal protein talin and vinculin in integrin-mediated adhesion.
Biochem. Soc. Trans. 32, 831–836 (2004).
6. Nieswandt, B. et al. Loss of talin1 in platelets abrogates integrin activation, platelet
aggregation, and thrombus formation in vitro and in vivo. J. Exp. Med. (in the press).
7. Calderwood, D.A. et al. The phosphotyrosine binding-like domain of talin activates
integrins. J. Biol. Chem. 277, 21749–21758 (2002).
8. Calderwood, D.A. et al. The talin head domain binds to integrin bsubunit cytoplasmic
tails and regulates integrin activation. J. Biol. Chem. 274, 28071–28074 ( 1999).
9. Wegener, K.L. et al. Structural basis of integrin activation by talin. Cell 128, 171–182
(2007).
10. Kloeker, S. et al. The Kindler syndrome protein is regulated by transforming growth
factor-band involved in integrin-mediated adhesion. J. Biol. Chem. 279, 6824–6833
(2004).
11. Tu, Y., Wu, S., Shi, X., Chen, K. & Wu, C. Migfilin and Mig-2 link focal adhesions to
filamin and the actin cytoskeleton and function in cell shape modulation. Cell 113,
37–47 (2003).
12. Ussar, S., Wang, H.V., Linder, S., Fa
¨ssler, R. & Moser, M. The Kindlins: subcellular
localization and expression during murine development. Exp. Cell Res. 312,
3142–3151 (2006).
13. Weinstein, E.J. et al. URP1: a member of a novel family of PH and FERM domain-
containing membrane-associated proteins is significantly over-expressed in lung and
colon carcinomas. Biochim. Biophys. Acta 1637, 207–216 (2003).
14. Hodivala-Dilke, K.M. et al. b3-integrin-deficient mice are a model for Glanzmann
thrombasthenia showing placental defects and reduced survival. J. Clin. Invest. 103,
229–238 (1999).
15. Nieswandt, B. et al. Glycoprotein VI but not a2b1 integrin is essential for platelet
interaction with collagen. EMBO J. 20, 2120–2130 (2001).
16. Lenter, M. et al. A monoclonalantibody against an activation epitope on mouse integrin
chain b1 blocks adhesion of lymphocytes to the endothelial integrin a6beta1.Proc.
Natl. Acad. Sci. USA 90, 9051–9055 (1993).
17. Holtkotter, O. et al. Integrin a2-deficient mice develop normally, are fertile, but
display partially defective platelet interaction with collagen. J. Biol. Chem. 277,
10789–10794 (2002).
LETTERS
NATURE MEDICINE VOLUME 14
[
NUMBER 3
[
MARCH 2008 329
©2008 Nature Publishing Group http://www.nature.com/naturemedicine
18. Pfaff, M., Liu, S., Erle, D.J. & Ginsberg, M.H. Integrin bcytoplasmic domains
differentially bind to cytoskeletal proteins. J. Biol. Chem. 273, 6104–6109
(1998).
19. Chen, Y.P. et al. Ser-752-Pro mutation in the cytoplasmic domain of integrin b3
subunit and defe ctive activation of platelet integrin aIIbb3 (glycoprotein IIb-IIIa) in a
variant of Glanzmann thrombasthenia. Proc. Natl. Acad. Sci. USA 89, 10169–10173
(1992).
20. McCarty, O.J. et al. Rac1 is essential for platelet lamellipodia formation and aggregate
stability under flow. J. Biol. Chem. 280, 39474–39484 (2005).
21. Savage, B., Shattil, S.J. & Ruggeri, Z.M. Modulation of platelet function through
adhesion receptors. A dual role for glycoprotein IIb-IIIa (integrin aIIbb3) mediated by
fibrinogen and glycoprotein Ib-von Willebrand factor. J. Biol. Chem. 267,
11300–11306 (1992).
22. Takahashi, S. et al. The RGD motif in fibronectin is essential for development but
dispensable for fibril assembly. J. Cell Biol. 178, 167–178 (2007).
23. Fa
¨ssler, R. et al. Lack of b1 integrin gene in embryonic stem cells affects morphology,
adhesion, and migration but not integration into the inner cell mass of blastocysts.
J. Cell Biol. 128, 979–988 (1995).
LETTERS
330 VOLUME 14
[
NUMBER 3
[
MARCH 2008 NATURE MEDICINE
©2008 Nature Publishing Group http://www.nature.com/naturemedicine