The fibronectin synergy site re-enforces cell adhesion and mediates a crosstalk between integrin classes

Fibronectin (FN), a major extracellular matrix component, enables integrin-mediated cell adhesion via binding of α5β1, αIIbβ3 and αv-class integrins to an RGD-motif. An additional linkage for α5 and αIIb is the synergy site located in close proximity to the RGD motif. We report that mice with a dysfunctional FN-synergy motif (Fn1syn/syn) suffer from surprisingly mild platelet adhesion and bleeding defects due to delayed thrombus formation after vessel injury. Additional loss of β3 integrins dramatically aggravates the bleedings and severely compromises smooth muscle cell coverage of the vasculature leading to embryonic lethality. Cell-based studies revealed that the synergy site is dispensable for the initial contact of α5β1 with the RGD, but essential to re-enforce the binding of α5β1/αIIbβ3 to FN. Our findings demonstrate a critical role for the FN synergy site when external forces exceed a certain threshold or when αvβ3 integrin levels decrease below a critical level. DOI: http://dx.doi.org/10.7554/eLife.22264.001


Introduction
Fibronectin (FN) is a large extracellular matrix (ECM) glycoprotein that triggers biochemical and mechanical signaling via integrin binding. FN is essential for mammalian development and tissue regeneration, and can influence disease such as cancer progression. FN is abundant in blood and in most tissues and is present in provisional matrices of healing wounds and in the stroma of tumors. FN is secreted as a disulfide-bonded dimer, assembled into fibrils of variable diameters and then crosslinked into a fibrillar network of variable rigidity (Leiss et al., 2008) that binds to and serves as a scaffold for numerous other ECM molecules. FN consists of three different repeating Ig-like folded units, called type I-III modules. Whereas type I and II modules are stabilized by internal disulfide bonds, the 15 type III repeats of FN lack disulfide bonds, which confers elasticity to FN fibrils and the ability to modulate fibril rigidity (Erickson, 1994;Oberhauser et al., 2002). The major cell-binding site in FN is an arginine-glycine-aspartate (RGD) motif located in the 10 th type III module (FNIII10) that is recognized by a5b1, aIIbb3, and av-class integrins. In addition to the RGD motif, FN harbors the so-called FN synergy site in the FNIII9 module (Obara et al., 1988), which binds a5b1 and aIIbb3 integrins but not av-class integrins (Bowditch et al., 1994). The synergy site encompasses the DRVPPSRN sequence in mouse FN and site directed mutagenesis identified the two arginine residues to be essential for all synergy site-induced functions (Aota et al., 1994;Friedland et al., 2009;Chada et al., 2006;Nagae et al., 2012).
In vitro studies have shown that the synergy site increases cell spreading (Aota et al., 1991), FN fibril assembly (Sechler et al., 1997) and platelet adhesion to FN (Chada et al., 2006). Based on the crystal structures, the RGD motif forms a flexible loop that physically interacts with both the a5 and b1 integrin subunits, while the synergy site contacts only the head domain of the a subunit (Redick et al., 2000;Nagae et al., 2012). The synergy site has been studied using protein-and cellbased assays, which produced different results giving rise to diverse hypotheses regarding the mechanistic properties. One hypothesis based on ultra-structural analyses of the recombinant a5b1 ectodomain and the FNIII7-10 polypeptide proposes that the synergy site aligns the binding interface of the integrin heterodimer with the RGD motif to increase the on-rate constant (K on ) of a5b1 binding to FN (Leahy et al., 1996;García et al., 2002;Takagi et al., 2003). A combination of theoretical and cell-based studies with FRET sensors inserted into the linker region between FNIII9 and FNIII10 concluded that cell-induced forces reversibly stretch the linker, separate the FN-RGD motif from the synergy site and switch the binding of a5b1 integrins to av-class integrins (Grant et al., 1997;Krammer et al., 2002). Finally, using a spinning disk device, it was shown that the engagement of the synergy site allows FN-a5b1 bonds (or FN-aIIbb3 bonds on platelets) to resist shear forces, suggesting that force exposure allows to switch the bonds from a relaxed to a tensioned state, leading to an extension of the FN-integrin bond lifetime and to adhesion strengthening (Friedland et al., 2009). Although these in vitro studies highlight the importance of a5b1 and aIIbb3 integrin-binding to the synergy site, the mode of action is still unclear and the apparently important roles of these interactions have never been scrutinized in vivo using genetic loss-of-function approaches.
We decided to directly test the role of the synergy site in vivo by substituting critical residues of the FN-synergy site in mice. We report that mice carrying a homozygous inactivating mutation in the fibronectin gene (Fn1)-synergy site (Fn1 syn/syn ) are viable, fertile, show no overt organ defects, however, display a mild bleeding tendency and delayed thrombus formation after vessel injury. The lethal intercrosses of Fn1 syn/syn mice with b3 integrin (Itgb3)-deficient mice and in vitro assays with purified FN syn isolated from the blood of Fn1 syn/syn mice revealed three important findings: (i) the synergy site does not influence the K on of a5b1 integrin binding to FN-RGD, (ii) the synergy site strengthens a5b1/aIIbb3 integrin binding to FN upon application of external force such as blood flow or internal force such as actomyosin pulling forces, and (iii) during force-induced adhesion strengthening the synergy site binding to a5b1 and av-class integrin binding to FN compensate each other, at least in part, up to a certain force threshold.

Results
Normal development and prolonged trauma-induced bleeding in Fn1 syn/syn mice To directly test the in vivo role(s) of the FN synergy site, we generated the Fn1 syn allele by substituting the two arginines (R 1374 and R 1379 ) of the synergy motif (DRVPPSRN) in the FN-III9 module with alanines (A) ( Figure 1A and Figure 1-figure supplement 1A-C). Intercrossing of heterozygous mice (Fn1 +/syn ), which showed no apparent phenotype, gave rise to homozygous offspring (Fn1 syn/ syn ) with a normal Mendelian ratio before and after weaning. Fn1 syn/syn mice were fertile, had normal size and weight, and aged normally. The morphology, ultrastructure, and FN distribution in heart ( Figure 1B), liver, kidney, and lung (Figure 1-figure supplement 1D) were indistinguishable between Fn1 syn/syn and control littermates. Blood vessel organization in whole mount ear samples analyzed by anti-PECAM-1 and anti-aSMA immunostainings revealed no abnormalities ( Figure 1C), and the subendothelial matrix visualized with antibodies to laminin-1, collagen IV and FN, was also normally organized in Fn1 syn/syn mice (Figure 1-figure supplement 1E). Altogether, these data indicate that the FN synergy site is dispensable for development and postnatal homeostasis.
Importantly, plasma levels of fibrinogen were also similar in Fn1 +/+ (2.10 ± 0.17 mg/ml) and Fn1 syn/syn (2.08 ± 0.07 mg/ml) mice. In vitro aggregation of washed platelets, induced with either collagen I, thrombin or ADP, triggered normal shape changes and aggregations (Figure 1-figure supplement 2C-E). To quantitatively study the velocity of thrombus formation in vivo, thrombi induction was measured in the arterioles of the cremaster muscle upon vessel injury. The experiments revealed a small delay in the onset of thrombus formation in the Fn1 syn/syn mice (10.29 ± 9.04 min) that, however, was not significantly different compared to the Fn1 +/+ littermates (5.13 ± 3.89 min). In contrast, the time required for arteriole occlusion was significantly increased in Fn1 syn/syn mice (28.56 ± 10.24 min) compared to Fn1 +/+ mice (17.82 ± 9.74 min) ( Figure 1G). Notably, in 3 out of 11 Fn1 syn/syn mice no total occlusion was observed after 40 min ( Figure 1H), a defect that was never observed in control mice.
These results demonstrate that the synergy site is dispensable for development and postnatal homeostasis but is required to stabilize platelet clots in vivo and to prevent prolonged bleeding times.

Fibroblasts delay their focal adhesion maturation on FN syn
The assembly of FN into a fibrillar network depends on a5b1 binding to FN (Fogerty et al., 1990). To test whether FN assembly proceeds normally in the absence of the synergy site, we incubated FN-deficient (Fn1-KO) fibroblasts that express high levels of a5, av, b1 and b3 integrins on their cell surface (Figure 2-figure supplement 1A) with blood plasma derived from either Fn1 +/+ or Fn1 syn/ syn mice. In line with our immunostaining of FN in tissues from Fn1 syn/syn mice, Fn1-KO cells assembled fibrillar FN networks of indistinguishable complexity, fibril diameter and length with plasma from Fn1 syn/syn and Fn1 +/+ mice, respectively ( Figure 2A).
Next, we coated glass coverslips with plasma FN (pFN) purified from Fn1 +/+ or Fn1 syn/syn mice (Figure 2-figure supplement 1B-E), seeded Fn1-KO fibroblasts and measured adhesion and spreading ( Figure 2B-E). Adhesion of Fn1-KO cells to pFN wt and pFN syn began around 3 min after cell seeding and increased with time without noticeable differences (Figure 2-figure supplement 1F). While the formation of nascent adhesions (NAs) was similar on pFN wt and pFN syn ( Figure 2B), the numbers as well as percentage of paxillin-positive focal adhesions (FAs) linked to stress fibers were significantly reduced in Fn1-KO fibroblasts seeded for 30 min on pFN syn ( Figure 2D,E) indicating that the transition from NAs to mature, stress fiber-anchored FAs is delayed on pFN syn . Furthermore, cell spreading determined as cell area at different time points after cell seeding onto pFN syncoated substrates was also delayed in the first 30 min ( Figure 2C). Time-lapse video microscopy confirmed the delayed cell spreading on pFN syn and revealed unstable adhesions consisting of several cycles of binding and release from the substrate (see Video 1, Video 2 and still images in These findings indicate that the synergy site is dispensable for FN fibril formation but promotes the transition from NAs to FAs.   The FN synergy site is required to tension FN-a5b1 bonds and to resist shear forces It has been reported that HT1080 cells seeded on the FNIII7-10 polypeptide, increase adhesion strength to FN upon force application (Friedland et al., 2009). Therefore, we next tested whether the force-induced adhesion strengthening is FN-synergy site-dependent when Fn1-KO cells adhere to plasma-derived, purified full-length pFN syn . We seeded overnightstarved Fn1-KO fibroblasts for 1 hr onto substrates coated with pFN wt or pFN syn and recombinant FNIII7-10 wt or FNIII7-10 syn polypeptides, respectively, and applied a hydrodynamic shear force with a spinning disk device (García et al., 1998). Typically, the number of Fn1-KO fibroblasts adhering to pFN wt -coated coverslips and spun for 5 min decreased non-linearly with the applied force and followed a sigmoidal curve (Figure 3figure supplement 1), whose inflection point (t50) corresponds to the mean shear stress for 50% detachment, and hence to a quantitative measure of adhesion strength. Interestingly, the t50 values of Fn1-KO cells decreased on purified full-length pFN syn by 16% compared to pFN wt ( Figure 3A), and by 43% on FNIII7-10 syn fragment compared to FNIII7-10 wt , indicating that cells develop less adhesion strength on the synergy site-deficient pFN and that higher adhesion strengths arise on fulllength FN compared to FNIII7-10 fragments.
Simultaneous engagement of the RGD motif and the synergy site was suggested to enable a5b1 and aIIbb3 integrins to induce tensioned bonds, which form when receptor and ligand are in close proximity and hence, can be chemically cross-linked (Shi and Boettiger, 2003). To test the extent of bond tensioning on pFN syn , we seeded (15, 30 and 60 min) serum-starved Fn1-KO fibroblasts onto pFN wt -and pFN syn -coated substrates, respectively, spun them and treated them with 3,3'-dithiobis (sulfosuccinimidyl propionate; DTSSP) to crosslink extracellular secondary amines that are within 1.2 nm proximity to each other. We found that the amount of a5 integrins crosslinked to FN in Fn1-KO fibroblasts was reduced to 60% on pFN syn ( Figure 3B). Upon spinning, Fn1-KO cells increased the proportion of a5 integrins crosslinked to pFN wt . Importantly, in cells on pFN syn , the tension was unable to increase the number of crosslinked bonds upon spinning and their numbers remained at the same levels as before spinning ( Figure 3B), which altogether indicates that the spinning force strengthens a5b1-mediated adhesion to FN in a synergy site-dependent manner. Furthermore and in line with a report showing that the conversion of FN-a5b1 bonds from a relaxed to a tensioned state induces phosphorylation of focal adhesion kinase (FAK) on Y397 (Guan et al., 1991;Kornberg et al., 1992), pY397-FAK levels were reduced by 54% when cells were plated on pFN syn compared to pFN wt ( Figure 3C). Importantly, phosphorylation of Y861-FAK, which occurs independent of substrate binding (Shi and Boettiger, 2003), was indistinguishable in cells seeded on pFN wt or pFN syn ( Figure 3C).
Since the intensity of FAK Y397 phosphorylation was shown to operate as a sensor for ECM rigidity (Seong et al., 2013), we conclude that fibroblasts attached to pFN syn perceive insufficient information regarding substrate stiffness.
av-class integrins compensate for the absent FN synergy site which could, at least in part, compensate for the absence of the synergy site during adhesion strengthening ( Figure 3). To test this hypothesis, we seeded pan-integrin-null fibroblasts (pKO) reconstituted with b1-class integrins to express a5b1 (pKO-b1), or with av integrins (pKO-av) to express avb3 and avb5 integrins, or with both b1 and av integrins (pKO-av/b1) (Schiller et al., 2013) on pFN wt -and pFN syn -coated substrates and evaluated cell adhesion, spreading, and adhesion site formation. From the three cell lines, only pKO-b1 cells exhibited reduced adhesion on pFN syn compared to pFN wt at all-time points analyzed ( Figure 4A). Moreover, pKO-b1 cells had significantly fewer FAs, contained fewer stress fibers, and spread less on pFN syn compared to pFN wt ( Figure 4B-F, see Videos 3 and 4 and still images in Figure 4-figure supplement 1). Moreover, the areas of FAs determined with paxillin and b1 integrin stainings were significantly reduced on pFN syn compared to pFN wt ( Figure 4G,H), which altogether suggests that pFN syn -bound a5b1 integrins fail to organize functional adhesion sites and to induce contractile stress fibers required for cell spreading. pKO-av cells adhered and spread similarly on pFN wt and pFN syn , and developed comparably large, paxillin-positive FAs that were anchored to thick stress fibers ( Figure 4B,D). Importantly, pKO-av/b1 cells also showed the same adhesion and spreading behavior, and developed similar adhesion sites on pFN syn indicating that av-containing integrins compensate for the absence of a functional synergy site ( Figure 4B,E). Interestingly, the pKO-av/b1 cells do not show a delay in the transition from NAs to mature FAs on pFN syn , as we observed with Fn1-KO cells, which could be due to the significantly higher b3 and lower a5 integrin cell surface levels on pKO-av/b1 as compared to Fn1-KO cells ( The FN synergy site is dispensable for the on-rate of FN binding to a5b1 integrins Electron microscopy studies of the ligand-binding headpiece of integrin a5b1 complexed with fragments of FN indicated no contact with the synergy site region while kinetic data suggested a role of the synergy site for enhancing the K on of the complex (Takagi et al., 2003). These findings gave rise to the hypothesis that the synergy site contributes to accelerate the initial encounter of a5b1 with FN-RGD, which was in conflict with our observations that adhesion initiation was unaffected in Fn1-KO cells seeded on pFN syn (Figure 2-figure supplement 1F). To further test whether the synergy site is required for the FN binding on-rate, we quantified the probability of pKO-b1, pKO-av, pKO-a v/b1 and pKO cells binding to FNIII7-10 wt or FNIII7-10 syn fragments and to purified full-length pFN wt or pFN syn using single-cell force spectroscopy ( Figure 4I,J). To this end, a single cell was attached to the ConA-coated cantilever, lowered onto the FN with a speed of 1 mm/s until a contact force of 200 pN was recorded. After a very short contact time of » 50 ms, cell and substrate were separated to detect the rupture of the few specific bonds formed between integrins and FN. On FNIII7-10 wt , the experiments revealed a 3-fold higher binding probability of pKO-av and pKO-av/b1 cells compared to pKO-b1 cells, indicating that avb3 integrins have a higher affinity for FN-RGD than a5b1 integrins. Similar results were observed with full-length pFN wt or pFN syn ( Figure 4J). Interestingly, however, full-length pFN showed higher binding probability than fragments for all cell lines tested including the pKO cells that lack integrin expression, which altogether suggests that in addition to integrins also other FN-binding cell surface receptor(s) contribute to the initial binding.
These findings indicate that the FN synergy site promotes the maturation of FAs but accelerates neither the rates of FN binding to a5b1 integrins nor the formation of NAs.

The FN synergy site compensates for aIIbb3 integrin loss on platelets
To test whether the FN synergy site can also compensate for the loss of b3-class integrin expression in vivo, we generated homozygous compound mice carrying the Fn1 syn mutation and the Itgb3 null mutation (Itgb3 -/-) (Hodivala-Dilke et al., 1999). Itgb3-null mice fail to express the widely expressed avb3 integrins and the platelet-specific aIIbb3 integrin, and suffer from a bleeding disorder resembling human Glanzmann thrombasthenia. Around 87% of Itgb3-null mice are born and around 40% of them survive the first year of life (Hodivala-Dilke et al., 1999). To test how the Fn1 syn alleles affect development and survival of Itgb3 -/mice, we intercrossed Fn1 syn/+ ;Itgb3 +/as well as Fn1 syn/ syn ;Itgb3 +/mice and obtained a total of 245 and 90 live offspring at P21, respectively ( Table 1 and   Table 1-source data 1). Out of the 335 offspring altogether, one instead of the expected 38 compound homozygous Fn1 syn/syn ;Itgb3 -/mice survived to P21. The survivor died at the age of 5 months from excessive bleeding. To determine the time-point of lethality, embryos were collected at different gestation times and genotyped. While compound homozygous Fn1 syn/syn ;Itgb3 -/embryos were present at the expected Mendelian distribution until E15.5, no live embryos were present at E16.5 or later. Interestingly, mice with one wild-type Itgb3 allele (Fn1 syn/syn ;Itgb3 -/+ ) were normally Adhesion of pKO-b1, pKO-av and pKO-av/b1 fibroblasts seeded on pFN wt or pFN syn for indicated times (n = 3 independent experiments; mean ± sem). (B) pKO-b1, pKO-av and pKO-av/b1 fibroblasts were seeded on pFN wt or pFN syn , fixed at the indicated times and stained for total b1 integrin (green), paxillin (white) and F-actin (red). Scale bar, 50 mm. (C-E) Quantification of cell area of pKO-b1 (C), pKO-av (D) and pKO-av/b1 (E) cells seeded on pFN wt or pFN syn for indicated times. (F-H) Quantification of the number of FAs (F), the percentage of FA coverage measured as paxillin-positive area (G) and the percentage of b1 integrin-positive areas referred to the total cell area (H) in pKO-b1 cells (n = 25 cells for each measurement and three independent experiments; mean ± sem). The binding probability Figure 4 continued on next page distributed, which altogether indicates that one b3 integrin allele is sufficient to compensate for normal development.
Compound homozygous Fn1 syn/syn ;Itgb3 -/embryos displayed multiple cutaneous hemorrhages and edema, which were first visible at E11.5/12.5 ( Figure 5A and Figure 5-figure supplement 1A) and then spread over the whole body at E15.5 ( Figure 5B). Interestingly, at E12.5 the bleeds were visible at sites where the lymphatic vessels form (arrowheads in Figure 5A) and therefore, we hypothesized that the newly formed lymphatic vessels fail to separate from the cardinal vein, which occurs between E11-13 (Carramolino et al., 2010). In line with our hypothesis, Lyve1-positive lymphatic vessels in the skin of Fn1 syn/syn ;Itgb3 -/embryos were dilated and covered with ectopic asmooth muscle actin (a-SMA)-positive cells and filled with Ter119-positive erythroblasts ( Figure 5C-E). In contrast, lymphatic vessels in the skin of Itgb3-null or wild-type littermates neither contained Ter119-positive cells nor were surrounded with a-smooth muscle actin-positive cells.
The separation of the primary lymphatic sac from the cardinal vein is driven by platelet adhesion to and aggregation at the lymphatic endothelium (Carramolino et al., 2010;Uhrin et al., 2010). We therefore hypothesized that the platelet functions are severely compromised in Fn1 syn/syn ;Itgb3 -/embryos as they lack aIIbb3-mediated binding to fibrinogen and FN ( Figure 6A), as well as the ability to strengthen adhesion and signaling via a5b1 integrin-mediated binding to FN. To test the hypothesis, we performed spreading assays as well as adhesion assays under flow with wild-type or Itgb3 -/platelets. The mean spreading area of wild-type platelets seeded for 60 min on fibrinogen, pFN wt , and pFN syn was 15-16 mm 2 . As expected, Itgb3 -/platelets failed to spread on fibrinogen (mean spreading area of 4.6 mm 2 ). Furthermore, they showed a reduced mean spreading area of 8.9 mm 2 on pFN wt and failed to spread on pFN syn (mean spreading area of 3.8 mm 2 ) ( Figure 6B,C). Application of shear flow reduced adhesion of wild-type platelets to pFN syn by 10-fold compared to pFN wt , while adhesion of Itgb3-null platelets was lost on fibrinogen as well as pFN syn , and only slightly diminished on pFN wt ( Figure 6D,E). Importantly, adhesion and spreading of platelets isolated from Itgb3 -/mice to collagen were unaffected, irrespective of whether shear flow was applied or not ( Figure 6C,E).
These in vitro experiments demonstrate that adhesion of aIIbb3-deficient platelets to wild-type FN is partially compensated by a5b1 integrins in a FN synergy site-dependent manner, and that a5b1 as well as aIIbb3 integrins require the FN synergy site for stabilizing platelet adhesion to FN, under shear flow.

The FN synergy site compensates for avb3 during vessel maturation
The absence of a5b1 integrins leads to vascular defects (Abraham et al., 2008). To test whether vascular abnormalities due to an impaired a5b1 function contribute to the severe bleeds and the lethality of Fn1 syn/syn ;Itgb3 -/embryos, we analyzed the mural coverage and anchorage to the ECM. While immunostaining of E11.5 whole mount embryos with an anti-PECAM-1 antibody revealed that the vessels in the trunk of Fn1 syn/ syn ;Itgb3 -/embryos showed normal sprouting ( Figure 5-figure supplement 1B), the arteries and veins of the dermal vasculature of E15.5 embryos were tortuous and irregularly covered with a-SMA-positive cells ( Figure 7A). Furthermore, the vascular network was less intricate and had significantly fewer branching points in Fn1 syn/syn ;Itgb3 -/embryos compared to wildtype littermates ( Figure 7B). Interestingly, collagen IV immunostaining indicated that many small vessels in E15.5 Fn1 syn/syn ;Itgb3 -/embryos lacked a clear lumen and PECAM-1 immunosignals (see arrowheads in Figure 7C). They probably represent retracted vessels and were significantly more frequent in Fn1 syn/syn ;Itgb3 -/embryos compared to wild-type, Fn1 syn/syn ; Itgb3 +/+ and Fn1 +/+ ;Itgb3 -/littermates ( Figure 7D). Moreover, small vessels in Fn1 syn/ syn ;Itgb3 -/embryos were often less covered by pericytes. Instead, NG2-positive pericytes were either detached or formed patchy aggregates on the vessel surface (see arrowheads in Figure 7E). Altogether, these observations indicate that the vessel wall coverage and stability are decreased in the Fn1 syn/syn ;Itgb3 -/embryos and probably contribute to their severe hemorrhages.

Discussion
Although cell-based studies suggested that the FN synergy site is required for aIIbb3 and a5b1 integrin function, the in vivo evidence was missing and the mechanistic property controversial. We report here the characterization of a mouse strain, in which the synergy site of FN (Fn1 syn ) was disrupted. Contrary to expectations, the Fn1 syn/syn mice were born without developmental defects indicating that the synergy site is dispensable for organogenesis and tissue homeostasis. However, when Fn1 syn/syn mice are exposed to stress such as tail bleeding and arteriole injury, or the genetic ablation of the FN-binding b3-class integrins (avb3, aIIbb3), the synergy site becomes essential for cells that have to resist or produce high forces such as platelets and vascular smooth muscle cells (Figure 8).  Ablation of the Fn1 gene in mice, as well as the simultaneous ablations of the Itga5/Itgav integrin genes in mice arrests development at embryonic day 8.5 (E8.5) due to defects in the formation of mesoderm and mesoderm-derived structures (George et al., 1993;Georges-Labouesse et al., 1996;Yang et al., 1999). The replacement of the FNIII10 RGD motif with the RGE in mice also affects mesoderm development, although less severe and restricted to the vascular system and to the posterior region of the developing embryo (Takahashi et al., 2007;Giró s et al., 2011). Interestingly, these defects resemble those observed in Itga5-deficient mice indicating that the RGE mutation is sufficient to abrogate a5b1 integrin function and that the synergy site cannot compensate for a dysfunctional RGD motif. Furthermore, the normal development of Fn1 syn/syn mice also excludes an essential role of the synergy site for a5 integrin function in vivo (Grant et al., 1997;Krammer et al., 2002). A reduced a5b1 integrin function would probably have occurred if the synergy site would indeed guide the binding pocket of a5b1 towards the RGD motif and increase the FN-binding on-rate (Takagi et al., 2003). However, the absence of obvious 'a5b1-loss-of-function defects' (Yang et al., 1993) in Fn1 syn/syn mice and the normal FN-binding on-rates of pKO-b1 cells in single-cell force spectroscopy experiments indicate that the synergy site is probably dispensable to accelerate a5b1 integrin-FN binding. We also demonstrate an unexpected, compensatory role between the FN synergy site and avb3 integrins for the vascular coverage by smooth muscle cells. Apparently, the high myosin II-induced forces generated by these cells are only efficiently absorbed with either high av-class integrin surface levels or a fully functional FN. Whether a similar functional relationship operates also during paraxial mesoderm, whose formation critically depends on the expression of a5b1 and av-class integrins (Yang et al., 1999), cannot be deduced from our experiments. However, the normal mesoderm formation in Fn1 syn/syn mice indicates that mesodermal cells require a5b1 to bind the RGD motif but Figure 6. Shear flow exposed platelets fail to adhere to pFN syn . (A) Cartoon showing the platelet integrins that can be ligated to the different substrates used in the experiments. The color intensity of the integrin denotes whether the integrin is active or inactive. (B) Spreading of Itgb3 +/+ and Itgb3 -/platelets after 1 hr on fibrinogen, pFN wt , pFN syn and type I collagen. Scale bars, 10 mm. (C) Quantification of the platelet area at indicated times (n = 100 platelets per each condition in three independent experiments; mean ± sem). (D) Representative figures of fluorescently labeled Itgb3 +/+ or Itgb3 -/platelets seeded on indicated substrates and exposed to shear flow. Scale bar, 40 mm. (E) Platelet coverage after 10 min shear flow of 1000 s À1 . (n = 10 pictures per experiment, four independent experiments for each condition; mean ± sem). Statistical significances were calculated using the Student t-test; *p<0.05, **p<0.01 and ***p<0.001. DOI: 10.7554/eLife.22264.020 Figure 7. Malformed blood vessels in Fn1 syn/syn ;Itgb3 -/embryos. (A) PECAM-positive endothelial cells (red) and a-SMA-positive smooth muscle cells (green) in dermal whole mounts from E15.5 Fn1 +/+ ;Itgb3 +/+ and Fn1 syn/syn ; Itgb3 -/littermate embryos indicate veins (V) and arteries (A). (B) Quantification of the number of branching points (n = 10-15 images of 2-3 embryos; mean ± sem). (C) Vascular basement membranes in dermal whole mounts from E15.5 Fn1 +/+ ;Itgb3 +/+ , Fn1 +/+ ;Itgb3 -/and Fn1 syn/syn ;Itgb3 -/littermate embryos stained for type IV collagen (green) and PECAM-positive endothelial cells (red). Arrowheads show small vessels lacking lumen. (D) Quantification of retracted vessels (n = 14-23 from 4-7 embryos; mean ± sem). (E) PECAM-positive endothelial cells (red) and NG2positive pericytes (green) in dermal whole-mounts from E15.5 Fn1 +/+ ;Itgb3 +/+ , Fn1 +/+ ;Itgb3 -/and Fn1 syn/syn ;Itgb3 -/littermate embryos. Note pericytes are sparse, absent or aggregate on mutant vessels (arrowheads). Statistical significances were calculated using the Student t-test; *p<0.05, and ***p<0.001. Scale bars, 50 mm. DOI: 10.7554/eLife.22264.021 not the synergy site. Interestingly, we also find clear evidence for a compensatory role of the FN synergy site and av-class integrins for cell spreading and FA maturation in vitro (Figure 8), which differs from previous reports showing that a5b1 and av-class integrins have non-overlapping functions for inducing myosin contractility (Schiller et al., 2013) or controlling directional migration (Missirlis et al., 2016).
It is well known that platelet adhesion to pFN, mediated by aIIbb3 with contributions from a5b1 and other integrins, plays a critical role for hemostasis (Wang et al., 2014;Ni et al., 2003b). Our findings revealed that the FN synergy site is critically important for the adhesion of wild type platelets in in vitro flow chamber settings, while impaired spreading or defects under static adhesion only arise when aIIbb3 expression is lost. These findings indicate that the synergy site plays a central role as soon as force is applied to the bonds between FN and platelet integrins. A similar force-dependent requirement of the synergy site was also apparent after arteriolar injuries in vivo, which showed that Fn1 syn/syn mice display diminished platelet adhesion and delayed thrombus formation. This requirement of the synergy site for platelets to resist the shear forces of the blood flow also provides a rational explanation for the prolonged bleeding times observed in Fn1 syn/syn mice after tail tip excisions. Interestingly, the in vivo platelet dysfunctions in Fn1 syn/syn mice as well as Itgb3 -/mice profoundly aggravate in compound Fn1 syn/syn ;Itgb3 -/embryos where they cause fatal bleedings and an insufficient platelet-mediated separation of the lymphatic vessels from the cardinal vein. Since murine platelets contain around 100 times more aIIbb3 than a5b1 integrins (Zeiler et al., 2014) the compensation of the entire aIIbb3 pool by the minor a5b1-FN wt complexes underscores the fundamental role of the adhesion strengthening property of the synergy site in FN.
Our mouse strain will allow now to test how the synergy site-mediated adhesion re-enforcement affects the course of tissue fibrosis or cancer development and progression, which are heavily influenced by integrin surface levels as well as tissue rigidity that in turn is modulated by the strength of integrin-ligand bonds (Zilberberg et al., 2012;Laklai et al., 2016).

Animals
Mice were housed in special pathogen free animal facilities. All mouse work was performed in accordance with the Government of the Valencian Community (Spain) guidelines (permission reference A1327395471346) and with the Government of Upper Bavaria. Mice containing the integrin b3 deletion were bred under the permission reference 55.2-1-54-2532-96-2015. The tail-bleeding and cremaster muscle venules injury assays performed under the permission reference 55.2-1-54-2532-115-12.

Generation of Fn1 syn/syn knockin mice
A 129/Sv mouse PAC clone was used to construct the targeting vector ( Figure 1-figure supplement 1A), which consisted of a 2.1 kb fragment containing exons 26 and 27, a neomycin cassette flanked by loxP sites, a 2.4 kb fragment containing the exon 28 carrying the mutated nucleotides, and a 3.5 kb fragment with the exons 29 to 32. The targeting construct was linearized with NotI and electroporated into R1 embryonic stem (ES) cells. Approximately 300 G418-resistant clones were isolated and screened by Southern blot for homologous recombination. The genomic DNAs were digested with SacI, XbaI or BstEII and probed with external probes 1 and 2 (Figure 1-figure supplement 1A). Two correctly targeted clones were injected into C57BL/6 host blastocysts to generate germline chimeras. The Fn1 syn-neo/+ mice were crossed with a deleter-Cre strain to eliminate the loxP flanked neomycin cassette. The elimination of neomycin was analyzed by Southern blot, genomic DNA was digested with Eco RI and probed with probe 3 (Figure 1-figure supplement 1B). The Fn1 syn/+ mice were intercrossed to generate homozygous Fn1 syn/syn mice. The following primers were used to genotype the mouse strain by PCR: 5'-TCACAAGGAAACCAGGGAAC-3' (forward); 5'-CCGTTTTCACTCTCGTCAT-3' (reverse).

Cell lines
The mouse Fn1-KO cell line and the integrin pan-KnockOut fibroblast lines were isolated from a mouse kidney and immortalized by retroviral delivery of the SV40 large T. To generate Fn1-KO cells, the Fn1 gene was deleted from Fn1 flox/flox with the adenoviral transduction of the Cre recombinase. Integrin pKO fibroblasts were generated as described by Schiller et al. (2013). Itgav flox/flox ; Itgb1 flox/flox ; Itgb2 -/-;Itgb7 -/immortalized fibroblasts were treated with adenoviral Cre recombinase and reconstituted with mouse Itgb1 and/or Itgav integrin cDNAs to generate pKO-b1, pKO-av and pKO-av/b1 cells. The cells were provided by H. Schiller (Max-Planck Institute for Biochemistry, Martinsried, Germany). Cell lines were not tested for mycoplasma.

Purification of plasma fibronectin (pFN)
Blood was collected from Fn1 +/+ and Fn1 syn/syn mice using 0.5 M EDTA as anticoagulant in non-heparinized capillaries, centrifuged at 3000 rpm for 20 min and the pFN was purified from the supernatant (plasma) using Gelatin-Sepharose (GE Healthcare Life Sciences, Valencia, Spain) affinity chromatography (Retta et al., 1999) adapted to minicolumns (Poly-Prep, Bio-Rad). Briefly, the columns were washed with 0.5 NaCl in 10 mM Tris-HCl pH7.4 and pFN was eluted with 2 M urea in TBS (0.15 M NaCl in 10 mM Tris-HCl, pH 7.4) and dialyzed against TBS. Purified FN was analyzed by 8% SDS-PAGE and stained with Coomassie brilliant blue, and by Western blot.

Production of the FNIII7-10 fragment
We used the human cDNA encoding the FNIII7-10 fragment and subcloned in the expression vector pET-15b (Takahashi et al., 2007). To generate the FNIII7-10 syn we mutated by site-directed mutagenesis the two arginines in the synergy sequence: DRVPHSRN>DAVPHSAN. We performed two rounds of PCR using the following primers: 5´-GATGCGGTGCCCCACTCTCGGAAT-3´(forward) and 5´-GATGCGGTGCCCCACTCTGCGAAT-3´(forward) and the complementary reverse primers. The expression of recombinant FN fragments was done in the E. coli strain Rosetta T1R. Purification was performed using TALON Metal Affinity chromatography (Clontech, Saint Germain en Laye, France). Finally the protein was obtained by gel filtration chromatography using Superdex 200 10/300 GL columns (GE Healthcare) and Superdex Size Exclusion Media (GE Healthcare, Valencia, Spain) and eluted in PBS.

Adsorption of purified pFN onto glass
Glass coverslips of 18 mm diameter were poly-maleic anhydride-1-octadecene (POMA; Polysciences Inc)-treated (Prewitz et al., 2013) and coated with 0.1-10 mg/ml of purified mouse pFN during 1 hr at RT, followed by a blocking step of 1 hr using 1% BSA in PBS. To quantify the adsorbed FN, the coverslips were then incubated for 2 hr at RT with anti-FN antibodies (Ab; diluted 1/300 in blocking solution), washed, incubated with anti-rabbit Ab conjugated-HRP (diluted 1/500 in blocking solution) 1 hr at RT and finally treated with 50 ml of 2,2'-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid (ABTS; Peroxidase substrate kit, Vector SK-4500) for 30 min in the dark. The ABTS-containing solution was collected and the absorbance was measured at 405 nm.
Cell adhesion assay 96 well plates were coated with 10 mg/ml of pFN or poly-lysine (Sigma-Aldrich, Madrid, Spain) or 3% BSA in PBS during 1 hr at RT, followed by a blocking step of 30 min using 3% BSA in PBS. The cells were starved overnight in 9% serum replacement medium (SRM) composed of 46.5% AIM-V (Life Technologies, Madrid, Spain), 5% RPMI (Life Technologies) and 1% NEAA (Non-Essential Amino Acid Solution, Sigma-Aldrich) supplemented with 1% FN-depleted calf serum. 5 Â 10 4 cells were plated, allowed to adhere for the indicated times and medium was removed and wells washed three times with PBS. The cells were stained with crystal violet (20% Methanol, 0.1% Crystal Violet) overnight at 4˚C, washed, 0.1% triton X-100 was added and incubated for 2 hr at RT. Absorbance was measured at 595 nm.

Spinning disk assay
The spinning disk assay was done as previously described (Boettiger, 2007) on POMA-treated glass coverslips of 25 mm diameter, coated with a solution of 10 mg/ml of purified pFN during 1 hr at RT and afterwards blocked 1 hr with 1% BSA in PBS. The Fn1-KO or HT1080 cells were starved overnight in 9% SRM supplemented with 1% FN-depleted serum. 7 Â 10 5 cells were seeded, allowed to adhere for 1 hr and spun for 5 min at 6000 rpm in Dulbecco´s PBS supplemented with 80 mM CaCl 2 and 80 mM MgCl 2 . After spinning the cells were fixed with 4% PFA and nuclei stained with DAPI. The nuclei were counted with a Zeiss Axiovert (objective 10x) controlled by Metamorph software, which allows taking images at determined positions. Data were analyzed as described (Boettiger, 2007). We calculated for each condition the t50, which is the mean force for cell detachment.

pFN-integrin crosslinking assay
Cells were seeded onto pFN-coated glass coverslips and spun and non-spun cells were incubated with 1 mM 3,3'-dithiobis (sulfosuccinimidyl propionate) (DTSSP; Thermo Scientific, Madrid, Spain) during 15 min at 4˚C. Quenching was carried out with 50 mM Tris, pH 7.4 for 15 min at RT and cells were extracted with 20 mM Tris, pH 7.4, 0.1% SDS and proteinase inhibitors (Inhibitors cocktail, Roche , Barcelona, Spain). Cell lysates were collected and coverslips were thoroughly washed with 20 mM Tris, pH 8.5 followed by incubation with 20 mM Tris, pH 8.5, 0.1% SDS and 25 mM DTT for 1 hr at 37˚C to break the crosslinks. The whole crosslinked fractions and the cell lysates were separated by SDS-PAGE and transferred to a nitrocellulose membrane. Western-blots were analyzed with ImageJ and the levels of crosslinked integrins were calculated as the relation between the crosslinked and the total integrin fractions (cell lysates + crosslinked fraction).

Single-cell force spectroscopy (SCFS)
For cell attachment, cantilevers were plasma cleaned (PDC-32G, Harrick Plasma, Ithaca, NY, USA) and then incubated overnight at 4˚C in PBS containing ConA (2 mg/ml, Sigma-Aldrich) . For substrate coatings, 200 mm thick four-segmented polydimethylsilane (PDMS) mask fused to the surface of glass bottom Petri dishes (WPI, Sarasota, FL, USA) was used (Yu et al., 2015). Each of the four PDMS framed glass surfaces were incubated overnight at 4˚C either with the FNIII7-10 wt or FNIII7-10 syn fragments or full-length FN (50 mg/ml) in PBS. For SCFS, we mounted an AFM (Nanowizard II, JPK Instruments, Berlin, Germany) on an inverted fluorescence microscope (Puech et al., 2006) (Observer Z1/A1, Zeiss, Germany). The temperature was kept at 37˚C throughout the experiment by a Petri dish heater (JPK Instruments,Berlin, Germany). 200 mm long tip-less V-shaped silicon nitride cantilevers having nominal spring constants of 0.06 N/m (NP-0, Bruker, USA) were used. Each cantilever was calibrated prior the measurement by determining its sensitivity and spring constant using the thermal noise analysis of the AFM (Hutter and Bechhoefer, 1993). To adhere a single fibroblast to the AFM cantilever, overnight serum-starved fibroblasts with confluency up to » 80% were washed with PBS, trypsin-detached for up to 2 min, suspended in SCFS media (DMEM supplemented with 20 mM HEPES) containing 1% (v/v) FCS, pelleted and resuspended in serum free SCFS media. Fibroblasts were allowed to recover for at least 30 min from trypsin treatment. Adhesion of a single fibroblast to the free cantilever end was achieved by pipetting the fibroblast suspension onto the functionalized Petri dishes. The functionalized cantilever was lowered onto a fibroblast with a speed of 1 mm/s until a force of 1 nN was recorded. After » 5 s contact, the cantilever was retracted with 1 mm/s for 10 mm and the cantilever bound fibroblast was incubated for 7-10 min to assure firm binding to the cantilever. Using differential interference contract (DIC) microscopy, the morphological state of the fibroblast was monitored. For single molecule sensitivity, the fibroblast bound to the cantilever was lowered onto the coated substrate with a speed of 1 mm/s until a contact force of 200 pN was recorded for » 50 ms contact time. Subsequently, the cantilever was retracted at 1 mm/s and for !13 mm until the fibroblast and substrate were fully separated. After the experimental cycle, the fibroblast was allowed to recover for 0.5 s. For each measurement, the area of the substrate was varied. Force-distance curves were analyzed to determine binding probability using JPK software. Mann-Whitney tests were applied to determine significant differences between the binding probability of fibroblast lines at different conditions. Tests were done using Prism (GraphPad, La Jolla, USA).

pFAK analysis
Cells were plated on pFN-coated glass coverslips and spun in the spinning disk device, then lysed in RIPA buffer (50 mM Tris, pH 7.4; 1% NP-40; 0.5% Na-Deoxycolate; 0.1% SDS; 2 mM EDTA) supplemented with proteinase inhibitors (Complete Proteinase Inhibitor Cocktail tablet, Roche), phosphatase inhibitors (Protease Inhibitors Cocktail 2 Aqueous Solution and Cocktail 3, Sigma-Aldrich), 1 mM Na 3 VO 4 and 5 mM NaF for 10 min on ice, and sonicated for 1 min. The protein concentrations were quantified using the Pierce BCA Protein Assay Kit (Thermo Scientific) assay and 30-50 mg of protein were separated by SDS-PAGE gel, transferred to nitrocellulose membranes and hybridized with specific antibodies. Western-blots were analyzed with ImageJ and the levels of phospho-Tyr 397-FAK or phospho-Tyr 861 FAK were referred to the total FAK content.

FN matrix assembly assay
Fn1-KO fibroblasts were starved overnight in 9% SRM supplemented with 1% FN-depleted serum, trypsinized and transferred into 8-well Lab-Tek chambers (Thermo Scientific) coated for 1 hr with a solution of 20 mg/ml of Laminin (Roche) at RT. After 3 hr, the 9% SRM was supplemented with 1% mouse plasma and cells were incubated for 24, 48, 72 and 96 hr, fixed with 4% PFA and prepared for immunofluorescence staining.

Integrin expression analysis by FACS
Flow cytometry to analyse integrin levels on the Fn1-KO fibroblasts was carried out as previously described (Theodosiou et al., 2016).

Histological analysis
Adult mice were perfused with 4% parafolmaldehyde (PFA) in PBS or tissue pieces and embryos were fixed overnight with 4% PFA at 4˚C. Fixed tissues were dehydrated in graded alcohol series, embedded in paraffin (Paraplast X-tra, Sigma-Aldrich), sectioned into 8 mm thick sections and stained with Haematoxylin-Eosin (H and E) using standard protocols. For immunostainings, sections were hydrated with inverse graded alcohol series, unmasked by heating in 10 mM citrate buffer (pH 6) for 10 min, blocked for 1 hr with 3% BSA at RT and incubated overnight with the primary antibody, washed, incubated with secondary antibodies for 1 hr at room temperature, washed, DAPI stained and mounted on glass slides with elvanol.

Embryo and skin whole mount immunostaining
Embryos were isolated from pregnant mothers at the stages of E11.5, E15.5 and E16.5 and fixed overnight at 4˚C with DENT´s fixative consisting of 80% Methanol, 20% DMSO. The skin was dissected after fixation from the E15.5 and E16.5 embryos, washed 3 times with 100% methanol (5 min) and kept at À20˚C in 100% methanol. For staining, fixed pieces of skin or whole E11.5 embryos were hydrated in decreasing (75, 50 and 25%) methanol series, diluted in PBS supplemented with 0.1% Tween20 (PBST) and blocked for 2 hr at RT with 3% BSA in PBST. Incubations with primary and secondary antibodies were done overnight at 4˚C with gentle rocking and after washing with PBST, tissues were mounted with elvanol.

Platelet aggregation in vitro assays
Platelet aggregation was measured with 1 Â 10 8 washed platelets stimulated with 0.5 U/ml thrombin (Sigma-Aldrich) or 5 mg/ml fibrillar type I collagen (Nycomed, Munich, Germany) in the presence of 10 mg/ml pFN isolated either from Fn1 +/+ or Fn1 syn/syn mice. For platelet aggregation with 20 mM ADP, platelet rich plasma (PRP) was isolated. The mouse blood was collected with citrate buffer (1:9, buffer:blood), centrifuged at 110xg and the supernatant (PRP) was collected. A volume of 225 ml of PRP containing 6.75 Â 10 7 platelets was used for each experiment adding 20 mM ADP. Light transmission was recorded with a ChronoLog aggregometer over 15 min as arbitrary units with the transmission through buffer defined as 100% transmission.

FN and fibrinogen quantification in isolated platelets and blood plasma
Platelets were isolated from Fn1 +/+ and Fn1 syn/syn heparinized blood as described above. About 5 Â 10 6 platelets were lysed with 0,1% Triton in TBS with proteinase inhibitors (Complete Proteinase Inhibitor Cocktail tablet, Roche) during 10 min on ice. After centrifugation at 13,000 rpm, the supernatant was run in an 8% SDS-PAGE under reducing conditions, transferred to nitrocellulose membranes and incubated with anti-FN antibodies. To quantify the plasma content of FN and fibrinogen, 2 ml of plasma were loaded onto the 8% SDS-PAGE. As a reference, we used pure human pFN (Millipore) and human fibrinogen (Sigma-Aldrich). Western-blots were analyzed with ImageJ. To know the FN levels in platelets derived from the different mouse strains the FN levels were related to their vinculin contents.

Platelet spreading assay
To study platelet spreading, glass bottom dishes were coated with 10 mg/ml of pFN wt , pFN syn , fibrinogen (Sigma-Aldrich) or soluble collagen type I (PureCol, Advanced Biomatrix, San Diego, CA, USA) at RT for 1 hr and blocked with 1% BSA in PBS. Washed platelets (0.5-1 Â 10 6 ) were added to the dishes in a final volume of 1 ml and activated with 0.01% thrombin (Sigma-Aldrich). Images were taken after 15, 30 and 60 min under a differential interference contrast microscopy (Zeiss Axiovert 200M microscope with a Plan-NEOFLUAR,Â100, 1.45 oil objective; Zeiss, Jena, Germany) using the Metamorph software (Molecular Devices, Sunnyvale, CA, USA). The platelet spreading area was analysed using the ImageJ software.

Platelets adhesion assay under flow
Flow chamber experiments were carried out as described previously (Schulz et al., 2009) using the air-driven continuous flow pump system from Ibidi. Briefly, platelets were isolated, fluorescently labelled with 5 mM carboxyfluorescein succinimidyl ester (CFSE; Invitrogen) in Tyrodes buffer pH 6.5 for 15 min and then washed. To achieve near-physiological conditions during perfusion of the pFNcoated flow chamber slides, 2 ml of washed platelets with a platelet count of 1 Â 10 7 were combined with 1 ml of human erythrocytes isolated from the blood of a healthy volunteer.
For each experiment, 4 channels of a flow chamber slide (m-Slide VI 0.1 ibiTreat, Ibidi) were coated with 10 mg/ml fibrillar collagen, fibrinogen, pFN wt or pFN syn over night at 4˚C and blocked with 1% BSA the following day. The coated channels of one m-slide were connected in series with connector tubings for simultaneous perfusion. The platelet suspension was filled in one reservoir of a Perfusion Set Black (Ibidi) and the pump was started with unidirectional flow at the highest possible pressure 100 mbar) until all channels were filled with the blood-like fluid. Then, the experiment was started by adjusting the shear rate to approximately 1000/s. The channels were perfused for 10 min and subsequently washed by perfusing Tyrodes buffer for another 10 min. Platelets were imaged after performing the perfusion with a Zeiss Apotome microscope and platelet surface coverage was analysed using ImageJ.

Microvascular thrombus formation
The surgical preparation of the mouse cremaster muscle was performed as described (Baez, 1973). Mice were anesthetized using a mixture of 100 mg/kg ketamine and 10 mg/kg xylazin. The left femoral artery was cannulated in a retrograde manner to administer FITC-labeled dextran (MW 150 kDa; Sigma Aldrich). The right cremaster muscle was exposed through a ventral incision of the scrotum. The muscle was opened ventrally in a relatively avascular zone and spread over the pedestal of a custom-made microscopy stage. Epididymis and testicle were detached from the cremaster muscle and placed into the abdominal cavity. Throughout the surgical procedure and in vivo microscopy, the muscle was superfused with warm saline solution. At the end of each experiment, blood samples were collected by cardiac puncture to determine systemic cell counts using a hematology analysis system (ProCyte DX, IDEXX Laboratories ).
Microvascular thrombus formation was induced by phototoxic injury as described (Rumbaut et al., 2005) with slight modifications. Briefly, after surgical preparation of the cremaster muscle, 4 ml/kg body weight of a 2.5% solution of FITC-dextran was infused intraarterially and the exposed the vessel segment under investigation was continuously epi-illuminated with a wavelength of 488 nm (Polychrome II, TILL Photonics, Grä felfing, Germany). An Olympus water immersion lens (60 Â /NA 0.9) in an upright microscope (BX50; Olympus Microscopy, Hamburg, Germany) was used to focus the light onto the cremaster muscle and to visualize the microvascular thrombus formation in real-time. Thrombus formation was induced in one arteriole (25-35 mm) per experiment by analyzing the time until the first platelet adhesion to the vessel wall (defined as the onset of thrombus formation) occurred and the time until blood flow ceased (defined as the complete occlusion of the vessel).

Tail bleeding assay
The tail bleeding assay was performed in anesthetized mice directly after the analysis of microvascular thrombus formation. For this purpose, the distal 2 mm segment of the tail was removed with a scalpel. Bleeding was monitored by absorbing the bead of blood with a filter paper in 30 s intervals without touching the wound. Tail bleeding time was defined as the time until hemorrhage ceased.