Plasma fibronectin depletion enhances platelet aggregation and thrombus formation in mice lacking fibrinogen and von Willebrand factor

Platelet adhesion and subsequent aggregation at the site of vascular injury are key events required for hemostasis. However, excessive platelet accumulation and thrombus formation may result in myocardial infarction or stroke, the 2 leading causes of morbidity and mortality worldwide.^1,2  It has been shown that fibrinogen (Fg) and von Willebrand factor (VWF) are the 2 key molecules required for platelet adhesion and aggregation.³  Interestingly, we found that platelet aggregation and thrombus formation still occur in mice lacking Fg or VWF or both,^4,5  suggesting that additional molecule(s) can promote platelet aggregation and thrombus formation.

We subsequently found that platelet fibronectin (Fn) content was markedly increased in Fg-deficient mice⁴  and a severe hypofibrinogenemic human patient,⁶  although no alteration of plasma Fn (pFn) was observed in their Fg-deficient blood. The increase in platelet Fn content was found to be due to enhanced pFn internalization via β₃ integrin.⁷  Further experiments with pFn conditional deficient mice and Fn heterozygous mice showed that pFn promoted thrombus growth and stability in injured arterioles.^8,9  These data are consistent with recent in vitro studies that showed that Fn assembly on the platelet surface,^10,11  supported thrombus growth.^12-15  However, the role of pFn in Fg/VWF-independent platelet aggregation and thrombus formation has not been studied. It was expected that pFn may be required for these processes.

To address this question, we generated Fg, VWF, and pFn triple-deficient (TKO) mice by breeding Fg and VWF double-deficient (Fg/VWF^−/−) mice⁴  with pFn conditional-deficient (Fn^flox/flox, Cre^+/−) mice.¹⁶  pFn depletion was induced by intraperitoneal injections of polyinonic-polycytidylic acid in Cre⁺ TKO (Cre⁺, Fn^flox/flox, Fg/VWF^−/−) mice but not Cre⁻ control (Cre⁻, Fn^flox/flox, Fg/VWF^−/−) littermates. Unexpectedly, we found that platelet aggregation and thrombus growth were not abolished, but rather they were enhanced in these TKO mice compared with their littermate controls. These data indicate that pFn is not required for, but inhibits, platelet aggregation and thrombogenesis in the absence of Fg and VWF. We also found that pFn is able to inhibit wild-type mouse platelet aggregation. We propose that soluble pFn is an important inhibitory factor in platelet aggregation; however, insoluble pFn (formed either by covalent linkage between pFn and fibrin or VWF or by self-assembly) may support thrombus growth. The possible mechanisms that control this pFn functional switching and its implications in clinical patients are discussed.

Fg and VWF double-deficient (Fg/VWF^−/−) mice, pFn conditional deficient (Fn^flox/flox, Cre^+/−) mice, β₃ integrin–deficient (β₃^−/−), and GPIbα-deficient (GPIbα^−/−) mice were described before.^4,16-18  β₃^−/− mice and GPIbα^−/− mice were kindly provided by Dr Richard O. Hynes (Howard Hughes Medical Institute, Massachusetts Institute of Technology, Boston, MA), Dr Jerry Ware, and Dr Zaverio M. Ruggeri (The Scripps Research Institute, La Jolla, CA). Syngeneic BALB/c mice and C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME).

All mice were housed in the research vivarium of St Michael's Hospital in Toronto, and the experimental procedures were approved by the Animal Care Committee.

β3^−/− mice and GPIbα^−/− mice were backcrossed onto the BALB/c background at the Massachusetts Institute of Technology and St Michael's Hospital, respectively. Polyclonal antibodies against β₃ integrin and GPIbα were generated in these mice by transfusion of syngeneic wild-type platelets as previously described.¹⁹  The hybridoma of M1 anti–β₃ integrin monoclonal antibody was generated by fusion of immunized β₃^−/− splenocytes with Ag8.653 myeloma cells (ATCC, Manassas, VA). The effect of monoclonal antibody M1 (8 μg/mL) on platelet aggregation was determined by aggregometry as previously described.⁵ 

The Fg/VWF/pFn^−/− TKO mice were generated by breeding male Fg/VWF^−/− mice with pFn conditional knockout (Fn^flox/flox, Cre^+/−) female mice. Because 4 genes are involved in the generation of these TKO mice, heterozygous mice generated from the first round of breeding were bred, generating Cre^+/−, Fn^flox/flox, Fg/VWF^−/− background male mice and Cre^+/−, Fn^flox/flox, Fg^+/−, VWF^−/− female mice. The Fg heterozygous female (Cre^+/−, Fn^flox/flox, Fg^+/−, and VWF^−/−) mice were used for breeding because female Fg-deficient (Fg^−/−) mice die from uterine bleeding during pregnancy.²⁰  These mice were then bred to generate Cre⁺, Fn^flox/flox, Fg/VWF^−/− mice and Cre⁻, Fn^flox/flox, Fg/VWF^−/− littermates (negative controls). pFn depletion was induced by 3 intraperitoneal injections of 250 μg of polyinonic-polycytidylic acid (polyI-polyC; Sigma-Aldrich, St Louis, MO) into both the Cre⁺ and Cre⁻ control mice at 2-day intervals as previously described.^8,16  Mouse genotypes were detected by polymerase chain reaction and confirmed by Western blot, immunofluorescence microscopy, or both.

Gel-filtered platelets were prepared as previously described.^4,5,7,21  Mice (6-8 weeks old) were anesthetized and bled from the retroorbital plexus with the use of heparin-coated glass capillary tubes. The blood was collected into a tube containing 3% ACD (1/9, vol/vol). Platelet-rich plasma (PRP) was obtained by centrifugation at 300g for 7 minutes. Platelets were then isolated from the PRP with the use of a Sepharose 2B column in PIPES buffer (PIPES 5 mM, NaCl 1.37 mM, KCl 4 mM, glucose 0.1%, pH 7.0). Platelet-poor plasma (PPP; from either citrated or heparinized PRP) was prepared by centrifugation at 1500g for 20 minutes. The PPP was further centrifuged at 10 000g for 5 minutes to remove the remaining cells.

The plasma and platelet Fn levels in Cre⁺ TKO and Cre⁻ control mice were determined by Western blot analysis with the use of a rabbit anti–human Fn antibody, which cross reacts with mouse Fn as previously described.⁸  Briefly, 1 μL of plasma samples or 1 × 10⁸ gel-filtered platelet lysates were separated on a 7% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gel and transferred to Immobilon-P PVDF membrane (Millipore, Bedford, MA). Fn was detected by using a 1:500 dilution of the rabbit anti-Fn antibody followed by a 1:20 000 dilution of alkaline phosphatase–conjugated goat anti–rabbit polyvalent IgG (StressGen, Victoria, BC). Membranes were developed with the use of 5-bromo-4-chloro-3-indolyl phosphate nitroblue tetrazolium purple substrate (Sigma-Aldrich) to visualize immonoreactive bands. Results were scanned using a Canoscan D2400U Photo scanner (Canon, Beijing, China). Fn depletion was determined by comparing the density of the Fn bands using Adobe Photoshop 7.0 Limited Edition (Adobe Systems, San Jose, CA).

Cre⁺ and Cre⁻ platelets were examined for platelet surface expression of β₁ integrin, β₃ integrin, GPIbα, and P-selectin. Resting gel-filtered platelets (10⁶) from Cre⁺ TKO and Cre⁻ mice were incubated separately with a 1:100 dilution of FITC-conjugated hamster anti–rat CD29 (anti–β₁ integrin; BD PharMingen, San Jose, CA) or a 1:100 dilution of PE-conjugated rat anti–mouse CD61 (anti–β₃ integrin; BD PharMingen, Missisauga, ON). GPIbα was detected by using 10 μg/mL rat anti–mouse GPIbα antibody p0p5 and subsequently incubated with a 1:100 dilution of FITC-conjugated anti–rat IgG (Sigma-Aldrich). Platelet surface P-selectin expression was detected on resting or thrombin (1 U/mL)–activated platelets after incubation with FITC-conjugated rat anti–mouse CD62P (P-selectin; BD PharMingen Canada). To examine the activation of β₃ integrin, thrombin (1 U/mL)– or ADP (20 μM)–treated platelets were incubated with R-phycoerythrin–conjugated JON/A antibody (JON/A-PE; EMFRET Analytics, Eibelstadt, Germany), an antibody that recognizes only activated mouse αIIbβ₃.²²  All samples were analyzed by a FACScan flow cytometer (Becton Dickinson, San Jose, CA).

To detect peripheral blood cell counts in mice, 20 μL of whole blood was collected into an EDTA-coated Microvette from each polyI-polyC–treated Cre⁺ TKO and Cre⁻ control mouse. Peripheral blood cell counts were analyzed with the use of a HEMAVET HV950FS (Drew Scientific, Oxford, CT).

Platelets were prepared and aggregation was assayed as described previously.^5,21  TKO and Cre⁻ control PRP and PPP were obtained by centrifugation of heparin (25 IU/mL) anticoagulated whole blood. Platelet concentration in PRP was adjusted to 3 × 10⁸ platelets/mL with the use of autologous PPP, and aggregation was induced by various agonists, including 500 μM thrombin receptor activating peptide (TRAP; AYPGKF-NH2; Peptides Int, Louisville, KY), 20 μg/mL collagen (Nycomed Pharma, Ismaning, Germany), 20 μM ADP (Sigma-Aldrich), 20 μM U46619 (Cayman Chemical, Ann Arbor, MI). We also tested TKO and Cre⁻ control platelet aggregation after switched plasma from the opposite genotype. Briefly, gel-filtered TKO and Cre⁻ control platelets were suspended in PPP (3 × 10⁸ platelets/mL) from either the same genotype or from the opposite genotype. Platelet aggregation was then induced by 500 μM TRAP. In another set of experiments, gel-filtered platelet aggregation in PIPES buffer was induced by ADP, U46619, and different doses of TRAP, murine thrombin (Sigma-Aldrich), or collagen, as previously described.⁵  The change in light transmission was monitored and recorded with the use of a computerized Chrono-log aggregometer (Chrono-Log, Havertown, PA) with the stir bar rate set to 1000 rpm at 37°C. The images of platelet aggregates were recorded using a digital camera (DP70; Olympus, Tokyo, Japan) under a Zeiss Axiovert 135 inverted fluorescent microscope (Zeiss, Oberkochen, Germany).

We also performed platelet aggregation after adding purified murine plasma Fn (50-200 μg/mL; Molecular Innovations, Southfield, MI), or our newly developed mouse anti–mouse β₃ integrin antisera and monoclonal antibody (8 μg/mL; clone M1),²³  and mouse anti–mouse GPIbα antisera. A functional blocking rat anti–mouse GPIbα monoclonal antibody (40 μg/mL; clone Xia.B2; EMFRET Analytics) was also used to test its effect on TKO platelet aggregation.

A 35-mm Petri dish was coated with acid-soluble human fibrillar type III (type X) collagen (1 mg/mL; Sigma-Aldrich) overnight at 4°C as previously described.⁴  A parallel plate flow chamber (Glycotech, Rockville, MD) with a silicone rubber gasket was placed on the collagen-coated Petri dish. Heparin (25 IU/mL) anticoagulated whole blood was fluorescently labeled with the use of DiOC₆ (1 μM; Sigma-Aldrich) for 10 minutes at 37°C and 1 mL of the labeled blood was perfused over the collagen-coated surface under a controlled flow rate (shear rate of 500 seconds⁻¹) with the use of a syringe pump (Harvard Apparatus, Holliston, MA). The blood was perfused for 4 minutes followed by 2 minutes of perfusion with a rinsing buffer (NaCl 130 mM, KCl 2 mM, NaHCO₃ 12 mM, CaCl₂ 2.5 mM, MgCl₂ 0.9 mM, glucose 5 mM, pH 7.4) at 37°C. Platelet adhesion, aggregation, and thrombus formation were monitored under Zeiss Axiovert 135 inverted fluorescent microscope (32×/0.4 NA; Zeiss), and pictures were taken using the DP70 digital camera (Olympus). Three randomly selected fields were recorded in each experiment. The microscope settings, including contrast, brightness, and magnification, were kept constant to facilitate comparison between different experiments. The images of adherent platelets were recorded, and the percentage of surface coverage by platelets was analyzed by computer (IBM IntelliStation Z Pro; IBM, New York, NY) with the use of Slidebook software (Intelligent Imaging Innovations, Denver, CO).

The ex vivo perfusion chamber thrombosis model was also performed at high shear rate (1800 seconds⁻¹) with the use of a method previously described.²⁴  Briefly, rectangular glass microcapillary tubes (height 0.1 × width 1.0 × length 100 mm; Vitro Dynamics, Rockaway, NJ) were coated with 100 μg/mL type I collagen fibrils (equine collagen Horm; Nycomed, Roskilde, Denmark) overnight at 4°C. Fluorescently labeled whole blood was perfused through the collagen-coated microcapillary tubes for 4 minutes. Platelet adhesion, aggregation, and thrombus formation were recorded in real time over the course of perfusion under a Zeiss Axiovert 135 inverted fluorescent microscope by a computer (IBM IntelliStation Z Pro) using the Slidebook program (Intelligent Imaging Innovations). After 4 minutes of perfusion, 3-dimensional images of thrombi were captured with the use of a confocal cell imaging system (CARV; Atto Bioscience, Rockville, MD) attached to the microscope, and volumes of thrombi were analyzed using the Slidebook program (Intelligent Imaging Innovations).

The whole process of thrombus formation in arterioles (shear rates are 1466 ± 74 seconds⁻¹ in TKO mice and 1441 ± 62 seconds⁻¹ in Cre⁻ littermates; P > .05) was monitored in 3- to 4-week-old Cre⁺ TKO and Cre⁻ control mice injected with donor-matched fluorescently labeled platelets under a Zeiss Axiovert 135-inverted fluorescent microscope (Zeiss).^4,21  Injury was induced by topical application of 30 μL of 250 mM FeCl₃, and the characteristics of thrombus formation were compared based on (1) number of fluorescent platelets deposited on the vessel wall during the first 3 to 5 minutes after injury, (2) time required for the formation of the first 20-μm thrombus, and (3) time to complete vessel occlusion.^4,7,21,25 

To quantitatively study platelet accumulation within growing thrombi after the vascular injury in vivo, we used a laser to induce an arteriole thrombus in the cremaster muscle as previously described.^21,26  Briefly, male adult mice were anesthetized, and a trachea tube was inserted to facilitate breathing. Antibodies and anesthetic reagent (pentobarbital; Abbott Laboratories, Toronto, ON; 0.05 mg/kg) were administered by a jugular vein cannula. The cremaster muscle was prepared under a dissecting microscope and superfused throughout the experiment with preheated bicarbonate-buffered saline. Platelets were labeled by injecting a rat anti–mouse CD41 antibody (Leo.A1; EMFRET Analytics; 0.1 μg/g) detected with Alexa-660–conjugated goat anti–rat IgG (Invitrogen, Carlsbad, CA; 2 mg/mL). Multiple independent upstream injuries were performed on a cremaster arteriole with the use of an Olympus BX51WI microscope with a pulsed nitrogen dye laser. A total of 20 thrombi in 2 Cre⁺ TKO mice and 21 thrombi in 2 Cre⁻ control mice were studied. The dynamic accumulation of fluorescently labeled platelets within the growing thrombus was captured and analyzed using Slidebook software (Intelligent Imaging Innovations).

Data are presented as mean plus or minus SEM. Statistical significance was assessed by Student unpaired t test and χ² test.

To examine whether pFn is the key molecule mediating Fg/VWF-independent platelet aggregation and thrombosis, we generated Cre⁺, Fn^flox/flox, Fg/VWF^−/− and Cre⁻, Fn^flox/flox, Fg/VWF^−/− mice. To minimize genetic background differences, mice were delivered from Cre^+/− parents, and both Cre⁺ and Cre⁻ littermates were injected with polyI-polyC. Depletion of the Fn^flox/flox gene by Cre only occurred in Cre⁺, Fn^flox/flox, Fg/VWF^−/− mice but not in the Cre⁻, Fn^flox/flox, Fg/VWF^−/− littermate controls. As shown in Figure 1A, Western blot analysis showed that the pFn in Cre⁺ TKO mice was decreased greater than 98.8% and was nearly undetectable after 3 injections of polyI-polyC. Absence of Fg and VWF was also confirmed by both DNA genotyping (polymerase chain reaction) and Western blot (data not shown).

We previously showed that Fg is a competitive ligand of platelet β₃ integrin during pFn internalization.⁷  In the absence of Fg or both Fg and VWF, platelet Fn content is increased 3- to 5-fold.^4,6  It is therefore interesting to examine whether platelet β₃ integrins, in the absence of both Fg and VWF, are still able to capture and internalize the remaining pFn in TKO mice. As shown in Figure 1B, although some residual platelet Fn is still detectable, the platelet Fn content in Cre⁺ mice was markedly decreased (>80%) compared with Cre⁻ littermates. The residual of platelet Fn (probably a cellular form of Fn originating from megakaryocytes) in TKO mice is similar to that we observed in pFn single-deficient mice.^8,9  There is no significant difference in platelet Fn levels between Cre⁺ and Cre⁻ control platelets before injections of polyI-polyC (data not shown). This further showed that depletion of pFn was successful in the Cre⁺ mice, and platelet β₃ integrins are not able to internalize the trace amounts of pFn in TKO blood even without Fg competition. Therefore, we defined the Cre⁺, Fn^flox/flox, Fg/VWF^−/− mice as triple deficient (TKO) mice and Cre⁻, Fn^flox/flox, Fg/VWF^−/− littermates with normal plasma Fn levels as negative controls throughout this study.

It is notable that there was no detectable alteration in platelet content of vitronectin and thrombospondin-1 in TKO platelets compared with platelets from Cre⁻ control littermates after polyI-polyC treatment (data not shown).

To exclude intrinsic changes in platelet function after depletion of pFn, we examined several major adhesion receptors on Cre⁺ TKO and Cre⁻ control platelets by flow cytometry. As shown in Figure 2A, the expression levels of GPIbα and β₁(CD29) and β₃(CD61) integrins are almost identical. We also examined P-selectin (CD62P) expression (Figure 2A) and the active conformation of αIIbβ₃ integrin in resting and ADP (20 μM)– and thrombin (1 U/mL)–activated platelets (Figure 2B). No difference was found between Cre⁺ TKO platelets and the Cre⁻ control platelets. There is also no detectable difference in expression levels of these adhesion molecules on platelets between wild-type and TKO mice, except for P-selectin, which is decreased when Fg is absent (H.Y., S. Lang, Z. Zhai, L. Li, W. Kahr, P. Chen, C.M.S., A.R., M. Flick, J. Degen, J.F., and H.N., manuscript submitted). We did not observe significant alterations (data not shown) in peripheral blood cell counts in Cre⁺ TKO mice recruited in this study compared with their Cre⁻ littermates.

Surprisingly, we found that platelet aggregation in TKO (Cre⁺) PRP was not abolished, but rather it was significantly enhanced compared with Cre⁻ control PRP when stimulated with TRAP (P < .05). Gel-filtered Cre⁺ TKO platelet aggregation in PIPES buffer was also enhanced when stimulated with either thrombin (1 U/mL; P < .05; Figure 3A) or TRAP (500 μM; data not shown). The TKO platelet aggregates were larger with fewer individual platelets remaining in the PRP or PIPES buffer (Figure 3A). The trend toward enhancement of platelet aggregation was also reproduced when various concentrations of TRAP (from 125 μM to 1 mM) and thrombin (from 0.25 IU to 5 IU) were used (data not shown; no aggregation was induced when ≤ 62.5 μM TRAP or ≤ 0.125 IU thrombin was used in this study). The observed enhanced platelet aggregation was reproducible when collagen was used to induce platelet activation (Figure 3B). However, neither ADP nor U46619 was able to induce TKO or Cre⁻ control platelet aggregation in PRP or gel-filtered platelets in PIPES buffer (Figure 3B). This is consistent with our previous study, which indicated that Fg is required for ADP-induced platelet aggregation in routine anticoagulated blood, and released platelet granule proteins are essential for Fg-independent platelet aggregation.⁵ 

The enhancement of platelet aggregation in TKO mice was diminished when Cre⁺ TKO platelets were aggregated in pFn-positive control PPP, whereas enhanced aggregation was observed when control platelets were aggregated in pFn-depleted TKO plasma (Figure 3C). These data exclude the possibility of TKO platelet hyperactivity during platelet aggregation. Only partial correction was observed with respect to the enhanced platelet aggregation in these plasma switch experiments, potentially reflecting the inhibitory effect of α-granule pFn in Cre⁻ control platelets. This internalized platelet pFn may be released from platelets and inhibit platelet aggregation, as observed in gel-filtered Cre⁻ platelet aggregation (Figure 3A). In addition to inhibiting platelet aggregation in TKO mice, we also found that exogenous pFn significantly inhibited both thrombin-induced (Figure 3D) and TRAP-induced (data not shown) gel-filtered wild-type mouse platelet aggregation in a dose-dependent manner. This is consistent with earlier studies in wild-type rat and pig and in healthy human platelets.^27-30 

To examine whether β₃ integrin is essential for TKO platelet aggregation, we generated mouse anti–mouse β₃ integrin antisera and developed a mouse anti–mouse β₃ integrin monoclonal antibody (M1) with the use of β₃ integrin^−/− mice. High-titer (> 1:1600) mouse anti–mouse GPIbα antisera were also generated in GPIbα^−/− mice. The anti–β₃ integrin antisera and M1 are specific to β₃ integrin and are able to inhibit wild-type mouse platelet aggregation (data not shown). We found that TKO platelet aggregation can be completely blocked by this anti–mouse β₃ integrin antisera and the monoclonal antibody M1, but it was not affected by polyclonal anti–mouse GPIbα antisera and a functional blocking rat anti–mouse GPIbα monoclonal antibody (Figure 3E) in both PRP and gel-filtered platelets in PIPES buffer. This indicates that β₃ integrin, but not GPIbα, is required and that other ligands of β₃ integrin may mediate Fg/VWF/pFn-independent platelet aggregation.

To examine platelet adhesion and aggregation under flow conditions, heparin-anticoagulated whole blood from TKO mice or Cre⁻ control mice was perfused over collagen-coated surfaces in laminar perfusion chambers. Consistent with data from in vitro platelet aggregation, significantly more platelet thrombi formed when TKO blood was perfused at a shear rate of 500 seconds⁻¹. Platelet surface coverage was significantly higher in TKO blood (TKO, 14.50% ± 1.8%, Cre⁻ control, 6.2% ± 0.8%; P < .001; Figure 4A,). Because of the absence of VWF, platelet adhesion and aggregation at high shear (1800 seconds⁻¹) was significantly decreased in both TKO and Cre⁻ controls, but enhancement of platelet aggregation in TKO mice was still evident (TKO, 3.9 ± 0.3, Cre⁻ control, 2.2 ± 0.3; P < .01; Figure 4B,).

To test the effects of pFn depletion on Fg/VWF^−/− thrombus formation in vivo, we used 2 intravital microscopy models in different vascular beds. In the FeCl₃-injury mesenteric arteriole thrombus model, we studied thrombosis in response to injury in a total of 9 Cre⁺ TKO mice and 10 Cre⁻ Fg/VWF^−/− control mice. As shown in Figure 5A and B, the number of single fluorescently labeled adherent platelets per minute, determined at 3 to 5 minutes after vessel injury, was not significantly different between TKO and Cre⁻ control mice (Cre⁺ TKO, 80.75 ± 25.1 minute⁻¹; Cre⁻ control, 61.5 ± 24.9 minute⁻¹; P > .05). However, in TKO mice, the formation of the first thrombus greater than 20 μm in diameter was faster (Cre⁺ TKO, 10 ± 0.5 minutes; Cre⁻ control, 13.1 ± 1.1 minute; P < .05) and more stable, and the time required for complete vessel occlusion was significantly shorter (Cre⁺ TKO, 25.9 ± 2.8 minutes; Cre⁻ control, 34.1 ± 2.3 minutes; P < .05) compared with Cre⁻ controls. The thrombus did not result in vessel occlusion in 5 (50%) of 10 Cre⁻ control mice within a 40-minute recording time after vascular injury, whereas vessel occlusion occurred in 7 (78%) of 9 TKO mice, although a statistically significant difference was not reached (χ² test = 1.569; P > .05).

We also used a laser-injury cremaster muscle arteriole thrombosis model to quantitatively compare platelet accumulation within growing thrombi in TKO and Cre⁻ control mice. We found that thrombi in Cre⁻ control mice appeared fragile and unstable; frequent embolization and detachment of whole thrombi was evident. In sharp contrast, thrombi in TKO mice exhibited rapid platelet accumulation within the growing thrombus. The total amount of platelets within the Cre⁺ TKO thrombus was significantly greater and the rate of decrease in amount of platelets after maximal platelet deposition was lower compared with Cre⁻ control thrombi (P < .05; Figure 6).

It has been suspected for decades that pFn may contribute to thrombosis and hemostasis. However, its precise role in these processes has not been well developed. We previously showed that platelet aggregation and occlusive thrombi occurred in mice lacking Fg or VWF or both.^4,5  In these Fg-deficient animals and a severely hypofibrinogenemic human patient,⁶  platelet Fn content was markedly increased. It was speculated that pFn may be required for platelet aggregation and thrombus formation in the absence of Fg and VWF. To address this question, we generated TKO (Fg/VWF/pFn^−/−) mice in the present studies. We found these mice are viable. Interestingly, TKO platelets did not fail to aggregate, but they exhibited enhanced aggregation in both PRP and gel-filtered platelets in PIPES buffer. The enhancement of platelet aggregation was reproducible in perfusion chambers at different shear stresses. Furthermore, enhanced thrombogenesis in TKO mice was shown in both FeCl₃-injured mesenteric arterioles and laser-injured cremaster arterioles with the use of intravital microscopy thrombosis models. These data suggest that pFn is not required for, but inhibits Fg/VWF-independent platelet aggregation and thrombus formation.

The role of Fn in platelet aggregation is inconclusive and controversial, although pFn is a dimer that contains 2 RGD sites that could potentially cross-link adjacent platelets. Early experiments suggested that pFn may support platelet aggregation,³¹  and an anti-Fn monoclonal antibody decreased platelet aggregation.³²  Subsequent studies, however, showed that pFn may have either no effect or an inhibitory effect on platelet aggregation.^27-29  The inhibitory effect was thought to be due to pFn competition with more potent integrin ligands such as Fg and VWF.

It has been speculated that pFn may be the bridging molecule for platelet aggregation in the absence of Fg or of both Fg and VWF. In Fg/VWF^−/− mice, we found that (1) platelet Fn content increased 3- to 5-fold, (2) platelet-rich thrombi still formed in vivo after vascular injury, and (3) platelet aggregation can be induced in vitro under more physiologic conditions by strong agonists such as thrombin, TRAP, collagen, A23187, and phorbol myristyl acetate,⁵  indicating that pFn may play a supportive role in platelet aggregation. This hypothesis was further supported by our in vivo study showing that pFn promotes thrombus growth⁸  and studies showing that Fn assembly on the platelet surface^10,11  supports thrombus growth in ex vivo perfusion chamber experiments.^12-15,33  However, there is no direct evidence indicating that pFn supports Fg/VWF-independent platelet aggregation. The question cannot be addressed simply by removing Fn from the plasma during aggregation assays because Fn released from platelets may also contribute to platelet aggregation, and greater than 80% of platelet Fn comes from pFn internalization by β₃ integrin.⁷  TKO (Fg/VWF/pFn^−/−) mice are therefore required to answer this question.

The enhancement of TKO platelet aggregation both in the presence and absence of plasma was unexpected. After blindly repeating these results several times, we asked whether adhesion molecules potentially involved in platelet aggregation were up-regulated on TKO platelets. However, this is probably not the answer because the expression levels of β₃ and β₁ integrins, GPIbα, and P-selectin are almost identical between TKO and Cre⁻ control platelets (Figure 2A). The active conformations of αIIbβ₃ integrin, detected by a conformation sensitive monoclonal antibody JON/A, in resting and ADP- and thrombin-activated platelets are also comparable (Figure 2B). We further performed platelet aggregation after switching their plasma and found that aggregation of Cre⁻ littermate platelets was enhanced in TKO PPP, and aggregation of TKO platelets was decreased in pFn-positive plasma (Figure 3C). These results indicate that the observed enhanced platelet aggregation in TKO mice is unlikely due to an alteration in platelet surface adhesive receptors or other possible mechanisms that lead to hyperactivation of TKO platelets. Because enhanced platelet aggregation was also found in gel-filtered platelets in PIPES buffer, we therefore conclude that pFn, including internalized platelets pFn, may not be the bridging molecule required for Fg/VWF-independent platelet aggregation, but rather act as an inhibitory factor in this process.

Note that we also found that pFn inhibited wild-type gel-filtered platelet aggregation, which is consistent with earlier studies from other species, including healthy human platelets.^27-30  Although the mechanism is not fully understood, it is possible that pFn may inhibit β₃ integrin–mediated platelet aggregation by the same pathway as plasma vitronectin²¹  (ie, probably by RGD occupancy of the β₃ integrin ligand binding site). This occupancy may sterically prevent the bridging ligand(s) from cross-linking adjacent platelets. In addition, it is currently unknown whether soluble pFn can also deliver negative signals to platelets or attenuate platelet activation through blocking integrin outside-in signals delivered by more potent β₃ integrin ligand(s). We also cannot exclude the possibility that thrombospondin-1 is able to more efficiently block the antithrombotic activity of nitric oxide/cGMP signaling³⁴  in the absence of pFn. These questions should be of interest for further studies.

TKO platelet aggregation was completely blocked by our newly developed mouse anti–mouse β₃ integrin antisera and monoclonal antibody M1 (Figure 3E). However, aggregation was not affected by our anti–mouse GPIbα antisera and a functional blocking rat anti–mouse GPIbα monoclonal antibody. These data indicate that β₃ integrin, but not GPIbα, is essential for TKO platelet aggregation. They also exclude the possibility that mutual interactions, including indirect binding, between these 2 most abundant glyco-proteins (ie, β₃ integrin and GPIbα) on the platelet surface may mediate this aggregation, although engagement between β₂ integrin (Mac-1) and GPIbα occurs under certain circumstances.³⁵  Because TKO platelet aggregation cannot be induced by either ADP or U46619 (Figure 3B), other ligand(s) of β₃ integrin released from platelets may mediate Fg/VWF/pFn-independent platelet aggregation. However, it is currently unknown whether EDA domain-positive cellular Fn³⁶  (a megakaryocyte synthesized form of Fn present in platelets in a small proportion; 10%-20%) is necessary for this process. It is probably that the cellular Fn and other adhesion ligands (such as thrombospondin-1, vitronectin, and CD40L or combinations thereof)^21,34,37-42  synergistically contribute to this Fg/VWF/pFn-independent platelet aggregation.

Our data from ex vivo perfusion chambers and the 2 intravital microscopy models are consistent with our results from in vitro platelet aggregation assays (Figure 3). However, the enhancement of thrombus growth in these ex vivo and in vivo models may not only be due to the enhanced platelet aggregation, but also to enhanced platelet adhesion. There are significantly more adhesive platelets in the perfusion chamber experiments (Figure 4A), and there is also a trend toward increased platelet deposition at the site of vessel wall injury at the early stages of thrombus formation (Figure 5B). It is currently unknown whether this is due to pFn inhibition of platelet interactions with either collagen,²⁸  or to other subendothelial matrix proteins.

It is interesting that our results from the present studies contrast with our previous studies that used conditionally deficient pFn mice (single deficiency)⁸  and studies of Fn assembly.^11-13,15  One possible explanation is that Fg and VWF may be required for pFn to promote thrombus growth. Our recent finding that fibrin formation is important for Fn retention on the human platelet surface,⁶  together with findings from perfusion chamber studies that thrombogenesis is enhanced by Fn assembly in platelet thrombi by cross-linking to fibrin,¹³  supports this explanation. It is currently unclear whether Fg or VWF or both (in plasma or that released from activated platelets) are required for Fn assembly on aggregated platelets in vivo. However, it is likely that fibrin and subsequent covalently linked pFn-fibrin matrix formation at the site of vascular injury may promote thrombus growth. This may explain why impaired thrombus formation was observed in pFn single-deficient mice with the use of the intravital microscopy model⁸  because significant thrombin and fibrin were generated after vascular injury in the system.^4,7,21,25  The levels of severity of vascular injury, local concentration of thrombin, as well as other enzymes, including factor XIIIa and the possible protein disulfide isomerases (PDIs)^43,44  that enhance insoluble pFn matrix formation, may contribute to switching the function of pFn from an inhibitory role in platelet aggregation to a supportive effect on thrombus growth. The intrinsic PDI activities of both VWF propolypetide and pFn may also facilitate pFn insolubilization.^45,46  Understanding the mechanisms of this pFn functional “switch” may provide insights into the elevated pFn levels in patients with acute myocardial infarction and venous thromboembolism^47-51  and elucidate whether increasing pFn is a potential protective mechanism in these common and critical human diseases.

Another challenging question is whether increased platelet Fn content in Fg-deficient platelets plays a supportive role in hemostasis. Here, we found that pFn, including platelet pFn internalized from the plasma, inhibits platelet aggregation (an important hemostatic event). However, it is possible that high amounts of Fn released from adherent platelets at the site of vascular injury may dramatically boost the local pFn concentration, which facilitates Fn self-and nonself assembly on the injured vessel wall. These pFn-mediated extracellular matrix-like fibrils may significantly contribute to hemostasis.

In summary, we generated Fg/VWF/pFn^−/− triple-deficient mice. We clearly show that pFn is unlikely the key molecule that mediates Fg/VWF-independent platelet aggregation. In contrast, soluble pFn is a significant inhibitory factor during thrombus formation. We propose that pFn may play dual roles in thrombosis and hemostasis. By switching its soluble form (inhibitory) to insoluble extracellular matrix-like fibrils (supportive) by either self-assembly or nonself covalent linkage with fibrin or other matrix proteins, pFn sophisticatedly controls the vital physiologic and pathologic processes (ie, hemostasis and thrombosis).

We thank Drs Reinhard Fässler, Jay L. Degen, Denisa D. Wagner, and Cecile V. Denis for their earlier work preparing pFn conditional knockout mice, Fg and VWF gene-deficient mice. We thank Dr Pingguo Chen for his assistance with genotyping and backcrossing GPIbα^−/− and β₃^−/− mice onto BALB/c background and Michelle Lee Webster and Sean Lang for assistance with preparation of the manuscript.

This work was supported by the Heart and Stroke Foundation of Canada (Ontario), Canadian Blood Services and Canadian Institutes of Health Research, and by equipment funds from St Michael's Hospital, Canadian Blood Services, and Canada Foundation for Innovation. H.Y. is a recipient of the Heart and Stroke/Richard Lewar Excellence Award and a Canadian Blood Services postdoctoral fellowship award.

Contribution: A.R. designed the experiments, performed the research, analyzed the data, and wrote part of the manuscript; H.Y., G.Z., W.J., F.H., and X.B. performed the research; C.M.S. performed the research and edited the manuscript; P.L.G. analyzed the data and edited the manuscript; J.F. contributed a vital analytical tool (flow cytometer), analyzed the data, and edited the manuscript; and H.N. is the principal investigator who designed the experiments, analyzed the data, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Heyu Ni, Canadian Blood Services and Department of Laboratory Medicine and Pathobiology, St Michael's Hospital, University of Toronto, 30 Bond Street, Room 2-006, Bond Wing, Toronto, ON, Canada M5B 1W8; e-mail: nih@smh.toronto.on.ca.