Loss of matrix metalloproteinase 2 in platelets reduces arterial thrombosis in vivo

Zymography of plasma, platelet lysates, and lung extracts confirmed that MMP-2 is not expressed in MMP-2 knockout mice. On the contrary, in WT mice MMP-2 was detectable by zymography in all three types of biological samples (unpublished data). Pro–MMP-2 was present in platelet lysates (43.9 ± 2.1 ng/10⁸ platelets) from WT mice, and thrombin, collagen, or ADP induced a rapid increase in the amount of active MMP-2 in the supernatant to a maximal value of 36.4 ± 2.5 ng/10⁸ platelets. As expected, active MMP-2 was not detected in the supernatant of platelets from MMP-2^−/− mice (unpublished data).

RT-PCR of spleen extracts confirmed that WT mice, but not MMP-2^−/− mice, contain MMP-2 messenger RNA (mRNA). MMP-2 mRNA was reduced by 51% in spleen extracts from MMP-2^+/− mice, as compared with WT mice. The expression of mRNA for MMP-9, TIMP-1, and TIMP-2 was also evaluated in mouse spleens, and it was found that although mRNA for MMP-9 was not significantly different between MMP-2^−/−, MMP-2^+/−, and WT mice, mRNA for TIMP-1 and TIMP-2 was reduced both in heterozygous (30% reduction for both) and in homozygous MMP-2^−/− mice (36 and 33% reduction, respectively).

MMP-2^−/− mice had a normal platelet count (949.000 ± 68.000 vs. 1,047.000 ± 22.630 platelets/µl of WT mice; n = 10; P = NS) and platelets appeared ultrastructurally normal at electron microscopy (Fig. S1). Surface expression of the major platelet glycoproteins (GPIIb/IIIa, GPIbα, GPVI, and CD9) did not differ between MMP-2^−/− and WT mice. Platelet ATP content was similar in WT and MMP-2^−/− mice (17.5 ± 1.7 and 18.8 ± 0.75 nmol/10⁸ platelets, respectively; n = 5). On the contrary, although platelet lysates of WT mice contained MMP-2 (43.9 ± 2.1 ng/10⁸ platelets; n = 5), platelets from MMP-2^−/− were completely devoid of it (<0.1 ng/10⁸ platelets; n = 5). 0.1 U/ml thrombin induced a rapid release of active MMP-2 in the supernatant of WT platelets (36.4 ± 2.47 ng/10⁸ platelets; n = 3) but not of platelets of MMP-2^−/− mice. A time course of the appearance of active MMP-2 on the surface of platelets upon activation was performed with gel-filtered WT mouse platelets by flow cytometry, using an antibody that specifically recognizes active MMP-2. Resting platelets did not express active MMP-2 on their surface. The expression of active MMP-2 on mouse platelets was evident already at 30 s (baseline = 2.5 ± 0.3%, 30 s = 9.3 ± 0.2% of positive platelets) after thrombin stimulation, reached a maximum at 2 min (21.4 ± 1.04% of positive platelets), and remained expressed for up to 15 min (n = 3).

The ability of 0.5 ng/ml of human active MMP-2 to potentiate mouse platelet aggregation and the time course of this effect was evaluated with washed platelets from WT mice. With all the agonists used (0.8 µg/ml collagen, 3 µM U46619, and 0.05 U/ml of purified human α-thrombin), the potentiation of platelet aggregation was maximal within 2 min and decreased thereafter to disappear in 30 min (Table S1). Platelets of MMP-2^−/− mice revealed a significant functional defect. In fact, although platelets from WT mice, as well as from MMP-2^+/− mice, aggregated dose dependently in response to ADP (0.2, 0.4, 2, and 10 µM), collagen (2, 5, and 10 µg/ml), arachidonic acid (0.1, 0.4, and 0.6 mM), epinephrine (10 and 100 µM), U46619 (3 and 4 µM), and purified human α-thrombin (0.05 and 0.1 U/ml), platelets from MMP-2^−/− mice failed to aggregate in response to the lowest concentrations of each of the agonists tested, and only the highest concentrations induced aggregation (Fig. 1 A; Table S2).

Secretion of dense granules in response to ADP or arachidonic acid (evaluated by the release of ATP) was also impaired in platelets from MMP-2^−/− mice (Fig. 1 B; Table S2). Similarly, the expression on the platelet surface of P-selectin, an α-granule protein, in response to collagen, thrombin, or ADP was reduced in platelets from MMP-2^−/− mice (Fig. 1 C; Table S2).

Serum thromboxane B₂ levels did not differ between WT and MMP-2^−/− mice (WT, 95 ± 13.6 ng/ml; MMP-2^−/−, 92 ± 10.1 ng/ml; n = 5; P = NS), excluding defective arachidonic acid metabolism as a cause of impaired secretion. Fibrinogen binding to platelets stimulated in vitro with 10 µM ADP was significantly reduced in MMP-2^−/− as compared with WT mice, whereas platelets from MMP-2^+/− mice showed an intermediate value (Fig. 1 D). Platelet adhesion to collagen, under flow conditions at high shear rate, was significantly lower in MMP-2^−/− mice as compared with WT mice (Fig. 1 E). The phosphorylation of Akt, the principal downstream effector of the PI3K signaling pathway, induced by stimulation with thrombin was almost absent in platelets from MMP-2^−/− mice as compared with platelets from WT mice (Fig. 1, F and G).

The expression of P-selectin on circulating platelets was enhanced dose dependently by the i.v. injection of a mixture of collagen plus epinephrine, significantly more so in WT mice than in MMP-2^−/− mice (Fig. 1 H; Table I). Similarly, fibrinogen binding to platelets induced by the i.v. injection of collagen plus epinephrine was also lower in MMP2^−/− as compared with WT mice (Fig. 1 I; Table I). Finally, platelet/leukocyte aggregates circulating in whole blood were enhanced by the i.v. injection of collagen plus epinephrine significantly more in WT mice than in MMP-2^−/− mice (Fig. 1 J; Table I).

The i.v. injection of collagen plus epinephrine in WT mice induced platelet pulmonary thromboembolism, with a dose-dependent increase of mortality of the number of lung vessels occluded by platelet thromboemboli and a marked drop of the number of circulating platelets. MMP-2^−/− mice were partially protected against platelet pulmonary thromboembolism. This was more evident when low doses of the thrombogenic challenge were used, whereas the thromboembolic response was close to normal when a strong stimulus was used (Fig. 2 A). Histological analysis of slices of lungs obtained 2 min after the i.v. injection of collagen plus epinephrine showed that a significantly higher number of lung vessels were totally or partially occluded by platelet thrombi in WT animals than in MMP-2^−/− mice when low doses of the thrombotic challenge were used (Fig. 2, B and C). The injection of collagen plus epinephrine dose-dependently induced a drop of the platelet count that was stronger in WT mice as compared with MMP-2^−/− mice, indicating that in vivo activation and sequestration was more sustained in WT than in knockout mice (Fig. 2 D).

Photochemically induced thrombosis of the femoral artery was correlated with the dose of rose Bengal infused. An occluding thrombus occurred after 13 ± 2.4 min (n = 5) and 7.6 ± 0.5 min (n = 5) from the beginning of the irradiation with 20 or 50 mg/kg of rose Bengal, respectively, in WT mice. In MMP-2^−/− mice, instead, the infusion of the lowest dose of rose Bengal (20 mg/kg) did not lead to the development of an occluding thrombus (>30 min; n = 7; P < 0.01 vs. WT), whereas blood flow stopped in 8.6 ± 1.6 min (n = 5; P = NS vs. WT) when the strongest thrombotic stimulus was used (50 mg/kg of rose Bengal; Fig. 2 E). Thrombus size was 0.33 ± 0.06 mg with 20 mg/kg of rose Bengal and 0.86 ± 0.17 mg with 50 mg/kg of rose Bengal (+165%; n = 5) in WT mice. When the weaker thrombogenic stimulus was used (20 mg/kg of rose Bengal), thrombus weight was particularly reduced in MMP-2^−/− mice, to 78% smaller on average than in WT mice, whereas it was only 28% smaller with the strongest stimulus (Fig. 2 F). Bleeding time was 175 ± 14 s (median 155 s; 95% confidence intervals [CIs] 143.3–205.7 s) in WT mice, 306.5 ± 8.6 s (median 197.5 s; 95% CIs 124.2–488.8 s) in heterozygous mice (MMP-2^+/−; P < 0.01 vs. WT), and 555.5 ± 97.7 s (median 452.5 s; 95% CIs 334.5–776.5 s) in MMP-2^−/− (P < 0.0001 vs. WT and P < 0.001 vs. heterozygous; Fig. 2 G).

To confirm that pro–MMP-2 is rapidly transformed into active MMP-2 at the surface of activated mouse platelets, as was previously shown for human platelets (Kazes et al., 2000), gel-filtered platelets from MMP-2^−/− mice were preincubated with 100 ng/ml of human pro–MMP-2 and stimulated with human α-thrombin. Active MMP-2, which is not present on resting platelets, was already detectable at their surface 30 s after stimulation, reached a maximum at 2 min, and remained expressed at up to 15 min, as revealed by flow cytometry (Fig. 3 A). Gelatinolytic activity at 68 kD (active MMP-2) was visualized by zymography in the supernatant of gel-filtered platelets from MMP-2^−/− mice incubated with 100 ng/ml of human pro–MMP-2 and stimulated with α-thrombin 2 min after stimulation (Fig. 3 B).

The addition of 100 ng/ml of human pro–MMP-2 to mouse platelet-rich plasma (PRP), 2 min before challenge with thrombin, potentiated platelet aggregation in both WT and MMP-2^−/− mice (Fig. 3 C). Preincubation of mouse-washed platelets with human 100 ng/ml of pro–MMP-2, 2 min before 0.1 U/ml thrombin, enhanced Akt phosphorylation in platelets from both WT (+53%) and MMP-2^−/− (+61.6%) mice (Fig. 3, D and E). The plasmatic half-life of human recombinant pro–MMP-2, assessed after the i.v. injection of 500 ng of the proenzyme in MMP-2^−/− mice, was 19 min. Accordingly, in the subsequent experiments assessing the in vivo effects of human pro–MMP-2, all tests were performed between 2 and 15 min from its i.v. injection. The injection of 100 ng/mouse of human pro–MMP-2 into MMP-2^−/− mice, 2 min before blood sampling, normalized ex vivo platelet adhesion (from 37 ± 1.5 to 48.7 ± 1.5% of surface covered; n = 5; P < 0.01; Fig. 3 F). The i.v. injection of 100 ng/mouse of human pro–MMP-2, 15 min earlier, shortened the bleeding time in MMP-2^−/− mice (from 599.2 ± 74.1 to 222 ± 47 s; n = 10; Fig. 3 G). The thrombotic defect of MMP-2^−/− mice upon photochemically induced femoral artery injury was partially reversed by the i.v. injection of 100 ng/mouse of purified pro–MMP-2 (n = 5; Fig. 3 H). In WT mice that were infused with a low dose of Rose Bengal (20 mg/kg), the injection of 100 ng/mouse of human pro–MMP-2 significantly shortened the time to occlusion (from 13 ± 2.4 to 6.9 ± 0.6 min; n = 5; P < 0.01), whereas it did not exert any appreciable effect in WT mice infused with 50 mg/kg of rose Bengal (from 7.6 ± 0.5 to 6.4 ± 0.8 min; n = 5; P = NS).

To establish the contribution of MMP-2 derived from platelets in the hemostatic and thrombotic defect observed in MMP-2^−/− mice, cross-transfusion experiments were performed (Pitchford et al., 2004). WT mice that were made deeply thrombocytopenic (platelet count <5% of basal) by the i.v. injection of a rabbit anti–mouse platelet antiserum had a very prolonged bleeding time (≥900 s). The reconstitution of an almost normal platelet count by the transfusion of platelets from WT mice shortened the bleeding time by 47.5% (473 ± 89 s; P < 0.01; n = 11), whereas the reconstitution of the platelet count by the transfusion of platelets from MMP-2^−/− mice shortened the bleeding time by only 12% (785.5 ± 61 s; n = 6; Fig. 4 A).

To confirm that platelet-derived MMP-2 was the cause of the hemostatic effect of transfused WT platelets and to exclude that other factors, such as plasma components, were involved, we created chimeric mice with MMP-2^−/− blood cells and WT vessels. The tail bleeding time of WT mice engrafted with bone marrow from WT animals (controls) was 296 ± 71 s (n = 10), whereas MMP-2^−/− bone marrow chimeras had a prolonged bleeding time, which was similar to that of MMP-2^−/− mice. The bleeding time of MMP-2^−/− mice engrafted with bone marrow from WT mice was normal (Fig. 4 B).

The bleeding time of MMP-2^−/− mice transplanted with bone marrow from the same genotype (880 ± 144.6 s) was somewhat longer than that of MMP-2^−/− mice (555.5 ± 97.7 s). Considering that in cross-transplantation experiments the reconstitution of the platelet count reached ∼70% of the normal, we performed experiments to assess the effect of a mild reduction of the platelet count on the bleeding time in MMP-2^−/− mice. The reduction of the number of circulating platelets to ∼70% in MMP-2^−/− mice by the administration of a low dose of a rabbit anti–mouse platelet antiserum induced a prolongation of the bleeding time to a level similar to that seen in MMP-2^−/−/MMP-2^−/− cross-transplanted animals (755 ± 57 s; n = 5).

Femoral artery thrombosis was photochemically induced in chimeric mice upon infusion of a low dose of rose Bengal (20 mg/kg). Blood flow stopped after 10.6 ± 2.6 min (n = 5) from the beginning of irradiation in chimeric mice with MMP-2^+/+ blood cells. In contrast, in chimeric mice with MMP-2^−/− blood cells, blood flow stopped after 27.8 ± 2.2 min (n = 5; P < 0.001 vs. chimeric mice with normal bone marrow; Fig. 4 C). The reduced thrombotic phenotype observed in MMP-2^−/− bone marrow chimeras was associated with defective platelet function: ex vivo platelet adhesion to a collagen-coated surface under high shear was lower in MMP-2^−/− bone marrow chimeras than in normal bone marrow mice (Fig. 4 D); platelet P-selectin expression after the i.v. injection of 60 µl of collagen plus epinephrine was significantly lower in MMP-2^−/− bone marrow chimeras (Fig. 4 E); and similar findings were observed for platelet/leukocyte aggregates (Fig. 4 F).

When only a mild damage to the femoral artery was induced by the infusion of a low dose of rose Bengal (10 mg/kg; Fig. 5 A), no occluding thrombus occurred in WT mice (time to occlusion, >30 min; n = 5; Fig. 5 B). However, when a second arterial damage of identical intensity was induced downstream of the first insult (Fig. 5 A) a complete occlusion occurred (time to occlusion, 19.4 ± 2.3 min; n = 5; Fig. 5 B). When the sequence of the two arterial damages was inverted, first downstream and then upstream of the first insult, no occlusive thrombus occurred. These data suggest that the first arterial damage induces the release of mediators, possibly platelet derived, flowing downstream and potentiating the thrombotic response to a second stimulus. Indeed, a significant but localized platelet activation occurred after the first insult. In fact, the expression of P-selectin on the platelet surface rose from 2.4 ± 0.15% (baseline) to 19.9 ± 2.7% in blood collected immediately downstream of arterial damage (Fig. 5 C and Fig. S2), whereas it did not change on platelets from systemic blood tail vein (Fig. 5 C). One of the platelet activation–potentiating mediators may be MMP-2. Interestingly, the expression of MMP-2 on platelets from blood withdrawn from the femoral artery was clearly enhanced after the first arterial injury (Fig. 5 D). Moreover, when an identical experiment was performed in MMP-2^−/− mice, no occlusive thrombus formed either after the first or after the second downstream arterial insult (time to occlusion, >30 min; n = 4; Fig. 5 B). However, the cross-transfusion of WT platelets into platelet-depleted MMP-2^−/− mice restored the ability to form an occluding thrombus after the second downstream arterial insult (time to occlusion, 17.7 ± 2.65 min; n = 7; P < 0.05 vs. MMP-2^−/−; Fig. 5 E). On the contrary, the cross-transfusion of MMP-2^−/− platelets into platelet-depleted WT mice also abolished thrombus formation after the second downstream damage to the femoral artery (time to occlusion, >30 min; n = 4). The expression of P-selectin and of MMP-2 (MMP-2 expression rose from 2.5 ± 0.5 to 18.2 ± 4.6% of positive cells; n = 7) on platelets from blood collected immediately downstream of the arterial insult was significantly enhanced in MMP-2^−/− mice cross-transfused with WT platelets after the photochemically induced arterial damage (Fig. 5 F) as compared with MMP-2^−/− mice (Fig. 5 G).

To confirm the role of platelet-released MMP-2 in the thrombogenic response downstream to a first arterial damage, we infused systemically WT mice i.v. with 1 µg/mouse of the selective inhibitor of MMP-2, TIMP-2. Immediately, downstream of the first arterial insult, thrombus formation was abolished. Moreover, P-sel and MMP-2 expression at the surface of circulating platelets collected downstream of the second arterial insult were not different from baseline in TIMP-2–treated mice. In contrast, when the first weak, normally nonthrombogenic, arterial damage was performed in WT animals systemically infused with 100 ng/mouse of pro–MMP-2, a thrombus had already developed after the first arterial insult (time to occlusion, 15 ± 1.08 min).

All these data are consistent with a model in which localized platelet activation induced by a first arterial damage generates the release of platelet-stored pro–MMP-2, which gets rapidly activated to active MMP-2 and acts on the neighboring platelets, making them hyperresponsive to a second stimulus like that generated by an arterial damage downstream to the first, to produce the formation of an occluding thrombus. The fact that this priming effect is essentially limited to the local circulatory bed derives from the transient nature of the priming effect of MMP-2 on platelets and from the dilution and/or inactivation in the systemic circulation of thrombogenic substances generated by localized platelet activation. This was confirmed by the subsequent experiments. The potentiating activity of human active MMP-2 on mouse platelet aggregation was maximal after a 2-min preincubation and it almost disappeared after 10 min. The infusion of thrombin-activated platelets directly into the femoral artery led to the appearance of P-selectin–positive platelets in the systemic circulation within the first 2 min after the injection (baseline, 3 ± 0.1%; 2 min, 7.7 ± 1.3% of positive platelets) and, thereafter, the expression of P-selectin on platelets in systemic blood disappeared.

We show in this paper that MMP-2^−/− mice are strongly protected against thrombosis induced by mild thrombogenic stimuli and have only a moderate impairment of hemostasis, that these alterations are the consequence of an abnormal platelet function, and that platelet-derived MMP-2 plays a crucial role in platelet activation and thrombus formation in vivo. Mice in which the MMP-2 gene has been disrupted do not suffer spontaneous bleeding and have an only moderately prolonged tail tip bleeding time as compared with WT mice, which is at variance from mice in which mechanisms of essential importance for the hemostatic function of platelets have been disrupted, such as the GPIIb/IIIa-von Willebrand factor interaction (Hodivala-Dilke et al., 1999; Marx et al., 2008). In contrast, MMP-2^−/− mice are resistant to platelet-dependent thrombosis, like pulmonary platelet thromboembolism (Gresele et al., 1990; Momi et al., 2005) or photochemically induced thrombosis of the femoral artery (Kikuchi et al., 1998; Momi et al., 2005). Several data suggest that the hemostatic defect and impaired thrombus formation of MMP^−/− mice are the result of defective platelet activation.

Indeed, although the structure and number of platelets from MMP-2^−/− mice are normal, they show defective aggregation in response to a variety of stimuli in vitro, reduced adhesion to collagen, an impaired expression of platelet P-selectin, and diminished formation of platelet/leukocyte aggregates upon activation in vivo. Interestingly, defective platelet activation in vitro and in vivo is especially evident when weak stimuli are used, whereas it is much less apparent with strong stimuli, which is in agreement with an autocrine- or paracrine-potentiating action of platelet-released MMP-2, as had been previously observed in vitro with human platelets (Sawicki et al., 1997; Falcinelli et al., 2005; Santos-Martínez et al., 2008). Defective platelet function of MMP-2^−/− mice was restored by the in vitro addition of human pro-MMP-2; moreover, the i.v. injection of human pro–MMP-2 into MMP-2^−/− mice restored platelet function in terms of adhesion, aggregation, and P-selectin expression, indicating a pivotal role of MMP-2 for platelet activation. The human pro–MMP-2 injected into MMP-2^−/− mice was rapidly activated at the surface of platelets, and the concentrations of active MMP-2 measured in plasma 2 min after the i.v. injection of pro–MMP-2 were in the range of those potentiating platelet activation in vitro (Falcinelli et al., 2005) and, thus, sufficient to reverse the deficit of platelet function of knockout mice.

The hemostatic defect and defective thrombus formation observed in mice in which the gene for MMP-2 has been disrupted is mainly the result of the absence of MMP-2 from platelets, as shown by the observation that the cross-transfusion of MMP-2^−/− platelets into thrombocytopenic WT mice does not correct the prolonged bleeding time. Similarly, chimeric mice in which the hematopoietic cell line derives from bone marrow cells of MMP-2^−/− mice have moderately impaired hemostasis and defective photochemical injury-induced arterial thrombosis, which is similar to what observed in MMP-2^−/− mice, whereas MMP-2^−/− mice with bone marrow derived from WT mice, and, therefore, with circulating platelets containing normal amounts of MMP-2 but with blood vessels devoid of MMP-2, showed a normal thrombotic response upon arterial injury.

Altogether, our results are compatible with a model in which MMP-2 contained in platelets is released upon activation and acts in an autocrine and paracrine way on the neighboring platelets, facilitating their activation response to physiological or pathological stimuli (Gresele et al., 2008b). This may become crucial in transforming a normal hemostatic platelet activation response to mild vessel wall damage into an uncontrolled thrombotic response. This is confirmed by the observations that when two mild arterial insults are induced sequentially, a thrombus forms only at the level of the second arterial damage downstream, that this phenomenon does not take place when circulating platelets are devoid of MMP-2, and that a selective MMP-2 inhibitor blocks downstream thrombus formation. These observations might explain why in the coronary bed of patients dying from acute myocardial infarction, where arterial plaques are widely disseminated and more than one may be fissured or eroded, only a single acute occluding thrombus forms and is ultimately responsible for the acute ischemic event (Casscells et al., 2003). In contrast, pathological conditions in which circulating levels of MMP-2 are enhanced as a result of the release of MMP-2 from tissue, like ruptured inflammed atherosclerotic plaques (Galis et al., 1994; Galis and Khatri, 2002), may be associated with platelet hyperreactivity and with an enhanced tendency to thrombosis. Indeed, the release of MMP-2 in the coronary circulation of patients with acute coronary syndromes has been previously shown (Hojo et al., 2002; Falcinelli, E., M. Leone, G. Cimmino, T. Corazzi, P. Golino, and P. Gresele. 2005. Congress of the International Society on Thrombosis and Haemostasis. Abstr. OR370). Our data, showing that an increase of circulating MMP-2, which is obtained by the infusion of pro–MMP-2 in the systemic circulation of normal mice, transforms a mild nonthrombogenic stimulus into a strong thrombotic insult, are in favor of this hypothesis.

The important role played by platelet-released MMP-2 in arterial thrombosis identifies a new potential target for novel antithrombotic therapies. This is especially relevant if one considers that aspirin intake does not prevent the in vivo release of MMP-2 from activated platelets (Falcinelli et al., 2007). The modulation of platelet-released MMP-2 by the development of specific monoclonal antibodies, of selective small molecule inhibitors, or by the application of gene silencing techniques can be one of the strategies to overcome the narrow gap between insufficient efficacy and excessive bleeding typical of current antiplatelet drugs (Bhatt, 2007; Gresele et al., 2008b).

C57BL/6 mice were purchased from Charles River. Mice deficient in MMP-2 were generated by gene targeting (Itoh et al., 1997; Wielockx et al., 2001). The mutation heterozygous mice were obtained by crossing the chimeras to C57BL/6J one to five times, and then crossing to obtain the mutation homozygous mice. These mice, and controls with matching genetic backgrounds, were bred as homozygous lines and were used at the age of 8–12 wk. Mice were maintained with a 12:12-h light/dark cycle and had free access to food and water.

Genotyping of mice was performed by PCR analysis. Littermate tail tip DNA extraction was performed with the NucleoSpin tissue kit (M-Medical). Primer sequences used for mouse MMP-2 were 5′-AACGATGGAGGCACGAGTGGCC-3′ (forward) and 5′-TGTTTTAGGGGCGACATCGCCG-3′ (reverse). Primer sequences used for NEO detection were 5′-CAACGCTATGTCCTGATAGCGGTCC-3′ and 5′-CGTGTTCCGGCTGTCAGCGCAGG-3′. In addition, the absence of MMP-2 protein in MMP-2^−/− mice was evaluated by zymography in plasma (Fig. S3). All animal protocols were approved by the Committee on Ethics of the University of Perugia (protocol 186/2006 and 216/2007) and by the Italian Ministry of Public Health (authorization protocol 74/2007 and 76/2007-B).

MMP-2, MMP-9, TIMP-1, and TIMP-2 gene expression was assessed by RT-PCR in spleens taken from anesthetized WT, MMP-2^−/−, and MMP-2^+/− mice after perfusion with 20 ml of saline via left ventricular puncture at an infusion pressure ≤100 mm Hg, and immediately frozen in liquid nitrogen. Total RNA was extracted using an RNA purification kit (Absolutely RNA Miniprep kit; Agilent Technologies) and treated with DNase to avoid any genomic DNA contamination. Quantitative PCR was performed by monitoring in real time the increase in fluorescence of the Sybr Green Dye in a real-time PCR system (Mx 3000P; Agilent Technologies).

MMP-2, MMP-9, TIMP-1, and TIMP-2 segments of each resultant complementary DNA sample were PCR amplified in the presence of specific sense and antisense primers (200 nmol/liter each), using a power Sybr Green PCR Master Mix (Applied Biosystems) in a final reaction volume of 25 µl. Cycling conditions were 95°C for 10 min, followed by 40 cycles of 95°C for 15 s (denaturation) and 60°C for 1 min (annealing and extension). β2 microglobulin complementary DNA was amplified as an internal control. Each sample was tested in triplicate for each gene, and then a melting curve to ensure the specificity of amplification and the expression level of the gene was calculated by the ΔΔC_T method (Antonov et al., 2005).

The gelatinolytic activity of MMP-2 was measured in plasma, in the supernatant of activated platelets, or in tissue extracts (20 µg of protein extracts) by zymography (Kleiner and Stetler-Stevenson, 1994), as previously described (Falcinelli et al., 2005). In brief, samples were mixed with SDS sample buffer without 2-mercaptoethanol and applied to 10% polyacrylamide gels copolymerized with 2 mg/ml of gelatin. After electrophoresis, gels were incubated for 1 h with 2.5% Triton X-100 to remove SDS and then overnight in buffer (50 mM Tris-HCl buffer with 20 mM NaCl, 5 mM CaCl₂, and 0.01% Brij35, pH 7.5) at 37°C. Gels were then stained with 0.05% Coomassie Brilliant blue G-250 in a mixture of methanol, acetic acid, and water (3:1:6, vol/vol) and destained in 30% methanol with 10% acetic acid. Gelatinolytic activities were detected as transparent bands against the background of Coomassie Brillant blue–stained gelatin and quantified by analyzing the zymograms by densitometry. Intensity of pro–MMP-2 and MMP-2 bands was quantified by densitometry using the software Quantity One (version 4.1.1; Bio-Rad Laboratories). Gelatinolytic bands were compared with a standard curve built using increasing amounts of recombinant human MMP-2 (EMD). Tissue or plasma protein content was measured by a protein assay (Bio-Rad Laboratories).

Blood was collected by cardiac puncture from mice under ether anesthesia and anticoagulated with 1/10 vol of tripotassium EDTA. After thorough mixing, platelets were counted optically, using the Brecher-Cronkite method (Brecher and Cronkite, 1950), by an operator unaware of the experimental groups (Momi et al., 2005).

Resting washed platelets were fixed for 4 h at 4°C, using cacodylate-HCl buffer (pH 7.4) containing 4% wt/vol of glutaraldehyde. The samples were then washed, maintained in cacodylate buffer for a further 4 h, and placed in 1% osmium tetraoxide and pelleted by centrifugation at 10,000 g for 30 s. Ultrathin sections of the platelet pellets were stained with uranyl acetate and lead citrate and observed with a transmission electron microscope at 80 kv (Optic EM208; Phillips; Gresele et al., 2009).

The effect of MMP-2 on platelet aggregation was studied by the photometric method using either washed platelets (Vezza et al., 2002) or PRP (Momi et al., 2005). Whole blood, drawn in 4% sodium citrate, was centrifuged at 150 g for 15 min to obtain PRP, which was then adjusted to 300,000 platelets/µl using autologous platelet-poor plasma. Washed platelets were prepared in the presence of 0.2 µM PGI₂, 50 U/ml heparin, and 0.01 mg/ml apyrase. Finally, platelets were resuspended in standard Tyrode's buffer, pH 7.4, containing 0.35% BSA and 0.1% glucose, to give a final concentration of 250,000 platelets/µl. Gel-filtered mouse platelets were prepared as previously described (Pulcinelli et al., 1998) and adjusted to 10⁸/ml in Hepes-Tyrode's buffer containing 0.1% BSA, pH 7.4. Platelet aggregation in response to different doses of collagen, arachidonic acid, ADP, purified human α-thrombin (given by R. De Cristofaro, Catholic University of Rome, Rome, Italy), or U46619, a stable analogue of TxA₂ (Vezza et al., 1993), was measured using an automatic platelet aggregation analyzer (APACT 4; Alfa Wassermann).

Platelet ATP release was measured using firefly luciferase (7.3 µg/ml final concentration) and luciferin (800 U/ml final concentration). Luminescence generated by platelet-secreted ATP was monitored using a lumi-aggregometer (Chrono-Log Corporation) and compared with that obtained by adding a known concentration of ATP standard (Emiliani et al., 2006).

Platelet adhesion to a collagen-coated surface under flow conditions was studied in a parallel-plate perfusion chamber (Sixma et al., 1998), as previously described (Momi et al., 2005). In brief, 0.5 ml of citrated blood was passed through a perfusion chamber, over a plastic coverslip sprayed with ∼30 µg/cm² collagen from equine tendon (Hormon-Chemie), at a wall shear rate of 3000 s⁻¹. The chamber was then perfused with BSA-containing saline to remove all residual blood, and the coverslip was harvested, gently washed with 10 mmol/liter Hepes, and fixed with 0.25% glutaric dialdehyde in PBS. Adhering platelets were stained with May-Grünwald/Giemsa, observed under an optical microscope, and the area covered by platelets was measured using a computerized image analyzer (Scion Image; National Institutes of Health; Momi et al., 2005).

Platelet surface P-selectin, MMP-2, and fibrinogen expression and circulating platelet-leukocyte aggregates were measured in whole blood by flow cytometry (EPICS XL-MCL; Beckman Coulter) as previously described (Ciferri et al., 2000; Pitchford et al., 2003). Blood samples, collected in 4% sodium citrate and diluted 1:10 with PBS, were stimulated with a mixture of 0.5–5–20 µg/ml collagen plus 1–10–50 µM epinephrine, 30 µg/ml collagen, 0.05 or 0.1 U/ml of human α-thrombin, or 20 µM ADP for 30 min under continuous stirring. Aliquots of 5 µl were then stained with a PE-labeled anti-CD41 antibody (Leo.D2; Emfret Analytics) as a platelet identifier and with an FITC-labeled anti–P-selectin antibody (CD62P; Emfret Analytics) or with an FITC-labeled anti-fibrinogen antibody (P140-1; Emfret Analytics) for 30 min in the dark. The reaction was stopped with 1 ml of 1% PFA. P-selectin, MMP-2, and fibrinogen expression are reported as a percentage of positive cells. For platelet-leukocyte aggregates, 50 µl of undiluted whole blood were fixed with 1% PFA and incubated for 30 min with saturating concentrations of FITC-labeled anti-CD41/CD61 (α_IIbβ₃; Leo.D2; Emfret Analytics) and Peridin chlorophyl protein–conjugated anti–mouse CD45 (clone 30-F11; BD). Subsequently, red blood cells were lysed by adding 400 µl of water and, after 10 min, 1 ml of PBS.

Leukocytes were identified on the basis of their morphology and of the positivity for CD45. Platelet-leukocyte aggregates were quantified as the percentage of leukocytes positive for the platelet antigen CD41/CD61 or as the mean fluorescence intensity of the platelet antigen CD41/CD61 on the leukocyte surface. In some selected experiments, platelet P-selectin expression, fibrinogen binding, and platelet/leukocyte aggregates were assessed in blood collected from WT and MMP-2 ^−/− mice 2 min after the i.v. injection of a mixture of 150 µg/ml collagen plus 0.25 µM epinephrine.

For platelet MMP-2 expression, 5 µl of diluted whole blood was incubated for 30 min with saturating concentrations of rabbit PE-labeled anti–mouse CD41 (Leo.F2; Emfret Analytics) and goat anti–mouse MMP-2 (R&D Systems). Subsequently, an Alexa Fluor 488 anti–goat antibody (1:500; Invitrogen) was added. At the end of incubation period the reaction was stopped with 1 ml of 1% PFA.

For the study of the time course of expression of active MMP-2 on platelet surface, gel-filtered platelets from WT or MMP-2^−/− mice, incubated with 100 ng/ml of human pro–MMP-2, were activated by 0.1 U/ml of human α-thrombin, and the reaction was stopped at 30 s and 2 and 15 min after the addition of α-thrombin with 1 ml of 1% PFA. Active MMP-2 was assessed with a specific rabbit anti–mouse active MMP-2 antibody (Millipore) and a goat anti–rabbit antibody (1:500; Invitrogen) by flow cytometry.

500 ng of human recombinant pro–MMP-2 (Millipore) was injected i.v. in MMP-2^−/− mice (n = 15). After 2 min, and then every 30 min for a total of 120 min, whole blood was withdrawn in sodium citrate 4% (1:9 vol/vol) and quickly centrifuged to obtain platelet-poor plasma that was snap frozen for later determination of MMP-2 activity. The amount of active and total MMP-2 in mouse plasma was assed by ELISA (MMP-2 Biotrak activity assay system; GE Healthcare) and by gelatin zymography. The plasma half-life of MMP-2 was determined as the time required by human MMP-2 to be degraded to half the concentration found in plasma 2 min after MMP-2 injection.

For serum TxB₂ measurement, non-anticoagulated blood was collected in glass tubes and placed in a water bath at 37°C for 60 min and then centrifuged (Patrono et al., 1980). Immunoreactive TxB₂ was measured in unextracted highly diluted serum samples by a highly specific radioimmunoassay, as previously described (Gresele et al., 1987).

Akt phosphorylation was studied in washed platelets incubated for 2 min with activated human MMP-2 (Falcinelli et al., 2005) or with its vehicle before the addition of 0.05–0.1 U/ml thrombin. After 10 min, the reaction was stopped with SDS-PAGE sample buffer, platelets were lysed with 2% SDS, and samples were carefully mixed and stored on ice. Proteins (measured by the Bradford assay) were analyzed by SDS-PAGE followed by immunoblotting, using an antibody against Akt or phospho-Akt Ser⁴⁷³ (1:2,000). In each electrophoresis well, an identical amount of proteins was loaded (20 µg; Falcinelli et al., 2005). Data were quantified using the QuantiScan software (BIOSOFT) and expressed as percentage of control (i.e., phosphorylation of Akt in thrombin-stimulated platelets). Human pro–MMP-2 was activated by incubation with 1 mmol/liter APMA (para-aminophenylmercuric acetate), in a buffer containing 50 mmol L⁻¹ NaCl, 5 mmol L^-1 CaCl₂, and 0.01% Brij 35, pH 7.5, at 37°C for 30 min (Falcinelli et al., 2005). APMA was then removed by filtration through Amicon filters (Microcon YM-10; Millipore).

Collagen plus epinephrine–induced pulmonary thromboembolism was performed in MMP-2^−/− and in WT mice at 8–12 wk of age, using a method described previously (Gresele et al., 1990; Momi et al., 2005). Mice were caged and fed a regular diet for at least 1 wk before use. Thrombotic challenge was generated by the rapid i.v. injection of a mixture of 250 µg/ml collagen 1.5 µg/ml and epinephrine into one of the tail veins. MMP-2^−/− and WT mice were challenged with increasing volumes (from 20 to 100 µl) of the collagen plus epinephrine mixture and mortality was recorded. The effect of the i.v. challenge with collagen plus epinephrine was evaluated as described; the cumulative end point was death of the animal or paralysis of the hind limbs for more than 15 min (Gresele et al., 1990; Momi et al., 2005). In each session, at least five animals per experimental group were tested. Controls were run at the beginning and at the end of each experimental session. At the end of each experimental session, surviving animals were sacrificed by an overdose of anesthesia. Mice were accustomed to handling and the injections were performed by skilled investigators with minimal disturbance to the animals. Data are presented as a percentage of animals dead per total number of animals tested.

The total number of lung vessels occluded by platelet thrombi was counted by light microscopy in paraffin-embedded sections stained with hematoxylin and eosin (Momi et al., 2005).

Photochemical-induced femoral artery thrombosis was induced in anesthetized mice by a method described previously (Kikuchi et al., 1998; Momi et al., 2005). In brief, mice were anaesthetized by 60 mg/kg of intraperitoneal sodium pentobarbital and placed on a heated operating table. A 25G-needle venous butterfly was inserted in one of the tail veins for the infusion of rose Bengal. The left femoral artery was carefully exposed, and a laser Doppler probe (Transonic System Inc.) was positioned onto the branch point of the deep femoral artery, distal to the inguinal ligament, for monitoring blood flow. The exposed artery was irradiated with green light (wavelength, 540 nm) of a Xenon lamp (L4887; Hamamatsu Photonics) equipped with a heat-absorbing filter via a 3-mm diameter optic fiber attached to a manipulator. Light irradiation was protracted for 20 min. The infusion of 10–50 mg/kg of rose-Bengal was started 5 min after the beginning of irradiation and lasted for 5 min. The end point was set as the cessation of blood flow for >30 s. If no occlusion occurred after 30 min, the time was recorded as 30 min (Kikuchi et al., 1998; Momi et al., 2005). 1 h after the end of the irradiation period, the femoral artery was excised and opened, and the thrombus was removed, dried at 37°C for 24 h, and then weighed. In some experiments, arterial damage was induced sequentially at two different sites of the femoral artery by two irradiation injuries separated from one another by 30 min, with the second 0.5 cm downstream of the first. For flow cytometry measurement of platelet P-selectin or platelet MMP-2 expression, blood was collected immediately after the first or the second vessel injury by arterial puncture at a femoral artery site distal to the injury or systemically from a tail vein.

To assess the detectability on the systemic circulation of localized platelet activation in the femoral artery, platelets obtained from WT mice were isolated by gel filtration and activated with 0.1 U/ml of purified human α-thrombin for 15 min. Thrombin was neutralized with equimolar hirudin (Huo et al., 2003), and then 3 × 10⁸ activated platelets were injected directly into the femoral artery of recipient mice. Blood for the detection of platelet P-selectin expression by flow cytometry was collected after platelet injection systemically, from a tail vein, at 2, 5, 10, and 15 min.

Bleeding time was measured by the tail tip transection method, as previously reported (Momi et al., 2005). In brief, mice were positioned in a special immobilization cage that keeps the tail of the animal steady and immersed in saline thermostated at 37°C. After 2 min, the tip of the tail was transected with a razor blade at 2 mm from its end. The tail was immediately reimmersed in thermostated saline and time taken to stop bleeding was measured. The end point was the arrest of bleeding lasting for more than 30 s. Bleeding was recorded for a maximum of 900 s.

In some selected experiments, platelets were isolated by centrifugation at 150 g for 10 min from whole blood of donor mice anticoagulated with acid citrate dextrose (1:6, vol/vol). The PRP obtained was pooled and infused in recipient mice that had been previously rendered deeply thrombocytopenic by one single i.v. injection of a saturating amount (100 µl) of an anti–mouse anti-platelet serum (APAS), prepared as previously described (Pitchford et al., 2004, 2008), or by γ irradiation (10Gy; Clinac 600/C; Varian Medical Systems, Inc.). APAS was injected 15 min before platelet transfusion. Platelet counts decreased by ≥95% within 2 min, and a gradual increase in the number of circulating platelets was seen only 48 h after administration (Pitchford et al., 2004). In γ-irradiated mice, the number of circulating platelets dropped by 96% by day 5 and platelets were reinfused at that time. Approximately 1.0 × 10⁹ platelets were reinfused into each animal, leading to a restoration of the platelet count to ∼80% of basal (Pitchford et al., 2004). Bleeding time measurements or photochemically induced arterial damage was performed 1 h after platelet transfusion.

8–12-wk-old, male and female MMP-2^−/− (C57BL/6 background) and C57BL/6 (WT) mice were used. Recipient mice were exposed to a single lethal dose of 8 Gy from an 18-mV photon beam linear accelerator (Clinac 600/C) with a focus-to-skin distance of 63 cm and a dose rate of 0.2 Gy/min. After irradiation, all mice not receiving bone marrow transplantation died within 14 d. Donor bone marrow cells were collected into PBS by flushing the shafts of the femur and tibia. The cells were suspended and washed with PBS, and then red blood cells were lysed with ammonium chloride–potassium buffer and the remaining cells were washed twice with PBS. Finally, bone marrow cells were resuspended at a final concentration of 20 × 10⁶ cells per ml in PBS and, the day after total body irradiation, 10 × 10⁶ cells/mouse were injected into recipient mice i.v. via the lateral tail vein (Ruggeri et al., 2002).

In some experiments MMP-2^−/− mice were used as donors and WT as recipients, in others WT mice were used as donors and MMP-2^−/− as recipients. Control experiments were performed by transplanting bone marrow from WT donors into WT recipients and from MMP-2^−/− donors into MMP-2^−/− recipients. 1 mo after bone marrow transplantation, both chimeric and control mice had almost normal platelet counts (746.000 ± 39.975 and 800.000 ± 30.551 platelets/µl, respectively). MMP-2 activity by zymography and MMP-2 expression by RT-PCR were assessed in blood cells, aorta, kidney, and lung extracts obtained from recipients 8 wk after bone marrow transplantation to ensure that full engraftment had taken place.

Data are expressed as arithmetic means ± SEM and were analyzed by one way analysis of variance, followed by the Newman-Keuls multiple comparison tests between all groups. Differences in mortality were assessed by the Fisher's exact test. All analyses were performed by Prism 4.0 (GraphPad Software, Inc.). A p-value of <0.05 was considered to be statistically significant.

Fig. S1 shows electron microscopy of resting platelets from MMP-2^−/− and WT mice. Fig. S2 shows flow cytometry plots of the expression of P-selectin and active MMP-2 on the platelet surface in WT and MMP-2^−/− mice, in blood collected at baseline and immediately downstream of an arterial damage. Fig. S3 shows an example of zymography that confirms the absence of MMP-2 in MMP-2^−/− mice. Table S1 reports data on the potentiation of platelet aggregation induced by active MMP-2. Table S2 summarizes platelet aggregation, ATP release, and P-selectin expression in response to various agonists in WT and MMP-2^−/− mice.

The authors thank Dr. G. Guglielmini for his kind help with the platelet adhesion experiments.

This work was supported in part by a grant from Fondazione Cassa di Risparmio di Perugia (project 2007-0130-020), a grant from the Italian Ministry of University and Scientific Research (MIUR 2007, Prot. 20073KBBHC_003), and a grant from the Italian Ministry of Health (RFPS-2006-8-334062) to P. Gresele and by a fellowship from Regione Umbria to S. Momi (POR Ob.3 2000-2006 Misura D4 Risorse CIPE).

The authors have no conflicting financial interests.