Polyphosphate colocalizes with factor XII on platelet-bound fibrin and augments its plasminogen activator activity

The contact pathway comprises factor XII (FXII), prekallikrein (PK), factor XI (FXI), and a nonenzymatic cofactor, high-molecular-weight kininogen (HK). Reciprocal proteolytic activation of FXII and PK to their active forms, FXIIa and kallikrein, respectively, occurs via interaction with a negatively charged surface and is enhanced by Zn²⁺.^1-5  FXI and PK circulate in complex with HK,^6,7  which assembles these proteases on the activating surface. FXIIa can cleave FXI, stimulating the intrinsic pathway and downstream generation of thrombin, leading to its classification as a coagulation factor. However, FXIIa is reported to participate in multiple pathways, including inflammation, complement, and fibrinolysis.⁸ 

The profibrinolytic effects of the contact pathway are multifaceted, involving direct and indirect interactions. Kallikrein liberates the vasoactive peptide bradykinin from HK, which induces a host of vascular responses, including release of tissue plasminogen activator (tPA) from endothelial cells.⁹  Kallikrein also directly cleaves single-chain urokinase plasminogen activator^10,11  to active urokinase plasminogen activator (uPA). FXII shows distinct homology to tPA and uPA^12-14  and, accordingly, FXIIa exhibits plasminogen activator activity. The kinetics of the reaction are considered unfavorable^13,15  but FXII is present in plasma at 4 orders of magnitude higher concentrations compared with tPA and uPA.^16,17  During evolution, redundancy has developed in the fibrinolytic system, underscored by the relatively mild abnormalities associated with deficiency in tPA or uPA, whereas the double knockout exhibits a more acute phenotype.¹⁸  Under certain circumstances or within specific milieus, it is plausible that FXIIa contributes to plasmin generation to complement or compensate for tPA and uPA activity and may be potentially relevant in vivo.¹⁹ 

Several natural surfaces facilitate FXII activation, including polyphosphate (polyP),^20,21  RNA,²²  misfolded proteins,²³  and collagen.²⁴  PolyP is an ancient biomolecule that is ubiquitous in nature.²⁵  It is highly anionic and, consequently, acts as a propitious surface for activation of the contact pathway.^20,21  PolyP is localized in platelet dense granules²⁶  and is secreted upon activation²¹  alongside adenosine 5′-diphosphate (ADP), serotonin, and metal ions, including Zn²⁺. Platelet polyP is ∼60 to 100 mers,²¹  which is significantly shorter than the long-chain polymers (1000-2000 mers) in bacteria.²⁷  Polymer length is crucial to polyP’s biological activity, with shorter chains showing reduced capacity to activate the contact pathway.²⁸  PolyP also interacts with fibrin(ogen) and accumulates in clots, altering their structural properties and susceptibility to tPA-mediated fibrinolysis.^29,30  The half-life of polyP in plasma is relatively short (1.5-2 hours)²⁰  but could be preserved within the microenvironment of the thrombus.

The plasminogen activator function of FXIIa is augmented by artificial surfaces and Zn²⁺.³¹  Here, for the first time, we show that a natural surface, polyP, amplifies the plasminogen activator function of FXIIa. Indeed polyP, of approximate chain length of that found in platelets, binds both FXII(a) and plasminogen, indicative of a template mechanism of activation. Importantly, we also demonstrate that platelet-associated fibrin acts as a reservoir for FXII(a), plasminogen, and platelet-released polyP.

For platelet experiments, peripheral blood was collected into acid citrate dextrose solution A VACUETTE tubes (Greiner Bio-One); the first 3 mL was discarded. Platelets were washed and counted as described previously.³²  Pooled normal plasma that is essentially free of platelets was prepared from whole blood of 20 normal donors collected into 3.2% trisodium citrate.³³ 

PolyP was extracted from platelets as described previously.²⁸  Experiments were performed with platelet-derived polyP, synthetic polyP₆₅ (Sigma-Aldrich), or polyP₇₀ (a kind gift from BK Giulini, GmbH). Similar results were obtained with both synthetic preparations and, for simplicity, are described as polyP₇₀ throughout and the concentration quoted as phosphate monomer (monomer formula, NaPO₃).

Plasminogen-depleted fibrinogen (3.8 μM), Glu- or Lys-plasminogen (0.24 μM), and αFXIIa (200 nM) were added with or without polyP₇₀ (70 μM) in 10 mM Tris-HCl (pH 7.4), 140 mM NaCl, and 0.01% Tween²⁰  to 96-well plates (Greiner Bio-One). Clotting was initiated with 0.25 U/mL of human thrombin and 10 mM CaCl₂. In an analogous set of experiments, 200 nM FXII was preincubated with polyP₇₀ (140 μM) at ambient temperature for 30 minutes. In some cases, tPA (1 pM) was also included. Some experiments incorporated plasminogen activator inhibitor 1 (PAI-1; 0.1-1 nM) and PAI-2 (0.1-1 nM), C1 inhibitor (125-500 nM), or histidine-rich glycoprotein (HRG; 0.25-1 μM). PolyP₇₀ was replaced by HeLa RNA (10 μg/mL) or equine type 1 (Horm) collagen (5 μg/mL) in some assays. Absorbance at 405 nm was recorded every 1 minute using a BioTek PowerWave plate reader and BioTek Gen5 software.

Plasminogen activation was analyzed as described previously,³⁰  with the following modifications. αFXIIa (200 nM), Glu- or Lys-plasminogen (0-400 nM), and S2251 (0.5 mM) were added to microtiter plates in 10 mM Tris-HCl (pH 7.4), 140 mM NaCl, and 0.01% Tween,²⁰  with or without polyP₇₀ (0-70 μM). Alternatively, 200 nM FXII was added with or without polyP₇₀ (140 μM). Plasmin activity was detected by measuring the absorbance at 405 nm every 1 minute. In a parallel set of assays, the following inhibitors were incorporated: PAI-1 (0.1-1 nM), PAI-2 (0.1-1 nM), C1 inhibitor (125-500 nM), and HRG (0.25-1 μM). In some experiments, soluble fibrin (3.8 μM), prepared as described previously,³⁴  was coated onto plates overnight at 4°C before performing activity assays. In some cases, RNA (10 μg/mL) or collagen (5 μg/mL) was included in place of polyP₇₀.

FXIIa activity was analyzed as described previously.³⁵  Briefly, αFXIIa (50 nM) and S2302 (0.25 mM), with or without polyP₇₀ (70 μM), were added to microtiter plates and the absorbance was read at 405 nm for 4 hours, with addition of 5 μL (5 mM) S2302 or 5 μL (50 nM) protein at the 2-hour midpoint.

In autodegradation experiments, αFXIIa (50 nM) was incubated with or without polyP₇₀ (70 μM) for up to 360 minutes before resolving on 4% to 12% Bis-Tris gels (NuPAGE; Thermo Fisher Scientific) under nonreducing conditions and protein staining using InstantBlue.

PolyP₇₀ was bound to Sepabeads as described previously³⁶  and incubated with 5 μg of FXII, αFXIIa, βFXIIa, or plasminogen. Flow-through material and subsequent low-salt (50 mM NaCl) and high-salt (1 M NaCl) washes were collected, and fractions were analyzed by western blotting as described previously³²  using horseradish peroxidase–conjugated goat anti-human FXII or horseradish peroxidase–conjugated goat anti-human plasminogen (both Enzyme Research Laboratories).

Slides were coated with collagen (20 μg/mL) and thrombin (100 nM) and subsequently blocked with 5% bovine serum albumin. Platelets (5 × 10⁷/mL) were added to coated slides for 45 minutes with 4′,6-diamidino-2-phenylindole (DAPI; 25 μg/mL; excitation wavelength, 358 nm; emission wavelength, 525 nm), DyLight 488 (Thermo Fisher Scientific)-labeled human FXII (DL488-FXII; 365 nM; excitation wavelength, 493 nm; emission wavelength, 518 nm), and Alexa Fluor 647–conjugated annexin V (AF647-annexin V; 1/20, excitation wavelength, 594 nm; emission wavelength, 633 nm; Thermo Fisher Scientific) in the presence of CaCl₂ (2 mM). In control experiments, Benzonase Nuclease (Sigma-Aldrich) was included during stimulation to degrade contaminating DNA and RNA. Platelets were visualized by fluorescent confocal microscopy on a Zeiss LSM710 confocal microscope with ×63/1.40 oil immersion objective. Images were recorded on bright field and on separate channels for each wavelength and analyzed using Zen 2012 software.

Washed platelets (6.35 × 10⁸/mL; final concentration in clot, 1.5 × 10⁸/mL) were incubated with DAPI (20 μg/mL) and AF647-annexin V (1/20) before activating with collagen (100 μg/mL), thrombin receptor activator peptide 6 (TRAP-6; 100 µM), and CaCl₂ (4 mM). Activated platelets were added to pooled normal plasma (50% final concentration) in the presence of DL488-FXII (365 nM) and DyLight 550 (Thermo Fisher Scientific)-labeled human fibrinogen (120 nM; excitation wavelength, 562 nm; emission wavelength, 576 nm). Thrombin (24 nM) and CaCl₂ (10 mM) were added, and clots were allowed to form in μ-Slide VI^0.4 ibiTreat chambers (ibidi GmbH) for 2 hours. Images were recorded as detailed in “Visualization of FXII and polyP on single platelets.”

DL488-FXII (365 nM) and DAPI (10 μg/mL) were added to washed platelets (2 × 10⁷/mL) in N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid resuspension buffer (pH 7.45) containing 2 mM CaCl₂ and then were stimulated with either 100 ng/mL of convulxin (CVX) and 20 μM TRAP-6 or with 100 ng/mL of CVX and 100 nM thrombin for 45 minutes at ambient temperature. AF647-annexin V (1/20) was added 5 minutes before the end of stimulation. In some cases, 5 mM Gly-Pro-Arg-Pro (GPRP; Sigma-Aldrich) was included to inhibit fibrin polymerization. Analysis was performed on an LSR II flow cytometer with FACSDiva 6.1.3 software (Beckton Dickinson), with appropriate compensation applied and 10 000 events collected per sample. Data were analyzed using FlowJo V.X.0.6 software. Results are expressed as mean percentage of positive platelets and median fluorescence intensity ± standard error of the mean.

Data analysis was performed in GraphPad Prism 5.04. Clot lysis time (CLT) results are expressed as time to 50% lysis and were derived from the time taken from the maximal amplitude of the clot to reach the midpoint to baseline. Alternatively, graphs were normalized and data plotted as percentage turbidity. Plasmin generation was calculated as described previously.³⁷  Briefly, absorbance at 405 nm was plotted against time squared, and the slope from the initial linear portion was estimated. These values were used to calculate plasminogen activation rates using the specific activity of plasmin against S2251 that was experimentally determined to be 1.055 at A₄₀₅/min per μM (not shown). Fold-changes in lysis were calculated from mean CLT data. Statistical analysis was performed on CLT, plasmin activity assays, and flow cytometry data using the Student t test or, when multiple parameters were tested, 1-way analysis of variance with the Dunnett multiple comparison post hoc test. Values of P < .05 were considered significant.

Artificial surfaces can enhance the plasminogen activator capacity of αFXIIa.³¹  This prompted us to examine the effect of the “natural” activator polyP, of approximately the size (60-100 mers) of that found in platelets, in αFXIIa-mediated plasminogen activation. In line with previous observations, we show that αFXIIa is a relatively weak plasminogen activator.^13,15  However, inclusion of polyP₇₀ significantly augments the ability of αFXIIa to drive fibrinolysis, decreasing the CLT by 2.3-fold (105 ± 6 minutes vs 238 ± 14 minutes; P < .0001; Figure 1A). Similarly, polyP₇₀ accelerated αFXIIa-mediated plasmin generation (Figure 1B; P < .0001) in a dose-dependent manner, with concentrations as low as 4.4 μM, demonstrating a significant enhancement over αFXIIa alone (Figure 1C; P < .0001). PolyP had no direct effect on the activity of preformed plasmin (Figure 1D; P = .93). This finding indicates that polyP₇₀ accelerates αFXIIa-mediated conversion of plasminogen to plasmin. The derivative, βFXIIa, lacks the surface binding domain and is less efficient at stimulating plasminogen activation.¹⁶  The low level of plasmin generation observed with βFXIIa was not augmented by polyP₇₀, reflecting the requirement of the anion binding domain of αFXIIa for interaction with polyP₇₀ (Figure 1E; P = .71). Inclusion of Zn²⁺ in the reaction buffer did not further enhance the cofactor function of polyP₇₀ in this reaction (not shown). We found that platelet-derived polyP extracted from human platelets similarly enhanced the plasminogen activator capacity of αFXIIa (Figure 1F).

Because polyP is a known activator of FXII,^20,21,35  we examined its capacity to both stimulate FXII activation and subsequently stimulate its plasminogen activator function. CLTs were significantly longer with FXII (>300 minutes) compared with αFXIIa (238 ± 58 minutes) and, similarly, rates of plasmin generation were significantly lower with FXII compared with αFXIIa, presumably reflecting the time for transition of zymogen to protease. Nevertheless, polyP₇₀ exhibits significant cofactor activity in clot lysis and activity assays when initiated with FXII-polyP rather than with FXII alone (Figure 2A).

We next investigated the binding of polyP₇₀ to FXII derivatives and found it complexed with FXII and αFXIIa, but not with βFXIIa, confirming the critical role of the anion binding site for this interaction (Figure 2B). When in complex with polyP₇₀, autodegradation of αFXIIa was delayed (Figure 2C). The protective effect of polyP₇₀ on αFXIIa was analyzed using S2302 substrate (Figure 2D). Addition of excess substrate to the αFXIIa reaction at 2 hours generated a further increase in absorbance in the presence of polyP₇₀, whereas no change was observed with αFXIIa alone. No further increase in activity was observed when additional αFXIIa was added to reactions with polyP₇₀ at 2 hours, confirming that protein is not limiting (not shown). We further analyzed changes in αFXIIa activity in terms of cleavage of FXI but were unable to detect any changes in FXI activation after 360 minutes of incubation in the absence or presence of polyP₇₀ (not shown). These results indicate that although polyP₇₀ confers protection against autodegradation, the differences may be too minor to alter functional activity toward physiological targets.

We found that plasminogen bound immobilized polyP₇₀ and could be released by washing with high-salt buffer (Figure 3A). This is indicative of an electrostatic interaction with the polymer, as previously shown for other proteins.^30,34,36  Plasminogen circulates in 2 forms: the predominant Glu-plasminogen is described as the “closed” conformation, whereas the truncated Lys-plasminogen, cleaved at the C-terminus by plasmin, is in an “open” conformation and exhibits a shorter half-life. Enhanced binding of Lys-plasminogen to fibrin³⁸  and the activators tPA and uPA results in more rapid plasmin generation.^39,40  We observed faster plasmin generation and clot lysis with Lys-plasminogen (catalytic efficiency [CE] = 257.5 ± 21.0 M⁻¹s⁻¹) than Glu-plasminogen (CE = 18.5 ± 4.2 M⁻¹s⁻¹) when activated with αFXIIa (note the different scales in Figure 3B-C). PolyP₇₀ significantly augmented αFXIIa-mediated activation of Glu-plasminogen (CE = 41.1 ± 6.0 M⁻¹s⁻¹) and Lys-plasminogen (CE = 629.5 ± 75.6 M⁻¹s⁻¹) (Figure 3B) and accelerated clot lysis with both forms by ∼2.6-fold (Figure 3C). These data suggest that polyP does not impact on the transition of the closed (Glu) to open (Lys) conformation of plasminogen but directly facilitates cleavage to plasmin.

We have previously shown that polyP binds to fibrin(ogen) and alters the structure of the fibrin network³⁴ ; therefore, we assessed the impact on αFXIIa-mediated plasminogen activation. Fibrin significantly enhanced the plasminogen activator function of αFXIIa (Figure 4A) and, when combined, fibrin and polyP₇₀ acted in concert to further amplify αFXIIa-mediated plasminogen activation (Figure 4A; P < .001).

We analyzed whether other natural surfaces that stimulate FXII activation, such as RNA²²  and collagen,²⁴  modulated the plasminogen activator function of αFXIIa. RNA slightly enhanced plasmin generation by αFXIIa (Figure 4B; P < .001), but not to the same magnitude as polyP₇₀, whereas collagen was unable to stimulate activation. However, neither collagen nor RNA was as effective in shortening the CLT as polyP₇₀ (Figure 4C; P < .001).

We next addressed whether αFXIIa acted in concert with tPA to mediate fibrinolysis. Inclusion of αFXIIa with a concentration of tPA (15 pM) sufficient to induce clot lysis further shortened the CLT from 69 ± 8.6 minutes to 47 ± 2.7 minutes (P < .01; CLT of αFXIIa alone was >300 minutes). We then performed assays at a suboptimal dose of tPA (1 pM), which, alone, was not sufficient to induce lysis (CLT >300 minutes). Addition of αFXIIa to clots containing 1 pM tPA significantly shortened the CLT (180 ± 9.4 minutes), but not to the extent of polyP₇₀ (106 ± 6.5 minutes; P < .0001; Figure 4D). A marginal decrease in CLT was observed with both αFXIIa and tPA in the presence of polyP₇₀ (97 ± 1.4 minutes; P < .01). Similarly, visualization of lysis in real time (supplemental Video 1, available on the Blood Web site) demonstrated significantly faster lysis in the presence of both αFXIIa and tPA compared with tPA alone. Interestingly, the pattern of lysis in the presence of αFXIIa was different, with clots lysing from the edge inward, rather than proceeding with a defined lysis front as observed with tPA alone. These data illustrate that fibrin degradation is enhanced in the presence of both tPA and αFXIIa, compared with either activator alone.

Inclusion of PAI-1 and PAI-2 at plasma concentrations did not impact αFXIIa-mediated lysis (Figure 5A) or plasmin generation (Figure 5B) in the absence (not shown) or presence of polyP₇₀ (Figure 5A). In contrast, C1 inhibitor significantly attenuated αFXIIa-mediated clot lysis (Figure 5A; P < .0001) and plasmin generation (Figure 5B; P < .001), with or without polyP₇₀. HRG was also effective in downregulating CLT (Figure 5A; P < .0001) and plasminogen activation (Figure 5B; P < .001) by αFXIIa, with or without polyP₇₀. The inhibition of αFXIIa-mediated plasminogen activation by α₂-antiplasmin could not be examined due to its dominant inhibition of plasmin, but we found no discernible effects on direct inhibition of αFXIIa using S2302, with or without polyP₇₀ (not shown).

We examined the interaction of DL488-FXII with platelets by flow cytometry and found that binding to CVX/thrombin-stimulated platelets was significantly augmented compared with unstimulated platelets (62.6% ± 6.9% vs 22.4% ± 13.5%; P < .0001; Figure 6A). Platelets stimulated with CVX/TRAP-6 displayed a reduced capacity to bind DL488-FXII (44.7% ± 8.9%; P < .01). Inclusion of GPRP, to impede fibrin polymerization, markedly decreased the percentage of CVX/thrombin-stimulated platelets that bound DL488-FXII (33.4% ± 6.9%; P < .0001). Together, these data suggest that platelet-bound fibrin plays a crucial role in the association of FXII with the activated platelet surface.

We found a significantly higher degree of positivity for platelet-derived polyP on the surface of CVX/thrombin-stimulated platelets compared with unstimulated platelets (52% ± 2% vs 8.1% ± 4.0%; P < .0001). Surface-bound polyP was slightly decreased when platelets were stimulated with CVX/TRAP-6 (37.1% ± 7.2%; P < .01) and upon inclusion of GPRP (31.5% ± 12.1%; P < .01), suggesting a role for platelet-bound fibrin in its retention on the activated platelet membrane (Figure 6B).

Using fluorescence confocal microscopy, we examined the localization of FXII on the activated platelet surface stimulated with CVX/thrombin. Phosphatidylserine (PS)-positive platelets bound DL488-FXII in a single protruding “cap” on the platelet surface that is also rich in PS (Figure 6C; supplemental Video 2). DL488-FXII also bound to PS–negative spread platelets in a central diffuse pattern over the area of the granulomere (Figure 6D). Staining for platelet-derived polyP was dispersed over the activated membrane of PS-positive platelets (Figure 6E). Control experiments performed with nuclease, to degrade contaminating DNA and RNA, did not alter DAPI staining on activated platelets (supplemental Figure 1).

The location of platelet-derived polyP and FXII was studied in plasma clots formed in the presence of activated platelets. Platelet-derived polyP associates with the surface of activated platelets, particularly procoagulant PS-positive platelets and on platelet-bound fibrin (Figure 6F.) No DAPI staining was observed in clots formed in the absence of platelets (data not shown). DL488-FXII also decorated fibrin fibers extending from platelet aggregates, in clear colocalization with platelet-derived polyP.

In the present study, we reveal a cofactor function for platelet-derived polyP in modulating the plasminogen activator activity of αFXIIa. Platelet-derived polyP colocalized with αFXIIa on the fibrin matrix extending from platelet aggregates. Blocking fibrin polymerization with GPRP reduced binding of FXII, highlighting its importance in the accrual of FXII in platelet-rich areas. Our findings demonstrate that in the presence of platelet polyP, αFXIIa is an efficient plasminogen activator. In addition, we show that fibrin augments the plasminogen activator capacity of αFXIIa in a synergistic manner to polyP. There have been several reports on the modulation of fibrinolysis by polyP,^20,29,30  but to our knowledge, this is the first study to document a profibrinolytic function of platelet-derived polyP.

Although the plasminogen activator function of αFXIIa has been described previously,^15,16,31  it has been largely ignored due to unfavorable kinetics. Despite displaying a 20-fold lower catalytic efficacy for plasminogen than for uPA, the relative abundance of FXII (375 nM) in plasma implies that it may be relevant as a plasminogen activator.¹⁷  Nevertheless, studies on the role of αFXIIa as a direct plasminogen activator are limited.^19,41,42  Here, we demonstrate that polyP (of approximately the size found in platelets) and platelet-derived polyP significantly augment αFXIIa-mediated plasmin generation. PolyP accelerates αFXIIa-mediated activation of Glu- and Lys-plasminogen to a similar degree, indicating that it does not facilitate transition of the closed to the open conformation, nor does it directly impact on plasmin activity. We observed binding of polyP to FXII(a) and plasminogen, indicating that the cofactor function of this polymer is potentially mediated via a template mechanism on direct conversion of plasminogen to plasmin (Figure 7). PolyP also stimulates autoactivation of FXII to αFXIIa³⁵  and, indeed, affords some protection against autodegradation. This could be relevant in terms of time frame, because FXII may become activated during clot formation; however, by interacting with platelet-derived polyP, its activity may be protected, allowing it to subsequently participate in clot degradation. Indeed, a recent publication highlighted the dual role of FXII in supporting fibrin formation and degradation of the clot.⁴³  It is interesting to speculate on the complex role of this enzyme in modulating both fibrin formation (via generation of thrombin) and fibrin degradation (via plasmin). Further work is necessary to define the procoagulant and profibrinolytic properties of αFXIIa and the role of effector molecules such as polyP in these processes. Indeed, opposing functions of a hemostatic enzyme is not an uncommon phenomenon. The central enzyme thrombin is a prime example, because its activity can be directed from procoagulant to anticoagulant processes, depending on the effector molecule bound to exosite I or exosite II of the protease.⁴⁴ 

Fibrin also amplifies αFXIIa-mediated plasminogen activation, with maximal stimulation observed when both fibrin and polyP are present, suggesting that these molecules act synergistically to drive this process. PolyP has the capacity to bind to fibrin^29,30  and FXII(a), but these observations indicate that their respective binding sites on FXII(a) must be distinct. Fibrin is a well-established cofactor for tPA-mediated plasminogen activation. αFXIIa was able to act in conjunction with tPA to significantly accelerate fibrinolysis, suggesting that incorporation of these activators into the forming fibrin network may facilitate clot degradation, as depicted in the model shown in Figure 7. Interestingly, other surfaces that promote FXII activation and procoagulant function, specifically collagen²⁴  and RNA,²²  were not effective in promoting αFXIIa-mediated fibrinolysis. This finding suggests that cofactor molecules may drive αFXIIa activity toward distinct downstream target substrates.

Antifibrinolytic functions of polyP have been described in the past.^20,29,30  The first of these relates to enhanced activation of the thrombin activatable fibrinolysis inhibitor, due to acceleration of thrombin generation in the presence of polyP.²⁰  Activated thrombin activatable fibrinolysis inhibitor downregulates fibrinolysis by removing C-terminal lysine residues on fibrin that are important for binding plasminogen and tPA.⁴⁵  PolyP also exerts antifibrinolytic function by altering the structure of the forming fibrin network.^29,30  Our recent work has shown that this arises from impaired fibrin polymerization.⁴⁶  The modifications to the fibrin architecture by polyP alter its capacity to bind tPA and plasminogen, particularly that of partially degraded fibrin, thereby reducing tPA-mediated plasmin generation.³⁰  The profibrinolytic capacity of polyP described here, in terms of augmenting αFXIIa-mediated plasminogen activation, is intriguing, particularly in light of the fact that fibrin also plays a role in this process. It seems that the functions of polyP are more diverse than first thought and that this polymer may exert different levels of control over the hemostatic system, depending on timing and perhaps local concentrations of available reactants.

Of interest, PAI-1 and PAI-2 were ineffective in modulating polyP-αFXIIa-mediated plasminogen activation and fibrinolysis, whereas C1-inhibitor and HRG both effectively neutralized activity. C1-inhibitor is the predominant inhibitor of the contact pathway and regulates tPA-mediated plasminogen activation in situations when tPA is in excess over PAI-1.^47-49  C1-inhibitor has also been shown to bind to polyP.⁵⁰  Recent data highlighted the important role of HRG in modulating FXIIa activity and function.^51-53  HRG binds to FXIIa with incredibly high affinity in the presence of Zn²⁺ ions,⁵¹  and both are released from platelet α-granules upon activation.^54,55  HRG also associates with DNA and RNA and attenuates nucleic acid–driven activation of FXII.⁵²  Here, we show for the first time that HRG dampens αFXIIa activity directed toward the fibrinolytic pathway, adding to the complexity of this unusual adapter protein in regulation of hemostatic pathways.

To date, the binding sites for αFXIIa on platelets have not been elucidated, but unlike other coagulation factors, it does not bind directly to PS-positive platelets via Gla domains.⁵⁶  We have previously shown that FXII(a) binds to fibrin and is actively incorporated into clots,⁵⁷  and elegant flow studies confirmed that FXII(a) interacts with platelet-associated fibrin.⁵⁶  We observed significantly more FXII on the surface of platelets activated with CVX/thrombin than with CVX/TRAP-6. Blocking fibrin polymerization with GPRP during CVX/thrombin stimulation, but not CVX/TRAP-6 stimulation, significantly reduced the amount of FXII associated with platelets. Together, these data indicate that fibrin anchors FXII to the activated platelet membrane. PS-positive platelets display a cap of FXII on the activated surface that we have recently shown to be rich in fibrin(ogen) and plasminogen.⁵⁸  We also show that platelet-derived polyP is retained on the membrane of PS-positive platelets, but unlike FXII, fibrinogen, and plasminogen, it is homogenously distributed. Inhibition of fibrin polymerization attenuates the association of polyP with the platelet surface, consistent with its known affinity for fibrin,³⁰  but does not completely abrogate binding. This suggests the presence of an as-yet-unidentified second mechanism that mediates retention of polyP on activated platelets. Nevertheless, we also show that polyP can translocate from the “hot-spots” of PS-positive platelets into the surrounding fibrin network. FXII also decorates fibrin strands, accentuating the importance of platelet-bound fibrin in localizing FXII and plasminogen⁵⁸  in the vicinity of activated platelets. The release of polyP from stimulated platelets could stimulate activation of FXII on fibrin and enhance the plasminogen activator capacity of αFXIIa; in this sense, fibrin will be acting as a surface for its own destruction, as it does in tPA-mediated plasminogen activation.

Thrombi are composed of different regions, the inner “core” and the outer “shell,” which differ in their levels of platelet activation, aggregation, and packing.⁵⁹  The core comprises tightly packed, degranulated platelets encased in fibrin.⁵⁹  ADP released from dense granules regulates α-granule secretion,^60-62  forming dense regions that stabilize the platelet aggregate.⁶³  Concomitant release of polyP with ADP from dense granules suggests that it may be retained in these low-solute transport areas. α-Granule release occurs at lower agonist concentrations than dense granule release,^64,65  suggesting that platelets at the edge of the core and within the shell may release fibrinogen and bind fibrin before polyP secretion, thereby providing an anchor to retain polyP in the locale of the activated platelet aggregates.

Multiple questions remain over how the pleotropic effects of FXII(a) are mediated in biological systems. Further investigations on the role of αFXIIa as a plasminogen activator in vivo are warranted, particularly in light of the current interest in FXII as an antithrombotic target. It is plausible that different surfaces act as cofactors that direct the function of FXIIa to procoagulant and profibrinolytic pathways, analogous to mechanisms seen with other proteins in the hemostatic cascade. Platelet-bound fibrin acts as reservoir for polyP and αFXII(a) and, within the thrombus microenvironment, may preserve their functional activity. Clearly, there is still much to be learned about the mysterious FXII(a) and its contribution to various physiological processes.

The authors thank students Natasha Walker and Thomas Nolan for contributions to the project; the Microscopy and Histology Core Facility and the Iain Fraser Cytometry Centre at the University of Aberdeen for excellent advice and use of facilities; and Jeffrey Weitz from McMaster University, Canada, for the kind gift of HRG.

This work was supported by the British Heart Foundation grant FS/11/2/28579 (N.J.M. and A.S.L.) and the University of Aberdeen Development Trust (J.L.M. and N.J.M.). P.Y.K. is supported by an Early Career Award and by the New Investigator Fund from Hamilton Health Sciences.

Contribution: J.L.M. performed the research, analyzed the data, and wrote the manuscript; A.S.L., A.K., G.G., and C.B. performed the research and analyzed the data; P.Y.K. analyzed the data; and N.J.M. provided the initial concept of the study, supervised the research, analyzed the data, and wrote the manuscript.

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

Correspondence: Nicola J. Mutch, Institute of Medical Sciences, School of Medicine, Medical Sciences and Nutrition, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD, United Kingdom; e-mail: n.j.mutch@abdn.ac.uk.