Cationic Zinc is Required for Factor XII Recruitment and Activation by Stimulated Platelets and for Thrombus Formation in vivo

Abstract

Background:

Although divalent zinc (Zn²⁺) is known to bind FXII and affect its sensitivity to autoactivation, little is known about the role of Zn²⁺ in the binding of FXII to platelets, where FXII activation is thought to occur in vivo, and the function of Zn²⁺ during thrombus formation following vascular injury remains poorly understood.

Objectives:

To evaluate the role of Zn²⁺ in platelet-dependent FXIIa generation.

Methods:

FXII binding to platelets and FXII activation by stimulated platelets were assessed using flow cytometry and a platelet-dependent thrombin generation assay. The mouse cremaster laser injury model was used to evaluate the impact of Zn²⁺ chelation on thrombus formation in vivo.

Results:

Our data demonstrate that stimulated platelets support FXII-dependent thrombin generation and that FXII activation by platelets requires the presence of Zn²⁺. By contrast, thrombin generation by stimulated endothelial cells occurred independently of FXII and Zn²⁺. Using flow cytometry, we found that FXII-FITC binds to the surfaces of stimulated platelets in a specific and Zn²⁺-dependent manner, whereas resting platelets demonstrated minimal binding. Other physiologically-relevant divalent cations are unable to support this interaction. Consistent with these findings, the Zn²⁺-specific chelator CaEDTA confers thromboprotection in the mouse cremaster laser injury model without causing increased bleeding. We observed an identical phenotype in FXII null mice tested in the same system.

Conclusions:

Our results suggest a novel role for Zn²⁺ in the binding and activation of FXII at the platelet surface, an interaction that appears crucial to FXII-dependent thrombin generation but dispensable for hemostasis.

Introduction

Platelet-dependent thrombin generation is a key contributor to thrombosis and is the target of many currently available antithrombotic therapies. It is well-established that platelet procoagulant activity is due in part to the ability of stimulated platelet surfaces to ligate and co-localize humoral coagulation factors in a way that greatly accelerates thrombin generation. In addition to this function, it has been known since the 1970s that platelets can directly initiate the coagulation cascade via activation of factor XII (FXII), a circulating 80,000 kDa serine protease [1, 2]. Once FXII undergoes cleavage activation to factor XIIa (FXIIa), it is able to trigger the contact pathway of coagulation and downstream thrombin generation as well as activation of the kinin-kallikrein pathway [3, 4]. Interestingly, severe congenital deficiency of FXII in humans is not associated with a bleeding diathesis, and the physiologic relevance of FXII activation by platelets remained unclear for decades. More recently however, FXII blockade or deficiency has been shown in several animal models to protect against thrombosis [5–8], highlighting the potential therapeutic value of targeting FXII to prevent thrombosis without adversely affecting hemostasis. These observations suggest that the process of platelet-dependent FXIIa generation may be linked to pathologic processes in vivo and has intensified interest in understanding the interaction of platelets and FXII. Nevertheless, the mechanism by which FXII is recruited and activated by platelets remains poorly understood.

Cationic zinc (Zn²⁺) is a cofactor that plays an important role in several aspects of the contact system, including the binding of FXII and high molecular weight kininogen (HK) to endothelial cells [9, 10] and potentiation of FXII autoactivation upon binding to artificial surfaces [11–15]. FXII has been shown to bind weakly to stimulated washed platelets [16, 17], but the interaction of platelets and FXII in the presence of physiologic concentrations of Zn²⁺ has not been previously investigated, even though Zn²⁺ is likely required for FXII-dependent processes in vivo. Likewise, the function of Zn²⁺ in platelet-dependent FXIIa generation has not been studied. Here we show that FXII binding to the surfaces of stimulated platelets is greatly enhanced by concentrations of Zn²⁺ similar to those found in plasma. Further, we demonstrate that the Zn²⁺ typically available in normal platelet-rich plasma (PRP) is sufficient to support platelet-dependent FXIIa generation and that depletion of Zn²⁺ impairs thrombus formation in vivo. Taken together, our data indicate that Zn²⁺ plays a key role in the regulation of FXII recruitment and activation at the platelet surface during thrombus formation.

Materials and Methods

Materials

The PAR-1 agonist peptide (SFLLRN), ethylenediaminetetraacetic acid calcium disodium salt (CaEDTA), diethylenetriaminepentaacetic acid calcium salt (CaDTPA), fluorescein-5-isothiocyanate (FITC), soybean trypsin inhibitor (SBTI), and 2-deoxy-D-glucose (2DDG) were purchased from Sigma Aldrich. Anti-FXIIa antibody 3F7 was custom manufactured by Viva Biotech (Shanghai, China) based on complementarity-determining region (CDR) sequences obtained from the published international patent (WO2014207199A1), and its performance was confirmed by aPTT assay (Figure S1). Polyclonal anti-FXII western blot antibody was purchased from Affinity Biosciences. Purified human plasma-derived FXII was purchased from Enzyme Research Labs. C-terminal amide Gly-Pro-Arg-Pro (GPRP) was manufactured by the custom peptide division at Thermo Fisher Scientific. Anti-human tissue factor (clone IIID8) and anti-human factor VIIa (clone IIH2) antibodies were purchased from Biomedica Diagnostics. SN-20 fluorogenic thrombin substrate (BOC-L-FPR-ANSNH-C2H5) was purchased from Haematologic Technologies. Phycoerythrin-conjugated anti-human p-selectin (CD62P) was purchased from Thermo. S-2302 chromogenic FXIIa/kallikrein substrate was purchased from Diapharma. TNF-α was purchased from Millopore Sigma. The source of Zn²⁺ in all experiments was ZnCl₂.

Preparation of Human Platelets

Whole blood was drawn into 0.4% sodium citrate (final concentration w/v) and centrifuged at 160 rcf for 10 minutes to generate PRP. To generate washed platelets, acid citrate dextrose (ACD) solution was added to PRP (final concentration 20% v/v), and platelets were washed twice by centrifugation at 650 rcf in Hepes-buffered Tyrode’s solution with glucose (HTG, 134 mM NaCl, 0.34 mM Na₂HPO₄, 2.9 mM KCl, 12 mM NaHCO₃, 20mM HEPES, and 1mM MgCl₂). All procedures were performed with the approval of the Institutional Review Board at Beth Israel Deaconess Medical Center.

Platelet-dependent Thrombin and FXIIa Generation Assays

Platelet-dependent thrombin and FXIIa generation assays were performed using a protocol modified from Jurk, et al. [18]. In the first step, either 50 μM SFLLRN or vehicle was added to 200 ul single-donor PRP (platelet count 250 × 10³/μl) anticoagulated with 10 mM GPRP to prevent fibrin polymerization and containing increasing amounts of CaCl₂ (4, 6, 8, 10, and 12 mM). A calcium titration was necessary due to donor-dependent differences in plasma calcium levels as well as subtle variability in blood draw techniques. The PRP mixture was incubated at room temperature for one hour, after which it was diluted 1:5 in PBS assay buffer containing 0.5 mM of the FXIIa-specific substrate S-2302, 12 mM EDTA (Sigma), 30 μM SBTI, and 2 units/ml hirudin, and absorbance was measured at 405 nm for 20 minutes. The concentration of Ca²⁺ that provided maximum signal separation between resting and SFLLRN-stimulated platelets was then chosen for use in the second step. In this phase, the inhibitor being tested was added to PRP containing a fixed concentration of Ca²⁺ and anticoagulated with GPRP, which was then stimulated with SFLLRN, incubated, and read in a manner identical to that used in Step 1. The platelet-dependent thrombin generation assay was performed in a similar fashion, except PBS assay buffer (1.05 mM KH₂PO₄, 155 mM NaCl, 3 mM, Na₂HPO₄-7H₂O, pH 7.4, Gibco) containing the thrombin-specific substrate (SN-20) at 50 μM and 12 mM EDTA was substituted for S-2302 containing buffer, and fluorescence was read by excitation at 352 nm and emission at 470 nm.

For mouse platelet-dependent FXIIa generation assays, we obtained pooled citrated PRP from two groups of five animals each that were treated 30 minutes previously with either vehicle or 250 mg/kg CaEDTA (platelet count 225 × 10³/μl). GPRP (final concentration 10 mM) was added to the PRP as well as 0, 2, 4, 6, or 8 mM of CaCl₂ (total volume per condition: 100 μl), and the sample was then stimulated with 500 μM of the PAR-4 agonist AYPGKF or vehicle (2 conditions per calcium concentration, 10 conditions total). After incubation for two hours at room temperature, each condition was diluted 1:5 in PBS assay buffer containing 0.5 mM S-2302, 12 mM EDTA, 30 μM SBTI, 300 μg/ml apixaban, and 2 units/ml hirudin, and absorbance was measured at 405 nm for 30 minutes. For each calcium concentration, the level of FXIIa generation in the AYPGKF-stimulated condition was corrected against the corresponding resting platelet condition (background), and results are reported as the peak FXIIa generation across the five calcium concentrations.

Activated Partial Thromboplastin (aPTT) Assay

The PTT assay was performed using the PTT Automate assay (Stago) on the Start4 coagulometer (Stago) per the manufacturer’s instructions.

HUVEC Thrombin Generation Assay

HUVEC (Lonza, P1-P5) monolayers cultured in Endothelial Growth Medium (EGM)-2 (Lonza) on gelatin coated culture vessels were sub-cultured on 96-well plates. Confluent HUVEC monolayers were stimulated with vehicle or TNF-α (10 ng/mL) in endothelial basal medium (EBM)-2 (Lonza) for 4 hours at 37°C. For thrombin generation experiments, stimulation medium was removed from wells and replaced with a 25 μL of 80% platelet poor plasma (PPP) mixture composed of pooled PPP from 4–5 donors in 5 mM GPRP and HEPES Buffered Saline (20 mM HEPES, 150 mM NaCl, pH 7.4, HBS) containing 10 mM CaCl₂, with or without inhibitors for 15 minutes. Following incubation, thrombin levels were measured with the addition of SN-20 fluorogenic substrate in PBS with 2 mM CaCl₂. Fluorescence was measured every minute for 60 minutes using the Synergy4 plate reader (Biotek). Thrombin levels were determined by converting reaction rate to units/ml.

Generation of Recombinant FXII

Triple mutant FXII (R334A, R343A, and R353A, which we designated “FXII-3T”) [19] was cloned into a pcDNA^™ 3.4 backbone using Gibson Assembly® and was transiently transfected and expressed in Expi293F^™ cells for 96 hours. Following harvesting, conditioned media was treated with Sodium Acetate pH 5.2 to a final concentration of 50mM and applied to a 5mL HiTrap Heparin HP column (GE Lifesciences). Protein was eluted off the column using a linear gradient of 2M NaCl. Peak fractions were pooled and applied to a 5mL Strep-Tactin®XT Superflow® high capacity cartridge (IBA Lifesciences) followed by elution by buffer containing 50mM biotin. Peaks were pooled and concentrated for injection on a Superdex^™ S200 Increase 10/300 GL. The peak corresponding to soluble protein was combined and concentrated, and dialyzed into TBS storage buffer (20 mM Tris, 150 mM NaCl, pH 8.3) to be used as the final product.

Platelet Flow Cytometry

Washed platelets (400–500K/μl) were stimulated with 50 μM SFLLRN for 10 minutes. Platelets were then diluted in HTG at 10,000 per μl together with 2 mM CaCl₂, Zn²⁺ at 0, 3, or 10 μM, and 100 nM FITC-labeled FXII, FXIIa (Enzyme Research Labs), or FXII-3T. The mixture was incubated at room temperature for 20 minutes and then diluted 1:10 in HTG containing the condition-dependent Zn²⁺ concentration and analyzed by flow cytometry on a Gallios instrument (Beckman Coulter). For experiments with 2-deoxy-D-glucose (2DDG) and sodium azide, platelets were pre-incubated with these two agents for 30 minutes prior to activation with SFLLRN.

Mouse Laser Injury Model

The intravital mouse cremaster arteriole laser injury model was performed as described previously [20, 21]. The operator was blinded to the genotype and treatment assignment (vehicle vs. CaEDTA) of each subject animal. Digital images were captured with an Orca Flash 4.0v2 sCMOS camera (Hamamatsu Photonics K.K., Shizuoka Pref., Japan). Representative images are presented, and the median curves include the full data. The kinetics of platelet thrombus formation were analyzed using SlideBook 6.0 (Intelligent Imaging Innovations, Denver, CO).

Male wild-type C57/BL6J mice were purchased from Jackson Laboratory. Male FXII null mice were the kind gift of Dr. David Gailani. All protocols involving the use of animals were in compliance with guidelines from the National Institutes of Health and the Beth Israel Deaconess Medical Center Institutional Animal Care and Use Committee.

Size Exclusion Chromatography

Size exclusion chromatography was performed on a Superdex 200 column, and HBS with varying concentrations of Zn²⁺ was used as a running buffer. Prior to loading on the column, purified human FXII was dissolved at 0.66 mg/ml in HBS with the indicated concentrations of Zn²⁺ and incubated for 10 minutes at room temperature.

Measurement of Plasma Labile Zinc Levels (Heparinized PPP)

Labile Zn²⁺ levels in heparinized human plasma were measured using the chromogenic Zinc Assay Kit from Abcam (ab102507) per the manufacturer’s instructions.

FITC-labeling of FXII

Plasma-derived purified FXII (Enzyme Research Laboratories) was dialyzed into TBS overnight at 4° C, resulting in a FXII solution of approximately 1 mg/ml. FITC-Celite® (Sigma) was dissolved at 50 mg/mL in anhydrous DMSO (Thermo Fisher, D12345) to form a working solution. This working solution was added to the dialyzed FXII to a final concentration of 1 mg/mL of FITC. The reaction mixture was vortexed briefly and incubated in the dark at ambient temperature for 20 minutes. After incubation, the excess Celite was removed by centrifugation at 10,000 g for 5 minutes. The labeled FXII was then dialyzed against PBS overnight at 4°C to remove excess FITC.

Quantification of Platelet FXII Binding Sites

In order to quantify the number of platelet FXII binding sites, a calibration curve was created using standardized Quantum^™ FITC-5 Molecules of Equivalent Soluble Fluorochrome (MESF) beads (Bangs Laboratories). The standard curve was generated by plotting the fluorescent intensities of beads with known MESF values, resulting in the following equation:

$$MFI=m\left(\mathit{\text{MESF}}\right)+b$$

where MFI is the mean fluorescence intensity reading at any given MESF level, and m and b are derived from best-fit linear regression.

In the second step, SFLLRN-stimulated platelets were added at 10³/uL to a series of tubes containing 300 μM Zn²⁺ in HTG and either 0, 100, 300, 1000, 3000, or 5000 nM FXII-FITC and incubated for 20 minutes before collecting fluorescent reads (Gallios, Beckman Coulter). The value recorded as the curve leveled off was used as the saturation point for FXII binding. The fluorescent read at this saturation point was used to interpolate the MESF value from the standard curve that was previously derived. Subsequently, the number of binding sites could be determined by dividing the MESF value by the fluorophore per protein molecule ratio (F:P) of FXII-FITC:

$$\# \mathit{\text{binding}} \mathit{\text{sites}}= \frac{\mathit{\text{MESF}}}{F:P}$$

Results

Platelet-dependent Thrombin Generation Requires FXIIa and Zinc

The mechanism by which stimulated platelets directly activate the coagulation cascade to generate thrombin is not understood. In order to evaluate the role of FXII in platelet-dependent thrombin generation, we stimulated PRP with the PAR-1 agonist peptide SFLLRN (TRAP-6) in the presence of inhibitory antibodies directed against FXIIa (3F7), tissue factor (TF), or factor VIIa (FVIIa) and measured thrombin generation by use of a fluorogenic substrate. Blockade of FXIIa resulted in nearly complete abrogation of platelet-dependent thrombin generation, while inhibition of FVIIa and TF had no effect (Figure 1A).

Figure 1: Thrombin generation by stimulated platelets is FXII and zinc dependent.

(A) Thrombin generation was measured after platelet rich plasma (PRP) was stimulated with 50 μM of the PAR-1 receptor agonist peptide SFLLRN in the presence of either non-immune control antibody (SFLLRN) or inhibitory antibodies against FXIIa (3F7, left, N=12 replicates), factor VIIa (middle, N=12 replicates), or tissue factor (right, N=15 replicates). Each experiment was performed independently two times with 9 replicates per condition. (B) PRP was stimulated with 50 μM SFLLRN in the presence of increasing concentrations of the Zn²⁺-specific chelator CaEDTA and FXIIa generation was recorded (left, N=9 replicates per condition). P-selectin (CD62P) expression was measured in resting and SFLLRN-stimulated platelets in the presence or absence of 40 mM CaEDTA (right, N=3 replicates per condition). (C) By comparison, the activated partial thromboplastin time (aPTT) was recorded in the presence of increasing concentrations of CaEDTA, N=3 replicates per condition. **P<0.05 compared to activated platelet control by Kruskal-Wallis test with Dunn’s post-test for multiple comparisons.

In light of previous observations that cationic zinc (Zn²⁺) significantly enhances the activation of purified FXII in buffer and PPP [11–14], we next determined whether Zn²⁺ is required to support platelet-dependent FXIIa generation ex vivo. We found that at baseline, heparinized pooled human PPP contained labile Zn²⁺ at a mean ± SEM concentration of 12.16 ± 0.22 μM. PRP was incubated with increasing concentrations of the Zn²⁺-specific extracellular chelator ethylenediaminetetraacetic acid calcium disodium salt (CaEDTA) [22–24], followed by measurement of FXIIa generation. CaEDTA inhibited platelet-based FXIIa generation in a dose-dependent manner without affecting the activation state of platelets as reflected by P-selectin (CD62P) expression (Figure 1B). Correspondingly, plasma labile Zn²⁺ declined to undetectable levels in the presence CaEDTA (Figure S2A). We sought to confirm these results using a different Zn²⁺-specific chelator, calcium diethylenetriaminepentaacetic acid (CaDTPA) [25] and found that FXIIa generation was likewise inhibited (Figure S2B). The presence of CaEDTA did not appear to influence kaolin-mediated FXII autoactivation or the cleavage of FXII to FXIIa by kallikrein in vitro (Figure S3). By contrast, Zn²⁺ chelation did not prolong coagulation times as measured by the activated partial thromboplastin time (aPTT) performed on PPP (Figure 1C). These data demonstrate that stimulated platelets can directly activate coagulation by generating FXIIa through a process requiring the Zn²⁺ available in PRP.

Stimulated Endothelial Cells Do Not Require FXII to Generate Thrombin

Endothelial cells stimulated in vitro are known to support thrombin generation by expressing procoagulant molecules such as TF and phosophatidylserine [26]. FXII has been shown to bind endothelial cells via a trimolecular membrane receptor complex in the presence of Zn²⁺ [9], but the effect of the FXII-endothelial cell interaction on thrombin generation has not been described. We sought to understand whether, like platelets, endothelium supports thrombin generation via a FXIIa-dependent mechanism. We therefore interrogated the roles of zinc and FXIIa in thrombin generation by stimulated human umbilical vein endothelial cells (HUVECs) (Figure 2). HUVECs were overlaid with citrated human PPP and stimulated with tumor necrosis factor alpha (TNF-α) in the presence of CaEDTA, 3F7, or function-blocking antibodies against FVIIa or TF. Inhibition of either FVIIa or TF significantly impaired thrombin generation by TNF-α stimulated HUVECs, whereas neither CaEDTA nor 3F7 was able to inhibit HUVEC procoagulant activity, suggesting that stimulated endothelial cells and platelets utilize different arms of the coagulation system to initiate thrombin generation.

Figure 2: Thrombin generation by stimulated endothelial cells is not FXII or zinc dependent.

Human umbilical vein endothelial cells were stimulated with 10 ng/ml TNF-α, and thrombin generation was recorded in the presence of inhibitory antibodies (10 μg/ml) or calcium EDTA (40 mM) as indicated. Antibody 3F7 was used for FXIIa inhibition. N=4 replicates per condition, **P<0.05 for comparison to TNF-α condition.

FXII Binding to the Platelet Surface is Zinc-dependent and Specific

Given the finding that stimulated platelets can support FXII activation, we sought to assess FXII binding to the platelet surface. Resting or stimulated washed platelets were incubated with FITC-labeled FXII (FXII-FITC) and increasing concentrations of Zn²⁺ prior to analysis by flow cytometry (Figure 3A). Consistent with previous literature [16], in the absence of Zn²⁺ washed platelets stimulated with SFLLRN displayed increased binding of FXII-FITC relative to resting platelets. However, physiologic concentrations of Zn²⁺ greatly enhanced FXII-FITC binding to stimulated and, to a lesser degree, resting platelets. The binding of FXII-FITC was readily reversible with unlabeled FXII (Figure 3B), indicating the specific nature of the FXII-FITC/platelet interaction. We next recombinantly generated a single-chain form of FXII (FXII-3T) in which the major cleavage sites at R334, R343, and R353 have each been replaced with alanine [19]; this form of FXII cannot be converted to FXIIa. When we assayed the ability of FITC-labeled FXIIa and FXII-3T to bind the platelet surface in the presence of 10 μM ZnCl₂, we found that these two forms of FXII performed similarly to FXII-FITC zymogen (Figure S4). We next sought to assess the possibility that Zn²⁺ was inducing the formation of FXII aggregates that were non-specifically adhering to platelets in our assay. We found that the presence of Zn²⁺ even at supraphysiologic concentrations did not lead to the formation of detectable FXII aggregates in solution (Figure S5). Previous data have suggested that FXII binds to the GPIbα receptor on the platelet surface [27], though this finding has never been corroborated. We therefore evaluated whether platelets supported FXII-FITC binding in an energy-independent manner as would be expected with GPIbα, a constitutively active receptor. FXII-FITC binding was largely prevented when platelets were pre-incubated with 2-deoxy-D-glucose (2DDG) and sodium azide in order to deplete available energy stores (Figure 3C), suggesting that the Zn²⁺-mediated binding of FXII-FITC to stimulated platelets is regulated by an energy-dependent process. Because Zn²⁺ is itself a platelet activator [28], we evaluated the level of CD62P expression on resting platelets after exposure to Zn²⁺ (Figure S6). At 10 μM Zn²⁺, we noted a small but measurable increase in platelet CD62P expression, suggesting that platelet activation may be responsible in part for the increased binding of FXII-FITC observed with resting platelets in the presence of Zn²⁺.

Figure 3: FXII binding to the platelet surface is specific and zinc-dependent.

(A) The binding of FXII-FITC (100 nM) to resting platelets and platelets stimulated with 50 μM SFLLRN was measured by flow cytometry in the absence or presence of Zn²⁺. Each experiment was performed with 6 replicates, and representative scatter plots are shown here. (B) Binding of FXII-FITC (100 nM) to stimulated platelets in the presence of 10 μM Zn²⁺ and increasing molar equivalents of unlabeled FXII was measured by flow cytometry (representative plot from 3 independent experiments with 3 replicates for each condition). (C) Binding of FXII-FITC to stimulated platelets was measured in the presence or absence of 10 μM Zn²⁺ and with or without 30-minute preincubation with 0.2% sodium azide and 4 mg/ml 2-deoxy-D-glucose (2DDG), N=6 replicates, **P<0.0001 (D) FXII-FITC (100 nM) binding to stimulated washed platelets was recorded by flow cytometry at increasing concentrations of the indicated divalent metal cations (N=3 replicates per condition).

We next tested the ability of other physiologic divalent cations to support binding of FXII-FITC to the platelet surface. SFLLRN-stimulated washed platelets and FXII-FITC (100 nM) were incubated with increasing concentrations of divalent calcium, magnesium, manganese, or zinc, and FXII-FITC platelet binding was measured by flow cytometry (Figure 3D). Enhanced FXII-FITC binding could be readily detected when Zn²⁺ was included, but binding remained unchanged in the presence of Ca²⁺ (CaCl₂), Mg²⁺ (MgCl₂), and Mn²⁺ (MnCl₂). Using reagent beads containing known amounts of fluorochrome (Quantum^™ MESF), we estimated that each stimulated platelet maximally bound on average 135,922 ± 4117 molecules of FXII-FITC at 300 μM Zn²⁺. These data are consistent with the unique role that Zn²⁺ is known to play in the biology of the contact system.

Zinc Chelation Prevents Thrombus Formation in vivo Without Increasing Bleeding

In order to interrogate the role of Zn²⁺ in thrombus formation in vivo, we used CaEDTA to deplete Zn²⁺ in the mouse cremaster arteriole laser injury model of thrombosis. Mice received intravenous infusion of CaEDTA (250 mg/kg) or vehicle control immediately prior to laser injury. Treatment with CaEDTA led to significantly decreased platelet accumulation and fibrin formation at sites of vascular injury (Figure 4A–E). When Zn²⁺ levels were assessed in mouse plasma ex vivo, we found that CaEDTA completely eliminated detectable levels of labile Zn²⁺ from mouse plasma (Figure 4F) in a manner similar to that seen with human plasma (Figure S2A). Despite preventing thrombus formation in this model, CaEDTA did not lead to an increased tendency for bleeding as gauged by a tail transection assay (Figure 4G and S7). In a separate experiment, we assayed FXIIa generation ex vivo in stimulated pooled PRP taken from mice treated with either vehicle or 250 mg/kg CaEDTA (N=5 animals per group). Our results confirm that platelet-dependent FXIIa generation was significantly reduced in mice treated with CaEDTA (Figure S8).

Figure 4:

Zinc chelation prevents thrombus formation in vivo. (A) Mice infused with 250 mg/kg of CaEDTA or vehicle control underwent quantitative assessment of thrombus formation (platelet accumulation in red, fibrin formation in green) in the cremaster arteriole laser injury model, and representative intravital images are shown, along with median curves and AUC distributions for platelet accumulation (B-C) and fibrin formation (D-E). A total of 56 thrombi were evaluated in each group. (F) Mice were treated with either CaEDTA (250 mg/kg) or vehicle control (N=4 in each group), and plasma Zn²⁺ concentrations were measured at 1 hour. (G) Tail bleeding times were compared between CaEDTA and vehicle treated animals (N=6 per group). **P<0.05 compared to vehicle control.

Using homozygous FXII knockout (FXII^−/−) mice, we compared the in vivo effects of Zn²⁺ chelation with those of FXII deficiency in this model. FXII^−/− mice significantly protected from thrombosis following laser-induced injury to the cremaster arterioles (Figure 5A–E), despite these mice having no demonstrable bleeding diathesis [7]. These data suggest that Zn²⁺ plays a role in thrombosis but not hemostasis and support the concept that Zn²⁺ may be necessary to facilitate FXIIa generation in vivo.

Figure 5:

Factor XII deficiency inhibits thrombus formation in vivo. (A) Wild-type (WT) or FXII knockout mice (FXII KO) mice underwent quantitative assessment of thrombus formation in the cremaster arteriole laser injury model (platelet accumulation in red, fibrin formation in green). Representative intravital images are shown, together with median curves and AUC distributions for platelet accumulation (B-C) and fibrin formation (D-E). **P<0.05 compared to WT.

Discussion

Given the longstanding observation that platelet function is critical to initiating and propagating arterial thromboses, platelets are likely a major contributor of FXIIa at sites of vascular injury [29]. Nevertheless, the interaction between platelets and FXII remains uncharacterized. Cationic zinc lies at the intersection of the twin processes of FXII surface binding and FXIIa generation; therefore, understanding the function of Zn²⁺ in this setting is key to elucidating the broader mechanism of FXII in thrombus formation. Our data provide direct evidence that Zn²⁺ plays a crucial role in the process of FXII ligation and activation at the stimulated platelet surface and demonstrate the importance of Zn²⁺ in thrombus formation.

An inhibitory anti-FXIIa antibody completely abrogated platelet-dependent thrombin generation in plasma, whereas blockade of TF and FVIIa appeared to have no effect, indicating that direct activation of coagulation by stimulated platelets occurs via the contact pathway exclusively. By contrast, endothelial cells stimulated with TNF-α mediate thrombin generation by initiating the extrinsic arm of the coagulation cascade, with little need for FXII. These findings provide a previously unrecognized basis for thrombin generation assays performed in PRP [18, 30, 31] during which addition of thrombin to PRP leads to significant downstream thrombin generation. Our use of SFLLRN as a platelet agonist rules out an enzymatic contribution to thrombin generation from the small amounts of thrombin used as the initiator in some versions of this assay, which is thought to mimic the physiologic process of thrombus propagation driven by feedback activation from the “thrombin burst.” During this phase, coagulation factors V and VIII undergo activation and greatly increase the reaction rates of the tenase and prothrombinase complexes at the platelet surface [32]. While this model assumes that the initial source of thrombin is the TF-FVII pathway, our data indicate that FXIIa generation quickly becomes an important, perhaps the dominant, contributor once stimulated platelets are present in large numbers at sites of vascular injury. Platelet-dependent FXIIa generation therefore provides a plausible explanation for the observation that FXII blockade or deficiency prevents arterial thrombosis in preclinical models [7, 33, 34].

Given the importance of FXIIa generation for platelet-dependent thrombin generation, we further explored the FXII-platelet interaction. Platelet FXIIa generation in plasma was abolished in the presence of Zn²⁺-specific chelators, indicating that Zn²⁺ is required for the binding and/or activation of FXII at the platelet surface. Similarly, recent work by Wang and colleagues has demonstrated that FXII autoactivation by polyphosphate as well as FXII activation by kallikrein in buffer are Zn²⁺-dependent processes, with peak Zn²⁺ cofactor activity occurring between 5 and 10 μM of the cation [15]. Resting platelets demonstrated minimal binding to FXII, whereas stimulated platelets were able to bind FXII in a manner that was greatly enhanced by the presence of Zn²⁺ and appeared to be specific. Moreover, because FXII binding to stimulated platelets required metabolic energy, it is intriguing to speculate that an activatable platelet surface protein and/or phospholipid exposure may serve as the principal platelet binding partner for FXII. At variance with a previous report demonstrating an in vitro interaction between FXII and platelet glycoprotein 1bα (GP1bα) [27], it appears unlikely based on our data that this role is played by a constitutively active surface molecule such as the GP1b-V-IX complex.

It has been proposed that FXII has between 4 and 8 discrete binding sites for Zn²⁺ localized to the fibronectin type II domain of its heavy chain [11, 12, 35, 36], and Zn²⁺ potentiates contact-driven FXIIa generation, likely by inducing FXII to assume a conformational state that is amenable to autoactivation [11, 12]. FXII almost certainly circulates bound to one or more Zn²⁺ atoms. Further studies will be needed to determine whether Zn²⁺ acts by facilitating FXII autoactivation, by enabling FXII binding to the platelet surface, or via both of these mechanisms. By contrast, zinc is not required for FXII activation in standard clinical coagulation assays such as the activated partial thromboplastin time (aPTT), suggesting that platelet-mediated FXIIa generation may represent a physiologic process that is distinct from that triggered by artificial surfaces. In interpreting our results, it is important to note that while CaEDTA is considered a highly specific Zn²⁺ chelating agent under most physiologic circumstances [22, 23, 37], it does have activity as a chelator for several non-physiologic and/or trace metal cations (e.g. thorium, palladium, uranium, cobalt, etc.), and we cannot rule out the possibility that chelation of one or more of these other elements could be contributing to our findings.

An important consideration is whether endogenous Zn²⁺ is sufficient under physiologic conditions to support the binding and activation of FXII by platelets. The availability of adequate Zn²⁺ in platelets and plasma for these reactions is strongly corroborated by our observation that Zn²⁺ chelation in PRP abrogates platelet-dependent FXIIa generation. It is also noteworthy that the concentration of Zn²⁺ in platelets is approximately 25- to 60-fold higher than plasma and that this zinc resides largely in alpha granules [9, 38–40]. Upon stimulation at sites of vascular injury, platelets are thought to secrete Zn²⁺ into the local microenvironment [9, 38], which allows them to serve as an ancillary source of Zn²⁺ and perhaps also provide an extra layer of regulatory control for FXIIa generation. Consistent with this idea, mice treated with CaEDTA were protected from thrombosis without experiencing impaired hemostasis, a phenotype identical to that observed with FXIIa blockade in the same model.

Although total plasma Zn²⁺ levels are in the range of 10–20 μM [41, 42], a remarkable degree of disagreement persists regarding the level of protein unbound or “free” Zn²⁺, which has proven difficult to measure precisely. Estimates range between 1 nM and 2 μM depending on the method used [38, 43–45]. However, free Zn²⁺ does not accurately reflect the true exchangeable or “labile” pool of zinc that is available to participate in biological reactions; it has been estimated that plasma in fact contains on average 8.1 ± 4.8 μM of labile zinc [46], similar to our own readings of ~12 μM. This is because approximately 80% of plasma Zn²⁺ is bound loosely to albumin [41] and serves as a ready source of Zn²⁺ in the body. Albumin contains two Zn²⁺ binding sites, including a low-affinity site with a K_d of ~65 μM [47]. By comparison, the high-affinity Zn²⁺ binding site on FXII is associated with a K_d of approximately 0.1 μM [12], suggesting that FXII is able to draw on the pool of labile plasma Zn²⁺. Therefore, our use of 10 μM Zn²⁺ in these experiments is likely similar to the plasma concentration of Zn²⁺ available in vivo. Taken together, these results indicate that Zn²⁺ is central to the process of platelet-dependent FXIIa generation and form the basis for further experiments aimed at elucidating the interaction between FXII and potential platelet surface binding partners.

Essentials.

• The mechanism of FXII recruitment and activation and the platelet surface remains unclear

• We find that cationic zinc (Zn²⁺) is necessary for platelet-dependent FXII activation

• Stimulated endothelium generates thrombin via a FXII and Zn²⁺-independent mechanism

• Zn²⁺ chelation blocks thrombus formation and phenocopies FXII deletion in vivo