Neutralizing blood-borne polyphosphate in vivo provides safe thromboprotection

Abstract

Polyphosphate is an inorganic procoagulant polymer. Here we develop specific inhibitors of polyphosphate and show that this strategy confers thromboprotection in a factor XII-dependent manner. Recombinant Escherichia coli exopolyphosphatase (PPX) specifically degrades polyphosphate, while a PPX variant lacking domains 1 and 2 (PPX_Δ12) binds to the polymer without degrading it. Both PPX and PPX_Δ12 interfere with polyphosphate- but not tissue factor- or nucleic acid-driven thrombin formation. Targeting polyphosphate abolishes procoagulant platelet activity in a factor XII-dependent manner, reduces fibrin accumulation and impedes thrombus formation in blood under flow. PPX and PPX_Δ12 infusions in wild-type mice interfere with arterial thrombosis and protect animals from activated platelet-induced venous thromboembolism without increasing bleeding from injury sites. In contrast, targeting polyphosphate does not provide additional protection from thrombosis in factor XII-deficient animals. Our data provide a proof-of-concept approach for combating thrombotic diseases without increased bleeding risk, indicating that polyphosphate drives thrombosis via factor XII.

Figure 1. Cloning and expression of PPX mutants.

(a) Scheme of full-length PPX and PPX deletion mutants lacking various domains. Dark C-terminal squares represent the stop codons, and numbers on top indicate residues. All constructs were fused to an N-terminal 6xHis-tag. Affinity purified proteins were separated by SDS–polyacrylamide gel electrophoresis and visualized by (b) Coomassie brilliant blue staining or (c) western blotting with an antibody against the 6xHis-tag. A representative photographic film of three independent experiments is shown.

Figure 2. PPX variants binding to polyanions.

(a,b) Immobilized long-chain (LC, a) or short-chain polyP (SC, b) was incubated for 60 min with 10 nM purified, full-size PPX, PPX deletion mutants or PPX_Δ12 preincubated with polyP. Bound PPX variants were detected using an antibody against the 6xHis-tag, an HRP-coupled secondary antibody and substrate reaction. The cartoon shows the enzyme-linked immunosorbent assay set-up with ——=LC polyP and —=SC polyP, =PPX variant, =6xHis-tag antibody and =HRP-coupled detection antibody. Shown are relative amounts of PPX variants binding to polyP. Data blotted relative to full-size PPX, set to 1.0. Mean±s.e.m., n=4, ***P<0.001 by Student's t-test. (c) Gel mobility shift assay of full-size PPX, PPX_Δ1, PPX_Δ2, PPX_Δ12 and PPX_Δ124 binding to polyP. Increasing concentrations of PPX mutants (0–400 nM) were incubated with polyP (7.5 μg ml⁻¹) for 30 min, and reaction mixtures containing 0–8 pmol mutant protein per lane were resolved on 1% agarose gels. PolyP was visualized with DAPI-negative staining and synthetic polyP with mean chain length of 383 and 637 phosphate units served as molecular size standard. (d) PPX_Δ12 (400 nM) was incubated for 30 min with increasing concentrations of various polyanions including LC polyP, SC polyP, dextran sulfate (DXS), oversulfated chondroitin sulfate (OSCS), DNA, RNA, chondroitin sulfate (CS), dermatan sulfate (DS) and heparan sulfate (HS). PPX_Δ12/polyanion complexes were dissolved on urea-polyacrylamide gels and PPX_Δ12 protein was stained with Coomassie brilliant blue. The dashed line gives PPX_Δ12/polyanion complex formation assessed by densitometric scans. A representative gel of three independent experiments is shown.

Figure 3. PPX specifically degrades polyP.

Concentration- and time-dependent hydrolysis of LC and SC polyP by PPX. (a) LC and SC polyP (50 μg ml⁻¹ each) were incubated for 30 min with PPX (0–160 μg ml⁻¹). Reaction mixtures were separated on 10% urea-polyacrylamide gels, polyP was DAPI-negative stained and signals were blotted relative to buffer-treated polyP (100%). Mean±s.e.m., n=4. (b,c) LC and (d,e) SC polyP (50 μg ml⁻¹ each) were incubated with 10 μg ml⁻¹ PPX. Aliquots of 10 μl were taken at indicated time points, resolved on urea-polyacrylamide gels and visualized by negative DAPI staining. Synthetic polyP with mean chain length of 39, 97, 383 and 637 phosphates were loaded as molecular size standard. PolyP incubated with PPX in the presence of the inhibitor heparin (50 μg ml⁻¹) is shown in the last lanes. Bars are mean±s.e.m., from four independent experiments, **P<0.01, ***P<0.001 versus 0 min by one-way analysis of variance (ANOVA) and by Student's t-test versus heparin addition. (f) DNA (1 μg) and RNA (1 μg) were treated with buffer (w/o), PPX (10 μg ml⁻¹), DNase (0.1 mg ml⁻¹) or RNase (0.5 mg ml⁻¹) for 30 min, respectively, and resolved on agarose gels. (g,h) Heparin (10 μg ml⁻¹) or polyP (50 μg ml⁻¹) was treated with buffer (w/o), PPX (10 μg ml⁻¹), heparinase I (1 U ml⁻¹) or RNase (0.5 mg ml⁻¹) for 30 min, respectively, separated on urea-polyacrylamide gels and negative DAPI stained. (i) ATP (1 μM) was incubated for 30 min with increasing concentrations of PPX (0–10 μg ml⁻¹) or shrimp alkaline phosphatase (PSP, 0–2.3 μg ml⁻¹) and quantified using a luciferase-based bioluminescence assay. Data are mean±s.e.m., from three independent experiments. (j) ADP (5 μM) was treated for 30 min with buffer, apyrase (0.1 U ml⁻¹) or increasing concentrations of PPX (0–10 μg ml⁻¹). Platelet aggregation in human platelet-rich plasma stimulated by the reaction mixtures. Representative curve of n=4.

Figure 4. PPX and PPX_Δ12 interfere with polyP-induced coagulation.

(a,c) PPX and (b,d) PPX_Δ12 inhibit polyP-initiated thrombin formation. Real-time thrombin generation in the absence or presence of increasing concentrations of PPX or PPX_Δ12 in PPP stimulated with long-chain (LC; 1 μg ml⁻¹) polyP or short-chain (SC; 10 μg ml⁻¹) polyP. Representative thrombin generation curve of n=6 is shown. (e) FXIIa formation in human plasma was stimulated with buffer, LC (1 μg ml⁻¹), SC (10 μg ml⁻¹) or LC and SC polyP preincubated with PPX or PPX_Δ12 (100 μg ml⁻¹ each). FXIIa was measured by conversion of the chromogenic substrate D–Pro–Phe–Arg–p nitroanilide (S-2302) at λ=405 nm and t=60 min in the presence of inhibitors specified in the methods. Mean±s.e.m., n=6, ***P<0.001 by one-way analysis of variance (ANOVA). (f–h) Targeting polyP interferes with activated platelet-driven coagulation. (f) Real-time thrombin generation in collagen- (3.3 μg ml⁻¹) stimulated PRP in the absence or presence of PPX or PPX_Δ12 (500 μg ml⁻¹ each). (g) Recalcification clotting times in Trap6- (30 μM) or collagen- (33 μg ml⁻¹) stimulated human PRP dependent on addition of anti-FXIIa antibody (3F7; 375 nM), PPX, PPX_Δ12 or polyP pre-bound-PPX_Δ12 (500 μg ml⁻¹ each). Mean±s.e.m., n=4, ***P<0.001 versus buffer by one-way ANOVA. (h) Recalcification clotting times triggered by Ca²⁺ ionophore (A23187, 5 μM) or collagen (33 μg ml⁻¹) in PRP of WT or F12^−/− mice in the presence (+) or absence (−) of PPX or PPX_Δ12 (500 μg ml⁻¹ each). Clotting time reduction is given relative to untreated plasma. Mean±s.e.m., n=4. Insert: collagen failed to initiate zymogen FXII activation in PPP, while dextran sulfate (DXS; 100 μg ml⁻¹) activated all plasma FXII. (i) Role of polyP in FXI activation in the plasma. F12^−/− mouse plasma was incubated with α-thrombin (5 nM) in the absence or presence of polyP (LC or SC, 10 μg ml⁻¹ each) and FXII (30 μg ml⁻¹). Formed FXIa was measured by conversion of the chromogenic substrate S-2366 at λ=405 nm in the presence of inhibitors specified in the methods. Mean±s.e.m., n=4, ***P<0.001 versus buffer by one-way ANOVA.

Figure 5. PPX and PPX_Δ12 reduce thrombus formation in blood under flow.

Citrated whole blood from human and mice, readjusted to physiologic Ca²⁺ and Mg²⁺ concentrations, was perfused for 4 min over a collagen-coated surface at an arterial (a,c) or venous (b,d) shear rate. Representative phase-contrast images of thrombi formed during perfusion in the absence or presence of indicated PPX (a,b) or PPX_Δ12 (c,d) concentrations. Scale bars, 20 μm. Columns give the percentage of surface area covered by thrombi. Mean±s.e.m., from four independent experiments, ***P<0.001 by one-way analysis of variance.

Figure 6. PPX and PPX_Δ12 do not alter fibrin composition.

Scanning electron microscopy images of thrombi formed in Fig. 5 under flow. Scanning electron micrographs of thrombi formed in full blood under arterial (a–d) or venous (e–j) shear rates in the presence of buffer (w/o), PPX or PPX_Δ12 (1 mg ml⁻¹ each). White squares denote the area that is enlarged in b, f, h and j). Scale bar, 25 μm (e,g,i); 5 μm (a,c,d,f,h,j); 2 μm (b). (k) Fibre thickness measured from scanning electron micrographs of 25 fibres in three representative areas. Mean±s.e.m., NS=non-significant by one-way analysis of variance. (l) Human platelets were incubated for 20 min with buffer or collagen (10 μg ml⁻¹). Released polyP was measured by increase of phosphate in PPX- (50 μg ml⁻¹) versus buffer-treated platelet supernatants. Phosphate was calculated from malachite green absorbance at 650 nm from a standard curve. Mean±s.e.m., n=10, ***P<0.001 by Student's t-test.

Figure 7. PPX and PPX_Δ12 interfere with arterial and venous thrombosis in mice.

(a) Thrombosis was induced in the left carotid artery by topical application of 5% FeCl₃ applied for 3 min in F12^−/− or WT mice that were previously injected with buffer, PPX_Δ12 or PPX (300 mg kg⁻¹ BW each). Artery patency was monitored by a flow probe until complete occlusion occurred and zero flow was recorded for >10 min. Representative curves are shown from five independent experiments. (b) Pulmonary embolism induced by intravenous infusion of collagen–epinephrine. The survival time of WT or F12^−/− mice pretreated with buffer, PPX or PPX_Δ12 (150 mg kg⁻¹ BW each) was monitored. Mortality was assessed in each group of mice, and animals alive 30 min after challenge were considered survivors; **P<0.01, *P<0.05 versus buffer-treated WT by one-way analysis of variance, NS=non-significant. (c) Collagen–epinephrine-challenged mice were intravenously infused with Evans blue shortly after the onset of respiratory arrest, while the heart was still beating or after 30 min for those animals that survived. Lungs were excised and perfusion defects were analysed. Occluded parts of the lungs remain their natural pinkish colour. Scale bar, 5 mm.

Figure 8. PPX and PPX_Δ12 do not interfere with haemostasis.

Wild-type mice were intravenously injected with buffer, PPX, PPX_Δ12 (150 mg kg⁻¹ BW each) or heparin (200 U kg⁻¹ BW). Bleeding times and blood loss from clipped tail injury assessed the haemostatic capacity of treated and F12^−/− mice. (a) Bleeding time and (b) total haemoglobin loss determined by absorbance of haemoglobin in 37 °C PBS at λ=575 nm. (c) Tail-bleeding times were analysed by gently absorbing blood with a filter paper. Each symbol represents one animal; bars within each column indicate the mean, ***P<0.001, *P<0.05, NS=non-significant by one-way analysis of variance.