Felix A. Ruiz, Unidad de Investigación, 9ª Planta, Hospital Universitario Puerta del Mar, Avenida Ana de Viya 21, 11009 Cádiz, Spain. Tel.: +34 956003156; fax: +34 956002347. E-mail: email@example.com
Summary. Background: Polyphosphate, a phosphate polymer released by activated platelets, has recently been described as a potent modulator of blood coagulation and fibrinolysis. In blood plasma, polyphosphate binds to and alters the biological functions of factor XII, fibrin(ogen), thrombin and factor VII activating protease.
Objectives: The aim of the present study is to investigate whether polyphosphate also binds to von Willebrand factor (VWF) and alters some of its activities.
Methods/Results: When studying patients with type 1 von Willebrand disease (VWD) and their healthy relatives, we discovered a significant correlation between von Willebrand factor (VWF) and platelet polyphosphate levels. We have also found polyphosphate in preparations of VWF isolated from normal platelets and plasma. Surface plasmon resonance and electrophoretic mobility assays indicated that polyphosphate interacts with VWF in a dose- and time-dependent manner. Treatment of normal plasma with active exopolyphosphatase decreased the VWF ristocetin cofactor (VWF:RCo) activity, a functional measure of VWF binding to platelet glycoprotein receptor Ib. VWF collagen binding and multimerization were unaltered after polyphosphate depletion. Moreover, addition of polyphosphate increased the deficient VWF:RCo activity presented by plasma from patients with type 1 VWD.
Conclusions: Our results reveal that a new role is played by polyphosphate in hemostasis by its interaction with VWF, and suggest that this polymer may be effective in the treatment of some types of VWD.
The von Willebrand factor (VWF) is a multimeric protein that plays key roles in the hemostatic process through its interactions with platelet glycoprotein receptor Ib (GP1b), factor (F) VIII and subendothelial collagen (recently reviewed in [1,2]). VWF is present in plasma, in platelet alpha-granules and in Weibel-Palade bodies of endothelial cells . Deficits, in quantity or function, of VWF are responsible for von Willebrand disease (VWD), which is the most common inherited bleeding disorder in humans. . Among the different varieties of VWD, type 1 is the most frequent form and is characterized by reduced levels of VWF with nearly normal protein function and structure .
Since our discovery that inorganic polyphosphate (polyP) is abundant in platelet dense granules and is released by activated platelets , there has been increasing interest in this polyanion in hematology. PolyP is a linear polymer, formed by phosphate (Pi) residues linked by ATP-like bonds, which modulates blood coagulation in vitro  and in vivo , enhances fibrinolysis  and increases vascular permeability , among other effects. Most of the hemostatic effects of polyP may occur through the union of this polymer with various different plasma proteins. It has recently been reported that PolyP binds to, and alters the biological functions of, FXII , fibrin(ogen) [9,10], thrombin  and FVII activating protease .
In a recent paper we reported that patients with congenital defective platelet dense granules have depleted levels of platelet polyP, in comparison with healthy individuals . Here, we have continued to study platelet polyP in other genetic bleeding disorders, by testing type 1 VWD patients and their healthy relatives. The results obtained have led us to hypothesize that VWF could also be a polyP-regulated protein. In this article, we describe how polyP is involved in the activity of union between VWF and GP1b.
Materials and methods
Blood samples were provided by individuals with a positive family history for type 1 VWD, obtained by the Hematology Service of the ‘Puerta del Mar’ Hospital (Cadiz, Spain) for screen diagnosis. Plasma and platelets were separated from blood samples by centrifugation. Approval for this study was obtained from the institutional review board (Ethics Committee). Informed consent was provided according to the Declaration of Helsinki. At the time of donating the blood, individual donors confirmed that they had not taken any platelet-affecting medication during the preceding week.
Type 1 VWD diagnosis
Plasma von Willebrand factor antigen (VWF:Ag) and plasma von Willebrand factor activity (VWF:Act) were measured routinely in the Hematology Service of the ‘Puerta del Mar’ Hospital. VWF:Ag was determined using an immuno-turbidimetric method (STA Liatest vWF, Diagnostica Stago, Asnieres, France) and a Triturus analyser with dedicated software (Grifols Diagnostics, Barcelona, Spain). VWF:Act was determined by an ELISA assay that detects levels of functional VWF epitope (DG-EIA vWF activity, Grifols Diagnostics). Ratios of VWF:Ag/VWF:Act were higher than 0.7 for all individuals included in this study. Normal range was established as individuals with values of VWF:Ag between 60–150% and VWF:Act higher than 60%. Lower values of VWF:Ag and VWF:Act define the patients with type 1 VWD.
Measurement of polyP levels
Commercial polyP65 (Sigma Chemicals Co., St Louis, MO, USA) was previously isolated by gel filtration with Sephadex G25 to remove orthophosphate, pyrophosphate and lower-chain polyP . Exact polyP contents in platelet acidic extracts, VWF samples and commercial polyP65 preparations were determined from the amount of Pi released upon treatment with an excess of purified recombinant exopolyphosphatase from Saccharomyces cerevisiae (PPX), as described before . PPX is a highly specific and active enzyme against polyP . For all experiments described here, polyP concentrations are expressed in terms of their phosphate residues (Pi). Denaturized exopolyphosphatase (dPPX) was prepared by an incubation of native PPX for 60 min at 95 °C, and depletion of their enzymatic activity was confirmed by a standard assay, as described above. Purified VWF without polyP (VWF ‘apo’) was obtained by treatment of isolated plasma VWF with PPX, as described above, and the subsequent separation of PPX using Amicon Ultra-0.5 100 k filter devices (Millipore, Billerica, MA, USA) (Fig. S2). No detectable PPX activity was seen in VWF ‘apo’ preparations.
Detection of polyP in immuno-precipitated VWF from human platelets
Immuno-precipitation of VWF was performed using platelet-rich plasma units from the Hematology Service at the ‘Puerta del Mar’ Hospital. Five milliliters of platelet-rich plasma was centrifuged at 200 g for 10 min, to eliminate contaminating red blood cells and leukocytes, followed by centrifugation at 1000 g for 15 min. The resulting pellet was resuspended with 110 μL of digitonin lysis buffer (25 mmol L−1 Hepes pH 6.5, 0.25 mol L−1 sucrose, 12 mmol L−1 sodium citrate, 1 mmol L−1 ethylene diamine tetraacetic acid (EDTA), 1% digitonin, and a 1/500 dilution of Sigma protease inhibitor cocktail for mammalian cell extracts) and left to stand for 30 min on ice. Lysates were centrifuged at 13 000 g for 2 min to remove cell debris, and the supernatants were diluted 10X in NET buffer (50 mmol L−1 Tris pH 7.5, 5 mmol L−1 EDTA, 150 m mol L−1 NaCl and 0.05% Nonidet P40). VWF was immuno-precipitated with protein A coupled to magnetic beads (Dynabeads, Invitrogen, Oslo, Norway), according to the manufacturer’s directions. Samples were incubated for 1 h with rabbit IgG anti-human VWF (100 μg mL−1) or a similar volume of purified rabbit IgG (Sigma). Following the immuno-precipitation, the supernatants from non-denaturing elutions were resuspended in SDS sample buffer (for 8% SDS–polyacrylamide gel electrophoresis, PAGE), or with 0.5 mol L−1 perchloric acid (for polyP measurement with PPX). Identification of proteins in bands sliced from SDS-PAGE gels was performed by ion trap mass spectrometry at the Proteomics Laboratory (CIMA/Universidad de Navarra, Navarra, Spain) as described in the notes to Supplementary Table 1.
PolyP localization using a polyP-binding protein and confocal microscopy
PolyP localization on permeabilized platelets was performed using the recombinant PPBD (polyP-binding domain) of E. coli PPX linked to an Xpress epitope tag , with some modifications. Briefly, fixed platelets, purified as described by Sehgal & Storrie , were bound to poly-lysine coated coverslips, fixed in 2% paraformaldehyde, permeabilized with 0.2% Triton X-100 for 5 min, and treated with 100 mmol L−1 ammonium chloride for 5 min. Then, coverslips were blocked with 3% BSA for 30 min. Staining was carried out by incubations for 1 h with: (a) PPBD (8 μg mL−1), anti-Xpress epitope monoclonal antibody (10 μg mL−1) and rabbit IgG anti-human VWF (1/400); and (b) 1/100 diluted Cy3-labeled anti-mouse and 1/400 Alexa488-labeled anti-rabbit IgGs as secondary antibodies. All dilutions were carried out with Tris-buffered saline (100 mmol L−1 Tris/HCl (pH 7.2) and 150 mmol L−1 NaCl), and we added 1% BSA to the media in both incubations with antibodies. Negative controls were prepared by a similar procedure but without PPBD or primary-specific Abs. Coverslips, in DABCO antifading agent, were mounted on slides for imaging using a laser TCS-SL confocal imaging system (Leica Microsystems, Mannheim, Germany).
Isolation of platelet alpha-granules
Platelet alpha-granules were isolated from platelet-rich plasma units, using a sucrose gradient centrifugation method previously described  with modifications. Briefly, washed platelets were resuspended (2 × 109 platelet mL−1) in lysis buffer (25 mmol L−1 Hepes pH 6.5, 0.25 mol L−1 sucrose, 12 mmol L−1 sodium citrate, 1 mmol L−1 EDTA, and a 1/500 dilution of Sigma protease inhibitor cocktail for mammalian cell extracts). Platelets were sonicated five times on ice (5 s on and 15 s off at 15% intensity in a Branson Sonifier, model 150; Branson Ultrasonics, Danbury, CT, USA), and unbroken cells were separated by centrifugation. Supernatants were centrifuged for 20 min at 19 000 g at 4 °C. The pellet was resuspended in 1 mL of lysis buffer. Suspension was applied at the top of a continuous gradient of 27 mL of 30–60% sucrose (w/v) in lysis buffer. The gradient was centrifuged for 60 min at 100 000 g. Fractions of 1.5 mL were obtained from the top of the gradient, and VWF, polyP content and total protein of gradient fractions were measured, as previously described .
Detection of polyP bound to VWF by native agarose gel electrophoresis
Aliquots of 10–15 μL (around 8 μg) of plasma-purified VWF (FVIII-free) (Haematologic Technologies Inc., Essex Junction, VT, USA) were incubated in 30 mmol L−1 MgCl2 with 1.6 μg of active PPX, or a similar volume of water, for 150 s at 37 °C. A similar volume of native-sample buffer was added (50% glycerol, 1% bromophenol blue), and samples were separated in a 0.7% agarose gel prepared in 40 mmol L−1 Tris-acetate, pH 7.8 and 1 mmol L−1 EDTA. Electrophoresis was run at 30 mA for 30 min and then at 50 mA for 5 h. Gels were stained with a polyP-staining solution (0.05% Toluidine blue, 25% methanol and 5% glycerol) and de-stained with a similar solution without Toluidine blue. The gels were then stained with a protein-staining solution (0.25% Coomassie blue, 45% methanol and 10% acetic acid), and de-stained with a similar solution without Coomassie blue.
Surface plasmon resonance
The binding kinetics was measured with a XPR36 ProteOn system (BioRad, Hercules, CA, USA). Isolated VWF (FVIII-free, 0.125 mg mL−1) was immobilized on a GLM Sensor chip (BioRad), according to the supplier’s protocol, using an amine coupling kit (BioRad), in ProteOn-acetate buffer pH 5.0. Phosphate-buffered saline pH 7.4 containing 0.005% Tween 20 (PBS-T) was used as running buffer. Interaction analysis experiments were performed at a flow rate of 30 μL min−1 and 1 min contact time at 25 °C. Different concentrations of isolated polyphosphate of 65 residues (polyP65, 0.084-6.67 μmol L−1) were injected. Immobilization of isolated VWF on sensor chips, and association and dissociation with interacting polyphosphate, were followed in real time by monitoring the change in plasmon resonance signal, expressed in resonance units (RU). Sensorgrams were analyzed using the ProteOn Manager software. Kinetic constants were obtained using the Langmuir analysis model.
Separation of polyP by urea-polyacrylamide gel electrophoresis
PolyP size was studied in a 6% urea-polyacrylamide gel electrophoresis, as described . Using this method we analyzed: (i) total polyP isolated from aliquots of purified VWF; and (ii) commercial polyP65 after incubation with purified VWF. Total polyP isolation was performed as described , and gels were stained with DAPI-negative staining . In the case of commercial polyP, aliquots of 0.46 μg (or indicated amounts) of purified VWF were incubated with 0.8 mmol L−1 of isolated polyP65 for 30 min at 37 °C. Then, gels were stained with a polyP-staining solution (0.05% Toluidine blue, 25% methanol and 5% glycerol) and de-stained with a similar solution without Toluidine blue.
Other VWF determinations
VWF ristocetin cofactor (VWF:RCo) activity was determined with the Ristocetin Cofactor Assay Kit (Helena Laboratories, Beaumont, TX, USA) and an optical platelet aggregometer (Chronolog Lumi Aggro-meter model 400; Chronolog Corporation, Havertown, PA, USA). Collagen binding to VWF was determined using a commercially-available ELISA kit (Life Diagnostics, Clarkston, GA, USA). The VWF multimer assay was performed in SDS-agarose gel electrophoresis and immunoblotting, as previously described , using a rabbit anti-human VWF as a primary antibody (DakoCytomation, Glostrup, Denmark). Before all these measurements, samples were incubated in the absence or presence of active PPX or polyP65, as described.
Platelet polyP in type 1 VWD
We have previously shown that platelet polyP is depleted in patients with inherited defective platelet dense granules . Interested in examining platelet polyP levels in other genetic bleeding disorders, we have now tested samples from 50 untreated individuals with a positive family history for type 1 VWD (Fig. 1). According to the values of VWF:Ag and VWF:Act obtained, the individuals were diagnosed and classified as either normal or type 1 VWD (Fig. 1A). Values of normal platelet polyP (Fig. 1A) were similar to those reported in a previous study , which correspond to polyP concentrations of around 1.2 mmol L−1 (assuming platelet mean volumes of 7-11 fl ). We observed that the platelet polyP level was around 40% lower in individuals with type 1 VWD (Fig. 1A). In addition, we found a statistically-significant correlation between VWF:Ag and platelet polyP levels in all the individuals studied (Fig. 1B). These results suggest to us that there could be a relationship between levels of platelet polyP and VWF.
PolyP is present in platelet VWF
Isolated platelets from healthy human subjects were lysed and VWF was immuno-precipitated by using a specific anti-VWF antibody and protein A coupled to magnetic beads (Fig 2A). The immuno-precipitate was subjected to SDS/PAGE and stained for protein detection (Fig 2A). Two main bands were seen (Fig. 2A), the first one at 220 kDa (which is the molecular mass of VWF ) and another one at around 35 kDa. Proteins contained in these two gel bands were identified by mass spectrometry: the 220 kDa band contained VWF factor and platelet myosin; and the 35 kDa band contained cytoplasmic actin isoforms (Table S1). Presence of VWF in the band at 220 kDa was further confirmed by Western blotting (result not shown). We measured polyP levels in these immuno-precipitates obtained from platelet lysates (Fig. 2B). Immuno-precipitates obtained with the anti-VWF antibody have four times more polyP than control samples, in which purified standard rabbit IgG was used (Fig. 2B), suggesting that VWF from human platelets co-precipitates with polyP.
PolyP in platelet alpha-granules
We have further studied the localization of polyP in platelets. We have previously reported that platelet polyP is localized preferentially in platelet dense granules . We have found that polyP present in dense granules is released from fixed platelets after permeabilization with the nonionic detergent Triton-X 100 (data not shown). In these permeabilized platelets, we have found that the remaining polyP is largely co-localized with VWF (Fig. 2C). As platelet alpha granules are the principal reservoir of VWF , we measured whether polyP is present in these organelles. Using a well-known method of alpha granules isolation , we obtained in these organelles a fraction that is enriched by a factor of 20 (Fig. S1). This fraction had low, but detectable, levels of polyP (of 0.37 ± 0.02 pmol of polyP per μg protein).
PolyP is present in isolated VWF from human plasma
We incubated plasma-isolated VWF with an excess of recombinant yeast exopolyphosphatase (PPX), which specifically releases phosphate from polyP (Fig. 3A). We found levels of around 12 pmol of polyP (expressed in phosphate residues) per each μg of purified VWF analyzed (Fig. 3A). We further examined VWF-PolyP interaction using low-resolution agarose gel electrophoresis (without sodium dodecyl sulfate), which permits the separation of native VWF in a discrete band (Fig. 3B). Isolated VWF that had been pre-incubated with PPX presented a decreased labeling for polyP, but there was no change in its labeling for total protein (Fig. 3B). Additionally, we performed an extraction of total polyP from plasma-isolated VWF (Fig. 3C). Analysis of the extracted polyP with urea-polyacrylamide gel electrophoresis (PAGE) and specific staining for polyP showed an intense band which appeared to have an electrophoretic mobility similar to that of polyP released from platelets to the plasma [6,8], with an estimated length of 70–80 phosphate residues (Fig. 3C). We also studied the interaction between polyP and VWF in real time, using surface plasmon resonance (Fig. 4). Isolated VWF was immobilized on a chip by amine coupling chemistry and exposed to increasing concentrations of polyP (Fig. 4). With the data obtained, a binding constant of 2.7 μmol L−1 for polyP65 (in terms of its phosphate residues) was calculated. Direct examination of the affinity of polyP for proteins using plasmon resonance has been reported before only for thrombin . The characteristics of this interaction were further studied using urea- PAGE (Fig. 5). Isolated plasma VWF was incubated with polyP65 in different conditions, and the resultant electrophoretic mobility of polyP in the mixtures was analyzed (Fig. 5). Addition of VWF modified the electrophoretic mobility of that polyP, which migrated faster than free polyP, and this property permitted us to observe that the polyP-VWF interaction is dose (Fig. 5A), time (Fig. 5B) and temperature (Fig. 5B) dependent. Moreover, we determined that the interaction of polyP and VWF depends on divalent cations (Fig. 5C), and concentrations of up to 10 mmol L−1 of manganese acted to improve the binding (Fig. 5C). Similar results were obtained when non-denaturing gels were used (results not shown).
Incubation of normal human plasma with the enzyme exopolyphosphatase (PPX), which specifically degrades polyP , produced a decrease in the von Willebrand factor ristocetin cofactor (VWF:RCo) activity (Fig. 6). The reduction of VWF:RCo activity does not occur if a denatured enzyme is used (Fig. 6A), and is dependent on the doses of enzyme included in the assay (Fig. 6B,C). In addition, repurified VWF, in which polyP had previously been removed (VWF ‘apo’), also presents a decreased VWF:RCo activity (Fig. S2), thus discounting the possibility that the decrease occurred as a non-specific effect of PPX in the assay. Other in vitro activities of VWF, such as multimerization assembly (Fig. S3) and collagen binding (Fig. S4), were unaltered by incubation with an excess of active PPX.
PolyP increases VWF ristocetin cofactor activity on type 1 VWD
We tested the effect on the VWF:RCo activity of adding increasing concentrations of polyP65 to normal human plasma (Fig. 7A, C). In most of the normal individuals, addition of polyP did not produce a significant change, and in others, an inhibition of less than 20% of the activity was observed (Fig. 7A, C). In contrast, when plasmas of type 1 VWD patients were used, an increase, in excess of 20% in most of the cases, was recorded in the VWF:RCo activity (Fig. 7B, C). Statistical analysis showed significant differences in the VWF:RCo activity of normal and type 1 VWD plasmas, after the addition of 2.5 μmol L−1 of polyP65 (Fig. 7C).
The findings presented in this study provide strong evidence for a newly-identified role of polyP as a new binding partner of human VWF. VWF can therefore be added to the recent list of proteins, with key roles in hemostasis, that are modulated by PolyP, including FXII , fibrin(ogen) [9,10], thrombin  and FVII-activating protease , among others.
We found that VWF, isolated from platelet and plasma, included detectable amounts of polyP (Figs 2 and 3). As total polyP levels in normal platelets are around 1.2 mmol L−1, and this level is only 40% lower in individuals with type 1 VWD (Fig. 1), polyP associated with platelet VWF should be in the micromolar range. In addition, whole blood polyP may also reach micromolar levels [6,7]. All of these are consistent with the functional assays and with the binding affinity that we measured, and suggest that polyP may act as a physiological modulator of VWF. As the polyP bound to VWF has around 70–80 phosphate residues (Fig. 3C), we estimate that there is just one molecule of polyP for every 20 VWF monomers of plasma. Further, the isolated plasma VWF could bind more polyP, as we demonstrated with the plasmon resonance and gel mobility experiments (Figs 4 and 5A). As polyP promotes the binding of VWF to platelets, measured in cases of type 1 VWD (Fig. 7), this makes it an attractive focus for prospective studies for new VWD therapies.
For polyP, we observed an electrophoretic acceleration of polyP that was dose, time and temperature dependent after incubation with VWF (Fig. 5). Similar approaches of gel mobility ‘acceleration’ have been observed in the study of other protein-polyanion interactions, such as histone-DNA  and protein-RNA . This acceleration in electrophoretic mobilities is usually attributed to conformational changes in the charged polymer . PolyP is polymorphic and its conformation depends on its coordination with other surrounding charged moieties ; thus our experiments suggest that the interaction of VWF with polyP is ionic in nature.
VWF is a large protein and many of its binding functions have been localized to discrete domains in its monomeric structure . To resolve which particular domain(s) of VWF are specific for binding polyP, further functional mutagenesis studies are needed. However, we hypothesize that the polyP-binding site of VWF should overlap with its heparin-binding site, as has recently been described for thrombin . In both proteins, the interaction with heparin is ionic and involves basic residues of arginine and lysine [11,28]. The heparin-binding site for VWF is localized at domain A1, which also mediates the binding with platelet GPIb, and with the antibiotic ristocetin (which in vitro induces the binding of VWF to GPIb ). In a manner analogous to that which occurs with thrombin, we found that polyP produces effects contrary to those produced by heparin in the biological activity of VWF. Thus, whereas polyP promotes ristocetin-induced platelet agglutination (or binding of VWF to platelets) (Figs 6 and 7), heparin has been reported as inhibitory for this activity . We believe that the data of Figs 6 and 7 are complementary. PolyP depletion in normal plasma decreases VWF:RCo activity (Fig. 6), but polyP addition cannot increase this VWF:RCo activity in normal plasmas because they have already reached the maximum activity (Fig. 7A). Consequently, polyP addition can increase VWF:RCo activity in type 1 VWD plasmas because they have low basal activity (Fig. 7B).
Patients with some variants of type 2 VWD (such as 2B or 2M) have mutant VWF with altered affinity for platelet GPIb . As polyP promotes specifically the binding of VWF to platelet GPIb (not multimerization or collagen binding), the study of polyP on these VWF mutants could provide excellent models for understanding the interaction between these molecules. In addition, delta storage pool disease patients, who have depleted levels of platelet polyP , could be an attractive model to study how VWF activities take place in the absence of polyP.
We have previously reported that platelet polyP is localized preferentially in platelet dense granules . In this present study, we find that platelet alpha granules also have polyP (Figs 2C and S1), but levels of this polymer are 500 times lower than those existing in dense granules. It is well known that the ionic nature of platelet dense granule content differs with the main constituents of alpha granules, which are mostly proteins . In addition, we have found that the polyP present in alpha granules is resistant to platelet permeabilization (Fig. 2C), in contrast to the polyP in dense granules, which is released in the permeabilization process. In accordance with these data, we hypothesize that the pool of polyP in platelet alpha granules could be strongly associated with proteins (mostly with VWF), and the polyP in dense granules could be associated with ions (mostly calcium), and so could be more available for release into the plasma, whereas polyP on alpha granules could modulate the functions of proteins therein.
We thank the late Professor Arthur Kornberg for providing us with the E. coli CA38 pTrcPPX1, and all the personnel of the Hematology Service of the ‘Puerta del Mar’ Hospital for their assistance. This work was supported in part by the EU (FEDER) and the Spanish Ministry of Science and Innovation (Grant FIS/PI10/01222 to FAR.), the Junta de Andalucia (Grants P07-CTS-02765 and C.S.0257/09 to FAR.), and Plan Andaluz de Investigacion (Cod. CTS-554). We also acknowledge the help of F. Corrales from the Proteomics Laboratory, CIMA/ Universidad de Navarra (part of the network of the Spanish National Institute of Proteomics Facilities, ProteoRed).
Disclosure of conflicts of Interests
The authors state that they have no conflict of interest.