Ravindra Sarode, Associate Director Blood Bank/Medical Director Coagulation Laboratory, University Hospitals of Cleveland, 11100 Euclid Ave, Cleveland, OH 44106, USA. E-mail: email@example.com
ABO (H) blood group antigens are covalently linked to the oligosaccharide side-chains of von Willebrand factor (VWF). In this study, we investigated the role of the A and B antigens in the expression of VWF adhesive activity. VWF of type A, B or O was purified from fresh frozen plasma. Presence of A or B antigen on the VWF was confirmed by enzyme-linked immunosorbent assay (ELISA) and by immunoblotting with monoclonal anti-A or anti-B. The A or B antigen was also detected in the 48/52-kDa fragment of the respective VWF after trypsin digestion. Removal of A antigen with α-N-acetylgalactosaminidase or B antigen with α-galactosidase did not affect its multimer size or antigenic level, but decreased the ristocetin cofactor (RCoF) activity of the respective VWF by 33–39% (P < 0·01–0·002). Removal of A or B antigen from VWF did not affect the binding of the VWF to immobilized type III collagen. A and B antigens were not detected in platelet VWF. These results indicate that AB structures play a role in platelet aggregating activity of VWF.
von Willebrand factor (VWF), a multimeric glycoprotein (GP) synthesized in endothelial cells and megakaryocytes, is critical for normal haemostasis as it promotes platelet adhesion and aggregation at sites of vessel injury. A deficiency in VWF leads to a bleeding diathesis, which can be fatal in severe cases. Studies in animal models of von Willebrand disease have suggested that VWF is also involved in the development of arterial thrombosis and atherosclerosis (Blann & McCollum, 1994). Some studies have shown that a high plasma level of VWF in patients with coronary artery disease is an independent risk factor for myocardial infarction and death (Jansson et al, 1991; Thompson et al, 1995). VWF may also promote cancer metastasis (Nierodzik et al, 1995) and vaso-occlusion in sickle cell disease (Kaul et al, 1993).
VWF, secreted from endothelial cells as a disulphide-linked polymer of a 2050-residue polypeptide, is cleaved in the circulation by a plasma metalloproteinase to become the series of multimers found in normal plasma (Tsai, 1996). The loop structure formed by the disulphide bond between Cys509 and Cys695 in the A1 domain is critical in modulating VWF binding to platelet GP Ib/IX (Meyer & Girma, 1993). The A3 domain, which contains a disulphide-linked loop between Cys923 and Cys1109 as well as the A1 domain, is involved in VWF binding to collagen (Meyer & Girma, 1993).
Plasma VWF contains 12 N-linked and 10 O-linked oligosaccharide chains, which account for approximately 15% of the total weight (Titani et al, 1986). Recombinant VWF deficient in the O-glycans binds normally to collagen but less effectively to platelets in the presence of ristocetin (Carew et al, 1992). Although removal of sialic acid does not affect the ristocetin-induced binding of VWF to platelets, it increases the susceptibility of VWF to N-terminal proteolytic cleavage (Berkowitz & Federici, 1988). Asialo-VWF is also rapidly cleared by the liver (Sodetz et al, 1977). Platelet VWF contains 50% less carbohydrates and has a lower ristocetin cofactor activity than plasma VWF (Williams et al, 1994).
In humans and anthropoid apes, A and B antigens are found on the red cells as well as on epithelial and endothelial cells, whereas in lower mammals they are expressed only in the epithelial cells of the gastrointestinal (GI) tract (Kominato et al, 1992; Yamamoto et al, 1992). ABO (H) antigens are also present in the N-linked oligosaccharide chains of human plasma VWF and α2-macroglobulin (Matsui et al, 1993). Although the ABO (H) gene is highly conserved in the genomic DNA of various mammals (Kominato et al, 1992), the ABO (H) structures have not been associated with any biological functions.
Recombinant green coffee bean α-galactosidase and α-N-acetylgalactosaminidase purified from chicken livers have been used to produce type O cells from type B and type A cells by removing A and B antigens from the respective red blood cells (Zhu & Goldstein, 1995; Zhu et al, 1996). In this study, we investigated the role of AB antigens in the expression of VWF adhesive activity by enzymatically removing A or B antigen from blood type specific VWF.
Materials and methods
Reagents Monoclonal anti-A and anti-B antibodies were obtained from Ortho Diagnostic System (Raritan, NJ, USA), lyophilized platelets were from Biodata (Hartboro, PA, USA) and the DAB-metal enhancer kit was from Pierce (Rockford, IL, USA). Rabbit anti-human VWF polyclonal antibody (Dako, Carpinteria, CA, USA) was labelled with 125I using Iodogen beads (Tsai, 1996). Recombinant green coffee bean α-galactosidase (Bzyme) and chicken liver α-N-acetylgalactosaminidase (Azyme) were prepared as previously described (Zhu & Goldstein, 1995; Zhu et al, 1996). Electrophoresis reagents were obtained from Bio-Rad (Richmond, CA, USA). All other reagents were from Sigma Chemicals Co. (St. Louis, MO, USA).
Preparation of VWF from plasma Type A-, B- or O-specific VWF was purified from respective fresh frozen plasma by glycine salt precipitation and gel filtration (Tsai & Lian, 1998). Multimeric analysis of VWF was performed by SDS agarose gel electrophoresis as previously described (Tsai & Lian, 1998).
Preparation of platelet lysates Whole blood (25 ml) from normal volunteers of A, B or O blood type was collected in 5 mm EDTA. Platelet-rich plasma (PRP) was prepared by centrifugation at 2500 r.p.m. for 3 min. Platelet pellets were obtained from PRP by centrifugation at 2500 r.p.m. for 15 min and were washed three times with Tris-buffered saline (TBS) containing 1% EDTA. Platelets (3·1–4·2 × 107/µl) were then lysed with 1% Triton X in TBS in the presence of 5 mm EDTA, 1% phenylmethylsulphonyl fluoride (PMSF) and 100 U/ml aprotinin by incubating at 37°C for 30 min with frequent vortexing. After centrifugation at 12 000 g for 3 min to remove insoluble elements, the supernatants were frozen at −70°C until further analysis.
48/52-kDa fragment of VWF by trypsin digestion The 48/52-kDa fragment of plasma VWF generated by trypsin contains the GPIb binding domain of VWF (Fujimura et al, 1987). This fragment was prepared by incubating 2500 U of trypsin per mg of VWF at 37°C for 3 h (Fujimura et al, 1987) in a buffer containing 0·01 m Tris, 0·15 m NaCl and 0·02% NaN3, pH 7·4. The fragment was isolated with a heparin–sepharose CL-6B column and eluted with 0·01 m Tris/0·5 m NaCl buffer, pH 7·4 (Andrew et al, 1989). The fractions containing the 52/48-kDa fragment were identified by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting with antibodies against VWF (Tsai, 1996).
Detection of A or B antigen on VWF To detect A or B antigen on VWF molecules, VWF was separated by 6% SDS-PAGE under reducing conditions. VWF bands were located by immunoblotting with antibodies against VWF as previously described (Tsai & Lian, 1998). Duplicate lanes were probed with IgM monoclonal antibodies against A or B antigen, followed by 125I-labelled goat anti-mouse IgM and autoradiography.
The difference in AB antigenic structure between the VWF purified from plasma and the VWF in platelet extracts was also investigated by an enzyme-linked immunosorbent assay (ELISA) method. VWF samples were serially diluted and incubated in duplicate wells with polyclonal anti-VWF immobilized on the surface of microtitre plates. In one set of the wells, the amount of bound VWF was developed with horseradish peroxidase-labelled anti-VWF. In the other set, bound VWF was developed with monoclonal anti-A or -B antigen and horseradish peroxidase-labelled anti-mouse IgM. For each sample, the optical densities developed with anti-A or -B were then plotted against the optical densities of the same sample at the same dilutions but developed with anti-VWF.
Removal of A or B blood group antigens from VWF To remove A or B antigen, purified VWF was incubated at 37°C for 3 h in PBS, pH 6·0, with Azyme or Bzyme at a ratio of 2–6 U of enzymatic activity per 15–25 µg/ml of purified VWF containing only large multimers. Control VWF was incubated with PBS alone. At the end of incubation, the treated VWF was immediately diluted with TBS to final concentrations of 0·25–0·75 U/ml of ristocetin cofactor activity and analysed for RCoF and GPIb and collagen binding or diluted in appropriate buffers for gel analysis.
Ristocetin cofactor activity of VWF The platelet-aggregating activity of VWF was studied by ristocetin cofactor (RCoF) assay as previously described (Tsai, 1996). The RCoF in normal control plasma was defined as 1 U/ml.
VWF binding to platelets The ristocetin-dependent binding of VWF to platelets (GPIb) was studied by incubating formalin-fixed platelets (200 × 109/l) in TBS with enzyme-treated VWF (1 U/ml) and varying concentrations of ristocetin (0–1·6 mg/ml). The same VWF treated with buffer served as control. After incubation at 37°C for 20 min with frequent vortexing, the tubes were centrifuged to remove platelets and bound VWF. The amount of unbound VWF, represented by the concentration of VWF in the supernatant, was determined by ELISA. Normal pooled plasma with VWF at 1 U/ml was used as a reference standard.
VWF binding to collagen A polyvinyl microtitre plate was coated with 30 µg/ml of type III placental collagen in carbonate buffer at pH 9·6 at room temperature for 18 h. After the wells were incubated with enzyme-treated VWF at designated concentrations, the collagen-bound VWF was visualized with horseradish peroxidase-conjugated anti-VWF. The same VWF treated with buffer served as a control. Normal pooled plasma (VWF antigen = 1 U/ml) and its dilutions served as the reference standard.
Detection of A or B blood group antigen on purified plasma VWF
The VWF isolated from type A plasma, when separated by SDS under reducing conditions and probed with anti-VWF, appeared as a single polypeptide of 225 kDa (Fig 1A, lane 1). Treatment of the VWF with Azyme did not affect the size of the VWF polypeptide (Fig 1A, lane 2). When a duplicate of these two lanes was probed with the monoclonal anti-A antibody, only the untreated VWF was visualized (Fig 1B, lane 1) and the Azyme-treated type A VWF lost its reactivity with the monoclonal anti-A antibody (Fig 1B, lane 2). Similar results were obtained with type B VWF treated with Bzyme (Fig 1C and D).
A or B antigen on the 48/52-kDa fragment of VWF after trypsin digestion
The 48/52-kDa fragments of VWF after trypsin digestion have been shown to contain the A1 loop that is critical in regulating the interaction of VWF with platelet GPIb. To determine whether these fragments contain AB antigenic structure, the fragments were isolated from trypsin-digested plasma VWF with a heparin–agarose column, subjected to 12·5% SDS-PAGE and immunoblotted with rabbit anti-human VWF and 125I-labelled goat anti-rabbit IgG. A duplicate of these two lanes was also probed with the monoclonal anti-A or anti-B and 125I-labelled anti-mouse IgM. The results with type A and type B VWF are shown in Fig 2A and B. When detected with polyclonal anti-VWF, tryptic fragments of type A VWF, before and after treatment with Azyme, reacted similarly (Fig 2Aa). However, only the fragments from control VWF reacted with monoclonal anti-A (Fig 2Ab). The species smaller than the 48/52-kDa fragment probably resulted from excessive digestion.
The results of type B VWF are shown in Fig 2B. The 225-kDa VWF polypeptides, before or after treatment with Bzyme (Fig 2Ba) or their tryptic fragments (Fig 2Bb and c), reacted similarly to polyclonal anti-VWF. However, only the 48/52-kDa fragment generated from control type B VWF and eluted from heparin–sepharose columns reacted with monoclonal anti-B, whereas the same fragment from Bzyme-treated type B VWF was not reactive (Fig 2Bd).
Effects of removal of AB antigens on VWF ristocetin cofactor activity
The ristocetin cofactor activity of purified VWF was determined after incubation with buffer or either Azyme or Bzyme. Compared with control incubation with buffer alone, incubation of type A VWF with Azyme decreased the RCoF by 36 ± 13% (n = 3, P < 0·01, Fig 3A). Similarly, incubation with Bzyme decreased the RCoF of type B VWF by 39 ± 3% (n = 3, P < 0·002, Fig 3B). In separate studies, incubation with either enzyme did not result in a change in the RCoF activity of type O VWF (Azyme 10 ± 10%, n = 3, P > 0·5; Bzyme 5 ± 6%, n = 3, P > 0·5; Fig 3C and D). Similarly, there was no effect of Azyme on the RCoF activity of type B VWF or Bzyme on the RCoF of type A VWF (Fig 3E and F).
VWF binding to platelets
To determine whether removal of A or B antigen affected platelet binding, the same VWF sample used in the ristocetin cofactor assays or separate VWF at similar concentrations were incubated with platelets in the presence of various concentrations of ristocetin. The concentration of unbound VWF, represented by the VWF remaining in the supernatant of the reaction mixture, was determined by ELISA and plotted against the ristocetin concentration. As shown in Fig 4A, incubation of type A VWF with Azyme did not affect the ristocetin-induced binding of VWF to platelets. Similarly, the binding of type B VWF to platelets was not affected by incubation with Bzyme (Fig 4B)
VWF binding to collagen
Type A or B VWF, after incubation with the respective enzyme as used in RCoF assay, was serially diluted and incubated with collagen immobilized on the surface of microtitre plate wells. The amount of bound VWF, represented by the optical density generated from the peroxidase substrate, correlated with the concentration of VWF (Fig 5). Compared with the control VWF incubated with buffer, removal of A or B antigen did not affect the binding of the VWF to collagen.
Multimeric structures of VWF
SDS agarose gel electrophoresis showed that incubation of type A VWF with Azyme or type B VWF with Bzyme did not alter the multimeric structure of VWF (Fig 6).
Antigenic reactivity to polyclonal antibodies
Studies by ELISA showed that enzyme treatment did not decrease the antigenic reactivity of VWF to polyclonal antibodies (Fig 7). No difference in VWF concentration was detected by electroimmunoassay of Laurell either (data not shown).
Absence of AB structures on platelet VWF
VWF extracted from platelets of normal donors of type A, B or O, when subjected to SDS PAGE under reducing conditions and immunoblotted with anti-VWF, appears as a 225-kDa band that is not different from plasma VWF. However, it was not visualized when probed with monoclonal anti-A or anti-B (data not shown). To confirm that AB structures were absent, platelet extracts of various concentrations were incubated in duplicate with the polyclonal anti-VWF immobilized on microtitre plate wells. The bound VWF in one set of wells was developed with anti-VWF, whereas that in the duplicate set of wells was developed with anti-A or -B. Figure 8A shows the results of VWF isolated from type A plasma and platelets. With plasma VWF, the optical density developed with anti-A correlated with the optical density of the same VWF developed with anti-VWF. In contrast, the optical density of type A platelet VWF never rose above baseline when it was developed with anti-A. Similar results were obtained with VWF isolated from type B plasma and platelets (Fig 8B).
A, B and O (H) blood group structures have been shown to be covalently linked to N-glycans of VWF (Matsui et al, 1992). This study confirms the presence of covalently bound AB antigens on VWF in a blood group-specific manner. The study further demonstrates that AB structures are present on the 48/52-kDa fragments of trypsin-treated VWF, which contains the A1 domain critical in regulating the binding of VWF to platelet glycoprotein Ib/IX (Meyer & Girma, 1993).
Removal of A or B antigen from type A or B VWF with Azyme or Bzyme decreased the RCoF of the VWF. The reduction of VWF activity resulted specifically from the removal of the A or B antigen because cross-treatment of type A VWF with Bzyme, type B VWF with Azyme or type O VWF with Azyme or Bzyme did not affect the RCoF activity of the VWF. Blocking of A or B antigen with polyclonal anti-A or anti-B also decreased its RCoF (data not shown).
Both RCoF- and collagen-binding activities of VWF are affected by the size of the VWF multimers. The selective decrease in RCoF but not in collagen binding suggests that the observed decrease in RCoF in association with AB antigen removal is not caused by a loss of the large multimers. SDS agarose gel electrophoresis confirms that treatment with the enzymes does not alter the multimeric structure of VWF.
How does removal of AB antigen from VWF cause a decrease in its RCoF activity? Platelet aggregation is determined not only by the binding of VWF molecules to platelets but also by their capacity to bind multiple platelets. A decrease in the VWF RCoF activity might be caused by a decrease in VWF–platelet binding. However, our data demonstrated that the same VWF or VWF at concentrations similar to those used in ristocetin cofactor assays did not exhibit a difference in platelet binding compared with control VWF. This suggests that the decrease of VWF RCoF after AB antigen removal cannot be attributed to deceased VWF–platelet binding. However, these data do not preclude the possibility that AB antigens affect VWF–platelet binding under different conditions.
Unfolding of VWF by shear stress has been shown to enhance the capacity of VWF to support ristocetin-induced platelet aggregation (Tsai & Lian, 1998), without increasing its binding to platelets (unpublished observations). Thus, we hypothesize that removal of AB antigens may cause the opposite changes in VWF; it renders the VWF more compact and less capable of binding multiple platelets. More studies are needed to determine the validity of this hypothesis.
Sialic acid or galactose moieties have been suspected to play a role in the adhesive activity of VWF (Sodetz et al, 1977; Gralnick, 1978; Kao et al, 1980). However, the observed decrease may have been due to proteolysis caused by contaminating proteases in the enzyme preparations because other studies failed to demonstrate a change in VWF activity by either neuraminidase or β-galactosidase when the experiments were conducted in the presence of protease inhibitors (Federici et al, 1984; Goudemand et al, 1985). On the other hand, recombinant VWF specifically lacking O-linked carbohydrates exhibits less binding to platelets and a diminished capacity to promote ristocetin-dependent platelet agglutination (Carew et al, 1992). Platelet VWF has been found to be less effective than plasma VWF in supporting platelet agglutination, despite its particularly large multimeric composition (Williams et al, 1994). Sweeney & Hoernig (1992) reported that platelet VWF does not contain AB antigens. Our studies confirm the absence of AB antigens on platelet VWF. Based on the findings in this study, we speculate that the lack of AB structures may contribute to the lower RCoF activity of platelet VWF than that of plasma VWF.
A difference in VWF activity may have important clinical implications. ABO blood type has been found to be an important determinant of various bleeding and thrombotic disorders. For example, high A/O and B/O relative incidences are found in patients with thrombotic strokes, whereas the opposite is found in patients with haemorrhagic strokes (Ionescu et al, 1976). Ischaemic heart disease also occurs more frequently in non-O blood group people (Medalie et al, 1971; Erikssen et al, 1980; Meade et al, 1994). On the other hand, type O is overrepresented by patients of type I von Willebrand disease (Gill et al, 1987) and by patients of Hermansky–Pudlak syndrome with bleeding manifestations (Witkop et al, 1993). Our findings provide a plausible explanation of why type O blood group, compared with non-O blood group, is associated with a lower plasma VWF RCoF activity, a higher risk of bleeding and a lower risk of thrombosis.
This study was supported in part by grants nos HL 62131 and HL 38655 from the National Heart, Lung and Blood Institute of NIH (H-MT).