Mitsuru Murata, Department of Medicine, School of Medicine, Keio University, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan. E-mail: email@example.com
Interaction of platelet glycoprotein (GP) Ibα with von Willebrand factor (VWF) is essential for thrombus formation, particularly under high shear conditions. Previous case–control studies indicated that two GPIbα polymorphisms, 145Thr/Met and/or variable number (1–4) tandem repeats of 13 amino-acid sequences, are associated with arterial thrombosis. The 145Met-allele and the 3R- or 4R-allele is associated with increased risk. However, there is little clear experimental data to support this association. To elucidate the functional effects of these polymorphisms, we prepared recombinant GPIbα fragments and tested them in vitro. The dissociation constants of ristocetin-induced 125I-labelled VWF binding to two forms of soluble recombinant GPIbα [1His–302Ala, either 145Thr (145T) or 145Met (145M)] were not different. Four types of Chinese hamster ovary cells expressing full-length GPIbαβ/IX, 145T with one repeat (T1R), 145M with one repeat (M1R), 145T with four repeats (T4R), and 145M with four repeats (M4R), were prepared, and cell interactions with immobilized-VWF were examined under various shear conditions. The cell rolling velocity of M4R under a shear condition of 114/s was significantly slower than that of T1R. Intermediate values were obtained with M1R and T4R. The results suggest that M4R interacts more strongly with VWF under flow conditions.
Glycoprotein (GP) Ib-IX-V complex is a platelet membrane receptor for von Willebrand factor (VWF) (Lopez, 1994; Clemetson, 1997; Ware, 1998; Andrews et al, 2003). This receptor consists of four subunits, GP Ibα, Ibβ, IX, and V. The largest subunit of the complex, GPIbα, has a VWF-binding site within the N-terminal 45-kDa extracytoplasmic domain of approximately 300 amino acids (Titani et al, 1987; Huizinga et al, 2002; Uff et al, 2002). Interaction of GPIbα with VWF mediates high shear-stress-dependent platelet activation, which is a critical step for thrombus formation (Ikeda et al, 1997; Dopheide et al, 2001; Ruggeri, 2003). The VWF/GPIbα interaction is not observed under static conditions in vitro, but only under shear conditions. Assessment under static conditions requires the presence of non-physiologic inducers, such as ristocetin or botrocetin.
In previous case–control studies, two genetic polymorphisms within the coding region of GPIbα were reportedly associated with arterial thrombosis, such as coronary artery disease and stroke (Hato et al, 1997; Murata et al, 1997; Sonoda et al, 2000; Simmonds et al, 2001; Yamada et al, 2002; Afshar-Kharghan et al, 2004). The first polymorphism is an amino acid dimorphism, Thr/Met, at residue 145 (Murata et al, 1992). The second polymorphism is a variable number tandem repeat [1–4 repeats (1R–4R)] of the 13-amino acid sequence, residues 399–411 (VNTR polymorphism) (Moroi et al, 1984; Ishida et al, 1991; Simsek et al, 1994). These two polymorphisms are in linkage disequilibrium (Ishida et al, 1991; Simsek et al, 1994). The 145Met-allele, which is tightly linked to the 3R- or 4R-allele, is associated with increased risk. There is a race difference in the genotype distribution of the VNTR polymorphism; although 3R is observed in Caucasians, African-Americans (Afshar-Kharghan et al, 2004), Japanese, and Koreans (Ishida et al, 1996), 4R is observed in Japanese and Koreans (Ishida et al, 1996). Epidemiologic data indicate that GPIbα polymorphisms are clinically significant. Molecular mechanisms for the association between thrombus formation and those polymorphisms, however, are not yet clearly understood. To elucidate the effects of 145Thr/Met and/or VNTR polymorphisms on interactions with VWF, we performed two series of experiments; (a) ristocetin-induced 125I-labelled VWF binding to two recombinant fragments containing a partial GPIbα sequence (1His–302Ala) with either 145Thr (145T) or 145Met (145M), and (b) the interaction between immobilized VWF and four types of Chinese hamster ovary (CHO) cells expressing full-length GPIbα/β/IX, 145Thr/1R (T1R), 145Met/1R (M1R), 145Thr/4R (T4R), or 145Met/4R (M4R), under flow conditions.
Preparation of recombinant GPIbα fragments
The GPIbα insert of a pBluescript KS (−) construct, which contained a cDNA encoding a partial GPIbα sequence (1His–302Ala) with 145Thr, was cloned (Murata et al, 1991). The 145Thr/Met substitutions on the GPIbα insert were performed using Quick ChangeTM (Stratagene, La Jolla, CA, USA), and each insert was ligated with a expression vector pcDNA3·1Zeo (−) (Invitrogen, Groningen, The Netherlands). Each construct was sequenced to ensure that the introduced mutation was restricted to residue 145 and then transfected into CHO cells (Dainippon Pharmaceutical Co., Osaka, Japan) using FuGENETM 6 Transfection Reagent (Roche, Nutley, NJ, USA). Cells were cultured in the presence of 300 μg/ml of zeocine (Invitrogen) for selection of stable transfectants. Culture medium containing secreted soluble protein was collected from 145T-, 145M-, or mock-transfected cells after serum-free culture medium for 48 h.
Quantitation and immunologic evaluation of recombinant proteins
The expression of each recombinant protein was confirmed by Western blot analysis with anti-GPIbα monoclonal antibody, LJ-Ibα1 (a generous gift from Dr Z.M. Ruggeri, The Scripps Research Institute, La Jolla, CA, USA), which recognizes an epitope within the 45-kDa domain and reacts strongly with the reduced GPIbα fragment, using an enhanced chemiluminescence (ECL) Western blotting systemTM (Amersham Pharmacia Biotech, Buckinghamshire, UK). The amounts were measured by dot-blot analysis with LJ-Ibα1 and 125I-anti-mouse IgG (Daiichi Pure Chemicals, Tokyo, Japan). Purified recombinant GPIbα (Moriki et al, 1997) was used as a standard. Equivalent amounts of 145T and 145M were evaluated for their immunologic reactivity toward a panel of anti-GPIbα monoclonal antibodies, LJ-P3 (a generous gift of Dr Z.M. Ruggeri, The Scripps Research Institute), GUR83–35 (Takara Shuzo, Shiga, Japan), and GUR20–5 (Takara Shuzo), which recognize conformation-specific epitopes within the 45-kDa domain (Handa et al, 1986; Vicente et al, 1988; Kawasaki et al, 1995; Ikeda et al, 2000). 125I-anti-mouse IgG was used as a secondary antibody in this quantitative analysis. A constant concentration (2 ng/μl) of 145T and 145M was used for subsequent functional analysis.
Ristocetin-induced 125I-VWF binding to immobilized recombinant GPIbα fragment
Human VWF was provided by WelFide Co. (Osaka, Japan), and was radiolabelled with 125I (Amersham Pharmacia Biotech) according to the IODO-GEN procedure (Fraker & Speck, 1978). The analysis of ristocetin (final concentration, 1·0 mg/ml)-induced 125I-labelled soluble VWF (final concentration, 1·0 ug/ml) binding to immobilized 145T or 145M (400 ng/spot) was performed using the enzyme-linked immunofiltration assay apparatus (Pierce Chemical Co., Rockford, IL, USA). Details of this assay were described previously (Murata et al, 1991; Moriki et al, 1997). In the Scatchard plot analysis for the binding of 125I-VWF (0·5–16 ug/ml) to 145T or 145M, the dissociation constant was analysed by a simple regression model. It was assumed that the same proportion of VWF multimers binds to each recombinant protein, and that the molecular weight of VWF was 220-kDa.
Establishment of CHO cells expressing the GPIbαβIX complex
A stable transfectant for GPIbβIX-expressing CHO cells was established, as described previously (Suzuki et al, 1999). A cDNA encoding the GPIbα sequence was cloned into a pBluescript KS (−) as described previously (Suzuki et al, 1999) and was subcloned into a mammalian expression vector pcDNA 3·1 Hygro (+) (Invitrogen) using the restriction sites for Kpn I (Takara Shuzo) and Not I. We prepared four types of plasmids for expression, T1R, M1R, T4R, and M4R. 145Thr/Met substitution was created by polymerase chain reaction (PCR)-based site-directed mutagenesis using Quick ChangeTM (Stratagene), and then inserted into the 1R or 4R sequences; PCR was performed on genomic DNA with the 1R and 4R alleles, followed by subcloning using a TA cloning kit (Invitrogen). Each insert was subsequently cloned into the GPIbα-pcDNA 3·1 Hygro (+) using Xba I restriction sites and sequenced. Each plasmid was transfected into GPIbβIX-expressing CHO cells using FuGENETM 6. These cells were grown in culture medium with 800 μg/ml of G418, 300 μg/ml of zeocine, and 400 μg/ml of hygromycine for selection of GPIX, GPIbβ, and GPIbα respectively.
Measurement for GPIbαβIX expression on CHO cells
Expression of GPIbα on CHO cells was confirmed by flowcytometry analysis with either the anti-GPIbα antibody, LJ-P3, or the anti-GPIX antibody, SZ1 (Immunotech, Marseille, France), which reacts with the GPIbIX complex, but does not react with GPIb or GPIX alone. Four types of cells were independently sorted by fluorescence activated cell sorter (FACS) analysis using SZ1. Subsequently, quantitation of GPIbα on each cell was performed using an enzyme immunosorbent assay (EIA) with a glycocalicine EIA kit using GUR83–35 and GUR20–5, according to the supplied protocol. Purified glycocalicine of GPIbα (Lopez, 1994), which has a GPIbα extracytoplasmic domain that contains sites for the 145Thr/Met and VNTR polymorphisms, was used as a standard in this assay.
Perfusion studies: analyses for the interaction of GPIbαβIX-expressing CHO cells with immobilized VWF under flow conditions
Glass cover slips were incubated with 10 μg/ml of VWF at 4°C overnight and then blocked with 0·5% bovine serum albumin (BSA) (Sigma-Aldrich, Tokyo, Japan) at room temperature for 1 h. CHO cells were harvested with 0·5 mmol/l EDTA, washed twice with phosphate buffered saline, and resuspended in 0·5 mmol/l EDTA/HEPES-Tyrode's buffer without Ca2+ and Mg2+ to a final concentration of 2 × 105/ml. The interaction between 106 cells expressing GPIbαβIX and immobilized VWF was examined using a recirculating flow chamber system (Nishiya et al, 2000). Cells interacting with the surface were monitored for a 4-min period. Data were stored on video tape. Single-frame video images were analysed using an image processor, an Argus 50 image processor (Hamamatsu Photonics, Hamamatsu, Japan), and rolling velocities of the cells were analysed using an image processor, an Argus 20 image processor (Hamamatsu Photonics). Rolling velocity was determined as the distance cells rolled per second. To confirm the specificity of the cell rolling, experiments under flow conditions were performed in the presence of 10 μg/ml of soluble-VWF or after incubation with 50 μg/ml of anti-GPIbα antibody GUR83–35 at room temperature for 15 min.
Differences in immunologic reactivity between 145T and 145M were assessed using Student's t-test. Analysis of covariance (ancova) was used to compare the influence of three variables, GPIbα sequence (145T vs. 145M), VWF binding, and day of the experiment, in VWF binding between 145T and 145M. Analysis of variance (anova) was used for analysis of GPIbα expression on CHO cells and the perfusion studies among four types of GPIbα-expressing cells. The Bonferroni post hoc test for multiple comparisons was performed to compare the rolling velocity among the cells. ancova was used to analyse three variables; GPIbα sequence (T1R vs M4R), rolling velocity, and day of experiment. A P-value of less than 0·05 was considered to be statistically significant.
Comparison of immunologic reactivity of recombinant GPIbα fragments
To characterize the 145Thr/Met polymorphism, we established stable cell lines that secreted recombinant GPIbα fragments (1His–302Ala), 145T or 145M, into the culture medium. The secretion of each fragment into the culture medium was confirmed by Western blot analysis using the LJ-Ibα1 antibody under reduced conditions. An LJ-Ibα1-positive species with a molecular mass of approximately 45-kDa was observed in 145T and 145M, but not in the culture medium from mock-transfected cells. Subsequently, we quantified each fragment by dot-blot analysis with LJ-Ibα1 under reduced conditions. There was similar immunologic reactivity to LJ-Ibα1 between 145T and 145M. An equivalent amount of each fragment was tested in dot-blot analysis for their immunologic reactivity to several anti-GPIbα monoclonal antibodies that recognize confirmation-specific epitopes within the 45-kDa domain and 125I-anti mouse IgG as a secondary antibody. The binding of each antibody, as measured by counts per minute, was not significantly different between 145T and 145M (Table I), suggesting that the 145Thr/Met polymorphism does not affect immunologic reactivity for LJ-P3, GUR20–5, and GUR83–35. Similar results by dot-blot analysis using the ECL detection system with anti-mouse IgG antibody as a secondary antibody were observed in seven independent experiments (data not shown).
Table I. Immunologic reactivity (cpm count) in dot-blot analysis.
Values are mean ± SD of duplicate determinations in one experiment.
2594·0 ± 17·0
2536·0 ± 130·1
2532·0 ± 93·3
2771·0 ± 168·3
2777·0 ± 75·0
2730·0 ± 435·6
2930·0 ± 45·3
2622·0 ± 260·2
Ristocetin-induced 125I-labelled VWF binding to immobilized recombinant GPIbα fragments
To elucidate the effect of the 145Thr/Met polymorphism on VWF binding, we performed an experiment using recombinant GPIbα fragments without the VNTR polymorphism site, containing residues 1–302. This binding was examined in the absence or presence of ristocetin under static conditions (Fig 1). The binding levels (pmol/well; mean ± SD) were 0·011 ± 0·005 for BSA, 0·018 ± 0·008 for 145T, and 0·016 ± 0·004 for 145M in the absence of ristocetin, indicating no specific binding in this experimental condition. In the presence of ristocetin, specific 125I-VWF binding was observed in 145T and 145M, but not BSA; 0·015 ± 0·008 for BSA, 0·039 ± 0·015 for 145T, and 0·047 ± 0·017 for 145M, and the binding levels were not different between 145T and 145M (P = 0·1185).
To evaluate the binding affinities of 145T and 145M, a Scatchard plot analysis was performed in the presence of 1·0 mg/ml of ristocetin. Both samples had saturable binding of 125I-VWF, and the dissociation constants (nmol/l; mean ± SD) were 620·5 ± 176·1 for 145T and 406·0 ± 40·2 for 145M (P = 0·0551). The maximum number of binding sites (Bmax, nmol/l; mean ± SD) was 68·0 ± 11·4 for 145T and 56·1 ± 11·3 for 145M (P = 0·1889). These results suggest that 145T and 145M have a similar binding affinity for VWF in the presence of ristocetin.
Quantitation of GPIbα expression on the GPIbβIX-expressing CHO cells
We prepared four types of GPIbαβIX-expressing CHO cells; two naturally occurring sequences, T1R and M4R, and two artificial or extremely rare sequences, T4R and M1R, to investigate which GPIbα polymorphisms affect the interaction with VWF under flow conditions. First, the surface density of the GPIbαβIX complex on each cell was examined by flowcytometry analysis with LJ-P3 or SZ-1, and the reactivity was not different among cells. Subsequently, four types of GPIbαβIX-expressing CHO cells were isolated using FACS to obtain cells with an equivalent amount of the GPIbαβIX complex on the surface. Moreover, quantitative determination of each sorted cell was performed by EIA using the anti-GPIbα antibodies, GUR83–35 and GUR20–5. The GPIbα expression levels, calculated by standard curves for absorbance and glycocalicine as a standard protein, are shown in Table II. There was no statistically significant difference in the GPIbα expression levels among the four types of cells, although T4R had a slightly higher expression level.
Table II. GPIbα molecule on single CHO cell.
GPIbα molecule (×106)
Values are mean ± SD of duplicate determinations in three independent experiments.
4·26 ± 0·31
4·46 ± 0·34
5·26 ± 0·48
4·36 ± 0·17
Interaction between immobilized VWF and CHO cells expressing the GPIbαβIX complex under flow conditions
To determine the optimal shear condition, preliminary studies were performed using T1R cells to monitor the interaction between immobilized VWF and GPIbα-expressing CHO cells. Rolling cells per unit area (mm2) of VWF-immobilized glass were counted under various shear conditions. The number of rolling cells was 26·91, 16·34, 8·65, 0·96, and 0 for shear conditions of 32/s, 64/s, 114/s, 214/s, and 600/s respectively. In CHO cells without GPIbα, the rolling cell number was 3·84, 0·96, 0, 0, and 0 for shear conditions of 32/s, 64/s, 114/s, 214/s, and 600/s respectively. We then examined the effect of the anti-GPIbα antibody GUR83–35 on rolling cell number per minute and rolling velocity. Under a shear condition of 32/s, the number of rolling cells per minute was 14 and 16 in the absence and presence of GUR83–35 respectively. Under a shear condition of 114/s, the number was 134 and 41 in the absence and presence of GUR83–35 respectively. The rolling velocity under a shear condition of 32/s was similar between the two experimental conditions; 240·61 ± 67·49 (μm/s, mean ± SD) in the absence of GUR83–35 and 291·98 ± 60·92 in the presence of GUR83–35. In contrast, the rolling velocity under a shear condition of 114/s was different in the absence and presence of GUR83–35; 704·74 ± 153·5 and 959·10 ± 172·1 respectively. These findings indicated that the observed interaction was specific for GPIbα under a shear condition of 114/s, but not of 32/s. Thus, we adopted the shear condition of 114/s for the subsequent experiments.
As shown in Fig 2 and Table III, the rolling velocity of M4R under shear conditions of 114/s was significantly slower than that of T1R (P = 0·00042). The rolling velocities of M1R and T4R were similar, with values that were intermediate between those for T1R and M4R. Moreover, we analysed the rolling velocity between T1R and M4R by ancova to test whether the difference identified by anova was affected by different day of experiment. The rolling velocity was 1077·7 ± 20·9 (mean ± SD) for T1R and 918·4 ± 17·3 for M4R (P < 0·0001), indicating that the significant difference between T1R and M4R was not influenced by interactions with other variables among the four independent experiments. The number of rolling cells interacting with the immobilized VWF per minute was 32·0 ± 5·6 for T1R, 34·7 ± 2·1 for M1R, 26·7 ± 3·1 for T4R, and 35·7 ± 2·5 for M4R. These values were not significantly different among the four types of cells (P = 0·0574). In GPIbβIX-expressing CHO cells, the rolling cell number was 5·2 ± 0·5. Under this flow condition, the cell rolling velocity was inhibited by GUR83–35; i.e. the velocities became faster and with soluble-VWF addition (Table III). Thus, the specificity of interaction between immobilized VWF and GPIbα-expressing CHO cells was confirmed under a shear condition of 114/s.
Table III. Rolling velocity for GPIbα-expressing cells interacting with VWF under 114/s flow condition
Soluble VWF (−)
Soluble VWF (+)
Values for rolling velocity (μm/s) are mean ± SD. NS, not significant.
1035·1 ± 40·5*
1291·1 ± 61·9
1385·7 ± 82·5
951·1 ± 26·4**
1486·8 ± 109·5
1165·4 ± 62·9
952·5 ± 36·9**
1638·9 ± 384·6
1276·6 ± 86·5
902·2 ± 30·8*
1521·5 ± 86·6
1293·5 ± 126·4
The present study demonstrated that the 145Met and 4R polymorphisms of GPIbα facilitate interaction with immobilized VWF under flow conditions, which is a highly adaptive physiologic response. To date, molecular mechanisms for the functional differences of the 145Thr/Met and/or VNTR polymorphisms in GPIbα have not been fully understood whereas numerous epidemiologic data have been reported. We report the first experimental data obtained using recombinant proteins to determine the functional differences of 145Thr/Met and VNTR GPIbα polymorphisms. Previously, 145Met and/or 3R/4R polymorphisms were demonstrated to be associated with an increased risk for arterial thrombosis, such as coronary artery disease or stroke (Simmonds et al, 2001; Yamada et al, 2002). Because the 145Thr/Met and VNTR polymorphisms are in linkage disequilibrium, focus on either 145Met or 3R/4R allele was likely to be sufficient to examine the association between GPIbα polymorphisms and arterial thrombosis in the epidemiological study. We reported that the frequency of either 145Met- or 4R-allele among patients with coronary artery disease was higher than that among control subjects and that the genotypes with the 145Met-allele were more frequently found in the patients with cerebrovascular disease than in control subjects (Murata et al, 1997; Sonoda et al, 2000). A large case-cohort study (Afshar-Kharghan et al, 2004) showed the relationship of the 2R/2R genotype with a lower risk of coronary heart disease in African-Americans. However, conflicting data have also been published (Hato et al, 1997; Simmonds et al, 2001). In experimental studies of these polymorphisms, Boncler et al (2002) demonstrated that the inhibitory effect of the VWF antagonist on ristocetin-induced agglutination was higher in 145Met/3R-positive platelets than in 145Met/3R-negative platelets. Ulrichts et al (2003) reported that platelets with 145Thr or recombinant GPIbα (residues 1–289) with 145Thr had a higher VWF binding affinity than 145Met. These findings are not consistent with our results although the experimental conditions of the present study differed from those of previous studies: use of ristocetin or botrocetin or use of an assay system. Other studies have also used various methods with inconsistent results (Mazzucato et al, 1996; Li et al, 2000; Jilma-Stohlawetz et al, 2003). These reports suggest that the functional analyses of GPIbα polymorphisms seem to be easily affected by several factors in relation to platelet activation or experimental conditions. Therefore, in this study, recombinant GPIbα and purified human VWF were examined under two experimental conditions to focus on the relationship between GPIbα polymorphisms and interactions with VWF. The first study, using soluble GPIbα lacking the VNTR polymorphism site, did not show the effect of the 145Thr/Met polymorphism on the major conformation because the immunoreactivity to anti-GPIbα antibodies that recognize confirmation-specific epitopes were not significantly different between these polymorphisms. The 145Thr/Met polymorphism did not affect the 125I-VWF binding in the presence of ristocetin under static conditions. Although ristocetin provides a convenient method to investigate the VWF/GPIbα interaction in vitro, it is not a physiologic substance. Thus, the second study was designed with an alternative approach, an in vitro assay for VWF/GPIbα interaction under flow conditions. Cells expressing GPIbα were prepared as a GPIbαβIX complex because expression of a full-length GPIbα alone was unstable in the cell culture system (Lopez et al, 1992). Two types of cells with naturally occurring sequences (T1R and M4R) and two types of cells with artificial or extremely rare sequences (T4R and M1R) were used to determine which polymorphism was more closely related to the VWF/GPIbα interaction. We carefully measured the GPIbα expression level on each cell because these levels were reported to affect the VWF/GPIbα interaction under flow conditions (Nishiya et al, 2000). After using FACS to obtain cells expressing similar GPIbα levels, EIA assay was performed using GUR83–35 and GUR20–5. Because these two antibodies were shown not to be influenced by the 145Thr/Met polymorphism (Table I), we used these antibodies in this assay. Perfusion analyses of the quantified cells indicated that M4R, which is a risk factor for arterial thrombosis, had a high ability to interact with VWF under a flow condition of 114/s, as compared with T1R. This flow condition of 114/s may correspond to wall shear rate for large veins in vivo (Bevan et al, 1995), where VWF-dependent platelet phenomena may not take place. Compared with platelets, however, CHO cells have 2·5- to threefold larger diameters, and the GPIbα-expressing CHO cells are approximately 20-fold higher in GPIbα density. The cell size and receptor density are likely to affect the sensitivity of cells to flow conditions. Also, we were unable to determine the order of effectiveness of the polymorphisms among the four sequences, 145Thr, 145Met, 1R, and 4R, in VWF/GPIbα interactions because T4R and M1R had a similar ability to interact with VWF. Although the synergistic effect of the 145Thr/Met and VNTR polymorphisms on GPIbα function remains unclear, the present data are compatible with previous speculations (Lopez, 1994; Murata et al, 1997) that GPIbα with 4R is longer in size and thus places the VWF-binding global domain further away from the platelet plasma membrane. Thus, VWF would be more easily accessible to the binding site on the receptor under high shear conditions. Functional polymorphisms of GPIbα might be responsible for the increased prevalence of arterial thrombosis. Our observations might explain the molecular basis for the previous epidemiologic studies. Further studies to examine the interactions between GPIbα polymorphisms and other ligands are necessary. The present data support a potentially new therapeutic approach to arterial thrombosis by targeting specific GPIbα polymorphisms.