Analysis of Structure-Function Relationships in the Platelet ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 268, No. 16, Issue of June 5, pp. 12185-12192,1993 Printed in U.S.A.

Analysis of Structure-Function Relationships in the Platelet Membrane Glycoprotein Ib-binding Domain of von Willebrand’s Factor by Expression of Deletion Mutants*

(Received for publication, November 16,1992, and in revised form, February 18,1993)

Mitsuhiko Sugimoto, Judith Dent, Richard McClintock, Jerry Ware, and Zaverio M. RuggeriS From the Roon Research Center for Arteriosclerosis and Thrombosis, Division of Experimental Hemostasis and Thrombosis, Department of Molecular and Experimental Medicine and Committee on Vascular Bwbgy, The Scripps Research Institute, LaJolla, California 92037

We have used a series of Escherichia coli-expressed deletion mutants of the glycoprotein (GP) Ib-binding domain of von Willebrand factor (vWF) to study the structural basis of its function. In addition to the prototypic molecule (~vWF~~~-” ’ ) , we constructed 11 mutants; seven had deletions of sequence on the amino- and/or carboxyl-terminal side of the intrachain disulfide loop, and four had limited deletions inside the loop. Other cysteine residues in addition to 609 and 695, when present in the corresponding native sequence, were mutated to glycine; all molecules were purified in the oxidized as well as reduced and alkylated state. The smallest species retaining the ability to interact with GP Ib in the absence of modulators was the oxidized ~ v W F ~ ~ * - ~ ~ ~ ; the latter, as well as rvWF441- ee6, became inactive after reduction and alkylation. In contrast, all the other fragments with deletions outside of the loop, but extending at least to residue 700, showed better binding to platelets after reduction and alkylation than when the Cyssoe-Cysee6 disulfide bond was oxidized. Any limited deletion of sequence inside the loop caused complete loss of G P Ib-binding function both in the absence or in the presence of botrocetin, and this persisted even after reduction and alkylation. In contrast, all mutants with intact sequence between residues 609 and 696 bound to GP Ib in the presence of botrocetin, regardless of whether the 2 cysteine residues were oxidized or reduced and alkylated. Ris- tocetin, unlike botrocetin, appeared to have no effect in modulating the binding of any of the expressed fragments to platelets. Our findings suggest that the GP Ib-binding domain of vWF contains multiple interaction sites, but integrity of the sequence 509-696 is important for function.

The interaction of von Willebrand’s factor (vWF)’ with the glycoprotein (GP) Ib receptor is critical for the initiation of platelet deposition at sites of vascular injury both during

* This work was supported in part by Grant HL-48728 from the National Institutes of Health. This is Paper 7679-MEM/CVB from The Scripps Research Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom correspondence should be addressed The Scripps Re- search Inst., SBR-8, 10666 North Torrey Pines Rd., La Jolla, CA 92037. Tel.: 619-554-8950; Fax: 619-554-6779.

The abbreviations used are: vWF, von Willebrand‘s factor; GP, platelet membrane glycoprotein; HPLC, high performance liquid chromatography; PAGE, polyacrylamide gel electrophoresis.

normal hemostasis (1) and in the development of acute arte- rial thrombosis (2-4). The isolation and characterization of fragments generated by limited proteolytic digestion of purified plasma vWF has led to the demonstration that the GP Ib-binding domain is localized between residues 449 and 128 of the mature vWF subunit (5,6). Experiments with synthetic peptides and monoclonal antibodies have suggested that two short segments of sequence within the domain, spanning residues 474-488 and 694-708, may be responsible for GP Ib- binding activity ( 7 ) . The general validity of this conclusion has been challenged by the observation that the two peptides fail to inhibit vWF binding to platelets induced by the modulator botrocetin (8,9), even though this interaction, like that supported by ristocetin, is mediated by the same functional domain of vWF comprising residues 449-728 (10). Moreover, others have suggested that the segment of sequence comprising residues 514-542 has a crucial role in supporting vWF binding to GP Ib regardless of the conditions under which the interaction is modulated (11).

It has become apparent, therefore, that several distinct structures of vWF may be necessary to support its binding to GP Ib, and that the process is more complex than previously thought. In order to probe in detail the structure-function relationships relevant for this interaction, we have performed selected mutagenesis experiments utilizing the recently reported bacterial expression system for the corresponding isolated domain (9). The results obtained provide information on the minimal structures required for activity and may help in the design of a new class of selective anti-thrombotic agents.

EXPERIMENTAL PROCEDURES

Expression of the Parent vWF Fragment Containing the GP Ib- binding Site and of Deletion Mutants-A cDNA fragment coding for residues 441-733 of the mature vWF subunit was synthesized in a polymerase chain reaction using two oligonucleotides that add an EcoRI restriction site at the 5’ end of codon and a Hind111 restriction site at the 3’ end of codon (12). The amplified fragment was cloned into the filamentous phage M13mp18 as an EcoRI-Hind111 restriction fragment, and the presence of the correct coding sequence for mature vWF residues 441-733 was confirmed (13, 14). Within this stretch of sequence, the native vWF cDNA contains seven codons for Cys at positions 459, 462, 464, 471, 474, 509, and 695; in the cloned fragment, the first five Cys codons were replaced with Gly codons by successive site-directed mutagenesis procedures utilizing a uracil template of the M13 mp18/441-733 construct (15). Thus, a cDNA fragment coding for residues 441-733 was generated that contained only two Cys codons at positions 509 and 695 (15). Following mutagenesis, the EcoRI-Hind111 fragment was modified with BamHI linkers and cloned into pET3a, an Esche- richia coli expression vector containing a T7 RNA polymerase promoter (16); the resultant plasmid construct was designated

12185

12186 Structure of the GP Ib-binding Domain of vWF ~ v W F ~ ~ ~ - ~ ~ ~ . The vWF polypeptides were synthesized in the host E. coli BL21-DE3 by induction with isopropyl-0-D-thiogalactopyrano- side as described previously (16, 17). Immunoblotting analysis confirmed that the vWF fragment synthesized under direction of the plasmid was contained mainly in inclusion bodies. Because of the cloning strategy used, the expressed vWF fragment, designated

, contains 20 non-vWF residues on the amino-terminal side of Arg4" and 22 non-vWF residues on the carboxyl-terminal side of Val7=. The presence of this extraneous sequence had no conse- quence on the functional properties of rvWP1-733, as shown by comparison with an analogous recombinant fragment that contained only native vWF sequence (9).

For mutagenesis of P v W F ~ " - ~ ~ ~ , an XbaI-Hind111 restriction fragment was removed from the parent plasmid and cloned into MI3 mp18. The XbaI-Hind111 fragment contained the T7 RNA promoter sequence, the coding sequence for rvWF141-733, and the translation termination sequences within pET3a. Mutagenesis was performed using uracil-containing templates and oligonucleotides designed to generate the desired recombinant molecules. The amino-terminal deletions of NWF'"~~ were engineered to remove the short non- vWF DNA coding sequence present in the parent fragment. This was achieved by placing the initiating Met codon of the gene 10 capsid protein within pET3a immediately 5' to the vWF codon for Thr"'

should be noted that the initiating methionine may or may not be removed from the expressed polypeptide, depending in a manner not yet firmly established on the primary structure of the amino-terminal sequence. The carboxyl-terminal truncations were engineered by adding two translation termination codons, TGA, immediately following vWF codons for Pro7@' (NWP"-~'" and NWF'~~'"), Glu7@'

cated molecules lacked the short non-vWF coding sequence present at the carboxyl terminus of the parent fragment. Finally, internal deletions of N W F " " ~ ~ ~ were engineered to delete residues 512-526, 540-560,560-580, or 629-643. After mutagenesis, the sequence of all the constructs was verified to confirm the presence of the desired mutations as well as the integrity of the remainder of the coding sequence. The mutated constructs were then cloned into pET3a as XbaI-Hind111 restriction fragments and expressed in host E. coli

Purification and Characterization of Recombinant u WF Fragments Containing Either Oxidized or Reduced and Alkylated Cysteine Resi- dues-Inclusion bodies obtained from the high level expression of recombinant vWF fragments in E. coli were solubilized by resuspen- sion in 6 M guanidine hydrochloride, 100 mM dithiothreitol, 50 mM Tris, pH 8.8. After incubation for 30 min at 37 "C, the mixture was filtered through a 0.45-pm filter (Gelman Acrodisc), and the soluble fraction was applied to a reversed phase Vydac C8 column (1 X 30 cm, 10-pm particle, and 3-nm pore size; Separations Group) heated to 40 'C in a column oven (model 925; Tracor) and connected to a Perkin-Elmer Cetus series 400 high performance liquid chromatography (HPLC) system. Elution was attained with a 60-min acetonitrile gradient, from 30 to 60% in 0.1% trifluoroacetic acid and 10% n-propyl alcohol. Detection was at 210-229 nm with a Perkin-Elmer Cetus LC-95 spectrophotometer. Fractions containing the vWF fragment, as judged by polyacrylamide gel electrophoresis (PAGE) in the presence of SDS (9), were pooled and separated into two aliquots. One was subjected to reduction and alkylation of the cysteine residues using dithiothreitol and iodoacetamide according to a previously described procedure (9). The reduced and alkylated fragment was further purified by reversed phase HPLC using the same conditions described above. In the other aliquot, oxidation of the intrachain disulfide bond between Cysm and Cys6@' was attained as follows. The pool of partially purified vWF fragment was adjusted to a concentration of less than 200 rg/ml by adding Milli Q water (Millipore); CuSO4-5H20 was then added to a concentration of 10 mM. After rotation on a circular mixer for 30 min at 22-25 "C, the sample mixture was further purified by reversed phase HPLC to remove any previously co-eluting species or contamination of multimeric molecules generated during this oxidation step. The major protein peak eluting from the column during the acetonitrile gradient was collected. TO evaluate whether the final product contained any trace amount of nonoxidized molecules, the eluted fractions were mixed in a 1:1 volume ratio with a solution composed of 6 M guanidine hydrochloride and 50 mM Tris, pH 8.8, and applied to a column of 5-thio-2- nitrobenzoic acid thiol-agarose (2-ml gel volume; Pierce Chemical Co.), a substance that binds free thiol groups. The column was pre- equilibrated with 0.1 M sodium phosphate, 1 mM EDTA, pH 6.8

rvWP41-733

( N ~ ~ 4 9 Z - 7 3 3 ) or Tyrm ( N W P ' ~ ~ , NWF~~?@', and rvWFm*). It

(rVWP41-7W ), or Aspa (mWF'"- and rvWF-*). All the trun-

BL21-DE3.

(Buffer A) and developed at flow rate of 1 ml/min. The flow-through

as the final product. fractions, exhibiting significant absorbance at 280 nm, were pooled

Both oxidized and reduced/alkylated purified vWF fragments were concentrated (SpeedVac Concentrator, Savant), dialyzed extensively with 1 mM citric acid, 1.5 mM NaCl, 5% mannitol buffer, pH 3.5, and stored at -70 'C until used. The purity and quality of all vWF fragments were assessed by 1) analytical reversed phase HPLC on a 25 X 0.46-cm Vydac C8 column under conditions similar to those described above; 2) SDS-PAGE in a 12% gel, according to Laemmli (18); and 3) immunoblotting (19, 20) and dot-blotting (21) using various anti-vWF monoclonal antibodies.

Monoclonal Antibodies-The anti-vWF monoclonal antibodies used in this study have all been characterized in detail in previous publications. LJ-RG46 and LJ-52K2 react with vWF residues 474- 488 and 694-708, respectively (7). NMC-4 reacts with the GP Ib- binding domain comprising vWF residue 449-728 and recognizes preferentially a disulfide bond-dependent conformation (7). The anti- GP Ib antibody, LJ-Ibl, interacts with the 45-kDa domain of the platelet receptor (22) and is a competitive inhibitor of vWF binding to platelets. All the mouse IgG used in this study was purified by chromatography on protein A-Sepharose (Pharmacia LKB Biotech- nology Inc.), according to established procedures (23).

Radioiodination of Proteins-The purification of vWF from cry- oprecipitate has been described in detail previously (24, 25). Botro- cetin was isolated from the crude venom of Bothrops jararaca as reported (21). Purified proteins (vWF, LJ-Ibl, botrocetin, polyclonal rabbit anti-mouse IgG from Zymed) were labeled with "'I (Amersham Corp.) using IODO-GEN (Pierce) according to the method of Fraker and Speck (26). The specific activities of radiolabeled ligands were between 1.2 X IO9 and 6.5 X IO9 cpm/mg.

Dot Blot Analysis of Recombinant u WF Fragments with Anti-u WF Antibodies-All the steps of this procedure were performed at 22- 25 "C. Various concentrations of purified recombinant fragments in a constant volume of 15 pl were applied as a drop onto dry nitrocellulose (0.45-pm pore size; Pierce) and air -dried. The nitrocellulose strips were then soaked for 15 min in a solution composed of 5% non- fat dry milk in Hepes buffer (Blotto). The strips were then incubated for 1 h in a Blotto solution containing a 1:lOOO dilution of LJ-RG46 or NMC-4 ascitic fluid, followed by three successive incubations, for 10 min each, in fresh aliquots of Blotto solution. The nitrocellulose was then incubated for 1 h in a Blotto solution containing 1261-labeled rabbit IgG against mouse IgG (1 X IO5 cpm total radioactivity count), followed by three successive incubations, 10 min each, in fresh aliquots of Blotto solution. The nitrocellulose strips were then dried and exposed to x-ray film (Kodak AR) to obtain an autoradiograph. Individual radioactive dots were then cut out from the nitrocellulose strip in patterns of identical size and the associated radioactivity was measured in a y-scintillation counter.

Inhibition of Anti-GP I b Monoclonal Antibody Binding to Platelets by Recombinant u WF Fragments-This assay, described in a previous publication (9), was based on the demonstrated ability of native vWF to inhibit the binding to platelets of an anti-GP Ib monoclonal antibody LJ-Ibl. Since the antibody, in turn, inhibits vWF binding to platelets, the corresponding epitope is thought to overlap with the vWF binding site in GP Iba (Fig. 1). Indeed, in studies not reported here, we could demonstrate that the antibody inhibits vWF binding to platelets in a competitive manner. Briefly, washed platelets (1 x 10s/ml final count) prepared by the albumin-gradient method (27) were incubated with various concentrations of recombinant fragments and a constant amount of '251-labeled LJ-Ibl (10 pg/ml) for 30 min at 22-25 "C. In some experiments, the modulators ristocetin (Sigma) or botrocetin were also added to the sample mixture at the final concentration of 1 mg/ml or 5 pg/ml, respectively. In the mixtures containing ristocetin, citrated plasma from a patient with severe von Willebrand disease (vWF antigen levels <0.1% of normal average) was added in a volume corresponding to 40% of the total to prevent nonspecific interactions between the modulator and the recombinant fragment (9). At the end of the incubation, platelets were centrifuged through a layer of 20% sucrose to separate bound from free ligand; the platelet-associated radioactivity was then determined in a y- scintillation counter. Nonsaturable binding was estimated in mixtures containing the same reagents as test mixtures, with the addition of a 100-fold excess of unlabeled ligand the corresponding bound counts (<lo% of total bound) were subtracted from all data points. Binding was expressed as a percentage of that measured in the control mixture containing Hepes buffer instead of recombinant fragments.

Complex Formation between Botrocetin and v WF-This assay was

Structure of the GP Ib-binding Domain of vWF 12187

GPIbIX Complex vWF-binding

GPlb-binding 4 domain of vWF

Platelet

FIG. 1. Schematic representation of the assay used to measure the binding of vWF fragments to GP Ib in the absence of modulators. Platelets express on their membrane the GP Ib-IX complex (formed by GP Iba, GP ID and GP IX). Native vWF in solution does not bind to GP Ib in the absence of modulators. In contrast, an anti-GP Iba antibody (LJ-Ibl), reacting with an epitope overlapping the vWF-binding site, interacts with GP Ib even in the absence of modulators. Thus, the inhibition of antibody binding by vWF fragments can be taken as a measure of their ability to interact directly with the vWF binding site in the absence of modulators.

performed using purified vWF bound to microtiter polystyrene plates and biotinylated botrocetin, according to the procedure described in detail elsewhere (21). In essence, the binding of biotinylated botrocetin to immobilized vWF was detected using peroxidase-conjugated streptavidin (Zymed) and the substrate o-phenylenediamine. The inhibition of botrocetin binding to vWF was measured by incubating a fixed amount of biotinylated botrocetin (0.2 rg/ml) and recombinant vWF fragments before addition to vWF-coated plates. The residual botrocetin binding was expressed as percentage of that measured in a control mixture containing citrate buffer instead of recombinant fragment.

Measurement of u WF Binding to Platelets-The binding of radiolabeled purified vWF to washed platelets in the presence of the modulator ristocetin was measured according to a procedure described previously in detail (24).

RESULTS

Expression of the GP Zb-Binding Domain of uWF in E. coli-The prototypic molecule expressed for these studies was modeled on the GP Ib-binding domain isolated after tryptic digestion of purified multimeric plasma vWF and known to encompass residues 449-728 of the mature subunit (5, 6); in its native state, this fragment of vWF is a homodimer (28) (Fig. 2). The recombinant molecule, containing residues 441- 733 and designated rvWF441-733, corresponds essentially to the A1 domain of vWF with the addition of a short segment of the D3 domain on the amino-terminal side (29) (Fig. 2). Five of the cysteine residues in the sequence (all pertaining to the D3 domain) were mutated to glycine; this allowed efficient refolding and formation of the intrachain disulfide bond between Cys- and Cys6” reducing the occurrence of disordered aggregates. The recombinant fragment was purified to ho- mogeneity either oxidized or after reduction and alkylation of the 2 cysteine residues in the sequence; the two forms could be differentiated by the distinct retention times in reversed phase HPLC and the different mobility in SDS-PAGE analysis (Fig. 3).

The presence or absence of the intrachain disulfide bond caused a marked difference in the way rvWF441- 733 reacted with the monoclonal anti-vWF antibody NMC-4. This antibody, which recognizes the GP Ib-binding domain of vWF only under nonreducing conditions (7), exhibited a much stronger reactivity with the oxidized form of rvWF441-

than with the reduced and alkylated fragment; in contrast, antibody W-RG46, specific for the linear sequence 474-488 of mature vWF (7,30), reacted equally well with both oxidized and reduced rvWF441-733 (Fig. 4). Thus, the differential im- munoreactivity of antibody NMC-4 provided an additional test to distinguish between the oxidized and reduced forms of rvWF441-733

n

733

Native vWF Sequence

W r vWF44”733 (oxidized)

FIG. 2. Schematic model of the prototypic recombinant vWF fragment. Upper panel, the GP Ib-binding domain of vWF is located between residues 441-733 of mature vWF. This segment of sequence contains 7 cysteine residues (C), identified here with the corresponding residue numbers in the sequence of the mature vWF subunit; in native vWF, 4 of the cysteines are engaged in two intrachain disulfide bonds (39,40) and at least 1 of the remaining 3 forms an intersubunit bond (41) (all 3 are indicated here as involved in interchain bonds, but two could form an additional intrachain bond; this point has not been clarified to date). Lower panel, in the expressed prototypic recombinant fragment, the 5 cysteine residues pertaining to the D3 domain were mutated to glycine residues (GI. The fragment contained only 2 cysteine residues with possibility of forming the Cysm-Cysag5 intrachain disulfide bond.

Interaction of ruWP41-733 with Platelet GP Zb: Effect of the Zntrachain Disulfide Bond Cy~~~’-Cys~~~-The interaction of

fragment to inhibit the binding to platelets of an anti-GP Ib monoclonal antibody LJ-Ibl. This assay offers the advantage that the binding of the antibody, unlike that of native vWF, is not dependent on the presence of modulators (like ristocetin or botrocetin) and allows evaluation of the direct binding of

(Fig. 1). Under these conditions, the reduced and alkylated form of rvWF441-733 interacted with G P Ib with a much greater affinity than the oxidized form; the concentration required to inhibit 50% of antibody binding (IC5,,) was over 10-fold lower for the reduced and alkylated fragment than for the oxidized one (Fig. 5, top panel). There was no decrease in the ICso of either form of rvWF441-733 when ristocetin was added to the experimental mixture (Fig. 5, middle panel). In contrast, botrocetin had a profound effect on the GP Ib-binding affinity of the oxidized form of rvWF441-733 and increased also the affinity of the reduced and alkylated form; in the presence of botrocetin, the oxidized species inhibited LJ-Ibl binding to platelets with an affinity similar to that of the reduced and alkylated form (Fig. 5, lower panel).

Mutational Analysis of the GP Zb-binding Domain of u WF- In order to obtain information on the minimal structural requirements necessary to support vWF binding to GP Ib, we

rvWF441-733 with GP Ib was measured by the ability of the

rvWF441-733 to GP Ib in the absence of exogenous substances

12188 Structure of the GP Ib-binding Domain of vWF

n

E 0.4

c!. Q 0 C m e s 0.2

0 0

0 I 0 0 5 10 15 20 25

Retentlon Tlme (min) L 3 0 m

66 -

29 - 24 -

- 66

- 45 - 36 - 29 -24 - 20

1 2 3 4 5 6 FIG. 3. Characterization of recombinant fragments by

HPLC and SDS-PAGE. Upper panel, separation of the oxidized (Or) and reduced and alkylated (RIA) forms of ~ v W F ~ ” ~ ’ ‘ by HPLC using a 30-60% acetonitrile gradient; 20 pg of sample were applied to the column. Lower panel, analysis of purified by SDS- PAGE (12% polyacrylamide gel). Cathode is at the top. Lanes 1-3, nonreduced samples ( N R ) ; lanes 4-6, reduced samples (R) . Molecular mass standards were applied into lanes 1 and 6; the corresponding molecular mass values are shown (in kDa) on the left and right of the gel. Lanes 2 and 4 contain the recombinant fragment purified in the oxidized form; lanes 3 and 5 contain the recombinant fragment purified after reduction and alkylation. Note the small but reproduc- ible change in mobility of the purified oxidized sample analyzed under nonreducing conditions (lane 2 ) as compared with the same sample analyzed under reducing conditions (lane 4 ) . Note also that the purified reduced and alkylated fragment has the same mobility both under nonreducing (lane 3 ) and reducing conditions (lane 5). Proteins in the gel were stained with Coomassie Blue.

expressed a series of recombinant molecules containing limited deletions of the sequence present in the prototypic rv-

shown in Fig. 6. The mutant molecules contained deletions in either one or both of the two segments of sequence flanking the C y ~ ‘ ~ - C y s ~ ~ ~ disulfide loop or limited deletions within the loop sequence itself. Some of the mutant molecules had complete or partial deletions of one or both of the two short segments of sequence, residues 474-488 and 694-708, that have previously been indicated as being involved in the binding of vWF to GP Ib (7). All molecules were purified both in the oxidized state, i.e. with the Cys509-Cys695 intrachain disulfide bond present, or after reduction and alkylation of the two cysteine residues, i.e. without the intrachain disulfide bond. In all cases, the oxidized and reduced forms of a given molecule could be differentiated by the distinct retention time in

WF441-733. , their designations and corresponding sequences are

2xlQ10 1x10.10 3 ~ 1 0 ~ ~ 2xlV10 lxlV10 3X1U11 -0x- -RIA-

r vWF441-733 (molestdot)

FIG. 4. Dot blot analysis of recombinant vWF fragments using anti-vWF antibodies. Three different amounts (as indicated) of either oxidized (On) or reduced and alkylated ( R I A ) purified

were applied to the nitrocellulose, followed by the appropriate anti-vWF mouse monoclonal antibody and by IZ5I-labeled rabbit anti-mouse IgG. Upper panel, Reactivity with the anti-vWF antibody W-RG46 which recognizes a linear epitope contained within residues 414-488 of the mature subunit. An autoradiograph of the dots is shown along with the measurement of the radioactivity bound to each dot. Lower panel, reactivity of the same fragments with the monoclonal antibody NMC-4 which recognizes an epitope strongly influenced by the presence of the intrachain Cys509-Cys695 disulfide bond. Note, in fact, that the reduced and alkylated samples fail to react significantly with antibody NMC-4, whereas they show strong reactivity with antibody W-RG46.

r v ~ ~ 1 4 1 - 7 3 3

reversed phase HPLC, with a difference similar to that seen with and/or by the distinct electrophoretic mobility (see above). The reactivity with antibody NMC-4, however, was abnormal for some of the mutant molecules even in their oxidized state; rvWF4414s, rvWF5094s, and all four molecules with different partial deletions in the loop region showed no reactivity; rvWF50a704 had markedly reduced reactivity (Fig. 6).

The interaction of each of the mutant molecules with GP Ib was evaluated using the assay based on inhibition of antibody W-Ibl binding to platelets; the corresponding IC, values (taken to represent the GP Ib-binding affinity of the inhibitory fragments) are presented in Table I. In the absence of modulators, all the molecules with intact C y ~ ~ - C y s ~ ~ ~ disulfide loop, including those with partial or complete deletions of the two segments of sequence previously indicated as part of the GP Ib-binding site of vWF, inhibited antibody binding like oxidized rvWF441-733 or better; the lowest IC,


120, I

120'

80

40

r vWF441-733 (pM)

FIG. 5. Interaction of recombinant vWF fragments with GP Ib. These experiments were based on measuring the ability of rv- WF'"733, used at various concentrations as indicated on the abscissa, to inhibit the binding to plateIets of the 'BI-IabeIed monoclonal anti- GP Iba antibody, LJ-Ibl, used at a final concentration of 10 pg/ml. In all experimental mixtures, washed platelets were used at a final count of 1 X 108/ml and were incubated with the other reagents for 30 min at 22-25 "C before separating platelet-bound from free antibody. The results are expressed as percentage of residual antibody binding (on the ordimte) relative to a control mixture containing Hepes buffer instead of fragment. Upperpanel, experiment performed in the absence of modulators. Middle panel, experiment performed in the presence of 1 mg/ml ristocetin. Lower panel, experiment performed in the presence of 5 pg/ml botrocetin. Each point represents the mean and range of values observed in two separate experiments.

values were measured with the mutants having complete deletion of the sequence on the amino-terminal side of the loop. In contrast, the four molecules with internal deletions within the C y ~ ~ - C y s ~ ~ ' disulfide loop failed to interact with GP Ib, as judged by lack of inhibition of W-Ibl binding to platelets (Table I).

After reduction and alkylation of the 2 cysteine residues, all but three of the molecules with intact sequence between Cyssw and CysSg5 interacted with GP Ib as efficiently as reduced rvWFu1-733 or with lower ICso. Exception to this were the two molecules with complete deletion of the sequence on the carboxyl-terminal side of CYS~'~, which were significantly impaired in their ability to inhibit LJ-Ibl binding to platelets

even at the highest concentration tested, and the molecule with deletion of the sequence following residue 700, which required higher concentrations to inhibit antibody binding than the reduced and alkylated molecules extending to residue 704 or beyond (Table I). As in the case of the oxidized fragments, progressive deletions of sequence on the amino- terminal side of the C y ~ ' ~ - C y s ~ ~ ~ loop resulted in molecules with better inhibitory activity. The four fragments with internal deletions between Cysm and failed to interact with GP Ib even after reduction and alkylation of the 2 cysteine residues, thus behaving like the corresponding oxidized forms (Table I).

In the presence of the modulator botrocetin, the mutants with intact sequence between CysSW and Cys6" bound to GP Ib with affinity similar to that of rvWp41-733; even under these experimental conditions, however, the mutants with deletions within the loop failed to interact with GP Ib (Table 11). The latter results can be explained by the fact that none of the inactive molecules could form a complex with botrocetin, a necessary requisite for the activity of this modulator (Fig. 7).

To confirm that the assay based on inhibition of antibody W-Ibl binding to platelets reflected the interaction with the vWF-binding site on GP Ib, selected recombinant molecules were tested for their ability to inhibit the ristocetin-dependent binding of vWF to platelets. The results obtained were in agreement with those observed for the inhibition of LJ-Ibl binding in the absence of modulators, namely the recombinant molecules with an intact intrachain disulfide bond were less effective inhibitors that those with reduced and alkylated cysteine residues (results not shown). This is in agreement with the fact that, as demonstrated above (Fig. 5), ristocetin has no appreciable effect on the interaction between the E. coli-expressed fragments and GP Ib.

DISCUSSION

The results presented here provide information on two important aspects of the GP Ib-binding domain of vWF: 1) the structures involved in modulating its affinity for the platelet receptor, and 2) the structures necessary to support the interaction with the receptor. Crucial for these studies has been the expression in E. coli of the monomeric GP Ih- binding domain of vWF containing, like the native molecule, the intrachain disulfide bond between Cys509 and Cy@". This was obtained by mutating 5 of the 7 cysteine residues to glycine, a procedure that facilitated oxidation of the 509-695 intrachain bond while limiting the formation of random disulfide-linked aggregates. The Cys -+ Gly mutations, however, cannot be responsible for any of the results reported here since: 1) after reduction, the expressed fragment behaved like a similar molecule, characterized previously, having no sub- stitutions in the native sequence (9); and 2) one of the mutant molecules described here ( rvWF492-733), missing the segment of sequence with the substituted cysteines, exhibited functional attributes indistinguishable from those of the prototypic fragment rvWF441-733. The latter was used as the initial model for the GP Ib-binding site of vWF, because it corresponds to the domain extending between residues 449 and 728 derived from tryptic digestion of the native molecule ( 5 , 6).

The prototypic ~ v W F ~ ~ ~ - ~ ~ ~ fragment exhibited markedly lower affinity for GP Ib when the Cys509-Cys695 intrachain disulfide bond was oxidized than after reduction and alkylation of the 2 cysteine residues; this was apparent for the interaction occurring in the absence of modulators but not in the presence of botrocetin. The Cys509-Cys695 bond, therefore,

12190 Structure of the GP Ib-binding Domain of vWF

FIG. 6. Schematic representation of deletion mutant fragments of vWF. Each of the fragments is identified with the first and last residue in its sequence (referred to the sequence of the mature vWF subunit). For fragments containing limited deletions within the 509-695 disulfide loop, the extent of the deletion is indicated by the first and last residue of the missing sequence. Shaded areas within each fragment represent the two segments of sequence, residues 474- 488 and 694-708, previously identified as possible GP Ib interaction sites (7). The deleted sequences within the intrachain disulfide loop represent: A, a region indicated as part of a possible GP Ib interaction site (11); B, C, and D, three re- gions indicated as possible binding sites of the modulator botrocetin (21). The reactivity of each oxidized fragment with the monoclonal antibody NMC-4 is shown on the right (+ indicates strong reactivity; - indicates no detectable reactivity; +- indicates markedly reduced reactivity). Note that no reduced and alkylated fragment showed any reactivity with this antibody.

DESIGNATION SEQUENCE REACTIVITY WITH

NMC4 (Dot blot)

+

+

+

f

TABLE I Inhibition of LJ-261 binding to platelets by recornbinant v WF

fragments in the absence of modulators The values reported (in micromoles/liter) represent the range of

results observed in the indicated number of experiments. The Cys6@- CyssB6 intrachain disulfide bond was present only in the oxidized fragments.

Oxidized S-S Reduced and Designation bond alkylated

ICm n IC, n

,.,WF44-733 rvWp82-733 rvWFw733 ~vWF"" '~ ~vWFU"~* rvWFy4'- ~ v W F ~ ~ ~ rvWF608-688 rVWFlll-733 A512-526 rvWF441-733 ,.,wF441-733

rVWF441-733 A629-643

11540-560 A560-580

6lM 3-5

1.2-4 0.1-0.3 2.6-3.2 2.0-3.1

0.2-0.6 0.4-0.7

w 4 4 4 4 >4

1.0-1.3

6 3 4 2 2 3 3 2 1 1 1 1

P M

0.07-0.3 0.04-0.05 0.02-0.03 0.03-0.04

0.5 >4

0.04-0.09 >4 >4 >4 >4 >4

6 2 3 2

TABLE I1 Inhibition of LJ-lbl binding to platelets by recombinant v WF

fragments in the presence of botrocetin The values reported (in nanomoles/liter) represent the range of

results observed in the indicated number of experiments. The Cys609- intrachain disulfide bond was present only in the oxidized

fragments. Oxidized S-S Reduced and

Designation bond alkylated

ICs0 n rc, n nM nM

rvWF441-733 13-25 3 14-24 3 rVWF492-733 36 1 28 rvWF*a733

1 2.6 1 17 1

1 6 1 ~ V W F " I ~ - ~ ~ 8 rvWF441-696 33 1 20 1

1 30-36 2

6 1 31 1 rvWFsops* 160-200 2 51 1

rVWF441-704 6

rVWF441-733 A512-526 >4,000 1 14,000 1 rVWF441-733 A540-560 74,000 1 14,000 1 rvWF441-733 A560-580 >4,000 1 24,000 I rvWF441-733 A629-643 >4,000 1 14,000 1


g 100

E Y

3 0

C m .- 5

50 m

0 a

a

0

m - cn 9)

0

0 A512-526 0 A540-560

0.1 1 10 100 1,000

Concentration of Recombinant Fragment (nM) FIG. 7. Interaction of recombinant vWF fragments with bo-

trocetin. The assay was based on the ability of various concentrations (indicated on the abscissa) of different recombinant vWF fragments (as shown) to inhibit the binding of biotinylated botrocetin to native vWF insolubilized onto polystyrene microtiter plates. The binding of biotinylated botrocetin was measured using peroxidase- conjugated streptavidin and o-phenylenediamine. The residual binding measured in the presence of recombinant vWF fragments is indicated on the ordinate as percentage of that measured in control mixture containing Hepes buffer instead of fragment. Each point represents the mean and range of values obtained in three separate experiment.

may contribute to stabilize a conformation of native vWF in solution that prevents its interaction with circulating platelets but allows expression of full binding activity in the presence of appropriate modulators. The residual, modulator-inde- pendent interaction of oxidized rvWF441-733 with GP Ib, albeit with a markedly lower affinity than seen with the corresponding reduced fragment, is likely to be due to the lack of other structural elements important in regulating the binding of native vWF to GP Ib. These may include the presence of other intra- and interchain disulfide bonds (28) and/or the carbohydrate side chains located on both sides of the Cys509- Cysee6 disulfide loop (31). In fact, we have demonstrated previously that a mammalian-expressed homodimeric glyco- sylated fragment, with primary structure essentially identical to that of the E. coli-expressed fragment, has no demonstrable direct binding to GP Ib in the absence of modulators (32).

Progressive deletion of sequence on the amino-terminal side of the Cys609-Cys695 disulfide loop resulted in fragments with increasing affinity for GP Ib, whereas deletion of sequence on the carboxyl-terminal side of the loop had less effect in this regard. In interpreting these results it should be considered that all the recombinant fragments representing vWF sequence beginning at residue 441 contained 20 non- vWF amino acid residues preceding the vWF sequence and all those in which the vWF sequence terminated at residue 733 contained 22 non-vWF residues following the vWF sequence (see "Experimental Procedures"). It is unlikely that these additional residues caused any significant change in the functional properties of the prototypic recombinant fragment, since the two molecules rvWF492"33 (lacking the amino-terminal extension) and rvWFU"704 (lacking the carboxyl-terminal extension) had a behavior essentially similar to that of

(containing both extensions). It appears, therefore, that residues 441-507, and particularly 493-507, may limit exposure of the GP Ib-binding site in the isolated recom-

mWF441-733

binant fragment when the Cys509-Cys695 intrachain disulfide bond is oxidized in contrast, residues 697-733 appear not to have this modulating effect. Whether the same is also true in native vWF remains to be established; however, it is intriguing to speculate that the sequence 441-507 may contain a site involved in modulating the affinity for platelets. Indeed, it has been suggested that ristocetin may induce vWF binding to GP Ib by interacting with the sequence Glu-Pro-Gly-Gly, corresponding to residues 476-479 of the mature vWF subunit (33). These residues are part of a limited sequence (residues 474-488) that was tentatively identified in a previous publication as a component of the GP Ib-binding site (7); however, since E. coli-expressed vWF fragments cannot be modulated by ristocetin in the same way as native vWF, it appears that the action of this modulator is more complex than just interacting with specific linear sequences.

Regardless of whether the C y ~ ~ - C y s ~ ~ ~ bond is oxidized or reduced, integrity of the 185-residue loop between the 2 cysteine residues appears essential for expression of GP Ib- binding function. In fact, four distinct limited deletions of loop sequences thought to be involved in interacting with GP Ib (11) or botrocetin (21) resulted in fragments exhibiting undetectable interaction with GP Ib, whether measured directly or in the presence of botrocetin, a result that may be explained by inability of the mutant molecules to form a complex with the modulator (21). The lack of direct interaction, however, indicates that the loop sequence may contain elements necessary for vWF binding to GP Ib, in agreement with the role proposed for residues 514-542 ( l l ) , and/or may contribute to maintaining an appropriate conformation of contact sites possibly located both inside and outside of the loop. The latter conclusion is supported by the observation that, when the intrachain Cys509-Cys695 bond is oxidized, complete deletion of the sequences flanking the disulfide loop on both the amino- and carboxyl-terminal side has no dele- terious effect on the interaction with GP Ib. These results imply that none of the residues within the sequences 441-507 and 697-733 is strictly necessary for expression of the GP Ib- binding function of vWF. In contrast, when the Cys509-Cys695 loop is disrupted by reduction and alkylation of the cysteine residues, deletion of the segment of sequence comprising residues 697-733 results in fragments with markedly de- creased interaction with GP Ib. By comparison, a molecule extending to residue 700 on the carboxyl-terminal side of the loop was only partially dysfunctional and molecules extending to residue 704 were fully functional, suggesting that residues 697-704 of vWF may also contribute to optimal function of the GP Ib-binding site of vWF. Indeed, the segment of vWF sequence 694-708 has been indicated as part of the GP Ib- binding site on the basis of experiments performed both with synthetic peptides and monoclonal antibodies (7).

It is evident that the necessity of using exogenous modulators under experimental conditions in vitro complicates the understanding of the mechanisms that support the binding of vWF to GP Ib. For example, it is not known how the effects of the two modulators most commonly used, ristocetin and botrocetin, correlate to the events that are relevant for in vivo function. Nevertheless, it is clear that ristocetin and botrocetin act through distinct mechanisms and may induce vWF binding to GP Ib through contact sites not necessarily identical (8, 21), suggesting that the GP Ib-binding domain of vWF may have multiple functional conformations involving distinct residues in the interaction with platelets. The same may occur during normal hemostasis and/or pathological thrombosis, depending on the conditions that initiate platelet thrombus formation. This notion may help in the critical

12192 Structure of the GP Ib-binding Domain of vWF evaluation of the numerous studies, including the present one, devoted to the definition of the GP Ib-binding site of vWF.

Indeed, the results presented here limit the general validity of the previously proposed concept that the site of interaction for platelet GP Ib is formed by two discontinuous segments of sequence in the mature vWF subunit, residues 474-488 and 694-708, kept in spatial proximity by the intrachain disulfide bond between Cys509 and Cysags (7). These conclusions, supported by experimental evidence obtained with synthetic peptides and monoclonal antibodies, must be reconsidered in view of the data obtained with recombinant fragments of vWF, leading to the following proposed reconstruction of the organization of the GP Ib-binding site of vWF. With regard to the structures that support the ligand-receptor interaction, the sequence delimited by Cys609 and appears to contain important functional elements that, in conjunction with the additional segment of sequence between residues 696 and 704, provide for optimal binding affinity. Residues in the sequence 474-488 may be involved in modulating the interaction, but in the context of the isolated domain, they are not essential for optimal affinity of binding. The proposed functional role of the sequence in the Cys509-Cys695 loop is in agreement with findings on the location of type IIB mutations in vWF, which result in increased affinity for GP Ib (32, 34, 35), as well as on the location of the discontinuous sequences within the loop that may be involved in interacting with the modulator botrocetin (21). The applicability of these conclusions to the function of intact native vWF remains to be established. It is certainly intriguing that a missense mutation of residue 561, as reported in a patient with the rare type B variant form of von Willebrand disease (36), leads to a loss of ristocetin- dependent vWf binding to platelets without affecting botrocetin-dependent binding. This finding confirms both the distinct effects of modulators and the role of the C y ~ ~ " $ - C y s ~ ~ ~ loop in regulating the vWF-GP Ib interaction; it is also compatible with the concept that there may be distinct structures in vWF capable of binding independently to GP Ib.

With regard to the potential use of isolated recombinant fragments of vWF as therapeutic anti-thrombotic agents, the minimal structure necessary to achieve functional inhibition of the vWF-binding site on GP Ib appears to be the cyclic 508-696 molecule containing the disulfide bond. Better inhibitory activity can be obtained with molecules having additional sequence on the carboxyl-terminal side of Cysasa; in the latter case, however, reduction of the intrachain disulfide bond results in more active compounds. The latter modification may, indeed, overcome one of the structural features regulating the affinity of vWF for GP Ib. The use- fulness of recombinant vWF fragments in preventing platelet thrombus formation is currently being evaluated (37,38). The present results provide information that may lead to the design of more active compounds.

Acknowledgments-We thank Dr. Akira Yoshioka and Dr. Midori Shima, Nara Medical College, Nara, Japan for allowing us to use the monoclonal antibody NMC-4; James Roberts for preparing other monoclonal antibodies; and Eileen Bristow and Lynn LaCivita for excellent secretarial assistance.

REFERENCES

2. Fuster, W., Bowie, E. J., Lewis, J. C., Fass, D. N., Owen, C. A., Jr., and 1. Ruggeri, Z. M., and Zimmerman, T. S. (1987) Blood 70,895-904

Brown, A. L. (1978) J. Clin. Invest. 61.722-730 3. Nichols, T. C., Bellinger, D. A., Johnson, T. A., Lamb, M. A., and Griggs,

T. R. (1986) Circ. Res. 59,15-26 4. Bellinger, D. A., Nichols, T. C., Read, M. S., Reddick, R. L., Lamb, M. A.,

Brinkhous, K. M., Evatt, B. L., and Griggs, T. R. (1987) Proc. Natl. Acad.

5. Fujimura, Y., Titani, K., Holland, L. Z., Russell, S. R., Roberts, J. R., Elder, Sei. U. S. A. 84,8100-8104

J. H., Ruggeri, Z. M., and Zimmerman, T. S. (1986) J. Biol. Chem. 2 6 1 ,

6. Fujimura, Y., Titani, K., Holland, L. Z., Roberts, J. R., Kostel, P., Ruggeri, Z. M., and Zimmerman, T. S. (1987) J. Biol. Chem. 262,1734-1739

7. Mohri, H., Fujimura, Y., Shima, M., Yoshioka, A., Houghten, R. A., Ruggeri, Z. M., and Zimmerman, T. S. (1988) J. Biol. Chem. 263,17901-17904

8. Girma, J. P., Takahashi, Y., Yoshioka, A., Diaz, J., and Meyer, D. (1990) Thromb. Haemostasis 64,326-332

9. Sugimoto, M., Ricca, G., Hrinda, M. E., Schreiber, A. B., Searfoss, G. H., Bottini, E., and Ru geri, Z. M. (1991) Biochemistry 30,5202-5209

10. Fujimura, Y., Hollanrf L. Z., Ruggeri, Z. M., and Zimmerman, T. S. (1987) Blood 70,985-988

11. Berndt. M. C.. Ward. C. M.. Booth. W. J.. Castaldi. P. A.. Mazurov. A. V..

381-385

and Andrews, R. K. (1992) Biochemist& 31,11144-11151 12. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G.

T., Mullis. K. B., and Erlich, H. A. (1988) Science 239,487-491 13. Sanger, F., Nicklen, S., and Coulson,.A. R. (1977) Proc.. Natl. Acad. Sei.

U. S. A. 74,5463-5467 14. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloni

Laboratory Man&, Cold Spring Harbor Laboratory, Cold Spring He%; NV

15.

16.

17. 18. 19. 20.

Kiikel, T. A., Roberts, J. D., and Zakour, R. A. (1987) Methods Enzymol.

Rosenberg, A. H., Lade, B. N., Chui, D. S., Lin, S. W., Dunn, J. J., and

Studier, F. W., and Moffatt, B. A. (1986) J. Mol. Biol. 189,113-130 Laemmli, U. K. (1970) Nature 227,680-685 Burnette, W. N. (1981) Anal. Biochem. 112,195-203 Johnson. D. A.. Gautsch. J . W.. SDortsman. J. R.. and Elder. J. H. (1984)

154,367-383

Studier, F. W. (1987) Gene ( A m s t ) 56,125-135

Gene Ad. Tech. 1,3:8

BWL Chem. 266,18172-18178

(1986) J. Biol. Chem. 261,12579-12585

429-436

I . , , . , 21. Sugimoto, M., Mohri, H., McClintock, R. A., and Ruggeri, Z. M. (1991) J.

22. Handa, M., Titani, K., Holland, L. Z., Roberts, J. R., and Ruggeri, Z. M. 23. Ey, P. L., Prowse, S. J., and Jenkm, C. R. (1978) Immunochemistry 15 ,

24. Ruggeri, Z. M., De Marco, L., Gatti, L., Bader, R., and Montgomery, R. R.

25. De Marco, L., Girolami, A., Russell, S., and Ruggeri, Z. M. (1985) J. Clin.

26. Fraker, P. J., and Speck, J. C. (1978) Biochem. Biophys. Res. Commun. 80,

." ."

(1983) J. Clin. Invest. 72.1-12

Invest. 75,1198-1203

849-857

281 -29R 27. Walsh, P. N., Mills, D. C. B., and White, J. G. (1977) Br. J. Haematol. 36 ,

28. Azuma, H., Dent, J. A,, Sugimoto, M., Ruggeri, Z. M., and Ware, J. (1991)

29. Shelton-Inloes, B. B., Titani, K., and Sadler, J. E. (1986) Biochemistry 2 6 ,

"_ "_ J. Biol. Chem. 266,12342-12347

30.

31.

32.

33.

34.

35.

36.

37.

Mohri, H., Yoshioka, A,, Zimmerman, T. S., and Ruggeri, Z. M. (1989) J. 3164-3171

Titani, K., Kumar, S., Takio, K., Ericsson, L. H., Wade, R. D., Ashida, K., Biol. Chem. 264,17361-17367

Walsh, K. A., Chopek, M. W., Sadler, J. E., and Fujikawa, K. (1986)

Ware, J., Dent, J. A., Azuma, H., Sugimoto, M., Kyrle, P. A., Yoshioka, A., Biochemistry 26,3171-3184

and Ruggeri, Z. M. (1991) Proc. Natl. Acad. Sci. U. S. A . 8 8 , 2946-2950 Scott. J. P.. Montgomery. R. R., and Retzinger, G. S. (1991) J. Bml. Chem.

266,8149-8156 '

- .

Randi. A. M.. Rabinowitz. I.. Mancuso. D. J.. Mannucci, P. M., and Sadler. J. E: (1991) J. Clin. Invest. 87,1220-1226

Cooney, K. A., Nichols, W. C., Bruck, M. E., Bahou, W. F., Sha iro, A. D., Bowie, E. J. W., Gralnick, H. R., and Ginsburg, D. (1991) J. &in. Inuest.

Rabinowitz, I., Tuley, E. A., Mancuso, D. J., Randi, A. M., Firkin, B. G., 87, 1227-1233

Howard, M. A., and Sadler, J. E. (1992) Proc. Natl. Acad. Sei. U. S. A. 89,9846-9849

Gralnick, H. R., Williams, S., McKeown, L., Kramer, W., Krutzsch, H:, Gorecki. M.. Pinet. A.. and Garfinkel. L. I. (1992) Proc. Natl. Acad. Scr. U. S. A.'89,'7880-18&k

38. Alevriadou, R. B., Moake J. L., Turner, N. A,, Ru geri, Z. M., Folie, B. J., Philli s, M. D., Schreiber, A. B., Hrinda, M. E., and McIntire, L. V. (1993PBIood 81,1263-1276

39. Marti, T., Roesselet, S., Titani, K., and Walsh, K. A. (1987) Biochemistry

40. Andrews, R. K., Gorman, J. J., Booth, W. J., Corino, G. L., Castaldi, P. A.,

41. Azuma, H., Hayashi, T., Dent, J. A., Ruggeri, Z. M., and Ware, J. (1993) J.

26,8099-8109

and Berndt, M. C. (1989) Biochemistry 28,8326-8336

Biol. Chem. 268,2821-2827

Analysis of Structure-Function Relationships in the Platelet ...

Documents

Transcript of Analysis of Structure-Function Relationships in the Platelet ...