Dimeric Ristocetin Flocculates Proteins, Binds to ... - Semantic Scholar

VOl. 266, No. 13, Issue of ' May 5,pp. 8149-8155,1991 Printed in lJ S.A .

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Dimeric Ristocetin Flocculates Proteins, Bindsto Platelets, and Mediates von Willebrand Factor-dependent Agglutination of Platelets* (Received for publication, November 20, 1990)

J. Paul ScottSgII, Robert R. MontgomeryS$II, and Gregory S. Retzinger 11 ** From the Departments of $Pediatrics and IIPathology, the Medical College of Wisconsin, Milwaukee, Wisconsin 53226 and the §Blood Center of Southeastern Wisconsin, Milwaukee, Wisconsin 53233

Ristocetin in aqueous solution dimerizes with an M, i.e. equilibrium dissociation constant of 5.0 X -1.1 mgml" (Waltho,J. P., and Williams, D. H. (1989) J. Am. Chem. SOC.11 1,2475-2480). At concentrations of about 1.0 mg ml" ristocetin flocculates many proteins, lyses plateletsand, in the presence of von Willebrand factor, agglutinates both fresh and formalinfixed platelets. Because ristocetin exists asboth monomericand dimeric species, we sought to determine which of these forms flocculates proteins and agglutinates platelets. We found that: 1) the initial rate of flocculation of certain proteins, 2) the initial rate of agglutination of formalin-fixed platelets, and 3) the binding of ristocetin to formalin-fixed platelets are higher order solely with respect to the concentration of ristocetin dimers. As to the operative mechanism, it appears that bifunctional dimers cross-link proteins that possess multiple copies of a common recognition site. Preliminary evidence indicates that a recognition site is a &turn of the form X-P-G-X'.

Although high dose ristocetin is only twice the concentration of the lower dose, the addition of high, but not low, dose ristocetin to platelet-rich plasma from normal individuals agglutinates the platelets. High, but not low, dose ristocetin also flocculates certain proteins (4,7) and lyses platelets (3). Recently, this same dependence on ristocetin concentration was observed for ristocetin-dependent bindingof glycocalicin (8).We wondered whether these rather to surface-bound vWF narrow dose-response relations might be due to concentration-dependent multimerizationof the glycopeptide. Supporting this notion, ristocetinA in aqueous media dimerizes with an equilibrium dissociation constant, &, of 5.0 X io-4 M, i.e. -1.1 mg ml" (9). The similarity of the Kd to the concentration of ristocetin that flocculates proteins, lyses platelets and, in the presence of vWF, agglutinates plateletsis striking andled us to hypothesize that dimers of ristocetin operate during ristocetin-dependent phenomenaassociated withproteins and cells pertinent to hemostasis. Herein we demonstrate that ristocetin dimers do indeed mediate biologic activities of the glycopeptide, and we propose a mechanism for these processes.

In the late 1950s, ristocetin, a glycopeptide synthesized by the actinomycete Nocardia lurida, was shown to have potent antibacterial activity (1). Shortly thereafter, ristocetin was introducedasanantibiotic for use in vivo. Drug-induced thrombocytopenia, a frequent complication of ristocetin therapy, prompted discontinuation of the use of the drug as a therapeutic agent (2, 3). Investigations addressing the mechanism(s) by which ristocetin causes thrombocytopenia led to the discoveries that the glycopeptide flocculates fibrinogen, lyses platelets and, in the presence of von Willebrand factor (vWF),' agglutinates both formalin-fixed and fresh platelets (3, 4). Indeed, a consequence of the latter discovery was the development of the ristocetin-induced platelet agglutination assay, a screening test for von Willebrand's disease (4-6). As performed routinely in the clinical hemostasis laboratory, this assayincludes ''low,'' i.e. -0.5 mgml" (-2.2 X M), and M), doses of ristocetin. "high," i.e. -1.0 mg ml" (-4.4 X

MATERIALSANDMETHODS

Reagents-Ristocetin sulfate of specified purity >90% ristocetin A was a gift from H. Lundbeck, Copenhagen, Denmark. Human fibrinogen,gradeL, was fromKabi AB, Stockholm, Sweden, and was dialyzedexhaustively against 0.02 M Tris-HC1, pH 7.40, aliquoted and stored at -20 ' until use. The fibrinogen concentration of these stock solutions was determined using the molar absorptivity of the protein at 280 nm, 5.12 X lo5 M" cm" (10). Bovine fibronectin was from Calbiochem. Poly L-Pro (M, = 6400) and poly (L-Pro-Gly-LPro) (M,= 5300) were from Sigma. Bacitracin, a cyclic undecapeptide containing both D- and L-amino acids,was from Fluka Chemical, Ronkonkoma, NY. Citrated plasma was prepared from the blood of healthy individuals. Plasma from an individual with congenital afibrinogenemia (11)was obtained using plasmapheresis. The fibrinogen concentration of citrated plasma was determined according to the method of Clauss (12). vWF was purified from Humate-P, Behringwerke AG,Marburg/Lahn, FederalRepublic of Germany, asdescribed elsewhere (13). AVWl, amonoclonal antibody that bindsto all multimers of vWF, was produced in our laboratory as already described (14, 15). Isotopic labeling of proteins in 0.02 M Tris-HC1, pH 7.40, was performed according to an establishedprocedure (16) using * Supported in partby the American Heart Association, Wisconsin NalZ5Ifrom Amersham Corp. and IODO-GEN from Pierce Chemical affiliate, Grants 88-GA-27 and89-GA-30, by a Research Foundation Co. Following iodination, radiolabeled proteins were isolated using of the Blood Center of Southeastern Wisconsin grant, by the National gel permeation chromatography (14). A defined mixture of bovine thyroglobulin, bovine y-globulin, chicken ovalbumin, horse myogloInstitutes of Health GrantsR01-HL33721-06 and POl-H244612-01, and by a Lucille P. Markey Charitable Trust Foundation grant. The bin, and vitamin Blz was obtained from Bio-Rad. Organic solvents costs of publication of thisarticle were defrayed inpart by the were of a grade suitable for high performance liquid chromatography payment of pagecharges. Thisarticlemusttherefore be hereby (HPLC). Water was deionized and then distilled using an all-glass marked "aduertisement" in accordance with 18 U.S.C. Section 1734 apparatus. All other chemicals andreagents were of the highest solely to indicate this fact. quality available commercially. 1TOwhom requests for reprints should be addressed. Purification of Ristocetin A-Ristocetin A was purified from the ** Lucille P. Markey Scholar. commercial mixture of ristocetins A(Fig. 1) andB.Preliminary The abbreviations used are: vWF, von Willebrand factor; GPIb, assessment using gel permeation chromatography confirmed that the platelet glycoprotein Ib; HPLC, high performance liquid chromatog- commercial preparation is a mixture of at least two materials, Fig. raphy. 2 A . The ratioof peak heights remains constantover a broad range of

8149

Ristocetin Dimeric

8150

and Hemostasis

OH

10

15

TIME, min

CHZOH

& J O I l

TIME, min

011

FIG. 1. Structure of ristocetin A. Numbering of aromatic rings after Ref. 9. applied concentrations, 0.05-100 mg ml-I, suggesting that the duality of peaks is not a consequence of self-association. By reducing the 30 flow rate of the chromatographic procedure to 0.25 ml min" base20 line resolution of the two peakscan be achieved. Isolationand subsequent reapplication of the material corresponding to each peak 10 yields a material that migrates as a single, sharp peak. The purity of 4 the commercial ristocetin preparation was also assessed using reverse 10 15 20 phase HPLC. By this method there are more than two components TIME, min in the parent mixture, Fig. 2B. Material corresponding to the leading FIG. 2. Purification of ristocetin A. A, gel permeation HPLC peak of the reverse phase method comigrates with leading the material of the gel permeation method.Likewise, the materialof the dominant of commercial ristocetin, 0.45 mg, using an analyticalmolecular sizing column, TSK 3000SW (Hewlett Packard, Avondale, PA), of dimenpeak isolated by reverse phase HPLC comigrates as a single peak with the dominant peakof material fractionated by gel permeation, sions 300 X 7.5 mm. The flow rate of the eluent, 0.05 M Tris-HC1, Fig. 2C. This material, collected within the time limits shown inFig. pH 7.40, containing 0.15 M NaCI, was 1.0 ml min". Absorbance at 280 nm was used to detect materials eluting from the column. B, 2B, fraction B, accounts for 80.4% by mass of the starting material and wasdesignated chromatographicallypureristocetin A. The reverse phase HPLC of commercial ristocetin, 2.5 mg, using a MaIL) of pooled material eluting prior to ristocetin A, i.e. Fig. 2B, fraction A, crosphere 300A c-4 column(AlltechAssociates,Deerfield, dimensions 250 X 10 mm. Stock eluents included trifluoroacetic acid contains presumably ristocetin B and accounts for about 15-18% of the mass of the starting material.Lyophilized crystals corresponding in water, 0.1%, v/v; and trifluoroacetic acid in acetonitrile, 0.1%, to chromatographic fractionsA and B are white, while crystals cor- v/v. The initial and final eluent compositions of the 10-min linear responding to chromatographic fraction C retain the amber color of gradient were, respectively, acetonitrile:water, 1:19, v/v; and acetothe starting material. Unless otherwise specified, chromatographically nitrile:water, 1:9, v/v. The flow rate was 2.5 ml min". Absorbance a t 280 nm was used to detect materials eluting from the column. The pure ristocetin A was used for all of our experiments. Spectrophotometric Analyses-For some experiments, absorbance time limitsof collection of three fractions designatedA, B and C were was used to quantitate ristocetin. There is appreciable an and discrete as shown. C, gel permeation HPLC of ristocetin A, 0.02 mg. The maximum in the absorbance spectrum of ristocetin A at about 280 conditions were as described for (A). nm (Amax = 280.3 nm). Using absorbance a t 280 nm and a molecular volume of solution after sedimentationof the flocculate. Quantitation weight of 2260 for the disulfated form of ristocetin A, the molar absorptivity of the chromatographically pureglycopeptide was deter- of fibrinogen and vWF was facilitated by the addition of a trace amount of the corresponding'251-labeledprotein to medium contain= 40.6 +- 0.2. For comparison, mined to be 9175 & 34 "' cm", or the commercial preparation of ristocetin was determined to have an ing unlabeled protein. Ristocetin was quantitated using a standard of at 280 nm of 46.6 2 0.1. This latter value is in good agreement known concentration and reverse phase HPLC as described.lZ5IAVWl was used to quantitate vWF flocculated fromplasma according with that determined by others, 49.0 & 1.0 (17). to an established technique (14). In brief, this method requires the Flocculation of Proteins-As will be shown, the use of the term "flocculation" is appropriate (18) since ristocetin forms cross-links addition of a trace amount of the radiolabeled antibody to reaction between peptides (9).A typical experiment was performed as follows. medium prior to the addition of ristocetin. In the absence of vWf, '251-AVW1does not sediment even after exposure to a high concenTo a volume of solution containing protein(s) was added either an equivalent or a lesser volume of a solution containing 0.05 M Tris- tration of ristocetin, 4.1 mg ml-'. In the presence of vWF, however, HCl, pH 7.40, 0.15 M NaCI, and ristocetin. This mixture was then the sedimentation of lZ5I-AVW1parallels exactly the loss of vWF from solution as determinedusing an immunoelectrophoretic method incubated a t room temperature for a period, the flocculate pelleted (19). Therefore, in order to quantitate rapidly the flocculation of vWF by centrifugation a t 1500 X g for 2 min, and the quantity of protein and/or ristocetin in the flocculate determined. We used as the meas- from plasma, the sedimentation of the bound, radiolabeled antibody ure of analyte in the flocculate the difference between analyte in a was monitored. fixed volume of solution prior toflocculation and analyte in the same The effect of several synthetic peptides on flocculation of fibrino-

8151

Dimeric Ristocetinand Hemostasis gen was also assessed. Forthese experiments,a test peptide was added to the reaction medium prior to the addition of ristocetin. Binding Studies-The binding of commercial ristocetin to twicewashed formalin-fixed platelets was assessed usinga modified version of a method for studying the binding of peptides to phospholipidcoated beads (20). Formalin-fixedplatelets were prepared as described elsewhere (21). Modifications of the binding assay included:1)use of formalin-fixed platelets of stock concentration 5.93 X 10' platelets ml" as substrate; 2) 10-min incubationof peptide, i.e. ristocetin, with platelets; 3) isolation of platelets using centrifugationat 1500 X g for 10 min; and 4) quantitation of unbound peptide using absorbance at 280 nm. As an added control, formalin-fixed plateletswere incubated in buffer in the absence of ristocetin, and the absorbance at 280 nm of this reaction medium, once freed of platelets, was subtracted from each test preparation. Agglutination of Formalin-fined Platelets-Agglutination was monitored using a platelet aggregometer in which 480 p1 of 0.05 M TrisHC1, pH 7.40, containing 0.15 M NaC1,0.5 unit of purified vWF, and 1.25 X 10' platelets was warmed to 37 "C and stirred continuously at 1200 rpm. Change in relative turbidity as a function of time after addition of 20 pl of buffered solution containing ristocetinwas monitored using a linear chart recorder, and the maximal slope of the turbidimetric tracingwas used as the measure of the maximal rate of agglutination. Tensiometry-The surface tension of buffered solutions containing various concentrations of ristocetin A weremonitoredusing a du Nouyring attached to a recordingelectronicmicrobalance(Cahn Instruments, Cerritos, CA).Allmeasurements were performed at

Ristocetin x 10

4

,M

FIG. 3. Ristocetin-induced flocculation of lz6I-fibrinogen (0)and lZ61-vWF(0)Solid . lines represent theoretical fits of the data to an equationdescribed in the text.

the percentage of radioactivity associated with the flocculate after 5.0 min and kfibis a constant of proportionality allows for independent determination of &. As shown in Fig. 3, the fit of the data to the theoretical equationis excellent. Using the paired values of R, and V we calculate kfib= 5.5 x lo9 f 1.7 X lo9 % M - ~and Kd = 5.6 X zk 2.2 X M. Within 24.2 "C. Analysis of Data-Time- or dose-dependentdata were paired with experimental error, the value of Kd we calculate is identical the corresponding time or dose and fit to equations described in the to that determined by others using a different method (9). text. The best values forthe parameters of these equations were then We conclude that the rate-limiting stepin the flocculation of determinedusing the paired data and a nonlinearleast squares fibrinogen by ristocetin involves two dimers of the glycopepregression method (22). Sequence Data-The sequences of proteins relevant to this study tide. Because ristocetindimers formbridgesbetween certain were obtained using computer-facilitated search of the Protein Sequence Database, release 20, of the Protein Identification Resource, tripeptides (9), itwas reasonable to assume thatsuch dimers the National Biomedical Research Foundation, Washington,DC. function similarly to flocculate plasma proteins. Since flocculation would necessarily involve the bindingof dimers to at RESULTS least two binding sites onsusceptible proteins, we determined Flocculation of Proteins-Ristocetin flocculates several pro- the stoichiometryof the ristocetin-fibrinogen interaction. For M, teins from aqueous media (4, 23). Using 1251-labeledproteins thispurpose, we incubatedpureristocetin A,5.91 x withlZ5I-fibrinogen, 6.43 X M, in 0.02 M Tris-HC1, pH andourstandardassayconditions, we observed that the amount of protein flocculated by a fixed concentration of 7.40, for 4 h. Following sedimentation of the flocculate, we ristocetin increases linearly with time during the first 5-10 determined that the molar ratio of ristocetin removed from min of incubation. This linearity is maintained even up to the solution to fibrinogen removed from solution was 32:l. Thus removal of as much as 70-80%of the protein in solution. at equilibrium 32 half-dimers of ristocetin bind on average to Therefore,therate of flocculation isnotfirstorderwith each fibrinogen. respect to the concentration of protein in solution, rather with We probed next interactions between ristocetin and vWF respect to the concentration of a species that remains constant since, at the present time, the clinical utility of ristocetin during nearly the entire time course of the reaction. At this relates to itsuse in vitro in the diagnosis of von Willebrand's time we can only speculate that this species is a nucleating disease, a condition characterizedby quantitative and/or qualof vWF used surface that changes little after the initial nucleation event. itative abnormalitiesof vWF. The concentration In order to determine whether monomeric or dimeric risto- for this studywas 75 pg ml-'. The rateof flocculation of vWF, cetin flocculates fibrinogen, we monitored the protein floc- like that of fibrinogen, appeared linear with respect to the culated after 5.0 min, i.e. the initial rate of flocculation, as a square of the concentration of ristocetin dimers. For this function of analytic, monomeric and dimeric ristocetin con- reason, these data, too, were analyzed in terms of the equacentrations. For preliminary analysis, the concentrations of tions given above. Thedata were not sufficiently robust, monomeric and dimeric ristocetin were calculated based on however, to allow statistically relevant determinationof both the Kd reported by others, 5.0 X M (9). The concentration the constant of proportionality, &wF, and Kd. Therefore, in of fibrinogen used for this study was 2.0 mg ml" (5.9 X order to facilitate convergence of the nonlinear least squares M). As shown in Fig. 3, the dose-responseprofile is markedly regression method the value of K d was fixed at 5.6 X curved withrespect to the analytic concentration of ristocetin. based on the analysisof the fibrinogen data. Again, the fit of Plots of initial rate uersus the concentrations of monomeric the data to the theoretical relation is very good (Fig. 3). Using and dimeric ristocetin are also curved (data not shown). A the paired values of Vn,, and R, we calculate from these data plot of the initial rate versus the square of the dimer concen- &wF = 1.25 X lo9 f 0.04 X 109 % "2. tration is linear ( r = 0.99), however, indicating that flocculaBecause ristocetin interactswith vWF in the plasma milieu, tion is second orderwithrespecttotheconcentration of we assessed flocculation of vWF in citrated plasma. Using dimers. At equilibrium, the molarity of ristocetin dimers, D, '*'I-AVWl we monitored flocculation of vWF from normal in terms of the analytic molarityof ristocetin, R,, is given by plasma of vWF concentration 1.0 unit ml-' (Fig. 4). Clearly, (4R, + Kd - ((4R, + Kd)* - 16R,2)1'2)/8. Substituting this the dose-response profile in plasma is different from that in expression into the linear equationVn,, = k , O 2 where Vn,, is buffer. Whereas flocculation in buffer is obvious as the con-

Ristocetin Dimeric

8152

60.0 50.0 40.0

30.0 20.0 10.0

0.0 0.0

Ristocetin x l o 4 , M

FIG. 4. Flocculation of vWF from normal humanplasma (0) and from the plasma of an afibrinogenemic individual (0). The flocculation of vWF after 5.0-min incubationwith ristocetin was monitored as describedin the text. Solid lines represent least squares fits of the data.

and Hemostasis ovalbumin and myoglobin increased by about IO%, that of vitamin BI2was unaffected, and 3)while there was no perceptible change in the apparent molecular weight of either yglobulin or vitamin BIZ, the apparent molecular weight of myoglobin increased by about 4,000, that of ovalbumin by about 10,000. On the basisof calibration standards, themolar ratio of ristocetin no longer free in solution to thyroglobulin removed from solution was about 150:l. From these data it appears that ristocetin binds to all the proteins of the test mixture but flocculates only thyroglobulin. In addition to their rather high molecularweight, fibrinogen (MI -340,000), vWF (Mr >680,000), and thyroglobulin (Mr -670,000) are homomultimers. If multiple homologous subunitsare sufficient for flocculation, thenothersuchproteins-provided they bind the ristocetindimer-should flocculate in the presenceof ristocetin. We tested whether ristocetin flocculates fibronectin, a protein composed of multiple homologous subunits. Indeed, fibronectin, 0.6 mg ml-' (1.36 X 10+ M), does form a microdispersion in the presence of ristocetin and, by visual inspection, the dependency in the ristocetin concentration appeared to be the same as that for fibrinogen and vWF. Binding of Ristocetin to Formalin-fixed Platelets-In the presence of vWF, concentrations of ristocetin in excess of about 1.0 mg ml-' cause the rapid agglutination of formalinfixed platelets and this is the basis of the most widely used assay for vWF activity (21). For this reason, we investigated the bindingof ristocetin toformalin-fixed platelets using the commercial preparation of the glycopeptide. As shown in Fig. 5, binding is not linear with respect to the analytic concentration of ristocetin; rather, binding appears to bea cooperative process. The best fit of these data to a typical hyperbolic adsorption isotherm was achieved when the amount of ristocetin boundwas plotted, again, asa function of the square of the dimer concentration. Thus, we analyzed these data in the context of dimerization. At equilibrium, the concentration of occupied binding sites, C , expressed in terms of the number of half-dimers of ristocetin per platelet, isgiven by P,,D"/(K, + D"), where Po is the number of half-dimers bound to each platelet at saturation, K, is the dissociation constant of this cooperative binding in M', and n is the minimum order in the cooperative binding sites, i.e. the Hill parameter(24). Substituting theequilibrium expression of the dimer concentration for D in this equationallows calculation of Po,K,, Kd, and n. Due to the limited and dispersed data of this experiment,we fixed the values of Kd and n at 5.6 X M and 2, respectively, based on our previous results. We then calculate Po = 6.43 x IO7k 0.5 X IO7half-dimers per platelet andK, = 1.37 X lo-'

centration of ristocetin approaches Kd, no such flocculation occurs in plasma until the concentration of ristocetin is about twice Kd. Once reaching this concentration there appears to be a lineardependence of flocculation ontheamount of ristocetin added. An obvious explanation for this discontinuity would be the presence in plasma of molecules that bind ristocetindimers moreavidly than does vWF. Fibrinogen undoubtedly represents a virtual "sink" for ristocetin dimers since this protein is flocculated from normal plasma by ristocetin at concentrations near Kd (4).Indeed, given the fibrinogen concentration of the plasma used for this study(-7.7 X M) andthestoichiometry of theristocetin-fibrinogen interaction (32:1), thedimerconcentration (-3.2 X M) at the point of the interpolated discontinuity is near the concentration of binding sites on the availablefibrinogen (-2.5 x 10-4 MI. If fibrinogen alone sequesters ristocetin dimersfrom vWF, then plasmalacking fibrinogen should provide a medium akin to buffer. This is not the case, however, as shown in Fig. 4. Using afibrinogenemic plasma of vWF concentration approximately that of normal plasma, we found that flocculation of vWF in this medium is even less sensitive to ristocetin. We conclude that not only fibrinogen but also other components of plasmamust compete with vWF for ristocetin dimers. Furthermore, supplementing the afibrinogenemic plasma with fibrinogen to a concentration equivalent to that of the normal plasma yields a medium equivalent to the normal plasma. Thus, vWF mustcoflocculate with fibrinogen as the amount of the latter protein becomes depleted, and the formation of 1 this coflocculate is favored over flocculation of vWF alone. We wondered whether flocculation of proteins by ristocetin is a generic process or whether the glycopeptide flocculates preferentially a particular type of protein. To address thiswe t incubated an aqueous mixture of thyroglobulin, 5.0 mg ml-' x E 40.0 (7.46 X M); y-globulin, 5.0 mg ml-' (3.16 X lo-' M); .-01 ovalbumin, 5.0 mg ml" (1.13 X M); and myoglobin, 2.5 30.0 1 mg ml" (1.47 X M) with a high concentration of com.- n 20.0 mercial ristocetin, 10mg ml-I. Vitamin Blz,0.5 mg ml-' (3.37 w . l! x M ) , provided a convenientcontrol of nonproteinmaterial. A dense microdispersionformedimmediately upon addition of ristocetintothemixture.Comparison of the 0.0 0.0 1 .o 2.0 3.0 4.0 5.0 chromatograph of the residual material left in solution to that Ristocetln x l o 4 , M of the mixture treated similarly but without the addition of ristocetin revealed that after adding the glycopeptide: 1)the FIG. 5. Binding of ristocetin to formalin-fixed platelets. peak due to thyroglobulin was abolished, 2) the area of the Solid line represents the theoretical fit of the data to an equation peak due to y-globulin increased by about 1%,those due to given in the text. L

E . ;

8153

Dimeric Ristocetinand Hemostasis f 0.4 X

lo-'

M*.The fitof the data to this theoretical relation

,j

10.0

is reasonably good, consistent with the notion that ristocetin 8.0 dimers bind to platelets in cooperative fashion. Agglutination of Formalin-fixed Platelets-Having evidence that ristocetin dimers bind to both formalin-fixed platelets and vWFwe next determined whether dimers mediate agglutination of formalin-fixed platelets in the presence of vWF. For this purpose,we monitored the rateof agglutination asa function of ristocetin concentration, Fig. 6. Unlike flocculation of proteins and binding to platelets, the rateof platelet agglutination is not linear with respect to the square of the 0.047. 0 200 400 600 800 1000 1200 dimer concentration. It is, however, linear with respect to the cube of the dimer concentration ( r = 0.99). Thus, these data TIME, S where were analyzed according to the equation Vagg= kapgD3 FIG. 7. Surface pressure of ristocetin A as a function of Vagg isthe observed percentage of the maximal velocity time. The surface tension of a stock solution of ristocetin A, 4.42 X M ristocetin)and k,, (achieved inthepresence of 7.1 X M, was measured as described in the text. Solid line represents is a constant of proportionality. Using the data of Fig. 6 we the theoretical fit of the data to an equation given in the text. The calculate kagg= 2.12 X 1014f 0.07 X ioL4% M-3 and Kd = 5.6 parameters of this analysis are given in Table I. X k 0.2 X low4M. Ingratifyingagreement,the value of Kd derived from these data is identical to that derived from TABLEI the flocculation data. Weconclude that the rate-limiting step Parameters of surface tension studies of vWF-dependent agglutination of formalin-fixed platelets Parameters of surface tension studies were derived directly from requires three ristocetin dimers. the data (i.e. a,) or using an equation given in the text. The fit of the Surface Properties of Ristocetin A-It occurred to us that dataset of ristocetin A concentration 4.42 X M tothe equation dimerization of ristocetin, an amphiphilicmolecule, might be given in the text is shown by the solid line in Fig. 7. catalyzed by amphiphilic surfaces where the concentration of [Ristocetin A] X 10' r. T. k x 103 A~ x 10-10 the glycopeptide would be expected exceed to its concentration M dyn cm" dyn cm" s-' cm2 mole" in solution. Surface-catalyzed dimerization might then oper2.21 0.45 ND" ND ND ate during ristocetin-dependent processes. As a preliminary 4.42 0.83 6.67 & 0.25 4.0 -t 0.8 2.87 & 0.57 step toward elucidating the role of surfaces in the activities ND ND 1.72 ND 8.85 of ristocetin, we investigated the surface properties of the Not done. glycopeptide. As shown in Fig. 7 and Table I, the instantaneous surface pressure, A,, of solutions of ristocetin A that we prepared was low but measurable, consistent with formation 8.31 X lo7 dyn cm mol" OK"; T i s the temperature in "K,AT of a hydrated soluble monolayer (25). Doubling the bulk phase is the total area of the surface in cm', and A, is the area per concentration resulted ina doubling of A,, indicating that,for mole of ristocetin at the interface in cmzmol-'. We assumed that the rate of change of n is a firstorder process, the those concentrations tested, the surface was not saturated with the glycopeptide. With time, thesurface pressure of the integrated form of which would then be given by n,(l - e-kt) solutions increased tending to an equilibrium value, re.This where n, is the numberof moles at the surface at equilibrium, s-', and t i s time time-dependent increase in surface pressurelikely represents i.e. at A.; K is the first order rate constant in in s. At equilibrium ne = AT/[A, + (RT/*,)].Rearranging and thespontaneousformation of amore stable,dehydrated monolayer. Such a result is most consistent with self-associ- substituting for ne yields n = nAT(l - e-k')/(Aor, R T ) . ation of ristocetin in aqueous solution. Since A is a function Substituting this expression for n into the two-dimensional of the number of moles, n, of ristocetin at the interface, time- gas equation gives A = (1 - e-")/[(l/*,) + (A,e-kt/RT)]. r were analyzed interms of time- Finally, since A, # 0 the right hand side of this last equation dependentchangesin dependent changes inn. One canuse the two-dimensionalgas is increased by r0 yielding K = [(l- e - k f ) / ( ( l / r e ) (A,e-kr/ equation A = n R T / ( A r - nA,) to relate r to n, where R is RT))] + K,. The paired values of A and t were fit to this equation and used to derive parameters k , A, and re (Table 100.0 ....................................... I). The theoreticalfit of these data is showr? by the solid line 90.0 in Fig. 7. From A, we calculate a nominal cross-sectional area of-464 A' for the interfacial species of ristocetin A. The 2 10.0 0 nominal area per amin?acid residue of peptides a t interfaces 0 7 0 . 0 -a is between 10 and 40 A' (20, 26). Accordingly, ristocetin A, > 60.0 which has as its peptide "core" seven tyrosge-like subunits, - 50.0 should have a maximal area of about 280 A*. Indeed, using 40.0 Corey-Pauling-Koltun molecularmodels the peptide backX bone of monomeric ristocetin is approximatedobya rectangle I" 10.0 of dimensions of about 12 X 18 A, or 216 A'. Given that K 20.0 ristocetin A forms dimers in aqueous solution, the most par10.0 simonious explanation for these results is the formation of 0.0 stable dimers at the air-water interface. 0.0 1 .O 2.0 3.0 4.0 Investigation of the PeptideRecognition Site of the Ristocetin 4 Dimer: Inhibition of Ristocetin-dependent Flocculation of FiRistocetin x 10 , M brinogen by Poly (L-Pro-Gly-L-Pro)-That so many proteins FIG. 6. Dependence of the percentage of maximal velocity fibrinogen and of agglutination of formalin-fixed plateletson ristocetin con- bind ristocetin dimers, and proteins such as centration. Solid line represents the theoretical fit of the data toan vWF coflocculate in the presenceof ristocetin dimers indicate equation given in the text. that a dimer recognition sitemust be common tomany

+

+

!

8154

Ristocetin Dimeric

proteins. We used our data and that of others (9, 17, 23) to deduce this site: 1) The antibiotic activity of ristocetin, like that of vancomycin, is thought to be related tocomplexation with mucopeptide precursors of bacterial cell walls terminating in L-X-D-Ala-D-Ala (9, 17). Because mammalian proteins contain no D-amino acids and incorporationof L-amino acids into thesecond position of tripeptide cell wall analogs reduces their affinityfor ristocetin (17), we reasoned that glycine, which lacks a chiral center, might be a component of the recognition site. 2) The stoichiometryof ristocetin binding to fibrinogen is high, 32:l. It is unlikely that amino or carboxyl termini of peptide subunits composing proteins of the type flocculated by ristocetinare necessarilyrecognition sites since, using fibrinogen as anexample, thesewould be too few in number. It seems more likely that the recognition site is a recurring structuralmotif expressed along thesurface of suswhich glycine ceptible proteins. 3 ) The secondary structure in is most frequently found is the @-turn (27). As to a specific turn sequence, collagen, a rather monotonous protein rich in turns containing X-Pro-Gly-X' is flocculatedby ristocetin (23). 4) Within fibrinogen, there are 13 X-Pro-Gly-X' each in of the two a chains and one in eachof two 0 chains yielding a stoichiometry, 281, similar to that determined from the binding data.All of the X-Pro-Gly-X'of the a chain are found within the so-called a chain protruberance. 5) Proteins that we investigated that are flocculated by ristocetin dimers including thyroglobulin, fibronectin and vWF all have multiple copies of X-Pro-Gly-X' and, as will be elaborated for vWF under "Discussion," this sequence occurs in a region critical to the ristocetin-dependent bindingof the protein to itscellular receptor. 6) Proteins that we investigated that bind to but are not flocculated by ristocetin dimers contain only one copy of X-Pro-Gly-X' (horse myoglobin and chicken ovalbumin). Based on these facts and observations, we hypothesized that a recognition site of the ristocetin dimer is a @-turn of the form X-Pro-Gly-X'. A reasonable test of this hypothesis would be to ascertain whether peptides containing the sequence X-Pro-Gly-X inhibit ristocetin-dependent phenomena. Others have already shown that proteins-collagen in particular-containing multiple copies of this sequence inhibit ristocetin-dependent agglutination of platelets (23). We chose to study the effect of relevant peptides on theflocculation of fibrinogen. The commercial peptides that we assessed included the tetrapeptide Gly-L-Pro-Gly-Gly, poly (L-Pro-Gly-L-Pro) and, as controls, poly-L-Pro and bacitracin. As a preliminary screen of inhibition,peptides at rather high discreteconcentrations were M, just added to medium containing lZ5I-fibrinogen,5.9 X priortotheaddition of ristocetin, 4.9 X M. Radioactivity remaining in solution 5.0 min after the additionof ristocetin was then monitored. The results of thesestudies were as follows: in the absence of added test peptide, ristocetin flocculated 72% of the fibrinogen from solution; in the presence of 1.4 X M bacitracin, 72%; in 1.7 X M Gly-L-ProM poly-L-Pro, 52.5%; and in 3.1 Gly-Gly, 56.5%; in 3.1 X X M poly (L-Pro-Gly-L-Pro),13%. Thatthetetrapeptide Gly-L-Pro-Gly-Gly had little effect on flocculation was not unexpected since its short length precludes formation of a stable turn. The relatively marked inhibition due topoly (LPro-Gly-L-Pro), however, prompted further investigation of the inhibitory effect of this polypeptide on flocculation. We found that in the presence of poly (L-Pro-Gly-L-Pro) and a fixed concentration of ristocetin, the amount of fibrinogen flocculated from solutions of fibrinogen ranging from 2.9 X to 1.8 X M is constant and, thus, independent of the protein concentration. In contrast and asshown in Fig. 8, the

and Hemostasis 90.0 80.0

70.0

60.0 50.0

40.0 30.0 20.0 10.0

0.0 0.0

50.0

100.0

150.0

5

,M FIG. 8. Inhibition of ristocetin-dependent flocculation of '261-fibrinogenby poly (Pro-Gly-Pro). The concentration of fibrinogen was 5.9 X 1O"j M. The concentrations of ristocetin A were: 2.4 X 10" M ( A ) ,4.9 X M ( B ) ,and 9.7 X M (c). P o l y( P r o - G l y - P r o )

x 10

degree of inhibition for a fixed fibrinogen concentration, 5.9 X M, is sensitive to the ristocetin concentration indicating that inhibition is a consequence of the interaction of poly (LPro-Gly-L-Pro) with the ristocetin dimer. Taken together, these data and all other available evidence support the proposal that a p-turn of the form X-Pro-Gly-X' represents a dimer recognition site. DISCUSSION

Others have shownthat ristocetinA in aqueous mediaforms M (9). At aboutthesame dimers witha K d of 5.0 X concentration asK d ristocetin elicitsseveral phenomena pertinent to hemostasis. For this reason we sought to identify the activeform of the glycopeptide relevant tohemostasis. In summary, we found that ristocetin dimers flocculate certain proteins,bindto formalin-fixed platelets,andagglutinate formalin-fixed platelets in the presence of vWF. Eachof these processes will be addressed in turn. Our data indicate that while ristocetin dimers bind to many proteins, these dimers flocculate only certain proteins. Proteins that are composed of homologous subunits and that themselves polymerize or form colloids seem to be especially susceptible to flocculation. While the bindingof many copies of ristocetin dimers to proteins, e.g. fibrinogen, may occur, the rate-limiting stepof flocculation needs involve only a few dimers. We propose that the mechanism by which ristocetin dimers flocculatesusceptible proteins is analogous to that already described for the cross-linkingby ristocetin of tripeptide analogs of bacterial cell walls (9). Our data and that of others (23) indicate that a peptide recognition site of the dimer is a 0-turn of the form X-Pro-Gly-X. While as yet untested, 0-turns structurally related to X-Pro-Gly-X' may also be recognized by the ristocetin dimer and function similarly. In contradistinction to the self-limiting ternary complexesformedbetween ristocetindimersandmonovalent analogs of bacterial cell walls, propagation of complexes formed between dimers and proteins containing multiple copies of a dimer recognition site will not be self-limiting, and protein flocculation will be an inevitable consequence. Application of the flocculation phenomenon may prove a very simple and convenient meansby which to isolate susceptible proteins from protein mixtures, the purity of the flocculate dependingonthe relative affinitiesandcapacities of the proteins in the mixturefor dimeric ristocetin. At saturation, the equivalent of approximately 6.4 X lo7 half-dimers of ristocetin bind to aformalin-fixed platelet. This value is reasonably close to that, -1.60 X 10' ristocetin

Hemostasis and Ristocetin Dimeric

8155

REFERENCES “monomers” platelet-’, determined by othersusingfresh, Philip, 1. J. E., Schenck, J. R., and Hargie, M. P. (1957) in Antibiwashed platelets as substrate (28). From the tensiometry data otics Annual: 1956-1957 (Welch, H., and Marti-Ibanez, F., eds) the noominal cross-sectional area of dimeric ristocetin A is -464 A2. Using thisvalue we calculate from our binding data 2. pp. 699-705, Medical Encyclopedia, New York Gangarosa, E. J., Landerman, N. S., Rosch, P. J., and Herndon, a total area of about 1.5 X lolo A* attributable to dimers E. G. (1958) N. Engl. J. Med. 259, 156-161 bound to the platelet at saturation. Approximating, albeit 3. Gangarosa, E. J., Johnson, T. R., and Ramos, H. S. (1960) A. M. A. Arch. Znt. Med. 105, 107-113 roughly, the shapeof theplatelet by a smooth oblateellipsoid 4. Howard, M.A., and Firkin, B. G. (1971) Thromb. Diath. Haeof major axis 3.2 X lo4 A (29), one calculates a surface area morrh. 26, 362-369 for the “platelet” of about 3.2 x lo9 A*, or an area about20% 5. Weiss, H. J., Hoyer, L. W., Rickles, F. R., Varma, A., and Rogers, that of the bound dimers. It seems likely, therefore, that most J. (1973) J. Clin. Znuest. 52, 2708-2716 of the dimers bind to proteins ramifying from the plasma 6. Brinkhous, K. M., Graham, J. E., Cooper, H. A., Allain, J. P., and Wagner, R. H. (1975) Thromb. Res. 6, 267-272 membrane surface,a surface thatincludes an open canalicular 7. Ts’ao, C., Green, D., and Rossi, E. C. (1975) Blood 45, 621-629 system (30). Supporting this, others have demonstrated ris8. Vicente, V., Kostel, P. J., andRuggeri, Z. M. (1988)J. Biol. Chem. tocetin-dependent formation of electron dense accumulations 263,18473-18479 of proteinaceous material-presumed fibrinogen-within the 9. Waltho, J. P., and Williams, D. H. (1989) J. Am.Chem. SOC. canalicular system (7, 31). 111,2475-2480 10. Mihalyi, E. (1968) Biochemistry 7, 208-223 We have shown that in the presence of vWF ristocetin dimers mediate the agglutinationof formalin-fixed platelets. 11. Rodman, N. F., Mason, R. G., Painter, J. C., and Brinkhous, K. M. (1966) Lab. Znuest. 15,641-656 On the basis of our data, the simplest mechanism for risto- 12. Clauss, V. A. (1957) Acta Haematol. 1 7 , 237-246 cetin-induced agglutination would be that vWFcoflocculates 13. Montgomery, R. R., and Zimmerman, T. S. (1978)J. Clin. Invest. 61,1498-1507 on the surface of the platelet with the vWF receptor, the a 14. Schullek, J., Jordan, J., and Montgomery, R. R. (1984) J. Clin. chain of platelet glycoprotein Ib (GPIb). Subsequently, risZnuest. 73,421-428 tocetin-dependent “tethers” of vWF emanatingfrom GPIb at 15. Ruggeri, Z. M., De Marco, L., Gatti, L., Bader, R., and Montgomthe surface of theplatelet would complex with similarly ery, R. R. (1983) J. Clin. Znuest. 72, 1-12 disposed tethers on neighboring platelets. In support of such 16. Fraker, P. J., and Speck, J. C., Jr. (1978) Biochem. Biophys. Res. Commun. 80,849-857 a mechanism, the reported ristocetin-dependent bindingdomains of bothvWFandGPIbcontainX-Pro-Gly-X’ se- 17. Nieto, M., and Perkins, H. R. (1971) Biochem. J. 124,845-852 18. Seaman, G. V. H., and Goodwin, J. W. (1986) Am. Clin. Prod. quences (8, 32, 33). For vWF thesequence is G l ~ ~ ~ ~ - P r o - G l y - Reu. 5, 25-31 Gly; for the (Y chain of GPIb the sequence is Tyr259-Pro-Gly- 19. Laurell, C. B. (1971) Anal. Biochem. 15,45-52 Lys. Other copies of X-Pro-Gly-X’ occur in close proximity 20. Retzinger, G . S., Meredith, S. C., Lau, S. H., Kaiser, E. T., and KBzdy, F. J. (1985) Anal. Biochem. 150, 131-140 to the purported binding domains of these proteins, e.g. for K. M., and Read, M. S . (1978) Thromb. Res. 13,591vWF, Gly716-Pro-Gly-Leu,but the sequences that we identify 21. Brinkhous, 597 explicitly are those withinregions of the proteinsbelieved, at 22. Yamaoka, K., Tanigawara, Y., Nakagawa, T., and Uno, T. (1981) the present time, be to involved in the binding of ligand, vWF, J. Pharmacobio-Dyn. 4 , 879-885 23. Stibbe, J., and Kirby, E. P. (1976) Thromb. Res. 8, 151-165 to its receptor, GPIb. In closing, we have presented evidence that the biologic 24. Segel, I. H. (1976) Biochemical Calculations, 2nd Ed., pp. 309312, John Wiley & Sons, New York activities of ristocetin pertinent tocells and proteinsinvolved 25. Adamson, A. W. (1976) Physical Chemistry of Surfaces, 3rd Ed., in hemostasis are mediated by dimers of the glycopeptide. It pp. 68-92, John Wiley & Sons, New York 26. Gaines, G. L (1966)Insoluble Monolayersat Liquid-GasInterfaces, appears that such dimers bind to proteins at @-turns containp. 173, Interscience, New York ing thesequence X-Pro-Gly-X’ and, depending on valency the 27. Chou, P. Y., and Fasman, G. D. (1978) Anrzu. Reu. Biochem. 47, of proteins with respect to these turns, can form homogene251-276 ous/heterogeneoussupramolecular complexes by bridging 28. Senogles, S. E., and Nelsestuen, G. L. (1983) J. Biol. Chem. 258, 12327-12333 neighboring molecules. In the contextof this bridging mechanism, one can now design rational experiments that should 29. Lentner, C. (ed) (1984) Geigy Scientific Tables, Vol.3, p. 212, Ciba Publications Dept., Summit, NJ answer questions relevant to the therapeutic and diagnostic 30. White, J.G. (1987) in Hemostasis and Thrombosis. Basic Princiuses of ristocetin. We believe that the answers to these quesples and Clinical Practice, (Colman, R. W., Hirsh, J., Marder, tions will, in most instances, be related to the affinity, stoiV. J., and Salzman, E. W., eds) 2nd Ed., pp. 537-554, J. B. Lippincott, Philadelphia chiometry or accessibility of the ristocetin dimer for its binding sites. Perhaps relevant,a @-turnof the form Gly-Pro-Gly- 31. Escolar, G., Monteagudo, J., Villamor, N., Garrido, M., Castillo, R., and Ordinas A. (1988) Br. J. Haematol. 69, 379-386 X is nearly invariant within the principal neutralizing deter- 32. Titani, K., Takio, K, Handa, M., and Ruggeri, Z. M. (1987) Proc. minant of the human immunodeficiency virus (34-36). Thus, Natl. Acad. Sci. U. S. A. 84,5610-5614 mechanistic understandingof the selectivity of the ristocetin 33. Vicente, V., Houghten, R. A., and Ruggeri, Z. M. (1989) Blood 74, 170a dimer for @-turnsmay leadto a usefultherapy for the acquired 34. Rusche, J. R., Javaherian, K., McDanal, C., Petro, J., Lynn, D. immunodeficiency syndrome. Also appealing is the possibility L., Grimaila, R., Langlois, A., Gallo, R. C., Arthur, L. O., that studyof the interaction of ristocetin dimerswith @-turns Fischinger, P. J., Bolognesi, D. P., Putney,S. D., and Matthews, T. J. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 3198-3202 of thetypecommonto collagen will revealsome asyet unappreciated albeit fundamental relationship between this 35. LaRosa, G. J., Davide, J. P., Weinhold, K., Waterbury, J. A., Profy, A. T., Lewis, J. A., Langlois, A. J., Dreesman, G.R., protein and theadhesive proteins and their cellular receptors. Boswell, R. N., Shadduck, P., Holley, L. H., Karplus, M., Acknowledgments-The authors gratefully acknowledge Dr. F. J. KBzdy for insight and discussion, and E. Vokac, L. Zumwalt, M. C. McGinnis, and B. C. Cook for technical support. Thanks to Ruth Mary Retzinger for inspiration.

Bolognesi, D. P., Matthews, T. J., Emini, E. A,, and Putney, S. D. (1990) Science 249, 932-935 36. Nara, P. L., Smit, L., Dunlop, N., Hatch, W., Merges, M., Waters, D., Kelliher, J., Gallo, R. C., Fischinger, P. J., and Goudsmit, J. (1990) J. Virol. 64, 3779-3791