von Willebrand factor: Investigation of the structural requirements for interaction

THROMBOSIS RESEARCH 32; 545-556, 1983 0049-3848/83 $3.00 + .OO Printed in the USA. Copyright (c) 1983 Pergamon Press Ltd. All rights reserved.

TRE INTERACTIONBETWEEN COLLAGENSAND FACTOR VIII/VONWILLEBRAED FACTOR: INVESTIGATIONOF THE STRUCTURALREQUIREMENTSFOR INTERACTION.

L.F.Morton+, B.Griffin*, D.S.Pepper+ and M.J. Barnes+. @rangeways ResearchLaboratory,Warts'Causeway,Cambridge,CBl 4RN and ScottishNationalBlood TransfusionService,HeadquartersUnit Laboratory,Forrest Road, Edinburgh,Eli12QN.

(Received 19.7.1983; Accepted in revised form 5.10.1983 by Editor P.J. Gaffney)

ABSTRACT The blood proteinFactor VIII/van Willebrandfactor (FVIII/VWF)has been shown to bind to a variety of collagenpolymers including(i), the native-typefibres (of collagenstypes I and III), (ii), segmentlong-spacing(SLS)aggregates(of collagenstypes I, III, IV and V), (iii),the insolublepolymer obtainedby random cross-linkingof the type I monomer and (iv),the non-striatedfibril (of type I) produced by alcohol precipitation.Relativelylittle binding of FVIII/VWF to the amorphous,non-fibrillarform of collagen(type I) producedby salt precipitationfrom acid solutionwas observed. No significant bindingeither to elastinor to the insolublepolymer derived by random cross-linkingof bovine serum albumin was noted. The absorptionof FVIII/VWFto collagenswas affectedby ionic concentrationand FVIII/VWF was only totallybound at relativelylow ionic strength. Bindingof radiolabelledFVIII/VWFcould be largely inhibitedby an excess of the unlabelledprotein. The interactionof FVIII/VWFwith collagenfibres was inhibitedin a concentrationdependentmanner by monomeric collagenwhen presentat relatively high concentrations.Gelatindid not appear to inhibit binding significantly.The structuralrequirementsof collagenfor binding to occur appear to resemble those requiredfor collagen-induced plateletaggregationin which collagenquaternarystructurerather than collagentype per se is the important factor. Loss of secondaryor hiser orders of structureof FVIII/VWFas a result of heat denaturationor reductionof disulphidebonds decreasedor preventedbinding. In accord with the associationof biologicalactivitywith FVIII/VWFaggregates,optimal binding appeared to require the presenceof aggregatedFVIII/VWF. $ey Words: Factor VIII/vanWillebrandfactor, collagena,platelets. Member of the externalstaff of the Medical Research Council.

545

546

COLLAGEN-FVIII/VWF

INTERACTION

Vo1.32, No.6

INTRODUCTION The plasma protein Factor VIII/van Willebrand factor (FVIII/VWF) is considered to fulfil an important function in primary haemostasis by promoting the interaction of platelets with the subendothelium of the injured blood vessel wall (1,2,3). It may, therefore, also play an important part in the events underlying thrombosis and atherosclerosis. The involvement of FVIII/VWF in the interaction of platelets with the vessel wall may infer an ability of this protein to interact both with specific platelet surface receptors and, at the same time, the platelet-reactive constituents of the subendothelium, in particular collagen fibres which are known to possess potent platelet-aggregatory activity (4). A number of studies have indeed provided evidence for the occurence of an interaction between collagens and FVIII/VWF. In particular, the rapid absorption from plasma of FVIII/VWF, detected by measurement of either FVIII-related antigen or ristocetin cofactor activity, upon exposure of plasma to fibres of collagens types I, II and III has been described (5-g). More directly, binding between collagens (either as fibres in suspension or immobilised in air-dried films or on Sepharose) and isolated FVIII/VWF has also been reported (g,lO,ll). In our earlier investigations (IO), we noted the ability of purified FVIII/VWF to bind to basement membrane collagen type IV and collagen type V in addition to the interstitial collagen8 type I and III. All four collagens are known to occur in the subendothelial region of the vessel wall (12). Binding, in this instance, was measured from the absorption of 12%-labelled FVIII/VWF to airdried collagen films. The absorption was specific in as much as it was inhibited by an excess of unlabelled FVIII/VWF but was unaffected by an excess of fibronectin, another protein bown to complex with collagens. The absorption to air-dried films was, however, slow whilst binding to collagen (type I) fibres in suspension was found to be very rapid. Furthermore the precise degree of fibrillar structure within the air-dried film or the type of fibre formed, striated or otherwise, is uncertain especially in the case of films derived from collagens types IV and V. Our further investigations as described here, have therefore involved measurement of the binding of l25I_ FVIII/VWF to collagens in suspension, using various insoluble polymers of relatively well-defined and uniform quaternary structure. We report further on the specificity of the interaction, on its sensitivity to ionic strength and present some observations on the structural requirements of both reactants for binding to occur. MATERIALS AND METHODS Sources -of collagens Native type I collagen fibres from bovine deep flexor tendon were received as a finely-dispersed suspension (lOmg/ml) generously donated by Ethicon Inc., Somerville; New Jersey, USA and details relating to which are as previously presented (10). Prior to use a sample of the suspension was diluted to the required concentration with O.OlM acetic acid and dialysed against the same. The preparation remained finely-dispersed during this procedure. Collagens types ----I, III, IV and V were extracted from human placenta by pepsin digestion and purified by differential salt precipitation as described before (10,ll). The purity and identity of these collagen8 was confirmed by SDS-polyacrylamide gel electrophoresis (13).

vo1.32, No.6

COLLAGEN-FVIII/VWF INTERACTION

547

Reconstituted native-type fibres of collagens types I and III were prepared by dialysing a solution of each collagen (from human placenta) against 0.021 Na2HP0 at 4'. Prior to testing for binding activity, the fibres were disperset with a Potter glass-teflon homogeniser and dialysed against the appropriate sodium phosphate buffer (0.02 ionic strength, pH 7.6). The concentration of fibres was determined by hyclroxyprolineanalysis (as described below). Randomly cross-linked insoluble type I collagen was prepared from a solution of collagen type I-human pl%mecisely as described by Santoro and Cunningham (14) using glutaraldehyde (0.25%)as cross-linking agent. Prior to use, the insoluble product was dispersed in buffer (0.4 ionic strength phosphate, pH 7.6) at known concentration as just described. SLS aggregates of collagens types I, III, IV and V (from human placenta) wereobtained by dialysis against an acid solution of ATP and stabilised by cross-linking with formaldehyde exactly as described before (10). Aggregates were dispersed in appropriate phosphate buffer at known concentration, as detailed above. Non-striated collagen fibrils were prepared by alcohol precipitation of a solution of collagen type I-human placenta) as described by Muggli (15). The precipitate was suspended in 0.02 ionic strength phosphate buffer at known concentration as above. Insoluble amorphous collagen was prepared from collagen type I (from human placenta) by precipitation from acid solution (0.5M acetic acid) by addition of NaCl to 0.7M. The precipitate was dispersed at known concentration (in 0.02 ionic strength phosphate) as above. Monomeric collagen for use as a competitor in binding asays was a solution of collagen type I (from human placenta) in 0.2 ionic strength phosphate buffer and centrifuged prior to use to ensure the absence of any aggregated species. Concentration was determined by hydroxyproline analysis. Denatured collagen was obtained by dialysing a suspension of native type I collagen fibres (from bovine tendon) against 0.2 ionic strength phosphate and then exposing the suspension to 100' for 20 min. The denatured product was centrifuged, the supernatant collected and the collagen concentration estimated by hydroxyproline analysis. Cyanogen bromide (CNBr)-treated collagen. Monomeric collagen type I (from human placentamssolved in 70% formic acid (at lOmg/ml), the solution gassed with N and CNBr added at 20mg/ml. The mixture was incubated for 5h at 30°, then 8 iluted with 10 vol of water and freeze-dried. The mixture of peptides was dissolved in 0.2 ionic strength phosphate at a concentration determined by hydroxyproline analysis. Sources -of FVIII/VWF Purified ?.%-FVIII/VWF was prepared precisely as described previously (10). A stock solution in 130mM NaCl, 15mM trisodium citrate, pH 6.7 containing 3mM sodium azide was stored until use as frozen aliquots.

548

COLLAGEN-FVIII/VWF INTERACTION

Vo1.32,No.6

Heat-denaturedFVIII/VWF. A sample of the stock solutionof '*%FVIII/VWFwas diluted 1:10 with 0.4ionic strengthphosphatebuffer and then denaturedby placing the solutionat 100° for 20min. Reduced FVIII/VWF An aliquot of the stock solutionof 1251-FVIII/VWFwas diluted 50 fold with 0.36Mtris-HCl,pH 8.6 containing0.2% EDTA. Reduction of diaulphidebonds with mercaptoethanolwas carriedout as describedby Bailey et al (16),but in the absence of urea, using three different concentrationa,0.01, 0.1 and l.O%, of reducingagent (17) to yield differing degrees of reduction. Samples were then alkylated(16)and dialysedagainst the appropriatephosphatebuffer. Cryoprecipitated FVIII/VWFcontaininga known amount of FVIII/VWFprotein was preparedas previouslyby the procedureof Mason et al (18). Other proteins Insolublebovine serum albumin was obtainedby treatinga solutionof the prote-maedfrom Chemical8Ltd.,Poole, England)with glutaraldehydea8 describedabove for the random cross-linkingof collagen type I monomer. The insolubleproduct wae, as previously,dialyaedagainst 0.05%acetic acid and then freeze-dried,weighed and dispersedat known concentrationby weight in the appropriatephosphatebuffer. Insolubleefaatin was obtainedfrom porcineaorta by the procedureof Starcherand Gam9) based on the use of CNBr to solubiliseall other componentsof the tissue. The resultantelastin preparationwas dieperaedat known concentrationby weight in appropriatephosphatebuffer. Bindingaaaae This was undertakenessentiallyas describedpreviously(IO). Microcentrifugetubes (1.5ml capacity,from Alpha Labe.,Eastleigh,Ha&s., UK) were precoatedas before,prior to use, with 0.5 ml of Dulbecco'sMinimal EesentialMedium containing10% foetal calf serum. Excess liquid was removed and the tubes allowed to air-dry. The standardreactionmixture comprised0.5ml of appropriatephosphate buffer containingin auapensiona known amount (usually50 pg) of insoluble collagen(or other) binding agent and 0.01 ml of a solutionof 1251-FVIII/VWF containing 0.1 w of FVIII/VWFprotein of known specificactivityand equivalentto a minimum of 5000 cpm. Samples were incubatedon an orbital shakerat room temperaturefor a specifiedperiod of time (generally60 min) and then centrifugedfor 5 min at full speed in a micro-haematocrit centrifuge. Radioactivityin the pellet and eupernatantwaa estimatedin a Prias r-counter (PackardInstrumentCo. Inc, III. U.S.A.).Estimationswere usually performed in duplicateand results expressedas the mean. A control tube containingappropriatebuffer and 1251-FVIII/VWFbut no bindingagent was incorporatedwith each assay. In competitiveinhibitionstudiee,1251-FVIII/VWFwas preincubatedfor 60 min with a known amount of inhibitor(denaturedcollagenfibres,monomeric collagen8or the CNBr-treatedmonomer) dissolvedin 0.2 ionic strength phosphatebuffer prior to the additionof insolublecollagen (bovinetendon) fibrea. Incubationwas then continuedfor a further60 min.

COLLAGEN-FVIII/VWF INTERACTION

Vo1.32, No.6

549

Inhibition of binding of 125I-FVIII/VWF by unlabelled FVIII/VWF was studied by including known amounts of the unlabelled protein (cryoprecipitated) in the collagen fibre suspension prior to the addition of labelled FVIII/VWF. Hyfiroxyprolineanalysis Collagen concentrations were generally determined by hydroxyproline analysis. Samples were first hydrolyaed in 6M-IX!1in sealed tubes at 108' for l6h. Hydrolysatea were dried in vacua and hydroxyproline then estimated calorimetrically by the procedure of Bergman and Loxley (20). RESULTS The extremely rapid binding of purified '251-FVIII/VWF to collagen (type I) native-type fibrea (from bovine tendon) is illustrated in the data presented in Figure 1. In this particular experiment, binding was essentially complete at the earliest time point measurable (when centrifugation was undertaken immediately upon the addition of one component to the other). As shown in Figure 2, binding of radiolabel increased with increasing fibre

.

804 ??

60 .

40

20 F

L

I

I

20

I

1

I

I

60

40

I

I

80

I

I

100

Time (mid

FIG. 1 The binding of '251-FVIII/VWF to insoluble polymeric collagena. Binding is shown to suspensions of collagen type I native-type fibres (from bovine tendon; M) and SLS aggregates of collagena types I (o----O), III (_)andV(t--r ) from human placenta. Binding with time was estimated under the standard conditions described in Materials and Methods, using 0.02 ionic strength phosphate buffer, pH 7.6. Binding is expressed as the % of the total radioactivity in the sample that is associated with the insoluble collagen polymer following centrifugetion.

550

COLLAGEN-FVIII/VWF INTERACTION

vo1.32, No.6

concentration. Although it was found that binding to native fibres was invariably a rapid process (occurring maximally in less than 10 mins usually) irrespective of conditions, the actual level of binding achieved at saturating concentrations of collagen appeared to be dependent upon the ionic strength of the medium in which binding occurred (Figure 2). At relatively high values (as when 0.4 ionic strength sodium phosphate buffer, pH 7.6 was used in the binding assay) the amount of radiolabel bound, reaching a maximum of approximately 60% of the total (with collagen fibres present in very large excess) was less than that bound at lower ionic strengths (when 0.2 or 0.02 ionic strengh phosphate buffer was employed) where up to approximately 80% of TABLE 1. Binding -of a-FVIII/VWF

to Insoluble Collagens of Varying Structure.

Collagen form Expt. 1. Native fibre (type I from bovine tendon)

44 (100)

!TypeI SLS

36

Type III SLS

49 (111)

Type IV SLS

38

(86)

!Qpe V SLS

39

(89)

Artificial randomly croee-linked polymer (type I>

44 (100)

(82)

Expt. 2. Native fibre (as in Expt. 1.)

75 (100)

Reconstituted native fibre (type I)

72

(96)

Reconstituted native fibre (type III)

73

(97)

Non-striated fibril, alcohol precipitated (type I)

69

(92)

Non-fibrillar, salt precipitated (type 1).

18

(24)

Binding was estimated under the standard conditions described in Materials and Methods using in Expt. 1, phoephate buffer, ionic strength 0.4 and an incubation of 45 min and in Expt. 2, ionic strength of 0.02 and a binding time of 60 min. % Binding refers to the $ of the total radioactivity in the sample that is associated with the insoluble collagen upon centrifugation. The figures in parentheses are the relative binding values compared to that for the native type fibre (bovine tendon) expressed as 100%.

551

COLLAGEN-FVIII/VWF INTERACTION

Vo1.32, No.6

the radioctivitywas abeorbed. The 20% of radioactivityremainingunbound at the lower ionic strengthswae consideredto be unrelatedto FVIII since of the total radioactivityassociatedwith the 1251-FVIII/VWFaround 20% failed to be precipitatedby a specificanti-FVIII/VWFantieerum. The binding of 1251FVIII/VWF occurringat 0.4 ionic strengthwas almost totally inhibitedby an excess of unlabelledFVIII/VWF. That occurringat 0.2 or 0.02 ionic strength however could not be completelyprevented(see Figure 3) suggestingthat the additionalbindingat lower compared to higher ionic concentrationswas at

Amount

of Collagen

FIG.

fibres

(pg)

2.

The binding of I251-FVIII/VWFto type I collagenfibres. Binding is shown to euapeneioneof native-typefibrea from bovine tendon. Binding with increasingfibre concentrationwas estimatedunder standardconditionsusing phosphatebuffer, ionic strength0.4 ( M), 0.2 (M) and 0.02 ( > and an incubation period of 60min. Binding is expressedas in Fig.1. 100

FIG. 3.

1 SO-I

The inhibitionof binding of 1251-

FVIII/VWF to type I collagenfibres by unlabelledFVIII/VWF. Binding to bovine tendon native fibres was estimatedunder standardconditions using phoaphatebuffer, ionic strength 0.4 ( M ), 0.2 (w) and 0.02 (C--r) and in the presenceof varying amounts of unlabelledFVIII/VWF. Binding is expressed aa in Fig. I. 5

10

15

Amount of unlabelled

20

F XUI

25

(pg)

COLLAGEN-FYIII/VWF INTERACTION

552

Vo1.32, No.6

TABLE 2. Effect of Disul hide Bond Reductionon the Binding of '25_-FVIII/VWFto -Collagen=_ -her 1-s Reducingagent.-1 0

1

0.1

0.01

Experiment. 1

74

-

18

2

30

21

11

3

75

56

38

4

61

45

-

4

Results are shown for four separateexperiments. Binding to collagentype I native fibres (bovinetendon)was estimatedas deecribed in Materialsand Methods using phosphatebuffer, ionic strength0.4 in Expt. 1. and 2, 0.2 in Expt.4 and 0.02 in Expt. 5. Incubation was for 15 min in Expt. 1, otherwisefor 60 min. Reductionof FVIII/VWFusing the specifiedconcentrationsof mercaptoethanolwas undertakenas describedin the text. The values presentedrefer to the % binding as defined in Table I.

80

60

40

II

0 ??

.

20

I

1

. I

I

I

I

2

1

Concentration

of competitor

1

3

I

1

4

(mg/ml)

FIG. 4 The inhibitionof binding of '251-FVIII/VWFto type I collagenfibresby monomeric type I collagen. Bindingwas estimatedunder standardconditions using 0.2 ionic strengthphosphatebuffer. Bovine tendon native fibres were preincubated(for 60 min) with solubledenaturedfibres (H), monomeric or a CNBr digest of the monomer collagentype I from human placenta -) (M), priox to the additionof '4 51-FVIII/VWF. Binding is expressedas in Fig. 1.

Vol.32,

No.6

COLLAGEN-FVIII/VWFINTERACTION

553

least in part of an unspecificnature. In accord with the inverse relationshipbetween bindingand ion concentration,no binding was observedin the presence of 1M NaCl (assayedin 0.05M tris-HClbuffer, pH 7.4). This agrees with the findingsof Santoro and Cowan (9) who reportedthat the complex between FVIII/VWFand collagenfibres was partiallydissociatedby 1M NaCl. As indicatedin Figure 1 and Table I, it was possibleto record binding of 1251-FVIII/VWFto a varietyof insolubleoollagensof differingpolymeric structure. Binding of a comparableorder to that observedwith the nativetype bovine tendon fibres was noted with SLS aggregatesof collagen6types I, III, IV and V, reconetitituted native-typefibres of collagenstypes I and III, an insolublepolymer derivedfrom monomeric collagentype I randomly cross-linkedwith glutaraldehydeand non-striatedtype I fibrilsobtainedby alcohol precipitation.Amorphousnon-fibrillartype I collagenobtainedby salt precipitationfrom acid solutionexhibitedmuch lower bindingactivity. Insolubleelastinand bovine serum albumin randomly cross-linkedinto an insolublepolymer both showed no significantbindingactivityirrespectiveof the ionic strength employed (resultsnot shown). Binding to insolublenative-typecollagenfibres was only very slightly inhibitedin the presenceof relativelylarge amounts of either soluble denaturedcollagen(gelatin)obtainedby heat denaturationof the fibres or the solublemixture of peptidesobtainedby CNBr digestionof the type I monomer. However very high concentrationaof the intactmonomer appearedable to inhibit the binding to fibres in a concentrationdependentmanner (see Figure 4). Heat denaturationof 1251-FVIII/VWFcaused a decreasein binding to bovine tendon fibres of around 00$ (measuredin 0.4 ionic strengthphosphate buffer). Reductionof disulphidebonds under controlledconditions(17) designedto yield primarilyeither the basic protomericmolecule (composedof two subunits)by use of 0.1% mercaptoethanolor the subunit itself (obtained with 1% reducingagent)caused a losa of binding of the order of 50% in the former case and almost total in the latter (see Table 2). Even reduction under extremelymild conditions(0.01%reducingagent) resulted in substantial loss of binding. DISCUSSION The reaults presentedherein have demonstratedthe abilityof FVIII/VWF to bind to various polymericforma derived from a variety of collagentypes. The importanceof collagenquaternarystructurein bindingis emphasisedby the need for very high concentrationsof the monomer relativeto that of the polymer to produce significantinhibitionof binding to the native-typefibre. The actual collagentype appears to be unimportantsince,for example,SLS aggregatesof all four collagentypes teeted exhibitedcomparablebindingand reconstitutednative type fibres of collagentypes I and III showed similar binding activityto each other. The native-typepolymericstructure,in which moleculesare alignedin a highly specificmanner to yield striatedfibres of 67nm periodicity(as observedby electronmicroscopy),is not essentialfor bindingsince activitywas also observedwith other polymeric forms, including SLS aggregates,an insolublepolymer producedby random cross-linkingof the monomer and non-striatedfibrilsproducedby alcohol precipitation.These data are in accord with the known structuralrequirementsfor collagen8to

554

COLLAGEN-FVIII/VWF INTERACTION

vo1.32, No.6

induce platelet aggregation which is consistent with a possible role for FVIII/VWF in the interaction of collagens with platelets. Thus all collagen types in suitable polymeric form can induce platelet aggregation including besides the native-type fibre, SLS aggregates (21). the polymer derived from random cross-linking of the monomer with glutaraldehyde (14), and the alcoholprecipitated non-striated fibrils (15). Interestingly the amorphous product produced by salt precipitation from acid solution does not cause platelet aggregation (14,22)and reveals only poor binding activity. Gelatin also fails to induce the aggregation of platelets or to prevent their aggregation by collagen fibres (14) and is unable, as shown here, to inhibit the binding of isolated FVIII/VWF to collagen fibres or, as shown by Santoro and Cowan (q), to prevent the absorption from plasma by collagen fibres of FVIII-related ristocetin co-factor activity. In our previous studies (10) we also noted that binding to an air-dried gelatin film was substantially less than to the equivalent film obtained with native collagen. The binding of FVIII/VWF then, like collagen-induced platelet aggregation (15,21), appears to require an assembly of the native collagen monomer into some form of polymer but not necessarily the highly-ordered assembly associated with the native-type fibre. Although we have shown here that collagens types IV and V, as I and III, when in a suitable polymeric form can bind FVIII/VWF and can also, shown previously (21), induce platelet aggregation, it is not certain to what extent these two collagens possess a quaternary structure in vivo that would support such activity. Collagen types I and III exist in vivo as highly characteristic fibres (exhibiting a striation of 67nm periodicity) the activity of which can be demonstrated either by their direct isolation as fibres or by their reconstitution as such in vitro from collagen solutions. However the precise quaternary structure of collagens types IV and V in vivo has yet to be ascertained. The queation of their binding ability in vivo will remain unresolved until more is known of their polymeric form and the possibility of generating this in vitro is realised. Importance of the native structure of FVIII/VWF for binding is emphasised by the substantial loss of activity on heat denaturation. Furthermore, the decrease in activity associated with the reduction of the polymer to the dimeric protomer, or the total loss of binding upon more complete reduction to the subunit structure stresses the importance of the quaternary structure of FVIII/VWF as well as collagena for binding to occur. Our results are entirely consistent with the observation of Santoro (23) that collagen fibres preferentially remove from plasma the very large molecular weight aggregates of FVIII/VWF leaving a proportion of the total FVIII-related antigen activity unbound and they support the notion that the biological activity of FVIII/VWF reaides in the more highly aggregated species. It might be supposed that the radioactivity we observed remaining unbound to collagen at low ionic str ngth, representing around 20% of the total, might be due to the presence of 12$_ labelled forms of FVIII/VWF of relatively low molecular weight. However, this seems unlikely since, as already intimated, approximately this proportion of the total radioactivity appeared anyway to be immunologically distinct from FVIII/VWF and presumably it is this material which fails to bind to collagen. In our experiments, binding of FVIII/VWF was increased at lower relative to higher ionic strengths; at the higher values only a proportion (approx three-quarters) of the total FVIII/VWF was bound. The reason for this is unknown but presumably it may reflect the heterogeneity of the FVIII/VWF preparation in terms of the degree of aggregation, conceivably the lower ionic

vo1.32, No.6

COLLAGEN-FVIII/VWF INTERACTION

555

strengths altering the state of aggregation to larger forms which favour

binding. The specificity of the interaction is demonstrated both by the fact that the binding of radiolabelled FVIII/VWF could largely be prevented by an excess of unlabelled FVIII/VWF and the fact that binding to elastin or albumin was not observed.

ACKNOWLEDGEMENTS The authors are grateful to Dr. D.M. Scott for helpful discussion of this work and for his criticism of the manuscript. B.G. is supported by Grant No. K/MRS/50C387 from the Scottish Home and Health Department. REFERENCES BAUMGARTNER H.B., TSCHOPP T.B. and WEISS H.J. Platelet interaction with collagen fibrils in flowing blood. II Impaired adhesion-aggregation in bleeding disorders. Thromb. Haem. 'J7, 1728, 1977. 2. BAUMGARTNER H.R., TSCHOPP T.B. and MEYER D. Shear rate dependent inhibition of platelet adhesion and aggregation on collagenous surfaces by antibodies to human factor VIII/van Willebrand factor 1980. Brit.J.Haematol. 44, 127-139, -3. BOLHUIS, P.A., SAKAXASSEN, K.S., SANDER, H.J., BOUMA, B.N. and SIXMA J.J. Binding of factor VIII-von Willebrand factor to human arterial subendothelium precedes increased platelet adhesion and enhances platelet spreading. --J. Lab. Clin. Mea. 97, 568-576, 1981. 4. BARNES, M.J. The collagen-platelet interaction. In: Weiss, J.B. and Jayson, M.I.V. (e&s). Collagen in Health and Disease, Churchill Livingstone, Edinburgh, ppl79-197, 1982. Interaction of collagen with the factor VIII antigen5. NYMAN, D. activity-von Willebrand factor complex. Thromb. -Res. II, 433-438, 1977. 6. NYMAN, D. Von Willebrand factor dependen? platelet aggregation and adsorption of factor VIII-related antigen by collagen. Thromb. Res. 17, 209-214, 1980. 7. LEGRAND, Y.J., RODRIGUEZ-ZEBALLOS, A., KARTALIS, G. and CAEN, J.P. Adsorption of factor VIII antigen-activity complex by collagen. Thromb.Res.13, 909-911, 1978. -8. SANTORO, S.A. Adsorption of von Willebrand factor/factor VIII by the genetically distinct interstitial collagens. Thromb. Res. 21, 689-693, 1981. 9. SANTORO, S.A. and COWAN, J.F. Adsorption of von Willebrand factor by fibrillar collagen - implications concerning the adhesion of platelets to collagen. Collagen Rel. Res. 2, 31-43, 1982. 10. SCOTT, D.M., GRIFFIN, B., PEPPER,m.andgARNES, M.J. The binding of purified factor VIII/van Willebrand factor to collagens of differing type and form. Thromb. -Res. 24, 467-472. 1981. 11. SCHOSSLER, W and DITTRICH, C. Crossed immuno-affinity electrophoresis - a method for analysing the interaction of factor VIII-related antigen with vessel wall proteins. Thromb. -Res. 28, 677680, 1982. 12. BARNES, M.J. and SCOTT, D.M. Glycoproteins secreted by the endothelium 1.

556

COLLAGEN-FVIII/VWF INTERACTION

Vo1.32, No.6

and their involvement in specific interactions at the subendothelium, In: Cryer, A. (ea.), Biochemical Interactions at the Endothelium, Elsevier Science Publishers B.V., Amsterdam, pp 111-165, 1983. 13. LAEMMLI, U.K. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (Land) 227, 680-685, 1970. 14. SANTORO, S.A. and CUNNINGHAM, L.W.maEmed platelet aggregation : The role of multiple interactions between the platelet surface and collagen. Thromb. Haem, 43 , 158-162, 1980. 15. MUGGLI, R. Collagen-induced platelet aggregation : Native collagen quaternary structure is not an essential structural requirement. Thromb.Res. 13,829-843,1978. -16. BAILEY, A.J., SIMS, T.J., DUANCE, V.C. and LIGHT, N.D. Partial characterisation of a second basement membrane collagen in human placenta. --FEBS Lett 99, 361-366, 1979. 17. COUNTS, R.B., PASKELL, S.L. and ELGEE, S.K. Disulphide bonds and the quaternary structure of factor VIII/van Willebrand factor. -J. Clin.Invest.62,702-709, 1978. 18. MASON, E.C., PEPPER, D.S. and GRIFFIN, B. Production of cryoprecipitate of intermediate purity in a closed system, thawsiphon process. Thromb. Haem. 46, 543-546, 1981. 19. STARCHER, B.C. and GALION-.Purification and comparison of elastins from different animal species. Analyt. Biochem. 74, 441-447, 1976. 20. BERGMAN, I. and LOXLEY, R. Two improved and simolified methods for the spectrophotometrid determination of hydroxyproline. Analyt. Chem. 35, 1961-1965, 1963. 21. BARNES, M.J., BAILEY, A.J., GORDON, J.L. and MACINTYRE, D.E. Platelet aggregation by basement membrane-associated collagens. Thromb.Res.18, 375-388, 1980. -22. BARNES, M.J., GORDON, J.L. and MACINTYRE, D.E. Platelet-aggregating activity of type I and type III collagens from human aorta and chicken skin. Biochem. J. 160, 647-651, 1976. 23. SANTORO, S.A. PreferentslTnding of high molecular weight forms of von Willebrand factor to fibrillar collagen. Biochim. Biophys. Acta.756,123-126, 1983. a --