Evidence for a Structural Requirement for theAggregation of Platelets by Collagen

Russell Jaffe, Daniel Deykin

J Clin Invest. 1974;53(3):875-883. https://doi.org/10.1172/JCI107628.

This study investigates whether soluble collagen can initiate platelet aggregation orwhether a higher degree of polymerization is required. Purified rat skin collagen wasprepared in four states. Soluble monomeric collagen, containing 2 µM calcium chloride, wasmaintained at 4°C until use. A previously uncharacterized form of collagen, solublemicrofibrillar collagen, was prepared from monomeric collagen containing calcium chlorideby allowing it to polymerize at 23°C. Viscometric and electron microscopic characterizationof microfibrillar collagen indicated polymerization to ordered native filaments. Particulatenative macrofibrillar collagen was prepared from monomeric collagen by allowing it topolymerize at 37°C in the absence of calcium. Particulate collagen, in which the fibers wererandomly associated, was prepared by salt precipitation of calcium-free monomericcollagen. Microfibrillar and native macrofibrillar collagen initiated platelet aggregation, witha lag phase of approximately 60 s. Monomeric collagen initiated aggregation with a lagphase of approximately 180 s. The duration of the lag phase for platelet aggregationinitiated by monomeric collagen was independent of the dose. Salt-precipitated particulatecollagen did not initiate platelet aggregation. Agents which prolong the transition frommonomeric collagen to fibrillar collagen (urea, arginine) retarded or prevented theaggregation of platelets by monomeric collagen. Sodium borohydride, which stabilizes theintraand intermolecular cross-links of collagen did not affect platelet aggregation.Penicillamine, which displaces the intermolecular cross-links and binds the […]

Research Article

Find the latest version:

http://jci.me/107628/pdf

Pdf

Evidence for a Structural Requirement for

the Aggregation of Platelets by Collagen

RussELL JAFFE and DANIEL DEYKIN

From the Department of Medicine, Boston Veterans Administration Hospitaland Boston University Medical School, Boston, Massachusetts 02130

A B S T R A C T This study investigates whether solublecollagen can initiate platelet aggregation or whether ahigher degree of polymerization is required. Purifiedrat skin collagen was prepared in four states. Solublemonomeric collagen, containing 2 /hM calcium chloride,was maintained at 4°C until use. A previously unchar-acterized form of collagen, soluble microfibrillar collagen,was prepared from monomeric collagen containing cal-cium chloride by allowing it to polymerize at 230 C.Viscometric and electron microscopic characterization ofmicrofibrillar collagen indicated polymerization to or-dered native filaments. Particulate native macrofibrillarcollagen was prepared from monomeric collagen byallowing it to polymerize at 37°C in the absence of cal-cium. Particulate collagen, in which the fibers were ran-domly associated, was prepared by salt precipitation ofcalcium-free monomeric collagen. Microfibrillar and na-tive macrofibrillar collagen initiated platelet aggrega-tion, with a lag phase of approximately 60 s. Monomericcollagen initiated aggregation with a lag phase of ap-proximately 180 s. The duration of the lag phase forplatelet aggregation initiated by monomeric collagenwas independent of the dose. Salt-precipitated particulatecollagen did not initiate platelet aggregation. Agentswhich prolong the transition from monomeric collagento fibrillar collagen (urea, arginine) retarded or pre-vented the aggregation of platelets by monomeric col-lagen. Sodium borohydride, which stabilizes the intra-and intermolecular cross-links of collagen did not affectplatelet aggregation. Penicillamine, which displaces theintermolecular cross-links and binds the intramolecularcross-links of collagen, did not prevent platelet aggrega-tion. The data suggest that an architectural requirementexists for the initiation of self-perpetuating platelet ag-

This study was presented in part at the Federation ofAmerican Societies for Experimental Biology, April 1972.

Dr. Deykin is the recipient of a Career DevelopmentAward (HL-25261).

Received for publication 17 May 1973 and in revised form1 November 1973.

gregation; that tropocollagen units do not fulfill thisrequirement; that a soluble collagen preparation, micro-fibrillar collagen, contains the minimal structural unit;and that cross-linkages within collagen do not play acritical role in platelet aggregation.

INTRODUCTIONThe normal hemostatic sequence begins when a bloodvessel is injured and culminates in the formation of afibrin-platelet meshwork that is a structural barrier tothe escape of blood. The trigger that initiates these hemo-static reactions is the separation or disruption of theendothelium, thereby allowing flowing blood to contactsubendothelial connective tissue. Platelets immediatelyadjacent to the site of injury adhere to the connectivetissue. Subsequent reactions initiated by the adherenceof platelets to connective tissue lead to the formation ofa definitive hemostatic plug. Initially, Hugues (1) sug-gested, as has been repeatedly confirmed in other labora-tories (2-5), that collagen was the specific componentof connective tissue to which platelets first adhere.

There have been several reports which have examinedthe properties of collagen necessary to promote plateletadherence. Initially, it was shown that heat-denaturedor collagenase-digested preparations were inactive (2,5). Nossel and his associates (6-8) showed that treat-ment of collagen with reagents that either blocked oroxidized the free e-amino groups of collagen, impairedplatelet-aggregating activity. They suggested a specificrole for these amino groups, rigidly spaced in the col-lagen molecule, in the reaction between platelets andcollagen. Jamieson, Urban, and Barber (9, 10) havedescribed UDPGlucosyltransferases bound to plateletmembranes which link glucose or galactose to collagen.The suggestion that the carbohydrate side chain of col-lagen is essential in the platelet-collagen reaction hasbeen supported by Chesney, Harper, and Colman (11).

Within the past few years the structure of collagenhas been clarified (12, 13). Schmitt (14) has proposed

The Journal of Clinical Investigation Volume 53 March 1974- 875-883 875S

that the reconstitution of collagen in vitro proceedsthrough tlhree states of increasing complexity: mono-meric, microfibrillar, and particulate. Monomeric col-lagen (tropocollagen, mol wt 300,000) is a triple-stranded coiled coil composed of a- (single clhain) and1-(double chain intramolecularly cross-linked) subunits.Microlibrillar collagen is postulated to form when mono-meric collagen polymerizes between 12 and 27°C. It isthought to exist as a highly ordered structure of varyingmolecular weights ranging from 106 to 10'. Solublemicrofibrils are capable of additional association to formmacrofibrils which are particulate. This further poly-merization occurs slowly at 23°C but it is acceleratedat 37°C. Macrofibril formation can be inhibited by lowconcentrations of calcium (15).

Wehave prepared purified rat skin collagen in severalstates of association and have utilized them to examinewhich can initiate platelet aggregation. Our data indi-cate that a soluble, transitional structure more complexthan monomeric collagen is required to initiate plateletaggregation. Our findings suggest that modifications ofcollagein which impair its platelet-aggregating abilitymay do so at least in part by preventing the polymeriza-tioni of collagen into a critical conformation which sup-ports aggregation.

METHODSNeutral salt and acetic acid-soluble rat skin collagen wasprepared and purified by a minor modification of themethod of Gallop and Seifter (16). Amino acid analysisof the collagen revealed residue content consistent withpublished values for purified rat skin collagen. The puri-fied collagen contained 320 residues of glycine per 1,000amino acids and had a hydroxyproline to proline ratio of0.9, characteristic of collagen (17). Purified, lyophyllizedcollagen was reconstituted at a final concentration of 1 ,Ig/,ud (based on hydroxyproline content) (18) in 0.05 N aceticacid at 4°C. This material was treated in four ways toyield different states of collagen. Monomeric collagen wasprepared immediately by adjusting the p11 of the recon-stituted collagen slowly to pH 7.2 with 0.1 N NaOH, andthen making the solution 2 joM in CaC12, to preventtransition to macrofibrils. The monomeric collagen wasmaintained at 4'C until used. Microfibrillar collagen wasprepared from monomeric collagen by allowing portions tostand at 23°C for 90 min. At the concentration of collagen(1 mg/ml) and calcium (2 juM) employed, macrofibrilformation does not occur at 37'C, and the collagen solutionremains soluble indefinitely. In plasma, no fibril formationwas detected over the range 0.01-0.50 mg/ml. Macrofibrillar,native particulate collagen was prepared by a modificationof the previous techniques. Reconstituted collagen wasprepared at 4°C and adjusted to pH 7.2 in the absence ofcalcium. It was then allowed to polymerize at 37°C for20 min. After centrifugation at 15,000 g the supernatantmaterial was discarded. The pellet was resuspended in dis-tilled water, finely minced, and utilized as a particulatesuspension. The preparation was maintained at 4°C untilused. Material so prepared has been characterized as hav-ing ultrastructure and cross-links indistinguishable fromnative collagen (19, 20). Another form of particulate

collagen was prepared f rom monomeric collagen by thea(ldition at 4°C of 1 part of a solution of cold NaCl (20g/100 ml) to 4 parts of monomeric collagen. The fluffywNhite precipitate which formed on slow stirring for anlhoulr was centrifuged at 2,500 g for 20 min. The precipitclte' \was xvashed tlhree titlics wx-ith cold water and wasfinally resuspended in deionized \vatce at a final concentra-tion of 1 jag/jul. This material was used as a fine particu-late suspension, maintained at 4°C until used.

The physical states of the various preparations of col-lagen were assessed by electron microscopy and by vis-cometry. The collagen preparations were prepared for elec-tron microscopy by a modification of the method of Rauter-berg and Kuhn (21). Wet preparations were staineddirectly on a Formvar-coated grid with phosphotungsticacid (0.4 g/100 ml, pH 3.5) for 12 min. The grid wasthen washed with vater and was counterstained withuranyl acetate (1 g/100 ml) for 10 min. Micrographs weremade with an RCA EMU-3G electron microscope. Mono-meric preparations gave barely detectable staining patternsat the resolution employed. Macrofibrillar, native particu-late collagen appeared as broad sheets of typical collagenfibers, arranged in a dense meshwork. Particulate, salt-precipitated collagen appear as an amorphous material, inwhich no native, cross-banded fibrils could be seen. Thisappearance was consistent with previous descriptions ofsimilarly prepared material (22). The electron micro-graphic appearance of microfibrillar collagen is shown inFig. 1. The fibrils were approximately 600 A wide, withlengths ranging from 3 to 30 jum (mean of 50 measure-ments, 11.5 ,tm+2.7 SD).

The viscometric properties of the two soluble collagenpreparations, monomeric and microfibrillar, were carriedout in a size 100 Ostwald-Cannon-Fenske capillary vis-cometer. Temperature was controlled to within 0.05'C ina constant temperature, circulating water bath. Beforebeing tested, all collagen solutions were centrifuged at30,000 g for 30 mim. The solutions were then pre-equili-brated at the temperature to be tested. Each preparationwas tested, in triplicate, at five concentrations. The reducedintrinsic specific viscosity was determined by extrapolation(Fig. 2). When reconstituted, monomeric collagen wasassessed at 4'C, p11 3.8, an intrinsic specific viscosity of15.2 dl/g was observed, in accordance with publishedvalues for purified native tropocollagen (23). Adjustingthe pH of monomeric collagen to 7.1 at 4°C increased thespecific intrinsic viscosity to 18.2 dl/g. When the prepara-tion was allowed to equilibrate for 30 min at 23°C theintrinsic viscosity rose to 27.4 dl/g. There was no furtherincrease in viscosity on standing for as long as 90 min at23°C. The increase in viscosity was reversible. Reducing thetemperature of the collagen (after 60 min of standing at 23'C)to 4°C lowered the intrinsic viscosity to its original valueof 18.2 dl/g. After maintaining the temperature at 40Cfor a further 60 min, the temperature wvas then raisedlagain to 23°C, and the reduced intrinsic viscosity rose to27.0 dl/g as before.

TFhe combined electron microscopic and viscometric datatlhus established the four forms of collagen as distinct fromeachl other.

Platelet-rich plasma (PRP) 1 was prepared from normalsubjects by a two-syringe technique with plastic disposablesyringes and tlhin-walled 19-gauge needles. Whole bloodwas transferred to plastic centrifuge tubes containing 0.5

1 Abbreviationt uised in this paper: PRP, platelet-richplasma.

876 R. Jaffe and D. Deykin

w _ .*. j_ s,*. L _ . .._t 8 *t

; .e

#s i ."

-S-We.E+: . X.si: ;Y ::

_rEeIf. 44

. ffi ..S .t

_ fn :,'FJX j-: #

/

I

FIGURE 1 Microfibrillar collagen. Acid- and salt-soluble collagen (1 /ug//.l) containing 2MuM CaCl2 was adjusted to pH 7.2 at 4°C and allowed to polymerize at 23°C. Wet prepara-tions stained directly on grid. Insert: magnification, X 10. Original magnification, X 20,300.

ml of 3.8% trisodium citrate. Sufficient blood was added toreach a volume of 5 ml, and the anticoagulated blood wascentrifuged at room temperature at 180 g for 10 min. ThePRP was then removed with silicone-coated Pasteur pip-ettes. The platelet count of the PRP ranged from 150,000to 400,000/,l. Platelet-poor plasma was prepared in a simi-lar fashion except that the whole blood was centrifugedat 2,400 g for 20 min.

The interaction of collagen with platelets was assayedby the technique of platelet aggregation (24, 25). Thechange in optical density of stirred PRP induced by plate-let aggregation was measured at 37°C in a platelet aggre-gometer (Chrono-Log Corp., Broomall, Pa.) with a 609-nmred filter. 0.5 ml of PRP was added to a 0.312-inch di-ameter cuvette containing a 1 X 4-mm stirring bar (cutsegment of a paper clip). All collagen preparations were in0.05 M Na acetate, pH 7.1. When used, calcium concen-tration was 2 ,M. A measured volume of the collagenpreparation to be tested was added, and changes in opticaldensity were continuously recorded.

The final change in optical density is a complex func-tion, one of platelet adhesion to collagen and then of aseries of self-perpetuating platelet-aggregation reactionsinitiated by the adherence of platelets to collagen. Thistechnique, therefore, is only an indirect assay of platelet

adherence to collagen, the necessary first step in thesequence.

In some experiments urea was added to the PRP. Ureasolutions were freshly prepared and filtered through amixed bed resin (Dowex 50 X 8, Dowex 1 X 8) to re-move cyanate.

Sodium borohydride reduction (26) (0.18 mg/mg ofcollagen) of both monomeric and microfibrillar collagenwas performed under carefully controlled conditions toprevent fibril formation during the exothermic reaction.Reduction was performed at pH 7.0, 230C. Excess boro-hydride was removed by dialysis. To prevent fibril forma-tion during dialysis, the pH of the dialysis solution (0.05 MNa acetate) was raised gradually from 7.0 to 7.2 by se-quential dialysis against solutions of pH 7.0, then 7.1, andfinally pH 7.2, each with sufficient CaCl2 to attain 2 uM.Reduction of monomeric collagen was carried out as de-scribed above except that all reactions were carried out at40C. Dialysis was then conducted as with the microfibrillarcollagen. The reduced monomeric collagen was maintainedat 40C until use.

To assay the degree of reduction, standardized tritiatedsodium borohydride (26) was added to both forms ofsoluble collagen. The nonspecific radioactivity was removedby dialysis, and the bound radioactivity was measured after

Aggregation of Platelets by Collagen 877

z.-

A.-z..T 1.

.4:. i

acid and alkaline hydrolysis of the collagen, separation ofthe amino acids on an amino acid analyzer equipped witha stream-splitter, and detection of the radioactivity in thealcohols and secondary amines formed during the reductionof the native cross-links by liquid scintillationspectrometry.

RESULTS

Fig. 3 depicts representative tracings produced by theaddition of 30,ug of the native particulate, microfibrillar,or monomeric collagen. Both particulate macrofibrillarand soluble microfibrillar collagen initiate aggregationin 60 s or less. Monomeric collagen also initiated aggre-gation, but only after a delay in excess of 3 min. In con-trast, particulate, salt-precipitated collagen (not shown)did not aggregate platelets at any concentration employed(30-300 /Ag).

To test the reproducibility of these observations, weexamined the time required for the initiation of aggrega-tion (lag phase) after the addition of various forms ofcollagen to PRP from 30 different individuals. The lagphases observed were: macrofibrillar collagen, 52±20 s(mean ±2 SD); microfibrillar collagen, 53±14 s; andmonomeric collagen, 179±14 s. There was no significantdifference between macrofibrillar and microfibrillar col-lagen. The lag phase observed with monomeric collagenwas highly significantly different (P <0.001, Student'st-test) from those of particulate and microfibrillar col-lagen.

A representative dose-response curve for increasingamounts of monomeric collagen is shown in Fig. 4.1 gg of monomer did not initiate aggregation, but higher

42

'38 _3

3434

303

26

22

l 4-(3 18

0 1 2 3 4 5

COLLAGENCONCENTRATION(mg/lOOm/IFIGuRE 2 Reduced intrinsic specific viscosity of solublecollagen preparations. All determinations performed intriplicate. 0, Acid-soluble collagen (2 ,uM CaCl2), 4°C,pH 3.8; *, same preparation, pH 7.1, 4°C; 5, samepreparation, pH 7.1, 23°C; A, same preparation, tempera-ture lowered to 4°C after 60 min at 23°C; *, same prepa-ration, temperature restored to 23°C after 60 min at 4°C.

pg ( Ca ) PARTICULArE

06F-

04k

0.2k

3jopg

14J.

(3i

( b ) MICROF/BRILLAR

0.8 r

0.6F

0.4k

0.2

( C ) MONOMERIC

MINUTESFIGURE 3 Representative tracings of platelet aggregationinduced by 30 ,g of collagen added to 0.5 ml of PRP -con-tinuously stirred at 37°C. (a) Particulate native prepara-tion; (b) soluble microfibrillar collagen; (c) soluble mono-meric collagen.

doses produced increasing degrees of platelet aggrega-tion. At all doses, however, the lag phase remainedconstant.

Studies were undertaken to assess the effect of ureaon the transition from monomeric to microfibrillar col-lagen, and in parallel experiments on the initiation ofplatelet aggregation. The effect of urea on the poly-merization of monomeric collagen (containing 2 yiMCaCl2) was studied by viscometry (Fig. 5). The pH ofa solution of monomeric collagen was raised from 3.8to 7.1 at 4° C, and the specific intrinsic viscosity rosefrom an extrapolated value of 15.6 to 16.0 dl/g, con-sistent with our prior observations (Fig. 2). The

878 R. Jaffe and D. Deykin

0.

I..4tJ

(3KiCA

0.

0.

~~~~~~~~~3.2 _ \4,_. ^ 8

10

o I lo 2 4 6 8 10 12 18

MINUrES

FIGURE 4 Dose-response curve for monomeric collagen.Collagen was added with a chilled syringe.

solution was then made 0.1 M in urea, and the tempera-ture was raised to 230C. In contrast to the prompt riseto 27.0 dl/g observed in the absence of urea (Fig. 2),the intrinsic viscosity had not changed at 30 min androse slowly thereafter reaching 21.0 dl/g at 90 min.Urea had no effect on preformed microfibrillar collagen(Fig. 6). The specific intrinsic viscosity of a mono-meric solution of collagen containing 2 ,uM CaCl2,measured at pH 3.8, 4°C, was 15.6 dl/g. The solutionwas adjusted to pH 7.1 and allowed to equilibrate at23°C for 30 min. The intrinsic viscosity was now 27.8dl/g. The solution was rendered 0.1 M in urea, and theviscosity was measured after 30, 60, and 90 min stand-ing at 23°C. There was no change in the intrinsic spe-cific viscosity.

To assess the effect of urea on collagen-initiatedplatelet aggregation, sufficient urea was added to PRPto reach a final concentration of 0.1 M (no attempt wasmade to assess the degree of protein binding of urea).Urea did not impair the aggregation of platelets initiated

8..

K

~It(341

38rQ..

(It-

IIJi8::

341

30

26

22 -

18

14

0 1 2 3 4 5

aCLGENV caZCENTRAT/ION (mg/lOOm/J

FIGURE 6 Effect of urea on preformed microfibrillar col-lagen: Mean of triplicate determinations. 0, acid-solublecollagen (2 ,AM CaCl2), pH 3.8, 4°C; El, same preparation,pH 7.1, 230C for 30 min; *, same preparation rendered0.1 M in urea, 23°C for 90 min.

by the addition of 10 Ag of microfibrillar collagen (Fig.7a). By contrast, 0.1 M urea completely prevented theaggregation initiated by the addition of an equal amount

(a) MICROFIBRILLAR COLLAGEN

0.6k

(I)Zt

0tl~

K40R

0.6

0.2k

Urea

Control

i Opg (b) MONOMER/CCOLLAGEN

04 V

0.2 -

COLLAGENCONCENTRATION(mg/IOOV)

FIGURE 5 Effect of urea (0.1 M) on reduced intrinsicspecific viscosity of soluble collagen. Mean of triplicatedeterminations. 0, acid-soluble collagen (2 AM CaC,2),4°C, pH 3.8; 0, same preparation rendered 0.1 M inurea, pH 7.1, 4°C; EO, same preparation after 30 min at23°C; *, same preparation after 60 min at 23°C; A,same preparation after 90 min at 23°C.

0 4 8 12 16 20MINUV/TES

FIGURE 7 Effect of urea (0.1 M final concentration) onplatelet aggregation. (a) 10 ;Lg microfibrillar collagenadded at arrow. (b) 10 ,ug monomeric collagen added atarrow.

Aggregation of Platelets by Collagen 879

O.

04k

TABLE ILag Period for NaBH4-Reduced Monomeric and

Microfibrillar Reagent Colkagens

Lag period*

Collagen preparationi Intact Reducedt

Mloiioimieric 210 195Microfibrillar 75 60

* Interval between additioni of (ollagen anid first detectabledecrease in optical density.t Sodium borohydride redtictioni (0.18 miig NaBH4/11mg colla-gen) was carried out as described in Methods.0.5 ml PRP was prewarmed at 37°C for 1 mim. 3 uig of eachcollageii preparationi was added to the PRP. For monomericcollagen, the syringe was precooled.

of monomeric collagen (Fig. 7b). The defect was notabsolute. When 50 Ag of monomieric collagen was used,aggregation was delayed and did not begin until 9 minafter the addition of collagen. Arginine (0.12 M1) pro-duced similar effects to urea at both concentrations ofcollagen.

In a further experiment monomeric collagen (10 ,ug)was incubated for 20 min in 0.5 ml of PRP containing0.1 M urea (final concentration), after which 5 ,ug ofmicrofibrillar collagen was added. As a control, 5 Agof microfibrillar collagen was added to a similarly incu-bated sample of PRP (0.1 M urea) to which no mono-meric collagen had been added. In both instances, briskaggregation occurred withl the expected lag period ofapproximately 1 min.

D-Penicillamine, between 0.02 and 0.2 nmMI, did notaffect aggregation initiated by microfibrils, but increas-ing amount of penicillamine progressively prolonged thelag period after the addition of monomeric collagen. Thefirst detectable prolongation of aggregation was ob-served at a final D-penicillamine concentration of 0.02mM. Maximum prolongation (11 min) was observed at0.1 mM.

In further experiments, the cross-links of both mono-meric and microfibrillar collagen were reduced withsodium borohydride (Table I). There was no differencein the onset of aggregation nor in the degree of aggrega-tion between the reduced and nonreduced forms of col-lagen.

DISCUSSIONThe structure of collagen and the sequence of the for-mation of collagen fibers from soluble monomeric pre-cursors have been summarized recently (12, 13). Mono-meric skin and tendon collagen consists of three poly-peptide chains (two a, and one a2) coiled in a helix.Each chain has an approximate molecular weight of

95,000. Throughout most of the length of the chainglycine is present as every third amino acid. Hydroxy-lysine residues are subsequently glycosylated by glucosyl-and galactosyltransferases to form O-galactosyl-P-gly-cosyl side chains. Carbonyl groups are introduced by anenzyme(s) (27) which forms aldehydes from lysine andh)ydroxylysine. Thrlese aldeldeydes r-eact w\ ith aninlo aci(dside chains onl other collagen molecules to formi inter-molecular cross-links and react with each other to formlaldols which serve as intramolecular cross-links. Collagenmolecules in solutioni under physiologic conditions self-orient to facilitate the formationi of aggregates.

Wehave utilized the previous observationi of Bensusanand Hoyt (15) that low concentrations of calcium retar(dthe formation of fibrils to prepare a form of collagenithat has undergone partial but reversible polymeriza-tion. The electron microscopic and viscometric data pre-sented in Figs. 1 and 2 clearly identify microfibrillar col-lagen as a more higlhly ordered structure than monomolericcollagen. Because of the mlarked axial assymmetry (ob-served on the electron photomicrographs) of the micro-fibrillar preparation, the measurements of viscosity mayactually underestimate the degree of polymerization byas much as 50% (28, 29), but the qualitative differencesbetween monomeric and microfibrillar collagen are clear-cut. The reversibility of the microfibrillar preparation, itssolubility, and its electron microscopic appearance dif-ferentiate it as well from native particulate collagenformed in the absence of calcium.

Our data demonstrate that our preparation of micro-fibrillar collagen is a potent initiator of platelet aggrega-tion, as effective as native particulate collagen. This ob-servation suggests that the platelets adhere as well tomicrofibrillar as to particulate collagen. Our observa-tion that particulate salt-precipitated collagen does notinitiate platelet aggregation indicates that randomlyoriented native collagen aggregates do not afford anappropriate surface for platelet aggregation. The dif-ference in effectiveness between monomeric and micro-fibrillar collagen, therefore, is not simply a functionof the size of the collagen polymers.

The difference in lag phase between monomeric andmicrofibrillar collagen (Fig. 3) could reflect either oftwo possibilities. The monomeric preparation may beitself a less effective initiator of aggregation in thatfewer platelets adhere to the collagen compared withthe same microfibrillar preparation. Alternatively, themonomeric preparation may have no effect but mustfirst polymerize to form microfibrils which then initiateaggregation. To examine these alternatives we deter-mined the dose-response curve of platelet aggregationinitiated by increasing amounts of monomeric collagen.It was our hypothesis that if monomers were inefficientaggregators, then increasing the dose should progres-

880 R. Jaffe and D. Deykin

sively shorten the delay before aggregation becomes ap-parent. Conversely, since in the doses of collagen weemployed the rate of polymerization of monomers tomicrofibrils is independent of the concentration of col-lagen, then if polymerization is a requirement for theinitiation of aggregation by monomers, increasing thedose should not affect the lag phase. The data in Fig.4 suggest that monomers themselves do not initiate self-perpetuating platelet aggregation but must first poly-merize to microfibrils.

To test further our hypothesis that a molecular struc-ture more complex than that present in monomeric col-lagen is required to initiate aggregation, experimentswere carried out withl urea. The viscometric data pre-sented in Figs. 5 and 6 indicate that 0.1 Murea retardsthe formation of microfibrillar collagen but does notdissociate preformed microfibrillar collagen. These ob-servations extend prior findings that urea retards fibrillarcollagen formation (30). Similarly, arginine has beenshown to impair the polymerization of purified collagenin a manner identical to urea (30). Our observation,that urea (Fig. 7) and arginine prevent aggregation ofplatelets by monomeric collagen but do not influence theeffect of microfibrillar collagen, strengthens our con-cept that monomeric tropocollagen must first polymerizeto microfibrillar collagen to initiate aggregation.

The experiment in which monomeric collagen waspreincubated with PRP before the addition of micro-fibrillar collagen further suggests that platelets do notadhere to monomeric collagen or that if they do, theydo not undergo the release reaction, since the subsequentaddition of microfibrillar collagen resulted in briskaggregation. Since the exposure of both forms of col-lagen to platelets occurred in the presence of urea, thelack of aggregation with monomeric collagen cannot beascribed solely to the urea.

Penicillamine displaces the Schiff-base intermolecularcross-links in fibrillar collagen (31). Since penicillaminedid not diminish the ability of microfibrillar collagen toinitiate platelet aggregation, these links are not criticalto the platelet-collagen interaction. The results of peni-cillamine treatment of monomeric collagen are morecomplex, since there was a prolongation of the lag phase.The fact that aggregation did occur indicates that thebinding of the aldehydic groups in monomeric collageninterfered with the rate of fibril formation, but did notprevent polymerization.

Additional evidence that neither intra- nor intermolec-ular cross-links are essential for collagen-initiated plate-let aggregation was obtained from the sodium boro-hydride experiments (Table I). Stabilization of bothclasses of bonds results from the introduction of hydro-gens by sodium borohydride, converting the aldehydes

to alcohols and forming covalent secondary amines fromthe Schiff bases. Therefore, neither displacing bondswith penicillamine nor fixing them by reduction inter-fered with platelet aggregation. This evidence, takentogether, strongly suggests that the lysine and hydroxy-lysine-derived cross-links do not have a specific func-tion for platelet aggregation.

Nossel and his associates (6-8) treated acid-dispers-able collagen with reagents (nitrous acid, glacial acetic,and acetic anhydride, 2,4-dinitrofluorobenzene, and 2,4,6-trinitrobenzenesulfonic acid) which blocked the freeamino groups of lysine. They found that the treatedcollagen lost its ability to aggregate platelets. They con-cluded that free amino groups in general and specificiallythe e-amino groups of lysine were critical for the plate-let-aggregating activity of collagen. As shown byRauterberg and Kuhn (21), however, such forms oftreatment may lead to random orientation of the col-lagen chains in the particulate collagen that results. Ourdata show that particulate collagen treated in this wayis ineffective in initiating aggregation. Therefore, otherfactors in addition to the blocking of free amino groupsof lysine may have contributed to the loss of platelet-aggregating activity.

Jamieson and his co-workers (9, 10) have suggestedthat a platelet membrane-bound glucosyltransferase en-zymatically links the platelet membrane to a glycosidicreceptor on collagen. They postulate that the link formsthe basis for the platelet-aggregating ability of collagen.Chesney, Harper, and Colman, (11) extended thesefindings by demonstrating that treatment of acid-solublecollagen with galactose oxidase impaired the subsequentability of that material to aggregate platelets. They con-cluded that galactose was essential in platelet aggrega-tion by serving as a receptor for platelet glucosyl-transferase. However, Muggli and Baumgartner (32)have presented data which agree with our original re-port (33) and show further that galactose oxidasetreatment of soluble collagen monomers impairs theirability to polymerize.

Our observations lead us to propose an alternativehypothesis to explain much of the previously reporteddata. We suggest that the chemical modifications ofcollagen which impair its ability to initiate platelet ag-gregation may do so by impairing the polymerizationof collagen into an oriented meshwork.

In a recent report, Katzman, Kang, and Beachy (34)stated that isolated al-monomers and the al-CB5 frag-ment of car-monomers purified from denatured collagenaggregate platelets. It is difficult to assess their data inthe absence of ultrastructural or physical evidence thatreassociation had not taken place in these preparations.Since a large excess of denatured protein was added in

Aggregation of Platelets by Collagen 881

their experiments, small amounts of renatured collagenmay have been included. Their work is, as they acknowl-edged, contrary to that of Wilner, Nossel, and LeRoy(6), Chesney et al. (11), as well as to our own. Katz-man and his associates (34) also claim that the inhibi-tions of platelet aggregation that they observed withglucosamine and glucosyl-galactosyl-hydroxylysine re-flected binding of a specific platelet receptor site by theseagents. However, it has been previously shown thatglucosamine retards collagen fibril formation (35) whichmay explain the inhibition of aggregation by the prepara-tion of collagen they used. Glucosyl-galactosyl-hydroxy-lysine also contains a sugar-protonated group and maybehave as does glucosamine.

Our demonstration that a material less complex thanmacrofibrillar collagen can initiate platelet aggregationin vitro may be pertinent to hemostasis. Baumgartnerand Haudenschild (36) have shown that platelets canadhere to subendothelial amorphous material. When thesubendothelial amorphous material is treated with col-legenase, platelets no longer adhere. They point out thatthe amorphous material differs from fibrillar collagenin being sensitive to trypsin, in that it is less reactivewith platelets than fibrillar collagen, and it is morpholoc-ially distinct from collagen fibers. They recognize thattheir arguments do not exclude the presence of collagen-like proteins in the amorphous material. To the extentthat microfibrillar collagen exists as a component inthe amorphous subendothelial material, it may serve asan important thrombogenic surface.

ACKNOWLEDGMENTSWe wish to thank Professor Carl Franzblau for his help-ful criticisms.

This study was supported by g-rants f rom the NationalHeart and Lung Institute (HL-14182 and HL-11414).

REFERENCES1. Hugues, J. 1960. Accolement des plaquettes au collagen.

C. R. Soc. Biol. 154: 866.2. Zucker, M. B., and J. Borelli. 1962. Platelet clumping

produced by tissue suspensions and collagen. Proc. Soc.Exp. Biol. Med. 109: 779.

3. Hovig, T. 1963. Aggregation of rabbit blood plateletsproduced in vitro by saline "extract" of tendons. Thromb.Diath. Haemorrh. 9: 248.

4. Spaet, T. H., J. Cintron, and M. Spivack. 1962. Someproperties of the platelet-connective tissue mixed ag-glutination reaction. Proc. Soc. Exp. Biol. Med. 111:292.

5. Spaet, T. H., and M. B. Zucker. 1964. Mechanism ofplatelet plug formation and role of adenosine diphosphate.Am. J. Physiol. 206: 1267.

6. Wilner, G. C., H. L. Nossel, and E. C. LeRoy. 1968.Aggregation of platelets by collagen. J. Clin. Invest.47: 2616.

7. Nossel, H. L., G. C. Wilner, and E. C. LeRoy. 1969. Im-portance of polar groups for initiating blood coagulationand aggregating platelets. Nature (Lond.). 221: 75.

8. Wilner, G. C., H. L. Nossel, anid T. L. Procupez. 1971.Aggregation of platelets by collagen: polar active sitesof insoluble human collagen. Am. J. Physiol. 220: 1074.

9. Jamieson, G. A., C. L. Urban, and A. J. Barber. 1971.Enzymatic basis for platelet: collagen adhesion as theprimary step in haemostasis. Nat. New Biol. 234: 5.

10. Barber, A. J., and G. A. Jamieson. 1971. Platelet colla-gen adhesion characterization of collagen glucosyltrans-ferase of plasma membranes of human blood platelets.Biochim. Biophys. Acta. 252: 533.

11. Chesney, C. M., E. Harper, and R. W. Colman. 1972.Critical role of the carbohydrate side chains of collagenin platelet aggregation. J. Clin. Invest. 51: 2693.

12. Grant, M. E., and D. J. Prockop. 1972 The Biosyntlhesisof collagen, N. Engl. J. Med. 286: 194; 242; 291.

13. Tanzer, M. L. 1973. Cross-linking of collagen. Endo-genous aldehydes in collagen react in several ways toform a variety of unique covalent cross-links. Science(Wash. D. C.). 180: 561.

14. Schmitt, F. O., J. Gross, and J. H. Highberger. 1955.The crystallization of collagen in fiber formation. Sni/p.Soc. Exp. Biol. 9: 148.

15. Bensusan, H. A., and B. L. Hoyt. 1958. Effect of variousparameters on the rate of formation of fibers in collagensolutions. J. Am. Chem. Soc. 80: 719.

16. Gallop, P. M., and S. Seifter. 1963. Preparation andproperties of soluble collagen. Methods Enzymol. 6: 635.

17. Ramachandran, G. K. 1967. Treatise on Collagen. Aca-demic Press, IInc., New York. 284.

18. Prockop, D. J., and S. Udenfriend. 1960. A specificmethod for the analysis of hydroxyproline in tissues andurine. Anial. Biochemtl. 1: 228.

19. Gross, J. 1964. Organization and disorganization of col-lagen. Biophys. J. 4: 63.

20. Franzblau, C., A. H. Kang, and B. Faris. 1970. In vitroformation of intermolecular crosslinks in chick skinicollagen. Biochem. Biophys. Res. Comniunn. 40: 437.

21. Rauterberg, J., and K. Kuihn. 1968. The renaturation be-haviour of modified collagen molecules. Iloppe-Seler's-Z. Physiol. Chem. 349: 611.

22. Pease, D. C., and M. Bouteille. 1971. The tridimensionalultrastructure of collagenous fibrils cytochemical evidencefor a carbohydrate matrix. J. Ultrastrutct. Res. 35: 339.

23. Kahn, L. D., and L. P. Witnauer. 1966 The viscometricbehaviour of solubilized calf skin collagen at low shearrates. J. Biol. Chem. 241: 1784.

24. Born, G. V. R., and M. J. Cross. 1963. The aggregationof blood platelets. J. Physiol. (Lond.). 168: 178.

25. O'Brien, J. R. 1962. Platelet aggregation. Part I. Someeffects of the adenosine phosphates thrombin and cocaineupon platelet adhesiveness. II. Some results of a newmethod of study. J. Clin. Pathol. 15: 446, 452.

26. Gallop, P. M., and 0. 0. Blumfeld. 1966. Amino alde-hydes in tropocollagen: the nature of a probable cross-link. Proc. Natl. Acad. Sci. U. S. A. 56: 1260.

27. Narayanan, A. S., R. C. Siegel, and G. R. Martin. 1972.On the Inhibition of lysyl oxidase by B-amino proprio-nitrile. Biochem1. Biophys. Res. Commun. 46: 745.

28. Scheraga, H. A. 1955. Observations on the viscometricproperties of soluble collagen. J. Chem. Phys. 23: 1526.

882 R. Jaffe and D. Deykin

29. Simha, R. 1940. Physical studies of purified collagen.J. Phys. Chem. 44: 25.

30. Gross, J., and D. Kirk. 1958. The heat precipitation ofcollagen from neutral salt solutions: some rate-regulatingfactors. J. Biol. Chern. 233: 355.

31. Nimni, M. E. 1965. Effect of penicillamine on collagen.Biochim. Biophys. Acta. 111: 516.

32. Muggli, R., and H. R. Baumgartner. 1973. Collagen in-duced platelet aggregation requirement for tropocollagenaggegates. Proceedings of the 4th International Congresson Thrombosis and Haemostasis, Vienna. 86.

33. Jaffe, R., and D. Deykin. 1972. Platelet aggregation ini-tiated by microfibrillar collagen. Fed. Proc. 31: 241.

34. Katzman, R. L., A. H. Kang, and E. H. Beachy. 1973.Collagen-induced platelet aggrgation involvement of anactive glycopeptide fragment (al-CB5). Science (Wash.D. C.). 181: 670.

35. Le Gande, Y., J. P. Caen, and L. Robert. 1968. Effect ofglucosamine on platelet-collagen reaction. Proc. Soc.Exp. Biol. Med. 127: 941.

36. Baumgartner, H. R., and C. Haudenschild. 1972. Adhe-sion of platelets to subendothelium. Ann. N. Y. Acad. Sci.201: 22.

Aggregation of Platelets by Collagen 883