Studies of the articular cartilage proteoglycan aggrecan in health and osteoarthritis Evidence for molecular heterogeneity and extensive molecular changes in disease

Studies of the articular cartilage proteoglycan

aggrecan in health and osteoarthritis. Evidence

for molecular heterogeneity and extensive

molecular changes in disease.

G Rizkalla, … , E Bogoch, A R Poole

J Clin Invest. 1992;90(6):2268-2277. https://doi.org/10.1172/JCI116113.

Changes in the structure of the proteoglycan aggrecan (PG) of articular cartilage were determined immunochemically by RIA and gel chromatography and related to cartilage degeneration documented histologically by the Mankin grading system. Monoclonal antibodies to glycosaminoglycan epitopes were used. In all cartilages, three chondroitin sulfate (CS)-rich populations of large size were observed in addition to a smaller keratan sulfate (KS)-rich population. In grades 7-13 OA cartilages (phase II), molecules were

significantly larger than the equivalent molecules of grades 2-6 (phase I). CS chain lengths remained unchanged. In most OA cartilages, a CS epitope 846 was elevated in content, this being most marked in phase II (mean: fivefold). Loss of uronic acid, KS, and hyaluronic acid were only pronounced in phase II OA because of variations in normal contents. Aggregation of PG was unchanged (50-60%) or reduced in OA cartilages, but molecules bearing epitope 846 exhibited almost complete aggregation in normal cartilages. This study provides

evidence for the capacity of OA cartilage to synthesize new aggrecan molecules to replace those damaged and lost by disease-related changes. It also defines two phases of PG change in OA: an early predominantly degenerate phase I followed by a net reparative phase II accompanied by net loss of these molecules.

Research Article

Find the latest version:

http://jci.me/116113/pdf

Studies

of the

Articular

Cartilage Proteoglycan

Aggrecan in

Health

and

Osteoarthritis

Evidence forMolecular Heterogeneity and Extensive MolecularChanges in Disease

GeihanRizkalla,*Agnes Reiner,* Earl

Bogoch,t

and A. RobinPoole*

*JointDiseasesLaboratory, ShrinersHospitalfor Crippled Children, Division ofSurgical Research, Department ofSurgery, McGill University,Montreal, Quebec, Canada, H3G IA6; and*Departmentof OrthopaedicSurgery,

WellesleyHospital, University of Toronto, Toronto, Ontario, Canada, M4YIJ3

Abstract

Changes in thestructureof the proteoglycanaggrecan(PG)of articularcartilageweredeterminedimmunochemicallybyRIA

andgelchromatography andrelatedtocartilage degeneration

documented histologically by the Mankin grading system.

Monoclonal antibodies to glycosaminoglycan epitopes were

used. In allcartilages,threechondroitin sulfate(CS)-rich

popu-lations of large sizewereobserved inadditionto asmaller

kera-tansulfate(KS)-richpopulation.Ingrades 7-13 OAcartilages

(phaseII), moleculesweresignificantlylargerthan the equiva-lent moleculesof grades 2-6 (phaseI). CS chain lengths

re-mained unchanged. In mostOA cartilages,a CSepitope846

was elevated in content,this beingmost marked inphase II (mean:fivefold). Loss of uronicacid, KS, andhyaluronic acid

wereonlypronouncedinphaseIIOA because of variationsin

normalcontents.Aggregationof PGwasunchanged(50-60%)

orreduced inOA cartilages,butmoleculesbearing epitope846 exhibited almost complete aggregation in normal cartilages.

Thisstudyprovidesevidence for thecapacityofOA cartilageto

synthesize newaggrecan moleculestoreplacethosedamaged

and lostbydisease-relatedchanges.Italso definestwophases

ofPG change in OA:anearlypredominantly degenerate phase

Ifollowed bya netreparativephase II accompanied by net loss

.ofthesemolecules.(J. Clin. Invest. 1992. 90:2268-2277.)Key

words: cartilage-proteoglycan* osteoarthritis * aggrecan *

im-munochemistry

Introduction

Osteoarthritis isadegenerative jointdiseasethatinvolves

im-pairmentandeventuallylossofjointfunction. It is

character-ized by extensive degeneration of the articular cartilages in bothlargeandsmalljoints, usuallyinthepresenceof low level intraarticularinflammation (1).

Thearticular cartilageofadiarthrodialjointiscomposedof

anextensive networkof collagenfibrils that

provide cartilage

Address allcorrespondenceto Dr. A.RobinPoole,ShrinersHospital

forCrippled Children, Division ofSurgical Research, Departmentof

Surgery,McGillUniversity,1529 Cedar Ave.,Montreal,Quebec, Can-ada, H3G1A6.

Geihan Rizkalla'scurrent address is Department ofPsychiatry,

Royal VictoriaHospital,McGillUniversity.

Receivedfor publication26February1992andinrevisedform31

July1992.

with its tensile strength and stiffness (2, 3). This is composed

predominantly oftypeII collagen. Type XI collagen is present

withinthefibril(4). On its surface, covalently attached to type IIcollagen in thenonhelical telopeptide regions, resides type IXcollagen (5, 6).Interactingwith this network, either directly orindirectly, isa"meshwork" of highmolecular weight

hyal-uronic acid (HA)'(7). A large proteoglycan (PG), called ag-grecan, whichon synthesishas a 2-4 X 106

Mr,

binds to HA

throughanamino-terminal GIglobulardomain (8). Each

at-tachment is stabilized by a single link protein (9, 10) that

sharessequencehomology withthe G 1 domain(11, 12). The core protein ofaggrecan contains a second globular

domain called G2 which hasconsiderable homology with GI ( 12):its function is unknown.BetweenG2and another

globu-lar domain called G3 (12, 13) situated at the

carboxyl-ter-minal, there are the glycosaminoglycan attachment regions ( 12, 13).AdjacenttoG2 thereisakeratansulfate (KS) attach-ment region composed of repeating hexapeptide motifand

containingup to 30 KSchains ( 13, 14). Between this region and G3 there isahigh concentration ofup to 100 CSchains (13). There is evidence fromstudies of humanaggrecan, based on coreproteinstructure, that thisregionmaybe composed of

twosubdomains called CS-I and CS-2 (13). The G3 domain has beenshowntohavelectin-likeproperties,since it bindsto

galactosyl residues (15). This property may permit interac-tions with moleculesinthe extracellular matrix.

Antibodies have been preparedtounsaturatedsulfated and unsulfated disaccharides of chondroitin sulfate(CS),someof which remainasstubs attachedtothe coreproteinofaggrecan

after eliminativedigestion of CS (16, 17). These antibodies

react preferentially with the "stubs" thatremain attached to coreprotein,andaresulfationspecific (disaccharide

chondroi-tin4-sulfate stubs attachedto coreprotein [AdiC4S-s], disac-charide chondroitin 6-sulfate stubs attached to core protein

[AdiC6S-s], and disaccharide chondroitin stubs attached to coreprotein

[AdiCOS-s]).

Morerecently,antibodiestonative CSepitopeshave beenprepared ( 16, 18, 19).Theseareusually

mostcommonly found in fetalorembryonic tissues.The

ma-jority oftheepitopes theyrecognizeremaintobecharacterized. AntibodiestoKS(20,21)havebeen showntorecognize highly

sulfated domainsonthisglycosaminoglycan (22).Useofthese

antibodies in immunoassays permitsthedetection and

mea-surementof PG andfragmentsthereofincartilagesinthe pres-enceofother macromolecules( 17).

1.Abbreviations usedinthis paper: CS, chondroitinsulfate; AdiC4S-s,

disaccharide chrondroitin 4-sulfate stubs attached to core protein;

AdiC6S-s, disaccharide chondroitin 6-sulfate stubs attached to core protein; GuCl,guanidiniumchloride;HA,hyaluronicacid; KS, kera-tansulfate;OA,osteoarthritis; PG, proteoglycans; UA,uronic acid. J.Clin. Invest.

© The AmericanSocietyforClinicalInvestigation,Inc.

0021-9738/92/12/2268/10 $2.00

Theseproteoglycans constituteup to 10%of thewetweight ofcartilage. Because of their highcontentof

glycosaminogly-cans,they becomeveryhydrated andcanabsorbup to50 times theirweight ofwater. Butthis hydration is constrained by the tensile strength ofthecollagen network. Thisresultsina

swell-ing orosmoticpressurethatendowscartilage with its

compres-sive stiffness and hence its reversibledeformability (23-25).As

humanarticular cartilageages,there isanaccumulationof the GIglobular domain (26) and of smaller KS-richaggrecan mol-ecules(27, 28). These proteoglycandegradation products

in-creaseincontentmainly in the deepzone(28, 29). The degra-dation probablyoccurs over aperiod of time inthe

extracellu-lar matrix. Further evidence for extracellular degradation is providedby the progressive accumulation withaging of link

protein of reduced size (30). Thisprobably resultsinpartfrom the extracellularaction of proteinases,such as the

metallopro-teinase stromelysin (31 ).

There havebeenmanystudiesof the PGaggrecanin

osteo-arthritic articularcartilages. These havesometimes produced

contradictoryresults.Studies of molecular size andaggregation

with HA invariably involved measurementof uronic acidor

hexosaminecontent.Theseinvestigationsconcludedthat there is eithernochange (32), areduction (33, 34),or anincrease (35) in molecular size inosteoarthritis (OA).Lossof

aggrega-tion in advanced disease has been reported(36), but others foundnochange inaggregation (32).Thereis, however,a

re-duction in KS relativeto CScontent(33). Synthesis isoften

increased in earlydegeneration and later decreasedascartilage is further destroyed(37, 38). These observations indicate that therearesignificant changesinthis PG inOA,but theprecise

molecularevents involved andimplications ofthesechanges

areunclear, since these studieswere madeon manydifferent

specimens, concentrating on chemical analysesofthe chon-droitin sulfate of isolatedhighbuoyantdensityPG.Moreover,

inmostofthesestudies,withafew

exceptions

in which

synthe-siswasstudied (37, 38), therewasafailuretorelatemolecular changes to the different stages of degenerative change that Mankinet al.havedescribed (38). Cartilageswereusually inap-propriatelytreated as ahomogeneousgroup.

Wetherefore decidedto usenewlydevelopedquantitative immunochemical assays for the study of these molecules in total cartilage extracts so that aggrecan size, heterogeneity, composition, degradation products, andaggregationcould be clearly identified in normal and OAcartilagesthat had been

age- andtissue-matched andhistologicallygraded for degener-ative changes.Thestudypresented hereprovidesevidence for

even more molecularheterogeneity than hasbeenpreviously recognized. It also reveals biosynthetic and degradative changes in OA that are indicative ofan initial phase ofnet

degradationfollowed byaphase of extensive replacement with

new aggrecanmolecules thatarethenpartially degraded:

Even-tually, there isanetlossofthese molecules in advanced disease.

Methods

Cartilage. Humanarticular cartilage that appeared macroscopically normalwas obtainedatautopsy within 12 hof death from femoral condyles of adults (ages 47-81 yr) withnoevidenceof joint trauma, connective tissue abnormality,orarthritic disease. The cartilage was frozenat -20'C until processed. Human OAfemoralcondyles

ob-tainedatsurgeryforkneearthroplastywerefrozenimmediatelyafter

removal. The age range of the patientswas59-77 yr.Altogether, six normal and sevenOA joints were studied for total contents, molecular

sizes,andaggregation.

1-cm2specimens of the whole depth of cartilage ( 100-200 mg in wetwt) from thedistal weight bearing areas of the femoral condyles of normal and OA jointswereremoved, takingcareto exclude underlying subchondral bone. Insomeof the normal andOA samples,two speci-menswerepreparedatdifferent sites ofthe weight-bearing region ofthe samefemoralcondyletostudyvariations within different sites of the samejoint. Specimenswerefrozen sectionedat20 Mmwithacryostat

(TissueTEK II; Miles Scientific Div., Miles Laboratories Inc., Naper-ville, IL)beforeextraction. Fetal, epiphyseal, and articular cartilages wereobtained within 16 h (at 40C) of therapeutic abortion.

Histology.Afull depth sampleadjacenttothespecimenchosenfor

immunochemical and biochemical analyseswasremovedand fixed for 24 h in10%formalin in 25 mM sodium phosphate, pH 7.0, for histol-ogy. Afterwaxembedding,6-Amthick sections takenat twodifferent levelsperpendiculartothe articular surfacewerecutand stained with

hematoxylinand eosinorwithsafranin-O and fast green-iron

hema-toxylin.Sectionsweregraded accordingtoMankin'shistological

grad-ingsystem for OAcartilages (38).This methodassignedscores tothe intactness of thestructureof thetissue, the sizes andgroupingof the cells and the degree of safranin-O(proteoglycan) staining.Thescores weretotaledfor each sample andagradewasassigned.Since sections were cutup to,butnotincludingthetide mark, thiswasexcluded from thegrading.The maximumscorepossible, when tide mark isnot in-cluded, is 13.

ExtractionofPG.The20-Atmthick frozen sections from each

sam-plewereextractedwith30 vol (wt/vol) of4 Mguanidinium chloride

(GuCl)in 100 mMsodium acetate, pH 6.0(containing1mMEDTA, 1 mMiodoaceticacid,1 mMphenylmethylsulphonylfluoride with 5

,ug/mlofpepstatintoinhibitmetallo, cysteine, serine, and aspartate

proteinases, respectively)at4°C for 48 h with constantstirring.The PGextracts wereclarifiedby centrifugationinaSorval GLC-2B

centri-fugeat 3,500 rpm (2,000 g)for 15 min to removeany particulate

material.Supernatantandpelletwerestoredseparatelyat-20°C. Be-fore assay for uronic acid(see below) pelletswererinsed with distilled water, dissolved in 2 ml7NHClandincubatedat70°C for 20 min.

Isolationandpurificationofproteoglycanstandardsfor

radioimmu-noassays.Aggrecanmonomer waspreparedfromextractsof fetal and adultcartilages (10volof4 MGuCl withproteinaseinhibitors), by density gradient centrifugationingradientsof cesiumchloride,as de-scribedpreviously (39):-Fraction Dl ofhighestdensity (> 1.54)was isolated and usedasthesourceof aggrecan.

Hyaluronicacid.Aradiometricassay was usedas describedbythe manufacturer(PharmaciaFineChemicals,Uppsala, Sweden).

Uronic acid. Thiswasdeterminedbythe carbazolereaction(40)on

pellets(nonextractable)andafterdialysisofextractsagainstdeionized water.

Immunoassays. Radioimmunoassayswereperformed using anti-bodiestoKS(AN9P1)(21, 27),andtounsaturateddisaccharides pro-ducedbydigestionof CS with chondroitinaseABC,namely AdiC4S-s monoclonal 6B ( 17) and AdiC6S-s (monoclonal3B33)(16).Amouse monoclonalantibody846wasused that reactswith anasyet unidenti-fiedfetal-common epitopeonthenative CS of aggrecan

(chondroitin-ase ABClabile) (Poole,A.R., andA. Reiner,unpublished

observa-tion)andnot oncoreproteinaspreviouslyconcluded (18). Theradioimmunoassayswere asdescribedpreviously forAN9P1

(27),6B(17), and 846 (18). These used both '25I-labeledand unla-beled(fortracerand standard,respectively)humanfetal (846 and 6B) and adult(AN9P1and3B3)human aggrecan. Chondroitinase ABC-treated aggrecan (17)wasusedfortracerandstandardsfor assays with antibodies 6B and 3B3 to detect primarily unsaturated stubs of AdiC4S-s andAdiC6S-s, respectively,thatremain bound to core pro-tein afterdigestion with chondroitinase ABC ( 17) (Hirata, S., and

Table I. Summary ofRadioimmunoassayswithDifferentAntibodies

Standard and Radiolabled Standard

Antibody and recognition site sample treatments proteoglycans proteoglycan Precipitation step

3-B-3

MouseIgM recognizes AdiC6S-s left Chase ABC and '25I-HAPG/chase HAPG R191 (rabbit anti-mouse

oncoreprotein after chase ABC SDS ABC IgM) and protein A

treatment 846

Mouse IgM recognizes CS epitope SDS '25I-HFPG/native HFPG R191 (rabbit anti-mouse

expressed maximally on HFPG IgM and protein A)

6B

Mouse IgG recognizes AdiC4S-s Chase ABC and '25I-HFPG/chase HFPG ProteinA stubsleft on coreproteinafter SDS ABC

chase ABC treatment AN9P1

MouseIgGrecognizes aKS-rich SDS '25I-HAPG/ HAPG ProteinA

PG, native native

Additional abbreviations used are chase ABC, chondroitinase ABC; HAPG and HFPG, human adult and fetal aggrecan,Dl preparation.

Wherever necessary, all samples were dialyzed before assay (by

mi-crodialysis witha3,500Dcutoffmembrane)for 48 hatroom tempera-ture to remove 4MGuCl ( 150 vol of 200 mM sodiumacetate,pH 5.5

containing0.05%sodium azide). Subsequently, beforeassay,all sam-ples ( 100

_gl)

weretreated with 100gl200 mMTris/acetate, pH 7.5,

containing0.04 U/ml chondroitinase ABC (ICN, Montreal), and were incubated for 6 h at370C. Atthe end of incubation, 50Mil of 0.125% SDS in 100 mMTris/acetate buffer, pH 7.6,wasaddedtogive

afinalconcentrationof 0.025% SDS. This mixturewasincubatedat 80'C for 15 mintodisaggregateany PG aggregates(17)to ensurethat epitopesonPG moleculesweremaximallyexposedtoantibody. Hu-manadult and PG used for standardswerediluted inassociative col-umnbufferin the range 0.05-200Mg/mland treated in thesameway.

Radioimmunoassay buffer contained7.5 mMpotassium dihydrogen

phosphate, 134.9 mMdisodium hydrogen phosphateatpH 8.0,

con-taining0.1% BSA, 0.5% sodiumdeoxycholate,0.25% Nonidet P-40 and0.05% sodium azide. Resultsofallimmunoassayswereexpressed

inequivalents ofintact adultorfetalPG. Theuseofboth adult and fetal PGwasnecessitated bythefactthatantibodies6B and 846 show mini-malreactivitywith adultproteoglycans,whereas AN9PI and 3B3 ex-hibit minimalreactivitytofetal PG.

Asummary of the antibodies used and main features of their assays is shown in Table I. The results oftypicalinhibitionprofilesareshown inFig. 1.

Gel

chromatography

to determine

molecular sizes,

aggregation with hyaluronic acid, and

chondroitin

sulfate chain length

Dissociative conditions. Todeterminethe PGmonomersizes in total extractsofcartilage samples,500 Ml ofextractin4MGuCI with pro-teinase inhibitorswerechromatographedunder dissociative conditions onSepharoseCL-2B(100cmX 1.25cmindiameter,6ml/hflow rate, 1mlfractions, Pharmacia FineChemicals, Montreal, Canada), eluting

with4MGuCI, 50mMTris/HCI, pH 7.3. The columnwascalibrated with blue dextran (3 mg/ml) and glucuronolactone (2 mg/ml) for determination of void volume and total volume, respectively. Frac-tionsweremicrodialyzed against 150 volof 200 mMsodiumacetate, 0.05%sodiumazide, pH5.5for 48h at roomtemperature before assay. Associative conditions. TodeterminetheaggregatabilityofPG,HA wasadded in excesstotheextract to ensuremaximalreaggregation. Hyaluronicacid(highmolecularweight,humanumbilicalcord, kindly supplied byDr. P.J.Roughley,GeneticsUnit,ShrinersHospital,

Mon-treal)at 10%(wt/wt) of the PG content (the proteoglycan content in extracts of adult cartilage was estimated as being 6.5 times the total uronic acid (UA) content, which is about 15% ofthe total proteoglycan

forthis age group [ 39]),wasadded to the PG extract in 4 M GuCl. The

z 10

0 F f

FS m

z 4

10

2

m 6

z 0

m

6

E ₄

z 8

wo _AdiC6S-s

B0 /"

60 /

40 20

o ₈₄₆

;oi

40

o ~~/

zo

j/l

)o _AdiC4S-s

--30

/O/

20 i

oo-

KS,.-80.

/-60.

40U

20 /

' . -'

0.01 0.1 1 10o 100 1600 ggPG/ml

Figure1.Standard inhi-bitioncurvesfor the RIAofAdiC6S-s

(anti-body3B3)and

AdiC4S-s(antibody 6B) linkage regionsto coreprotein;

of CSepitope846 (anti-body 846)and of KS

(antibodyAN9P1).For further detailssee text

and TableI.

sample wasthendialyzed for 24 hat4VCagainst 150 vol 200mM

sodiumacetate, 0.05% sodiumazide, pH5.5. Thedialyzedsamplewas

chromatographed(500 ul)on aSepharoseCL-2B column(same di-mension as above) and eluted with 200 mM sodium acetate, 0.5% sodiumazide,0.25% BSA,pH5.5,atthesameflowrateandfraction volume as above.Aggregation was recorded as that PGelutingatequal toorlessthan Kav=0.15.

SepharoseCL-6B gelchromatographywasusedtoinvestigatechain length ofCS. Proteoglycan extracts weredialyzedovernight at4VC

against0.2 Macetatebuffer,pH5.0,and thendigestedin the presence of 5 mMcysteinewith0.2mg/mlpapainat370C. After4h,afurther

0.2 mgpapainwasadded.Thedigestionwasterminatedafter24 hby

theaddition of10 mMiodoacetamide.Digests (1 ml)were chromato-graphed on Sepharose CL-6B(100 cm X 1.25cm) in 0.2 Macetate

buffer, pH 5.5, at a flow rateof 6 ml/hand fraction vol of 1 ml. Chondroitinsulfatewasdetermined byUAanalyses.

Statisticalanalyses.Means±SD were recordedfornsamples. Stu-dent'sunpairedttest wasusedwhereindicated.P<0.05was

consid-eredsignificant.

Results

Histology. Histological examination of tissues permitted

as-signment ofa Mankin gradeto each of the normal and OA

tissues examined. Normal cartilage grades ranged from 0to3, whereasthosefromOAjoints ranged from 2to 13,according

to thedegree of degenerative change (Table II). Even within thesamespecimengrade couldvary.Therefore,thegrades

as-signedwererepresentative of the overall changes and indicated

the changes shown by the majority of the tissue. The grade assignedwaswithin therange0-6or7-13, permittingan

arbi-traryclassification of tissues into early (phase I) (grade 0-6)or

advanced(phase II) (grades 7-13) degenerative changes.

Efficiencyofproteoglycan extraction. Tomeasurethe

per-centageof PG extracted from frozen-sectioned articular carti-lage, UAcontentsof cartilageextractsandpelletswere

deter-mined(Table II). ThepercentageofUA extracted ranged from

Table II. TotalContentsofUA, theDifferentPG EpitopesandHAofNormal andOACartilageExtracts

Mankin Uronic

Sampleage grade AdiC6S-s &diC4S-s 846 KS acid HA

yr Normal N#l 47a 47b N#273a 73b N#379a 79b N#481 N#579 N#681 Mean SD Osteoarthritic

Mankin grade 2-6 phaseIOA

OA#l 74a 74b OA#2 70a OA#3 74 OA#4 74 Mean SD

Mankin grade 7-13 phaseIIOA OA#5 77a 77b OA#6 73a 73b OA#7 59a 59b OA#2 70b Mean SD

Ag/mgwet wt Ag/mgwet wt ng/mgwet wt Mg/mgwet wt Ag/mgwet wt jug/mgwetwt

0 0 3 32 3 2 2 2 2 S 6 9 7 10 8 13 10 7 56 78 36 15 17 8 56 17 10 33 25 17 14 10 15 44 20 14 45 10 11 16 7 6 26 17* 14 0.73 0.63 0.87 0.60 0.72 0.22 1.10 1.30 0.61 0.75 0.31 2.79 5.58 6.00 1.47 0.90 3.35* 2.34 1.51 3.20 4.50 4.10 1.27 1.26 2.93 2.68* 1.36 12 33 29 70 70 50 49 103 56 52 27 42 108 261 243 108 152* 95 144 4 225 216 306 324 261 255* 65 41 40 35 18 23 5 32 23 23 27 12 44 20 11 9 15 20 14 31 5 18 25 15 5 11 16* 10 3.17 4.16 2.76 2.52 3.03 1.75 3.37 1.61 1.71 2.68 0.87 2.99 2.60 3.51 1.78 3.14 2.80 0.66 3.91 2.20 2.28 2.78 1.37 1.40 3.51 2.49 0.98 0.53 0.49 0.60 0.58 0.48 0.48 0.36 0.34 0.40 0.47 0.09 0.45 0.30 0.41 0.43 0.40 0.40* 0.06 0.58 0.26 0.30 0.30 0.26 0.24 0.41 0.34* 0.12

56 to 82% (data not shown). Compared to normal cartilage (mean 68.6±8.7 SD), the percentage extracted from OA carti-lage with grades . 6 (mean 69.8±5.4 SD) and grades 7-13 (mean 73.2±8.4 SD)didnotchange.

Totalcontent of extractable proteoglycan epitopes, uronic acid andHAin normal and osteoarthritic cartilages and

Man-kin grade. The need to use different PG standards (fetal and adult), prevented a comprehensive analysis of the absolute

compositionsofthe different molecules identified in this study. But these analyses did permit comparisons to be made of the relative content of fetal epitopes (AdiC4S-s and 846), adult

epitopes(AdiC6S-s and KS), and epitope differences between healthy and OA cartilages. There were differences in epitope contentswhen samples from different sites on the same femo-ralcondyles were examined. But these contents were often

re-cognizably different fromthose in OA cartilages. In OA carti-lages, several trends(Fig.2) and significant differences (Table

II) were observed. UA content, AdiC6S-s, KS, and HA con-tentstended to decrease, whereas AdiC4S-s and epitope 846 increased with Mankin grade (Fig. 2). To permit a clearer

anal-ysis,thesechanges in OA wereclassifiedas phase I (grades 2-6) orphaseIIOA(grades7-12) (TableII). Chondroitin 6-sulfate stubs were significantly decreased in phase II more than in phase I. Keratan sulfate and HA contents were also signifi-cantly reduced in phase II more than in phaseIwhencompared with normals. Uronic acid content was not significantly

changedinphaseIandphaseIIOA. AdiC4S-s and epitope 846 were significantly increased in both phase I and II when com-pared with normals. Theseobservationsclearly reveal compo-sitional changes in OAcartilagesthat are relatedto thedegree ofdegenerationrevealed byhistological grading.

i ,*

'l

5 10 MANKINGRADE

15

Figure2.Relationshipofcontent of PGepitopesand HA to Mankin

grade.Values for normal(.)andOA samples (o)areindicated.To

indicategeneraltrends thebest fit linesareshownwhicharederived

fromregression analysesofthe all values.Sinceanalysesweremade

ofboth normaland OAsamples regression analysescannot be made.

A

0

'U

a 0

2

E

't

0

'U

0

0.

14 AdiC6S-s

12.

10

°8.

6

₁

4. 2 0

0.06 846

0.04

0.02

0.00

2. AdiC4S-s

20

15 K

10

S. o

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Kav

B

16 AdiC6S-s

14 12

10 B

6 4 2 0

0.06 846

0.04 0.02 0.00

1.2 AdiC4S-s

1.0

0.6-0.6.

0.4

0.2-0.0c

20 KSK

15 10

5--0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Kav

Figure 3. Gel chromatography of aggrecan PG populations from nor-malcartilageson Sepharose CL-2B under dissociative conditions (with 4 M GuCl) to demonstrate heterogeneity of aggrecan in a 81-yr-old,grade 1(A) and a 79-yr-old grade 2 (B) are shown.

Identification

of

different

proteoglycan populations in

normal and osteoarthritic cartilages using dissociative

(4MGuCl)

and

associative

chromatography on

Sepharose

CL-2B

The percentage recoveryof PG epitopes fromchromatography undereither conditionsrangedfrom70 to85%.

Monomer size. Column fractions of chromatography

underdissociativeconditionswereassayed with each antibody

toinvestigatemonomersize,and the resultswereexpressedas

intact PGequivalents. Thechromatograms forrepresentative

normal samplesare shown in Fig. 3, and representative OA

samplesareshown inFig. 4.Theseanalysesrevealed the

pres-enceof differentpopulationsof PG innormal and OA carti-lagescharacterizedbydifferencesincompositionandsize.For

example,in normalcartilage,thelargestpopulationelutedat a Kavy 0.35. It wasrecognized usingtheanti-AdiC6S-s anti-body.ElutingatapproximatelythesameKav, and present ina

relatively very small amountwas apopulation bearing epitope

846.Slightlysmallerthan theAdiC6S-spopulationwasa popu-lationelutingatKav- 0.45. Thiswasrecognizedbythe anti-AdiC4S-s antibody, 6B. This suggeststhattheAdiC4S-s

popula-tionis smaller than and distinct fromtheAdiC6S-s and

epitope

846population(s). ElutingatKavy 0.6was aPG

population

recognizedby the anti-KSantibody,AN9P1. This isthesmall

KS-rich

PG population previously described by us (27) and

URONIC ACID

4' 0

0 0

.0

3'

2

0 0

80 AdIC6s-6

60

0 0

40

20 8 0

0. 0 ₀

0~~~~~~

7 AdIC4S-s 6 0

5-0

4 0

3 o 8

2=

1 !. 0 00

'i

0

I

-0.2

I

F

'a_=L

i

If

U

v

Kav

Figure5. Method of de-terminationof relative hydrodynamic sizes of

aggrecanpopulations by assigning arbitrary Kavvalues, aboveor

belowwhichthe

per-centageofthetotal

con-tentintheshadedarea wasrecorded.The ex-ample givenisOA

sam-ple 73-yr-old,grade 10.

Figure 4. Gel

chromatogra-phyofaggrecan populations

from OAcartilageson

Seph-aroseCL-2B under dissocia-tive conditionsto demon-strateheterogeneity of aggre-can inphaseI:(A)a

70-yr-old, grade2; (B)

74-yr-oldgrade 5;andphaseII:(C)

a59-yr-old, grade13; (D)a

73-yr-old,grade 9.

recognizedas adegradationproduct ofaggrecancontainingthe KS-rich region. In general, the elution for AdiC6S, AdiC4S,

and 846 was often but not always unimodal. A small "shoulder" ofhighermolecularweightPGwasseen in the

elu-tionprofilefor KS.SimilarpopulationswereidentifiedinOA

cartilages, irrespective of grade.

To compare these populations in normal and OA carti-lages, thefollowing approachwasused asoutlined in Fig. 5.

Therelativecontentsofepitopes in columnfractions less or greaterthanagiven Kav, whichwasfixed for each epitope in normaland OAcartilages,weredeterminedand expressedas a percentageofthe totalepitopecontentin all columnfractions. Therefore, for epitopes AdiC6S-s, 846, and AdiC4S-s (Kav

<0.35)andfor KS (Kav20.55),relativehydrodynamicsizes

werecalculatedas percentages.Theseresults are shownin

Ta-ble III. Thedata shownaretherefore forthepercentage amount

ofa PG ofa given hydrodynamic size. Examination ofthe

differentpopulationsin normalcartilages confirmedthe pres-ence oftwo populationsof similar size containing zAdiC6S-s

and846epitopes.Theseweresignificantlylarger than the

popu-lation containing AdiC4S-s (Table III). The presence ofa

much smallerKS-richpopulation is againapparent.

When the sizesofthedifferentPGin all the OA samples

were compared with the normal samples, there was no

evi-denceforanysignificant differences. Butwhen the OA

speci-mens were divided into two groups by their grades, phase I (grade <6) and phase II (grades 2 7),significantdifferences

wereobserved(Table III). TheequivalentPG proteoglycans in

phaseI OA weregenerally smaller than innormalcartilagebut thedifferenceswere notsignificant. But in phase IIOA, mole-culesbearing epitopesAdiC6S-s,846,andAdiC4S-swere signif-icantly larger in sizethan thecorrespondingPG in phase I OA (grades 7-13). The predominant small KS-rich PG species

showednosignificantchanges (TableIII),although there was a

A B

3

2

E

ED

Tol

0

E

E'

a

I

a

a

I

0

E-LL I

C 'U

a

LL I

CL

'-X,

0.

RI

TableIII. Comparison ofRelative HydrodynamicSizesof DifferentAggrecanPopulationsinNormal, PhaseI,andPhase II

OA, DeterminedasShownin Fig. 5

Epitope

Tissue and mean±SD Significance

epitope (Kav) (%) P values

AdiC6S-s<0.35

Normals 53.0±12.7 0.1889

Phase I OA 40.7±12.7

PhaseIIOA 62.4±12.6 0.0372*

846<0.35

Normals 47.2±11.5 0.0551

PhaseIOA 32.3±6.4

Phase II OA 62.4±17.8 0.0145*

AdiC4S-s<0.35

Normals 31.4±9.6 0.1916

PhaseI OA 24.3±1.5

PhaseIIOA 44.2±11.6 0.0121*

KS>0.55

Normals 50.6±4.7 0.1114

PhaseI OA 58.2±7.8

Phase II OA 43.8±15.1 0.1297

Pvaluesareshown indicating significant differences(*)between phase

I(Mankin grades 2-6) and phaseII(Mankin grades 7-13) OA. Five normal, fourphaseI,and five phase II sampleswereanalyzed. In

ad-ditiontothestatisticalanalysesindicated, analysesof normal

popu-lations richinCAweremade: 846wasnotsignificantlysmaller in size than AdiC6S-s(P=0.4708),but AdiC4S-swassignificantly differentinsize than 846(P=0.0461)andA\diC6S-s(P=0.0162).

trend towardsadecreaseinsizeinphaseIOAandanincrease insize inphaseII OA.

Chondroitin sulfate chain length. To determine whether thesechangesinPGmonomersize resulted fromachange in

synthesis resultinginaltered CS chain length, PGweredigested with papain and chromatographed under associative condi-tionsonSepharose CL-6B and analyzed for UA. The resultsare

showninFig. 6. Therewas noevidence foranychange in chain

length in OA cartilage of various Mankin grades.

Monomer aggregation. Analysisofthesepopulations under associative conditions revealed that in normal cartilages,

be-tween50and 59%ofeachofthesepopulations aggregatedwith

the exception of the population identified by epitope 846, which showed almost complete aggregation (TableIV,Fig. 7). Therefore, there existtwopopulations of the largest size,one

identifiableby the AdiC6S-s stubson coreprotein anda more

minorpopulation bearing epitope 846.

Analysis of aggregation of PG fromOAcartilages withHA

revealed similar aggregation for AdiC6S-S and KS epitopesto

that seen in normal cartilage (Table IV). But epitopes 846

(phaseIand II) and AdiC4S-s (phaseIIonly)exhibited much

lessaggregation inOA cartilage.

Discussion

By using immunochemical methodstostudyPGintotal

carti-lageextracts wehave characterized recognizablydifferent

ag-0.16 0.12

0.08

0.04 0.1;

0.01

0.04 0.14

0.1: 0.01 0.1

0.1

0.0

0.0 0.1

0A

0.c 0.c

0.1

C 0.1

0,0.C

0.1

0.'

0.4

0.10.1

SAMPLEAGE /MAWN GPAE

N47/1-2

N79/1

N81/2

QA74/2

OA74/5

OA74/6

OA77/9

OA73/10

OA59/13

-0.2 0 0.2 0.4 0.6 0.8 1 1.2

Kav

Figure 6. AnalysisonSepharose CL-6B of hydrodynamic sizes of CS

chains in normal(N) and OA cartilages. Theageand grade of the sampleis indicated.

TableIV.PercentageAggregationofProteoglycans BearingtheEpitopeIndicated

Epitopes

AdiC6S-s KS AdiC4S-s 846

Normals

Mean 59.0 50.0 52.0 96.1

SD= ±12.6 ±10.0 ±5.3 ±3.5

(n=5) (n=5) (n=4) (n=6)

Phase IOA (grades 2-6)

Mean 65.0 61.3 49.0 58.3

SD= ±6.9 ±4.9 ±7.0 ±7.6

(n=₄₎ (n=₃₎ (n=₂₎ (n=3)

Phase II OA (grades 7-13)

Mean 56.0 48.7 33.5 58.5

SD= ±6.0 ±9.6 ±2.1 ±13.4

(n=3) (n=4) (n=2) (n=2)

Aggregationwasmeasuredasthat PGelutingat Kav=0.15.

o0 rrrrTTrrT

so

.TrYr

6 i2 )8

4 12

48

12 D8 D44 D.2

- ).1-0 12

)8-D8

'

12 08 04

rprIrrr1rjrr-a Figure 7.Gel

chroma-cs tographyon Sepharose

w

00.6- CL-2B under associative

m_

conditions in the

pres-0.4 enceofexogenousHA

E X of aggrecan populations

-0.2- from twonormal

carti-lagestodemonstrate

ag-.o-

n

gregation

-

ofepitope

1.0 846-containing

mole-M_{0.8- cules with hyaluronic}

W acid in(a)a49-yr-old

0 0.6- female and(b) a

54-yr-a.

0. old male. Percent of

ag-E 0.4- >

gregation (shaded

area)

is(a)91.8 and(b) 92.1.

0.- _{1Recoveries from}

col-0

,

umn fractions of added

-0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 sampleswere (a)112%

Kav and(b)92%.

grecan

populations

in normal and OA

cartilages.

The

previ-ously

described smaller KS-rich molecules

(27,

29, 41)

are

thought

to represent

degradation products

derived from the

larger

populations

thatareretained in

cartilage (in

partat

_least)

aspartoftheir

aggregation

with

hyaluronic

acid. In

_addition,

largerAdiCS-s-rich

populations

wereobservedwhichhavenot

previously

been described. Proteoglycans

bearing

AdiC6S-s

(the

linkage region

ofCSto core

protein)

were

_larger

than those inwhich AdiC4S-Swasthe

predominant "linkage"

sulfation.

Molecules

_{bearing epitope}

846

(a

veryminor

population

com-paredwith those

bearing

C4S

(the

samestandardwasusedfor these two

assays)

cochromatographed

with

PG-bearing

AdiC6S-s. But innormalcartilages,moleculesbearing epitope

846 showed almost

complete aggregation

with HA, unlike those

_bearing

AdiC6S-s. This indicated the presence offour

recognizable

populations:twoof similarlargesize,an interme-diate size AdiC4S-S-rich

population,

and the small KS-rich molecules. Thesedifferences in the largemolecules mayalso result fromdifferentdominant CSsubstitutions,assuggested

by

earlierstudies frommanylaboratories.

This

heterogeneity

displayed

bytheselarger

molecules,

re-vealedbytheuseof

antibodies,

was not

previously

detectable

by

thetraditionaluronic acidorhexosamine

_{analyses (32-35,}

41

).

Itcould result inpartfrom differences in sulfation of CS

linkage regions along

theCS-attachmentzone.The

_recognition

oftwodistinct domains in thecore

_protein

of the

CS-attach-ment

region

of human aggrecan

(

13)

has ledtothe

assignment

oftwo

adjacent

CS

domains, CSl

and

_CS-2,

the latter

being

closest to the G3globulardomain. It is conceivablethat the

AdiC6S-sand AdiC4S-s maybe found predominantlyin the CS2andCS1 domains, respectively.Thus,ifallor partofthe

CS2 domainwas removed by proteolysis inthe extracellular

matrix,

thiswouldresultin a smallermolecule enrichedin the

CS1domainandAdiC4S-s linkages.In normalcartilage, mole-cules

bearing

epitope846exhibit essentiallycomplete

aggrega-tionwith HA. Fewnonaggregatingmoleculeswere observed.

This wasdue tolack ofsensitivity oftheimmunoassay. It is

possible,

therefore, in view of their large size and complete

aggregation,

that moleculesbearingepitope846represent

re-centlysynthesizedmolecules. If theseepitopes arepresent on

thesamemolecule,theverylowcontentofepitope846 relative

to AdiC4S-s suggeststhat theCS chains bearingthisepitope

may be in a partofthemoleculethatisreadilyremovedfrom

PG inthe extracellularmatrix: This would arguablybecloseto

the G3 domain(since there isnoevidence for CS in the G 1 and G2domains). Alternatively, epitope846maybepresenton a

distinctveryminorpopulation of PG. Clearly further careful experimental work isrequiredtoinvestigatefurther the struc-turesandheterogeneity ofthese molecules.

Inany studyof articularcartilages in OA,itis extremely importantto takenoteof the wide variation in degenerative

changes that are a feature of this disease. Althoughit is not

perfect, the grading system introducedby Mankin etal. has permitted the identification ofcompositionalandbiosynthetic

changes in PG in OA.Thus, inearlydisease(low grade)

synthe-sisofPGwasfoundtobeincreased,whereas in advanced

dis-easeitwasdecreased(37,38).This loss ofsynthesis

accompa-nieda netlossofPG thatwasonlyclearlyobserved in advanced disease. Usingthisapproach,weexamined PGsize, composi-tion,aggregation,andcontent asthediseaseprogresses.

AllourPGpopulationswereobserved in OAcartilage

irre-spective of grade. However, the aggregation differences

be-tweenthepopulations bearingepitope846andAdiC6S-swere

nolongerobserved.Changeswerenoted in thesetwoepitopes

indicative of changes in CScomposition.AdiC6S-scontent was

reduced, whereasepitope846 andAdiC4S-scontents were in-creased. This probably occurred when there was no overall change in uronic acid content, although KS content was

re-duced. Theassignmentoftwophasesof OApermittedtheclear recognition of pathology-related changes that have not

previ-ously beenreported.InphaseI, there is evidence forageneral reduction (althoughnotsignificant)in thesizesofall the PG

populationscomparedtotheir normalcounterparts.Since this

isaccompaniedby the lossof AdiC6S-s, thismayindicatethe lossof the G3 and CS2 domains, accordingto ourproposed

assignmentof AdiC6S-stotheCS2 domain. This reduction in size was not accompanied by a change in CS chain length, suggesting that increased proteolysis occurred involving the coreprotein. Moreover, thelossofaggregationshown bythe

population bearingepitope846 is alsoindicative ofproteolysis.

InphaseIIof the disease(grades7-13),whenfibrillation is

moreextensive andcartilage becomesverydegenerate in ap-pearance,the CS-rich and KS-rich PG populationswerestill detected. But theywere all larger than their normal counter-parts and(except for the KS-rich molecules)weresignificantly larger than theirphase Icounterparts. These increases in size didnotresultfromchanges in CS chainlength, (observed

previ-ously, reference 42), suggesting that these larger molecules

contained longercoreproteinsand were less degraded.

Collectively, these observations indicate that in OA, two

phasescanbeidentifiedin thepathological changes involving aggrecan that occur in the articular cartilages. The initial phase I is characterized by a general degradation of PG molecules

combined with replacement with new molecules of altered

composition, as indicated earlier (37, 38). Then in phase II there is extensive replacement of degraded molecules with

larger compositionallydifferentmolecules, probably as a result

of the increased synthesis described previously. Evidence for

contentof thenative CSepitope846: The latter wouldbe

ex-pected

if,

as wesuggest,

epitope

846 is presentonrecently

syn-thesized molecules and is locatedon CS close tothe G3 do-main. Asimilarincrease incontentof another native CS

epi-tope

recognized by

monoclonal

antibody

3B3 has also been observedin

experimental

OAindogs (43).Since these native CS

epitopes

are most

commonly

found in fetal and

embryonic

tissues,

their reappearance in OA suggests thatchangesin the environment of the

chondrocyte

cause

changes

in synthesis

and CS

assembly.

Eventually,

as aresultofextensivedamagetothecartilage,

PGis lost. This

probably

resultsfrom extensivedamagetothe

collagen

fibrillar

organization

we described

previously

(44)

leading

totheloss ofHA(45, 46)and of PG(37, 38)observed in the present

study

aswellas

previously.

But,contraryto

com-mon

belief,

before this

happens,

ourobservations indicatethat OA

cartilage

hasaconsiderable

capacity

forextensiveturnover

ofthe PG component of the extracellular matrix. Ifcollagen

damage

could be

prevented

and its

repair promoted

then it may be

possible

for

cartilage

to

repair

itself.

Acknowledaments

Wethank Renee Leiberman andAudreyWheeler,whoprocessed this

manuscript,and MarkLepikand JaneWishart,who prepared the

fig-ures.

Thestudywasfundedby the Medical Research Council of Canada andtheShrinersofNorthAmerica(toDr. A.Robin Poole).

References

1. Sokoloff, L., and A. J. Hough, Jr. 1985. Pathologyofosteoarthritis. In

ArthritisandAllied Conditions. ATextbookofRheumatology. 10th ed. D. J.

McCarty, editor.Lea andFebiger, Philadelphia.1377-1399.

2. Kempson, G. E. 1979.Mechanical properties ofarticular cartilage. In Adult

ArticularCartilage, M. A. R.Freeman,editor. PitmanBooks Ltd. Tunbridge Wells, Kent, England. 333-414.

3. Mow,V.C.,D. S.Howell,and J. A.Buckwalter. 1990. Structure and

functionrelationshipsof articularcartilageand theeffectsofjoint instabilityand trauma oncartilage function. InCartilage Changes in Osteoarthritis. K. D. Brandt,editor. IndianaUniversity School of Medicine, Indianapolis,IN.22-42. 4. Mendler,M.,S.G. Eich-Bender, L.Vaughan, K.Winterhalter, and P. Bruckner. 1989.Cartilagecontains mixedfibrils of collagentypesII, IX,and XI. J.CellBiol.108:191-197.

5.Eyre, D. R., S.Apone,J. J.Wu, L. H.Ericsson,and K. A. Walsh. 1987.

Collagentype IX:evidence forcovalentlinkagestotype IIcollagenincartilage.

FEBS(Fed.Eur.Biochem. Soc.)Lett.220:337-341.

6.vanderRest, M.,and R.Mayne.1988.TypeIXcollagenproteoglycan from cartilageis covalently cross-linkedtotype IIcollagen.J.Biot. Chem.

263:615-1618.

7.Poole,A.R., I.Pidoux,A.Reiner,andL.Rosenberg. 1982.An immuno-electronmicroscope studyof theorganizationofproteoglycanmonomer, link

protein,andcollagenin the matrixof articularcartilage.J.CellBiol.93:921-937. 8.Heinegard, D.,andV.C. Hascall. 1974.Aggregationofcartilage

proteogly-cans.III.Characteristics oftheproteinsisolatedfromtrypsin digestofaggregates. J.BioL. Chem.249:4250-4256.

9.Tang, L-H.,L.Rosenberg,A.Reiner,andA.R.Poole.1979.Proteoglycans

from bovine nasalcartilage.Properties ofasoluble formof linkprotein.J.Biol.

Chem.254:10523-10531.

10.Mow, V. C., W. Zhu, W.M.Lai,T. E.Hardingham,C.Hughes,and H. Muir. 1989. Theinfluence of linkproteinstabilizationonthe viscometric proper-tiesof proteoglycan aggregate solutions.Biochim.Biophys.Acta 992:201-208.

11.Neame, P. J., J. E.Christner,and J.R.Baker.1987.Cartilage

proteogly-canaggregates. The linkproteinandproteoglycanamino-terminalglobular do-mains have similarstructures. J.BioL.Chem. 262:17768-17778.

12. Doege,K., M. Sasaki, E. Horigan, J. R. Hassell, and Y. Yamada. 1987. Complete primary structure of the rat cartilage proteoglycan core protein de-ducedfromcDNAclones. J.Biol. Chem. 262:17757-17767.

13.Doege, K. J., M. Sasaki, T. Kimura, and Y. Yamada. 1991. Complete

codingsequence and deducedprimarystructureofthehuman cartilage large

aggregatingproteoglycan aggrecan. Human-specific repeats and additional alter-nativelysplicedforms.J.Biol.Chem. 266:894-902.

14. Antonsson,P., D.Heinegard, and A. Oldberg. 1989. The keratan

sulfate-enrichedregion of bovine cartilageproteoglycanconsists of a consecutively re-peatedhexapeptide motif.J.Biol. Chem.264:16170-16173.

15.Halberg, D. F., G. Proulx, K. Doege, Y. Yamada, and K. Drickamer. 1988. A segmentofthecartilage proteoglycan core protein has lectin-like activity.

J.Biol. Chem.263:9486-9490.

16. Caterson, B., J. E. Christner, J. R. Baker, and J. R. Couchman. 1985. Production and characterization of monoclonal antibodies directed against

con-nective tissueproteoglycans.Fed.Proc. 44:386-393.

17.Matsui,Y., M.Alini,C.Webber, and A. R. Poole. 1991. Characterization

of aggregatingproteoglycansfromtheproliferative,maturing, hypertrophic and

calcifying zones of the cartilage physis (growth plate). J. Bone Jt. Surg. 73A: 1064-1074.

18.Glant,T.T., K. Mikecz, P. J. Roughley, E. Buzas, and A. R. Poole. 1986. Age-related changesin protein-related epitopes of human articular cartilage pro-teoglycans.Biochem.J. 236:71-75.

19. Sorrell, J. M., A. M. Lintala, F. Mahmoodian, and B. Caterson. 1988. Epitope specific changes in chondroitin sulfate/dermatan sulfate proteoglycans asmarkers in thelymphopoieticandgranulopoietic compartments ofdeveloping bursaeofFabricus.J.Immunol. 140:4263-4270.

20.Caterson, B., J. E. Christner, and J. R. Baker. 1983. Identification of a monoclonalantibodythatspecificallyrecognizes corneal and skeletal keratan

sulfate.J.Biol. Chem. 258:8848-8854.

21.Poole, A. R.,C.Webber, A.Reiner,and P. J.Roughley. 1989. Studies of a monoclonalantibody to skeletal keratan sulfate. Importance ofantibody valency.

Biochem.J.260:849-856.

22. Mehmet, H., P.Scudder, P. W. Tang, E. F. Hounsell, B. Caterson, and T.

Feizi.1986. Theantigenic determinantsrecognized by three monoclonal

antibod-iestokeratansulfate involvesulphated hepta or largeroligosaccharides of the

poly-(N-acetylactosamine)series.Eur. J.Biochem. 157:385-391.

23.Maroudas, A. 1980.Physicalchemistry ofarticular cartilage and the inter-vertebraldisc. InThe Joints andSynovialFluid. Vol. II.L.Sokoloff, editor.

AcademicPress, New York.239-291.

24. Kempson,G.E., H.Muir,M. A.R.Freeman, and S. A. V.Swanson. 1970.

Correlationsbetween thecompressive stiffnessandchemicalconstituentsof hu-manarticularcartilage. Biochim.Biophys.Acta.215:70-77.

25.Lai,W.M.,J.S.Hou,and V. C. Mow. 1991.Abiphasictheory of the

swellinganddeformation behavioursofarticularcartilage. J.Biomech. Eng.

113:245-258.

26.Roughley,P.J., R. J.White,andA.R.Poole. 1985.Identificationof a

hyaluronic acid proteinthatinterferes withthepreparation of high-buoyant

den-sity proteoglycanaggregatesfromadult humanarticularcartilage. Biochem.J 231:129-138.

27. Webber,C.,T. T. Glant,P.J.Roughley,and A. R. Poole.1987.The

identificationandcharacterization oftwopopulations ofaggregating

proteogly-cansofhigh buoyant density isolated from post-natalhumanarticularcartilages of differentages.Biochem.J.248:735-740.

28.Franzen,A., S.Inerot,S-O.Hejderup,and D.HeinegArd.1981. Variations in thecomposition of bovinehiparticularcartilagewithdistance fromthe

articu-larsurface. Biochem.J. 195:535-543.

29.Bayliss,M.T.,and S.Y.Ali. 1978.Age-related changesinthecomposition

and structure of human articularcartilage proteoglycans. Biochem.J. 176:683-693.

30.Mort,J.S.,A.R.Poole,and P. J.Roughley. 1983.Age-related changesin thestructureof proteoglycanlinkproteininnormal humanarticularcartilage.

Biochem. J. 214:269-272.

31. Nguyen,Q.,G.Murphy,P. J.Roughley,and J. S.Mort.1989. Degrada-tion ofproteoglycanaggregateby acartilage metalloproteinase. Biochem. J. 259:61-67.

32.Brocklehurst, R.,M.T.Bayliss,A.Maroudas,H. L.Coysh,M.A.R.

Freeman,P. A.Revell,andS.Y.Ali. 1984. Thecompositionofnormal and

osteoarthriticarticularcartilagefrom human kneejoints.J.BoneJt.Surg.Am.

Vol.66A:95-106.

33.Sweet,M. B.E.,E.J-M.A.Thonar,A.R.Immelman,and L.Solomon. 1977. Biochemical changes in progressive osteoarthritis. Ann. Rheum. Dis. 36:387-398.

34. Vasan, N. 1980.Proteoglycans in normalandseverely osteoarthritic hu-mancartilage. Biochem.J. 187:781-787.

36.Palmoski, M.,and K. Brandt.1976.Hyaluronate-binding by

proteogly-cans:comparison of mildlyandseverelyosteoarthritic regions of humanfemoral cartilage. Clin. Chim.Acta.70:87-95.

37. Thompson, R. C., and T. R.Oegema,Jr. 1979. Metabolic activity of articular cartilage in osteoarthritis. J. BoneJ1.Surg. 61 A:407-416.

38.Mankin,H. J., H.Dorfman,L.Lippiello, and A. Zarins. 1971.

Biochemi-cal and metabolic abnormalities in articular cartilage from osteo-arthritic human hips. II. Correlationofmorphology with biochemical and metabolic data. J. Bone Jt.Surg. Am. Vol. 53A:523-537.

39. Roughley, P. J., and R. J. White. 1980. Age-related changes in the struc-tureoftheproteoglycansubunits fromhumanarticular cartilage.J.Biol. Chem.

255:217-224.

40.Bitter, T., and H. M. Muir. 1962. Amodifieduronic acid carbazole

reac-tion.Anal.Biochem.4:330-334.

41.Bayliss,M. T., M. Venn, A. Maroudas, and S. Y. Ali. 1983. Structureof

proteoglycansfrom differentlayers of human articular cartilage. Biochem. J. 209:387-400.

42.Bollet, A. J., and J. L. Nance. 1966. Biochemicalfindingsin normal and

osteoarthritic articular cartilage.II.Chondroitin sulfate concentrationandchain

length, water and ash content. J.Clin.Invest.45:1170-1177.

43. Caterson, B. G., F. Mahmoodian, J. M. Sorrell, T. B. Hardingham, M. T. Bayliss, S. L. Carney, A. Ratcliffe, and H. Muir. 1990. Modulation of native

chondroitin sulphate structure in tissue development and in disease. J. Cell

Science. 97:411-417.

44.Dodge, G. R., and A. R. Poole. 1989.Immunohistochemical detection

andimmunochemicalanalysis oftype II collagendegradationin human normal,

rheumatoid,andosteoarthriticarticularcartilagesand inexplantsofbovine artic-ularcartilagecultured withInterleukin1.J.Clin.Invest.83:647-661.

45. Thonar, E. J-M. A., M. B. E. Sweet, A. R. Immelman, and G. Lyons. 1978. Hyaluronate inarticularcartilage age-related changes.CalcifTissue Res.

26:19-21.

46.Sweet, M. B. E., E. J.-M. A. Thonar, A. R.Immelman,and L.Solomon.

1977. Biochemical changes in progressive osteoarthritis. Ann. Rheum. Dis.