The structure of aggrecan fragments in human synovial fluid Evidence for the involvement in osteoarthritis of a novel proteinase which cleaves the Glu 373 Ala 374 bond of the interglobular domain

The structure of aggrecan fragments in human

synovial fluid. Evidence for the involvement in

osteoarthritis of a novel proteinase which

cleaves the Glu 373-Ala 374 bond of the

interglobular domain.

J D Sandy, … , P J Neame, L S Lohmander

J Clin Invest. 1992;89(5):1512-1516. https://doi.org/10.1172/JCI115742.

Synovial fluid was collected from patients with recent knee injury and from patients with early or late stage osteoarthritis. Chondroitin sulfate-substituted aggrecan fragments present in these fluids, and in normal bovine synovial fluid, were purified by cesium chloride

gradient centrifugation, enzymically deglycosylated and fractionated by gel filtration on Superose-12. Each sample contained two major aggrecan core protein populations with apparent molecular masses of approximately 90 kD and 150 kD. For all samples, NH2-terminal analysis of both populations gave a single major sequence beginning ARGSV. This NH2 terminus results from cleavage of the human aggrecan core protein at the Glu 373-Ala 374 bond within the interglobular domain between the G1 and G2 domains. Cleavage at this site also occurs during control and interleukin-1 stimulated aggrecan catabolism in bovine cartilage explant cultures (Sandy, J., P. Neame, R. Boynton, and C. Flannery. 1991. J. Biol. Chem. 266:8683-8685). These results indicate that the major

aggrecan fragments present in both osteoarthritic human synovial fluid and in normal bovine synovial fluid are large, being composed of a short NH2-terminal stretch of the interglobular domain, the G2 domain, the keratan sulfate domain, and variable lengths of the chondroitin sulfate domain(s). We conclude that the release of aggrecan fragments from articular cartilage into the synovial fluid seen at all stages of human osteoarthritis (Lohmander, L. S. 1991. Acta […]

Research Article

Find the latest version:

The

Structure

of Aggrecan Fragments in Human Synovial Fluid

Evidence for the Involvement in

Osteoarthritis

of a Novel Proteinase Which Cleaves the Glu 373-Ala 374 Bond of the

_{Interglobular}

Domain

John D.Sandy,*tCarl R.Flannery,*Peter J.Neame,**andL.StefanLohmanderl

*ShrinersHospitalforCrippled Children, Tampa Unit, Tampa, Florida 33612; *Department of Biochemistry and Molecular Biology, School ofMedicine, University ofSouth Florida, Tampa, Florida 33612; and IDepartment ofOrthopedics, University Hospital, S-22185 Lund, Sweden

Abstract

Synovial fluidwascollected from patients with recent knee

in-juryand from patients withearly or late stage osteoarthritis.

Chondroitin sulfate-substitutedaggrecan fragments present in thesefluids,and in normal bovine synovialfluid,werepurified bycesium chloride gradientcentrifugation, enzymically degly-cosylated and fractionated by gel filtration on Superose-12.

Eachsample contained twomajoraggrecan coreprotein

popula-tionswithapparentmolecularmassesof 90 kD and 150 kD.

For all samples, NH2-terminal analysis of both populations

gave a single major sequence beginning ARGSV. This NH2

terminus results fromcleavageof the human aggrecancore pro-tein at the Glu373-Ala374 bond within theinterglobular do-mainbetween the G1 andG2 domains.Cleavage at this site also

occursduringcontrol andinterleukin-1stimulated aggrecan

ca-tabolism in bovine cartilage explant cultures (Sandy, J., P.

Neame, R. Boynton, and C. Flannery. 1991. J. Biol. Chem.

266:8683-8685).

These resultsindicate that the majoraggrecan fragments

present in bothosteoarthritichumansynovial fluidand in

nor-malbovine synovial fluidarelarge,being composed ofashort

NH2-terminalstretchof theinterglobular domain,the G2 do-main, the keratan sulfatedomain,and variablelengthsof the

chondroitinsulfatedomain(s).Weconclude that the release of aggrecanfragments from articularcartilage into the

synovial

fluidseenatall stages of human osteoarthritis(Lohmander,L.

S. 1991.ActaOrthop. Scand.

62:623432)

is

promoted by

the action ofanormalcartilage

proteinase

which cleaves the Glu 373-Ala 374 bond of theinterglobular domain. (J. Clin.Invest. 1992.89:1512-1516.) Keywords:articularcartilage* proteo-glycan*coreprotein*cartilage

proteinase

*

NH2-terminal

se-quence

Introduction

Thedisease mechanisms activeinosteoarthritis

(OA)'

are un-clear,butchanges in thebiochemicalandbiomechanical prop-erties ofjoint cartilage, changesinchondrocytematrix

synthe-AddresscorrespondencetoDr. J. D.Sandy, ShrinersHospitalfor

Crip-pledChildren,TampaUnit,Tampa, FL 33612.

Receivedfor publication 15 October1991 and in revisedform8 January 1992.

1. Abbreviations used in ?his paper: CS, chondroitin sulfate; OA,

osteoarthritis.

sis, and, finally, a gradual destruction of the matrix are

charac-teristicofthediseaseprocess. This sequence of events has been best documented by the use ofanimal models ofposttraumatic OA (1-3). Cartilage proteoglycan fragments are released to

jointfluid afterkneeinjury and in early stages ofboth

posttrau-maticandprimaryOAinthe human(4-6). An important role

for matrix metalloproteinasesin both normal turnover of

con-nective tissuematrix and in the tissue destruction seen in, for

example,OA has been suggested (7, 8), and an imbalance be-tweentissuemetalloproteinases and inhibitors has been demon-stratedinanimalmodel OAcartilage (9),injoint fluids from patients withrecentjoint injury,and in OA joint fluids (10,11).

Further, increased expression of mRNAs for collagenase,

stro-melysin, and metalloproteinase inhibitor has been shown in

synovial cells stimulated with interleukin-1 and in synovial

cellsfrom rheumatoid orosteoarthritic joints(12-15). How-ever, nodefinitive evidencehas yet been presentedwhich dem-onstrates that theseenzymesaredirectly involvedincartilage matrix breakdownin OA. Adetailed characterization ofmatrix fragmentsreleasedfromjoint cartilage aftertraumaand in OA

ofthe human couldhelp identifythedegradative mechanisms. Characterization ofaggrecan fragments released from

bo-vine explants treated with interleukin-1 in vitrohas demon-stratedamajor cleavage site withintheinterglobular domain of

the proteoglycan core protein which releases large aggrecan

fragments fromthetissue(16). In the presentinvestigation,we

have isolated aggrecan fragments from joint fluids obtained frompatients withrecent kneejoint injuryorearly or late stage knee OA. We show that themajoraggrecanproductspresent in thesesynovial fluidsarerelativelylarge andarecomposed ofa

segment ofthe

interglobular

domain attachedto the G2

do-main,the KSdomain,andvariable lengths ofthechondroitin sulfatedomain(s).(Aggrecan structuraldomainsaredefinedin

refs. 17and18.)The resultsareconsistentwithanaccelerated lossofaggrecanfrom articularcartilageatall stagesofhuman

osteoarthritis

(19), thisloss

being

promoted

by

theaction ofa cartilage

proteinase

which cleaves human aggrecanatthe Glu 373-Ala 374 bond.

Methods

Materials. Superose-12(HR 10/30) and fastdesalting(HR10/10)

col-umnswerefromPharmacia LKBBiotechnologyInc.,Piscataway,NJ.

ChondroitinaseABC(protease free) fromProteusvulgarisand

kera-tanasefromPseudomonas speciesIFO-13309werefrom ICN

Immun-obiologicals, Lisle,IL.Keratanase IIfrom BacillusspeciesKS 36was

fromSeikagakuAmerica Inc.,Rockville,MD. Allother reagentswere

analytical gradefrom commercialsources.

Synovialfluids.Humansynovialfluidwascollected from the knee

jointsofsevenpatientsinconnection withdiagnosticortherapeutic

arthrocentesis. Fluidswerecentrifugedat3,000gto removecells and

particulate material and then frozen at -70°C. The diagnosis was

J.Clin. Invest.

© The AmericanSocietyfor ClinicalInvestigation,Inc.

0021-9738/92/05/1512/05 $2.00

madethrougharthroscopyandradiological and clinicalexamination

anddetailsaregiven in TableI.Bovinesynovial fluidwascombined fromthenormal metacarpalphalangeal joints of ninesteers.

Isolationofaggrecanfragmentsfromjointfluid. Synovial fluid sam-ples were adjusted to 4 Mguanidine-HCl,0.05 Msodiumacetate,0.1

mM PMSF, 5 mM EDTA, 5 mMiodoacetamide, I gg/mlpepstatin,

pH 6.8, andcentrifugedat20,000 gfor30 min to remove any

precipi-tate.SolidCsCl was then added to adensity of1.45g/ml andgradients

established bycentrifugationat40,000rpm for 48 hat10C.Thetubes were sliced in three equal parts to yieldfractionsDl(bottom) to D3, andsamplesweredialyzedat4VCagainst0.05 MsodiumacetatepH

6.8withproteaseinhibitors (see above). Afterafinaldialysisagainst

water,the samples werelyophilizedandportionswereassayed for total proteinorpapain digested forassayofglycosaminoglycan.The

insolu-blegelcollected fromthe topofthegradienttubeswasdialyzedand

Iyophilizedtogether with the topgradient fraction.

Assay and sequencing procedures. Proteoglycanand

glycosamino-glycan weredeterminedaschondroitin sulfate(CS) equivalents by

di-methylmethylene blue (20), usingshark chondroitin sulfateas stan-dard.Protein wasdeterminedusingthebicinchoninic acidassaykit

fromPierce ChemicalCo.,Rockford,IL,with bovineserumalbumin as standard. Amino acidsequence analysiswasperformed on a

se-quencer(473Aor477A;Applied Biosystems, Inc.,FosterCity, CA)

with on-linephenylthiohydantoin analysis. For deglycosylation, Dl

samples in 50 mMTrisacetate,10mMEDTA,pH7.6,weretreatedat

370CwithchondroitinaseABC(0.075 U/mg CS)for 2 h.Sampleswere

adjustedto5 mM PMSF and 10 mMN-ethylmaleimideandincubated for 1hwith keratanase(0.1 U/mg starting CS)andfinallywith keratan-ase 11(0.0025U/mg starting CS) forafurther 1 h.Thisprocedure

resulted ina >90%reductionin thereactivity ofallsampleswith di-methylmethylene blue. Coreproteinpreparations from Superose-12

wereconcentrated to 0.5 ml and desalted on afast desaltingcolumn elutedwithwaterbefore drying forsequenceanalysis.

Results

Isolation ofaggrecanfragments from synovial fluid. Synovial fluids fromseven patients(see TableIforclinical data) were

subjectedtoCsClgradient centrifugationand thegradient frac-tionswereassayedfor glycosaminoglycan and protein. Inall

samples, > 94%oftheglycosaminoglycan and <2.5% ofthe total proteinwas recovered from the gradientin the bottom

fraction(D1,density - 1.55g/ml). This indicates that the bulk

of theaggrecanfragmentspresent in these synovial fluid sam-ples wereof highbuoyant density and therefore rich in chon-droitin sulfate relativetoprotein.

Aportion of the Dl fractions from patientsAandF(Table

I) were analyzed on Sepharose CL-2B in the presence of 5%

(wt/wt) hyaluronan and 4% (wt/wt) linkprotein(not shown). The twoprofiles were very similar and showed the presence ofa

polydisperse population ofCS-bearing fragments (peak Kav

0.55)which did notform aggregateswith hyaluronanand

linkprotein. The fragments eluted later on Sepharose CL-2B

than aggrecan monomerfrom maturehuman cartilage(peak

Kavy 0.38), but not as late as a tryptic digest of monomer

(peakKavy 0.70)orapapain digestofmonomer(peakKav

- 0.79). The results indicate that aggrecan fragments from

thesejoint fluidscompletely lack a functional G1domainand also that they have been cleaved, although not extensively,

within the CS-bearing domains.

Consistent with these findings, fractionation ofsynovial fluidDl aggrecanfragments from patientsAthrough G (Table I) onSuperose- 12 showed thatinall casestheCS-bearing spe-cies werelarge enough to beexcluded fromSuperose-12 (see

Fig. 1, upper panels, for examples), which has an exclusion limit for dextransof - 300 kD. Inaddition, fractionation of

deglycosylated Dl sampleson Superose-12 (seeFig. 1, lower panels, for examples) indicatedthat themajor aggrecan core

protein speciesin all samples were in the apparent molecular mass range of 90-150 kD.

Characterization

ofcore

proteins.Adetailedanalysisof the

Dl aggrecanfragments fromthe threepatientsA,D, and F was

carriedout(seeFig. 1, a, b, and c, respectively). Superose-12

profilesareshownfortheDl samples(Fig. 1, upperpanels)and anequivalent portion ofthecorresponding deglycosylated sam-ples(Fig. 1,lowerpanels). It can be seen that theDl samples werecontaminated to a varying degree with nonaggrecan

pro-tein, shown as 214 nm absorbance, which eluted in the

in-cluded fractions withanapparentsizebelow 66kD.This

mate-rialwaspresumably "trapped" with theDl aggrecan during ultracentrifugation, possiblyas aresultofahigh concentration ofhyaluronan andprotein in these samples. Theidentity of

theseproteinswas notfurther investigated.

Table I. Clinical DataforPatientSamples

Age,

Sample sex Diagnosis* Radiologyt Cartilage at arthroscopy TimeP Vol' ml

A 45 F ACL Normal Chondral lesion 3 d 20

B 27 M ACL + MM Normal Clefts,softening 2 wk 10

C 38 M ACL+MM +LM Normal Superficial fibrillations 5 yr 20

D 50 M Medial Osteophytes, minor decrease No data 15 yr 20 meniscectomy of medialjoint space

E 64 F Primary OA 50% decrease of medial joint Clefts,softening 5 yr 10

space

F 59 M PrimaryOA 100% decrease of medial No data 10 yr 100

jointspace

G 36 M MM Normal Normal Iwk 10

*Traumaandrupture of anterior cruciate ligament (ACL), medial meniscus (MM), lateral meniscus (LM). Primary osteoarthritis with no history of kneejointinjury (OA). t Standing x-ray films. §At time of joint fluid sampling. 11Time afterinjury or meniscectomy or duration of

(b)

I

ia

C4x

to

a

ISOK UK 29K ~~~~~~~~~~2.0

2. 1.5

a2 1.0

20 0.5

0 I -;;10.0

0 2 4 6 6 10 12 14 10 16 20 22 226 2 20

2 4 6 o10 12 14 16 16 2022 24 26 20 20 Broau

I

I

I

II

i.

I,41

14

e4 2<S

_ IO t 29K 2.0

2 1-_.5

,1.0

0 2 4 6 6 10 12 14 10 16 20 22 24 2620 20

FPr mon

2 4 6 6 10 12 14 16 16 20 22 24 26 28

Rmyan

Figure 1. Superose-12 fractionation ofaggrecanfragmentsfromthesynovial fluid of (a) patient A, (b) patient D, and (c) patient F. (Upper panels)

Aggrecan fragments (typically 400-600jig)purified byCsClgradient centrifugation from synovial fluidswereappliedtoSuperose-12 and eluted in0.5 M guanidine-HClat0.5ml/minwithcollection of0.5 mlfractions.Theeluantwasmonitoredat214nm(continuous line) and fractions

wereassayed forCS(connected dots). (Lower panels) Deglycosylatedaggrecanfragments (see Methods)wereappliedtoSuperose-12asdescribed

aboveand the eluantwasmonitoredat214nm(continuous line).Coreprotein speciesinpoolI(fractions 4-8) and pool II (fractions 9-13)

weretaken for NH2-terminal sequencing. Molecularmassmarkerswerecarbonicanhydrase (29 kD),bovineserumalbumin(66 kD), and alcohol

dehydrogenase (150kD).

After deglycosylation, the 214nmabsorbanceprofilesfor

each sample (Fig. 1, lowerpanels) showedtwo major peaks

between fractions4and 13(labelledasIandII).The214-nm

absorbingspecieseluted between fractions14and 22were

ap-parently not generated bythe deglycosylation step butwere present in the starting material (note change in absorbance

scalebetweenupperand lowerpanels).Thehighabsorbanceat

around fraction 26 correspondstothe column Vt and

repre-sentscomponents andproductsof thedeglycosylation

proce-dure.

Theproteinseluted inpoolsIandII(Fig. 1,lowerpanels) wereboth derived fromthehighmolecularweightCS-bearing

aggrecanfragmentsshownintheupperpanels (fractions 4-10). Thus,when otherportionsofDI samplesfrompatients A, D,

and FwerepurifiedonSuperose-12, andfractions 4-10were pooled fordeglycosylation, the 214-nm profile of thesecore samplesonSuperose-12againshowed the characteristic double peakinfractions 4-11. TheDl preparationfromnormal

bo-vine synovial fluid showed essentiallythe same Superose-12 profilesasthehumanmaterial,forboth theuntreated and the deglycosylated samples(profilesnotshown).

NH2-terminal analysis ofaggrecancoreproteinsfrom

syno-vialfluids. Theaggrecan corepreparations from Superose-12

chromatography (pools I and II)of both human and bovine

synovial fluid samplesweretaken forNH2-terminal analysis

and theresults,withapproximate picomole yields, areshown inTable II.

For all synovialfluids analyzed, and for both pool I and

poolIIcoreprotein species,themajorsequenceobtained(see

for example, patient A Superose pool I, sequence ARGS-VILXVKP) clearly corresponded to the human aggrecan

se-quencebeginning ARGSVILTVKP, which initiatesatAla 374

withintheinterglobular domain(from the cDNA, ref. 17).

In-terestingly,theeighth residuecould notbe detected in any

sam-pleandsince this isathreoninein the cDNA-derived sequence,

itseemslikelythatthis residue ismodified, probably by O-sub-stituted carbohydrate.Thefindingof the same NH2terminus

onpoolIandpoolII corespeciessuggests thattheirseparation

Table II. NH2-NTerminalSequence DataObtained forDI Aggrecan FragmentsIsolatedfrom Synovial Fluidsand

Fractionatedby Superose-12 Gel Filtration (Superose Pools ShowninFig. 1,LowerPanels)

Synovial Approximate fluid Superosepool Sequence yield

pmol

Patient A I ARGSVILXVKP 33

II ARGSV 32

Patient D I AXGXVIL 10

II ARGSVILXVK 39

PatientF I ARGSV 11

II ARGSVILXVKPIFEVXP 145

PatientG II ARGSVIL 11

Bovine I ARGSVI 13

II AXGSVIL 15

1514 J. D.Sandy,C. R.Flannery,P.J.Neame,and L. S.Lohmander

I

I

2 4 6 6 10 12 14 10 16 20 22 24 20 26 20 Frslo

(a) (c)

Plea"

i

E

I

i

.011

4

A

,a

4 j

I

bysizeis due to different COOH termini, consistent with lim-ited cleavage within the CS domain(s) as discussed above.

Discussion

Large aggrecan fragments have previously been shown to be present in joint fluids from patients with inflammatory joint

disease or OA (21). We now extend these findingsand show that posttraumatic and osteoarthritic human synovial fluid

containsatleast twopopulations ofaggrecanfragments,both

of which are relatively large and both of which carry the

NH2-terminal sequence ARGSVILXVK, situatedwithinthe

interglobular domain. A diagrammatic representation of two

possible structures is given in Fig. 2. It is clear that this NH2 terminus (Ala 374) was not generated by proteolysis during

deglycosylation with chondroitinase and keratanase, nor at subsequent stages in theisolation ofcoreproteins;thus in con-trolstudies withdeglycosylated calf articular AlDl aggrecan

theonly detectable NH2-terminal sequence obtained, VEVS,

corresponded to that expectedfortheintactmolecule ( 16). In

addition, the NH2-terminal sequence FFGVGGEEDIXVQ

(initiating at Phe 342) was obtained after deglycosylation of

stromelysin-generated CS-bearinghumanaggrecanfragments,

with no evidence for anNH2-terminal sequence beginning at Ala 374 (22).

Suchspecies(Fig. 2) would appear to represent the bulk of theCS-bearing fragmentspresent in osteoarthritic human syno-vial fluid for the following reasons. Firstly, greater than 94% of thefragmentsrecoveredfromthe CsClgradientwererecovered in the

Dl

fraction. Secondly, all of the fragments in suchDl fractionswerepresent in thevoidvolume on Superose- 12

frac-tionation anddeglycosylation ofsuch samplesgenerated only

twopopulations ofaggrecancoreprotein (termed poolsIand II). Thirdly, the only detectable sequence in both pool I and

poolIIsampleswasARGSVILXVK and the molaryield ofthis

sequence was generally about 25%ofthestarting proteinused inpreparation ofthesamples (assuming a molecular mass of 150kDfor these coreproteins).This appears to be a reasonable yield, since two chromatographic steps, with attendant losses, were used in sample preparation for sequencing and the yield in Edman degradation is often only 40-80% of expected. It should be noted, however, that these analyses are confined to

CS-bearing fragmentsof aggrecan and do not address the

struc-ture or abundanceof nonglycosylated fragmentssuch asthe G 1

domainshown to be present in humansynovial fluidsby Wit-terand co-workers (21).

The results also show that essentially the same productsare

released into synovial fluid from normal mature bovine articu-larcartilage in vivo. In addition, our earlierstudies (16)showed thatthe sameNH2-terminalsequencewaspresentonaggrecan fragments released from bovine cartilage in explant cultures bothwithandwithoutIL- I treatment. Similar studieson ag-grecan fragments in the synovial fluid fromrabbitjoints

in-jectedwithIL- Ihave also shown thesamemajorNH2-terminal

sequence(McDonnell,J., V. Moore, R. Boynton, C.Flannery,

P. Neame, and J. Sandy, unpublishedobservation).

When taken together, these studies indicate that the pro-teaseresponsiblefor cleavageofthe Glu 373-Ala 374 bond in theinterglobular domain ofaggrecan iswidely distributed,and would appear to be part of a general mechanism for the catabo-lism of aggrecan in both normal turnover and inpathological

situations.This impliesthat this enzyme is normallyexpressed

bychondrocytesand thatoverexpression of this activitymay beakeyeventin the

pathogenesis

of diseasestates

involving

acceleratedcartilage degradation, suchasosteoarthritis.

The finding ofthis predominant NH2-terminal sequence

onaggrecanfragmentsin allofthesesituationsdoesnotimply

that the enzyme which cleaves the Glu 373-Ala 374 bond is the only, or indeedprimaryenzyme involved in aggrecan catabo-lism. The results do suggest, however, thatthisenzymeacts on

themajorityof catabolized molecules and that it catalyzes the most COOH-terminal cleavage within the interglobular do-main. Itisthuspossible thatthisenzyme actssecondarily to other proteaseswhich cleave theinterglobular domainnearer totheNH2terminus.

Inthis regarditmay be relevant thatstromelysin-1 cleaves theinterglobular domain ofboth human(22)andporcine ag-grecan (23) at the Asn 34 1-Phe 342 bond. Further, a propor-tion of theG1domain which accumulates in human articular

cartilagehas Asn341 as aCOOHterminus,indicative of

stro-melysin- 1 action in situ;stromelysin-1 does not, however, ap-pear to catalyze the cleavageoftheGlu 373-Ala 374 bond(22).

Theprotease which cleaves the Glu 373-Ala 374 bondof

aggrecanremainstobeidentified.Itmay be a memberofthe

family ofmatrixmetalloproteinases,manyof whichhave now been shown todegrade aggrecan as effectively asstromelysin

KS

CS

G2

ARGSVLTVK

IELE

Residue374 Residue 1545

KS

G2

ARGSVLTVK-K

Residue374

CS

1111lEEEI

i

Residue1714

Figure 2. Diagrammatic representationof the

possiblestructureofaggrecanfragmentsin syno-vialfluid.Twopopulationsof aggrecanfragments

areshown,bothlackingtheGIdomain,and both with theNH2-terminalsequence ARGSVILTVK

adjacenttotheG2domain. Thelength oftheCS attachmentregiononthesespecieshasnotbeen

directlydeterminedbutispredicted fromthe ap-parent size of the deglycosylatedcoreproteins.

TheCOOH termini as drawn also coincide with cleavagepositionsin theCS-domain described by

Manigliaetal.(ref.28) in bovine explant studies. Keratansulfate (KS) chains(smallarrows)are

shownonly in the KS domain butsome arealso probably interspersed in the CS domain. CS chains

(large arrows)areshownatallputative

(24). Alternatively, it might be a member of the cathepsin

group of lysosomal cysteine proteinases (25). There isdirect

evidence thatin humancartilageinvivo,aproportion of the

link protein has been cleaved by cathepsinBwhich is presum-ably derivedfrom thechondrocytes (26).Whatever the nature

of this protease,it is almostcertainlyinvolved ingenerationof

theshort half-lifepoolofaggrecanseenincartilageboth in vivo

and in vitro(18,27). This wouldindicate that itsactivitymight beparticularlyhighinthepericellularmatrix ofcartilagewhere

it would have accesstonewlysecreted aggrecan,thereby

imply-ing thatthis enzyme plays a central role in the control of aggre-candepositionincartilagematrix in both normal and diseased

states.

Thefinding thatproducts of this enzymearepresentinthe synovial fluidofpatientA(3 daysafter anterior cruciate

liga-ment rupture), patient D (15 years after medial

meniscec-tomy), and patient F(10 years of primary OA with extensive cartilage loss), suggests thatexpression of this activity may be both an early and persistent feature of the chondrocyte re-sponse to the jointchanges associated with bothprimaryand posttraumatic OA. Identification of the enzyme could allowa

rational approach towards the development of a therapy with theaim ofretardingorinhibiting cartilage matrix destruction in osteoarthritis.

Acknowledgments

The authors wish toacknowledge the excellent technical assistance of Raymond Boynton andCarmen Young.

Thiswork was supported by a grantfrom the Shriners of North

America, NationalInstitutes of Health, grant AR 38580, the Swedish

MedicalResearchCouncil(8713), the King Gustaf V 80-year fund, the Ax:sonJohnson and KockFoundations and the Medical Faculty of LundUniveristy.

References

1.Hoch, D. H., A. J.Grodzinsky,T. J.Koob,M. L.Albert,and D. R. Eyre.

1983.Early changes in materialproperties of rabbit articular cartilage after meni-sectomy. J.Orthop.Res. 1:4-12.

2.Carney,S. L., M. E. J.Billingham,H.Muir,andJ. D.Sandy.1984.

Demon-stration ofincreased proteoglycanturnoverincartilage explants from dogs with experimental osteoarthritis.J.Orthop.Res.2:201-206.

3.Carney,S.L.,M.E. J.Billingham,H.Muir,and J. D.Sandy. 1985.

Struc-tureof newly synthesized(35S)-proteoglycansand(35S)-proteoglycanturnover

products ofcartilage explant cultures from dogs with experimental osteoarthritis.

J.Orthop.Res.3:140-147.

4.Lohmander,L.S.,L.Dahlberg,L.Ryd,and D.Heinegard.1989.Increased

levels of proteoglycanfragmentsin kneejoint fluid after injury. Arthritis Rheum.

32:1434-1442.

5.Lohmander,L.S.,L.Dahlberg,L.Ryd,and D.Heinegird. 1990.Joint markers insynovial fluid in humanosteoarthritis. Trans. Orthop.Res. Soc. 15:212.(Abstr.)

6.Lohmander, L.S.,and L.Dahlberg. 1991.Proteoglycan epitopesinjoint fluidin human osteoarthritis. Trans.Orthop.Res. Soc. 16:227. (Abstr.)

7. Docherty, A. J. P., and G.Murphy. 1990. The tissue metalloproteinase family and the inhibitor TIMP: a studyusing cDNAs and recombinant proteins. Ann.Rheum.Dis.49:469-479.

8. Murphy, G., R. M. Hembry, C. E.Hughes, A. J. Fosang, and T. E. Har-dingham. 1990. Role andregulationofmetalloproteinasesin connective tissue turnover.Biochem. Soc.Trans. 18:812-815.

9. Dean,D. D., J.Martel-Pelletier,J.-P.Pelletier,D. S.Howell,and J. F. J. Woessner.1989.Evidence for metalloproteinaseandmetalloproteinase inhibitor

imbalancein humanosteoarthriticcartilage.J.Clin. Invest.84:678-685.

10. Lohmander, L. S., L. A. Walakovits, and M. W. Lark. 1992. Metallopro-teinases, tissue inhibitor and proteoglycan fragments in knee synovial fluid in human osteoarthritis. Trans. Orthop.Res. Soc. 17:273. (Abstr.)

11.Lohmander, L. S., L. A.Walakovits, L. Dahlberg, H. Roos, and M. W. Lark. 1992. Temporal patterns ofrelease of stromelysin, tissue inhibitor and proteoglycan fragments in synovialfluid after knee injury. Trans. Orthop. Res. Soc. 17:651. (Abstr.)

12.MacNaul, K. L., N.Chartrain,M.Lark, M. J.Tocci,and N. I. Hutchin-son.1990. Discoordinateexpression ofstromelysin, collagenase, and tissue

inhibi-torofmetalloproteinases-1 inrheumatoidsynovial fibroblasts. J.Biol. Chem. 265:17238-17245.

13. Gravallese,E. M., J. A.Darling, A.L.Ladd, J. N. Katz,and L. H.

Glimcher. 1991.Insituhybridization studiesofstromelysin and collagenase

mes-senger RNAexpression in rheumatoidsynovium.Arthritis Rheum. 34:1076-1084.

14.McCachren, S. S. 1991.Expression ofmetalloproteinasesand

metallopro-teinase inhibitorin humanarthriticsynovium.Arthritis Rheum. 34:1085-1093. 15.Firestein, G.S., M. M.Paine,and B. H.Littman. 1991. Geneexpression

(collagenase, tissue inhibitor ofmetalloproteinases,complementandHLA-DR) inrheumatoid arthritisandosteoarthritis synovium. Arthritis Rheum. 34:1094-1105.

16. Sandy, J. D., P. J. Neame, R. E.Boynton, andC.R.Flannery. 1991. Catabolism ofaggrecan incartilage explants. Identification ofamajorcleavage site withintheinterglobular domain.J.Biol.Chem. 266:8683-8695.

17.Doege, K., M.Sasaki,T.Kimura, andY.Yamada.1991.Complete

cod-ingsequence and deducedprimarystructureofthehumancartilage large aggregat-ing proteoglycan, aggrecan. J.Biol.Chem.266:894-902.

18. Sandy, J. D. 1992. Extracellular metabolism of aggrecan. In Articular

Cartilage and Osteoarthritis.K.Kuettner,etal.,editors.Raven Press, Ltd., New York.21-33.

19. Lohmander,L.S. 1991. Markersofcartilage matrix metabolismin

arthro-sis.Areview.ActaOrthop. Scand. 62:623-632.

20.Farndale,R.W., C. A.Sayers,and A. J. Barrett.1982. A direct

spectropho-tometricmicroassay for sulfatedglycosaminoglycanincartilagecultures. Con-nect. Tissue Res.98:247-248.

21.Witter,J., P. J.Roughley, C. Webber, R. Roberts, E. Keystone, and A. R. Poole.1987.Theimmunologicdetection andcharacterizationofcartilage proteo-glycandegradationproducts insynovial fluids of patients with arthritis.Arthritis Rheum.30:519-526.

22.Flannery, C. R., M. W. Lark, and J. D. Sandy. 1992.Identificationofa

stromelysin cleavagesite within theinterglobulardomainofhumanaggrecan: evidence forproteolysisatthissiteinvivoinhumanarticularcartilage.J. Biol. Chem. 267:1008-1014.

23. Fosang, A. J., P. J. Neame, T. E.Hardingham, G. Murphy, and J. A.

Hamilton. 1991. Cleavage ofcartilage proteoglycanbetween G 1 and G2domains bystromelysins.J.Biol. Chem. 266:15579-15582.

24. Murphy,G.,M. I.Cockett, R. V. Ward, and A. J. Docherty. 1991.Matrix

metalloproteinasedegradation ofelastin,type IVcollagen and proteoglycan.

Bio-chem.J. 277:277-279.

25. Barrett, A. J., D. J.Buttle, and R. W. Mason. 1988. Lysosomalcysteine proteinases. ISI AtlasofScience Biochem. 256-260.

26. Nguyen,Q., J.Liu,P. J.Roughley,and J.S. Mort.1991.Linkproteinas a monitor in situof endogenous proteolysisin adult humanarticularcartilage. Biochem. J.278:143-147.

27.Lohmander, L. S. 1977. Turnover ofproteoglycans in guinea pig costal

cartilage. Arch. Biochem.Biophys.180:93-101.

28.Maniglia, C. A., P. P. Loulakis,A. V.Shrikhande,and G. Davis. 1991.

IL-1 elevated PGdegradationrevealsNH2-terminal sequence homology. Trans.

Orthop.Res.Soc. 16:193.(Abstr.)