ultrastructural localisation of fibronectin in human

Annals of the Rheumatic Diseases, 1987; 46, 816-822

Ultrastructural localisation of fibronectin inhuman osteoarthritic articular cartilageJ A REES, S Y ALI, AND R A BROWN

From the Experimental Pathology Unit, Institute of Orthopaedics (University of London), Royal NationalOrthopaedic Hospital, Stanmore, Middlesex

SUMMARY A protein A-gold immunolocalisation technique has been used on sections of femoralhead articular cartilage to localise fibronectin. Chondroitinase treatment enhanced gold staining,particularly when tissue samples were digested before fixation. The greatest accumulations offibronectin were seen in the surface zone of osteoarthritic cartilage. Disease free cartilagecontained very little fibronectin in this region. Cells which appeared to produce fibronectin were

rare in normal specimens but common in the superficial region of osteoarthritic cartilage. Thesechrondrocytes appeared to release fibronectin as part of an amorphous material whichaccumulated in the pericellular region. This is the first ultrastructural demonstration offibronectin synthesis by articular cartilage chondrocytes.

Key word: chrondrocytes

A number of groups have demonstrated the pres-ence of fibronectin in normal cartilage tissues. -5Some studies identified fibronectin using immuno-fluorescence, in a pericellular halo around thechondrocytes in normal hyaline cartilage,' whereasothers found no staining in the pericellular regionbut rather in the interterritorial matrix.2 Thelocation of fibronectin in cartilage is important asthis may provide clues to its source and its role in thematrix. This is highlighted by reports of increasedlevels of fibronectin in cartilage from osteoarthritichuman6 and canine joints.The aims of this ultrastructural study were to

identify where in the cartilage matrix fibronectin islocated, where it accumulates in chondrocytes, andhow these features differ in normal and osteo-arthritic cartilage. The findings reported heresupport the contention that fibronectin is synthesisedlocally by chondrocytes.

Materials and methods

Osteoarthritic articular cartilage, type IV-residualcartilage as described by Ali and Bayliss8-wastaken from the femoral heads peripheral to the areaof exposed bone9 immediately after total hip

Accepted for publication 1 May 1987.Correspondence to Dr R A Brown, Experimental Pathology Unit,Institute of Orthopaedics (University of London), Royal NationalOrthopaedic Hospital, Stanmore, Middlesex HA7 4LP.

replacement. Normal femoral head cartilage (type Ior II) was obtained after amputation or surgery forsubcapital fracture of the femoral neck. Thin slicesof tissue were fixed in 2-5% glutaraldehyde in 0-085M sodium cacodylate buffer, pH 7-2, for one hour atroom temperature. Secondary fixation was for onehour in 1% osmium tetroxide in the same buffer.Specimens were dehydrated, embedded in Araldite,and sections placed on Formvar coated grids andexamined unstained or stained with uranyl acetateand lead citrate using a Phillips 300 electronmicroscope.

IMMUNOELECTRON MICROSCOPY

Specimens were taken as described above. Pre-fixation digestion was with chondroitinase ABC(Sigma, Dorset) at 2 units/ml in 50 mM TRIS-HClbuffer, pH 8, for 45 min at 37°C. Both predigestedand non-digested specimens were fixed in 1%glutaraldehyde in 01 M phosphate buffer, pH 7-2,for one hour at room temperature and washed inbuffer for 15 minutes. Excess glutaraldehyde wasneutralised by exposing the tissue to 0 5 Mammonium chloride for two hours, and specimenswere partially dehydrated at 4°C in graded concen-trations of methanol. Specimens were transferred to70% methanol at -30°C for one hour, then into a1:1 mixture of Lowicryl K4M resin:70% methanolovernight and finally, into four changes of pureresin for 48 hours. The resin was polymerised at

816

copyright. on M

ay 22, 2022 by guest. Protected by

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.46.11.816 on 1 N

ovember 1987. D

ownloaded from

http://ard.bmj.com/

Fibronectin in human osteoarthritic articular cartilage 817

-30°C for 24-48 hours with an ultraviolet lightsource. The blocks were stored desiccated, insunlight, before sectioning. Gold/silver sections(70-100 nm thick) were mounted on Formvar coatedcopper grids and stained by the protein A-goldmethod as described by Mapp and Revell. "' Gridswere treated for five minutes with 1% bovine serumalbumin (Armour) in phosphate buffered saline-10mM sodium phosphate, 0-15 M NaCl, pH 7-5(BSA/PBS). Undigested specimens were digestedwith chondroitinase at this stage as described above,but for .30 minutes. Grids were then rinsed inBSAIPBS and treated with rabbit antihuman plasmafibronectin (Behring AG, Marburg, FDR), diluted1:50 in BSA/PBS for one hour at room temperature,and washed repeatedly in BSA/PBS. Sites ofantibody binding were localised by treatment forone hour at room temperature with protein A-goldusing either 5 or 20 nm diameter particles (Janssen,Beerse, Belgium and E-Y Laboratories, San Mateo,USA). Grids were washed thoroughly in BSA/PBSand in distilled water and examined unstained orcounterstained with aqueous uranyl acetate and leadcitrate. Control sections were prepared (a) byomitting the primary antifibronectin antibodytreatment, (b) by incubating sections with non-immune rabbit serum in place of antifibronectin,and (c) by pretreating the rabbit antifibronectin withexcess, purified human fibronectin.

Results

Specimens of femoral head articular cartilage wereexamined from the periphery of eroded areas' offour osteoarthritic (two men aged 78 and 60 years,

two women aged 61 and 55 years) and two normaljoints (one man aged 22 years, one woman aged83 years). Type IV8 full depth and slope zonecartilage showed many of the typical degenerativechanges associated with osteoarthritis, such assurface fibrillation, loss of matrix staining forproteoglycans, cell clusters, and a somewhat dis-orientated collagen network in the superficialzone. I

CONNECTIVE TISSUE LOCALISATIONIn normal articular cartilage the matrix staining forfibronectin was negligible. A deposit of electrondense material, probably a layer of adsorbedprotein, was seen in a normal (man aged 22 years,amputation) cartilage specimen at the surface (Fig.1). This material stained strongly with gold particles,indicating a high and localised concentration offibronectin. It was notable, however, that there wasno evidence of penetration of fibronectin from thissurface film into the underlying cartilage matrix.The absence of staining in the superficial zone is incontrast with staining in osteoarthritic specimens.The collagen matrix appeared normal with a regulararrangement of fibrils of constant diameter, whichwere orientated tangentially to the articular surface(Fig. 1). This contrasted with the osteoarthriticspecimens, where orientation was more variable anda range of fibril diameters was present (Fig. 2).The heaviest accumulations of fibronectin were

seen in superficial zone cartilage matrix from all ofthe osteoarthritic specimens examined (Fig. 2a).This accumulation appeared as a band of intensegold staining reaching 20 to 60 ,tm below thearticular surface. No specific accumulations were

II -q-, AwI in

K -. 9.i2 . *%f.f

Fig. 1 Normal surface articularcartilage matrix stainedforfibronectin. Electron micrographofthe surface ofnormal (managed 22 years, amputationspecimen) femoral head cartilageshowing the absence ofgoldparticles, hence fibronectin, inthe matrix. Note that the thinband ofnon-collagenous materialoverlying the cartilage (arrow)is stained strongly with 20 nmgold particles (see inset). Sectionswere treated with chondroitinaseABC and counterstained withuranyl acetate and lead citrate.

-05tsm

----w,-I ITWwk-.-T-.;, 4.*f-

copyright. on M


http://ard.bmj.com

/A

nn Rheum


ovember 1987. D

ownloaded from

http://ard.bmj.com/

Fig. 2 Osteoarthritic surfacecartilage matrix stained forfibronectin. Electron micrographof the surface region ofosteoarthritic (61 vear old woman)femoral head cartilage. Thecartilage matrix stained stronglywithl gold particles and had beenincubated with (a) rabbitantihuman fibronectin anid(b) non-immune rabbit serumbefore treatment with 20 nmprotein A -gold particles. Sectionswere treated with chondroitinaseABC and counterstained withuranyl acetate and lead citrate.

seen, for example, adjacent to surface fibrillationsor clefts. Light counterstaining of this area withuranyl acetate and lead citrate indicated that themain matrix component in this area was fibrillarcollagen. Although there was a tendency for goldparticles to lie on collagen fibrils, no pattern ofassociation was visible, either with fibrils of a

particular size or with the collagen striations. Somegold staining was also seen in deeper areas of matrixin some osteoarthritic specimens. In all the abovecases control sections showed only small amounts ofnon-specific gold binding (Fig. 2b).

INTRACELLULAR AND PERICEILLULARLOCA LI SATION

Intracellular and pericellular staining for fibronectin

were most pronounced in the superficial zone ofosteoarthritic specimens, though local accumulationswere seen in and around a few cells of disease freecartilage. Clusters of three or four biosyntheticallyactive cells producing fibronectin were seen in thesuperficial zone and occasionally in the mid or deepzones of osteoarthritic cartilage. Figs 3 and 4illustrate chondrocytes producing an amorphousmaterial which contained fibronectin, as shown bythe localisation of gold particles. Similar materialwas also seen intracellularly in secretory vacuoles.Although this fibronectin did not appear to be themain constituent of the secreted material, itslocalisation within the amorphous material andsecretory vacuoles was striking. The same amor-

phous material, seen in control sections treated with

818 Rees, Ali, Brown

a

1 [Im

b

1 jnm

a_ .0-

copyright. on M


http://ard.bmj.com

/A

nn Rheum


ovember 1987. D

ownloaded from

http://ard.bmj.com/


a

.,..b4.s.,,, . p e

rA ~ eWA?

OA-

12tl- P

085pm

bj ~.'.'** I..',

C t%; ~= t40~~7

* -44 ab

Fig. 3 Secretion offibronectincontaining material:chondroitinase digestion ofthesection. (a) Lowpowerappearance ofa cluster ofchondrocytes in the superficialzone ofhuman osteoarthriticarticular cartilage. ger=granularendoplasmic reticulum. Noteamorphous material (arrows) inthe lacuna (pericellular space), seebelow and Fig. 4. (b) Detail ofthesurfacefrom an active secretorychondrocyte releasing anamorphous material (arrowed)which stainedforfibronectin with5 nm gold particles. p=cellprocesses. (c) Non-immune serumtreated control section showing thesame amorphous material whichdid not bind gold. Specimens weredigested on the grid withchondroitinase ABC, treated with5 nm gold, and counterstained withuranyl acetatellead citrate.

:,;, "..;'-

'Ill'4

non-immune serum, did not bind gold particles (Fig.3c). The amorphous areas of secretory product wereintact further from the cell (Figs 4a and 4c), thoughthere was no evidence of an enclosing membrane.Even where this amorphous material was somedistance from the chondrocyte, fibronectin was

clearly localised as a component (Fig. 4c). A fewcells surrounded by fibronectin containing materialwere seen in a specimen from a 22 year old normalman.

Figs 3 and 4 illustrate the effects on the section ofdigesting the tissue with chondroitinase eitherbefore fixation and embedding or after processing.Although localisation was similar by both methods,

prefixation digestion appeared to give a generallyhigher level of staining. Staining for fibronectin was

very weak in the absence of any digestion stage.Fibronectin was localised specifically in the

endoplasmic reticulum, Golgi region, dilatedcisternae of the rough endoplasmic reticulum, and inmembrane bound and apparently secretory vesicles(Fig. 5) of active chondrocytes in osteoarthritic,surface zone cartilage. Similar membrane boundsecretory vesicles have been described in chondro-cytes by other workers. 1 Treatment of serial sectionsof fibronectin positive cells with non-immuneserum, in place of antifibronectin, failed to producestaining above a low background level (Fig. Sc).

b3

t

4.

Z&Wtqwl;.

-1 ';.I ,

. Z:

...I .i4il

e

copyright. on M


http://ard.bmj.com

/A

nn Rheum


ovember 1987. D

ownloaded from

http://ard.bmj.com/


a C

_05[im

b

Fig. 4 Secretion and release offibronectin containing material:chondroitinase digestion on block.(a) Electron micrograph showingcell surface (lower right hand) andpericellular area (lacuna) ofanactive chondrocvte. Amorphousmaterial was visible bothsurrounded by cell processes acndfurtherfrom the cell (78 year oldman, osteoarthritis). (b) Detailfrom (a) (lower box) of cellprocesses surrounding amorpholsmaterial which stained forfibronectin with 5 nm gold(arrowed). Note the absence of5 nm gold in areas away from theamorphous material. (c) Detail ofpericellular amorphous material ata distancefrom the cell surfaceseen in (a) (upper box) showingintense localisation offibronectin(see also Fig. 3a). Once again,gold bound only to this amorphousmaterial. Specimens were digestedwith chondroitinase A BC beforefixationlembedding sectionswere stained using 5 nm proteinA -gold and counterstainedwith uranyl acetatellead citrate.

..k."A ..

4 ,,

0-5wil

Discussion

Fibronectin was present in much larger amounts in

cartilage from osteoarthritic, rather than fromnormal joints. Cells producing fibronectin were seen

only rarely in normal specimens. The greatestaccumulations were seen in the superficial zone ofosteoarthritic cartilage and this may, in part, havebeen synthesised locally by chondrocytes. Certainly,considerable numbers of these cells, in osteoarthriticcartilage, synthesised and released fibronectin as

part of a characteristic, amorphous material whichaccumulated in the pericellular region. A proportionof the fibronectin at the surface may also originatefrom synovial fluid, which is rich in this protein.'2The proportional contribution by synovial fluid isunknown, however. Penetration of fibronectin intothe surface of normal articular cartilage was minimalin both normal specimens. This, and the observation

of a fine layer containing fibronectin on the surfaceof normal cartilage, are consistent with a previousreport that labelled fibronectin did not penetrate thesurface of normal cartilage explants.'13The nature of other components present in the

amorphous secretory material is unknown, althoughthe improvement in antigen location after chon-droitinase digestion implies that some proteoglycanis present. The nature of this material is currentlyunder investigation. Similar amorphous material hasbeen described by Geerts et al at or near the surfaceof liver cells.'4 In this case both fibronectin andheparan sulphate were present.The superficial cartilage matrix is composed

largely of collagen, aligned tangentially to thesurface. Although gold particles tended to lie oncollagen fibrils in osteoarthritic specimens, therewas no evidence of a regular or symmetrical patternof distribution on collagen, as reported by Shiozawa

r-,--, I

1 .lm 1-

copyright. on M


http://ard.bmj.com

/A

nn Rheum


ovember 1987. D

ownloaded from

http://ard.bmj.com/


c *?.I'

%, *.. s I S~~~~. .%

I

et al in rheumatoid joints.'t Postembedding local-isation used in this study is not ideal for staining ofan antigen on collagen fibrils.

In any attempt to determine the relation betweenfibronectin accumulation and degenerative cartilagechanges it is important to distinguish betweenpenetration of exogenous protein and its synthesisby chondrocytes. Increased synthesis of fibronectinby superficial zone chondrocytes may represent partof a cellular response aimed at repair of the matrix,perhaps by stabilising local proteoglycan loss. Thiswould be analogous to its production as part of thewound repair response in other tissues. 6 Alter-natively, cartilage fibronectin may be produced, inconjunction with cell surface proteoglycans, as part

Fig. 5 Intracellular localisation-.e. offibronectin with 20 nm gold.'- V (a) Low power view ofa cluster

.*; ofactive chondrocytes in thesuperficial zone ofhumanosteoarthritic articular cartilage(61 year old woman).ger=granular endoplasmic

. reticulum. (b) High,power detailofchondrocytes synthesisingfibronectin. Note the localisationoffibronectin on the granularendoplasmic reticulum and in adilated cisterna (arrow).Inset shows the concentration ofgold particles on Golgi vacuoles.(c) Control treatment, withnon-immune serum, ofa similarregion in the same cluster of cells.Sections were treated withchondroitinase ABC, stained with20 nm protein A-gold, andcounterstained with uranyl acetateand lead citrate.

0 5pm

of a pericellular matrix similar to that isolated fromfibroblasts. '7 18 This seems less likely in view of ourobservation that discrete areas of fibronectin-richamorphous material were often some distance fromthe nearest cell. In addition, it has been suggestedthat the increased synthesis of fibronectin representsa subtle change in phenotype (perhaps with nodirect function),7 reflecting the presence of de-differentiated chondrocytes.

Penetration and accumulation of exogenousfibronectin, in contrast, have a different signifi-cance. Molecules of the proportions of fibronectinhave negligible access to the matrix of normalcartilage,'9 as demonstrated in this and previousstudies.9"3 Thus penetration of fibronectin for short

F_ ..- s

e,. A.I£. - 14

.&. .-e.Ar.

.~~~~~~~~~~~~~~~~~~~~~~~~~~~.

5 pmn 05tim

r

r"p

q#

'IsIlk

.I

I

f .-,; I... .

"ji,.

,

I

4

47.0

a

copyright. on M


http://ard.bmj.com

/A

nn Rheum


ovember 1987. D

ownloaded from

http://ard.bmj.com/


distances into canine osteoarthritic cartilage'3 islikely to be a result of disruption of the collagennetwork. The consequences of this accumulation areunknown, though there is some evidence thatchondrocyte behaviour may be altered.2t'

This is the first description of intracellular fibro-nectin in osteoarthritic cartilage at the ultra-structural level and provides confirmation ofprevious radiotracer and biochemical evidence thatfibronectin is synthesised in vivo by chondrocytes.4 7The differences in pericellular distribution

between normal and osteoarthritic cartilage areconsistent with the suggestion that there is a localsignal for its synthesis as actively synthesising cellswere seen in isolated groups,9 mainly in one region,the superficial zone. Furthermore, cells with andwithout fibronectin were frequently adjacent to oneanother. Initiation of this localised response may besignalled by changes in the structural matrix, forexample through altered permeability or mechanicalproperties. The questions of how fibronectinproduction is stimulated and what its function orconsequences are to the matrix are presently underinvestigation.

We are grateful to Miss Linda Adams for prcparation of themanuscript. This work was supportcd hs the Arthritis andRheumatism Council.

References

I Glant T T. Hadhazv Cs, Mikecz K. Sipos A. Appecirancc andpersistence of fibronectin in cartilagc. HiotochemnitrY 1985: 82:149-58.

2 Evans H B, Ayad S, Ahcdin M Z. et al. Locailisation of collagentypes and fibronectin in cartilagc hy immunofluorescence. Ati,Rheum Dis 1983; 42: 575- 81.

3 Weiss R E. Reddi A H. Appecirancc of fibroncctin during thedifferentiation of cartilagc. honc and honc marrow. J Cell Biol1981; 88: 631-6.

4 Wurstcr N B. Lust G. Deposition of fibroncctin in articulaircartilage of canine osteoarthritic joints. Ain J Vet Res 1985; 46:2542-5.

S Wurster N B, Butler M. Ilarter M. et al. Prcscncc of fihroncctinin articular cartilagc in two animal models of ostcoarthritis. JRheumatol 1986; 13: 175-82.

6 Miller D R. Mankin 11 J. Shoji H. D'Amhrosia R D.Identification of fibronectin in preparations of ostcoarthritichuman cartilage. Coontiect Tissue Res 1984; 12: 267-75.

7 Wurster N B. Lust G. Fibronectin in ostcoairthritic cartlalgc- ipossible indication of phenotypic modulation of the chondro-cyte. In: Verbruggen G. Veys E M. eds. Degeoeratire Joiott.Vol. 2. Amsterdam: Elsevier. 1985: 141-7.

8 Ali S Y, Bayliss M T. Enzymic changes in human osteoarthriticcartilage. In: Ali S Y, Elves M W, Leaback D H, eds. Normaland osteoarthritic articular cartilage. London: Institute ofOrthopaedics, 1974: 189-205.

9 Jones K L, Brown M, Ali S Y. Brown R A. An immuno-histochemical study of fibronectin in human osteoarthritic anddisease free articular cartilage. Ann Rheum Dis 1987; 46: 809-15.

1() Mapp P 1. Revell P A. Fibronectin production hv synosialintimal cells. Rheu,natol Int 1985; 5: 229-37.

11 Stockwell R A. Meachim G. The chondrocvtes. In: FreemainM A R, ed. Adult airticulair cartilage. Tunbridge Wells: PitmanMedical. 1979: 69-144.

12 Scott D L, Farr M. Crockson A P, Walton K W. Svnoviail fluidand plasma fibronectin levels in rheumaitoid arthritis. Cliii Sci1982: 62: 71-6.

13 Wurster N B. Lust G. Incorporation of purified fibronectin intoexplants of articulair cartilage from disecise-free and osteo-arthritic canine joints. J Ortholp Res 1986; 4: 4119-19.

14 Geerts A, Geuze H J, Slot J W et al. Immunogold localisationof procollagen III, fibronectin and heparan sulphate proteo-glycan on ultra thin frozen sections of normal rat liver.Histochemistrv 1986; 84: 355-62.

15 Shiozawa K, Shiozawa S. Shimizu S, Fujita T. Fibronectin onthe surface of articular cartilage in rheumatoid arthritis.Arthritis Rheum 1984; 27: 615-22.

16 Grinneil F. Fibronectin and wound healing. J Cell Biochemn1984: 26: 10)7-16.

17 Yamagata Ii. Yamada K M, Yoneda M. Suzuki S. Kimatat K.Chondroitin sulphaite proteoglvcan (PG-M-like protcoglvcan)is involved in the hinding of hyailuronic acid to ccllulairfibronectin. J Biol Chemn 1986; 261: 13526-35.

18 f1edman K. Kurkinen M. Alitalo K. V\ahcri A. Johansson S.Hook M. Isolation of the pericellular mattrix of humain fibro-hlast cultures. J Cell Biol 1979; 81: 83-91.

19 Maroudas A. Distribution and diffusion ot solutes in articularcartilage. Btophys J 197t0: 10: 36.5-79.

211 West C M. L'anza R. Rozenbloom J. Lowc M. f-oltzcr IH.Fihronectin altcrs the phenotypic propertics of cultured chickembrvo chondrohlasts. (ell 1979; 17: 491-501.

21 Gibson G J. KieIlv C M. Garner C. Schor S L. Graint M F.ldentification and partial charaicterisaition ot three low, molc-cular weight collagenous polypeptides synthcsised hv chondro-cvtes cultured within collagen gels in the absence aind in thepresence of fibronectin. Biochemn J 1983: 211: 417-2(1.

copyright. on M


http://ard.bmj.com

/A

nn Rheum


ovember 1987. D

ownloaded from

http://ard.bmj.com/

ultrastructural localisation of fibronectin in human

Documents