Subpopulations of chondrocytes from different zones of pig articular

cartilage

Isolation, growth and proteoglycan synthesis in culture

MARTIN SICZKOWSKP

Department of Anatomy and Developmental Biology, University College and Middlesex School of Medicine, Windeyer Building,Cleveland Street, London W1P 6DB, UK

and FIONA M. WATT

Keratinocyte Laboratory, Imperial Cancer Research Fund, PO Box 123, Lincoln's Inn Fields, London WC2A 3PX, UK

* Present address: Leukaemia Research Fund Centre, Institute of Cancer Research, Chester Beatty Laboratories, Fulhatn Road,London SW3 6JB, UK

Summary

Articular cartilage varies in ultrastructure andcomposition with distance from the articular surface.We have cultured chondrocytes from different zonesof pig articular cartilage and investigated whetherthere are intrinsic differences in their behaviour thatmight account for the variation observed in intacttissue. On isolation, cells from the upper third of thecartilage were smaller than those of the lower third,but this difference was not maintained in culture.Upper zone cells attached and spread more slowlythan lower zone cells; morphological differencesbetween the two populations could be seen forseveral weeks. The growth rates of the two popu-lations were similar, but upper zone cells reached alower confluent density. Levels of protein synthesiswere similar for both populations, but upper zonecells deposited less proteoglycan in the cell layer. Onisolation, the percentage of upper zone cells thatstained positive with MZ15, a monoclonal antibody tokeratan sulphate, was smaller than the percentage oflower zone cells, but this difference was lost afterseveral days in culture. Nevertheless, the keratansulphate content of proteoglycan synthesised by

lower zone chondrocytes at high density was greaterthan of that synthesised by upper zone cells. Theproportion of nonaggregating proteoglycan wasgreater in upper than lower zone cartilage and thisdifference was also observed in long-term cultures.Proteoglycans were further characterised by compo-site and polyacrylamide gel electrophoresis and byimmunoblotting; differences detected in cartilageextracts were not, however, maintained in culture;instead, the small proteoglycans synthesised by bothupper and lower zone cells varied with platingdensity. Finally, alkaline phosphatase, a marker ofhypertrophic, calcifying cartilage, was only ex-pressed in lower zone cultures. We conclude thatsome of the observed heterogeneity of articularcartilage reflects intrinsic differences between thecells of different zones, whereas some may reflect theresponse of chondrocytes to different environmentalconditions.

Key words: chondrocytes, cartilage, proteoglycan.

Introduction

Articular cartilage covers the ends of bones where theymeet at joints and absorbs forces that are brought to bearon the joints. It consists of a single cell type, thechondrocyte, that secretes an extensive extracellularmatrix, the most-abundant components of which are typeII collagen and large proteoglycan molecules aggregatedwith hyaluronic acid.

Several distinct zones of cartilage can be recognised byhistology (Palfrey and Davies, 1966). The superficial zone,furthest away from the bone, is a thin zone of elongatedchondrocytes. The transitional, or middle, zone consists ofrandomly oriented cells that are oval or round. In the deepJournal of Cell Science 97, 349-360 (1990)Printed in Great Britain © The Company of Biologists Limited 1990

zone, the cells are larger than in the other zones; they areround and may be arranged in vertical columns. Finally,between the deep zone and bone there is a layer ofhypertrophic cartilage.

In addition to variation in cell morphology and densitywith distance from the articular surface, there is alsovariation in the biochemical composition and physicalproperties of the extracellular matrix. Thus, the packingand orientation of collagen fibres change with depth fromfine, tangentially orientated in the superficial zone tothick and vertically orientated in the deep zone (Gha-dially, 1978; Poole et al. 1982). The amount of proteogly-can, its size, glycosaminoglycan composition and aggre-gation properties also change with distance from the

349

articular surface (Maroudas et al. 1969; Franzen et al.1981; Bayliss et al. 1983; Ratcliffe et al. 1984; Kuijer et al.1986; Manicourt et al. 1988). Two proteins that arespecifically expressed in hypertrophic, calcifying, carti-lage are type X collagen (Kielty et al. 1985; Schmid andLinsenmayer, 1985;- Kwan et al. 1986) and alkalinephosphatase (Fortuna et al. 1980; Hsu et al. 1985).

An important question that has been largely unexploredis the extent to which the differences between differentcartilage zones reflect intrinsic differences between thecells of those zones and the extent to which they are due toenvironmental regulation. One approach to answeringthis question was taken by Zanetti et al. (1985), whoshowed that a monoclonal antibody that recognizeskeratan sulphate, MZ15 (Zanetti et al. 1985; Mehmet et al.1986), stained the upper third of pig articular cartilagemore weakly than the deeper zones, and that theproportion of cells isolated from the upper zone withsurface keratan sulphate was less than the proportion ofcells from the deeper zone. With time in culture, theproportion of upper zone cells that stained positive withMZ15 increased, so that the two subpopulations could nolonger be distinguished on the basis of MZ15 staining.This suggested that keratan sulphate expression can bemodulated by the environment. However, differences inthe morphology of cells from different zones did persistwith time in culture, indicating that intrinsic differencesbetween chondrocyte subpopulations might exist (Zanettiet al. 1985).

In this paper, we have carried out more extensivestudies on the properties of subpopulations of pig articularchondrocytes in culture, which have enabled us to identifyboth inherent and environmentally regulated differencesbetween the cells of different zones.

Materials and methods

Isolation of chondrocytes from different cartilage zonesThe method used to isolate chondrocytes was essentially thatdescribed by Zanetti et al. (1985). Cartilage was dissected from themetacarpo-phalangeal joints of trotters from bacon pigs, rinsedtwice in calcium/magnesium-free Dulbecco's phosphate-bufferedsaline (PBS) containing 400 units ml"1 penicillin/streptomycinand 200 units ml"1 nystatin and stored at 4°C overnight in PBScontaining 200 units ml"1 penicillin/streptomycin and67 units ml"1 nystatin. The cartilage was finely chopped and thechondrocytes were released from their extracellular matrix bysequential digestion at 37 °C with 0.05% hyaluronidase (Boehr-inger-Mannheim) in PBS for 20min, 0.1% Pronase (Sigma) inPBS for l h and 0.25% collagenase (Boehringer-Mannheim) inDulbecco's modified Eagle's medium (DMEM) containing 1 %foetal calf serum (FCS) for 1-3 h. If the cartilage was notcompletely digested by this stage, a further incubation withcollagenase was carried out. The small numbers of cells releasedby hyaluronidase and Pronase were discarded. Cells obtainedfrom the collagenase digests were pooled and passed through asterile 60 fan aperture nylon screen (Nitex) to remove anyundigested cartilage fragments.

Chondrocytes were isolated from different depths of cartilageby sequentially dissecting away the upper, middle and lower thirdof the tissue. Cells were then isolated separately from the upperand lower thirds and the middle third was discarded.

For every chondrocyte isolation, cartilage from several trotters(6-24) was pooled. The yield of cells from unfractionated cartilagewas typically 107 per gram wet weight.

Culture conditionsThe culture medium consisted of 3 parts of DMEM to 1 part ofHam's F-12 medium, supplemented with 24.3 mg I"1 adenine,

2xl05unitsl 1 penicillin/streptomycin and 10% FCS (Newmanand Watt, 1988; Watt and Dudhia, 1988). The same batch of FCSwas used for any one experiment.

Cells were seeded on tissue culture plastic at high (1.25 xlO5

cells per cm2) or low (1.25x 104 cells per cm2) density, as describedpreviously (Watt, 1988; Watt and Dudhia, 1988). For someexperiments cells were suspended in medium made viscous byaddition of methylcellulose (Dow Chemical Corp.) to a finalconcentration of 1.2% (Stoker, 1968). Cells were suspended inmethylcellulose at a concentration of 5xlO5 or 5xl04ml"1,corresponding to the same cell/medium ratio as in high- and low-density cultures on tissue culture plastic.

Determination of cell sizeCells were fixed in 3.7% formaldehyde for 8min at roomtemperature and resuspended in PBS. The cells were photo-graphed under bright field using a Zeiss Photomicroscope III andcell diameter determined from the photographs. More than 800cells were scored for each determination.

Proteoglycan synthesisCultures were incubated with S^Ciml"1 [35SJsulphate (Amer-sham, carrier-free, specific activity 25-40 Cimg"1) for 24 h. Themedium was recovered and mixed with an equal volume of 8 Mguanidinium chloride, 0 . 1 M sodium acetate, pH6.8, containingthe protease inhibitors 10 mM phenylmethylsulphonyl fluoride,10 mM EDTA, 10 mM benzamidine hydrochloride and 50 HMra-caproic acid (Oegema et al. 1975; Pearson and Mason, 1977). Thecell layer was extracted for fluorimetric determination of DNAcontent (see below) and then mixed with an equal volume of the8 M guanidinium chloride extraction buffer, for proteoglycananalysis.

After incubation for 48 h at 4°C with rotation, [35S]sulphateincorporation into proteoglycans in the medium or cell extractswas determined by precipitation on Whatman 3 MM filter paperwith 1 % cetyl pyridinium chloride, as described previously(Wasteson et al. 1973; Newman and Watt, 1988), and counted in ascintillation counter. All samples were analysed in duplicate andthe radioactivity was corrected to allow for decay.

Protein synthesisCultures were labelled with lO^Ciml"1 [35S]methionine (Amer-sham, specific activity llSeCimmol"1) for 24h. Cells andmedium were extracted as described above for proteoglycananalysis and precipitated, in the presence of excess BSA (bovineserum albumin; as carrier), onto Whatman GF/C glass microfibrefilters with 10% trichloroacetic acid, TCA (4°C). Filters werewashed twice with 10% TCA, then with absolute ethanol, air-dried, and counted in a scintillation counter. Each sample wasassayed in duplicate and the radioactivity extrapolated back tothe activity date.

Fluorimetric determination of DNA contentCell number was measured by fluorimetric determination of DNAcontent (Karsten and Wollenberger, 1972; Newman and Watt,1988). The cell layer was rinsed twice in PBS; incubated for20 min at 37 °C in 20 fig ml"1 Pronase (DNase-free, Sigma) in PBS;sonicated for 20 s on ice; and stored at -20°C until assayed. Forthe assay, samples or diluted DNA standards were mixed with0.08 ml Pronase, 0.08 ml RNase (Sigma, DNase-free, 125 /(gml"1),and PBS to give a volume of 0.8 ml. After 1 h incubation at 37 °C,0.2 ml ethidium bromide (25^gml"1 in PBS) was added. Fluor-escence was determined in a fluorimeter with an excitationwavelength of 360 nm and an emission wavelength of 580 nm,while stirring the sample. All samples were read within 5-60 minof adding the ethidium bromide. Standard curves of fluorescenceversus known amounts of DNA or numbers of cells wereconstructed.

ImmunofluorescenceCell suspensions were fixed in 3.7 % formaldehyde in PBS at roomtemperature for 8-10 min, then stained with MZ15, a mouse

350 M. Siczkowski and F. M. Watt

monoclonal antibody recognising keratan sulphate (Zanetti et al.1985; Mehmet et al. 1986). The second antibody, fluoresceinatedrabbit anti-mouse immunoglobulin, was purchased from MilesScientific. Antibody incubations were for l h at room tempera-ture. Controls were carried out in which second antibody alonewas used. Preparations were mounted in Gelvatol and examinedwith a Zeiss Photomicroscope III.

Determination of [35S]sulphate incorporation intokeratan sulphateExtracts of proteoglycan in guanidinium hydrochloride weredialysed against 40mM Tris-HCl, pH7.3, for 48h with fourchanges of dialysis buffer. The total amount of incorporated[35S]sulphate was determined by cetyl pyridinium chlorideprecipitation, as described above. The amount of [35S]sulphatepresent in keratan sulphate side-chains was determined by usingchondroitinase ABC to remove the chondroitin sulphate side-chains, leaving chrondroitin sulphate stubs in addition to intactkeratan sulphate side-chains. The validity of this method wasconfirmed by a second method in which chondroitinase ABCtreatment was followed by papain digestion to release individualkeratan sulphate chains with a small core protein peptideattached. The amount of [35S]sulphate in the keratan sulphatechains was determined by applying a 0.5 ml sample to a 5 mlSephadex G50 column (Pharmacia), collecting 10x0.5 ml frac-tions and counting radioactivity in the void volume peak.

Samples were digested with chondroitinase ABC for 5 h at 37 °Cwith 0.05 unit enzyme (Miles Scientific) per 0.5 ml dialysedmedium in 40 mM Tris-HCl buffer, pH 7.3. In the second method,the sample was then digested for 20 h at 60 "C with 2mgml~papain (Sigma, twice recrystallised) in 0 . 1 M sodium acetate,0.02 M EDTA, 0.2 M cysteine hydrochloride, 4 M sodium chloride,pH5.5(Oldberge*aZ. 1977).

Sepharose CL2B gel filtration chromatographyThe method was based on that described by Mitchell andHardingham (1981). Two gel nitration columns of approximatesize 100 cm length by 0.6 cm diameter were packed withSepharose CL2B (Pharmacia). One column, to be run underassociative conditions, was equilibrated with 0.5 M sodiumacetate, pH6.8. This column was used to determine whatproportion of proteoglycans was capable of forming aggregateswith hyaluronic acid. The second column was set up underdissociative conditions in 0.5 M sodium acetate, pH6.8, containing2 M guanidinium hydrochloride (Fluka), to determine proteogly-can monomer size.

Sepharose CL2B has a fractionation range of 7xlO4 to4xlO7 Mr. Molecules larger than 4xlO7 Mr are excluded from theSepharose beads and elute in the void volume (Vo), whilemolecules smaller than 7xlO4Afr enter the column matrix andelute at the total volume (Vt) of the column. The Vo and Vt of eachcolumn were determined using high molecular weight DextranBlue (Pharmacia) and Phenol Red solutions, respectively.

Proteoglycan extracts in guanidinium chloride were dialysedagainst 0.5 M sodium acetate, pH 6.8, for 48 h with four changes ofdialysis buffer. The sample was then divided into two. One halfwas mixed with an equal volume of 0.5 M sodium acetate, pH 6.8,containing 8 M guanidinium chloride and rotated overnight at4°C before application to the dissociative column. A 10 fig portionof hyaluronic acid was added to the second aliquot and the sampleleft overnight at 4°C to aggregate before application to theassociative column. Samples were applied to each column in atotal volume of no greater than 0.7 ml; buffer was pumped out ofthe bottom of each column at a speed of 3-4 ml per h and fractionsof 0.7 ml were collected. The distribution coefficient (ifav) andrelative molecular mass (Mr) of each species resolved on thecolumn were determined as described by Ohno et al. (1986).

Agarose-polyacrylamide composite gel electrophoresisThe method is based on that of McDevitt and Muir (1971), asmodified by Carney et al. (1986), with the further modificationthat a sheet of Gelbond (for agarose gels; Pharmacia) was added to

the glass plate to facilitate handling of the gel. The Gelbond wasomitted when gels were to be immunoblotted.

The gels were 2 mm thick and usually consisted of 1.2%polyacrylamide and 0.6 % agarose, although for some experimentsthe acrylamide concentration was increased to 2.4 %. The runningbuffer was 4 M urea, 10 mM Tris-HCl, 0.25 mM sodium sulphate,lmM EDTA, pH6.8. Gels were run at 60 V for 5min, followed by120 V for 1 h, or until the dye front had migrated 3 cm. Sampleswere usually dialysed extensively against distilled water, freeze-dried, taken up in sample buffer of 8 M urea, 10 % sucrose, 10 mMTris-HCl, 0.25mM sodium sulphate, lmM EDTA, 0.001%Bromophenol Blue, pH6.8, and left at 4°C overnight. In someexperiments the cell layer was extracted directly in sample buffer.

Gels were stained with 0.2 % Toluidine Blue in 0.1 M acetic acidfor 20min; destained in 3% acetic acid for 90min and left indistilled water until clear. For fluorography, gels were impreg-nated with 0.4 % 2,5-diphenyloxazole in absolute ethanol for 16 h,washed in distilled water for 1 h, then dried down under vacuumand exposed to preflashed Fuji X-ray film at —70°C.

In some experiments composite gels were lightly stained withToluidine Blue and destained to visualise the proteoglycan bands.Individual bands were cut out of the gel with a scalpel andelectroeluted using an ISCO electroelution apparatus withcomposite gel running buffer at 3 W overnight. The electroelutedbands were rerun on composite gels to confirm the presence ofsingle bands.

Polyacrylamide gel electrophoresisLinear gradient (4 % to 20 %) polyacrylamide gels were preparedusing the buffer system of Laemmli (1970). Electrophoresis was ata constant voltage of 60 V while the sample was in the stackinggel and at 120 V thereafter. Gels were stained with CoomassieBlue or prepared for fluorography, as described by Bonner andLaskey (1974).

ImmunoblottingBlotting of composite gels or polyacrylamide gels was carried outessentially as described by Towbin et al. (1979), except that thetransfer buffer contained 0.01 % SDS, to aid transfer of largemolecules to nitrocellulose (Erickson et al. 1982). Transfer wascarried out at 50 V, 4°C, overnight. The transfer buffer was thenreplaced with the same buffer without SDS and transfercontinued for a further 2 h.

Nitrocellulose was stained for protein using AuroDye Forte(Janssen), according to the manufacturer's instructions. Antigenswere visualised using AuroProbe BL plus second antibody andsilver enhancer IntenSE II (Janssen), also according to themanufacturer's instructions (Brada and Roth, 1984; Moeremanset al. 1984). When the first antibody recognised chondroitinsulphate stubs, blots were incubated with 0.025 unit ml"1

chondroitinase ABC (Miles Scientific) in blocking buffer for60min at 37 °C, with shaking after the blocking step.

The following mouse monoclonal antibodies were used: l-B-5(recognising nonsulphated chondroitin sulphate), 2-B-6 (chon-droitin 4-sulphate), 3-B-3 (chondroitin 6-sulphate), 5-D-4 (kera-tan sulphate), all generous gifts from Bruce Caterson (Caterson etal. 1983; Couchman et al. 1984; Caterson et al. 1985; Caterson etal. 1986). In addition, MZ15 (keratan sulphate), another mousemonoclonal, was used (Zanetti et al. 1985; Mehmet et al. 1986).Rabbit antiserum to the proteoglycan hyaluronic acid bindingregion was a gift from T. Hardingham (Ratcliffe and Hardingham,1983).

Histochemical staining for alkaline phosphataseCultures were fixed in 70 % ethanol at 4°C for 5 min, rinsed in50 mM Tris-base, pHIO, and stained for 20 min in 50 mMTris-base, pHIO, containing 0.1% a-naphthyl acid phosphate(sodium salt), 0.1% magnesium chloride and 0.1% fast blue RRsalt, which had been mixed and filtered immediately before use(Kiernan, 1981). After staining, the samples were rinsed indistilled water and mounted in glycerol. Controls consisted ofadding the alkaline phosphatase inhibitor levamisole at 4 mM tothe incubation mixture (Reynolds and Dew, 1977) or omitting thesubstrate, tv-naphthyl acid phosphate.

Chondrocytes from different cartilage zones 351

Results

Cell isolationChondrocytes were isolated from two regions of cartilage,designated the upper and lower zones. The upper zone(100-200 ;Um thick) corresponded to the histologicallyrecognised superficial zone plus some of the transitionalzone; the lower zone (200-350/tm thick) corresponded tothe deep zone and occasionally a small amount of thehypertrophic zone. The middle zone that was discardedcontained some transitional and, mainly, deep zonecartilage (Palfrey and Davies, 1966). In order to obtain auniform sample of cartilage from each zone only cartilagefrom the metacarpus of the metacarpo-phalangeal jointwas used, excluding the ridges. The cell yield (per gramwet weight) from upper zone cartilage was higher thanfrom lower zone, as would be expected since there are morecells per unit volume in the upper zone.

The diameters of chondrocytes from each zone weremeasured on isolation and after growth in culture. Freshlyisolated upper zone cells had a modal diameter of 10 [im,

range 8-12 /an. Lower zone cells had the same modaldiameter on isolation, but there appeared to be anadditional size class with a modal diameter of about 15 /im;there were also a few large cells of 17-20 ,um. After 32 daysin culture at high or low density the modal diameter ofboth populations had increased to 17 //m and the range insizes had increased to 14-22 ^an; thus the two subpopu-lations could no longer be distinguished on the basis of cellsize distribution.

Cell morphology in cultureAs previously reported (Zanetti et al. 1985), upper zonecells adhered and spread more slowly in culture than lowerzone cells. After 5 days upper zone cells that had attachedat high or low seeding density were still rounded (Fig. 1 A).In contrast, cells from lower zone cartilage attached andspread within 5 days (Fig. IB). Less than 5 % of cells ineach population failed to attach. The initial failure ofupper zone cells to spread in culture was not due to celldamage during isolation, because reducing the collagen-

Fig. 1. Morphology of cells grown in culture at high density for different lengths of time. (A,B) 5 days; (C,D) 14 days; (E,F) 32days. (A,C,E) Upper zone; (B,D,F) lower zone. Bar: A-D, 200 ^m; E,F, 114 fim.

352 M. Siczkowski and F. M. Watt

E

U

104

co«18C

12000-

10000-

8000-

6000-

4000-

2000-

8000-

6000-

4000-

2000-

0-

D

a j

r* 1

20 25 30

10 15Days

20 30

Fig. 2. Replicate wells of 24-well (2.4 cm2) plates were seeded at high (A,C,E,G) or low (B,D,F,H) density and assayed for: A,B, cellnumber per well; points are means of duplicate determinations from 2 wells, which typically varied by less than 5 % in A and lessthan 8% in B. C-F, [35S]sulphate incorporation per 104 cells; C,D, total counts incorporated into medium and cell layer; E,F, ratioof counts incorporated into medium versus cell layer; points are means of counts from 2 wells. G,H, [36S]methionine incorporation(medium plus cell layer) per 104 cells; one well assayed per point. (•) Upper zone; ( • ) lower zone cells.

ase digestion time to 1.5 h and including 15 % FCS in thecollagenase solution had no effect on spreading.

Cells were grown at high or low density. When cultured

at high density chondrocytes continue to express type IIcollagen and large aggregating proteoglycans for severalweeks, whereas at low density the differentiated pheno-

Chondrocytes from different cartilage zones 353

type of the cells is less stable (Watt, 1988; Watt andDudhia, 1988). After 14 days at high density, differences incell morphology were still apparent. Upper zone culturesconsisted of small clusters of rounded cells separated byflattened cells (Fig. 1C), while lower zone cells were moreuniformly spread and polygonal (Fig. ID). By 32 days athigh density, upper zone cultures no longer containedclusters of rounded cells, but were spread, mainly as amonolayer, with only a few rounded cells (Fig. IE). Inlower zone cultures most of the cells were rounded and thecultures appeared to be multilayered with large amountsof refractile matrix between adjacent cells (Fig. IF).

At low density, differences between the two subpopu-lations were less pronounced, both types of cultureconsisting mainly of flattened cells at 14 days. At 32 daysthe cells were still mainly flattened, but nodules ofrounded cells were evident. Nodules covered a greaterproportion of the lower zone than the upper zone cultures(data not shown).

Growth of chondrocytes from upper and lower zonecartilageChondrocytes isolated from upper and lower zone cartilagewere plated at high or low density in 24-well multiwelldishes, and every 2-3 days for 28 days cell number perwell was determined by DNA fluorimetry. At high densitythe average increase in cell number was less than 2-fold(Fig. 2A), while at low density cell number increased bymore than 10-fold (Fig. 2B). At high density, cell numberper well in upper zone cultures was consistently lowerthan in lower zone cultures. At low density, the growthcurves for the two subpopulations were initially the same,but after day 6 the number of cells per dish in the upper

70

zone cultures was consistently less than in lower zonecultures.

Proteoglycan synthesisProteoglycan synthesis was measured by incorporation of[35S]sulphate into glycosaminoglycans. Upper and lowerzone cells were grown at high or low density andincorporation of label into the medium and cell layer wasmeasured at intervals for 28 days (Fig. 2C,D). At highdensity, incorporation was similar in upper and lower zonecultures for the first 12 days, but thereafter incorporationwas greater in lower zone cultures. At low density,incorporation was initially greater in upper zone cultures,but from day 15 onwards incorporation was similar in bothzones. The level of [35S]sulphate incorporation wasinitially greater in high density than low density cultures,as observed previously (Watt, 1988).

More pronounced differences between upper and lowerzone cells were observed when the ratio of [35S]sulphatecounts incorporated into the medium versus the cell layerwas plotted (Fig. 2E,F). The ratio was consistently higherin upper than in lower zone cultures, a difference that wasmore pronounced at high (Fig. 2E) than at low (Fig. 2F)density.

In all of these experiments cells from full-depthcartilage were included for comparison and consistentlygave levels of incorporation intermediate between upperand lower zone cells.

Protein synthesisThe differences in proteoglycan synthesis did not reflectdifferences in total protein synthesis. There were no majordifferences in total [35S]methionine incorporation between

Fig. 3. Percentage of cells stained with MZ15 after seeding at: (A,C) high density; (B,D) low density. (A,B) Cells grown on tissueculture plastic; (C,D) cells grown in suspension in methylcellulose. In A,B at least 250 cells were counted per time-point and atleast 400 in C,D. Results in A,B are means of duplicate dishes. (•) Upper zone; (•) lower zone cells.

354 M. Siczkowski and F. M. Watt

Fig. 4. [35S]sulphate incorporation into keratan sulphate,expressed as a percentage of total [35S]sulphate incorporation.Cells were grown at: (A) high density; (B) low density.(•) Upper zone cells; ( • ) lower zone cells.

upper, lower and full-depth cells, at high or low density(Fig. 2G,H and data not shown). In all situations there wasa decrease in incorporation during the first two weeks,followed by a transient increase during the third week.

Keratan sulphate expressionThe keratan sulphate content of cartilage proteoglycanincreases with depth (Stockwell and Scott, 1967; Kempsonet al. 1973; Venn, 1978). Zanetti et al. (1985) reported thatthe proportion of cells from upper zone cartilage thatstained positively with MZ15, a monoclonal antibody tokeratan sulphate, was less than the proportion of cellsfrom deep cartilage when freshly isolated or incubatedovernight in suspension, but that with prolonged culti-vation at high density the difference disappeared. In orderto extend these observations, we measured both thepercentage of MZ15-positive cells and the keratan sul-phate content of proteoglycan synthesised at differenttimes in culture (Figs 3, 4).

In high-density lower zone cultures the percentage ofMZ15-positive cells fell from approximately 55 % to 42 %within 2 days and declined only slightly thereafter. Inhigh-density upper zone cultures the number of keratansulphate-positive cells rose from 16% to 30% within 2days and remained between 30 and 38% thereafter(Fig. 3A). In low density cultures the number of keratansulphate-positive cells initially rose in the upper zonecultures and fell in the lower zone cultures to reach about33 % by day 7. In upper zone cultures this fell to 23 % by 28days, whereas in lower zone cultures, the proportion ofpositively stained cells remained approximately constantat 35% (Fig. 3B).

Fig. 3C,D shows the percentage of MZ15-positive cells in

upper and lower zone cultures maintained in suspensionin methylcellulose for up to 7 days. In each case theproportion of positively stained upper zone cells rapidlyincreased to the same level as in lower zone suspensions,subsequently remaining constant at high density, butfalling at low density.

The keratan sulphate content of proteoglycan synthe-sised by upper and lower zone cells in culture is shown inFig. 4. [35S]sulphate-labelled proteoglycan extracted fromthe culture medium was digested with chondroitinaseABC and the amount of incorporated sulphate wasdetermined by cetyl pyridinium chloride precipitation.The majority of incorporated [35S]sulphate remainingafter digestion would be in keratan sulphate, although asmall proportion would be in the chondroitin sulphatestubs. In high-density cultures the keratan sulphatecontent of the proteoglycan was consistently greater inlower than upper zone cultures (Fig. 4A); this could reflectan increase in the number of keratan sulphate chains, inchain length or in sulphation. In low-density cultures,however, the keratan sulphate content of proteoglycansynthesised by both zones decreased with time and therewas no significant difference between the two cellpopulations (Fig. 4B).

Sepharose CL2B profiles of'3S'S-labelled proteoglycansextracted from different cartilage zones and fromcultures of upper and lower zone chondrocytesCartilage slices from different depths of cartilage werelabelled for 24 h with lO^Ciml"1 [35S]sulphate in culturemedium containing 10 % FCS. The medium was removed,the cartilage slices rinsed twice in PBS and extracted for48 h in 4 M guanidinium hydrochloride buffer. Afterextensive dialysis the samples were run on SepharoseCL2B columns under associative and dissociative con-ditions.

Under associative conditions most of the proteoglycanfrom both zones formed aggregates with hyaluronic acid,but the proportion of nonaggregating proteoglycan wasgreater in upper zone than lower zone cartilage (30% ofincorporated label compared with 14%) (Fig. 5A,B). Underdissociative conditions the proteoglycan that failed toaggregate under associative conditions appeared as ashoulder to the main monomer peak and was morepronounced in upper than lower zone cartilage (data notshown). From the elution positions of the main proteo-glycan species and the less-abundant species, molecularweights of 1.68±0.26xl06Mr (#av=0.22) and4.4±O.9xlO5 Mr (.Kav=0.57), respectively, were calculated(Mr values are means of 5 determinations ± standarddeviation).

[35S]sulphate-labelled proteoglycans secreted into themedium by cultures of upper and lower zone chondrocyteswere also analysed by Sepharose CL2B chromatography.After 4 days at low (Fig. 5C,D) or high density (data notshown) there was a single aggregating species with a Kavof approximately 0.21 (1.71xlO6Mr). Thus the low Mrproteoglycan that was a feature of upper zone cartilageprofiles (Fig. 5A) was not observed after 4 days in culture.

With time in low density culture, changes in theproteoglycans synthesised were observed. At day 12 andfor the remainder of the culture period a small non-aggregating proteoglycan was synthesised by both types ofculture (Fig. 5E,F, and results not shown). This elutedfrom Sepharose CL2B columns with a Kav of 0.56,corresponding to a Mr of about 4.5 xlO5. At 42 days themajority of the proteoglycan still aggregated with hyalu-

Chondrocytes from different cartilage zones 355

20000

25000

20000

10 20 30

Elution volume (ml)

40

25000

20000-

15000-

10000-

5000-

10 20 30

Elution volume (ml)

40

10 20 30Elution volume (ml)

40

10 20 30Elution volume (ml)

40

10 20 30

Elution volume (ml)40

20000-

15000-

10000-

5000-

0-

F

1

il

10 20 30Elution volume (ml)

40

Fig. 5. Sepharose CL2B chromatography under associative conditions. [35S]sulphate-labelled proteoglycans were extracted from:(A) upper zone (B) lower zone cartilage or from the medium of upper zone (C,E) or lower zone (D,F) chondrocytes, that had beencultured for 4 days (C,D) or 42 days (E,F) at low density. Vo and Vt are indicated by arrows (left and right, respectively).

ronic acid, but upper zone cultures contained a higherproportion of the non-aggregating proteoglycan thanlower zone cultures (27 % of total incorporated [35S]sul-phate versus 16%). Thus with time in culture, the columnprofiles came to resemble those of proteoglycans synthe-sised in cartilage slices (Fig. 5A,B,E,F).

Composite gel electrophoresis of proteoglycans fromdifferent cartilage zonesProteoglycan extracted from the different cartilage zoneswas analysed by composite gel electrophoresis. ToluidineBlue staining snowed that there was a major band of highmolecular weight in both samples and an additional minorspecies of slightly faster mobility in extracts of upper, butnot lower, zone cartilage. A third, faint, band, migratingclose to the dye front, was present in both cartilageextracts, but was more abundant in upper zone extracts(Fig. 6A).

The relative levels of the two high molecular weightproteoglycans varied between preparations, but the fastermobility band was always less abundant. Both bands werepresent in extracts of freshly isolated upper zone cartilage,in cartilage slices cultured for 24 h and in medium inwhich the slices had been incubated (Fig. 6B).

Mobility on composite gels is a function not only of size,but of glycosaminoglycan composition (McDevitt andMuir, 1971; Carney et al. 1986) and so molecular weightscannot be assigned solely on the basis of mobility.However, if the two high molecular weight bands are bothlabelled with [35S]sulphate, it is likely that they elute inthe major peak on dissociative Sepharose CL2B columnsCK"av=0.22), whereas the low molecular weight band wouldcorrespond to the material that elutes with a Kav of 0.57.

Immunoblotting of cartilage proteoglycansImmunoblotting was used to characterise the two high

356 M. Siczkowski and F. M. Watt

1 2 3

B2 3

c l

i- M6*•

8

5

7

Fig. 6. Composite gel electrophoresis of cartilageproteoglycans. (A) Toluidine Blue staining; proteoglycan wasextracted from: 1, upper zone; 2, lower zone; 3, full-depthcartilage. Arrows mark the bands discussed in the text.(B) Toluidine Blue-stained gel of upper zone proteoglycansextracted from: 1, cartilage maintained in culture for 24 h;2, freshly isolated cartilage; 3, culture medium in whichcartilage had been incubated for 24 h. (C) Immunoblotting ofproteoglycan from full-depth cartilage. The two high molecularweight bands (see A,B) were dissected from the gel,electroeluted and then re-run in separate tracks of a secondgel. In each case the slower migrating band is on the left-handside of each pair of tracks: 1, stained with AuroDye; 2, MZ15;3, 5-D-4; 4, 3-B-3 after chondroitinase ABC predigestion;5, 3-B-3 without chondroitinase digestion; 6, 2-B-6 withchondroitinase ABC predigestion; 7, l-B-5 with chondroitinaseABC predigestion; 8, rabbit antiserum to the hyaluronic acid-binding region.

molecular weight proteoglycans resolved by composite gelelectrophoresis. Since the two bands migrated very closetogether they were cut out of the gels, electroeluted and re-run prior to transfer to nitrocellulose. Staining thenitrocellulose for protein showed that the individual highmolecular weight proteoglycans migrated to the samerelative positions as in the unfractionated sample. Thelower band was slightly contaminated with the upperband, but could be clearly distinguished by its positionrelative to the upper band on the same gel (Fig. 6C, 1).

Antibodies MZ15 and 5-D-4, which recognise keratansulphate, reacted with both bands, although the upperband appeared to stain more intensely (Fig. 6C, 2,3). Whenthe proteoglycans on nitrocellulose were digested with

chondroitinase ABC, 3-B-3 (recognising chondroitin 6-sulphate) reacted strongly with both bands (Fig. 6C, 4);however, if the digestion step was omitted no reactivitywas observed (Fig. 6C, 5). After chondroitinase digestion,2-B-6 (recognising chondroitin 4-sulphate) stained theupper band more strongly than the lower band (Fig. 6C, 6)while l-B-5 (recognising non-sulphated chondroitin sul-phate) reacted very weakly with both bands (Fig. 6C, 7).Rabbit antiserum against the hyaluronic acid-bindingregion reacted more strongly with the upper band than thelower band (Fig. 6C, 8). The results show that bothproteoglycans contained keratan sulphate and chondroitinsulphate side-chains, but that the lower band containedrelatively less keratan sulphate, chondroitin 4-sulphateand hyaluronic acid-binding region than the upper band.

Electrophoretic analysis of proteoglycans synthesised incultureMedium from [35S]sulphate-labelled upper and lower zonecultures was analysed by composite gel electrophoresis atintervals of from 2 to 27 days (Fig. 7A and results notshown). No significant differences between upper andlower zone cultures were observed. In high-densitycultures some of the samples resolved into two bands thatmigrated close together (Fig. 7A), while at low densityonly one band could be detected (results not shown).

In order to compare the small non-aggregating proteo-glycans (Kav=0.56) detected after prolonged culture at lowdensity with the smallest proteoglycan synthesised inorgan culture of cartilage slices, [35S]sulphate labelled celland cartilage extracts were subjected to electrophoresis ongradient gels of 4% to 20% acrylamide (Fig. 7B). Themajor high molecular weight proteoglycans CKav=0.22) donot enter this type of gel. As shown in Fig. 7B, extracts ofupper and lower zone cartilage contained only a singlesulphated species that had an Mr of approximately300 xlO3. Cultured chondrocytes synthesised a300xl03Mr band when grown at high density (Fig. 7B).However, in low-density cultures, an additional band of125xlO3Afr was observed (Fig. 7B,C).

Alkaline phosphataseAlkaline phosphatase is a marker of calcifying cartilage.32-day- and 42-day-old high- and low-density upper andlower zone cultures were stained for alkaline phosphatase.Upper zone cultures at both densities were virtuallyunstained, while lower zone cultures at both densitiesstained intensely (Fig. 8 and results not shown).

Discussion

In this report we have compared the behaviour of upperand lower zone chondrocytes in culture at high and lowdensity. On isolation upper zone cells were smaller thanlower zone cells, as reported previously (Zanetti et al.1985), but with time in culture this difference disappeared.Again, as reported previously, upper zone cells initiallyshowed less attachment and spreading (Zanetti et al.1985). Morphological differences between the two cellpopulations persisted for several weeks and were morepronounced in cultures seeded at high than those seeded atlow density: the lower zone cells deposited a moreextensive extracellular matrix than the upper zone cells.The growth rates for both populations were similar, butupper zone cultures reached a lower final density.[ S]methionine incorporation into total protein was

Chondrocytes from different cartilage zones 357

B2 3 4 5 1 23 4 5 6 7 8

LUJ

Fig. 7. Electrophoresis of [35S]sulphate-labelled proteoglycans. (A) Composite gels of proteoglycans secreted into the medium byhigh-density cultures of (top row) upper or (bottom row) lower zone chondrocytes. Cells were in culture for: 1, 9 days; 2, 16 days;3, 25 days. (B,C) [35S]sulphate-labelled proteoglycans resolved on 4 % to 20 % polyacrylamide gels. Equal numbers of counts wereloaded per track. Arrowheads indicate Mr (xlO~3) markers: 214, 111, 68, 45, 24 and 18. (B) 1,2. Extracts of chondrocytes grown atlow density for 70 days: 1, upper zone; 2, lower zone; 3-5. Cartilage extracts: 3, upper zone; 4, lower zone; 5, full depth. (C) Mediumfrom cultures of full-depth chondrocytes grown at low density (1-4) or high density (5-8) for 7 days (1,5), 20 days (2,6), 28 days(3,7), 37 days (4,8).

similar for both populations, but upper zone cells depositedless proteoglycan in the cell layer. The latter observationfits well with the in vivo data showing that superficialcartilage contains a lower concentration of proteoglycanthan deep cartilage (Maroudas et al. 1969; Franzen et al.1981; Bayliss et al. 1983; Ratcliffe et al. 1984). Our resultsare consistent with the observations of Aydelotte and co-workers (Aydelotte and Kuettner, 1988; Aydelotte et al.1988) on chondrocytes from different zones of bovinearticular cartilage: they found that superficial zone cellswere smaller on isolation than deep zone cells and thatdeep zone cells produced more proteoglycan and retained agreater proportion in the extracellular matrix.

The keratan sulphate content of proteoglycan synthe-sised by lower zone chondrocytes in high density culturewas greater than that of upper zone cells, consistent withthe observation that the keratan sulphate content ofcartilage increases with depth (Stockwell and Scott, 1967;Kempson et al. 1973; Venn, 1978). At low density thekeratan sulphate content of proteoglycan synthesised byboth subpopulations was not significantly different anddeclined with time. It is interesting that the twopopulations could be distinguished in this way, when theycould not be distinguished by MZ15 staining (Zanetti et al.1985; and Fig. 3); this is presumably because cell surfacestaining with the antibody can give an indication of thepresence or absence of keratan sulphate, but not of relativeabundance.

The proteoglycans of upper and lower zone cartilage andof chondrocyte cultures were compared in more detail. Theproportion of proteoglycan that did not aggregate withhyaluronic acid was greater in upper than lower zonecartilage; this difference was lost in early cultures, butreappeared in long-term cultures. Aydelotte et al. (1988)have also reported that superficial bovine articularchondrocytes in culture synthesise less aggregatingproteoglycan than deep cells.

Composite gel electrophoresis resolved two high molecu-lar weight proteoglycans in cartilage, the less abundant ofwhich was unique to the upper zone. However, in culturetwo bands, when resolved, were present in both upper andlower zone proteoglycans. The cartilage bands werefurther studied by immunoblotting: we established thatboth bands contained chondroitin sulphate and keratansulphate side-chains. The less-abundant band stainedmore weakly for keratan sulphate, chondroitin 4-sulphateand hyaluronic acid-binding region. One explanation forthese results is that the lower band is generated from theupper band by limited proteolysis (Hascall et al. 1983;Campbell et al. 1984), but further experiments arerequired to test this possibility. A number of other studieshave revealed proteoglycan heterogeneity, by usingcomposite gel electrophoresis. Changes in gel bandpatterns as a function of cartilage depth (Franzen et al.1981; Bayliss et al. 1983), anatomical site (Stanescu et al.1980) and age (Bayliss and Ali, 1978) have all been

358 M. Siczkowski and F. M. Watt

differences between the cells of different zones, whereassome probably reflects the response of chondrocytes tolocal differences in their environment.

This project was initiated at the Kennedy Institute ofRheumatology and supported by a grant to Fiona Watt from theArthritis and Rheumatism Council. We are grateful to PaulNewman, Tim Hardingham and Mike Bayliss for advice and toLewis Wolpert for providing laboratory space for Martin Sicz-kowski. We thank Wendy Senior for expert typing of themanuscript.

8A

Fig. 8. Cultures stained for alkaline phosphatase activity.Cells were grown for 42 days at high density. (A) Upper zone;(B) lower zone. Bar, 200 fim.

reported and there is also variation between animalspecies.

Small, non-aggregating proteoglycans were analysed bygradient polyacrylamide gel electrophoresis. In cartilageand in cultures seeded at high density, a single diffuseband of 300 x 103 Mr was observed, but in low densitycultures an additional band of 125 x 103 MT was found. Thedifferentiated phenotype of chondrocytes is less stable inlow-density cultures (Watt and Dudhia, 1988; Watt, 1988)and this may be the explanation for the appearance of the125 x 103 Mr band. Further studies are necessary toestablish the composition of both the 300 x 103 Mr and the125 xlO3Mr proteoglycans.

Chondrocytes in culture can undergo hypertrophy andcalcification (see, for example, Tacchetti et al. 1987), aprocess that can be regulated by growth factors or otherserum components (Bruckner et al. 1989; Kato andIwamoto, 1990). We examined expression of one marker ofcalcification, alkaline phosphatase, in upper and lowerzone cultures and found that it was only expressed bylower zone cells, in keeping with their relative proximity,in vivo, to the underlying bone.

In conclusion, some of the differences between upper andlower cartilage zones, such as cell size and percentage ofMZ15-positive cells, were lost in culture, whereas others,such as the keratan sulphate content of proteoglycan,proportion of proteoglycan deposited in the cell layer andproportion of nonaggregating proteoglycan synthesised,were retained. In addition, differences between upper andlower zone cultures, which did not have a strict in vivocorrelate, including cell morphology and confluent den-sity, were observed. We therefore conclude that some of theheterogeneity of articular cartilage reflects intrinsic

References

AYDELOTTE, M. B., GEEENHILL, R. R. AND KUETTNER, K. E. (1988).Differences between sub-populations of cultured bovine articularchondrocytes. II. Proteoglycan metabolism. Conn. Tiss. Res. 18,223-234.

AYDELOTTE, M. B. AND KUETTNER, K. E. (1988). Differences between sub-populations of cultured bovine articular chondrocytes. I. Morphologyand cartilage matrix production. Conn. Tiss. Res. 18, 205-222.

BAYLISS, M. T. AND ALI, S. Y. (1978). Age-related changes in thecomposition and structure of human articular-cartilage proteoglycan.Biochem. J. 176, 683-693.

BAYLISS, M. T., VENN, M., MAROUDAS, A. AND ALI, S. Y. (1983).Structure of proteoglycans from different layers of human articularcartilage. Biochem. J. 209, 387-400.

BONNER, W. M. AND LASKEY, R. A. (1974). A film detection method fortritium-labelled proteins and nucleic acids in polyacrylamide gels.Eur. J. Biochem. 46, 83-88.

BRADA, D. AND ROTH, J. (1984). 'Golden Blot' - detection of polyclonaland monoclonal antibodies bound to antigens on nitrocellulose byProtein A-gold complexes. Analyt. Biochem. 142, 79-83.

BRUCKNER, P., HORLER, I., MENDLER, M., HOUZE, Y., WINTERHALTER, K.H., EICH-BENDER, S. G. AND SPYCHER, M. A. (1989). Induction andprevention of chondrocyte hypertrophy in culture. J. Cell Biol. 109,2537-2545.

CAMPBELL, M. A., HANDLEY, C. J., HASCALL, V. C, CAMPBELL, R. A. ANDLOWTHER, D. A. (1984). Turnover of proteoglycans in cultures ofbovine articular cartilage. Archs Biochem. Biophys. 234, 275—289.

CARNEY, S. L., BAYLISS, M. T., COLLIER, J. M. AND MUIR, H. (1986).Electrophoresis of 35S-labelled proteoglycans on polyacrylamide-agarose composite gels and their visualisation by fluorography.Analyt. Biochem. 156, 38-44.

CATERSON, B., CALABRO, T., DONOHUE, P. J. AND JAHNKE, M. R. (1986).Monoclonal antibodies against cartilage proteoglycan and linkprotein. In Articular Cartilage Biochemistry (ed. Kuettner, K.), pp.59-73. Raven Press, New York.

CATERSON, B., CHRISTNER, J. E. AND BAKER, J. R. (1983). Identificationof a monoclonal antibody that specifically recognizes corneal andskeletal keratan sulphate. J. biol. Chem. 258, 8848-8854.

CATERSON, B., CHRISTNER, J. E., BAKER, J. R. AND COUCHMAN, J. R.(1985). Production and characterization of monoclonal antibodiesdirected against connective tissue proteoglycans. Fedn Proc. Fedn Am.Socs exp. Biol. 44, 386-393.

COUCHMAN, J. R., CATERSON, B., CHRISTNER, J. E. AND BAKER, J. R.(1984). Mapping by monoclonal antibody detection ofglycosaminoglycans in connective tissue. Nature 307, 650—652.

ERICKSON, P. F., MINIER, L. N. AND LASHER, R. S. (1982). Quantitativeelectrophoretic transfer of polypeptides from SDS polyacrylamide gelsto nitrocellulose sheets: a method for their re-use inimmunoautoradiographic detection of antigens. J. Immun. Meth. 51,241-249.

FORTUNA, R., ANDERSON, H. C, CARTY, R. P. AND SAJDERA, S. W. (1980).Enzymatic characterization of the matrix vesicle alkaline phosphataseisolated from bovine fetal epiphyseal cartilage. Calc. Tiss. Int. 30,217-225.

FRANZ£N, A., INEROT, S., HEJDERUP, S-O. AND HEINEGARD, D. (1981).Variations in the composition of bovine hip articular cartilage withdistance from the articular surface. Biochem. J. 195, 535-543.

GHADIALLY, F. N. (1978). Fine structure of joints. In The Joints andSynovial Fluid, vol. 1 (ed. L. Sokoloff), pp. 105-176. Academic Press,NY, London.

HASCALL, V. C, MORALES, T. I., HASCALL, G. K., HANDLEY, C. J. ANDMCQUILLAN, D. J. (1983). Biosynthesis and turnover of proteoglycansin organ culture of bovine articular cartilage. J. Rheum. (Suppl.) 11,45-52.

Hsu, H. H. T., MUNOZ, P. A., BARR, J., OPPLIGER, I., MORRIS, D. C,VAANANEN, H. K., TARKENTON, N. AND ANDERSON, H. C. (1985).Purification and partial characterization of alkaline phosphatase of

Chondrocytes from different cartilage zones 359

matrix vesicles from fetal bovine epiphyseal cartilage. J. biol. Chem.260, 1826-1831.

KARSTEN, U. AND WOLLENBEROER, A. (1972). Determination of DNA andRNA in homogenized cells and tissues by surface fluorometry. Analyt.Biochem. 46, 135-148.

KATO, Y. AND IWAMOTO, M. (1990). Fibroblast growth factor is aninhibitor of chondrocyte terminal differentiation. J. biol. Chem. 265,5903-5909.

KEMPSON, G. E., Mum, H., POLLARD, C. AND TUKE, M. (1973). Thetensile properties of the cartilage of human femoral condyles relatedto the content of collagen and glycosaminoglycans. Biochim. biophys.Ada 297, 456-472.

KIELTY, C. M., KWAN, A. P. L., HOLMES, D. F., SCHOR, S. L. AND GRANT,M. E. (1985). Type X collagen, a product of hypertrophic chondrocytes.Biochem. J. 227, 545-554.

KIERNAN, J. A. (1981). Histological and Histochemical Methods: Theoryand Practice. Pergamon Press, Oxford.

KUIJER, R., VAN DE STADT, R. J., VAN KAMPEN, G. P. J., DE KONING, M.H. M. T., VAN DE VOORDE-VISSERS, E. AND VAN DER KORST, J. K.(1986). Heterogeneity of proteoglycans extracted before and aftercollagenase treatment of human articular cartilage. Arthritis Rheum.29, 1248-1255.

KWAN, A. P. L., FREEMONT, A. J. AND GRANT, M. E. (1986).Immunoperoxidase localization of type X collagen in chick tibiae.Biosci. Rep. 6, 155-162.

LAEMMLI, U. K. (1970). Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature 227, 680-685.

MANICOURT, D. H., PITA, J. C, MODEVITT, C. A. AND HOWELL, D. S.(1988). Superficial and deeper layers of dog normal articular cartilage.Role of hyaluronate and link protein in determining thesedimentation coefficients distribution of the nondissociativelyextracted proteoglycans. J. biol. Chem. 263, 13121-13129.

MAROUDAS, A., MUIR, H. AND WINGHAM, J. (1969). The correlation offixed negative charge with glycosaminoglycan content of humanarticular cartilage. Biochim. biophys. Ada 177, 492-500.

MCDEVITT, C. A. AND MUIR, H. (1971). Gel electrophoresis ofproteoglycans and glycosaminoglycans on large pore compositepolyacrylamide agarose gels. Analyt. Biochem. 44, 612-622.

MEHMET, H., SCUDDER, P., TANG, P. W., HOUNSELL, E. F., CATERSON, B.AND FEIZI, T. (1986). The antigenic determinants recognized by threemonoclonal antibodies to keratan sulphate involve hepta or largeroligosaccharides of the poly (N-acetyllactosamine) series. Bur. J.Biochem. 157, 385-391.

MITCHELL, D. AND HARDINGHAM, T. E. (1981). The effects ofcycloheximide on the biosynthesis and secretion of proteoglycans bychondrocytes in culture. Biochem. J. 196, 521-529.

MOEREMANS, M., DANEELS, G., VAN DUCK, A., LANGANGER, G. AND DEMEY, J. (1984). Sensitive visualization of antigen-antibody reactionsin dot and blot immune overlay assays with immunogold andimmunogold/silver staining. J. immun. Meth. 74, 353-360.

NEWMAN, P. AND WATT, F. M. (1988). Influence of cytochalasin D-induced changes in cell shape on proteoglycan synthesis by culturedarticular chondrocytes. Expl Cell Res. 178, 199-210.

OEGEMA, T. R., HASOALL, V. C. AND DZIEWIATKOWSKI, D. D. (1975).Isolation and characterization of proteoglycans from the Swarm ratchondrosarcoma. J. biol. Chem. 250, 6151-6159.

OHNO, H., BLACKWELL, J., JAMIESON, A. M., CARRINO, D. A. ANDCAPLAN, A. I. (1986). Calibration of the relative molecular mass ofproteoglycan subunit by column chromatography on Sepharose CL2B.Biochem. J. 235, 553-557.

OLDBERG, A., HOOK, M., OBRINK, B., PERTOFT, H. AND RUBIN, K. (1977).Structure and metabolism of rat liver heparan sulphate. Biochem. J.164, 75-81.

PALFREY, A. J. AND DA VIES, D. V. (1966). The fine structure ofchondrocytes. J. Anat. 100, 213-226.

PEARSON, J. P. AND MASON, R. M. (1977). The stability of bovine nasalcartilage proteoglycans during isolation and storage. Biochim.biophys. Ada 498, 176-188.

POOLE, A. R., PIDOUX, I., REINER, A. AND ROSENBERG, L. (1982). Animmunoelectron microscope study of the organization of proteoglycanmonomer, link protein, and collagen in the matrix of articularcartilage. J. Cell Biol. 93, 921-937.

RATCLIFFE, A., FRYER, P. R. AND HARDINGHAM, T. E. (1984). Thedistribution of aggregating proteoglycans in articular cartilage:Comparison of quantitative immunoelectron microscopy withradioimmunoassay and biochemical analysis. J. Histochem. Cytochem.32, 193-201.

RATCLIFFE, A. AND HARDINGHAM, T. (1983). Cartilage proteoglycanbinding region and link protein. Radioimmunoassays and thedetection of masked determinants in aggregates. Biochem. J. 213,371-378.

REYNOLDS, J. J. AND DEW, G. W. (1977). Comparison of the inhibition ofavian and mammalian bone alkaline phosphatases by levamisole andcompound. R8231. Experientia 33, 154-155.

SCHMID, T. M. AND LINSENMAYER, T. F. (1985). Immunohistochemicallocalization of short chain cartilage collagen (type X) in avian tissues.J. Cell Biol. 100, 598-605.

STANESCU, V., MAROTEAUX, P. AND SOBCZAK, E. (1980). Proteoglycanpopulations of baboon (Papio papio) cartilages from differentanatomical sites. Gel electrophoretic analysis of dissociatedproteoglycans and of fractions obtained by density gradientcentrifugation. Biochim. biophys. Ada 629, 371-381.

STOCKWELL, R. A. AND SCOTT, J. E. (1967). Distribution of acidglycosaminoglycans in human articular cartilage. Nature 215,1376-1378.

STOKER, M. (1968). Abortive transformation by polyoma virus. Nature218, 234-238.

TACCHETTI, C, QUATRO, R., NITSCH, L., HARTMANN, D. J. ANDCANCEDDA, R. (1987). In vitro morphogenesis of chick embryohypertrophic cartilage. J. Cell Biol. 105, 999-1006.

TOWBIN, H., STAEHELIN, T. AND GORDON, J. (1979). Electrophoretictransfer of proteins from polyacrylamide gels to nitrocellulose sheets:procedure and some applications. Proc. natn. Acad. Sci. U.S.A. 76,4350-4354.

VENN, M. F. (1978). Variation of chemical composition with age inhuman femoral head cartilage. Ann. rheum. Dis. 37, 168-174.

WASTESON, A., UTHNE, K. AND WESTERMARK, B. (1973). A novel assayfor the biosynthesis of sulphated polysaccharide and its application tostudies on the effects of somatomedin on cultured cells. Biochem. J.136, 1069-1074.

WATT, F. M. (1988). Effect of seeding density on stability of thedifferentiated phenotype of pig articular chondrocytes in culture. J.Cell Sci. 89, 373-378.

WATT, F. M. AND DUDHIA, J. (1988). Prolonged expression ofdifferentiated phenotype by chondrocytes cultured at low density on acomposite substrate of collagen and agarose that restricts cellspreading. Differentiation 38, 140-147.

ZANETTI, M., RATCLIFFE, A. AND WATT, F. M. (1985). Two subpopulationsof differentiated chondrocytes identified with a monoclonal antibody tokeratan sulfate. J. Cell Biol. 101, 53-59.

(Received 8 May 1990 - Accepted 22 June 1990)

360 M. Siczkowski and F. M. Watt