hyaluronic acid in cartilage and proteoglycan aggregation

565Biochem. J. (1974) 139, 565-581Printed in Great Britain

Hyaluronic Acid in Cartilage and Proteoglycan Aggregation

By TIMOTHY E. HARDINGHAM and HELEN MUIRKennedy Institute ofRheumatology, Bute Gardens, London W6 7D W, U.K.

(Received 21 November 1973)

1. Dissociation of purified proteoglycan aggregates was shown to release an interactingcomponent of buoyant density higher than that of the glycoprotein-link fraction ofHascall & Sajdera (1969). 2. This component, which produced an increase in hydro-dynamic size of proteoglycans on gel chromatography, was isolated by ECTEOLA-cellulose ion-exchange chromatography and identified as hyaluronic acid. 3. The effectofpH of extraction showed that the proportion of proteoglycan aggregates isolated fromcartilage was greatest at pH4.5. 4. The proportion of proteoglycans able to interact withhyaluronic acid decreased when extracted above or below pH4.5, whereas the amountof hyaluronic acid extracted appeared constant from pH 3.0 to 8.5. 5. Sequential extrac-tion of cartilage with 0.15M-NaCl at neutral pH followed by 4M-guanidinium chloride atpH4.5 was shown to yield predominantly non-aggregated and aggregated proteoglycansrespectively. 6. Most of the hyaluronic acid in cartilage, representing about 0.7% of thetotal uronic acid, was associated with proteoglycan aggregates. 7. The non-aggregatedproteoglycans were unable to interact with hyaluronic acid and were of smaller size,lower protein content and lower keratan sulphate content than the disaggregatedproteoglycans. Together with differences in amino acid composition this suggested thateach type of proteoglycan contained different protein cores.

Proteoglycans of cartilage are complex hetero-polysaccharides in which a large number ofchondroitin sulphate and keratan sulphate chains arelinked to a polypeptide backbone. A large proportioncan be extracted with 4M-guanidinium chlorideand have been shown to contain both aggregatedand non-aggregated molecules (Sajdera & Hascall,1969; Rosenberg et al., 1970; Tsiganos et al., 1971;Mashburn & Hoffman, 1971). Equilibrium density-gradient centrifugation of the purified proteoglycansfrom bovine nasal septum in the presence of4M-guanidinium chloride was shown to dissociateaggregates and separate a protein-rich fractionfrom the majority of the proteoglycans (Hascall &Sajdera, 1969). This fraction was found to be neces-sary for the reassociation of proteoglycans intoaggregates and was therefore termed 'glycoprotein-link'. A similar fractionation was applied toproteoglycans from pig laryngeal cartilage (Tsiganoset al., 1971), but the protein-rich fraction, althoughbinding to proteoglycans, did not promote aggrega-tion (Tsiganos et al., 1972). However, a fractionfrom the gradient, of density (1.45-1.57g/ml)intermediate between the protein-rich componentand the proteoglycan, was found to interactwith the proteoglycan, producing an increase inits apparent size on gel chromatography (Tsiganoset al., 1972). This effect was also produced bycomparable fractions of proteoglycans from varioustypes ofcartilage, including bovine nasal septum, ThisVol. 139

suggested that there was a component in the protein-rich fraction of Hascall & Sajdera (1969) ofhigher density that could be separated under theconditions used by Tsiganos et al. (1971). Theincrease in size of proteoglycans shown by gelchromatography was not accompanied by an increasein rate of sedimentation in the ultracentrifuge(R. Pain, personal communication). The effect thusdid not constitute the complete re-formation ofproteoglycan aggregates as described by Hascall &Sajdera (1969), which was characterized by a largeincrease in sedimentation coefficient. The isolationand characterization of the component of higherbuoyant density and its distribution in aggregated andnon-aggregated proteoglycan fractions is describedhere, and its interaction with proteoglycans examined.A preliminary report of part of this work has beenpublished (Hardingham & Muir, 1973a).

Materials

All reagents were of analytical grade exceptfor galactosamine hydrochloride, glucosamine hydro-chloride, glucuronolactone, carbazole, guanidiniumchloride and acetylacetone. The guanidinium chloridewas purified with activated charcoal (Norit N.K.,Hopkin and Williams, Chadwell Heath, Essex, U.K.)and acetylacetone was redistilled (b.p. 133-134°C).

Cetylpyridinium chloride, special grade (lot30601), was obtained from AB Kabi, Stockholm,

T. E. HARDINGHAM AND H. MUIR

Sweden. Chondroitin 4-sulphate was prepared frompig laryngeal cartilage as described by Hardingham& Muir (1970). Hyaluronic acid (from umbilical cord)and twice-crystallized papain (4units/mg of protein)were obtained from BDH Chemicals Ltd., Poole,Dorset, U.K. Crude papain (type II) was obtainedfrom Sigma Chemical Co., St. Louis, Mo., U.S.A.,and testicular hyaluronidase (l000units/mg) fromBoehringer, Mannheim, W. Germany. Gelatin(Bacto) was obtained from Difco Laboratories,Detroit, Mich., U.S.A., and Visking tubing, width24/32 in and 8/32 in, for dialysis from ScientificInstruments Centre, London W.C.1, U.K.

Methods

Analytical methods

Uronic acid was determined by an automatedmodification (Heinegard, 1973) of the method ofBitter & Muir (1962) with glucuronolactone asstandard. Hexosamine was determined by the pro-cedure of Kraan & Muir (1957) with galactosaminehydrochloride as standard. Samples were hydrolysedin 8m-HCl for 3h at 95°C (Swann & Balazs, 1966) inglass tubes sealed under N2. Excess of acid wasremoved by rotary evaporation at 40°C and thehydrolysate was adjusted to a known volume withwater. Glucosamine/galactosamine molar ratioswere determined in the same hydrolysate by usingthe short column of a Locarte automated aminoacid analyser. Sulphate was determined after hydro-lysis of the sample in 60% (v/v) formic acid for 8hat 100°C. Acid was removed by rotary evaporationand sulphate determined by the method of Dodgson(1961) after the addition of a known volume of water.Protein was measured by an automated modification(Heinegard, 1973) ofthe method ofLowry etal. (1951)with bovine serum albumin (fraction V) as standardand values were generally about 20-30% higher thanthe protein estimated by summation of amino acidanalyses. Hydroxyproline was measured by themethod of Woessner (1961) after hydrolysis in6M-HCI at 1050C for 24h in screw-top Teflon-linedglass tubes (Kimax, A. R. Horwell Ltd., LondonN.W.6, U.K.) Excess of acid was removed by rotaryevaporation and the sample was dissolved in a knownvolume of water for analysis. Amino acid analyseswere performed on a Locarte amino acid analyserby using a three-buffer single-column elution system.Samples containing 200-300,ug of protein in 2ml of6M-HCI were hydrolysed at 105°C for 24h in glasstubes sealed under N2. Excess ofacid was removed byrotary evaporation at 400C and the residue was dis-solved in 1.Oml of water and a sample (0.5ml) wasapplied to the column of the analyser. No corrections

were made for the destruction of amino acids duringhydrolysis.

Preparative methods

Extraction of proteoglycans with 4M-guanidiniumchloride. Pig laryngeal cartilage was obtained freshfrom the slaughterhouse, dissected free of adheringtissue and perichondrium and finely shaved with aStanley Surform (woodwork tool). The cartilage wassuspended in 10 times its weight of cold 4M-guanidinium chloride buffered at pH5.8 or at pH4.5with 0.05M-sodium acetate and was agitated for24h at 4°C. The extract was filtered through twolayers of lint on a Buchner funnel and the cartilagewashed with a small volume of the buffered4M-guanidinium chloride. The combined extractand washings were dialysed against 7vol. of 0.05M-sodium acetate, pH5.8. The proteoglycans were thenpurified by equilibrium density-gradient centri-fugation as described below.

In separate experiments to examine the effect ofpH, proteoglycans were extracted from cartilageby the same procedure with the following changes.The 4M-guanidinium chloride was buffered topH 3.0, 4.5, 5.8 and 7.0 with sodium citrate-phosphatebuffer (Mcllvaine, 1921) used at 25% of therecommended strength and to pH8.5 with 0.05M-Tris-HCI. After extraction the combined extractsand washings were adjusted to pH5.8 with acetic acidor with 4M-NaOH as necessary and then dialysedagainst 7vol. of 0.05M-sodium acetate, pH5.8.The proteoglycans were purified as described below.

Sequential extraction ofproteoglycans with 0.15M-NaCI followed by 4M-guanidinium chloride. Slicedcartilage prepared as described above was suspendedin 10 times its weight of0.15M-NaCl and agitated for3h at 4°C. Although unbuffered the pH remainedat about pH7. The extract was filtered on a coarseglass sinter (grade 1) and the proteoglycans wereprecipitated by the dropwise addition of 10% (w/v)cetylpyridinium chloride. The precipitate was centri-fuged and washed twice with 0.05% (w/v) cetyl-pyridinium chloride containing 5mM-Na2SO4 andonce with 0.05% (w/v) cetylpyridinium chloridealone. The precipitate was dissolved in a minimumvolume of propan-1-ol and re-precipitated as thesodium salt after the addition of a few drops ofsaturated sodium acetate and 5vol. of ethanol. Theprecipitate was left overnight at 4°C and thencentrifuged, washed with 80% (v/v) ethanol, then withethanol and dried in vacuo.The cartilage residue from the extraction with

0.15M-NaCl was then suspended in 10 times itsweight of 4M-guanidinium chloride-0.05M-sodiumacetate, pH4.5, and extracted for 24h at 4°C. Theproteoglycans were isolated as described above for

1974

566

HYALURONIC ACID AND PROTEOGLYCAN AGGREGATION

the single-step extraction with 4M-guanidiniumchloride.

Purification ofproteoglycans by equilibrium density-gradient centrifugation under 'associative' conditions.The dialysed 4M-guanidinium chloride extractcontaining about 5mg of proteoglycan/ml wasadjusted to a density of 1.60g/ml by the additionof solid CsCl. Equilibrium density-gradient centri-fugation was performed in an MSE 65 centrifugein an 8x25ml angle head at 95000ga,. for 48h at20°C. The tubes were then quickly frozen in asolid-CO2-acetone bath and cut into two fractions.The bottom fraction (5ml), containing a thickproteoglycan gel, accounted for 91-96% of the totaluronic acid. It was separated from the upper fraction(13ml), which contained collagen and other protein-rich components. The proteoglycan gel was dis-persed by using a Pasteur pipette and was theneither used directly for dissociative density-gradientcentrifugation or dialysed to remove CsCl and usedas the starting material for other experiments.

Fractionation of proteoglycans by equilibriumdensity-gradient centrifugation under 'dissociative'conditions. (1) To the bottom fraction of purifiedproteoglycans, prepared as described above, wasadded an equal volume of 7.5M-guanidiniumchloride-0.05M-sodium acetate, pH5.8, and thedensity adjusted to 1.50g/ml by the addition of solidCsCl. Sufficient 4M-guanidinium chloride-CsClsolution of density 1 .50g/ml was then added to dilutethe proteoglycan concentration to 4-6mg/ml. Thesolution was then centrifuged under the same con-ditions as described above. After centrifugation thetubes were quickly frozen in a solid C02-acetonebath and cut into three fractions, i.e. bottom, 4ml,middle, 10ml, and top, 4ml. Samples of the fractionswere either used directly for experiments or thefractions were dialysed against 0.5M-sodium acetate,pH6.8, for subsequent analysis.

(2) The proteoglycans extracted with 0.15M-NaClweredissolved overnight in 4M-guanidiniumchloride-0.05M-sodium acetate, pH5.8, adjusted to a densityof 1.50g/ml with solid CsCl and then centrifugedand fractionated as described above.When isolating the three fractions from the

dissociative gradient, it was found that by freezingand cutting the tubes there was less cross-con-tamination of the fractions than when the tubeswere pierced and drained from the bottom, or whena dense inert fluorocarbon liquid was pumped intothe bottom of the pierced tube and fractions werecollected from the top. It was impractical to cut thefrozen tubes into a large number of fractions,however, so when the distribution ofmaterial throughthe gradient was examined in more detail, the tubeswere pierced at the bottom and the liquid was slowlydrained. Twelve fractions (each 1.5ml) were collected.

Recombination of fractions separated by 'dis-Vol. 139

sociative' density-gradient centrifugation. Samplesof undialysed fractions were mixed in the samerelative proportions in which they occurred in thegradient and were then dialysed overnight against alarge volume of 0.5M-sodium acetate, pH6.8.Alternatively, before mixing, each fraction wasdialysed separately against 4M-guanidinium chlorideto remove CsCl and then mixed with the othercomponent(s) and dialysed against 0.5M-sodiumacetate, pH6.8. Both methods gave the same results.The interacting component was assayed as

follows. Samples to be tested were adjusted to 4M-guanidinium chloride by adding concentrated guani-dinium chloride solution containing 0.05M-sodiumacetate, pH5.8. They were then mixed with an excessof disaggregated proteoglycan in 4M-guanidiniumchloride-0.05M-sodium acetate, pH 5.8, and dialysedas described above. The samples were then examinedby gel chromatography on Sepharose 2B. Sincedisaggregated proteoglycan contained very littlematerial that was eluted in the region of the voidvolume of the column, the increase in material elutedin this region was taken as a measure of the extent ofinteraction. This was estimated by cutting andweighing a tracing ofthe uronic acid elution profile.

Preparation of glycosaminoglycans. Samples offreshly sliced cartilage or the washed cartilage residueafter 4M-guanidinium chloride extraction weresuspended in 3-5 times their weight of 0.1M-sodiumacetate, pH5.5, containing 0.05M-EDTA (disodiumsalt) and 0.01M-L-cysteine. Activated (Kimmel &Smith, 1954) crude papain (20mg/g wet wt. ofcartilage) or crystalline papain (0.Smg/g wet wt. ofcartilage) was added to the suspension, which wasincubated at 60°C for 4h. The insoluble residue wasdiscarded. After dilution of the digest with an equalvolume of water, the glycosaminoglycans wereprecipitated by the dropwise addition of 10% (w/v)cetylpyridinium chloride. The precipitate was isolatedand converted into the sodium salt, washed and driedas described for the proteoglycan extracted with0.15M-NaCI.

Chromatographic techniques

ECTEOLA-cellulose ion-exchange chromatography.ECTEOLA-cellulose (Serva, Heidelberg, Germany)was washed with 4M-HCI followed by water until thewashings reached pH5. Fines and dissolved airwere then removed and the cellulose was packed in acolumn (16.Ocmxl.1cm). Samples of the middlefraction from 'dissociative' density-gradient centri-fugation containing 1.0-10.Omg of uronic acid weredialysed exhaustively against water and then conc.HCI was added to a final concentration of 0.02M.Samples were applied to the column, washed in with20ml of 0.02M-HCI and then eluted with 20ml eachof 0,5M-NaCl 2.5M-NaCl and 4M-HCI The acidiQ

567


fraction was neutralized with NaOH as soon as itemerged from the column. The last two fractionswere dialysed against a large excess of 0.5M-sodiumacetate, pH6.8. After concentration by pressuredialysis by using an Amicon Diaflo apparatus witha UM-10 membrane the uronic acid and hexosaminecontents and galactosamine/glucosamine molar ratiosofeach fraction were determined, and the interactionwith proteoglycan was tested as described above.Fractions containing more than 1mg of the inter-acting component were concentrated and the materialwas then precipitated at 4°C by the addition of Svol.of ethanol. The precipitates were centrifuged, washedwith ethanol and dried in vacuo.

Gel chromatography. (1) Samples of proteoglycancontaining about lmg of uronic acid/ml in 0.5M-sodium acetate, pH6.8, were applied to a column(165cmxl.1cm) of Sepharose 2B (Pharmacia,Uppsala, Sweden) which was eluted upwards with0.5M-sodium acetate, pH6.8, at 4°C at the rate of6ml/h by using a peristaltic pump. Fractions(about 2.5ml; 40 drops) were collected and theiruronic acid and/or protein contents determined.Proteoglycan aggregates and glucuronolactone wereused as markers of the void volume and total columnvolume respectively.

(2) A sample of the middle fraction from thedissociative density gradient was chromatographedon a column (9OcmxO.9cm) of Sepharose 2B,eluted with 4M-guanidinium chloride, pH5.8. Ninefractions of equal volume were collected from theregion of the void volume (Vo) of the column to thetotal volume (Vi) of the column. The ability ofthe fractions to interact with proteoglycans wastested by gel chromatography as described above,and their uronic acid contents were determinedafter dialysis against several volumes of 0.5M-sodiumacetate, pH6.8.

(3) Samples of the purified interacting componentcontaining about 100lug of uronic acid were appliedto a column (120cmxO.9cm) of Sephadex G-200eluted with 0.2M-sodium acetate, pH6.8, by using ahydrostatic head of 60cm. Fractions (2ml) werecollected and pooled into nine equal fractions fromthe void volume to the total volume of the column.

The ability of these fractions to interact with proteo-glycan was tested as described above.

Experiments on the interacting component

Hyaluronidase digestion. A sample of the purifiedinteracting component was incubated overnight at37°C in 0.1M-sodium acetate buffer, pH5.3, con-taining lOO,ug of testicular hyaluronidase/ml. Boiledenzyme was added in place of active enzyme in acontrol incubation. The ability of the digested andcontrol samples to interact with disaggregatedproteoglycans was tested as described above.Papain digestion. A sample of the purified fraction

was incubated with activated crystalline papain asdescribed for the digestion of cartilage. At the end ofthe incubation it was placed in a boiling-water bathfor 5min and a drop of saturated CaC12 added. Itwas then cooled and chromatographed directly ona column of Sephadex G-200 as described above.

Alkali treatment. To a sample of the purifiedfraction in 0.5M-NaCl was added an equal volume of1.OM-NaOH. The solution was gassed with N2 for5min, stoppered and left at 4°C overnight. The solu-tion was then neutralized with the calculated volumeof acetic acid and chromatographed on a columnof Sephadex G-200 as described above.

Infrared spectroscopy. A portion (1.0mg) of thepurified interacting component from ECTEOLA-cellulose ion-exchange chromatography and 100mgof KCI were dissolved in 2ml of water and the solu-tion was freeze-dried. The resulting powder wascompressed into a disc and the i.r. spectrum recordedon a Unicam SP.200 spectrophotometer. Spectrawere also obtained from samples of hyaluronicacid from umbilical cord and chondroitin 4-sulphatefrom pig laryngeal cartilage by the same procedure.

Results and Discussion

Identification of the interacting component

Proteoglycans extracted from pig laryngeal cartil-age with 4M-guanidinium chloride, pH5.8, werepurified by equilibrium density-gradient centri-fugation under 'associative' conditions. They were

Table 1. Analysis offractions obtained by equilibrium density-gradient centrifugation of proteoglycans, under dissociativeconditions

For conditions see the text.

Fraction Density (g/ml)Top <1.43Middle 1.43-1.57Bottom >1.57

Volume (ml)4104

Contents (Y. of total)

Uronic acid Protein (Folin)2.3 31.85.0 7.2

92.7 61.0

Galactosamine/glucosaminemolar ratio

2.24.312.6

1974

568


then separated into three fractions by a seconddensity gradient under 'dissociative' conditions inthe presence of 4M-guanidinium chloride at pH5.8.The fraction ofhighest density (>1.57g/ml) containedmost of the proteoglycan, and a fraction rich in pro-tein separated at the top of the gradient (<1.43 g/ml).A fraction that accounted for only 5% of the totaluronic acid and 7.2% of the total protein wasobtained from the middle of the gradient (1.43-1.57g/ml). Analyses of the three fractions are givenin Table 1.The effect of combining samples from the top and

the middle of the gradient on the hydrodynamic sizeof the proteoglycan from the bottom of thegradient was assessed by gel chromatography onSepharose 2B (Figs. la-Id). The samples were mixed

0._-

004'-40

10

*_-

*R

.1

in the same relative proportions in which theyoccurred in the gradient.The fraction from the middle of the gradient

increased the hydrodynamic size of about 50% of theproteoglycans (Fig. lb), whereas the protein-richfraction from the top of the gradient changed theelution of the proteoglycan very little, even thoughprotein became bound to the proteoglycan and waseluted with it (Fig. lc) (Tsiganos et al., 1972).Combining all three density-gradient fractions didnot enhance the effect (Fig. ld), implying that theactive component was present exclusively in themiddle fraction. There was a linear relationshipbetween the amount of the middle fraction added tothe proteoglycan and the percentage of the totaluronic acid excluded by gel chromatography onSepharose 2B (Fig. 2), up to a proportion of middlefraction to proteoglycan of 0.75:1 (w/v) (1:1 corre-sponded to the proportion in which they occurred inthe gradient). When this proportion was increased nomore proteoglycan was excluded from the gel,and the amount remained constant at 52 %, suggestingthat 48% was unable to interact. The middle fractionthus contained more activity than was required toproduce the maximum effect on the proteoglycan,although subsequent experiments showed that withoptimal conditions of preparation a larger propor-tion of proteoglycans was able to interact. Theresults showed that the interacting componentcould be assayed by mixing with disaggregatedproteoglycans from the bottom fraction of the density

60r

0 30 40 50 60 70t tVo Vt

Fraction no.

Fig. 1. Gel chromatography on Sepharose 2B of proteo-glycanfractions after disaggregation

Proteoglycans were fractionated in a dissociative densitygradient as shown in Table 1. Fractions were recombinedin the same proportions in which they occurred in thegradient, dialysed against 0.5M-sodium acetate, pH6.8,and samples applied to a column (165cmx 1.1cm) ofSepharose 2B as described in the text. Fractions wereanalysed for uronic acid (-) and protein (----).(a) Disaggregated proteoglycan (bottom fraction);(b) disaggregated proteoglycan+middle fraction; (c) dis-aggregated proteoglycan+top fraction; (d) disaggregatedproteoglycan+middle and top fractions.

Vol. 139

Rl, *a 50a

40-

- 30

cu8 200 0

I0 1

0 1.0 2.0

Ratio of middle fraction/proteoglycan

Fig. 2. Interaction between disaggregated proteoglycansand the middle fraction from the dissociative gradient

Various proportions of the middle fraction were mixedwith a constant amount of disaggregated proteoglycan(bottom fraction). Samples were chromatographed on acolumn (165cmx 1.1cm) of Sepharose 2B and theproportion of uronic acid excluded from the gel wasdetermined as described in the text. A ratio of middle frac-tion to proteoglycan of 1.0 corresponded to theproportion in which they occurred in the gradient.

569


gradient and measuring the change in elution profileby gel chromatography. When the relative pro-portion of interacting component to proteoglycanwas below that required to give the maximum effectthe method could be used to determine quantitativelythe amount of activity.From its analysis (Table 1) and behaviour on gel

chromatography (Tsiganos et al., 1971) the middledensity-gradient fraction contained proteoglycansof higher protein content than those in the bottomfraction. Attempts were thus made to subfractionatethis material to discern whether ability to interactwith proteoglycans was a property of the materialas a whole or of a minor constituent. It seemedlikely that if it were a minor constituent, it would bebound to the proteoglycans at low ionic strengthand dissociated at high ionic strength. An attemptwas therefore made to separate the active componentfrom proteoglycans (containing uronic acid) by gelchromatography on Sepharose 2B in the presence of4M-guanidinium chloride, pH5.8, but the proteo-glycans were eluted as a broad retarded peak, whichlargely coincided with the activity (Fig. 3). It wasevident, however, that the activity was not propor-tional to the uronic acid content ofthe fractions. Thusalthough the fractionation failed to separate activityfrom proteoglycan it suggested that the former wasnot a property of the latter.

Ion-exchange chromatography on ECTEOLA-cellulose, however, successfully separated the activityfrom the proteoglycan, since only 10% of theuronic acid was eluted in the 0.5M-NaCl fraction,which contained 73.2% of the activity. Analysis ofthis fraction showed that it contained equimolaramounts of uronic acid and hexosamine and had aglucosamine/galactosamine molar ratio of 25:1(Table 2). The fractions eluted later accounted for83% ofthe total uronic acid and from the compositionappeared to contain proteoglycans with relativelyhigh protein and glucosamine contents (Table 2)

compared with proteoglycans in the bottom fractionof the density gradient (Table 1).The carbohydrate composition of the active

fraction isolated by ECTEOLA-cellulose ion-exchange chromatography suggested that it was aglycosaminoglycan, which, since it was mostlyeluted in the void volume on Sephadex G-200, waslarge. Moreover, the position of elution was un-altered after treatment with papain or exposure toalkaline conditions that induce fl-elimination of

0.5

o 0.4

,4 0.3

i 0.2

oR 0.1.o o!; I

0.3~ ~~~~~~~~.

cs o

a %-

30UD)

Fraction no.

Fig. 3. Gel chromatography on Sepharose 2B in 4M-guanidinium chloride of the middle fraction from the dis-sociative gradient

A sample of the middle fraction containing about 1mg ofuronic acid was applied to a column (9OcmxO.9cm) ofSepharose 2B eluted with 4M-guanidinium chloride-0.05M-sodium acetate, pH5.8. Nine fractions of equalvolume from the region of the void volume (VO) to thetotal volume (Vi) were collected. One-tenth of eachfraction was mixed with disaggregated proteoglycan(containing 1mg of uronic acid) and the interactionmeasured by gel chromatography as described in the text(m). After dialysis against 0.5M-sodium acetate, pH6.8,the uronic acid content of each of the nine fractions wasdetermined (0).

Table 2. Analysis offractions obtained by ECTEOLA-cellulose ion-exchange chromatography of the materialfrom the middle(density 1.43-1.56g/ml) ofthe dissociative density gradient (as shown in Table 1)

A sample containing about 1mg ofuronic acid was applied to a column (16.0cmx 1.1 cm) ofECITEOLA-cellulose and elutedas described in the text. Acid fractions were neutralized immediately they were eluted from the column and pooled fractionsfrom each eluate were concentrated by pressure dialysis (Diaflo) before analysis.

Molar ratios

Eluent Uronic acid Protein (Folin)fraction content (jug) content (jig)

0.02M-HCI 40 1400.5M-NaCl 78 502.5M-NaCl 244 67014M-HCI 467 2120 1* Analysis on comparable fraction from larger sample.

Sulphateuronic acid

0.1*1.06

Glucosaminegalactosamine

25.00.16

Interaction withproteoglycan (% ofrecovered activity)

1.073.210.016.8

1974

570


polysaccharide chains linked to hydroxyamino acids.Thus the activity ofthe component was not dependenton an integral protein moiety. On the other hand, theability of the component to interact with theproteoglycans was completely abolished by digestionwith testicular hyaluronidase. These propertiestogether with the analysis suggested that the activecomponent was hyaluronic acid. This was confirmedby the i.r.-absorption spectrum, which showed noabsorption at 1240cm-1 characteristic of sulphategroups and which resembled the spectrum ofauthentic hyaluronic acid (Fig. 4).A sample of the interacting component containing

0.5mg of uronic acid was fractionated by gelchromatography on a column of Sepharose 2B

(Fig. Sa) and showed a broad distribution of sizecomparable with that observed in the presence of4 M-guanidinium chloride (Fig. 3). Theeluted materialwas arbitrarily divided into three fractions and eachtested separately with proteoglycan (Fig. Sb). Theactivity of each fraction was directly proportionalto the amount of uronic acid in it and not to theposition of elution, which further suggested that theactive component was hyaluronic acid, and not someother compound. Moreover, hyaluronic acid fromother sources, such as umbilical cord and cock'scomb, also interacted in a similar way with cartilageproteoglycans (Hardingham & Muir, 1972b).The results showed that during the fractionation

of proteoglycan aggregates in a dissociative density

(C)

1800 1600 1400 1200 1000 800 650Wavenumber (cm-')

6 7 8 9 10 11 12 13 14 15

Wavelength (nm)

Fig. 4. Infrared absorption spectraI.r. spectra of(a) ECIEOLA-cellulose 0.5M-NaCl fraction; (b) hyaluronic acid fromhuman umbilical cord (BDH ChemicalsLtd.); (c) chondroitin 4-sulphate from pig laryngeal cartilage. For experimental details see the text.

Vol. 139

571


60O

40 (a)

II-4 20 - 1 2 8, 3I.)J0I 1

*g 20 30 40 50 60 70$ t t

~D vo t

Fraction no.

m a,0 m

o .0-4 X

0 2 4 6Interacting component (,pg)

8

Fig. 5. (a) Fractionation by gel chromatography on

Sepharose 2B of the isolated interacting component and(b) interaction of the subfractions with disaggregated

proteoglycan

(a) A sample of the ECTEOLA-cellulose 0.5M-NaClfraction containing 1 mg of uronic acid was applied to a

column (165cmx 1.1cm) of Sepharose 2B eluted as

described in the text. The fractions were analysed foruronic acid and pooled as shown to give three fractionsof decreasing size, 1, 2 and 3 respectively. (b) Theability of fractions 1 (@), 2 (m) and 3 (A) to interact withdisaggregated proteoglycan (containing 1mg of uronicacid) was measured by gel chromatography as describedin the text.

gradient, a small amount of hyaluronic acidseparated from the proteoglycans. Its detaileddistribution in the gradient is shown in Fig. 6(a).The gradient was divided into twelve equal fractionsand the ability of each to interact with disaggregatedproteoglycan was assessed by gel chromatography.There was a fairly broad band of activity in themiddle of the gradient. The buoyant density of 90%of the material was between 1.43 and 1.56g/ml. Thediffuseness of the band may have resulted fromfailure to achieve true equilibrium in the gradient,but may also have been contributed to by thepolydispersity of the hyaluronic acid. Purifiedhyaluronic acid from umbilical cord also distributedin the gradient in a similar way (Fig. 6b), althoughits peak was at a slightly lower buoyant density,i.e. 1.48g/ml compared with 1.51 g/ml.

The interaction of hyaluronic acid with proteo-glycans, as assessed by gel chromatography, provideda sensitive method for the assay of hyaluronic acidin glycosaminoglycan fractions from cartilage(T. E. Hardingham & H. Muir, unpublished work).Preliminary results showed that pig laryngeal cartil-age contained 0.35mg of hyaluronic acid/g wet wt.(accounting for 0.7% of the total uronic acid), andthat most of this (about 95%.) was associated withpurified aggregated proteoglycan. That hyaluronicacid is essential for the formation of aggregates canbe deduced from the results of Gregory (1973) asdiscussed below.

Effect of pH on the extraction of proteoglycanaggregates

Sajdera & Hascall (1969) showed that extraction ofcartilage with 4M-guanidinium chloride dissociatedproteoglycan aggregates and the proportion thatre-formed when the extract was dialysed to low ionicstrength depended on the pH of extraction andwas maximal at pH5.8. Since hyaluronic acid isessential for the formation ofaggregates, the observedeffect of pH may have been on the amount of hyal-uronic acid extracted. The proportion of aggregatesin proteoglycans extracted in 4M-guanidiniumchloride between pH3.0 and pH8.5 was thereforeassessed and the relative amount of hyaluronic acidin each extract measured. After extraction and beforedialysis to 0.5M-guanidinium chloride, the extractswere adjusted to pH5.8, so that in each extractaggregates were re-formed under similar conditions.The proteoglycans were then purified in an associativedensity gradient and the proportion of aggregatesin each preparation was estimated by gel chromato-graphy on Sepharose 2B. The proportion varied withpH as shown in Fig. 7; the proteoglycans extractedat pH4.5 contained the largest proportion ofaggregates, amounting to about 75% of the totaluronic acid.The proteoglycans from each extract were then

separated into three fractions in a dissociativegradient. The hyaluronic acid content of the middlefractions from the gradient was determined bymeasuring interaction with a 1.5-fold excess of thebottom fraction of disaggregated proteoglycansextracted at pH5.8. The extent of interaction wassimilar for all the fractions (Fig. 8), showing that theamount of hyaluronic acid extracted varied littlewith pH. In contrast, the proteoglycans extracted atdifferent pH values differed widely in their abilityto interact with an excess of the hyaluronic acidfraction from the pH5.8 extract (Fig. 8). Theproteoglycans extracted at pH4.5 contained thelargest proportion able to interact. The proportionof aggregates in extracts prepared at differentpH values was thus related to the ability of the

1974

572


0

C-

o

-4)1 c is

15

10

5

0L

4)

0C

00t.4

7

IIIINC3

. 0.-4

u. .,a

cl.tiie

0

aua)

' I' I 1 41 2 3 4 5 6 7 8 9 10 11 12 13

Fraction no.

Fig. 6. Equilibrium density-gradient centrifugation under dissociative conditions of(a)purifiedproteoglycans and (b) hyaluronicacid

Purified proteoglycans from an associative gradient were fractionated in a dissociative CsCl gradient (starting density1.50g/ml) containing 4M-guanidinium chloride-0.05M-sodium acetate, pH5.8, at 95000ga,. for 48h at 20°C as described inthe text. Twelve fractions (each 1 .5ml) were collected by piercing and draining the tubes. A portion (1OO,ul) of each fractionwas mixed with disaggregated proteoglycan and the interaction (o) measured by gel chromatography as described in thetext. The uronic acid (0) and protein (Folin method) contents (u) of each gradient fraction were determined after dialysisagainst O.5M-sodium acetate. Density is shown by the dashed line.

proteoglycan to interact with hyaluronic acid andnot to the amount ofhyaluronic acid in the extract.The proportion of aggregates obtained was

greatest at pH4.5, which differs from the optimumpH of 5.8 reported by Hascall & Sajdera (1969).The present experiments, however, differ in severaldetails from those of Hascall & Sajdera (1969),the most important being that in the present studythe 4M-guanidinium chloride extracts were alladjusted to pH5.8 before dialysis and hence theability of all the extracts to form aggregates wastested under identical conditions. In the experimentsof Hascall & Sajdera (1969), on the other hand, theproportions of aggregates in each extract weredetermined at pH5.8, but only after dialysis at thepH ofextraction. Since otherexperiments showed thataggregates became increasingly dissociated as the pHfell below 5.8 the formation of aggregates may wellhave been decreased in those extracts dialysed belowpH5.8. That this may be the explanation is shown bya further experiment of Hascall & Sajdera (1969), inwhich the proportion of aggregates in the extract

Vol. 139

prepared at pH5.8 was much decreased when the pHwas adjusted to pH3.8 before dialysis. Theresults of Hascall & Sajdera (1969) thus reflect acombination of two effects, (a) the effect of pH onextraction and (b) the effect of pH on reaggregation.The experiments described here were designed toexamine only the effect of pH on extraction. Thepresent and previous results were thus comparableat pH5.8 and above when reaggregation wasfavoured, but different at lower pH values, when itwas inhibited. However, even at low pH the tworesults could be compatible if the stabilizing influenceof the protein-rich fraction (see 'General discussion'),but not the proteoglycan-hyaluronic acid interaction,was adversely affected by low pH, because gelchromatography cannot distinguish between aggre-gates and the complex resulting from the interactionof proteoglycans with hyaluronic acid.As with the present results, Hascall & Sajdera

(1969) showed that above pH5.8 there was aprogressive decrease in the proportion of aggregatesin the extract. Extraction at high pH values appeared

573


80mCl404)0

4)

o00

o4'

CU

la= 0

C.)*g:

70 F60 F

50 -

40 H

30 F

20 F

I0

2 3 4 5 6pH of extraction

7 8 9

Fig. 8. Interaction measured by gel chromatographybetween dissociative density-gradient fractions of proteo-

glycans extracted at differentpH values

Purified proteoglycans extracted at different pH valueswere fractionated in a dissociative density gradient as

/, , , , - _, shown in Table 1. Samples of the middle fractions were30 40 50 60 70 each mixed with a standard preparation of disaggregated

t t proteoglycan (extracted at pH5.8) present in 1.5-foldvo vt excess (relative to the original proportions in the

gradient), and the interaction was measured by gelFraction no. chromatography as described in the text (W). Samples of

Ge1 chromatography on Sepharose 2B of proteo- the bottom fractions were mixed with a standard pre-extracted in 4M-guanidinium chloride, pH 3.0-8.5 paration of middle fraction (from a pH5.8 extract)

present in twofold excess (relative to the originalycans extracted in 4M-guanidinium chloride proportions in the gradient) and the interaction wasat (a) pH3.0, (b) pH4.5, (c) pH5.8, (d) pH7.0, similarly measured by gel chromatography (M). Thevere purified in an associative gradient at pH 5.8 proportion of aggregates in each preparation beforeibed in the text. Samples of the purified proteo- disaggregation (Fig. 8) is shown for comparison (A).

glycans containing 1 mg of uronic acid were applied to acolumn (165cm x 1.1 cm) of Sepharose 2B eluted asdescribed in the text. The fractions were analysed foruronic acid. Vo and V4 mark the void volume and totalvolume of the column respectively.

to affect irreversibly the ability ofsome proteoglycansto interact with hyaluronic acid. If the factors thatbring about this 'denaturation' of proteoglycan canbe identified more ideal conditions of extractionmay be devised. Under the conditions used in thepresent study, however, the effects were minimalat pH4.5 and this pH was therefore used for allsubsequent extractions of proteoglycans in 4m-guanidinium chloride.When hyaluronic acid was present in excess, the

proportion of disaggregated proteoglycans thatinteracted with it never approached the proportionthat was excluded from Sepharose 2B in the originalaggregated preparation, although the difference was

not always as large as that found in the pHexperiments (Fig. 8). This suggested that subsequenttreatment of the proteoglycan involving furtherexposure to 4M-guanidinium chloride and/or CsClduring dissociative gradient centrifugation may havedecreased their capacity to interact. Moreover, as theproteoglycans had been extracted in 4M-guanidiniumchloride and purified in the presence of concentrated(SM) CsCl before their degree of aggregation wasassessed, it could be argued that proteoglycans weremore aggregated in vivo than in the extracts. On theother hand, there is evidence thatnot all proteoglycansare aggregated in vivo. Proteoglycans of relativelysmall molecular size have been preferentiaIlyextractedfrom cartilage with iso-osmotic sodium acetate orNaCI (siganos & Muir, 1969a; Brandt & Muir,1969, 1971; Simiinek & Muir, 1972). Since theseconditions of extraction would not dissociate

1974

(a)60

40

20

(b)

40

(c)

(d)

ED 40

r 20U

. 0

40

(e)

20 -

0*

40 -

20 -

O20

Fig. 7. (glycans d

Proteoglbuffered(e) 8.5, vas descri

574

7


aggregates it implies that these smaller proteo-glycans were present in cartilage in a non-aggregatedform and that they probably account for some of thenon-aggregated proteoglycan in the 4M-guanidiniumchloride extract. The extraction of the cartilage withiso-osmotic NaCl followed by 4M-guanidiniumchloride appeared to be a simple method forseparating aggregated and non-aggregated proteo-glycans, thus providing a further assessment of theirrelative abundance and also enabling their charac-teristics to be examined in more detail.

Preparation ofnon-aggregated and aggregatedproteo-glycans

Preliminary experiments showed that extractionwith O.15M-NaC1 at 4°C brought into solution amaximum of 12% of the total uronic acid (Fig. 9a),which was largely all in solution after 3h. Subse-quent extraction with 4M-guanidinium chloride,buffered at pH4.5, for 24h at 4°C then brought intosolution a further 71 % of the total uronic acid(Fig. 9b). The combined yield of the two extracts(83 Y.) was thus comparable with that obtained froma single extraction with 4M-guanidinium chloride,suggesting that the population of proteoglycansextracted by the sequential method was similar to thepopulation extracted by a single-step procedure.

0.3

0.2

, 0.1

r 2..0*E

1.0

t(a)

1 2 3 4(b) Time (h)

5 6

I(-

0 4 8 12 1 6 20 24 42 44Time (h)

Fig. 9. Time-course ofextraction ofproteoglycansfrom piglaryngeal cartilage

(a) Weighed samples of sliced cartilage (5.0g) preparedas described in the text were extracted with 50ml ofunbuffered 0.15M-NaCl at 4°C. Samples were withdrawnat timed intervals up to 6h and the uronic acid contentwas measured. (b) The cartilage residue from samplesextracted for 3 h with 0.15M-NaCl were suspended in 50mlof 4M-guanidinium chloride buffered at pH4.5. Sampleswere withdrawn at timed intervals and the uronic acidcontents measured.

Vol. 139

This form of sequential extraction was used toprepare two proteoglycan fractions as shown inScheme 1. Proteoglycans extracted first (fraction N)resembled those extracted from the same type ofcartilage by using 0.15M-sodium acetate with low-speed homogenization (Tsiganos & Muir, 1969b).The protein content was low and galactosamine/glucosamine molar ratio was high (24:1), indicating alowkeratan sulphatecontent (Table 3). Gel chromato-graphy on Sepharose 2B showed a broad distributionof size (Kay. 0.51; Fig. 10a) and as only 7% wasexcluded from the gel, few aggregates were present.The proteoglycans extracted subsequently with4M-guanidinium chloride (fraction Al) had a highprotein and keratan sulphate content (Table 3)and consisted of a large proportion of aggregates,since about 90% was excluded from Sepharose 2B(Fig. 10d). This extract was thus enriched inaggregates compared with the single-step 4M-guanidinium chloride extract, presumably becausemost of the smaller non-aggregated proteoglycanshad already been extracted with 0.15M-NaCl.

Further evidence for the aggregated and non-aggregated nature of the proteoglycan fractions wasobtained by equilibrium density-gradient centri-fugation under dissociative conditions. Proteo-glycans in fraction Al were disaggregated in thegradient and a protein-rich fraction D3 was separatedfrom the majority of the proteoglycans (Dl), whichwere thereby decreased in protein and keratan sul-phate content (Table 3). Gel chromatography of thedisaggregated proteoglycans showed a single retardedpeak on Sepharose 2B (Kay. 0.24; Fig. 10e), as foundwith other preparations of disaggregated proteo-glycans from non-sequential extracts (see Fig. 1).In contrast, the molecular size of the proteoglycansextracted with 0.15M-NaCl was unaffected byfractionation in a dissociative density gradient (Fig.lOb). Only a small amount of protein separated at thetop of the gradient (Fig. 11), so that the compositionofthe proteoglycans in thebottom fractionwas almostunchanged (ND1; Table 3) and the middle fraction(ND2) contained no detectable hyaluronic acid.Comparison of the non-aggregated proteoglycans

(ND1) with the disaggregated fraction (DI) showedthat the latter was larger on gel chromatography(Kay. 0.24 compared with Ka,. 0.50 on Sepharose 2B,Figs. 10b and 10e) and that it contained more proteinand keratan sulphate (Table 3). This result is notcompatible with the existence of a protein corecommon to the two groups of molecules, because,on the basis ofa simple 'bottle brush' model ofproteo-glycan structure (Mathews & Lozaityte, 1958;Partridge et al., 1961), size should increase with thenumber of chondroitin sulphate chains, whenceproteoglycans of small size would have a highprotein content. As previously proposed (Tsiganos& Muir, 1969b), the low protein content of the small

575


Freshsliced cartilage

Extracted in Q.S.-NaCI for 3 h at 4°C

Filtered ) Extract

Residue

Extracted in 4 M-guanidiniumchloride-O.05 M-sodium acetate,pH 4.5, for 24 h at 4°C.

Non-aggegated proteoglycansProteoglycans precipitated withcetylpyridinium chloride and isolated asthe sodium salt [Fcon¶jj

II

Dissociative density gradient

ND3

ND2

N| FractionND1

Aggregated proteoglycansI Proteoglycans

Filtered Extract dialysed A2 purified in anassociative

-density gradientL2|FractionfAl

Residue. Washed, digested withnanain and the aElvcosaminosclvcansprecipitated with cetylpyridinium chlorideand isolated as the sodium salt Dissociative density gradient

DIFractionfDl

Scheme 1. Sequential extraction ofnon-aggregated and aggregatedproteoglycans from pig laryngeal cartilageFractions ND2 and ND3 were not analysed in detail. The distribution of fraction N in the gradient is shown in Fig. 11.

Table 3. Composition offractions ofproteoglycans extractedfrom pig laryngeal cartilage

Fractions are as shown in Scheme 1. Protein content is obtained from summation of amino acid analyses.

Uronic acid/protein(Folin) wt. ratio

2.232.481.510.1332.240.400.063

Galactosamine/glucosaminemolar ratio

24.225.39.54.312.51.791.10

Uronic acid content(% of dry wt.)

25.226.023.88.8

24.915.25.3

Protein content(% of dry wt.)

7.216.88

11.05

8.1842.079.0

1974

FractionNNDIAlA2DID2D3

Uronic acidcontent (% oftotal in tissue)

12.211.670.25.9

67.32.20.7

576

I


60

40

20

0

'40

(a)

/

- (b)

20-

abo3la0

.4_

0

0

U

40 - (c)20

0

40 - (d) A

20 /-

0

40 (e)20 -

-_

0

40 (f)20- %

20 30 40 50 6- t

-20 30 40 50 60 70

Fraction no.

Fig. 10. Gel chromatography on Sepharose 2B ofproteo-glycans extracted sequentially (see Scheme 1)

(a) Proteoglycan extracted with 0.15M-NaCl (N); (b) frac-tion N after dissociative gradient fractionation to givefraction ND1; (c) fraction ND1 mixed with l%y (w/w)hyaluronic acid; (d) proteoglycan extracted with 4M-guanidinium chloride (Al); (e) fraction Al after dis-sociative gradient fractionation to give fraction Dl;(f) fraction Dl mixed with l%o (w/w) hyaluronic acid.Samples containing about 1 mg of uronic acid wereapplied to a column (165cmxl.1cm) of Sepharose 2Beluted as described in the text. The uronic acid (-) andprotein (Folin method) (-) contents of the fractionswere determined. VO and Vt mark the void volume andtotal volume of the column respectively.

non-aggregated proteoglycans implies that they havea smaller protein core than the disaggregatedproteoglycans.The amino acid composition of the non-aggre-

gated and aggregated proteoglycans provided furtherevidence that the protein cores were different.Although the analyses show a general similarity(Table 4), the non-aggregated proteoglycans con-sistently contained more serine and glycine and fewercysteine, tyrosine and arginine residues than did the

Vol. 139

400

.- 300

" 200 "I

100 _-0

0Do 1 2 3 45 6 7 8 9l0 12Fraction no.

I1.6 --

1.5

.4

Fig. 11. Equilibrium density-gradient centrifugation underdissociative conditions of the proteoglycan extracted with

0.15M-NaCI (N)

The purified proteoglycan in 4M-guanidinium chloride-0.05 M-sodium acetate, pH 5.8, containing CsCl at astarting density of 1.50g/ml was centrifuged at 95000g.,.for 48h at 20°C as described in the text. Twelve sampleswere collected, and the uronic acid (0) and protein (m)contents determined after dialysis against 0.5M-sodiumacetate, pH6.8. Density is shown by the dashed line.

disaggregated proteoglycans. The analyses of theother fractions produced by density-gradient frac-tionation of fraction Al are also included in Table 4for comparison. The protein-rich fraction D3 con-tained much more aspartate, cysteine and aromaticand basic amino acids than the proteoglycan Dl,and fraction D2 had an analysis intermediate betweenthat of fractions Dl and D3. Further examinationshowed that it was not a mixture of the componentsof Dl and D3, as the analysis was unaltered afterre-running in the gradient (T. E. Hardingham &H. Muir, unpublished work). Its analysis and posi-tion in the gradient suggests that this fraction con-tained some proteoglycans of high protein contentin addition to hyaluronic acid. Similar differencesin amino acid analyses were previously notedbetween density-gradient fractions of disaggregatedproteoglycans extracted in one step (Tsiganos et al.,1971). The present results showed that differences inamino acid composition were still present within theproteoglycan population even when most of the non-aggregating fraction had been removed. The signifi-cance of these differences has not yet been examinedin detail.

Chondroitin sulphate chains are attached to theprotein core of proteoglycans via alkali-labile link-ages to serine residues (Muir, 1958). The chainsattached to both small and large proteoglycansfrom pig laryngeal cartilage had been shown to beof similar size (mol.wt. about 15000) (Tsiganos &Muir, 1969b), enabling the proportion of serineresidues with chains attached to be calculated fromthe analyses of the non-aggregated and disaggregated

T

577

tVt


Table 4. Amino acidcomposition ofproteoglycanfractionsfrompig laryngeal cartilage

Fractions are as shown in Scheme 1. Each value given is the average of two separate analyses. No corrections were appliedfor losses during hydrolysis.

Composition (residues/1000 residues)

FractionAspThrSerGluProGlyAlaCysValMetlIleLeuTyrPheLysHisArg

... ND168.356.2150.0145.879.8

142.971.9Trace61.83.9

38.880.713.025.310.28.0

29.1

Al82.959.7

103.8127.592.2

122.080.17.5

58.74.4

36.382.127.229.921.513.750.1

DI76.362.3

122.7140.098.5

129.579.42.4

58.93.4

36.576.719.828.011.59.8

44.1

D293.362.673.3

121.283.294.581.813.356.08.3

34.390.735.439.633.819.159.7

D3125.845.756.595.168.5

102.176.224.453.44.9

31.088.342.642.650.023.964.5

fractions. Although non-aggregated proteoglycanscontained less protein they contained a higherproportion ofserine residues, such that the proportionlinked to chondroitin sulphate chains was very similarin both, i.e. 52% in the non-aggregated and 49% inthe disaggregated proteoglycans. These are, however,average values and some range of variation is likelyin each fraction. A similar calculation applied tofraction D2, which contained proteoglycans of highprotein content, showed that only 15% of serineresidues appeared to have chondroitin sulphatechains attached.

Interaction ofproteoglycan fractions with hyaluronicacid

The behaviour on gel chromatography on Sepha-rose 2B of the non-aggregated and disaggregatedproteoglycans when mixed with excess of hyaluronicacid is shown in Figs. 10(c) and 10(f). Disaggregatedproteoglycans interacted with hyaluronic acid sothat 74% of the uronic acid became excluded fromthe gel, whereas less than 5% of the non-aggregatedproteoglycan interacted either before or afterdissociative density-gradient centrifugation, nor didproteoglycans extracted with 0.15M-NaCl interactwith hyaluronic acid before cetylpyridinium chlorideprecipitation. Hence neither exposure to 4M-guani-dinium chloride nor cetylpyridinium chloride pre-cipitation was responsible for failure of these proteo-glycans to interact.

The fact that proteoglycans extracted with0.15M-NaCl were not aggregated and unable tointeract with hyaluronic acid would explain whythey are present in cartilage in a non-aggregatedform and, together with their small size, why theywere extracted with iso-osmotic NaCl. Presumablythey are retained within the cartilage in vivo onlybecause of the presence of a membrane, theperichondrium, which limits diffusion out of thetissue.

General discussion

It was previously shown that hyaluronic acidinteracted specifically with disaggregated cartilageproteoglycans (Hardingham & Muir, 1972b). Thecharacteristics of this interaction suggested that alarge number of proteoglycans were bound to eachhyaluronic acid chain, and that each proteoglycancontained only a single binding site. The inhibitionof proteoglycan-hyaluronic acid interaction byoligosaccharides derivedfromhyaluronicacidshowedthe binding site to have a high affinity for a unit assmall as a decasaccharide (Hardingham & Muir,1973b). The properties of the proteoglycan-hyaluronic acid system were remarkably similar tothose of the reversible aggregation of proteoglycanreported by Hascall & Sajdera (1969), the majordifference being that the proteoglycan-hyaluronicacid complex did not have a high coefficient ofsedimentation in the ultracentrifuge.

1974

578

HYALURON1C ACID AND PROTEOGTLYCAN AGGREGATION

Gregory (1973), extending the work of Hascall &Sajdera (1969), presented evidence that two differentcomponents were required for the formation ofproteoglycan aggregates; one was a fraction richin protein of low buoyant density (about 1.25g/ml)equivalent to fraction D3 in the present work; theother was a fraction with a buoyant density similarto that of hyaluronic acid. These results, togetherwith the present identification of hyaluronic acid inproteoglycan aggregates, show that the interactionbetween proteoglycan and hyaluronic acid isessential for the formation of aggregates, which inaddition requires the protein-rich fraction D3 tostabilize the complex in some way that alters itsbehaviour in the ultracentrifuge. Confirmation of therole ofhyaluronic acid and the protein-rich fraction inaggregation has also been established in proteo-glycans from bovine nasal septa and bovine trachealcartilage (V. C. Hascall & D. Heinegard, personalcommunication).The presence of hyaluronic acid in the middle

density-gradient fraction D2 explains why whenreturned to normal associative conditions for gelchromatography, the proteoglycans in this fractionwere largely excluded from Sepharose 2B (Tsiganoset al., 1971), whereas the disaggregated proteoglycansat the bottom of the gradient were not. A similarresult was reported in the fractionation of proteo-glycans from bovine trachea (HeinegArd, 1972a).In the light of this explanation it appears unnecessaryto consider that these proteoglycans are basicallydifferent from those in the bottom of the gradient.The glucosamine content of purified proteoglycans

has been shown to be attributable mainly to keratansulphate attached to the protein core (Tsiganos &Muir, 1967; Heinegard & Gardell, 1967), but thepresent results show that some of the glucosamine inpurified aggregates is present in hyaluronic acid.However, it accounted for only a small proportion ofthe total glucosamine (about 7.0%Y.) and if presentin similar amounts in other cartilage would notinvalidate the use ofglucosamine as a general measureof keratan sulphate content.

Various factors affect the proportion of aggregatesin proteoglycan preparations. Thus there is someevidence that the proportion decreased with ageand may vary from one type of cartilage to another(Tsiganos & Muir, 1973). Moreover, conditions ofisolation are critical, since in addition to theeffects of pH shown here, proteoglycan aggregateswere unstable to high-speed homogenization orsonication (Sajdera & Hascall, 1969). It wouldappear that the region of the molecule necessary foraggregation is sensitive to denaturation and ismaintained by disulphide bridges, since proceduresfor the reduction and alkylation of cystine residuesabolished the ability to interact with hyaluronic acid(T. E. Hardingham & H. Muir, unpublished work)

Vol. 139

and caused disaggregation (Hascall & Sajdera,1969).Small non-aggregated proteoglycans are not

peculiar to one type of cartilage. They havepreviously been characterized from pig laryngealcartilage (Tsiganos & Muir, 1969b), pig articularcartilage (Brandt & Muir, 1971; gimbinek & Muir,1972) and bovine nasal septum (Mayes et al., 1973).Brandt et al. (1973) showed that small proteo-glycans, extracted with 0.15M-sodium acetate,reacted directly with antiserum raised against wholeproteoglycans, whereas larger proteoglycans (includ-ing disaggregated fractions) required hyaluronidasedigestion before any interaction was possible. Thusalthough small and large proteoglycans shared mostantigenic determinants, these were more accessiblein the former.The inability of non-aggregated proteoglycans to

interact with hyaluronic acid suggests that they arestructurally different from those able to aggregate.Moreover, comparison of their analysis and hydro-dynamic size is incompatible with the presence of aprotein core common to both types of proteoglycan.Further, they do not appear to have resulted fromdegradation of larger proteoglycans either in vivoor during preparation. There was no detectableproteolytic activity in iso-osmotic sodium acetateextracts of cartilage (Tsiganos & Muir, 1969a) andthere was a limit to the amount of non-aggregatedproteoglycans extracted with time (Fig. 9), whichsuggested that they were not being formed by enzymeactivity during the extraction procedure. Hardingham& Muir (1972a), using pulse-chase experiments,showed that smaller proteoglycans were neitherprecursors nor degradation products of largeproteoglycans, both beingsynthesized at thesame timeand independently of each other, although possiblerelationships within the chondrocytes during theirformation have not yet been ruled out.

All preparations of cartilage proteoglycans arevery polydisperse. This is partly the result ofextensivevariation in the number and type of polysaccharidechains attached to the protein core (see Muir, 1969).Whether it also arises from the existence of differentprotein cores remains a source of controversy.There is evidence both for and against this proposaland it is apparent that the type of preparation studiedand the methods used to examine it are mostimportant (as discussed by Tsiganos & Muir,1969b; Hascall & Sajdera, 1970). Since fractionationin a 'dissociative' gradient separates non-covalentlybound protein from purified proteoglycans littlereliance can be put on deductions made with the use ofmaterial in which the possibility of similarly boundprotein has not been excluded.

It was suggested that proteoglycans extracted frombovine nasal septum in 4M-guanidinium chloride andsubfractionated in a dissociative density gradient

579

580 T. E. 4ARDlNGHAM AND H. MUIR

consisted of a single polydisperse species (Hascall &Sajdera, 1970). A similar conclusion was drawn whenchondroitinase-digested proteoglycans were frac-tionated in a similar way (Hascall & Riola, 1972).On the other hand, 0.15M-KCI extracted 15-20%of the total proteoglycans from the same type ofcartilage; these proteoglycans had a different proteincontent and amino acid analysis from the majorityextracted with 4M-guanidinium chloride (Mayeset al., 1973). This suggested that dissociative gradientcentrifugation used by Hascall & Sajdera (1970),which fractionates molecules largely according totheir relative carbohydrate/protein content, hadfailed to detect these differences. This may be becauseeach proteoglycan species varies in its relative carbo-hydrate/protein content and thus overlaps extensivelyin the gradient. Other techniques such as zone electro-phoresis (Mashburn & Hoffman, 1971; Mayes et al.,1973) also failed to detect some aspects of hetero-geneity revealed by other methods such as sequentialextraction or gel chromatography, thus emphasizingthe importance of the methods chosen to examinethis problem.The existence of proteoglycans of aggregating and

non-aggregating types in many varieties of cartilagesuggests that there are a minimum of two types ofprotein core. Further, examination of the proteincores of proteoglycans from bovine tracheal cartilagedigested with hyaluronidase and fractionated by gelchromatography revealed at least three fractions(Heinegard, 1972b). The differences in amino acidcomposition among disaggregated proteoglycanfractions reported here (i.e. Dl and D2) also suggeststhat there were more than two types of coreprotein.Although there is considerable homology in pro-

tein structure between proteoglycans prepared bysimilar methods from cartilage from differentanatomical sites and species (Mathews, 1971), thisis not incompatible with the existence of more thanone type of proteoglycan, because homology couldapply equally well to a group of polypeptides as to asingle species. The existence of distinct types ofproteoglycan suggests that age-related and patho-logical changes in proteoglycan composition may havetwo separate origins; one may be the result ofpreferential synthesis ofparticular proteoglycan cores,the amino acid sequence of which may determinethe number and type of polysaccharide chainsattached (see Baker et al., 1972); the other mayinvolve some change in the production andorganization of the enzymes and substrates necessaryfor polysaccharide chain synthesis (see Roden, 1970).Such changes in composition could lead to changes inthe quality of the proteoglycans, such as alterationsin the proportion and size of the aggregates theyform, the importance of which needs to be evaluatedin aged and diseased tissue.

We are grateful to T. Wall and Son Ltd., LondonN.W.10, for supplying fresh tissue for this study. Wethank Mr. R. J. F. Ewins and Mr. Y. Saloojee for technicalassistance and the Arthritis and Rheumatism ResearchCouncil for financial support.

References

Baker, J. R., Rod6n, L. & Stoolmiller, A. C. (1972)J. Biol. Chem. 247, 3838-3847

Bitter, T. & Muir, H. (1962) Anal. Biochem. 4, 330-334Brandt, K. D. & Muir, H. (1969) FEBSLett. 4,16-18Brandt, K. D. & Muir, H. (1971) Biochem. J. 121,261-270Brandt, K. D., Tsiganos, C. P. & Muir, H. (1973) Biochim.

Biophys. Acta 320, 453-468Dodgson, K. S. (1961) Biochem. J. 78, 312-319Gregory, J. D. (1973) Biochem. J. 133, 383-386Hardingham, T. E. & Muir, H. (1970) FEBS Lett. 9,

145-148Hardingham, T. E. & Muir, H. (1972a) Biochem. J. 126,

791-803Hardingham, T. E. & Muir, H. (1972b) Biochim. Biophys.Acta 279, 401-405

Hardingham, T. E. & Muir, H. (1973a) Biochem. Soc.Trans. 1, 282-284

Hardingham, T. E. & Muir, H. (1973b) Biochem. J. 135,905-908

Hascall, V. C. & Riola, R. L. (1972) J. Biol. Chem. 247,4529-4538

Hascall, V. C. & Sajdera, S. W. (1969) J. Biol. Chem. 244,2384-2396

Hascall, V. C. & Sajdera, S. W. (1970) J. Biol. Chem. 245,4920-4930

HeinegArd, D. (1972a) Biochim. Biophys. Acta 285,181-192

Heinegird, D. (1972b) Biochim. Biophys. Acta 285,193-207

Heinegard, D. (1973) Chim. Scripta 4, 199-201HeinegArd, D. & Gardell, S. (1967) Biochim. Biophys. Acta

148, 164-171Kimmel, J. R. & Smith, E. (1954) J. Biol. Chem. 207,

515-531Kraan, J. & Muir, H. (1957) Biochem. J. 66, 55PLowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall,R. J. (1951) J. Biol. Chem. 193, 265-275

Mashburn, T. A., Jr. & Hoffman, P. (1971) J. Biol. Chem.246, 6497-6506

Mathews, M. B. (1971) Biochem. J. 125, 37-46Mathews, M. B. & Lozaityte, I. (1958) Arch. Biochem.

Biophys. 74, 158-174Mayes, R. W., Mason, R. M. & Griffin, D. C. (1973)Biochem. J. 131, 541-553

Mcllvaine, T. C. (1921) J. Biol. Chem. 49, 183-186Muir, H. (1958) Biochem. J. 69, 195-204Muir, H. (1969) Amer. J. Med. 47, 673-690Partridge, S. M., Davis, H. F. & Adair, G. S. (1961)

Biochem. J. 79, 15-26Rod6n, L. (1970) in Metabolic Conjugation and Metabolic

Hydrolysis (Fishman, W. H., ed.), pp. 346-442,Academic Press, London and New York

Rosenberg, L., Pal, S., Beale, R. & Schubert, M. (1970)J. Biol. Chem. 245, 4112-4122

Sajdera, S. W. & Hascall, V. C. (1969) J. Biol. Chem. 244,77-87

1974

HYALURONIC ACID AND PROTEOGLYCAN AGGREGATION 581

gimfinek, Z. & Muir, H. (1972) Biochem. J. 126, 515-523

Swann, D. A. & Balazs, E. A. (1966) Biochim. Biophys.Acta 130, 112-129

Tsiganos, C. P. & Muir, H. (1967) Biochem. J. 104, 26cTsiganos, C. P. & Muir, H. (1969a) Biochem. J. 113,

879-884Tsiganos, C. P. & Muir, H. (1969b) Biochem. J. 113,

885-894

Tsiganos, C. P. & Muir, H. (1973) in Connective Tissueand Ageing (Vogel, H. G., ed.), vol. 1, pp. 132-137,Excerpta Medica, Amsterdam

Tsiganos, C. P., Hardingham, T. E. & Muir, H. (1971)Biochim. Biophys. Acta 229, 529-534

Tsiganos, C. P., Hardingham, T. E. & Muir, H. (1972)Biochem. J. 128, 121 P

Woessner, J. F. (1961) Arch. Biochem. Biophys. 93,440-447

Vol. 139

hyaluronic acid in cartilage and proteoglycan aggregation

Documents