A chondroitin sulphate proteoglycan from human cultured glial and

Biochem. J. (1984) 221, 845-853Printed in Great Britain

845

A chondroitin sulphate proteoglycan from human cultured glial and gliomacells

Structural and functional properties

Borje NORLING,*t Bengt GLIMELIUS: and Ake WASTESONt*Department of Medical and Physiological Chemistry, Swedish University of Agricultural Sciences, Uppsala,tDepartment of Medical and Physiological Chemistry and Clinical Research Centre, University of Linkoping,

Linkoping, tDepartment of Pathology, University of Uppsala, Uppsala, Sweden

(Received 15 March 1984/Accepted 24 April 1984)

A chondroitin sulphate proteoglycan capable of forming large aggregates withhyaluronic acid was identified in cultures of human glial and glioma cells. The glial-cell- and glioma-cell-derived products were mutually indistinguishable and had somebasic properties in common with the analogous chondroitin sulphate proteoglycan ofcartilage: hydrodynamic size, dependence on a minimal size of hyaluronic acid forrecognition, stabilization of aggregates by link protein, and precipitability with anti-bodies raised against bovine cartilage chondroitin sulphate proteoglycan. However,they differed in some aspects: lower buoyant density, larger, but fewer, chondroitinsulphate side chains, presence of iduronic acid-containing repeating units, andabsence (<1%) of keratan sulphate. Apparently the major difference betweenglial/glioma and cartilage chondroitin sulphate proteoglycans relates to the glycanrather than to the protein moiety of the molecule.

Chondroitin sulphate proteoglycan (CSPG)from cartilage is capable of reacting specificallywith hyaluronic acid (HA) with the formation oflarge aggregates (Hardingham, 1981). These aggre-gates are deposited in the extracellular matrix ofcartilage and are held to be of importance for thestructure and function of this compartment; theywere long considered to be unique to cartilage(Wiebkin et al., 1975). The demonstration thatcultures of human glial cells manufacture aggregat-ing CSPG became an indicator that CSPG-HAinteraction may be a more generalized phenom-enon, not restricted to the extracellular matrix ofcartilage (Norling et al., 1978). Support for thisidea has since accumulated, as new tissues and celltypes have been found to be producers of aggregat-ing CSPG (Wight & Hascall, 1983).Whereas the structural features of cartilage

CSPG have been studied in detail (Hardingham,1981), those of the analogous non-cartilage productare still relatively unknown (Oegema et al., 1979;McMurtrey et al., 1979; Coster et al., 1979; Poole

Abbreviations used: CS, chondroitin sulphate; CSPG,chondroitin sulphate proteoglycan; HA, hyaluronicacid; GuHCI, guanidinium chloride.

et al., 1982; Wight & Hascall, 1983). CartilageCSPG has a large hydrodynamic size (Ka,. = 0.3 onSepharose CL-2B) and a high buoyant density(> 1.70g/ml) due to a high degree of substitutionwith CS chains (Mr about 20000). It interacts withHA via a globular CS-poor portion of the coreprotein; the binding is stabilized by link proteins.Only HA fragments consisting of five or morerepeating disaccharide units are recognized by thebinding site. These properties were utilized in thepresent work for a comparative analysis of CSPGderived from cultures of glial and glioma cells; theyboth were similar to, but not identical with,cartilage CSPG.

ExperimentalChemicals

Carrier-free [35S]sulphate and [ 251]iodide wereobtained from The Radiochemical Centre, Amer-sham, U.K. Sepharose and Sephadex gels andDEAE-Sephacel were purchased from PharmaciaFine Chemicals, Uppsala, Sweden. WhatmanDEAE-cellulose (DE-52) was a product of What-man Biochemicals, Maidstone, Kent, U.K. Bacte-

Vol. 221

B. Norling, B. Glimelius and A. Wasteson

rial chondroitinase ABC (EC 4.2.2.4) and chon-droitinase AC (EC 4.2.2.5), and the unsaturatedreference disaccharides from chondroitin sul-phate (CS), 2-acetamido-2-deoxy-3-O-(f-D-gluco-4-enepyranosyluronic acid)-4-O-sulpho-D-galact-ose (A-di-4S) and 2-acetamido-2-deoxy-3-O-(P-D-gluco-4-enepyranosyluronic acid-6-O-sulpho-D-galactose (A-di-6S), were provided by MilesLaboratories, Elkhart, IN, U.S.A. Testicular hyal-uronidase (EC 3.2.1.35) was provided by AB Leo,Helsingborg, Sweden, Practical-grade guanidin-ium chloride (GuHCl) was obtained from SigmaChemical Co., St. Louis, MO, U.S.A.; it wasfurther purified as described by Norling et al.(1978).

Preparation of standardsHigh-molecular-mass HA was prepared by

chromatographing HA on Sepharose CL-2B in 4M-GuHCl; the excluded portion (HA,) was pooled.HA oligosaccharides of specific size classes were

prepared by partial digestion ofHA with testicularhyaluronidase, followed by chromatography on aSephadex G-50 (superfine grade) column(2cm x 200cm) in 1 M-NaCl (Wasteson et al.,1973). Fractions corresponding to the even-num-bered oligosaccharides tetrasaccharides to dodeca-saccharides (HA4 to HA12) were pooled, desaltedand freeze-dried. Standards of HA8, HA10 andHA12 were generously given by Dr. T. C. Laurentof this Department.

Cartilage proteoglycan was prepared from bo-vine nasal septum, essentially as described byHascall & Sajdera (1969). A number of proteinaseinhibitors were included during the preparation, assuggested by Oegema et al. (1975). The procedureincluded centrifugation first under associativeconditions (fraction. Al), then under dissociativeconditions (fraction Al-DI) (Heinegard, 1972).The product was dialysed against 0. 15M-NaCl andkept at - 20°C until used. Cartilage CSPG waslabelled with 1 25I by using the chloramine-Tmethod (Hunter & Greenwood, 1962).

Link proteins (Hardingham, 1979) were recov-ered from the lightest fractions of the dissociativestep mentioned above. It contained two proteinswith Mr values about 41000 and 44000, as shownby sodium dodecyl sulphate/polyacrylamide-electrophoresis under reducing conditions(Laemmli, 1970). The material was dialysedagainst 4M-GuHCl and stored at - 20°C; a portionof the preparation was labelled with 125I (Hunter&Greenwood, 1962). Fibronectin was kindly givenby Dr. S. Johansson and collagen types I and III byDr. K. Rubin of this Department; laminin waspurchased from Bethesda Research Laboratories,Rockville, MD, U.S.A. The preparations werelabelled with 1251 as described above.

CS fractions of known M, values were availablein our laboratory (Wasteson, 1971). 14C-labelledhexa- or tetra-saccharides were obtained afterhyaluronidase treatment of 14C-labelled CS, fol-lowed by Sephadex G-25 chromatography (Amadoet al., 1974; Ingmar & Wasteson, 1979). Theanalogous penta- and tri-saccharides were ob-tained from the respective higher oligosaccharidesby digestion with f-glucuronidase. Treatment of14C-labelled CS with chondroitinase ABC yieldedthe unsaturated disaccharides, A-di-4S and A-di-6S.Cell culture and labelling conditionsThe normal human glial-cell lines U-787 CG and

U-1508 CG and the human established malignantglioma-cell line U-251 MG were kept underconditions previously described (Glimelius et al.,1978a). The experimental details for incubation ofcultures with [35S]sulphate have been describedpreviously (Glimelius et al., 1978a).AH-Sepharose chromatography

35S-labelled samples were applied to a1.0cm x 2.0cm column of AH-Sepharose 4B,equilibrated with 10mM-Tris/HCl buffer, pH8.0,containing 0. 15M-NaCl, 1 mM-Na2SO4 and 0.1%bovine serum albumin and operated at 4°C at aflow rate of 18 ml/h. The column was then washedwith the same buffer, supplemented with 0.4M-GuHCl, to decrease the radioactivity in the eluateto below 100c.p.m./ml. Elution was then madewith 10mM-Tris/HCl buffer, pH8.0, containing4M-GuHCl, 0.15M-NaCl, 1 mM-Na2SO4 and 0.1%bovine serum albumin (buffer A). Under theseconditions both intact [35S]proteoglycans and free35S-labelled glycosaminoglycan chains were elutedfrom the column; gradient elution showed that thecritical concentration of GuHCl was about 1.OM,indicating a stronger binding of CSPG to AH-Sepharose 4B than to DEAE derivatives (DE-52DEAE-cellulose and DEAE-Sephacel); in thelatter case desorption occurred at about 0.4M-GuHCl. The strong binding of CSPG conferredselectivity on to the AH-Sepharose step; e.g. 1251-fibronectin or 125I-laminin adsorbed-on the col-umn were desorbed at 0.35M-GuHCI, i.e. at amarkedly lower concentration of GuHCl thanrequired for the elution of glycan components(results not shown). Analogous experiments with125I-labelled (Hunter & Greenwood, 1962) calfserum proteins showed that about 50% of theapplied radioactivity was retained on the column;virtually all of this material was desorbed at 0.20M-GuHCl.Gel chromatography

Gel chromatography on Sephadex G-25(1 cm x 190cm column) and Sephadex G-50

1984

846

Chondroitin sulphate proteoglycan from cultured glial cells

(2cm x 200cm or 1 cm x 190cm column) was car-ried out in 1 M-NaCl. The Sephadex G-50 columnswere calibrated with known standards of HAoligosaccharides (HA8, HA1O and HA12). Sephar-ose CL-2B chromatography was either on a1 cm x 90cm column, equilibrated with 10mM-Tris/HCl buffer, pH8.0, containing 0.15M-NaCl,1 mM-Na2SO4 and 0.1% bovine serum albumin, ora 1 cm x 190cm column, equilibrated with bufferA. Sepharose CL-4B (1 cm x 190cm column) andSepharose CL-6B (1 cm x 140cm or 1 cm x 190cmcolumn) chromatography was performed withbuffer A. The former Sepharose CL-6B columnwas calibrated with CS fractions of known M,values; a calibration curve was constructed byplotting log Mr versus Ka,. (Wasteson, 1971).[Desalting of oligosaccharides was carried out bygel chromatography on a Sephadex G-15 column(2cm x 90cm) equilibrated with 10% (v/v) eth-anol.] All columns were operated at 4°C.

Density-gradient centrifugation in CsClCentrifugation was carried out in an MSE SS65

ultracentrifuge with an 8 x 14ml angle rotor.Samples were prepared for centrifugation bymixing with weighed amounts of CsCl (finalconcentration 0.28, 0.45 or 0.70g/ml). Runs wereperformed at 1000OOg for 60-72h at 18°C. Afterthe run the tubes were either punctured andevacuated via a steel needle through the bottom ofthe tubes or emptied in the reverse direction byintroducing a dense solution (Fluorinert FC-77;3M, St. Louis, MO, U.S.A.) at the bottom of thetubes, driving the lighter fractions through a tight-fitting cap at the top. Fractions of volume 0.5-0.8ml were collected; the density of individualfractions was determined by weighing 200 4u1samples in a glass pipette.

Isolation and identification ofCSPGfrom cell culturemedium

In order to minimize the influence of possiblepreparation artifacts, three different methods wereused to isolate CSPG from glial-cell or glioma-cellculture medium; the resulting products turned outto be similar in all respects examined. Theharvested medium was used either after 10-foldconcentration under reduced pressure (methods Iand II) or used directly (method III). The recoveryof CSPG in individual steps was 90% or better.

In method I the sample of concentrated mediumwas first acidified by the addition of 0.3 ml of 0.1 M-sodium acetate buffer (pH 5.5)/ml of concentratedmedium and then digested with platelet heparitin-ase (Oldberg et al., 1980) (loig of plateletprotein/2ml of concentrated medium). After addi-tion of an equal volume of 8M-GuHCl the samplewas chromatographed on a Sepharose CL-4B

column (1 cm x 190cm) eluted with buffer A. 35S-labelled material eluted with the void volume ofthe column was pooled. It was identified as CSPGby the following criteria. It was susceptible topapain digestion (Glimelius et al., 1978a) ortreatment with alkali (Norling et al., 1978),yielding components with K3v. 0.83 on SepharoseCL-2B and Kav. 0.30 on Sepharose CL-6B. Further,on DEAE-cellulose ion-exchange chromato-graphy, the products migrated like a standard offree CS chains, carrying, on an average, one sul-phate residue per disaccharide unit. Digestion withchondroitinase AC converted 90% of the materialinto products chromatographing in the included-volume fraction of a Sephadex G-25 column elutedwith 1 M-NaCl.

In method II, HA, was first added to the sample(10pg/2ml of concentrated medium). After incu-bation at room temperature for 6h or more, thesample was chromatographed on a Sepharose CL-2B column (1 cm x 90cm) eluted with 10mM-Tris/HCl buffer, pH8.0, containing 0.15M-NaCl,lmM-Na2SO4 and 0.1% bovine serum albumin.35S-labelled material eluted in the void volume waspooled and concentrated to 2 ml in a Minicon cell;then an equal volume of 8 M-GuHCl was added andthe sample chromatographed on a Sepharose CL-2B column (1 cm x 190cm) in buffer A. The 35S-labelled material was now shifted to an includedposition, whereas HA, remained in the void-volume fraction. The pooled 35S-labelled materialwas shown to be CSPG by the same criteria asgiven above for method I.

In method III the unprocessed culture mediumwas passed through a column of AH-Sepharose 4B(see above). 35S-labelled material eluted withbuffer A was pooled and purified by density-gradient centrifugation in CsCl (final concentra-tion 0.28g/ml); the material recovered in thebottom 5/20 of the gradient was pooled. The poolwas rechromatographed on AH-Sepharose 4B; theradioactivity eluted from the column with buffer Awas collected and rechromatographed on Sephar-ose CL-2B. Fractions corresponding to Kay. 0.20.5were pooled, mixed with 2mg of fraction A 1-Dlcarrier and re-centrifuged in a CsCl densitygradient (final concentration 0.70g/ml); materialhaving p = 1.48-1.59g/ml was collected. It wassubjected to another cycle of AH-Sepharose 4Bchromatography and finally chromatographed onSepharose CL-2B, producing a peak with Ka,. 0.3.This material was pooled and identified as CSPGby the criteria described above.

Aggregation assayThe capability of glial-cell or glioma-cell CSPG

to form large aggregates in the presence of HA,under different conditions was tested by gel

Vol. 221

847


chromatography. HA, (about 2pug in 10ul of 4M-GuHCl) was added to 35S-labelled CSPG(40000c.p.m. in lOmI of lOmM-Tris/HCl buffer,pH8.0, containing 0.15M-NaCl) and the mixturewas left at room temperature for 6h. The samplewas then chromatographed on a column ofSepharose CL-2B operated under associative con-ditions. The radioactivity eluted with the voidvolume was ascribed to [35S]CSPG-HA1 aggre-gates. HA oligosaccharides of different size classeswere tested for their effect on HA-induced aggre-gate formation. A 150-fold excess (about 300pg) ofoligosaccharides was then added at the same timeas HAI.

The. stabilizing effect of link protein was testedeither with a preparation of purified link protein(15pg per incubation) or with CSPG Al fraction(HeinegArd, 1972; 15pg per incubation), addedtogether with HAI. The mixtures were made to 4M-GuHCl and then dialysed against an associativebuffer (IOM-Tris/HCl buffer, pH8.0, containing0.15M-NaCl). At that time HA oligosaccharideswere added, and their effect on HAI-CSPGaggregates was examined as described above.

Immunological techniquesAntibodies against cartilage proteoglycan were

raised in a rabbit by injecting 130Mg portions of theantigen (isolated as described above) into a lymphnode on the back of each thigh. The first dose wasmixed with an equal volume of Freund's completeadjuvant; booster doses given 2 and 4 weeks laterwere mixed with Freund's incomplete adjuvant. At2 weeks after the last injection the rabbit was bledto death; the recovered serum was stored at - 200Cuntil used.

Antiserum reacting only with cartilage linkproteins but not with cartilage CSPG was preparedby passing antiserum against bovine link proteins(kindly provided by Dr. A. Tengblad of thisDepartment) over a column of Sepharose 4B withbound chondroitinase ABC-digested cartilageCSPG.

Immunoprecipitation was performed by mixingthe sample (5 MI, containing 35S-labelled CSPG orI25I-labelled cartilage CSPG in 4M-GuHCI con-taining 0.1% bovine serum albumin) with 50pl ofantiserum or pre-immune serum. Incubation wasat room temperature for 1 h and at 4°C overnight;then 100p1 of suspended fixed Staphylococcus Abacteria (Arvidsson et al., 1970; Kessler, 1976)[15% (v/v) in 10mM-sodium phosphate buffer,pH7.4, containing 0.15M-NaCl, 0.1% Tween 80and 0.02% NaN3] was added and incubationprolonged with agitation at room temperature for1 h. The suspension was finally diluted with 2ml oflOmM-Tris/HCl buffer, pH8.0, containing 0.15M-NaCl, 0.1% Tween 80, 0.1% bovine serum albu-

min, 1 mM-Na2SO4 and 0.02% NaN3, and centri-fuged. The resulting supernatant and pellet wereanalysed for radioactivity.

Degradative methodsTreatment with alkali (fl-elimination), digestion

with chondroitinase AC or chondroitinase ABCand reductive alkylation were carried out as pre-viously described (Glimelius et al., 1978a; Norlinget al., 1978). The preparation of chondroitinaseABC, used to produce CS-free CSPG protein core,was first shown to have no degrading activity on35S-labelled heparan sulphate proteoglycan, amarkedly proteinase-susceptible molecule. Degra-dation with testicular hyaluronidase or fl-glucuron-idase was as described by Ingmar & Wasteson(1979). Periodate oxidation and subsequent alka-line elimination was carried out as described byFransson & Carlstedt (1974); non-labelled derma-tan sulphate (kindly given by Dr. U. Lindahl ofthis Department) was added as a carrier.

High-voltage paper electrophoresisHigh-voltage paper electrophoresis was per-

formed as described by Glimelius et al. (1978b).Disaccharides migrating half-way between thestandards of monosulphated disaccharidcf (Adi-4/6-S) and inorganic sulphate were assigned twosulphate residues per repeating disaccharide unit.

Paper chromatographyThe distribution of Adi-4-S and Adi-6-S in

disaccharides obtained after chondroitinase treat-ment was determined by descending paperchromatography as described by Ingmar & Waste-son (1979) with solvent C (butan-I-oIaceticacid/1 M-NH3, 2:3 :1, by vol.); standards of Adi-4-S or Adi-6:-S were located by a silver-dip technique(Smith, 1960).

Radioactivity35S radioactivity was measured in a Nuclear-

Chicago Isocap/300 liquid-scintillation counterwith Scintillator 299 as the scintillation medium;125I was measured in a Packard model 5166 Auto-gamma spectrometer.

Results

Analysis of intact CSPGHydrodynamic size and buoyant density. Gel

chromatography on Sepharose CL-2B in thepresence of 4M-GuHCl indicated that CSPGisolated from glial-cell or glioma-cell culturemedium had similar hydrodynamic size; they bothproduced symmetrical peaks with a Kay. of 0.30(Fig. 1). A standard of '251-labelled CSPG from

1984

848


VO V,

2.1.0

Cu

Cu

0.5x

0

50 75 100 125 150 1Effluent volume (ml)

Fig. 1. Gel chromatography on a Sepharose CL-2B column(1cm x 190cm) ofCSPG isolatedfrom the medium of35S-

labelled glial-cell (U-787CG) culturesFor full experimental details see the text. In this andsubsequent Figures V0 and V, indicate void volumeand total volume respectively.

E 3(CLQ

C)

Cu.5CZ*° 2tCL

x

0IC

E0-

._

Fraction no.

Fig. 2. Density gradient centrifugation in CsCl (0.70g/ml)ofCSPG isolatedfrom the medium of 35S-labelled glioma-

cell (U-251 MG) culturesFor full experimental details see the text.

bovine nasal cartilage migrated with a similar Kay.,0.29.

Isopycnic centrifugation in CsCl in the presenceof 4M-GuHCl showed buoyant densities of about1 .55g/ml for both glial-cell and glioma-cell derivedCSPG (Fig. 2). In contrast, the standard of 1 251-labelled cartilage CSPG under the same conditionsdid not form a band in the gradient but appeared inthe high-density bottom fractions (> 1.67g/ml).

Aggregation with HA. Titration of different sizeclasses of HA oligosaccharides (HA4-HA12) fortheir capacity to inhibit complex-formation be-tween glial-cell or glioma-cell CSPG and HAshowed that the smallest active oligosaccharidewas the decasaccharide, HA1o (Fig. 3). Althoughthe octasaccharide fraction also had some effect,this was ascribed to decasaccharide, known tocontaminate the octasaccharide preparation. Thisview was supported by the findings of controlexperiments, in which the HA-induced aggrega-tion of cartilage CSPG was similarly affected bythe respective oligosaccharide preparations. It wasconcluded that the HA-binding region of glial-cellor glioma-cell and cartilage CSPG had the samerequirement for minimal size of the HA molecule.

In contrast, the aggregate formed by HA andglial-cell or glioma-cell CSPG in the presence oflink proteins derived from bovine nasal cartilagewas not affected by HA oligosaccharides (results

not shown). This would indicate that the cartilage-derived link proteins stabilized the complex inwhich glial-cell or glioma-cell CSPG was involved.Similar conclusions were obtained in analogousexperiments in which solubilized cartilage CSPG-HA aggregate (fraction Al; Heinegard, 1972)was reconstituted in the presence of small amountsof glial-cell CSPG. The latter was apparentlyintegrated in the newly formed link-stabilizedcomplex, since HA oligosaccharides failed todisplace any glial-cell CSPG from the high-molecular-mass material (results not shown).

Reduction and alkylation of glial-cell or glioma-cell CSPG completely abolished the formation ofaggregates with HA (results not shown). Thusthese CSPGs, in this respect, were similar tocartilage CSPG.

Immunoprecipitation. Rabbit antiserum raisedagainst bovine nasal-cartilage CSPG precipitated54% (over the control) of the 35S-labelled CSPGisolated from glial-cell culture medium. In ananalogous experiment 67% of cartilage CSPG wasprecipitated by the same antiserum. The antiserumdid not react with a number of isolated extra-cellular-matrix components, such as human orbovine '25I-labelled fibronectin, laminin or colla-gen type I, III or V, but showed some activityagainst CSPG link proteins of bovine origin. Thelatter finding raised the possibility that link

Vol. 221

849


1.51(a) VO V,

S

C)

1-1

e

.2

C)

cd:._

Cd

x0en

In

0 25 50 75 100

Effluent volume (ml)0 25 50 75 100

Effluent volume (ml)

Fig. 3. Effect ofHA oligosaccharides on HA-induced aggregation of CSPG isolated from glial cells (U-787 CG)Under associative conditions addition ofHA to glial-cell CSPG led to the formation of aggregates as revealed by theexcluded-volume material on Sepharose CL-2B chromatography (1cm x 90cm column), eluted with lOmM-Tris/HCIbuffer, pH 8.0, containing 0.15M-NaCl, mM-Na2SO4 and 0.1% bovine serum albumin (a). This phenomenon was

inhibited in the presence of HA10, as shown by the marked increase in included material (b). For full experimentaldetails see the text.

proteins caused the effect on [35S]CSPG, i.e. thatthe [35S]CSPG were co-precipitated with linkproteins without being specifically recognized bythe antiserum. Therefore an antiserum against linkprotein was tested for its ability to precipitate theCSPG; it had no activity in this respect. It wasconcluded that the glial-cell-derived CSPG hadantigenic determinants in common with thecartilage-derived analogue.

Analysis of chondroitinase ABC digested CSPGChondroitinase ABC digestion of 35S-labelled

glial-cell CSPG left 2% of the radioactivity in ahigh-Mr form, representing protein core withremnants of CS chains still attached (Hascall et al.,1972). Chromatography of this material on Sephar-ose CL-6B produced one major peak, as shown inFig. 4(a). In contrast, cartilage CSPG subjected tothe same treatment (Fig. 4b) showed a more hetero-geneous distribution, presumably as a result ofthe presence of keratan sulphate. However, itsmajor included component was similar in size tothat of glial-cell CSPG. After alkali treatment of

the chondroitinase ABC-resistant glial-cell mate-rial half of it was eluted as a homogeneous peakslightly ahead of a disaccharide on chromato-graphy on Sephadex G-50. The other half of thematerial showed Mr values in the range 5000-20000, as determined by chromatography onSepharose CL-6B. It may represent incompletelydegraded CS chains, or possibly keratan sulphate.Therefore, if present, keratan sulphate could onlyamount to a maximum of 1% of the total 35S radio-activity in glial-cell CSPG.

Analysis of CS chains obtained by fl-elimination ofCSPG

Hydrodynamic size and charge properties. Ion-exchange chromatography of [35S]CS chains de-rived from the CSPG of the medium of 35S-labelled glial-cell or glioma-cell cultures yieldedindistinguishable, sharp and symmetrical peaks,eluted only slightly after the standard of CS,carrying one sulphate group per disaccharide unit(results not shown). Gel chromatography of thesame components on Sepharose CL-6B similarly

1984

):

.2C)0co

x

0

850


(a) VO Vt

2.0 - (b) Vo V,

100

Effluent volume (ml)150 0


Fig. 4. Gel chromatography on a Sepharose CL-6B column (1cm x 190cm) ofchondroitinase ABC-digested glial-cell (35S) or

cartilage CSPG (A280)Only the macromolecular chondroitinase products were applied; the low-M, products had been removed bySephadex G-50 chromatography. For full experimental details see the text.

produced identical elution profiles (results notshown); the Ka, values of the peak fraction were

0.30 and 0.31 for the glial-cell- and glioma-cell-derived CS chains respectively. Judged from theKav. values of calibration standards (Wasteson,1971), these values correspond to Mr values of44000 and 42000 respectively

Susceptibility to chondroitinase. ChondroitinaseABC digestion of the glial-cell- and glioma-cell-derived 35S-labelled CS chains converted about98% of the glial-cell material as well as the glioma-cell material into disaccharides, as demonstratedby Sephadex G-25 chromatography (results notshown). The remaining non-disaccharide portion(2%) was eluted slightly ahead of the standardtetrasaccharide. Since most of it was degraded todisaccharides by chondroitinase AC, it probably

represented material located close to the carbo-hydrate-protein linkage region; the analogousposition in cartilage-derived CS is partially resis-tant to chondroitinase ABC (Hascall et al., 1972).Treatment of the glial-cell or glioma-cell CS

chains with chondroitinase AC resulted in identi-cal product patterns, as indicated by Sephadex G-25 chromatography (Fig. 5). A major part (about80%) was converted into disaccharides. A minorportion (9%) appeared at a position slightly afterthat of a standard tetrasaccharide. Periodate/alkalitreatment (known to degrade unsulphated but not2-sulphated iduronic acid; Fransson & Carlstedt1974) of this material caused later elution on theSephadex G-25 column, half of the productsmigrating slightly ahead of the disaccharidefraction, and half being eluted at the position of

Vol. 221

E 1.5CiU

'tuCZ

0

Cd

CT

In 1.0)

0 '' 50

851

0

0.5 _-


V0 hexa tetra di V,(b)

VO hexa tetra di V,

0.6_.

C)

0-

0Cd

CZ

x

0.2

I .175 100

Effluent volume (ml)125 0 50 75 100 125


Fig. 5. Gel chromatography on Sephadex G-25 ofglial-cell (a) or glioma-cell (b) CS digested with chondroitinase ACFor full experimental details see the text. The position of oligosaccharides are indicated.

inorganic sulphate (results not shown). Theseproducts should be derived from a tetrasaccharidecontaining a non-sulphated iduronic acid residueat its penultimate position. The remaining portionof the chondroitinase AC products (11%) was inthe excluded-volume fraction on Sephadex G-25; itstayed in that position after the periodate/alkalistep (results not shown) and therefore should becomposed of contiguous blocks of disaccharidescontaining 2-sulphated iduronic acid residues.

Constituent disaccharides. High-voltage paper

electrophoresis of the disaccharide portions of thechondroitinase ABC or chondroitinase AC digestsrevealed no differences between the glial-cell- andglioma-cell-derived products. Whereas essentiallyonly monosulphated disaccharides were found inchondroitinase AC digests, significant amounts(5%) of disulphated disaccharide were demon-strated in the material obtained after digestionwith chondroitinase ABC. Paper chromatographyof the disaccharide fraction showed approximatelyequal amounts of mono-4- and mono-6-sulphatedspecies. The average disaccharide composition ofan individual CS chain that could be deduced fromthe present analyses was the same for the glial-cellproduct as for the glioma-cell product (Table 1). As

Table 1. Calculated average distribution ofconstituent dis-accharides in glial-cell or glioma-cell CS

CSPG was isolated from the culture medium ofglial-cell or glioma-cell cultures; CS chains were

then obtained by alkali-induced fl-elimination. Theresults are based on degradative analysis of the CSchains, assuming a uniform chain length (92 disac-charides) and a uniform composition in the respec-tive populations. Abbreviations: IdUA, iduronicacid; GlcUA, glucuronic acid.

Number per CS chain

Disacharide typeIdUA(2-S)-GalNAc(4-S)IdUA-GalNAc(4-S)GlcUA-GalNAc(4-S)GlcUA-GaINAc(6-S)

Glial-cellCs53

4242

Glioma-cellCS53

4242

can be seen from the Table the observed pattern ofiduronic acid-containing to glucuronic acid-con-taining disaccharides (10:90) was different fromthat held to govern the process of self-aggregationamong CS/dermatan sulphate chains (50:50)(Fransson, 1976).

1984

0.8 k (a)

0.61-

U

C3

-o._

0C13

U)V)

x

8

0+4

0.2 I-

0'' 50

852

f

Chondroitin sulphate proteoglycan from cultured glial cells 853

Discussion

The present results demonstrate that the CSPGsproduced by human normal glial or malignantglioma cells in culture have structural and func-tional properties in common with cartilage-derivedCSPG. The hydrodynamic sizes of the monomericmolecules were the same, as judged by SepharoseCL-2B chromatography under dissociative condi-tions, and large-size aggregates were similarlyformed with HA under associative conditions. Thesame principles were apparently operating in theformation of these aggregates. Thus, for therecognition of HA, the glial-cell- or glioma-cell-derived CSPG showed the same minimal require-ment for HA chain length as the analogouscartilage product. Further, their binding to HAwas similarly abolished by reducing agents, andstabilized by link proteins. This suggests that thedifferent CSPGs may have similar HA-bindingstructures. The common antigenic determinantsindicated by the immunoprecipitation experi-ments may therefore reside in this region of theCSPG molecule; the CSPG antibodies wereprobably directed against the globular portion ofthe molecule, since intact rather than chon-droitinase-treated CSPG had been used forimmunization.However, significant structural differences were

noted between glial-cell/glioma-cell CSPG andcartilage CSPG. Whereas they all had the samehydrodynamic size, they differed with respect tobuoyant density (p = 1.55g/ml versus > 1.67g/ml)and size of the chondroitin sulphate side chains(Mr 40000 versus 20 000). Since the size of the poly-peptide core was similar, it can be inferred thatglial-cell/glioma-cell CSPG contains fewer CS sidechains than does cartilage CSPG. Additionaldifferences relate to the composition of thesubstituent glycosaminoglycan chains in the re-spective types of CSPG. In contrast with theanalogous cartilage CSPG, glial-cell/glioma-cellCSPG showed significant amounts of iduronicacid-containing repeating units (about 10% of thetotal) and virtually lacked keratan sulphate.

It is concluded that the glial-cell CSPG isindistinguishable from the glioma-cell CSPG, andthus much more related to the latter than tocartilage CSPG, i.e. a product of a cell with adifferent origin. The major difference betweenglial-cell/glioma-cell CSPG and cartilage CSPGrelate to the glycan part of the molecule; theprotein cores may well be identical. If so, it wouldappear as if the particular biosynthetic machineryof each cell type would determine the subsequentsubstitution with carbohydrate.

This work was supported by the Swedish MedicalResearch Council (1 3X-4486 and 1 3X-2309) and KonungGustaf V:s 80-Arsfond. The skilful technical assistance ofMs. M. Lindstrom, Ms. S. Wennergren and Ms Y.Ohgren is gratefully acknowledged.

References

Amad6, R., Ingmar, B., Lindahl, U. & Wasteson, A.(1974) FEBS Lett. 39, 49-52

Arvidsson, S., Holme, T. & Wadstr6m, T. (1970) J.Bacteriol. 104, 227-233

C6ster, L., Carlstedt, I. & Malmstrom, A. (1979) Bio-chem. J. 183, 669-681

Fransson, L.-A. (1976) Biochim. Biophys. Acta 437, 106-115

Fransson, L.-A. & Carlstedt, I. (1974) Carbohyd. Res. 36,349-358

Glimelius, B., Norling, B., Westermark, B. & Wasteson,A. (1978a) Biochem. J. 172, 443-456

Glimelius, B., Norling, B., Westermark, B. & Wasteson,A. (1978b) Exp. Cell Res. 117, 179-189

Hardingham, T. (1979) Biochem. J. 177, 237-247Hardingham, T. (1981) Biochem. Soc. Trans. 9, 489-497Hascall, V. C. & Sajdera, S. W. (1969) J. Biol. Chem. 244,

2384-2396Hascall, V. C., Riolo, L. R., Hayward, J., Jr. &

Reynolds, C. C. (1972) J. Biol. Chem. 247, 4521-4528Heinegard, D. (1972) Biochim. Biophys. Acta 285, 181-

192Hunter, W. M. & Greenwood, F. C. (1962) Nature

(London) 194, 495-496Ingmar, B. & Wasteson, A. (1979) Biochem. J. 179, 7-13Kessler, S. W. (1976) J. Immunol. 117, 1482-1490Laemmli, U. K. (1970) Nature (London) 227, 680-685McMurtrey, J., Radhakrishnamurthy, B., Dalferes,

E. R., Jr., Berenson, G. S. & Gregory, J. D. (1979) J.Biol. Chem. 254, 1621-1626

Norling, B., Glimelius, B., Westermark, B. & Wasteson,A. (1978) Biochem. Biophys. Res. Commun. 84,914-921

Oegema, T. R., Jr., Hascall, V. C. & Dziewiatkowski,D. D. (1975) J. Biol. Chem. 250, 6151-6159

Oegema, T. R., Jr., Hascall, V. C. & Eisenstein, R.(1979) J. Cell Biol. 254, 1312-1318

Oldberg, A., Heldin, C.-H., Wasteson, A., Busch, C. &Hook, M. (1980) Biochemistry 19, 5755-5762

Poole, A. R., Pidoux, I., Reiner, A., Coster, L. & Hassel,J. R. (1982) J. Cell Biol. 93, 910-920

Smith, I. (1960) in Chromatographic and ElectrophoreticTechniques (Smith, I., ed.), vol. 1, pp. 246-260,Interscience, New York

Wasteson, A. (1971) Biochem. J. 122, 477-485Wasteson, A., Westermark, B., Lindahl, U. & Ponten, J.

(1973) Int. J. Cancer 12, 169-178Wiebkin, O., Hardingham, T. & Muir, H. (1975) in

Extracellular Matrix Influences on Gene Expression(Slavkin, H. C. & Greulich, R. C., eds.), pp. 209-222,Academic Press, New York

Wight, T. N. & Hascall, V. C. (1983) J. Cell Biol. 96, 167-176

Vol. 221

Documents

A chondroitin sulphate proteoglycan from human cultured glial and