Biochem. J. (1981) 193, 143-153 143Printed in Great Britain
Isolation and characterization of dermatan sulphate proteoglycans frombovine sclera
Lars COSTER and Lars-Ake FRANSSONDepartment ofPhysiological Chemistry 2, University ofLund, S-220 07 Lund, Sweden
(Received 19 May 1980/Accepted 12 August 1980)
1. Proteoglycans were extracted from sclera with 4M-guanidine hydrochloride in thepresence of proteinase inhibitors and purified by ion-exchange chromatography anddensity-gradient centrifugation. 2. The entire proteoglycan pool was characterized bycompositional analyses and by specific chemical (periodate oxidation) and enzymic(chondroitinases) degradations. The glycan moieties of the molecules were exclusivelygalactosaminoglycans (dermatan sulphate-chondroitin sulphate co-polymers). Inaddition, the preparations contained small amounts of oligosaccharides. 3. The scleralproteodermatan sulphates were fractionated into one larger (I) and one smaller (II)component by gel chromatography. Proteoglycan I was eluted in a more excludedposition on gel chromatography in 0.5 M-sodium acetate than in 4.0M-guanidinehydrochloride. Reduced and alkylated proteoglycan I was eluted in the same position (in0.5M-sodium acetate) as was the starting material (in 4.0M-guanidine hydrochloride).The elution position of proteoglycan II was the same in both solvents. Proteoglycans Iand II had s° values of 2.8 x 10-13 and 2.2 x 10-13 s respectively in 6.0M-guanidinehydrochloride. 4. The two proteoglycans differed with respect to the nature of theprotein core and the co-polymeric structure of their side chains. Also proteoglycan Icontained more side chains than did proteoglycan II. The dermatan sulphate side chainsof proteoglycan I were D-glucuronic acid-rich (80%), whereas those of proteoglycan IIcontained equal amounts of D-glucuronic acid and L-iduronic acid. Furthermore, theco-polymeric features of the side chains of proteoglycans I and II were different. Theprotein core of proteoglycan I was of larger size than that of proteoglycan II. The latterhad an apparent molecular weight of 46000 (estimated by sodium dodecyl sulphate/polyacrylamide-gel electrophoresis), whereas the former was >100000. In addition, theamino-acid composition of the two core preparations was different. 5. As proteoglycan Ialtered its elution position on gel chromatography in 4M-guanidine hydrochloridecompared with 0.5 M-sodium acetate it is proposed that a change in conformation or adisaggregation took place. If the latter hypothesis is favoured, aggregation may be dueto self-association or mediated by an extrinsic molecule, e.g. hyaluronic acid.
Fibrous connective tissues contain an abundanceof collagen fibres and, in the interfibrillar matrix,characteristic macromolecular glycoconjugates,which include dermatan sulphate proteoglycans.These molecules are difficult to extract quanti-tatively in an undegraded state. By extraction withurea at elevated temperatures Toole & Lowther(1965, 1968) succeeded in isolating proteodermatansulphates from heart valves, skin and tendon. By asimilar approach Obrink (1972) prepared a proteo-glycan from pig skin. Proteoglycans have also beenextracted from tendon, arterial wall and sclera using4M-guanidine (Anderson, 1975; Antonopoulos etal., 1974; Ehrlich et al., 1975; Eisenstein et al.,
Vol. 193
1975; Radhakrishnamurthy et al., 1977). Morerecently, Damle et al. (1979) separated pig skinproteoglycan into one proteochondroitin sulphateand one proteodermatan sulphate. These studieshave demonstrated that dermatan sulphate is co-valently linked to a non-collagenous protein.
By the introduction of efficient procedures for theextraction of fibrous connective tissues (Antono-poulos et al., 1974) and the utilization of proteinaseinhibitors (Oegema et al., 1975) it became possibleto isolate proteoglycans quantitatively from suchtissues in a more native state. The present study isconcerned with the extraction, purification andcharacterization of proteoglycans from bovine
sclera. Preliminary reports have been published(C6ster & Fransson, 1975, 1976).
Experimental
MaterialsEyes from cows were obtained from the local
slaughterhouse and the sclerae were excised at 40Cwithin 6 h and immediately frozen in liquid N2.
Sepharose and Sephadex gels, as well as a proteincalibration kit, were obtained from Pharmacia FineChemicals, Uppsala, Sweden. ECTEOLA-cellulosewas purchased from Serva, Heidelberg, Germany.Dialysis tubing (nominal molecular-weight cut-off12000-14000) was a product of Spectrum MedicalInc., Los Angeles, CA, U.S.A. Chondroitinase-ACand chondroitinase-ABC (EC 4.2.2.5) were ob-tained from Miles Laboratories, Elkhart, IN, U.S.A.Papain (twice-crystallized, 16-40 BAEE units/mg)and guanidine hydrochloride (practical grade) werepurchased from Sigma Chemical Co., St. Louis,MO, U.S.A. Stock solutions of guanidine hydro-chloride (7M) were treated with activated charcoalunder stirring overnight. After filtration throughfilter paper the A280 of the clarified solution was 0.05(with water as blank). Stock solutions of urea (8 M)were passed through a mixed anion-cation exchangeresin to remove traces of cyanate immediately beforeuse. Dithiothreitol ('for biochemistry') was boughtfrom Merck, Darmstadt, Germany. All otherchemials were analytical grade.
Analytical methodsHexosamine was quantified by a modified Elson-
Morgan procedure (Antonopoulos et al., 1964).Hexose was estimated by the method of Goa (1955).Hexuronic acid was determined by the carba-zole-borate method (Bitter & Muir, 1962) andprotein by the method of Lowry et al. (1951).Automated versions of these methods were alsoemployed (Heinegard, 1973).Amino acids, glucosamine and galactosamine
were estimated by using a Durrum automaticamino-acid analyser. Hydrolysis was performed(under argon) in 6M-HCI at 1100C for 24h (foramino acids) or in 8M-HCl at 950C for 3h (forhexosamines). Norleucine was used as the internalstandard. The amounts of cysteine and cystine weredetermined as cysteic acid (on the amino-acidanalyser) after oxidation with performic acid(Moore, 1963). Hydroxyproline was quantified bythe method of Stegemann & Stalder (1967).
Neutral sugars were determined by g.l.c. of theiralditol acetates after hydrolysis in 2M-trifluoro-acetic acid at 1000C for 4h under N2 (Axelsson &Heinegard, 1975).
UltracentrifugationSedimentation velocity measurements were per-
formed in an MSE Centriscan 75 ultracentrifuge byusing the Schlieren detection system. Serial dilutionsof stock solutions were each dialysed with stirringagainst the buffer to achieve Donnan equilibrium.Sedimentation coefficients were determined as de-scribed by Schachman (1957). The s%O,w values wereobtained by extrapolation to infinite dilution.
Degradations and modificationsThe glycan side chains of proteoglycans were
released by alkaline elimination of the xylose-serinelinkage or by exhaustive proteolytic digestion of theprotein core. Alkaline cleavage was performed in0.1M-LiOH (2mg of proteoglycan/mg) at 4°C for72h. The reaction was terminated by neutralizationwith b.5M-acetic acid. Proteolysis was carried outwith papain [O.1mg/lOmg of substrate in 1ml of1.OM-NaCl/0.05 M-EDTA (disodium salt)/O.Ol M-cysteine hydrochloride/0.05 M-sodium phosphatebuffer, pH 7.01 at 65 IC for 24 h.
Periodate oxidation-alkaline elimination (selec-tive cleavage of L-iduronic acid residues) at pH3.0and 40C was used to degrade the liberated dermatansulphate side chains as described previously (Frans-son, 1974; Fransson & Carlstedt, 1974; Fransson &Coster, 1979).
Cleavage of hexosaminidic bonds to D-glucuronicacid or to all hexuronic acids in free chains orproteoglycans was accomplished by chondroitinase-AC or chondroitinase-ABC digestions. Samples(2mg/ml) were dissolved in 0.lOM-Tris acetate,pH7.3, and treated with enzyme at 370C for 4h.The amounts of enzyme used were 0.1 unit/mg(chondroitinase-AC) and 0.01 unit/mg (chon-droitinase-ABC) respectively.
Ratios between L-iduronic acid and D-glucuronicacid were obtained by measuring the amount ofunsaturated glycuronosyl residues formed afterchondroitinase digestions by the periodate-thiobar-biturate assay of Hascall et al. (1972).
Reduction and alkylation ofproteoglycans (1Omg/ml) was performed in 4 M-guanidine hydrochlor-ide/0.005 M-dithiothreitol/0.05 M-Tris/HCl, pH 8.5.After incubation at 370C for 5 h iodoacetamide wasadded to a final concentration of 0.015 M. Thesolution was kept in the dark at 40C overnight. Thesample was then dialysed against several changes of0.5 M-sodium acetate, pH 7.0.
ChromatographyGel chromatography of proteoglycans, polysac-
charides and oligosaccharides was performed onSepharose CL4B, Sephadex G-200 or G-50 in0.5 M-sodium acetate, pH 7.0, or 4 M-guanidine
1981
144
Scleral proteodermatan sulphates
hydrochloride/0.05 M-Tris/HCl, pH 7.0. Samples(3-5 mg/0.5 ml) were dissolved in the elution bufferand centrifuged to remove, if present, small amountsof insoluble material before chromatography. Fordetails see the legends to the appropriate Figures.
Separation of glycopeptides (oligosaccharide-pep-tides) and peptidoglycans (glycosaminoglycan-pep-tides) was accomplished by chromatography onECTEOLA-cellulose of papain digests of the pro-teoglycans (Anseth, 1961).
Polyacrylamide-gel electrophoresis was con-ducted on 5 or 8% (w/v) gels in 0.1% sodiumdodecyl sulphate by the procedure of Neville (1971).The gels were stained with 0.25% Kenacide R andscanned with a Zeineh soft-laser densitometer.
Isolation andfractionation ofproteoglycansSclera was powdered under liquid N2 and ex-
tracted with 15 vol. of 4 M-guanidine hydrochloride/0.05M-sodium acetate, pH5.8, containing protein-ase inhibitors (0.10M-6-aminocaproic acid/0.01M-disodium EDTA/0.005M-benzamidine hydrochlor-ide) at 40C under gentle shaking for 15h. Thesuspension was centrifuged at 18OOOg for 45minand the pellet was re-extracted once. The extract wasconcentrated by ultrafiltration in an Amicon cellwith a PM 10 filter, dialysed against 7M-urea/0.05 M-Tris/HCl, pH 6.5, with inhibitors, and passedthrough a DEAE-cellulose column (Antonopoulos etal., 1974) equilibrated with the same buffer. Thecolumn was eluted with (a) the starting buffer, (b)0.15 M-NaCl in the same buffer and (c) 2M-NaCl inthe same buffer. The 2M-NaCl fraction yielded acrude proteoglycan preparation free of collagen.Further purification was achieved by density-gradient centrifugation.
acetate, pH 5.8/CsCl solution containing proteinaseinhibitors with a starting density of 1.30g/ml wasprepared and the crude proteoglycan preparation(5 mg/ml) was dissolved in this solution. Centri-fugation was performed in an MSE Superspeed 65centrifuge fitted with an 8 x 25 ml angle rotor at10°C and 34000rev./min for 48h. Tubes wereemptied (2 ml fractions) with the aid of an MSE tubepiercer. The densities of the fractions were deter-mined with a 200,u1 pipette as pycnometer. Uronicacid was estimated directly on the fractions andprotein was estimated by measuring the A280 afterdiluting 10-fold with water. Material was recoveredby dialysis against several changes of water followedby freeze-drying. Final purification and fractiona-tion of the proteoglycan was performed by gelchromatography on a column (16mm x 1300mm)of Sepharose CL4B, which was eluted with 0.5 M-sodium acetate, pH 7.0, or 4M-guanidine hydro-
chloride/0.05 M-Tris/HCl, pH 7.0, at a rate of 13 ml/h; fraction size, 3.5 ml.
Preparation ofthe protein coreIsolation of the protein core of the proteoglycans
was accomplished by digestion with chondroitin-ase-ABC followed by gel chromatography (seeabove). Keiser & Hatcher (1977) claim that com-mercial chondroitinase-ABC preparations containsmall amounts of proteinases. Therefore, chon-droitinase-ABC digestions of proteoglycans werecarried out on a time scale of 3, 6, 12, 24 and 48h.In addition, digestions were performed in the pres-ence of 2 mM-di-isopropyl fluorophosphate.
ResultsIsolation and characterization of scleral proteo-glycans
Analyses of tissue extracts and residues byECTEOLA-cellulose chromatography of papaindigests (Anseth, 1961) indicated that at least 85% ofthe tissue glycosaminoglycans were extracted by4M-guanidine hydrochloride. The crude proteogly-can preparation (2M-NaCl fraction from DEAE-cellulose chromatography) was obtained in a yield ofapprox. 80%. This fraction was further purifiedby density-gradient centrifugation in 4M-guanidinehydrochloride/CsCl. Fractions 1-5 from the bottomof the tube (densities>1.30g/ml) corresponded to90% of the hexuronic acid content of the materialsubjected to centrifugation. The purified proteo-glycan contained 16% hexuronic acid and 50%protein. No hydroxyproline was detected. Theoverall yield after these steps was approx. 60%.
The purified proteoglycan was characterized bygel chromatography on Sepharose 2B and 4B (Figs.la and lb). Two proteoglycan components wereobserved, the larger of which was included onSepharose 2B (Fig. la) but partially excluded fromSepharose 4B (Fig. lb). To demonstrate the proteo-glycan nature of these components, chromato-graphy was performed after release of the glycanside chains by alkali (Fig. lc) or by papain digestion(Fig. ld). The polysaccharides were eluted in a moreretarded position than the proteoglycans. The resultssuggest that the smallest proteoglycans (V, = 40-55 ml in Fig. lb) contained very few glycan sidechains.The carbohydrate portion of the proteoglycan
consisted of 87% glycosaminoglycan and 13% smallneutral oligosaccharides (ECTEOLA-cellulose chro-matography of papain digests and expressed ashexosamine). In the glycosaminoglycan portion thehexosamines were exclusively galactosamine. Thusno heparan sulphate could be present. To ascertainthe uronic acid composition the galactosamino-
Vol. 193
145
L. Coster and L.-A. Fransson
A o z x
0.4
0.2-
(d)
0.4
0.2
30 40 50 60 70 80Effluent volume (ml)
Fig. 1. Gel chromatography of purified proteoglycan (aand b), alkali-treated proteoglycan (c) andpapain-treated
proteoglycan (d)The proteoglycan was purified by DEAE-cellulosechromatography and density-gradient centrifuga-tion (bottom fractions 1-5). Alkaline eliminationand papain digestion were performed as describedin the Experimental section. The column(9mm x 1300mm) contained Sepharose 2B (a) or4B (b-d), with 0.5M-sodium acetate, pH7.0, aseluent at a rate of 4 ml/h. , Uronic acid(carbazole); - , protein (Folin). A minor por-tion of the alkali-treated material (c) had the elutionvolume of proteoglycans, although the proteincontent of this material was very low. We have noadequate explanation for this result.
glycan side chains were treated with periodate atpH3.0 and 4°C (selective oxidation of L-iduronicacid) followed by scission in alkali (Fig. 2). As mostof the material was eluted in a more retardedposition compared with the untreated chains it maybe concluded that all side chains contained L-iduronicacid and the original proteoglycan may be classified
Effluent volume (ml)Fig. 2. Gel chromatography on Sephadex G-200 ofpapain-digested proteoglycans before (a) and after (b)
periodate oxidation-alkaline eliminationThe purified proteoglycan was digested with papain(see the Experimental section) and chromato-graphed (a). The glycosaminoglycan-peptides weresubsequently oxidized with periodate (L-iduronicacid residues), cleaved in alkali and chromato-graphed (b). Column size, 9mm x 1400mm; eluent,0.5M-sodium acetate, pH 5.0; rate, 5 ml/h; analysis,uronic acid.
as a dermatan sulphate proteoglycan. This wasconfirmed by chondroitinase-ABC digestion of theside chains. They were completely degraded todisaccharides (see below).
Subfractionation ofthe proteoglycanAs indicated above (Fig. 1) the purified scleral
proteoglycan was heterogeneous in size and twosubpopulations could be discerned. On chromato-graphy on Sepharose CL4B in 0.5M-sodium acetate(Fig. 3a) the larger proteoglycan subpopulation wasalmost entirely excluded (proteoglycan I). However,on gel chromatography in 4 M-guanidine hydro-chloride (Fig. 3b) proteoglycan I was partiallyincluded in the gel, whereas the position of proteo-glycan II was unchanged. The two proteoglycansubfractions were isolated by preparative gel chro-matography (see the Experimental section) either in0.5 M-sodium acetate (Fig. 3a) or in 4M-guanidinehydrochloride (Fig. 3b). In both cases the yieldswere 40 and 60% respectively, on a w/w basis. The
1981
146
Scleral proteodermatan sulphates
A
Effluent volume (ml)
Fig. 3. Gel chromatography on Sepharose CL 4B ofpurified proteoglycans in (a), 0.5M-sodium acetate,pH- 7.0, (b) 4 M-guanidine hydrochloridelO.05 M-Tris/HCI,pH7.0 and .(c) reduced and alkylated proteoglycan in
0.5M-sodium acetate,pH 7.0Reduction and alkylation was performed as de-scribed in the Experimental section. Samples weredissolved in 4 M-guanidine hydrochloride anddialysed against the elution buffer before chroma-tography. For column size and technical details,see the legend to Fig. 1. , Uronic acid(automated carbazole in a and c, manual carbazolein b); ---, protein (Folin in a and c, A280 in b). The
Vol. 193
chemical composition was also the same whether theseparation was carried out in acetate or guanidine(see below). Rechromatography of proteoglycan Iisolated as in Fig. 3(a) on the same column elutedwith 4M-guanidine hydrochloride produced theprofile of a partially included proteoglycan (resultsnot shown). Conversely, rechromatography ofproteo-glycan I isolated as shown in Fig. 3(b) on thesame column but in acetate resulted in exclusion ofthe material (as shown in Fig. 3a). Proteoglycansubfraction II was eluted in the same position in bothsolvents. After reduction and alkylation of theproteoglycans, fraction I was partially included inthe gel (Fig. 3c) when chromatographed in 0.5 M-sodium acetate. The profile was not altered bychromatography in 4 M-guanidine hydrochloride(results not shown). The elution position of proteo-glycan II was unaffected by reduction and alkyla-tion.
Ultracentrifugation oftheproteoglycansSolutions of the two proteoglycan subfractions in
6M-guanidine hydrochloride were studied by ultra-centrifugation in the sedimentation-velocity mode.Both proteoglycans I and II showed the pattern of apolydisperse material, but neither were resolved intomore components. The sedimentation coefficientsobtained (2.8 x 10-13 and 2.2 x 1033s) suggest thatboth proteoglycans are small.
Chemical characterization of the proteoglycanspecies
Both the glycan side chains and the protein coreof the two proteoglycan species were subjected tocompositional analyses. As shown in Table 1proteoglycan I had a higher glycan/protein ratiothan did proteoglycan II. There were no significantdifferences in chemical composition between thepreparations obtained under associative or disso-ciative conditions (Figs. 3a and 3b). Proteoglycan Icontained a slightly smaller proportion of neutral
carbazole reactivity of fraction II was higher in themanual procedure (b) than in the automated pro-cedure (a). Although chondroitin sulphate gives thesame colour yield in both procedures (Heinegard,1973), dermatan sulphates with L-iduronic acid con-tent should give lower colour yield in the automatedprocedure as the boiling time is shorter (7-8minversus 20min). As shown in Table 1 proteoglycan IIcontains two to three times more L-iduronic acidthan proteoglycan I. Fractions (I and II in a and b)were pooled as indicated by the vertical lines andmaterial was recovered by freeze-drying andweighed. The weight ratio of the proteoglycansubfractions was approx. 2:3, irrespective ofwhether the separation was performed in acetate (a)or guanidine (b).
147
L. Coster and L.-A. Fransson
Table 1. Analyses ofproteoglycans I and IIThe purified proteoglycan was subfractionationed on Sepharose CL 4B (Fig. 3) either in 0.5M-sodium acetate,pH 7.0, or in 4M-guanidine hydrochloride, pH 7.0. In both cases, the yields of proteoglycans I and II were 40 and60% respectively on a w/w basis. The fractions were digested with papain and the liberated carbohydrate pros-thetic groups were separated Into glycopeptides and glycosaminoglycans (expressed as the percentage of totalhexosamine). The uronic acid composition (expressed as the percentage of total uronic acid) was determined afterchondroitinase-AC (D-glucuronic acid) and chondroitinase-ABC digestions (L-iduronic acid + D-glucuronic acid)followed by quantification of unsaturated glycuronosyl residues. L-Iduronic acid content = total uronic acid -D-glucuronic acid. Protein (Folin), uronic acid (carbazole) and hexosamine are expressed as the percentage ofdry weight.
Table 2. Analyses ofoligosaccharide-peptides derivedfrom the proteoglycanThe proteoglycan subfractions were degraded by proteolysis and oligosaccharides were isolated by ion-exchangechromatography. Neutral sugars were determined by g.l.c. The amounts of neutral sugars and hexosamines inthe oligosaccharides are expressed as the percentage of the dry weight of the proteoglycan. Individual sugars areexpressed as the percentage of neutral sugars or hexosamines respectively.
AnalysesNeutral sugars
L-FucoseD-XyloseD-MannoseD-Galactose
HexosaminesD-GlucosammeD-Galactosamine
oligosaccharide side chains than did proteoglycan II(Tables 1 and 2). Both oligosaccharide-peptidescontained large amounts of D-mannose and D-glucosamine. The composition of neutral sugars wasquite similar. However, oligosaccharide I had asignificantly higher D-galactosamine content thandid oligosaccharide II.
The glycan side chains of the individual proteo-glycans were chromatographed on Sephadex G-200as in Fig. 2a (results not shown). Both preparationsshowed a similar degree of polydispersity. Thedermatan sulphate chains of proteoglycan I wereD-glucuronic acid-rich, whereas in proteoglycan IIthe side chains contained almost equal proportionsof the two uronic acids (Table 1).
Content (%)
Oligosaccharides from proteoglycan ... I II1.7 2.79.5 8.14.8 3.3
51.8 52.533.9 36.02.0 2.1
57 8443 16
Selective periodate oxidation of L-iduronic acidresidues in the glycan side chains followed byscission in alkali and gel chromatography affordedthe elution profile shown in Fig. 4. In accordancewith the high content of D-glucuronic acid, chainsfrom proteoglycan I were moderately degraded (Fig.4a). The degradation products that were excludedfrom the gel (fraction 1) were further degraded withchondroitinase AC (Fig. 4b). The vast majority ofthe material was degraded to disaccharides. Theperiodate-degradation products that were partiallyincluded in the gel (fraction 2 in Fig. 4a) yieldedboth disaccharides and oligosaccharides on chon-droitinase-AC digestion (Fig. 4c). These oligosac-charides represent sequences where the L-iduronic
1981
148
Scleral proteodermatan sulphates
Effluent volume (ml)Fig. 4. Gel chromatography on Sephadex G-50 ofdegradation products derivedfrom the dermatan sulphate side chains
of(a-c) proteoglycan I and (d-f) proteoglycan IIProteoglycan subfractions I and II (Fig. 3a) were digested with papain (see the Experimental section) and theliberated side chains (isolated by ECTEOLA-cellulose chromatography) were subsequently degraded by periodateoxidation-alkaline elimination (a and d). The excluded and the partially included degradation products (fractions 1and 2 respectively) were separately digested with chondroitinase-AC and re-chromatographed on the same column(the results for fraction 1 are shown in b and e and those for fraction 2 in c andf). Column size 8mm x 1500mm;eluent, 0.2 M-pyridine/acetate, pH 5.0; rate, 6 ml/h; analysis, carbazole.
acid residues are 0-sulphated (Coster et al., 1975).Thus the side chains of proteoglycan I are charac-terized by the presence of large block regionscomposed of D-glucuronic acid-containing repeats.Periodate oxidation-alkaline degradation of the sidechains of proteoglycan II produced largely oligo-saccharide fragments (Fig. 4d). The fragmentseluted with the void volume (fraction 1) contained anappreciable quantity of chondroitinase-AC-resistantmaterial (Fig. 4e). The main oligosaccharide frac-tion obtained after periodate oxidation (fraction 2 in
Fig. 4d) was further degraded by chondroitin-ase-AC (Fig. 4f) into disaccharides and higheroligosaccharides. The latter saccharides containO-sulphated L-iduronic acid residues. A comparisonbetween the two sets of results indicates thatsegments containing L-iduronic acid 0-sulphate weremore frequent in the dermatan sulphate side chainsof proteoglycan II than in those of proteoglycan I. Inthe present scheme of degradation regions com-posed of alternating short segments of L-iduronicacid- and D-glucuronic acid-containing repeats
Vol. 193
149
L. Coster and L.-A. Fransson
would yield short oligosaccharides of the generalcarbohydrate structure N-acetylgalactosamine-(D-glucuronic acid-N-acetylgalactosamine).-R, where Ris the remnant of an oxidized and degradedL-iduronic acid residue. Such oligosaccharidesshould be present in fraction 2 in Figs. 4(a) and 4(d).Although the side chains of proteoglycan II affordedlarge amounts of saccharides containing periodate-resistant uronic acids (Fig. 4d) a considerable pro-portion of these saccharides contained bothL-iduronic acid 0-sulphate and D-glucuronic acid(Fig. 4f). The corresponding saccharide fractionfrom proteoglycan I (fraction 2 in Fig. 4a) con-tained largely D-glucuronic acid residues (Fig. 4c).
The amino-acid compositions of the parent scleralproteoglycan, the subfractions I and II and proteo-glycans prepared from skin, cornea and cartilage byother workers are given in Table 3. In proteoglycan Ithe serine, threonine, glutamate, proline, alanine andglycine contents were higher, whereas the aspartate,isoleucine, leucine and lysine contents were lowerthan in proteoglycan II; methionine was only foundin the latter species, whereas cysteine was present inboth.
The protein cores of the two subpopulations wereprepared by chondroitinase-ABC digestion followedby preparative gel chromatography. As shown inFig. 5 the protein core of proteoglycan I (Fig. 5a)was of a larger size than that of proteoglycan II (Fig.
5b). On polyacrylamide-gel electrophoresis intactproteoglycan II had a polydisperse appearance withan apparent average molecular weight of 85 000(results not shown). After chondroitinase-ABCdigestion one distinct band was observed with anapparent molecular weight of 46000. The resultswere the same whether the chondroitinase-ABCdigestion was carried out for 3 h or 48 h or ifdi-isopropyl fluorophosphate was present in thedigestion mixture. Thus our preparation of chon-droitinase-ABC did not seem to contain anyproteinases active on the core protein. Identificationof the core protein of proteoglycan I after gelelectrophoresis was difficult owing to low affinity forthe staining reagent. When large amounts of materialwere applied, core preparation I did not seem topenetrate into a 5% gel, suggesting a molecularweight >100000.
Discussion
Dermatan sulphate-containing proteoglycanshave been isolated from heart valves, skin, tendon,sclera and arterial wall. The proteoglycans of thearterial wall are generally of higher buoyant densityand higher carbohydrate content than those of theother tissues. Furthermore, the arterial wall (R.Kapoor, C. F. Phelps, L. Coster & L.-A. Fransson,unpublished work) and skin (Damle et al., 1979)
Table 3. Amino-acid composition ofscleralproteoglycansValues are expressed as residues per 1000 residues.
Unfractionatedproteoglycan Proteoglycan I Proteoglycan II
Effluent volume (ml)Fig. 5. Gel chromatography on Sepharose 4B of chon-droitinase-ABC digests of (a) proteoglycan I and (b)
proteoglycan IIProteoglycan subfractions I and II (Fig. 3a) were
digested with chondroitinase-ABC and the proteincore and liberated disaccharides were separated onthe same column as in Fig. 1; eluent, 0.5 M-sodiumacetate, pH 7.0; , uronic acid (carbazole);---~-, protein (Folin). The protein cores (V, =40-60ml and 60-75 ml respectively) were isolatedby dialysis and freeze-drying. The positions of theintact proteoglycan subfractions are indicated in therespective panels. The anthrone-positive compon-ents of proteoglycans I and II were quantitativelyrecovered in the respective core fractions.
contain separate proteochondroitin sulphate andproteodermatan sulphate species. The proteogly-cans of sclera, heart valves, tendon *and skinresemble each other with respect to protein contentand amino-acid composition. Attempts to determinethe molecular weight of proteodermatan sulphateshave yielded values ranging from approx. 105 (Toole& Lowther, 1968; Anderson, 1975; Damle et al.,1979) to 3 x 106 (6brink, 1972).
In the present investigation scleral proteoglycans(all of which are proteodermatan sulphates) were
separated into two populations, one larger in size (I;so = 2.8 x 10-13s) and one smaller (II; s =2.2 x 1013 s). The yield of the two components mayvary if proteinase inhibitors are omitted from thetissue extractant and in some cases even free glycanchains may be obtained. However, in the presenceof proteinase inhibitors the results were highly re-producible. Scleral proteoglycans are comparablewith corneal proteoglycans in size (1.7S; Mw=72000; Axelsson & Heinegird, 1978) whereas cartil-
Vol. 193
age proteoglycans are considerably larger (25 S; Mw= 2.5 x 106; Hascall & Sajdera, 1970). The 2.1 Sscleral proteoglycan (II) contained 60% protein andthe apparent molecular weight of the core proteinwas 46000. Thus a molecular weight of 80000 maybe calculated for the intact proteoglycan II. This isto be compared with a value of 85000 obtained forthe intact material by gel electrophoresis. It was notpossible to obtain a molecular-weight estimate forthe 2.8 S proteoglycan (I) in gel electrophoresis.Results of gel chromatography and ultracentri-fugation of proteoglycan I show that it is larger thanproteoglycan II, but still considerably smaller thancartilage proteoglycan (cf. S values above; see alsoFigs. 1 and 5).
Both proteoglycans from sclera seemed to carrytwo types of carbohydrate side groups, i.e. man-nose- and glucosamine-rich oligosaccharides andgalactosaminoglycans. These side groups must belocated at the s'ame protein core, as digestion of theglycans by chrondroitinase-ABC alters the elutionposition in gel chromatography (Fig. 5) and themobility in sodium dodecyl sulphate/polyacryl-amide-gel electrophoresis of the entire material,including the anthrone-positive components. Fur-thermore, only one component could be detectedon electrophoresis.The two scleral proteoglycans are both exclu-
sively proteodermatan sulphates. However, the twosubfractions differed with respect to the co-poly-meric structure of their dermatan sulphate sidechains. The results obtained here are slightly atvariance with those reported by Antonopoulos et al.(1974). We obtain two proteoglycan subpopula-tions instead of four, our fractions contain moreprotein and we do not find any chondroitin sulphateproper in the material. This can be ascribed todifferences in the methods ofextraction and analyses.In the present study the use of proteinase inhibitorsminimize degradation and therefore decreases theheterogeneity and increases the protein content ofthe preparation. In the previous study (Antono-poulos et al., 1974) the method used to quantifychondroitin sulphate and dermatan sulphate did notdistinguish between D-glucuronic acid-rich co-poly-meric dermatan sulphate and chondroitin sulphateproper (Fransson et al., 1970). The dermatansulphate side chains of proteoglycan I were largelycomposed of D-glucuronic acid-N-acetylgalactos-amine sulphate repeats. The L-iduronic acid con-taining units were few (20%), and occasionallyarranged in short sequences intercalated betweenshort sequences containing D-glucuronic acid. Al-though the dermatan sulphate side chains of proteo-glycan II contained a large proportion of shortsequences with D-glucuronic acid, the intercalatedL-iduronic acid-containing repeats often carried anester sulphate on the uronic acid moiety.
151
152 L. Coster and L.-A. Fransson
The two proteodermatan sulphates described inthe present report contained markedly differentprotein cores, as indicated by distinctly differentamino-acid composition and molecular size (Fig. 5).Proteoglycan I, which carried more side chains thandid proteoglycan II (60 and 45% carbohydraterespectively) had a higher serine content. As shownpreviously dermatan sulphate chains are linked toprotein via the sequence D-glucuronic acid-D-galac-tose-D-galactose-D-xylose-serine (Fransson, 1968;Stern et al., 1971). In general, proteoglycan I, whichhad more chondroitin sulphate-like side chains, hadan amino-acid composition similar to the chon-droitin sulphate-rich proteoglycans from skin(Damle et al., 1979) and cartilage (Heinegird,1972). In contrast, proteoglycan II, which had ahigher protein content than did proteoglycan I, wasmore akin to the protein-rich comeal proteokeratansulphates described by Axelsson & Heinegird(1975).As proteoglycan I altered its elution position on
gel chromatography in 4M-guanidine hydrochloridecompared with 0.5 M-sodium acetate it is proposedthat a change in conformation or a disaggregationtook place. We favour the latter hypothesis, asproteoglycan I appears smaller in 4.0M-guanidineand after reduction and alkylation. A conforma-tional change induced by guanidine or by thecleavage of disulphide bonds might rather expandthe molecule. Proteoglycan II showed no aggrega-tion phenomenon under the present conditions. Theaggregation seen with proteoglycan I must be due toself-association or mediated by an extrinsic mole-cule. Self association may be mediated via the sidechains and/or protein core in three different ways,i.e. carbohydrate-carbohydrate, carbohydrate-pro-tein and protein-protein. One extrinsic moleculeknown to multimerize proteoglycans is hyaluronate(see Heinegard & Hascall, 1974). The finding thatreduction and alkylation abolished aggregationmight suggest that the protein core was involved.The same treatment of cartilage proteoglycanabolishes the interaction with hyaluronate (Harding-ham & Muir, 1974; Heinegird & Hascall, 1974).The scleral proteoglycan was of rather low buoyantdensity. Therefore it cannot be excluded thatendogenous hyaluronic acid was present in thepurified material. As the proteoglycans were puri-fied by ion-exchange chromatography and den-sity-gradient centrifugation in a dissociative mediumthe presence of an aggregating protein seems ratherunlikely.
Self association of proteoglycan I could be medi-ated by the dermatan sulphate side chains. Freedermatan sulphate chains from various sourcesself-associate (Fransson, 1976). These species ofdermatan sulphate possess alternating or mixedsequences of L-iduronic acid- and D-glucuronic
acid-containing repeats. Non-aggregating chainslack this feature and contain considerable amountsof O-sulphated L-iduronic acid residues (Fransson &Coster, 1979). At present one cannot definitelydistinguish between the various possibilities formultimerization of proteodermatan sulphate. Anyexplanation, however, must take into account thedistinct chemical differences between scleral proteo-dermatan sulphates I and II, both with regard to thenature of the protein core and the co-polymericfeatures of their side chains.
Mrs. Lena Aberg provided skilful technical assistance.Dr. A. Malmstrom and Dr. I Carlstedt are thanked formany stimulating discussions. We received grants for thiswork from the Swedish Medical Research Council (567),'Greta och Johan Kocks Stiftelser', 'Gustaf V:s 80-irsfond' and the Medical Faculty and the Royal Physio-graphic Society, University of Lund.
References
Anderson, J. C. (1975) Biochim. Biophys. Acta 379,444-455
Antonopoulos, C. A., Gardell, S., Szirmai, J. A. &DeTyssonsk, E. R. (1964) Biochim. Biophys. Acta 83,1-14
Antonopoulos, C. A., Axelsson, I., Heinegard, D. &Gardell, S. (1974) Biochim. Biophys. Acta 338,108-119
Anseth, A. (1961) Exp. Eye Res. 1, 106-112Axelsson, I. & Heinegird, D. (1975) Biochem. J. 145,
491-500Axelsson, I. & Heinegard, D. (1978) Biochem. J. 169,
517-530Bitter, T. & Muir, H. (1962) Anal. Biochem. 4, 330-334Coster, L. & Fransson, L.-A. (1975) Ital. J. Biochem. 24,
vxIIC6ster, L. & Fransson, L.-A. (1976) Arch. Int. Physiol.
Biochim. 84, 14Coster, L., Malmstrom, A., Sjoberg, I. & Fransson, L.-A.
(1975) Biochem. J. 145, 379-389Damle, S. P., Kieras, F. J., Tzeng, W.-K. & Gregory,
J. D. (1979) J. Biol. Chem. 254, 1614-1620Ehrlich, K. C., Radhakrishnamurthy, B. & Berenson,
G. S. (1975)Arch. Biochem. Biophys. 171, 361-369Eisenstein, R., Larsson, S.-E., Kuettner, K. E., Sorgente,
N. & Hascall, V. C. (1975)Atherosclerosis 22, 1-17Fransson, L.-A. (1968) Biochim. Biophys. Acta 156,
106-1 15Fransson, L.-A. & Carlstedt, I. (1974) Carbohydr. Res.
36, 349-358Fransson, L.-A. & Coster, L. (1979) Biochim. Biophys.
Acta 582, 134-144Fransson, L.-A., Anseth, A., Antonopoulos, C. A. &
Gardell, S. (1970) Carbohydr. Res. 15, 73-89Goa, J. (1955) Scand. J. Clin. Lab. Invest. Suppl. 22,
7-45
1981
Scleral proteodermatan sulphates 153
Hardingham, T. E. & Muir, H. (1974) Biochem. J. 139,565-581
Hascall, V. C. & Sajdera, S. W. (1970) J. Biol. Chem.245,4920-4930
Hascall, V. C., Riolo, R. L., Hayward, J. & Reynold,C. C. (1972) J. Biol. Chem. 247. 4521-4528
Heinegard, D. (1972) Biochim. Biophys. Acta 285,181-192
Heinegird, D. (1973) Chem. Scr. 4, 199-201Heinegard, D. & Hascall, V. C. (1974) J. Biol. Chem.
249,4250-4256Keiser, H. D. & Hatcher, V. B. (1977) Connect. Tissue
Res. 5, 147-155Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall,
R. J. (195 1) J. Biol. Chem. 193, 265-275Moore, S. (1963) J. Biol. Chem. 238, 235-237
Neville, D. (197 1) J. Biol. Chem. 246, 6328-6334Cbrink, B. (1972) Biochim. Biophys. Acta 264, 354-361Oegema, T. R., Hascall, V. C. & Dziewiatkowski, D. D.
(1975) J. Biol. Chem. 250, 6151-6159Radhakrishnamurthy, B., Ruiz, H. A. & Berenson, G. S.
(1977) J. Biol. Chem. 252,4831-4841Schachman, H. K. (1957) Methods Enzymol. IV, 32-
103Stegemann, H. & Stalder, K. (1967) Clin. Chim. Acta 18,
267-273Stern, E. L., Lindahl, B. & Roden, L. (1971) J. Biol.
Chem. 246, 5707-5715Toole, B. P. & Lowther, D. A. (1965) Biochim. Biophys.
Acta 101, 364-366Toole, B. P. & Lowther, D. A. (1968) Biochim. Biophys.
