Biochem. J. (1971) 125, 37-46 37Printed in Great Britain

Comparative Biochemistry of Chondroitin Sulphate-Proteins ofCartilage and Notochord

By MARTIN B. MATHEWSDepartments of Pediatrics and Biochemistry, The La Rabida-University ofChiWago InstituWe, and theJoseph P.Kennedy Jr. Mental Retardation Research Center, The University of Chicago, Chicago, Ill. 60637, U.S.A.

(Received 27 April 1971)

Fragments that consisted mainly of two polysaccharide chains joined by a shortpolypeptide bridge (doublets) were prepared from chondroitin sulphate-proteins oflamprey, sturgeon, elasmobranch and ox connective tissues after hydrolysis withtrypsin and chymotrypsin. Consideration of molecular parameters, compositionsand behaviour on gel electrophoresis and density-gradient fractionation leads to aproposed parent structure for chondroitin sulphate-proteins. A single polypeptidechain of about 2000 amino acid residues contains alternating short and long repeat-ing sequences. Ashort sequence consists ofless than 10 amino acid residues with oneN-terminal and one C-terminal serine residue, each ofwhich carries a polysaccharidechain linked glycosidically to its hydroxyl group. This structure constitutes thedoublet subunit. Some variation is introduced when the doublet subunit carriesonly a single polysaccharide chain. The long sequence contains about 35 amino acidresidues and is subject to cleavage by trypsin and chymotrypsin. The main poly-peptide is probably homologous in the vertebrate sub-phylum with strong con-servation of structure suggested for the short sequence. However, polymorphismof polypeptide structures cannot be excluded.

Mathews & Lozaityte (1958) proposed a comb-like model for the chondroitin sulphate-proteinmacromolecules of bovine hyaline cartilage. Thebasic unit consisted ofan extended protein backboneto which side chains of acidic polysaccharide werecovalently bound at intervals to form a macro-molecule with a molecular weight of about 4 x 106.It was subsequently shown that each linear poly-saccharide chain was linked glycosidically, via alinkage region containing glucuronic acid, galactoseand xylose, to the hydroxyl group ofa serine residue(Rod6n, 1968). Electron-microscopic studies sup-port a branched structure for bovine chondroitinsulphate-protein (Serafini-Fracassini & Smith,1966; Rosenberg, Hellmann & Kleinschmidt, 1970).To account for the observation (Mathews, 1956)

that chondroitin sulphate prepared by alkalineextraction ofcartilage had a molecular weight aboutone-half of that ofpreparations obtained by trypsindigestionofchondroitinsulphate-protein, Anderson,Hoffman & Meyer (1965) suggested that two 'single'chains of chondroitin sulphate were joined by ashort peptide sequence as 'doublets' containing analkali-labile bond. Luscombe & Phelps (1967)found that the weight-average molecular weight(43 300) ofthe product obtained by trypsin digestionwas decreased to 21000 by the action of papain oralkali, and thus supported this proposal.

The present paper deals with the composition andstructure of doublets obtained by the action of bothtrypsin and chymotrypsin. Number-average mole-cular weights (M.) were obtained so that amino acidcontents could be related reliably to numbers ofmolecular chains under conditions of some poly-dispersity of polysaccharide chain weight. Sincechondroitin sulphate-protein molecules in carti-lage of many vertebrates are very similar inchemical composition, and in molecular size para-meters (Mathews, 1967), it was decided to investi-gate the extent of homology at the doublet levelof structure.

MATERIALS AND METHODS

Isolation of chondroitin 8ulphate-protein. The pre-parations of chondroitin sulphate-protein from ox nasalseptal cartilage (SCSP4E) and from ray cartilage (W-2B)were described previously (Mathews & Lozaityte, 1958;Mathews, 1962). The physical constants for the prepara-tions of chondroitin sulphate-protein from cartilage andnotochord oflamprey and ofsturgeon have been published(Mathews, 1967). However, the procedure for isolationfrom tissue of the latter preparations was sufficientlymodified from the procedure employed earlier to merit abrief description.About lOg of tissue was homogenized with a Virtis

blender in 200 ml of water, kept cold by an ice bath. Four

M. B. MATHEWSperiods of blending at high speed of 5min each werealternated with periods at low speed of 30min each. Thehomogenate was diluted with 1.8 litres of 0.11m-NaCl,heated within 10min to 75-80°C and, after 1 min, was cooledto 45°C. The mixture was filtered with suction through apad of Celite filter powder and aq. 4% (w/v) cetylpyridi-nium chloride was added to the filtrate to obtain completeprecipitation. After 1 h at 300C, the suspension wascentrifuged at 2000g. The precipitate was stirred with300ml of 0.1M-NaCl containing 0.1% cetylpyridiniumchloride and recovered by centrifugation. To obtain thechondroitin sulphate-protein as the sodium salt, theprecipitate was stirred at 40-50°C with a mixture of 160 mlof 2.5 m-NaCl and 80 ml of methanol for several hours andthe resulting solution clarified by centrifugation at roomtemperature. The supernatant was diluted with 400 ml of60% (v/v) ethanol and added with stirring to 1 litre of90%ethanol. After setting overnight, the precipitate wasrecovered by centrifugation, washed successively withethanol and ether and dried over P205 in a vacuum. Asmall content of hydroxyproline (less than 0.1%) wasfurther decreased by the following steps. The precipitate,dissolved in 500ml of 0.4M-NaCl, was filtered through aCelite pad, diluted to 2 litres with water, re-precipitatedwith cetylpyridinium chloride and converted into thesodium salt as described above.Enzyme treatment8. A sample of 300mg of chondroitin

sulphate-protein was dissolved in 60ml of 0.05M-sodiumphosphate buffer, pH 7.6, and placed in a Ubbelohdesuspended-level viscometer in a water bath at 25.0000.Three drops of toluene and 1.5mg of crystalline trypsin(Calbiochem, Los Angeles, Calif., U.S.A.) were added.The viscosity of the solution decreased rapidly andreached a limiting value in 24h. On addition of 1.5mg ofcrystalline ohymotrypsin (Calbiochem) a further fall inviscosity occurred for about 1 h. Subsequent changes invisoosity were small and became negligible after 48h.After dialysis of the solution against 0.1 m-NaCl andprecipitation with cetylpyridinium chloride, the productwas recovered as a dry sodium salt. Further treatmentas described above with 1.5mg of chymotrypsin for 24hproduced changes in reduced viscosity of less than 3%.The recovered product in aqueous solution was thenadsorbed on to a column of 250 ml of Dowex 1 resin (X2;200-400 mesh; Cl- form) and the resin eluted successivelywith 400-500 ml batches of water, 1.0M-NaCl, 1.5M-NaCl,1.7M-NaCl and 5.0M-NaCl. The bulk of the product,recovered as dry sodium salt, was eluted by 1.5M- and1.7M-NaCl. These two fractions represented a yield of70-90% of the hexosamine of chondroitin sulphate-protein, and were substantially free of keratan sulphate(Rosenblum & Cifonelli, 1967). The 1.7 M-NaCl fractionsfrom preparations of ox, sturgeon and ray cartilage (60-70% yield) were used further. The 1.5m- and 1.7M-NaClfractions from preparations of lamprey tissues and ofsturgeon notochord were combined in each case.

Hydrolyses with crystalline papain (Nutritional Bio-chemicals Corp., Cleveland, Ohio, U.S.A.) were carriedout at 60°C for 1Bh with concentrations of 1mg oftryptic-chymotryptic hydrolysate product and 0.1 mg of enzyme/ml of 0.1M-sodium phosphate buffer, pH6.5, containing0.01M-cysteine and 0.01M-EDTA. The hydrolysate wasdialysed against water, clarified by filtration, precipitated

with cetylpyridinium chloride and the product finallyrecovered as the sodium salt by alcohol precipitation.

Hydrolyses with Pronase (B grade; Calbiochem) werecarried out at 60°C for 2 days with concentrations of 2 mgof tryptic-chymotryptic hydrolysate product and 0.07 mgof enzyme/ml of 0.1M-sodium phosphate buffer, pH7.6.The principal product was recovered as the sodium salt(see above).

Alkali treatment. A solution of 50mg of tryptic-chymotryptic hydrolysate product in 0.1 M-NaCl wasmade 0.0BM in NaOH by addition of 2.5M-NaOH and keptat 25°C for 24h. The solution was neutralized with 1M-acetic acid and the product precipitated with cetylpyridi-nium chloride and converted into the sodium salt. Duringthe period of exposure to alkali, the reduced viscositydecreased rapidly, at about an exponential rate, to reacha limiting value in about 6 h, with about 30min requiredto produce one-half of the final decrease in viscosity.Although similar limiting values ofreduced viscosity werefound after 24h exposure to 0.01 M-NaOH, the rate of fallof viscosity was about one-sixth of that in 0.05M-NaOH.No appreciable changes of solution viscosity were foundafter 24h in the presence of 0.02M-Na2CO3 (pH approx.10.5).

Analy8es. The hexosamine content of preparationswas determined after hydrolysis of samples in 4M-HCI for14h at 100C by a modification (Boas, 1953) of themethod of Elson & Morgan (1933) and corrected for lossesby comparison with a standard sample run concurrently.An 'anhydrous' content of polysaccharide was thencalculated from the hexosamine content and the structuralformula for the disaccharide repeating unit (Mathews &Dorfman, 1953). Amino acids, galactosamine and glucos-amine were determined with the Technicon Amino AcidAnalyzer after hydrolysis of samples for 24h at 100°C in6m-HCl. Control experiments with known mixtures ofamino acids and chondroitin sulphate indicated that lossesof individual amino acids were less than 5%. Hence theamino acid results were not corrected for losses. However,since amino sugar losses during hydrolysis were about 30%,corrections were made on the basis of the previouslydetermined values of hexosamine content. Values ofintrinsic viscosity and Mn, obtained by previouslydescribed procedures (Mathews & Dorfman, 1953), werebased also on the 'anhydrous' polysaccharide content ofpreparations. Tryptophan was not determined, butappears to be present only in very small amounts in bovinechondroitin sulphate-protein (Mathew s& Lozaityte,1958).

Cystine was determined as cysteic acid after performicacid oxidation of samples by the method of Hirs (1967).Crystalline bovine albumin was used as a referencestandard in each analysis in the presence of 10 times itsweight of chondroitin sulphate to approximate theproportion of protein and polysaccharide in chondroitinsulphate-protein. After the oxidation step, the sampleswere hydrolysed for 24h at 100(C in 6M-HCI and evapor-ated repeatedly to remove HCI. The amino acid residueswere then spotted on Whatman 3MM paper together withstandards of cysteic acid; and electrophoresis was per-formed at 10V/cm in a buffer that was prepared by mixing6ml of pyridine, 20ml of acetic acid and 1 litre of waterand adjusting to pH 3.9. After the paper was stained with

38 1971

STRUCTURE OF CHONDROITIN SULPHATE-PROTEINSninhydrin, the densities of the cysteic acid spots weremeasured with a Photovolt model 525 densitometer. Thealbumin reference standard gave values for cystine within10% of theoretical.

Solutions of chondroitin sulphate-protein in 4.0M-guanidinium chloride-.05m-acetate buffer, pH5.8, andCsCl, of density 1.50g/ml, were fractionated by thedensity-gradient procedure of Sajdera & Hascall (1969).A Spinco model L2-65 ultracentrifuge with a no. 65 headwas operated for 48 h at 20°C and an average force of75000g.

Electrophoresis. The disc-electrophoresis apparatus ofthe Buchler Instrument Co. was employed, with poly-acrylamide gels made with different concentrations ofmonomer in sodium phosphate buffer, pH7.5, I 0.13. Theacidic polysaccharide bands were stained with AlcianBlue. The use of gels of different pore size permittedresolutions of homologous polymers based on molecularsize (Mathews & Decker, 1971).

RESULTS

Homogeneity. Since preparations of chondroitinsulphate-protein may contain degradation productsderived by the action of native proteolytic enzymes(Ali, 1964), they were examined by gel electro-phoresis. Products with the molecular size oftryptic-chymotryptic hydrolysate products andlower-molecular-weight products were essentiallyabsent. The tryptic-chymotryptic hydrolysateproducts from ray, ox and sturgeon cartilage

contained only trace amounts of chondroitinsulphate chains of the size of those obtained bypapain hydrolysis of the respective chondroitinsulphate-protein preparations. However, the 1.5M-sodium chloride eluate fractions from the resinpreparation of the tryptic-chymotryptic hydro-lysate products from ox and sturgeon cartilagecontained appreciable proportions (near 20% byvisual estimate of stain in gel electrophoresis) ofchondroitin sulphate of lower molecular size. Simi-lar proportions of polysaccharide of low molecularweight were found in the tryptic-chymotryptichydrolysate products from lamprey tissues andfrom sturgeon notochord.

Molecular parameters. Table 1 shows the mole-cular weights and the intrinsic viscosities of theproducts of hydrolysis of the chondroitin sulphate-protein with trypsin and chymotrypsin and of theproducts after further degradation by sodiumhydroxide, papain and by Pronase. The M. valuesfor the products of sodium hydroxide and ofpapaintreatments ofthe tryptic-chymotryptic hydrolysateproducts of ox, sturgeon and ray cartilage are closeto one-half of the initial values. These results areconsistent with additional scission of only one bondper molecule of the tryptic-chymotryptic hydro-lysate products. The close agreement of M. valuesof the tryptic-ehymotryptic hydrolysate productand of the product from additional hydrolysis by

Table 1. Molecular parameters of tryptic-chymotryptic hydrolysate of chondroitin sulphate-protein andproducts after various treatments

M7, viscosity-average molecular weight, was estimated from relationship [X] - 3.1 x 10-2 MO.74 for chon-droitin sulphate A and [q] = 5.8 x 10-2 MO-74 for chondroitin sulphate C. Ox cartilage and the notochords ofsturgeon and lamprey contain mainly chondroitin sulphate A; cartilage of the sturgeon contains primarilychondroitin sulphate C; cartilage of the ray contains an over-sulphated chondroitin sulphate C (Mathews,1959, 1967).

Treatment

TissueOx cartilage

Sturgeon cartilage

Ray cartilage

Lamprey cartilage

Sturgeon notochord

Lamprey notochord

Molecularparameter

[77]Mnmv[77]M.Mu[77]Mnmv[77]MnMv[77]Mnmv[,q]Mnmv

1024300056000

2155000066000

31080000110000

387000(6600)

381000015000

451140018000

NaOH60

2100027000

1352540035000

2054010062000

295100

(4600)25

59008100

2559008300

Papain60

2180027000

1302300034000

1694020052000

285200(4400)

2371007400

2865009600

Pronase96

51000200

61000310

11000027

(4300)33

1200039

14600

Vol. 125 39

Table 2. Fractionation of chondroitin sulphate-protein in a caesium chloride density gradient

The volumes and average densities at 20°C of zones were: A, 3 ml, 1.42; B, 5 ml, 1.47; C, 4 ml, 1.54. Proteinin chondroitin sulphate-protein from each zone was calculated from amino acid analysis.

Percentage of totalZone IProtein Chodroitin sulphat

Zone Protein Chondroitin sulphateABC

ABC

ABC

ABC

ABC

ABC

102070101080101080203050404020305020

0

<5>95

0

<4>96

0

<4>96

730636

3064174736

Protein in chondroitin sulphate-proteinfrom each zone (%)

10.4

7.0

5.3714632693021585240

papain for the ox preparations with the weight-average molecular weights of 43300 and 21000,respectively for comparable preparations obtainedby Luscombe & Phelps (1967) suggests a highdegree of molecular size homogeneity. Evidently,the cleavage of the tryptic-chymotryptic productsdivides them into two macromolecular fragments ofapproximately equal size. It is likely that similarstatements apply to the homologous sturgeon andray molecules. The values of the viscosity-averagemolecular weight (Me) support these conclusions.The results for the remaining three tryptic-

chymotryptic hydrolysate products shown inTable 1 are not amenable to a simple interpretation.Although the products from notochord of sturgeonand of lamprey and from cartilage of lamprey are

similarly degraded by alkali and by papain, M.values are decreased by less than 50%. Also,Pronase shows an appreciable effect on [,q] of allthree of these tryptic-chymotryptic hydrolysateproducts, in contrast to a lack of effect on the firstthree preparations in Table 1. The values of M,are only approximations, since they have beencalculated from intrinsic-viscosity values outsidethe range of the values used to establish theempirical relationships.

Composition of chondroitin sulphate-protein. Forpreparations of chondroitin sulphate-protein fromcartilage of ox, sturgeon and ray, density-gradientfractionation (Table 2) achieved separation ofextraneous protein with virtually complete con-

centration of the polysaccharide content in the

densest fraction. Analyses are given for thesefractions (zone C) in Table 3, computed for assumeddoublets (two average polysaccharide chains).These computations were based on values of M. ofthe products derived from alkali treatments oftryptic-chymotryptic hydrolysate products (Table1). The gram-molecular weight for the parentdisaccharide, which is defined (Mathews, 1967) as

consisting of one residue of glucuronic acid, one

residue of acetylgalactosamine and one residue ofester sulphate, was 503 g; for the over-sulphated raypolysaccharide (Mathews, 1962) the correspondingvalue was 515g. Relatively small proportions ofthe total protein in each tube were contained inzones of low density. These results differed fromthose of Hascall & Sajdera (1969) and may beattributed to a comparatively low content ofcontaminating protein in the preparations.

Considerable variation of protein content withzone density is shown by the results in Table 2 forchondroitin sulphate-protein fractions of theremaining preparations. Zone B as well as zone Cwas analysed but analytical results for zone Bfractions only are given in Table 3. The correspond-ing zone C fractions have lower contents of totalamino acid residues per doublet, as indicated byresults on protein contents of the chondroitinsulphate-protein of each zone. Although theprotein of a zone C fraction was similar in aminoacid composition to the protein of the correspondingzone B fraction, the differences in content for someamino acids, especially cystine, clearly indicated

SourceOx cartilage

Sturgeon cartilage

Ray cartilage

Lamprey cartilage

Sturgeon notochord

Lamprey notochord

197140 M. B. MATHEWS

STRUCTURE OF CHONDROITIN SULPHATE-PROTEINSTable 3. Molar composition per doublet of chondroitin sulphate-protein fractions from density-gradient

centrifugation

ResidueGalactosamineGlucosamineSerineGlycineGlutamic acidAspartic acidProlineThreonineAlanineValineIsoleucineLeucineTyrosinePhenylalanineLysineHistidineArginineHalf-cystineSum of aminoacid residues

Oxcartilagezone C8665.35.45.33.23.32.33.22.21.53.40.71.51.50.71.81.0

42

Sturgeoncartilagezone C100

54.14.64.42.71.71.82.32.13.61.42.50.81.00.61.00.635

Raycartilagezone C15554.06.26.23.95.22.43.22.83.82.40.81.21.80.51.90.9

46

Lampreycartilagezone B200.210.88.0

15.37.05.44.75.25.23.33.50.01.22.24.93.20.580

Sturgeonnotochordzone B231.04.24.88.14.24.53.53.52.51.63.31.31.32.21.41.90.9

49

Lampreynotochordzone B230.08.26.2

24.89.3

10.612.07.63.78.18.23.22.32.54.63.61.0

116

Table 4. Molar composition of tryptic--chymotryptic hydrolysatesulphate doublets

,of chondroitin sllphate-protein as chondroitin

ResidueGalactosamineGlucosamineSerineGlycineGlutamic acidAspartic acidProlineThreonineAlanineValineIsoleucineLeucineTotal amino acids

heterogeneity. The half-cystine contents of zone Cfractions of chondroitin sulphate-protein fromlamprey cartilage, sturgeon notochord and lampreynotochord were respectively 0.1, 0.2 and 0.4mol/doublet.

Composition of tryptic-chymotryptic hydrolysateproduct8. The amino acid compositions of thetryptic-chymotryptic hydrolysate products com-

puted on the basis of chondroitin sulphatedoublets are shown in Table 4. Several composi-tional features are common to all preparations, i.e.

approx. 2mol each ofserine and glycine, a minimumof mol of glutamic acid or glutamine and a totalamino acid content less than one-third of thenumber of amino acids per doublet in chondroitinsulphate-protein indicated in Table 3. Apparently,each serine residue is bound via a hydroxyl groupto a polysaccharide chain.

Effects of enzyme and alkali treatments. Thetreatment ofthe trypsin-chymotrypsin hydrolysateproduct with alkali or with enzymes results inproducts containing polysaccharide with altered

Oxcartilage

850.71.91.91.30.50.90.40.60.60.40.89.3

Sturgeoncartilage100

0.82.02.31.00.8

0.60.6

0.50.88.6

Raycartilage155

1.92.12.41.91.12.80.3

1.91.8

14.3

Lampreycartilage200.22.22.21.90.60.50.40.70.90.40.3

10.1

Sturgeonnotochord

230.22.12.32.60.81.71.11.51.30.70.7

14.8

Lampreynotochord

23

2.02.03.51.11.11.30.80.4

12.2

Vol. 125 41

Table 5. Molar composition per doublet of produd8 from treatment of tryptic-chymotryptic hydrolysate ofchondroitin sulphate-proteins of OX, sturgeon and ray cartilages

Ox cartilage

NaOH Pronase Papain83 86 860.7 0.7 0.40.7 2.0 2.20.9 1.8 1.70.7 1.5 0.70.4 0.5 0.50.7 1.1 0.50.2 0.3 0.30.3 0.6 0.60.3 0.6 0.5

Sturgeon cartilage

NaOH Pronase100 100

1.0 1.00.3 2.00.9 1.50.5 1.10.3 1.0

0.3 0.70.2 0.7

0.5 0.4

Table 6. Molar composition per doublet ofproductsfrom treatment of tryptic-chymotryptic hydrolysate of otherchondroitin sulphate-protein8

Lamprey cartilage

NaOH200.21.81.81.80.60.40.40.80.8

Pronase200.22.22.21.60.70.60.40.60.6

Sturgeon notochord

Papain200.22.01.50.80.40.30.30.50.4

NaOH230.20.60.70.70.30.70.70.40.5

Pronase230.21.91.62.00.61.00.90.80.8

Lamprey notochord

Papain280.31.71.00.80.40.70.70.50.6

NaOH230.00.40.41.20.20.20.60.40.1

Pronase230.01.81.81.40.90.40.50.50.1

Papain260.01.71.61.00.90.30.30.1

amino acid compositions, as well as molecularparameters. Tables 5 and 6 show the compositionsof these products. The structural interpretationof the compositional results, as was the case for theresults on molecular parameters, is relatively simplefor the products listed in Table 5. The productslisted in Table 6 cannot be interpreted in the same

way and are discussed below.To account for the observation that papain

bisects the 'doublet fraction', presumably bypeptide-bond cleavage, one may assume thepresence in this portion of the polypeptide chain ofaminimum peptide sequence with one serine residueas the N-terminal group and a second serineresidue as the C-terminal group. For purposes ofbrevity, this hypothetical peptide sequence betweentwo serine residues, including the two terminalserine residues as well, will be referred to as sequenceX. A polysaccharide chain is linked to the hydroxylgroup of each serine residue. In the tryptic-chymotryptic hydrolysate products some amino

acid residues may be present that are linked

to the serine residue and extemal to sequence X.Also, all of the amino acid residues in Table 4 maybe components of sequence X, including thosepresent in fractional molar amounts, since theextent of polymorphism of sequence X is unknown.Papain digestion (Table 5) not only bisects the

tryptic-chymotryptic hydrolysate product but alsopartially removes amino acid residues, except forserine. These results do not permit the assignmentof any residue to sequence X with certainty. How-ever, the retention of glycine suggests that at leastone glycyl residue is vicinal to the serine. Pronasedigestion, which does not bisect the tryptic-chymotryptic hydrolysate product, removes pri-marily leucine and isoleucine, which are apparentlyexternal to sequence X. Consequently, a probablemaximum of nine amino acid residues comprisesequence X.Treatment of the tryptic-chymotryptic hydro-

lysate product with alkali may be expected toresult in ,B-elimination of the polysaccharide chainsfrom serine residues, accompanied by conversion

ResidueTreatment ...

GalactosamineGlucosamineSerineGlycineGlutamic acidAspartic acidProlineThreonineAlanineValineIsoleucineLeucine

Ray cartilage

Papain921.01.81.80.60.6

0.2

NaOH155

1.90.81.31.30.71.5

Pronase155

1.92.22.21.81.02.6

Papain1550.81.91.01.00.80.50.3

2.11.2

ResidueTreatment ...

GalactosamineGlucosamineSerineGlycineGlutamic acidAspartic acidProlineThreonineAlanineValine

197142 M. B. MATHEWS

STRUCTURE OF CHONDROITIN SULPHATE-PROTEINSof the serine residues into dehydroalanine residuesprovided that both the carboxyl group (Anderson etal. 1965) and the amino group (Bella & Danishefsky,1968) of each serine residue are substituted.Alkali treatment of the tryptic-chymotryptichydrolysate products shown in Tables 5 and 6,except for the preparation from lamprey cartilage,produces a loss of slightly more than one serineresidue and about one glycine residue. It wouldtherefore appear that at least one serine residue ofsequence X in each molecule of the tryptic-chymotryptic hydrolysate products is substitutedat both amino and carboxyl positions. However,the recovery of polysaccharide containing someserine shows that an appreciable amount of serineis also in a terminal position of peptide sequences.In addition, the supernatant solution obtained afterprecipitation ofalkali-treated tryptic-chymotryptichydrolysate products with cetylpyridinium chlorideis free of amino sugar, but contains amino acidsincluding 0.1-0.2mol of serine per doublet. Thelatter result indicates the presence in the tryptic-chymotryptic hydrolysate products of some serinethat is not glycosylated.

DISCUSSION

Structure of chondroitin sulphate-protein. Fig. 1summarizes the observations on the tryptic-

chymotryptic hydrolysate products and suggests astructure for chondroitin sulphate-protein. Thiselementary schematic formulation assumes thepresence of a single polypeptide chain to which sidechains of chondroitin sulphate of uniform lengthare attached at regular intervals; it will be utilizedhere as a parent model structure.

Structure of the tryptic-chymotryptic hydroly8ateproduct. The hypothesis that the product of trypticanid chymotryptic hydrolysis of chondroitin sul-phate-protein contains a polypeptide sequence X,comprising less than ten amino acid residues, withpolysaccharide chains bound glycosidically to thehydroxyl groups of two serine residues, is in accordwith the results for preparations from cartilage ofox,sturgeon and ray. The persistence of a glutamateresidue in association with polysaccharide chainsaftervarioustreatmentsofthetryptic-chymotryptichydrolysate products (Tables 5 and 6) suggests thatat least one such residue is close to a serine residueand is possibly included in sequence X. Indeed, apeptide of the structure Glu-Gly-dehydroAla-Glyhas been isolated by Katsura & Davidson (1966)from a mixture of peptides resulting from alkalitreatment of a papain digest of porcine chondroitinsulphate-protein. However, the results of alkalitreatment ofthe tryptic-chymotryptic hydrolysateproducts are incompatible with the assignment ofthis sequence about both serine residues of thedoublet. This is so since the observed ,-elimination

Short sequence, X< 10 amino acid residues

Long sequence, Yav. -35 amino acid residues

I

----A Ser B---A'Ser'B'Z.- -- A Ser B --A'Ser'B'----I

Polypeptide resistant toenzymes except papain

Polypeptide split byproteolytic enzymes

I

Chondroitin sulphate Chondroitin sulphate'

Polysaccharide chain

Fig. 1. A segment of the proposed structure for a chondroitin sulphate-protein molecule which contains a

single polypeptide chain of 2000-3000 amino acid residues to which side chains of chondroitin sulphate are

attached. In the simplest, parent, structure the polypeptide chain consists of alternating X and Y sequencesand every X sequence bears two polysaccharide side chains.

Doublet, D

l

Vol. 125 43

M. B. MATHEWS

reactionwould require minimally either thesequence-Gly-Ser-Gly-An-Gly-Ser- or the sequence -Ser-Gly-An-Gly-Ser-Gly-, where An represents other aminoacid residues of sequence X. Thus if each serineresidue had been flanked by two glycine residues inchondroitin sulphate-protein, then the tryptic-chymotryptic hydrolysate product should havecontained minimally 3mol of glycine instead of theassayed value of 2mol per doublet.

Polypeptide 8tructure. Considerations of thereduction of protein contents from starting materialobserved in zone C after density sedimentationrequires a revision of the molecular weight for thebasic molecular units of chondroitin sulphate-protein from bovine cartilage (Mathews & Lozaityte,1958) and from chondrichthye (Mathews, 1962) toabout 3.5 x 106. Hascall & Sajdera (1970) obtaineda mean molecular weight of 2.5 x 106 for the bovinechondroitin sulphate-protein. Therefore the poly-peptide chains contain 2000-3000 amino acidresidues per molecule of chondroitin sulphate-protein. Of the many possible different ways ofvisualizing the main polypeptide chain in terms ofsubunit sequences only the simplest ones will beconsidered. Although Serafini-Fracassini, Peters &Floreani (1967) had suggested that peptide subunitsmay be assembled in a random manner, there is noevidence for covalent linkages other than peptidebonds. Accordingly, the polypeptide chain of amolecule of chondroitin sulphate-protein is as-sumed to be a single biosynthetic entity whosestructure is determined by a single structural gene.Division of the chain into subunits is solely fordescriptive purposes.The structure shown in Fig. 1 may be composed

ofonly two types ofsubunits in regular order so thatthe polypeptide chain is represented as a repeatingpolymer (-X-Y-),. Alternatively, the order ofXandY sequences may not be regular, or more than twotypes of sequences may be present. The availableresults do not warrant further detailed speculationat this moment. The situation is complicated furtherby the presence in chondroitin sulphate-proteinpreparations of variable proportions of keratansulphate-peptides, in part covalently linked to thechondroitin sulphate-protein molecule (Rosenblum& Cifonelli, 1967; Tsiganos & Muir, 1967).

It is evident from the variation in composition ofchondroitin sulphate-proteins with buoyant densitythat the molecules are heterogeneous. Tsiganos,Hardingham & Muir ( 1971) showed that chondroitinsulphate chains were of uniform size in comparablefractions from pig laryngeal cartilage and con-cluded that the heterogeneity was due to thepresence of different polypeptide chains differingin length and also in the type, distribution andnumber ofcarbohydrate chains attached. However,Hascall, Riolo, Hayward & Reynolds (1971) showed

that the polypeptide chains ofchondroitin sulphate-proteins of bovine nasal cartilage were of closelysimilar size and composition differing primarily inthe number of keratan sulphate chains attached.These findings are compatible with the presencegenerally of more than one type of polypeptidechain and of a variable extent of heterogeneity ofchondroitin sulphate-proteins dependent on tissue,metabolic state etc.The different polypeptide chains, in accord with

the results of the present paper, may be similar toeach other in amino acid composition and to somedegree in amino acid sequences. Their similaritiesin structure and in structural relationships topolysaccharide chains suggests that these poly-peptides may be polymorphically related. Thepresence of polymorphic polypeptides could ac-count not only for the small variations in amino acidcomposition of the protein of fractions of differentbuoyant densities but also for the presence of non-integral amino acid residues in the doublet fractionsshown in Table 4. However, such non-integralvalues may also arise as a consequence ofincompleteproteolysis, as indicated by the action of Pronase ondoublets.The degree of difference in amino acid sequence

between polymorphic or homologous proteins isusually variable along the polypeptide chains withthe smallest number of residue differences occurringgenerally in regions associated with strict functionalrequirements. Since X sequences are involved inlinkage to carbohydrate we might expect littlevariation of structure in this region of the poly-peptide chains of molecules from the same tissue.Thus in chondroitin sulphate-protein (J. Baker &L. Roden, unpublished work), as in the structurallyrelated dermatan sulphate-protein molecule (Bella& Danishefsky, 1968; Lindahl & Roden, 1971), aglycine residue is linked to the carboxyl group ofserine residues that bear a chondroitin sulphatechain. As a consequence, if the amino terminus ofthe peptide chain is chosen to lie to the left of thesequence shown in Fig. 1, then both B and B' areglycine residues. The probable presence of at leastone glutamate residue in sequence X has alreadybeen noted.

Polysaccharide chains. Chondroitin sulphate-protein preparations (Table 3) contain a minimumof two serine residues per doublet that are notinvolved in linkage to chondroitin sulphate chains.Thus it is possible that glycosylation of serineresidues may be limited to specific regions of thepolypeptide chain by requirements for enzymespecificity or by steric factors. These conditionsset an upper limit to the number of chondroitinsulphate chains that may be linked to a given poly-peptide. Also, gel electrophoresis showed thepresence of single chondroitin sulphate chains as

44 1971

STRUCTURE OF CHONDROITIN SULPHATE-PROTEINSwell as doublets in tryptic-chymotryptic hydro-lysate products, particularly those from lampreytissues and from sturgeon notochord. In addition,the relative proportions ofchondroitin sulphate andprotein vary in zone B and zone C fractions.To explain these observations wemay consider the

hypothetical examples of chondroitin sulphate-protein structure of Fig. 2. The structures shownfor zoneB reflect a chondroitin sulphate/polypeptideweight ratio lower than that of the structure forzone C, in accordance with the lower sedimentationdensity. Since the intrinsic viscosities ofchondroitinsulphate isolated by papain digestion of material inzones A, B and C of preparations from sturgeonnotochord and from lamprey tissues are virtuallyidentical, the chain sizes of chondroitin sulphatemust be the same. Hence structure 1 of zone Bcannot be of major significance. It is probabletherefore that structures similar to those of types2 and 3 are present in regions of the chondroitinsulphate-protein molecule, particularly in chon-droitin sulphate-protein from lamprey tissues.Fig. 1 shows that the two serine residues of thedoublet are distinguishable and that one of theseresidues may be preferentially glycosylated. Theparticular examples in Fig. 2 were selected toaccount for the presence of single chondroitin

Polypeptide

Chondroitin sulphateZONE C side chains

ZONE B -

7ETF1111N

I--

L1 1 1-

2

3

Fig. 2. Some hypothetical and schematic structures ofrepresentative segments of chondroitin sulphate-proteinmolecules illustrating variation in number and arrange-

ment of polysaccharide chains attached to a given poly-peptide sequence.

sulphate chains of uniform size in chonclroitinsulphate-protein and should not imply (see above)that the polypeptide chain necessarily consists ofX and Y sequences in regular altemating order.A similar type of variation in numbers of chains

may account for the observed variation in peptidecontent, mean sedimentation density and mole-cular weight of fractions of chondroitin sulphate-protein from heart valves (Lowther, Preston &Meyer, 1970). Hascall & Sajdera (1970) observedsimilar variations in these characteristics of frac-tions of chondroitin sulphate-protein from bovinecartilage. With the exception of half-cystine,amino acids were present in essentially the samerelative proportions in all fractions. Hascall &Sajdera (1970) suggested that polydispersity inmolecular weight of bovine chondroitin sulphate-protein was due primarily to differences in thenumber of chondroitin sulphate chains attached toone type of protein. However, Serafini-Fracassini(1968) obtained results suggesting that bovinechondroitin sulphate-protein contains three dif-ferent polypeptide chains. Also, Brandt & Muir(1971) found appreciable polydispersity in size ofchondroitin sulphate chains from chondroitinsulphate-proteins of pig cartilage.

Phylogeny and ontogeny. The hypothesis that thechondroitin sulphate-protein molecules from dif-ferent vertebrate species form an homologous classis supported indirectly by structural data showingthe occurrence ofvery similar, ifnot identical, poly-saccharide chains bound to polypeptide in anidentical fashion (Mathews, 1967). Strong evidencefor this hypothesis, however, requires the demon-stration that the polypeptide structures, which arereferable directly via the genetic code to structuralgenes, are also very similar. My present results onthe short peptide sequence of the doublet areindicative of homology and also suggest conserva-tion over an extended period ofevolution. Immuno-logical results, which show the localization of acommon antigenic determinant in the region of theshort sequence, are consistent with this conclusion(Sandson, Damon & Mathews, 1970). The generalsimilarity in number and type of amino acidresidues composing the long polypeptide sequencesof chondroitin sulphate-protein molecules of ox,sturgeon, ray and embryonic chick (Z. Nevo,unpublished work) cartilage suggests homology forthis section of the chemical structure as well.The chemistry of chondroitin sulphate-protein

of vertebrates varies with stage of embryonicdevelopment and maturation in extent and site ofsulphation of the amino sugar residues of carbo-hydrate chains (Mathews, 1967). Concomitantvariation in the extent of glycosylation of serineresidues may also occur. In addition, the possiblepresence of variable proportions of two or more

Vol. 125 45

46 M. B. MATHEWS 1971types of polypeptide, differing mainly in half-cystine content, may be of particular significance,since this residue is involved in aggregation ofchondroitin sulphate-protein molecules (Sajdera,Hascall, Gregory & Dziewiatkowsky, 1970). Allsuch variations in molecular structure influenceinteractions of chondroitin sulphate-protein withcollagen and hence affect the molecular organizationand properties ofconnective tissues (Mathews, 1970;Obrink & Wasteson, 1971).

I am indebted to Miss LaVerne Decker for able labora-tory assistance and to Dr Lennart Rod6n ofthe Universityof Chicago for constructive criticism of the manuscript.This work was supported by Grants AM-05996 and HD-04583 of the U.S. Public Health Service and by a grantfrom the Chicago Heart Association.

REFERENCES

Ali, S. Y. (1964). Biochem. J. 93, 611.Anderson, B., Hoffman, P. & Meyer, K. (1965). J. biol.Chem. 240, 156.

Bella, A., jun. & Danishefsky, I. (1968). J. biol. Chem. 243,2660.

Boas, N. F. (1953). J. biol. Chem. 204, 553.Brandt, K. D. & Muir, H. (1971). Biochem. J. 121,261.Elson, L. A. & Morgan, W. T. J. (1933). Biochem. J. 27,

1824.Hascall, V., Riolo, R., Hayward, J. & Reynolds, C. (1971).Fedn Proc. Fedn Am. SoC8 exp. Bil. 30, 1278.

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

Hasoall, V. C. & Sajdera, S. W. (1970). J. biol. Chem. 245,4920.

Hirs, C. H. W. (1967). In Methods in Enzymology, vol. 11,p. 197. Ed. by Hirs, C. H. W. New York: AcademicPress.

Katsura, N. & Davidson, E. A. (1966). Biochim. biophys.Acta, 121, 120.

Lindahl, U. & Rod6n, L. (1971). In Glycoproteins, 2nd ed.Ed. by Gottschalk, A. Amsterdam: Elsevier (in thePress).

Lowther, D. A., Preston, B. N. & Meyer, F. A. (1970).Biochem. J. 118, 595.

Luscombe, M. & Phelps, C. F. (1967). Biochem. J. 103,103.

Mathews, M. B. (1956). Archs Biochem. Biophys. 61, 367.Mathews, M. B. (1959). Biochim. biophy8. Acta, 35, 9.Mathews, M. B. (1962). Biochim. biophys. Acta, 58, 92.Mathews, M. B. (1967). Biol. Rev. 42, 499.Mathews, M. B. (1970). In Chemistry and Molecular

Biology of the Intercellular Matrix, vol. 2, p. 1155. Ed.by Balazs, E. A. London: Academic Press.

Mathews, M. B. & Decker, L. (1971). Biochim. biophys.Acta, 244, 30.

Mathews, M. B. & Dorfman, A. (1953). Archs Biochem.Biophy8. 42, 41.

Mathews, M. B. & Lozaityte, I. (1958). Archs Biochem.Biophys. 74, 158.

Obrink, B. & Wasteson, A. (1971). Biochem. J. 121, 227.Rod6n, L. (1968). In Biochemistry of Glycoproteins and

Related Substances, p. 185. Ed. by Rossi, E. & Stoll,E. New York: S. Karger.

Rosenberg, L., Hellmann, W. & Kleinschmidt, A. K.(1970). J. biol. Chem. 245, 4123.

Rosenblum, E. L. & Cifonelli, J. A. (1967). Fedn Proc.Fedn Am. SoCs exp. Biol. 26, 282.

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

Sajdera, S. W., Hascall, V. C., Gregory, J. D. & Dziewiat-kowsky, D. D. (1970). In Chemistry and MolecularBiology of the Intercellular Matrix, vol. 2, p. 851. Ed. byBalazs, E. A. London: Academic Press.

Sandson, J., Damon, H. & Mathews, M. B. (1970). InChemistry and Molecular Biology of the IntercellularMatrix, vol. 3, p. 1563. Ed. by Balazs, E. A. London:Academic Press.

Serafini-Fracassini, A. (1968). Biochim. biophys. Acta,170, 289.

Serafini-Fracassini, A., Peters, T. J. & Floreani, L. (1967).Biochem. J. 105, 569.

Serafini-Fracassini, A. & Smith, J. W. (1966). Proc. R.Soc. B, 165, 440.

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

Tsiganos, C. P. & Muir, H. (1967). Biochem. J. 104, 26c.