Glycosaminoglycans in cervical connective tissue during pregnancy and parturition.

OBJECTIVE: To investigate the content and distribution patterns of glycosaminoglycans in the human cervix during pregnancy and parturition. METHODS: We obtained a total of 87 specimens from nonpregnant and pregnant women. Biopsies (weight 50-200 mg) were taken from the posterior lip of the cervix. Hyaluronic acid, dermatan sulfate, chondroitin sulfates, and heparan sulfate were separated on a Dowex 1 x 2 column and identified. RESULTS: The total amount of glycosaminoglycans increased during pregnancy from 2800 to 5000 nmol/g dry weight. The highest values were observed at the onset of labor (7100 nmol/g dry weight), followed by a sharp decrease during parturition. The clinical features of cervical ripening and dilatation were also associated with remarkable changes in glycosaminoglycan patterns. CONCLUSION: Besides collagenolysis during pregnancy, the glycosaminoglycans are also important regulators of cervical function. The different clinical features of the human cervix are characterized not only by variation in the total glycosaminoglycan content but also by changes in the proportions of the different glycosaminoglycans.

Quantitative analysis of glycosaminoglycans in urothelium and bladder wall of calf.

Glycosaminoglycans (GAG) were described by histochemical staining procedures in the normal urothelium of the urinary bladder; they are supposed to be involved in the antibacterial defense mechanism. Our quantitative analysis, however, demonstrated only heparan sulfate in the normal calf urothelium (less than 600 nmol/Gm d.wt.); only trace amounts of other GAG were to be analyzed. High concentrations of GAG were found in the submucosa and muscle layers as to be expected in mesoderm originating tissues. According to these results there were no GAG with the exception of heparan sulfate at the surface of the normal urothelium; therefore, glycoproteins detected in mumol/Gm d.wt. ranges are more likely to be involved in the antibacterial defense mechanism.

Examination of corneal proteoglycans and glycosaminoglycans by rotary shadowing and electron microscopy.

Proteoglycans (PGs) from cornea and their relevant glycosaminoglycan (GAG) chains, dermatan sulphate (DS) and keratin sulphate (KS), were examined by electron microscopy following rotary shadowing, and compared with hyaluronan (HA), chondroitin sulphate (CS), alginate, heparin, heparan sulphate (HS) and methyl cellulose. Corneal DS PG had the tadpole shape previously seen in scleral DS FG, and the images from corneal KS PG could be interpreted similarly, although the GAG (KS) chains were very much fainter than those of DS PG GAG. Isolated GAG (KS, DS, CS, HA, etc.) examined in the same way showed images that decreased very significantly in clarity and contrast, in the sequence HA greater than DS greater than CS greater than KS. The presence of secondary and tertiary structures in the GAGs may be at least partly responsible for these variations. HA appeared to be double stranded, and DS frequently self-aggregated, KS and HS showed tendencies to coil into globular shapes. It is concluded that it is unsafe to assume the absence of GAGs, based on these techniques, and quantitative measurements of length may be subject to error. The results on corneal DS PG confirm and extend the hypothesis that PGs specifically associated with collagen fibrils are tadpole shaped.

[Cystine stone therapy with mercapto-propyonyl-glycine (MPG) (Thiola) (author's transl)]. / Cystinsteintherapie mit Mercaptopropyonylglycin (MPG) (Thiola)

Six patients with recurrent cystine stones and cystinuria were treated with MPG. Three patients were treated for 3 years and 3 patients for 6 months. A relatively high dosage resulted in a reduction of the cytinuria. Since MPG has almost no side effects it represents a valuable aid in the treatment of cystine stones.

[The biochemical mechanism of action of proquazone with special reference to the metabolism of connective tissue (biochemical studies for the determination of antiinflammatory effectiveness)]. / Zum biochemischen Wirkungsmechanismus von Proquazon unter besonderer Berücksichtigung des Bindegewebsstoffwechsels (Biochemische Untersuchungen zur Prüfung der antiphlogistischen Wirksamkeit).

A method is proposed to determine the specific radioactivity of proteokeratan sulfate and proteochondroitin sulfate after incorporation of 35SO4 and 3H-glucosamine. Proquazone inhibits the incorporation of 35SO4 in proteochondroitin sulfate and also in proteokeratan sulfate. Also the incorporation of 3H-glucosamine in proteokeratan sulfate and proteochondroitin sulfate is inhibited. A mechanism is proposed in which Proquazon inhibits the core protein synthesis of proteoglycans and the secondary biosynthesis of the glycosaminoglycan chains. In our opinion the action of an anti-inflammatory drug is based on the inhibition of the proteoglycan synthesis in the anabolic phase of the inflammation.

Constant and variable domains of different disaccharide structure in corneal keratan sulphate chains.

Four peptidokeratan sulphate fractions of different Mr and degree of sulphation were cut from the pig corneal keratan sulphate distribution spectrum. After exhaustive digestion with keratanase, the fragments were separated on DEAE-Sephacel and Bio-Gel P-10 and analysed for their Mr, degree of sulphation and amino sugar and neutral sugar content. It was found that every glycosaminoglycan chain is constructed of a constant domain of non-sulphated and monosulphated disaccharide units and a variable domain of disulphated disaccharide units. Total neuraminic acid of the four peptidokeratan sulphates was recovered from their isolated linkage-region oligosaccharides. In kinetic studies, the four peptidokeratan sulphates were investigated for Mr distribution after various incubation times with keratanase. There was a continuous shift towards lower Mr and no appearance of a distinct intermediate-sized product at any degradation time. The linkage-region oligosaccharide was already being liberated after a very short incubation period. From the results of these kinetic investigations in connection with the results of neuraminic acid analyses it is suggested that there exists only one disaccharide chain per peptidokeratan sulphate molecule. A model of corneal keratan sulphate is postulated. One of the alpha-mannose residues in the linkage region is bound to an oligosaccharide consisting of a lactosamine and a terminal sialic acid. The other alpha-mannose residue is attached to the disaccharide chain. This chain contains one or two non-sulphated disaccharide units at the reducing end, followed by 10-12 monosulphated disaccharide units. The disulphated disaccharide moiety of variable length is positioned at the non-reducing end of the chain.

Laser nephelometric determination of glycosaminoglycans--method and application.

Light scattering due to the formation of insoluble complexes between the long-chain quaternary ammonium salt N-cetylpyridinium chloride and glycosaminoglycans was utilized for a relative simple, sensitive and precise determination of total and specific types of glycosaminoglycans by laser nephelometry. The addition of the ammonium salt to solutions of various glycosaminoglycans in 0.03 mol/l NaCl produces a time-dependent increase in light scattering, which reaches a maximum between 14 and 18 h of complex formation, irrespective of the type of glycosaminoglycan studied. Only keratan sulphate does not generate light scattering, and is therefore not detectable by the procedure. The scattering of laser light by certain types of sulphated glycosaminoglycans (e.g. heparan sulphate, heparin) depends more on the degree of sulphation (charge density) than on chain length within a certain range. Optimum light scattering was found at 28 mmol/l N-cetylpyridinium chloride and at a ionic strength around 0.03 mol/l NaCl. The detection limits and linear ranges of the individual glycosaminoglycans were evaluated. For the determination of chondroitin sulphate, laser nephelometry is at least 8 times more sensitive and much more simple than the modified carbazole method (glucuronic acid). The intra-assay and inter-assay coefficients of variation are about 4% and 7%, respectively. Laser nephelometry is much more sensitive than turbidimetry. Complex synthetic mixtures of glycosaminoglycans and biological fluids were accurately differentiated by successive chemical and enzymatic degradation of the respective glycosaminoglycans followed by the measurement of the resulting reduction of laser light scattering. In synovial fluids from non-inflammatory joint diseases, light scattering (units/ml) was about 4.5 times higher than in synovial fluids from inflammatory articular lesions. In both pathologic conditions nearly all of the light scattering can be attributed to hyaluronic acid.

Glycosaminoglycans in urothelial carcinomas.

The glycosaminoglycan (GAG) content in human urothelial carcinomas was biochemically determined and compared to that of normal urothelium and bladder wall of the calf. The total GAG content was elevated in urothelial carcinomas, and the distribution pattern differed from that of normal urothelium and bladder wall. Whereas urothelial carcinomas contained heparan sulphate, dermatan sulphate, chondroitin 4-sulphate and chondroitin 6-sulphate, only heparan sulphate could be detected in the normal urothelium. The GAG determination was based on hexosamine analysis and thin layer chromatography after elution on Dowex 1 X 2 columns.

[Paraproteinaemic coma in multiple myeloma (author's transl)]. / Das Coma paraproteinaemicum beim multiplen Myelom.

The diagnosis of paraproteinaemic coma was made in 5 out of 74 patients with plasmocytoma. Compared with the other patients these 5 showed a nearly twofold increase of total serum and CSF protein concentration, a higher percentage of plasma cells in sternal marrow and extreme hypergamma-globulinaemia in the immunoelectrophoresis. In both groups CSF and serum amino acid concentrations showed no pathological changes. Psychopathologically the patients with paraproteinaemic coma showed the picture of an exogenous psychosyndrome. Electroencephalography demonstrated moderate general disturbances.

Increased urinary excretion of keratan sulfate in fucosidosis.

In two children exhibiting the clinical symptoms of fucosidosis, the diagnosis was biochemically ascertained by the demonstration of a profound altpha-L-fucosidase deficiency in cultured skin fibroblasts. The non-dialysed urines of these fucosidosis patients were separated into two fractions by chromatography on Biogel P-2. The first fraction containing the glycosaminoglycans was further fractionated on Dowex 1 X 2 by stepwise elution with increasing NaCl concentrations. Keratan sulfate-chondroitin sulfates attached to the same peptide core were assayed and characterised mainly in the fractions eluted with 1.25, 1.5, 2.0 and 3.0 mol/1 NaCl. Whereas the excretion of normal children of the same age was found to be 0.77 mumol glucosamine equivalents per day in the 2 mol/1 and 3 mol/1 NaCl fraction, the two patients excreted 6.7 (M. C.) and 3.5 (M. S.) mumol glucosamine equivalents per day, respectively. Since keratan sulfate contains alpha-fucose at the non-reducing terminal, this increase in excretion of long chain keratan sulfate in fucosidosis could result from impaired degradation of keratan sulfate, due to the alpha-fucosidase deficiency.

Mercaptopropionylglycine: a progress in cystine stone therapy.

During a 4-year period 9 cases of severe recurrent cystine stones and cystinuria were treated with mercaptopropionylglycine. The oral long-term administration revealed the high effectiveness of the drug and resulted in no further stone formation. Mercaptopropionylglycine is free of side effects and more effective than D-penicillamine.

Interactions of glycosaminoglycans with DNA and RNA synthesizing enzymes in vitro.

The sulfated glycosaminoglycans chondroitin 4-sulfate, chondroitin 6-sulfate, keratan sulfate, dermatan sulfate, heparin, and glycosaminoglycan polysulfate are competitive inhibitors of the DNA-dependent RNA polymerase, the DNA-dependent RNA polymerase and the RNA-dependent DNA polymerase (reverse transcriptase). The unsulfated glycosaminoglycans chondroitin and hyaluronate are without any influence on the synthesis of DNA and RNA. The strongest inhibitor is a glycosaminoglycan polysulfate with four sulfate groups per disaccharide unit. It has the following inhibitor constants: DNA polymerase, Ki = 1.5 X 10(-6) M; RNA polymerase, Ki = 0.9 X 10(-6) M; reverse transcriptase, Ki = 1.1 X 10(-6) M. The inhibition is closely correlated to the degree of sulfation of the glycosaminoglycans. There is a relationship between the sulfate/hexosamine ratio and the degree of inhibition. The inhibition of the DNA and RNA synthesizing enzymes by sulfated glycosaminoglycans depends on the nature of the template. With double-stranded DNA as template, inhibition occurs only when sulfated glycosaminoglycans are added before or shortly after (30 s) initiation of the synthesis. There is no inhibition if the inhibitors are added after the onset of the synthesis. On the other hand, with a single-stranded template synthesis was blocked completely at each phase of reaction.

Studies on the mechanism of hyaluronate lyase action.

Hyaluronate lyase from streptococci group A was purified by chromatography on DEAE-Sephadex A-50 and on Biogel P-150, thereby enriching it about 1,000-fold and separating it into two enzyme fractions with the same amino acid composition. The photooxidation of hyaluronate lyase in the presence of methylene blue results in rapid inactivation of the enzyme. The histidine content of the enzyme is decreased considerably, but also the content of methionine, tyrosine, and lysine is lowered. The enzyme is inhibited, but incompletely so, by N-tosyl-L-phenyl-alanine-chloromethyl ketone (TPCK) and N-alpha-p-tosyl-L-lysine chloromethyl ketone (TLCK). Hyaluronic acid methyl ester, prepared form hyaluronic acid and diazomethane, is not split by hyaluronate lyase (EC 4.2.2.1.). Hyaluronic acid methyl ester is not a competitive inhibitor of hyaluronate lyase. For the mechanism of the enzymatic elimination reaction a proton transfer between histidine of the enzyme and the carboxylate group of hyaluronate is proposed.

[Biochemical principles of cervix ripening and dilatation]. / Biochemische Grundlagen der Zervixreifung und Muttermundseröffnung.

The central function of the cervix to maintain pregnancy is biochemically characterized by an increased synthesis of collagen, proteins, glycosaminoglycans (GAG) and fibronectin within the extracellular matrix, thus leading to an increase of cervical volume without significant changes of cervical consistency. During the time of cervical ripening we found a marked reduction of collagen concentration, a 2.5-fold increase in GAG content, a significant fall in dermatan sulfate concentrations from 41% to 15% of total GAG content, a 12-fold increase in hyaluronate concentrations, and a marked reduction in fibronectin, demonstrated by immunhistochemical methods. Thus, the loss of collagen and sulfated GAGs may facilitate distensibility in the ripened cervix, while the significant gain in hyaluronate associated with hydratation may explain the soft and swollen consistency. In this connection increased hyaluronate concentrations and degradation of fibronectin may play a trigger role for subsequent cervical dilatation. The dramatic changes of the cervix during parturition occurring within a few hours require the rapid activation and action of catabolic enzyme systems. Our studies showed a significant increase of sialidase-, collagenase- and elastase activities during cervical dilatation. These proteinases originate from polymorphonuclear leucocytes (pml), which accumulate in cervical capillaries at the onset of labor; this is followed by a massive leucocyte infiltration of the cervical stroma at the beginning of cervical dilatation and a degranulation of the pml at further dilatation, thus releasing collagenase and other proteinases. This process is limited by the immediate post partum insudation of the cervix by plasma containing highly potent proteinase inhibitors. The clinical aim of our basic biochemical studies is to develop new concepts in the causal treatment of cervical pathology during pregnancy.

Composition of proteoglycan fragments from hyaline cartilage produced by granulocytes in a model of frustrated phagocytosis.

An in vitro model of frustrated phagocytosis was developed in which granulocytes interact with well-defined slices of hyaline cartilage. The composition of the purified proteoglycan fragments released from the cartilage slices by N-formyl-methionyl-leucyl-phenylalanine-stimulated granulocytes was studied after 30, 60 and 90 min incubation time. It was shown that the proteoglycan fragments do not change their composition during incubation. The only change observed during incubation was an increase in the quantity of the fragments. The protein content of the proteoglycan fragments is 7.0-8.6%, corresponding to a peptide chain of 24-28 amino acids, and the relative molecular mass of the total fragment is Mr = 37,600-39,200. On average, each proteoglycan fragment contains two chondroitin sulphate chains (Mr = 22,000-22,400), every fourth fragment contains a keratan sulphate chain (Mr = 7000-7200) and every seventh to eighth contains an O-glycosidic oligosaccharide, whereas no N-glycosidic oligosaccharide could be detected. The results of the disaccharide analysis show that the galactosaminoglycan chains contain 76.2-83.6% chondroitin 4-sulphate, 12.9-19.4% chondroitin 6-sulphate, 3.5-3.8% chondroitin and no dermatan sulphate. Since composition and relative molecular mass of the chondroitin sulphate and keratan sulphate chains from the proteoglycan fragments resemble those of native proteoglycans, the conclusion may be drawn that the degeneration of the proteoglycans occurs by proteases that attack preferably the chondroitin sulphate-rich region of the core protein. This is the first inflammation model of joint destruction, which demonstrates the elution of soluble specific proteoglycan degradation products of defined size.

Secreted and cellular proteochondroitin sulfates of a human B lymphoblastoid cell line contain different protein cores.

Proteoglycans of the human B lymphoblastoid cell line LICR-LON-HMy2 were metabolically labeled with [35S]sulfate. High-density fractions of 35S-labeled material separated by CsCl gradient ultracentrifugation were further purified by anion exchange chromatography and gel filtration. Two proteoglycans, isolated from cell lysates and culture supernatants, were characterized by gel filtration and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in combination with enzymatic degradation. Treatment with chondroitinase AC completely degraded the glycosaminoglycan moiety of the proteoglycans. Three to 4 chondroitin sulfate chains (average molecular mass = 26 kDa) were estimated for each of the two proteoglycans. Differences between the proteochondroitin sulfates (CSPG) were observed in the content of N-linked oligosaccharides. After chondroitinase AC treatment the resulting band in SDS-PAGE of the secreted CSPG was sensitive to treatment with endoglycosidase F (Endo F) which further reduced the molecular mass from 30 to 21.5 kDa, whereas the band of the cellular CSPG after chondroitinase AC treatment (molecular mass = 30 kDa) remained resistant to Endo F treatment. The composition of amino acids was different in the protein cores, suggesting differences in the primary structure. Both CSPG contained a high percentage of glycine and serine. For both CSPG a molecular mass of approximately 135 kDa was deduced from the hydrodynamic sizes of the glycosaminoglycan chains obtained after alkaline/borohydride treatment and the migration of the protein/oligosaccharide complexes in SDS-PAGE. 75% of all [35S]sulfate-labeled molecules were found in the culture supernatant and 25% in the cellular fraction. 35S-Labeled material in the culture supernatant consisted exclusively of intact CSPG, whereas 35S-Labeled molecules in the cellular preparation consisted largely of free chondroitin sulfate chains. Only 8.3% of the cellular material, isolated from the microsomal fraction, was intact CSPG. In pulse-chase experiments maximal secretion of CSPG was found after 4 h, comprising approximately 40% of totally synthesized CSPG. From these experiments we tentatively conclude that a small proportion of CSPG synthesized by LICR-LON-HMy2 cells is membrane-associated, a larger portion is secreted, and another portion is intracellularly degraded.

Changes in the catalytic activities of proteoglycan-degrading lysosomal enzymes in parenchymal and non-parenchymal liver cells and in serum during the development of experimental liver fibrosis.

The catalytic activities of 4 glycosidases (hyaluronate-4-glycanohydrolase (EC 3.2.1.35), beta-N-acetyl-D-glucosaminidase (EC 3.2.1.30), beta-glucuronidase (EC 3.2.1.31), alpha-L-iduronidase (EC 3.2.1.76)), of the arylsulphatases A and B (EC 3.1.6.1) and of the protease cathepsin D (EC 3.4.23.5) were measured in extracts from hepatocytes and non-parenchymal cells and in serum during the development of thioacetamide-induced rat liver fibrosis (22 weeks). In non-parenchymal liver cells the catalytic activities of beta-N-acetyl-D-glucosaminidase, beta-glucuronidase, alpha-L-iduronidase and cathepsin D were increased significantly during chronic liver damage, but that of hyaluronate-4-glycanohydrolase was reduced by 40 to 65% during the period of application of thioacetamide. The catalytic activities of the arylsulphatases were lowered by 65% compared to control values in the 12th week but with advancing liver damage the catalytic activities returned to nearly normal values. Parenchymal cells of rats, which had been liver-damaged for 6 months, contained strongly elevated activities of beta-glucuronidase, beta-N-acetyl-D-glucosaminidase, arylsulphatases A and B, and cathepsin D but only slightly increased activities of hyaluronate-4-glycanohydrolase and alpha-L-iduronidase, respectively. In the serum of liver-damaged rats the activity of alpha-L-iduronidase was strongly elevated, while that of N-acetyl-beta-D-glucosaminidase was only slightly increased. The activities of beta-glucuronidase and of arylsulphatases A and B were decreased during the whole period of treatment. The catalytic functions of hyaluronate-4-glycanohydrolase and of cathepsin D, respectively, were decreased initially, but both enzyme activities were elevated during the more advanced stages of long term thioacetamide treatment.

Changes in the glycosaminoglycans distribution pattern in the human uterine cervix during pregnancy and labor.

The glycosaminoglycans distribution pattern of uterine cervix samples obtained from 42 women of reproductive age was determined by means of proteolytic digestion and subsequent chromatographic separation. The following glycosaminoglycans were detected: chondroitin 4- and 6-sulfates, dermatan sulfate, hyaluronate, chondroitin, and keratan sulfate. The connective tissue of the uterine cervix shows a characteristic distribution pattern with regard to glycosaminoglycans which does not correspond to that found in any other tissue studied so far. Based on dry weight, the content of keratan sulfate increases during pregnancy while the concentration of chondroitin remains unchanged. The chondroitin sulfates and dermatan sulfate drop simultaneously. During labor chondroitin increases threefold. The hyaluronate content of the postpartum cervix is higher than that of the cervix in nonpregnant women. Both changes in the solubility of collagen as well as in the distribution pattern of the glycosaminoglycans seem to be related to cervical dilatation.

PIP: This study attempted to quantitate the presence of glycosaminoglycans in the uterine cervix during pregnancy and labor. Distribution patterns, within the cervical samples, of the glycosaminoglycans were determined by proteolytic digestion and subsequent chromatographic separation. Cervical samples studied were from 42 women of reproductive age. During pregnancy, the keratan sulfate content increased significantly; the chondroitin concentration remained unchanged; and the contents of the chondroitin sulfates and dermatan sulfate decreased. During labor, a considerable rise of chondroitin was noted, whereas the concentrations of keratan sulfate, chondroitin 4- and 6-sulfates, and dermatan sulfate did not change. Hyaluronate content of a postpartum cervix was significantly higher than in nonpregnant women. The total glycosaminoglycans content (related to dry weight of the cervix) was not larger during pregnancy as compared with samples of nonpregnant women. During labor, however, there was an increase from 8.57 mcmol/gm of dry weight. The hydroxyproline content dropped during pregnancy and labor. The ratio of total glucosaminoglycans content to that of hydroxyproline changed in favor of the glycosaminoglycans during pregnancy, and it became more pronounced during labor. Distribution patterns of serial transverse cervical sections showed an increasing content of hyaluronate from the external os upward to the lower uterine segment. The other glucosaminoglycans did not show any definite trend. Both the changes in the solubility of collagen as well as glycosaminoglycans distribution patterns were related to cervical dilatation.

Studies on the polydispersity and heterogeneity of proteokeratan sulfate from calf and porcine cornea.

After proteolysis of the calf and porcine corneae with papain 66 (calf) and 56 (hog) polysaccharide-containing fractions were obtained by chromatography on Dowex 1X2 and fractionating precipitation with ethanol (calf) and by chromatography on Dowex 1X2 and CPC-cellulose (hog). The sulfatation degrees (mol sulfate/mol hexosamine) and molecular weights (Mw) of 13 peptidokeratan sulfate fractions from calf cornea and of 9 such fractions from porcine cornea were 0.41-1.25 (Mw 3200-21 500), and 0.42-1.41 (Mw = 4900-25 600), respectively. It was found that the sulfatation degree increases more than proportionally with the chain-length. More than 90% of total keratan sulfate in both cases contain 2-3 mannose and 5-9 amino acid molecules per peptidokeratan sulfate molecule. About 30% of the corepeptide amino acids were asparagine or aspartic acid. More than 90% of the peptidokeratan sulfates from porcine cornea exhibits chain lengths of 31-47 disaccharide units. Two oligosaccharide peptides were isolated from porcine cornea, both being sulfate-free. Besides a core peptide of about 9 amino acids (2 Asx) one of them showed a sugar composition very similar to that of the binding-region from corneal keratan sulfate accord. to Keller, R., Stein, T., Stuhlsatz, H.W., Greiling, H., Ohst, E., Müller, E., & Scharf, H.-D. (1981) Hoppe-Seyler's Z. Physiol. Chem. 362, 327-336: 3 Man, 2 Gal, 4 GlcNAc. The calculated molecular weight based on 3 mannose residues (Mw = 2 600) agrees with that determined by gel chromatography. The second oligosaccharide peptide was homogeneous on electrophoresis and contained the constituents of the linkage region from chondroitin sulfate beside those from keratan sulfate. Based on 4 mannose residues the calculated molecular weight (4 300) is in agreement with that from gel chromatography. The sequence Asn-X-Asx starting with the binding-Asn is postulated in the core protein of proteokeratan sulfate from amino acid analyses of the corneal peptidokeratan sulfate fractions.