Hyaluronic acid enhances peripheral nerve regeneration in vivo.

Hyaluronic acid has been shown to enhance peripheral nerve regeneration in vitro. It has been proposed that, during the fibrin matrix phase of regeneration, hyaluronic acid organizes the extracellular matrix into a hydrated open lattice, thereby facilitating migration of the regenerating axons. Hyaluronic acid solutions and saline control solutions were injected into a nerve guide spanning a transected gap in the sciatic nerve of Sprague-Dawley rats (five in each group). Nerve conduction velocities were measured at 4 weeks by electromyography (EMG) before sacrifice of the animals. These studies demonstrated increased conduction velocities in the hyaluronic acid group compared with control animals (P = 0.006). After the animals were sacrificed, regenerated axon cables were quantified histologically, and axon branching was delineated by retrograde tracer analysis. In addition, the hyaluronic acid group showed an increase in myelinated axon counts at 4 weeks (P= 0.03). An increase in retrograde flow was demonstrated in the hyaluronic acid groups compared with animals receiving saline solution.

Use of sodium hyaluronate in treating temporomandibular joint disorders: a randomized, double-blind, placebo-controlled clinical trial.

This study assessed the efficacy of high-molecular-weight sodium hyaluronate as a treatment for certain intracapsular temporomandibular joint (TMJ) disorders. One hundred twenty-one patients were studied at three test sites using a randomized, double-blind, placebo-controlled experimental design. Patients were selected on the basis of 1) confirmed diagnosis of either degenerative joint disease (DJD), reducing displaced disc (DDR), or nonreducing displaced disc (DDN); 2) nonresponsiveness to nonsurgical therapies; and 3) severe dysfunction as established by the Helkimo indices (HI), visual analog scales (VASs), and physical measurements of joint movement and joint noise (arthrophonometry [APM]). Subjects received a unilateral upper joint space injection of either 1) 1% sodium hyaluronate in physiologic saline (MedChem Products, Woburn, MA) or 2) USP physiologic saline. Clinical evaluations were performed using HI, VAS, and APM at weekly intervals for the first month and then at monthly intervals up to 6 months postinjection. Statistical analyses for both categorical and continuous variables were performed for each diagnostic category at each examination interval. For DJD, no difference in outcome was seen between treatment groups. For DDN, significant between-group differences were seen through 1 month; however, beyond this time point, the number of DDN patients was insufficient to draw meaningful conclusions concerning efficacy. For DDR, statistically significant within-group and between-group improvement in all three measures (HI, VAS, APM) was seen for the hyaluronate group compared to the saline group throughout the 6-month test period. At the month-2 and month-3 examination intervals, twice as many patients treated with hyaluronate (90%) showed improvement compared to patients given placebo. Further, only 3% of patients with DDR who were treated with hyaluronate relapsed compared with 31% of patients with DDR given placebo.

Chemical modification of hyaluronic acid by carbodiimides.

Hyaluronic acid (HA) is a linear polysaccharide with repeating disaccharide units of glucuronic acid and N-acetylglucosamine and is found in the extracellular matrix of connective tissues. Reaction of high molecular weight sodium hyaluronate (NaHA, MW approximately 2 x 10(6] with EDC at pH 4.75, either in the presence or absence of a primary diamine, gave the N-acylurea and O-acylisourea as NaHA-carbodiimide adducts. None of the expected intermolecular coupling with the amine component was observed. On the basis of this new observation, this method for chemical modification of HA was used in conjunction with new synthetic carbodiimides to prepare HA derivatives bearing lipophilic, aromatic, cross-linked, and tethered functional groups. The degree of conversion to NaHA-acylurea products appears to depend upon both the characteristics of various carbodiimides and the conformational structure of NaHA.

Isolation and some structure analyses of a copolymeric chondroitin sulfate-dermatan sulfate proteoglycan from post-burn, human hypertrophic scar.

A D-glucuronic acid rich, copolymeric chondroitin sulfate (CS)-dermatan sulfate (DS) proteoglycan (PG) from post-burn hypertrophic scar tissue (HSc) was obtained by DEAE-cellulose chromatography and differential ethanol fractionation, and further purified on a Sepharose CL-6B column. CS-DS-PG protein content was 14% (w/w). The amino-terminal amino acid sequence of the first ten residues was as follows: NH2-Asp-Glu-Ala-B-Gly-Ile-Gly-Pro-Glu-Val. This sequence is identical to that of human embryonic fibroblast cell (IMR-90) CS-DS-PG, as well as to human HSc-DS-PG. After chondroitinase ABC treatment, two peptides (Mr 22,000 and 16,000 daltons) were detected by sodium dodecyl sulfate-(polyacryl)amide gel electrophoresis (SDS-PAGE). ELISA analysis using rabbit antiserum raised against a synthetic peptide that contained 15 amino acids in the same sequence as the amino terminus of human fetal membrane PG showed significant reactivity with HSc CS-DS-PG. HSc CS-DS-PG had an apparent Mr of approximately 78,000 daltons, as determined by Sepharose CL-6B chromatography and SDS-PAGE. Alkaline borohydride treatment of CS-DS-PG liberated CS-DS glycosaminoglycan (GAG) chains having an Mr of 29,000 daltons. The conversion of xylose to xylitol indicated that the GAG chains are attached to the PG protein core at O-3 through a xylosyl-seryl linkage. CS-DS-PG also contained both N and O-linked oligosaccharides and did not aggregate with hyaluronic acid. These results, together with those reported previously, showed that HSc CS-DS-PG and DS-PG have the same A1-A15 amino acid sequence at the amino terminus but different protein cores. HSc CS-DS-PG was completely digested with chondroitinase AC and is, therefore, distinctly different from HSc DS-PG.

Kinetics of the chondrocyte biosynthetic response to compressive load and release.

To gain insight regarding the rate at which cartilage tissue can sense and respond to a dynamic mechanical stimulus, we have examined the time-course of changes in biosynthetic activity following both the application and release of a static compressive stress. Cartilage harvested from the reserve zone of calf epiphyseal plate was subjected to unconfined static compressive stresses of 0, 0.25 and 0.5 MPa. Incorporation of [35S]sulfate and [3H]proline was measured during loading periods of less than 1 to 26 h and after preloading periods of 0.5, 2 or 12 h. During loading, total incorporation decreased to steady levels with time constants estimated to be 0.25-4 h (proline) and 1-5 h (sulfate). Proline incorporation exceeded control levels for 3 h after release of a 2 or 12 h preload. Sulfate incorporation remained depressed for at least 4 h after release of a 12 h preload and remained at control levels following release of 0.5 and 2 h preloads. We conclude that the modulation of proline incorporation by both loading and load release is faster than the modulation of sulfate incorporation. Furthermore, the response to unloading is not just the inverse of the response to loading; this nonlinearity suggests that the response to dynamic loading would not be determined simply by the time average component of the dynamic load.

Age-related changes in the synthesis of matrix macromolecules by bovine articular cartilage.

Calf and mature cow articular cartilage was labeled in vitro with [35S]SO4 and [3H]glycine and kinetics of incorporation of both isotopes by cartilage fragments was determined by scintillation spectroscopy. The cartilage fragments were then extracted in sequence with 4M GuHCl (Guanidium chloride) and pepsin. The pepsin digest was adjusted to 1.3 M NaCl and pepsin-solubilized collagen salted out. The 4M GuHCl extract, collagen and pepsin-resistent residue were then freeze-dried. The 4M GuHCl extract was further fractionated by DEAE (Diethylaminoethyl) 52 ion exchange chromatography to obtain protein and PG (Proteoglycan) fractions. The protein fraction was also characterised by SDS-PAGE and PG fraction by Sepharose C1-2B chromatography under associative conditions in the presence and absence of an exogenous HA (Hyaluronic acid). The GAG (Glycosaminoglycan) side chains of the PG samples were analysed by Sephadex G-200 column chromatography and their composition determined by paper chromatography after chondroitinase ABC digestion. Linear incorporation of both isotopes was observed from 1 to 18 hours of incubation and roughly equal amounts of [35S]SO4 counts were found on per cell bases in both cartilages although less [3H]glycine was incorporated by cow chondrocytes. It was also found that calf chondrocytes synthesize much greater proportion of the collagen whereas the cow cells synthesize PGs of smaller hydrodynamic sizes, bearing shorter GAG side chains that are enriched in KS (Keratan sulfate) and Ch-6S (Chondroitin-6 sulfate isomer). A failure of cow 35S-PGs monomers to interact with an exogenous HA in the presence of other extracted components was also demonstrated. The relevance of these findings for the mechanism of cartilage damage in aging and osteoarthritis is discussed.

Isolation and partial characterization of dermatan sulfate proteoglycans from human post-burn scar tissues.

Dermatan sulfate (DS) proteoglycans (PGs) were extracted from human post-burn scar (Sc) tissues with 4M guanidinium chloride and isolated from the extracts by DEAE-cellulose chromatography and by differential ethanol precipitation. The DS.PGs were further purified by Sepharose CL-6B column chromatography. The average molecular weight (Mr) of hypertrophic scar (HSc) tissue DS.PGs was 39,000 based on sedimentation equilibrium measurements. Alkaline borohydride treatment of DS.PGs liberated glycosaminoglycan (GAG) chains and the presence of xylitol indicated that these chains were attached to protein core by xylosyl residues. The average Mr of the DS.GAG chain from HSc and normal scar (NSc) samples were 23,500 and 20,000 respectively. After digestion of the HSc and NSc, DS.PGs with chondroitinase ABC in the presence of proteinase inhibitors, two peptide components with Mr values of 21,500 and 17,000 were detected by SDS-polyacrylamide gel electrophoresis using reducing conditions. Analysis of the protein core fractions derived from NSc and HSc DS.PGs by Sepharose CL-6B column chromatography showed the presence of a single NH2-terminal amino acid (aspartic acid) and also that the fractions with different KAV values had an identical NH2-terminal sequence (A1-A5). The A1-A23 sequence of NSc DS.PG (major fraction, C): NH2Asp-Glu-Ala-O-Gly-Ile-Gly-Pro-Glu-Val-Pro-Asp-Asp-Arg-Asp-Phe-G lu-Pro- Ser-Leu-Gly-Pro-Val was the same as reported for a DS.PG isolated from human fetal membrane (HFM) tissue (Brennan et al., 1984). ELISA inhibition assay using monoclonal antibodies raised in rabbit against the NH2-terminal peptide (containing 15 amino acids) of human fetal membrane tissue were found to cross-react with HSc and NSc DS.PGs. Monoclonal antibodies to bovine skin DS.PGs protein core (Pearson et al., 1983) did not show any cross-reactivity with scar DS.PGs. These results show that the scar DS.PGs described here are different from normal bovine skin DS.PGs in the size and type of the protein core, and that in all the samples, the peptide components have the same NH2-terminal amino acid sequence.

The structures of N- and O-glycosidic carbohydrate chains of a chondroitin sulfate proteoglycan isolated from the media of the human aorta.

A large Mr chondroitin sulfate proteoglycan was extracted from the media of human aorta under dissociative conditions and purified by density-gradient centrifugation, ion-exchange chromatography, and gel filtration chromatography. Removal of a contaminating dermatan sulfate proteoglycan was accomplished by reduction, alkylation and rechromatography on the gel filtration column. After chondroitinase ABC treatment, the proteoglycan core was separated from a residual heparan sulfate proteoglycan by a third gel filtration chromatography step. As assessed by radioimmunoassay, the isolated proteoglycan core was free of link protein, but possessed epitopes that were recognized by antisera against the hyaluronic acid binding region of bovine cartilage proteoglycan as well as those that were weakly recognized by anti-keratan sulfate antisera. Following beta-elimination of the protein core, the liberated low Mr oligosaccharides were partially resolved by Sephadex G-50 chromatography, and their primary structure was determined by 500-MHz1H NMR spectroscopy in combination with compositional sugar analysis. The N-glycosidic carbohydrate chains, which were obtained as glycopeptides, were all biantennary glycans containing NeuAc and Fuc; microheterogeneity in the NeuAc----Gal linkage was detected in one of the branches. The N-glycosidic glycans have the following overall structure: (Formula: see text). The majority of the O-glycosidic carbohydrate chains bound to the protein core were found to be of the mucin type. They were obtained as glycopeptides and oligosaccharide alditols, and possessed the following structures: NeuAc alpha(2----3)Gal beta(1----3)GalNAc-ol, [NeuAc alpha(2----3)Gal beta(1----3)[NeuAc alpha(2----6)]GalNAc-ol, and NeuAc alpha-(2----3) Gal beta(1----3)[NeuAc alpha(2----3)Gal beta(1----4)GlcNAc beta(1----6)] GalNAc-ol. The remainder of the O-glycosidic carbohydrate chains bound to the isolated proteoglycan were the hexasaccharide link regions of the chondroitin sulfate chains that remained after chondroitinase ABC treatment of the native molecule. These latter glycans, which were obtained as oligosaccharide alditols, had the following structure (with GalNAc free of sulfate or containing sulfate bound at either C-4 or C-6): delta 4,5GlcUA beta(1----3)GalNAc beta(1----4)GlcUA beta(1----3)Gal beta(1----3)Gal beta(1----4)Xyl-ol.

A comparison of glycosaminoglycan synthesis by human fibroblasts from normal skin, normal scar, and hypertrophic scar.

Fibroblasts isolated from normal skin, normal scar, and hypertrophic scar tissues were compared with respect to their growth curves, protein contents, and abilities to synthesize glycosaminoglycans (GAGs). While no significant differences were found with respect to protein content or population doubling times, we did find significant differences in the proportions of radiolabel incorporated into the various GAGs among the 3 groups of cell lines. Using a dual-label technique to label both hyaluronic acid and the sulfated GAGs, we isolated labeled constituents from the extracellular, the pericellular, and the cellular fractions by pronase digestion and gel filtration and identified the various GAGs by electrophoresis and selective digestion with enzymes. Of the GAGs isolated from the extracellular fraction, hypertrophic scar fibroblasts incorporated proportionately more 35S into chondroitin sulfate and less into heparan sulfate and more [3H]glucosamine into hyaluronic acid than did normal skin fibroblasts. Of the GAGs isolated from the cellular fraction, hypertrophic scar fibroblasts incorporated proportionately more 35S into heparin and less into dermatan sulfate and more [3H]glucosamine into hyaluronic acid than did normal skin fibroblasts. These differences in biosynthesis may help to explain the differences in GAG content in skin and scars found in vivo and to give insight into the development of hypertrophic scars.

Studies on human scar tissue proteoglycans.

The proteoglycans (PGs) in pooled normal scars and pooled hypertrophic scars were extracted with 4 M guanidinium chloride and isolated by DEAE-cellulose chromatography. The PG samples were then fractionated further by dissociative CsCl density gradient sedimentation. Following cleavage of the density gradient PG fractions with alkaline NaB3H4, the glycosaminoglycan (GAG) constituents were isolated and analyzed by quantitative cellulose acetate electrophoresis. In addition, single samples of normal skin and a keloid scar were also analyzed. The results showed that the hypertrophic scars had a higher average content of extractable and also residual PGs than did the normal scars but a wide range of values was obtained for each type of scar. Some differences were noted in the amounts and distribution of the GAGs in CsCl gradient fractions, in the different types of scar tissue. The PGs in tissues were distributed in low-, medium-, and high density fractions and are, therefore, heterogeneous. Dermatan sulfate (DS) was the major GAG in each tissue and smaller quantities of chondroitin sulfate (CS), heparan sulfate (HS), and heparin (HP) were also present. In addition, 2 other GAG constituents were also detected. Based on the susceptibility of these GAGs to enzymes and nitrous acid treatments, one was a HS or HP while the second was a DS. The major differences in the PG composition of the scar tissues were the higher proportions of low-density CS-PGs in the keloid scar and of low density DS-PGs in hypertrophic and keloid scars.

Mechanical analysis of hypertrophic scar tissue: structural basis for apparent increased rigidity.

The mechanical behavior of normal human skin and hypertrophic scar tissue (HST) is compared using constant-strain-rate and successive stress-relaxation uniaxial loading programs in vitro. HST is less extensible, requires more energy to be stretched in the physiologic range, and stores strain energy less efficiently than normal skin. The explanations for the differences observed between the mechanical behavior of normal skin and HST are based on the differences in their composition and structure. We suggest that the collagen fiber network is partially "prealigned" in a crimped tendon-like organization in HST, which reduces its extensibility and raises the strain energy required to stretch it. It is further hypothesized that an incomplete elastic fiber network, an abnormal glycosaminoglycan content, and/or abnormal collagen fiber slippage are responsible for the reduced capacity to return strain energy in the hypertrophic scar tissue. The results of these studies indicate that although HST has been described as stiffer than normal skin, the maximum stiffness of skin and HST are similar. The "apparent" increased rigidity of HST is a result of reduced extensibility rather than a change in its stiffness. This inexensibility may manifest itself by limiting joint mobility in the patient with HST.

The molecular structure and lubricating activity of lubricin isolated from bovine and human synovial fluids.

Lubricin was isolated from bovine ankle, metacarpophalangeal and knee and human knee synovial fluids. The lubricins isolated from the bovine joint fluids had the same amino acid and carbohydrate compositions, but differences were observed in the relative molecular masses. The Mr values of bovine metacarpophalangeal and ankle lubricin determined by light-scattering measurements were about 200 000, whereas values of 132 000 and 143 000 were obtained for the bovine knee lubricin. The human knee lubricin had a similar carbohydrate composition to bovine knee lubricin except for the higher glucosamine content, and the amino acid composition differed slightly. The human sample had a lower glutamic acid content and a leucine/isoleucine ratio of 2:1 compared with 1:1 in the bovine. The Mr value of the human knee lubricin (166 000) was also lower than that of the bovine metacarpophalangeal and ankle samples. The Mr value of the bovine knee lubricin determined by sedimentation-equilibrium measurements was 171 000. The length measurements determined by electron microscopy and also the sedimentation measurements showed considerable polydispersity and indicate that the degree of extension of lubricin molecules can vary. Friction measurements showed that the human knee synovial-fluid lubricin had equivalent lubricating ability in a test system in vitro to that observed for lubricin isolated from normal bovine synovial fluids. The lubricating ability of lubricin was concentration-dependent, and each lubricin sample was able to act as a lubricant in vitro in an equivalent manner to whole synovial fluid at concentrations that are thought to occur in vivo.

On the structure of bovine articular cartilage high density proteoglycans. Isolation of the keratan sulfate and chondroitin sulfate side chains.

The structure of adult bovine articular cartilage high density proteoglycans (PG-I) was studied by degradation with Pronase, chondroitinase ABC, and alkaline borohydride treatments and fractionation and analysis of the products. The keratan sulfate (KS) peptides were rich in glutamic acid, proline, and serine and had a low glycine content. The chondroitin sulfate (CS) peptides had a high content of serine, glycine, and glutamic acid and a much lower proline content than the KS peptides. The data indicate that the KS and CS chains occur in more distinct regions of the protein core(s) than in bovine nasal cartilage PG. After alkaline borohydride treatment there was an almost quantitative conversion of xylose to xylitol and galactosaminitol was the only hexosaminitol detected in KS fractions. The results obtained indicated that the alkali-labile bonds linking the CS and KS chains are the same as those reported to occur in other cartilage PGs. The Mr of the KS chains calculated from the glucosamine and galactosaminitol contents gave values of 6,000-7,000, although gel chromatography and light scattering measurements indicated considerable heterogeneity. The KS and CS chains were quantitatively precipitated by cetylpyridinium chloride and the KS and a portion (15%) of the CS chains were found to be soluble in 1% cetylpyridinium chloride. The abnormal solubility properties of the CS chains in the presence of 1% cetylpyridinium chloride is thought to be due to their low sulfate content. The molecular weight of the remainder of the CS chains, based on the ratio of xylitol to galactosamine, varied from 6,500 to 16,000. The low Mr CS chains were rich in 6-sulfated disaccharides whereas the higher Mr chains had a higher content of 4-sulfated disaccharides. The ratio of galactose to xylitol also varied with Mr. These results indicate similarities in the structure of the adult bovine articular cartilage PG-Is to other cartilage high density PGs. The heterogeneities observed in the composition of the KS and CS chains, and their occurrence in relatively distinct regions of the protein core(s) indicate, however, that there is still much to be learned about the structure of these complex macromolecules.

The lubricating activity of human synovial fluids.

The lubricating abilities of human synovial fluids were measured using a rotating cartilage-on-glass apparatus. A total of 247 human fluids were lubrication-tested. Of these, 20 of the 180 knee fluids from patients with degenerative/traumatic joint disease lubricated less well than normal bovine synovial fluid. The remainder of the fluids from the knee and other joints were equivalent to normal bovine synovial fluid in their lubricating properties. The concentrations of hyaluronic acid, protein, and sialic acid and the relative viscosity of 117 human fluid samples were determined, but no relationships between the gross composition and the lubricating abilities were apparent.

Glycosaminoglycans of normal and hypertrophic human scar.

The glycosaminoglycan (GAG) contents of hypertrophic scars, normal scars, and human skin from cadavers of matched ages were compared. Cellulose acetate electrophoresis, chondroitinase digestions, and reaction product and infrared analyses were used to characterize the component GAGs. DEAE-cellulose chromatography was used to separate hyaluronic acid (HA) and sulfated GAGs. Chondroitinase analysis was improved under these conditions. HA was determined enzymatically. Results showed an elevation of HA in hypertrophic scar. Dermatan sulfate was the major GAG in both scars and a slightly greater quantity was observed in the hypertrophic scar. Small amounts of chondroitin 4-sulfate and chondroitin 6-sulfate disaccharide constituents were also detected by the chondroitinase assay method and these were also elevated in hypertrophic scar. These results suggest that the GAGs of hypertrophic scar differ from normal scar and normal skin.

Comparison of the thermal stabilities and solubilities of human hypertrophic scar, normal scar and normal skin collagens.

The thermal stability and solubility profiles of human hypertrophic scar collagen have been measured and compared with those of the collagens present in human normal scar, human normal skin and a lathyritic dermal tissue. The thermal stability profiles of the hypertrophic scar collagens were found to be very similar to each other and to closely resemble that of lathyritic collagen but differ significantly from those obtained with the other dermal collagens. The hydrothermal solubility profiles did not, however, show a clear distinction between normal and hypertrophic scars.

Age-related changes in chemical composition and physical properties of mucus glycoproteins from rat small intestine.

Mucus glycoproteins from newborn and adult rat small intestine were radiolabelled in vivo with Na2 35SO4 and isolated from mucosal homogenates by using Sepharose 4B column chromatography followed by CsCl-density-gradient centrifugation. Non-covalently bound proteins, lipids and nucleic acids were not detected in the purified glycoproteins. Amino acid, carbohydrate and sulphate compositions were similar to chemical compositions reported for other intestinal mucus glycoproteins, as were sedimentation properties. There were, however, important differences in the chemical and physical characteristics of the mucus glycoproteins from newborn and adult animals. The buoyant density in CsCl was higher for the glycoproteins from newborn rats (1.55 g/ml versus 1.47 g/ml). On sodium dodecyl sulphate/polyacrylamide/agarose-gel electrophoresis, the glycoprotein from newborn rats had a greater mobility than the adult-rat sample. Although both preparations had similar general amino acid compositions, variations were observed for individual amino acids. The total protein content was greater in the glycoprotein from newborn animals (27%, w/w, versus 18%, w/w). The molar ratio of carbohydrate to protein was less in the newborn, primarily owing to a decreased fucose and N-acetylgalactosamine content. Comparison of the molar ratio of fucose and sialic acid to galactose for both glycoproteins demonstrated a reciprocal relationship similar to that described by Dische [(1963) Ann. N.Y. Acad. Sci. 106, 259-270]. The sulphate content was greater in the glycoprotein from newborn rats (5.5%, w/w, versus 0.9%, w/w). Both had similar sedimentation coefficients in a dissociative solvent. These results suggest an age-related difference in the types of mucus glycoproteins synthesized by small intestine.

Isolation and partial characterization of low density proteoglycans from bovine articular cartilage.

Low density proteoglycans (PG-III) were isolated from bovine articular cartilage by extraction with 4 M guanidinium chloride followed by sedimentation in a dissociative CsCl density gradient and fractionation by chromatography on DEAE-cellulose and Sepharose CL-6B columns. The PG fractions obtained were analyzed to determine the amino acid composition, the content of sulfate and carbohydrate constituents, the molecular weight, and sedimentation properties under associative and dissociative conditions and the types of glycosaminoglycan components. The results show that the major type of low density PG in adult bovine metacarpophalangeal cartilage is a proteochondroitin sulfate with a Mr approximately 44,000. A single glycosaminoglycan component was detected following alkaline borohydride cleavage that was completely digested by chondroitinase ABC treatment. The disaccharide composition of this glycosaminoglycan was 4% O-sulfate, 22% 6-sulfate, and 74% 4-sulfate. Variations were observed in the content of chondroitin sulfate and the other carbohydrate constituents, indicating that the low density PG in this tissue probably occurs as a family of molecules with a variable composition. Sedimentation velocity analysis showed that under associative conditions PG-III formed high molecular weight complexes by self-association.