Search

Decorin-binding protein A (DBPA) is an important surface adhesin of the bacterium Borrelia burgdorferi, the causative agent of Lyme disease. DBPA facilitates the bacteria's colonization of human tissue by adhering to glycosaminoglycan (GAG), a sulfated polysaccharide. Interestingly, DBPA sequence variation among different strains of Borrelia spirochetes is high, resulting in significant differences in their GAG affinities. However, the structural mechanisms contributing to these differences are unknown. We determined the solution structures of DBPAs from strain N40 of B. burgdorferi and strain PBr of Borrelia garinii, two DBPA variants whose GAG affinities deviate significantly from strain B31, the best characterized version of DBPA. Our structures revealed that significant differences exist between PBr DBPA and B31/N40 DBPAs. In particular, the C-terminus of PBr DBPA, unlike C-termini from B31 and N40 DBPAs, is positioned away from the GAG-binding pocket and the linker between helices one and two of PBr DBPA is highly structured and retracted from the GAG-binding pocket. The repositioning of the C-terminus allowed the formation of an extra GAG-binding epitope in PBr DBPA and the retracted linker gave GAG ligands more access to the GAG-binding epitopes than other DBPAs. Characterization of GAG ligands' interactions with wild-type (WT) PBr and mutants confirmed the importance of the second major GAG-binding epitope and established the fact that the two epitopes are independent of one another and the new epitope is as important to GAG binding as the traditional epitope. [ABSTRACT FROM AUTHOR]

The structure of the GAG (glycosaminoglycan) chain of recombinantly expressed decorin proteoglycan was examined using a combination of intact-chain analysis and domain compositional analysis. The GAG had a number-average molecular mass of 22 kDa as determined by PAGE. NMR spectroscopic analysis using two-dimensional correlation spectroscopy indicated that the ratio of glucuronic acid to iduronic acid in decorin peptidoglycan was 5 to 1. GAG domains terminated with a specific disaccharide obtained by enzymatic degradation of decorin GAG with highly specific endolytic and exolytic lyases were analysed by PAGE and further depolymerized with the enzymes. The disaccharide compositional profiles of the resulting domains were obtained using LC with mass spectrometric and photometric detection and compared with that of the polysaccharide. The information obtained through the disaccharide compositional profiling was combined with the NMR and PAGE data to construct a map of the decorin GAG sequence motifs.

The interactions of the ECM (extracellular matrix) protein asporin with ECM components have previously not been investigated. Here, we show that asporin binds collagen type I. This binding is inhibited by recombinant asporin fragment LRR (leucine-rich repeat) 10-12 and by full-length decorin, but not by biglycan. We demonstrate that the polyaspartate domain binds calcium and regulates hydroxyapatite formation in vitro. In the presence of asporin, the number of collagen nodules, and mRNA of osteoblastic markers Osterix and Runx2, were increased. Moreover, decorin or the collagen-binding asporin fragment LRR 10-12 inhibited the pro-osteoblastic activity of full-length asporin. Our results suggest that asporin and decorin compete for binding to collagen and that the polyaspartate in asporin directly regulates collagen mineralization. Therefore asporin has a role in osteoblast-driven collagen biomineralization activity. We also show that asporin can be expressed in Escherichia coli (Rosetta-gami) with correctly positioned cysteine bridges, and a similar system can possibly be used for the expression of other SLRPs (small LRR proteoglycans/proteins).

Fibulin-5 plays an important role in elastic fibre formation in vivo. We have investigated the molecular interactions between fibulin-5 and components of fibrillin-rich microfibrils which form a template for elastin. Fibulin-5 interacted in a dose-dependent manner with a fibrillin-1 N-terminal sequence and with tropoelastin, but not with MAGP-1 (microfibril-associated glycoprotein-1) or decorin. Fibulin-5 did not inhibit interactions between fibrillin-1 N- and C-terminal fragments, or fibrillin-1 interactions with tropoelastin. Fibulin-5 may provide a link between tropoelastin and microfibrils in the pericellular space during elastic fibre assembly.

The aim of the present study was to characterize the proteoglycans and catabolic products of proteoglycans present in the tensile region of ligament and explant cultures of this tissue, and to compare these with those observed in the tensile region of tendon. Approx. 90% of the total proteoglycans in fresh ligament was decorin, as estimated by N-terminal amino acid sequence analysis. Other species that were detected were biglycan and the large proteoglycans versican (splice variants V(0) and/or V1 and/or V2) and aggrecan. Approx. 23% of decorin detected in the matrix was degraded. Intact decorin and decorin fragments similar to those observed in the matrix that retained the N-terminus were also observed in the medium of ligament cultures. Intact biglycan core protein was detected in the matrix and medium of ligament cultures, and two fragments originating from the N-terminal region of biglycan were observed in the matrix of cultured ligament. Versican and versican fragments that retained the N-terminus of versican core protein were detected in fresh matrix and medium of tendon cultures. Approx. 42% of versican present in the fresh ligament was degraded. Aggrecan catabolites appearing in the culture medium were derived from aggrecanase cleavage of the core protein. An intact link protein and a degradation product from the N-terminal region of type XII collagen were also detected in the medium of the ligament explant.

Transforming growth factor-beta (TGFbeta) is a key mediator of extracellular matrix (ECM) accumulation in sclerotic kidney diseases such as diabetic nephropathy. One of the main target cells for TGFbeta in the kidney are glomerular mesangial cells, which respond by increasing expression of ECM proteins, such as collagens, laminin and fibronectin, while suppressing the expression of ECM-degrading proteases and increasing the synthesis of ECM protease inhibitors, including plasminogen activator inhibitor-1. Previous studies have shown that exposure of mesangial cells to chronic high-glucose conditions, such as those seen in diabetes, increases ECM deposition in a mechanism involving glucose-mediated up-regulation of TGFbeta expression. Naturally occurring inhibitors of this TGFbeta-dependent fibrotic response include decorin, a small leucine-rich proteoglycan. While the mechanism by which TGFbeta stimulates gene expression via the Smad signal-transduction pathway is becoming clear, the precise mechanism by which decorin may impinge upon TGFbeta activity remains to be established. In this study, for the first time we provide evidence that decorin can disrupt glucose- and TGFbeta/Smad-dependent transcriptional events in human mesangial cells through a mechanism that involves an increase in Ca(2+) signalling, the activation of Ca(2+)/calmodulin-dependent protein kinase II and ensuing phosphorylation of Smad2 at Ser-240. We show that decorin also induces Ser-240 phospho-Smad hetero-oligomerization with Smad4 and the nuclear localization of this complex independently of TGFbeta receptor activation. Thus, in human mesangial cells, the mechanism of decorin-mediated inhibition of TGFbeta signalling may involve activation of Ca(2+) signalling, the subsequent phosphorylation of Smad2 at a key regulatory site, and the sequestration of Smad4 in the nucleus.

It has been reported that the disaccharide composition of dermatan sulphate shows transient changes after epicutaneous application of the hapten 2,4-dinitrofluorobenzene to mouse skin, and that these changes are most conspicuous in healing skin on day 15 after chemical insult [Kuwaba, Nomura, Irie and Koyama (1999) J. Dermatol. Sci. 19, 23-30]. In the present study it was found that the molecular size of dermatan sulphate was increased on day 15 after hapten application. The molecular size of decorin increased in healing skin, whereas the size of dermatan-sulphate-depleted core protein did not increase. The length and localization of decorin dermatan sulphate were investigated by electron microscopy. Dermatan sulphate filaments oriented orthogonally to collagen fibrils were longer in healing skin than in control skin. In control skin, dermatan sulphate filaments were found among tightly packed collagen fibrils. In contrast, the interfibrillar gaps between each collagen fibril were enlarged in healing skin; elongated dermatan sulphate filaments extended from the surface of collagen fibrils across the enlarged gap. These results suggest that the increase in molecular size of decorin dermatan sulphate is important in organizing collagen fibrils separated by enlarged interfibrillar gaps in healing skin.

The genomic structure of murine fibromodulin was determined, and its age-related expression and distribution were characterized in knee epiphyses, with decorin studied for reference. Fibromodulin, as well as decorin, have roles in collagen fibrillogenesis both in vitro and in vivo. The murine fibromodulin gene, Fmod, was similar with that in other species, with three exons and 86% of the translated sequence in exon 2. The 2.7 kb long cDNA contains an open reading frame of 1131 nt. Fibromodulin mRNA levels were highest in tissues rich in fibrillar collagens type I or type II. During growth, the distribution of fibromodulin mRNA was similar with that of type II collagen, with the highest levels between 5 days and 1 month of age. Thereafter, the expression of type II collagen declined to a level near the detection limit, whereas the fibromodulin expression decreased less markedly to a level of approx. 35% of maximum, and remained constant throughout the rest of the observation period. In contrast, decorin mRNA levels were the highest in old animals. Pericellular deposition of fibromodulin was strong around the late-hypertrophic chondrocytes of the secondary ossification centre and in the growth plate. In young epiphyses, both fibromodulin and decorin were found interterritorially, mainly in the uncalcified and deep-calcified cartilage. In the old mice, calcified cartilage became enriched with regard to fibromodulin, while, in contrast, decorin deposition diminished, particularly near the tidemark. In the subchondral bone trabeculae, decorin was found in the endosteum of growing, but not in the mature, epiphyses. Differences in the expression and distribution profiles suggest different roles for fibromodulin and decorin in the regulation of collagen fibrillogenesis, maintenance of the fibril organization and matrix mineralization. As fibromodulin is deposited closer to cells than decorin, it may have a primary role in collagen fibrillogenesis, whereas decorin might be involved in the maintenance of fibril structures in the interterritorial matrix.

We have examined the catabolism of the proteoglycans aggrecan, decorin and biglycan in fresh tendon samples and in explant cultures of tissue from the tensional and compressed regions of young and mature bovine tendons. A panel of well-characterized antibodies that recognize glycosaminoglycan or protein (linear or neoepitope) sequences was used to detect proteoglycans and proteoglycan degradation products that were both retained within the tissue and released into the culture medium. In addition, a reverse-transcriptase-mediated PCR analysis was used to examine the mRNA expression patterns of tendon proteoglycans and aggrecanases. The results of this study indicate a major role for aggrecanase(s) in the catabolism of aggrecan in bovine tendon. The study also provides a characterization of glycosaminoglycan epitopes associated with the proteoglycans of tendon, illustrating age-related changes in the isomers of chondroitin sulphate disaccharides that remain attached to the core protein glycosaminoglycan linkage region after digestion with chondroitinase ABC. Evidence for a rapid turnover of the small proteoglycans decorin and biglycan was also observed, indicating additional molecular pathways that might compromise the integrity of the collagen matrix and potentially contribute to tendon dysfunction after injury and during disease.

Decorin is a small leucine-rich extracellular matrix proteoglycan, the expression of which is down-regulated in proliferating and malignantly transformed cells. In the present study we show that the expression of decorin in fibroblasts is suppressed by epidermal growth factor (EGF) and PMA, and that the effect of both is potently inhibited by blocking the extracellular signal-regulated protein kinase (ERK)1,2 signalling pathway (Raf/MEK1,2/ERK1,2) with the specific MAPK/ERK kinase (MEK)1,2 inhibitor, PD98059. In addition, specific activation of ERK1,2 by adenovirus-mediated expression of constitutively active MEK1 in dermal fibroblasts results in marked reduction in decorin mRNA abundance and production. Co-transfection of NIH-3T3 fibroblasts with human decorin promoter/chloramphenicol acetyltransferase (CAT) construct (pDEC--879/CAT) in combination with the expression vectors for constitutively active Raf-1 and MEK1 markedly suppressed decorin promoter activity. Co-transfections of human decorin promoter 5'-deletion constructs with constitutively active MEK1 expression vector identified the region -278 to -188 as essential for ERK1,2 mediated down-regulation of decorin promoter activity. These results show that activation of the ERK1,2 signalling pathway by a mitogenic growth factor, a tumour promoter or transformation suppresses decorin gene expression in fibroblasts, which in turn may promote proliferation and migration of normal and malignant cells.

Receptor-mediated endocytosis of decorin depends on its core-protein-mediated interaction with a 51 kDa membrane protein, which, in addition to its core-protein-binding site, carries a binding site for glycosaminoglycan chains. Membrane-associated heparan sulphate as well as heparin are known to have an inhibitory effect on decorin endocytosis by cultured skin fibroblasts. In this study, structural features of both glycosaminoglycans required for binding to the 51 kDa protein and for inhibiting decorin endocytosis, were investigated. Upon digestion of [(3)H]glucosamine-labelled heparan sulphate with heparinase III, dodeca- and higher saccharides were able to interact with the receptor protein. In comparison with unbound fragments of the same size, bound fragments were enriched in N-sulphated disaccharides carrying one or two sulphate ester groups. Using heparinase III-generated fragments from [(35)S]sulphate-labelled heparan sulphate chains, binding of fragments as small as octasaccharides could be detected. Competition experiments between dermatan sulphate and chemically modified heparin revealed that N- and 6-O-sulphation of glucosamine residues are important structural elements for binding to the receptor, whereas iduronate-2-O-sulphate groups contribute to binding only to a limited extent. However, with respect to the inhibition of decorin endocytosis, 2-O-desulphation had a quantitatively similar effect to 6-O-desulphation. Furthermore, for maximal inhibition of decorin endocytosis, longer fragments were required than for binding to the receptor. Thus, it appears that heparin/heparan sulphate has to interact with additional component(s) for effective inhibition of decorin uptake.

A bovine nasal-cartilage culture system has been utilized to analyse the catabolic events occurring in response to interleukin-1beta over a 14-day period. An early event following the start of interleukin-1 treatment was the release of glycosaminoglycan into the culture medium. This release was accompanied by the appearance in the tissue, and shortly thereafter also in the culture media, of a globular domain (G1)-containing aggrecan degradation product generated by the action of aggrecanase. Link protein was also released from the cartilage with a similar timeframe to that of the G1 fragment, although there was no evidence of its proteolytic degradation. By comparison with aggrecan, the small leucine-rich repeat proteoglycans decorin, biglycan and lumican showed a resistance to both proteolytic cleavage and release throughout the culture period. In contrast, fibromodulin exhibited a marked decrease in size after day 4, presumably due to proteolytic modification, but the major degradation product was retained throughout the culture period. Also in contrast with the early changes in the components of the proteoglycan aggregate, type II collagen did not display signs of extensive degradation until much later in the culture period. Collagen degradation products compatible with collagenase action first appeared in the medium by day 10 and increased thereafter. These data demonstrate that the leucine-rich repeat proteoglycans are resistant to proteolytic action during interleukin-1-stimulated cartilage catabolism, compared with aggrecan. This resistance and continued interaction with the surface of the collagen fibrils may help to stabilize the collagen fibrillar network and protect it from extensive proteolytic attack during the early phases of cartilage degeneration.

Dermatopontin, a recently found low-molecular-mass component of the extracellular matrix, was studied for its interaction with decorin and transforming growth factor beta (TGF-beta) and its influence on TGF-beta bioactivity. Dermatopontin reacted with decorin with an apparent Kd of 100 nM in a solid-phase assay. Dermatopontin inhibited the formation of the decorin-TGF-beta1 complex. Decorin also competed with dermatopontin for the binding of this cytokine. The dermatopontin-decorin complex bound 3-fold more TGF-beta1 than did each component individually, and binding was inhibited more strongly by decorin preincubated with dermatopontin than by dermatopontin or decorin alone. Dermatopontin augmented the biological activity of TGF-beta1, as analysed by the expression of luciferase in mink lung epithelial cells transfected with a plasminogen activator inhibitor-promoter-luciferase construct, although dermatopontin itself did not show apparent induction of luciferase. Dermatopontin showed weak inhibitory activity on the proliferation of mink lung epithelial cells, and it enhanced the growth-inhibitory activity of TGF-beta on these cells. Thus dermatopontin increases the cellular response to TGF-beta. These findings strongly suggest that dermatopontin modifies the behaviour of TGF-beta through interaction with decorin in the microenvironment of the extracellular matrix in vivo.

Cloned immortalized MC615 mouse chondrocytic cells were used to examine their capability to produce multiple types of matrix proteoglycans. Immunofluorescence staining indicated a uniform expression of aggrecan, biglycan and decorin by all cells. After culture with [35S]sulphate, proteo[35S]glycans secreted by the cells were found to elute in two peaks from a Sepharose CL-4B column. The first peak, at the void volume of the column, contained a large proteoglycan with an estimated average hydrodynamic mass of 10(3) kDa. The glycosaminoglycan chains of this proteoglycan had an average hydrodynamic size of 17 kDa, estimated by Sepharose CL-6B chromatography, indicating the presence of 30-70 glycosaminoglycan chains per core protein, which was consistent with the characteristics of aggrecan. Biglycan and decorin were immunoisolated from the second Sepharose CL-4B peak, and had average glycosaminoglycan hydrodynamic sizes of approx. 25 kDa and 32 kDa respectively. Glycosaminoglycan chains of the aggrecan, biglycan and decorin were treated with chondroitin ABC lyase, chondroitin AC lyase and chondroitin B lyase to determine the positions of sulphation and the degree of uronic acid epimerization. The aggrecan glycosaminoglycan chains were found to contain a 4-sulphate/6-sulphate ratio of 7:3, with no epimerization of glucuronic acid to iduronic acid. The biglycan glycosaminoglycan chains were found to contain a similar ratio of 4-sulphate/6-sulphate, but with approx. 40-45% of the glucuronic acid epimerized to iduronic acid. The decorin glycosaminoglycan chains were found to contain 4-sulphate but no detectable 6-sulphate, and approx. 30-35% epimerization of the glucuronic acid to iduronic acid. The results, using these cloned cells, indicated that a single MC615 cell is able to make all three proteoglycans with distinctive differences between the glycosaminoglycans of aggrecan, biglycan and decorin. These data indicate that a mechanism must exist for a single MC615 cell to regulate the sizes and fine structures of glycosaminoglycans on simultaneously produced, different proteoglycans in a core-protein-specific manner.

Decorin (DCN) is a ubiquitous proteoglycan comprised of a core protein attached to a single dermatan/chondroitin sulphate glycosaminoglycan chain. It may play a role in regulation of collagen fibrillogenesis and function as a reservoir of transforming growth factor beta (TGF-beta) in the extracellular milieu. We have examined the susceptibility of DCN to five different matrix metalloproteinases (MMPs): MMP-1 (tissue collagenase), MMP-2 (gelatinase A), MMP-3 (stromelysin 1), MMP-7 (matrilysin) and MMP-9 (gelatinase B). MMP-2 and MMP-3 digest DCN into seven major fragments in a similar pattern. The N-terminal sequence of the two fragments generated by MMP-2 and MMP-3 is Leu211-Lys-Gly-Leu-Asn, but that of the others is Asp1-Glu-Ala-Ser-Gly. MMP-7 cleaves DCN into three major fragments which have the N-termini Asp1-Glu-Ala-Ser-Gly, Glu2-Ala-Ser-Gly-Ile and Leu244-His-Leu-Asp-Asn. Activities of MMP-1 and MMP-9 against DCN are negligible. The values of Km for the MMPs capable of degrading DCN are very similar (10-12 microM), but the kcat/Km value for MMP-7 (30.5 microM-1.h-1) is 4.5-fold higher than those for MMP-2 and MMP-3. Incubation of a DCN-TGF-beta1 complex with MMP-2, -3 or -7 results in release of TGF-beta1 from the complex. These data indicate proteolytic degradation of DCN by MMP-2, MMP-3 and MMP-7, and suggest the possibility that, under pathophysiological conditions, the digestion by the MMPs may induce tissue reactions mediated by TGF-beta1 released from DCN in the connective tissues.

The proteoglycans decorin and biglycan in extracts of human articular cartilage were analysed by SDS/PAGE and immunoblotting, using antisera raised to peptide sequences present in the pro-regions and the mature core proteins. In adult cartilage, both pro-forms and mature processed forms of the proteoglycan core protein were observed for both decorin and biglycan. In the case of biglycan, it was also shown that additional proteolytic processing takes place after removal of the propeptide and that this accounts for the presence of non-glycanated forms of the molecule. For both decorin and biglycan, the relative abundance of the pro-forms was much less in the juvenile than the adult. Different adult connective tissues, including meniscus, tendon and intervertebral disc were also examined for the presence of pro-forms of the proteoglycans. While the mature form of decorin was present at a similar level in extracts of all tissues examined, the pro-form was only detected in the articular cartilage. In the case of biglycan, the abundance of the mature form was more varied, with high levels in articular cartilage, intermediate levels in meniscus and the annulus fibrosus of the intervertebral disc, low levels in the nucleus pulposus of the intervertebral disc, and non-detectable levels in the patellar tendon. The pro-form of biglycan was detected in the disc tissue extracts, albeit at a lower level than in articular cartilage, but was not detected in the meniscus or tendon. The proportion of the pro-form relative to the mature form of biglycan was, however, higher in the nucleus pulposus of the intervertebral disc than in articular cartilage. Thus, the persistence of pro-forms of both decorin and biglycan is a feature of the extracellular matrix of some connective tissues, although their abundance is both tissue- and age-dependent, with adult articular cartilage being a particularly rich source.

To elucidate the correlation between vascular cholesterol metabolism and proteoglycan (PrGl) biosynthesis, we investigated PrGl synthesis in human aortic smooth muscle cells (SMCs) after cholesterol enrichment with cationized low-density lipoproteins (LDL). Compared with normal SMCs, total PrGl synthesis by cholesterol-enriched cells decreased 2.4-fold (11874 +/- 530 d.p.m. per 10(5) cells compared with 4890 +/- 385 d.p.m. per 10(5) cells). This was the net result of a 6.9-fold reduction in medium PrGl (11000 +/- 490 d.p.m. per 10(5) cells compared with 1580 +/- 246 d.p.m. per 10(5) cells) and a 3.8-fold increase in cellular PrGl over controls (874 +/- 27 d.p.m. per 10(5) cells compared with 3310 +/- 193 d.p.m. per 10(5) cells). Prior incubation of SMCs with native LDL had no effect on PrGl synthesis by these cells. The decrease in PrGl synthesis in cholesterol-enriched cells correlated with a 90% and 20% reduction in the steady-state level of mRNA for biglycan and decorin respectively, and a virtual elimination of the steady-state level of mRNA for versican over controls. Despite the down-regulation of PrGl synthesis, cholesterol-loaded cells produced a 2-fold increase in a PrGl subfraction with high affinity for LDL. Compared with the corresponding PrGl subfraction from normal cells, that from the cholesterol-enriched cells exhibited increased charge density and a higher molecular mass and contained relatively larger proportions of chondroitin 6-sulphate and dermatan sulphate. These results show that PrGl metabolism is dramatically altered in cholesterol-enriched human SMCs.

The small dermatan sulphate protein decorin interacts via its core protein with fibrillar collagens, and its glycosaminoglycan chains were proposed to be capable of self-association. It was therefore of interest to study the role of decorin in the contraction of cell-populated collagen lattices. Stable transfection of dihydrofolate reductase-deficient CHO cells with decorin cDNA resulted in impaired collagen lattice contraction. Using normal human skin fibroblasts in serum-free cultures, inclusion of 0.3 microM decorin in the culture medium also led to a delayed collagen gel contraction. Protein-free dermatan sulphate and the dermatan sulphate-degrading enzyme chondroitin ABC lyase were ineffective. Potential interactions between dermatan sulphate chains were studied by gel filtration. A shift in the elution position of [35S]sulphate-labelled decorin-derived glycosaminoglycans by unlabelled decorin could be observed only when the chains were prepared by trypsin. Chains liberated by beta-elimination or by cathepsin C were eluted at identical positions in the presence or absence of decorin. It is therefore unlikely, that the effect of decorin on collagen-gel retraction is brought about solely by glycosaminoglycan-glycosaminoglycan interactions.

Interleukin-4 (IL-4) is a pleiotropic cytokine expressed by inflammatory cells. Previous work from our laboratory has shown that it stimulates collagen synthesis in fibroblasts. Here we report the effects of recombinant human IL-4 on glycosaminoglycan (GAG) and proteoglycan synthesis in normal dermal fibroblasts from adult donors. IL-4 (10 and 100 units/ml) induced a dose-dependent increase of [3H]glucosamine and [35S]sulphate incorporation into total GAGs. The analysis of the different GAG fractions indicated the enhanced synthesis of dermatan/chondroitin sulphates. IL-4 had no effect on hyaluronan synthesis. The increase of sulphated GAG synthesis was correlated with an increase of proteoglycans in the culture medium. Decorin was identified as the major chondroitin/dermatan sulphate-containing proteoglycan in the culture medium of fibroblasts. Its synthesis was strongly stimulated by IL-4. Both the core-protein synthesis and mRNA expression were enhanced, indicating that the cytokine acted, at least in part, at the pre-translational level. These results indicate that IL-4 is able to modulate not only collagen, but also proteoglycan, production by human fibroblasts. Their implications in physiopathological processes such as wound healing or fibrosis is suggested.

We have analysed the interactions of three proteoglycans of the decorin family, decorin, biglycan and fibromodulin, with transforming growth factor beta (TGF-beta). The proteoglycan core proteins, expressed from human cDNAs as fusion proteins with Escherichia coli maltose-binding protein, each bound TGF-beta 1. They showed only negligible binding to several other growth factors. Intact decorin, biglycan and fibromodulin isolated from bovine tissues competed with the fusion proteins for the TGF-beta binding. Affinity measurements suggest a two-site binding model with Kd values ranging from 1 to 20 nM for a high-affinity binding site and 50 to 200 nM for the lower-affinity binding site. The stoichiometry indicated that the high-affinity binding site was present in one of ten proteoglycan core molecules and that each molecule contained a low-affinity binding site. Tissue-derived biglycan and decorin were less effective competitors for TGF-beta binding than fibromodulin or the non-glycosylated fusion proteins; removal of the chondroitin/dermatan sulphate chains of decorin and biglycan (fibromodulin is a keratan sulphate proteoglycan) increased the activities of decorin and biglycan, suggesting that the glycosaminoglycan chains may hinder the interaction of the core proteins with TGF-beta. The fusion proteins competed for the binding of radiolabelled TGF-beta to Mv 1 Lu cells and endothelial cells. Affinity labelling showed that the binding of TGF-beta to betaglycan and the type-I receptors in Mv 1 Lu cells and to endoglin in endothelial cells was reduced, but the binding to the type-II receptors was unaffected. TGF-beta 2 and 3 also bound to all three fusion proteins. Latent recombinant TGF-beta 1 precursor bound slightly to fibromodulin and not at all to decorin and biglycan. The results show that the three decorin-type proteoglycans each bind TGF-beta isoforms and that slight differences exist in their binding properties. They may regulate TGF-beta activities by sequestering TGF-beta into extracellular matrix.

Polyclonal anti-peptide antibodies were raised to the C-terminal regions of human biglycan and decorin. These antibodies were used in immunoblotting to study structural variations with age in the proteoglycan core proteins present in extracts of human articular cartilage and intervertebral disc. Three forms of the biglycan core protein were identified. The largest form was detected only after chondroitinase treatment and represents the proteoglycan form of the molecule from which the glycosaminoglycan chains have been removed. However, chondroitinase treatment did not alter the electrophoretic mobility of the two smaller proteins, which appear to represent non-proteoglycan forms of the molecule, resulting either from a failure to substitute the intact proteoglycan core protein with glycosaminoglycan chains during its synthesis or from proteolytic processing of the intact proteoglycan causing removal of the N-terminal region bearing the glycosaminoglycan chains. The non-proteoglycan forms constituted a minor proportion of biglycan in the newborn, but were the major components in the adult. A similar trend was seen in both articular cartilage and intervertebral disc. In comparison, decorin appears to exist predominantly as a proteoglycan at all ages, with two core protein sizes being present after chondroitinase treatment. Non-proteoglycan forms were detected in the adult, but they were always a minor constituent.

This study investigates the effects of interleukin (IL)-1 beta on proteoglycan metabolism by fibroblasts surrounded by endogenous extracellular matrix. In both three-dimensional matrix cultures and long-term monolayer cultures IL-1 beta caused a significant decrease in synthesis and deposition of sulphated proteoglycans, but had no effect on release of deposited material. The decrease in synthesis became successively more pronounced, and corresponded to 40-60% of the control after 72 h incubation. The reduction was almost totally accounted for by an effect on the chondroitin ABC-lyase-sensitive proteoglycans. Gel electrophoresis showed a significant decrease in a high-molecular-mass chondroitin ABC-lyase-sensitive proteoglycan after incubation with IL-1 beta. Northern-blot analyses of total RNA revealed a pronounced decrease in the steady-state mRNA levels of versican, the large chondroitin sulphate, with levels corresponding to 10-30% of controls. In comparison, the steady-state mRNA level for decorin, the major sulphated proteoglycan synthesized by the cells, was only slightly affected. The prominent decrease in synthesis of sulphated proteoglycans induced in long-term fibroblast cultures, including the pronounced decrease in versican steady-state mRNA levels, is likely to have a significant effect on the structure of the extracellular matrix. Induction of this type of change may constitute a significant mechanism whereby IL-1 beta can affect the properties of connective tissue during inflammation and wound healing.

Immunological studies revealed the presence of several different forms of biglycan and decorin in human intervertebral-disc tissues (annulus fibrosus, nucleus pulposus and cartilage end-plate). In the young intervertebral disc, glycosaminoglycan-containing (glycanated) forms of both biglycan and decorin represented a greater proportion of the total proteoglycan population present in extracts of annulus fibrosus and cartilage end-plate compared with extracts of nucleus pulposus, in which they were barely detectable. In older discs the glycanated forms of biglycan and decorin represented only a small proportion of the total proteoglycan present. Immunochemical analyses with an antibody to chondroitin/dermatan sulphate isomers indicated differences in the glycosaminoglycans substituted on glycanated forms of small proteoglycans found in different disc tissues. Dermatan sulphate was the predominant glycosaminoglycan present on biglycan and decorin in annulus fibrosus extracts, whereas chondroitin 4-sulphate was present in both small proteoglycans isolated from cartilage end-plate. In addition, immunochemical analyses with antibodies against core protein epitopes identified two non-glycanated forms of both biglycan and decorin. These non-glycanated forms of the small proteoglycans were found in all three regions of the disc. The two nonglycanated forms of biglycan had estimated molecular masses of 37 and 41 kDa and those of decorin were 43 and 45 kDa, respectively. These non-glycanated forms of biglycan and decorin increased in proportion with aging. N-terminal sequence analysis indicated that the larger non-glycanated form of decorin was a degradation product of its glycanated precursor. However, no N-terminal sequence information was obtainable from the other non-glycanated form of decorin or the two non-glycanated forms of biglycan. These data are consistent with the hypothesis that some of the non-glycanated forms of decorin and biglycan are degradation products of native precursors. However, the possibility remains that several different post-translationally modified forms of decorin and biglycan are synthesized by intervertebral-disc tissues.

Immunostaining of adult human skin shows that the small dermatan sulphate proteoglycan decorin is abundant in the whole dermal layer but absent from the epidermis. In the papillary layer adjacent to the dermal-epidermal border, more decorin was detected than in the reticular layer of the dermis. Expression of decorin mRNA by cells in the papillary dermis could also be shown by in situ hybridization. In contrast, biglycan, another small chondroitin sulphate/dermatan sulphate proteoglycan, is found only at the dermal-epidermal border. Therefore the biosynthesis of these two proteoglycans by papillary and reticular fibroblasts from two different donors was compared in tissue culture. Papillary fibroblasts secrete up to 5.9 times more decorin than reticular fibroblasts, while the amounts of cell-associated decorin in both cell types are similar. By Northern blot analysis as well as by in situ hybridization it was shown that papillary fibroblasts contain more mRNA coding for decorin than do reticular cells. In addition, no mosaic pattern of decorin expression was found in the cultured cells. The expression and synthesis of biglycan compared with decorin was about 10 times lower and did not show any significant differences for the two cells types. The kinetics of secretion and the rate of endocytosis of decorin were similar for both types of fibroblasts. These results were found with fibroblasts between the 9th and 15th passage from a newborn subject as well as from a 78-year-old donor, indicating that the pattern of decorin synthesis is not age-dependent in the range investigated. These results further show that fibroblasts from different layers of the dermis have a specific pattern of synthesis of small chondroitin sulphate/dermatan sulphate proteoglycans, and they also maintain these patterns in cell culture.

Proteoglycans synthesized by periodontal (gingival, periodontal ligament, dental follicle) fibroblasts were analysed by SDS/polyacrylamide and agarose gel electrophoresis after being labelled with radioactive sulphate. Medium, cell membrane and extracellular matrix fractions were analysed separately. Samples were treated with chondroitinase AC, chondroitinase ABC, heparitinase or a combination of chondroitinase ABC and heparitinase before electrophoretic separation of proteoglycans. Antibodies to versican and decorin were used to identify these molecules by Western immunoblots. For steady-state metabolic radiolabelling of fibroblasts, medium and cell membrane fractions contained about equal proportions of radiolabelled proteoglycans (about 43%), whereas less radioactivity (about 14%) was found in proteoglycans of the matrix fraction. Periodontal fibroblasts produce six major proteoglycans: versican, a high-molecular-mass chondroitin sulphate proteoglycan (CSPG); decorin, a dermatan sulphate proteoglycan (DSPG); a membrane-associated heparan sulphate proteoglycan (HSPG); two medium- or matrix-associated HSPGs; and a 91 kDa membrane-associated CSPG. Variation in decorin molecular size was observed in mass cultures of fibroblasts. Similar polydispersity in molecular size of decorin was seen in several clones established from one mass culture.

Decorin, an interstitial small proteoglycan, was shown to interact with fibronectin via its core protein. In a solid-phase assay, both high-affinity (KD values between 10 and 20 nM) and low-affinity (KD values between 110 and 130 nM) binding sites were found. The central position of decorin core protein is made up of several repeats containing NKISK in positions 85-89 and similar sequences in other repeats. The pentapeptide inhibited, albeit not completely, the high-affinity interaction between decorin and fibronectin in a specific charge-independent manner. Half-maximal inhibition occurred at a peptide concentration of 10 microM. Core-protein-derived peptides that had been produced by endoproteinase Lys-C digestion were not inhibitory, but endoproteinase Arg-C-generated peptides served as inhibitors of binding. These results suggest that NKISK as a component of repetitive sequences of decorin is involved in the interaction between the proteoglycan and fibronectin.

We have previously reported that the connective tissue stroma of human colon carcinoma contains elevated amounts of decorin, a small proteoglycan involved in the regulation of matrix formation and cell proliferation. These biochemical changes were correlated with increased mRNA levels and general hypomethylation of the decorin gene in human colon cancer DNA. In this report we use a quantitative polymerase chain reaction method coupled with digestion of the DNA template by methylation-sensitive restriction endonucleases to investigate in detail the location of hypomethylated sites in decorin gene. We demonstrate that a specific site in the 3' region of the gene, encompassing codons 360-361, is specifically hypomethylated in both colon carcinoma and benign polyp. In contrast, three HpaII sites, clustered in the 5' untranslated region, show full methylation in normal and neoplastic DNA. The lack of such changes in ulcerative colitis DNA suggests that chronic inflammation alone is not sufficient to alter cytosine methylation in the decorin gene. These results suggest the possibility that the 3' region of the decorin-coding sequence may be involved in the control of decorin gene expression.

The influence of transforming growth factor-beta (TGF-beta) on the expression of different forms of small proteoglycans was investigated in human skin fibroblasts and in a human osteosarcoma cell line. TGF-beta was not found to act as a general stimulator of small proteoglycan biosynthesis. In both cell types, an increased expression of the core protein of proteoglycan I was found. However, there was a profound decrease in the expression of a 106 kDa core protein, and either no alteration or a small decrease in the biosynthesis of the collagen-binding small proteoglycan II core protein. These results show that the production of individual members of the small proteoglycan family is differentially regulated.

The cDNA for the full-length core protein of the small chondroitin sulphate proteoglycan II of bovine bone was cloned and sequenced. A 1.3 kb clone (lambda Pg28) was identified by plaque hybridization with a previously isolated 1.0 kb proteoglycan cDNA clone (lambda Pg20), positively identified previously by polyclonal and monoclonal antibody reactivity and by hybrid-selected translation in vitro [Day, Ramis, Fisher, Gehron Robey, Termine & Young (1986) Nucleic Acids Res. 14, 9861-9876]. The cDNA sequences of both clones were identical in areas of overlap. The 360-amino-acid-residue protein contains a 30-residue propeptide of which the first 15 residues are highly hydrophobic. The mature protein consists of 330 amino acid residues corresponding to an Mr of 36,383. The core protein contains three potential glycosaminoglycan-attachment sites (Ser-Gly), only one of which is within a ten-amino-acid-residue homologous sequence seen at the known attachment sites of related small proteoglycans. Comparisons of the published 24-residue N-terminal protein sequence of bovine skin proteoglycan II core protein with the corresponding region in the deduced sequence of the bovine core protein reveals complete homology. Comparison of the cDNA-derived sequences of bovine bone and human embryonic fibroblast proteoglycans shows a hypervariable region near the N-terminus. Nucleotide homology between bone and fibroblast core proteins was 87% and amino acid homology was 90%.