Structure and deduced amino acid sequence of the human fibromodulin gene.

We have determined the structure and partially sequenced the human fibromodulin gene. The translated region of the gene is composed of two exons. An exon in the 5'-non translated region is separated from the next exon by a 1 kb intron. This exon, which encodes the major part of the translated region, is 983 bp and is followed by an approx. 5 kbp intron. The last 50 nucleotides of the translated region as well as the 3'-nontranslated region are located on the last exon. This exon arrangement is different from the arrangement reported for the gene of the structurally related biglycan. The translated region of the gene was sequenced and compared with bovine fibromodulin. The amino acid sequences of human and bovine fibromodulin show an overall homology of 90%.

Stabilization of the alpha-helical coiled-coil domain in laminin by C-terminal disulfide bonds.

The long arm of laminin in which three polypeptide chains alpha, beta, and gamma are assembled in an alpha-helical coiled-coil structure is stabilized by non-covalent interactions and disulfide bridges. The stabilizing role of the disulfide linkage between the beta and gamma-chains at the C-terminal region of the assembly domain was investigated with about 100-residue long recombinant fragments. Circular dichroism spectra and electron micrographs were identical for linked and non-linked species and indicated two-stranded coiled-coil structures with about 100% alpha-helicity at 20 degrees C. Thermal transition profiles revealed an increase of the melting temperature from 42 degrees C to 60.4 degrees C upon disulfide formation at a chain concentration of 25 microM. The enthalpy of interaction was identical for the two species but the negative entropy involved in joining the two chains was reduced by the disulfide bonds. At chain concentrations of 10 microM the Gibbs free energy delta G was by 17.5 kJ/mol more negative for the disulfide-linked than for the unlinked chains. Because of the concentration dependence of the entropy of the non-linked chains, this difference decreased with increasing concentration and, by extrapolation at chain concentrations of 10 mM, the stability of both structures would be the same. As a competing reaction, beta-chains associated to four-stranded bundles which probably consist of pairs of two-stranded coiled-coils. After disulfide formation a biphasic transition curve was observed which indicated two different ways of connecting the chains in the bundle.

Selective chain recognition in the C-terminal alpha-helical coiled-coil region of laminin.

Recombinant fragments alpha, beta, and gamma were prepared comprising about 100 C-terminal residues of the corresponding polypeptide chains in the three-stranded alpha-helical coiled-coil domain of laminin. Circular dichroism spectra, thermal transition profiles, non-denaturing gels, analytical ultracentrifugation, and calorimetry indicated alpha-helicity and high thermal stabilities for the beta gamma heterodimer and homoassociates of beta. Very little or no coiled-coil formation was found for alpha and gamma. The thermal melting profiles and their concentration dependencies were quantitatively described by a two-state mechanism in which two unfolded chains combine to a fully alpha-helical dimer. For the beta gamma dimer the melting temperature was Tm = 42 degrees C at a chain concentration of 25 microM in 5 mM sodium phosphate buffer (pH 7.4). Addition of 100 mM NaCl decreased the Tm slightly but the relative stability of beta gamma and beta beta coiled-coils was not significantly changed, indicating that electrostatic interactions alone are not responsible for chain selection. Upon addition of 1 M urea the Tm value dropped by about 10 degrees C. The enthalpy changes for the formation of the coiled-coil were delta H degrees = -304(+/- 30) kJ/mol for the beta gamma heterodimer and -198(+/- 20) kJ/mol for the beta-homoassociates. Gibbs free energies and entropies amounted to delta G degrees = -42.8 kJ and delta S degrees = -876 J/mol K for the heteroassembly and -37.8 kJ/mol and -537 J/mol K for the homoassembly of beta. This low preference for heteroassociation of the fragment is smaller than the chain selectivity observed for larger fragments and intact laminin. Deletion of ten residues from the C-terminal region of the gamma-fragment which were recently reported as an essential assembly-site was not sufficient to abolish heteroassociation. Interaction of alpha-fragment with double-stranded beta gamma coiled-coils reflected the formation of a three-stranded coiled-coil in laminin but for the small recombinant fragments association between alpha and beta-homoassociates was also observed. The C-terminal 100 residues in the coiled-coil domain are therefore not alone responsible for the high specificity of chain selection in laminin.

The crystal structure of staphylococcal enterotoxin H: implications for binding properties to MHC class II and TcR molecules.

The X-ray structure of the superantigen staphylococcal enterotoxin H (SEH) has been determined at 1.69 A resolution. In this paper we present two structures of zinc-free SEH (apoSEH) and one zinc-loaded form of SEH (ZnSEH). SEH exhibits the conventional superantigen (SAg) fold with two characteristic domains. In ZnSEH one zinc ion per SEH molecule is bound to the C-terminal beta-sheet in the region implicated for major histocompatibility complex class II (MHC class II) binding in SEA, SED and SEE. Surprisingly, the zinc ion has only two ligating amino acid residues His206 and Asp208. The other ligands to the zinc ion are two water molecules. An extensive packing interaction between two symmetry-related molecules in the crystal, 834 A(2)/molecule, forms a cavity that buries the zinc ions of the molecules. This dimer-like interaction is found in two crystal forms. Nevertheless, zinc-dependent dimerisation is not observed in solution, as seen in the case of SED. A unique feature of SEH as compared to other staphylococcal enterotoxins is a large negatively charged surface close to the Zn(2+) site. The interaction of SEH with MHC class II is the strongest known among the staphylococcal enterotoxins. However, SEH seems to lack a SEB-like MHC class II binding site, since the side-chain properties of structurally equivalent amino acid residues in SEH and those in SEB-binding MHC class II differ dramatically. There is also a structural flexibility between the domains of SEH. The domains of two apoSEH structures are related by a 5 degrees rotation leading to at most 3 A difference in C(alpha) positions. Since the T-cell receptor probably interacts with both domains, SEH by this rotation may modulate its binding to different TcR Vbeta-chains.

Antagonistic effects of the staphylococcal enterotoxin a mutant, SEA(F47A/D227A), on psoriasis in the SCID-hu xenogeneic transplantation model.

Psoriasis is a T-cell-mediated immune dermatosis probably triggered by bacterial superantigens. This pathomechanism has been experimentally reproduced in a SCID-hu xenogeneic transplantation model. We analyzed the effects of different bacterial superantigens on the induction of psoriasis in this model. Staphylococcal enterotoxin B and exfoliative toxin triggered the onset of psoriasis when administered repetitively intracutaneously over a period of 2 wk, whereas staphylococcal enterotoxin A representing a distinct subfamily of staphylococcal enterotoxins only mimicked certain aspects of psoriasis. The biologic effects of staphylococcal enterotoxin A were more pronounced when a mutated form, SEA(H187A), of this superantigen with reduced affinity to major histocompatibility complex class II was coinjected. Another mutated variant, SEA(F47A/D227A), exhibiting no measurable major histocompatibility complex class II affinity blocked the effects triggered by wild-type staphylococcal enterotoxin A when injected in a 10-fold higher dose. Inhibition was specific as induction of psoriasiform epidermal changes by staphylococcal enterotoxin B could not be blocked. As staphylococcal enterotoxin A, in contrast to the other superantigens tested, is capable of inducing epidermal thickening but not the typical appearance of psoriasis, we conclude that bacterial superantigens may differ with regard to their effects on human nonlesional psoriatic skin. Staphylococcal-enterotoxin-A-mediated effects were blocked by a genetically engineered superantigen highlighting the potential therapeutic use of mutated superantigens.

Amino-terminal deletions in the decorin core protein leads to the biosynthesis of proteoglycans with shorter glycosaminoglycan chains.

Analysis of the N-terminal sequence of decorin purified from connective tissues and comparison with the sequence deduced from the cDNA indicate that the nascent proteoglycan has a 14 amino acid residue N-terminal propeptide. Mammalian expression vectors encoding wild-type decorin and decorin with deletions in the propeptide were used to transform COS and CHO cells. Cells transformed with vectors encoding deletion variants of decorin synthesize proteoglycans with shorter galactosaminoglycan chains than cells transformed with wild-type decorin. This effect on the polysaccharide chain length may be due to a lower affinity between the core protein and the glycosyltransferases synthesizing the linkage region. Alternatively, the deletions may affect the intracellular transport of decorin. An antiserum prepared against the N-terminal propeptide immunoprecipitated decorin secreted by cultured cells, showing that decorin is exported with the N-terminal region intact.

Posttranslational modifications of fibromodulin.

Tyrosine sulfate residues were identified in fibromodulin produced by tracheal chondrocytes, by tendon and sclera fibroblasts in primary culture, as well as in Chinese hamster ovary cells transfected with a construct containing fibromodulin cDNA. The tyrosine sulfate residues were located in the N-terminal part of fibromodulin. Thus, Chinese hamster ovary cells expressing a deleted variant of fibromodulin lacking the N-terminal 52 amino acids following the predicted signal peptide did not contain any tyrosine sulfate residues. The substitution with keratan sulfate chains was not restricted to chondrocytes, but was also identified in fibromodulin synthesized by bovine tendon fibroblasts and sclera fibroblasts, as well as in fibromodulin isolated from tendon. Digestion of fibromodulin with N-glycosidase F reduced the apparent size of fibromodulin to that of the core protein, as predicted from sequence analysis (Oldberg, A., Antonsson, P., Lindblom, K., and Heinegård, D. (1989) EMBO J.8, 2601-2604). Thus fibromodulin from cartilage, tendon, and sclera contains N-glycosidically linked oligosaccharides, some of which are extended to keratan sulfate chains.

The keratan sulfate-enriched region of bovine cartilage proteoglycan consists of a consecutively repeated hexapeptide motif.

We have determined the sequence of a cDNA clone encoding the keratan sulfate-rich domain of the large aggregating cartilage proteoglycan core protein. The C-terminal portion of the deduced amino acid sequence is homologous to the chondroitin sulfate-rich region (domain CS1) of the rat chondrosarcoma proteoglycan, and the N-terminal portion is homologous to the second globular domain (G2) of the rat proteoglycan (Doege, K., Sasaki, M., Horigan, E., Hassell, J. R., and Yamada, Y. (1987) J. Biol. Chem. 262, 17757-17767). We could identify, inserted between these regions, a region absent in the rat proteoglycan. This domain corresponds to the keratan sulfate-enriched region of the bovine proteoglycan. It consists of a highly conserved hexapeptide motif consecutively repeated 23 times. Transfer blot analysis of genomic DNA indicated a single gene. The coding region for the keratan sulfate-enriched region was present both in human and bovine DNA, whereas the coding region for this domain appears to be absent in the rat genome. Transfer blot analysis of RNA showed that the keratan sulfate-rich region is present in proteoglycans from fetal as well as adult sources. Furthermore, RNA protection assays of RNA isolated from adult and fetal bovine articular cartilage showed that no alternative splicing occurs within this keratan sulfate-enriched region. These experiments show that the fetal bovine cartilage proteoglycan contains the keratan sulfate attachment domain, although it lacks the keratan sulfate side chains.

The partial amino acid sequence of bovine cartilage proteoglycan, deduced from a cDNA clone, contains numerous Ser-Gly sequences arranged in homologous repeats.

We have determined the sequence of a partial cDNA clone encoding the C-terminal region of bovine cartilage aggregating proteoglycan core protein. The deduced amino acid sequence contains a cysteine-rich region which is homologous with chicken hepatic lectin. This lectin-homologous region has previously been identified in rat and chicken cartilage proteoglycan. The bovine sequence presented here is highly homologous with the rat and chicken amino acid sequences in this apparently globular region. A region containing clusters of Ser-Gly sequences is located N-terminal to the lectin homology domain. These Ser-Gly-rich segments are arranged in tandemly repeated, approx. 100-residue-long, homology domains. Each homology domain consists of an approx. 75-residue-long Ser-Gly-rich region separated by an approx. 25-residue-long segment lacking Ser-Gly dipeptides. These dipeptides are arranged in 10-residue-long segments in the 100-residue-long homology domains. The shorter homologous segments are tandemly repeated some six times in each 100-residue-long homology domain. Serine residues in these repeats are potential attachment sites for chondroitin sulphate chains.

A collagen-binding 59-kd protein (fibromodulin) is structurally related to the small interstitial proteoglycans PG-S1 and PG-S2 (decorin).

We have determined the primary structure of a 59 kd collagen binding protein which is present in many types of connective tissues, e.g. cartilage, tendon, skin, sclera and cornea. The amino acid sequence, deducted from a 2662 bp cDNA clone, predicts a 42 kd protein with a high content of leucine residues. Most of the protein consists of homologous 23 amino acid residues repeats with predominantly leucine residues in conserved positions. Similar leucine rich repeats have been identified in a number of proteins including the small interstitial proteoglycans decorin and PG-S1. The 59 kd protein and the two proteoglycans are homologous in their entire sequences suggesting that they have evolved from a common ancestral gene. The 59 kd protein and decorin are also functionally related in that both bind to collagen type I and II, and affect their fibrillogenesis. The substitution with glycosaminoglycan chains appears to be a feature shared by all three members of this family of leucine rich motif extracellular proteins, since the 59 kd protein isolated from cartilage is substituted with at least one keratan sulfate chain.

COMP (cartilage oligomeric matrix protein) is structurally related to the thrombospondins.

Cloning and sequence analysis of cartilage oligomeric matrix protein (COMP) cDNA, representing a cartilage pentameric protein, revealed a protein of 755 amino acid residues with a calculated molecular mass of 82,700 Da. Expression of the cDNA in COS cells showed that COMP is a homopolymer composed of five identical disulfide-linked subunits. COMP is homologous to the carboxyl-terminal half of thrombospondin, and the homologies include 89% and 54% of the residues in COMP and thrombospondin, respectively. The similarities are most pronounced in the carboxyl-terminal domains and in the calcium binding type 3 repeat domains in which about 60% of the amino acid residues are identical. In the type 2/epidermal growth factor repeat domains the two proteins contain 41% identical residues. The sequence of the amino-terminal 84-amino acid residues is unique for COMP. Comparison of the amino acid sequences in the type 2 and type 3 repeat domains of COMP and the thrombospondins shows that COMP is the product of a unique gene and not the result of an alternatively spliced thrombospondin gene.

Staphylococcal enterotoxin H displays unique MHC class II-binding properties.

Staphylococcal enterotoxin H (SEH) has been described as a superantigen by sequence homology with the SEA subfamily and briefly characterized for its in vivo activity. In this study, we demonstrate that SEH is a potent T cell mitogen and inducer of T cell cytotoxicity that possesses unique MHC class II-binding properties. The apparent affinity of SEH for MHC class II molecules is the highest affinity ever measured for a staphylococcal enterotoxin (Bmax1/2 approximately 0.5 nM for MHC class II expressed on Raji cells). An excess of SEA or SEAF47A, which has reduced binding to the MHC class II alpha-chain, is able to compete for binding of SEH to MHC class II, indicating an overlap in the binding sites at the MHC class II beta-chain. The binding of SEH to MHC class II is like SEA, SED, and SEE dependent on the presence of zinc ions. However, SEH, in contrast to SEA, binds to the alanine-substituted DR1 molecule, betaH81A, believed to have impaired zinc-bridging capacity. Furthermore, alanine substitution of residues D167, D203, and D208 in SEH decreases the affinity for MHC class II as well as its in vitro potency. Together, this indicates an MHC class II binding site on SEH with a different topology as compared with SEA. These unique binding properties will be beneficial for SEH to overcome MHC class II isotype variability and polymorphism as well as to allow an effective presentation on APCs also at low MHC class II surface expression.

Cooperative zinc binding in a staphylococcal enterotoxin A mutant mimics the SEA-MHC class II interaction.

The structure of a mutant form of staphylococcal enterotoxin A (SEA) has been determined to 2.1 A resolution. The studied SEA substitution H187-->A187 (SEAH187A) leads to an almost 10-fold reduction of the binding to major histocompatibility complex (MHC) class II. H187 is important for this interaction since it coordinates Zn2+. The zinc ion is thought to hold MHC class II and SEA together in a complex. Interestingly, only one of two molecules in the asymmetric unit binds Zn2+. H225, D227, a water molecule, and H44 from a symmetry-related molecule ligate Zn2+. The symmetry-related histidine is necessary for this substituted Zn2+ site to bind to Zn2+ at low zinc concentration (no Zn2+ added). Since a water molecule replaces the missing H187, H44 binds Zn2+ at the position where betaH81 from MHC class II probably will bind. Dynamic light scattering analysis reveals that in solution as well as in the crystal lattice the SEA(H187A) mutant forms aggregates. The substitution per se does not cause aggregation since wild-type SEA also forms aggregates. Addition of EDTA reduces the size of the aggregates, indicating a cross-linking function of Zn2+. In agreement with the biological function, the aggregation is weak (i.e. not revealed by gel filtration) and non-specific.

Cartilage matrix proteins. A basic 36-kDa protein with a restricted distribution to cartilage and bone.

A non-collagenous quantitatively prominent protein was purified from guanidine hydrochloride extracts of bovine tracheal cartilage. Purification was achieved by cesium chloride density gradient centrifugation and chromatography on DEAE-cellulose at pH 7.0 followed by CM-cellulose at pH 5.0. The protein has a marked tendency to form aggregates in denaturing solutions of high ionic strength, e.g. 6 M guanidine hydrochloride. The purified protein contains a single, Mr 36,000 polypeptide chain, with a particularly high content of leucine. It contains about 1% carbohydrate with a remarkable absence of hexosamines and sialic acid, whereas xylose, galactose, mannose, and fucose were identified in the preparation. The protein was identified in extracts of cartilage and bone and could be shown to be primarily extracellular. Tendon may contain trace amounts of the protein, whereas extracts of several other tissues showed no immunoreactivity in enzyme-linked immunosorbent assay.

The synthesis of a family of structurally related proteoglycans in fibroblasts is differently regulated by TFG-beta.

Fibroblasts synthesize a variety of proteoglycans among which is a family of structurally related small proteoglycans, i.e. PG-S1 (biglycan) and PG-S2 (decorin). Fibromodulin, which is present in some tissues as a keratan sulfate proteoglycan, also belongs to this family. We have used primary fibroblasts from fetal skin and bovine sclera in culture to study the metabolism of proteoglycans. In particular the regulatory effect of transforming growth factor-beta (TGF-beta), interleukin-1 (IL-1) platelet-derived growth factor (PDGF) and dexamethasone was determined by studies of mRNA levels for these structurally related proteoglycans. Furthermore the synthesis and secretion of these macromolecules was studied using radioactive precursors. TGF-beta induced a 3-fold increase of mRNA for PG-S1, collagen I and III in both types of fibroblasts. mRNA for PG-S2 increased only slightly (1.7-fold) in human skin fibroblasts; while no effect was noticed in sclera fibroblasts. The expression of fibromodulin mRNA was not effected in any of the cells investigated. IL-1, PDGF and dexamethasone had no significant effects on the levels of proteoglycan and collagen mRNA, respectively. Synthesis and secretion of PG-S1, -S2 and fibromodulin wa studied by labeling with [3H]-leucine and [35S]-sulfate. Final separation of PG-S1 and -S2 was achieved by hydrophobic interaction chromatography. TGF-beta induced a 3- to 6-fold increase of [3H]- and [35S]-labeled PG-S1; while PG-S2 only increased 1.3- to 1.4-fold in both types of fibroblasts. No effect on synthesis and secretion of immunoprecipitated fibromodulin was noted.(ABSTRACT TRUNCATED AT 250 WORDS)

The crystal structure of staphylococcal enterotoxin type D reveals Zn2+-mediated homodimerization.

Bacterial superantigens, including the staphylococcal enterotoxins, are the most potent activators of T cells known and have been suggested as a causative factor in Gram-positive shock in humans. Staphylococcal enterotoxin D (SED) is dependent upon Zn2+ for high affinity interactions with MHC class II molecules and thus SED was co-crystallized with Zn2+. The crystal structure of SED has been determined in two different space groups, at 2.3 and 3.0 A resolution respectively. The three-dimensional structure of SED is similar to structures of other bacterial superantigens, although this study has revealed that SED has the unique capability of forming dimers in the presence of Zn2+. The high affinity Zn2+ site used in dimer formation is located on the surface of the beta-sheet in the C-terminal domain. Two bound metal ions are coordinated by residues from both molecules in the dimer interface and thus contribute directly to formation of the dimer. A second Zn2+ site is located on the surface close to the domain interface of the molecule. The unique feature of SED in forming a Zn2+-dependent homodimer seems to facilitate novel and biologically relevant multimeric interactions with MHC class II molecules, as shown by the induction of cytokine mRNA in human monocytes when exposed to SED and SED mutants.

A mutation of F47 to A in staphylococcus enterotoxin A activates the T-cell receptor Vbeta repertoire in vivo.

The bacterial superantigen staphylococcal enterotoxin A (SEA) binds with high affinity to major histocompatibility complex (MHC) class II molecules and subsequently activates T cells bearing particular T-cell receptor (TCR) Vbeta chains. Structural and mutational studies have defined two distinct MHC class II binding sites located in the N-terminal and C-terminal domains of SEA. The N-terminal F47 amino acid is critically involved in a low-affinity interaction to the MHC class II alpha-chain, while the C-terminal residues H187, H225, and D227 coordinate a Zn2+ ion and bind with moderate affinity to the beta-chain. In order to analyze whether the SEA-MHC class II alpha-chain interaction plays a role in dictating the in vivo repertoire of T-cell subsets, we studied distinct Vbeta populations after stimulation with wild-type SEA [SEA(wt)] and SEA with an F47A mutation [SEA(F47A)]. Injections of SEA(wt) in C57BL/6 mice induced cytokine release in serum, strong cytotoxic T-lymphocyte activity, expansion of T-cell subsets, and modulated expression of the T-cell activation antigens CD25, CD11a, CD44, CD62L, and CD69. SEA-reactive TCR Vbeta3+ and Vbeta11+ T cells were activated, while TCR Vbeta8+ T cells remained unaffected. The SEA(F47A) mutant protein induced a weaker T-cell response and failed to induce substantial interleukin-6 production compared to SEA(wt). Notably, SEA(F47A) failed to activate TCR Vbeta11+ T cells, whereas in vivo expansion and modulation of T-cell activation markers on TCR Vbeta3+ T cells were similar to those for SEA(wt). A similar response to SEA(F47A) was seen among CD4+ and CD8+ T cells. Activation of TCR Vbeta3+ and TCR Vbeta11+ T-cell hybridomas confirmed that SEA(F47A) activates TCR Vbeta3+ but not TCR Vbeta11+ T cells. The data support the view that the SEA-N-terminal MHC class II alpha-chain interaction defines a topology that is required for engagement of certain TCR Vbeta chains in vivo.

Functional characterization of the interaction between the superantigen staphylococcal enterotoxin A and the TCR.

In this report, we show that despite an overall amino acid residue identity of more than 80% between the staphylococcal enterotoxins (SE) A and E, these proteins markedly differ in their absolute requirement for the MHC class II during T cell activation. The superantigens were produced as C215Fab-SE fusion proteins and analyzed for their ability to activate T cells in a MHC class II-independent manner, using C215 Ag expressing cell lines as pseudo super-APCs. C215Fab-SEA, but not C215Fab-SEE, induced T cell cytotoxicity and proliferation in these MHC class II-independent systems. Introduction of a region from SEA, comprising amino acids 20-27, to SEE transferred the ability to engage T cells in the absence of MHC class II. Analysis of the Vbeta specificity of the chimeric SEA/SEE molecules and a panel of SEA mutants demonstrated that the site for TCR interaction covers the edge surrounding the shallow cavity on top of the SEA molecule.

The spectral and thermodynamic properties of staphylococcal enterotoxin A, E, and variants suggest that structural modifications are important to control their function.

The superantigens staphylococcal enterotoxin A and E (SEA and SEE) can activate a large number of T-cells. SEA and SEE have approximately 80% sequence identity but show some differences in their biological function. Here, the two superantigens and analogues were characterized biophysically. SEE was shown to have a substantially higher thermal stability than SEA. Both SEA and SEE were thermally stabilized by 0.1 mM Zn(2+) compared with Zn(2+)-reduced conditions achieved using 1 mM EDTA or specific replacements that affect Zn(2+) coordination. The higher stability of SEE was only partly caused by the T-cell receptor (TCR) binding regions, whereas regions in the vicinity of the major histocompatibility complex class II binding sites affected the stability to a greater extent. SEE exhibited a biphasic denaturation between pH 5.0-6.5, influenced by residues in the TCR binding regions. Interestingly, enzyme-linked immunosorbent assay, isoelectric focusing, and circular dichroism analysis indicated that conformational changes had occurred in the SEA/E chimerical constructs relative to SEA and SEE. Thus, it is proposed that the Zn(2+) binding site is very important for the stability and potency of SEA and SEE, whereas residues in the TCR binding site have a substantial influence on the molecular conformation to control specificity and function.

Cartilage matrix proteins. An acidic oligomeric protein (COMP) detected only in cartilage.

An Mr = 524,000 oligomeric protein was isolated from bovine cartilage and designated COMP (Cartilage Oligomeric Matrix Protein). The protein is composed of disulfide-bonded subunits with an apparent Mr of 100,000 each. It is markedly anionic, probably due to its high contents of aspartic acid and glutamic acid, as well as to its substitution with negatively charged carbohydrates. COMP was found in all cartilages analyzed, but could not be detected in other tissues by enzyme-linked immunosorbent assay of guanidine HCl extracts. Within a given cartilage, COMP shows a preferential localization to the territorial matrix surrounding the chondrocytes.