Identification of fibronectin neoepitopes present in human osteoarthritic cartilage.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 54, No. 9, September 2006, pp 2912–2922 DOI 10.1002/art.22045 © 2006, American College of Rheumatology Identification of Fibronectin Neoepitopes Present in Human Osteoarthritic Cartilage Marc D. Zack, Elizabeth C. Arner, Charles P. Anglin, James T. Alston, Anne-Marie Malfait, and Micky D. Tortorella C-terminus VRAA271 were detected following cytokine treatment of human cartilage extracts. These neoepitopes localized with areas of aggrecan loss in OA cartilage. Conclusion. Human OA cartilage contains fibronectin fragments with catabolic activity and a major cleavage site within fibronectin. This study is the first to characterize fibronectin neoepitopes in OA cartilage, suggesting that they may represent a novel biomarker of arthritis. Objective. Fibronectin fragments are present at high concentrations in the cartilage of patients with rheumatoid arthritis and patients with osteoarthritis (OA) and have been shown to promote cartilage catabolism in human cartilage cultures, suggesting that fibronectin fragments participate in the initiation and progression of arthritic disease. This study was undertaken to 1) identify the major fibronectin fragments in human OA cartilage and confirm their ability to elicit cartilage catabolism, 2) identify the cleavage sites in fibronectin and generate the corresponding neoepitope antibodies, and 3) explore the utility of fibronectin neoepitopes as biomarkers. Methods. Fibronectin fragments were purified from human OA cartilage using affinity chromatography; their N-termini were then identified by sequencing. Bovine nasal cartilage was treated with affinity-purified fibronectin fragments and assayed for aggrecan breakdown by monitoring the release of glycosaminoglycans and the aggrecan neoepitope 1771AGEG. Fibronectin neoepitopes were detected by Western blotting in cytokine-treated media of human cartilage explants, and by immunohistochemical analyses of human OA cartilage. Results. Multiple fibronectin fragments were isolated from human OA cartilage, and all contained the N-terminus 272VYQP. These fragments induced aggrecanase-mediated cartilage catabolism in bovine cartilage explants. Fibronectin fragments with the N-terminus 272 VYQP and fragments with the Fibronectin is a multimeric glycoprotein found in human plasma and tissues. Alternative splicing of fibronectin messenger RNA reveals many different species of fibronectin monomers that comprise plasma and tissue fibronectin (1,2). The plasma form of fibronectin is a soluble dimer composed of 200–220-kd subunits bound together through C-terminal disulfide bonds, yielding a total molecular weight of ⬃440 kd. Tissue fibronectin exists in the form of large insoluble multimers within the extracellular matrix (ECM) bound together through disulfide bonding, as well as interactions through self-association domains (2). Monomers contain functional domains that can bind various integrins, heparin, fibrin, and collagen/gelatin (for review, see ref. 2). As a result of the activity of these many functional domains, fibronectin is involved in a wide range of biologic processes, such as cell migration, wound healing, angiogenesis, and cell differentiation (3–6). In normal cartilage, fibronectin is a minor component of the ECM. However, osteoarthritic (OA) cartilage explants can contain up to 10-fold more fibronectin than is present in normal cartilage (7,8). This is due to an increase in both synthesis and accumulation of fibronectin (9). Increases in cartilage fibronectin are accompanied by a corresponding accumulation of fibronectin in the synovial fluid of both rheumatoid arthritis (RA) and OA patients (10). Fibronectin can be Marc D. Zack, MS, Elizabeth C. Arner, MS, Charles P. Anglin, MS, James T. Alston, BS, Anne-Marie Malfait, MD, Micky D. Tortorella, BS: Pfizer Global Research & Development, Chesterfield, Missouri. Address correspondence and reprint requests to Marc D. Zack, MS, Pfizer Global Research & Development, BB334-K, 700 Chesterfield Parkway, Chesterfield, MO 63017. E-mail: marc.d.zack @pfizer.com. Submitted for publication February 6, 2006; accepted in revised form May 16, 2006. 2912 FIBRONECTIN NEOEPITOPES IN OA CARTILAGE proteolysed into smaller fragments by multiple enzymes, and fragmentation is seen in the synovium and cartilage of both patients with OA and patients with RA (11,12). Xie et al (12) have shown that synovial fluid fibronectin fragments can reach levels higher than 1 M, which is ascribed to the increase in fibronectin production and the concomitant elevation of proteolytic enzymes in these disease states. It is well established that the products of fibronectin degradation, and not intact fibronectin, are able to induce a catabolic phenotype in chondrocytes and cartilage explants. Fragments have been shown to induce metalloproteinases, including matrix metalloproteinase 3 (MMP-3) (13,14), MMP-13, and aggrecanases (13), as well as serine proteinases such as urokinase plasminogen activator (14). These proteases have the ability to break down the major cartilage matrix components, including type II collagen and aggrecan, which is a hallmark of OA. It has also been shown that prolonged exposure of cartilage to high levels of fibronectin fragments can suppress matrix synthesis (15), thus limiting the anabolic reparative response to cartilage damage. In addition, fibronectin fragments have been demonstrated to up-regulate proinflammatory mediators such as interleukin-1␤ and inducible nitric oxide synthase (16), which are thought to play roles in the onset and progression of OA and RA. Therefore, degradation of fibronectin, and the resulting fibronectin fragments, may play an important role in the initiation and progression of arthritic disease. To date, only fibronectin fragments present in the synovial fluid have been characterized. Peters et al (17) observed that several fibronectin fragments, ranging in molecular weight from 47 kd to 170 kd, contained an intact N-terminus, in both patients with RA and those with OA. These studies focused on the structural and ligand-binding properties of the fragments that contained the N-terminal heparin-binding domain of fibronectin. Although these experiments helped characterize the components of fibronectin associated with arthritic diseases, the cleavage sites within fibronectin were not determined. Furthermore, these experiments analyzed the presence of fragments in the synovial fluid, which may not accurately reflect that in cartilage. The present study was conducted to 1) isolate fibronectin fragments from human OA cartilage using affinity chromatography, in order to determine whether these fragments are capable of inducing catabolism in cartilage explant cultures, 2) identify the N-terminus of the fibronectin fragments and generate antibodies to both the N-terminal and the C-terminal neoepitopes, 2913 and 3) begin to explore these fragments as potential biomarkers of cartilage breakdown. PATIENTS AND METHODS Materials. All common laboratory chemicals, the silver staining kit (ProteoSilver Plus), and rabbit polyclonal antifibronectin antibody (F-3468) were purchased from Sigma (St. Louis, MO). Protease inhibitor cocktail tablets, purchased from Roche Diagnostics (Indianapolis, IN), were used to inhibit a broad spectrum of proteases, including cysteine, serine, aspartyl, and metalloproteinases. Gelatin–Sepharose 4-B was purchased from Amersham Biosciences (Upsala, Sweden). Recombinant human oncostatin M (OSM) and recombinant human tumor necrosis factor ␣ (TNF␣) were purchased from R&D Systems (Minneapolis, MN). All electrophoresis reagents, cell culture media, and supplements were purchased from Invitrogen (Carlsbad, CA). Alkaline phosphatase–conjugated anti-rabbit IgG antibodies (Promega, Madison, WI) were used according to the manufacturer’s directions. Deglycosylating enzymes keratanase I, keratanase II, and chondroitinase were purchased from Seikagaku (East Falmouth, MA). Chemicals and reagents used for N-terminal protein sequencing were purchased from Applied Biosystems (Foster City, CA). Human cartilage extraction. The human OA cartilage used for extraction, identification, and Western blotting of the fibronectin fragments was obtained from OA patients at the time of total knee replacement (a gift from Dr. Kosei Ijiri, Kagoshima University, Kagoshima, Japan). All donors were men between the ages of 64 and 82 years. Cartilage was removed from the femoral condyles and tibial plateau and then minced into small pieces with a scalpel. Thereafter, all steps were performed in the presence of protease inhibitor (used as suggested by the manufacturer). Proteins were extracted from human OA cartilage at 4°C with constant inversion for 7 days in extraction buffer (4M guanidine HCl, 50 mM sodium acetate, pH 6.8; 10 ml per 1 gm cartilage). Samples were then subjected to cesium chloride density-gradient centrifugation. Supernatants were adjusted to a density of 1.5 gm/ml with cesium chloride, centrifuged at 39,400 revolutions per minute for 72 hours at 4°C, and the D4 fraction (top 25%) of the sample was retained and extensively dialyzed (molecular weight cutoff point of 10 kd) against Tris buffered saline (TBS) (50 mM Tris, pH 7.6, 100 mM NaCl) in the presence of protease inhibitors. Fibronectin fragments were quantified using Lowry’s modified protein assay according to the manufacturer’s directions (Pierce, Rockford, IL). Silver staining was used to assess the purity of the fragments, with analyses carried out by electrophoresis of 30 l dialyzed extract under reducing conditions. Molecular weight was estimated by Western blotting with molecular weight standards (Bio-Rad, Hercules, CA). Human cartilage explant cultures. Macroscopically normal cartilage was obtained postmortem from the femoral condyles of 2 male donors (National Disease Research Interchange, Philadelphia, PA). Cartilage was cut into ⬃100-mg sections and allowed to equilibrate at 37°C in 5% CO2 for 2 days in serum-free Dulbecco’s modified Eagle’s medium (DMEM)–F12 media (1:1 ratio) supplemented with 100 units/ml penicillin, 100 g/ml streptomycin, and 250 ng/ml 2914 amphotericin B. Following equilibration, cartilage was incubated in DMEM–F12 (100 mg cartilage/1.5 ml media) containing TNF␣ and OSM at 10 ng/ml each. Supernatants were collected after 14 days of culture. For time-course experiments, cartilage from a single donor was divided into separate wells for independent cultures over 4, 6, 8, and 10 days, without media change. Neoepitope antibody generation. Polyclonal neoepitope antibodies (Quality Controlled Biochemicals, Hopkinton, MA) were prepared against the peptide sequences PFTDVRAA and VYQPQPHP, representing the C-terminus and N-terminus, respectively, of the fibronectin fragments that were generated by cleavage at the Ala271/Val272 bond. Antibodies recognizing the spanning peptide (PFTDVRAAVYQPQPHP) were removed by passing sera through a spanning peptide agarose column. Antibodies flowing through the spanning peptide column were then positively selected based on their ability to bind C-terminal (PFTDVRAA) or N-terminal (VYQPQPHP) peptides conjugated to agarose. Antibodies were further characterized using Western blotting. Purification of fibronectin fragments. After dialyzing into TBS, fibronectin and fibronectin fragments were isolated using gelatin chromatography. Gelatin–Sepharose was equilibrated with 5-column volumes of TBS. Extracts were loaded onto the column and washed with 5-column volumes of TBS, and gelatin-binding proteins were eluted with TBS containing 8M urea. Fractions were then dialyzed against TBS and subjected to Western blot analysis. In some cases, fractions were concentrated from 1 ml to ⬃50 l, using centrifugal concentrations (molecular weight cutoff of 10 kd). Elution of fibronectin fragments with 8M urea ensured that ADAMTS-4 and ADAMTS-5 were not copurified with the fibronectin fragments, since neither of these enzymes refold efficiently (⬍7%) after denaturation with 8M urea (Tortorella MD, et al: unpublished observations). Silver staining provided additional confirmation of the purity of these fibronectin fragments (results not shown). Western blotting. Samples (30 l) were subjected to electrophoresis in denaturing and reducing conditions (NuPage lithium dodecyl sulfate [LDS] 4⫻ sample buffer, 2.5% ␤-mercaptoethanol) using 10% Bis-Tris 10-well gels (Invitrogen). Proteins were then transferred to 0.2-m polyvinylidene difluoride membranes and probed with antibody. For analysis of aggrecan fragments, samples were enzymatically deglycosylated, prior to electrophoresis, with chondroitinase ABC (0.1 unit/50 ml media), keratanase (0.1 unit/50 ml media), and keratanase II (0.002 unit/50 ml media) for 3 hours at 37°C in buffer containing 50 mM sodium acetate, 0.1M Tris HCl, pH 6.5 (18). After digestion, the aggrecan was precipitated with 5 volumes of acetone and reconstituted in 30 l 4⫻ LDS sample buffer (Invitrogen) containing 2.5% ␤-mercaptoethanol, and heated for 10 minutes at 95°C. Fibronectin neoepitope antibodies were used at 2 g/ml, and antifibronectin antibodies (Sigma) were diluted 1:5,000. Aggrecan neoepitope antibodies, as described by Tortorella et al (18), were diluted 1:1,000. The AGEG neoepitope measures a preferred aggrecanase cleavage site within aggrecan and is much more reactive than other antibodies to the other 4 aggrecanase cleavage sites. Thus, this antibody is more likely to detect early levels of aggrecanase-mediated aggrecan breakdown. Incubation and detection were performed with the ZACK ET AL Western Breeze alkaline phosphatase detection system (Invitrogen) using protocols provided by the manufacturer. N-terminal sequencing. Proteins for N-terminal sequencing were stained using 0.025% Coomassie R-250 stain in 40% methanol, and destained with 50% methanol to allow visualization of proteins. Bands were then cut from the membrane and sequenced via the Edman degradation method, using a Procise-cLC (Applied Biosystems) N-terminal sequencer. Results were analyzed using SequencePro software (Applied Biosystems). Bovine nasal cartilage cultures. Cartilage disks (1 mm ⫻ 8 mm) from bovine septa (Covance, Princeton, NJ) were incubated for 6 days in DMEM containing 5% heatinactivated fetal bovine serum and penicillin, streptomycin, and amphotericin B (100 units/ml, 100 g/ml, and 0.25 g/ml, respectively). Disks were cut into 8 slices of ⬃12–15 mg each. Each piece was weighed, placed into a 96-well plate, and allowed to equilibrate for 2 days in 180 l serum-free DMEM (containing antibiotics). After equilibration, media were replaced with 180 l media containing human OA fibronectin fragments. Treatments were performed in duplicate. Fibronectin fragments were diluted in DMEM and sterile filtered (0.2 m) before application to explant cultures. Detection of sulfated glycosaminoglycan (GAG) by 1,9-dimethylmethylene blue (DMMB assay). Aggrecan degradation was monitored by assessing the release of sulfated GAG into the media, using the DMMB assay with shark chondroitin sulfate as a standard. This assay was adapted from the method described by Farndale et al (19). Results are expressed as micrograms of GAG per milligram of wet cartilage. Histology and immunohistochemistry. Human OA cartilage obtained at the time of knee replacement from a 57-year-old female patient (Washington University Department of Orthopedics, St. Louis, MO) was fixed overnight with 10% neutral buffered formalin and processed for paraffin embedding. Cartilage was graded using the macroscopic Collins score (20). Sections were stained with Safranin O for histologic evaluation. Slides prepared for immunohistochemistry were deparaffinized, rehydrated through graded ethanols, and then pretreated with 0.1 units/ml chondroitin ABC lyase (MP Biomedicals, Aurora, OH). Slides were then washed in TBST (50 mM Tris, 138 mM NaCl, 2.7 mM KCl, 0.05% Tween 20, pH 8.0). Endogenous peroxidase was blocked using Peroxidase Block (DakoCytomation, Carpinteria, CA), and nonspecific antibody binding was blocked using Protein Block (DakoCytomation). Slides were incubated with primary antibody (antiVYQP at 1 g/ml or anti-VRAA at 1 g/ml). Antibody binding was detected using the Envison Plus horseradish peroxidase system (DakoCytomation). Sections were counterstained with hematoxylin. Negative controls contained no primary antibody but an equivalent dilution of rabbit serum. As antibodyspecificity controls, the primary antibody was preabsorbed with its immunizing or blocking peptide at 50⫻ molar excess. RESULTS Purification of fibronectin fragments present in OA cartilage. Fibronectin was purified from OA knee cartilage extracts using gelatin–Sepharose chromatogra- FIBRONECTIN NEOEPITOPES IN OA CARTILAGE 2915 Figure 1. Purification of fibronectin fragments from osteoarthritic (OA) cartilage. A, Silver staining of isolated fibronectin fragments from OA cartilage extracts assessed using gelatin chromatography. A comparison of staining pre– and post–gelatin chromatography demonstrates that the majority of the protein in the extract did not bind gelatin. Four bands were detected in the eluted fraction, and 3 of the 4 bands were visualized after transfer to a polyvinylidene difluoride membrane and subsequent Coomassie staining. Asterisks denote bands that were N-terminally sequenced. The N-terminus of each fragment was 272VYQPQP. The diagram adjacent to the gel predicts the fibronectin modules contained in the fibronectin fragments, based on molecular weight. B, Structure of intact fibronectin monomer and known cleavage sites. A major site of cleavage within fibronectin is predicted to be at Ala271/Val272. The known cleavage sites of other mammalian fibronectin-degrading enzymes are also denoted. Squares represent type I modules (labeled 1–12). Circles represent type II modules (labeled 1 and 2). Ovals represent type III modules (labeled 1–15). Shaded ovals represent type III modules found in splice variants of fibronectin. EDA ⫽ extra domain A; EDB ⫽ extra domain B; CS ⫽ variable connecting segment; MMP-2 ⫽ matrix metalloproteinase 2. phy. Eluates contained multiple gelatin-binding proteins with molecular weights ranging from 50 kd to 85 kd, as revealed by silver staining (representative gel in Figure 1A). After concentration, 2 of these proteins (⬃60 kd and ⬃85 kd) were identified as fibronectin fragments by N-terminal sequence analysis. Both of the fragments contained the amino-terminus 272VYQPQPHP, suggesting a site of cleavage within fibronectin at Ala271/Val272. This was repeated on knee cartilage extracts from 4 different OA donors, and fibronectin fragments containing this N-terminus were seen at molecular weights ranging from ⬃50 kd to ⬃85 kd (results not shown). The N-terminus 272VYQPQPHP is found in the spacer region between the fifth and sixth type I modules within the intact molecule (Figure 1B). All of the fragments purified contained an intact gelatin-binding domain near the N-terminus, thus explaining the affinity for gelatin. Because of the relatively low abundance of the other gelatin-binding proteins that were purified from OA cartilage, N-terminal sequencing and identification of these could not be achieved. Characterization of neoepitope antibodies to fibronectin fragments. Neoepitope antibodies were raised against the new N-terminus 272VYQP and the new 2916 ZACK ET AL Figure 2. Detection of fibronectin fragments generated by cleavage at Ala271/Val272 with a neoepitope antibody to 272VYQP and to VRAA271. Cartilage extract (30 l) was separated on a 10% Bis-Tris gel and analyzed for fibronectin fragments containing the N-terminus 272VYQP and C-terminus VRAA271 by Western blot analysis using the 272VYQP (A) and VRAA271 (B) neoepitope antibody in the absence or presence of 10 M immunizing peptide (IP). C-terminus VRAA271 generated following cleavage of intact fibronectin at Ala271/Val272. Analysis of a human OA knee cartilage extract from a 64-year-old male donor using the anti-272VYQP antibody revealed multiple bands ranging in molecular weight from ⬃50 kd to ⬃100 kd (Figure 2A). This antibody did not detect intact fibronectin, and inhibition of reactivity with the corresponding blocking peptide (VYQPQPHP) confirmed antibody specificity. The presence of numerous fibronectin fragments having the identical N-terminus, 272 VYQP, suggests that there are multiple cleavage sites downstream from Val272. The anti-VRAA271 antibody detected a distinct 30-kd fragment (Figure 2B). Less intense staining was also found at 250 kd, suggesting that this antibody may recognize the intact molecule when present at high concentrations. Inhibition of reactivity with the corresponding blocking peptide confirmed that this antibody was specific for fibronectin fragments with the free C-terminal sequence VRAA271 (Figure 2B). The blocking peptide PFTDVRAA was able to block immunoreactivity of the 250-kd band, and since this antibody did not react with other fibronectin fragments, this may indicate that the N-terminal VRAA271 30-kd fibronectin fragment is bound to intact fibronectin, even under denaturing and reducing conditions. Stimulation of aggrecan catabolism in bovine nasal cartilage by isolated OA fibronectin fragments. Endogenous fibronectin fragments were purified from human OA knee cartilage by gelatin chromatography, as described above. Western blot analysis using the 272 VYQP neoepitope antibody confirmed that these FIBRONECTIN NEOEPITOPES IN OA CARTILAGE 2917 monitor fibronectin degradation, macroscopically normal human knee cartilage explants were incubated with and without TNF␣ and OSM (10 ng/ml each) for 2 weeks. Media were collected from treated and untreated cartilage explants and probed for total fibronectin and for the fibronectin neoepitopes VRAA271 and 272VYQP by Western blot analysis. Media samples collected from stimulated ex- Figure 3. Induction of aggrecanase-mediated glycosaminoglycan (GAG) release in bovine nasal cartilage by isolated osteoarthritic (OA) fibronectin (Fn) fragments (Fn-f). Top, Bovine nasal cartilage explants were treated for 4 days with either isolated OA Fn fragments (purified by gelatin chromatography) or with intact Fn, and aggrecan catabolism was monitored with the use of 1,9-dimethylmethylene blue assay to measure GAG release into the media. Bars show the results in individual wells. Replicate assays could not be performed due to the limited amount of fibronectin fragments that could be isolated from human OA cartilage. Bottom, Detection of the aggrecanase-generated aggrecan neoepitope 1771AGEG. Detection of the neoepitope was achieved by analyzing 30 l of media on a 10% Bis-Tris gel followed by Western blot analysis using the 1771AGEG neoepitope antibody. fragments were generated through cleavage at the Ala271/Val272 bond (results not shown). These isolated OA fibronectin fragments, at a concentration of ⬃1 M, induced the release of GAG in bovine nasal cartilage after 4 days of stimulation; this was not seen in cultures treated with 1 M intact fibronectin (Figure 3). Media with bovine nasal cartilage were also analyzed for aggrecan fragments generated by aggrecanase-mediated cleavage of the aggrecan core protein at the Glu1771/ Ala1772 bond, assessed using Western blotting with the anti-1772AGEG antibody as previously described (18). The AGEG Western blot showed increased production, over control levels, of a 120-kd aggrecan fragment (Figure 3, bottom), suggesting that the observed fibronectin fragment–induced catabolism was mediated by aggrecanases. Presence of fibronectin neoepitopes VRAA271 and 272 VYQP in human cartilage stimulated with TNF␣/ OSM. In an effort to determine whether detection of the neoepitopes VRAA271 and 272VYQP may be used to Figure 4. Release of fibronectin (Fn) fragments from normal human knee cartilage in cultures stimulated with tumor necrosis factor ␣ (TNF␣) and oncostatin M (10 ng/ml each) for 14 days. Supernatant (30 l) was analyzed for Fn fragments by sodium dodecyl sulfate– polyacrylamide gel electrophoresis under reducing conditions on a 10% Bis-Tris gel. Western blotting was used to detect A, total Fn by using a Fn polyclonal antibody, B, Fn fragments containing the N-terminus 272VYQP using the 272VYQP neoepitope antibody, and C, Fn fragments containing the C-terminus VRAA271 using the VRAA271 neoepitope antibody. 2918 ZACK ET AL Presence of fibronectin neoepitopes VRAA271 and VYQP in human OA cartilage at sites of cartilage degradation. Localization of the fibronectin neoepitopes 272 VYQP and VRAA271 in cartilage was determined by immunohistochemistry using neoepitope antibodies. Macroscopically normal (Collins score 0) human knee cartilage (Figures 6a–c) with very mild aggrecan depletion at the surface (Figure 6a) was compared with severely affected (Collins score IV) OA knee cartilage (Figures 6d–f). In normal cartilage, small amounts of VRAA271 were detected in the regions of aggrecan depletion (Figure 6b), whereas no 272VYQP staining was observed (Figure 6c). In contrast, both neoepitopes were abundantly present in OA cartilage, with the VRAA271 epitope primarily detected in the ECM, especially in the fibrillated areas at the cartilage surface (Figure 6e). The 272 VYQP neoepitope was also seen in the cartilage matrix and at sites of fibrillation (Figure 6f). In addition, 272 VYQP staining was very strong at the surface of chondrocytes (Figure 6i). Specificity of the neoepitope antibodies was confirmed by demonstrating that the respective immunizing peptide blocked staining (Figure 6g for VRAA271 and Figure 6h for 272VYQP). 272 Figure 5. Release of fibronectin fragments containing the C-terminal neoepitope VRAA271 from human osteoarthritic knee cartilage over time. Cartilage explants were treated with tumor necrosis factor ␣ (TNF␣) and oncostatin M (10 ng/ml each) for 10 days. Media were collected every other day on days 4–10. Supernatant (30 l) was run under reducing conditions on a 10% Bis-Tris gel and analyzed for fibronectin fragments by Western blotting using the VRAA271 neoepitope antibody. plants contained more total fibronectin than did control explants. Moreover, stimulated samples contained a greater abundance of degraded fibronectin when compared with controls (Figure 4A). An ⬃45-kd OA fibronectin fragment with the N-terminus 272VYQP (Figure 4B) was detected in 3 of 5 treated samples (weak detection in the first lane), but was not detected in any of the untreated samples. The anti-VRAA271 antibody revealed an ⬃30-kd fibronectin fragment in all treated samples, but not in any of the untreated samples (Figure 4C). Time-dependent increase in OA fibronectin fragments with TNF␣/OSM treatment. Our preliminary experiments showed production of both of the OA fibronectin neoepitopes, VRAA271 and 272VYQP, after 14 days of stimulation with TNF␣ and OSM. Since this was a relatively long incubation time, the experiment was repeated and the media were monitored for neoepitopes on days 4, 6, 8, and 10. The fibronectin fragments with the C-terminus VRAA271 were detected by Western blotting by day 6, and increased in a time-dependent manner at days 8–10 (Figure 5). Fibronectin fragments with the N-terminus 272VYQP were not detected at these time points. The apparent absence of this fragment could be due to lower sensitivity of the 272VYQP antibody. However, it could also be due to further C-terminal degradation of 272VYQP-containing fibronectin fragments, which would dilute the neoepitope among multiple species. In addition, the 272VYQPcontaining fibronectin fragments have multiple cell binding sites, and may be retained in the cartilage ECM and not readily released into the media. DISCUSSION Cartilage degradation is the hallmark of arthritic disease. It occurs when steady-state homeostasis of the main structural cartilage macromolecules, aggrecan and type II collagen, is altered, disturbing the fine balance between anabolism and catabolism of the ECM. Accelerated breakdown of cartilage is an early and sustained feature of OA and can be ascribed to proteolytic enzymes, including the aggrecanases, which mediate aggrecan breakdown, and the collagenases, which mediate collagen breakdown. Accumulating evidence shows that fragments of fibronectin function as a major stimulus responsible for the induction of these proteolytic enzymes (14,21–23). Thus, identifying the fibronectin fragments responsible for stimulating catabolism in cartilage could lead to a better understanding of arthritic diseases. In this study we have shown that several fibronectin fragments found in cartilage contain the N-terminus 271 VYQPQPHP, suggesting a major cleavage site within fibronectin at Ala271/Val272. The Ala271/Val272 bond is located in the middle of a spacer region that separates the fifth type I module from the sixth type I module. The sixth type I module demarks the beginning of the gelatin-binding domain (Figure 2). Thus, as expected, these fragments were retained on a gelatin–Sepharose column (Figure 1). This region within fibronectin con- FIBRONECTIN NEOEPITOPES IN OA CARTILAGE 2919 Figure 6. Immunohistochemical analyses of fibronectin neoepitopes. Serial sections of macroscopically normal (Collins score 0) (top) and severely osteoarthritic (Collins score IV) (middle) cartilage were prepared and stained for loss of aggrecan with Safranin O (SAF O) (a and d), probed for the fibronectin fragment neoepitope VRAA271 in the absence (b and e) or presence (g) of blocking peptide (immunizing peptide [IP]), or probed for the fibronectin fragment neoepitope 272VYQP in the absence (c and f) or presence (h) of blocking peptide (IP). A higher magnification view of f is shown in i (original magnification ⫻ 10), highlighting the cell-associated staining of this epitope. tains a stretch of hydrophobic aliphatic amino acids that confer linearity to this spacer domain. In addition, this region of the molecule is particularly susceptible to proteolytic attack, since it has been reported that multiple enzymes can hydrolyze fibronectin in the region between amino acids Ser254 and His286. Proteinases that have been shown to cleave fibronectin within this region include bacterial subtilisin, meprins, plasmin, thrombin, and trypsin (24–27). Fibronectin fragments generated by cleavage at Ala271/Val272 have been identified by N-terminal sequencing and by Western blot analysis using neoepitope antibodies that detect the new C-terminus VRAA271 and the new N-terminus 272VYQP. The 30-kd fibronectin fragments containing the C-terminus VRAA271 were present in multiple OA donors, as determined by Western blotting and immunohistochemistry. This particular fragment may be the most abundant fragment found in OA cartilage, because it is relatively small in size and has a compact modular structure that may be less prone to further proteolytic degradation. Ingham et al (28) have shown that similar fragments generated by neutrophil elastase digestion are thermally stable. Interestingly, this fragment is very similar to the N-terminal heparinbinding fragment investigated by several research groups and shown to be a potent inducer of cartilage catabolism (29). The N-terminal fragment that was described by Homandberg and Erickson is generated from thrombin digestions to produce a 29–30-kd fragment with the C-terminus FTDVR269 (26). However, fibronectin fragments with this neoepitope or the corresponding N-terminal neoepitope 270AAVY have not been investigated in human cartilage to date, and we did not identify this neoepitope at any time in the studies presented 2920 herein. The OA fibronectin fragment that we identified is ⬃30 kd with a C-terminal sequence of DVRAA271. Thus, nonphysiologic fragments commonly used to elicit catabolic responses in human cartilage differ by only 2 amino acids from the physiologic fibronectin fragment that we identified. The gelatin-binding fibronectin fragments that were isolated from human OA cartilage explants, which included fragments containing the 272VYQP neoepitope, stimulated cartilage catabolism at physiologically relevant concentrations in bovine nasal explant cultures after 4 days of stimulation. Aggrecan breakdown, as measured by GAG release, was mediated by the aggrecanases, as indicated by the release of the 1771 AGEG aggrecan neoepitope. Thus, it is hypothesized that fibronectin fragments produced by cleavage of fibronectin at Ala271/Val272 are able to induce the production and/or activation of the aggrecanases ADAMTS-4 and ADAMTS-5. The role of fibronectin fragments in inducing aggrecanase-mediated aggrecan catabolism is supported by findings from Stanton et al (13), who demonstrated that a 45-kd fibronectin fragment stimulated the production of the neoepitope ITEGE373, which is generated when aggrecanase cleaves aggrecan at Glu373/Ala374. However, this particular fibronectin fragment was generated by cathepsin D and trypsin digestion, and was therefore different from those that we have identified in OA cartilage. Nevertheless, the ability of the 45-kd fibronectin fragment used by Stanton and colleagues to induce cartilage catabolism suggests a common feature shared by fibronectin fragments generated by in vitro digestion and those isolated from human OA cartilage. We have also shown that the OA fibronectin neoepitopes were generated in healthy human cartilage explants after stimulation with a combination of the catabolic cytokines TNF␣ and OSM. These cytokines are frequently used to induce cartilage degradation in bovine and human explant cultures. In these systems, TNF␣ in combination with OSM has been shown to induce MMP-1, MMP-3, and MMP-13, which degrade multiple ECM components (30). OSM alone has been shown to increase fibronectin production in human breast cancer cell lines (31), consistent with the possibility that these cytokines may promote cartilage degradation through a fibronectin fragment–mediated pathway. Fibronectin fragments containing the C-terminus VRAA271 were released and detected in explant media by day 6. However, proteolysis of fibronectin may occur before this time, and the delay in detection may simply be a function of diffusion of fibronectin fragments from ZACK ET AL the cartilage. In fact, we have previously demonstrated that the majority of fibronectin fragments in nonstimulated OA cartilage are retained in the matrix even after 14 days of culture (Tortorella MD, et al: unpublished observations). Thus, locally generated fibronectin fragments may stimulate cartilage catabolism for a prolonged period of time. The apparent lack of the 272VYQP neoepitope in the media of cytokine-treated cultures may have been due to 1) the decreased ability of the fragment to diffuse out of the cartilage because of increased cell binding mediated by an intact central cell binding domain, and/or 2) increased susceptibility of the 272VYQPcontaining fibronectin fragments to further proteolytic cleavage, resulting in a lack of detection of multiple species by Western blot analysis. The consistent detection of fragments containing the C-terminus VRAA271 in cytokine-stimulated cartilage cultures may have been due to the stability of this particular fibronectin fragment. The 30-kd fragment is smaller in size and is composed entirely of type I modules. Because type I modules each contain 2 disulfide bonds, this N-terminal fragment is predicted to be more stable than larger fragments containing the 272 VYQP N-terminus, which contain (at most) 7 type I modules, 2 type II modules (2 pairs of disulfide bonds), and 15 type III modules (no disulfide bonds). Therefore, the 30-kd fibronectin fragment is predicted to be more stable and less susceptible to further proteolytic cleavage, and thus to be more abundant in cartilage. In these studies, fibronectin fragments with an affinity for gelatin were purified from OA cartilage, but we cannot exclude the possibility that other fibronectin fragments exist in OA cartilage that are important in the initiation and progression of disease. Future studies will focus on the isolation of fibronectin fragments that do not bind gelatin. However, at this point, antibodies that detect the fibronectin neoepitopes VRAA271 and 272 VYQP may be valuable biomarkers of cartilage degradation in OA and RA. Immunohistochemical analysis showed that both neoepitopes localize in areas of aggrecan depletion and fibrillation. Since fibronectin fragments induce catabolic factors that lead to the destruction of cartilage, detection of the VRAA271 and 272 VYQP neoepitopes may be useful for identifying the initiation of arthritis prior to the onset of significant cartilage destruction, and may be used to monitor more established disease. Our data suggest that fibronectin fragments containing the N-terminal neoepitope 272VYQP may be more readily retained in the cartilage matrix when FIBRONECTIN NEOEPITOPES IN OA CARTILAGE compared with fibronectin fragments containing the C-terminal VRAA271 neoepitope, which is more likely to diffuse into the synovial fluid and bloodstream. Future studies will determine if these neoepitopes are found in the synovial fluid, blood, and urine of patients with OA and those with RA. Moreover, in future studies, we should be able to characterize the stability of the neoepitopes in these types of fluids. The results presented herein raise the question as to which enzyme is responsible for generating the fibronectin fragments found in OA. Bacterial subtilisin is the only enzyme known to cleave fibronectin at Ala271/ Val272, bringing up the possibility that the human proteinases responsible for this cleavage are structurally related. Thermolysin, plasmin, and thrombin have all been shown to cleave fibronectin within the spacer domain located between the fifth and sixth type I fibronectin modules (24,26). N-terminal sequence analysis of thrombin digests, which are commonly used to study the effects of fibronectin fragments, revealed a major cleavage site at Arg269/Ala270 within fibronectin (26). Thus, thrombin is not likely to be a contributor to the pool of fibronectin fragments in OA. Urokinase has also been shown to cleave at the same site as thrombin, as well as at Arg2309/Thr2310 (27). Recently, it was demonstrated that meprin-␣ preferentially cleaves at Tyr273/Gln274 and Asn689/Thr690, while meprin-␤ cleaves at Leu701/Val702 and Arg883/Ser884 (13). MMP-2 has been shown to cleave fibronectin near the C-terminus at Gly1786/Val1787 (24), and Wilhelm et al (32) have shown that MMP-3 can degrade fibronectin in both neutral and acidic conditions. However, the major MMP-3 site of cleavage was identified as Pro700/Leu701, which generates fibronectin fragments with molecular weights of 160–180 kd (26). Many other mammalian enzymes have also been shown to cleave fibronectin, but thus far none have been shown to cleave at Ala271/ Val272. The identification of the human proteinase responsible for cleavage of fibronectin at Ala271/Val272, currently referred to as endogenous fibronectinase, will be critical for understanding its role in fibronectin degradation and in the pathogenesis and progression of OA. REFERENCES 1. Hershberger RP, Culp LA. Cell-type-specific expression of alternatively spliced human fibronectin IIICS mRNAs. Mol Cell Biol 1990;10:662–71. 2. Pankov R, Yamada KM. Fibronectin at a glance. J Cell Sci 2002;115:3861–3. 2921 3. George EL, Baldwin HS, Hynes RO. Fibronectins are essential for heart and blood vessel morphogenesis but are dispensable for initial specification of precursor cells. Blood 1997;90:3073–81. 4. Hines KL, Kulkarni AB, McCarthy JB, Tian H, Ward JM, Christ M, et al. Synthetic fibronectin peptides interrupt inflammatory cell infiltration in transforming growth factor ␤1 knockout mice. Proc Natl Acad Sci U S A 1994;91:5187–91. 5. Muro AF, Chauhan AK, Gajovic S, Iaconcig A, Porro F, Stanta G, et al. Regulated splicing of the fibronectin EDA exon is essential for proper skin wound healing and normal lifespan. J Cell Biol 2003;162:149–60. 6. Schlaepfer DD, Broome MA, Hunter T. Fibronectin-stimulated signaling from a focal adhesion kinase-c-Src complex: involvement of the Grb2, p130cas, and Nck adaptor proteins. Mol Cell Biol 1997;17:1702–13. 7. Brown RA, Jones KL. The synthesis and accumulation of fibronectin by human articular cartilage. J Rheumatol 1990;17:65–72. 8. Wurster NB, Lust G. Synthesis of fibronectin in normal and osteoarthritic articular cartilage. Biochim Biophys Acta 1984;800: 52–8. 9. Burton-Wurster N, Lust G. Incorporation of purified plasma fibronectin into explants of articular cartilage from disease-free and osteoarthritic canine joints. J Orthop Res 1986;4:409–19. 10. Carnemolla B, Cutolo M, Castellani P, Balza E, Raffanti S, Zardi L. Characterization of synovial fluid fibronectin from patients with rheumatic inflammatory diseases and healthy subjects. Arthritis Rheum 1984;27:913–21. 11. Clemmensen I, Andersen RB. Different molecular forms of fibronectin in rheumatoid synovial fluid. Arthritis Rheum 1982;25: 25–31. 12. Xie DL, Meyers R, Homandberg GA. Fibronectin fragments in osteoarthritic synovial fluid. J Rheumatol 1992;19:1448–52. 13. Stanton H, Ung L, Fosang AJ. The 45 kDa collagen-binding fragment of fibronectin induces matrix metalloproteinase-13 synthesis by chondrocytes and aggrecan degradation by aggrecanases. Biochem J 2002;364:181–90. 14. Xie DL, Hui F, Meyers R, Homandberg GA. Cartilage chondrolysis by fibronectin fragments is associated with release of several proteinases: stromelysin plays a major role in chondrolysis. Arch Biochem Biophys 1994;311:205–12. 15. Homandberg GA, Wen C, Hui F. Cartilage damaging activities of fibronectin fragments derived from cartilage and synovial fluid. Osteoarthritis Cartilage 1998;6:231–44. 16. Gemba T, Valbracht J, Alsalameh S, Lotz M. Focal adhesion kinase and mitogen-activated protein kinases are involved in chondrocyte activation by the 29-kDa amino-terminal fibronectin fragment. J Biol Chem 2002;277:907–11. 17. Peters JH, Carsons S, Yoshida M, Ko F, McDougall S, Loredo GA, et al. Electrophoretic characterization of species of fibronectin bearing sequences from the N-terminal heparin-binding domain in synovial fluid samples from patients with osteoarthritis and rheumatoid arthritis. Arthritis Res Ther 2003;5:R329–39. 18. Tortorella MD, Pratta M, Liu RQ, Austin J, Ross OH, Abbaszade I, et al. Sites of aggrecan cleavage by recombinant human aggrecanase-1 (ADAMTS-4). J Biol Chem 2000;275:18566–73. 19. Farndale RW, Sayers CA, Barrett AJ. A direct spectrophotometric microassay for sulfated glycosaminoglycans in cartilage cultures. Connect Tissue Res 1982;9:247–8. 20. Collins D. The pathology of articular and spinal disease. London: E. Arnold & Co.; 1949. 21. Homandberg GA. Potential regulation of cartilage metabolism in osteoarthritis by fibronectin fragments. Front Biosci 1999;4: D713–30. 22. Homandberg GA, Hui F, Wen C, Purple C, Bewsey K, Koepp H, et al. Fibronectin-fragment-induced cartilage chondrolysis is associated with release of catabolic cytokines. Biochem J 1997;321: 751–7. 2922 23. Homandberg GA, Wen C. Exposure of cartilage to a fibronectin fragment amplifies catabolic processes while also enhancing anabolic processes to limit damage. J Orthop Res 1998;16:237–46. 24. Balian G, Click EM, Crouch E, Davidson JM, Bornstein P. Isolation of a collagen-binding fragment from fibronectin and cold-insoluble globulin. J Biol Chem 1979;254:1429–32. 25. Gold LI, Garcia-Pardo A, Frangione B, Franklin EC, Pearlstein E. Subtilisin and cyanogen bromide cleavage products of fibronectin that retain gelatin-binding activity. Proc Natl Acad Sci U S A 1979;76:4803–7. 26. Homandberg GA, Erickson JW. Model of fibronectin tertiary structure based on studies of interactions between fragments. Biochemistry 1986;25:6917–25. 27. Kruse MN, Becker C, Lottaz D, Kohler D, Yiallouros I, Krell HW, et al. Human meprin ␣ and ␤ homo-oligomers: cleavage of basement membrane proteins and sensitivity to metalloprotease inhibitors. Biochem J 2004;378:383–9. 28. Ingham KC, Brew SA, Broekelmann TJ, McDonald JA. Thermal ZACK ET AL 29. 30. 31. 32. stability of human plasma fibronectin and its constituent domains. J Biol Chem 1984;259:11901–7. Homandberg GA, Meyers R, Xie DL. Fibronectin fragments cause chondrolysis of bovine articular cartilage slices in culture. J Biol Chem 1992;267:3597–604. Hui W, Cawston T, Rowan AD. Transforming growth factor ␤1 and insulin-like growth factor 1 block collagen degradation induced by oncostatin M in combination with tumour necrosis factor ␣ from bovine cartilage. Ann Rheum Dis 2003;62:172–4. Zhang F, Li C, Halfter H, Liu J. Delineating an oncostatin M-activated STAT3 signaling pathway that coordinates the expression of genes involved in cell cycle regulation and extracellular matrix deposition of MCF-7 cells. Oncogene 2003;22:894–905. Wilhelm S, Shao Z, Housley T, Seperack P, Baumann A, GunjaSmith Z, et al. Matrix metalloproteinase-3 (stromelysin-1): identification as the cartilage acid metalloprotease and effect of pH on catalytic properties and calcium affinity. J Biol Chem 1993;268: 21906–13.