fibronectin fragments upregulate insulin-like growth factor binding proteins in chondrocytes

Osteoarthritis and Cartilage (2002) 10, 734–746© 2002 OsteoArthritis Research Society International. Published by Elsevier Science Ltd. All rights reserved. 1063–4584/02/$35.00/0doi:10.1053/joca.2002.0808, available online at http://www.idealibrary.com on

InternationalCartilageRepairSociety

Fibronectin fragments upregulate insulin-like growth factor bindingproteins in chondrocytesC. R. Purple*, T. G. Untermann†, R. Pichika* and G. A. Homandberg**Department of Biochemistry, Rush Medical College at Rush-Presbyterian-St. Luke’s Medical Center, 1653West Congress Parkway, Chicago, IL 60612-3864, U.S.A.†Rm. 5A122A Research (MP 151), Veterans Affairs Chicago Health Care System (West Side), 820 SouthDamen Ave., Chicago, IL 60612, U.S.A.

Summary

Addition of fibronectin fragments (Fn-fs) to cultured cartilage explants has been shown to mediate extensive cartilage matrix degradationfollowed by anabolic responses.

Objective: To determine whether specific Fn-fs regulate cartilage metabolism through a mechanism, in part, involving insulin-like growthfactor (IGF) and insulin-like growth factor binding proteins (IGFBPs).

Methods: Primary bovine articular chondrocyte cultures were treated with Fn-fs. mRNA from the cultures was analysed by Northern blotting.Changes in the levels of IGFBPs in cellular extracts and conditioned media were analysed by Western ligand blotting. Explant cultures ofbovine articular cartilage were used to assay release of exogenous IGF-I and IGFBP-2. An analog of IGF-I with altered affinity for IGFBPswas used to assay the effect of IGFBPs on proteoglycan synthesis.

Results: The Fn-fs increased protein levels of IGFBPs-2, -3 and -5 in conditioned media and of IGFBP-2 in cell extracts by as much asnine-fold. Conversely, the protein level of constitutively expressed IGBP-4 was decreased in conditioned medium. Northern blot analysisreflected increased IGFBP-3 mRNA but not decreased IGFBP-4 mRNA. The IGF-I analog was more effective at restoring PG synthesissuppression by Fn-fs than was wild type IGF-I.

Conclusions: The Fn-fs increased levels of IGFBPs in cultures of bovine articular chondrocytes and elicited release of IGFBP-2 and IGF-Ifrom articular cartilage. The increased level of IGFBPs may trap IGF-I and account in part for the initial suppression of PG synthesis. Inducedproteinases may subsequently liberate IGF-I and cause greatly enhanced anabolic processes, contributing to cartilage repair. © 2002OsteoArthritis Research Society International. Published by Elsevier Science Ltd. All rights reserved.

Key words: Fibronectin, Fibronectin fragments, IGF binding proteins, Cartilage, Chondrocytes.

Abbreviations: Fn-fs, fibronectin fragments; Fn, fibronectin; PG, proteoglycan; MMP, matrix metalloproteinase; MMP-3, stromelysin-1; OA,osteoarthritis; IGF-I, insulin-like growth factor; IGFBP, IGF-I binding proteins; HBSS, Hank’s balanced salt solution; DMEM, Dulbecco’smodified Eagle’s medium; TBS, Tris-buffered saline; BAC, bovine articular chondrocytes; DMB, dimethylmethylene blue.

Received 17 October 2001; accepted 13 March 2002.Address correspondence to present address: Gene A.

Homandberg, Ph.D., University of North Dakota School ofMedicine and Health Sciences, Department of Biochemistry andMolecular Biology, Grand Forks, ND 58202, U.S.A., Tel.: (701)777-6422.

Supported by the Seikagaku Corporation (Tokyo, Japan), theNational Institute of Arthritis and Musculoskeletal DiseasesSpecialized Center of Research Grant AR39239-03, the Dr. Ralphand Marian C. Falk Medical Research Trust Fund and, in part, bythe Department of Veterans Affairs Merit Review Program and theNational Institutes of Health Grant R01 DK41430.

Introduction

There are many observations of enhanced anabolic activi-ties in early osteoarthritic cartilage as well as in explantcultures of human cartilage. For example, cartilage fromdonors with mild and moderate osteoarthritic (OA) showsenhanced proteoglycan (PG) synthesis rates, while carti-lage from donors with severe OA shows low PG synthesisrates1. Further, osteoarthritic cartilage that shows initiallyenhanced PG synthesis rates, later shows decreased syn-

734

thesis2. Little is known of how this anabolic response isinitiated. A thorough understanding of the mechanism forthis early attempted but failed repair response may allowus to bolster its magnitude and allow completion of thereparative program.

As a means of understanding attempted repair, we haveused a cartilage explant culture system in which matrixdamage is initiated by addition of 29-kDa amino-terminal or140-kDa Arg-Gly-Asp containing, proteolytic fragments offibronectin (Fn-fs). These Fn-fs elevate synthesis of thecatabolic cytokines IL-1 and TNF-�3,4 which in turncause increased expression of matrix metalloproteinases(MMPs), including MMP-33,5–8 and aggrecanase-like activi-ties9,10 resulting in PG degradation3–8,11,12. These Fn-fs athigh concentrations also temporarily suppress synthesis ofPG in cartilage tissue11,12. The PG synthesis suppressionis followed by a compensatory reparative response.Mechanistically, this Fn-f activity requires penetration ofintact cartilage6 and probably involves the �5�1 fibronectinreceptor since analog peptides of the Arg-Gly-Asp integrin-binding sequence block activities of both the 29-kDa and140-kDa Fn-fs13. The physiological significance of these

Osteoarthritis and Cartilage Vol. 10, No. 9 735

activities is supported by our observations that amongother Fn-fs, the 29-kDa and 140-kDa Fn-fs are found athigh levels in synovial fluids of patients with rheumatoidarthritis and OA14 and that injection of Fn-fs into rabbit kneejoints causes severe and apparently irreversible loss ofarticular cartilage PG within 7 days15.

One of the most interesting features of this model is thatas the concentration of Fn-fs is reduced, severe cartilagematrix damage is no longer evident. Instead, PG synthesisrates are immediately enhanced as well as steady statelevels of the PG content of the cartilage11. However, withextended time in culture, the explants ultimately begin toshow depletion of PG content. Thus, at low concentrationsof Fn-fs the treated tissue behaves like early OA tissue asshown by an anabolic response that precedes matrixdamage. Further, if cartilage is treated with high concen-trations of Fn-f for a period of time too brief to allow severematrix damage, and the Fn-fs are removed, an anabolicresponse is initiated16. The culturing conditions used toinitiate these anabolic responses have also been shown tomake the tissue more refractory to subsequent insults11,16.Thus, the anabolic response–catabolic response observedin human osteoarthritic cartilage can be mimicked in thisFn-f treatment model.

We have proposed that this anabolic response is due toIGF-I, a major PG synthesis-promoting factor in cartilagetissue17. This was based on observations that Fn-f treat-ment of cartilage enhances release of IGF-I into themedia4. When added to explant or monolayer cultures,IGF-I can enhance synthesis of aggrecan, link protein, typeII collagen18 and hyaluronan19; it can also counter theeffects of IL-1 or TNF-�20, two of the proinflammatorycytokines associated with reduced anabolic processes andincreased catabolic processes, characteristic of OA21,22.Our earlier work suggested that the ability of the Fn-f toenhance IGF-I release is not due to up-regulation of IGF-Imessage or protein synthesis since IGF-I protein releasewas enhanced in the presence of metabolic inhibitors4. Amore likely possibility is that enhanced proteolysis of thematrix enhanced release of IGF-I stored in the cartilagetissue. We also considered that there may be a secondevent, prior to proteolysis, that bolstered an IGF-Iresponse; Fn-fs might enhance synthesis of IGF-I bindingproteins (IGFBPs) that would trap IGF-I around the cellsand enhance the effect of IGF-I once the IGFBPs aredegraded. IGFBPs have properties that would support sucha role. IGFBPs belong to a family of six differentiallyregulated proteins that bind IGFs specifically, with highaffinity and thereby alter the activity of IGFs. IGFBPs caneither potentiate or attenuate IGF-I activity, dependingupon the type of IGFBP and on the involvement of extra-cellular matrix (ECM). The activities of the IGFBPs havebeen reviewed23.

Our objectives were to investigate whether Fn-fs mightaffect changes in the levels of IGFBPs expressed byarticular chondrocytes. We conclude that some of theability of Fn-fs to initially suppress PG synthesis and to laterenhance PG synthesis as observed in developing OA mightbe due to increased IGFBPs.

from the Nichols Institute (San Juan Capistrano, CA).IGF-I and Des(1-3)-IGF-I were purchased from GroPep(Adelaide, Australia). The IGF-I was a recombinant humanform. Additional IGF-I was purchased from AustralBiologicals (San Ramon, CA). Antisera against IGFBPswere purchased from Upstate Biotechnology, Inc. (LakePlacid, NY). A 2.2 kb bovine IGFBP-3 cDNA was providedby Dr Cheryl Conover (Mayo Clinic, Rochester, MN). A505 bp human IGFBP-4 cDNA probe was provided by DrShunichi Shimasaki (Dept. of Reproductive Medicine, Uni-versity of California, San Diego). Statistical analysis wasperformed by the unpaired two-way Student’s t-test whereP<0.05 was considered significant. For most analysis, N≥4.

Isolation of the Fn-fs from human plasma fibronectin (Fn)was performed as described5. The two Fn-fs studied werea well-defined thrombin-generated heparin and fibrin-binding 29-kDa amino-terminal Fn-f, which is the mostactive on cartilage tissue, and a mixture of carboxyl-terminal Fn-fs of 100–140-kDa that contain the cell-bindingregion and which we refer to as the 140-kDa Fn-f5.

PREPARATION OF HIGH-DENSITY BOVINE ARTICULAR

CHONDROCYTE MONOLAYER CULTURES AND CARTILAGE EXPLANT

CULTURES

Bovine articular cartilage was removed from the meta-carpophalangeal joints of adult bovine animals (18–20months old) obtained from the slaughterhouse within 1–2 hof death. Cartilage was washed twice with Hank’s balancedsalt solution (HBSS) that contained penicillin and strepto-mycin. The cartilage was either cultured or chondrocyteswere released for monolayer culture. To release the bovinearticular chondrocytes (BAC), the cartilage was incubatedfirst in a solution of 0.4% pronase (Calbiochem Corp., LaJolla, CA) w/v in DMEM with 5% fetal bovine serum for90 min at 37°C under 5% CO2. The cartilage slices werethen washed three times with serum-free DMEM thenincubated under the same conditions in a solution of 0.02%collagenase P (Roche Diagnostics Corp., Indianapolis, IN)w/v in DMEM, 5% serum for 18 h. Following digestion, thecells were washed in serum-free DMEM, counted using ahemacytometer and seeded at densities of 5.0×105 or7.5×105 cells/cm2 in 10% serum/DMEM. The medium waschanged every other day.

LIGAND BLOT ANALYSIS OF IGFBPS

Cultured cells were washed twice with HBSS and thenadjusted to serum-free DMEM. One day later, Fn-fs wereadded in fresh serum-free DMEM. After a 48-h incubation,the media were collected and concentrated 10-fold; thecells were extracted with non-reducing SDS samplebuffer24. Samples of cell extracts and concentrated mediacontaining 150 �g protein were separated by electrophore-sis through a 13% polyacrylamide gel24 and transferred toa nitrocellulose membrane at 200 V for 2 h. The IGFBPswere analysed by ligand blotting performed essentially asdescribed25. Briefly, the membranes were then rinsed inTBS (0.05 M Tris, 150 mM NaCl, pH 7.4) containing 3%NP-40, blocked with TBS, 0.1% Tween-20, 1% BSA for 2 hand then incubated overnight with 2×105 cpm/ml 125Ilabeled human recombinant IGF-I, prepared as described26

with a final specific activity of 295–305 �Ci/�g. Membraneswere then rinsed with TBS, 0.1% Tween-20 and finally with

Materials and methods

All common chemicals, except where noted, were pur-chased from Sigma Chemical Co. (St Louis, MO). Sodium35S-sulfate (43 Ci/mg S) was from ICN Biomedicals, Inc.(Costa Mesa, CA). The IGF-I assay kits were purchased

736 C. R. Purple et al.: Fibronectin fragments upregulate IGF binding proteins

TBS. After drying, membranes were dried, autoradio-graphed and the X-ray sheets analysed using a StormPhosphorimager and ImageQuant software (MolecularDynamics, Sunnyvale, CA). In initial ligand blot exper-iments 100-fold excess unlabeled IGF-I was included as acontrol for binding specificity. Binding was found to bespecific and these controls were discontinued (data notshown).

analysed by the method of Masuda et al.

IMMUNOPRECIPITATION OF IGFBP-2 AND IGFBP-4

To form the immune complexes, rabbit anti-bovineIGFBP-2 or rabbit antihuman IGFBP-4 (UBI, Lake Placid,NY) was diluted 1:20 with Pansorbin (Calbiochem Corp., LaJolla, CA) solution and incubated, with agitation, for 4 h at4°C. The immune complexes were then washed four timeswith TBS and then re-suspended in 0.5 ml of this buffer. A100 �l volume of the immune complexes was incubated for18 h with pre-cleared conditioned media from Fn-f treatedBAC cultures. The immune complexes were then washedwith TBS. The Pansorbin suspension was then boiled inSDS-PAGE sample buffer and the bound proteins analysedby ligand blotting. As per the vendor (UBI): the antihumanIGFBP-4 antiserum cross-reacts with the bovine proteinand there is approximately 50% cross-reactivity of theIGFPB-4 antiserum with IGFBP-2 protein.

NORTHERN BLOT ANALYSIS

Total RNA was extracted from monolayer cultures ofBAC with RNAzol B (Tel-Test, Inc., Friendswood, TX) usingthe manufacturer’s protocol. However, to obtain sufficientlypure RNA from BAC, re-extraction of the aqueous phasewas required. RNA (10 �g per lane) was electrophoresedthrough a 1% agarose formaldehyde gel and then trans-ferred to uncharged nylon membrane (NEN, Inc., Boston,MA). Membranes were probed with 2.2 kb bovine IGFBP-3or with a 505 bp human IGFBP-4 cDNA probe. The probeswere labeled with alpha 32P-CTP using Random PrimersKit (Gibco BRL, Inc., Rockville, MD). The membranes werepre-hybridized overnight at 48°C in a solution of 5× NaCl,sodium phosphate EDTA (SSPE), 50% deionized forma-mide, 10% dextran sulfate, 5× Denhardt’s solution and 1%and SDS. Following pre-hybridization, the membraneswere hybridized for 18 h at 50°C with 1×106 cpm/ml ofcDNA probe, then washed at 55°C sequentially with 2×SSPE, 0.1% SDS, 1× SSPE, 0.1% SDS and 0.1× SSPE,0.1% SDS. Autoradiography was performed at −70°C witha Cronex intensifying screen and X-Omat film from Kodak.

IMMUNOFLUORESCENT LOCALIZATION OF IGFBP-2

Cells were plated into eight-well chamber slides at lowdensity (25 000 cells/cm2) in 10% serum/DMEM adjustedto 250 �g/ml hyaluronidase. Cells were allowed to adhereovernight and then the media replaced with 10% serum/DMEM. After an additional 18 h, cells were rinsed five timeswith HBSS and cultured in serum-free DMEM containing100 nM 29-kDa Fn-f for 48 h. Cells were then rinsed withPBS and fixed for 5 min with 1.5% paraformaldehyde inPBS. The fixed cells were then incubated 1 h with 3% BSAin PBS, followed by incubation for 1 h with a 1000-folddilution of rabbit antibovine IGFBP-2 antisera in 1% BSA/PBS. The cells were rinsed with PBS and incubated for 1 hwith a 1000-fold dilution of a rhodamine conjugated goat

antirabbit IgG. After rinsing with PBS, antibody labeling wasvisualized by fluorescence microscopy.

EFFECT OF DES (1-3) IGF-I AND IGF-I ON PG SYNTHESIS

To test the effect of mutant IGF-I, BAC in DMEM wereincubated 24 or 48 h with 100 nM Fn-f, followed by adjust-ment of the media to 20 ng/ml of IGF-I or Des (1-3) IGF-I(GroPep, Adelaide, Australia). After an additional 20 h ofincubation, the media were adjusted to 10 �Ci of 35SO4 fora 2-h pulse-chase labeling. The cells were then extractedwith 4 M guanidine HCl. The cell extracts were exhaustivelydialysed against H2O and 35SO4 incorporation was

27

EFFECT OF ANTI-IGFBP-2 AND ANTI-IGFBP-4 ON PG SYNTHESIS

SUPPRESSION BY THE 29-KDA AND 140-KDA FN-FS

Polyclonal antisera against IGFBP-2 (UBI, Lake Placid,NY) and IGFBP-4 (Austral Biologicals, San Ramon, CA),with or without Fn-f, were incubated with monolayer cul-tures in serum-free DMEM for 48 h. During the final 2 h thecultures were labeled with 35SO4 as described above.Sulfate incorporation was quantified using the method ofMasuda et al.27.

FN-F MEDIATED RELEASE OF RHODAMINE LABELED HRIGFBP-2

AND HRIGF-I FROM CARTILAGE

Both IGFBP-2 and IGF-I were labeled using NHS-rhodamine according to the manufacturer’s protocol(Pierce Chem. Co., Rockford, IL), using quantities of 1 �gof IGFBP-2 and 10 �g of IGF-I. Explant cultures of bovinearticular cartilage were incubated for 24 h with either100 ng/ml IGF-I or 50 ng/ml IGFBP-2. The media wereremoved and the tissue was rinsed twice with serum-freeDMEM and the media were replaced with either serum-freeDMEM or serum-free DMEM with 100 nM 29-kDa Fn-f.Following a further 48-h incubation the tissue was rinsedtwice and prepared for cryosectioning. Fluorescencewas photographed on a Nikon microscope equipped withepifluorescence.

TESTS OF EFFECTS OF MMP-3 ON PG CONTENT, PG SYNTHESIS

AND IGF-I RELEASE

ProMMP-3 was isolated from chondrocyte monolayercultures treated with 100 nM Fn-f in serum-free DMEM asdescribed13. Activation of proMMP-3 was performed byadjustment to 10 mM aminophenylmercuric acetate, fol-lowed by exhaustive dialysis. MMP-3 was added to carti-lage in DMEM to a final concentration of 5 �g/ml. Cartilagetissue was removed periodically to measure cartilage PGcontent as well as measure rates of PG synthesis. Aliquotsof media were removed to measure IGF-I content.

ASSAYS FOR IGF-I

To assay the Fn-f conditioned media, aliquots (0.5 ml) ofconditioned media were collected and, without freezing,immediately dialysed against 10 mM NaCl, 10 mM Tris,pH 7.4, and concentrated 10× on a Speed-O-Vac concen-trator without application of heat. Assays were performed

T

S

dMp3ccIasI


using an RIA kit from the Nichols Institute (San JuanCapistrano, CA) as directed by the manufacturer. This kitrecognizes bovine IGF-I as well as human IGF-I. Linearequations for the standard curves had r2 values of greaterthan 0.98. The data shown are the means and S.D. values ofat least three separate tissue culture wells. Day 0 timepoints correspond to the amount of material found in themedia at the start of the experiment, immediately afteraddition of the Fn-f. For kinetic points, the contentsreported correspond to accumulation up to the respectivetime point.

Results

THE FN-FS INCREASED IGFBPS

In addition to native Fn, two Fn-fs were tested, theamino-terminal 29-kDa fragment and a cell binding 140-kDa fragment. As shown in Fig. 1, addition of either of theFn-fs, or, to a lesser extent, native Fn, to cultured chondro-cytes resulted in increased levels of several IGFBPs both inthe conditioned media and in the cell extracts as comparedwith untreated cells. Densitometric data summarized inTable I illustrate that the greatest effect of the Fn-fs was onthe 30-kDa IGFBP, which is probably IGFBP-5, based on itsmolecular weight. The experiments summarized in Table Iare representative of three separate experiments. Con-versely, the Fn-fs did not alter levels of a 24-kDa IGFBP inthe cell extracts but decreased its levels in the conditionedmedia. The 24-kDa IGFBP was detected by the phoshorim-ager during densitometric analysis; however the band is notapparent in the photograph of the autoradiogram shown inFig. 1. Native Fn had weaker effects than the Fn-fs. Theassignments, as based on masses28 were: IGFBP-4 for the24-kDa band, IGFBP-5 or glycosylated IGFBP-4 for the30-kDa band, IGFBP-2 for the 32-kDa band and IGFBP-3for the 38/42-kDa doublet. The ligand blot analysis shownin Fig. 2 demonstrates that the effect of the Fn-fs on IGFBPlevels was greater at 1 �M than 100 nM. While the magni-tude of the increase of the 30-kDa protein was greater at0.1 �M, it was at the 1 �M concentration that both the 30and 32-kDa proteins were increased. There was adetectable increase as low as 1 nM 29-kDa Fn-f (data notshown).

Fig. 1. Effect of Fn-fs on IGFBP activity as shown by ligand blotanalysis. Bovine articular chondrocytes in monolayer cultures werecultured for 48 h in serum-free DMEM and the conditioned mediaand cell extracts analysed by ligand blotting. Shown are cultureswith no addition (Con) or addition of 100 nM 29-kDa (29), 140-kDaFn-fs (140) or with 100 nM Fn (Fn). Markers on left show positions

of molecular weight standards.

Table ISummary of the effects of Fn-fs on changes in IGFBP levels in CM and SDS extracts of bovine articular

chondrocytes in serum free high-density monolayer cultures

Cell extracts Conditioned media

IGFBP Mr29-kDa

Fn-f140-kDa

Fn-f Fn29-kDa

Fn-f140-kDa

Fn-f Fn

39 2 3 1 4 2 134 3 4 1.3 5 2 232 5 5 2 4 2 230 9 9 4 4 2 324 1 1 1 0.6 0.6 0.8

Cells were cultured for 48 h with 100 nM 29-kDa Fn-f, 100 nM 140-kDa Fn-f or 100 nM fibronectin (Fn). Theconcentrated media and the SDS extracts were analysed by ligand blotting and quantified by densitometry. Thechanges in IGFBP levels are expressed as a fold increase of IGFBP under the experimental conditions, withrespect to the untreated control.

HE IGFBPS ELEVATED BY THE FN-FS INCLUDED IGFBP-2 AS

HOWN BY IMMUNOPRECIPITATION

To simply identify the 24- and 32-kDa IGFBPs, con-itioned media were immunoprecipitated, as described inethods, with antisera against IGFBP-2 or IGFBP-4; therecipitated protein was analysed by ligand blotting. Figure

(left lane) shows that the 24-kDa band found in theultures and which was decreased in intensity in theonditioned media by the 29-kDa Fn-f (Fig. 1) wasGFBP-4. However, since the IGFBP-4 antiserum hasbout 50% cross-reactivity with IGFBP-2 (Per UBI, Inc.),ome 32/34-kDa IGFBP-2 was also detected in the anti-GFBP-4 lane. As shown in the right lane, the precipitate


contained 34-kDa IGFBP-2, one of the proteinsup-regulated by the 29-kDa Fn-f. Notably missing from theIGFBP-4 precipitate is a band of glycosylated IGFBP-4 at30-kDa, further suggesting that the 30-kDa IGFBPincreased by the Fn-fs is IGFBP-5.

THE FN-FS INCREASED CELL BOUND IGFBP-2 AS SHOWN BY

IMMUNOFLUORESCENCE

To confirm the effect of Fn-fs on IGFBP-2 levels and todetermine sites of deposition, monolayer cultures treated

with the 29-kDa Fn-f for 48 h in DMEM were reacted withanti-IGFBP-2 followed by incubation with rhodaminelabeled antirabbit IgG and fluorescence microscopy. Figure5 shows that IGFBP-2 was strongly increased by the Fn-fand that the binding protein was found surrounding thechondrocytes. These data are consistent with the dataabove showing up-regulation of IGFBP-2 and the strongpresence of IGFBP-2 in the cell extracts. When similarcultures were probed with anti-IGFBP-4, the Fn-f treatedcells showed only background fluorescence, at the level ofuntreated cells (data not shown); confirming that IGFBP-4was not localized around the chondrocytes.

Fig. 2. Effect of varying Fn-f concentration on IGFBPs as shown byligand blot analysis. Cultures were established as described in Fig.1. The ligand blot shows extracts of cells cultured with no Fn-f(Control) or with 100 nM or 1 �M 29-kDa Fn-f or 140-kDa Fn-f. The

positions of 30-kDA and 32-kDa are shown on the left.

Fig. 3. Immunoprecipitation of IGFBP-2 and IGFBP-4 and visuali-zation of IGFBPs by ligand blot analysis. Media conditioned byculturing chondrocytes with 100 nM 29-kDa Fn-f were precipitatedwith antisera against IGFBP-2 or -4 and the precipitates wereanalysed by ligand blotting. The positions for IGFBP-4 and -2 are

shown on the left, respectively, at 24-kDa and 34-kDa.

ANTISERA AGAINST THE �5�1 FN RECEPTOR ENHANCED IGFBP

LEVELS AND SUPPRESSED PG SYNTHESIS

Since other properties of the Fn-fs such as MMP-3release were affected by antibodies to this Fn receptor(unpublished data), we tested the effect of rabbit antihuman�5�1 antiserum first on IGFBP levels and then on PGsynthesis. PG synthesis was tested because suppressionof PG synthesis is a significant activity of the Fn-fs11,12.

THE FN-FS INCREASED IGFBP-3 MRNA, BUT HAD NO EFFECT ON

IGFBP-4 MRNA

Since no antibodies were available against bovineIGFBP-3, the effect of Fn-fs on IGFBP-3 was confirmed byNorthern blot analysis. IGFBP-4 mRNA was also analysedsince the effects of the Fn-fs decreased the level of thisprotein in the previous experiments. As shown in top panelof Fig. 4, the IGFBP-4 cDNA hybridized to a 2.2 kb mRNA,consistent with the published size of the bovine IGFBP-4mRNA29. Densitometric scanning and normalization ofIGFBP-4 mRNA to 18s rRNA indicated that there was nosignificant difference between the Fn-f treated cultures andthe untreated control, based on N=3. The cDNA probe forbovine IGFBP-3 (Fig. 4, bottom panel), hybridized to asingle 1.6 kb mRNA. This size corresponds to the sizereported for IGFBP-330. Normalization of the IGFBP-3mRNA to 18s rRNA showed a 2.3 fold increase in IGFBP-3mRNA.


Fig. 6 shows that culturing chondrocyte monolayers withantihuman �5�1 for 48 h resulted in up-regulation of the 32and 30-kDa IGFBPs, which were probably IGFBP-2 and -5,respectively (panel A), as well as dose-dependent suppres-sion of PG synthesis (panel B). Antigoat IgG dilutions hadno effect on either activity. Therefore, the Fn-f effects onIGFBPs may be occurring through this receptor.

IGF-I ALSO ENHANCED IGFBP LEVELS

Since the Fn-fs enhance release of IGF-I and IGF-I isknown to elevate IGFBPs, we tested the effect of IGF-I inthis system. Figure 7 shows that within 48 h, 50 ng/ml IGF-I(I) increased levels of IGFBPs in the cell extract (CE), of30-, 32- and 34-kDa bands that are probably IGFBPs-4, -5and -2, respectively. The corresponding conditioned media(CM) show in the treated cells the additional presence of anintense 24-kDa band, which is probably IGFBP-4, and lightbands of 34- and 42-kDa, which probably correspond to

IGFBPs-2 and one of the bands of the IGFBP-3 doublet.This is consistent with the observation that IGF-I caninduce IGFBPs31. However, comparison with Fig. 1 showsthat a major difference between the Fn-f and IGF-I effects isthat Fn-fs decreased IGFBP-4, while it was increased byIGF-I.

Fig. 4. Effect of Fn-fs on IGFBP-4 and IGFBP-3 mRNA levels.Chondrocytes were cultured for 48 h without Fn-f (C) or with100 nM 29-kDa or 140-kDa Fn-fs. Total RNA was analysed with anIGFBP-4 (top panel) or an IGFBP-3 cDNA probe (bottom panel).The autoradiogram of the Northern blot is shown in top part of eachpanel and the ethidium bromide stained 18s rRNA is shown in the

lower part.

Fig. 5. Effect of Fn-fs on cellular bound IGFBP-2 epitopes. Bovinearticular chondrocytes were cultured in chamber slides for 48 h,with or without 100 nM 29-kDa Fn-f in serum-free DMEM. Cellswere fixed with 1.5% paraformaldehyde in PBS, then incubatedwith antisera against IGFBP-2 and finally with rhodamine-labeledantirabbit IgG secondary antibody. Binding was visualized by UVfluorescence microscopy. The left panels are phase contrast of−Fn-f and +Fn-f, while the right panels are the corresponding

fluorescent images. The final magnification was 150×.

THE EFFECT OF FN-F ENHANCED IGFBP LEVELS WAS TO

SUPPRESS PG SYNTHESIS AS SHOWN BY USE OF IGF-I MUTANTS

To address the role of the IGFBPs in 29-kDa Fn-fmediated PG synthesis suppression, we asked whether thetruncated IGF-I analog, des (1-3) IGF-I was more or lesseffective at reversing PG synthesis suppression that wildtype IGF-I. Des (1-3) IGF-I binds to the type I IGF-I receptorwith the same affinity as IGF-I, but des (1-3) IGF-I hasgreatly reduced affinity for the IGFBPs32. Monolayer cul-tures were first incubated for either 48 or 72 h with the29-kDa Fn-f, then either the des(1-3) IGF-I or IGF-I wasadded for an additional 24 h. Finally, sulfate incorporationinto the cell layer was assayed.

Figure 8, panel A, shows that the 48-h incubation withthe 29-kDa Fn-f suppressed PG synthesis to 20% of controllevels (P≤0.01), while addition of IGF-I (20 ng/ml final) orDes (1-3) IGF-I (20 ng/ml final) during the last 24 h ofculture restored PG synthesis to approximately 50%(P≤0.01) of the untreated control level. Since, as shown inpanel B, 20 ng/ml of either factor should have increasedPG synthesis to nearly four-fold of control untreated cells,this suggests that a component(s), probably IGFBPs,up-regulated by the Fn-fs, is offsetting the effects of thegrowth factors on PG synthesis. Surprisingly, the Des IGF-Idid not reverse the effect of the Fn-f more than did IGF-I.

However, in cultures incubated for 72 h with the 29-kDaFn-f and with the growth factors (panel C), IGF-I restoredPG synthesis to the level of the control (P≥0.05) (controlsshown in Panels B and D), while the des (1-3) IGF-Ielevated the PG synthesis to 140% of control (P≤0.05).Thus, with the longer incubation, the 29-kDa Fn-f inducedfactors, probably IGFBPs, that suppress PG synthesis andwhich can be effectively counteracted by IGF-I. Even more


effective in reversing PG synthesis suppression was des(1-3) IGF-I with its reduced affinity for IGFBPs and there-fore greater availability to the IGF-I receptor. Panel Dshows the effects of the growth factors incubated in the last24 h in the 72-h cultures.

THE EFFECT OF FN-F ENHANCED IGFBP LEVELS WAS TO

SUPPRESS PG SYNTHESIS AS SHOWN BY USE OF ANTISERA

AGAINST IGFBPS

To provide confirmatory data for the role of IGFBPs in PGsynthesis suppression, antisera against IGFBP-2 and -4were tested for their effects on PG synthesis by monolayercultures incubated for 48 h with 100 nM 29-kDa [Fig. 9(A)]or 140-kDa Fn-f [Fig. 9(B)] with or without the antisera.

Anti-IGFBP-2 was tested because IGFBP-2 is one of themajor IGFBPs up-regulated by the Fn-f and found in the cellextracts. Anti-IGFBP-4 was tested because the data to thispoint had suggested that it did not have a major role in Fn-fmediated activities and, in fact, it was not up-regulated atthe mRNA level.

As shown in Fig. 9(A), the 29-kDa Fn-f significantlysuppressed PG synthesis, while the IGFBP-2 antiserumalone increased PG synthesis 63% above control levels(P≤0.01). The IGFBP-2 antiserum that was added to thecultures 24 h prior to the addition of the 29-kDa Fn-fsignificantly reversed or blocked the PG synthesis suppres-sion caused by the 29-kDa Fn-f (P≤0.05). Therefore, a rolefor IGFBP-2 in Fn-f mediated PG synthesis suppression issuggested. In contrast, the IGFBP-4 antiserum alone didnot enhance PG synthesis, nor did the IGFBP-4 antiserumblock the effect of the Fn-f on suppressing PG synthesis[Fig. 9(A)]. Neither IGFBP-2 nor IGFBP-4 antisera had asignificant effect on PG synthesis suppression by the140-kDa Fn-f [Fig. 9(B)].

Fig. 6. Effect of anti-�5�1 antiserum on IGFBP levels and PGsynthesis. Chondrocytes were cultured for 48 h with anti-�5�1

antiserum or an irrelevant antigoat IgG antiserum. Panel A: cellextracts were analysed for IGFBPs by ligand blotting. Shown areuntreated control (left), antigoat IgG (middle) and anti Fn receptor(�5�1) (right). The positions for 32-kDa and 30-kDa IGFBPs (top tobottom) are shown in the margin on left. Panel B: newly synthe-sized PG was labeled with a two-hour pulse of 35SO4 in culturestreated with increasing concentrations of antibody. Sulfate incor-poration was quantified using the alcian blue binding assay. Cellextracts and media were assayed and added to reflect total

labeled PG.

Fig. 7. Effect of IGF-I on IGFBP expression. Chondrocytes werecultured for 48 h with IGF-I in serum-free DMEM. Cell extracts(CE) (left panel) and conditioned media (CM) (right panel) wereanalysed by ligand blotting. Each panel shows cell extracts fromcontrol cells (C) and IGF-I (I) treated cells. The locations of 24-kDa,30-kDa and 32-kDa molecular weight standards are shown on left.

THE 29-KDA FN-F ENHANCED RELEASE OF IGFBP-2 AND IGF-I

FROM CULTURED CARTILAGE AS SHOWN BY FLUORESCENCE

MICROSCOPY

Since IGF-I and IGFBP-2 can accumulate in the cartilagematrix, we investigated whether the Fn-f could causerelease of exogenous rhodamine labeled IGF-I or IGFBP-2.Rhodamine labeled IGFBP-2 was incubated for 24 h with


bovine cartilage explants in serum-free DMEM and themedia were replaced with serum-free DMEM with or with-out 100 nM 29-kDa Fn-f. After 48 h, 15 sections from threedifferent explants were examined. Figure 10 shows thataddition of labeled IGFBP-2 resulted in enhanced fluor-escence around the cells and that Fn-f treatment enhancedrelease of the labeled IGFBP-2. Figure 11 shows that undersimilar conditions, when rhodamine labeled IGF-I wasadded to cartilage for 24 h, followed by addition of Fn-f,there was also an initial accumulation around chondrocytesand that Fn-f treatment decreased the labeled IGF-I in thetissue.

MMP-3 AND TRYPSIN ALSO FIRST SUPPRESSED PG SYNTHESIS,

THEN ALLOWED INCREASES TO SUPERNORMAL RATES WHILE IGF-I

RELEASE WAS ENHANCED

Since it is known that proteolysis of binding proteins canliberate IGF-I, we tested whether MMP-3 alone could mimicthe action of the Fn-f. Figure 12 shows that adjustment ofcartilage cultures to 5 �g/ml MMP-3 in DMEM led to areduction of PG content as well as an initial suppression ofPG synthesis. However, after the MMP-3 was removed, PGsynthesis rates increased to above normal levels. This wasassociated with enhanced release of IGF-I into the media.However, this anabolic response was insufficient to restorePG content by day 9.

Fig. 8. Effect of IGF-I and Des (1-3) IGF-I on PG synthesis by BACtreated for 48 or 72 h with Fn-fs. (A) BAC were incubated for 24 hwith 100 nM 29-kDa Fn-f. IGF-I or Des(1-3) IGF-I was added to afinal concentration of 20 ng/ml and the cultures were incubated fora further 24 h. During the final 4 h of each incubation 10 �Ci of35SO4 were added. (C) Same as (A) except the Fn-f treatment was48 h with the growth factors added for the final 24 h. Totalincubation was 72 h. Panels B and D show control incubations with

growth factor alone for 48 and 72 hours, respectively.

Discussion

Our objective was to test the hypothesis that Fn-fs might,in part, regulate PG synthesis through a mechanism involv-ing IGF-I and IGFBPs. The hypothesis was based on ourobservations that the Fn-fs suppress PG synthesis11,12 andenhance release of IGF-I and catabolic cytokines fromexplant cultures of human articular cartilage4. Since bothIGF-I32 and catabolic cytokines regulate IGFBPs33, it wasalso of interest to test for induced IGFBPs in our Fn-fmodel. An additional rationale was that this Fn-f modelexpresses certain characteristics of OA chondrocytes andIGF-I and IGFBPs are enhanced in OA34; in situ hybridiz-ation35 and immunohistochemical36 studies have shownthat IGF-I mRNA and protein are increased in chondrocytesthat surround osteoarthritic lesions. The increased expres-sion of IGF-I protein was found to be proportional to theseverity of the lesion36. Paradoxically, cultured osteo-arthritic chondrocytes are resistant to the effects of IGF-I37–

39. Although there is no evidence of damage to or reductionin the number of IGF-I receptors on osteoarthritic chondro-cytes, osteoarthritic cartilage shows increased binding sitesfor IGF-I39,40. Thus, it has been proposed that theenhanced number binding sites in osteoarthritic cartilage isdue to enhanced levels of IGFBPs. Bhakta et al. havedemonstrated that IGFs bind the cartilage matrix indirectlythrough IGFBPs41.

To investigate the effects of Fn-fs on IGFBPs, bothconditioned media and cell extracts of monolayer cultureswere analysed. Serum-free monolayer cultures werestudied rather than explant cultures, because this systemallowed us more easily to isolate RNA and the serum-freeculture conditions eliminated IGF-I and IGFBPs containedin serum. However, the use of serum-free monolayer cul-tures did not allow us to study long-term cultures orlong-term kinetics of expression. Data obtained frommonolayer cultures of chondrocytes are valuable for under-standing the response of chondrocytes to stimuli, however,care must be taken in the extrapolation of the data tocartilage; the matrix of the monolayer is not the same ascartilage. Nonetheless, we have shown that MMP-3 andother MMPs are up-regulated not only in explant cul-tures3,6,7, but also in monolayer cultures8. We also knowthat PG synthesis suppression occurs in explant cul-tures11,12,16 and as shown here, in monolayer cultures aswell. Thus, extrapolation of our results to explant culturesmay have some validity, although we would not expect thatthe activities of the Fn-fs would be of the same magnitudein both culture systems.

We had to use a combination of methods to identify theIGFBPs, since ligand blot analysis only provided masses ofIGFBPs, not their exact identities. Another parameter wasthe choice of Fn-fs. In order to ensure that our conclusionswould reflect Fn-fs in general, we tested two distinct typesof Fn-fs, a 29-kDa amino-terminal Fn-f and a central140-kDa Fn-f. The amino-terminal Fn-f is the most potent5

but does not bind to Fn receptors through the Arg-Asp-Ser-Gly recognition sequence, while the 140-kDa Fn-f is lesspotent5 but does bind directly to Fn receptors through thissequence. Although these Fn-fs are dissimilar, both typessuppress PG synthesis12, elevate MMPs3,5–8 and causecartilage matrix degradation3. We have recently shownthat the activities of both types of Fn-fs also can besuppressed by antisense oligonucleotides to the �5 subunit(Osteoarthritis Cart, in press). Therefore, our conclusionshere probably represent activities of Fn-fs in general.


Fig. 9. Effect of IGFBP-2 or -4 antiserum on PG synthesis by BAC. Chondrocytes were incubated for 24 h with a 1:1000 dilution ofanti-IGFBP-2 or IGFBP-4 antiserum before the addition of 100 nM 29-kDa Fn-f. The cells were incubated for an additional 48 h with Fn-f

and/or antisera. Newly synthesized PG was labeled with a 2-h pulse of 35SO4 and incorporation quantified.

Fig. 10. Effect of Fn-fs on association of rhodamine labeledIGFBP-2 with cartilage. Rhodamine labeled IGFBP-2 was incu-bated for 24 h with cartilage tissue. The IGFBP-2 media wereremoved and replaced with either SF DMEM (-Fn-f) or with SFDMEM+100 nM 29-kDa Fn-f (+Fn-f). The explant cultures wereincubated for a further 48 h with these media. The tissues wererinsed and prepared for frozen sectioning. The phase contrastmicrographs on the left were taken with an automatic exposurewhile the fluorescent micrographs on the right were exposed for

30 s. Images are representative of 30 sections examined.

Fig. 11. Effect of Fn-fs on association of rhodamine labeled IGF-Iwith cartilage. Bovine articular cartilage explant cultures wereincubated for 24 h with 100 nM rhodamine labeled IGF-I. The IGF-Imedium was replaced with either serum free DMEM (-Fn-f) or thesame with 100 nM 29-kDa Fn-f (+Fn-f) and the cultures incubatedfor a further 48 h. The tissue was rinsed and prepared for frozensectioning. The phase contrast micrographs on the left were takenwith an automatic exposure while the fluorescent micrographs onthe right were exposed for 30 s. Images presented are represen-tative of 15 sections examined. Final magnification for these was

50×.

We found that the Fn-fs increased levels of IGFBPs of24-kDa to 42-kDa, corresponding to IGF-2, -3 and -5 asshown by ligand blot analysis. IGFBPs-2 and -4 wereidentified by immunoprecipitation followed by ligand blot-ting. Because specific antisera against bovine IGFBP-3were not commercially available, the identity of the 43/39-kDa doublet as IGFBP-3 was inferred from its molecularweights and from Northern blot analysis that confirmed

enhanced levels of IGFBP-3 mRNA. The presence of theseIGFBPs in bovine chondrocyte cultures is consistent withthe work of Olney et al.31 that showed IGFBP-2, -3, and -4in cultures of BAC. It was interesting that an antibodyagainst the �5�1 Fn receptor also increased IGFBPs.Induction by this antibody suggests that the effect of theFn-fs on IGFBPs may be mediated through the Fn receptor.

The observation that the Fn-fs enhanced levels ofIGFBP-2 and -3 can be compared to a report that inosteoarthritic synovial fluid only IGFBP-3 is elevated, whilein rheumatoid arthritic synovial fluid, BPs -2 and -4 areelevated in addition to BP-337. While the Fn-fs increased


Fig. 12. Effect of MMP-3 and trypsin on PG content, PG synthesisrate and IGF-I release from cartilage explants into the media.Bovine articular cartilage in serum-free DMEM was incubated with5 �g/ml MMP-3 for days 0–6. The media were then removed andreplaced with 10% serum/DMEM. Panel A shows the PG contentduring the culture. Curves are control cartilage (C) and MMP-3treated cartilage (h). Panel B shows the PG synthesis rates.Curves are control cartilage (C) and MMP-3 treated cartilage (h).Panel B shows the IGF-I released into the media after correctingfor the serum contribution beyond day 6. Curves are control

cartilage (C) and MMP-3 treated cartilage (h).

IGFBPs -2, -3 and -5, they decreased the level of IGFBP-4.As shown by ligand and Northern blotting respectively,IGFBP-4 protein was decreased while the level of IGFBP-4mRNA remained unchanged. The decrease in IGFBP-4protein may have been a result of increased proteolysisinduced by the Fn-fs. Proteolysis is a well-documentedmechanism of post-translational regulation of IGFBPactivity; serine proteases as well as MMPs can degradeIGFBPs to non-IGF-I binding fragments42–45. Such serineproteases have been identified in several different types ofcells42,46–48. Additionally, a kallikrein-like serine proteasethat cleaves IGFBP-5 into low-mass non-IGF-I bindingfragments has been identified36. Since IGFBP-4 has canbe degraded by serine proteinases39, it is possible that aserine proteinase activity might also be involved in Fn-feffects on cartilage tissue. Interestingly, IGF-I itself maypartly regulate degradation of IGFBPs since binding of

IGF-I by IGFBP-5 has been shown to stabilize this IGFBPagainst proteolysis by a specific IGFBP-5 protease42.

The effect of the up-regulated IGFBPs on chondrocytemetabolism must be discussed in relation to the reporteddualistic activities of the IGFBPs. IGFBPs can both blockand enhance the activity of IGF-I. In this context, the role ofthe extracellular matrix in regulating the IGF-I bindingactivity of IGFBPs also must be considered. For example,while soluble IGFBPs can attenuate the activity of IGF-I,IGFBPs associated with the cell surface PGs or matrix PGscan potentiate the effect of IGF-I49. Perhaps this can beexplained by observations that the interaction with thematrix may facilitate dissociation of IGF-I from complexeswith IGFBPs. For example, binding of the IGFBP-2/IGF-Icomplex to chondroitin-6-sulfate reduces the affinity ofIGFBP-2 for IGF-I 3-fold50. In addition, other extracellularmatrix components that bind IGFBPs, such as heparin,heparan sulfate or dermatan sulfate can also reduce theaffinity of IGFBP-3 or -5 for IGF-I51. This regulation by thematrix can also be explained by observations that PGsslow degradation of IGFBP-5 by an IGFBP-5 specificprotease52.

Based on the above observations, it is likely that at leastsome IGFBPs serve to enhance delivery of IGF-I tochondrocytes. Within the cartilage tissue, de novo synthe-sized IGFBPs may serve to trap de novo synthesized IGF-Iand to sequester it as an inactive complex until needed anduntil activated through protease activity. Current work isdirected toward testing for enhanced degradation ofIGFBPs in the Fn-f treated cultures.

One of the rationales for investigating whether Fn-fsinduce IGFBPs was based on observations that IGF-Irelease is elevated in conditioned media of Fn-f treatedcartilage4 and that IGF-I enhances IGFBP levels53,54. Forexample, in FRTL-5 cells, IGF-I enhances IGFBP-5 in theconditioned media with a corresponding increase inIGFBP-5 mRNA54. However, when we tested IGF-I onBAC, it produced a different IGFBP profile than did Fn-fs. Incontrast to the Fn-fs, IGF-I increased IGFBP-4. Thus, theFn-f effects on IGFBPs cannot be explained solely througheffects on enhancement of IGF-I activity. Similarly, thesuppression of PG synthesis by Fn-fs is unlikely to occursolely through modulation of the IGF/IGFBP axis.

The availability of IGF-I analogs that bind with normalaffinity to the type I IGF-I receptor, but have negligibleaffinities for the IGFBPs32 allowed us to test the effect ofthe up-regulated IGFBPs on IGF response of Fn-f treatedcells. The IGF-I analogue des (1-3) IGF-I was more activein restoring PG synthesis following treatment with the29-kDa Fn-f than was IGF-I. This result, coupled with ourobservation that the PG synthesis was partially restoredfollowing treatment of 29-kDa Fn-f cultured cells with theIGFBP-2 antiserum, suggests that the overall effect of theincreased levels of the IGFBPs in this system is initially tosuppress PG synthesis. These data are consistent with ourobservations that Fn-fs added to cartilage results in im-mediate suppression of PG synthesis12 and suggest thatIGFBPs may play a role in this effect. The inability to causefull restoration may suggest that another pathway, such ascatabolic cytokines, also contribute to PG synthesis sup-pression. Since catabolic cytokines can also mediate PGsynthesis suppression by Fn-fs3,4, it is possible that bothcytokines and IGFBPs may be involved in PG synthesissuppression, although they may not be effective at thesame time. It would be of interest to determine how much ofthe suppressive role that has been assigned to cataboliccytokines could occur through IGFBPs. However, since


References

1. Mankin HJ, Johnson ME, Lippiello L. Biochemical andmetabolic abnormalities in articular cartilage fromosteoarthritic human hips. J Bone Joint Surg (Am)1981;63:131–40.

2. Lafeber FPG, van Roy H, Wilbrink B, Huber-Bruning O,Bijlsma JWJ. Human osteoarthritic cartilage is syn-thetically more active but in culture less vital thannormal cartilage. J Rheum 1992;19:123–9.

3. Homandberg GA, Hui F, Wen C. Association of prote-oglycan degradation with catabolic cytokine andstromelysin release from cartilage cultured withfibronectin fragments. Arch Biochem Biophys1996;334:325–31.

4. Homandberg GA, Hui F, Wen C, Purple C, Bewsey K,Koepp H, et al. Fibronectin fragment induced carti-lage chondrolysis is associated with release of cata-bolic cytokines. Biochem J 1997;321:751–7.

5. Homandberg GA, Meyers R, Xie DL. Fibronectinfragments cause chondrolysis of bovine articularcartilage slices in culture. J Biol Chem 1992;267:3597–604.

6. Xie DL, Homandberg GA. Fibronectin fragments bindand penetrate cartilage tissue resulting in proteaseexpression and cartilage damage. Biochim BiophysActa 1993;1182:189–96.

7. Xie DL, Hui F, Meyers R, Homandberg GA. Cartilagechondrolysis by fibronectin fragments is associatedwith release of several proteinases; stromelysinplays a major role in chondrolysis. Arch BiochemBiophys 1994;311:205–12.

8. Bewsey K, Wen C, Purple C, Homandberg, GA.Fibronectin fragments induce the expression ofstromelysin-1 mRNA and protein in bovine chondro-cytes in monolayer culture. Biochim Biophys Acta1996;1317:55–64.

9. Homandberg GA, Hui F, Manigalis C, Shrikhande A.Cartilage chondrolysis caused by fibronectin frag-ments causes cleavage of aggrecan at the samesites as in osteoarthritis. Osteoarthritis Cart1997;5:450–3.

10. Kang Y, Koepp H, Cole AA, Kuettner KE, HomandbergGA. Cultured human ankle and knee cartilage differIn susceptibility to damage mediated by fibronectinfragments. J Orthop Res 1998;16:551–6.

11. Homandberg GA, Hui F. High concentrations offibronectin fragments cause short term cataboliceffects in cartilage tissue while lower concentrationscause continuous anabolic effects. Arch BiochemBiophys 1994;311:213–8.

12. Xie DL, Hui F, Homandberg GA. Fibronectin fragmentsreversibly alter matrix protein synthesis in cartilage invitro. Arch Biochem Biophys 1993;307:110–18.

13. Homandberg GA, Hui F. Arg-Gly-Asp-Ser analogs sup-press cartilage chondrolysis activities of integrin-binding and non-binding fibronectin fragments. ArchBiochem Biophys 1994;310:40–8.

14. Xie DL, Meyers R, Homandberg GA. Fibronectinfragments in osteoarthritic synovial fluid; pathologicimplications. J Rheumatol 1992;19:1448–52.

15. Homandberg GA, Meyers R, Williams J. Intra-articularinjection of fibronectin fragments induces cartilageproteoglycan release. J Rheumatol 1993;20:1378–82.

16. Homandberg GA, Wen C. Exposure of cartilage to afibronectin fragment amplifies catabolic processeswhile also enhancing anabolic processes to limitdamage. J Orthop Res 1998;16:237–46.

17. Luyten FJP, Hascall VC, Nissley SP, Morales TI, ReddiAH. Insulin-like growth factors maintain steady-statemetabolism of proteoglycans in bovine articular

catabolic cytokines also up-regulate IGFBPs as well as anIGFBP-3 protease activity33 the answer may be difficult toobtain.

Throughout these studies we considered a model inwhich the Fn-fs might initially up-regulate IGFBPs and trapIGF-I within the matrix at a time when PG synthesis is alsosuppressed. We also considered that during the damagephase, increased levels of proteases eventually liberateIGF-I, perhaps through proteolysis of IGFBPs, leading togreatly enhanced PG synthesis. Thus, we tested whetheraddition of Fn-fs to cartilage tissue that had been incubatedwith labeled IGFBP-2 or IGF-I would enhance release ofthe labeled proteins. Although release occurred, we cannotbe certain that these experiments mimicked the physiologicscenario and that the release occurred through proteolysisof IGFBPs. In fact, it is possible that the enhanced releaseof IGF-I or IGFBPs was not due to proteolysis of IGFBPs,but due to the proteolytic release of PGs, which can trapthe IGFBP/IGF-I complex in the matrix. Nonetheless, theexperiments did prove that the Fn-fs could enhance releaseof IGF-I and IGFBP-2 from the cartilage matrix. This isconsistent with our earlier work showing that Fn-fs enhancerelease of IGF-I and that this is associated with enhancedPG synthesis3,4, another component of our model. It shouldbe noted that we showed earlier that the enhanced releaseof IGF-I was not due significantly to de novo synthesissince metabolic inhibitors did not block the enhancedrelease4. Thus, although it is possible that Fn-fs mayenhance IGF-I synthesis albeit at low levels, the bulk of theenhanced IGF-I levels measured in conditioned media arelikely from IGF-I stored in the matrix. Thus, the effects of theFn-fs on mobilization of IGF-I are most likely occurringthough indirect effects on the matrix.

In summary, Fn-fs elevated IGFBPs and this was associ-ated with release of IGF-I and IGFBPs (namely IGFBP-2)from cartilage tissue, consistent with a model in which Fn-fsgenerated during cartilage damage might up-regulateIGFBPs which trap newly synthesized IGF-I around thechondrocytes cell matrix, at a higher level of depositionthan found in non-damaged cartilage, and in a way as toprotect IGF-I from degradation. However, as the matrixdegradation proceeds, IGF-I now highly concentratedaround the cells might be liberated slowly and enhance PGsynthesis to a level greater than that in undamaged carti-lage. Thus, this model might explain the delayed anabolicresponse observed with Fn-f treatment of cartilage11,16 inwhich PG synthesis rates increase after Fn-f exposure tolevels higher than in untreated cartilage. Since these ana-bolic events of enhanced IGF-I release and PG synthesiscould be mimicked by exogenous MMP-3, it is plausiblethat matrix degradation in this IGF-I activation model mightplay a role. Thus, addition of proteinases at an amount toolow to cause extensive matrix damage, under the rightconditions, ironically might enhance reparative processes.While this proposed model might be useful in explaining theability of Fn-fs to enhance anabolic activities after the initialcartilage damage, there are likely other anabolic pathwaysinitiated in damaged cartilage in general and especially incartilage damaged through other catabolic pathways.


cartilage explants. Arch Biochem Biophys 1988;267:416–25.

18. Hill DJ, Logan A, McGarry M, De Sousa DJ. Control ofprotein and matrix-molecule synthesis in isolatedovine fetal growth-plate chondrocytes by the inter-actions of basic fibroblast growth factor, insulin-likegrowth factors-I and -II, insulin and transforminggrowth factor-beta 1. J Endocrinol 1992;133:363–73.

19. Curtis AJ, Ng CK, Handley CJ, Robinson HC. Effect ofinsulin-like growth factor-I on the synthesis and dis-tribution of link protein and hyaluronan in explantcultures of articular cartilage. Biochim Biophys Acta1992;1135:309–17.

20. Tyler JA. Insulin-like growth factor 1 can decreasedegradation and promote synthesis of proteoglycanin cartilage exposed to cytokines. Biochem J1989;260:543–8.

21. Arner EC, Pratta MA. Modulation of interleukin-1-induced alterations in cartilage proteoglycan metab-olism by activation of protein kinase C. ArthritisRheum 1991;34:1006–13.

22. Chandrasekhar S, Harvey AK, Hrubey PS. Intra-articular administration of interleukin-1 causes pro-longed suppression of cartilage proteoglycansynthesis in rats. Matrix 1992;12:1–10.

23. Hwa V, Oh Y, Rosenfeld RG. The insulin-likegrowth factor-binding protein (IGFBP) superfamily.Endocrine Reviews 1999;20:761–87.

24. Laemmli UK. Cleavage of structural proteins during theassembly of the head if bacteriopage T4. Nature1970;227:680–5.

25. Hossenlopp P, Seurin D, Segovia-Quinson B,Hardouin S, Binoux M. Analysis of serum Insulin-likegrowth factor binding proteins using western blotting:use of the method for titration of the binding proteinsand competitive binding studies. Anal Biochem1986;154:138–43.

26. Unterman TG, Patel K, Mahathre VK, Rajamohan G,Oehler DT, Becker RE. Regulation of low molecularweight insulin-like growth factor binding proteins inexperimental diabetes mellitus. Endocrinology1990;126:2614–24.

27. Masuda K, Shirota H, Thonar E-JMA. Quantification of35S-labeled proteoglycans complexed to alcian blueby rapid filtration through multiwell plates. AnalBiochem 1994;217:167–75.

28. Kelley KM, Oh Y, Gargosky SE, Gucev Z, MatsumotoT, Hwa V, et al. Insulin-like growth factor-bindingproteins (IGFBPs) and their regulatory dynamics. IntJ Biochem Cell Biol 1996;28:619–37.

29. Moser DR, Lowe WL Jr, Dake BL, Booth BA, Boes M,Clemmons DR, Bar RS. Endothelial cells expressinsulin-like growth factor-binding proteins 2 to 6. MolEndocrinol 1992;6:1805–14.

30. Spratt SK, Tatsuno GP, Sommer A. Cloning andcharacterization of bovine insulin-like growth factorbinding protein-3 (bIGFBP-3). Biochem Biophys ResCommun 1991;177:1025–32.

31. Olney RC, Smith RL, Kee Y, Wilson DM. Productionand hormonal regulation of insulin-like growth factorbinding proteins in bovine chondrocytes. Endocrinol-ogy 1993;133:563–70.

32. Oh Y, Muller HL, Lee DY, Fielder PJ, Rosenfeld RG.Characterization of the affinities of insulin-like growthfactor (IGF)-binding proteins 1-4 for IGF-I, IGF-II,IGF-I/insulin hybrid, and IGF-I analogs. Endocrinol-ogy 1993;132:1337–44.

33. Olney RC, Wilson DM, Mohtai M, Fielder PJ, Smith RL.Interleukin-1 and tumor necrosis factor-alphaincrease insulin-like growth factor-binding protein-3(IGFBP-3) production and IGFBP-3 protease activityin human articular chondrocytes. J Endocrinology1995;146:279–86.

34. Tardif G, Reboul P, Pelletier JP, Geng C, Cloutier JM,Martel-Pelletier J. Normal expression of type 1insulin-like growth factor receptor by human osteo-arthritic chondrocytes with increased expressionand synthesis of insulin-like growth factor bindingproteins. Arthritis Rheum 1996;39:968–78.

35. Middleton, JFS, Tyler, JA. Upregulation of insulin-likegrowth factor I gene expression in the lesions ofosteoarthritic human articular cartilage. Annal RheumDis 1992;51:440–7.

36. Middleton JF, Tyler JA. Insulin like growth factor geneexpression by osteoarthritic and normal humanarticular chondrocytes. Trans Orthop Res Soc1991;16:384.

37. Tavera C, Abribat T, Reboul P, Dore S, Brazeau P,Pelletier JP, et al. IGF and IGF-binding protein sys-tem in the synovial fluid of osteoarthritic and rheuma-toid arthritic patients. Osteoarthritis Cart 1996;4:263–74.

38. Schalkwijk J, Joosten LA, van den Berg WB, van dePutte LBA. Chondrocyte nonresponsivness toinsulinlike growth factor 1 in experimental arthritis.Arthritis Rheum 1989;32:894–900.

39. Dore S, Pelletier JP, DiBattista JA, Tardif G, Brazeau P,Martel-Pelletier J. Human osteoarthritic chondrocytespossess an increased number of insulinlike growthfactor 1 binding sites but are unresponsive to itsstimulation. Possible role of IGF-I-binding proteins.Arthritis Rheum 1994;37:253–63.

40. Joosten LA, Helsen MM, van den Berg WB. Transientchondrocyte nonresponsiveness to insulin-likegrowth factor-1 upon H2O2 exposure is not related toIGF receptor damage. J Rheumatol 1991;18:585–90.

41. Bhakta NR, Garcia AM, Frank EH, Grodzinsky AJ,Morales TI. The insulin-like growth factors (IGFs) Iand II bind to articular cartilage via the IGF-bindingproteins. J Biol Chem 2000;275:5860–66.

42. Nam JT, Busby WH, Clemmons DR. Human fibroblastssecrete a serine protease that cleaves insulin-likegrowth factor binding protein-5. Endocrinology1994;135:1385–91.

43. Fowlkes J, Freemark M. Evidence for a novel insulin-like growth factor (IGF)-dependent protease regulat-ing IGF-binding protein-4 in dermal fibroblasts.Endocrinology 1992;131:2071–6.

44. Fowlkes JL, Enghild JJ, Suzuki K, Nagase H. Matrixmetalloproteinases degrade insulin-like growth fac-torbinding protein-3 in dermal fibroblast culture. J BiolChem 1994;269:25742–6.

45. Fowlkes JL, Suzuki K, Nagase H, Thrailkill KM. Pro-teolysis of insulin-like growth factor binding protein-3during rat pregnancy: a role for matrix metalloprotei-nases. Endocrinology 1994;135:2810–3.

46. Conover CA, Bale LK, Clarkson JT, Torring O. Regu-lation of insulin-like growth factor binding protein-5messenger RNA expression and protein availabilityin rat osteoblast-like cells. Endocrinology 1993;132:2525–30.


47. Cohick WS, Gockerman A, Clemmons DR. Vascularsmooth muscle cells synthesize two forms of insulin-like growth factor binding proteins which are regu-lated differently by the insulin-like growth factors. JCell Physiol 1993;157:52–60.

48. Durham SK, Riggs BL, Harris SA, Conover CA. Altera-tions in insulin-like growth factor (IGF)-dependentIGF-binding protein-4 proteolysis in transformedosteoblastic cells. Endocrinology 1995;136:1374–80.

49. Tesch GH, Handley CJ, Cornell HJ, Herington AC.Effects of free and bound insulin-like growth factorson proteoglycan metabolism in articular cartilageexplants. J Ortho Res 1992;10:14–32.

50. Russo VC, Bach LA, Fosang AJ, Baker NL, WertherGA. Insulin-like growth factor binding protein-2 bindsto cell surface proteoglycans in the rat brain olfactorybulb. Endocrinology 1997;138:4858–67.

51. Arai T, Parker A, Busby W Jr, Clemmons DR. Heparin,heparan sulfate, and dermatan sulfate regulate for-mation of the insulin-like growth factor-I and insulin-like growth factor-binding protein complexes. J BiolChem 1994;269:20388–93.

52. Arai T, Arai A, Busby WH, Clemmons DR. Gly-cosaminoglycans inhibit degradation of insulin-likegrowth factor binding-protein-5. Endocrinology1994;165:2358–63.

53. Chevalier X, Tyler JA. Production of binding proteinsand role of the insulin-like growth factor I bindingprotein 3 in human articular cartilage explants. Br JRheumatol 1996;35:515–22.

54. Backeljauw P, Dai Z, Clemmons D, D’Ercole AJ. Syn-thesis and regulation of insulin-like growth factorbinding protein 5 in FRTL-5 cells. Endocrinology1993;132:1677–81.

fibronectin fragments upregulate insulin-like growth factor binding proteins in chondrocytes

Documents