Binding of the Proteoglycan Decorin to Collagen Type VI*

THEJOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 267, No. 8,Issue of March 15, pp. 5250-5256,1992 Printed in U.S.A.

Binding of the Proteoglycan Decorin to Collagen Type VI* (Received for publication, July 15, 1991)

Deborah J. Bidanset$#, Clyde Guidryll, Lawrence C. Rosenbergll, HaingU. ChoiII, RupertTimpl**, and MagnusHook$ll From the Departments of $Cell Biology and (Biochemistry, University of Alabama at Birmingham, Birmingham, Alabama 35294, the 11Montefwre Medical Center, Orthopedic Research Laboratories, Bronx, New York 10467-2490, and the **Max-Planck-Znstitutfur Bwchemie, 0-8033 Martinsried/Munich, Federal Republic of Germany

We haveexaminedtheinteractionsbetweenthe found in the cornea appear to be associated with collagen small dermatan sulfate proteoglycan decorin and col- fibrils and are believed to participate in regulating the conlagen typesI-VI using solid phase binding assays. The trolled spacing between collagen fibrils that is necessary for results of these studies showed that 12‘I-decorin bound tissue transparency (3). The keratan sulfate proteoglycan most efficiently tocollagentype VI inatime-and fibromodulin binds to fibrillar collagen and modulates collaconcentration-dependentmanner.Furthermore, this gen fibrillogenesis in vitro (4, 5). Chondroitin/dermatan sulinteraction was specific and of moderately high affin- fate glycosaminoglycans have also been shown to interact ity ( K d -3 X lo” M). Binding of decorin to collagen with collagens (6).Ultrastructural studies on rat and bovine type VI appears to involve the decorin core protein tendon (7-lo),human and rat sclera (11),rabbit cornea and rather than the glycosaminoglycan side chains, since skin (3,8), bovine and rat bone, and various noncalcifying the isolated core protein as well as a recombinant fuconnective tissues (8) have revealed that sulfated proteoglysion protein containing amajor segment (66%)of the cans (e.g. dermatan sulfate proteoglycans) are often associated human decorin core protein inhibited binding of 1261decorin to collagen type VI. Other related proteogly- with the major collagen fibrils (types I and 11) with an axial cans and their respective core proteins also inhibited distance that is D-periodic. Using various ultrastructural techniques, the dermatan sulfate proteoglycans were further lothe binding of 1261-decorin to collagen type VI, whereas calized to the“d” and “e” bands of the D-period. Biochemical unrelatedproteinsandisolatedglycosaminoglycan studies have indicated that dermatan sulfate proteoglycans chainswerewithout effect. In additiontodecorin, associated with the collagen fibrils may regulate fibrillogenesis collagen typeI1 was also shown to bind to immobilized and could 13) possibly prevent collagen type VI. Both interactions were effectively of fibrillar collagens (6, 9, 12, calcification of soft connective tissues (8). inhibited by preincubation of the immobilized collagen VI with decorin or collagen type 11. These results sug- A small dermatan sulfate proteoglycan, later identified as gested that the collagen typeVI molecule has binding decorin, has been shown to bind to fibrillar collagens and sites for collagen typeI1 and decorin which are locatedinhibit fibrillogenesis in vitro (9, 12). The molecular mass of in close proximity on the collagen type VI molecule. decorin, isolated from articular cartilage, ranges from 87 to Possible functional roles of these interactions are dis- 120 kDa (14,15)and arises from a core protein of 43 kDa and cussed. a single galactosaminoglycan side chain of variable size. The core protein contains 10 repeats of a leucine-rich motif also found in several other proteoglycan core proteins (e.g. biglycan and fibromodulin) and other apparently unrelated proThe extracellular matrix is composed mainly of collagens, teins (5, 16). Decorin is found in the extracellular matrix of proteoglycans, and various glycoproteins. In recent years, a numerous tissues, including the adventitia of blood vessel large number of novel connective tissue macromolecules have walls, the dermis of the skin, tendon, ligament, sclera, articbeen described and their structures elucidated. The assembly ular cartilage, as well as other interstitial tissues (17, 18). of these macromolecules into extracellular matrices is thought Decorin has also been shown to bind a variety of other to be dependent on specific highaffinity interactions between proteins, including fibronectin (19, 20), transforming growth individual components. One presumably important butpoorly factor-@(21),as well as a high affinity cell surface receptor understood interactionis that between proteoglycans and found on human osteosarcoma cells and fibroblasts, which collagens. mediates endocytosis of the proteoglycan (22).In this study, Proteoglycan interactions with fibrillar collagens have been we have examined the interactions between decorin and colimplicated in the regulation of extracellular matrix assembly lagen types I-VI using solid phase assays. Our results dem(for review see Refs. 1 and 2). Keratan sulfate proteoglycans onstrate aspecific moderately high affinity binding of decorin to collagen type VI. * This investigation was supported by National Institutesof Health The collagen type VI monomer, which is composed of three Grants AM27807 (to M. H.) and AR34614 and AR21498 (to L. R.), distinct LY chains, has a molecular mass of greater than 420 the Helen Keller Eye Research Foundation (to C . G.), andthe National Arthritis Foundation (to C . G.). The costs of publication of kDa (23-26). Electron microscopic examinations of rotary this article were defrayed in part by the payment of page charges. shadowed preparations (27),biochemical studies (28), and This article must therefore be hereby marked “advertisement” in analysis of the amino acid sequences deduced from the coraccordance with 18 U.S.C. Section 1734 solelyto indicate this fact. responding cDNA clones (29-35)indicate that the collagen Helen Keller Eye Research Foundation predoctoral fellow. To type VI monomer is composed of large globular domains at whom correspondence should be addressed Dept. of Biochemistry, University of Alabama at Birmingham, 1918University Blvd., BHSB each end linked by a triple helical domain of 105 nm. Soon 508, Birmingham, AL 35294-0005. Tel.: 205-934-3722;Fax: 205-934- after synthesis, collagen type VI monomers assemble into well defined oligomers which become the building blocks of the 1359. 5250

Binding of the Proteoglycan Decorin to Collagen Type VI microfibrils found in theextracellular matrix of many tissues and cell culture systems (36-38). In the extracellular matrix, collagen type VI may serve as a substratum for cell adhesion as several Arg-Gly-Asp-containing sequences found in the triple helical segment appear to berecognizedby cellular receptors (39,35). In addition, the globular domains in collagen type VI contain several modules similar to the collagen binding type A motifs present in von Willebrand factor (40, 26, 29, 31-35). In fact a direct binding of collagen type VI to immobilized collagentype I has been demonstrated previously (35). EXPERIMENTALPROCEDURES

Materials-Decorin (dermatan sulfate proteoglycan 11) and biglycan (dermatansulfate proteoglycan I) were isolated from bovine articular cartilage as described previously (14,15). A truncated recombinant form of the human decorin core protein was generously provided by Dr. David M. Mann (La Jolla Cancer Research Foundation, San Diego, CA) in the form of a prokaryotic expressed TrpE fusion protein, containing the 10 leucine-rich repeats of the human decorin core protein (amino acids 38-282). Collagen type VI was isolated from pepsin extracts of human placenta (28, 36). Human collagen types I, 111, IV, and V were obtained from Southern Biotechnology Associates, Inc., Birmingham, AL, and bovine collagen type I1 isolated from cartilage was generously provided by Dr. Edward J. Miller (Department of Biochemistry, University of Alabama a t Birmingham). All five collagentypes (I-V) were isolated by methods described by Miller and Rhodes (41). Fibromodulin isolated from calf articular cartilage (42) was generously provided by Dr. Anna Plaas (Shriners Hospital for Crippled Children, Tampa, FL). Na"'1 (16.9 mCi/mg) was purchased from Amersham Corp. Falcon 3072 microtiter tissue culture plateswere obtained from Becton Dickinson and Co., Lincoln Park, NJ, andImmulon 2 Removawells were purchased from Dynatech Laboratories Inc., Chantilly, VA. All other reagents used were procured from Sigma. Adsorption of Protein to Microtiter Wells-Microtiter wells were coated with either 2 pg collagen/well or 1% (0.5 mg/well) BSA' in 0.05 ml of PBS (8 mM NazHP04-7Hz0,1.5 mM KHZPO4,pH 7.2, containing 0.137 M NaC1, 2.7 mM KCl, 0.5mM Mg& 0.9 mM CaC122HZO) byincubation at 4 "C, overnight. After a brief rinse, nonspecific binding sites in the wells were blocked by incubation at room temperature for 1 h with 0.2 ml of 1%BSA in PBS. The wells were then rinsed with PBS andprepared for the appropriate assay. To determine the amount of collagen adsorbed to microtiter wells, collagen types I, 111, and VI were '251-labeled(see below) and diluted with unlabeled collagen to known specific activities. Two micrograms of each collagen type in 50 p1 of PBS were then allowed to adsorb in microtiter wells overnight followed by a brief rinse. The amounts of collagen adsorbed to the microtiter wells were determined and calculated to be 0.968 k 0.054 pg for collagen type I, 1.292 k 0.068 pg for collagen type 111, and 1.008 k 0.041 for collagen type VI. Decorin Binding Assay-Decorin (and the various collagens) was labeled with using "Enzymobeads" in a modified lactoperoxidase reaction which was performed according to the manufacturer's instructions (Bio-Rad). Labeled macromolecules were then separated from free iodine bygel permeation chromatography on a PD-10 column (Pharmacia LKBBiotechnology Inc.). The estimated specific activity of the isolated proteoglycan was 2-6 X lo6cpm/pg. Iodinated decorin was diluted into PBS and 2.5-5 X IO4 cpm in 0.05 ml/well were added to wells coated with the indicated proteins and incubated at 37 "C (see figure legends for more details). Unbound decorin was removed and combined with three successive rinses (0.1-0.2 ml of PBS containing 0.1% Tween 20) for radioactivity determination. Individual wells containing bound radiolabeled protein were removed and analyzed. Radioactivity associated with the bound and unbound radiolabeled decorin was then quantified with an LKB y counter. The recovery in the unbound and bound pools using this method was over 90%. Variations of this binding assay were occasionally used and are indicated in the text and figure legends. Preparation of Proteoglycan Core Proteins and Glycosaminoglycan

5251

Chins-Core proteins from decorin and biglycan were prepared by digestion of the proteoglycans with chondroitinase ABC (Seikagaku America, Inc., St. Petersburg, FL) according to Oike et al. (43, 44). Briefly, 1-2 mg of proteoglycans were dissolved in 0.1 ml of 0.1 M Tris-HCI, pH 8.0, containing 30 mM sodium acetate and protease inhibitors (0.2 mg/ml pepstatin A, 1 pg/ml leupeptin, 5 mM phenylmethanesulfonyl fluoride, and 10 mM EDTA). This mixture was incubated overnight at 37 "C with chondroitinase ABC at a concentration of 0.1 unit/mg proteoglycan. The mixture was diluted to 1 ml with 50 mM Tris-HC1, pH 8.0, containing 6 M urea and 6 mM CHAPS, and core protein was separated from the enzyme on a Mono Q column fitted on a fast protein liquid chromatography system (PharmaciaLKB). The column was eluted with a gradient of 0-1 M NaCl in the above buffer, and fractions containing core protein were combined and dialyzed extensively against PBS. Glycosaminoglycan side chains from decorin were prepared and isolated as described previously (14) Antibody Production and Isolation-Polyclonal antibodies against decorin were raised in rabbits receiving four intramuscular injections each containing 200pgof proteoglycan emulsified in 0.5 mlof complete Freund's adjuvant (initial injection) or in incomplete Freund's adjuvant (subsequent injections). The IgG fraction of the collected serum was isolated by affinity chromatography on a protein A-Agarose (Pierce Chemical Co.) column eluted with 0.1 M glycine HCl, pH 2.8. The purified antibodies were then extensively dialyzed against PBS. Binding Assay of Type II Collagen-Collagen type I1 was labeled with biotin essentially following the method described by Orr (45). Briefly, collagen type I1 was dissolved in PBS (1 mg/ml), and a 100fold molar excess of NHS-biotin (20 mM in N,N-dimethylformamide) was added. A tube with this mixture was fitted onto anend-over-end mixer and incubated a t 4 'C for 4 h. Unincorporated NHS-biotin was removed by extensive dialysis against PBS. Binding of biotinylated collagen type I1 to various substrates was examined in enzyme-linked immunosorbent-type assays. Various proteins were adsorbed in Falcon 3072 microtiter wells essentially as described above. Biotin-conjugated collagen type 11, diluted in PBS, was added to the wells in concentrations varying from 0.2 to 4 pg/ well and incubated for 2 h a t room temperature. The wells wererinsed three times with PBS and incubated with an avidin-alkaline phosphatase conjugate (Cappel, West Chester, PA), diluted in PBS containing 1%BSA, for 2 h a t room temperature. The wells were rinsed three times with PBS containing 0.1% Tween 20 and thendeveloped with a chromogenic substrate solution containing 1 mg/ml p-nitrophenyl phosphate (Sigma 104 phosphatase substrate) in a diethanolamine buffer (9.6% diethanolamine, 0.25 mM MgClZ,pH 9.8). After a 15-20-min incubation a t 37 "C,the absorbance at 405 nm was determined with a UVmax Kinetic Microplate Reader (Molecular Devices). RESULTS

Decorin Binds Preferentially toType VI Collagen-The binding of '251-labeled decorin to microtiter wells previously coated with collagen types I-VI or BSA was examined. Protein-coated wellswere incubated with -3 X 10' cpm "'1decorin for 3 h at 37 "C.The wells werethen washed, and the amount of bound lZ5I-labeledligand was determined. The results indicated that wells coated with collagen type VI bound higher levels of '251-de~~rin than did wells coated with the other collagen types (Fig. 1). At the specific activity used, close to 40% of the 1251-labeleddecorin added bound to collagen type VI coated wells. Wells coated with BSA alone bound approximately 7.5% of added radioactivity, and collagens I-V bound intermediate amounts of '251-decorin. Characterization of the Binding of Decorin to Type VI Collagen-The kinetics of decorin binding to collagen type VI were examined by incubating labeled decorin at 37 "C in microtiter wells coated with either collagen type VI or BSA for various time periods. The results indicate that thespecific binding of decorin to collagen type VI was a time-dependent ' The abbreviations used are: BSA, bovine serum albumin; PBS, process which reached maximal levels after 180 min of incuphosphate-buffered saline; CHAPS, 3-[(3-cholamidopropyl)dimeth- bation (Fig. 2 A ) . The binding of decorin to collagen type VI was partially ylammoniol-1-propanesulfonic acid; NHS-biotin, D-biotin-N-hydroxysuccinimidester. reversible. Radiolabeled decorin was added and incubated in

Binding of the Proteoglycan Decorin to Collagen Type VI

5252

microtiter wells coated with collagen type VI for 3 h. Unbound proteoglycan was removed, and excess amounts of unlabeled decorin were added. The incubation was then continued for varying time periods. Unbound material was removedand the amount of lZ5I-decorindisplaced as well as that which remained bound to theprotein coated well werequantified. The results indicate that about 50% of the labeled decorin initially bound to theimmobilized collagentype VI was displaced in a time-dependent process (Fig. 2B). The binding of decorin to collagen type VI was found to be a saturable process. Radiolabeled decorin was combined with unlabeled decorin to give a specific activity -35,000 cpm/pg, and increasing amounts of this mixture were added to collagen CI CII Clll CIV CV CVI BSA type VI or BSA-coated microtiter wells in 0.2 ml of PBS ADSORBED PROTEIN containing 0.1% BSA. After 3 h at 37 "C,the wells wererinsed FIG. 1. Binding of '*'I-decorin to collagen types I-VI. Mi- two times with 0.2 ml of PBS containing 0.1% BSA, and the crotiter Removawells, coated with the indicated collagen types or l% amounts of bound and unbound ligand were determined. The BSA in PBS, were incubated with 3 X 10' cpm '251-decorin in 0.05 ml amount of '261-decorinbound to immobilized collagentype VI of PBS for 3 h at 37 "C. After washing, the amount of bound radioactivity was quantified. Results are expressed as the means of and BSA increased as the amount of lZ5I-decorinadded was determinations taken from triplicate wells. Bars represent the range increased. However, when the amount of decorin specifically bound to collagen type VI coated wellswas analyzed, the of '261-decorinbound. binding appeared to reach a maximum of 0.2 pg bound when 2 5 pg of decorin was added to the wells (Fig. 3A). By using the methodology of Scatchard (46),assuming the presence of only one binding site on the collagen type VI molecule and a " 20 molecular mass of 1 x 10' kDa for decorin, a dissociation 2 constant of 3.0 X M was calculated for the binding of X

-

15

n

(A

2 3

10

z a 0

2 0

5

I 0

0 0

60 240 180 120

300

TIME (MINUTES) 0

I00

500

400

200 300

DECORIN ADDED

(nM)

0.1

B

0

60 240 180 120

300

TIME (MINUTES)

FIG. 2. The binding of decorin to collagen type VI is timedependent ( A ) and reversible ( B ) . Microtiter Removawells, coated with collagen type VI (closed circles) or 1% BSA (closed squares), were incubated with 5 X lo4cpm '261-labeleddecorin in 0.05 ml of PBS for the indicated lengths of time. At the indicated times, the wells were washed, and the amount of bound radioactivity was determined ( A ) .In B, the '251-decorin wasincubated for 3 h in wells coated with collagen type VI, and wells were washed prior to the addition of 25 pg of unlabeled decorin. The wells were incubated for the indicated length of time and were once again washed, and the amount of bound radioactivity was determined. Results are expressed as the means of determinations taken from triplicate wells. Open circles ( A ) indicate the amount of decorin bound to wells coated with collagen type VI minus that bound to wells coated with BSA. Bars represent the range of "'I-decorin bound.

O

0

.

O

:10

DECORIN BOUND

L

(nW

FIG. 3. The binding of decorin to collagen type VI is a saturable process. Microtiter Removawells, coated with collagen type VI (closed circles) or 1% BSA (closed squares), were incubated with "'I-decorin a t concentrations from 0 to 20 pglwell in 0.05 ml of PBS (specific activity = 35,000 cpm/pg). After a 3-h incubation, the wells werewashed, and theamount of bound radioactivity was determined ( A ) . Specifically bound decorin (open circles) was calculated as the amount bound to collagen type VI coated wells minus that which bound to BSA coated wells. In B, the dataare plottedaccording to themethodology of Scatchard. Results are expressed as themeans of determinations from quadruplicate wells. Bars representthe range of "'I-decorin bound.

Binding of the Proteoglycan Decorin to Collagen Type VI decorin to collagen type VI (Fig. 3B). The Specificity of Decorin Binding to Collagen Type VIRabbit polyclonal antibodies raised against decorin were tested for the ability to inhibit the binding of decorin to collagen type VI. The results (Fig. 4) indicate that thedecorinspecific antibodies effectively inhibited this interaction with an ICwof 3.15 pg of IgGin 0.1 ml.Antibodies from preimmune serum, obtained from the same animal prior to the immunization series, were not effective in inhibiting this interaction. A number of unrelated proteins,including fibrinogen, ovalbumin, fibronectin, BSA, lysozyme, and gelatin, along with collagen type VI and decorin, were tested as potential inhibitors of the decorin-collagen type VI interaction. The proteins were preincubated at a concentration of 0.1 mg/ml, in wells coated with collagen type VI, for 1 h prior to the addition of 5 x lo4 cpm '251-decorin, and the incubation was continued for an additional 3 h. Of the unlabeled proteins tested, only collagen type VI and decorin effectively blocked binding of '251-decorinto theadsorbed collagen (Table I). Preincubation

0

20

40

60

100 60

ug ANTIBODYADDED

FIG. 4. The binding of decorin to collagen type VI is inhibited by decorin-specific polyclonal antibodies. Microtiter Removawells, coated with collagen type VI, wereincubated with 5 X lo4 cpm '2SI-decorinand 0-100 pg of IgG purified from either preimmune (open circles) or decorin-specific antisera (closed circles). The antibody-'*'I-decorin mixtures were incubated together for 1 h at 37 "C before being added to the collagen-coated wells. Subsequent incubation was for 3 h after which the wells were extensively washed, and the amount of bound radioactivity was quantified. Results are expressed as the means of determinations from quadruplicate wells. Bars represent the range of 'Z51-decorinbound.

TABLE I Binding of '"I-decorin to collagen type VI is not inhibited by unrelated proteins Microtiter Removawells, coated with collagen type VI or 1%BSA, were incubated in the absence or presence of the indicated inhibitor (100 pg/ml) for 1 h at 37 "C and subsequently with 5 X lo4 cpm 'Tdecorin for 3 h at 37 "C. The wellswere washed, and the bound radioactivity was determined. Determinations from quadruplicate wells were made for each inhibitor tested. The amount of labeled decorin bound to collagen type VI was set a t 100% (15,182 cpm), whereas that bound to BSA was set at 0% (4023 cpm). The relative Dercentaees of labeled decorin bound are shown (ff3.D.). Competitor

Relative percentage lZ6I-decorinbound ~~~~

None Fibrinogen Ovalbumin Fibronectin BSA Lyzozyme Gelatin Collagen type VI Decorin

~~

100 98.2 f 8.0 87.2 6.9 125.0 f 10.2 87.2 _t 3.4 96.0 f 9.1 108.6 f 5.2 -1.6 f 5.4 2.4 k 1.0

5253

of the collagen type VI-coated wells with fibronectin resulted in a noticeable enhancement of '251-decorinbinding. This is possibly due to a binding of fibronectin to the immobilized collagen and a subsequent binding of '251-decorin to both proteins. Decorin Binds via Its Core Protein to Collagen Type VITo identify the component of the decorin proteoglycan responsible for binding to the immobilized collagen type VI, precoated microtiter wellswere incubated with varying amounts of unlabeled intact decorin, decorin core protein, or isolated dermatan sulfate side chains. After 60 min at 37 "C, labeled decorin was added, and theincubation was continued for an additional 3 h. The results (Fig. 5) suggest that the isolated core protein aswell as the intact proteoglycan inhibit the binding of lZ5I-labeleddecorin to collagen type VI with IC50values of 2.5 and 2.4 pg/O.l ml, respectively. The isolated dermatan sulfate chains hadno significant effect. Several related proteoglycans and the corresponding isolated core proteins were also examined for the ability to inhibit the binding of lZ5I-decorinto collagen type VI. Both biglycan and decorin from cartilage as well as the core proteins from these proteoglycans inhibited this interaction in a concentration-dependentmanner (Fig. 6). In addition, fibromodulin isolated from calf articular cartilage and arecombinant fusion protein containingthe leucine-rich segment of human decorin linked to the amino terminus of TrpE were also effective inhibitors of the binding of labeled decorin to collagen type VI. Taken together, these results show that the core protein and not the glycosaminoglycan chains of decorin contain the primary binding site(s) for collagen type VI and that this binding site is located in the segment containing the leucinerich repeats. Furthermore, core proteins of other proteoglycans similarly containing leucine-rich motifs interfere with the binding of decorin to collagen type VI possibly by binding to the collagen molecule. This suggests that one or several leucine-rich modules may constitute the collagen type VI binding site in these proteins. Collagen Type II Inhibits the Binding of Decorin to Type VI Collagen-The inhibitory activities of various collagens on the binding of '251-decorinto immobilized collagen type VI

10

-

8-

64-

2-

0

, 0

I

10

20

30

40

50

ug INHIBITOR

FIG. 5. The binding of decorin to collagen type VI-coated substrates is mediated by the decorin protein core. Microtiter Removawells, coated with collagen type VI, were incubated with the indicated amounts of either intact decorin (closed circles), purified decorin core protein (open circles),or protein-free glycosaminoglycan chains from decorin (closed squares) for 1 h at 37 "C. Subsequently, 5 X lo4 cpm '"I-decorin was added, and the mixtures were incubated for an additional 3 hat 37 'C. The wells werewashed, and theamount of bound radioactivity was determined. The results are expressed as the means of determinations from triplicate wells. Bars represent the range of '251-decorinbound.

5254

Binding of the Proteoglycan Decorin to Collagen Type VI -m

b.-

20

X

n

Z 3

15

0

m

10

za

0 5

n

30

0

o

20

40

60

a0

1000

20

40

60

a0

loo

ug/ml INHIBITOR FIG. 6. The binding of decorin to collagen type VI coated substrates is inhibited by other related leucine-rich proteoglycans. Microtiter Removawells, coated with collagen type VI, were incubated with 0-100 pg/ml of the indicated unlabeled proteoglycans or core proteins for 1 h at 37 "C. This was followed bythe addition of 5 X lo' cpm '2SI-labeleddecorin which was incubated for an additional 3 h at 37 "C. The wells were washedand the amount of bound radioactivity was determined. Results are expressed as the means of determinations from quadruplicate wells. Bars represent the range of '"I-decorin bound. A shows inhibition by intact bovine biglycan (closed circles), intact bovine decorin (open circles), and recombinant fusion protein containing the leucine-rich segment of human decorin (closed squares), and B shows inhibition by isolated core protein from bovine decorin (open squares), isolated core protein from bovine biglycan(closed triangles), and bovine fibromodulin (open triangles).

L

CI

i

CII Clll CIV CV

I CVI BSA

INHIBITORS

1

I

0

10

I

I

20

30

50

40

ug COLLAGENTYPE

II

FIG. 7. The binding of decorin to collagen type VI-coated substrates is inhibited by both soluble collagen type VI and collagen type 11. Microtiter Removawells, coated with collagen type VI, were incubated with 20 pg of collagen types I-VI or BSA for 1 h at 37 "C. Subsequently, '"I-decorin was added, and the incubation was continued for 3 h at 37 "C, after which the wells were washed, and the bound radioactivity was determined. Results are expressed as themeans of determinations from triplicate wells. Bars represent the range of '261-decorinbound.

FIG. 8. The inhibitory activity of collagen type I1 is dosedependent. Microtiter Removawells, coated with collagen type VI, were incubated with varying amounts of collagen type 11, ranging from 0 to 50 pg, for 1 h at 37 'C. Subsequently, 5 X lo4 cpm "'1labeled decorin was added, and the incubation was continued for an additional 3 h at 37 "C. The wells were washed, and the amount of bound radioactivity was determined. Results are expressed as the means of determinations from quadruplicate wells. Bars represent the range of '"I-decorin bound.

were examined by preincubating collagen type VI-coated wells with 20 pg of collagen types I-VI or BSA,each diluted in 50 pl of PBS,for 60 min prior to the addition of 5 X lo4 cpm 1251-dec~rin. After 3 h at 37 "C,the wells wererinsed, and the amount of bound radioactive ligand was determined. As expected, collagen type VI was found to effectively inhibit the binding of '251-decorinto immobilized collagen type VI (Fig. 7). In addition, collagen type I1 inhibited 1251-decorinbinding to collagen type VI. The other soluble collagen types analyzed and BSA were without effect. The inhibitory activity of collagen type I1 was further examined and shown to be dosedependent (Fig. 8) with an ICs0 of 7.7 pg in 0.1 ml. The mechanism whereby soluble collagen type I1 inhibits the binding of labeled decorin to immobilized collagen type

VI was further analyzed. Since '"I-decorin did not seem to bind effectively to immobilized collagen type I1 (see Fig. l), we examined the possibility that theobserved inhibition was a consequence of an interaction between collagen type I1 and collagen type VI molecules. To test thispossibility, microtiter wells coated with collagen type VI were incubated for 1h with 20 pg of soluble collagen type I1 in 50 pl of PBS. Some wells were washed to remove unbound material prior to theaddition of 1251-decorin,whereas in others, the labeled proteoglycan was added directly to the collagen mixture. The inhibitory activity of collagen type I1 was similar in both sets of experiments, indicating that collagen type I1 was in fact binding to collagen type VI (data not shown). To examine the specificity of collagen type I1 binding to

Binding of the Proteoglycan Decorin to Collagen Type VI collagen type VI microtiter wells coated with collagen types I, 111, VI, or 1%BSA in PBS, were incubated with varying amounts of biotin-labeled collagen type I1 for 2 h. The wells were rinsed, and theamount of biotin-labeled material bound was detected using an avidin-alkaline phosphatase conjugate. The results indicate that biotin-labeled collagen type I1 bound specifically to wells coated with collagen type VI (Fig. 9). Binding to adsorbed collagen types I or I11 was at levels similar to those seen for binding to wells coated with BSA. Type II Collagen Binds to Type VI CollagenCloseto the Decorin Binding Site-The results presentedabove showthat collagen type VI contains binding sites for both collagen type I1 and the proteoglycan decorin. Furthermore, collagen type I1 inhibits the binding of 1251-decorinto adsorbed collagen type VI. To test the reverse situation, soluble decorin along with soluble collagen type I1 were tested as inhibitors of the binding of biotin-labeled collagen type I1 to microtiter wells coated with collagen type VI. The results indicated that both collagen type 11and decorin effectively inhibited the binding of biotin-labeled collagen type I1 to collagen type VI in a concentration-dependentmanner (Fig. 10) with calculated IC60 values of 0.5 and 0.9 pg/O.l ml for decorin and collagen type 11, respectively. Taken together, these data suggest that the binding sites for decorin and collagen type I1 could be located close together on the type VI molecule.

0.4

5255

1

0.3 -

E : : 0

0.2-

W

0

z a ma

0.1

-

0.0

-

8m 4:

0

20

40

60

ug INHIBITOR

FIG. 10. The binding of collagen type I1 to collagen type VI is inhibited by the presence of decorin. Microtiter wells, coated with collagen type VI, were incubated with the indicated amounts of either collagen type I1 (closed circles) or decorin (open circles) for 1 h at room temperature. After a rinsing, 1 pg of biotin-labeled collagen type I1 was added and an enzyme-linked immunosorbent-type assay was performedas described in the legend to Fig. 9.

mediated by unidentified linking proteins. Alternatively, the conformation and structural organization of the major fibrilIn this report, we have analyzed the postulated interaction lar collagen moleculesrequired for decorin binding is different between decorin and various collagen types using solid phase from the monomeric forms used in this study. binding assays. We report on a time- and concentrationThe binding of decorin to collagen type VI is saturable with dependent high affinity binding of decorin to collagen type a calculated dissociation constant of 3 X M. Binding of VI. Under the conditions of the experiments, the other colla- decorin to immobilized collagen type VI seems to involve a gen types tested bound significantly lower amounts of 1251- site in the core protein, since isolated core protein, but not decorin than did collagen type VI. Theseresults do not protein-free glycosaminoglycan chains, inhibited the binding support earlier ultrastructural studies indicating a direct as- of decorin to immobilized collagen type VI. A recombinant sociation of decorin with fibrillar collagen types (6-11). How- decorin fragment containing only the leucine-rich repeat modever, this apparent conflict could be explained by an interac- ules was also shown to inhibit this interaction. In addition, tion of proteoglycans with fibrillar collagens (types 1-111) proteoglycans with related leucine-rich core proteins acted as effective inhibitors of the binding, suggesting that a common structural motif is recognized by collagen type VI. In the proteoglycan core proteins containing these leucine-rich repeats, e.g. biglycan, decorin, and fibromodulin, this motif is 2.0 believed to participate in the binding to fibrillar collagens in at least the lattertwo proteoglycans (5, 16). In other proteins 5 which also contain leucine-rich repeats like ribonuclease/ 0 angiogenin inhibitor (47,48),platelet glycoprotein l b (49,50), 3 w yeast adenylate cyclase (51-53), and receptors for luteinizing 0 z hormone/chorionic gonadotropin (54),this motif has, in some 2 1.0 cases, been thought to mediate protein-protein interactions. a The biological implications of the interactions between 5: m collagen type VI and decorin are unclear. Bonaldo et al. (35) a has suggested that the cell attachment and thecollagen type I binding properties of collagen type VI are indicative of a bridgmg role for collagen type VI whereby the protein anchors 0.0 the cell to the collagen fibrils in the extracellular matrix. In 0.0 0.4 0.8 1.2 1.6 2.0 fibroblast cultures decorin is primarily located on the cell ug COLLAGEN TYPE II ADDED surface; and little of the proteoglycan is found in the extraFIG. 9. The binding of collagen type I1 to collagen type VI cellular matrix. The cell-associated proteoglycan could then is specific. Microtiter wells were coated with collagen type VI (closed participate in cellular interactions with the collagen type VI circles),collagen type I (open circles),collagen type 111 (close squares), microfibrils. In fact, an interaction between a membraneor 1% BSA (closed triangles). After the wells were blocked with 1% bound chondroitin sulfate proteoglycan found on several cell BSA, varying amounts of biotinylated collagen type 11, ranging from 0 to 2 pg, were added and incubated for 2 h at room temperature. types, known as NG2, and collagen type VI has been described After a washing, alkaline phosphatase-conjugated avidin was added, (55). It is tempting to speculate that the binding of decorin to and the incubation was continued for another 2 h at room temperaDISCUSSION

u)

ture. After a final washing, the wells were developed with a chromogenic substrate, and the absorbance was readat 405 nm.

~~

_____

~~

~

~~

* D. J. Bidanset and M. Hook, unpublished observations.

5256

Binding of the Proteoglycan Decorin to Collagen Type VI

Rauch.. U.., and Kresse.H. 119891 ~. collagen typeVI may also play a role inregdating the struc- 22. Hausser. H..HOD"W.. ~ w J. 26%; ~ 137-142 ~ m tural organi~tionof the extracellular matrix. If collagen type 23. Chung. E.. Rhodes. R. K.. and Miller. E. J. f1976) . , Biochem. VI plays a role in controlling the structural organization of Bw&s.'Res. Comrnun. 71,1167-117k fiber forming collagens has as been suggested(35),the binding 24. Trueb, B., and Winterhalter, K. H. (1986)EMBO J. 6, 2815of decorin to collagen type VI may regulate this activity. In 2819 our studies, binding of collagen type I1 to collagen type VI 25. Colombatti, A., Bonaldo, P., Ainger, K., Bressan, G. M., and Volpin, D. (1987)J. Biol. Chem. 262, 14454-14460 was shown to inhibit the bindingof decorin to collagen type 26. Doliana, R., Bonaldo, P., and Colombatti, A. (1990)J. Cell Bwl. VI, whereas collagen typeI was without effect. Bonaldoet al. 11 1,2197-2205 (35) demonstrated a binding of collagen type I to collagen 27. Engel, J., Furthmayr, H., Odermatt, E., von der Mark, H., Aumailley, M., Fleischmajer, R., and Timpl, R. (1985)Ann. N. Y. type VI whichcouldbelocalized to a globulardomain of Acad. Sci. 460,25-37 cuS(V1). Theseobservationssuggestthattherearebinding E., Risteli, J., Van Delden, V., and Timpl, R. (1983) sites on collagen type VI for both collagen type I and collagen 28. Odermatt, Biochem J. 211,295-302 type 11. Preliminaryexperimentsanalyzingthebinding of 29. Chu, M.-L., Mann, K., Deutzmann, R., Pribula-Conway, D., Hsudecorin to pepsin-solubilized collagen type VI by rotary shadChen, C.-C., Bernard, M.P., and Timpl, R (1987)Eur. J. Bwchem. 168,309-317 owingindicatethat the proteoglycanbinds to aterminal domainof the collagen molecule. Hence, it seems possible 30. Chu, M.-L., Conway, D., Pan, T., Baldwin, C., Mann, K., Deutzmann, R., and Timpl, R. (1988)J. Biol. Ckem. 263, 18601that the binding of proteoglycans and collagen types I and I1 18606 involve the globular domains the of collagen typeVI molecule 31. Bonaldo, P., and Colombatti, A. (1989)J. Bwl. Chem. 264, and that interactions between collagen type VI and the fibril20235-20239 32. Koller, E., Winterhalter, K. H., and Trueb, B. (1989)EMBO J. larcollagensareregulatedbyproteoglycans.Whateffects 8,1073-1077 such hypothetical regulatory mechanisms may have on the 33. Chu, M.-L., Pan, T.-C., Conway, D., Kuo, H.-J., Glanville, R. W., structural organization of collagen fibrilsis presently unclear. Timpl, R., Mann, K., and Deutzmann, R. (1989)EMBO J. 7,

Acknowkdgments-We thank Drs. David Mann, Anna Plaas, and Edward J. Miller for providing various proteins. REFERENCES 1. Scott, J. E. (1986)C h F o u ~S.y w . 124,104-124 2. Scott, J. E. (1988)Biochem J. 262,313-323 3. Scott, J. E.,and Haigh, M. (1985a)Biosci Rep. 5,765-774 4. Hedbom, E., and Heinegard, D. (1989)J. Biol. Chem. 264,68986905 5. Oldberg, A,, Antonsson, P., Lindblom, K., and Heinegard, D. (1989)EMBO J. 8,2601-2604 6. Obrink, B. (1973)Eur. J. Bwchem. 33,387-400 7. Scott, J. E., and Orford, C. R. (1981)Biochem. J. 197,213-216 8. Scott, J. E., and Haigh, M. (1985b)Biosci. Rep. 5, 71-81 9. Vogel, K. G., and Trotter, J. A. (1987)Collagen Rel. Res. 7 , 105114 10. Pringle, G.A., and Dodd, C.M. (1990)J. Histochem. Cytochern. 38.1404-1411 11. Young, R. D. (1985)J. Celt Sci. 74,95-104 12. Vogel, K. G., Paulsson, M., and Heinegard, D. (1984)Biochem. J. 223,587-597 13. Uldbjerg, N., and Danielsen, C. C. (1988)Biochem. J. 261,643648 14. Rosenberg, L. C., Choi, H. U., Tang, L.-H., Johnson, T. L., Pal, S., Webber, C., Reiner, A., and Poole, A. R. (1985)J. Bwl. Chem. 260,6304-6313 15. Choi, H.U., Johnson, T. L., Pal, S., Tang, L.-H., Rosenberg, L., and Neame, P. J. (1989)J. Bwl. Chem. 264,2876-2884 16. Fisher, L. W., Termine, J. D., and Young, M. F. (1989)J . BwL Chem. 264,4571-4576 17. Rosenberg, L.C., Choi, H. U., Poole, A. R., Lewandowska, K., and Culp, L. A. (1986)Ciba Found. Symp. 124,47-68 18. Poole, A. R., Webber, C., Pidoux, I., Choi, H. U., andRosenberg, L. C. (1986)J. Histochem. Cytmkm. 34,619-625 19. Schmidt, G., Robenek, H., Harrach, B., Glossl, J., Nolte, V., Hormann, H., Richter, H., and Kresse, H. (1987)J. CeM Bwl. 104,1683-1691 20. Lewandowska, K., Choi, H. U.,Rosenberg, L. C., Zardi, L., and Culp, L. A. (1987)J. Cell Biol. 106,1443-1454 21. Yamaguchi, Y., Mann, D., and Ruoslahti, E. (1990)Nature 346, 281-284

1939-1946 34. Chu, M.-L., Zhang, R.-Z., Pan, T.-C., Stokes, D., Conway, D., Kuo, H.-J., Glanville, R., Mayer, U., Mann, K., Deutzmann, R., and Timpl, R. (1990)EMBO J. 9,385-393 35. Bonaldo, P., Russo, V., Bucciotti, F., Doliana, R., and Colombatti, A. (1990)Biochemistry 29, 1245-1254 36. ~ r t h m a H., ~ , Wiedemann, H., Timpl, R., Odermatt, E., and Engel, J. (1983)Biochem. J. 211,303-311 37. Bruns, R. R.,Press, W., Engvall, E., Timpl, R., and Gross, J. (1986)J. Cell Bbl. 103,393-404 38. Colombatti, A., Ainger, K., and Colizzi, F. (1989)Matrix 9, 177185 39. Aumailley, M.,Mann, K., von der Mark, H., and Timpl, R. (1989) Exp. Cell Res. 181,463-474 40. Pareti, F.I., Niiya, K., McPherson, J. M., and Ruggeri, Z. M. (1987)J. Bid. Chem. 262,13835-13841 41. Miller, E. J., and Rhodes, R. K. (1982)Methods Enzymol. 82, 33-64 42. Plaas. A. H. K.. Ison. A. L.. and Ackland., J. (1989) . . J. Biol. Chem. 26i,14447-i4454 43. Oike. Y.. Kimata. K.. Shinomura, T.. Nakazawa, K.. and Suzuki, S. (1980)Biocf;em.'J. 191,193-207 44. Oike, Y.,Kimata, K., Shinomura, T., Suzuki, S., Takahashi, N., and Tanabe, K. (1982)J. Bioi. Chem. 267,9751-9758 45. Orr, G. A. (1981)J. BWZ. Chem. 256, 761-766 46. Scatchard, G. (1949)Ann. N. Y. Acad. Sei. 61,660-672 47. Hofsteenge, J., Kieffer, B., Matthies, R., Hemmings, B.A., and Stone, S. R. (1988)Biochemistry 27,8537-8544 48. Lee, F. S., and Vallee, B. L. (1990)Proc. Natl. Acad. Sci. U.S. A. 87,1879-1883 49. Titani, K., Takio, K., Handa, M., Ruggeri, Z.M. (1987)Proc. Natl. Acad. Sci. U. S. A. 84,5610-5614 50. Lopez, J. A., Chung, D. W., Fujikawa, K., Hagen, F. S., Davie, E. W., and Roth, G. J. (1988)Proc. NatL Acad. Sei. U. S. A. 86, 2135-2139 51. Katoaka, T., Broek, D., and Wigler, M. (1985)Cell 43,493-505 52. Colicelli, J., Field, J., Ballester, R., Chester, N., Young, D., and Wigler, M. (1990)Mol. Cell. Biol. 10,2539-2543 53. Field, J., Xu, H.-P., Michaeli, T., Ballester, R., Sass, P., Wigler, M., and Colicelli, J. (1990)Science 247,464-467 54. Braun. T.. Schofield. P. R.. and SDreneel. EMBO J. - " R. (1991) . IO,i&-isgo ' 55. StallcuD. W. B.. Dahlin.. K... and Healv. P. (1990)J. CeU Biol. 111,*3177-3188 ,~

~

~~

~