Enhancement of Cell Adhesion and Spreading by a Cartilage-specific Noncollagenous Protein, Cartilage Matrix Protein (CMP/Matrilin-1), via Integrin α1β1 *

Cartilage matrix protein (CMP; also known as matrilin-1), one of the major noncollagenous proteins in most cartilages, binds to aggrecan and type II collagen. We examined the effect of CMP on the adhesion of chondrocytes and fibroblasts using CMP-coated dishes. The CMP coating at 10–20 μg/ml enhanced the adhesion and spreading of rabbit growth plate, resting and articular chondrocytes, and fibroblasts and human epiphyseal chondrocytes and MRC5 fibroblasts. The effect of CMP on the spreading of chondrocytes was synergistically increased by native, but not heated, type II collagen (gelatin). The monoclonal antibody to integrin α1 or β1 abolished CMP-induced cell adhesion and spreading, whereas the antibody to integrin α2, α3, α5, β2, α5β1, or αVβ5had little effect on cell adhesion or spreading. The antibody to integrin α1, but not to other subunits, coprecipitated125I-CMP that was added to MRC5 cell lysates, indicating the association of CMP with the integrin α1 subunit. Unlabeled CMP competed for the binding to integrin α1with 125I-CMP. These findings suggest that CMP is a potent adhesion factor for chondrocytes, particularly in the presence of type II collagen, and that integrin α1β1 is involved in CMP-mediated cell adhesion and spreading. Since CMP is expressed almost exclusively in cartilage, this adhesion factor, unlike fibronectin or laminin, may play a special role in the development and remodeling of cartilage.

taining pure CMP (14 g; 0.08 ml/mouse) was mixed with Ribi adjuvant (0.14 ml/mouse; Ribi ImmunoChemical Research, Hamilton, MT) and injected subcutaneously into two female BALB/cAnnNCrj mice once every 2 weeks. The serum was obtained 5 weeks after the first injection.
Immunostaining of CMP-Rib cartilage obtained from 10-day-old pigs was fixed with periodate/lysine/paraformaldehyde for 24 h at 4°C, washed with water for 12 h, dehydrated, and then embedded in paraffin wax as described previously (24). Sections (6 m) were incubated with 250 units/ml hyaluronidase (Sigma) at 37°C for 30 min and then with 4% skim milk at room temperature for 30 min. After being washed with PBS, the sections were incubated with nonimmune serum or anti-CMP serum (300-fold dilution) at room temperature for 30 min, washed with PBS containing 0.05% Tween 20, and then incubated with fluoresceinconjugated goat IgG fraction to mouse IgG F(abЈ) 2 at room temperature for 30 min in the absence of light. After washing with PBS containing 0.05% Tween 20, the histology was observed under a confocal laser microscope (LSM410, Carl Zeiss, Inc., Oberkochen, Germany).
Cells-Chondrocytes were isolated from the femur articular cartilage at knee joints and the rib growth plate of 4-week-old Japanese White rabbits or from the epiphyseal cartilage of human embryos as described previously (25). Human embryonic epiphyseal cartilage was obtained from the Department of Pathology, Norman Bethune University of Medical Sciences (Changchun, China) (25). Rabbit fibroblasts were isolated from soft connective tissue of the rabbits described above (26). A human embryonic lung fibroblast line (MRC5) was obtained from RIKEN (Tsukuba, Japan). All cells were seeded at a density of 5 ϫ 10 5 cells/10-cm plastic tissue culture dish and maintained in ␣-modified Eagle's medium containing 10% fetal bovine serum (Mitsubishikasei, Tokyo, Japan), 25 g/ml ascorbic acid, 32 units/ml penicillin, and 60 g/ml kanamycin. Cultures were incubated in an atmosphere of 5% CO 2 in a humidified incubator.
Coating Dishes with CMP-CMP, bovine type II collagen (acid-soluble, pepsin-resistant; Koken, Osaka, Japan), or both at various concentrations were incubated in 50 l of PBS containing 10 mM NaHCO 3 in 6-mm plastic microwells (Falcon-3072, Becton Dickinson, or Sumilon MS-8096, Sumitomo Bakelite, Osaka) at 4°C for 18 h. The substrata were washed three times with PBS and then incubated with 50 l of PBS containing 10 mg/ml bovine serum albumin (BSA; Sigma) at room temperature for 2 h to block nonspecific cell attachment.
Cell Adhesion-When the cultures became 80% confluent, the chondrocytes or fibroblasts were preincubated with 10 g/ml cycloheximide for 2 h and then harvested with PBS containing 0.1% trypsin and 0.1% EDTA. The cells were seeded at 5 ϫ 10 3 cells/6-mm plastic tissue culture microwell (Falcon-3072) coated with CMP and incubated at 37°C for 15 min to 2 h in 0.1 ml of ␣-modified Eagle's medium containing 1 mg/ml BSA, 25 g/ml ascorbic acid, 32 units/ml penicillin, 60 g/ml kanamycin, and 10 g/ml cycloheximide (medium A) or in 0.1 ml of 10 mM Hepes (pH 7.4) containing 0.9% NaCl, 1 mg/ml BSA, and 10 g/ml cycloheximide. All cells dropped to the bottom of the dishes within 15 min after seeding if they did not attach to the culture surface. Spread cells were distinguishable by their cellular projections. Round and spread cells were separately counted under a phase-contrast microscope. At least 60 -80 cells were evaluated, and the percentage of spread cells to total cells was calculated.
In some experiments, cells were seeded at a density of 5 ϫ 10 4 cells/6-mm plastic non-tissue culture microwell (Sumilon MS-8096) coated with CMP and then incubated at 37°C for 1-2 h in 0.1 ml of medium A. Nonadherent cells were removed by gentle rinsing, and the number of adherent cells in each microwell was then quantified using an aqueous soluble tetrazolium/formazan assay (Promega, Madison, WI) (27).
Labeling of CMP with 125 I-CMP was labeled with 125 I using chloramine T (ICN Biomedicals, Costa Mesa, CA). CMP (0.1 mg) was dissolved in 0.25 ml of 0.1 M sodium phosphate (pH 7.5), mixed with 10 l of Na 125 I solution (100 mCi/ml in 0.1 M sodium phosphate (pH 7.5)), and then incubated with 2 l of chloramine-T solution (1 mg/ml in 0.1 M sodium phosphate buffer (pH 7.5)) for 1 min. The reaction was terminated by the addition of 2.5 l of Na 2 S 2 O 5 solution (1 mg/ml in 0.1 M sodium phosphate (pH 7.5)). The reaction mixture was applied to a Sephadex G-10 column (1 ml) that was equilibrated with 0.1 M sodium phosphate (pH 7.5). SDS-PAGE analysis showed that this preparation of 125 I-CMP migrated at a position corresponding to 60 kDa even under nonreducing conditions.
Preparation of Lysates from Isolated Single Cells-When human fibroblasts (MRC5) were grown to 80% confluence, the cells were incubated for 5 h with PBS containing 0.5 mM CaCl 2 , 2.5 mM N-ethylmaleimide, and 0.8 mg/ml pure bacterial collagenase (type VII, Sigma) at 37°C. The cells were washed twice with PBS and then incubated with PBS containing 0.1% EDTA and 0.1% trypsin for 3 min. The dispersed cells were transferred to small plastic tubes, washed three times with PBS, and then incubated for 30 min with RIPA buffer (1% Nonidet P-40 and 10 mM Tris-HCl (pH 7.4) containing 0.1% sodium deoxycholate, 0.1% SDS, 0.15 M NaCl, 1 mM MgCl 2 , and 1 mM phenylmethanesulfonyl fluoride) at 4°C. Insoluble materials were removed by centrifugation at 13,000 ϫ g for 20 min at 4°C. The supernatant was precleared by protein G-Sepharose beads (Pharmacia, Uppsala, Sweden) to remove proteins that nonspecifically bound to the beads. Lysates were analyzed for protein using a protein assay kit (Bio-Rad).
Precipitation of 125 I-CMP in the Presence of Cell Lysates with Antiintegrin Antibodies-The MRC5 cell lysate (100 l, 450 g/ml protein) was incubated at 4°C with 125 I-CMP (2000 cpm, 150 ng of protein in 1 l of 0.1 M sodium phosphate (pH 7.5)) for 2 h and then mixed with rabbit antiserum to integrin (␣ 1 , ␣ 2 , ␣ V , ␤ 1 , or ␤ 5 ), rabbit antiserum to human type I collagen, or rabbit nonimmune serum (1 l) and a 1:1 suspension of protein G-Sepharose in RIPA buffer (40 l). The suspension was incubated for 2 h at 4°C. The protein G-Sepharose beads with bound immune complexes were washed five times with RIPA buffer and then boiled in Laemmli buffer (40 l). Laemmli buffer-resolved proteins were fractionated by SDS-PAGE. The gel with the fractionated proteins was dried and exposed to x-ray film (Eastman Kodak Co.).
Surface Labeling with Biotin and Immunoprecipitation-When MRC5 fibroblasts were grown to 80% confluence, the cells were harvested with PBS containing 0.1% EDTA and 0.1% trypsin. Cells in suspension were incubated with 10 ml of PBS containing 10 g/ml NHS-LC-biotin (Pierce) on ice for 90 min and then washed twice with PBS. The cells were dissolved with RIPA buffer and triturated intermittently for 30 min. Insoluble materials were removed by centrifugation at 13,000 ϫ g for 20 min at 4°C. The supernatant was precleared by protein G-Sepharose beads. Lysates were analyzed for protein using the Bio-Rad protein assay kit mentioned above.
The biotin-labeled cell lysates (100 l) were incubated with 1 l of rabbit anti-integrin ␣ 1 antiserum or rabbit nonimmune serum at 4°C for 2 h. Immune complexes were then incubated with a 1:1 suspension of protein G-Sepharose in RIPA buffer (40 l) at 4°C for 2 h. The Sepharose beads with bound immune complexes were washed five times with RIPA buffer and then boiled in Laemmli buffer (40 l). Laemmli buffer-resolved proteins were fractionated by SDS-PAGE and transferred to a nitrocellulose membrane (28). The membrane was blocked with BSA and incubated for 1 h in streptavidin-conjugated horseradish peroxidase (Amersham International, Buckinghamshire, United Kingdom). Biotinylated proteins were visualized using enhanced chemiluminescence (Amersham International) and then exposed to x-ray film.

Localization of CMP-Conventional immunohistochemistry
showed that CMP was present in the territorial and interterritorial matrixes in fetal bovine rib cartilage (10). The amount of CMP in the interterritorial matrix was less than that in the territorial matrix (10). In human arthritic articular cartilage, CMP was present near or within chondrocytes, but was barely detectable in the interterritorial matrix (8). In this study, we examined CMP in the resting cartilage ( Fig. 2A) and the matrix-forming (prehypertrophic) zone of the growth plate ( Fig.  2B) of newborn pigs, using the anti-CMP antiserum and confocal microscopy. In these cartilages, CMP was concentrated near the cell surface (Fig. 2, A and B). No stain was observed with the control serum (Fig. 2C).
Stimulation of Cell Adhesion and Spreading by CMP-CMP is present in the pericellular matrix ( Fig. 2). In addition, CMP has type A-like domains that may be involved in cell adhesion (13). We therefore hypothesized that CMP may be an adhesion factor. To test this hypothesis, we seeded rabbit articular chondrocytes on plastic tissue culture dishes coated with various concentrations of CMP and incubated them at 37°C for 2 h in the presence of cycloheximide, an inhibitor of protein synthesis. Round and spread cells were separately counted using a phasecontrast microscope. The chondrocytes spread rapidly on the CMP-coated dishes, whereas only a few cells spread on the dishes not coated with CMP (Fig. 3A, inserts a and b). The percentage of spread cells to total cells on the CMP (20 g/ml)coated dishes was 42% compared with 3% on the control dishes (Fig. 3A). This stimulation of spreading was induced when dishes were preincubated with CMP at 2 g/ml and became almost maximal at 20 g/ml (Fig. 3A). In another experiment, adherent cells were quantified using an aqueous soluble tetrazolium/formazan assay. CMP enhanced the adhesion of chondrocytes in a similar dose-dependent fashion (Fig. 3B). Fig. 4 shows that CMP was effective in stimulating the spreading of rabbit growth plate chondrocytes (rGC), rabbit articular chondrocytes (rAC), human embryonic chondrocytes (hEC), rabbit fibroblasts (rFB), and human MRC5 fibroblasts (hFB).
Synergism between CMP and Type II Collagen-Since CMP binds to type II collagen (2), we examined whether type II collagen modulates the effect of CMP on the spreading of chondrocytes. A low concentration of CMP (0.5 g/ml) enhanced the spreading of chondrocytes in the presence, but not absence, of type II collagen (3 g/ml) (Fig. 5A). This synergism between CMP and collagen was observed 15 min after cell seeding and was sustained for at least 60 min (Fig. 5A). The concentration of CMP required for cell spreading was 20 -200-fold higher in the absence of type II collagen than in its presence (Fig. 5B). At high concentrations (3-10 g/ml), type II collagen alone stimulated cell spreading (Fig. 5C). CMP at 0.5 g/ml increased this effect of type II collagen by 3-4-fold. The synergism between CMP and collagen was not observed with denatured type II collagen (gelatin) that was boiled for 5 min at pH 3.0, although gelatin alone had a greater effect on cell spreading than type II collagen alone (Fig. 5D).
Effects of EDTA and Divalent Cations on CMP-mediated Cell Spreading-Although CMP plus collagen produced a synergistic stimulation of cell spreading, our subsequent studies of CMP-recognizing integrins were carried out without collagen because collagen alone modulates integrin activity. The pres-ence of collagen makes the interpretation of the data difficult.
Since integrins require divalent cations to bind ligands, we examined whether EDTA inhibits cell adhesion to CMP. The addition of EDTA at 2 mM (but not 1 mM) to ␣-modified Eagle's medium suppressed the spreading of chondrocytes on CMPcoated dishes (Fig. 6A). The concentrations of Mg 2ϩ and Ca 2ϩ in ␣-modified Eagle's medium are 0.8 and 1.8 mM, respectively. When chondrocytes were suspended in 10 mM Hepes (pH 7.4) containing 0.9% NaCl, few cells were attached on CMP-coated dishes. However, the addition of Mg 2ϩ to the buffer markedly enhanced the cell spreading on CMP-coated dishes (Fig. 6B). This was induced at a Mg 2ϩ concentration of 1-2 mM and increased dose-dependently at least until 5 mM. Mn 2ϩ had a greater effect on cell spreading on CMP-coated dishes than Mg 2ϩ at 1-2 mM (Fig. 6C). Ca 2ϩ had less effect on cell spreading than Mg 2ϩ and Mn 2ϩ (Fig. 6D).
Inhibition of Cell Adhesion and Spreading on CMP-coated Dishes by Anti-integrin Antibodies-Since antibodies to rabbit integrins are rather difficult to obtain, we used antibodies to human integrins. Unless otherwise specified, a human fibroblast cell line (MRC5) was used because it was difficult to obtain human chondrocytes in primary cultures.
The spreading of MRC5 fibroblasts on CMP-coated dishes was suppressed by the mAb against integrin ␣ 1 or ␤ 1 (Fig. 7). The mAb against ␣ 2 , ␣ 3 , ␣ 5 , ␤ 2 , ␣ 5 ␤ 1 , or ␣ V ␤ 5 , as well as control IgG, had little effect on cell spreading on CMP-coated dishes. The inhibition of the CMP-induced cell adhesion by the mAb to integrin ␣ 1 or ␤ 1 , but not other mAbs, was observed with human chondrocytes using an aqueous soluble tetrazolium/ formazan assay (Fig. 8). These findings suggest that the integrin ␣ 1 and ␤ 1 subunits were involved in the cell adhesion and spreading on CMP-coated dishes.
Binding of CMP to Integrins-We next examined whether 125 I-CMP added to the cell lysates is coprecipitated with integrin subunits using antibodies to the ␣ 1 , ␣ 2 , ␣ V , ␤ 1 , and ␤ 5 subunits. After iodination, CMP did not form trimers even under nonreducing conditions during electrophoresis in the presence of SDS, as described under "Experimental Proce-dures." However, 125 I-CMP enhanced cell adhesion at 10 g/ ml. 2 The lysates of MRC5 cells that were dispersed with bac- terial collagenase and trypsin were incubated with 125 I-CMP in the presence of rabbit antiserum to a human integrin subunit (␣ 1 , ␣ 2 , ␣ V , ␤ 1 , or ␤ 5 ), rabbit antiserum to human type I collagen, or nonimmune rabbit serum, and then the material precipitated with protein G-Sepharose was analyzed by SDS-PAGE. Of these sera, only the antiserum to integrin ␣ 1 precipitated 125 I-CMP in the lysates (Fig. 9A, panel a). No 125 I-CMP directly bound to the antibodies in the absence of the lysates (Fig. 9A, panel b). The antiserum to integrin ␣ 1 precipitated the 200-kDa integrin ␣ 1 subunit in MRC5 lysates (Fig.  9B). The molecular mass of the integrin ␣ 1 subunit in human chondrocytes is ϳ200 kDa under nonreducing conditions (15). In other experiments, the mAb to integrin ␣ 1 also coprecipitated 125 I-CMP in the cell lysates, although the 125 I-CMP level precipitated with this mAb was lower than the precipitated level with the polyclonal antibodies to integrin ␣ 1 (data not shown).
In the presence of increasing concentrations of 125 I-CMP and the antiserum to integrin ␣ 1 , the precipitation of 125 I-CMP was detectable at 15 ng/ml 125 I-CMP and increased with the increase in 125 I-CMP, at least until 1.5 g/ml (Fig. 10A). In contrast, the control serum did not precipitate 125 I-CMP even at the highest concentration of 125 I-CMP (1.5 g/ml) (Fig. 10A). Next, the cell lysates were incubated in the presence of 1.5 g/ml 125 I-CMP, the antiserum to the integrin ␣ 1 subunit, and increasing concentrations of unlabeled native CMP. The native CMP decreased the precipitated level of 125 I-CMP dose-dependently (Fig. 10B). DISCUSSION The biological role of CMP is not known, although it is one of the major noncollagenous proteins in growth plates, tracheal cartilage, and other cartilages (4,8,29). It constitutes up to 5% of the wet weight of tracheal cartilage (4). CMP may stabilize the cartilage matrix and alter the tensile strength and elasticity of the matrix by binding to collagen and aggrecan. Although articular chondrocytes do not usually synthesize CMP, CMP synthesis is markedly enhanced in arthritic joints (8), suggesting that CMP may be involved in the destruction and/or remod-eling of cartilage.
Our immunohistochemical analyses showed that unlike aggrecan and type II collagen, CMP is concentrated near the chondrocyte surface in vivo. This observation, as well as the presence of the type A-like domain on CMP, suggested that CMP could be involved in the matrix-cell interaction. We tested this hypothesis and showed for the first time that CMP is an adhesion protein for chondrocytes and fibroblasts.
The physiological significance of the CMP-induced spreading of fibroblasts is not known. However, CMP may function as an adhesion factor for fibroblast cells in eye tissues, notochord, and tendon that contain CMP at low levels (5)(6)(7).
Biochemical studies have shown that CMP binds to type II collagen in vitro (2), and immunostained transmission electron microscopy has shown that CMP binds to the exterior of the collagen fibril in vivo (2). CMP distributes along type II colla-with 4 Ϯ 2 or 18 Ϯ 3% on dishes not coated or coated with native CMP (10 g/ml), respectively. The values are the means Ϯ S.D. of triplicate determinations. gen fibers with a periodicity of 59 nm (2). However, the biological significance of the CMP-type II collagen complex is not known. We showed here that native, but not denatured, type II collagen increases the effect of CMP on the spreading of chondrocytes. This finding suggests that CMP attached on type II collagen fibers more efficiently concentrates CMP receptors/ integrins at adhesion plaques than CMP alone attached uniformly on a plastic or gelatin surface. We also observed that fibronectin and laminin did not produce a synergistic stimulation of cell spreading in the presence of type II collagen (data not shown), although fibronectin and laminin bind to collagen; this distinguishes the role of CMP in cartilage from those of fibronectin and laminin.
CMP forms two types of filamentous networks in the pericellular matrix: one that contains type II collagen and another that does not contain type II collagen (30). The collagen-independent CMP filaments may also serve as a scaffold for cell adhesion in vivo because CMP at high concentrations (10 -20 g/ml) enhanced cell spreading in the absence of type II collagen in vitro.
In tracheal cartilage, CMP binds to aggrecan noncovalently and covalently, and the covalent cross-linking to the aggrecan core protein increases with age (3). The CMP-aggrecan complex is unlikely to enhance the adhesion or spreading of chondrocytes because aggrecan suppresses cell adhesion in vitro (31).
In this study, we examined whether integrins are involved in the CMP-induced cell adhesion. In our assays, the CMP-induced cell spreading required Mg 2ϩ or Mn 2ϩ . Ca 2ϩ had less effect on the cell spreading than Mg 2ϩ and Mn 2ϩ . This is consistent with results from extensive studies of the effect of divalent cations on integrin activity (32,33). The mAb to human integrin ␣ 1 or ␤ 1 suppressed the adhesion and spreading of human chondrocytes and human MRC5 fibroblasts on dishes coated with CMP. The other tested mAbs to various integrins had a marginal effect on cell adhesion or spreading on CMPcoated dishes. These findings suggest that integrin ␣ 1 ␤ 1 plays a pivotal role in the CMP-mediated cell adhesion and spreading.
Of the various antibodies to integrin subunits examined here, only the antibody to integrin ␣ 1 coprecipitated 125 I-CMP in cell lysates. Unlabeled native CMP competed for the binding to integrin ␣ 1 with 125 I-CMP. Although the anti-integrin ␤ 1 mAb inhibited the effect of CMP on cell adhesion, this mAb (data not shown), as well as the antiserum to integrin ␤ 1 , did not consistently precipitate 125 I-CMP in cell lysates. Under these conditions, the antibody to integrin ␤ 1 may coprecipitate the ␣ 1 subunit only at low levels. In addition, the affinity of integrin ␤ 1 for CMP may decrease in the absence of the intact plasma membrane. In any case, our findings suggest that CMP selectively binds to the integrin ␣ 1 subunit even in the presence of detergents (Nonidet P-40 and SDS).
Integrin ␣ 1 ␤ 1 , as well as ␣ 2 ␤ 1 , binds to collagen under some experimental conditions. However, the precipitation of CMP with anti-integrin ␣ 1 antibodies is not due to the binding of CMP to collagen because 125 I-CMP used in this study did not bind to type I, II, III, IV or V collagen. 3 Furthermore, the lysates were prepared after cells were dispersed from the cell layer with bacterial pure collagenase at a high concentration (0.8 mg/ml) and trypsin. No type I collagen was detected in the cell lysates by immunoblotting. 3 Chondroadherin is also an adhesion protein prominently expressed in cartilage and binds to integrin ␣ 2 ␤ 1 (34). Chondroadherin can promote cell adhesion, but not spreading (34), whereas CMP enhances both cell adhesion and spreading.
These findings suggest that CMP and chondroadherin have different roles in cartilage.
Whether CMP has a special role in cartilage in the presence of other adhesion proteins is not known. It is noteworthy, however, that genetic variation at the CMP gene locus was found to be significantly associated with hip radiographically evident osteoarthritis in 55-65-year-old men, whereas a significant association between hip or knee radiographically evident osteoarthritis in men or women and the cartilage link protein gene was not observed (35). Typically, hip radiographically evident osteoarthritis is most frequently present in men in the 55-65-year age group. Radiographically evident osteoarthritis particularly of the hip is often considered to arise due to anatomic abnormalities. During endochondral bone formation, the length and shape of the bone are determined. If CMP protein plays a distinctive role in the matrix-cell interaction during endochondral bone formation, genetic variation at the CMP gene locus may alter the shape of the skeleton. In addition, CMP may modulate the matrix-cell interaction in articular cartilage of osteoarthritic joints (8).
cDNAs encoding CMP-like proteins (matrilin-2 and -3) were recently cloned (36 -38). Matrilin-2 is expressed in a variety of organs, but not in cartilage (36), whereas matrilin-3 is expressed in a cartilage-specific manner (37,38). Matrilin-3 and CMP form disulfide-linked hetero-oligomers in bovine epiphyseal cartilage (39). These proteins may also function as adhesion factors since their modular structure is similar to that of CMP.
In conclusion, CMP was found to be an adhesion factor for fibroblasts and chondrocytes. CMP is the first protein observed to synergistically increase the effect of collagen on adhesion and spreading. Integrin ␣ 1 ␤ 1 seems to be involved in the CMPinduced cell adhesion. The cartilage-specific adhesion protein CMP may play an important role in the development and repair of skeletal tissues.