Integrins are cell surface receptors involved in cellular processes including adhesion, migration, and matrix assembly. In the present study, we analyzed the possible involvement of α5β1 integrin in the regulation of chondrocyte adhesion, spreading, and proliferation. We found that rabbit growth plate chondrocytes were able to attach to substrates coated with type I collagen, type II collagen, or fibronectin within 24 h of culture. During this time period, attachment to fibronectin appeared to be dependent on α5β1 integrin, whereas adhesion to collagens was not. By day 3 of culture, chondrocytes spread onto all the substrates tested. We found that regardless of the nature of the substrate, cell spreading was reversed by treatment with RGD peptide or antibodies against α5β1 or fibronectin, indicating that cell spreading involved α5β1 and fibronectin endogenously produced and deposited by the chondrocytes themselves. Colony formation by chondrocytes in soft agar was inhibited by treatment with RGD peptides or BIIG2, an antibody that interferes with α5β1 integrin–ligand interactions. Furthermore, DNA content was decreased by treatment with anti-fibronectin antibody in micromass culture of chondrocytes. Immunohistochemical analysis on tissue sections revealed that the α5 subunit was particularly abundant in the proliferative and hypertrophic zones of growth plate. The results of the study indicate that α5β1 integrin plays multiple roles in chondrocyte behavior and function and appears to be involved in the regulation of both chondrocyte–matrix interactions and proliferation.
Integrins are dominant receptors for extracellular matrix components including collagens, fibronectin, laminin, and vitronectin.(1–3) They are heterodimers composed of α and β subunits, and the combination of α and β subunits determines ligand specificity.(2) The interaction of cell surface integrins with the extracellular matrix induces cell adhesion, spreading, migration, and matrix assembly.(1,2) Through these cellular actions, integrins are involved in a variety of biological responses, including cell proliferation and differentiation, tissue organization, immune response, hemostasis, and metastasis and invasion of tumor cells.(1,2)
Recently, we demonstrated that β1 integrins mediate the interaction of chicken embryo chondrocytes with type I collagen, type II collagen, and fibronectin.(4) A similar role of β1 integrins in substrate adhesion of human chondrocytes was described by Durr et al.(5) These authors suggested that α2β1 integrin functions as a receptor for type II collagen and is involved in the assembly of collagen fibrils in cartilage.(5) Both human(5,6) and chicken(4) chondrocytes have been found to express relatively high amounts of α5β1, but the physiological role of this integrin in chondrocytes remains unclear. This integrin has been suggested to function as a fibronectin receptor(2) and to be involved in a variety of biological processes. For example, mutations in the α5 subunit cause defects in posterior trunk and mesoderm formation.(7) Interaction of α5β1 with fibronectin may play important roles in the morphogenesis of muscle cells,(8,9) wound healing, and inflammation.(10) In contrast, the loss of α5β1 function is closely related to terminal differentiation of keratinocytes.(11)
In this study, we investigated the possible roles of α5β1 in chondrocytes. We have found that α5β1 integrin is involved in the adhesion and spreading of chondrocytes, and that α5β1-mediating signals are required for colony formation of chondrocytes, suggesting that α5β1 is a positive regulator of proliferation in these cells.
MATERIALS AND METHODS
A rat monoclonal antibody to human α5β1 integrin, BIIG2, which is the same as B1E5,(12) was generously provided by Dr. C. Damsky. The supernatant of BIIG2 hybridoma cell cultures, rather than purified antibody, was used for the experiments because of limited supply. As controls, we used normal rat immunoglobulins and the supernatant fraction that did not bind to protein A Sepharose (Pharmacia LKB Biotechnology, Piscataway, NJ, U.S.A.); both showed no effect on rabbit chondrocytes. Rabbit polyclonal antibodies against the cytoplasmic domains of β1, α3, α6, and αv integrin subunits were prepared according to the method of Marcantonio and Hynes.(13) A rabbit polyclonal antibody against the cytoplasmic domain of α5 and an anti-fibronectin rabbit polyclonal antibody purified IgG fraction were purchased from Chemicaon International Inc. (Temecula, CA, U.S.A.) and Gibco BRL (Grand Island, NY, U.S.A.), respectively.
Rabbit chondrocytes were isolated from the growth plate of 4-week-old male New Zealand white rabbits as described previously.(14) Cells were grown in α-minimal essential medium (α-MEM) containing 0.5 ng/ml basic fibroblast growth factor (bFGF), 1.0 × 10−8 M hydrocortisone, 30 μg protein/ml egg lipoprotein, and 30 μg/ml transferrin (Medium A).
For agar cultures, cells were suspended in 0.25 ml of 0.41% bacto agar in Ham's F-12/Dulbecco's modified Eagle's medium (DMEM) (1:1) containing 10% fetal bovine serum (FBS), and the cell suspension was overlaid on a 0.25 ml base layer of 0.72% bacto agar in the same medium at a density of 1.0 × 104/ml.(15) For micromass cultures,(16) the isolated rabbit chondrocytes were inoculated at the density of 1.0 × 105 cells per tube in α-MEM containing 10% fibronectin-depleted FBS with or without anti-fibronectin antibodies and centrifuged at 1800 rpm for 5 minutes. The cells were cultured for 7 days without changing medium. From day 8, when the cell pellets were easily detectable, the medium was changed every other day.
Preparation of extracellular matrix protein substrates
Tissue culture plates (48-well plate; Costar, Cambridge, MA, U.S.A.) were coated with bovine plasma fibronectin (Wako Pure Chemical Industry, Osaka, Japan), bovine plasma vitronectin (Koken, Tokyo, Japan), or mouse EHS laminin (Koken) by incubation overnight at 37°C in 0.2 ml of phosphate-buffered saline (PBS) containing 10 μg of the protein. For collagen substrates, the plates were coated with type I collagen or type II collagen (Koken) solution (0.6 mg/ml in 0.02 N HCl) for 5 minutes, excess solution was removed, and the plates were dried at room temperature. The coated plates were blocked with 0.3% bovine serum albumin (BSA) solution for 30 minutes at 37°C and washed twice with PBS.
Cell attachment and spreading assays
Isolated chondrocytes were inoculated on substrate-coated plates (48-well plate) at the density of 2.5 × 104/well and maintained in Medium A. For cell attachment assay, various reagents were added to the cultures at the time of plating. After 16 h, cultures were washed with PBS three times, fixed with 3.7% formaldehyde, and stained with 0.5% crystal violet solution. The number of attached cells was determined. The freshly isolated chondrocytes hardly attach on the substrates within 12 h. Even after 12 h, the cells were easily detached by washing with PBS. Therefore, we carried out cell attachment assay 16 h after plating, although endogenous production of extracellular matrix (ECM) may contribute to adhesion. At this time point, the number of the attached cells was 60–70% of initially plated cells.
For cell spreading assay, cultures were grown for 2 days. On day 3, cultures were treated with reagents, washed with PBS, and fixed with 3.7% formaldehyde at the appropriate time. The numbers of round and spread cells were determined microscopically.
Cells were plated on glass coverslips coated with type I collagen in 35-mm tissue culture dishes at 1.5 × 105 cells/dish in Medium A. Cultures were washed with PBS three times, fixed with 3.7% formaldehyde for 8 minutes at 22°C, and permeabilized with 0.5% NP-40 in PBS when necessary. The cultures were then blocked by incubation in blocking buffer (PBS containing 5% FBS and 1% BSA) for 1 h at 37°C and incubated in primary antibody for 2 h at 22°C. The staining was visualized by incubation with a rhodamine-conjugated secondary antibody (Capel Lab., Cochranville, PA, U.S.A.). For F-actin staining, cultures were incubated with rhodamine-conjugated phalloidine (Molecular Probes, Eugene, OR, U.S.A.) for 30 minutes. Microphotographs were taken using T-max film (Kodak, Rochester, NY, U.S.A.).
Cross-linking and extraction techniques
Cultures were washed with PBS three times and incubated with 0.4 mM BS3 (Bis(sulfosuccinimidyl)suberate) (Pierce, Rockford, IL, U.S.A.) in PBS containing 2 mM polymethylsulfonylfluoride (PMSF) for 10 minutes at 22°C. The cross-linking reaction was stopped by the addition of PBS containing 10 mM Tris-HCl (pH 7.4) and 2 mM PMSF for 2 minutes at 22°C. Cultures were washed four times with PBS and extracted for 5 minutes at 22°C with RIPA buffer (0.1% SDS, 0.1% deoxycholic acid, 1.0% NP-40, 150 mM NaCl, and 50 mM Tris-HCl, pH 7.4). Cultures were then fixed with 3.7% formaldehyde for 8 minutes at 22°C and examined by immunofluorescence staining.
Immunoblotting and precipitation
Chondrocyte cultures were lysed in RIPA buffer containing 2 mM EDTA and 2 mM PMSF for 30 minutes on ice. Cell lysates were clarified by centrifugation at 12,000 rpm, and the resulting supernatants were frozen at −70°C. Proteins were separated by sodium dodecyl sulfide polyacrylamide gel electrophoresis (SDS-PAGE) on 7.0% gels, loading the same amount of protein per lane, and transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore Japan, Tokyo, Japan). After blocking the membrane with 10% horse serum containing 0.02% NaN3 overnight at 22°C, the membrane was incubated with primary antibody. Bound antibody was visualized using vectastain ABC kits (Vector Laboratories, Inc., Burlingame, CA, U.S.A.).
For immunoprecipitation studies, the cells were labeled with 100 μCi/ml [35S] methionine (1074 Ci/mmol; ICN Biomedicals, Inc., Irvine, CA, U.S.A.) in methionine-free MEM overnight. The cell lysate was incubated with BIIG2 (1:20) overnight at 4°C and then with protein A-Sepharose (Pharmacia LKB Biotechnology), which had been preabsorbed with unlabeled cell extracts for 2 h at 4°C. Samples were eluted from protein A-Sepharose by incubation at 60°C for 15 minutes in Laemmli sample buffer, and proteins were separated by SDS–PAGE on 7% gels. Gels were treated with Amplify (Amersham Corporation, Arlington, Heights, IL, U.S.A.) and exposed to X-ray films (Kodak X-Omat AR, Kodak, Rochester, NY, U.S.A.).
Growth plate cartilages were isolated from ribs of 4-week-old male New Zealand white rabbits. Tissue fragments were fixed by 4% paraformaldehyde in PBS, decalcified with 7.5% ethylene diamine tetraacetic acid (EDTA), and embedded in O.C.T. compound (Miles, Elkhart, IN, U.S.A.). Samples were kept at −70°C until use. Ten-micrometer-thick frozen sections were thawed on poly-L-lysine-subbed glass slides, air-dried, treated with 3% H2O2 in PBS for 30 minutes to block endogenous peroxidase, and treated with 600 U/ml of bovine testis hyaluronidase in 1% BSA and 3% normal swine serum for 30 minutes at 37°C. Sections were then processed for immunohistochemistry with antibodies described above and the peroxidase-labeled-streptavidin-biotin staining kit (Histofine, Nichirei, Tokyo, Japan).
Adhesion and spreading of rabbit chondrocytes are dependent on α5β1 integrin
In a first series of experiments, we investigated which substrate can induce cell attachment of rabbit chondrocytes. Chondrocytes isolated from the ribs of 4-week-old male rabbits were plated on a variety of extracellular matrix-coated substrates and cultured in Medium A (see Materials and Methods). After 16 h, the number of cells attached to the different substrates was determined. Type I collagen, type II collagen, and fibronectin strongly promoted cell attachment of rabbit chondrocytes (Fig. 1), whereas vitronectin and laminin substrate were less effective (data not shown). Cell attachment to every substrate tested was inhibited by the addition of 5 mM EDTA, but attachment was restored by the addition of divalent cations such as Ca2+ or Mg2+ (data not shown).
Attachment to fibronectin but not to type I collagen was inhibited when GRGDS peptide or anti-fibronectin antibody was added at the time of plating (Fig. 1A). But addition of inactive GRGES peptide had no effect (Fig. 1A). Because the interaction of α5β1 integrin with fibronectin is dependent on RGD sequences,(17,18) the above data suggested that α5β1 integrin was involved in adhesion of chondrocytes to fibronectin. To confirm this prediction, we carried out cell attachment assays using a monoclonal antibody against human α5β1 integrin (BIIG2). Addition of the BIIG2 antibody inhibited chondrocyte attachment to fibronectin but not to type I collagen. As negative control, we used normal rat IgGs at a concentration of 50 μg/ml, which is comparable to the concentration of immunoglobulins contained in the BIIG2 hybridoma supernatant used above. Both the normal rat IgGs and the supernatant fraction that did not bind to protein A-Sepharose had no effects (Fig. 1B). As shown in Fig. 2A (lane 1), the BIIG2 antibody precipitated α and β integrin subunits, which were recognized by antibodies specific for β1 and α5 (Fig. 2A, lanes 2 and 3, respectively). These data indicate that the BIIG2 specifically recognizes rabbit α5β1 integrin and that α5β1 mediates attachment of chondrocytes to fibronectin.
In the above experiments, treatment with antibodies to α5β1 and fibronectin and with GRGDS peptide at the time of plating did not affect cell attachment to type I collagen substrate (Fig. 1). In contrast, when the anti-α5β1 antibody or the anti-fibronectin antibody was added to cultures grown on type I collagen on day 3, the cells were detached from the substrate during the following 24 h (Table 1). Cell detachment was also observed when GRGDS peptide was added on day 3 (Table 1). The same results were obtained when type II collagen was used instead of type I collagen (data not shown). These data indicated that chondrocytes plated on type I or II collagen initially attach to the provided substrate and that the initial attachment is not dependent on α5β1 integrin, but cell attachment changes and becomes dependent on α5β1 integrin within 3 days of culture.
Table Table 1. Attachment and Detachment of Chondrocytes Plated on Type I Collagen Substrate
When rabbit chondrocytes attach to the substrate, the cells initially remain round but become fully spread on day 3. Spreading of the cells seeded on type I collagen substrate was reversed by treatment with the antibody to α5β1 or GRGDS peptide from day 3 to day 4 of culture (Table 2). The inactive peptide GRGES and nonimmune IgGs had no effects (Table 2). The reversal of cell spreading by these reagents was also observed in cultures plated on type II collagen and fibronectin substrates (data not shown). The addition of anti-fibronectin antibody also changed cell morphology from flat to round within 3 h (data not shown).
Evaluation of integrins involved in substrate adhesion in chondrocytes by immunofluorescence staining
The above data suggest that adhesion and spreading of chondrocytes depend on α5β1 integrin by day 3 of culture and probably require fibronectin produced by the cells themselves. Immunofluorescence staining using anti-fibronectin antibody revealed that chondrocytes produced and accumulated fibronectin in their pericellular matrix, and that fibronectin was localized to the terminus of actin fiber (Figs. 3A and 3B), indicating that fibronectin was one of the components of adhesion plaques in chondrocytes. The involvement of α5β1 in the adhesion and spreading of chondrocytes was confirmed using a method that we developed recently.(19) In this method, the integrins involved in cell adhesion are first cross-linked to the substrate by incubation with a cross-linker, and then the integrins not involved in adhesion and not cross-linked are removed by detergent extraction. Accordingly, 4-day-old cultures grown on type I collagen were treated with the cross-linker and then extracted with detergent. The integrins cross-linked to the substrate were then examined by immunofluorescence staining. Even though type I collagen was provided to chondrocytes as the initial substrate, β1 and α5 (Figs. 3C and 3D) were cross-linked to the substrate, while α3 and αv (Figs. 3E and 3F), although also expressed in chondrocytes (Fig. 2B), were not. Staining of α5 revealed streak-like structures underneath the cells (Fig. 3D), quite similar to staining by anti-β1 antibody (Fig. 3C). Figures 3G and 3H show double immunofluorescence staining of fibronectin and α5 after cross-linking and extraction in 3-day-old cultures where the chondrocytes had just spread. Fibronectin, produced by the cells, colocalized with α5, indicating that both molecules are involved in cell adhesion.
Involvement of α5β1 integrin in agar colony formation by chondrocytes
Next, we examined the possibility that α5β1 is involved in the regulation of chondrocyte proliferation. We had observed that in the presence of anti-α5β1 antibody, chondrocytes detached from the substrate and their growth was slow (data not shown). However, it was not clear from these experiments whether the inhibition of growth was a direct effect of the anti-α5β1 antibody on cell proliferation or an indirect effect caused by cell rounding and detachment. Chondrocytes have the unique ability to proliferate in agar cultures in response to basic fibroblast growth factor (bFGF) treatment without losing their phenotype.(15) Thus, we used a suspension culture system in agar to investigate the involvement of α5β1 integrin in proliferation of chondrocytes, which allowed us to distinguish effects on proliferation from effects on cell adhesion and morphology.
Chondrocytes were seeded in soft agar, and colony formation was induced by the addition of bFGF (0.3 ng/ml) (Figs. 4A and 4D). When the anti-α5β1 antibody was added to the cultures at the time of inoculation, colony formation was strongly inhibited (Fig. 4B), but nonimmune IgG had no effect (Fig. 4C). Likewise, GRGDS peptide inhibited colony formation (Fig. 4E), while the inactive peptide GRGES did not (Fig. 4F). Colony formation was reduced 20 and 50% by the addition of anti-α5β1 and GRGDS peptide, respectively, compared with control (Fig. 5).
Disruption of interaction with fibronectin inhibited cell proliferation in micromass cultures.
In agar cultures, we investigated whether anti-fibronectin antibody also inhibited colony formation of chondrocytes. However, the anti-fibronectin antibody showed only a marginal inhibitory effect. Because the accessibility of antibodies to newly secreted proteins may be imperfect in agarose cultures, we carried out the same experiment using micromass cultures in centrifuge tubes.(16) In this culture system, chondrocytes also proliferate in an anchorage-independent manner like agar cultures, but antibodies added to the liquid medium should more effectively reach the target proteins accumulating around the cells. As shown in Table 3, treatment with anti-fibronectin antibody decreased DNA content by 50% of control value without affecting glycosaminoglycan synthesis, suggesting that the interaction of chondrocytes with fibronectin is required for cell proliferation.
Table Table 3. Antifibronectin Antibody Inhibited Cell Proliferation of Chondrocytes in Micromass Cultures
Distribution of α5β1 integrin and fibronectin
To correlate our in vitro data to the in vivo condition, we examined the distribution of α5β1 integrin in growth plate cartilages. Interestingly, the α5 integrin subunit was particularly abundant in the proliferative and hypertrophic zones and less obvious in the maturation zone (Fig. 6B). In contrast, β1 subunit was homogeneously distributed (Fig. 6A). Fibronectin was concentrated to the territorial matrix in growth plate cartilage (Fig. 6C).
α5β1 integrin mediates cell adhesion and spreading of chondrocytes
The results of our study indicate that α5β1 integrin mediates the interaction of chondrocytes with extracellular components, fibronectin in particular, and that this interaction may play a key role in the control of chondrocyte proliferation. The rabbit chondrocytes display the ability to attach initially to type I and type II collagens and fibronectin substrates. However, within the following 2–3 days of culture, cell adhesion becomes dependent on fibronectin and α5β1 integrin. This dependency on α5β1 and fibronectin is observed regardless of the substrate that is initially provided to the cells. Because we used serum-free conditions in this study, this dependency cannot be attributed to contamination by serum components. The dependency of cell spreading and attachment on α5β1 integrin was not due to changes of integrin expression because no changes were observed in the expression levels of α subunits associated with β1 subunit on the cell surface (data not shown). Therefore, it appears that the chondrocytes initially attach to the substrate provided, but with time they produce and accumulate fibronectin in their pericellular matrix and use it for adhesion. However, the possibility remains that other endogenous products may play roles in cell adhesion in chondrocytes and that α5β1-fibronectin interactions may mediate the function of these unknown molecules as well.
Signals through α5β1 integrin are required for colony formation of chondrocytes
Except for hematopoietic cells, most normal cells attach and spread onto the substrate in monolayer cultures, and cell attachment and spreading are essential for their survival and proliferation. When the integrins are responsible for cell adhesion and spreading, disruption of their function causes inhibition of both cell spreading and proliferation. Therefore, it is normally difficult to discriminate between these two events and it is not clear whether ECM signals through integrins are required for cell proliferation directly or not. In the case of chondrocytes, however, suspension cultures and the accompanying round cell configuration allow the cells to express a stable phenotype.(20–23) In soft agar, even single chondrocytes can survive and express a differentiated phenotype, indicated as the synthesis of aggrecan and type II collagen.(21) When stimulated by FGF treatment, the chondrocytes actively proliferate and produce cell colonies in agar.(15) We demonstrate here that colony formation is inhibited when the interaction of α5β1 with its ligand is blocked by treatment of the antibody BIIG2. These findings indicate that a signal from the extracellular matrix through α5β1 integrin is required for colony formation. The chondrocytes remain round and do not spread in soft agar cultures, whereas they spread with well-organized cytoskeleton and adhesion plaques in monolayer cultures. Thus, it is likely that the suppression of colony formation results from an inhibitory effect on proliferation, rather than an indirect effect due to inhibition of cell adhesion and spreading.
The interaction of α5β1 integrin with fibronectin is a positive regulator in proliferation of chondrocytes
The RGD sequence of fibronectin is defined as a binding site of α5β1.(17,18) The addition of GRGDS peptides inhibited colony formation by chondrocytes, suggesting that the RGD sequence is equally necessary for recognition of the ligand by α5β1 in these cells. Because fibronectin is currently the only known ligand for α5β1,(2) it is likely that the ligand of α5β1 involved in colony formation of chondrocytes is fibronectin. This is consistent with our data that addition of antibodies to fibronectin inhibited cell proliferation of chondrocytes in micromass cultures. In agar cultures, the antibody to fibronectin failed to inhibit colony formation. We raise the following two points as the reasons for the discrepancy: (1) the accessibility of the antibody was incomplete in agar and (2) large enough amounts of the antibody are not able to be provided in the soft agar culture system.
It was originally thought that fibronectin is a product of dedifferentiated chondrocytes,(24–26) and that the amount of fibronectin increases in degenerative cartilage conditions such as osteoarthritis.(27,28) However, fibronectin synthesis is now known to characterize normal chondrocytes.(29,30) We have confirmed that fibronectin is a product of normal chondrocytes and is particularly abundant in the matrix of chondrocytes in growth plates. We propose that the interaction of α5β1 with fibronectin could exert a positive regulatory role in proliferation of chondrocytes in growth plates.
Several investigations have suggested that α5β1 is a negative regulator of cell proliferation. Overexpression of α5β1 in CHO cells suppresses the formation of colonies in soft agar and the growth of tumors in nude mice.(31) Furthermore, a negative correlation has been found between tumorigenesis or the efficiency of colony formation and the level of α5β1 expression.(32) These findings appear to contradict our conclusion that α5β1 is required for the proliferation of chondrocytes in soft agar culture and regulates proliferation in the growth plate. Varner et al.(33) have reported, however, that transformed cells selected for overexpression of α5β1 decrease proliferation and lose the transformed phenotype and that ligation of α5β1 with fibronectin substrate in these cells reverses the growth inhibition. They have suggested that the absence of ligation of α5β1 with fibronectin inhibits cell proliferation of transformed cells, whereas the occupancy of this receptor with fibronectin contributes to their proliferation. This might explain the controversy about the role of α5β1 in cell proliferation. We suggest that chondrocytes secrete fibronectin, that the interaction of α5β1 with fibronectin triggers cell proliferation, and that interference of α5β1 interaction with fibronectin by the anti-α5β1 antibody and RGD peptide may produce a negative signal for cell proliferation. This would be reminiscent of the absence of ligand for α5β1 in transformed cells which would result in the cessation of cell proliferation.
We thank Dr. C. Damsky for the generous gift of the antibody. This work is supported by grants-in-aid for scientific research from the Ministry of Education, Science and Culture of Japan.