Chondrocyte cell death may play an important role in the development of arthritis. The goal of the present study was to evaluate the role of the extracellular matrix (ECM) in promoting chondrocyte survival via signals through the integrin family of ECM receptors.
Chondrocytes were isolated by sequential enzymatic digestion from normal ankle cartilage of organ donors and from osteoarthritic (OA) knee tissue obtained from patients undergoing total knee replacement. Cell survival in monolayer and in suspension culture was measured using fluorescent labels after treatment with specific integrin-blocking antibodies and echistatin, a disintegrin peptide. A quantitative enzyme-linked immunosorbent assay for histone-associated DNA fragments and morphologic evaluation by electron microscopy were used to evaluate apoptosis.
Freshly isolated chondrocytes died when plated in serum-free media at low density on poly-L-lysine, but showed >95% survival on fibronectin (FN). A monoclonal blocking antibody to the α5-integrin subunit (FN receptor) significantly inhibited survival on FN, whereas control antibodies had no effect. Likewise, treatment of freshly isolated chondrocytes in serum-free alginate-suspension culture with the α5-blocking antibody resulted in cell death in a dose-dependent manner, with 20 μg/ml of the antibody reducing normal chondrocyte survival to 20% of that in controls, and OA chondrocyte survival to 23% of that in controls. Antibody inhibition of αv and α1 integrins or treatment with echistatin did not cause cell death. Addition of insulin-like growth factor 1 (IGF-1; 100 ng/ ml) was not able to improve survival of α5-antibody–treated cells. However, treatment with 10% fetal bovine serum improved normal chondrocyte survival to 98% (a 5.1-fold increase) and OA chondrocyte survival to 64% (a 2.8-fold increase). Cell death due to α5 inhibition was associated with apoptosis.
These results demonstrate that chondrocyte survival signals are transmitted via the α5β1 FN receptor. Inhibition of matrix survival signals mediated by α5β1 also inhibits the ability of IGF-1 to promote survival, suggesting that IGF-1–mediated survival signaling may require a cosignal from α5β1.
Integrins are a large family of glycoprotein receptors that mediate cell–cell and cell–extracellular matrix adhesion (1, 2). Integrins are heterodimeric transmembrane proteins consisting of an α and a β subunit. Both subunits have large extracellular domains that interact with extracellular matrix proteins and relatively small cytoplasmic domains that interact with cytoskeletal proteins. The extracellular matrix is an active, information-rich environment. It plays a pivotal role in regulating several biologic functions, such as cell growth, differentiation, adhesion, and migration. Signals from the extracellular matrix via the β 1-subunit–containing integrins have been shown to play a role in chondrogenesis in vitro, although the exact function of integrins during chondrogenesis is not clear (3, 4).
In many cell types, the extracellular matrix functions to promote cell survival. Epithelial and endothelial cells undergo programmed cell death if they are prevented from attaching to an extracellular matrix substrate (5, 6). Both integrins and some of their downstream effectors are involved in the regulation of this process. Depending on the cell type, several integrins (e.g., α 1β1, α5β1, αvβ3, and α6β1) have been shown to mediate cell survival (7–10). However, studies on integrin-mediated survival of adult chondrocytes have not been reported.
Apoptotic cell death has been noted in cultures of chicken sternal chondrocytes treated with anti–β1-integrin antibodies (11, 12). Transgenic mice lacking type II collagen demonstrated high levels of chondrocyte death by apoptosis, suggesting that a loss of collagen somehow disrupts matrix-mediated survival; however, the role of integrins was not determined (13). The objective of the present study was to further evaluate the role of the extracellular matrix in promoting chondrocyte survival via signals through specific integrin receptors, using cells isolated from adult human articular cartilage.
When chondrocytes are removed from cartilage by enzymatic digestion, expression of fibronectin (FN) is dramatically up-regulated to levels 8-fold higher than the RNA level found in the tissue (14). We hypothesized that one purpose for the increase in FN expression was to provide matrix survival signals through the α5β1 integrin. Therefore, chondrocytes isolated from articular cartilage were cultured in the presence of antibodies that specifically block the α5β1 integrin. Cells isolated from normal cartilage were compared with cells isolated from osteoarthritic (OA) cartilage to determine if changes in cell function due to OA would make chondrocytes more susceptible to death when matrix survival signals were blocked. Because matrix survival signals likely work in concert with signals provided by growth factors, we also investigated the ability of insulin-like growth factor 1 (IGF-1) and serum growth factors to promote survival when integrin signals were inhibited.
PATIENTS AND METHODS
Primary chondrocyte cultures.
Normal human chondrocytes were obtained from ankle cartilage of organ donors within 24 hours postmortem. Normal-appearing cartilage (Collins grade 0–1) (15) was used for these experiments. OA cartilage was obtained from the distal femur and the proximal tibia of tissue from patients undergoing total knee replacement surgery for OA. Full-thickness cartilage slices were digested in a 5% CO2 incubator at 37°C with continuous agitation for 1 hour in Dulbecco's modified Eagle's medium (DMEM) /F-12 (1:1) culture media containing 0.2% pronase (Calbiochem, San Diego, CA), and then overnight with 0.025% collagenase P (Roche, Indianapolis, IN) in DMEM/F-12 supplemented with 5% fetal bovine serum (FBS). After isolation, the cells were counted, and initial viability was assessed using trypan blue exclusion and was determined to be >90%.
Cells were cultured under serum-free conditions by either plating on coated coverslips or by placing the cells in suspension in alginate beads. For the coverslip experiments, sterile coverslips (13 mm; Nunc, Naperville, IL) were placed in wells in a 24-well plate and preincubated overnight at 4°C with 10 μg/ml FN (Gibco, Grand Island, NY) or 10 μg/ml poly-L-lysine (Sigma, St. Louis, MO). The wells were gently washed 3 times with phosphate buffered saline (PBS) and then 7.5 × 104 chondrocytes were plated in 400 μl per well of serum-free DMEM/F-12 media with antibiotics, for a final cell density of 3.75 × 104 cells/cm2. Alginate beads were prepared at a density of 2 × 105 cells/bead and were cultured in serum-free DMEM/F-12 supplemented with “mini-ITS+” (consisting of 5 nM insulin [Calbiochem], 2 μg/ml transferrin [BD Biosciences, Bedford, MA], 2 ng/ml selenous acid, 420 μg/ml/2.1 μg/ml bovine serum albumin [ BSA]/linoleic acid [Collaborative Biomedical Products, Bedford, MA]) plus 25 μg/ml ascorbate (Wako Pure Chemical Industries, Richmond, VA) and antibiotics as previously described (16). In experiments testing the effects of IGF-1 and FBS on survival, 100 ng/ml human recombinant IGF-1 (a gift from Chiron, Emeryville, CA) was added to DMEM/F-12 with mini-ITS+, or 10% FBS (Gibco) was added to DMEM/F-12 containing antibiotics. For the 7-day cultures, the medium was changed every other day.
Inhibition of specific integrin subunits. Monoclonal antibodies known to block α1-, αv-, and α5-integrin subunits were used, and included anti-α1 (5E8D9) from Upstate Biologicals (Lake Placid, NY), anti-α5 (NKI-SAM-1) from Chemicon (Temecula, CA), and anti-αv (M9; from Chemicon). Integrin inhibition was also tested using echistatin (Bachem, King of Prussia, PA), a disintegrin that has been shown to be a potent inhibitor of αvβ3 (17, 18). Controls included cultures incubated without antibodies, with nonspecific isotype control IgG (Sigma), or with a nonblocking anti-α5 antibody (HA5; Chemicon). In addition, because some of the antibodies were provided in a buffer containing a preservative (0.001% merthiolate or 0.1% azide), controls were tested with equal final concentrations of the preservative alone. Neither preservative, at the concentrations used in this study, caused chondrocyte death.
For treatment of chondrocytes plated on coated coverslips in 24-well plates, antibodies were added in the media with the cells as they were plated. The plates were gently shaken for 15 minutes at room temperature to evenly distribute the cells and antibody, prior to overnight incubation at 37°C in a 5% CO2 incubator. For treatment of cells cultured in alginate, the freshly made beads were transferred to 96-well plates with 1 bead per well in 100 μl of serum-free DMEM/F-12 media supplemented with mini-ITS+. Integrin and control antibodies or echistatin were added 30 minutes later to each well in duplicate, and then cells were incubated overnight at 37°C in a 5% CO2 incubator. Cell survival was assessed the next day, as described below.
Chondrocyte survival assay. Cell survival was measured by incubating cultured cells with calcein AM and ethidium bromide homodimer 1 (Molecular Probes, Eugene, OR). Calcein AM is a nonfluorescent cell-permeable dye that is cleaved by a ubiquitous intracellular esterase present in live cells to produce a highly green-fluorescent, cell-impermeable product. Ethidium bromide homodimer 1 is a cell-impermeable dye that enters cells following the compromise of plasma membrane integrity and stains nuclear DNA red. Media from cells cultured on coverslips were gently removed so as not to disturb any loosely attached cells and replaced with 400 μl of PBS containing 1 μM calcein AM and 4 μM ethidium bromide homodimer 1. The coverslips were then incubated at room temperature and protected from light. After 45 minutes, the coverslips were placed on a microscope slide and the numbers of dead and live cells were counted using an Olympus (New Hyde Park, NY) fluorescence microscope. Calculations of the percentage of live cells were determined by a total cell count performed in triplicate, with the number of cells counted always exceeding 100 for each data point.
For alginate cultures, after overnight incubation with integrin antibodies or echistatin, the media were removed and replaced with 100 μl of PBS containing 1 μM calcein AM/4 μM ethidium bromide homodimer 1 and incubated at room temperature and protected from light. After 30–45 minutes, the staining solution was removed and the beads were dissolved by shaking for 10 minutes in 20 μl 55-mM sodium citrate with 0.15-mM NaCl. Finally, 15 μl of the dissolved bead was placed on a microscope slide for counting, as described above.
FN immunocytochemistry. Chondrocytes were cultured overnight on poly-L-lysine–coated coverslips in serum-free media as described above. The cells were gently rinsed with PBS and then fixed for 20 minutes with 4% paraformaldehyde at 4°C followed by washing with Tris buffered saline (TBS). Slides were blocked for 3 hours with TBS plus 1% high-purity BSA (Sigma) at 4°C, and then incubated overnight at 4°C with anti-FN monoclonal antibody (P1H11; Chemicon) at 1:100 in TBS plus 0.1% Tween plus 1% high-purity BSA. The slides were then washed with TBS plus 0.1% Tween and incubated for 90 minutes with fluorescein-conjugated anti-mouse secondary antibody (ICN/Cappel, Aurora, OH) at 1:100 in TBS plus 0.1% Tween plus 0.1% high-purity BSA at 4°C. The slides were then washed again with TBS plus 0.1% Tween followed by double-distilled H2O and then mounted with anti-fading aqueous mounting medium (Dako, Carpinteria, CA) and viewed on an Olympus BX 60 microscope equipped for detection of fluorescence.
Apoptosis detection. The presence of apoptotic cells was based on the detection of histone-associated DNA fragments (mono- and oligonucleosomes) using the Cell Death Detection enzyme-linked immunosorbent assay (ELISA) kit (Boehringer Mannheim, Indianapolis, IN) and confirmed using electron microscopy on cells from a selected experiment. Ten beads (∼2 × 106 cells) in 400 μl of media per well of a 24-well plate were treated overnight with anti-α5 antibodies as described above. For the ELISA, the media were removed and the cells were released from the alginate beads by shaking in 1.5 ml 55-mM sodium citrate in 0.15-mM NaCl. The cell suspension was centrifuged to obtain a cell pellet which was then lysed with buffer from the kit. The samples were then used for the ELISA following the manufacturer's directions. The assay is a sandwich enzyme immunoassay for detection of histone-associated DNA fragments using anti-histone (H2A, H2B, H3, and H4) monoclonal antibodies coated onto a 96-well plate, cell-lysate samples from control and treated cultures, and anti-DNA monoclonal antibodies conjugated to peroxidase for detection.
For electron microscopy, the alginate beads were fixed with 2% glutaraldehyde in 0.1M sodium cacodylate buffer and postfixed in 1% osmium tetroxide in 0.1M sodium cacodylate. The cells were dehydrated in ethanol and propylene oxide and embedded in epoxy. Thin sections of ∼600 angstroms were mounted on grids and stained with 0.2% Reynolds lead hydroxide stain plus 2.5% uranyl acetate. Cells were imaged on a JEOL (Palo Alto, CA) 100-CX electron microscope, and representative cells were photographed.
Statistical analysis. Where indicated, results were compared by analysis of variance using StatView software (SAS Institute, Cary, NC). Significance was determined as a P value less than 0.05, and post hoc testing was performed using Fisher's protected least significant difference test.
Promotion of chondrocyte death by specific inhibition of α5-containing integrins. When freshly isolated chondrocytes from normal ankle cartilage were plated in serum-free media at low density on poly-L-lysine, only 34% of the cells survived an overnight incubation, whereas >95% survived when plated on FN (Figure 1A). Chondrocytes primarily utilize the α5β1 integrin for attachment to FN (19). A monoclonal blocking antibody to the α5-integrin subunit significantly inhibited survival in a dose-dependent manner in cells plated on FN, whereas a nonblocking α5 antibody at 50 μg/ml had no effect.
The α5-blocking antibody also reduced survival on poly-L-lysine from 34% to <5%, indicating that the cells that had survived on poly-L-lysine–coated coverslips likely did so through endogenous FN production. In order to test this possibility, chondrocytes incubated overnight on poly-L-lysine in serum-free media were immunostained with anti-FN antibodies. The presence of FN was detected in the pericellular region of surviving cells (Figure 1B).
In the next set of experiments, survival was evaluated using chondrocytes cultured in serum-free suspension culture in alginate. When freshly isolated cells, from either normal ankle cartilage or OA knee cartilage, were incubated overnight with the α5 monoclonal blocking antibody, significant cell death was noted in a dose-dependent manner (Figure 2). Incubation with 20 μg/ml of antibody reduced survival to 20% of that in the nonspecific IgG–treated controls, both in normal chondrocytes and in OA cells, and 50 μg/ml of the antibody reduced survival to < 5%. When the chondrocytes were maintained for 1 week in serum-free media supplemented with mini-ITS+ and then treated overnight with anti-α5 blocking antibodies, a similar dose-dependent decrease in survival was noted, although the proportion of surviving cells was better than that seen in day-1 cells from both normal and OA cartilage (Figure 2). For example, at 20 μ g/ml of anti-α5, 73% of cells from normal cartilage survived at day 7 compared with 20% at day 1 (P = 0.0001), and 61% of cells from OA cartilage survived at day 7 compared with 23% at day 1 (P < 0.02).
Chondrocytes express α1β1 and αvβ3 integrins, both of which have been shown to promote survival in some cell types (7–10). Treatment of alginate-cultured chondrocytes at day 1 of culture with blocking antibodies for the α1 or the αv subunits, in concentrations up to 50 μg/ml, did not cause cell death in cells from normal or OA cartilage (Figure 3). In addition, treatment with the disintegrin echistatin did not cause significant chondrocyte death (Figure 3).
Promotion of chondrocyte survival by IGF-1 or 10% FBS following blocking of the α5-integrin subunit. In previous work, we showed that autocrine signaling by endogenous IGF-1 promoted chondrocyte survival in serum-free alginate cultures (16). Our results, showing >90% cell death with α5-integrin inhibition, indicate that autocrine survival signaling from IGF-1 is not sufficient to maintain chondrocyte survival when α5 is blocked. Therefore, we tested whether addition of exogenous IGF-1 could prevent cell death in the presence of α5-blocking antibodies. Testing of freshly isolated chondrocytes from normal ankle cartilage of 4 donors showed that addition of IGF-1 (100 ng/ml) provided only a modest improvement in survival, which was not statistically significant (Figure 4A). Examining the results for individual donors revealed that cells from 2 donors (ages 63 and 72 years) had some improvement in survival with IGF-1 (average survival went from 17% without IGF-1 to 48% with IGF-1), whereas cells from 2 other donors (ages 63 and 73 years) had no response to IGF-1. There was no benefit from addition of IGF-1 in cells treated with 50 μg/ml of antibody (Figure 4A). In contrast, incubation in the presence of 10% FBS (Figure 4B) improved survival to >98% in the presence of 20 μg/ml anti-α5 (P = 0.0002) and to > 40% with 50 μg/ml anti-α5 (P = 0.002).
Survival of OA chondrocytes did not improve in the presence of IGF-1 and anti-α5 (Figure 4A), whereas 10% FBS was able to improve the survival of OA cells, although survival was lower than among the cells from normal cartilage (Figure 4B). Since it may have been possible that the cell-isolation procedure damaged the IGF-1 receptor and it had not fully recovered at day 1, cells from normal and OA cartilage cultured for 7 days were studied. Day-7 cells also did not respond to IGF-1 when α5 was blocked (data not shown). We also tested whether adding purified human plasma FN to the sodium alginate before the chondrocytes were entrapped or adding FN to the media with the anti-α5 antibody would promote chondrocyte survival. Neither treatment improved survival over that in the controls (data not shown). This could be due to the possibility that the FN molecules may not have formed the proper secondary structure to interact with the FN receptor once the cells were entrapped in the alginate beads. Purified FN when added in solution is known to be in a compact form which unfolds when plated on plastic surfaces, but would be unlikely to unfold in alginate.
Chondrocyte death after incubation with α5-blocking antibodies associated with apoptosis. In chondrocytes isolated from the ankle cartilage of 2 normal donors and cultured in alginate, overnight incubation with the α5-blocking antibody resulted in an increase in the production of histone-associated DNA fragments. As shown in Figure 5, inhibition of the α5-integrin subunit with 20 μg/ ml and 50 μg/ml antibody treatment resulted in a dose-dependent increase in DNA fragments relative to untreated controls. In order to further document apoptotic cell death, chondrocytes treated with the α5-blocking antibody were examined by electron microscopy. Unlike control, untreated cells, α5-antibody–treated cells showed features of apoptosis, including nuclear condensation and membrane budding (Figure 6A versus 6B).
In various cell types studied to date, growth factor signals that promote survival have been found to work in concert with signals from the extracellular matrix (9, 20, 21). Our findings in adult human articular chondrocytes are consistent with an important survival function for the matrix and also demonstrate that loss of a matrix signal may prevent IGF-1 from promoting survival. The term anoikis, from the Greek word for homelessness, has been used to denote apoptosis that occurs in cells detached from their native extracellular matrix (6). In many cell types, matrix binding mediated by the integrin family of receptors has been shown to prevent anoikis. Although chondrocytes survive when cultured in suspension, such as in alginate, they are still surrounded by and attached to an extracellular matrix, which they generate in situ. A pericellular matrix can be visualized around chondrocytes within 4 hours after isolation of the cells (22).
In the present study, cell death associated with evidence of apoptosis was seen in isolated suspension-cultured chondrocytes incubated with α5-integrin–blocking antibodies, suggesting that α5β1 signaling is at least one means of preventing anoikis in suspended chondrocytes. Since α5β1 is a primary chondrocyte FN receptor and blocking the α5 subunit also inhibited survival of chondrocytes plated on FN, these results indicate that the mechanism by which α5β1 promotes chondrocyte survival is most likely through binding to FN. Since the present studies were performed with isolated cells, the findings do not preclude the likely possibility that additional signals provided by other matrix proteins, such as collagen, may be at work in intact cartilage in vivo.
The specific role of α5-subunit–containing integrins in survival signaling has been studied previously in several cell lines. In Chinese hamster ovary cells, attachment to FN through α5β1 integrin, but not through αvβ1 integrin, promoted cell survival under serum-free conditions (8). Similar results demonstrating the importance of α5β1 for survival signaling have been shown with endothelial cells (23), HT29 colon carcinoma cells (24), and neuronal cells (25), to name just a few. The present study is the first, that we are aware of, to directly assess the role of α5-containing integrins in promoting chondrocyte survival. Since the only known partner for α5 is the β1 subunit, the results are consistent with a previous study (11) that demonstrated chondrocyte death in chick sternal organ cultures treated with a β1-blocking antibody. Chondrocytes do express other α subunits that can associate with β1, including α1, α2, α3, α6, and α10 (for review, see ref. 26). We did not see significant cell death with inhibition of α1. Although we did not test the effects of inhibition for all of the potential α subunits, the finding of <5% survival when α5 was completely blocked suggests that other β1-containing integrins do not have a key role, at least in early-suspension culture. Chondrocytes also express αv-containing integrins such as α vβ3, but inhibition of αvβ3 with blocking antibodies or with echistatin did not cause cell death.
Promotion of chondrocyte survival by the α5β1 integrin was particularly important at early time points after cell isolation, when there has not been enough time for the chondrocytes to regenerate a complete matrix. The survival signals provided by the limited amount of extracellular matrix could not counteract the effect of the FN-receptor blockade.
The finding that day-7 cells were less sensitive to the α5 inhibition suggests that either the matrix was preventing the ability of the antibody to completely block the receptor or that other survival signals were able to help protect chondrocytes from cell death at that time. The ability of receptor antibodies to penetrate the pericellular matrix formed by cultured chondrocytes has been shown in previous work, using antibodies to CD44 (27). These studies suggest that lack of antibody penetration is probably not the reason for better survival at day 7.
There was no significant difference between the survival of normal and OA chondrocytes after treatment with increasing doses of anti-α5 antibodies. In previous work, we found that α5β1 is expressed by both normal and OA chondrocytes, and immunohistochemical staining of monkey cartilage did not detect a significant difference in levels of the receptor between normal-appearing cartilage and cartilage with OA-like changes (19). When testing the ability of IGF-1 to promote survival when the α5-integrin subunit was blocked, a response to IGF-1 was detected in 2 of 4 cultures from normal donors and in none of 5 cultures from subjects with OA. The normal donors were of similar ages, making age an unlikely reason for the difference in donor-to-donor response. However, since all of the cultures were obtained from older adult cartilage, it is possible that the known age-related decline in anabolic chondrocyte responsiveness to IGF-1 (28) could have contributed to a reduced survival response to IGF-1. Additional studies using cells from younger donors are ongoing to examine the effects of age in more detail.
A lack of a survival response to IGF-1 in chondrocytes from OA tissue is consistent with previous studies of OA chondrocytes that showed a lack of stimulation of proline incorporation by IGF-1 in human OA chondrocytes in monolayer culture (29) and reduced sulfate and proline incorporation in response to IGF-1 in OA chondrocytes from cynomolgus monkeys with histologic changes of OA (28). Taken together, the results from the normal and OA cultures indicate that IGF-1 survival signaling may require a cosignal from the matrix provided by α5β1. In support of this, a recent study has provided evidence that IGF-1 and β1-containing integrins, including α5β1, can collaborate to activate an Shc–Grb2 complex in chondrocytes (30), and studies in other cell types have associated α5β1 signaling through Shc–Grb2 with activation of mitogen-activated protein kinase signals which regulate control of proliferation and apoptosis (23).
Incubation in the presence of 10% bovine serum was much more effective than IGF-1 in promoting survival when the α 5-integrin subunit was blocked. Serum contains many different growth factors and other proteins that could have promoted survival, and it is difficult to predict which specific factors were responsible. The ability of serum to promote survival in these experiments suggests that the cell death seen with α5 inhibition was not due to some unknown toxic effect of the antibody. Although OA chondrocytes responded to serum when α5 was blocked, they were less responsive than the cells from normal cartilage. The serum and IGF-1 results would suggest that OA cells are less responsive to other survival signals when matrix survival signals are lost.
In summary, the results of this work demonstrate a pivotal role for the extracellular matrix in promoting chondrocyte survival via FN receptors using cells isolated from adult human articular cartilage. Involvement of the α5β1 integrin in preventing cell death may be particularly important at early time points in culture, when the extracellular matrix is being reformed. Compared with serum, IGF-1 was unable to significantly promote survival when α5β1 was blocked, suggesting that IGF-1 survival signaling requires a costimulus from α5β1.
The authors would like to acknowledge the Regional Organ Bank of Illinois and Dr. Arkady Margulis, Department of Biochemistry, Rush Medical College, for providing normal donor tissues. We also thank Dr. Gabriella Cs-Szabo, Department of Biochemistry, and Dr. Richard Berger, Department of Orthopaedic Surgery, Rush Medical College, for providing OA tissue.