Integrin-mediated adhesion of human articular chondrocytes to cartilage

Authors


Abstract

Objective

To determine 1) the kinetics and strength of adhesion of human articular chondrocytes to a cut cartilage surface, and 2) the role of specific integrins in mediating such adhesion, using an in vitro model.

Methods

Human articular chondrocytes isolated from cadaveric donors (mean ± SD age 38 ± 13 years) were cultured in high-density or low-density monolayer. Following release from culture with trypsin and a 2–2.5-hour recovery period, chondrocytes were analyzed either for adhesion to cartilage or for integrin expression by flow cytometry.

Results

Following culture in monolayer, adhesion of chondrocytes to cartilage increased with time, from 6–16% at 10 minutes to a maximum of 59–82% at 80–320 minutes. After 80 minutes of adhesion, the resistance of cells to flow-induced shear stress (50% detachment) was ∼21 Pa. Chondrocyte adhesion to cartilage decreased with pretreatment of cells with monoclonal antibodies that bound to and blocked certain integrins. After an 80-minute incubation time, adhesion of chondrocytes cultured in high-density monolayer decreased from the value of IgG1-treated controls (55%) with blocking of the β1 integrin subunit (to 23%) or with blocking of α5β1 (to 36%). Following expansion of chondrocytes in low-density monolayer, the mechanisms of adhesion to cartilage were generally similar. After an 80-minute incubation time, adhesion of chondrocytes cultured in low-density monolayer decreased from the value of IgG1-treated controls (62%) with blocking of the β1 integrin subunit (to 30%) or with blocking of α5β1 (to 44%). Additionally, adhesion of these cells decreased to 46% by blocking of αvβ5, with a similar trend in effect for chondrocytes cultured in high-density monolayer. Blocking of the α1 or α3 integrin subunits or αvβ3 had no detectable effect on adhesion, even though these receptors were detected by flow cytometry.

Conclusion

Under the culture and seeding conditions studied, β1, α5β1, and αvβ5 integrins mediate human chondrocyte adhesion to cartilage. These chondrocyte integrins have a potential role in the initial adhesion and retention of chondrocytes at a cartilage defect site following clinical procedures of chondrocyte transplantation.

Autologous chondrocyte implantation (ACI) has become a popular clinical therapy for articular cartilage repair (1–3). One hypothesized mechanism for repair in ACI is that the injected chondrocytes are retained within the defect, synthesize and deposit cartilaginous matrix components to fill the defect, and integrate the newly formed tissue with the surrounding host tissue. The initial adhesion of transplanted cells to the surrounding host cartilage may be one mechanism by which cells are retained within a defect, and therefore may be important for integration and long-term success. However, the findings from in vivo animal models (4–8) and an in vitro model (9) of cartilage defect repair suggest that chondrocyte adhesion to cartilage develops slowly and can be deficient.

Binding of certain cell-surface receptors to tissue-matrix ligands appears to be the molecular basis of the initial adhesion of transplanted chondrocytes to cartilage. Articular chondrocytes express receptors, including the integrins α1β1, α3β1, α5β1, α10β1, αvβ3, and αvβ5, that may facilitate binding to these ligands (10–14). In traditional in vitro assays of cell adhesion to coated surfaces, chondrocytes adhere to certain cartilage extracellular matrix proteins, including type II collagen (14, 15), type VI collagen (16), and fibronectin (15, 17), as well as other matrix proteins such as laminin (14, 18) and matrix γ-carboxyglutamic acid protein (15, 19). Although such studies implicate certain receptor–ligand interactions in chondrocyte attachment, steric and interaction effects may exist at the cartilage surface, due to the collection of extracellular matrix components. This makes it difficult to extrapolate from such adhesion studies to the in vivo situation.

Other studies have examined the characteristics and mechanism of chondrocyte attachment to cartilage in vitro (20–23). Several studies examined the attachment of bovine chondrocytes, grown in primary high-density monolayer culture, to a cut surface of bovine cartilage, where adhesion increased with attachment time, in a process mediated primarily by β1 integrin (20–22). However, it is unclear whether β1 integrin facilitates binding of human chondrocytes to human cartilage, and if so, which specific α subunits are involved in β1 integrin–mediated adhesion. In addition, it is possible that other integrins expressed by chondrocytes, such as αvβ3 and αvβ5, may also play a role.

The mechanisms of chondrocyte adhesion to articular cartilage may be influenced by cell-culture conditions prior to transplantation. Chondrocytes maintained in high-density monolayer culture undergo relatively limited proliferation and remain phenotypically stable, as assessed by type II collagen production, for up to 2 weeks (15). In ACI, articular chondrocytes are expanded in low-density monolayer, a culture method that leads to dedifferentiation and requires phenotypic recovery of the transplanted cells in vivo to form cartilaginous matrix components (1, 2, 24). Culture conditions affect integrin expression, and dedifferentiation of the chondrocytes may lead to altered patterns of integrin expression (10, 11, 25). The role of chondrocyte culture conditions prior to transplantation on the subsequent mechanisms of adhesion of chondrocytes to cartilage is unknown.

The objectives of this study were to determine, in human articular chondrocytes cultured in low- and/or high-density monolayer, the characteristics of attachment to cartilage in a transplantation-type procedure, and in particular, 1) the kinetics and strength of adhesion, and 2) the role of specific integrins in mediating adhesion.

MATERIALS AND METHODS

Materials.

Materials for tissue explant, cell isolation, culture, flow cytometry, and biochemical analysis were obtained as described previously (22). Primary monoclonal antibodies (mAb) that block ligand binding to integrin subunits or integrin heterodimers were used (11, 14, 16, 25, 26). The β1 integrin subunit mAb 4B4 and mouse IgG1 were from Coulter Immunology (Hialeah, FL). The α3 integrin subunit mAb P1B5, α5β1 mAb JBS5, αvβ3 mAb LM609, αvβ5 mAb P1F6, mouse IgG, and mouse IgG2a were from Chemicon International (Temecula, CA). The α1 integrin subunit mAb 5E8D9 was from Upstate Biotechnology (Lake Placid, NY). Phycoerythrin-conjugated goat anti-mouse IgG and 5-chloromethylfluorescein diacetate (CMFDA) were from Molecular Probes (Eugene, OR). Bovine serum albumin (BSA) and L-ascorbic acid were from Sigma-Aldrich (St. Louis, MO). Trypsin was from Life Technologies (catalog no. 25095-019; Rockville, MD). Non–tissue-culture–treated plates were from Becton Dickinson (catalog no. 351147; Franklin Lakes, NJ).

Chondrocyte isolation and culture.

The human articular cartilage used in this study was obtained from a tissue bank. Cartilage with a similar appearance was analyzed histologically and exhibited either a normal, smooth or slightly roughened surface, with a Mankin-Shapiro score that was typically 0–1 (27). Chondrocytes were isolated from human articular cartilage from the distal femurs or tali of cadaveric donors (mean ± SD age 38 ± 13 years; n = 19). Chondrocytes were isolated from the cartilage by digestion with pronase and collagenase (28). The isolated chondrocytes were plated at 200,000 cells/cm2 for high-density culture or 20,000 cells/cm2 for low-density culture, and incubated for 9–10 days in a humidified 37°C and 5% CO2 atmosphere in medium (Dulbecco's modified Eagle's medium [DMEM], 10 mM HEPES, 0.1 mM nonessential amino acids, 0.4 mML-proline, 2 mML-glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin, and 0.25 μg/ml amphotericin B) supplemented with 10% fetal bovine serum (FBS). For low-density cultures, medium was supplemented with 50 μg/ml ascorbate. The medium (0.6 ml per cm2 for a 12.5-cm2 or 25-cm2 flask) was changed every 3 days for the duration of culture.

To facilitate cell tracking, some cultures were radiolabeled with 5–50 μCi/ml 3H-thymidine on days 0–3 or days 3–6 of culture, and then rinsed with medium 3 times over 30 minutes and maintained in culture (22, 29). Other cells were fluorescently labeled overnight on the final day of culture with 10 μM CMFDA; this was previously found to have no detectable effect on chondrocyte adhesion to cartilage (20).

Prior to use, cells were released with trypsin (0.05% solution in phosphate buffered saline [PBS] for 5–10 minutes), rinsed twice with medium with 10% FBS, collected, and resuspended in medium with 10% FBS. Proliferation of chondrocytes in low-density monolayer was 3.2 ± 1.1 times greater (mean ± SD; n = 7) than in high-density monolayer, as assessed by hemocytometer counting. Cell suspensions were transferred to non–tissue-culture–treated plates, and the cells were allowed to recover from trypsin for 2 hours on an Adams Nutator (Becton Dickinson, Franklin Lakes, NJ) in an incubator. Chondrocytes were used either for flow cytometry analysis of integrin expression or for an adhesion assay, of which there were 2 types, a rapid-screening adhesion assay and a parallel-plate shear-flow adhesion assay.

Flow cytometry.

Flow cytometry was used to assess receptor expression and completeness of receptor binding. To determine the concentration of antibodies required to maximize receptor binding, cells were incubated in medium with 10% FBS and the following primary antibodies: 4B4 (1, 10, or 20 μg/ml), 5E8D9 (1.7, 17, or 34 μg/ml), P1B5 (1, 10, or 20 μg/ml), JBS5 (1:1,000, 1:100, or 1:50 dilution), LM609 (1, 10, or 20 μg/ml), or P1F6 (1, 10, or 20 μg/ml). Following primary-antibody incubation, cells were rinsed with medium with 10% FBS at 4°C, incubated 30 minutes at 4°C in medium with 10% FBS and 1.5 μg/ml phycoerythrin-conjugated secondary antibody, rinsed twice in 0.1% NaN3 in PBS at 4°C, and then fixed in 0.5% paraformaldehyde in PBS. For each sample, 1,000–10,000 cells were analyzed on a Becton Dickinson FACScan, with voltage to the fluorescence detector set at 469 volts.

To assess receptor expression, cells were incubated in medium with 10% FBS for 30 minutes with primary antibodies against the β1 integrin subunit (10 μg/ml 4B4), α1 integrin subunit (10–17 μg/ml 5E8D9), α3 integrin subunit (10 μg/ml P1B5), α5β1 (1:100 dilution JBS5), αvβ3 (10 μg/ml LM609), or αvβ5 (10 μg/ml P1F6). Control cells for these experiments were incubated similarly with isotype control antibodies. Following primary antibody incubation, the cells were prepared for flow cytometry and analyzed as described above.

Adhesion assays.

Cartilage tissue sections for the adhesion assays were prepared from the patellofemoral groove of 12 cadaveric donors (mean ± SD age 33 ± 9 years). Osteochondral blocks (∼1 cm3) were cut using a bone saw and copious irrigation was performed with PBS. Cartilage sections were then cut 100-μm thick, approximately parallel to the articular surface, on a microtome (Microm, Walldorf, Germany). The first few sections, including the articular surface, were discarded. All remaining sections were rinsed in PBS and stored at −70°C until used in adhesion assays.

Rapid-screening adhesion assay.

The rapid-screening assay of chondrocyte adhesion to cartilage was performed as described previously (22). Briefly, individual cartilage sections were secured onto the end of a Teflon tube, creating a well with cartilage forming the bottom surface. Aliquots of chondrocytes, tagged with 3H-thymidine, were seeded onto the cartilage surface within each well. Assuming 7.7 × 10−12 gm DNA/chondrocyte (30), the calculated number of cells seeded was 38,000 ± 8,000 (mean ± SD; n = 63) and equivalent to a cell density of 210,000 ± 45,000 cells/cm2. Cells were allowed to settle to the cartilage surface before additional medium was added to fill the well, and the samples were left undisturbed in a 37°C incubator for a prescribed duration. The samples were then gently inverted, and the cartilage substrate with attached cells and media containing detached cells were each solubilized with proteinase K and analyzed for 3H-radioactivity (RackBeta Liquid Scintillation Counter; Wallac, Gaithersburg, MD). Chondrocyte adhesion to cartilage was calculated as the fraction of 3H-radioactivity present on the cartilage relative to the total 3H-radioactivity.

The rapid-screening adhesion assay was used to investigate 1) the time course of adhesion, 2) the effects of various media used during the seeding process, and 3) the role of specific integrins in mediating adhesion.

To assess the time course of adhesion, chondrocytes were allowed to adhere by incubation for 10, 20, 40, 80, 160, or 320 minutes. Control experiments confirmed that the majority of the 3H-radioactivity was associated with the cells after the longest incubation (320 minutes), since the percentage of total radioactivity that was present in the cell pellet following centrifugation (5 minutes at 750g) of a cell suspension averaged 97 ± 1% (mean ± SD; n = 15).

To investigate the effects of seeding medium on adhesion, chondrocyte preparations were rinsed twice and resuspended in DMEM, DMEM + 1 mg/ml BSA, DMEM + 0.5% FBS, or DMEM + 10% FBS. Chondrocyte suspensions were seeded and settled to the cartilage surface before more of the same medium was added to fill the well. Chondrocyte adhesion to cartilage was determined following a 10-, 80-, or 320-minute incubation.

To investigate the role of specific integrins in mediating adhesion, chondrocytes were rinsed twice in serum-free medium and incubated for an additional 30 minutes with primary mAb that block ligand binding to the β1 integrin subunit (final concentration of 10 μg/ml mAb 4B4), α1 integrin subunit (17 μg/ml 5E8D9), α3 integrin subunit (10 μg/ml P1B5), α5β1 (1:100 dilution JBS5), αvβ3 (10 μg/ml LM609), or αvβ5 (10 μg/ml P1F6). Control cells were similarly incubated in 10 μg/ml IgG1. Chondrocyte suspensions were seeded and settled to the cartilage surface before additional medium containing a blocking reagent, when appropriate, was added to fill the well. Chondrocyte adhesion to cartilage was determined following an 80-minute incubation.

Parallel-plate shear-flow adhesion assay.

The parallel-plate shear-flow adhesion assay was performed as described previously (22). Briefly, a suspension of fluorescence-labeled cells in serum-free medium was infused into the chamber, the cells were allowed to settle to the cartilage surface by gravity and left undisturbed for 80 minutes, and then the chamber was inverted and affixed to the stage of a microscope (Diaphot 300 or Eclipse TE 300; Nikon, Melville, NY). Images of the chondrocytes on the cartilage surface were captured. The total number of cells in the image field (NT) was defined to be the sum of those cells present in the preflow images of the cartilage surface (NS) and those cells present on the glass slide directly below the cartilage surface (NG). The total number of cells present in the image field (1.4 mm2) just after chamber inversion was 189 ± 59 (mean ± SD; n = 8). The fraction of attached cells (FA) at a shear stress of 0 Pa (with chamber inversion) was computed as

equation image

Medium was then driven through the chamber to produce shear stresses of 6, 28, and 70 Pa at the cell surface for 30 seconds (a duration sufficient to reach a steady state of cell detachment [20]). The flow rate was adjusted according to the measured channel height (mean ± SD 173 ± 20 μm; n = 8) for each assembly of the chamber. The fraction of attached cells (FA) for each level of applied shear stress was similarly computed from images of the cartilage surface captured after each increment of flow.

Statistical analysis.

Data are presented as the mean ± SD. Comparisons of fractional attachment data were made using analysis of variance after transforming the data with a modified arcsine function to improve normality (31). Incubation duration, type of medium used during seeding, and treatments with blocking reagents were analyzed as fixed effects. The donor from which the cells were isolated was analyzed as a random effect. Where required, Tukey's tests were used for post-hoc analysis for comparison with controls. Statistical analysis was implemented with Systat 5.2.1 (Systat, Evanston, IL). Data relating to the median fluorescence were analyzed using an unpaired t-test implemented with Excel (Microsoft, Redmond, WA).

RESULTS

Chondrocytes became increasingly adherent to cartilage with increasing incubation time, as assessed by the rapid-screening adhesion assay. Adhesion of chondrocytes to cartilage following culture in high-density monolayer and seeding in medium with 10% serum increased (P < 0.001) with longer durations of incubation, from a minimum value of 8% at 10 minutes to a maximum of 79% at 160 minutes (Figure 1A). For these cells, the change in cell adhesion between successive time points was significant (each P < 0.001) from 10 minutes up to 80 minutes, but not between 80 minutes and 160 minutes nor between 160 minutes and 320 minutes (P = 0.99 and P = 0.94, respectively). Similarly, adhesion to cartilage of chondrocytes following culture in low-density monolayer and seeding in medium with 10% serum increased (P < 0.001) with longer durations of incubation (Figure 1B).

Figure 1.

Effect of incubation duration on chondrocyte adhesion to cartilage using the rapid-screening adhesion assay. 3H-thymidine–labeled chondrocytes, cultured in high-density (A) or low-density (B) monolayer and suspended in Dulbecco's modified Eagle's medium + 10% fetal bovine serum, were seeded onto a cartilage surface and left undisturbed at 37°C for the duration indicated (in minutes [min]) before chamber inversion. Cell adhesion was determined as the fraction of total radioactivity present on the cartilage. Data are the mean and SD of 8 (A) or 4–5 (B) samples.

The kinetics of chondrocyte adhesion to cartilage were affected only slightly by the medium used during transplantation, with a slight interactive effect (P < 0.05) (Figure 2) between medium type and incubation duration. Following an 80-minute incubation, adhesion of chondrocytes cultured in high-density monolayer was relatively high (59–75%) for all groups and not markedly different from that at 320 minutes (69–76%). Therefore, an 80-minute incubation time was selected for subsequent experiments.

Figure 2.

Effect of seeding medium on the kinetics of chondrocyte adhesion to cartilage using the rapid-screening adhesion assay. 3H-thymidine–labeled chondrocytes, cultured in high-density monolayer and suspended in Dulbecco's modified Eagle's medium (DMEM), DMEM + 1 mg/ml bovine serum albumin (BSA), DMEM + 0.5% fetal bovine serum (FBS), or DMEM + 10% FBS were seeded onto a cartilage surface and left undisturbed at 37°C for the duration indicated (in minutes [min]) before chamber inversion. Cell adhesion was determined as the fraction of total radioactivity present on the cartilage. Bars show the mean and SD (n = 7–8).

The strength of adhesion of chondrocytes to cartilage after the 80-minute attachment time was quantified using the shear-flow adhesion assay (Figure 3). The fraction of chondrocytes cultured in high-density monolayer that remained adherent to cartilage decreased with increasing levels of applied shear stress, from a maximum of 0.69 ± 0.13 following chamber inversion (0 Pa) to a minimum of 0.18 ± 0.15 after application of 70 Pa shear stress. The adhesive strength, defined as the shear stress required to cause 50% cell detachment (20), was 21.3 Pa. Furthermore, 50% of the initially adherent cells were resistant to the application of 67.4 Pa of shear stress.

Figure 3.

Shear-induced detachment of human chondrocytes, cultured in high-density monolayer, from cartilage. Cells suspended in Dulbecco's modified Eagle's medium were infused into a shear-flow chamber and left undisturbed for 80 minutes before the application of increasing levels of shear stress. Data are the mean and SD (n = 8).

Flow cytometry was used to assess receptor expression and completeness of receptor binding. Analysis with various concentrations of the primary antibodies qualitatively confirmed that receptor binding was near maximum levels at the antibody concentrations described in the adhesion assays. With the primary antibody concentrations used in the adhesion assays, the fluorescence values during flow cytometry were a median of 77–100% of the maximum fluorescence values for all the concentrations tested.

Flow cytometry analysis of receptor expression showed that after high- or low-density primary monolayer culture, trypsinization, and a 2-hour recovery, chondrocytes expressed all of the integrins tested. Analysis of the median fluorescence data (Table 1) showed staining with antibodies against the β1 integrin subunit, α1 integrin subunit, α3 integrin subunit, α5β1, αvβ3, and αvβ5, as compared against the values for the corresponding controls (mean ± SD arbitrary units 18 ± 14 [n = 7], 4 ± 1 [n = 3], and 17 ± 13 [n = 8] for IgG1-, IgG2a-, and IgG-treated controls, respectively). The median fluorescence was significantly higher for chondrocytes cultured in low-density monolayer than for those cultured in high-density monolayer with antibodies against the β1 integrin subunit (P < 0.01; n = 3–4), α3 integrin subunit (P < 0.05; n = 3–4), α5β1 (P < 0.001; n = 3–5), αvβ3 (P < 0.01; n = 4), and αvβ5 (P < 0.01; n = 2–4). Median fluorescence with antibodies against the α1 integrin subunit (P = 0.2; n = 3–4) did not depend on culture density.

Table 1. Flow cytometry analysis of integrin expression by human chondrocytes*
 β1α1α3α5β1αvβ3αvβ5
  • *

    Cells were cultured in high- or low-density monolayer culture and then analyzed after trypsinization and a 2-hour recovery period. Values are the mean ± SD arbitrary units of median fluorescence (no. of samples).

  • P < 0.01 versus low-density culture.

  • P < 0.05 versus low-density culture.

  • §

    P < 0.001 versus low-density culture.

High-density culture550 ± 370 (4)120 ± 60 (4)81 ± 39 (4)260 ± 220 (5)§25 ± 15 (4)310 ± 230 (4)
Low-density culture1,900 ± 180 (3)200 ± 70 (3)290 ± 140 (3)990 ± 150 (3)72 ± 10 (4)1,400 ± 240 (2)

Chondrocyte adhesion to cartilage was blocked by treatment with certain integrin-specific antibodies. With the rapid-screening adhesion assay, chondrocyte adhesion to cartilage following an 80-minute incubation was dependent on antibody pretreatment of chondrocytes cultured in high-density (Figure 4A) or low-density (Figure 4B) monolayer (each culture P < 0.001; n = 14–17). Pretreatment to block the β1 integrin subunit (each culture P < 0.001) or α5β1 (high-density P < 0.001, low-density P < 0.05) significantly decreased chondrocyte adhesion to cartilage in comparison with the values for IgG1-treated controls. Blocking of αvβ5 decreased adhesion (P < 0.05) of chondrocytes cultured in low-density monolayer and had a similar trend in effect on chondrocytes cultured in high-density monolayer (P = 0.1). There was no significant effect on adhesion for chondrocytes treated to block the α1 integrin subunit, α3 integrin subunit, or αvβ3 (P = 0.7–1.0).

Figure 4.

Effect of reagents that specifically bind and block the β1 integrin subunit (monoclonal antibody [mAb] 4B4), α1 integrin subunit (mAb 5E8D9), α3 integrin subunit (mAb P1B5), α5β1 (mAb JBS5), αvβ3 (mAb LM609), or αvβ5 (mAb P1F6) on chondrocyte adhesion to cartilage using the rapid-screening adhesion assay. Suspensions of chondrocytes, cultured in high-density (A) or low-density (B) monolayer, were preincubated in Dulbecco's modified Eagle's medium with a blocking reagent, seeded onto a cartilage surface, and left undisturbed at 37°C for 80 minutes before chamber inversion. Cell adhesion was calculated as the fraction of total radioactivity present on the cartilage. Bars show the mean and SD (n = 14–17). ∗∗∗ = P < 0.001 and ∗ =P < 0.05 compared with IgG1-treated controls.

DISCUSSION

This study investigated the kinetics, strength, and integrin-mediated mechanisms of adhesion of human articular chondrocytes to a cartilage surface in a transplantation-type procedure. With increasing duration of attachment, chondrocytes became increasingly adherent to cartilage (Figures 1 and 2) in a manner that was generally independent of the composition of the culture medium during the attachment period (Figure 2). At the time of transplantation, chondrocytes expressed the α1, α3, and β1 integrin subunits, as well as α5β1, αvβ3, and αvβ5 (Table 1). The inclusion of antibodies that specifically block integrin binding (11, 14, 16, 25, 26) indicated that chondrocyte adhesion to cartilage was mediated by β1 integrins, specifically α5β1, as well as by αvβ5, with only slight differences between chondrocytes cultured at high (Figure 4A) and low (Figure 4B) density.

A variety of factors may affect the extent or mechanisms of chondrocyte adhesion to a cartilage surface, either in vitro as examined herein or in vivo in a clinical scenario. The expression of surface receptors on chondrocytes can be modulated by the method of culture. The increased expression of integrins during culture, noted previously for β1 integrin during high-density monolayer culture (11), appears even more pronounced for other integrins after cell proliferation during low-density culture (Table 1). Extensive passaging of chondrocytes may further affect the levels and patterns of integrin expression (10).

Higher levels of integrin expression by chondrocytes expanded in culture may be a result of various related factors, including the proliferative state, plating density, and inclusion of ascorbate in the culture medium. Ascorbate tended to enhance the levels of β1 integrin expression after 7 days of culture (11), although the subsequent effect on chondrocyte adhesion to cartilage requires further investigation. The levels and patterns of integrin expression have been positively correlated with chondrocyte attachment to protein-coated surfaces (25). In the present study, however, elevated levels of integrin expression by chondrocytes following low-density culture with ascorbate-supplemented media did not markedly affect the kinetics or extent of adhesion to a cartilage substrate. Although they may be expressed, some integrins may be inactive and therefore may not participate in adhesion (32). In addition, cell adhesion may be limited by ligand density, which may vary over territorial and interterritorial regions of the cartilage surface and with depth from the articular surface (33, 34).

Heterogeneity in integrin expression and adhesion properties may reflect differences between subpopulations of chondrocytes. The lack of full cell detachment after treatment to block a single integrin receptor is consistent with the possibility that multiple adhesion mechanisms may be operative. Annexin V, a receptor for type II collagen (35), as well as CD44, a receptor for hyaluronan (36), are expressed by chondrocytes and mediate interactions with cartilage matrix components. However, in a study using the same adhesion assays, these molecules did not appear to play a major role in the adhesion of bovine chondrocytes to bovine cartilage (22). Integrin or nonintegrin receptors, which were not studied herein, may also mediate chondrocyte adhesion to cartilage. Nevertheless, under the conditions used in our study, several integrins were implicated in adhesion of human chondrocytes to cartilage, whereas others were not.

Integrins containing the β1 subunit appear to be responsible for chondrocyte adhesion to cartilage, as shown in studies of human (Figure 4) and bovine (22) species. Although α5β1 appeared to be the primary mediator of chondrocyte adhesion to cartilage, the exact nature of the receptor–ligand interaction remains to be clarified. Fibronectin is the primary ligand for α5β1 and is a component of articular cartilage that is elevated in osteoarthritis (37). In addition to mediating chondrocyte attachment, α5β1 can mediate cell spreading (38), transduce mechanical stimuli (39), and also interact with fibronectin or fibronectin fragments to regulate chondrocyte signaling, cytokine production, and matrix synthesis/degradation (26, 40, 41).

Since the blocking of β1 integrin appeared to inhibit chondrocyte adhesion slightly more than did blocking of α5β1, and blocking of the α1 and α3 integrin subunits did not alter adhesion (Figure 4), other β1 integrins may be involved in chondrocyte adhesion to cartilage. β1 integrins have been shown to bind many ligands present in cartilage, including type II collagen (12, 14, 25, 35, 42). While the α integrin partner for the β1 subunit responsible for chondrocyte adhesion to type II collagen has not been determined, it may be α10β1, a type II collagen receptor expressed by articular chondrocytes (12, 25).

As shown in Figure 4, αvβ5 also appeared to mediate chondrocyte adherence to cartilage. Again, the exact nature of the receptor–ligand interaction remains to be clarified. It is known that αvβ5 is expressed not only by chondrocytes (10), but also by other cells of the skeletal system, including synovial cells (43) and osteoblasts (44). The natural ligands for αvβ5 in osteoblasts appear to be vitronectin and osteopontin (44). Osteopontin is present in osteoarthritic cartilage (45) and may also serve as a ligand for αvβ5 in normal cartilage. Alternatively, the ligand for αvβ5 in chondrocytes may be fibronectin, which has been implicated as a ligand for this integrin in metastatic tumor cells (46).

Although flow cytometry confirmed the presence of the α1, α3, and αvβ3 integrins on the cultured chondrocytes, these receptors did not mediate attachment of chondrocytes to cartilage under the same test conditions. The α1β1 integrin has been characterized as the primary chondrocyte receptor for type VI collagen (16, 25). Type VI collagen has been localized to the pericellular matrix surrounding chondrocytes in vivo (47) and can be a substrate to which chondrocytes attach (16), suggesting that it is important in cell–matrix interactions in vivo. However, type VI collagen may not be accessible in the cartilage matrix, and thus its interaction with α1β1 may not be significant in the attachment of chondrocytes to a cut cartilage surface. The apparent lack of involvement of the α3 integrin subunit in chondrocyte adhesion to cartilage is consistent with previous findings in which the effects of blocking the α3 integrin subunit were negligible in attachment of isolated chondrocytes to type I collagen–, type II collagen–, or fibronectin-coated culture plates (11, 14). Expression of αvβ3 may be limited to chondrocyte subpopulations, and the contribution of this receptor may not be detectable in the homogenous population of chondrocytes isolated from full-thickness articular cartilage used in the adhesion assay (10).

Quantitative measures, such as those obtained with the shear-flow adhesion assay, allow for analysis of the possible induction of cell detachment in vivo by mechanical stress during and following transplantation. In these studies, the cells were exposed to shear stresses over a broad range (0–70 Pa), which is likely to include the level of shear present within the joint under a variety of normal and abnormal conditions, such as following transplantation procedures. Chondrocytes embedded within a normal articular cartilage matrix are predicted to encounter a shear stress on the order of 0.1 Pa, due to fluid flow past the cells (20). However, in cases of softer and more permeable repair tissue (48–50), and in partially filled defects where gaps may exist, fluid velocities and corresponding shear stresses may be substantially higher (22). The quantitative data of the parallel-plate shear-flow assay may be useful for predicting whether or not transplanted chondrocytes remain attached at the repair site.

Although investigations using additional culture conditions and substrates are necessary to characterize chondrocyte adhesion in specific clinical procedures, the observation that human chondrocyte adhesion to cartilage develops with time, in a manner dependent on β1 integrins, α5β1, and αvβ5, may have important implications. In cartilage repair procedures involving transplantation of chondrocytes, the attachment of cells to the surrounding host tissue may be the first of a series of events that are important for a successful repair. Delineation of the governing mechanisms of chondrocyte adhesion to cartilage may allow cell manipulations or matrix treatments to enhance chondrocyte adhesion, retention, and subsequent function at a defect site.

Acknowledgements

We thank Dr. Scott Ball, Dr. Seth Williams, and Professor David Amiel for facilitating acquisition of cartilage specimens, and Dr. Kelvin Li, Ms Lisa Lottman, Mr. August Sage, and Dr. Michael Voegtline for technical assistance.

Ancillary