Growth-related oncogene α induction of apoptosis in osteoarthritis chondrocytes




To evaluate the apoptotic effect of the chemokine growth-related oncogene α (GROα), which we recently reported to be up-regulated in osteoarthritis (OA) chondrocytes. Chondrocyte apoptosis is considered to be a major determinant of cartilage damage in OA, a disease resulting from the aberrant production of inflammatory mediators (cytokines and chemokines) and effectors (matrix metalloproteinases and reactive oxygen and nitrogen species) by chondrocytes.


We investigated the apoptotic effect of GROα on isolated human cells and on in vitro–cultured cartilage explants by conventional methods (morphology, detection of DNA fragmentation in situ and in solution, exposure of phosphatidylserine) and by analysis of “early” biochemical events (plasma membrane depolarization, activation of caspase 3, and phosphorylation of c-Jun N-terminal kinase/stress-activated protein kinase).


We clearly demonstrated that GROα was able to initiate a series of morphologic, biochemical, and molecular changes that led to chondrocyte apoptosis. Moreover, we found that additional signals delivered from the extracellular matrix (ECM) were essential in the control of chondrocyte susceptibility to GROα-induced apoptosis, since cell death was detected only when cells were stimulated after reestablishment of their proper interactions with the ECM, or in cartilage explant samples with reduced ECM, as indicated by decreased Safranin O staining.


GROα can induce apoptosis in articular chondrocytes, and the induction is dependent upon additional signals from the ECM. These findings are relevant to understanding the pathogenesis of OA, in view of the availability of the GROα chemokine in the joint space in the course of this rheumatic disease.

Apoptosis has been recognized to play a relevant role in cartilage damage in osteoarthritis (OA) (1), although the extent of the phenomenon varies from study to study (2, 3). Apoptosis also has secondary degradative effects due to the presence of postapoptotic cell byproducts that are not effectively removed from the tissue, such as pyrophosphate and calcium crystals (4). Furthermore, apoptotic bodies have been reported to have high contents of matrix-degrading enzymes (4, 5). Apoptotic cells have been mainly localized in two distinct zones of OA cartilage: the superficial layer of cartilage in late-stage disease and the clusters containing proliferating chondrocytes (1).

Hashimoto et al (6) have reported a linkage between chondrocyte apoptosis and cartilage degradation. It is likely that in this case, apoptosis induction occurs following the loss of chondrocyte interaction with the extracellular matrix (ECM) (7), a mechanism typical of adherent cells. Other investigators (8) have suggested that apoptosis is the counterpart of cell proliferation, which is typical of the regenerating/proliferating cartilage. There is a close link between cell proliferation and apoptosis: when a cell picks up the machinery to proliferate, it also acquires an abort pathway (9). One of the first reports of chondrocyte apoptosis identified it in the hypertrophic region of growth plates (10), and it has recently been documented that in OA cartilage, the distribution of molecules relevant to apoptotic death (Bcl-2, Bax, and Fas) is correlated with regions of larger “cloning of chondrocytes” (8). On the other hand, it has been shown that the cells of the upper layer of OA cartilage present some features that are common to growth cartilage development and mineralization, such as the formation of mineral deposits, chondrocyte hypertrophy, terminal differentiation, and apoptosis (11).

Molecular signaling via soluble factors has been shown to be crucial to cartilage homeostasis, having been implicated not only in chondrocyte differentiation, but also in normal maintenance, as well as in aging and disease (12, 13). Therefore, it is likely that it also plays a relevant role in the induction of apoptosis. Two mechanisms of chondrocyte apoptosis have been described thus far: the Fas pathway and nitric oxide (NO) (1). Although chondrocytes in the superficial and upper middle zones of cartilage were shown to express Fas antigen, it is not known whether these cells express Fas ligand in intact cartilage (14). The only source of Fas ligand that has been described is inflammatory cells in the synovial tissue and synovial fluid (1), and inflammatory cells are found only occasionally in OA (1). Furthermore, many investigators have linked the occurrence of apoptosis to increased levels of NO, which acts as second messenger for prototypical cytokines (interleukin-1 [IL-1] and tumor necrosis factor [TNF]) (15, 16). Although induction of apoptosis via NO generated from exogenous NO donors has been demonstrated, the possibility that IL-1 or TNF induces apoptosis by means of such a reactive radical is a subject of controversy (17–20). Moreover, apoptosis has been successfully induced in several immortalized chondrocyte cell lines (21), which lack significant expression of inducible NO synthase (22). Thus, although the relevance of apoptosis in OA has been acknowledged, no soluble factors have yet proved to be a convincing protagonist in its induction.

A novel autocrine loop has been reported by our group: chondrocytes express and release chemokines (IL-8, growth-related oncogene α [GROα], monocyte chemoattractant protein 1 [MCP-1], macrophage inflammatory protein 1α [MIP-1α], MIP-1β, and RANTES) (23, 24) and present the specific receptors (CCR1, CCR2, CCR3, CCR5, CXCR1, and CXCR2) (25). Receptor expression tends to be enhanced in OA chondrocytes, and interaction of chemokine receptors with the corresponding ligands induces the release of matrix-degrading enzymes (matrix metalloproteinase and lysosomal glycosidase), which is also augmented in OA compared with normal chondrocytes (25). This supports the hypothesis that some of these chemokine/receptor loops are likely to contribute to the induction of the catabolic program with a potency comparable with that of IL-1. It has recently been reported that aside from the interaction with their receptors, chemokines have a second important interaction with cell surface glycosaminoglycans (GAGs), particularly sulfated GAGs (26). Furthermore, ECM GAGs might play a role in governing the local chemokine concentration, supporting the formation of solid gradients around the cells (26).

A link between chemokines and apoptosis is strongly suggested by recent reports suggesting a proapoptotic role for CXCR1, CXCR2 (27), and CXCR4 (28) as well as an antiapoptotic role for CCR2 (29) and CCR8 (30). In view of the hypothesized correlation of apoptosis and cell proliferation, it is noteworthy that the same proapoptotic receptors induce cell proliferation in other models (29, 31–34). Moreover, taking into account the linkage between apoptosis and loss of survival signals from the ECM, it is also important that chemokines have been shown to induce the production of matrix metalloproteinases by chondrocytes (25).

This study is the first to provide evidence that the GROα chemokine triggers apoptosis in chondrocytes and that this phenomenon is adhesion dependent and is associated with a marked depolarization of the cells. Apoptosis was confirmed by findings of TUNEL analysis, by evidence of DNA laddering in apoptotic cells, morphologic examination of nuclear modifications (May-Grünwald-Giemsa staining), annexin labeling, caspase 3 activation, and plasma membrane depolarization, and colocalization of TUNEL positivity, activation of caspase 3, and phosphorylation of c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK).


Cells and culture conditions.

We evaluated the apoptotic effect of GROα on isolated chondrocytes as well as on cells cultured in an intact ECM (cartilage explants). In 14 patients, confirmation of the apoptotic effect of GROα was performed on primary cultures of chondrocytes in at least 2 independent experiments. In 4 patients, first- or second-passage cultures were also used.

With the approval of the ethics committee of the Istituti Ortopedici Rizzoli, cartilage specimens were obtained from 20 OA patients undergoing hip and knee replacement surgery. Primary chondrocytes were isolated by sequential enzymatic digestion, as described by Pulsatelli et al (24), and cells were counted and cell viability was assessed.

In 2 patients, the role of the ECM was examined by comparing the apoptotic effect of GROα in chondrocytes derived from residual degenerating cartilage (type IV cartilage) versus nascent regenerating cartilage (type V) obtained from the same patient. Cartilage type was determined based on published criteria (35). Briefly, type IV cartilage has a yellow color and represents the residual tissue that may show all possible stages of degradation and Safranin O staining from slightly to severely impaired. Type V cartilage is white and contains hyperactive chondrocytes involved in reparative processes, with strong pericellular staining for Safranin O.

Chemokines and other reagents.

GROα and the other chemokines were prepared in the laboratory of Dr. Ian Clark-Lewis (Biomedical Research Center, University of British Columbia, Vancouver, British Columbia, Canada) by automated solid-phase chemical synthesis using an Applied Biosystems 430A peptide synthesizer (Applied Biosystems, Foster City, CA) and the tertiary butyloxycarbonyl and benzyl protection strategy (36). Most experiments were performed overnight with 100 nM GROα, as described previously (25). This dose was selected as being optimal for the release of matrix-degrading enzymes based on time-course and dose-kinetic experiments. Anti-Fas monoclonal antibody (clone CH-11; Medical and Biological Laboratories, Nagoya, Japan) and staurosporine (Sigma, St. Louis, MO) were also used as controls.

Cell morphology.

Analysis of morphologic evidence of apoptosis was performed in May-Grünwald-Giemsa–stained chamber slides of chemokine-treated and control cultures. Second passage OA chondrocytes were used since the higher morphologic homogeneity is better suited for the evaluation of apoptosis-specific features. These cells were also used to examine the effect of cell adhesion on the induction of apoptosis.

An 8-well chamber slide was seeded at 11 × 103 cells/well in 400 μl of 3% fetal calf serum (FCS)–Dulbecco's modified Eagle's medium (DMEM). Two series of samples were prepared. The first series was stimulated 24 hours after seeding, and the second series was stimulated 4–6 days after seeding, after the cells had recovered their proper interactions with the ECM (37, 38). All samples were stimulated for either 4 hours or overnight with 100 nM or 500 nM GROα and 100 nM or 500 nM RANTES.

Labeling of the 3′ end for electrophoresis of DNA fragmentation.

Primary chondrocytes were seeded at a density of 80 × 103 cells/well in 96-well plates in 200 μl 3% FCS–DMEM. After 4–6 days, the cells were stimulated overnight with 500 nM GROα. To collect sufficient cells for subsequent analysis, 5 wells per condition were pooled. Supernatants were collected, and detached cells were recovered by centrifugation and added to the adherent cells in 200 μl of proteinase K buffer (0.1M Tris HCl, 50 mM EDTA, pH 8.0, with 500 μg/ml of proteinase K [Roche Molecular Biochemicals, Basel, Switzerland]). Digestion continued overnight at 37°C. The suspension was extracted by phenol/chloroform and concentrated in a Microcon 100 device (Amicon, Beverly, MA) to ∼10 μl.

Each extract was 3′ end-labeled with digoxigenin-11-dUTP using terminal deoxynucleotidyl transferase for 60 minutes at 37°C to generate a short tail, using a Digoxigenin Oligonucleotide Tailing kit (Roche). DNA was loaded on a 2% agarose gel in Tris–borate–EDTA buffer. After ∼2 hours, an overnight Southern transfer was set up according to the alkaline method, and digoxigenin immunodetection was performed according to the manufacturer's instructions.

In situ 3′ end-labeling of DNA for detection of cell death.

Isolated chondrocytes.

After isolation, 400 × 103 cells/well were seeded in a 24-well plate in 500 μl of 3% FCS–DMEM. After 4 days, cells were stimulated overnight with 500 nM GROα or with 0.5 μM staurosporine (positive control). Supernatants were then collected to recover detached cells and added to the trypsinized cells obtained from the adherent monolayer. After washing in phosphate buffered saline, cells were fixed with 4% paraformaldehyde for 15 minutes on ice. To maximize cell recovery, a 5-minute incubation with trypsin at 37°C was performed. The cells were postfixed in 70% ethanol, stored at −20°C, and processed with the Apo-Bromodeoxyuridine kit (Alexis, Lausen, Switzerland) as instructed by the manufacturer. Samples were analyzed on a FACStar Plus flow cytometer (Becton Dickinson, Mountain View, CA) equipped with an argon ion laser at 488 nm, 200 mW. To monitor nonspecific retention of the labeled nucleotide, a no-enzyme control was included, using the staurosporine-treated cells.

Organ cultures.

It has previously been demonstrated that compared with chondrocytes isolated by digestion from native cartilage and cultured in vitro, the response of these cells to different stimuli may be dramatically different when chondrocytes are placed in culture (ex vivo) as explants with an intact articular cartilage ECM (19). Therefore, we set up a series of explant cultures utilizing cartilage specimens obtained at the time of knee arthroplasty from 4 OA patients.

From a homogeneous area, we cut full-thickness pieces of tissue ∼5 × 5 mm. The explants were allowed to recover for 24 hours in DMEM with 10% FCS in a 24-well plate and were then stimulated in culture for 7 days with the following factors: 100 units/ml of IL-1β (Roche), 100 nM MCP-1, 100 nM RANTES, and 100 nM GROα. Cultures were then washed and fixed with 4% paraformaldehyde for 3 hours, placed in 10% buffered formalin, and embedded in paraffin.

Histologic evaluation was performed on 5-μm sections of cartilage comprising all tissue layers. The induction of apoptosis was evaluated using the In Situ Cell Death Detection AP kit (Roche) according to the manufacturer's instructions. Sections derived from an unstimulated control were used to characterize OA-specific alterations, particularly the loss of ECM (Safranin O) or its mineralization (alizarin red S).

Molecular markers of apoptosis (activation of caspase 3 and phosphorylation of JNK/SAPK).

Isolated chondrocytes.

Chondrocytes stimulated in vitro with 100 nM GROα, along with recognized apoptotic factors, were recovered by trypsinization and immediately fixed for 25 minutes at room temperature with 10% buffered formalin. After permeabilization (phosphate buffered saline, 0.5% Triton X-100) and blocking of nonspecific binding (Tris buffered saline, 2% bovine serum albumin, 0.1% sodium azide, 5% normal swine serum), the cells were immunolabeled overnight at 4°C with 1 μg/ml of anti–active caspase 3 polyclonal antibody (Promega, Madison, WI) and then incubated with 25 μg/ml of fluorescein isothiocyanate (FITC)–conjugated swine anti-rabbit immunoglobulin secondary antibody (Dako, Glostrup, Denmark). The cells were counterstained with 1 μg/ml of propidium iodide and examined under a fluorescence microscope (Axiophot; Zeiss, Oberkochen, Germany); 100–300 cells were counted for each condition.

We also confirmed the activation of caspase 3 in chondrocytes stimulated with 100 nM GROα using the PhiPhiLux G1D2 kit (Oncolmmunin, Gaithersburg, MD), a unique class of fluorogenic substrate for the detection and measurement of enzyme activity in living cells. Freshly isolated chondrocytes (80 × 103/well) were plated in a 96-well plate. After 4 days, duplicate wells were treated overnight with 100 nM GROα or 0.75 μg/ml of anti-Fas and then processed according to the application note on the kit for adherent cells. Fluorescence intensities were analyzed on the SpectraMax Gemini plate fluorometer (Molecular Devices, Sunnyvale, CA). To maximize accuracy of the readings, the instrument was set in the well-scan mode with 488 nm excitation and 530 emission (excluding, in the emission, wavelengths preceding 515 nm). Results are expressed as the percentage of increase compared with unstimulated cells, calculated according to the following formula: [(fluorescence intensity of stimulated cells − fluorescence intensity of unstimulated cells)/fluorescence intensity of unstimulated cells] × 100.

Organ cultures.

Sections (5 μm) obtained from the paraffin-embedded organ cultures were dewaxed with xylene, rehydrated with a graded ethanol series, and then processed according to the immunohistochemistry procedure described previously (25). In serial sections, the coordinated activation of caspase 3 and the phosphorylation of JNK/SAPK were evaluated using 5 μg/ml of anti–active caspase 3 polyclonal antibody and anti–phospho-JNK/SAPK (Thr183/Tyr185) antibody (New England Biolabs, Beverly, MA), respectively.

In situ localization of apoptotic cells by annexin V staining.

Staining of apoptotic cells was performed with an FITC-labeled annexin V apoptosis detection kit (Oncogene Research Products, Boston, MA) exploiting the Rapid protocol, which allows direct binding of FITC-labeled annexin V in tissue culture media. Both primary chondrocytes (80 × 103/well) and first or second passage chondrocytes (4.4 × 103/well) were seeded in a 96-well plate in 200 μl of DMEM with 3% FCS. After 72 hours, the medium was exchanged, and duplicate wells were stimulated overnight with 100 nM GROα or 0.75 μg/ml of anti-Fas antibody. FITC-labeled annexin V was then added at 0.5 μl/well and maintained at room temperature for 30 minutes in medium containing the media-binding reagent.

Cells were washed by centrifugation, and 200 μl/well of 1× binding buffer was added to ensure retention of annexin V. Fluorescence intensities were read by using the SpectraMax Gemini plate fluorometer (Molecular Devices) set in the well-scan mode, at 488 nm excitation and 560 nm emission (excluding, in the emission, wavelengths preceding 530 nm). The results were expressed as described for the detection of active caspase 3 in living chondrocytes using the PhiPhiLux G1D2 Kit.

Confocal microscopy of live cells.

Freshly isolated chondrocytes (300 × 103 cells) were left to adhere to the central part of a 24 × 50–mm coverslide for 2–4 hours and then cultured in petri dishes in DMEM. Four days later, the coverslide was positioned onto a flow chamber and kept at 37°C during analysis with a Multiprobe 2001 confocal laser scanning microscope from Molecular Dynamics (Sunnyvale, CA) (39). Samples were observed through a 60× objective with a pinhole set at 50 μm. Cells were dye loaded by adding 400 nM bisoxonol (Molecular Probes, Eugene, OR) in DMEM for ∼1 hour until stabilization of the signal, with laser intensity, photomultiplier setting, and gain kept constant throughout the experiment (40). Each series of measurements was acquired at an equatorial section chosen at a vertical height suitable for focusing on round cells.

GROα and RANTES were both added at a concentration of 100 nM in DMEM supplemented with the fluorescent probe. After triggering, serial images of a preselected microscopic field were acquired at fixed intervals. Images were rendered in a 256 discrete scale of intensities, each corresponding to a pseudocolor. Fluorescence intensity was expressed as the average of the intensities of the pixels pertinent to cells within a microscopic field.

Statistical analysis.

As appropriate to nonparametric analysis, data distributions are presented as the median (25th and 75th percentiles), and data were analyzed for statistical significance using the Wilcoxon matched pairs test and the Mann-Whitney U test. Statistical computations were performed using CSS Statistica statistical software (StatSoft, Tulsa, OK).


Dependence of GROα induction of apoptosis on adequate cell adhesion.

Only chondrocyte cultures allowed to adhere for 4–6 days before the experiment showed morphologic evidence of apoptosis upon treatment with GROα, such as aggregated chromatin, fragmented nuclei, and condensed basophilic cytoplasm (Figure 1). The cells that underwent these alterations appeared to form clusters, the number and size of which increased with dose and time. (Cluster size was measured by counting the number of cells per cluster.) Data are therefore reported as the mean ± SD of at least 8 clusters counted.

Figure 1.

Morphologic evidence of chondrocyte apoptosis induced by growth-related oncogene α (GROα). Light microscopy of May-Grünwald-Giemsa–stained unstimulated (UN) and chemokine-treated cultures of osteoarthritis chondrocytes at the second passage is shown after a, 4 hours and b, 24 hours. RANTES was completely ineffective, while GROα exerted an apoptotic action that was kinetically dependent on time and concentration. Morphologic markers of apoptosis (shown at higher magnification in the insets) include the round shape, condensed basophilic cytoplasm, aggregated chromatin, and fragmented nuclei. (Original magnification × 100 in a and b; × 400 in insets.)

We examined the entire 0.8-cm2 area of each well of an 8-well chamber slide and counted 165 clusters of apoptotic cells (16.11 ± 6.66 cells/cluster) in cells stimulated overnight with 500 nM GROα, 26 clusters (6.4 ± 4.15 cells/cluster) in cells stimulated overnight with 100 nM GROα, 11 clusters (5.56 ± 2.65 cells/cluster) in cells stimulated for 4 hours with 500 nM GROα, and 8 clusters (3.29 ± 1.50 cells/cluster) in cells stimulated for 4 hours with 100 nM GROα. RANTES-treated cells were completely unreactive.

GROα-induced DNA fragmentation assessed by in solution and in situ 3′ end-labeling of isolated cells and chondrocytes cultured as explants in an intact ECM.

DNA laddering.

The presence of a DNA degradation pattern composed of both classic ladder-like fragments and higher molecular weight DNA was documented by immunodetection of digoxigenin incorporation by terminal deoxynucleotidyl transferase in solution (Figure 2), after agarose gel electrophoresis and Southern transfer.

Figure 2.

Endonuclease activation in growth-related oncogene α (GROα)–treated chondrocytes compared with unstimulated (UN) chondrocytes. GROα treatment of chondrocytes induced a DNA degradation pattern composed of classic ladder-like fragments as well as higher molecular weight DNA, as documented by immunodetection of digoxigenin incorporated by terminal deoxynucleotidyl transferase in solution.

Isolated chondrocytes.

The TUNEL experiment performed on primary OA chondrocytes (Figure 3) indicated the presence of a basal level of apoptosis (20% in primary culture) and a strong induction of apoptosis by GROα (80%), which paralleled the effect of 0.5 μM staurosporine. TUNEL staining was specifically associated with DNA fragmentation, as indicated by the no-enzyme control.

Figure 3.

Quantification of apoptosis in isolated chondrocytes (by TUNEL). Chondrocytes freshly isolated from osteoarthritic cartilage were cultured for 4 days and treated with a, 500 nM growth-related oncogene α (GROα) and b, 0.5 μM staurosporine. c, “No enzyme” control (performed on the staurosporine-treated cells to monitor nonspecific retention of the labeled nucleotide). Open histograms in a and b represent unstimulated cells.

Organ cultures.

A marked increase in the percentage of apoptotic chondrocytes was found only in 1 of 4 GROα-treated cartilage explants, while IL-1β, RANTES, and MCP-1 were ineffective (Figure 4a). This sample was also characterized by an overall remarkable loss of Safranin O staining, suggesting that the effect of GROα is dependent on the coexistence of other typical features of OA, such as the loss of ECM, and thus of survival signals for the cells (Figure 4b, panel 1). On the other hand, no difference in the level of calcium deposits was noted between the 4 samples examined using the alizarin red S method (data not shown).

Figure 4.

Localization of apoptotic cells in 1 of 4 organ cultures (corresponding to the case shown in panel 1 of 4b). a, Quantification of chondrocyte apoptosis in cartilage explants. Full-thickness pieces of tissue were kept in culture for 7 days with either 100 nM growth-related oncogene α (GROα) or control stimuli. Sections (5 μm) of paraformaldehyde-fixed paraffin-embedded tissue were cut and subjected to TUNEL. GROα treatment increased apoptosis, particularly in the cells of the deeper layers of the tissue. Inset, Higher magnification view of the apoptotic cells. The other stimuli (interleukin-1β [IL-1β], monocyte chemoattractant protein 1 [MCP-1], and RANTES) did not induce an increase in the apoptotic rate compared with the unstimulated organ culture (data not shown). b, Requirement of matrix degradation for induction of apoptosis in cartilage explants. GROα (100 nM) was added to the culture medium of full-thickness pieces of cartilage obtained from 4 different osteoarthritis patients. Safranin O staining revealed that the GROα-sensitive sample (panel 1), as for apoptosis induction, also showed severe proteoglycan depletion as compared with the other specimens, none of which were responsive to the chemokine (panels 2–4). c, Colocalization of molecular markers of apoptosis in GROα-stimulated explants. In the deeper layers of the tissue, active caspase 3 and phosphorylated stress-activated protein kinase/c-Jun N-terminal kinase (SAPK/JNK)–positive cells colocalized (arrow). (Original magnification × 125 in a; × 160 in b; × 250 in c; × 500 in inset.)

GROα-induced activation of caspase 3 and phosphorylation of JNK/SAPK.

Isolated chondrocytes.

The percentages of cells expressing active caspase 3 after an overnight incubation were as follows: unstimulated cells 1.64%, cells treated with 100 nM GROα 6.35%, and cells treated with 0.75 μg/ml of anti-Fas 9.3%. In a complementary series of 3 experiments with duplicate wells for each condition, GROα led to a statistically significant increase in caspase 3 activity (PhiPhiLux G1D2 kit), as evaluated based on fluorescence intensity (P = 0.04; n = 6). The median percentage increase in fluorescence in GROα-treated cells compared with unstimulated cells was 52 (25th percentile 34; 75th percentile 88).

Organ cultures.

Whereas a basal level of apoptosis was detected in the upper layer of tissue, apoptosis triggered by GROα was observed in the deeper layers. The colocalization of activated caspase 3 and phosphorylated JNK/SAPK was also noted in this portion of the tissue (Figure 4c).

GROα-induced translocation of phosphatidylserine as evaluated by annexin V labeling.

Chondrocytes are adherent cells, and although single-cell analysis techniques (i.e., flow cytometry) are valuable for studying phenomena that affect only a small fraction of the population, the detachment of the cells from resynthesized ECM proteins potentially introduces bias, such as an enhancement of the apoptotic index (2). Therefore, we also examined the loss of asymmetry in cell membrane phospholipids on chondrocytes kept in adhesion cultures in 96-well plates, exploiting the possibility of quantitating the increase in FITC-labeled annexin V–dependent fluorescence intensity by means of a plate fluorometer.

Addition of 100 nM GROα overnight induced a statistically significant increase in FITC-labeled annexin V signal compared with unstimulated cells (P = 0.01; n = 9) and a median percentage increase of fluorescence compared with unstimulated control of 9% (25th percentile 6; 75th percentile 14), whereas 0.75 μg/ml anti-Fas induced a median percentage increase of 10% (25th percentile 9; 75th percentile 14). Of interest, a notable difference was observed between GROα-stimulated chondrocytes derived from degenerating and regenerating cartilage obtained from the same patient: 7% versus 2% median percentage increase in fluorescence signal in 2 independent experiments with duplicate wells.

If, as reported in the literature, chondrocytes left in culture for 4–6 days re-create an ECM environment of intact cartilage similar to that in vivo, it is conceivable that cells derived from degenerating cartilage produce a scanter ECM that does not have the potential to provide enough “survival signals” to the cells when they are exposed to apoptogenic stimuli (37, 38). This further supports the hypothesis that GROα induces apoptosis in the absence of survival signals from the ECM, as probably happens in cultures of cells derived from degenerating and degraded cartilage.

GROα-induced plasma membrane depolarization without repolarization.

Plasma membrane depolarization is an early and sustained event that precedes the loss of cell volume and takes place in the initial activation phase of the cell death process (41). This phenomenon leads to both the activation of effector caspases and changes in cell volume.

We studied the effect of the addition of chemokines on membrane potential by confocal laser scanning microscopy of live cells cultured on coverslides and loaded with bisoxonol (40). This powerful technique allows visual monitoring of the time course of the membrane potential at the single-cell level, which is particularly useful when only a fraction of a cell population is reactive.

Stimulation with GROα induced a marked and progressive depolarization of a minority of the cells (Figures 5a–c), which was not sensitive to removal of the chemokine and continued up to cell death. Notably, the reactivity of the cells was heterogeneous as to the time kinetics and intensity, which is consistent with the heterogeneity of CXCR2 expression in primary chondrocytes. The more reactive cells quickly (∼30 minutes after addition of GROα) displayed morphologic changes consistent with apoptosis, such as cell shrinkage, blebbing, and fragmentation (Figure 5c, inset).

Figure 5.

Chondrocyte plasma membrane depolarization induced by growth-related oncogene α (GROα) in vitro. GROα (100 nM) induced chondrocytes depolarization and shape modifications consistent with apoptosis in time-lapse confocal microscopy at a, time 0 (bar = 5 μm), b, 10 minutes, and c, 30 minutes. A progressive membrane depolarization in 2 elements paralleled the increase in signal. Inset, A cell showing morphologic signs of apoptosis (bar = 2.5 μm). d, Time course of bisoxonol fluorescence intensity after stimulation with GROα and RANTES (100 nM). The time course of bisoxonol fluorescence intensity after the addition of RANTES indicates hyperpolarization. Values are the mean intensity of pixels pertinent to cell areas within the measured field in the same experiment.

Figure 5d shows the time course of the bisoxonol fluorescence increase due to rapid depolarization induced by GROα stimulation. To examine the specificity of the effect induced by this CXC chemokine, we performed a parallel experiment to test the effect of the addition of RANTES, which binds to CC receptors. Following the addition of RANTES, the cells showed a persistent decrease in signal (Figure 5d), corresponding to a generalized hyperpolarization, which at least within the first 10 minutes was reversible by removal of the chemokine and substitution of fresh medium. Cell morphology was still preserved up to 30 minutes after addition of the cytokine. Figure 5d shows the values obtained from a representative experiment of 4 independent experiments performed.


The role of chemokines in the pathophysiology of OA is now starting to come to light. This is the first study to show the ability of GROα to induce chondrocyte apoptosis.

Chondrocytes are long-lived cells that adhere to the ECM and depend on anchorage for growth and metabolism (13). Indeed, chondrocytes are sensitive to growth factors and cytokines provided that there is a proper network of interactions with the ECM. It has been reported that IL-1 signaling requires proper attachment and proximity of the receptor to focal adhesions, structural links between the ECM and the cytoskeleton, as well as specialized sites of signal transduction (42). We demonstrated here that the functional effects induced by chemokines are secondary to a complex and interdependent series of signal transduction pathways that start at the cell membrane, since GROα-induced apoptosis occurred only when the cells were left for an adequate period of time to recover and reestablish a proper network of interactions with ECM proteins (37, 38).

GROα-induced apoptosis is likely to occur through so-called anoikis, that is, death dependent on the loss of normal cell–substratum contact (43). It has been reported for epithelial cells (44) and, recently, for chondrocytes (45) that α5β1 integrin bound to fibronectin protects from apoptosis. It is possible that chondrocytes undergo apoptosis when the loss of such a contact occurs as the result of the induction of matrix-degrading enzymes (mainly, stromelysin 1, which is able to degrade fibronectin) upon chemokine stimulation (25). In keeping with such observations is the antiapoptotic effect of hyaluronan on anti-Fas–treated chondrocytes, as recently reported by our group (46), and the notion that the apoptotic index is higher in the superficial layer of OA cartilage (11), where matrix-degrading activity is enhanced compared with deeper areas (6). Furthermore, it has also been reported that apoptosis is highly increased in chondrocytes in suspension compared with whole tissue (2), possibly because detachment induced strong activation of caspases 8 and 3 (47). The results of our study further support this hypothesis, given the comparison between the estimates of the apoptotic index of adherent (annexin V–labeled; May-Grünwald-Giemsa–stained) versus suspended (TUNEL-labeled) cells.

An issue that deserves further investigation is the association between apoptosis and cell proliferation. GROα was first described as the “growth-related oncogene.” It belongs to the ELR chemokines. Chemokines are a large family of molecules containing a cluster that shares the ELR motif, which is associated with the induction of cell proliferation (48). A close link between cell proliferation and apoptosis has been reported (49), and it is probably dependent on a broad spectrum of responses to oxidative stress, ranging from a mitogenic response to growth arrest to apoptosis or necrosis, depending on the stress level encountered.

Apoptosis encompasses multiple characteristic morphologic and biochemical features, including chromatin aggregation, nuclear and cytoplasmic condensation, DNA fragmentation, alteration in membrane asymmetry, and activation of apoptotic caspases (50, 51). Furthermore, it is a dynamic process in which a characteristic morphologic or biochemical event used in an assay as a specific marker of apoptosis may be observed over a limited period of time (50). To strengthen the sensitivity and the specificity of apoptosis detection in chondrocytes, we therefore used multiple technical approaches.

The most specific assays appear to be those based on the detection of DNA strand breaks (50). The apoptotic effect of GROα was noted not only on isolated chondrocytes, but also on cells cultured within an intact ECM and, to our knowledge, this is the first report of apoptosis induced by a cytokine upon culture of human cartilage explants. When set within an intact ECM, chondrocytes are protected from apoptosis by virtue of survival signals received through the interaction of integrins and ECM proteins. Nevertheless, this model more closely resembles in vivo conditions, and furthermore, it may also allow the identification of differential effects across the various layers of cartilage, in keeping with a functional and phenotypic heterogeneity of the tissue. By using this model, apoptosis triggered by GROα was observed in the deeper layers, along with the coordinated activation of caspase 3 and the phosphorylation of JNK/SAPK, the molecular markers of the phenomenon. Safranin O labeling experiments suggest that the threshold for apoptosis is probably related to cross-talk between signaling pathways triggered by soluble factors and signals received from the ECM.

Of particular interest is the notion that GROα stimulation of chondrocytes also leads to a rapid and irreversible depolarization of the plasma membrane. It has recently been reported that the movement of intracellular monovalent cations plays a critical role in many cellular alterations related to apoptosis. The loss of intracellular potassium and sodium contributes to cell shrinkage and intracellular activation of caspases and endonucleases (41). This phenomenon occurs very rapidly after application of the apoptotic stimulus, when visualized at the single-cell level (52). Results of our confocal microscopy experiments clearly indicated that the cells are heterogeneous with respect to their response to GROα. Given the complexity of the functional behavior of adherent cells (42), it is likely that the threshold of GROα-induced apoptosis is determined not only by the presence of the specific receptor, but also by a peculiar repertoire of adhesion molecules known to regulate this phenomenon (44) and by a defined commitment to cell proliferation. The depolarizing effect of GROα is rapid and sustained and cannot be reversed by removal of the chemokine, as has been shown to occur during anti-Fas–induced apoptosis in Jurkat cells (41) as the result of a blockage of the Na+/K+-ATPase or of the inhibition of the voltage-activated potassium channels. Many recent reports strongly suggest that cellular depolarization could act as a checkpoint in the activation of apoptosis (53).

Our group of investigators has provided the first demonstration of the presence of chemokine receptors on chondrocytes (25), but the downstream signaling pathways remain to be clarified. However, pertussis toxin sensitivity and membrane potential experiments have suggested that chemokine receptors in chondrocytes are of the ion channel–linked G protein–coupled type (54) and that receptors for RANTES and GROα are associated with anion and cation channels, respectively, given the opposite sign of potentiometric effects. This hypothesis is also consistent with an antagonism observed in the experiments performed with first passage chondrocytes, with GROα inducing chondrocyte apoptosis and RANTES resulting in no morphologically evident effect compared with unstimulated controls. It is also likely that RANTES exerts an antiapoptotic action, by virtue of its hyperpolarizing effect on the cells, as has been reported for many members of the Bcl-2 protein family (55).

The ability of GROα to induce chondrocyte apoptosis is a finding of particular relevance in light of the availability of this chemokine in the joint space. We and other investigators have previously reported that chondrocytes constitutively express and release GROα (23, 24). Furthermore, GROα is readily up-regulated by inflammatory cytokines. GROα is also one of the most abundantly secreted chemokines by chondrocytes as well as by several other resident or immigrating cells within the joint, such as monocytes (56) and fibroblasts (57) of the synovial tissue, mononuclear cells and neutrophils of the synovial fluid (57), and even osteoblasts (58) and stromal cells (59) of the subchondral bone, particularly in inflamed joints.


The authors wish to express their gratitude to Dr. Ian Clark-Lewis (Biomedical Research Center and Department of Biochemistry, University of British Columbia, Vancouver, British Columbia, Canada) for his generous gift of synthetic chemokines, the Centro Interfacoltà Misure (University of Parma, Parma, Italy) for use of the Molecular Dynamics confocal laser scanning microscope, and Ms Graziella Salmi and Patrizia Rappini for editorial assistance.