Osteoarthritis (OA) is the most common degenerative disease of human articular cartilage, especially in the population aged over 65 years.1 It is characterised by extracellular matrix damage and an important loss in tissue cellularity.2-4Apoptosis, or programmed cell death, is a physiological process for maintaining homeostasis in both embryogenic and adult tissue but may also have a role in diseases involving articular cartilage degeneration, such as rheumatoid arthritis (RA)5-8 and OA.9-12 Apoptosis can be distinguished morphologically and biologically from necrosis.1314 The characteristic morphological feature of apoptosis is cell shrinkage with preservation of organelles. The chromatin appears compacted in homogeneous dense masses in contact with the nuclear membrane. The nucleus may also break up, and the cell emits material that often contains pyknotic nuclear fragments. This material tends to break off and become apoptotic bodies.15 Biochemically, fragmentation of DNA into regular fragments which are multiples of approximately 200 bases pairs, owing to specific cleavage between nucleosomes, seems to be specific for apoptosis. Once started, apoptosis proceeds rapidly. Cells undergoing apoptosis may completely disappear in minutes to hours,16but in the cartilage the duration of the apoptosis of chondrocytes is unknown. Moreover, removal of apoptotic bodies does not induce an inflammatory response, which is a crucial feature of apoptosis.
In adult articular cartilage, cell loss increases with aging. Recently, Adams and Horton found that chondrocyte apoptosis increases with age in the articular cartilage of adult mice and rats.17Moreover, others studies have shown that apoptosis is greater in human OA cartilage than in normal cartilage.10-12 On the other hand, among normal human cells, chondrocytes are known to produce inducible nitric oxide synthase (iNOS) in response to interleukin 1β (IL1β),1718 which is a well known proinflammatory cytokine in OA. The iNOS catalyses formation of nitric oxide,19 which depresses the synthesis of cartilage proteoglycan and type II collagen,20 damages DNA strands, and triggers apoptosis21 in normal human articular chondrocytes.22-24 Articular chondrocytes break down their extracellular matrix at increased rates while synthesising new matrix molecules at reduced rates, in response to IL1β.
The purpose of this report is, firstly, to compare the rate of apoptosis in normal and OA human articular chondrocytes both in vivo and in vitro and examine a possible correlation with patients' age and some characteristics of their pathology. Secondly, this paper aims at examining the effect of human recombinant interleukin 1β (hrIL1β) on human cultured chondrocytes.
Patients and methods
PATIENTS AND CARTILAGE SAMPLING
Samples of OA human cartilage from the femoral head were obtained during surgery from 14 patients (median age 63 years, range 37–88) with clinical and radiological features of OA according to the criteria of American College of Rheumatology (formerly, the American Rheumatism Association).25 Table 1 summarises the relevant clinical, radiographic, and biological features of these patients. The radiological features were graded 4 or severe according to the Kellgren and Lawrence grading system of OA.
Samples of normal human cartilage from the femoral head were obtained during surgery from four patients (median age 77 years, range 67–87) with osteoporotic femoral neck fractures. None of the subjects had a clinical history of inflammatory or non-inflammatory joint disease, or chronic systemic inflammatory disease.
Samples of normal and OA cartilage were removed from the anterior, posterior, lateral, and medial aspects of the perifoveal, central, and peripheral areas of each femoral head with a circular trephine (diameter 4 mm, depth 1 cm). Half the samples were processed by a classic histological assessment and frozen cartilage sections were prepared from the other half. Samples were immediately transported to the histopathology laboratory on ice.
Femoral heads were then immersed in Iscove modified Dulbecco's medium (IMDM) (Gibco-BRL, Scotland) and transported to our laboratory within the next four hours.
The remaining cartilage from each femoral head used to isolate chondrocytes was removed carefully and pooled. Tissue was minced into small pieces (1–3 mm3), washed in IMDM, and digested with 2 mg/ml of clostridial collagenase (Boehringer Mannheim, Germany) overnight at 4°C. Thereafter, the cell suspension was collected by centrifugation at 1000 × g for 15 minutes. The pellet was suspended in complete culture medium (CCM) (IMDM supplemented with 10% (v/v) fetal calf serum (Gibco-BRL, Germany)). Cell viability was quantified by trypan blue exclusion assay and was always >90%. The freshly isolated chondrocytes were seeded into Lab-Tek chambers (Nunc Inc, Naperville, IL) at 30 000 cells/cm2 and incubated in CCM for 24 hours at 37°C in an humidified air atmosphere containing 5% (v/v) CO2 to allow them to adhere. The adherent isolated chondrocytes, in primary culture, were simultaneously analysed using the “In situ cell death detection, fluorescein” kit (Boehringer Mannheim, Germany) and Annexin-V fluos (Boehringer Mannheim, Germany) (see below). In the supernatant we found no non-binding apoptotic chondrocytes using an apoptotic marker.
INDUCTION OF CHONDROCYTE APOPTOSIS BY HRIL1β
The resting cell suspension was cultured in 25 cm2culture flasks (Corning, Cambridge, USA) for 24 hours. Once confluent, chondrocytes were harvested with 0.2% (w/v) trypsin solution (Boehringer Mannheim, Germany), centrifuged at 1000 ×g for 10 minutes, and resuspended in fresh CCM. Chondrocytes, at the first passage, were seeded at a density of 20 000 cells/cm2 in Lab-Tek chambers and cultured in CCM for 24 hours at 37°C, at which time various concentrations ranging from 1 to 10 ng/ml of hrIL1β (Sigma Chemical, St Louis) were added to each well. Apoptosis was detected after 24 hours with Annexin-V-fluos (see below).
IN VITRO DETECTION OF APOPTOTIC CHONDROCYTES
The DNA cleavage was assessed by the terminal deoxynucleotidyl transferase mediated dUTP nick end labelling (TUNEL) as described by Garvieli et al26; consequently the “In situ cell death detection, fluorescein” kit was used. Cells were gently washed with phosphate buffered saline (PBS) and fixed with 250 μl of 4% (w/v) paraformaldehyde prepared freshly in PBS pH 7.4 for 30 minutes at room temperature. After washing twice in PBS, cells were incubated in 200 μl of permeabilisation solution (0.1% (v/v) Triton X-100 in 0.1% (w/v) sodium citrate) for two minutes on ice. Cells were again washed in PBS before incubation with 55 μl of TUNEL reagents (terminal deoxynucleotide transferase and fluorescein labelled nucleotide mixture) or 55 μl of LABEL (without terminal transferase) as negative control for two hours at 37°C in a humidified chamber.
Annexin-V is a Ca2+ dependent phospholipid binding protein with high affinity for phosphatidylserine (PS), and binds to cells with exposed PS.26 In the early stage of apoptosis, PS is translocated from the inner part of the plasma membrane to the outer layer, becoming exposed at the external surface of the cell. Annexin-V can hence be used as a sensitive probe for PS exposure upon the outer leaflet of the cell membrane and can therefore detect apoptotic cells.27-29 Cells were washed with PBS and incubated with 150 μl of Annexin-V in Hepes buffer (10 mM Hepes/NaOH pH 7.4; 140 mM NaCl; 5 mM CaCl2) containing propidium iodide (50 μg/ml) for 15 minutes at room temperature. After three rinses with PBS, slides were mounted and analysed under a fluorescent microscope (Zeiss Axioskop D-7082) (magnification ×200).
The total number of chondrocytes and the number of chondrocytes staining positively were quantified in 10 microscopic fields, which were randomly chosen, from 100 cells from each patient. The final result was expressed as the percentage of positive chondrocytes.
DETECTION OF APOPTOSIS IN FROZEN CARTILAGE SECTIONS
Samples of articular cartilage were collected from each femoral head, snap frozen in liquid nitrogen, and stored in cryovials at −30°C. The samples were cut in 5 μm thick cryostat sections and mounted on Superfrost slides (Fisher Scientific, Pittsburgh), then fixed with acetone for 10 minutes and stored at −20°C until further use. Apoptotic cells in the tissue samples were identified, as described above, by end labelling of nuclei with the TUNEL method and by membrane labelling with Annexin-V-fluos.
Histological evaluation was performed on sagittal sections of cartilage from perifoveal, central, and peripheral areas of each femoral heads. Specimens collected from each femoral head were fixed in 10% (v/v) formaldehyde, processed for paraffin embedding, and aligned so that the sections were perpendicular to the surface. Samples, 5 μm thick, were then cut, mounted on slides, and stained with haematoxylin and eosin and safranin O fast green. Staining was performed at room temperature.
Data are expressed as means (SEM). Statistical comparisons were assessed by Student's unpaired t test and Kruskal-Wallis test for comparing three OA groups. Values of p<0.05 were considered significant.
Clinical, radiographic, and biological findings in 14 patients with osteoarthritis (OA)
IN VITRO DETECTION OF APOPTOSIS IN HUMAN ARTICULAR CHONDROCYTES BY IMMUNOFLUORESCENCE
To evaluate apoptotic cells, cartilage was harvested and digested with collagenase. The adherent freshly isolated chondrocytes were stained by the TUNEL reaction and Annexin-V and analysed under fluorescent microscopy after 24 hours' culture.
On the one hand, the TUNEL reaction mixture (fig 1A) showed apoptotic OA chondrocytes distinguished by a more or less marked labelling of the nucleus; labelling was more marked for OA than normal chondrocytes. No labelling was detected when terminal deoxynucleotidyl transferase was omitted from the reaction (negative control). On the other hand, Annexin-V showed a specific membrane labelling, which was also more marked in OA (fig 1C) than in normal chondrocytes.
Detection of apoptosis by immunofluorescence on isolated OA chondrocytes by collagenase digestion (A and C) and in situ on fresh frozen OA cartilage (B and D). (A and B) Chondrocytes were fixed, permeabilised, and processed by the TUNEL reaction mixture. (C and D) Samples were treated with Annexin-V-fluos for 15 minutes and analysed immediately. An example of an apoptotic cell is identified by the arrows. (Original magnification A, B, C ×200; D ×100).
For the TUNEL reaction the apoptotic rates for normal and OA chondrocytes were 5.2 (2.1)% and 20.7 (5.8)%, respectively (p<0.01). For Annexin-V the results for normal and OA chondrocytes were 3.9 (2.3)% and 19.1 (4.3)%, respectively (p<0.01).
The quantitative study of normal and OA chondrocytes, and the two types of label used, indicated (fig 2) a significant difference in the percentage of apoptotic chondrocytes between normal (4–5%) and OA chondrocytes (19–21%). The percentage of stained nuclei was almost the same in OA chondrocytes with the two different markers.
Comparative study of normal and OA chondrocytes using the TUNEL reaction and Annexin-V labelling in chondrocytes isolated by collagenase digestion. The ratio of apoptotic cells/total cells was calculated as a percentage. Values are means (SEM) from four experiments for normal chondrocytes and 11 experiments for OA chondrocytes (p<0.01).
DETECTION OF APOPTOSIS IN FROZEN CARTILAGE SECTIONS
To verify the data obtained with freshly isolated chondrocytes, we used OA and normal cryopreserved cartilage sections for in situ analysis of apoptosis by the TUNEL reaction and the Annexin-V method.
Figure 1B, for the TUNEL test, shows a nuclei labelling of dystrophic chondrocytes located in their lacunae. The nuclei from OA cartilage showed more labelling than those of normal cartilage. Figure 1D shows membrane labelling by Annexin-V for normal and OA cartilage. Again, membrane labelling in OA cartilage was greater than in normal cartilage.
For the TUNEL reaction the OA cartilage contained 18.1 (2.8)% apoptotic chondrocytes and the normal cartilage contained only 1.5 (1.2)% (p<0.01). For Annexin-V the OA cartilage contained 19 (3.2)% apoptotic chondrocytes, the normal cartilage contained only 2.4 (1.2)% (p<0.01).
The quantitative study of normal and OA cartilage and the two patterns of labelling indicate (fig 3) a significant difference in apoptotic rate between normal and OA cartilage—apoptotic chondrocytes representing respectively 2–3% and 18–19% of the chondrocyte population.
Comparative study of normal and OA cartilage using the TUNEL reaction and Annexin-V labelling on frozen cartilage sections. The ratio of apoptotic cells/total cells was calculated as a percentage. Data are representative results from four experiments with normal cartilage and 13 experiments with OA cartilage (p<0.01).
INDUCTION OF CHONDROCYTE APOPTOSIS BY HRIL1β
To determine whether chondrocyte apoptosis is inducible by hrIL1β and at which concentration, seven confluent chondrocyte cultures from the first passage were exposed to various concentrations (1 ng/ml, 3 ng/ml, 10 ng/ml) of hrIL1β for 24 hours.
The percentage of chondrocytes undergoing apoptosis in response to hrIL1β treatment increased in a dose dependent manner (table 2). In OA chondrocyte cultures the maximum induction of apoptotic cells was 29.5 (6.8)% at 10 ng/ml, in comparison with a rate of 5.0 (2.2)% apoptotic cells in non-treated cultures (fig 4) (p<0.01).
Induction of apoptosis in OA chondrocytes by human recombinant interleukin 1β (hrIL1β). (A) Untreated control OA human chondrocytes. (B) OA human articular chondrocytes in primary culture stimulated with 10 ng/ml of hrIL1β for 24 hours and processed with Annexin-V-fluos for 15 minutes. Arrows identify examples of apoptotic cells. (Original magnification ×200.)
In normal chondrocyte cultures, hrIL1β induced apoptosis in a dose dependent manner (table 2) but at a lower rate than OA chondrocyte cultures. The maximum induction of apoptotic cells was 11.4 (2.3)% at 10 ng/ml and only 2.9 (1.1)% in non-treated cultures (p<0.01).
Figure 5 indicates a correlation between patients with OA aged less than 56, who were carriers of a secondary OA, and apoptosis as detected by the TUNEL reaction. Moreover, in a second group including patients older than 62, and representing essentially primary OA, a gradual increase of apoptotic rate with age was seen. In frozen cartilage sections there was no evidence of a correlation between the apoptotic rate and the age of the patients.
Percentage of apoptotic chondrocytes as a function of aging in the patients with OA and normal patients. Cells were isolated by collagenase digestion, and apoptosis in these isolated cells was analysed by TUNEL reaction labelling. Values are means (SEM) for 11 patients with OA and four normal patients.
Four patients with OA (Nos 3, 5, 13, 14—see table 1) had an apoptotic rate equal to or greater than 25%. Three patients (Nos 3, 5, 13) were overweight and two (Nos 3, 13) had a body mass index greater than 29. Three patients (Nos 5, 13, 14) had generalised OA.
Effect of human recombinant interleukin1β (hrIL1β) on apoptosis of osteoarthritis (OA) and normal chondrocytes
The loss of articular cartilage in OA is the central event. Moreover, there is an important decrease in chondrocyte number owing to increased cell death. Apoptosis is an important basic biological phenomenon in the regulation of the cell death. Our aim was to elucidate whether OA chondrocyte death might, in part, be due to apoptosis.
This study showed that the apoptotic rate was more important for OA chondrocytes (19–21%) than for normal chondrocytes (4–5%) isolated by collagenase digestion. The apoptotic rate was also more important in OA cartilage (18–19%) than in normal cartilage (2–3%) in cryopreserved cartilage sections. Therefore, in vitro apoptotic rates of adherent chondrocytes were almost the same as in situ apoptotic rates in cartilage sections. Thus, the possibility that the tissue digestion isolation procedure may cause DNA damage, and so create an artefactual apoptosis, can be ruled out.
In addition, our study showed that the extent of apoptosis by the Annexin-V staining method correlates well with that by the TUNEL method, though Aizawa et al have reported that the TUNEL method easily detects apoptotic cells but tends to overestimate their real number.30 Annexin-V is thought to identify cells at an earlier stage of apoptosis than assays based on DNA fragmentation because externalisation of PS seems to occur earlier than the nuclear changes associated with apoptosis.
As far as we know, our study is the first comparing the apoptotic rate, using two labels and two methods—in situ and in vitro. The data obtained using both labels are comparable in cartilage sections and chondrocyte cultures.
Moreover, in the analysed OA cartilage samples, apoptotic cells were mainly detected in the mid-zone and in the superficial area when this zone was still present (data not shown).
Three previous studies91012 have shown the presence of apoptosis in both normal and OA human articular cartilage. Blancoet al showed in vitro that OA cartilage displayed a higher rate of apoptotic chondrocytes (51%) than normal cartilage (11%).10 These data are not consistent with our results, but the authors speculated that their enzymatic digestion process might have accelerated apoptosis in OA cells. In situ, however, the apoptotic rate in OA cartilage was low (6%) in comparison with our results (19%). Moreover, Kouri et al showed in OA cartilage tissue an apoptotic rate varying from 30 to 88% for four patients with OA.9 This percentage is higher than we obtained, but with a limited number of patients with OA. Hashimotoet al, in an OA knee cartilage study, showed similar results to our rate by the TUNEL reaction.12 More recently, the importance of the potential role of apoptosis in the pathogenesis of articular loss in both OA and RA has been reported by Hashimoto et al.31 The authors showed that OA chondrocytes are the target for two independent pathways in the induction of apoptosis. One is associated with synovial inflammation and the pathway is Fas mediated, the other without synovial inflammation is NO dependent. In addition, NO had previously been shown to induce apoptosis when produced by IL1β stimulated chondrocytes.21 Our study also shows that hrIL1β increases apoptosis in both OA and normal chondrocytes in a dose dependent manner. Surprisingly, the level of apoptotic cells dramatically decreased in culture and this was seen in both normal and OA cultures. The in vitro detection of apoptosis was carried out on cells released using collagenase and incubated for only 24 hours, whereas the study with IL1β was performed using the first passage. One can suppose that apoptotic cells arising from the OA and normal cartilage died during the amplification pathway preceding the first passage. Moreover, OA chondrocytes seem more sensitive to IL1β than normal chondrocytes as apoptosis was increased fivefold in OA cultures, and by threefold in normal cultures. These observations can be correlated with the study of Martel-Pelletier et al, who reported that the levels of IL1β receptors expressed in OA chondrocytes are higher than in normal chondrocytes.32
This phenomenon seems to be the consequence of either a typical feature of chondrocytes or activation of chondrocytes by proinflammatory mediators arising from the synovial fluid, or both. In the OA articular cartilage, as in the growth plate cartilage, the OA chondrocytes may be induced to undergo hypertrophic differentiation by external factors. The expression of this hypertrophic phenotype may induce chondrocyte apoptosis and, therefore, the loss of articular cartilage.33
An important question is to know if the apoptotic population is present in OA at an early stage or, particularly, in late stage OA. In human OA, at an early stage, apoptotic cells are found in the OA fibrillated cartilage and in clusters of proliferating chondrocytes. In experimental OA, induced in rabbits by ligament transection, apoptotic cells are present in the early phases of OA.11 Moreover, Hashimoto et al describes an increasing relation between chondrocyte apoptosis and OA grade.12Inducers, such as Fas ligand, or survival factors, such as IL1β and tumour necrosis factor α, might induce apoptosis by the NO pathway.2234 In late stage OA, in our study, the average apoptotic chondrocyte rate is surprisingly similar and remains at a relatively constant level whatever the clinical and radiological findings may be. The highest rates are not correlated with the greatest cartilage losses. In advanced OA it seems that extracellular matrix degradation by metalloproteinases might lead to apoptosis of the remaining cells.
The clinical and radiographical findings do not allow us to establish a significant correlation with apoptosis. Additionally, it is not possible to establish a particular correlation with apoptotic rate for the two subsets—primary and secondary OA. This is due to the complex classification of OA and the small number of samples (n=14). Furthermore, the lack of correlation may perhaps be influenced by the grade of OA severity (grade 4 of the Kellgren and Lawrence system). However, two patients had an apoptotic level higher than 30% (Nos 5 and 14) and two others a level equal to or greater than 25% (Nos 3 and 13). These patients had one or more of the following characteristics: overweight, obesity, and generalised OA.
The relation between increased body weight and hip OA is not as strong as with knee OA. Some studies show obesity is a risk factor for hip OA among men and women, as indicated by the presence of Heberden's nodes.3536 In men, OA predisposes to coxarthrosis.37 Felson and Zhang suggest hypothetical abnormal levels of certain hormones or growth factors generated by excess adipose tissue, but without evidence.38 Generalised OA raises the same problems. Unfortunately, we have no evidence for an excessive quantity of NO and other cytokines. Clearly, more work needs to be done in this important area.
In conclusion, the results of our study suggest that upregulated apoptosis is probably present in the late stage of OA disease. However, apoptosis may play a significant part in the physiopathology of articular cartilage and in the course of OA and may be a new and attractive target for OA treatment.
The authors are deeply grateful to Dr Martine Julien for preparation of cryopreserved frozen cartilage samples and for performing histological studies of the synovium and cartilage samples at Robert Boulin Hospital, Libourne. Drs P Tramond and JL Chatelan are acknowledged for the sampling of osteoarthritic and normal femoral heads at the Department of Orthopaedics, Robert Boulin Hospital, Libourne.
Osteoarthritis (OA)3 is a degenerative disease characterized by several structural changes including the degradation of cartilage matrix (1). In normal mature cartilage, chondrocytes synthesize sufficient amounts of macromolecules to maintain the integrity of the matrix, whereas in response to OA changes, they do not synthesize sufficient matrix to repair significant tissue defects (2). The chondrocyte is the only cell type found in mature cartilage and is responsible for the synthesis and the maintenance of the extracellular matrix. Therefore, factors that limit the adequate cartilage formation and repair may include the lack of chondrocytes in the tissue. There is a well-documented decline in the number of articular chondrocytes and an increase in the number of empty lacunae with age (3). Although some chondrocytes proliferate during OA, the chondrocytes do not migrate through the matrix to enter the site of tissue defect (2). Furthermore, there is increasing evidence suggesting that chondrocyte death may contribute to the progression of OA. Several studies have shown that OA cartilage has a higher number of apoptotic chondrocytes than does normal cartilage in animal models (4) and humans (5, 6). The presence of increased numbers of apoptotic cells may correlate with the extent of cartilage matrix loss (5).
The production of NO may represent an important component in the pathogenesis of OA. NO is produced in large amounts by chondrocytes upon proinflammatory cytokine stimulation (7). High levels of nitrite/nitrate have been found in the synovial fluid and serum of arthritis patients (8). Both mRNA and protein for inducible NO synthase (iNOS), the enzyme responsible for NO production, have also been detected in synovial tissue from OA patients (9). Besides causing degradation (10) or inhibiting the synthesis of cartilage matrix (11), NO may also induce chondrocyte apoptosis. We have previously reported that the systemic administration of iNOS inhibitor, N-iminoethyl-l-lysine (l-NIL), in experimentally induced OA in dogs has resulted in a reduction of articular cartilage damage and the levels of cell apoptosis and caspase-3, as determined immunohistochemically (12, 13). In addition, there is a significant correlation between the level of nitrite production and the prevalence of apoptotic cells in cartilage tissue during experimentally induced OA in rabbits (14). In fact, NO generated from sodium nitroprusside (SNP) has been shown to induce apoptosis in cultured human articular chondrocytes (15). However, the mechanisms regulating the chondrocyte death have not been well characterized.
This study focused on the characterization of the signaling cascade during SNP-generated NO-induced cell death in human OA chondrocytes. We evaluated DNA fragmentation and cell viability to quantify the SNP-induced cell death in human OA chondrocyte culture and used various pharmacological inhibitors to study the different intracellular signaling pathways involved in this phenomenon. Caspase-3 activity and Bcl-2 level in the chondrocytes were also determined.
Materials and Methods
SNP, PGE2, pyrrolidine dithiocarbanate (PDTC), and indomethacin were purchased from Sigma-Aldrich Canada (Oakville, Ontario, Canada). Z-Asp(OCH3)-Glu(OCH3)-Val-Asp(OCH3)-CH2F (Z-DEVD-FMK), Z-Leu-Glu(OCH3)-His-Asp(OCH3)-CH2F (Z-LEHD-FMK), PD98059, and SB202190 were the products of Calbiochem-Novabiochem (San Diego, CA). SN-50 and NS-398 were obtained from Biomol (Plymouth Meeting, PA) and Cayman Chemical (Ann Arbor, MI), respectively. All other chemicals were of the analytical grade of purity and commercially available.
Specimen selection and chondrocyte cultures
Cartilage specimens were obtained from 12 patients with OA (eight females, four males, aged 67 ± 9 years, mean ± SD) undergoing total knee joint replacement. Diagnosis was established according to the American College of Rheumatology criteria (16). The OA cartilage (femoral condyles and tibial plateaus), obtained under aseptic conditions, was carefully dissected from the underlying bone in each specimen. Approximately 2–5 g of cartilage were obtained from each dissected specimen. Gross morphology of the cartilage specimens used in this study was classified as moderate to severe OA.
Specimens were then dissected and washed in PBS containing antibiotics (500 U/ml penicillin, 500 μg/ml streptomycin) and again extensively washed in PBS. Chondrocytes were released from articular cartilage by sequential enzymatic digestion as described (17): 1 h with 2 mg/ml pronase (Boehringer Mannheim Canada, Laval, Quebec, Canada) followed by 18 h with 1 mg/ml collagenase (type IV; Sigma-Aldrich Canada) at 37°C in DMEM (Life Technologies, Canadian Life Technologies, Burlington, Ontario, Canada) with 10% heat-inactivated FCS (Life Technologies) and antibiotics (100 U/ml penicillin, 100 μg/ml streptomycin). The digested tissue was centrifuged and the pellet was washed. The isolated chondrocytes were seeded at high density in tissue culture flasks (no. 1-56502; Nunc, Roskilde, Denmark) and cultured in DMEM supplemented with 10% FCS (10% FCS-DMEM) and antibiotics at 37°C in a humidified atmosphere of 5% CO2 and 95% air. At confluence, the cells were detached and passaged once, then seeded at 1 × 104 and 3 × 105 cells in a 96-well plate (Falcon 3072; Becton Dickinson, Franklin Lakes, NJ) and a 12-well plate (Costar 3513; Corning, Corning, NY), respectively. The cells were allowed to grow until confluence and then used in the following experiments.
Experimental culture conditions
SNP was used as a generator of NO. For the experiments on the SNP dose response, chondrocytes were treated with various concentrations of SNP for 24 h in 10% FCS-DMEM. To determine the time course of the response, cells were incubated with SNP (1 and 2 mM) for the indicated period (5–72 h).
To explore the signaling cascade on SNP-induced cell death, we used Z-DEVD-FMK (100 μM), Z-LEHD-FMK (100 μM), PD98059 (50 μM), SB202190 (10 μM), SN-50 (50 μg/ml), PDTC (10 μM), NS-398 (50 μM), and indomethacin (100 μg/ml). Chondrocytes were preincubated with each inhibitor for 2 h, followed by the coincubation of SNP (1 and 2 mM) for 24 h. Preliminary results confirmed that the effects of these inhibitors were dose dependent, and each inhibitor at the indicated concentration induced maximal response in our study (data not shown).
To examine the role of PGE2 on SNP-mediated chondrocyte death, cells were first pretreated with various concentrations (1–1000 ng/ml) of PGE2 for 48 h and then incubated with SNP (1 and 2 mM) in the absence of PGE2 for an additional 24 h.
DNA fragmentation ELISA
To assay DNA fragmentation ELISA, chondrocytes were seeded at 1.0 × 104 per well in a 96-well culture plate in 100 μl 10% FCS-DMEM and cultured until confluence. Cells were then synchronized by 0.5% FCS-DMEM for 1 day. To label DNA, the medium was replaced with 10% FCS-DMEM and 10 μM 5-bromo-2′-deoxyuridine was added to each well and incubated for 20 h. Following 5-bromo-2′-deoxyuridine incorporation, the cells were cultured in 10% FCS-DMEM according to the experimental culture conditions as mentioned above. After the incubation, the cells were lysed in 200 μl incubation buffer (Roche Diagnostics, Laval, Quebec, Canada). Labeled DNA fragments were separated from labeled intact genomic DNA by centrifugation (10 min at 1000 × g). Soluble DNA fragments present in the supernatant were quantified using the Cellular DNA Fragmentation ELISA (Roche Diagnostics) according to the manufacturer’s instructions. Results were expressed as OD units per 104 adherent cells.
Cell viability in a 96-well culture plate (see above) was evaluated using a modification of the MTT assay (18). For the colorimetric MTT assay, 10 μl MTT, a soluble tetrazolium salt solution (5 mg/ml in PBS), was added to the wells, containing 100 μl medium, and the plate was incubated for an additional 4 h. Thereafter, 100 μl solubilization solution (0.04 M HCl-isopropanol) was added to dissolve the water-insoluble formazan salt. Quantitation was then conducted with an ELISA reader at 590 nm. Results were expressed as OD units per 104 adherent cells.
Measurement of caspase-3 activity and Bcl-2 protein level
Chondrocytes were seeded at 3 × 105 cells per well in a 12-well culture plate in 2 ml 10% FCS-DMEM. After confluence, cells were treated with various concentrations of SNP for the indicated times.
For measurement of caspase-3 activity, adherent cells were washed with ice-cold PBS and resuspended in 100 μl lysis buffer (caspase-3 fluorometric assay; R&D Systems, Minneapolis, MN). The cell suspension was lysed by two cycles of freezing and thawing. Cell lysates (10 μg of total protein) were added to reaction mixtures containing 25 μM synthetic substrate Z-Asp-Glu-Val-Asp-AFC (Z-DEVD-AFC; Calbiochem-Novabiochem), 100 mM HEPES, 10% sucrose, 10 mM DTT, 1 mM PMSF, 10 μg/ml pepstatin, and 10 μg/ml leupeptin, pH 7.5, in a total volume of 100 μl. Caspase-3 activity was measured by the release of 7-amino-4-trifluoromethyl-coumarin (AFC) from the synthetic substrate Z-DEVD-AFC using a microplate spectrofluorometer in the kinetic mode with excitation and emission wavelengths of 400 and 505 nm, respectively. The reactions were inhibited by the addition of 100 μM Z-DEVD-FMK. The enzymatic activity was expressed in units per milligram of total protein, with 1 U corresponding to the amount of enzyme required to release 1 nmol AFC per min at 37°C. Protein content was determined with the bicinchoninic acid protein assay (Pierce, Rockford, IL).
To assay the Bcl-2 protein level, adherent cells were washed with ice-cold PBS and resuspended in 100 μl 50 mM Tris, containing 5 mM EDTA, 0.2 mM PMSF, 1 μg/ml pepstatin, and 0.5 μg/ml leupeptin, pH 7.4. Ag Extraction Agent (20 μl; Oncogene Research Products, Cambridge, MA) was added, and the cell suspension was incubated on ice for 30 min to lyse the cells. Bcl-2 level in the cell lysate was assayed using Bcl-2 ELISA (Oncogene Research Products) according to the manufacturer’s directions. The level of Bcl-2 was expressed in units per milligram of total protein, in which 1 U corresponded to the Bcl-2 protein level in 5.6 × 104 cells of HL60.
Western immunoblots for cyclooxygenase (COX)-2
Chondrocytes were seeded at 3 × 105 cells per well in a 12-well culture plate in 2 ml 10% FCS-DMEM and cultured until confluence. Cells were treated with various concentrations of SNP for 24 h. After this, the adherent cells were washed in ice-cold PBS once and cells were lysed in RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 1 mM PMSF, 10 μg/ml each of aprotinin, leupeptin, and pepstatin, 1% Nonidet P-40, 1 mM sodium orthovanadate, and 1 mM NaF). The cell lysate was boiled for 5 min in 20 μl lysis buffer (1% SDS, 10 mM Tris, pH 7.4) and centrifuged for 5 min. The supernatant (10 μg protein) was subjected to SDS-PAGE through 9% gels (final concentration of acrylamide) under reducing conditions and transferred onto nitrocellulose membranes (Amersham, Oakville, Canada). After blocking with Superblock blocking buffer in TBS (20 mM Tris-HCl, 150 mM NaCl, pH 7.5) and washing, the membranes were incubated overnight at 4°C with primary Ab in the blocking buffer as above and 0.5% Tween 20. The Ab used was a rabbit polyclonal anti-human COX-2 (1:5000 dilution; Cayman Chemical). A second anti-rabbit Ab (HRP conjugated, 1:20,000 dilution; Pierce) was subsequently incubated with membranes for 1 h at room temperature and then washed extensively (six times for 10 min each) with TTBS (20 mM Tris-HCl, 150 mM NaCl), pH 7.5, 0.1% Tween 20 at room temperature. Following incubation with the SuperSignal Ultra Chemiluminescent substrate (Pierce), membranes were prepared for autoradiography and exposed to Kodak X-Omat film (Eastman Kodak, Rochester, NY).
Detection of nuclear lamin degradation
Expression of nuclear lamin and its degradation fragments were measured by Western blotting. Cells were lysed in 0.5% SDS, protein was determined, and Western immunoblots were performed as described above. After blocking, the membranes were incubated overnight at 4°C with mouse mAbs to lamin A and lamin C (a gift from Dr. Yves Raymond, Research Center, Center Hospitalier de l’Université de Montréal–Hôpital Notre-Dame, Montréal, Québec, Canada). A second anti-mouse Ab (HRP conjugated, 1:20,000 dilution; Pierce) was subsequently incubated with membranes for 1 h at room temperature and, finally, incubated with the SuperSignal Ultra Chemiluminescent substrate (Pierce).
PGE2 was determined on the culture medium with the PGE2 Enzyme Immunoassay Kit (Cayman Chemical). This assay uses the competition between PGE2 and a PGE2-acetylcholinesterase conjugate (PGE2 tracer) for a limited amount of PGE2 mAb. The sensitivity was 9 pg/ml, and the working range was between 10 and 1000 pg/ml, based on a logarithmic transformation.
All statistical analyses were accomplished using InStat Statistical Software (GraphPad, Sorrento Valley, CA). Results are expressed as mean ± SEM when at least three independent experiments were performed. Statistical comparisons were performed with an ANOVA followed by Dunnett’s multiple comparison method. Values of p < 0.05 were considered statistically significant.
SNP causes chondrocyte death, caspase-3 activation, and Bcl-2 down-regulation
Human OA chondrocytes were treated with the NO generator, SNP. The cell viability and the extent of nuclear DNA fragmentation were determined by the MTT assay and ELISA, respectively. Treatment with SNP for 24 h caused chondrocyte death in a dose-dependent manner (Fig. 1⇓, A and B). Western blot analysis using antilamin A and C also confirmed that the cells contained the degradation fragments of nuclear lamin (data not shown), which is one of the characteristic changes during apoptosis (19). An initial 5-h exposure to SNP (1 and 2 mM) showed a significant increase in the extent of nuclear DNA fragmentation without any reduction in cell viability (Fig. 2⇓, A and B).
Dose-dependent effect of SNP on cell viability (A) and DNA fragmentation (B) in human OA chondrocytes. Cells were cultured in medium with or without various concentrations of SNP for 24 h. Mean ± SEM (n = 6). †, p < 0.05; ∗, p < 0.01 vs control; Dunnett’s multiple comparison test.
Time-dependent effect of SNP on cell viability (A) and DNA fragmentation (B) in human OA chondrocytes. Cells were cultured in medium with or without SNP (control, □; 1 mM, ▨; 2 mM, ▪) for 5, 24, 48, and 72 h. Mean ± SEM (n = 6). ∗, p < 0.01 vs control at each indicated time; Dunnett’s multiple comparison test.
Because caspase-3 is an executioner of apoptosis by a variety of stimuli (19), we examined whether SNP-generated NO activates caspase-3 in human OA chondrocytes. We also evaluated the effect of SNP on the level of apoptosis suppressor, Bcl-2, that is an intracellular protein and has been shown to enhance cell survival in part by inhibiting cytochrome c efflux from mitochondria, while protecting cells from apoptosis (20). The activity of caspase-3 and the level of Bcl-2 were assessed after treatment with SNP for 24 h. The treatment with SNP induced a dose-dependent increase in caspase-3 activity at the same time as a dose-dependent decrease in Bcl-2 level (Table I⇓).
Dose-dependent effect of SNP on caspase-3 activity and Bcl-2 level in human OA chondrocytesa
Effects of caspase, mitogen-activated protein kinase (MAPK), NF-κB, and COX inhibitors on SNP-induced chondrocyte death
To examine the signaling cascade on NO-induced cell death in human OA chondrocytes, we used various pharmacological inhibitors that affect different intracellular signaling. In this set of experiments, cells were preincubated with each inhibitor for 2 h, followed by the coincubation of SNP for 24 h. Cell death was initiated by the addition of 1 or 2 mM SNP and was analyzed on the extent of nuclear DNA fragmentation and cell viability. Neither DNA fragmentation nor cell viability in unstimulated controls was affected by each inhibitor used at indicated concentrations ( Figs. 3–6⇓⇓⇓⇓).
Effects of caspase inhibitors, Z-DEVD-FMK and Z-LEHD-FMK, on SNP-induced chondrocyte death. Cell viability (A) and DNA fragmentation (B). Cells were pretreated with Z-DEVD-FMK (100 μM), Z-LEHD-FMK (100 μM), or Z-DEVD-FMK (100 μM) + Z-LEHD-FMK (100 μM) for 2 h, followed by the coincubation of SNP (1 and 2 mM) for an additional 24 h. Mean ± SEM (n = 6). †, p < 0.05; ∗, p < 0.01 vs control without each inhibitor; Dunnett’s multiple comparison test.
Effects of MAPK inhibitors, PD98059 and SB202190, on SNP-induced chondrocyte death. Cell viability (A) and DNA fragmentation (B). Cells were pretreated with PD98059 (50 μM) or SB202190 (10 μM) for 2 h, followed by the coincubation of SNP (1 and 2 mM) for an additional 24 h. Mean ± SEM (n = 6). ∗, p < 0.01 vs control without each inhibitor; Dunnett’s multiple comparison test.
Effects of NF-κB inhibitors, SN-50 and PDTC, on SNP-induced chondrocyte death. Cell viability (A) and DNA fragmentation (B). Cells were pretreated with SN-50 (50 μg/ml) or PDTC (10 μM) for 2 h, followed by the coincubation of SNP (1 and 2 mM) for an additional 24 h. Mean ± SEM (n = 6). ∗, p < 0.01 vs control without each inhibitor; Dunnett’s multiple comparison test.
Effects of COX inhibitors, NS-398 and indomethacin, on SNP-induced chondrocyte death. Cell viability (A) and DNA fragmentation (B). Cells were pretreated with NS-398 (50 μM) or indomethacin (100 μg/ml) for 2 h, followed by the coincubation of SNP (1 and 2 mM) for an additional 24 h. Mean ± SEM (n = 6). †, p < 0.05; ∗, p < 0.01 vs control without each inhibitor; Dunnett’s multiple comparison test.
To define the role of caspases on SNP-induced chondrocyte death, we used caspase-3 inhibitor Z-DEVD-FMK (100 μM) and caspase-9 inhibitor Z-LEHD-FMK (100 μM). Incubation of chondrocytes with the caspase inhibitors alone or in combination for 2 h followed by the subsequent addition of SNP (1 mM) totally prevented both SNP-mediated DNA fragmentation and reduction in cell viability (Fig. 3⇑, A and B). Both DNA fragmentation and reduction in cell viability in response to 2 mM SNP were also prevented but partially by the addition of each or both caspase inhibitor (Fig. 3⇑, A and B). These indicate that both SNP-initiated DNA fragmentation and reduction in cell viability depend on the activity of these caspases. In addition, combined treatment with Z-DEVD-FMK and Z-LEHD-FMK had no additive inhibitory effects on SNP-induced DNA fragmentation nor reduction in cell viability, suggesting that both caspase-3 and caspase-9 participate in the same sequence of cascade during NO-induced chondrocyte death.
To elucidate the role of the extracellular signal-regulated protein kinases (ERK)1/2 and p38 kinase during NO-mediated cell death, we interrupted ERK1/2 and p38 kinase signaling by using the MAPK kinase (MEK)1/2 inhibitor PD98059 (50 μM) and the p38 kinase inhibitor SB202190 (10 μM), respectively. As shown in Fig. 4⇑, A and B, both PD98059 and SB202190 significantly inhibited DNA fragmentation in response to treatment with SNP, accompanied by an increase in cell survival. These results point to possible requirements of both ERK1/2 and p38 kinase during NO-elicited cell death.
We also tested the effects of NF-κB inhibitors, SN-50 (cell-permeable inhibitory peptide) and PDTC, because this transcription factor has also been implicated in the regulation of apoptosis (21). Treatment with SN-50 prevents nuclear translocation of the activated NF-κB complex (22), whereas PDTC inhibits NF-κB activation (23). SN-50 (50 μg/ml) tended to enhance DNA fragmentation and reduce cell viability following SNP addition, but this effect was not significant (Fig. 5⇑, A and B). A control peptide for SN-50 (SN-50 M, 50 μg/ml) had no effect (data not shown). PDTC (10 μM) significantly enhanced DNA fragmentation with a further reduction in cell viability (Fig. 5⇑, A and B). This implies an apoptosis-enhancing capability of the NF-κB inhibitor PDTC.
Because NO has been shown to stimulate PG biosynthesis in vitro and in vivo (24, 25), we also examined the effects of the COX-2-specific inhibitor NS-398 and the COX-1/COX-2 inhibitor indomethacin on SNP-induced chondrocyte death. Incubation of chondrocytes with NS-398 (50 μM) for 2 h followed by the subsequent coincubation of SNP (1 mM) completely blocked both SNP-induced DNA fragmentation and reduction in cell viability (Fig. 6⇑, A and B). Both DNA fragmentation and reduction in cell viability in response to 2 mM SNP were also inhibited by NS-398, but this inhibitory effect was less marked (Fig. 6⇑, A and B). NS-398 at this concentration caused maximal response regarding inhibition of the cell death (data not shown). Treatment with 100 μg/ml indomethacin was equally effective in attenuating both SNP-mediated DNA fragmentation and reduction in cell viability, as was NS-398 treatment (Fig. 6⇑, A and B). These data show that COX-2 appears to be one of the key regulators of NO-induced cell death in human OA chondrocytes.
SNP induces COX-2 expression and PGE2 production in chondrocytes
To clarify whether SNP-generated NO induces COX-2 expression in human OA chondrocytes, we examined the level of COX-2 protein by Western blot analysis. SNP induced COX-2 expression and PGE2 release in a dose-dependent manner (Fig. 7⇓). The COX-2 protein was not expressed in unstimulated controls.
Dose-dependent effect of SNP on COX-2 expression (A) and PGE2 production (B). Cells were cultured with or without various concentrations of SNP for 24 h. The COX-2 expression in adherent cells was determined by Western blotting. PGE2 released into culture medium was measured by enzyme immunoassay. Mean ± SEM (n = 6). ∗, p < 0.01 vs control; Dunnett’s multiple comparison test.
To evaluate the relationship between caspases, MAPK, and PGE2 synthesis during SNP-induced cell death in chondrocytes, we measured PGE2 release after treatment of chondrocytes with SNP (1 mM) in the presence or absence of Z-DEVD-FMK (100 μM), Z-LEHD-FMK (100 μM), PD98059 (50 μM), SB202190 (10 μM), NS-398 (50 μM), or indomethacin (100 μg/ml). The MEK1 inhibitor PD98059, the p38 kinase inhibitor SB202190, the COX-2 specific inhibitor NS-398, and the COX-1/COX-2 inhibitor indomethacin totally blocked the PGE2 release response to 1 mM SNP (Table II⇓). Neither the caspase-3 inhibitor Z-DEVD-FMK nor the caspase-9 inhibitor Z-LEHD-FMK had any effect on the PGE2 production (Table II⇓). Both PD98059 and SB202190 also inhibited the SNP-induced COX-2 expression (data not shown). This suggests that the ERK1/2 and p38 kinase pathways are upstream signaling of the PGE2 production, whereas the caspase cascade is not involved in PGE2 production during NO-induced cell death in human OA chondrocytes.
Effect of caspase inhibitors, MAPK inhibitors, and COX inhibitors on SNP-mediated PGE2 production in human OA chondrocytesa
PGE2 sensitizes chondrocytes to the cell death-inducing effect of NO
To determine the role of PGE2 on SNP-induced chondrocyte death, cells were pretreated with various concentrations of PGE2 (1–1000 ng/ml) for 48 h followed by a subsequent incubation of SNP (1 and 2 mM) without PGE2. As shown in Fig. 8⇓, A and B, pretreatment with PGE2 significantly enhanced the sensitivity of chondrocytes to both SNP-induced DNA fragmentation and reduction in cell viability. PGE2 alone did not induce chondrocyte death. Pretreatment of PGE2 for 24 h was long enough to cause its effect (data not shown). In addition, treatment with PGE2 affected neither the caspase-3 activity (control, 65.5 ± 15.6 U/mg protein; PGE2, 71.2 ± 20.1 U/mg protein, mean ± SEM, n = 6) nor the Bcl-2 level (control, 0.126 ± 0.014 U/mg protein; PGE2, 0.122 ± 0.013 U/mg protein, mean ± SEM, n = 6). These data suggest that exogenous PGE2 sensitizes human OA chondrocytes to the cell death-inducing effect of NO, and the mechanisms underlying the effect of PGE2 does not link directly to caspase-3 activity and Bcl-2 level.
Sensitization of SNP-mediated chondrocyte death by PGE2 pretreatment. Cells were pretreated with or without various concentrations of PGE2 for 48 h, followed by a subsequent incubation of SNP (1 and 2 mM) without PGE2. Mean ± SEM (n = 6). †, p < 0.05; ∗, p < 0.01 vs control without PGE2 pretreatment; Dunnett’s multiple comparison test.
In this study, we demonstrated that SNP-generated NO caused chondrocyte death through COX-2-mediated PGE2 production. Blanco et al. (15) have reported that SNP-generated NO-induced apoptosis in cultured human chondrocytes as determined by electron microscopy, 4′,6-dianidino-2-phenylindole dihydrochloride staining, flow cytometry, and histochemical detection of DNA fragmentation. We also supported these data pharmacologically, that is, both the caspase-3 inhibitor Z-DEVD-FMK and the caspase-9 inhibitor Z-LEHD-FMK completely blocked the SNP (1 mM)-induced chondrocyte death, suggesting that the cell death depends on the activity of caspases, which are largely absent in necrotic cells (26). In fact, the SNP-mediated chondrocyte death was accompanied by an increase in the activity of caspase-3. In contrast, these caspase inhibitors did not affect the chondrocyte death induced by hydrogen peroxide (Notoya et al., unpublished observation), which has been shown to induce necrosis in human chondrocytes (15). Therefore, this report provides a possible explanation for mechanisms by which NO induces apoptosis in human OA chondrocytes. However, a high dose of SNP (2 mM) also caused caspase-independent cell death, probably due to primary or secondary necrosis (27), although part of the chondrocyte death still depended on the activity of caspases.
SNP-generated NO is capable of stimulating the production of PGE2 via the induction of COX-2 in human OA chondrocytes. Our results parallel the studies by Hughes et al. (28) and by Kenthen and Brune (29) that indicate NO donors induced the expression of COX-2 protein as observed in osteoblasts and macrophages, respectively. In contrast, other studies show that NO may inhibit PGE2 release. Stadler et al. (30) have demonstrated that treatment with LPS and IFN-γ up-regulates the production of both NO and PGE2 in rat Kupffer cells. In these cells, the inhibition of NO production by a nonselective NO synthase inhibitor, NG-monomethyl-l-arginine (l-NMMA), further increased PGE2 production. This finding is consistent with the studies of Henrotin et al. (31) and Amin et al. (32), who have demonstrated that l-NMMA enhanced PGE2