All reagents were of analytical grade. The culture medium comprised 1:1 Dulbecco's modified Eagle's medium (DMEM) + Ham's F-12 with penicillin and streptomycin (all from Invitrogen Corporation, Carlsbad, CA, USA). Human recombinant OSM and recombinant human IGF were obtained from Sigma-Aldrich (Poole, UK), and human recombinant TNF-α was obtained from R&D Systems (Abingdon, UK).

Bovine articular cartilage explants were carefully harvested by cutting with a scalpel the outermost layer of articular cartilage without adherent calcified cartilage from bovine heifer stifle joints between 1 and 1.5 years of age. The cartilage explants (12 to 14 mg) were washed three times in phosphate-buffered saline (PBS), placed in 96-well plates, incubated at 37°C, 5% CO₂, and cultured under serum-free conditions in 200 μL of DMEM/F-12 containing cytokines in five replicates. As a control, articular cartilage explants and metabolically inactivated explants were cultured in DMEM/F-12. To deactivate the metabolism of the articular cartilage explants used for the 'metabolically inactive' (MI) condition (to investigate non-chondrocyte-mediated release of fragments), the explants were placed in cryo-tubes (Nunc, Roskilde, Denmark) and then frozen in liquid N₂ and thawed at 37°C in a water bath for three repeated freeze-thaw cycles.

All cell cultures with bovine articular cartilage explants were approved by the local ethics committee. Articular cartilage explants were stimulated for 7, 11, or 17 days with the cytokines OSM (10 ng/mL) and TNF (20 ng/mL). Each catabolic period was followed by either (a) no stimulation or (b) (100 ng/mL) IGF stimulation for 2 weeks, resulting in total culture times of 21, 25, or 31 days (Figure 1c). Between the catabolic and anabolic phases, the explants were washed three times in PBS. On the last day of culture, samples from each treatment were either formaldehyde-fixed or snap-frozen. For other controls, additional samples were cultured for either 7, 11, or 17 days and treated without stimulation (vehicle), OSM + TNF, and IGF, and these samples were also formaldehyde-fixed and snap-frozen. Control treatments were analyzed in parallel on the same plate for vehicle, MI, (100 ng/mL) IGF, and OSM (10 ng/mL) + TNF (20 ng/mL) for 21 days and, on the last day, were formaldehyde-fixed or frozen for protein extraction. All treatment conditions were refreshed three times a week with freshly prepared medium plus stimulants. The conditioned medium was collected and stored at -20°C for further analysis. The use of MI cartilage as a control serves to control for the passive physical-chemical release of proteins and other molecules into the culture medium. Thereby, the difference between MI and vehicle is the cell-mediated release.

Newly synthesized type II collagen was quantified as a marker of cartilage formation using a novel ELISA-based system 26. This ELISA detects an internal amino acid sequence (GPQGPAGEQGPRGDR) in the pro-peptide from the N-terminal of collagen type II, the pre-clinical PIINP (IDS Ltd., Bolton, UK), and the assay was used for the assessment of cartilage formation from the conditioned medium according to the manufacturer's instructions.

The amount of S-GAG in the cartilage explants after termination of the culture was determined by extraction of the proteins by liquid N₂ pulverization in quadruplicates. The explants were individually snap-frozen in liquid N₂ and transferred to frozen stainless-steel pulverization aggregates and, by means of the Bessman tissue pulverizer (Spectrum Laboratories, Inc., Rancho Dominguez, CA, USA), were pulverized and solubilized in 10 volumes of ice-cold buffer: 50 mM Tris-HCl, pH 7.4, containing 0.1 M NaCl and 0.1% Triton X-100 with 1:100 protease inhibitor cocktail III (Calbiochem UK, now part of Merck, Darmstadt, Germany) and 20 μM GM6001 (Biomol International L.P., Plymouth Meeting, PA, USA), a general MMP inhibitor. Compared with that of the traditional procedure, this procedure, using guanidine extraction and papain digestion, results in 95% of the total yield of S-GAG. This approach was specifically chosen as it allows for measurement of the proteins and neo-epitopes. Papain or other digestions destroy peptide sequences. We detected neither pro-peptides nor neo-epitopes in normal unstimulated cartilage.

MMP-2 and MMP-9 expression and activity were determined by gelatinase zymography as described previously 6. This technique allows for assessment of both pro-enzymes and active enzymes, which migrate differently according to their molecular weight during SDS-PAGE electrophoresis. This is important as all MMPs are synthesized as pro-enzymes, which then later are activated. The pro-enzyme is not activated under SDS-PAGE nor preparation but during overnight incubation in the activation buffer 630. Briefly, 5 μL of the samples was loaded onto 7.5% SDS-polyacrylamide gels containing 0.5 mg/mL gelatin. After electrophoresis, the gels were incubated overnight at 37°C in 0.1% Triton X-100, 5 mM CaCl₂, 1 mM ZnCl₂, 3 mM NaN₃, and 50 mM Tris pH 7.4 in a closed container, and then stained with coomassie blue, and finally destained, dried, and scanned for documentation.

One cartilage explant from each treatment was taken out of culture on the appropriate day, fixed in formaldehyde, and processed for standard histology. Alcian blue was used to stain the proteoglycans in 5-μm sections. The sections were stained in a 1% solution of Alcian blue (Sigma-Aldrich) in 3% acetic acid (pH 2.5) for 30 minutes and rinsed in tap water for 2 minutes, and the nuclei were counterstained with Ehrlich's hematoxylin. The sections were dehydrated and mounted in DPX. Digital histographs were captured using an Olympus BX60 microscope with × 60 magnification and an Olympus C5050-zoom digital camera (Olympus, Tokyo, Japan).

All graphs show one representative experiment of at least three, each with at least four replicates. Mean values and standard error of the mean were calculated using GraphPad Prism (GraphPad Software, Inc., San Diego, CA, USA) and compared by the Student two-tailed unpaired t test of statistical significance assuming normal distribution. Asterisks indicate the significance levels (*P < 0.05, **P < 0.01, ***P < 0.001).

A number of studies in different animal species have shown that OSM and TNF in combination induce cartilage degradation in vitro, in part through upregulation of both MMP and aggrecanase activities 67891011. IGF induces cartilage formation with regard to both collagen type II and proteoglycan synthesis 262728. To investigate the repair and formation potential of distinct levels of pathological chondrocytes, we used these well-described cytokines to induce three different levels of catabolic activity followed by anabolic stimulation. The experiments were designed such that different levels of chondrocyte catabolism were induced (OSM + TNF) for 7, 11, and 17 days followed by identical lengths of culture with either IGF or vehicle (14 days) to investigate the capacity for repair.

The total content of proteoglycan retained in the articular cartilage explants was measured to determine whether aggrecan lost from the explants during the catabolic phase could be replaced during the subsequent anabolic phase. As seen in Figure 1a, IGF treatment increased total S-GAG content by approximately 125% compared with the vehicle control, in agreement with previous reports 2831. OSM + TNF activation alone resulted in more than 95% (P < 0.001) depletion of the proteoglycan content. Interestingly, articular cartilage cultured alone in the absence of cytokine induction lost 50% of proteoglycan compared with that to the MI control. compared to the levels of the negative control, metabolic inactive (MI). Compared with t = 0, the MI control lost 40% (P < 0.001) of total S-GAG content, suggesting a substantial physical-chemical diffusion from the culture compared with that of the cell-mediated release when comparing the vehicle with the MI control.

Chondrocytes in the articular cartilage explants exposed to the different levels of catabolic treatment responded differently to IGF treatment. IGF significantly increased the proteoglycan content in all the catabolically depleted explants (Figure 1b). Furthermore, we found that anabolic stimulation restored the S-GAG content in the explants completely when initiated after 7 days of catabolic treatment (comparing Figure 1a vehicle with Figure 1b IGF-stimulated), whereas at later stages only incomplete anabolic responses were obtained. These results indicate that cartilage degradation until day 7 is close to fully reversible, whereas proteoglycan depletion at days 11 and 17 is less reversible. One important limitation of the extraction experiments is that extracted S-GAG may be the result of both newly synthesized proteoglycans and the inhibition of loss of proteoglycans. However, as presented below, retained proteoglycans in the presence of IGF are positioned as circles around the chondrocytes, suggesting new synthesis, although this needs to be documented further.

Proteoglycan degradation, in addition to the extraction of proteoglycan from the cartilage plugs, can be measured by S-GAG release into the conditioned medium, although S-GAG release is more the result of turnover, in contrast to the MMP- and aggrecanase-generated neo-epitopes discussed previously. Figure 1c shows accumulated S-GAG release in the conditioned medium from all treatments. Stimulation with OSM and TNF resulted in substantially increased S-GAG release until day 7 compared with non-stimulated and MI explants. However, after the first 7 days of stimulation with OSM and TNF, there were negligible changes in S-GAG release, with or without subsequent anabolic stimulation, most likely because nearly all of the S-GAG was released by day 7. IGF stimulation, without previous catabolic stimulation, decreased S-GAG loss into the conditioned medium, consistent with its anabolic actions in cartilage.

To investigate the anabolic potential of chondrocytes following the catabolic periods, we accumulated the S-GAG levels released to the conditioned medium for the anabolic periods (days 7 to 21, 11 to 25, and 17 to 31) (that is, during the 14-day anabolic period subsequent to the catabolic insult). As seen in Figure 1d, when the different levels of pathologies were investigated, only small differences in S-GAG release were detected, in contrast to the measurements performed on the cartilage matrix itself.

To further investigate the anabolic response of chondrocytes to IGF after different levels of catabolic stimuli, we measured the release of the N-terminal pro-peptide of pro-collagen type II, PIINP, as a marker of collagen type II synthesis 26. As expected, metabolically inactivated explants showed no type II collagen synthesis, whereas IGF stimulation throughout the culture period resulted in a 4-fold induction of collagen type II synthesis compared with the vehicle control (Figure 2a). In addition, the OSM + TNF-stimulated explants did not synthesize or release collagen II pro-peptides.

To investigate the anabolic potential of chondrocytes following the catabolic periods, we accumulated the PIINP levels released to the conditioned medium for the anabolic periods (days 7 to 21, 11 to 25, and 17 to 31) (that, is during the 14-day anabolic period subsequent to the catabolic insult). After 7 days of catabolic stimulation, collagen type II synthesis increased in response to IGF treatment. However, we observed a lower level of IGF-induced collagen II synthesis after 11 days of cytokine treatment and no IGF-induced collagen II synthesis after 17 days of cytokine treatment (Figure 2b).

Interestingly, under the current culture conditions, the cartilage did not lose the IGF-I responsiveness during prolonged culture periods. When IGF-I was added after 7, 11, or 17 days of culture, a similar induction of cartilage synthesis was observed (data not shown). In addition, these data suggest that cartilage has low levels of continuous collagen type II formation measured by the PIINP assay, however these levels could potently be stimulated by IGF-I exposure.

To further investigate the amount of PIINP that was retained in the cartilage compared with that which was released, we extracted articular cartilage either non-stimulated or stimulated with either catabolic or anabolic stimulation. We are not able to detect PIINP under any conditions (data not shown). These data suggest that n-telo-peptides of pro-collagen type II under the current culture condition are almost exclusively released during synthesis and thereby may be valid markers for collagen type II formation. These data further support our hypothesis that cartilage loss is reversible if the catabolic stimulation is short. Similarly, the potential for reversing cartilage degradation diminishes if cytokine treatment is extensive.

To further characterize the molecular mechanism underlying the loss of repair capacity, we measured levels of the catabolic biomarkers ³⁷⁴ARGSV, ³⁴²FFGVG, and CTX-II after the individual catabolic treatments. We found that OSM + TNF-stimulated degradation, mediated by aggrecanases and measured using the ³⁷⁴ARGSV-G2 assay, was high at day 7, intermediate at day 10, and almost absent at day 17 (Figure 3a). This is consistent with the S-GAG release data showing that the majority of S-GAGs are released at the early stages of catabolic stimulation. The levels of the MMP-generated fragment ³⁴²FFGVG-G2 showed that MMP-mediated aggrecan was undetectable at days 7 and 10 and high at day 17 (Figure 3b). The high aggrecanase activity at the early stages of culture may mask the MMP-mediated aggrecan epitope (³⁴²FFGVG-G2) by further processing in generating the aggrecanase (³⁷⁴ARGSV) site; however, Fosang and colleagues 32 have found that further processing of ³⁴²FFGVG to generate ³⁷⁴ARGSV cannot occur, at least not in vitro. High levels of MMP activity should have generated CTX-II fragments that are not further processed by other proteases, suggesting that MMP activity is present only at a lower level at early culture time points. This was verified by the use of a fluorescence substrate technique, in which MMP levels were detectable only in the presence of catabolic stimulation and only at late time points (data not shown), which correlate well with previous findings, documenting extensive MMP activities at later stages of catabolic induction but not at early stages 6932. Interestingly, most S-GAG is released at earlier time points than the ³⁴²FFGVG release, indicating that aggrecan loss is due primarily to aggrecanase activity, but later, aggrecanolysis shifts to an MMP-mediated degradation mode. Finally, we found that the release of the collagen type II degradation fragment CTX-II (Figure 3c) occurred with a pattern similar to that of ³⁴²FFGVG (Figure 3b), consistent with the fact that the CTX-II fragment is MMP-generated 33.

In summary, these data appear to mimic cartilage degradation in arthritis where aggrecanase activity on aggrecan precedes MMP mediated aggrecan degradation that is subsequently followed by MMP degradation of collagen, which has been reported with various techniques from other labs 32. In addition, these data show that there is a positive correlation between MMP activity (evidenced by the ³⁴²FFGVG-G2 and CTX biochemical markers) and the inability of cytokine-treated chondrocytes to initiate and/or maintain anabolic activity.

To further investigate the protease levels during anabolic and catabolic phases of chondrocyte stimulation, we measured MMP activity by gelatine zymography (Figure 4). The ³⁴²FFGVG-G2 and CTX-II peptides are generated by an array of MMPs, of which MMP-2 and MMP-9 are only a subset. On other occasions, the presence of these gelatinases has been a valid indication of total MMP activity and thereby the catabolic potential of the culture 6. Gelatinase activity at 7, 11, and 17 days after catabolic treatment was compared with gelatinase activity after 7 days of IGF treatment, corresponding to the middle of the anabolic stimulation period. We found that gelatinase activity and expression were attenuated by IGF, but not completely reversed, compared with gelatinase activity after 7 days with vehicle alone (Figure 4). The results with samples analyzed after 14 days of IGF or vehicle were similar to those for 7 days of IGF or vehicle (data not shown). The presence of active MMP-2 and MMP-9 at days 11 to 17 corresponds to the period when high levels of ³⁴²FFGVG-G2 and CTX-II are detected in Figures 3a and 3c. These data also indicate that, even in the presence of substantial MMP activity, chondrocytes are able to synthesize new aggrecan and proteoglycans (Figure 1b), but not collagen type II (Figure 2b).

To visualize the repair enhanced by IGF treatment, cultured cartilage was harvested at different time points. Proteoglycans in the cartilage were visualized using Alcian blue staining, the same dye used in the S-GAG assay. The control articular cartilage explants (shown in the bottom row of Figure 5) were cultured for 21 days with vehicle, OSM + TNF, IGF, or MI control for 21 days. In complete agreement with the S-GAG quantifications in Figure 2, IGF increased whereas OSM + TNF decreased GAG content compared with vehicle. MI control contained more GAG compared with vehicle as the cell-mediated loss of proteoglycan content was abrogated. With regard to the dynamics in the reversibility experiments presented in the upper panels, the vehicle control explants gradually lost S-GAG content over time, whereas the explants treated with IGF maintained the S-GAGs, even after 17 days in culture. OSM + TNF treatment depleted proteoglycans from the matrix maximally by day 7, consistent with the results in Figure 1c which show that S-GAG release into the medium is also maximal by day 7. Treatment with IGF stimulated GAG synthesis in the explants that were treated with cytokines for 7 and 11 days, but not for 17 days. IGF treatment of explants after 7 days of catabolic stimuli restored proteoglycan content throughout the entire cartilage matrix. IGF treatment after 11 days of catabolic treatment showed new proteoglycan synthesis around chondrocytes, indicative of repair. There was also evidence of repair in the absence of IGF treatment after 11 days of catabolic stimuli; however, the repair was substantially improved in the presence of IGF. Chondrocytes treated with catabolic cytokines for 17 days reinitiated, only to a very minor extent, proteoglycan synthesis in the presence of IGF compared with that of vehicle. These data support the idea that cartilage degradation may be more reversible before induction of MMP activity.