Effects of short-term glucocorticoid treatment on changes in cartilage matrix degradation and chondrocyte gene expression induced by mechanical injury and inflammatory cytokines

Cartilage disks were harvested from the femoropatellar grooves of one- to two-week-old bovine calf knee joints (obtained from Research 87, Hopkinton, MA, USA) as previously described [29]. A total of 16 joints from 13 different animals and 1 human were used. Briefly, cartilage-bone cylinders (9 mm diameter) were cored perpendicular to the surface. After a level surface was obtained by removing the most superficial layer (approximately 100 to 200 μm), one to two sequential 1 mm-slices of middle zone cartilage were cut from each cylinder. Five disks (3 mm-diameter, 1 mm-thick) were cored from each slice using a dermal punch. Cartilage from this middle zone in newborn calves was shown previously to have a reasonably homogeneous population of cells and matrix [30]. Cartilage disks for all treatment groups were matched for depth and location along the joint surface [31]. Disks were equilibrated in serum-free medium (low-glucose DMEM (Cellgro, Herndon, VA, USA)), 10 mM HEPES buffer (Invitrogen, Carlsbad, CA, USA), supplemented with 1% insulin-transferrin-selenium (10 μg/ml, 5.5 μg/ml and 5 ng/ml, respectively), 0.1 mM nonessential amino acids, 0.4 mM proline, 20 μg/mL ascorbic acid, 100 units/mL penicillin G, 100 μg/mL streptomycin, and 0.25 μg/mL amphotericin B (all from Sigma, St. Louis, MO, USA)) for two days prior to treatment in a 37°C, 5% CO₂ incubator.

Human donor knee cartilage (49-yr-old female, modified-Collins [32] grade-1 knee joint) was obtained from the Gift of Hope Organ and Tissue Donor Network (Elmhurst, IL, USA), approved by the Office of Research Affairs at Rush-Presbyterian-St. Luke's Medical Center and the Committee on Use of Humans as Experimental Subjects at MIT. Any fibrillated areas detected under visual inspection were excluded from the study. Human cartilage harvest and culture were identical to that of bovine, but included the intact superficial zone and each disk was approximately 0.8 mm thick. Human knee cartilage was obtained from both the femoropatellar groove and femoral condyles.

After equilibration in medium for three days, disks were injuriously compressed in a custom-designed incubator-housed apparatus [33, 34]. Each bovine disk was subjected to radially unconfined compression to 50% final strain at 1 mm/second velocity (100% per second strain rate), followed by immediate release of load at the same rate, as described [29]. Immediately after injury, some disks were deformed to an ellipsoidal shape (deformation score of 1 or 2 as described in [35]), but none exhibited gross fissuring. Adult human cartilage disks were thinner, had intact superficial zone and different effective biomechanical behavior compared to the immature bovine disks, reflecting the anisotropy and inhomogeneity associated with the presence of the superficial zone. The combined properties were such that higher strain and strain rate values were needed to produce levels of peak stress and visible deformation in human cartilage similar to that observed for immature bovine tissue [29]. Thus, a strain of 60% and strain rate of 300%/second were used, the same values utilized in our recent report with this in vitro injury plus cytokine stimulation system for adult human cartilage [11]. The resulting macroscopic tissue changes in human cartilage disks were similar (elliptical appearance) to those described previously using our human cartilage injury model and scoring system [36]. After injury, samples were immediately placed in treatment medium.

In a DEX dose-response study, bovine cartilage samples (70 disks from two joints of one animal) were treated either with or without rhTNFα (25 ng/mL) and incubated for six days with increasing concentrations of DEX (Sigma, St. Louis, MO, USA), from 0.1 nM to 100 μM.

Cartilage samples were either subjected to injurious compression or left uninjured, incubated with or without cytokines (all from R&D Systems, Minneapolis, MN, USA), and with or without DEX. Previously [11], we observed that treatments with TNFα, TNFα + injury, TNFα + IL-6/sIL-6R, and TNFα + injury + IL-6/sIL-6R caused significant release of GAGs from both human and bovine cartilage explants, with the latter condition causing the most severe loss of GAG. In this study, we first examined the effects of DEX on cartilage explants under these same conditions. For bovine cartilage (70 disks from two joints of another animal), DEX and recombinant human TNFα (rhTNFα) were used at 10 nM and 25 ng/mL, respectively, based on the results from the DEX dose response study. For human cartilage (36 disks from the distal femur), DEX and rhTNFα were used at 100 nM and 100 ng/mL, respectively. rhIL-6 (50 ng/mL) was always used in combination with soluble IL-6 receptor (sIL-6R, 250 ng/mL), since this combination was found previously to induce greater aggrecan degradation than when used separately in the presence of TNFα [37]. Bovine cartilage disks were cultured in these conditions for six days. Culture duration for human explants was extended to 10 days based on earlier studies showing that human cartilage released sGAG more slowly than bovine cartilage for these conditions [11]. Medium was replaced every two days and saved for analysis.

To test whether a short-duration pre-exposure of cartilage to DEX could prevent GAG loss and inhibition of biosynthesis induced by subsequent cytokine treatment, bovine cartilage disks (10 disks from a separate animal) were either pre-treated with DEX for two days or incubated in medium alone. Afterwards, both groups were incubated in medium containing TNFα but no DEX for an additional four days. To test whether post-treatment with DEX could diminish the effects of a pre-established cytokine insult, cartilage explants (10 disks from a different animal) were first treated with TNFα for two days, and DEX was then added to the medium in addition to continued treatment with TNFα for another four days. GAG loss and radiolabel incorporation were measured as above.

Two days before termination of the bovine cultures, the medium of each disk was supplemented with 5 μCi/ml Na₂³⁵SO₄ (Perkin-Elmer, Norwalk, CT, USA) as a measure of the rate of proteoglycan synthesis. The amount of radiolabeled sulfate was doubled in studies of human cartilage. Upon termination, disks were washed, weighed and solubilized (proteinase K, Roche, Indianapolis, IN, USA), and radiolabel incorporation was measured using a liquid scintillation counter [30]. The amounts of GAG lost to the medium and retained in the cartilage were measured using the dimethylmethylene blue (DMMB) assay, with shark chondroitin sulfate (Sigma) as the standard [38].

To examine the effects of DEX, injury and TNFα on chondrocyte gene expression, bovine cartilage disks from six different animals were cultured for four days under the eight treatment conditions: (1) no-treatment control, (2) DEX-only, (3) mechanical injury only, (4) DEX + injury, (5) TNFα, (6) TNFα + DEX, (7) TNFα + injury, and (8) TNFα + injury + DEX. A total of 48 disks per animal from each of six different joints (six different animals) were used. From each joint, RNA was pooled from the six disks assigned to each of the eight treatment conditions (matching disks from along the joint surface across treatment groups). Thus, there were six different repeats of this experiment in total, with each repeat corresponding to a different joint (animal). Samples were pulverized in liquid nitrogen and homogenized in TRIzol reagent (Invitrogen). The extract was spun at 13,000 g for 10 minutes in Phase Gel tubes (Eppendorf, Hamburg, Germany) with 10% chloroform (Sigma). After spinning, the clear supernatant was obtained and RNA was isolated using the RNeasy Mini columns (Qiagen, Chatsworth, CA, USA); genomic DNA was removed by a DNase digestion step (Qiagen) according to the manufacturer's protocol. Absorbance was read at 260 nm and 280 nm to measure the concentration of RNA and the purity of the extract. Reverse transcription of equal quantities of RNA (2.5 μg) from each condition was performed using the AmpliTaq-Gold Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) [39]. Genes of interests were those involved in cartilage homeostasis, including matrix molecules (aggrecan, collagen II and IX), cytokines (IL-1β, IL-6, TNFα), proteases and protease inhibitors (ADAMTS-4,-5, matrix metalloproteinase-3 (MMP-3), tissue inhibitor of metalloproteinase-3 (TIMP-3)), iNOS and a housekeeping gene (18 S). Bovine primer sequences for all genes except iNOS, collagen IX and IL-6 were reported in our previous studies [40, 41]; sequences for these latter three genes were reported in another study [42]; they were also designed using Primer3 software [43] on the basis of bovine sequences. A standard curve for amplification was generated for each of the primer. All primers demonstrated approximately equally efficiency, with standard curve slopes of approximately 1, indicating a doubling in complementary DNA quantity in each cycle [39]. Real-time PCR was performed using Applied Biosystems ABI 7900HT instrument and SYBR Green Master Mix (Applied Biosystems). Measured threshold values (Ct) were converted to RNA copy number according to primer efficiencies. Within each condition, the RNA copy numbers for each gene were normalized to that of 18 S from the same condition. To examine the effects of treatments, each gene was then normalized to its level in the no-treatment control group.

The role of chondrocyte GRs in the response to DEX was studied in bovine cartilage samples (30 disks from one animal) by treatment with the GR antagonist, RU486 (1 μM, Sigma), in the presence of TNFα and TNFα + DEX for six days. The role of proprotein convertases in matrix degradation was tested by the addition of the PC inhibitor decanoyl-RVKR-CMK (10 μM, Calbiochem, La Jolla, CA, USA) to bovine cartilage explants (35 disks from one animal) cultured with different combinations of TNFα, IL-6/sIL-6R and mechanical injury. The levels of iNOS protein were measured following four-day treatments of bovine cartilage disks with TNFα ± injury, in the presence or absence of DEX. The disks were then pulverized in liquid nitrogen and homogenized in buffer solution (20 mM pH 7.6 Tris, 120 Mm NaCl, 10 mM EDTA, 10% glycerol, 1% Nonidet P-40 (Sigma) with protease inhibitor cocktail (Roche)). Equal amounts of protein were collected from each condition, run on 4 to 15% gels (Invitrogen) and then transferred to polyvinylidene difluoride (BioRad, Hercules, CA, USA) for immunoblotting. Western blots were performed using anti-bovine iNOS antibody (1:1000, Millipore, Billerica, MA, USA), followed by secondary antibodies conjugated to horseradish peroxidase (1:4000, Cell Signaling Technology, Beverly, MA, USA). In another study, nitrite levels in the medium of 48 disks (one animal) were analyzed using the Griess Reagents (Invitrogen).

Experiments were performed to test the effect of DEX (0.1 nM to 100 μM) on both TNFα-treated and untreated control cartilage explants (Figure 1). TNFα treatment significantly increased GAG loss to the medium (to 16.2 ± 0.5% of total by six days) compared to that from controls (8.5 ± 0.2%, mean ± SEM), a finding consistent with previous studies [11]. DEX at concentrations of 1 nM or higher reduced GAG loss induced by TNFα treatment to levels that were not significantly different from controls. At concentrations 100 nM and higher, DEX treatment alone suppressed GAG loss to levels below those found in control cultures (Figure 1A).

All cartilage samples from Figure 1A were also radiolabeled with ³⁵S-sulfate to measure the rates of proteoglycan biosynthesis in response to treatment conditions. Compared to controls (having ³⁵S-sulfate incorporation = 51.6 ± 2.1 pmol/hour/mg wet weight), TNFα treatment significantly reduced sulfate incorporation to 25.3 ± 2.3 pmol/hour/mg (Figure 1B). In contrast, treatment with TNFα and DEX at concentrations of 0.1 nM and higher showed sulfate incorporation rates, which were not significantly different from controls. Moreover, concentrations of 0.1 μM to 100 μM DEX alone significantly increased sulfate incorporation above control levels (70.0 ± 1.6, 75.9 ± 3.5, and 73.0 ± 1.0 pmol/h/mg, respectively, Figure 1B).

Consistent with our previous findings, TNFα treatment together with mechanical injury or IL-6/sIL-6R or the combination of all three treatments, significantly increased GAG release from bovine cartilage (Figure 2A) [11]. The combined treatment with injury +TNFα IL-6/sIL-6R caused the most severe GAG loss by six days. The addition of 10 nM DEX significantly reduced GAG loss caused by injury +TNFα, TNFα + IL-6/sIL-6R, and injury +TNFα + IL-6/sIL-6R, the latter from 53.6 ± 9.8% down to 13.8 ± 1.5% compared to 7.3 ± 0.2% for controls.

TNFα treatment, either alone or together with mechanical injury, IL-6/sIL-6R or their combination, greatly reduced sulfate incorporation rates (Figure 2B), as seen in our previous study [11]. Importantly, DEX abolished the reduction in biosynthesis caused by all these treatments. For example, treatment with TNFα + IL-6/sIL-6R + injury reduced sulfate incorporation to 26.2 ± 7.2 pmol/hour/mg, whereas the addition of 10 nM DEX to this same condition significantly increased sulfate incorporation to 96.2 ± 13.44 pmol/hour/mg, a level that was not significantly different from no-treatment controls.

Treatment with injury + TNFα + IL-6/sIL-6R greatly increased GAG loss from human knee cartilage (to 36.0 ± 4% of total, Figure 3), consistent with our previous report [11]. Under these conditions, the addition of 100 nM DEX significantly reduced GAG loss to 20.5 ± 1.5%, but showed no effect on sulfate incorporation (data not shown).

Bovine cartilage samples were pre-incubated with 10 nM DEX for two days and then cultured in medium containing TNFα, without DEX, for an additional four days. The pre-treatment with DEX significantly reduced TNFα-induced GAG loss by Day 4 (Figure 4A), and significantly increased the sulfate incorporation rate compared to the condition without DEX pre-treatment (Figure 4B).

We next examined whether DEX would exert anti-catabolic effects in cartilage samples where matrix degradation had already been induced by cytokine stimulation. All cartilage samples were pre-incubated with TNFα for two days (starting at Day -2 in Figure 4C). Afterwards, starting at Day 0, one group of samples was cultured in medium with TNFα +10 nM DEX, while a second group was treated with TNFα alone. After the two-day pre-incubation with TNFα, disks from both groups had lost approximately 6% of total GAG (Day 0, Figure 4C). The addition of DEX significantly attenuated GAG loss and increased proteoglycan biosynthesis by Day 4 (Figure 4C, D).

To assess whether the inhibition of GAG loss and the increase in proteoglycan biosynthesis in DEX-treated cartilage were GR mediated, bovine explants were treated with the GR antagonist RU486 in addition to TNFα ± DEX. As shown in Figure 5, TNFα significantly increased GAG loss and reduced sulfate incorporation rate; the addition of DEX significantly reduced the release of GAGs and increased the sulfate incorporation rate. The effects of DEX on biosynthesis were significantly reversed by the presence of RU486, though the increase in GAG release upon addition of RU486 was not statistically significant. RU486 alone had no effect on either normal controls or TNFα-treated samples.

Real-time qPCR was performed to determine bovine chondrocyte gene expression responses to four-day treatments with DEX, TNFα and mechanical injury alone and in combinations (Figure 6). Matrix molecules collagen II and IX responded to both TNFα and TNFα + injury treatments with a significant decrease in mRNA levels. DEX treatment increased the expression of both genes in cartilage treated with TNFα to levels not significantly different than controls. Aggrecan core protein mRNA levels were significantly decreased in response to TNFα + injury; however, the addition of DEX resulted in mRNA levels not significantly different than controls. IL-6 mRNA levels were increased significantly by treatments involving TNFα, regardless of the presence of DEX or mechanical injury treatment. Treatment conditions had no significant effect on TNFα or IL-1β mRNA levels.

Analysis also showed that TNFα alone significantly up-regulated the levels of ADAMTS-4 and MMP-3 mRNA. TNFα + injury significantly increased ADAMTS-4 and ADAMTS-5 mRNA levels in the presence of DEX. Additional genes, related to protease and protease inhibition, were up-regulated in response to the TNFα + injury treatment, including TIMP-3 and MMP-3. Among the matrix proteases, only MMP-3 mRNA showed reduced expression in response to DEX + TNFα and DEX + TNFα+ injury treatments, whereas ADAMTS-5 mRNA levels were not down-regulated in the presence of DEX.

iNOS message expression was significantly elevated in response to all treatments with TNFα, but not by injury alone. The induction of iNOS mRNA was not abrogated by the addition of DEX (Figure 7A). However, DEX significantly reduced the amount of nitrite released to the conditioned medium caused by TNFα treatment alone (Figure 7B). Western blot analysis showed that iNOS protein levels in the TNFα + DEX and TNFα + injury + DEX conditions were markedly reduced compared to these same conditions without DEX (Figure 7C).

To assess the role of PC in cartilage degradation, a general PC inhibitor, decanoyl-RVKR-CMK, was added to the conditions of TNFα, TNFα + IL-6/sIL-6R, and TNFα + IL-6/sIL-6R + injury. 10 μM CMK significantly decreased GAG release induced by TNFα + IL-6/sIL-6R + injury (Figure 8).