Article Text

Download PDFPDF

Synthesis and release of human cartilage matrix proteoglycans are differently regulated by nitric oxide and prostaglandin-E2
  1. S C Mastbergen,
  2. J W J Bijlsma,
  3. F P J G Lafeber
  1. Department of Rheumatology & Clinical Immunology, University Medical Center Utrecht, Utrecht, The Netherlands
  1. Simon C Mastbergen, University Medical Center Utrecht, Room F02.127, PO Box 85500, 3508GA Utrecht, The Netherlands; s.mastbergen{at}


Objectives: Recent studies showed beneficial effects of COX-2 inhibition on proteoglycan turnover of both IL-1β/tumour necrosis factor α (TNFα) damaged cartilage and of osteoarthritic cartilage. Although proteoglycan release and content were normalised, proteoglycan synthesis was only partially influenced. Prostaglandin-E2 is the main product formed by COX-2. We therefore evaluate the role of prostaglandin-E2 in relation to nitric oxide in disturbing cartilage proteoglycan turnover.

Methods: Human healthy cartilage, alone or in the presence of IL-1β+TNFα, was cultured for 7 days with or without prostaglandin-E2 or the selective COX-2 inhibitor (celecoxib 10 μM). Changes in cartilage matrix proteoglycan turnover, levels of prostaglandin-E2 and nitric oxide were determined.

Results: Proteoglycan synthesis and release of the cartilage were not affected by prostaglandin-E2 alone. Addition of IL-1β+TNFα to healthy cartilage resulted in inhibition of proteoglycan synthesis and increase in proteoglycan release. When prostaglandin-E2 was added, in addition to IL-1β+TNFα, proteoglycan release increased further, but proteoglycan synthesis was not influenced further. Addition of a selective COX-2 inhibitor to the IL-1β+TNFα treated cartilage inhibited the enhanced prostaglandin-E2 production and almost completely normalised proteoglycan release, whereas synthesis remained unaffected. Also, the enhanced NO-levels remained elevated. Prostaglandin-E2 levels correlated significantly with proteoglycan release, whereas NO levels correlated significantly with proteoglycan synthesis.

Conclusion: The present results suggest involvement of prostaglandin-E2 in enhanced cartilage proteoglycan release but not synthesis, although healthy cartilage has to be sensitised by IL-1β+tumour necrosis factor α (TNFα). IL-1β+TNFα induced NO seems to be involved in inhibition of proteoglycan synthesis, independent of prostaglandin-E2, and thus seems insensitive to regulation by (selective) COX-2 inhibitors.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Osteoarthritis (OA), a chronic disease with slowly progressive destruction of the articular cartilage, results from the failure of chondrocytes to maintain the balance between synthesis and degradation of the extracellular matrix.1 Although osteoarthritis was characterised originally as a non-inflammatory arthropathy (osteoarthrosis), inflammatory responses in the synovial membrane caused by direct biomechanical perturbation and/or caused by reaction to cartilage matrix degradation products will contribute to disease progression.2 3 However, the major events in osteoarthritis pathogenesis are expectedly localised within the cartilage itself. There appears to be a central role for the cytokines interleukin-1β (IL-1β) and tumor necrosis factor-α (TNFα) produced by the chondrocytes and synovial tissue.4 5 These cytokines stimulate the synthesis of prostaglandin-E2 (PGE2) and nitric oxide (NO).6 7 More recent studies showed that the in vivo selective inhibition of iNOS, responsible for the production of NO, reduces the symptoms of inflammation and the biochemical abnormalities of osteoarthritic joints.810 Various roles of NO as mediator of other IL-1 induced responses, including the inhibition of aggrecan and collagen synthesis, enhancement of matrix metalloproteinase (MMP)-activity and chondrocyte apoptosis and reduction in the production of IL-1 receptor antagonist (IL-1RA) have also been suggested.1017

While it is quite clear from the literature that excessive production of NO exerts predominantly catabolic effects being harmful to cartilage, observations with respect to prostaglandin overproduction are less clear. In articular chondrocytes, IL-1β and TNFα synergistically induce COX-2, whereas COX-1 expression remains unchanged.18 Prostaglandin-E2 is elevated in the osteoarthritic joint in cartilage as well as in synovium.19 20 Prostaglandin-E2, a predominant product of COX-2, can exert both anabolic and catabolic effects in synovium and cartilage.21 22 For example, prostaglandin-E2 can reverse proteoglycan degradation induced by IL-1β in bovine and human cartilage explants.21 Prostaglandin-E2 also inhibits IL-1β induced MMP (collagenase, stromelysine) expression in human synovial fibroblasts.23 24 At nanomolar concentrations, prostaglandin-E2 enhances collagen type II and proteoglycan synthesis.25 26 In contrast, in studies using growth-plate chondrocytes, prostaglandin-E2 inhibited collagen synthesis but did not alter non-collagen protein synthesis.25 Prostaglandin-E2 is also capable of inhibiting chondrocyte differentiation: prostaglandin-E2 suppresses maturation of growth-plate chondrocytes and dose-dependently inhibits the expression of several maturation genes, such as type X collagen and metalloproteinase-13.27

Furthermore, indirect catabolic effects of prostaglandin-E2 are demonstrated by the use of selective COX-2 inhibition. Selective COX-2 inhibition, and with that prostaglandin-E2 production, was beneficial for cartilage under the influence of IL-1 and TNFα.28 Findings by El Hajjaji et al showed that selective COX-2 inhibition in vitro was able to increase proteoglycan synthesis and to diminish proteoglycan release of osteoarthritic cartilage obtained in joint-replacement surgery.29 Recent findings by our group confirmed these data and additionally demonstrated that selective COX-2 inhibition had in vitro a favourable effect on proteoglycan synthesis, retention, release and content of both degenerated (preclinical) and (late-stage) human osteoarthritic cartilage.30 As prostaglandin-E2 is a predominant product of COX-2, these data are in line with the catabolic role of prostaglandin-E2.

In the present study, we further evaluated the effect of prostaglandin-E2 in relation to NO in human articular cartilage explants with a focus on the differential effects on proteoglycan synthesis and release.


Cartilage culture technique

Human articular cartilage tissue was obtained post-mortem from knee condyles within 24 h of death. The donors were without any known history of joint disorders (mean age 65 (11) years, 7 males and 5 females (SEM)). Donors were without known predisposing conditions for joint disorders, and no signs of cartilage degeneration were present. Slices of cartilage were cut aseptically from the articular surface, excluding the underlying bone, and kept in phosphate-buffered (pH 7.4) saline (PBS). Within 1 h of dissection, the slices were cut into square pieces, weighed aseptically (range 5–15 mg, accuracy (0.1) mg) and cultured individually in 200 μl of culture medium in 96-well round-bottomed microtitre plates as described previously.31 The culture medium consisted of Dulbeco’s modified Eagle’s medium (D-MEM), supplemented with glutamine (2 mM), penicillin (100 IU/ml), streptomycin sulfate (100 μg/ml), ascorbic acid (0.085 mM) and 10% heat inactivated pooled human male AB+ serum. Cartilage was always precultured for 24 h (washout period), where subsequently culture medium was refreshed before the start of the experiment.

Experimental set-up

Healthy cartilage tissue was cultured for 7 days alone, in the presence of different concentrations prostaglandin-E2 (PGE2; Caymann Chemical, Ann Arbor, Michigan, USA), in the presence of IL-1β plus TNFα (200 pg/ml, Biosource, PHC0814, and 800 pg/ml, Pharmingen, 19761T, respectively), in the presence of IL-1β plus TNFα together with prostaglandin-E2, and in the presence of IL-1β plus TNFα with a selective COX-2 inhibitor (celecoxib; supplied by Pfizer, New York, USA; 10 μM) inhibiting prostaglandin-E2 production. The mean pharmacological plasma concentration is reported to be around 5 μM, making a concentration up to 10 μM in an in vitro set-up of pharmacological relevance.32 After 4 days, medium was refreshed and cartilage cultured for a further 3 days with the same additions. Changes in cartilage matrix turnover (proteoglycan-synthesis, -retention, -release and -content) and, when relevant, prostaglandin-E2 and NO production were determined.

Proteoglycan analyses

Sulphate incorporation rate was determined, as a measure of the proteoglycan synthesis rate, during the last 4 h of the first 4-day culture period, as described previously.30 After 4-h labelling, the cartilage explants were rinsed 3 times for 45 min in culture medium and incubated for an additional 3 days. After this second culture period, medium was removed and stored at −20°C for further analysis. Cartilage tissue samples were digested in papain buffer as described before.30 Glycosaminoglycans (GAGs) were stained and precipitated with Alcian Blue dye solution.33 35SO42– radioactivity of the samples was measured by liquid scintillation analysis. Proteoglycan synthesis rate is expressed as percentage change compared with untreated control values.

Release of newly formed proteoglycans as a measure of retention of these proteoglycans was determined similarly. GAGs were precipitated from the medium obtained from day 4 to 7 with Alcian Blue.30 The radio-labelled GAGs were measured by liquid scintillation analysis and normalised to the proteoglycan synthesis rate. Percentage release of newly formed proteoglycans is expressed as percentage change compared with untreated control values.

For the total release of proteoglycans, the GAGs in the medium obtained from day 4 to 7 were precipitated and stained with Alcian Blue as described above. The GAG content in the papain digest of cartilage samples, as a measure of proteoglycan content, was analysed in the same way. Values for content were normalised to the wet weight of the cartilage and expressed as percentage change compared with untreated control values. Values for release were normalised to the GAG content of the explants. Percentage release of GAGs is expressed as percentage change compared with untreated control values.

Prostaglandin E2 and NO determination

Prostaglandin-E2 (PGE2) and NO were determined in the 4-day culture medium after the first 4 days of culture. Prostaglandin-E2 was determined by Enzyme Immuno Assay (Caymann Chemical, Ann Arbor, Michigan, USA) and expressed as micrograms per millilitre per gram (wet weight) of cartilage tissue. Nitrite levels, as a measure for NO, in culture medium were determined using the standard Griess reaction and expressed as millimoles per gram (wet weight) of cartilage tissue.

Calculations and statistical analysis

Because of focal differences in composition and bioactivity of the cartilage in the knee joint, the results of 10 cartilage samples per parameter per donor, obtained randomly and handled individually, were averaged and taken as a representative value for the cartilage of that donor. Several experiments with each cartilage of a different donor were performed. Statistical evaluation of the effects of a single intervention compared with the untreated cartilage of the same donors was performed with a non-parametric test for paired data (Wilcoxon). For statistical evaluation of differences between different interventions the percentage change compared with untreated cartilage of the same donors was calculated and compared statistically by use of a non-parametric test for unpaired data (Mann–Whitney). Values of p than or equal to 0.05 were considered statistically significant.


Effect of prostaglandin-E2 on normal human articular cartilage

Prostaglandin-E2, at a concentration ranging from 3 to 500 pg/ml, did not change proteoglycan turnover of normal healthy human articular cartilage. Proteoglycan synthesis rate, release of newly formed proteoglycans, as a measure of retention of these newly formed proteoglycans, total release of proteoglycans, newly formed plus resident ones, as well as proteoglycan content were not influenced by exogenous prostaglandin-E2; effects remained with ±5%, not statistically significant (data not shown).

Effect of prostaglandin-E2 on human articular cartilage explants under the influence of IL-1β plus TNFα

To evaluate whether prostaglandin-E2 needed stimulated chondrocytes to display an effect, the cartilage was sensitised. Cartilage was exposed to two major mediators of cartilage degeneration, IL-1β and TNFα. Under the influence of IL-1β plus TNFα, the proteoglycan synthesis rate was significantly inhibited (on average almost 75% inhibition; fig 1A). Addition of prostaglandin-E2 showed no effect on this inhibited proteoglycan synthesis rate (still about 70% inhibition, p<0.05 compared with controls; fig 1A). The percentage release of newly formed proteoglycans (normalised to the proteoglycan synthesis rate) was significantly enhanced by the addition of IL-1β plus TNFα (about 25%, p<0.05; fig 1A), indicating an impaired retention of the newly formed proteoglycans. Remarkably, when prostaglandin-E2 was added, this release of newly formed proteoglycans was more than doubled (close to 100%, p<0.05; fig 1A), significantly different from the IL-1β plus TNFα condition without prostaglandin-E2 (p<0.04). Also, the total proteoglycan release, consisting of both newly formed proteoglycan and resident proteoglycans, was increased by the addition of prostaglandin-E2 in the same way. Addition of IL-1β plus TNFα alone displayed a total release of about 80% (p<0.05), which more than doubled when prostaglandin-E2 was added (more than 200%, p<0.01). IL-1β plus TNFα also induced a statistically significant decrease in proteoglycan content of the cartilage of a few per cent (p<0.05, fig 1B). Prostaglandin-E2 also for this parameter amplified the effect significantly to a 12% decrease in proteoglycan content (p<0.02).

Figure 1 Effect of prostaglandin-E2 on IL-1β plus TNFα induced cartilage damage. Percentage change was compared with healthy cartilage of IL-1β plus TNFα-treated cartilage in the absence (open bars) or presence (gray bars) of prostaglandin-E2. Effects on proteoglycan synthesis rate, percentage release of newly formed proteoglycans, percentages total release of proteoglycans (fig 1A) and proteoglycan content (fig 1B) are depicted. The absolute control values (average±SEM) were 6.33±1.05 nmol/h.g; 9.68±0.62%, 4.33±0.59% and 26.18±4.31 mg/g for these parameters, respectively). Open bars represent effects of IL-1β plus TNFα (200/800 pg/ml, respectively) on human healthy articular knee cartilage. Grey bars represent effects of prostaglandin-E2 (200 pg/ml) when added to IL-1β plus TNFα. The results are presented as means of at least 4 experiments with each cartilage of a different donor±SEM. Statistically significant differences with untreated control cartilage are indicated by asterisks. Statistically significant effects of prostaglandin-E2 are indicated by p values.

Effect of COX-2 inhibition on proteoglycan synthesis, release, prostaglandin-E2, and NO production of human articular cartilage under the influence of IL-1β plus TNFα

The addition of prostaglandin-E2 to cartilage under the influence of IL-1β plus TNFα specifically affected proteoglycan release, whereas proteoglycan synthesis rate was not changed by prostaglandin-E2. Conversely, recent studies showed that the beneficial effect of COX-2 inhibition (indirect prostaglandin-E2 inhibition) in cartilage was mainly reflected by a normalisation of the proteoglycan release, both for cartilage under the influence of pro-inflammatory cytokines and for osteoarthritic cartilage.28 30 34 Therefore, we studied whether proteoglycan synthesis and release were independently regulated, release being prostaglandin-E2-dependent, but not synthesis. Under the influence of IL-1β plus TNFα, proteoglycan synthesis was significantly inhibited (on average 75% inhibition, p<0.02; fig 2A) compared with healthy control cartilage, which could not be reversed by celecoxib. Release of proteoglycans was increased upon addition of IL-1β plus TNFα (on average 40%, p<0.02; fig 2B). Addition of celecoxib reversed this increased release almost completely.

Figure 2 Effect of COX-2 inhibition on proteoglycan synthesis, proteoglycan release, prostaglandin-E2, and NO production of human articular cartilage under the influence of IL-1β plus TNFα. Proteoglycan synthesis rate (A), total release of proteoglycans (B), prostaglandin-E2 production (PGE2, C) production and NO production (NO, D) is depicted. White bars represent the parameters of human healthy cartilage alone, grey bars represent the parameters of the articular cartilage under the influence of IL-1β plus TNFα, and the black bars represent the parameters of the cartilage under the influence of IL-1β plus TNFα in the presence of celecoxib (10 μM) as a selective COX-2 (and with that PGE2) inhibitor. The results are presented as means of at least 8 experiments (viz. 8 cartilage donors)±SEM. Statistically significant differences compared with untreated control cartilage and treated cartilage are shown by p values.

Prostaglandin-E2 production (fig 2C) and NO production (fig 2D) of cartilage were both significantly elevated under influence of IL-1β plus TNFα compared with control. Addition of celecoxib resulted only in a normalisation of prostaglandin-E2 levels, whereas NO levels were not influenced by celecoxib.

Interestingly, when the proteoglycan release was compared with prostaglandin-E2 production, we found a strongly positive correlation (fig 3A): the higher the prostaglandin-E2 concentration the higher the proteoglycan release. There was no correlation found between prostaglandin-E2 production and proteoglycan synthesis rate (fig 3C). However, there was a negative correlation found between proteoglycan synthesis rate and NO production (fig 3B): the higher the NO concentration, the lower the proteoglycan synthesis. In addition, there was also a positive correlation between NO production and proteoglycan release (fig 3D).

Figure 3 Correlations of proteoglycan release with prostaglandin-E2 and proteoglycan synthesis rate with NO production. Correlations between prostaglandin-E2 and proteoglycan release (A) and NO and proteoglycan synthesis (B) are depicted. Additionally, prostaglandin-E2 related to proteoglycan synthesis (C) and NO related to proteoglycan release (D) are depicted. Prostaglandin-E2 is converted to a natural logarithm (ln) to obtain a normal distribution of values. Each data point is the mean of 4 samples. Open circles represent data for human articular cartilage alone, grey dots represent human articular cartilage under the influence of IL-1β plus TNFα, and black dots represent human articular cartilage under the influence of IL-1β plus TNFα in the presence of celecoxib (10 μM).


In the present study, it has been demonstrated that prostaglandin-E2 alone has no effect on the proteoglycan turnover of healthy human articular cartilage. When cartilage is cultured in the presence of IL-1β plus TNFα, adverse changes in proteoglycan turnover and content are induced. When prostaglandin-E2 is added to this sensitised cartilage, prostaglandin-E2 amplifies mostly the adverse changes in proteoglycan release. Conversely, when prostaglandin-E2 production is inhibited (by selective COX-2 inhibition) the adverse effects of IL-1β and TNFα are mostly reduced for proteoglycan release. Most interestingly, for the first time, it is shown that prostaglandin-E2 exerts its effect mainly (exclusively) on proteoglycan release, supported by a clear correlation between the two, and not on proteoglycans synthesis. Proteoglycan synthesis seems to depend much more on NO production.

Recent work by our group and others showed that inhibition of COX-2, and thus inhibition of prostaglandin-E2 production, resulted in an improvement of proteoglycan turnover and hyaluronan synthesis in degenerated cartilage.2830 34 These studies indicate that significant prostaglandin E2 production (up to 200 pg/ml) is involved in disturbing the proteoglycan turnover of degenerated cartilage. However, in the present experiments, no direct catabolic effects of prostaglandin-E2 on healthy cartilage were found, although we used concentrations resembling those found in degenerated cartilage.30 Even higher concentrations (500 pg/ml) did not exert any effect.

A possible explanation for this lack of effect of prostaglandin-E2 on healthy cartilage could be the prostaglandin-E2 receptor profile on healthy chondrocytes by which the biological actions of prostaglandin-E2 are mediated. Prostaglandin-E2 can act through at least four different receptors, and it is not clear which of these contributes to cartilage degeneration. Each of these EP receptors, termed EP1–4, has a distinct pharmacological signature based on its prostaglandin-E2-evoked signal transduction.35 36

Literature is sparse with respect to the role of the different EP-receptors on chondrocytes. It is known that TNFα and IL-1β can upregulate different EP receptors on cultured cells.37 38 The presence of EP1 and 2 receptors has been suggested on rat growth plate chondrocytes, and EP4 receptors have been identified in bovine articular chondrocytes.3941 Another study showed that prostaglandin-E2 contributes to progression of cartilage damage, using an arthritis animal model, at least in part by binding to the EP4 receptor.42 Homozygous deletion of the EP1, EP2 or EP3 receptor did not affect the development of arthritis, whereas EP4-receptor-deficient mice showed decreased joint damage including decreased proteoglycan loss, and type II collagen breakdown in cartilage. These data suggest an important role of the EP4 receptor in cartilage degeneration. Nevertheless, a full evaluation of al subtypes of receptors and their role in articular cartilage has not been caried out, and certainly not for human chondrocytes.

In our study, it was found that adding prostaglandin-E2 to IL-1β plus TNFα sensitised cartilage resulted in a significant amplification of the adverse effects induced by IL-1β plus TNFα. Whether this effect is exerted by upregulated EP4 receptor expression needs further study.

Remarkably, the addition of prostaglandin-E2 to cytokine induced cartilage degeneration has a major effect on the release of proteoglycans, both newly formed and resident, whereas the proteoglycan synthesis rate was hardly influenced by prostaglandin-E2 addition. Conversely, when prostaglandin-E2 production was inhibited (by selective COX-2 inhibition), the adverse effects of IL-1β and TNFα were diminished. Again, the most significant effects were found on release of proteoglycans, and hardly any effects on proteoglycans synthesis were observed.We found a clear correlation between prostaglandin-E2 production and proteoglycan release, whereas the proteoglycan synthesis did not display such a correlation with prostaglandin-E2 production. The importance of prostaglandin-E2 in regulation of proteoglycans release is in concordance with recent studies using a selective COX-2 inhibitor where inhibition of COX-2 resulted in normalisation of proteoglycan release.28 30 34 It is known that prostaglandin-E2 is capable of inhibiting the synthesis of collagens.43 There is also evidence that prostaglandin-E2 has the capacity to activate matrix metalloproteinases (MMPs).44 Prostaglandin-E2 has been reported to stimulate collagenase gene expression in human synoviocytes and to increase MMP-3 production of osteoarthritic cartilage explants.45 46 These effects might play a role in the increased proteoglycan release and decreased proteoglycan content, as found in the present study. Apparently, under IL-1β plus TNFα conditions, prostaglandin-E2 (regulation) has no direct effect on proteoglycan synthesis. There are clearly other, prostaglandin-E2-independent, pathways that regulate chondrocyte proteoglycan synthesis.

Previous studies have shown that NO, being a product of affected chondrocytes as well, exerts a number of effects on synovial cell and chondrocyte functions that may promote degradation of cartilage including inhibition of collagen and proteoglycan synthesis, activation of MMPs and apoptosis.17 4749 Also, the present study showed that there is a clear correlation between the NO production and proteoglycan synthesis and also proteoglycan release. Inhibition of prostaglandin-E2 by using selective COX-2 inhibition had no effect on the NO production. There are a number of reports, confirming that inhibition of prostaglandin-E2 production has no influence on NO production.50 51 However, inhibition of NO (not studied in our set-up) can lead to a change in prostaglandin-E2 production.45 52 53 This may explain the correlation of NO production with proteoglycans release in addition to its correlation with synthesis.

In summary, this study demonstrates the involvement of prostaglandin-E2 in disturbing human articular cartilage turnover, especially proteoglycan release. Additional pathways such as NO production seem necessary to modulate proteoglycan synthesis. The exact mechanism by which prostaglandin-E2 and NO are involved in modulation of cartilage is still speculative, and further studies are needed to unravel these distinct pathways to provide tools that can modulate cartilage degeneration specifically with respect to synthesis. Although selective COX-2 inhibition displayed chondro-reparative activity, this is mostly restricted to inhibition of cartilage matrix proteoglycan release, not to enhancement of proteoglycan synthesis.


This study was supported by an unrestricted grant from Pfizer. Dr FP Lafeber is supported by the Dutch Arthritis Association.



  • Competing interests: None.