N-Acetylglucosamine Prevents IL-1β-Mediated Activation of Human Chondrocytes¹

Osteoarthritis (OA)³ is the most common joint disorder and has an immense socioeconomic impact (1, 2, 3). However, the conservative treatment of OA is still limited to a few classes of medications, such as acetaminophen, nonsteroidal anti-inflammatory drugs, injectable intraarticular corticosteroids, and hyaluronic acid, which provide primarily pain relief, but have not yet been demonstrated to interfere with the progression of the disease (4, 5, 6).

Many studies have demonstrated that cartilage from patients with OA is characterized by accelerated turnover of the cartilage matrix components and by inadequate repair (7, 8). Glucosamine (GlcN) salts (sulfate and chloride) represent a new generation of drugs, which possess potentially chondroprotective or disease-modifying properties (4, 9, 10), and were originally suggested to promote the repair of damaged cartilage. Since the first publication of W. Bohne in 1969 showing that GlcN can be used as a single pharmacologic agent to treat OA (11), the preparation has gained considerable popularity, and now is being consumed by many OA patients. Despite the increased use of GlcN in the treatment of OA, the mechanisms accounting for its in vivo and in vitro activity are still far from clear.

The current study presents experimental evidence that GlcN, and, to a higher degree, N-acetylglucosamine (GlcNAc), possess a unique range of anti-inflammatory activities and inhibit NO, cyclooxygenase-2 (COX-2), and IL-6 production induced in cultured human articular chondrocytes by IL-1β.

Normal cartilage was obtained from autopsy services and tissue banks. Articular cartilage was harvested from the femoral condyles and the tibial plateaus. All tissue samples were graded according to a modified Mankin scale (12), and only cartilage without evidence of OA was used as a source of chondrocytes. The interval between death and the time the cartilage was harvested from these knee joints in the laboratory was at least 24 h and ranged up to 96 h. Cartilage shavings were harvested by the tissue banks within 24 post mortem, placed in tissue culture medium (DMEM, 10% FBS, penicillin, streptomycin), and shipped to the laboratory at 4°C. This tissue was processed in the laboratory within 24 h after harvest.

Chondrocytes were isolated from the cartilage by collagenase digestion and maintained in continuous monolayer cultures in DMEM containing 10% FBS. Cell viability after chondrocyte isolation by collagenase digestion of normal cartilage is >95%. This level is maintained for at least 96 h post mortem. Studies on IL-1 effects as a catabolic response showed no apparent changes as a function of variations in the time between death and tissue processing when NO and IL-6 release were measured.

Experiments reported in this work were performed with primary or first passage cells.

GlcN, GlcNAc, glucose, and glucuronic acid N-acetylmannosamine were purchased from Sigma (St. Louis, MO). GlcNAc dimer (N,N′-diacetylchitobiose) and GlcNAc trimer (N,N′,N″-triacetylchitotriose) were purchased from TRC (Toronto, Canada).

Chondrocytes were plated at 40,000 cells/well in 96-well plates in the presence of 1% FBS. After 48 h, the medium was changed, and the cells were stimulated with IL-1β (Sigma) at a concentration of 5 ng/ml for 24 h. NO production was detected as NO₂⁻ accumulation in the culture supernatants by the Griess reaction, as described elsewhere (13).

IL-6 in the culture supernatants was measured by ELISA (R&D Systems, Minneapolis, MN) in accordance with the supplier’s protocol.

Whole cell extracts were prepared from 3 × 10⁶ chondrocytes stimulated as described in Results by lysing the cells on the plate with ice-cold lysis buffer (10 mM Tris-HCl (pH 7.6), 158 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, 1 μg/ml leupeptin, 1 μg/ml aprotinin, and 0.5 mM PMSF, which was added immediately before use). The lysates were transferred to Eppendorf tubes and centrifuged at 20,000 × g for 30 min at 4°C. The supernatants were transferred into fresh tubes, and the protein concentration was determined by Bradford assay. Similar amounts of protein were separated by 10% SDS-PAGE and transferred to a nitrocellulose filter (Schleicher & Schuell, Keene, NH) by electroblotting. The filter was blocked overnight in 5% milk powder/TBST solution and then further incubated with one of the following Abs: anti-inducible NO synthase (anti-iNOS; C-19; Santa Cruz Biotechnology, Santa Cruz, CA), anti-COX-2 (Cayman Chemical, Ann Arbor, MI), anti-COX-1 (H-3; Santa Cruz Biotechnology), anti-phospho-c-Jun N-terminal kinase (JNK; phospho-Thr¹⁸³/Tyr¹⁸⁵; New England Biolabs, Beverly, MA), anti-phospho-p38 mitogen-activated protein (MAP; phospho-Thr⁸⁰/Tyr¹⁸²; New England Biolabs), or anti-phospho-extracellular signal-regulated kinase (ERK; phospho-Thr¹⁸⁰/Tyr¹⁸²; New England Biolabs) for 2 h. The membranes were washed three times with TBST and then further incubated with the appropriate HRP-labeled secondary Ab in 5% milk powder/TBST and developed using an ECL system (Amersham, Arlington Heights, IL).

Total RNA was isolated from 2 × 10⁶ chondrocytes stimulated as described in Results using the STAT-60 reagent (Tel-Test, Friendswood, TX). The RNA from each sample was quantified photometrically, and 5 μg was separated on 1.2% agarose/6% formaldehyde gels. After electrophoresis, the gels were photographed, and the RNA was transferred onto Hybond-N nylon membranes (Life Technologies, Gaithersburg, MD) by capillary blotting. The membranes were air dried and incubated for 2 h at 80°C. Prehybridization was done for 2 h at 60°C in 5× SCC, 1 mM EDTA, 0.2% SDS, and 5× Denhardt solution. Radiolabeled probe was added and hybridization was conducted overnight at 60°C. After hybridization, the filters were rinsed twice in 2× SSC/0.1% SDS; washed once in 2× SSC/0.1% SDS at 60°C; and once in 0.2× SSC/0.1% SDS at 60°C. The membranes were covered with Saran wrap and exposed with intensifying screen for 12 h at −70°C. The probes used for the hybridization were prepared as described earlier (14, 15).

Nuclear protein extracts were prepared as follows: 2 × 10⁶ chondrocytes were stimulated as indicated in Results. The cells were harvested by trypsinization, washed once with ice-cold PBS, and lysed in 10 mM Tris-HCl buffer (pH 7.5) containing 2 mM MgCl₂, 140 mM NaCl, 0.5 mM DTT, 0.05% Triton X-100, 0.5 mM PMSF, 1 μg/ml leupeptin, and 1 μg/ml aprotinin. The nuclei were spun down, resuspended in 20 mM HEPES buffer (pH 7.9) containing 25% glycerol, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.5 mM DTT, 0.5 mM PMSF, 1 μg/ml leupeptin, and 1 μg/ml aprotinin, and rotated at 4°C for 30 min. After removal of the nuclear debris by centrifugation, the protein concentration of the lysate was determined using the Bradford assay. Equal amounts of the nuclear extracts (2 μg) were incubated for 15 min at room temperature with poly(dI-dC)poly(dI-dC) (0.1 mg/ml), BSA (1 mg/ml), 1 × 10⁵ counts of double-stranded radiolabeled oligodeoxynucleotide containing the NF-κB consensus DNA binding site (sequence: 5′-GATCGAGGGGACTTTCCCTAGC-3′) in 20 mM HEPES buffer (pH 7.9) containing 10% glycerol, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.5 mM DTT, and 0.5 mM PMSF. For competition experiments, unlabeled NF-κB oligodeoxynucleotide or oligodeoxynucleotide containing the Oct-1 consensus sequence was added at 100-fold molar excess to the binding reactions 10 min before the addition of radiolabeled NF-κB oligodeoxynucleotide. The binding reactions were loaded onto 6% TGE (50 mM Tris-HCl (pH 7.5), 380 mM glycine, and 2 mM EDTA) native polyacrylamide gel and electrophoresed for 2 h at 4°C. The gels were then dried and exposed for 16–48 h with intensifying screen at −80°C.

Statistical analysis of the generated data was performed with the aid of StatMost 32 program for Windows (Dataxiom Software, Los Angeles, CA).

IL-1β is known as a potent inducer of NO production in cultured human articular chondrocytes (16). In the first series of experiments, we demonstrated that both GlcN and GlcNAc were capable of suppressing NO production triggered by IL-1β (Fig. 1). The differences between NO production in chondrocyte cultures stimulated with IL-1β and chondrocyte cultures stimulated with IL-1β plus GlcN or GlcNAc were statistically significant, p < 0.001. When used in equimolar concentrations, GlcNAc demonstrated stronger inhibition of NO production than GlcN (the difference between these two groups was statistically significant, p < 0.01). Maximal inhibitory effect of GlcNAc was observed with a concentration of 20 mM; concentrations lower than 1 mM were insufficient in the suppression of NO production (Fig. 2). The IC₅₀ for GlcNAc was 4.1 ± 1.3 mM; the IC₅₀ for GlcN was 14.9 ± 2.1 mM, p < 0.01.

Both GlcN and GlcNAc at doses up to 20 mM did not affect cell viability measured by MTT assay (17) (data not shown).

To analyze the sugar specificity of the discovered phenomenon, we compared the effect of glucose, glucuronic acid, N-acetylmannosamine, N-acetylgalactosamine, GlcNAc, and GlcN on IL-1β-induced NO production. When used at a concentration of 10 mM, only GlcNAc and N-acetylgalactosamine demonstrated inhibitory activity, suggesting specificity of this effect (Fig. 3). GlcNAc polymers, including GlcNAc dimer and GlcNAc trimer, did not express any inhibitory activity against IL-1β-induced NO production (Fig. 4).

To investigate whether GlcNAc suppresses the enzymatic activity of iNOS, the expression of the corresponding protein, we analyzed the effect of GlcNAc on the expression of both iNOS protein (Western immunoblot) and iNOS mRNA (Nothern blot). Results of the experiments clearly demonstrated that GlcNAc strongly inhibited the expression of both iNOS mRNA and protein (Figs. 5 and 6).

As a part of the analysis of its anti-inflammatory activities, we studied the effect of GlcNAc on COX-2 expression in cultured human articular chondrocytes stimulated with IL-1β. Results of the experiments demonstrated that GlcNAc inhibited the expression of COX-2 protein measured in the Western immunoblot and COX-2 mRNA measured in the Northern blot (Figs. 7,A and 5). In contrast to COX-2, GlcNAc did not affect the expression of COX-1 protein (Fig. 7 B).

In addition to NO and COX-2, GlcNAc was capable of inhibiting IL-6 production in cultured human articular chondrocytes stimulated with IL-1β (Fig. 8). The differences were statistically significant (p < 0.001). Therefore, GlcNAc is capable of suppressing several IL-1β-inducible products of inflammation, but does not inhibit constitutively expressed molecules.

Intracellular signaling in the IL-1β pathway results in activation of several protein kinases, including the MAP kinases (18). GlcNAc residues participate in the dynamic process of protein O-glycosylation, which utilizes serine residues as anchoring sites. Therefore, by competing for the same binding sites, O-glycosyl residues could diminish the efficacy of serine phosphorylation and thus interfere with signal transduction. To address this potential mechanism, we analyzed the effect of GlcNAc on ERK, JNK, and p38 MAP kinase activation in chondrocytes induced by IL-1β. The experiments demonstrated that GlcNAc does not inhibit the ERK, JNK, and p38 MAP kinase activation (Fig. 9).

IL-1β-mediated induction of certain mediators of inflammation, including NO, COX-2, and IL-6, is associated with translocation of NF-κB dimers from the cytoplasm to the nucleus, where they bind target genes and regulate their transcription (19, 20, 21). The process of NF-κB activation depends on phosphorylation of two serines (Ser³² and Ser³⁶ in I-κBα (inhibitory protein that dissociates from NF-κB)) in the N-terminal regulatory domain of I-κB (22). To determine whether GlcNAc-mediated suppression of IL-1β-induced NO, COX-2, and IL-6 production depends upon suppression of the NF-κB activation, we studied nuclear translocation of NF-κB in chondrocytes stimulated with IL-1β alone in comparison with chondrocytes stimulated with IL-1β and treated with GlcNAc. These studies demonstrated that GlcNAc did not affect IL-1β-induced nuclear translocation of NF-κB (Fig. 10).

Despite the fact that GlcN represents one of the most commonly used drugs to treat OA, the molecular mechanisms of its activity are still poorly understood. Available experimental data indicate that GlcN possesses both chondroprotective and anti-inflammatory effects. The chondroprotective action of GlcN manifests as accelerated of glycosaminoglycan synthesis in cultured chondrocytes and cartilage tissue. This was demonstrated for cartilage tissue and cells isolated from various species and sources, including chicken embryo cartilage (23), rat acetabular cartilage (24), and chondrocytes from femoral heads of patients with OA (25, 26). In addition, GlcN was shown to restore mechanical properties of the bovine cartilage explants treated with IL-1α (27). GlcN-induced up-regulation of glycosaminoglycan synthesis represents a complex metabolic process, which is potentially mediated through several mechanisms, such as GlcN directly entering the hexosamine pathway and circumventing the negative feedback control from UDP-GlcNAc (28) and up-regulation of TGFβ1 production (29). Recently, a novel mechanism of GlcN-mediated chondroprotection was described, which involves the inhibition of aggrecanase activity in bovine cartilage explants and rat chondrosarcoma cells (30) via suppression of glycosylphosphatidylinositol-linked proteins (31).

Anti-inflammatory mechanisms, besides GlcN-induced up-regulation of glycosaminoglycan synthesis, are probably contributing to its antiarthritic activities as well. GlcN had anti-inflammatory activity and protected rats from paw edema induced by bradykinin, serotonin, and histamine (32). GlcN also protected animals against serositis induced by carragenan, rat peritonitis induced by Formalin, and mouse peritonitis induced by acetic acid (32). GlcN did not suppress COX or proteolytic enzymes in the inflamed rat paw, but it did suppress superoxide generation and lysosomal enzyme activities in rat liver (32). Orally administered GlcN also expressed anti-inflammatory activity in kaolin or adjuvant-induced arthritis in rats (33). However, in the studies cited above, antiexudative and anti-inflammatory activities of GlcN were lower as compared with those of acetylsalicylic acid or indomethacin. GlcN was found to be synergistic in its antiexudative activity with indomethacin, piroxicam, and diclofenac in a mouse model of aseptic inflammation (34).

The present study is the first to examine the effect of GlcN and GlcNAc on human chondrocyte response toward the stimulation with IL-1β, and it describes a novel mechanism of GlcN-mediated anti-inflammatory activity. Results of our experiments clearly indicated that GlcN, and to a higher degree, GlcNAc are capable of inhibiting IL-1β-induced NO production in cultured human articular chondrocytes. The effect of sugars on NO production is specific since several other monosaccharides, including glucose, glucuronic acid, and N-acetylmannosamine do not express this activity. Furthermore, we demonstrated that GlcNAc polymers, including the dimer and the trimer, also do not affect NO production. The observed suppression of IL-1β-induced NO production is the consequence of inhibition of iNOS protein and mRNA expression. In addition to its NO-inhibitory activity, GlcNAc also suppressed the production of IL-1β-induced COX-2 and IL-6. The expression of COX-1, however, was not affected by the sugar. Previously, Setnikar et al. (32) described a negative effect of GlcN on the COX activity of inflamed rat paw tissues. These data do not contradict our results for the following reasons. First, the authors used GlcN and not GlcNAc. Second, the dose of GlcN used to treat rats was much lower than the doses that express anti-inflammatory activities in vitro. Third, the authors did not make any distinction between COX-1 and COX-2.

GlcNAc did not suppress all responses in chondrocytes induced by IL-1β. For example, it did not suppress the IL-1β-mediated increase in hexosaminidase secretion (data not shown). Moreover, it was synergistic with IL-1β in the induction of TGFβ1 (data not shown). Collectively, these findings suggest that GlcNAc selectively inhibits cytokine-induced gene expression and the production of certain proinflammatory mediators.

Several aspects of the discovered GlcNAc-mediated activity require more detailed discussion. Our experiments showed that both GlcN and GlcNAc in the lower millimolar range measurably inhibited NO production; concentrations below 1 mM were not effective. This concentration range is identical to that previously described for GlcN-induced up-regulation of TGF-β production in cultured porcine mesangial cells (29). The relatively high concentrations of GlcN and GlcNAc required for the mediation of their anti-inflammatory activity most likely reflect the competition between these sugars and glucose from culture media for entering the cells via glucose transporter molecules (35). It is important to state that therapeutic concentrations of the aminosugars, which can be reached in humans upon oral administration of GlcN at the accepted dose of 1500 mg/day, are much lower than those used in the present publication. Therefore, the in vitro data regarding the anti-inflammatory mechanisms of GlcNAc and GlcN activities cannot be directly applied for explanation of the therapeutic efficacy of GlcN in patients with OA.

GlcNAc has several potential advantages over GlcN as a potential therapeutic anti-inflammatory agent. First, upon entering the cell, GlcN undergoes phosphorylation by glucokinase and competes with glucose for binding to glucokinase (36), which can result in GlcN-induced insulin resistance (37). GlcNAc, on the other hand, has much lower affinity toward glucokinase as compared with glucose and GlcN, and therefore does not significantly affect glucose metabolism (38). Second, the product of GlcN phosphorylation, GlcN-6 phosphate, is an allosteric inhibitor of glucokinase (39). This limits the flux of GlcN via the hexosamine pathway. Third, upon entering the cell, GlcNAc undergoes phosphorylation by GlcNAc kinase and does not compete with glucose for phosphorylation (40). This product of phosphorylation enters the hexosamine pathway more distally than GlcN-6 phosphate, and does not possess known negative allosteric effects toward glucokinase.

The present study also addressed potential mechanism involved in GlcNAc-mediated inhibition of the IL-1β response. One possibility is a GlcNAc-mediated inhibition of phosphorylation events in the IL-1β signaling cascade. One of the end products of the hexosamine pathway, UDP-GlcNAc, was shown to participate in the dynamic process of protein O-glycosylation, which utilizes serine or threonine residues as anchoring sides (41). Potentially, O-glycosylation of the serine residues can compete with the phosphorylation of the same residues, resulting in the impairment of intracellular signal transduction cascades (42). To address this possibility, we analyzed the effect of GlcNAc on ERK, JNK, and p38 MAP kinase activation, and on nuclear translocation of NF-κB in chondrocytes stimulated by IL-1β. The activation of these MAP kinases and of NF-κB are central events in the chondrocyte response to IL-1β and related cytokines. Results of the experiments revealed that measurable GlcNAc did not inhibit IL-1β-induced activation of ERK, JNK, and p38 MAP kinases, or the nuclear translocation of NF-κB. Additional experiments will be focused on possible targets for O-glycosylation interfering with IL-1β-activated signal transduction cascade.

In conclusion, the study demonstrates that GlcNAc expresses anti-inflammatory and chondroprotective activities by interfering with cytokine-inducible gene expression in chondrocytes.

We thank Diana C. Brinson, Jackie Quach, and Jean Valbracht for excellent technical support.