Differentiation status-dependent regulation of cyclooxygenase-2 expression and prostaglandin E2 production by epidermal growth factor via mitogen-activated protein kinase in articular chondrocytes.

Although large amounts of epidermal growth factor (EGF) are found in the synovial fluids of arthritic cartilage, the role of EGF in arthritis is not clearly understood. This study investigated the effect of EGF on differentiation and on inflammatory responses such as cyclooxygenase-2 (COX-2) expression and prostaglandin E(2) (PGE(2)) production in articular chondrocytes. EGF caused a loss of differentiated chondrocyte phenotype as demonstrated by inhibition of type II collagen expression and proteoglycan synthesis. EGF also induced COX-2 expression and PGE(2) production. EGF-induced dedifferentiation was caused by EGF receptor-mediated activation of extracellular signal-regulated protein kinases 1 and 2 (ERK1/2) but not p38 kinase, whereas the activation of both ERK1/2 and p38 kinase was necessary for COX-2 expression and PGE(2) production. Neither the inhibition of COX-2 expression and PGE(2) production nor the addition of exogenous PGE(2) affected EGF-induced dedifferentiation. However, COX-2 expression and PGE(2) production were significantly enhanced in chondrocytes that were dedifferentiated by serial subculture, and EGF also potentiated COX-2 expression and PGE(2) production, although these cells were less sensitive to EGF. Dedifferentiation-induced COX-2 expression and PGE(2) production were mediated by ERK1/2 and p38 kinase signaling. Our results indicate that EGF in articular chondrocytes stimulates COX-2 expression and PGE(2) production via ERK and p38 kinase signaling in association with differentiation status.

Although large amounts of epidermal growth factor (EGF) 1 are found in the synovial fluid of rheumatoid arthritic cartilage (1,2), the role of EGF in the pathophysiology of arthritic disease is not fully understood. EGF is known to inhibit chondrocyte differentiation in Meckel's cartilage (3) and chondrogene-sis in chick limb bud mesenchymal cells (4,5). EGF is also known to inhibit accumulation of sulfated proteoglycan in articular chondrocytes (6,7), suggesting that EGF not only inhibits chondrocyte differentiation but also causes loss of differentiated chondrocyte phenotype. Articular chondrocytes are characterized by their synthesis of cartilage-specific structural macromolecules such as type II collagen and aggrecan, and loss of differentiated chondrocyte phenotype or dedifferentiation causes cessation of type II collagen expression and induction of fibroblastic types I and III collagens (8). Although the contribution of dedifferentiation in the pathophysiology of arthritis is poorly understood, chondrocyte dedifferentiation is observed in arthritic cartilage, suggesting that EGF contribute to arthritis.
EGF stimulates expression of inflammatory molecules such as cyclooxygenase-2 (COX-2) and prostaglandin E 2 (PGE 2 ) in various cell types including cell lines originating from chondrocytes (9). However, there are no published data regarding EGF effects on COX-2 expression and PGE 2 production in primary culture articular chondrocytes. Although the role of EGF in PGE 2 production is not clear, pro-inflammatory cytokines such as interleukin-1␤ (IL-1␤) cause PGE 2 production, which is found in arthritic joints in large amounts. PGE 2 is a major mediator of cartilage inflammation, cartilage and juxta-articular bone erosion, and angiogenesis (10 -15). The rate-limiting steps in PG synthesis are hydrolysis of phospholipids to produce free arachidonic acid (catalyzed by phospholipase A 2 ) and conversion of arachidonic acid to PGE 2, which is catalyzed by the two isoforms of COX (16,17). COX-1 is constitutively expressed, whereas COX-2 is rapidly induced in response to a wide variety of cytokines such as IL-1␤ and growth factors including EGF. COX-2 expression is regulated at both transcription and post-transcription levels (18 -21). Although the molecular mechanism underlying COX-2 expression is not fully understood, several studies indicate that its expression is regulated by mitogen-activated protein kinase (MAPK) subtypes including extracellular signal-regulated protein kinases 1 and 2 (ERK1/2), p38 kinase, and c-Jun N-terminal kinase, depending on the types of extracellular stimuli and cells (20 -24). In addition to regulation of COX-2 expression and/or activity, ERK and p38 MAPK are involved in the regulation of chondrocyte dedifferentiation caused by serial subculture (25) or nitricoxide production (26,27), suggesting these MAPKs regulate both dedifferentiation and inflammatory responses such as COX expression and PGE 2 production.
The current study investigated the role of EGF in the maintenance of chondrocyte phenotype, COX-2 expression, and PGE 2 production. We also examined the functional relationship between EGF-induced dedifferentiation and COX-2 expression and characterized signaling pathways involved in EGF action. We report that EGF stimulates dedifferentiation, COX-2 expression, and PGE 2 production in articular chondrocytes via ERK1/2 and p38 kinase signaling and that EGF-induced dedifferentiation further potentiates inflammatory responses in articular chondrocytes.

EXPERIMENTAL PROCEDURES
Isolation and Monolayer Culture of Rabbit Articular Chondrocytes-Articular chondrocytes were isolated from cartilage slices of 2-week-old New Zealand White rabbits by enzymatic digestion as described previously (25). Cartilage slices were dissociated enzymatically in 0.2% collagenase type II (381 units/mg, Sigma) in Dulbecco's modified Eagle's medium (Invitrogen). Individual cells were suspended in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) bovine calf serum, 50 g/ml streptomycin, and 50 units/ml penicillin. Cells were plated on culture dishes at a density of 5 ϫ 10 4 cells/cm 2 . The medium was replaced every 2 days, and cells were confluent after approximately 5 days. The 2.5-day cell cultures were treated with EGF (Invitrogen). The following pharmacological agents were added 1 h prior to EGF: PD98059 (Calbiochem) to inhibit MEK1/2 (28); SB203580 (Calbiochem) to inhibit p38 kinase (29); and tyrphostin AG1478 (Biomol, Plymouth Meeting, PA) to antagonize EGF receptors (30). In some experiments, passage (P) 0 cells were subcultured to P3 by plating cells at a density of 5 ϫ 10 4 cells/cm 2 . Differentiation status of articular chondrocytes was determined by examining the accumulation of sulfated glycosaminoglycan with Alcian blue staining or expression of type II collagen by immunoblot analysis as described previously (25).
Three-dimentional Cultures of Chondrocytes-Dedifferentiated chondrocytes at P4 were three-dimensionally cultured in alginate gel beads or as pellets as described previously (25). Cells were suspended in 1.25% sodium alginate prepared in 20 mM HEPES (pH 7.4) containing 0.15 M NaCl at a cell density of 2 ϫ 10 6 cells/ml. The cell suspension was slowly dropped into a gelatin solution (5 mM HEPES (pH 7.4), 102 mM CaCl 2 ). Cells in alginate gel beads were cultured in complete Dulbecco's modified Eagle's medium and refed every other day and were recovered by solubilizing alginate with 2 volumes of 50 mM EDTA and 10 mM HEPES (pH 7.4). For pellet culture, dedifferentiated cells were resuspended in complete Dulbecco's modified Eagle's medium at a density of 2 ϫ 10 5 cells/ml and 1 ml of aliquots was pelleted by centrifugation. Cells in alginate gel beads and pellets were incubated for up to 8 days.
Cartilage Explant Culture and Immunohistochemistry-Rabbit joint cartilage explants (ϳ125 mm 3 ) were fixed in 4% paraformaldehyde for 24 h at 4°C, dehydrated with graded ethanol, embedded in paraffin, and sectioned into 4-m slices as described previously (31). The sections were stained by standard procedures using Alcian blue or antibody against type II collagen or COX-2 and visualized by developing with a kit purchased from DAKO (Carpinteria, CA).
Immunoblot Analysis-Whole cell lysates were prepared by extracting proteins using a buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Nonidet P-40, and 0.1% sodium dodecylsulfate supplemented with protease inhibitors (10 g/ml leupeptin, 10 g/ml pepstatin A, 10 g/ml aprotinin, 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride) and phosphatase inhibitors (1 mM NaF and 1 mM Na 3 VO 4 ). The proteins were size-fractionated by SDS-polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane. The nitrocellulose sheet was then blocked with 3% nonfat dry milk in Tris-buffered saline. COX-2 was detected using antibody purchased from Cayman Chemical (Ann Arbor, MI), and type II collagen was detected using antibodies purchased from Chemicon International Inc. (Temecula, CA). The bands were visualized using peroxidase-conjugated secondary antibodies and chemiluminescence.
MAPK Assay-ERK1/2 activation was examined using immunoblot analysis as described previously (25) using antibodies specific to activated threonine 202-and tyrosine 204-phosphorylated ERK1/2 (Cell Signaling Technology, Beverly, MA). p38 kinase activity was determined by immune complex kinase assays as described previously (26,27). Cells were lysed in a buffer containing 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, and inhibitors of proteases and phosphatases as described above. Total cell lysates were precipitated with polyclonal anti-p38 kinase antibody (Santa Cruz Biotechnology, Santa Cruz, CA), and the immune complexes were collected using protein A-Sepharose beads. The beads were resuspended in 20 l of kinase reaction buffer containing 25 mM Tris-HCl (pH 7.5), 5 mM dithiothreitol, 0.1 mM sodium orthovanadate, 10 mM MgCl 2 , 5 Ci of [␥-32 P]ATP, and 1 g of activating transcription factor-2 protein as a substrate (New England Biolabs, Beverly, MA). The kinase reaction was performed for 30 min at 30°C, and phosphorylated activating transcription factor-2 was detected by autoradiography following gel electrophoresis. PGE 2 Assay-PGE 2 production was determined by measuring the levels of cellular and secreted PGE 2 using an assay kit (Amersham Biosciences). P0 cells or dedifferentiated P4 cells were seeded in standard 96-well microtiter plates at 2 ϫ 10 4 cells/well. After the indicated treatments, total PGE 2 was quantified according to the manufacturer's protocol. PGE 2 levels were calculated against a standard curve of PGE 2 and normalized against the amount of genomic DNA.

EGF Causes Dedifferentiation of Articular Chondrocytes-To
examine the effects of EGF on articular cartilage chondrocyte differentiation, cartilage explant cultures were treated with 10 ng/ml EGF for 48 h, and expression of cartilage-specific matrix molecules were determined. EGF caused a dramatic loss of type II collagen and sulfated proteoglycan as determined by immunohistochemical staining and Alcian blue staining, respectively (Fig. 1A). EGF also inhibited type II collagen expression in primary culture articular chondrocytes in a dose-and time-dependent manner, and this inhibition was blocked when cells were pretreated with the EGF receptor antagonist AG1478 ( Fig. 1B). Similarly, EGF treatment of primary culture cells blocked the accumulation of sulfated proteoglycan in a dose-dependent manner (Fig. 1C), and the inhibition was also prevented by AG1478 (Fig. 1D). These results indicate that EGF induces dedifferentiation of articular chondrocytes in both cartilage explants and primary culture cells.
EGF-induced COX-2 Expression and PGE 2 Production-The effect of EGF on COX-2 expression and PGE 2 production was investigated using both cartilage explant cultures and primary culture chondrocytes. In cartilage explants, EGF stimulated COX-2 expression as determined by immunohistochemical staining (Fig. 2A). In primary culture chondrocytes, EGF increased protein levels of COX-2 in a time-and dose-dependent manner as determined by immunoblot analysis (Fig. 2B). COX-2 was detected 1 h after EGF treatment, and levels peaked at 12 h. The EGF effects on primary culture chondrocytes were completely blocked by AG1478 pretreatment (Fig.  2B). Consistent with the induction of COX-2 expression, EGF stimulated PGE 2 production (Fig. 2C) that was detectable 1 h after EGF addition (Fig. 2D) and was blocked by AG1478 (Fig.  2E). These data indicate that EGF not only causes dedifferentiation of articular chondrocytes but also stimulates COX-2 expression and PGE 2 production.
Differentiation Status-dependent COX-2 Expression and PGE 2 Production-Because EGF in articular chondrocytes causes both dedifferentiation and COX-2 expression/PGE 2 production, we next examined the functional relationship between EGF-induced dedifferentiation and inflammatory responses. Differentiation status-dependent COX-2 expression and PGE 2 production were examined first. Serial subculturing of chondrocytes caused dedifferentiation as demonstrated by cessation of type II collagen expression (Fig. 4A). This chondrocyte phenotype loss was accompanied by significantly increased levels of COX-2 protein (Fig. 4A) and PGE 2 production (Fig. 4B). When these dedifferentiated P3 cells were redifferentiated into chondrocytes by three-dimensional culture in alginate gel (Fig.  4C) or by pellet culture (Fig. 4D), COX-2 protein levels dropped to those observed in differentiated P0 cells. Therefore, COX-2 expression and PGE 2 production were inversely related to the differentiation status of articular chondrocytes.
COX-2 Expression and PGE 2 Production Do Not Mediate EGF-induced Dedifferentiation-Because PGE 2 production is known to modulate chondrocyte differentiation (32-34), we next examined whether COX-2 expression and the resulting PGE 2 production in EGF-treated cells are involved in dedifferentiation of chondrocytes. The COX-2 inhibitors indomethacin or NS398 (35) significantly blocked EGF-induced PGE 2 production as expected (Fig. 5A). However, EGF-induced inhibition of sulfated proteoglycan accumulation (Fig. 5B) and type II collagen expression (Fig. 5C) were not affected by indomethacin or NS398. In addition, treatment of P0 cells with exogenous PGE 2 did not significantly affect proteoglycan synthesis (Fig. 5B) or type II collagen expression (Fig. 5C). These data indicate that COX-2 expression and PGE 2 production are not involved in EGF-induced dedifferentiation of chondrocytes.
Dedifferentiation of Chondrocytes Potentiates COX-2 Expression and PGE 2 Production-We examined the effect of EGF on COX-2 expression and PGE 2 production in dedifferentiated cells. We found that EGF further increased COX-2 expression (Fig. 6A) and PGE 2 production in dedifferentiated P3 cells (Fig.  6B). Although the sensitivity of dedifferentiated chondrocytes to EGF was much less than that of P0 cells, the levels of COX-2 and PGE 2 were much higher in dedifferentiated chondrocytes. In an attempt to reveal the mechanisms underlying increased expression of COX-2 and PGE 2 production in dedifferentiated cells, the differentiation status-dependent activity of MAPK subtypes was examined. Serial subculture of chondrocytes to P3 caused sustained activation of both ERK1/2 and p38 kinase (Fig. 7A). The inhibition of ERK activity with PD98059 but not p38 kinase activity with SB203580 partially maintained type II collagen expression in P3 cells (Fig. 7B). Blocking ERK activity did not significantly affect dedifferentiation-induced increase of COX-2 protein level (Fig. 7B), but it did inhibit COX-2 activity as demonstrated by the inhibition of PGE 2 production (Fig. 7C). The inhibition of p38 kinase with SB203580 blocked both dedifferentiation and EGF-induced potentiation of COX-2 expression (Fig. 7B) and PGE 2 production (Fig. 7C). The above results suggest that dedifferentiation-induced activation of ERK1/2 and p38 kinase causes increased COX-2 expression and PGE 2 production in articular chondrocytes. DISCUSSION Chondrocytes in normal articular cartilage are a unique cell type in that their differentiated phenotype is reversible. Chondrocyte phenotype is regulated by a balance of anabolic and catabolic molecular reactions that are involved in maintaining homeostasis of cartilage tissue (8). Differentiated chondrocytes lose their phenotype and transform into fibroblast-like cells upon exposure to soluble factors such as IL-1␤ (36), retinoic acid (31,37), and nitric oxide (26,38) or during serial subculture in vitro (25,39). Such a destruction of homeostasis is believed to be involved in the pathophysiology of arthritis (8,40). Because large amounts of EGF are found in the synovial fluid of arthritic cartilage (1,2), it is likely that EGF contributes to the disease, although the role of EGF is poorly understood. In this study, we demonstrated that EGF caused both dedifferentiation and inflammatory responses such as COX-2 expression and PGE 2 production. We also demonstrated that dedifferentiation of chondrocytes by serial subculture caused significant increases in COX-2 expression and PGE 2 produc- 3. ERK1/2 and p38 kinase regulate dedifferentiation and COX-2 expression/PGE 2 production. A, chondrocytes were treated with 5 ng/ml EGF for the indicated periods (upper panel), or cells were treated with or without 250 nM AG1478 (AG) for 30 min prior to EGF treatment (lower panel). ERK1/2 phosphorylation was detected using immunoblot analysis with anti-phospho-ERK1/2 antibodies. p38 kinase activity was determined by immune complex kinase assay using activating transcription factor-2 as a substrate. Expression of ERK1/2 and p38 kinase was determined by immunoblot analysis. B and C, articular chondrocytes were treated with 5 ng/ml EGF in the absence or presence of the indicated concentrations of PD98059 (PD) or SB203580 (SB). ERK phosphorylation, type II collagen (Coll-II) and COX-2 were detected by immunoblot analysis. p38 kinase activity was determined by immune complex kinase assay using activating transcription factor-2 as a substrate (B). Accumulation of sulfated glycosaminoglycan was quantified by Alcian blue staining (C). ERK phosphorylation and p38 kinase activity were measured after cells were treated with EGF for 30 min, whereas the expression of type II collagen, COX-2, and sulfated glycosaminoglycan were determined after cells were treated with EGF for 48 h. Con, control. D, chondrocytes were untreated or treated with 20 M PD98059 or SB203580 and exposed to 5 ng/ml EGF for 24 h. PGE 2 levels were determined using an assay kit. The data are from a typical experiment (A and B) or are mean values Ϯ S.D. (C and D) from more than four independent experiments.

FIG. 4. Differentiation status dependent COX-2 expression and PGE 2 production.
A, P0 chondrocytes were serially subcultured to P3, and levels of type II collagen and COX-2 were determined by immunoblot analysis. B, PGE 2 production was compared between P0 cells and P3 cells by using an assay kit. C and D, dedifferentiated P3 cells were cultured three-dimensionally in alginate beads (C) or in pellets (D) for up to 8 days. Type II collagen and COX-2 expressions were determined by immunoblot analysis. The data represent typical results or mean values Ϯ S.D. from more than four independent experiments. M, monolayer. tion, even though COX-2 expression and PGE 2 production did not affect differentiation status. Therefore, our results indicate that EGF-induced dedifferentiation of articular chondrocytes amplifies EGF-induced inflammatory responses and suggest that dedifferentiation of chondrocytes may worsen arthritic cartilage inflammation.
A significant finding of this study is that EGF caused both dedifferentiation and COX-2 expression/PGE 2 production in articular chondrocytes and that dedifferentiation itself also resulted in these COX-2 and PGE 2 responses. Previous reports (41,42) indicate that dedifferentiation of chondrocytes caused reduction in responsiveness to the inflammatory cytokines such as IL-1␤ with respect to COX-2 expression and PGE 2  7. ERK1/2 and p38 kinase mediates dedifferentiation-induced increase of COX-2 expression and PGE 2 production. A, chondrocytes were subcultured to P3, and ERK1/2 and p38 kinase activity were determined by immunoblot analysis and immune complex kinase assay, respectively. B and C, differentiated P0 chondrocytes or dedifferentiated P3 cells were either untreated (Con) or treated with 20 M PD98059 (PD) or 20 M SB203580 (SB) for 1 h and then exposed to vehicle alone or 5 ng/ml EGF for 24 h. Type II collagen and COX-2 were detected using immunoblot analysis (B). PGE 2 production was measured using an assay kit and normalized by determining the amount of total genomic DNA. The data represent results of a typical experiment or mean value Ϯ S.D. from at least four independent experiments. production. Our current results also indicate that dedifferentiated cells are much less sensitive to EGF-induced increases in PGE 2 production. For instance, PGE 2 production increased Ͼ5-fold in differentiated P0 cells following EGF treatment, whereas only a ϳ2-fold increase was observed in dedifferentiated P3 cells (Fig. 6). The amount of PGE 2 produced per well by dedifferentiated cells either in the absence or presence of EGF was much less than that observed in differentiated P0 cells. However, when the amount of PGE 2 produced was normalized against genomic DNA to quantify PGE 2 production per cell, significantly higher amounts of PGE 2 were produced by dedifferentiated cells (Fig. 6). Similar results were observed when chondrocytes were treated with IL-1␤ (data not shown). Therefore, our observations clearly indicate that dedifferentiated chondrocytes produce more PGE 2 , although sensitivity to EGF or IL-1␤ is much lower.
In addition to PGE 2 production, the levels of COX-2 protein were much higher in serially subcultured dedifferentiated cells compared with differentiated P0 cells (Fig. 4). We also observed that dedifferentiation of chondrocytes by exposure to phorbol ester, nitric oxide, or retinoic acid caused significantly elevated expression of COX-2 (data not shown). The results of this study are different to those reported by Thomas et al. (42) in which IL-1␤-treated differentiated and dedifferentiated immortalized chondrocytes showed similar COX-2 mRNA levels. This discrepancy may be the result of different cell origins and culture systems. Those studies compared responses of an immortalized human articular chondrocyte cell line with responses of cells cultured as monolayers and three-dimensionally. This study used primary rabbit articular chondrocytes cultured in monolayers. Indeed, although EGF is known to induce COX-2 expression in various cell types including cell lines originating from chondrocytes (9), no data are available regarding the responses of primary culture articular chondrocytes to EGF. Thus, our current observation is the first to characterize the effect of EGF on primary chondrocytes.
In this study, we also demonstrated that EGF-induced COX-2 expression and PGE 2 production is regulated by ERK1/2 and p38 kinase signaling. The regulation of COX-2 expression has been shown to occur at both transcriptional and post-transcriptional levels, and structural information on the COX-2 gene promoter revealed binding sites for several transcription factors found in many cell types such as NFB and CAAT enhancer-binding protein, which are involved in transcription regulation of COX-2 (18,19,22,23). The stability of COX-2 mRNA regulated by sequences within the 3Ј-untranslated region is also an important regulatory mechanism of COX-2 expression (20,21). In addition, several studies linked COX-2 expression with MAPK subtypes including ERK1/2, p38 kinase, and c-Jun N-terminal kinase (20 -24). Therefore, our observation that EGF-induced expression and activity of COX-2 are regulated by ERK1/2 and p38 kinase signaling is consistent with the observations of others (20 -24). The increased COX-2 expression and PGE 2 production in dedifferentiated chondrocytes are mediated by the dedifferentiation-dependent activation of ERK1/2 and p38 kinase (Fig. 7). It has been shown that dedifferentiation of chondrocytes by exposure to IL-1␤ r retinoic acid significantly increased the number of cell surface EGF receptors without changing receptor affinity (43). This finding suggests that dedifferentiating chondrocytes by serial subculture may increase EGF receptor expression and serum factors such as EGF may stimulate COX-2 expression and PGE 2 production.
The effect of PGE 2 on chondrocytes depends on the culture system, microenvironment, and physiological conditions (14,33). PGE 2 exerts anabolic effects such as synthesis of proteo-glycan and type II collagen and catabolic effects such as enhancing matrix degradation (14,15). For example, PGE 2 has been shown to promote chondrocyte differentiation in addition to its role in inflammation by increasing type II collagen expression (32)(33)(34). However, the expression and activation of COX-2 and resultant PGE 2 production are also believed to contribute to cartilage destruction by altering matrix degradation via matrix metalloproteinases (15). Indeed, the administration of a COX-2 inhibitor can repress joint inflammation and cartilage destruction in animal models of arthritis (44,45). In our culture system, we observed that neither the inhibition of COX-2 expression and PGE 2 production nor the addition of exogenous PGE 2 affected EGF-induced modulation of chondrocyte phenotype, whereas dedifferentiation potentiated inflammatory responses, i.e. COX-2 expression and PGE 2 production. Therefore, it is likely that EGF-induced dedifferentiation further elevates COX-2 expression and PGE 2 production and that EGF found in arthritic cartilage synovial fluid contributes to inflammation and cartilage destruction during arthritic disease.