Basic Fibroblast Growth Factor Activates the MAPK and NFκB Pathways That Converge on Elk-1 to Control Production of Matrix Metalloproteinase-13 by Human Adult Articular Chondrocytes*

The pathology of joint destruction is associated with elevated production of basic fibroblast growth factor (bFGF) and matrix metalloproteinase-13 (MMP-13). In osteoarthritic joint disease, expression of bFGF and MMP-13 in chondrocytes and their release into the synovial fluid are significantly increased. We have previously found that the capacity for cartilage repair in human adult articular chondrocytes is severely compromised by minimal exposure to bFGF because bFGF reduces responsiveness to bone morphogenetic protein-7 and insulin-like growth factor-1 and induces MMP-13 through protein kinase Cδ-dependent activation of multiple mitogen-activated protein kinase (MAPK) signaling pathways. Here we show using biochemical and molecular approaches that transcription factor Elk-1, a direct downstream target of MAPK, is a critical transcriptional activator of of MMP-13 by bFGF in human articular chondrocytes. We also provide evidence that Elk-1 is a direct target of NFκB and induces MMP-13 expression upon activation of the NFκB signaling pathway. Taken together, our results suggest that elevated expression of MMP-13 occurs through Elk-1 activation of both MAPK and NFκB signaling pathways, thus revealing a two-pronged biological mechanism by which bFGF controls the production of catabolic enzymes that are associated with excessive degradation of the cartilage matrix in degenerative joint diseases such as osteoarthritis.

physiological significance and potential clinical impact on cartilage homeostasis, comparatively little is known about the role of growth factors, cytokines, and inflammatory mediators in the pathogenesis of the disease.
Articular cartilage is a relatively uncomplicated tissue composed of chondrocytes and extracellular matrix (ECM). Despite its limited cell diversity and absence of vasculature and innervation, it is a dynamic tissue in which maintenance of the ECM depends on the delicate and physiologically regulated balance between anabolic and catabolic responses of sparsely distributed chondrocytes in normal cartilage. However, in OA, chondrocyte metabolism is unbalanced due to excessive production of matrix metalloproteinases (MMPs), aggrecanases (ADAMTS), and other proteinases by chondrocytes (1,2). Among the MMPs, MMP-1 (collagenase-1) and MMP-13 (collagenase-3) play a significant role in the development of OA, because they are rate-limiting enzymes involved in collagen degradation (1,(3)(4)(5). MMP-1 is produced primarily by synovial cells, whereas MMP-13 is almost exclusively produced by chondrocytes that reside in the cartilage. In addition to collagen, MMP-13 also degrades the proteoglycan molecule aggrecan, suggesting a potential dual role in matrix destruction (5)(6)(7). Given the important role of MMP-13, it is vital to elucidate the key regulators and signaling pathways that control MMP-13 gene expression. Defining this mechanism is necessary to understand the molecular etiology of OA and to design new strategies for the development of therapautics to prevent the joint destruction seen in OA.
Basic FGF (or FGF-2) is a member of a large FGF family of structurally related proteins that bind heparin or heparin sulfate and modulate the growth, differentiation, migration, and survival of a wide variety of cell types (8). Basic FGF (bFGF) is expressed by chondrocytes and is stored in the ECM of adult articular cartilage (9,10). The mitogenic capacity of bFGF in growth plate cartilage (11,12) and adult articular cartilage (13) is well known. Nevertheless, previous studies on bFGF from a variety of species have yielded inconsistent results with regards to production of ECM and articular cartilage homeostasis. We recently showed that bFGF has a striking antagonistic effect on the well known cartilage anabolic activities of insulin-like growth factor-1 and osteogenic protein-1, as reflected by inhibition of proteoglycan synthesis and its deposition (14). Signals generated by bFGF stimulate MMP-13 expression in several cell types, including cultured human chondrosarcoma cells (15) and osteoblasts (16), as well as human adult articular chondrocytes (17,18), and the induction is at the transcriptional level (18). Hence, bFGF stimulation of articular chondrocytes may disturb cartilage homeostasis by increasing MMP-13 expression and ECM degradation. Despite these findings, though, the transcriptional mechanisms by which bFGF stimulates MMP-13 expression are poorly understood.
Immediate release of bFGF upon mechanical injury to cartilage leads to the rapid and sustained activation of Erk, and it occurs concomitantly with striking changes in chondrocyte gene expression (9,10). Recently, we have reported that the activation of multiple MAPK subgroups (Erk, p38, and JNK) is required for the bFGF-mediated stimulation of MMP-13 in human adult articular chondrocytes (18). Elk-1 is a direct downstream target gene of MAPK and is a member of the Ets domain-containing ternary complex factor subfamily of transcription factors (19 -23). bFGF-initiated transcriptional repression of elastin gene expression, an ECM protein produced by lung fibroblasts, has been reported to be mediated via the activation of Elk-1 (24). However, the precise physiological role of this transcription factor in chondrocytes has not yet been characterized after stimulation with bFGF. Similarly the critical role of the NFB pathway in the stimulation of MMP-13 production by various stimuli such as IL-1␤, fibronectin fragment, and hyaluranon oligonucleotide has been suggested in previous studies (25)(26)(27). Nevertheless, the precise cellular and molecular mechanism by which the NFB pathway stimulates MMP-13 is not clearly understood.
In this study, we investigated the concept that the transcriptional activities of Elk-1 and NFB may together control ECM homeostasis in cartilage. We identified Elk-1 as a key transcriptional regulator of MMP-13 expression by a two-pronged mechanism, because it acts both directly and indirectly. Our data show that activation of MMP-13 transcription occurs via Elk-1 directly, as well as via the NFB pathway, in which Elk-1 plays a role as an intermediate regulatory molecule to lead to the NFB-dependent induction of MMP-13 expression. These findings provide a mechanistic understanding of how the release of growth factors and inflammatory cytokines upon cartilage damage can result in the sustained activation of MMPs that may cause further destruction of articular cartilage.

EXPERIMENTAL PROCEDURES
Chondrocyte Isolation and Culture Conditions-Normal human ankle cartilage was obtained from tissue donors through the Gift of Hope Organ and Tissue Donor Network. Each donor specimen was graded for gross degenerative changes based on a modified version of the 5-point scale of Collins (28). Chondrocytes were isolated by enzymatic digestion of ankle articular cartilage (grade 0 or 1, which has no sign of cartilage degeneration) using Pronase followed by overnight digestion with collagenase-P as described previously (18). Isolated cells were resuspended in media at 2 ϫ 10 6 per milliliter and plated onto 12-well plates at 1 ml/well. Cells were cultured in Dulbecco's modified Eagle's medium/F-12 containing 10% fetal bovine serum and antibiotics (complete media) for 5 days before the experiments.
Chondrocyte Stimulation and Immunoblotting-Cells were serum-starved by changing media to serum-free Dulbecco's modified Eagle's medium/F-12 with antibiotics for 1 day. For inhibitor studies, cells were preincubated with individual pathway-specific chemical inhibitors for 30 min before stimulation with bFGF or IL-1␤. Experiments were terminated with removal of media and/or cell lysate preparation. The conditioned media was stored at 4°C with 0.1% NaN 3 and used for the experiments within 5 days. Cell lysates were prepared using modified cell lysis radioimmune precipitation assay buffer: 20 mM Tris (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Nonidet P-40, 0.25% deoxycholate, 2.5 mM sodium pyrophosphate, 1 mM glycerol phosphate, 1 mM NaVO 4 , with 2 mM phenylmethylsulfonyl fluoride (Sigma). Total protein concentrations of both media and cell lysates were determined by a bicinchoninic acid (BCA) protein assay (Pierce). Equal amounts of protein were resolved by 10% SDS-PAGE gels and transferred to nitrocellulose membrane for immunoblot analysis as described previously (18). Immunoreactivity was visualized using the ECL system (Amersham Biosciences) and the Signal Visual Enhancer system (Pierce), which magnifies the signal.
Transient Transfection by Nucleofection (Electroporation)-Nucleofection methods were optimized for use with human articular chondrocytes by minor modifications of the instructions for the Nucleofector TM kit (Amaxa Biosystems) that have been described previously (18,32). The Renilla vector (pRL-TK) was co-transfected as an internal control, and the luciferase activity representing promoter activity was measured using the Dual-Luciferase reporter assay system (Promega) and a lumi-bFGF Stimulates MMP-13 via Elk-1 Activation nometer (Berthold). In samples containing combinations of plasmids (i.e. co-transfections with Elk-1 or pCMV-IB␣M cDNA construct with MMP-13 promoter/luciferase reporter construct) we adjusted the total amount of DNA concentration to Ͻ5 g per 100 l of cell-Nucleofector solution complex for the entire set of experiments to minimize toxic effects observed at higher DNA concentrations. To introduce siRNA-targeting Elk-1, the validated TranSilent TF siRNAs targeting Elk-1 and GAPDH were purchased from Panomics (Fremont, CA) or SuperArray Bioscience Co. (Frederick, MD) and were introduced into human articular chondrocytes using nucleofection (as described above) by following the manufacturer's protocol. The effect of Elk-1 siRNA on bFGF-mediated MMP-13 expression was assessed by immunoblotting using anti-MMP-13 (R&D System) or anti-GAPDH antibody (Abcam, UK).
Reverse Transcription and PCR-Reverse transcription (RT) was carried out with 500 ng of total cellular RNA using either real-time or the One-Step RT-PCR System (Invitrogen) following the instructions provided by the manufacturer. For all experiments, optimal conditions were determined by initially generating cycle number-dependent expression curves and the linear range for the PCR amplification. The same amounts of total RNA or genomic DNA were subjected to One-Step RT-PCR simultaneously to minimize experimental variation due to differences in amplification efficiency. The assessment for GAPDH was performed in parallel (18,25). The One-Step RT-PCR was performed using 30 cycles of 95°C for 30 s, 62°C for 1 min, and 72°C for 40 s in the presence of 50 pmol of sense and antisense primers. The primer sequences and the conditions for their use are summarized in Table 1. The resulting PCR products were resolved in 1.5% agarose gels and visualized by staining with ethidium bromide and UV transillumination. Integrated density values for testing genes were normalized relative to GAPDH values to yield a semi-quantitative assessment. For real-time PCR, the primer sets specific to human genes and optimized conditions were used to generate Ͻ200-bp coding regions, spanning at least one exon and an intron as we performed previously (18).
Preparation of Nuclear Extracts-The nuclear extracts were prepared by using the nuclear extraction kit (Panomics) according to the manufacturer's protocol. Following stimulation, culture media was removed, and the cells were then lysed with 1 ml buffer A, incubated on ice for 10 min, and centrifuged at 15,000 ϫ g at 4°C for 3 min. The supernatants were discarded, and the pellets were resuspended in 150 l of buffer B. The samples were incubated on ice for 2 h and centrifuged, and the supernatants were aliquoted and stored at Ϫ80°C.
Gel Shift Assays (Electrophoretic Mobility Shift Assays)-Gel shift assays of Elk-1 and NFB were performed by using the commercially available kits (Panomics). Following stimulation, the nuclear extracts (prepared as mentioned above) were incubated with the biotin-labeled double-stranded transcription factor (Elk-1 and NFB) either consensus sequence or the sequence present in MMP-13 gene. The samples were then resolved on a 6% non-denaturing polyacrylamide gel and transferred to Pall Biodyne B membrane, blocked, and incubated with Streptavidin-horseradish conjugate. The bands were visualized with a chemiluminescence-imaging system. Unlabeled, double-stranded wild-type oligonucleotides were used to determine the binding specificity of the assay.
Nuclear Translocalization of Elk-1-Chondrocytes plated at a density of 100,000 in 4-well chamber slides were cultured for 24 h in complete media containing 10% fetal bovine serum. The medium was then changed to serum-free medium for 18 h. Chemical inhibitors were incubated for 1 h prior to treatment with bFGF (100 ng/ml). The treatment was then aborted by washing the cells with 1ϫ PBS three times followed by fixation in 4% paraformaldehyde for 15 min at room temperature. The cells were then washed with 1ϫ PBS and incubated with 0.2% Triton X-100 prepared in PBS for 10 min at room temperature. After permeabilizing the cells with Triton X-100, the cells were washed and blocked with 10% normal goat serum in PBS for 60 min at room temperature. After washing the cells with PBS, the primary antibody prepared in 1% bovine serum albumin made in PBS (1:200 dilution) was then added, and the mixture was incubated overnight at 4°C. The cells were then washed and incubated with green fluorescent-labeled anti-mouse secondary antibody (prepared in 1% bovine serum albumin-PBS) for 1 h at 4°C in the dark. The cells were then washed, the slides were prepared using the mounting media containing 4Ј,6-diamidino-2-phenylindole (Molecular Probes), which can counterstain the nucleus, and visualized using the fluorescent microscope (Nikon Eclipse E600 microscope) connected to a PC running MetaView Imaging Series 6.1. The nuclear image was visualized using a UV filter, whereas the cytoplasmic fluorescence staining was visualized by using a blue filter. Both images were overlaid using MetaView imaging software. Histology-Safranin Orange staining was performed using full thickness cartilage slices prepared from OA knee joint tissues and a 4-mm diameter puncher. These explants were cultured in 1.0 ml of Dulbecco's modified Eagle's medium/F-12 containing 10% fetal bovine serum. Following a 2-day recovery period, the cartilage explants were treated with or without bFGF (100 ng/ml) under serum-free conditions (mini-ITS). Following 14 days of incubation, the slices were fixed with 4% paraformaldehyde for overnight and embedded in paraffin, and 8 M sections were prepared. The paraffin sections were deparaffinized and stained with Safranin Orange staining to assess matrix proteoglycan loss.
Type II Collagen Digestion Assay-75 ml of culture media from each sample to be tested was activated with 1 M 4-aminophenylmercuric acetate (Sigma, catalog no. A9563) for 1 h at 37°C in the dark. After activation, 75 ml of 0.5 mg/ml chick sternal type II collagen (Sigma, catalog no. C9301) was added to the activated culture media, and the mixture was vortexed thoroughly followed by placing the mixture of culture media and type II collagen (150 l) in a heating block for 24 h at 37°C. The reaction was terminated by the addition of 150 l of 2ϫ reducing Laemmli sample buffer (Bio-Rad, catalog no. 161-0737), and the samples were then resolved on 8% SDS-PAGE followed by staining with Coomassie Brilliant Blue R-250 for 2 h. The gel was destained with water containing 10% methanol and 10% acetate acid.
Statistical Analyses-The statistical significance of results was determined by analysis of variance, using StatView 5.0 software (SAS Institute, Cary, NC). Data interpretation was performed by statistical normalization as assessed with histograms, and a 0.05 significance level was used for all statistical tests.

Identification of an Elk-1 Response Element in the Distal Promoter Region of the MMP-13 Gene That Regulates MMP-13
Expression in Human Adult Articular Chondrocytes-Our previous studies suggest that bFGF-mediated up-regulation of MMP-13 occurs via a transcriptional stimulation (18). To define elements controlling MMP-13 transcription, MMP-13 promoter-luciferase reporter constructs were transiently trans-fected into primary human articular chondrocytes. Deletion from Ϫ1600 (Ϫ1600MMP-13) to Ϫ736 (Ϫ736MMP-13) upstream of the ATG of the MMP-13 promoter region shows no significant change in the basal promoter activity (Fig. 1A), consistent with our previous report with an immortalized chondrocyte cell line (30,32). Further deletion from Ϫ736 to Ϫ370 of the MMP-13 promoter (Ϫ370MMP-13) significantly reduces basal promoter activity (up to ϳ40% reduction). This  decreased level of promoter activity is similar to the activity observed for the Ϫ186MMP-13 promoter that contains minimal core transcription factor response elements, including motifs for Runx2, ETS, and AP-1. These results suggest that the distal Ϫ736 to Ϫ370 promoter region of the MMP-13 gene encompasses key elements that regulate MMP-13 promoter activity in human adult articular chondrocytes. Bioinformatic analysis reveals the presence of an Elk-1 recognition motif (5Ј-CAGGAC) between positions Ϫ636 and Ϫ630 of the MMP-13 promoter region adjacent to putative serum response elements. Because bFGF activates the MAPK-AP-1 pathway to stimulate MMP-13 expression in human adult articular chondrocytes (18), and because Elk-1 is an MAPKresponsive factor in other biological contexts (19 -24), we examined the role of Elk-1 in MMP-13 expression. Cells were transiently co-transfected with the Ϫ1600MMP-13 promoterreporter construct, which contains the putative Elk-1 motif, and the Ϫ370MMP-13 construct, which lacks the Elk-1 binding site, in the presence or absence of a wild type Elk-1 expression vector. Forced expression of Elk-1 to mimic the activation of bFGF-signaling cascades significantly enhances the activity of the Ϫ1600 promoter by 2-fold but has no appreciable effect on the Ϫ370 promoter (Fig. 1B). Cells transfected with the empty expression vector were analyzed as negative controls (Fig. 1B, lane 1), whereas a c-fos-responsive promoter-reporter construct (c-fos-Luc) shows significant enhancement of luciferase activity (Fig. 1B, lane 7), which is consistent with the stimulation of c-fos expression by Elk-1 (24, 34 -36).
To examine whether Elk-1 levels are critical for the control of MMP-13 expression, we performed siRNA experiments using a commercially available Elk-1 siRNA (siElk-1, Panomics). Human primary articular chondrocytes transfected with siElk-1 were incubated in the absence or presence of bFGF. Cells with bFGF stimulation increase both MMP-13 and Elk-1 protein production. The introduction of siElk-1 significantly blocked these up-regulations in a dose-dependent manner while the level of internal control GAPDH remained unchanged (Fig. 1C, upper panel). Introduction of siRNA-targeting GAPDH was performed in parallel as a control in which siRNA-targeting GAPDH had no effect on bFGF stimulation of MMP-13 and Elk-1, whereas the GAPDH protein level was decreased dose dependently (lower panel). Our results suggest that Elk-1 plays a critical role as a primary transcriptional regulator of MMP-13 transactivation after stimulation with bFGF in human adult articular chondrocytes.
bFGF-mediated Phosphorylation of Elk-1 at Serine 383 Is Linked to Biological Activation of Elk-1 in Human Adult Articular Chondrocytes-Elk-1 was rapidly activated by bFGF (within 5 min), and the activation was sustained beyond 60 min as reflected by phosphorylation of Elk-1 at the amino acid residue serine 383 ( Fig. 2A). We examined the transcriptional activating ability of Elk-1 using a GAL4-responsive reporter gene in which multimerized GAL4 sites drive luciferase, and chimeric Elk-1 proteins fuse to the GAL4 DNA binding domain (GAL4 BD) (Fig. 2B). Human primary chondrocytes co-transfected with the GAL4-responsive promoter construct (GAL4 RE-Luc), and expression vectors for the GAL4-Elk-1 fusion protein (GAL4-Elk-1 1-428 wt) did not exhibit activation of the reporter activity (Fig. 2, B and C) until stimulation of cells with bFGF ( Fig. 2, B and D). In the presence of bFGF, the luciferease activity was significantly increased, reflected by an ϳ4-fold induction of GAL4 promoter activity, suggesting the stimulation of intrinsic transcriptional activity of Elk-1 via the biological action of bFGF. Co-expression of a vector containing only GAL4 BD with GAL4 RE-Luc was used as a negative control in the experiments (Fig. 2B, panels a and b).
The phosphoryaltion of Elk-1 at the serine 383 residue appeared to be essential for the transcriptional activating ability of Elk-1 in human adult articular chondrocytes. We observed that cells co-transfected with GAL4-Elk-1 1-428 S383A, in which the serine 383 residue was replaced with a non-phosphorylatable alanine, had no stimulatory effect on the reporter activity either in the absence (Fig. 2B, panel e) or presence of bFGF (Fig. 2B, panel f). These results suggest that bFGF exerts CMV-Elk11-428 CMV-Elk11-428383A Cell lysates were then prepared and analyzed by immunoblotting with phospho-specific anti-Elk-1 (Ser 383 ) antibody. Total Elk-1 was used as a control for normalization. B, cells were transiently co-transfected with a reporter construct, containing GAL4 element immediately upstream of the luciferase gene (GAL4RE-Luc) with the GAL4-Elk-1 fusion proteins (GAL4-Elk-1 1-428 wt or unphosphorylatable mutant, GAL4-Elk-1 1-428 S383A). 24 h after post-transfection, cells were treated with or without bFGF for an additional 24 h at which point the cell lysates were harvested and analyzed for the luciferase activity. Cells co-transfected with GAL4BD, the backbone of Elk-1 expression vectors, were used as controls. The data represent three different donors measured in triplicate for each experiment. C, cells were transiently co-transfected with the Ϫ1600MMP-13 or 562MMP-1 promoter construct and wildtype (CMV-Elk1 1-428 ) or mutant (CMV-Elk1 1-428 Ser383A) Elk-1 cDNA constructs. 24 h after post-transfection, cells were harvested and used for the analysis of luciferase activity. Co-transfection with empty CMV-Vector was used as a control. The data represent three different donors measured in triplicate for each experiment.
bFGF Stimulates MMP-13 via Elk-1 Activation OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 its biological effect by phosphorylating Elk-1 at serine 383, thereby promoting its transcriptional potential in enhancing expression of target genes such as MMP-13. We further examined whether phosphorylation of Elk-1 at serine 383 by bFGF is functionally linked to transactivation of MMP-13 in human adult articular chondrocytes. Cells were transiently co-transfected with the unphosphorylatable mutant Elk-1 expression vector (CMV-Elk-1 1-428 S383A) and the MMP-13 promoter-reporter construct that contains an Elk-1 response element (Fig. 2C). The MMP-13 promoter-driven luciferase activity of cells expressing wild type Elk-1 1-428 increased greater than 2-fold, but cells expressing the Elk-1 1-428 S383A mutant failed to stimulate MMP-13 transactivation. Unlike MMP-13, co-expression of Elk-1 with MMP-1 promoter (Ϫ562 upstream of ATG)-luciferase reporter construct (30) did not show a significant transactivation. These results suggest that phosphorylation of Elk-1 at the serine 383 residue is essential for the biological activity of Elk-1 and specific transactivation of MMP-13 in human adult articular chondrocytes.
Activation of Elk-1 by bFGF Is Specific to the Pathways Associated with MMP-13 Transactivation-Previously, we reported that bFGF activates PKC␦, which, in turn, transduces its signal to activate multiple MAPK subgroups (Erk, p38, and JNK) to regulate MMP-13 production (18). Because Elk-1 is a direct downstream target of the MAPK pathway in non-chondrocytic cells (19 -23), and because our data show that Elk-1 is critical for MMP-13 expression (see Figs. 1 and 2), we examined whether Elk-1 is a major effector of bFGF/PKC␦/MAPK-dependent pathways that control MMP-13 expression in human adult articular chondrocytes. Serum-starved chondrocytes cultured in monolayer were incubated with pathway-specific individual inhibitors for MAPK subgroups (Erki, p38i, and JNKi) and PKC isoforms (PKC␦i, PKC␣/␤i, and PKC⑀) 1 h prior to stimulation with bFGF for 5 min. We included a broad-spectrum PKC inhibitor (bisindolylmaleimide I), which can block signals generated from various PKC isoforms, including PKC␦, as a control (37). The modulation of Elk-1 activity was detected using phospho-specific anti-Elk-1 antibody (Ser-383). The blots were stripped and re-probed with total anti-Elk-1 antibody to normalize for equal loading of protein in each lane. The results showed that blocking MAPK subgroups (Erk, p38, and JNK) or the PKC␦ signaling pathways prevented phosphorylation of Elk-1 (Fig. 3A). Importantly, inhibition of PKC␣/␤ and PKC⑀ signaling pathways that are modulated by bFGF, but not associated with MMP-13 expression (18), failed to block Elk-1 activation.
Basic FGF mediates activation of the PI3K/Akt pathway in human adult articular chondrocytes as reflected by rapid phosphorylation of Akt at amino acid serine 473 within 5 min (Fig.  3B). This phosphorylation event renders Akt functionally active as a kinase, and consequently results in phosphorylation of its downstream targets, p70S6K and GSK3␤. However, blocking the PI3K/Akt pathway did not affect the bFGF-dependent stimulation of MMP-13 at both protein and mRNA levels (Fig. 3C). Inhibition of the MAPK pathway had no significant impact on bFGF activation of PI3K/Akt signaling (Fig. 3D) and vice versa (data not shown), suggesting that there was no discernable molecular cross-talk between the individual MAPK subgroups and the PI3K/Akt pathway. The MAPK pathway-dependent activation of Elk-1 was further supported by gel shift assay, using double-stranded Elk-1 sequences of MMP-13 and nuclear extracts prepared from primary chondrocytes that were preincubated with pharmacological inhibitors 1 h prior to stimulation with bFGF for 45 min. Consistent with our other results, we observed inhibition of Elk-1 protein-DNA interactions by pathway-specific inhibitors that are linked to MMP-13 expression, which includes MAPKs and PKC␦, after stimulation with bFGF (Fig. 3E). These results were further confirmed by using an Elk-1 activation-mediated GAL4 binding system in the presence of chemical inhibitors or co-expression of siRNA (Akt-siRNA) or dominant negative forms of MAPK subgroups (Erk-DN, p38-DN, and JNK-DN) (Fig. 3F). Again, consistent with our other data, blocking the Erk pathway either by pharmacological inhibitor or by overexpression of Erk-DN expression vector abolishes the bFGF-mediated activation of Elk-1, which is similar to the inhibitory level seen by blocking the FGF receptor 1 (either chemical inhibitor or overexpression of FGFR1-DN). Inhibition of the p38 and JNK pathways partially, but significantly, reduced bFGF-induced Elk-1 activation by both chemical inhibitors as well as the overexpression of dominant negative constructs. On the other hand, interfering with the PI3K/Akt pathway via either chemical inhibitor or introduction of Akt-siRNA vector (Mu6pro-Akt1/2) shows no influence on Elk-1 activation by bFGF. Control siRNA vector (Mu6pro) and CMV-vector (backbone of DNs) were used as a negative control for the experiment. Collectively, these data demonstrate that bFGF induction of MMP-13 production occurs via bFGF-responsive signaling pathways that mediate the activation of MAPK (preferentially Erk MAPK)-Elk-1 in human adult articular chondrocytes.
Basic FGF Activation of the NFB Pathway Regulates MMP-13 Expression in Human Adult Articular Chondrocytes-Our results show that stimulation of the cells with bFGF activates components of IB kinase (IKK) (IKK␣ Ser180 /IKK␤ Ser181 and IKK␥ (also known as NEMO)), which are important kinases for NFB activation (Fig. 4A). Phosphorylation of IKK, representing activation, is rapid (within 5 min), and the activation is sustained for Ͼ1 h after stimulation with bFGF.
To assess the role for NFB activation in bFGF stimulation of MMP-13 expression, serum-starved cells were incubated with NFB pathway-specific inhibitors 1 h prior to stimulation with bFGF. We used two different pharmacological inhibitors: peptide inhibitor SN50, which blocks the nuclear translocalization of activated NFB, and wedelolactone, a cell-permeable, selective inhibitor of IKK␣/␤ kinase activity (IC 50 Ͻ 10 M, Calbiochem). Conditioned media from cells treated with each of these inhibitors were analyzed for MMP-13 by immunoblotting. Pretreatment of chondrocytes with each NFB inhibitor significantly decreased the bFGF-mediated MMP-13 expression at both protein and mRNA levels (Fig. 4B, lanes 3 and 5). Because inhibition of the NFB pathway can cause apoptosis in chondrocytes (38), we performed DNA fragmentation analysis to ensure that the reduced MMP-13 production was not due to cell death. We found that cells incubated with chemical or peptide inhibitors of NFB showed no signs of apoptosis (data not shown). For comparison, experiments were performed in par-allel with IL-1␤, an inflammatory cytokine known to stimulate the NFB pathway (25), and we obtained similar results to those observed after stimulation with bFGF.
The pharmacological inhibitor studies were further validated by transient transfection studies. Human adult articular chondrocytes were transiently co-transfected with a construct (pCMV-IB␣M), expressing a dominant negative form of IB␣, and the MMP-13 promoter/luciferase reporter plasmid. Expression of IB␣M blocks endogenous NFB signaling by competitively binding to NFB and preventing the release of activated NFB (18,30,32). Our results showed that exogenous expression of IB␣M dramatically inhibited the bFGF-mediated stimulation of MMP-13 promoter activity (Fig. 4C, lane 4), which is consistent with our pharmacological inhibitor studies. Collectively, these results demonstrate the critical role of the NFB pathway in the stimulation of MMP-13 expression by bFGF in human adult articular chondrocytes.
NFB Pathway Regulation of MMP-13 Is Independent of NFB Protein-DNA Interaction-The promoter region of the MMP-13 gene from Ϫ1600 to Ϫ736 (relative to the ATG start site) contains a single putative NFB response element (Fig. 1A) (31). We analyzed the functional significance of this NFB motif using the Ϫ1600 and Ϫ736 MMP-13 promoter deletion constructs, which differ in the presence of the NFB response element (Fig.  5A). Cells transfected with the Ϫ1600 MMP-13 and Ϫ736 MMP-13 constructs did not exhibit a significant difference in stimulation of MMP-13 promoter activity by bFGF. We further verified the functional significance of this NFB motif present in the MMP-13 promoter by transiently transfecting a p3xNFBLuc construct that contains three copies of tandemly repeated NFB response element sequences of the MMP-13 promoter region into human adult   bFGF Stimulates MMP-13 via Elk-1 Activation OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 articular chondrocytes. Similar to those results obtained from the promoter deletion analysis, the transfected cells showed no response to the stimulation with bFGF (Fig. 5B). Our results suggest that the NFB response element present in the MMP-13 promoter region may not be functional and, thus, has no impact on bFGF stimulation of the MMP-13 promoter.

Elk-1 Is a Key Regulator That Mediates NFB Pathwaydependent Control of MMP-13 Expression upon bFGF
Stimulation-It is important to determine how bFGF-mediated activation of the NFB pathway can regulate MMP-13 expression independent of direct binding of NFB to its response element. Our real-time PCR results showed that stimulation of cells with bFGF rapidly increased elk-1 mRNA levels within 1 h (Ͼ2-fold), and this induction was abolished in the presence of a peptide inhibitor of the NFB pathway, which blocks the translocation of activated NFB (Fig. 6A). These results suggest that bFGF activation of NFB may regulate Elk-1 at the transcriptional level to control MMP-13 expression in human adult articular chondrocytes. Thus, we determined whether the NFB pathway is able to modulate the activation of Elk-1 protein after stimulation with bFGF. Serum-starved cells were treated with bFGF in a time course in the presence or absence of NFB inhibitor, and the cell lysates were analyzed for the phosphorylation of Elk-1 by immunoblot, using phospho-specific anti-Elk-1 antibody (Ser 383 ). Pretreatment of the cells with the NFB inhibitor significantly decreased the bFGFmediated activation of Elk-1 as represented by phosphorylation of Elk-1 at serine 383 (Fig. 6B).
NFB-dependent activation of Elk-1 after stimulation with bFGF was further supported by Elk-1 protein-DNA interaction in the presence of NFB inhibitors, which can block a specific regulatory step to activate the NFB pathway. Nuclear extracts were prepared from human adult articular chondrocytes treated with or without bFGF. Cells treated with bFGF showed a significant increase in Elk-1 protein-DNA-binding activity (Fig. 6C, lane 2) compared with untreated control chondrocytes (lane 1). Importantly, the presence of NFB inhibitors that can block either NFB unclear translocalization or IKK␣/␤ activity (lanes 3 and 4, respectively) significantly reduced the bFGFinduced Elk-1 protein-DNA interaction. In the absence of bFGF, these inhibitors decreased the basal level of Elk-1 protein-DNA interaction (data not shown). Specificity of binding was confirmed by incubating nuclear extracts with an excess of unlabeled Elk-1 oligonucleotides (lane 5).
The use of the GAL4 DNA binding system further validated our hypothesis that bFGF-mediated activation of Elk-1 is NFB-dependent (Fig. 6D). Chondrocytes co-transfected  Cell lysates were then prepared and used for dual luciferase assay system. B, cells were transiently transfected with p3xNFBLuc plasmid construct which contains three copies of tandemly repeated NFB response element sequences of MMP-13 promoter region. After 48 h of incubation, cells were stimulated with bFGF with additional incubation for 24 h. The relative luiferase units were calculated by normalizing the Renilla activity that was co-transfected. The data represent three different donors measured in triplicate for each experiment.

bFGF Stimulates MMP-13 via Elk-1 Activation
with the GAL4 RE-Luc and GAL4-Elk-1 1-428 wt exhibit dramatic increases in promoter activity in response to bFGF (Fig. 6D, c). In the presence of an inhibitor of IKK␣/␤ (d) or overexpression of IB␣M (e), which can prevent the release of activated NFB, this induction of Elk-1 activity was significantly reduced, whereas the overexpression of CMV-vector (backbone of IB␣M) had no inhibitory effect on the bFGFmediated Elk-1 activation (f). Collectively, our results suggest that bFGF-mediated activation of the NFB pathway modulates the biological function of Elk-1 at transcriptional and post-translational levels, because Elk-1 acts as a downstream target regulatory molecule. Thus, we conclude that, after stimulation with bFGF in human adult articular chondrocytes, Elk-1 functions as a key intermediate regulatory molecule in the NFB pathway that controls MMP-13 expression.

Nuclear Localization of Activated Elk-1 May Be Required for Biological Effect of bFGF in Human Adult Articular
Chondrocytes-Nuclear localization of activated transcription factors has been shown to be vital for fulfillment of many of their biological activities regulating target genes. Thus, we further elucidated the corresponding relationship between bFGFmediated regulation of nucleocytoplasmic trafficking of activated Elk-1 through phosphorylation of serine 383 and the stimulation of MMP-13 expression. The association of Elk-1 nuclear localization with the inhibitors that specifically block the bFGF-mediated signaling pathways leading to MMP-13 production was examined. In our time-course translocalization studies using anti-phospho-specific Elk-1 Ser383 antibody, we observed that the nuclear localization occurs within 5 min and reached the maximum nuclear localization level between 10 and 20 min after stimulation with bFGF (Fig. 7b). This level of localization was gradually decreased over 1 h (data not shown).  Thus, we selected 10 min for bFGF stimulation of cells for further studies. The translocalization of phosphorylated Elk-1 Ser383 to the nuclear region was significantly reduced by pharmacological inhibitors of FGF receptor (c), as well as the Raf (d), Erk MAPK (e), NFB (g), and PKC␦ (h) pathways. With less extent, we observed decreased Elk-1 nuclear localization in the presence of p38 and JNK inhibitors (data not shown). However, inhibition of the PI3K/Akt pathway (f), PKC␣/␤ and PKC⑀ (data not shown), which have no influence on the activation of Elk-1 nor on MMP-13 expression (Fig. 3), failed to modulate nuclear translocalization of activated Elk-1 after stimulation with bFGF. Our results suggest that the close association of biological activation of Elk-1 with the ability of Elk-1 to upregulate MMP-13 is concomitantly related to the nuclear translocalization of Elk-1 after stimulation with bFGF.
Normal and OA Comparison Studies on the Erk MAPK-Elk-1-MMP-13 Pathway-We examined the potential links between Elk-1 activity and cartilage-degenerative diseases by comparing the protein activity and expression of Elk-1 in normal and OA cells. Human adult articular cartilage tissue from normal knee (asymptomatic, grade 0 or 1) and degenerative cartilage obtained after knee replacement surgery were used for this study. Without culturing cells in monolayer, we directly isolated total RNA from the cartilage tissues and performed real-time PCR using human-specific primer sets. Compared with normal cells, the elevated basal level of elk-1 mRNA expression in OA cells was evident. Expression of both bfgf and c-fos, which are downstream target genes of Elk-1, was also significantly increased in OA compared with normal cells (Fig. 8A). Next, we examined whether Elk-1 DNAbinding activity, which represents the function of this transcription factor, is up-regulated in OA compared with normal cells. Serum-OA . Comparison studies between normal and OA cartilage. A, total RNA extracted from the cartilage tissue (normal and OA knees) were subjected to real-time PCR analysis for elk-1, bFGF, and c-fos gene expression. GAPDH mRNA was analyzed as a control for normalization purpose. B, Elk-1 protein-DNA interaction was tested using biotin-labeled probes incubated with nuclear extracts derived from untreated chondrocytes (lanes 1) or cells treated with bFGF for 45 min (lanes 2) of both normal and OA cells. Nuclear extracts incubated with excess amount of cold (unlabeled) probe were used for specificity purpose (lanes 3). C, total tissue protein extracted directly (without cell culture) from normal and OA knee cartilage (from three different donors) were resolved by 10% SDS-PAGE and immunoblotted with anti-phopshospecific Erk1/2 or MMP-13 antibodies. Total Erk was blotted as a control for loading equal amount of protein into each lane. D, cells isolated from normal and OA were serum-starved for 24 h followed by treatment with bFGF in different doses as indicated for 24 h. The conditioned media were then analyzed for MMP-13 production by immunoblot. For comparison purposes, cells were also treated with different doses of IL-1␤. Lanes: 1, control; 2, bFGF (10 ng/ml); 3, bFGF (50 ng/ml); 4, bFGF (100 ng/ml); 5, IL-1␤ (1 ng/ml); and 6, IL-1␤ (10 ng/ml). E, Safranin Orange staining was performed using full thickness cartilage explants prepared from OA knee joint tissues, and the explants were cultured in the presence or absence of bFGF (100 ng/ml) in mini-ITS. Following 14 days of incubation, the explants were fixed with 4% paraformaldehyde for overnight and embedded in paraffin, and 8 M sections were prepared for Safranin Orange staining to assess matrix proteoglycan loss. Representative results from three different donors are shown (n ϭ 6). F, type II collagen digestion assay was performed by using human knee chondorcytes (grade 2 and OA) cultured in different condition as indicated. The culture media was activated with 1 M 4-aminophenylmercuric acetate for 1 h at 37°C, and then incubated with chick sternal type II collagen. The samples were then resolved on 8% SDS-PAGE followed by staining with Coomassie Brilliant Blue R-250 for 2 h. The gel was destained with water, containing 10% methanol and acetate acid to determine collagen type II degradation ( 3 ⁄4 type II collagen fragment). Intact type II collagen was included as a control. Lanes: 1, Grade 2 control; 2, bFGF (100 ng/ml); 3, IL-1␤ (10 ng/ml); 4, tumor necrosis factor-␣ (10 ng/ml); 5, OA control; 6, bFGF (100 ng/ml); 7, IL-1␤ (10 ng/ml); 8, tumor necrosis factor-␣ (10 ng/ml); 9, OA control; 10, bFGF plus Erki; 11, bFGF plus NFBi; and 12, bFGF plus PI3K/Akti. starved cells in monolayers were incubated in the presence or absence of bFGF for 45 min and prepared for nuclear extract followed by gel-shift assay. A markedly increased basal level of Elk-1-DNA-binding activity was observed in OA compared with normal cells (Fig. 8B, Normal versus OA, lane 1). Treatment with bFGF (100 ng/ml) further elevated Elk-1 protein-DNA interactions in both normal and OA cells (Normal versus OA, lane 2). This elevated Elk-1 protein-DNA interaction in OA cells corresponded to the elevated basal activity of Erk MAPK, a direct upstream regulator of Elk-1 (Fig. 8C, upper  panel). Again, the enhanced basal activity of Erk was closely linked to stimulated production of MMP-13 at a basal level in OA compared with normal cells (Fig. 8C, lower panel). Furthermore, basal DNA binding activities of AP-1, which is a downstream target of Elk-1, were also significantly increased in OA cells compared with normal human articular chondrocytes (data not shown). These highly up-regulated abnormal basal activities of critical transcription factors in OA cells could explain why basal production of MMP-13 is higher in OA and why OA cells are hypersensitized to stimulation with bFGF ( Fig. 8D). We further examined the functional correlation between bFGF and articular cartilage degeneration by performing Safranin Orange staining, a marker to assess proteoglycan loss, using OA cartilage explant culture after incubation with or without bFGF for 14 days (Fig. 8E). Our results show accelerated Safranin Orange staining in the presence of bFGF in OA knee tissues (lower panel) compared with tissues not treated with bFGF (upper panel).
Using conditioned medium collected from bFGF-stimulated early-stage OA cells (grade 2) and normal cells, we also observed bFGF and its signaling pathway-mediated degradation of collagen type II. Incubation of cells with bFGF generated the 3 ⁄ 4 fragment of type II collagen (Fig. 8F, lanes 2 and 6), suggesting the potential links between bFGF and collagen digestion. Similar results were observed in the presence of well known catabolic pro-inflammatory cytokines (IL-1␤ and tumor necrosis factor-␣), which we included in our experiments in parallel for comparison (lanes 3, 4, 7, and 8). Importantly, the bFGF-induced collagen digestion was completely abolished by co-incubation with pathway-specific inhibitors of Erk (lane 10) and NFB pathway (lane 11), both upstream regulators of Elk-1 (Figs. 3, 6, and 7). Consistent with our other data, in the current study, blocking the PI3K/Akt pathway, which had no effect on Elk-1 activity and MMP-13 production (Figs. 3 and 7), showed no impact on the type II collagen degradation by bFGF (lane 12). Collectively, our results suggest that the bFGF/Erk/NFB/Elk-1 pathways co-operatively deregulate in OA and may represent an important pathophysiological link in cartilage degeneration.

DISCUSSION
This study focuses on the molecular mechanisms that mediate the biological impact of bFGF on human adult articular cartilage homeostasis and complements previous studies from our group (14,18). In human adult articular chondrocytes, binding of bFGF to its cognate receptor, FGFR1, activates multiple signaling cascades (18,38), including the MAPK pathway, PKC␦, NFB, and PI3K/Akt (Ref. 18 and current study). We have previously shown that NFB is one of the most critical activators of MMP-13 expression in response to fibronectin fragment (38). The present studies show that NFB pathway is also the ultimate target of bFGF signaling in human articular chondrocytes and is associated with the bFGF-mediated MMP-13 stimulation. The MMP-13 promoter region contains a putative NFB recognition motif (Ϫ1005 to Ϫ1014 upstream of ATG), and this element is not responsive to stimulation with bFGF. Accordingly, deletion of this motif did not affect basal or bFGF-stimulated levels of MMP-13 promoter activity. Similarly, Fujioka and colleagues (39) have demonstrated that vascular endothelial growth factor gene expression is regulated by the NFB signaling pathway, even though the promoter of vascular endothelial growth factor lacks the NFB binding motif. They proposed an indirect mechanism by which vascular endothelial growth factor stimulates Elk-1 expression which, in turn, controls the AP-1 component c-fos. Our current data suggest that NFB-mediated stimulation of MMP-13 by bFGF is mediated by NFB activation of the key ETS-domain-containing transcription factor Elk-1. Using human primary adult articular chondrocytes, we found that blocking the activation of the NFB pathway not only significantly reduces the bFGF-mediated phosphorylation and biological activation of Elk-1 but also elk-1 mRNA accumulation in response to bFGF. We propose that NFB activation, which may be under control of MAPK (JNK, Erk, and p38), may have a dual role as an upstream regulator of both Elk-1 and, thereby, c-Fos, resulting in the control of MMP-13 expression. Our results collectively indicate that multiple molecular networks converge to activate Elk-1, leading to the up-regulation of MMP-13 expression.
Blocking the biological activity of bFGF by inhibiting FGFR1, MAPKs, or NFB is associated with reduced Elk-1 activation, nuclear translocalization, and Elk-1 protein-DNA-binding activity, which may result in the down-regulation of MMP-13 expression. These findings suggest a close link between Elk-1 function and MMP-13 expression in human adult articular chondrocytes. Interestingly, our data suggest that the bFGFmediated activation of PI3K/Akt signaling cascades is completely independent from other known signaling events associated with MMP-13 stimulation. Investigation of the potential role of the PI3K/Akt signaling pathway after stimulation with bFGF may provide insight into the precise biological role of bFGF in human adult articular cartilage homeostasis (Fig. 9).
Using chimeric GAL4/Elk-1 proteins in transcriptional assays, we verified that bFGF biologically activates the inherent transactivation potential of Elk-1 in both a quantitative and sensitive manner. This bFGF-dependent stimulation was reduced by blocking the pathway controlled by FGFR1, the individual MAPK subgroups and NFB. The activation of multiple MAPK subgroups and NFB appeared to be required and essential to stimulate MMP-13 expression in a coordinated and integrated manner. Among the MAPK subgroups, Erk exhibited the most potent impact on Elk-1 activation, suggesting that Erk-Elk-1 is the preferential regulatory pathway utilized by bFGF in human articular chondrocytes. The bFGF-mediated activation of Elk-1 was completely abolished in the presence of an inhibitor of MEK1/2 (direct upstream regulator of Erk), bFGF Stimulates MMP-13 via Elk-1 Activation OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 whereas blocking the JNK, p38, and NFB pathways partially inhibited bFGF-mediated Elk-1 activity, suggesting that partial reduction of Elk-1 is sufficient enough to down-regulate MMP-13 expression.
Our results also show that overexpression of Elk-1 is selective for MMP-13, because we did not observe a significant induction of MMP-1. Because bFGF and inflammatory cytokines are present at elevated levels in the synovial fluids of patients with osteoarthritis, the possibility arises that both Erk-Elk-1 and NFB pathways are concomitantly activated in chondrocytes under pathological conditions, and NFB-directed modulation of Elk-1 may cooperate to activate shared target genes of these two rapidly inducible transcription factors.
The basal gene expression levels of bfgf, elk-1, and c-fos were increased in OA cartilage in comparison to that of normal articular cartilage. In addition, the basal activity of Erk and Elk-1 was highly up-regulated in OA compared with that of normal articular chondrocytes. Moreover, the increased basal activity levels of Erk-Elk-1 were correlated with increased basal levels of MMP-13 in OA cartilage as well as synovial fluids. In our separate protein-DNA interaction experiments, followed by individual gel shift assays, we observed dramatically elevated basal activity of NFB and AP-1 in OA cells (data not shown). The elevated basal activities of well known inflammatory pathwayassociated transcription factors perhaps can explain why the biological response to bFGF was accelerated in OA cells. Collectively, our results demonstrate pathophysiological links between abnormally high basal levels of activated Erk-Elk-1 and increased MMP-13 expression found in OA cartilage.
From our previous data and current studies, the anti-anabolic and catabolic effect on cartilage by bFGF is apparent in human adult articular cartilage (14,18). The negative biological action of bFGF on articular cartilage is significantly magnified when cartilage tissues are already in the process of degeneration (from the stage of early degeneration to OA condition). These findings suggest that naturally increased bFGF expression and its cognate receptor FGFR1 in OA cells may participate in the detrimental inhibition of cartilage repair resulting in tissue degeneration. Previously, we reported that bFGF stimulates MMP-13 expression together with up-regulation of various pro-inflammatory cytokines (18) and potent inhibition of proteoglycan synthesis in human adult articular cartilage (14). In the current study, we observed bFGF-mediated proteoglycan loss (by Safranin Orange staining) and type II collagen cleavage in an in vitro assay, which was rescued in the presence of inhibitors of Erk MAPK and NFB, further supporting our previous results. Inhibitors of the PI3K/Akt pathway, however, failed to rescue bFGF-mediated type II collagen degradation. This result is consistent with our other data in this study and confirms that bFGF-mediated stimulation of collagenase production occurs as the PI3K/Akt-independent pathway.
In conclusion, we have provided compelling evidence for the critical role of the Elk-1 transcription factor in the transcriptional stimulation of MMP-13 (collagenase-3), but not MMP-1 (collagenase-1), in response to bFGF in human adult articular chondrocytes. We further provide evidence for an indirect regulatory role of NFB on MMP-13 expression through its activation of Elk-1 as a critical intermediate transcriptional regulator. MAPK signaling (especially Erk) modulates Elk-1 activity, Elk-1 regulates AP-1, and AP-1 is one of the core transcription factors critical for MMP-13 gene expression. Furthermore, MAPK also controls RUNX2 activity, which is a key regulator of MMP-13 gene expression (40). Thus, bFGF may control MMP-13 by the coordinated molecular regulation of multiple regulatory factors and their cognate motifs present in the MMP-13 promoter: Elk-1, NFB, and the core factors, AP-1, ETS, and RUNX2.