Cooperation of Ras- and c-Myc–dependent pathways in regulating the growth and invasiveness of synovial fibroblasts in rheumatoid arthritis

Authors


Abstract

Objective

To study the specific contribution of MAP kinase activator c-Raf-1 and one of its downstream transcription factors, c-Myc, to the growth and invasive behavior of rheumatoid arthritis synovial fibroblasts (RASFs).

Methods

RASFs were transduced with retroviral constructs expressing dominant-negative mutants of c-Raf-1 or c-Myc (DN c-Raf-1 or DN c-Myc, respectively) or with the mock vector. The expression of wild-type and mutant proteins was confirmed by Western blotting. Growth curves of RASFs were recorded, and apoptosis was measured by flow cytometry. Invasiveness of RASFs was assessed in the SCID mouse model of RA. Immunohistochemistry was used to study the effects of DN c-Raf-1 on phosphorylated c-Jun and matrix metalloproteinase 1 (MMP-1) in RASFs implanted into SCID mice. The phosphorylation of ERK and JNK in DN c-Raf-1– and mock-transduced RASFs was determined in vitro by Western blotting. The levels of MMPs in these cells were measured by quantitative polymerase chain reaction (PCR).

Results

Neither DN c-Raf-1 alone nor DN c-Myc alone significantly altered proliferation or apoptosis of RASFs, but both mutants together rapidly induced apoptosis. Inhibition of c-Raf-1 or c-Myc significantly reduced the invasiveness of RASFs in the SCID mouse model. DN c-Raf-1 decreased the phosphorylation of ERK and JNK in vitro and reduced the in vivo expression of phosphorylated c-Jun as well as the expression of disease-relevant MMPs. As determined by quantitative PCR, the inhibition was most pronounced for MMP-1 and MMP-3.

Conclusion

The data demonstrate that Ras- and c-Myc–dependent signaling events cooperate to regulate the growth and invasiveness of RASFs. Targeting of both c-Raf-1 and c-Myc may constitute an interesting therapeutic approach in RA.

Growing evidence suggests that activated synovial fibroblasts (SFs) play a pivotal role in the pathogenesis of rheumatoid arthritis (RA), a systemic disabling disease that primarily affects the joints and results in their progressive destruction (1). Such activated fibroblasts are found predominantly in the synovial lining and appear largely responsible for the progressive destruction of articular cartilage. Apart from an altered morphology (2), changes in the expression of protooncogenes (3) and tumor suppressor genes (4, 5) have been demonstrated in RASFs. Among the signaling molecules that have been implicated in the activation of RASFs, MAP kinases (MAPKs) play a prominent role (6–8). MAPKs consist of 3 families: extracellular signal–regulated kinases (ERKs), c-Jun N-terminal kinases (JNKs), and p38 kinases. These differ with respect to their substrate specificities and their responses to extracellular signals. Activation of ERKs occurs predominantly as a consequence of mitogenic stimuli, while JNKs and p38 kinases are activated mainly by environmental stress as well as by cytokines such as tumor necrosis factor α (TNFα) (9, 10). Although the signals triggered by the individual MAPKs appear distinct, there is extensive crosstalk between these pathways.

In RA, pathways that link cytokine signaling to the expression of matrix-degrading enzymes as well as to cell growth and altered apoptosis have gained increasing attention, but the relevance of mitogenic signaling for the aggressive behavior of RASFs is still poorly understood. In this context, Ras-mediated signaling together with c-Raf-1, the primary upstream activator of ERKs, is of particular interest. Depending on the cell cycle and costimulatory events, activation of c-Raf-1 can result in a variety of counteracting cellular responses such as proliferation, growth arrest, induction of apoptosis, or differentiation (11–13). Of interest in RA, c-Raf-1 has also been associated with the up-regulation of adhesion molecules such as integrins (14). C-Raf-1 is activated mainly by Ras in response to signals from growth factor receptors such as epidermal growth factor (EGF), but it is also regulated by signaling pathways involving the tumor suppressor PTEN (15) and the phosphatidylinositol 3-kinase (PI 3-kinase)/Akt (protein kinase B) signaling pathway (16). Previously, we have demonstrated that PTEN is down-regulated in RASFs (17), while other studies revealed that alterations in the PI 3-kinase/Akt pathway are involved in the pathogenesis of RA (18).

Here, we investigated the effects of retroviral gene transfer of dominant-negative (DN) mutants of c-Raf-1 and c-Myc on growth and apoptosis of, and invasion by, RASFs. We report that inhibition of Ras-mediated signaling through DN c-Raf as well as inhibition of c-Myc reduce the invasiveness of RASFs into articular cartilage in vivo but fail to affect growth and apoptosis of these cells under in vitro conditions. In contrast, combined delivery of DN c-Raf-1 and DN c-Myc results in a significant increase of spontaneous apoptosis of RASFs in vitro. Furthermore, we provide evidence that Ras-dependent signaling events regulate the expression of disease-relevant matrix metalloproteinases (MMPs) through phosphorylation of both ERK and JNK and subsequent activation of the transcription factor c-Jun.

MATERIALS AND METHODS

Generation of replication-deficient retroviruses encoding DN c-Raf-1 and DN c-Myc mutants.

FLAG-tagged mutants lacking the C-terminal catalytic domain of c-Raf were generated as described previously (19). Briefly, the FLAG epitope–coding sequence was cloned 5′ to the human c-raf-1 complementary DNA in a pUC18 cloning vector. The retroviral pLFR(1-331)SN vector was then generated by inserting the Eco RI/Stu I fragment of this construct into the Eco RI/Hpa I sites of pLXSN. The pLM(Δ106-143)SN DN c-myc constructs that contained a deletion in the transactivation domain–encoding region (MycΔ106-143) have been described in detail previously (20). Stable cells producing DN c-Raf-1–carrying retroviruses were obtained by infecting 105 GP+envAm12 packaging cells with the viral supernatant of transient producer cells that had been generated through calcium phosphate precipitation of GP+E-86 cells with the retroviral construct pLFR(1-331)SN. Transduction of the GP+envAm12 packaging cells was performed in the presence of polybrene (8 μg/ml) according to standard procedures. The successfully transduced packaging cells were then selected by adding G418 (1 mg/ml).

To allow for cotransduction of target cells with both DN c-Raf-1 and DN c-Myc, replication-deficient retroviruses carrying the DN c-Myc construct were generated accordingly from PG13 packaging cells. Retroviruses carrying the empty (mock) construct were generated from GP+envAm12 packaging cells to be used for control experiments.

Isolation of human RASFs and transduction with the DN c-Raf-1 and DN c-Myc constructs.

RASFs were obtained from 3 patients with RA that met the 1987 American College of Rheumatology (formerly, the American Rheumatism Association) revised criteria for the disease (21). The tissue specimens were minced and digested enzymatically (1 mg/ml type VIII collagenase; Sigma, St. Louis, MO). The released cells were grown in Dulbecco's modified Eagle's medium (DMEM; Life Technologies, Gaithersburg, MD) with 10% fetal calf serum (FCS; Gemini Biological Products, Calabasas, CA) in a humidified 5% CO2 atmosphere. After allowing the cells to adhere overnight, nonadherent cells were removed, and the adherent cells were grown further over 4 passages. RASFs (106) were transduced by incubation with 10 ml of diluted retroviral supernatants for 12 hours in the presence of 8 μg/ml polybrene. This procedure was repeated 3 times at 12-hour intervals to increase retroviral infection. One day after the final transduction, selection with G418 (1 mg/ml) was started for at least 10 days. For double transduction, this procedure was repeated after 24 hours with the second construct. The number of passages was kept equal between the DN c-Raf-1–transduced, DN c-Myc–transduced, and mock-transduced RASFs.

Expression of DN c-Raf-1, DN c-Myc, ERK, and JNK.

The expression of DN c-Raf-1 in the transduced RASFs was assessed by immunoprecipitation and Western blotting as described previously (19). Briefly, 106 RASFs were grown to 80% confluence and lysed in radioimmunoprecipitation assay (RIPA) buffer, and Raf proteins from the supernatant were immunoprecipitated with rabbit anti-Raf serum 1617 (anti–N-terminus). Immune complexes were separated on a 10% sodium dodecyl sulfate polyacrylamide gel and blotted to a nitrocellulose membrane. The FLAG tag was detected using the M5 anti-FLAG monoclonal antibody (1:150 dilution) and a peroxidase-labeled anti-mouse antibody (Sigma) together with the ECL detection kit (Amersham, Braunschweig, Germany). In addition, the expression of wild-type c-Raf-1 as well as DN c-Raf-1 protein was determined directly by Western blotting using a rabbit anti-Raf serum. The expression levels of DN c-Myc were determined accordingly by direct immunoblotting using the 9E10 monoclonal antibody (Roche, Rotkreuz, Switzerland).

For the growth factor–dependent activation of ERK and JNK, cells were starved in the presence of 0.5% FCS for 20 hours. The cells were then stimulated for 5 minutes with 100 ng/ml EGF (Roche). Cells were lysed in RIPA buffer, protein content was determined using the Bio-Rad protein detection kit (Bio-Rad, Hercules, CA), and 15 μg total protein was analyzed by Western blotting using anti–phospho-ERK and anti–phospho-JNK antibodies (Cell Signaling Technology, Beverly, MA). Blots were stripped and reprobed with anti–ERK-2 and anti–β-actin antibodies (Santa Cruz Biotechnology, Palo Alto, CA) as a loading control.

Detection of apoptosis by flow cytometry analysis.

Cells were harvested at 80–90% confluence, washed with phosphate buffered saline (PBS), and fixed in 70% ethanol. Using a fluorescence-activated flow cytometer (FACSCalibur; BD Biosciences, Heidelberg, Germany), the DNA content of the cells was assessed by propidium iodide staining (40 μg/ml for 30 minutes) following RNase treatment. The rate of apoptosis was measured by flow cytometry in pLXSN (mock)–, pLFR(1-331)SN-, and pLM(Δ106-143)SN-transduced RASFs by double staining with fluorescein isothiocyanate (FITC)–labeled annexin V and propidium iodide.

The SCID mouse model of cartilage degradation.

Four-week-old, female SCID mice were provided by Charles River (Sulzfeld, Germany). Normal human articular cartilage was obtained from patients undergoing joint replacement surgery for severe trauma at the Department of Surgery, University Hospital Zurich, Zurich, Switzerland. Implantation of RASFs together with normal human cartilage was performed as described previously (22). After 60 days, mice were killed and the implants were removed. Tissue preparation included fixation in 4% buffered formalin and paraffin embedding according to standard procedures. Five animals were used per group (pLXSN, pLFR[1-331]SN, pLM[Δ106-143]SN), resulting in a total of 15 mice. Two animals died during anesthesia, and 1 could not be evaluated due to a dislocation of the cartilage.

Histologic evaluation.

Following hematoxylin and eosin staining of 4-μm paraffin sections, histologic evaluation was performed using semiquantitative scores for invasion and perichondrocytic cartilage degradation (described previously; see ref. 23). The score grades were defined as follows: 0 = no or minimal invasion; 1 = visible invasion (2–5 cell depths); 2 = invasion (6–10 cell depths); 3 = deep invasion (>10 cell depths); and 4 = overall deep invasion reshaping the cartilage surface. The sections were evaluated by 2 experienced scientists in a blinded manner.

Immunohistochemistry.

The expression of phosphorylated c-Jun was investigated by immunohistochemistry using antibodies that specifically recognized the site of Ser73 phosphorylation on c-Jun (Santa Cruz Biotechnology). For the immunohistochemical detection of collagenase 1 (MMP-1), we used rabbit anti-human antibodies that specifically recognize the hinge region of the human MMP-1 sequence (Research Diagnostics, Flanders, NJ). Negative controls involved substitution of the primary antibodies with isotype controls and 0.1% bovine serum albumin in PBS. Immunohistochemistry was performed according to standard procedures.

Expression of MMP messenger RNA (mRNA).

Expression levels of MMP-1, MMP-3, MMP-9, and MMP-13 mRNA in RASFs were analyzed by quantitative real-time polymerase chain reaction (PCR) using a fluorogenic 5′-nuclease assay (TaqMan; Applied Biosystems, Weiterstadt, Germany) on an ABI Prism 7900 HT Sequence Detection system. For each experiment, total RNA was extracted from 105 cells using the RNeasy system (Qiagen, Hilden, Germany). Total RNA was reverse transcribed using random hexamer primers.

For quantitative PCR, the primers and FAM-TAMRA–labeled probes were as follows: MMP-1 forward primer 5′-TGT-GGA-CCA-TGC-CAT-TGA-GA-3′, reverse primer 5′-TCT-GCT-TGA-CCC-TCA-GAG-ACC-3′, TaqMan probe 5′-CCA-ACT-CTG-GAG-TAA-TGT-CAC-ACC-TCT-GAC-ATT-CAC-C-3′; MMP-3 forward primer 5′-GGG-CCA-TCA-GAG-GAA-ATG-AG-3′, reverse primer 5′-CAC-GGT-TGG-AGG-GAA-ACC-TA-3′, TaqMan probe 5′-AGC-TGG-ATA-CCC-AAG-AGG-CAT-CCA-CAC-3′; MMP-9 forward primer 5′-GGC-CAC-TAC-TGT-GCC-TTT-GAG-3′, reverse primer 5′-GAT-GGC-GTC-GAA-GAT-GTT-CAC-3′, TaqMan probe 5′-TTG-CAG-GCA-TCG-TCC-ACC-GG-3′; MMP-13 forward primer 5′-TCC-TAC-AAA-TCT-CGC-GGG-AAT-3′, reverse primer 5′-GCA-TTT-CTC-GGA-GCC-TCT-CA-3′, TaqMan probe 5′-CAT-GGA-GCT-TGC-TGC-ATT-CTC-CTT-CAG-3′. For an internal standard, 18S ribosomal RNA was coamplified. Data were obtained from 3 different RASF cell lines and calculated with the ΔΔCt method as described (24).

Statistical analysis.

Differences in proliferation, invasiveness, and MMP expression between pLXSN (mock)–, pLFR(1-331)SN-, and pLM(Δ106-143)SN-transduced RASFs were analyzed for statistical significance using the Mann-Whitney U test. P values less than 0.05 were considered significant.

RESULTS

Retroviral gene transfer of DN mutants of c-Raf-1 and c-Myc in RASFs.

Human RASFs were stably transduced with retroviral constructs expressing DN c-Raf-1 (pLFR[1-331]SN, also called RafNT) or DN c-Myc (pLM[Δ106-143]SN, also called c-MycΔ106) or with the mock vector (pLXSN, also called mock) (Figure 1A). RafNT encodes the Ras-binding domain without the kinase domain and therefore acts as a competitor for Ras effectors. c-MycΔ106 is a DNA binding protein without the transactivation domain and competes with wild-type c-Myc for DNA binding, thus acting as an inhibitor. Following transduction, expression of the DN c-Raf-1 or c-Myc mutants was confirmed by Western blot analysis using a rabbit serum against the c-Raf-1 N-terminus as well as anti-FLAG and anti–c-Myc antibodies.

Figure 1.

Expression of dominant-negative (DN) mutants of c-Raf-1 and c-Myc in rheumatoid arthritis synovial fibroblasts (RASFs). A, Structure of the retroviral pLXSN vector (mock), the retroviral construct expressing DN c-Raf-1 (pLFR[1-331]SN, also called RafNT), and the retroviral construct expressing DN c-Myc (pLM[Δ106-143]SN, also called c-MycΔ106). B, Expression of the constructs RafNT and c-MycΔ106 as determined by Western blotting is shown at left. Immunoprecipitation of RafNT from mock- and RafNT-transduced RASFs is shown at right. C, Direct immunoblotting for c-Myc indicated the expression of c-MycΔ106. WT = wild-type.

RafNT-transduced cells expressed the mutant protein in lower amounts compared with endogenous wild-type Raf (Figure 1B). No RafNT or FLAG expression was seen in the mock- or c-MycΔ106–transduced cells (Figure 1B). As shown previously (19), transduction of the cells with RafNT clearly inhibited c-Raf activity, and the FLAG epitope did not interfere with Ras binding. This could be confirmed in our system with RASFs (see below). Transduction of RASFs with c-MycΔ106 resulted in the expression of the DN mutant c-Myc protein, as shown by immunoblotting (Figures 1B and C). Expression levels of the DN mutants were monitored for an extended period of time, and stable expression of both DN c-Raf-1 and DN c-Myc was seen in the RASFs for at least 60 days.

Growth characteristics and apoptosis of transduced RASFs.

Growth curves of RafNT-, c-MycΔ106–, and mock-transduced RASFs were obtained by counting the cells repeatedly after transduction. Under standard culture conditions (DMEM with 10% FCS and G418), c-MycΔ106–expressing cells showed a slight decrease in their rate of proliferation, but overall no significant differences in growth kinetics were seen between RASFs expressing RafNT, RASFs expressing c-MycΔ106, and control (mock-transduced) RASFs. Further, no differences in the morphologic appearance of these cells were observed even after prolonged culture periods (Figure 2A).

Figure 2.

Growth and apoptosis of mock-, RafNT-, and c-MycΔ106–transduced RASFs. A, As shown in the graph at left, no significant differences in growth were seen between RASFs expressing RafNT, RASFs expressing c-MycΔ106, and control (mock-transduced) RASFs. RASFs transduced with both RafNT and c-MycΔ106 did not proliferate but died rapidly within 3 days. Values are the mean ± SD. As shown in the panels at right, no differences in the morphology of surviving cells were observed after prolonged culture (original magnification × 630). B, There were no significant differences in apoptosis between RafNT- and c-MycΔ106–transduced RASFs. RASFs that received a double dose of c-MycΔ106 showed a slight increase in the number of apoptotic cells. RASFs transduced with both RafNT and c-MycΔ106 rapidly underwent apoptosis. FITC = fluorescein isothiocyanate (see Figure 1 for other definitions). Color figure can be viewed in the online issue, which is available at http://www.arthritisrheum.org.

No significant changes in apoptosis were found between the RafNT- and c-MycΔ106–transduced cells (Figure 2B). Transduction with double-dose RafNT or c-MycΔ106 did not alter this result significantly, but a slight increase in the number of apoptotic cells was seen in the RASFs that received double-dose c-MycΔ106. In contrast, RASFs that were transduced with both RafNT and c-MycΔ106 rapidly underwent apoptosis, as seen both by a characteristic pre-G1 peak and by staining with FITC-labeled annexin V and propidium iodide (Figure 2B).

Inhibition of RASF-mediated cartilage degradation.

The invasiveness of mock-transduced as well as RafNT- and c-MycΔ106–transduced RASFs into human articular cartilage was assessed in the SCID mouse model of RA. In this model, RASFs are coimplanted together with normal human cartilage into SCID mice and kept for 60 days. It has been shown previously that implantation of RASFs results in the invasion of these cells into the coimplanted cartilage in the absence of human inflammatory cells, whereas normal, non-RA synovial fibroblasts do not exhibit such behavior.

In accordance with these previous data, mock-transduced RASFs showed deep invasion into the cartilage (Figure 3A), with a mean ± SD invasion score of 2.50 ± 0.13. Expression of RafNT resulted in a significant reduction of invasiveness (to a mean ± SD score of 1.38 ± 0.12) (Figure 3A). The invasion of RASFs into the cartilage was not inhibited completely, but the highest invasion score in this group was only 2.0 and thus clearly below what was seen in the mock-transduced group (maximum score 3.0). Also, expression of c-MycΔ106 reduced the invasiveness of RASFs into the coimplanted cartilage (mean ± SD invasion score of 1.16 ± 0.14). Again, expression of DN c-Myc did not result in complete inhibition, but it did reduce the maximum invasion score to 2.0. As also seen in Figure 3A, there were no changes in cellularity and shape between the mock-, RafNT-, and c-MycΔ106–transduced RASFs. Due to the rapid onset of apoptosis following double gene transfer with RafNT and c-MycΔ106, RASFs that were transduced with both RafNT and c-MycΔ106 could not be used in the SCID mouse model.

Figure 3.

Regulation of invasiveness in the SCID mouse model of RA. A, As seen at histologic evaluation and by semiquantitative analysis, gene transfer of RafNT and c-MycΔ106 significantly reduced the invasiveness of RASFs in the SCID mouse model of RA compared with mock-transduced RASFs (∗ = P < 0.05) (original magnification × 400). B, Western blotting demonstrated that RafNT reduced the epidermal growth factor (EGF)–induced phosphorylation of ERK and JNK. No effects were seen on the expression of β-actin and total ERK-2, both of which were used as loading controls. C, RafNT significantly decreased the expression of matrix metalloproteinase 1 (MMP-1) and MMP-3 mRNA, as found by quantitative real-time polymerase chain reaction (Ct is the number of cycles needed to reach the amplification threshold, ΔCt represents the difference in the Ct values between the MMPs and the 18S ribosomal RNA that was used as an internal standard, and 2math image represents the mRNA levels of MMPs in RafNT-transduced RASFs as decimal fractions of those in mock-transduced RASFs). Some reduction was seen in the expression of MMP-9 mRNA. ∗ = P < 0.05 versus mock-transduced RASFs (n = 3 different cell lines). D, In immunohistochemistry studies, abundant staining for phosphorylated c-Jun was seen in mock-transduced RASFs, particularly at sites of cartilage destruction. RafNT markedly reduced the expression of phosphorylated c-Jun, which was accompanied by decreased expression of MMP-1 (original magnification × 400). Values in A and C are the mean and SD. Arrows and dotted lines in A and D indicate cartilage margins. H&E = hematoxylin and eosin (see Figure 1 for other definitions).

Inhibition of ERK and JNK phosphorylation by RafNT.

To test the inhibition of Ras and Raf signaling in RafNT-transduced RASFs, cells were stimulated with EGF, and the phosphorylation of ERK and JNK was determined by Western blot analysis using anti–phospho-ERK and anti–phospho-JNK antibodies. Compared with mock vector as control, RafNT inhibited ERK phosphorylation, whereas c-MycΔ106 had no effect on the Raf/MEK/ERK pathway (Figure 3B). In our analysis of two representative cell cultures, ERK-1/p44 appeared to be inhibited more strongly than ERK-2/p42, which may also indicate a lower MEK activity. Stable expression of RafNT in RASFs resulted in only partial inhibition of Raf signaling, corresponding to the moderate levels of expression of RafNT (Figure 1B). Of interest, EGF also induced the phosphorylation of JNK, which was nearly completely inhibited by RafNT. Such inhibition was not seen with c-MycΔ106 or the mock controls (Figure 3B).

RafNT inhibition of the expression of disease-relevant MMPs.

Next, we measured the expression of mRNA for disease-relevant MMPs in RafNT-transduced RASFs, by quantitative real-time PCR (Figure 3C). Compared with mock-transduced RASFs, expression of MMPs was reduced in RafNT-transduced RASFs. The inhibition was most pronounced for MMP-1 (to 46% of mock; P < 0.05), MMP-3 (to 53% of mock; P < 0.05), and MMP-13 (to 64% of mock). A weak reduction was seen in the expression of MMP-9 (to 85% of mock).

Effects of RafNT on the expression of phosphorylated c-Jun and MMP-1 in vivo.

We further investigated the effects of gene transfer with DN mutants of c-Raf-1 on the expression of phosphorylated c-Jun in the RASFs after implantation into the SCID mice. While mock-transduced RASFs showed abundant staining for phosphorylated c-Jun, particularly at sites of cartilage destruction, a marked reduction of the expression of phosphorylated c-Jun was seen in the RafNT-transduced cells that displayed a reduced invasiveness (Figure 3D). This was accompanied by a decreased expression of MMP-1 in the RafNT-transduced RASFs compared with the mock-treated cells (Figure 3D).

DISCUSSION

There is a continuing challenge in exploring the molecular triggers of synovial fibroblast activation in RA. Although it has been well established that the inflammatory environment contributes to the activation of RASFs through the action of cytokines such as TNFα and interleukin-1 (IL-1) (25), several clinical studies (26) as well as experimental data have demonstrated some dissociation between the actual inflammation and cartilage destruction. In particular, the SCID mouse model of RA has provided evidence that RASFs maintain their specific phenotype and their aggressive behavior in the absence of human inflammatory stimuli (22). It has therefore been suggested that activated RASFs are permanently imprinted or altered by the rheumatoid environment (27), although it remains to be determined how this imprint is maintained in the cells. In addition, alternative mechanisms have been discussed, such as cellular transformation by unknown triggers (1). In this context, the activation of cellular protooncogenes and their respective signaling pathways is of particular importance, because dysregulation of such pathways has been found in several tumors and has been shown at least in part for RA as well (3).

Based on these data, we hypothesized that the activation of signaling of Ras and the activation of one of its downstream transcription factors, c-Myc, contribute to the invasiveness of RASFs. Consequently, we sought to answer the question of whether delivery of DN c-Raf-1 and DN c-Myc, alone or in combination, would alter the growth rate and apoptosis of RASFs in vitro and reduce their invasiveness in vivo. Using retroviral vector delivery, we generated stably transduced RASFs that could be cultured and used in the SCID mouse model of RA.

Delivery of DN c-Raf-1 alone or DN c-Myc alone did not result in significant alterations of RASF growth characteristics in vitro. The failure of DN c-Raf-1 to alter the proliferation rate of RASFs is of special interest. Although increased activation of ERKs and expression of Ras- and ERK-inducible genes in RA synovial cells suggest an increased activity of Raf-1 in the rheumatoid synovium (6), there have been no functional studies on the contribution of Ras-mediated signaling pathways to the specific, destructive phenotype of RASFs. Data from other fibroblast lines indicate that c-Raf-1 can promote either cell proliferation or cell-cycle arrest depending on the level of activation (28). Specifically, low-level activation of c-Raf-1 results in the entry of quiescent cells into the cell cycle, while high levels of c-Raf-1 activity can mediate cell-cycle arrest through induction of cyclin-dependent kinase inhibitor p21Cip1 (28). It has been demonstrated that little p21Cip1 is expressed in RASFs, and that gene transfer of p21Cip1 inhibits synovial cell proliferation (29). These observations would suggest a low-level activation of c-Raf-1 and support the notion that proliferation of RASFs contributes significantly to synovial hyperplasia in RA (30). However, there is evidence that RASFs do not proliferate faster than either normal or osteoarthritis synovial fibroblasts (31–34).

Interestingly, it has been suggested that in NIH3T3 cells, a threshold for cell-cycle arrest is defined by the amount of c-Raf-1 activity required for the induction of p21Cip1 (28). Therefore, suppression of p21Cip1 through different pathways may result in a compensatory up-regulation of c-Raf-1. In this context, the tumor suppressor p53 is one molecule that interferes with downstream molecules of c-Raf-1 such as p21Cip1 (35) and modulates the response to c-Raf-1–mediated signaling. In previous studies, we have shown that inhibition of p53 results in an increased cellularity and aggressiveness of RASFs in the SCID mouse model of RA (36). Alterations in the expression of p53, as described previously (4), may therefore explain in part the low expression of p21Cip1 but high levels of c-Raf-1 activation.

In addition, not only has the transcription factor c-Myc been shown to be regulated by c-Raf-1 (37), but c-Myc itself suppresses the transcription of p21Cip1 (38). In RASFs, c-Myc has been demonstrated to be up-regulated (39), especially at sites of destruction in RA joints (40). Recently, it was shown that the delivery of c-myc antisense oligodeoxynucleotides significantly reduces the proliferation of RASFs and may induce apoptosis through caspase signaling (41). In our experiments, there were only minor effects on the proliferation rate of DN c-Myc–expressing RASFs, and no significant changes were observed in the percentage of apoptotic cells. Apart from the lack of antisense oligonucleotide toxicity, these differences are most likely due to the level of c-Myc inhibition. This notion is supported not only by the dose effects in the aforementioned study, but also by our observation that double transduction of RASFs with c-MycΔ106 increased the number of apoptotic cells. It is also consistent with most recent findings indicating that a modest reduction in c-Myc expression has minimal effects on cell growth and apoptosis but dramatically reduces susceptibility to Ras and Raf transformation (42). In our study, cells that expressed both DN c-Raf-1 and DN c-Myc rapidly underwent apoptosis. These data strongly suggest that both factors cooperate in promoting the survival of RASFs.

The concept of Ras activation but secondary modulation of more downstream signaling pathways is supported strongly by our data on cartilage destruction in vivo. Retroviral delivery of DN c-Raf-1 resulted in a clear suppression of invasiveness into human articular cartilage in the SCID mouse model of RA for a period of 60 days, illustrating the loss of destructive potential following gene transfer with DN c-Raf-1. In this context, we could show that DN c-Raf-1 reduced the phosphorylation of ERK-1/2 and JNK and inhibited the expression of MMPs. Of note, our DN c-Raf-1 construct constitutes a functional Ras-binding domain of c-Raf-1 and may therefore also act as a competitor for other Ras-mediated signaling events. This is of importance, because Ras has been implicated in the activation of all major MAPKs, ERK-1/2 (43), p38 (44), and JNK (45, 46), but the pathways that result in the activation of ERK and JNK appear distinct (47).

Therefore, our data indicate that DN c-Raf-1 interferes both with the Ras/Raf/ERK pathways (as seen from reduced ERK-1/2 phosphorylation) and with the Ras/JNK pathway (as demonstrated by the inhibition of JNK phosphorylation). However, Li and coworkers demonstrated that in vascular endothelial cells, activation of the Ras/JNK pathway can be inhibited by dominant-negative mutants of Ras, but not by kinase-deficient mutants of c-Raf-1 (48). In addition, data from studies of rheumatoid synovial fibroblasts indicate that ERK may be involved directly in the IL-1–mediated phosphorylation of c-Jun and subsequent up-regulation of MMPs (27). This notion is also supported by most recent data suggesting that the activation of Ras may result in the ERK-mediated phosphorylation of c-Jun in PC12 cells (49). Furthermore, Raf has been shown to interact directly with phosphorylated Jun (50). It is important to note that in our studies, DN c-Raf-1 inhibited the expression of phosphorylated c-Jun in vivo and resulted in a clear down-regulation of disease-relevant MMPs both in vitro and in vivo. This finding, taken together with the findings discussed above, indicates that at least two distinct Ras-mediated pathways are likely involved in this down-regulation of MMPs.

The suppression of cartilage degradation through inhibition of c-Myc is also of interest. So far, the question of whether alterations in the expression of c-Myc contribute specifically to the activation and invasiveness of RASFs (5) has not been investigated. Instead, there have been conflicting data, some of which have suggested a contribution of c-Myc to synovial proliferation (30), while others have pointed to apoptosis (51) or indicated no specific role of c-Myc in RA (52, 53). Although c-myc mRNA does not appear to be increased in RASFs, c-Myc protein has been detected at sites of destruction (40). The present data demonstrate clearly that down-regulation of c-Myc through gene transfer of dominant-negative mutants reduces the invasive behavior of RASFs, even at a level where alterations in growth and apoptosis cannot be detected.

Collectively, our data suggest that signaling pathways involving Ras and c-Myc constitute important features in the activation of synovial fibroblasts in RA. Inhibition of either molecule alone reduces the destructive potential of RASFs, but fails to affect growth and apoptosis of these cells significantly. In contrast, combined inhibition of both protooncogenes strongly induces apoptosis in RASFs and may therefore completely prevent cartilage damage. Tailored inhibition of Ras and c-Myc may therefore constitute a novel, promising goal in future treatment scenarios for RA.

Acknowledgements

The authors wish to thank F. Pataky and E. Jeisy for their expert technical assistance.

Ancillary