Control of fibroblast-like synoviocyte proliferation by macrophage migration inhibitory factor




The hyperplasia of fibroblast-like synoviocytes (FLS) is considered essential to the evolution of joint destruction in rheumatoid arthritis (RA), but the mechanisms underlying FLS proliferation remain poorly understood. Macrophage migration inhibitory factor (MIF) is a cytokine that has recently been shown to exert proinflammatory effects on RA FLS. This study sought to identify the mechanisms of activation of FLS by MIF, and to assess the effects of MIF on synovial cell proliferation.


Human RA FLS were treated with recombinant MIF, interleukin-1β (IL-1β), tumor necrosis factor α (TNFα), and/or anti-MIF monoclonal antibodies (mAb). Proliferation was measured with tritiated thymidine incorporation. Nuclear factor κB (NF-κB) and mitogen-activated protein (MAP) kinase activation were measured with immunohistochemistry and Western blotting, respectively.


FLS proliferation was significantly increased by MIF. IL-1β and TNFα also induced proliferation, but these effects were prevented by neutralization with anti-MIF mAb. Activation of NF-κB was induced by IL-1β, but not by MIF. Anti-MIF mAb had no effect on IL-1β–induced NF-κB nuclear translocation. By contrast, MIF induced phosphorylation of extracellular signal–regulated kinase (ERK) MAP kinase. ERK antagonism, but not NF-κB antagonism, prevented the effect of MIF on FLS proliferation.


These data suggest that MIF may regulate RA synovial hyperplasia by acting directly and via involvement in the effects of IL-1β and TNFα. In addition, the effects of MIF on FLS activation are independent of NF-κB, and dependent on ERK MAP kinase. These data suggest an important therapeutic potential for MIF antagonism in RA.

The synovial lesion of rheumatoid arthritis (RA) is characterized by the presence of chronic inflammation and synovial hyperplasia. Although cells originating in the bone marrow, including macrophages and CD4+ T cells, are increased in RA synovium, there is no evidence that these cells proliferate within the synovium. In contrast, evidence of in situ proliferation of fibroblast-like synoviocytes (FLS) suggests that these cells are locally expanded (1, 2). The hyperplasia of FLS is regarded as being of pivotal importance for the development of pannus, the expanded destructive synovial tissue responsible for cartilage and bone erosion in RA (3). Pathways controlling the proliferation of FLS are therefore of considerable interest. For example, interleukin-1 (IL-1) stimulates the proliferation of FLS in vitro (4) and is known to act via signal transduction pathways including nuclear factor κB (NF-κB) and mitogen-activated protein (MAP) kinases (5, 6).

Macrophage migration inhibitory factor (MIF) is increasingly recognized as an important regulatory cytokine in immune and inflammatory responses. MIF is a product of activated macrophages, T cells, and endothelial cells, and up-regulates the proinflammatory activity of these cells (7–9). In RA synovium, MIF is expressed by FLS as well as macrophages, and FLS-derived MIF up-regulates the release of monocyte tumor necrosis factor α (TNFα), suggesting that MIF acts as an upstream member of the network of cytokines operative in RA (10). Recent evidence also suggests that MIF directly activates the expression of phospholipase A2 (PLA2) and cyclooxygenase 2 (COX-2) by RA FLS, and is essential for the up-regulation of these molecules by IL-1 (11). Indirect evidence for a role of MIF in synovial proliferation is found in rodent models of RA, in which MIF monoclonal antibody (mAb) antagonism profoundly inhibits disease severity and synovial hypercellularity (12–14). We sought to test the hypothesis that MIF is an important regulator of synovial hyperplasia, by investigating the effects of MIF on FLS proliferation. We also examined the effects of MIF on synoviocyte NF-κB and MAP kinase activation.


Isolation and culture of FLS.

FLS were obtained from the synovium of patients with RA who were undergoing joint replacement surgery. All patients satisfied the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for the classification of RA (15). FLS were isolated using enzyme digestion and cultured in RPMI/10% fetal calf serum (FCS) (ICN, Melbourne, Australia) as previously described (10). In brief, a single cell suspension was obtained by digesting minced synovial tissue with 2.4 mg/ml dispase (grade II, 5 units/mg; Boehringer Mannheim, Melbourne, Australia), 1 mg/ml collagenase (type II; Sigma, Melbourne, Australia), and DNase (type I; Boehringer Mannheim). FLS were propagated in 10-cm culture plates in RPMI (ICN Biomedicals, Cincinnati, OH)/10% FCS (Trace Biosciences, Melbourne, Australia) at 37°C in a 5% CO2–humidified incubator. Cells beyond the third passage were >99% CD45-negative. Cells were used between passages 4 and 9. In each group of experiments, the number of samples studied is the number of individual human RA donor FLS used.

Measurement of FLS proliferation.

To determine the effect of recombinant human MIF (16) on FLS proliferation, DNA synthesis was measured by 3H-thymidine incorporation. FLS were seeded overnight at 0.5 × 105 cells per well in 24-well tissue culture plates in RPMI/10% FCS. Cell growth was synchronized by culturing FLS in RPMI/0.1% bovine serum albumin (Sigma) for 24 hours. FLS were treated with human recombinant MIF (5–500 ng/ml), anti–human MIF mAb (200 μg/ml; or isotype-matched negative control, IgG1), and/or human recombinant IL-1β (0.1 ng/ml) or human recombinant TNFα (1 ng/ml) for 54 hours prior to cells being pulsed for 18 hours with 1 μCi/ml 3H-thymidine (Amersham International, Sydney, Australia). Duplicate or triplicate cultures were used for each determination. FLS were detached using trypsin/EDTA and harvested using a cell harvester, and the radioactivity incorporated into DNA was determined using a Wallac 1409 liquid scintillation counter (Pharmacia, Turku, Finland).

Analysis of NF-κB.

FLS (1 × 105) were treated with human recombinant MIF (5–500 ng/ml), human recombinant IL-1β (0.1 ng/ml), and/or anti–human MIF mAb (50 μg/ml; or isotype-matched negative control, IgG1) for 30 minutes. Cells were then centrifuged onto Superfrost Plus microscope glass slides (Selby Laboratories, Melbourne, Australia) using a cytospin centrifuge (Shandon, Pittsburgh, PA).

Single immunohistochemical staining was performed using a streptavidin–biotin method as described previously (17), with the following modifications. The sections were incubated for 1 hour with blocking agent that consisted of 10% FCS/10% swine serum/0.05% azide in Tris buffered saline solution. The sections were then incubated for 1 hour at room temperature with the primary antibody, that is, either goat anti-human NF-κB p50 polyclonal antibody at 1:150 dilution (1.3 μg/ml; Santa Cruz Biotechnology, Santa Cruz, CA), goat anti-human NF-κB p65 polyclonal antibody at 1:800 dilution (0.25 μg/ml; Santa Cruz Biotechnology), or negative control goat anti-human NF-κB p50 (1.3 μg/ml) and p65 (0.25 μg/ml) polyclonal antibodies raised against p65 and p50 blocking peptides (Santa Cruz Biotechnology). The primary antibody was detected using a biotinylated swine anti-goat IgG (Dako, Sydney, Australia). This was followed by incubation with streptavidin–horseradish peroxidase (Dako) and the chromagen substrate 3,3′-diaminobenzidine tetrachloride (Dako). The sections were counterstained with Mayer's hematoxylin (Sigma) and then mounted and examined under light microscopy. Nuclear translocation of p50 or p65 was identified by nuclear staining, as previously described (18). To determine the percentage of cells exhibiting nuclear staining, 100 cells were counted on 3 fields of each cytospin with the use of a graticule, and the results are expressed as the mean (± SEM) percentage of nuclear-stained cells.

Nuclear translocation of NF-κB was blocked using SN50 (Biomol Research Laboratories, Plymouth Meeting, PA) or a negative control peptide, according to the manufacturer's instructions. SN50 inhibits translocation of the NF-κB active complex into the nucleus (19).

Analysis of MAP kinase activation.

The phosphorylation of p44/42 MAP kinase, or extracellular signal–regulated kinase (ERK), was assessed using Western blotting with mAb specific for the phosphorylated (activated) form of ERK, as described previously (20). In brief, cells were disrupted by repeated aspiration through a 21-gauge needle. After incubation on ice for 10 minutes and microcentrifugation at 3,000 revolutions per minute for 15 minutes (4°C), the supernatants were removed, the protein concentration was determined, and the lysates were stored at −80°C. Equal amounts of cellular proteins were fractionated on 10% sodium dodecyl sulfate–polyacrylamide electrophoresis gels and transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA). Immunoblotting was performed using antibodies directed against phospho-p44/42 (ERK) and total p44/42 according to the manufacturer's instructions (Cell Signaling Laboratories, Beverly, MA). The activation of ERK was antagonized with a specific inhibitor of MAP kinase kinase 2, PD98059 (Alexis Biochemicals, San Diego, CA). The activation of p38 MAP kinase was antagonized with SB203580 (kindly provided by Dr. Alison Badger from SmithKline Beecham Pharmaceuticals, King of Prussia, PA).

Statistical analysis.

Results are expressed as the mean ± SEM. Statistical analysis was performed using Student's t-test, with P values of less than 0.05 regarded as statistically significant.


Effect of MIF on FLS proliferation.

Constitutive proliferation of serum-exposed FLS was readily detected. FLS proliferation was significantly up-regulated by MIF (5–500 ng/ml) (P < 0.02) (Figure 1). A bell-shaped concentration-response curve was observed, with maximal MIF effect observed at 50 ng/ml.

Figure 1.

Effect of macrophage migration inhibitory factor (MIF) on proliferation of rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS). RA FLS were treated with recombinant MIF (rMIF) (5–500 ng/ml) for 54 hours and labeled with 1 μCi/ml 3H-thymidine for 18 hours. FLS proliferation was determined from duplicate cultures and expressed as 3H-thymidine incorporation (in counts per minute). Bars show the mean and SEM of 6 separate experiments. ∗ = P < 0.02 versus untreated FLS.

MIF has recently been shown to be essential for the induction of FLS cytosolic PLA2 (cPLA2) and COX-2 expression by IL-1β. The effect of FLS-derived MIF on IL-1β– and TNFα-induced proliferation was determined using an anti-MIF mAb. IL-1β and TNFα significantly induced FLS proliferation (P < 0.05) (Figure 2). Control IgG had no inhibitory effect on FLS proliferation (data not shown), but anti-MIF mAb prevented both the IL-1β– and the TNFα-induced proliferation (Figure 2). MIF mAb had no significant effect on constitutive FLS proliferation, although a trend toward reduced proliferation was observed (data not shown).

Figure 2.

Effect of anti-MIF monoclonal antibodies (mAb) on constitutive and stimulated FLS proliferation. FLS were treated with interleukin-1β (IL-1) (0.1 ng/ml), tumor necrosis factor α (TNF) (1 ng/ml), and anti-MIF mAb (200 μg/ml) or isotype-matched control IgG1. FLS proliferation was determined from triplicate cultures and expressed as 3H-thymidine incorporation (in counts per minute). Bars show the mean and SEM of 4 separate experiments. IL-1 and TNF significantly increased FLS proliferation (∗ = P < 0.05 versus medium-treated cells). In contrast, in the presence of anti-MIF mAb, no significant effect of IL-1 and TNF on FLS proliferation was detectable († = P not significant versus medium-treated cells). See Figure 1 for other definitions.

Effect of MIF and IL-1β on NF-κB activation.

The regulation of FLS proliferation has been previously linked to the activation of NF-κB (18). As has previously been reported, we detected constitutive cytoplasmic, but not nuclear, staining of the p50 and p65 subunits of NF-κB in untreated RA FLS (Figure 3 and Table 1). Induction of proliferation by IL-1β was associated with nuclear translocation of p50 and p65 (Figure 3 and Table 1). In contrast, MIF did not induce nuclear translocation of p50 and p65 (Figure 3 and Table 1), suggesting that the direct induction of proliferation of RA FLS by MIF is not dependent on the activation of NF-κB. To assess the influence of MIF on IL-1β–induced NF-κB activation in FLS, the inhibitory effect of anti-MIF mAb was investigated. IL-1β–induced p50 and p65 translocation was not inhibited by anti-MIF mAb, under conditions inhibitory of IL-1β–induced FLS activation (Figure 4 and Table 1). Treatment with a specific antagonist of NF-κB translocation, SN50, did not inhibit MIF-induced FLS proliferation (Figure 5).

Figure 3.

Nuclear translocation of nuclear factor κB (NF-κB) in IL-1β– or MIF-treated FLS. FLS were treated with medium alone (a–d), IL-1β (0.1 ng/ml) (e and f), or MIF (500 ng/ml) (g and h). Sections were immunostained with a negative control p50 antibody (a) or negative control p65 antibody (b) or antibodies directed against the p50 (c, e, and g) or p65 (d, f, and h) subunits of NF-κB. NF-κB activation is reflected as nuclear translocation of the p50 and p65 subunits, as seen in the IL-1β–treated cells (e and f) but not in the control or MIF-treated cells. Results are representative of 4 separate experiments; similar results were obtained for cells treated with 50 ng/ml MIF. See Figures 1 and 2 for other definitions.

Table 1. Nuclear translocation of NF-κB in IL-1β– or MIF-treated FLS*
 Nuclear p50Nuclear p65
  • *

    Values are the mean ± SEM percentage of nuclear-stained cells. Rheumatoid arthritis fibroblast-like synoviocytes (FLS) were treated with medium alone, macrophage migration inhibitory factor (MIF) at 50 or 500 ng/ml, or interleukin-1β (IL-1β; 0.1 ng/ml) in the presence of negative control (NC) IgG or anti-MIF monoclonal antibodies (mAb) (200 μg/ml). Cytospin slides were immunostained with antibodies directed against p50 or p65 subunits of nuclear factor κB (NF-κB). One hundred cells were counted on 3 fields of each cytospin to determine the percentage of positively stained cells, with the use of a graticule.

  • P < 0.01 compared with medium-treated cells.

  • P = not significant compared with IL-1β + NC mAb–treated cells.

MIF 50 ng/ml00
MIF 500 ng/ml00
IL-1β + NC mAb77 ± 275 ± 4
IL-1β + anti-MIF mAb76 ± 272 ± 4
Figure 4.

Nuclear translocation of NF-κB in IL-1β–stimulated FLS cotreated with anti-MIF mAb. FLS were treated with IL-1β (0.1 ng/ml) and negative control mAb IgG1 (50 μg/ml) (a and c, respectively) or IL-1β (0.1 ng/ml) and anti-MIF mAb (50 μg/ml) (b and d, respectively) for 30 minutes, and immunostained for the p50 (a and b) or p65 (c and d) subunits of NF-κB. Results are representative of 4 separate experiments. See Figures 1–3 for definitions.

Figure 5.

Effect of NF-κB inhibition on MIF-induced FLS proliferation. FLS were treated with MIF (50 ng/ml) plus control peptide/vehicle (C) or SN50 peptide (20 μg/ml) for 54 hours and labeled with 1 μCi/ml 3H-thymidine for 18 hours. FLS proliferation was determined from duplicate cultures and expressed as 3H-thymidine incorporation (in counts per minute). Bars show the mean and SEM of 4 experiments. MIF-induced FLS proliferation in the presence of control peptide (∗ = P < 0.05 compared with no treatment) and SN50 had no inhibitory effect. See Figures 1 and 3 for definitions.

Effect of MIF on ERK MAP kinase.

The involvement of ERK (also known as p44/42 MAP kinase) in the activation of 3T3 murine fibroblasts by MIF has been reported (20), but no study has examined this activation pathway in human cells. MIF induced the phosphorylation of ERK in RA FLS in a dose-dependent manner (Figure 6). Moreover, antagonism of ERK phosphorylation using the specific inhibitor PD98059 completely inhibited the stimulatory effect of MIF on FLS proliferation (Figure 7). PD98059 also inhibited constitutive proliferation of RA FLS, consistent with the role of ERK in the response to serum stimulation (6). In contrast, inhibition of p38 MAP kinase activity using the inhibitor SB203580 had no effect on MIF-induced proliferation (Figure 7).

Figure 6.

Induction of extracellular signal–regulated kinase (ERK) mitogen-activated protein (MAP) kinase phosphorylation by MIF in RA FLS. RA FLS were rendered quiescent by overnight culture in serum-free medium. Cells were stimulated for 4 hours with increasing concentrations of recombinant human MIF (rMIF). Lysates were examined for MAP kinase activation by Western blotting with a phospho-specific ERK MAP kinase monoclonal antibody. Total ERK-specific antibody was used to verify equal protein loading. Results are 2 representative examples of the findings from 4 independent experiments, with 33 and 37L representing RA tissue donor numbers. See Figure 1 for other definitions.

Figure 7.

Effects of inhibition of MAP kinase on the proliferation of FLS. RA FLS were treated with MIF with and without the addition of the ERK MAP kinase inhibitor PD98059 (50 μM) or the p38 MAP kinase inhibitor SB203580 (5 μM). MIF significantly increased RA FLS proliferation. Inhibition of ERK substantially inhibited FLS proliferation, whereas inhibition of p38 MAP kinase had no significant effect. Bars show the mean and SEM of 4 separate experiments. ∗ = P < 0.05 compared with vehicle (veh)–treated cells; † = P < 0.05 compared with MIF-treated cells without inhibitor; § = P < 0.05 compared with vehicle-treated cells. See Figures 1 and 6 for other definitions.


In recent years, the expression of many cytokines has been reported in RA synovium. IL-1β and TNFα have been regarded as critical cytokines in the evolution of RA, partly because application of specific treatment strategies in humans has been successful. Both of these cytokines contribute to inflammation in the synovial lesion, but effective inhibition of these cytokines still leaves significant inflammation and joint destruction in RA, which suggests that other critical factors may be present (21, 22).

The biologic functions of MIF in the immune system include activation of monocyte TNFα and innate immunity (23, 24), and MIF plays an essential role in T cell activation and models of T cell–directed immunity (25, 26). This range of functions suggests the involvement of MIF in diseases such as RA, in which activation of both innate and adaptive immunity may be involved. Considerable evidence directly supports the hypothesis that MIF is involved in RA. The contribution of MIF to arthritis has been demonstrated in numerous animal models of RA, including rat adjuvant arthritis (13), murine collagen-induced arthritis (12), and murine antigen-induced arthritis (14). MIF is expressed abundantly in human RA synovial FLS, macrophages, and, to a lesser extent, T cells, and is overexpressed in RA tissues and cells compared with that in healthy controls (10, 27). We have reported the induction of monocyte cell TNFα release by FLS-derived MIF (10), and MIF has also recently been shown to induce cPLA2 and COX-2 expression by FLS (11) and matrix metalloproteinase activity (28). MIF was also shown to be essential for the activating effects of IL-1β on FLS, in that inhibition of MIF using mAb prevented IL-1β activation of FLS cPLA2 and COX-2 (11).

Activation of inflammation is of considerable importance in RA, but the progressive destruction of articular cartilage is reliant on the evolution of hyperplastic and invasive synovial tissue. Hyperplasia of FLS is dependent on dysregulated proliferation and apoptosis (2, 3, 29). Factors considered important for the induction of proliferation include proinflammatory cytokines such as IL-1β, acting at least in part through the NF-κB and MAP kinase signal transduction pathways (4, 6, 30, 31). MIF has been reported to be involved in the control of cell proliferation in contexts other than RA, including tumors and cell lines (9, 19, 32). Moreover, MIF has been identified as capable of inducing phosphorylation of ERK MAP kinase in murine fibroblasts, and of overriding p53-dependent cell cycle control (20, 33).

The present study demonstrates the ability of MIF to stimulate the proliferation of cultured FLS derived from human RA tissues. The concentration range of MIF that induced proliferation is similar to the mean concentration of 20 ng/ml reported in human RA synovial fluid (10) and the range that induces FLS cPLA2 and COX-2 expression (11). A bell-shaped curve of the MIF effect has been previously noted with regard to other phenomena (16) and may relate to an apparent inhibitory effect of MIF in intracellular signal transduction events at very high concentrations (34). Moreover, the ability of IL-1β and TNFα to induce FLS proliferation was shown to be dependent on MIF, in that neutralization of MIF with mAb prevented activation of FLS proliferation by these stimuli. Because these cytokines coexist in rheumatoid synovium, possible additive effects of MIF and other cytokines would be of interest to investigate. These data suggest that, in addition to the important effects of MIF on T cell and monocyte activation, and induction of inflammatory eicosanoid synthesis by FLS, MIF is involved in the control of FLS proliferation. Reduced synovial cellularity was a feature of synovial tissue from rats and mice receiving anti-MIF mAb during the evolution of arthritis, although factors other than proliferation may have contributed to this effect (13, 14). Given that FLS proliferation is necessary for the generation of pannus, and given evidence of the ability of MIF to induce FLS metalloproteinase expression (28), these results suggest a critical contribution of MIF to the pathology of RA joint damage.

The mechanism of action of MIF has not been fully elucidated. Activation of synovial cells has variously been reported to be dependent on NF-κB and/or MAP kinase pathways (4, 6, 31, 35–38). The signal transduction pathways utilized by MIF in human cells have not been described. We have demonstrated the ability of MIF to activate ERK MAP kinase, and have shown that induction of FLS proliferation by MIF is dependent on ERK MAP kinase. Constitutive expression of activated ERK MAP kinase has been reported in RA tissues and cultured RA FLS, and the up-regulation of ERK phosphorylation by other cytokines has been identified (6). The current finding that ERK pathway inhibition inhibited constitutive (serum-induced) RA FLS proliferation is consistent with these observations. Although interpretation of the effects of ERK pathway inhibition on MIF-induced cell activation is difficult given the effects on constitutive proliferation, it is clear that MIF-induced proliferation requires the activation of ERK to proceed. MAP kinases, including ERK, can influence inflammatory gene expression without influencing NF-κB nuclear translocation and DNA binding (39). Our data suggest such a pathway is operative in MIF activation of FLS proliferation.

NF-κB has also been reported to be a pivotal factor in the regulation of FLS proliferation in RA (18, 40). Despite the effect of MIF on FLS proliferation, we found no evidence for the activation of NF-κB by MIF. Moreover, although antagonism of MIF potently inhibited IL-1β–induced activation of FLS, this antagonism had no effect on IL-1β–induced NF-κB nuclear translocation. In addition, antagonism of NF-κB activation had no effect on MIF-induced proliferation, further supporting the concept that cell activation by MIF is independent of NF-κB. Although not previously studied in FLS, this conclusion is supported by the findings of studies such as that of Kleemann et al (34), who found no involvement of MIF in TNF-induced NF-κB activation, and Daun and Cannon (41), who also reported that MIF did not directly influence NF-κB activation. The current results therefore imply that induction of cPLA2 and COX-2 expression by FLS, achieved with identical doses of MIF under similar conditions (11), is also independent of NF-κB activation, a finding supported by the lack of effect of NF-κB antagonism on MIF-induced PLA2 activity in FLS (Lacey D, et al: unpublished observations).

In conclusion, we have shown that MIF has significant effects on FLS proliferation, through mechanisms that are independent of NF-κB but involve activation of ERK MAP kinase. These data suggest that MIF is a significant contributor to synovial hyperplasia in RA. Taken together with the effects of MIF on TNFα and eicosanoid production, these data suggest that targeting MIF in the treatment of RA could have significant antiinflammatory and disease-modifying effects.