Sex and age matched patients with RA and healthy controls were used as a source of synoviocytes. Synovial tissue was obtained from four RA patients undergoing surgery for synoviectomy or joint replacement, and from four controls undergoing orthopaedic surgery for knee trauma. Informed consent was obtained from the patients enrolled in this study and ethics approval was given by the local ethics committee.

Synovia was removed from knee joints, cut and subjected to a mild proteolytic treatment (0.05% trypsin, 0.5 mM EDTA in phosphate buffer saline, for 10 minutes at 37°C under gentle shaking). Trypsin was neutralized with FCS (Celbio, Milano, Italy) and cells were plated in culture dishes with RPMI 1640 (Cambrex Bio Science, Milano, Italy) supplemented with 10% FCS, 2 mM glutamine (Cambrex Bio Science) and penicillin-streptomycin (Cambrex Bio Science). Cell monolayers were used within the seventh passage in culture. The cells were considered type B fibroblast-like synoviocytes if negative by staining with anti-CD69, anti-CD14, anti-CD11b and anti-CD11c (Santa Cruz Biotechnology, Santa Cruz, CA, USA), positive by staining for the enzyme uridine-diphospho glucose dehydrogenase, and if they had a spindle-shaped, fibroblast-like morphologic appearance.

RAL (Lilly, Sesto Fiorentino (FI), Italy) was dissolved in methanol 100 mmol and further diluted with culture medium. Samples were analyzed both in basal conditions and after treatment with RAL (48 h). In preliminary experiments, we tested the effects of different doses of RAL in order to identify which dose would be more effective on synoviocytes without a lethal effect. Thus, the concentrations of 0.5 μM and 1 μM RAL were chosen for further experiments and used in this study. These concentrations were approved by the manufacturer (Lilly) and seem to be comparable with normal therapeutic doses, even if it is impossible to know which dose really reaches the synovia.

A431 is a human epidermoid cancer cell line from a cervix squamous cell carcinoma. A431 cells were cultivated in DMEM with 4.5 g/L glucose (Cambrex Bio Science), supplemented with 10% FCS, 2 mM glutamine and penicillin-streptomycin. When cells were at confluence they were washed and incubated with DMEM 0.2% FCS for 48 h. Then the conditioned medium was centrifuged to remove particles and stocked at -80°C. It was used as a chemoattractant reagent in the chemoinvasion assay because of its known property to secrete chemoattractant soluble factors 25 26 27 .

Cell growth was quantified in subconfluent cell monolayers. Synoviocytes were seeded in 24 multi-well plates (15,000 cells/well) with 10% FCS in RPMI 1640. After 48 h incubation, cells were washed three times with serum-free medium and incubated in 0.2% FCS medium for an additional 48 h. Cells were then incubated for 48 h in 10% FCS medium (positive control); 0.2% FCS medium (negative control); 0.2% FCS with u-PA (Serono, Roma, Italy) 500 ng/ml with/without u-PA or u-PAR antagonists. The agonists were anti-human u-PA monoclonal antibody 5B4 (mAb 5B4) and anti-u-PAR monoclonal antibody 3936 (mAb 3936) (American Diagnostica, Montreal, Canada), which were used at 1.5 μg/ml. Both mAb 5B4 and mAb 3936 sterically impede the u-PA/u-PAR interaction and were developed against the u-PA-binding site of u-PAR. In preliminary experiments, we tested the effects of different doses of u-PA to identify what dose would be more efficacious in inducing synoviocyte proliferation. Thus, the concentration of 500 ng/ml u-PA was chosen.

Each experimental point was performed in triplicate. At the end of incubation cells were counted. Samples were analyzed both in basal conditions and after treatment with RAL 0.5 μM and 1 μM (48 h).

Samples were analyzed for u-PAR, u-PA and PAI-1 using commercially available ELISA kits (IMUBIND, American Diagnostica, Montreal, Canada) according to the manufacturer's instructions. Briefly, synoviocytes were seeded in six multi-well plates (25,000 cells/well) with 10% FCS in RPMI 1640. After 48 h of incubation, cells were washed three times with serum-free medium and incubated in 0.2% FCS medium for an additional 48 h. Cells were then treated with 0.5 μM or 1 μM RAL for 48 h. At the end of incubation, cells were detached, counted and lysed by a lysis buffer, as suggested by the manufacturer. The lysates were replaced in their original well and incubated for 1 h at 4°C to allow exhaustive extraction of undetached material. Cell extracts were centrifuged and stocked at -80°C until u-PAR analysis. Culture mediums were collected, centrifuged and stocked at -80°C until u-PA and PAI-1 determination. The results were correlated to the standard curve, within the range of linearity. Each sample was evaluated in triplicate and with two different dilutions. The u-PA assay reveals u-PA antigen, independent of its catalytic activity; the u-PAR and PAI-1 assays measure both occupied and unoccupied receptor, and both free and u-PA-coupled PAI-1, respectively.

The sensibility levels were: 10 pg of u-PA/ml of sample; 0.1 ng of u-PAR/ml of sample; 1 ng of PAI-1 /ml of sample.

u-PA enzymatic activity was evaluated by zymography. Culture medium samples were collected as described above and concentrated by centrifugation at 8,000 rpm for 30 minutes in centricon tubes (Amicon Division, Beverly, MA, USA) with 30 kDa molecular weight cut-off pores. The samples were subjected to SDS-PAGE (10%) under non-reducing conditions and migrated proteins were transferred onto 0.45 μm pore-size nitrocellulose filters (Bio-Rad Laboratories, Richmond, California, USA) in a 0.04 M phosphate buffer (pH 6.5) and run for 2 h under a current of 0.4 A. The nitrocellulose filter was removed and placed on an indicating layer containing casein and plasminogen. After overnight incubation at 37°C, u-PA digestion of plasminogen showed clear bands of lysis in the cloudy casein background, corresponding to the position of plasminogen activators in the polyacrylamide gel. u-PA 0.1 U/ml was used as a positive control. Zymograms were then scanned. Samples were analyzed both in basal conditions and after treatment with 0.5 μM or 1 μM RAL (48 h).

Because zymography was performed after protein separation by SDS-PAGE, which uncoupled u-PA from PAI-1, we used a direct fibrinolytic assay of cell membrane-associated plasminogen activators to study the final fibrinolytic balance of synoviocytes. Briefly, synoviocyte monolayers were treated with acidic wash (50 mM glycine-HCl buffer, pH 3.0, 0.1 M NaCl) to uncouple both u-PA and u-PA/PAI-1 complexes from cell surface u-PAR; Hepes buffer 0.5 M NaCl, pH 7.5, was added after 3 minutes. Aliquots of the acidic wash were dotted on fibrin plates prepared according to a method previously described 17 and fibrinolytic activity was evaluated after 16 h of incubation at 37°C, measuring the diameter of the lysis areas. Each sample was evaluated in triplicate.

u-PA was subjected to radioiodination with ¹²⁵I-Na, using Iodogen (Pierce Eurochemie BV, Holland), following the instructions of the manufacturer. Iodinated protein was separated from non-incorporated radioactivity by gel filtration on Sephadex G-25 (Pharmacia, Milan, Italy). The specific activity obtained in the preparation used in this study was 17.8 μCi/μg. Saturation binding experiments were performed as previously described 28 . Binding experiments were performed either on untreated cell monolayers or following acidic wash (50 mM glycine-HCl buffer, pH 3.0, 0.1 M NaCl) to uncouple u-PA from cell surface u-PAR; Hepes buffer 0.5 M NaCl, pH 7.5, was added after 3 minutes.

The Boyden chamber procedure was used to evaluate cell migration 14 . The method is based on the passage of cells across porous filters against a concentration gradient of the migration effector. A 48-well micro-chemotaxis chamber (Neuroprobe, Gaithersburg, MD, USA) was used. The two wells were separated by a polyvinyl-pyrrolidine-free polycarbonate filter with an 8 μm pore size (Neuroprobe). To evaluate chemoinvasion, the filter was coated with Matrigel (50 μg/filter) (Becton Dickinson Labware, Two Oak Park, Bedford, MA, USA). Test solutions were dissolved in serum-free medium and placed in the lower wells. 50 μl of cell suspension (12,500 cells) were added to each upper well. u-PA 100 ng/ml and/or RAL (0.5 μM or 1 μM) were added to the upper and/or lower well. In the experiment with neutralizing antibodies, the anti-u-PA mAb 5B4 (1.5 μg/ml) was placed in the lower wells, while the anti-u-PAR mAb 3936 (1.5 μg/ml) was incubated with the cell suspension. The chamber was incubated at 37°C for 5 h after which the filter was removed and fixed with methanol. Non-migrating cells on the upper surface of the filter were removed with a cotton swab. Cells were stained with Diff-Quick (Mertz-Dade AG, Dade International, Milan, Italy) and counted using a light microscope (40×) in 10 random fields for each well. Mobilization was measured by the number of cells moving across the Matrigel and the filter pores and spread on the lower surface of the filter. Each experimental point was performed in triplicate. Mean values of migrated cells for each experimental point were calculated.

A non-parametric Mann-Whitney test for independent samples was used to compare results from healthy and RA synoviocytes for the levels of u-PA, u-PAR and PAI-1. The results were expressed as mean ± standard deviation. The proliferation induced by u-PA and RAL was evaluated by two tailed t-test for independent samples and by analysis of variance (ANOVA with Bonferroni correction). Migration was measured as a percentage of the healthy basal response and was expressed as the mean ± standard error of the mean.

u-PA induced a dose-dependent proliferation in both normal and RA synoviocytes, reaching a maximum at 500 ng/ml (p < 0.001 for any dose in both cell lines) (Fig. 1). Proportionally, u-PA increased cell proliferation in RA and healthy synoviocytes, without reaching a significant difference between the two cell lines at any u-PA dose.

These results show that u-PA has a pro-proliferative effect on both healthy and RA synoviocytes and identify 500 ng/ml u-PA as the minimal dose providing the maximal proliferative effect. This dose has been used in the following experiments on u-PA-dependent proliferation.

RAL inhibited cell proliferation in a dose-dependent manner in both normal (p < 0.001) and RA synoviocytes (p < 0.001), reaching a maximal effect at 1 μM (Fig. 2a). The RAL induced decrease in cell proliferation did not significantly differ between RA and healthy synoviocytes for any RAL dose. These results indicate that RAL exerts an anti-proliferative effect on both healthy and RA synoviocytes that is effective at doses as small as 0.5 μM.

In both normal and RA synovial cells, RAL significantly reduced u-PA- and 10% FCS-dependent proliferation (p < 0.001) (Fig. 2b). RA synoviocytes are as prone as normal synoviocytes to spontaneous (0.2% FCS) proliferation and to proliferation challenged both with 10% FCS and 500 ng/ml u-PA (Fig. 2b). In particular, serum-dependent proliferation (10% FCS) did not significantly differ from u-PA-dependent proliferation in both healthy and RA cell lines.

The proliferative effect elicited by 500 ng/ml of u-PA in 0.2% FCS was significantly reduced by the mAb antagonists of u-PA (mAb 5B4) and u-PAR (mAb 3936) in healthy (p < 0.001) and RA synoviocytes (p < 0.001) (Fig. 2b). This indicates that u-PA/u-PAR interaction is required for the pro-proliferative effect of u-PA on normal and RA synoviocytes.

RAL blocked both FCS-dependent and u-PA-dependent proliferation to a similar extent. RAL reduction of u-PA-dependent proliferation was similar to the inhibition caused by the antagonist antibodies (p < 0.001 in normal and RA synoviocytes) (Fig. 2b) and was not increased by co-treatment with them, indicating a common target for RAL and anti u-PA/u-PAR antibodies (data not shown). This suggests that inhibition of the expression of critical members of the cell-associated fibrinolytic system is a likely target of RAL treatment.

As our ELISA assay for u-PA measures u-PA antigen independent of its interaction with PAI-1, the data shown in Fig. 3a indicate that u-PA released into the culture medium by RA synoviocytes was significantly lower than u-PA produced by healthy synoviocytes (3.1 ± 0.4 versus 10.05 ± 0.04 ng/10⁶ cells, respectivle; p < 0.05).

A zymographic assay of u-PA performed on aliquots of the culture medium confirmed the data obtained using antibodies. Culture mediums of RA synoviocytes display a lower u-PA activity than medium obtained from healthy synoviocytes (Fig. 4a).

The direct fibrinolytic assay, performed on fibrin plates as described, indicated the following: the culture medium of healthy synoviocytes showed a higher fibrinolytic activity than RA synoviocytes, thus confirming the data obtained with zymography, again indicating lower constitutive u-PA production in RA synoviocytes (Fig. 4b, left side); and the fibrinolytic activity exhibited by aliquots of the acidic wash was slightly higher in RA than in healthy synoviocytes (Fig. 4b, right side). This means that, regardless of the lower amount of secreted u-PA, a larger availability of u-PAR on the surface of RA synoviocytes enabled RA synovial cells to bind a higher amount of the ligand. These results have also been confirmed by radioligand binding experiments. Healthy synoviocytes exhibited 95 ± 15 × 10³ u-PAR/cell, with a Kd ranging from 1.75 to 1.85 nM, and the receptor number was 140 ± 25 × 10³/cell following acidic treatment prior to performing binding experiments. RA synoviocytes had 180 ± 30 × 10³ uPAR/cell before and 230 ± 48 × 10³ u-PAR/cell after acidic wash, while the Kd did not change. Higher levels of u-PAR are found in cell lysates of RA than of healthy synoviocytes (23.6 ± 0.10 versus 9.2 ± 0.05 ng/10⁶ cells, respectively; p < 0.05), also determined using ELISA (Fig. 3b).

Moreover, RA synovial cells release into culture medium higher amounts of PAI-1 than healthy synoviocytes (5.5 ± 0.1 versus 2.9 ± 0.1 μg/10⁶ cells, respectively; p < 0.05; Fig. 3c).

RAL reduces u-PA levels in healthy and RA synoviocytes (p < 0.0001; Fig. 3a). It does this dose dependently: 1 μM RAL is more efficient than 0.5 μM in healthy (p < 0.01) and in RA synoviocytes (p < 0.05). RAL-dependent reduction of u-PA levels at 0.5 μM and 1 μM is significantly higher (p < 0.05) in healthy than in RA synoviocytes.

A zymographic assay performed on culture medium treated with 0.5 μM and 1 μM RAL confirmed the data obtained with the u-PA assay. In fact, RAL reduces the u-PA enzymatic activity in a dose dependent manner, in both healthy and RA synoviocytes (Fig. 4a).

The direct fibrinolytic assay on fibrin plates, performed with aliquots of culture medium, indicated that RAL-dependent reduction of cell-associated fibrinolytic activity is more pronounced in RA than in normal synoviocytes (Fig. 4b, left side). Fig. 4b (right side) shows a RAL-dependent reduction of u-PA in the acidic wash of both normal and RA synoviocytes.

RAL significantly reduces u-PAR levels in cell lysates from RA synoviocytes (p < 0.0001; Fig. 3b); 1 μM RAL is more efficient than 0.5 μM at reducing u-PAR levels in healthy and RA synoviocytes (p < 0.01). The effect of 0.5 μM and 1 μM RAL is significantly higher in RA than in healthy synoviocytes (p < 0.05). Radioligand binding experiments showed that, upon RAL treatment, the receptor number before and after acidic treatment was similar in healthy synoviocytes, whereas a dramatic decrease of both unoccupied u-PAR (85 ± 18 × 10³ receptor/cell, before acidic wash) and total u-PAR (150 ± 25 × 10³ receptor/cell after acidic wash) was observed in RA synoviocytes.

RAL significantly increases PAI-1 levels in healthy and RA synoviocytes (p < 0.0001; Fig. 3c). This is true for both 0.5 μM and 1 μM RAL (healthy synoviocytes, p < 0.01 and p < 0.001, respectively; RA synoviocytes, p < 0.001 and p < 0.01, respectively). RAL 1 μM is more efficient than RAL 0.5 μM at increasing PAI-1 levels in both healthy and RA synoviocytes (p < 0.001). RAL 0.5 μM and 1 μM have a significantly higher effect on increasing PAI-1 levels in RA than in healthy synoviocytes (p < 0.05).

u-PA-dependent chemoinvasion was dose-dependent (Fig. 5a), with a maximal effect at 100 ng/ml for both healthy and RA synoviocytes. Basal migration was more pronounced (more than 40%) in RA than in healthy synoviocytes. This difference is maintained at each concentration of u-PA in the lower well of the migration chamber. An increase in invasion observed after a 5 h incubation with 100 ng/ml u-PA was counteracted by the incubation of invasive cells with mAb antagonists of u-PA (mAb 5B4) and u-PAR (mAb 3936) (Fig. 5b). This indicates that u-PA/u-PAR interaction is required for the pro-invasive effect of u-PA on normal and RA synoviocytes.

RAL inhibits cell chemoinvasion in a dose-dependent manner in both normal and RA synoviocytes, reaching a maximum at 1 μM (Fig. 5B); this decrease did not significantly differ between healthy and RA cells at any RAL dose.

RAL significantly reduced u-PA-induced chemoinvasion in healthy and RA synoviocytes; 1 μM RAL reduced chemoinvasion to basal values. When cells were incubated together with RAL in the upper chamber, it caused similar effects on chemoinvasion. In fact, RAL was able to inhibit migration induced by u-PA in a dose-dependent way with a maximal action at a dose of 1 μM and this decrease in mobility was proportionally similar in healthy and RA synoviocytes (data not shown).

Upon combining the results shown in Figs 3, 4 and 5, one is tempted to think that RAL works mainly through the reduction in the levels of u-PAR and, therefore, by blocking u-PA-dependent chemoinvasion. In fact, the RAL reduction of u-PA-dependent migration was similar to the inhibition caused by the antagonist mAbs (Fig. 5b) and is not increased by co-treatment with them, indicating that RAL and anti-u-PA/u-PAR antibodies have a common target (data not shown).