The u-PA^-/- mice, provided by Dr P Carmeliet (University of Leuven, Belgium), were backcrossed onto the C57BL/6 background for 11 generations. C57BL/6 CD45 congenic mice, expressing the Ly5.1 allotype, were obtained from Walter and Eliza Hall Institute Animal Supplies (Parkville, Victoria, Australia). All strains were bred in our onsite animal facility, fed standard rodent chow and water ad libitum, and housed in sawdust-lined cages in groups of five. Mice of both sexes, 8 to 12 weeks of age, were used in all experiments. All experiments were approved by The Royal Melbourne Hospital Research Foundation Animal Ethics Committee.

The u-PA^-/- or C57BL/6 recipient mice received total body irradiation (two exposures × 5.5 Gy, 3 hours apart). Bone marrow cells were harvested from the femurs and tibiae of C57BL/6 or u-PA^-/- donor mice expressing the Ly5.1 allotype of CD45. Recipient mice (expressing the Ly5.2 allotype of CD45) were injected intravenously with 5 × 10⁶ bone marrow cells. Effective bone marrow reconstitution was determined by flow cytometry analysis of peripheral blood leukocytes 6 weeks later using the different congenic CD45 allotypes (Ly5.1 and Ly5.2). Six weeks after irradiation, 95 ± 1% of circulating leukocytes expressed the phenotypic marker of the donor bone marrow.

Mice were immunized intradermally in the base of the tail with 100 μg chick CII (Sigma, St Louis, MO, USA), emulsified in an equal volume of complete Freund's adjuvant containing 5 mg/ml heat-killed Mycobacterium tuberculosis (H37 Ra; Difco, Detroit, MI, USA). This procedure was repeated as a boost 21 days later, as previously published 14 .

Animals were assessed for redness and swelling of limbs and a clinical score was allocated for each limb using an established scoring system with slight modifications 14 as follows: 0 = normal; 1 = slight swelling and/or erythema; 2 = extensive swelling and/or erythema; 3 = severe swelling; 4 = rigidity. Severity of arthritis is expressed in terms of the mean clinical score (range 0 to 16 per mouse).

The anti-CII mAb-producing hybridomas for M2139 and CIIC1 were a gift from Prof. R Holmdahl (Karolinska Institute, Stockholm, Sweden). The cocktail of M2139 and CIIC1 mAbs was prepared by mixing equal concentrations of each of the antibody, and mice were then injected intravenously with 4.5 mg mAb cocktail on days 0 and 1. On day 5, mice received intraperitoneally 50 μg lipopolysaccharide (Escherichia coli serotype 0127:B8; Sigma-Aldrich), 5 μg Pam-3-Cys (EMC Microcollections, Tübingen, Germany), or PBS. Mice were scored daily, using the same scoring system as for the CIA model.

K/BxN mice were bred as described previously 19 . Serum was collected up to 12 weeks of age and stored at -80°C. Serum (50 μl in 150 μl PBS) was injected intraperitoneally on days 0 and 2. Mice were scored daily, using the same scoring system as for the CIA model.

Fifty microliters of chicken egg albumin (ovalbumin, 20 mg/kg body weight; Sigma) was injected intravenously followed by intraperitoneal injection of 1 ml rabbit polyclonal IgG rich in antibody to chicken egg albumin (anti-ovalbumin, 800 μg/mouse; Sigma), as previously described 20 . Cells were harvested 4 hours later by lavage with 5 ml ice-cold, sterile PBS. Total and differential cell counts (Diff-Quik; Lab Aids, Narrabeen, NSW, Australia) were performed on the peritoneal exudate cells 21 22 .

In certain experiments, C5a (1.25 μg; HyCult Biotechnology, Uden, The Netherlands) was given intraperitoneally alone or administered together with the anti-ovalbumin, as above. In these experiments, cells were harvested 2 hours later.

At termination following arthritis induction, the rear limbs and ankles were removed, fixed, decalcified, and paraffin embedded, as previously described 14 . Frontal sections (5 μm) were stained with either H & E to examine joint architecture or with safranin O, fast green and hematoxylin for proteoglycan loss, and were evaluated without knowledge of the experimental groups, using the histologic assessment as published 14 . Briefly, infiltration of cells, cartilage damage and bone erosions were all scored separately from 0 (normal) to 3 (severe), and proteoglycan loss was scored from 0 (normal) to 3 (complete loss of staining). These scores were added to give an overall histologic score out of 12.

Fibrin(ogen) deposition was identified in rear limbs as before 14 . Briefly, paraffin-embedded sections were deparaffinized, incubated with 1% (w/v) BSA and 5% (w/v) skim milk powder for 1 hour, and then stained with a goat anti-mouse fibrinogen/fibrin antibody (Accurate Chemical & Scientific, Westbury, NY, USA) overnight at 4°C. Endogenous peroxidase activity was blocked with 0.3% (v/v) H₂O₂ (Sigma) in methanol. Following washing, sections were incubated with a biotinylated donkey anti-goat IgG (Jackson ImmunoResearch, West Grove, PA, USA), followed by a streptavidin-peroxidase conjugate (BD Pharmingen, San Diego, CA, USA). Peroxidase activity was demonstrated by incubation with 3-amino-9-ethylcarbazole (Sigma). Sections were counterstained with hematoxylin.

Following sacrifice, the tendons and synovium from the ankle joints of the hind limbs were dissected free from the surrounding tissue and washed in 200 μl DMEM, supplemented with HEPES (20 mM), L-glutamine (2 mM), and penicillin (50 U/ml)/streptomycin (50 μg/ml), and were incubated for 1 hour at room temperature to allow the elution of cytokines 14 . Supernatants were then removed and stored at -20°C until assayed.

TNFα and IL-1β levels were measured in ankle joint tissue washouts by ELISA (OptEIA ELISA kits; BD Pharmingen), as outlined previously 14 . TNFα and IL-1β ELISAs were sensitive down to 5 and 3 pg/ml, respectively. Keratinocyte-derived chemokine (KC) and macrophage inflammatory protein-2 (MIP-2) levels were measured in the peritoneal exudate fluid by ELISA (DuoSet; R&D Systems, Minneapolis, MN, USA), according to the manufacturer's instructions: KC and MIP-2 ELISAs were sensitive down to 2 pg/ml.

Zymography was used to assess protease expression in joint tissue washouts 23 . Briefly, joint washouts from mice with the same arthritic score were pooled and concentrated. SDS-PAGE mini-gels (10%) were prepared with the incorporation of gelatin (2 mg/ml; Labchem, Pittsburgh, PA, USA) before casting. The joint washouts (20 μl) were run into gels at a constant voltage of 200 V under nonreducing conditions. When the dye front reached the bottom, gels were removed and washed twice for 15 minutes in 2.5% Triton X-100 and incubated at 37°C overnight in zymography buffer (50 mM Tris-HCl (pH 7.5), 5 mM CaCl₂, 1 mM ZnCl₂ and 0.01% NaN₃). The gels were then stained for 45 minutes with Coomassie Brilliant Blue R-250 (Sigma) and extensively destained. Following destaining, zones of enzyme activity appeared clear against the Coomassie Blue background.

Quantitative PCR was performed as before 24 . Briefly, joints were crushed, RNA was extracted using the RNeasy Mini Kit (Qiagen, Valencia, CA, USA) and cDNA was prepared. Quantitative PCR was performed using Predeveloped TaqMan gene expression assays for TNFα, IL-1β, IL-6, MCP-1, t-PA, u-PA, u-PAR, MMP-3, MMP-9, MMP-13, ADAMTS-4 and ADAMTS-5 (Applied Biosystems, Foster City, CA, USA), and was read on an ABI Prism 7900H sequence detection system, followed by analysis using ABI Prism SDS 2.1 software. The TATA-binding protein (GeneWorks, Thebarton, SA, Australia) was used as the control gene. The comparative threshold method for relative quantification was used, and results are expressed as relative gene expression for each target gene.

For clinical and histologic scores and cytokine levels, the Mann-Whitney two-sample rank test was used to determine the level of significance between two experimental groups. For peritonitis and gene expression studies, an unpaired Student's t test was used; values are expressed as the mean ± standard error of the mean. P ≤ 0.05 was considered statistically significant.

Based on in vitro studies, several different cell types present in arthritic joints have been proposed to be potential sources of u-PA 10 . To determine the cellular source of u-PA important for the development of arthritis, bone marrow chimera experiments were performed followed by the induction of CIA.

We previously found 14 a reduced proliferative T-cell response to CII stimulation in vitro in CII-primed u-PA^-/- mice compared with CII-primed C57BL/6 mice even though the antibody response to CII was similar between the strains, suggesting normal immune complex formation in u-PA^-/- mice following CIA induction. In order to determine whether the reduced antigen-specific T-cell response could explain the mild CIA development in u-PA^-/- mice, the immune complex-mediated CAIA model 27 was initiated in u-PA^-/- and C57BL/6 mice. In contrast to CIA, this model bypasses the need for the induction of a T-cell response but requires LPS as a secondary stimulus to increase disease severity 16 . The TLR2 ligand, Pam-3-Cys, was used as a secondary stimulus, in place of LPS, as it gave more severe arthritis in C57BL/6 mice (average maximum clinical score 5. 3 ± 1.9 vs. 2.1 ± 1.3, Pam-3-Cys vs. LPS, P < 0.05).

u-PA^-/- mice were resistant to CAIA compared with C57BL/6 mice: 43% (three out of seven) of u-PA^-/- mice developed arthritis with an average maximum clinical score of 0.6 ± 0.2, compared with 80% (eight out of 10) of C57BL/6 mice with an average maximum clinical score of 5.3 ± 1.9 (P < 0.05). In fact, the three u-PA^-/- mice developed very mild arthritis, each with a maximum score of only 1.

Another immune complex-driven arthritis model is the K/BxN serum transfer model, in which serum from K/BxN mice, which develop spontaneous arthritis, is able to transfer the disease to naïve mice 19 . The arthritis that develops is more severe than the CAIA model and does not require an additional stimulus. C57BL/6 mice developed rapid and severe arthritis following serum transfer, whereas u-PA^-/- mice were essentially resistant (Figure 5b): 60% (three out of five) of u-PA^-/- mice developed arthritis with an average maximum clinical score of 1.2 ± 0.5, compared with 100% (seven out of seven) of C57BL/6 mice with an average maximum clinical score of 11.1 ± 1.3 (P < 0.005). Once again the three u-PA^-/- mice developed very mild arthritis, with a maximum score of only 2.

By histology, C57BL/6 mice show massive cellular infiltration, cartilage damage, proteoglycan loss, bone erosion and fibrin(ogen) staining following K/BxN serum transfer (Figure 5c, d). The joints from u-PA^-/- mice, on the other hand, look relatively normal, with minimal changes in these parameters (Figure 5c, d).

As already mentioned, the CIA model, the CAIA model and the K/BxN serum transfer arthritis model all involve immune complex formation and deposition in joints, and, as shown above, all depend on the presence of u-PA for disease progression. From the literature there is a report showing a requirement for u-PA in immune complex-driven lung inflammation 28 . In order to explore further the proposed involvement of u-PA in immune complex-dependent inflammation we required a model where we could assess u-PA dependence following direct application of an immune complex. As the immune-complex arthritis model is induced using an intra-articular injection 29 , we decided against this as such an injection leads to more severe arthritis in u-PA^-/- mice 12 13 , possibly due to the trauma involved. For this reason, and also because the peritoneal cavity is a convenient site for the isolation and quantification of both extravasated inflammatory cells and also of inflammatory mediators, we used the immune complex-mediated peritonitis model 20 .

We have previously shown that u-PA was not required for the development of murine peritonitis, as measured by neutrophil and macrophage infiltration, using either the nonspecific irritant, thioglycolate, as a stimulus, or an antigen (methylated BSA)-specific stimulus 30 . The lack of effect of u-PA deficiency with these stimuli therefore gave us the opportunity to once again see whether there is a particular association between immune complexes and u-PA.

In u-PA^-/- mice there were significantly fewer neutrophils present in their peritoneal cavity compared with C57BL/6 mice 4 hours post initiation of immune complex-mediated neutrophilia (P < 0.01) (Figure 6a). The levels of the chemokines, KC and MIP-2, shown previously to be produced following immune complex-induced neutrophilia in the peritoneal cavity 20 , were measured. KC (P < 0.01) and MIP-2 (P < 0.01) (Figure 6b) levels were lower in the peritoneal exudate fluid of u-PA^-/- mice compared with the levels in C57BL/6 mice. This suggests that u-PA is involved in the initiation of immune complex-mediated inflammatory response in the peritoneal cavity leading to the production of chemokines and recruitment of neutrophils.

There is evidence that immune complexes induce the bioactive complement component, C5a anaphylatoxin, which, in turn, interacts with the C5aR, thus reducing the threshold for Fcγ-receptor activation and leading to the recruitment of neutrophils into the peritoneal cavity 20 31 . Also, ablation of C5aR signaling abrogates neutrophil recruitment and production of KC and MIP-2 20 . Therefore, in order to determine whether u-PA might be required downstream of C5a signaling in acute peritonitis, C5a itself was given intraperitoneally and the inflammatory response was followed. A 2-hour time point was chosen for measurement of the neutrophilia based on published literature 32 . Following intraperitoneal injection of C5a there was no significant difference in the number of recoverable peritoneal exudate cells between u-PA^-/- and C57BL/6 mice (Figure 6c). As a control, once again, stimulation with the immune complex led to an increased number of recoverable peritoneal exudate cells from C57BL/6 mice, with a significantly reduced number of exudate cells recoverable from u-PA^-/- mice 2 hours post challenge (P < 0.001; Figure 6c). Injection of saline alone did not induce significant numbers of neutrophils at this time point (data not shown). Since, from the data above, C5a-triggered neutrophilia in the peritoneal cavity does not require u-PA, it suggests that the u-PA involvement in immune complex-mediated neutrophilia may not be downstream of C5a signaling.