Evaluation of protease-activated receptor 2 in murine models of arthritis


  • Nathalie Busso,

    1. Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland
    Search for more papers by this author
  • Matthias Frasnelli,

    1. Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland
    Search for more papers by this author
  • Roland Feifel,

    1. Novartis Institutes for BioMedical Research, Basel, Switzerland
    Search for more papers by this author
    • Drs. Feifel and Cenni own stock and/or hold stock options in Novartis Pharma. Dr. So has received consulting fees and/or honoraria (less than $10,000 each) from Abbott Laboratories and Pfizer.

  • Bruno Cenni,

    1. Novartis Institutes for BioMedical Research, Basel, Switzerland
    Search for more papers by this author
    • Drs. Feifel and Cenni own stock and/or hold stock options in Novartis Pharma. Dr. So has received consulting fees and/or honoraria (less than $10,000 each) from Abbott Laboratories and Pfizer.

  • Martin Steinhoff,

    1. University of Münster, Münster, Germany
    Search for more papers by this author
  • Justin Hamilton,

    1. University of California, San Francisco
    Current affiliation:
    1. Australian Centre for Blood Diseases, Monash University, Melbourne, Victoria, Australia
    Search for more papers by this author
  • Alexander So

    Corresponding author
    1. Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland
    • Centre Hospitalier Universitaire Vaudois, Service de Rhumatologie, Avenue Pierre Decker, 1011 Lausanne, Switzerland
    Search for more papers by this author
    • Drs. Feifel and Cenni own stock and/or hold stock options in Novartis Pharma. Dr. So has received consulting fees and/or honoraria (less than $10,000 each) from Abbott Laboratories and Pfizer.



Protease-activated receptor 2 (PAR-2) activation has been linked to pro- and antiinflammatory cellular responses. We undertook this study to explore the importance of PAR-2 activation in 4 murine models of arthritis and to analyze the expression of PAR-2 in human arthritic synovium.


Zymosan-induced arthritis (ZIA), K/BxN serum–induced arthritis, and Freund's complete adjuvant (CFA)–induced arthritis were generated in naive PAR-2−/− mice and PAR-2+/+ littermates. Antigen-induced arthritis (AIA) was generated in immunized mice using methylated bovine serum albumin (mBSA). The severity of arthritis was assessed by clinical scoring, technetium uptake measurement, and histologic analysis. Immune responses to mBSA were also evaluated from AIA. The expression of PAR-2 in synovial tissues from rheumatoid arthritis (RA) and osteoarthritis (OA) patients was compared.


In AIA, arthritis was significantly decreased in PAR-2–deficient mice and was associated with decreased levels of anti-mBSA IgG antibodies and lymph node cell proliferation. No difference in arthritis severity was seen in mice with ZIA, K/BxN serum–induced arthritis, and CFA-induced arthritis. Synovial biopsy specimens from RA patients demonstrated significantly increased expression of PAR-2 compared with those from OA patients.


PAR-2 deficiency was found to modulate articular inflammation in murine models of arthritis that require prior immunization and was associated with reduced levels of anti-mBSA IgG and lymph node cell proliferation in AIA. Expression of PAR-2 in RA synovium was significantly higher than that in OA synovium, and this suggests that PAR-2 is implicated in the pathogenesis of immune-mediated forms of arthritis.

The protease-activated receptors (PARs) are a family of transmembrane, G protein–coupled receptors that are activated by proteolytic cleavage of their extracellular N-terminus (1). Four PARs are currently known, 3 of which play important roles in the cross-talk between proteases of the coagulation cascade and cellular activation (2). PAR-2, unlike the other PARs, is activated by trypsin, mast cell tryptase, and leukocyte granzymes as well as by the nonthrombin coagulation proteases tissue factor–factor VIIa complex and factor Xa, and thus acts as a cellular sensor of inflammation and coagulation activation. Cleavage of the PAR-2 N-terminus unmasks a tethered ligand sequence (3) that activates the receptor and downstream signaling pathways, including inositol triphosphate production and Ca2+ mobilization.

Contrasting effects of PAR-2 in inflammation have been reported (2). PAR-2 may be proinflammatory, since PAR-2 activation increases vascular permeability, neutrophil infiltration (4), and proinflammatory cytokine secretion and stimulates the release of proinflammatory neurogenic peptides from neurons (5). In murine models of arthritis and multiple sclerosis, PAR-2–deficient mice showed a significant reduction of inflammation (6, 7). However, antiinflammatory effects of PAR-2 activation have also been reported in a murine model of mucosal inflammation (8).

We have previously reported high levels of thrombin–antithrombin complexes in rheumatoid arthritis (RA) synovial fluid and demonstrated that inhibition of factor VIIa and thrombin alleviates inflammation in animal models of arthritis (9, 10). We hypothesized that part of the proinflammatory effects of coagulation proteases in arthritis could be mediated via PAR-2 activation. In order to study the mechanisms and effects of PAR-2 signaling in arthritis, we analyzed 4 murine models of arthritis in PAR-2–deficient mice. Our results indicate that PAR-2 modulates immune-mediated joint inflammation in the antigen-induced arthritis (AIA) model, but not in models that did not require prior immunization. Biopsy specimens from patients showed that PAR-2 is prominently expressed in RA, but not in osteoarthritis (OA), synovial tissues, thus reinforcing the role of PAR-2 signaling in immune-mediated arthritis.


Animal studies.

AIA, zymosan-induced arthritis (ZIA), and Freund's complete adjuvant (CFA)–induced arthritis were generated in the same PAR-2–deficient mouse strain (11), whereas K/BxN serum–induced arthritis was generated in a different strain of PAR-2–deficient animals (12). All experiments used mice that were 8–12 weeks old at the start of the experiment. Age-matched PAR-2+/+ littermates were used as controls. Institutional approval was obtained for all in vivo experiments.

Models of experimental arthritis.

For AIA, mice were immunized as described elsewhere (13). Arthritis was induced on day 21 by intraarticular injection of 100 μg of methylated bovine serum albumin (mBSA) in 10 μl of sterile phosphate buffered saline (PBS) into the right knee; the left knee was injected with sterile PBS alone. Arthritis was assessed by measuring technetium uptake as described previously (13).

ZIA was induced by intraarticular injection of 180 μg (6 μl) of zymosan A (from Saccharomyces cerevisiae; Sigma, St. Louis, MO) through the suprapatellar ligament. The contralateral knee was injected with an equal amount of PBS as a control. Arthritis was assessed by measuring technetium uptake.

For K/BxN serum–induced arthritis, arthritogenic serum was obtained as described elsewhere (14). Arthritis was induced by intraperitoneal injection of 250 μl of K/BxN serum per recipient mouse on day 0. Arthritis was scored visually in each paw using a scale of 0 (no signs of inflammation) to 3 (maximal inflammation and swelling), with separate scores for the proximal and distal joints (maximum score 24). These assessments were performed by 2 trained individuals who were blinded to the study group.

CFA-induced arthritis was generated using the protocol described by Ferrell et al (6). Mice were injected in one knee with CFA (40 μl) and in the contralateral knee with PBS. Arthritis was assessed by measuring technetium uptake.

Histologic grading of arthritis.

Tissues and sections were prepared as described previously (13). Each section was graded independently by 2 observers who were unaware of the animal's genotype. For each histopathologic parameter, the mean ± SEM score of all slides was calculated.

Humoral and cellular immune response.

Measurements of serum levels of anti-mBSA antibodies and T cell proliferation assays were performed as described previously (15).

Sampling of human tissues.

Specimens of synovial tissue from 9 OA patients and 11 RA patients undergoing surgery of the knee or hip joint were obtained from the Department of Orthopedics, Centre Hospitalier Universitaire Vaudois. Tissues were cut into small pieces, immediately frozen in precooled hexane, and stored at −70°C until used. Analyses were performed on consecutive cryostat sections.

PAR-2 immunohistochemistry of human tissues.

Rabbit polyclonal anti–PAR-2 antibody (SC-5597; Santa Cruz Biotechnology, Santa Cruz, CA) at 10 μg/ml final concentration was applied overnight at 4°C on air-dried 5-μm– thick cryostat tissue sections. Sections had previously been fixed for 10 minutes in acetone at 4°C and then incubated for 30 minutes with 10% normal human serum, 10% normal goat serum, and 1% BSA. Bound primary antibodies were visualized with avidin–biotin–peroxidase complex (Vectastain Elite ABC kit; Vector, Burlingame, CA). The color was developed with 3,3′-diaminobenzidine (Sigma) containing 0.01% H2O2. As a control, nonimmune rabbit IgG was used. For double-staining of RA synovial tissues, mouse monoclonal antibodies against CD3, CD68, CD20, and vimentin (all from Sigma, Buchs, Switzerland) were detected by fluorescein isothiocyanate–labeled anti-mouse antibodies. Rabbit polyclonal anti-human PAR-2 antibodies were detected with rhodamine-labeled anti-rabbit antibodies.

Reverse transcriptase–polymerase chain reaction (RT-PCR) for human PAR-2.

RNA was extracted from cryostat tissue sections of OA and RA synovial membranes using TRIzol (Gibco, Basel, Switzerland). RT-PCR was performed using PAR-2 sense (5′-CGTCGGGGCTTCCAGGAG-3′) and antisense (5′-GACAGATGCAGAAAACTCATCC-3′) primers. As a reference control, GAPDH analysis by RT-PCR was performed in parallel.

Statistical analysis.

Data are reported as the mean ± SEM. The Wilcoxon rank sum test for unpaired variables was used to compare differences between groups with a non-Gaussian distribution. Student's unpaired t-test was used to compare groups with normally distributed values. The chi-square statistic was used to compare the frequencies. P values less than 0.05 were considered significant. All statistical calculations were performed using the JMP package (JMP version 4.02; SAS Institute, Cary, NC).


Effect of PAR-2 deficiency on experimental arthritis.

The role of PAR-2 was tested in 4 experimental models of arthritis, including AIA and 3 models not requiring prior immunization (ZIA, K/BxN serum–induced arthritis, and CFA-induced arthritis). For each model, PAR-2–deficient mice and their wild-type littermates (PAR-2−/− mice and PAR-2+/+ mice, respectively) were studied. The severity of arthritis was measured by technetium uptake at different time points up to day 7 after the onset of arthritis for AIA and ZIA and up to 29 days for CFA-induced arthritis. The severity of K/BxN serum–induced arthritis was assessed by clinical scoring. In AIA, technetium uptake (measured as the mean ± SEM ratio of the uptake in the arthritic knee joint to the uptake in the uninflamed knee joint) on days 1, 3, and 7 was lower in PAR-2−/− mice (n = 19) than in PAR-2+/+ mice (n = 22) (Figure 1A), with the results reaching statistical significance on day 3 (1.3 ± 0.05 versus 1.50 ± 0.06; P = 0.0031) and on day 7 (1.22 ± 0.05 versus 1.38 ± 0.05; P = 0.0032). PAR-2 deficiency did not attenuate arthritis in the ZIA or CFA-induced arthritis models (Figures 1B and D) and did not attenuate the clinical severity of arthritis in the K/BxN serum–induced arthritis model (Figure 1C).

Figure 1.

Effects of protease-activated receptor 2 (PAR-2) deficiency in experimental arthritis. Shown is the time course of knee joint inflammation in PAR-2–deficient mice with antigen-induced arthritis (AIA) (A), zymosan-induced arthritis (ZIA) (B), K/BxN serum–induced arthritis (KIA) (C), and Freund's complete adjuvant (CFA)–induced arthritis (CAA) (D). For AIA, ZIA, and CFA-induced arthritis, joint inflammation was measured by gamma counting of 99mTc uptake on different days after onset of arthritis. Results are expressed as the ratio of 99mTc uptake in the right (R) arthritic knee joint to that in the left (L) uninflamed knee joint. In A, B, and D, the mean and SEM R:L ratio is shown for each time point. In the AIA experiment, 22 PAR-2+/+ mice and 19 PAR-2−/− mice were analyzed, whereas 6 mice per genotype were used in the ZIA and CFA-induced arthritis models. For K/BxN serum–induced arthritis, inflammation of the distal and proximal joints of each paw was visually scored. In C, values are the mean ± SEM of 8 PAR-2+/+ mice and 7 PAR-2−/− mice. ∗ = P = 0.0031 versus PAR-2+/+ mice on day 3; ∗ = P = 0.0032 versus PAR-2+/+ mice on day 7.

The histologic features of AIA were examined on day 8 after arthritis onset. Two observers who were unaware of the animal's genotype independently graded synovial thickness and cartilage damage. The mean ± SEM synovial thickness score on day 8 of arthritis was significantly attenuated in PAR-2−/− mice (n = 20) compared with PAR-2+/+ mice (n = 22) (3.41 ± 0.3 versus 4.76 ± 0.24; P < 0.01) (results not shown). Mean ± SEM cartilage damage scores were slightly, but not significantly, decreased in PAR-2−/− mice (n = 20) compared with PAR-2+/+ mice (n = 22) (2.6 ± 0.3 versus 3.26 ± 0.19; P = 0.17) (results not shown). In ZIA and CFA-induced arthritis, there were no differences in histologic scoring, either for synovial thickness or cartilage damage, between wild-type and PAR-2–deficient mice (results not shown). Histologic analysis was not performed in mice with K/BxN serum–induced arthritis.

Effect of PAR-2 deficiency on immune responses in AIA.

We examined the immune responses to mBSA in AIA. Anti-mBSA antibodies were significantly lower in arthritic PAR-2−/− mice (n = 14) than in PAR-2+/+ mice (n = 22) (mean ± SEM 0.55 ± 0.09 optical density [OD] units versus 1.11 ± 0.15 OD units; P < 0.02) (Figure 2A). Total serum levels of IgG isotypes were identical in the 2 strains (Figure 2C), but levels of anti-mBSA IgG2b were significantly lower in PAR-2−/− mice (Figure 2B).

Figure 2.

PAR-2 effects on immune responses in AIA. Sera were collected on day 8 of AIA. A and B, Levels of anti–methylated bovine serum albumin (anti-mBSA) IgG antibodies (Ab) (A) and anti-mBSA–specific immunoglobulin isotypes IgG1, IgG2a, and IgG2b (B) were determined by enzyme-linked immunosorbent assay (ELISA). ∗ = P < 0.02 in A; ∗ = P = 0.036 in B. C, Total levels of immunoglobulin isotypes IgG1, IgG2a, and IgG2b in serum from nonarthritic, naive PAR-2+/+ and PAR-2−/− mice were determined by ELISA. At least 14 mice per group were used, and results are expressed as mean and SEM optical density (OD) units. D, Lymph node cell proliferation in response to mBSA (at 0, 1, 10, and 50 μg/ml) was tested in PAR-2+/+ and PAR-2−/− mice by 3H-thymidine incorporation 48 hours after addition of antigen. ∗∗ = P < 0.006 at 1 μg/ml mBSA; ∗ = P = 0.007 at 10 μg/ml mBSA. aBSA = anti-mBSA (see Figure 1 for other definitions).

The role of PAR-2 in T cell immune responses was also examined. Lymphocytes from draining lymph nodes were stimulated in vitro with mBSA. 3H-thymidine uptake was significantly decreased in cells isolated from PAR-2−/− mice compared with cells isolated from PAR-2+/+ mice (P < 0.006 and P = 0.007 at 1 and 10 μg/ml mBSA, respectively) (Figure 2D).

Expression of PAR-2 in RA synovial membranes.

PAR-2 expression was determined in synovial biopsy specimens from RA and OA patients was compared by RT-PCR (Figure 3A) and by immunohistology (Figure 3B). In RA synovial tissues, 5 of 11 specimens expressed PAR-2 by RT-PCR, whereas it was not detected in any of the 9 specimens from OA patients (χ2 = 7.33, P = 0.007). The differential expression of PAR-2 in RA tissues versus OA tissues was confirmed by immunohistochemistry. In RA tissues, lining and sublining synovial cells showed faint-to-moderate staining (Figure 3, parts d–f). Within the RA synovium, positive staining was also seen around some inflammatory aggregates. The vascular endothelium and/or smooth muscle was faintly positive in some vessels. In contrast, PAR-2 staining was absent in the 9 OA tissues examined (Figure 3B, parts a–c). Double-staining immunofluorescence with cell-specific markers (Figure 3C) demonstrated PAR-2 expression in a subset of vimentin-positive cells, CD68+ cells, and T and B cells.

Figure 3.

Protease-activated receptor 2 (PAR-2) expression in human synovial tissues. A, Synovial mRNA from osteoarthritis (OA) patients (lane 3) and rheumatoid arthritis (RA) patients (lane 4) was analyzed by reverse transcriptase–polymerase chain reaction (PCR) using PAR-2 primers. PCR with GAPDH primers was performed as a control for RNA quality. No RNA was added to the reaction in lane 1 (negative control). Gut mRNA was used as a positive control (lane 2). B, PAR-2 immunohistochemistry in OA and RA synovial membranes. In RA tissues, the lining (d) and sublining (e and f) layers showed faint-to-moderate staining. Within the inflammatory infiltrate, some cells were positive. In contrast, no PAR-2 staining was detected in OA tissues (ac). Negative controls on RA tissues, using nonimmune rabbit serum instead of primary antibody to rabbit PAR-2, showed no staining (g and h). C, Characterization of PAR-2–expressing cells within the RA synovial membrane. Staining for mouse anti-human CD3, CD68, vimentin, and CD20 was detected with fluorescein isothiocyanate–labeled anti-mouse antibodies (green fluorescence). Rabbit polyclonal anti-human PAR-2 antibodies were detected with anti-rabbit antibodies conjugated with rhodamine (red fluorescence). Double-staining immunofluorescence analysis demonstrated PAR-2 expression in some fibroblasts, a subset of macrophages, and T and B cells.


PAR-2 is capable of modulating inflammation by multiple mechanisms (2), but its role in arthritis has not been clearly established. Mast cell tryptase can activate PAR-2, and increased numbers of mast cells are present in synovial tissues from patients with RA. Although thrombin itself does not activate PAR-2, it is capable of amplifying the generation of upstream proteases (tissue factor–factor VIIa complex, factor Xa, and the cognate ternary complex) implicated in PAR-2 cleavage and could be one reason why the inhibitor hirudin is capable of reducing the severity of collagen-induced arthritis (10).

Based on the reported data, we expected that PAR-2 deficiency would abrogate joint inflammation. Unexpectedly, we found that PAR-2 deficiency attenuated joint inflammation only in the AIA model and not in the ZIA, CFA-induced arthritis, and K/BxN serum–induced arthritis models. The main difference is the requirement of prior immunization in AIA, which is not needed in the other 3 models. We therefore investigated immune responses to mBSA in the context of PAR-2 deficiency, and we observed a marked reduction in humoral and cellular responses to mBSA. These findings suggest that PAR-2 is capable of modulating antigen-specific immune responses but that it plays a negligible role in “innate immune” pathways of arthritis.

Our results differ from those obtained in a study of CFA-induced monarthritis, an unusual model of murine arthritis, which showed that PAR-2 deficiency virtually abolished inflammatory arthritis (6). We have tried to duplicate the reported results without success. In our experience, arthritis induced in this manner did not differ between PAR-2−/− and PAR-2+/+ animals, and histologic scores were equivalent (data not shown). Possible explanations for the discrepant results include differences in the genetic background of the deficient mice or in the gene construct of the PAR-2 deletion. Throughout our experiments, we used wild-type littermates as controls for the deficient animals in order to minimize the influence of background genetic and environmental factors.

Using immunohistochemistry, we showed that PAR-2 was expressed in RA synovial tissue, particularly in the lining layer and interstitial tissues and in a subset of fibroblasts and immune cells within the synovium. In contrast, no PAR-2 expression was seen in OA synovium. It appears that PAR-2 expression is not limited to a single cell type, as demonstrated by double-staining immunofluorescence. However, it is present on only a subset of macrophages and T and B cells. At present, we have no indication of what restricts expression to these subsets. Unfortunately, the available antibodies against murine PAR-2 were unsuitable for immunohistochemistry, so we are not able to provide details of its distribution in murine arthritis.

In conclusion, our data show that PAR-2 plays a role in immune-mediated joint inflammation but not in innate immune models of joint inflammation. The decreased severity of AIA in mice with PAR-2 deficiency and the prominent expression of PAR-2 in RA synovium suggest that PAR-2 activation is implicated in immune inflammatory diseases such as RA and may be an interesting target for future intervention studies.


Dr. So had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Drs. Busso, Cenni, Steinhoff, and So.

Acquisition of data. Drs. Busso, Frasnelli, Feifel, Cenni, and Hamilton.

Analysis and interpretation of data. Drs. Busso, Frasnelli, Feifel, Cenni, Steinhoff, and So.

Manuscript preparation. Drs. Busso, Cenni, Steinhoff, Hamilton, and So.

Statistical analysis. Drs. Busso and Feifel.

Provision of critical reagents. Dr. Hamilton.


We thank Carole Herkenne-Morard and Véronique Chobaz for excellent technical assistance and Eric Kolo for the confocal microscopy analysis.