Inflammatory bowel diseases (IBDs) such as Crohn’s disease and ulcerative colitis are chronic inflammatory conditions that affect mainly the intestine. The onset of IBD is unpredictable and patients can experience varying periods of active disease or flare-ups and remission. The aetiology of IBD is still not clear, but is likely to be mediated by genetic, environmental and immunological factors.1 Over the past few years, extensive evidence has been generated to indicate that certain proteases, such as thrombin, tryptase, trypsin and cathepsin G, are important mediators of the chronic inflammation observed in IBD.2^–7 These proteases can act as inflammatory mediators by cleaving and activating protease-activated receptors (PARs).7 8 PARs are novel members of the G-protein-coupled receptor family. The mechanism of PAR activation is unique, where proteolytic cleavage of the extracellular N-terminus generates a tethered ligand. PARs can be pharmacologically activated in the absence of proteolytic activity, by the use of the short synthetic peptides (PAR-activating peptide (PAR-AP)) corresponding to the sequence of tethered ligand domain. So far, four PARs have been identified, each with different N-terminal cleavage sites. Thrombin activates PAR₁, PAR₃ and PAR₄; human mast cell tryptase activates PAR₂; trypsin activates PAR₂ and PAR₄, and cathepsin G activates PAR₄.7 PARs can also be cleaved by various bacterial proteases.9 10

Of the four PARs, PAR₂ has been the most extensively studied in the context of inflammation. Interestingly, the levels of potential activators of PAR₂, such as trypsin and tryptase, are elevated in the colons of IBD patients.2 6 11 PAR₂ is expressed on many cell types of the gastrointestinal tract, including smooth muscle cells, endothelium, epithelium, enteric neurons and fibroblasts.7 The expression of PAR₂ by various cell types of the gut along with the elevated presence of PAR₂ activators during inflammatory pathological conditions, indicates that PAR₂ could be activated and play a role in the inflammatory processes of the gut. In previous studies, we demonstrated that local injection of PAR₂ agonists induced colonic inflammation characterised by tissue damage, granulocyte infiltration, increased intestinal permeability and T helper cell type 1 cytokine release.3 Further investigation indicated that PAR₂-induced colitis is mediated by a release of nitric oxide and neuropeptides neurokinin-1 and calcitonin-gene-related peptide.12 However, it has also been observed that systemic injection of a PAR₂ agonist during colitis induced by trinitrobenzene sulfonic acid (TNBS) can exert protective effects resulting in a significant reduction in mortality rate, tissue damage and the expression of inflammatory cytokines.5 The current data report on how PAR₂ agonists can modulate or induce colitis, but do not address what could be the role of PAR₂ in the course of inflammation. In order to define the role of PAR₂ in the initiation and progression of colitis, the effects of PAR₂ blockade on the development of inflammation in the gut still need to be investigated. To observe the development of colon inflammation in the absence of PAR₂ activity, we used PAR₂-deficient mice (PAR₂^−/−). We induced colitis by using three different murine models where inflammation was provoked by dextran sodium sulfate (DSS), TNBS and oxazolone in PAR₂^−/− and wild-type (PAR₂^+/+) mice. Further, we investigated the mechanism by which PAR₂ activation could influence the course of colitis, by studying the influence of PAR₂ inhibition on the expression of different inflammatory mediators, and the recruitment of inflammatory cells.

MATERIAL AND METHODS

Animals

C57BL6 mice (6–8 weeks old) were obtained from Charles River Laboratories (Montreal, Quebec, Canada). PAR₂^−/− and wild-type littermates (PAR₂^+/+) were originally provided by Johnson & Johnson Pharmaceutical Research Institute and bred at the University of Calgary animal care facility. Mice were kept at room temperature and had free access to food and water.

Intravital microscopy experiments

Mice were anaesthetised by intraperitoneal injection of a mixture of 10 mg/kg xylazine (MTC Pharmaceuticals, Cambridge, ON, Canada) and 200 mg/kg ketamine hydrochloride (Rogar/STB, London, ON, Canada). Intravital microscopy was performed on the large intestine (distal colon). Immediately after the anaesthesia, mice received an intravenous injection of leukocytes labelled with the fluorescent dye rhodamine 6G (0.3 mg/kg; Sigma, St. Louis, MO, USA). Then, a midline abdominal incision was performed, from the diaphragm, extending to the pelvic region. Segments from distal colon were carefully exteriorised and extended over a viewing pedestal.13 The exposed intestinal tissue was superfused with bicarbonate-buffered saline pH 7.4, to avoid tissue dehydration. The microcirculation was observed using an inverted microscope (Nikon, Mississauga, Canada) with a ×20 objective lens, and rhodamine 6G allowed visualisation and quantification of the number of rolling and adherent leukocytes, by epi-illumination at 510–560 nm, using a 590 nm emission filter. Single venules (20–40 μm in diameter) were selected for the study. After a 15 min equilibration period, the images of the selected venule were recorded for 5 min and the end of this 5 min interval was considered as time 0.13 14 Leukocyte adherence was determined upon video playback, on a vessel length of 100 μm. A leukocyte was considered adherent to the endothelium if it remained stationary for 30 s or more. Leukocyte flux was defined as the number of leukocytes per minute moving at a velocity less than that of erythrocytes, which passed a reference point in the venule. The change in leukocyte flux was measured as differences between the leukocyte flux at each interval and the basal flux of leukocytes. Additional images of the selected venule were then recorded for 5 min at times 15, 30, 45, 60 and 75 min. The parameters observed using the intravital study were plotted related to the time of the observation and the area under the curve was calculated.

Induction of colitis and study design

Colonic inflammation in mice was induced by TNBS, DSS or oxazolone. In detail, TNBS (1 or 2 mg per mouse dissolved in 40 or 50% ethanol) was injected intracolonically using a 1 ml syringe fitted with a catheter.5 All mice were fasted overnight before the intracolonic injection. Different concentrations of ethanol (40 and 50%) and different concentrations of TNBS (1 and 2 mg) were used in order to induce different degrees of inflammatory response. DSS was dissolved in drinking water (2.5% wt/vol) and the animals were free to drink the DSS solution for 7 days.5 Oxazolone-induced colitis was achieved by pre-sensitising mice with oxazolone dissolved in olive oil (3% wt/vol) applied to the skin of the abdomen. Mice were then challenged with the intracolonic injection of oxazolone (1% wt/vol, dissolved in 50% ethanol), 7 days after the sensitisation.4 The colitis induced by TNBS or DSS was given 7 days to develop, whereas for oxazolone treatment, animals were sacrificed 4 days after the induction of inflammation.

Body weight and survival rate were measured daily after the induction of colitis. On day 7 (for TNBS and DSS) or day 4 (for oxazolone), intravital microscopy was performed on the animals. Mice treated with lower dosage of TNBS (1 mg/mouse in 40% ethanol) were observed with intravital microscopy at day 7 before assessing the inflammatory parameters mentioned above. Administration of a lower dose of TNBS was preferred to prevent potential exacerbation of colitis and related lethality. Immediately after the intravital microscopy, various parameters of inflammation were assessed: bowel thickness measured using an electronic caliper (Mitutoyo, Mississauga, Canada, resolution 0.01 mm); macroscopic and microscopic damage scores; and myeloperoxidase (MPO) activity.

Assessment of macroscopic and microscopic damage

Macroscopic damage scores were evaluated as previously described.3 4 14 Briefly, when observed, the following parameters were given score of 1: haemorrhage, oedema, stricture, ulceration, faecal blood, mucus, and diarrhoea. Erythema was scored a maximum of 2 depending on the length of the area being affected (0, absent; 1, less than 1 cm; and 2, more than 1 cm). Adhesion was scored based on its severity (0, absent; 1, moderate; and 2, severe).

Microscopic damage scores were assessed on histological samples of colonic tissues.14 Cellular infiltration, submucosal oedema, damage/necrosis were given scores from 0 to 3: 0, absent; 1 mild; 2, moderate; and 3, severe. Vasculitis and perforation were scored from 0 to 1: 0, absent; and 1, present.

Measurements of myeloperoxidase activity

MPO activity was measured as an index of granulocyte infiltration as previously described.3 14 Briefly, tissue samples were homogenised in a solution of 0.5% hexadecyltrimethylammonium bromide dissolved in phosphate buffer solution (pH 6.0) for 1 min. The homogenised tissues were centrifuged at 13 000 g for 2 min. Supernatants were added to a buffer supplemented with 1% hydrogen peroxide, and O-dianisidine dihydrocholoride solution. Optical density readings were taken for 1 min at 30 s intervals at 450 nm.

Histology

Intestinal tissue samples were fixed in 10% neutral buffered formalin and embedded in paraffin. Sections of 6 μm were cut and mounted on colourfrost microslides (VWR Scientific, Edmonton, Alberta, Canada) were stained with haematoxylin & eosin (H&E).

Western blotting

Western blotting was performed on colon samples collected from mice treated with TNBS (2 mg in 40% ethanol). Proteins were extracted by homogenising the tissues in 1% Triton X-100, 500 mmol/l Tris-HCl, 200 mmol/l NaCl and separated on SDS 10% PAGE gel then transferred onto nitrocellulose membrane (Bio-RAD, Hercules, California, USA). Overnight at 4°C, membranes were blocked with 5% skim milk in Tris-buffered saline Tween (TBST) solution with primary antibody. For anti-COX-1 (Cedarlane, Burlington, Canada), a 1:10 000 dilution was used, and for all other primary antibodies (anti-COX-2, anti-ICAM-1, anti-VCAM-1, anti-alpha-4 and anti-alpha-M; all from Santa Cruz Biotechnology, Santa Cruz, California, USA) a 1:100 dilution was used. After washes in TBST, membranes were treated with the secondary antibodies in 5% skim milk TBST solution (at dilution of 1:10 000) for either anti-goat or anti-rabbit IgG antibody conjugated to horseradish peroxidase. After washes, membranes were treated with enhanced chemiluminescence (ECL) reagents (Amersham Biosciences, Piscataway, USA) and exposed onto film.

Statistical analysis

Statistical comparisons among groups were performed using a one-way analysis of variance followed by the Student–Newman–Keuls test. The results are expressed as mean with the standard error of the mean where a probability (p value) of less than 0.05 was considered to be significant.

RESULTS

PAR₂ deficiency protects against the development of colitis

Colitis induced by DSS, TNBS or oxazolone treatments provoked several signs of inflammation and disease activity, including weight loss, lethality, increased bowel thickness, granulocyte infiltration and tissue damage (figs 1, 2 and 3). After the induction of colitis by DSS, the weight loss observed in PAR₂^−/− mice was not different from that in PAR₂ ^+/+ mice (fig 1A). However, the macroscopic damage score was significantly reduced in PAR₂^−/− mice compared to wild-types (fig 1B). Differences for macroscopic damage scores were observed mainly for the parameters of oedema, faecal blood, and diarrhoea. Increased bowel thickness and MPO activity observed after DSS treatment were also significantly reduced in PAR₂^−/− mice compared to PAR₂^+/+ (fig 1C,D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 The effects of dextran sodium sulfate (DSS)-induced colitis on weight loss (A), macroscopic damage score (B), bowel thickness (C) and myeloperoxidase (MPO) activity in wild-type mice (PAR₂^+/+) and PAR₂-deficient mice (PAR₂^−/−). Colitis was induced by adding 2.5% DSS solution to the drinking water of mice for 7 days. Control mice (non-inflamed) received normal water. Inflammatory parameters were assessed 7 days after the induction of colitis. *Significantly different from wild-type mice counterpart, p<0.05. †Significantly different from the control (non-inflamed) group. Values are shown as mean with the standard error of the mean. PAR, protease-activated receptor.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 The effects of TNBS-induced colitis (2 mg in 40% ethanol) on changes in weight (A), bowel thickness (B), myeloperoxidase (MPO) activity (C), macroscopic damage score (D) and histological architecture (E, F) in wild-type mice (PAR₂^+/+) and PAR₂-deficient mice (PAR₂^−/−). Histological samples of mouse colon 7 days after colitis were collected in PAR₂^+/+ (E) and PAR₂^−/− (F), and microscopic damage scores were attributed on haematoxylin & eosin stained slides, scale bar is 100 μm. *Significantly different from wild-type mice counterpart, p<0.05. Values are shown as means with the standard error of the mean. PAR, protease-activated receptor. TNBS, trinitrobenzene sulfonic acid.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 The effects of oxazolone-induced colitis on weight loss (A), bowel thickness (B), myeloperoxidase (MPO) activity (C), macroscopic damage score, (D), and histological architecture (D–F) in wild-type (PAR₂^+/+) and PAR₂-deficient mice (PAR₂^−/−). Histological samples were obtained from PAR₂^+/+, PAR₂^−/− or both mice, 4 days after oxazolone-induced colitis. Microscopic damage scores were attributed on haematoxylin & eosin stained slides, scale bar is 100 μm. *Significantly different from wild-type mice counterpart, p<0.05. †Significantly different from the saline group. Values are shown as mean with the standard error of the mean. PAR, protease-activated receptor.

TNBS-induced colitis provoked mortality in C57Bl6 mice for different doses of ethanol and/or TNBS used. PAR₂^−/− mice injected with 2 mg of TNBS in 40 or 50% ethanol displayed a significantly lower percentage of mortality (8 and 21%, respectively, to 40 and 50% ethanol dosage) as compared to PAR₂^+/+ mice (64 and 71%, respectively, to ethanol dosages). The weight loss observed after TNBS treatment was also reduced in PAR₂^−/− mice compared to PAR₂^+/+ for several days after the induction of colitis (fig 2A). Increased bowel thickness and MPO activity after TNBS treatment were significantly lower in PAR₂^−/− mice compared to PAR₂^+/+ (fig 2B,C). Extensive erythema, oedema and adhesion were observed in mice 7 days after TNBS treatment, leading to a high macroscopic damage score (fig 2D). The extent of the increase in macroscopic damage was significantly lower in PAR₂^−/− compared to PAR₂^+/+ (fig 2D). Histology showed that TNBS-induced colitis affected all layers of the colon (transmural inflammation) with submucosal oedema, muscle thickening and strong granulocyte infiltration (fig 2E,F). The microscopic damage induced by TNBS was largely reduced in PAR₂^−/− mice especially with regards to damage/necrosis (fig 2D,E,F).

Oxazolone treatment induced significant weight loss in wild-type mice throughout the development of colitis (days 1–4, fig 3A), which was significantly reduced in PAR₂^−/−. Bowel thickness, MPO activity and macroscopic damage scores which were significantly increased by oxazolone challenge were significantly reduced in PAR₂^−/− mice compared to PAR₂^+/+ mice (fig 3B,C,D). At a microscopic level, oxazolone challenge caused extensive cellular infiltration (arrows, fig 3E), muscle thickening (double arrows, fig 3E), and damage/necrosis of the epithelium (arrow heads, fig 3E). All parameters of microscopic damage scores were significantly reduced in PAR₂^−/− mice compared to PAR₂^+/+ mice (fig 3E,F).

PAR₂ deficiency inhibits leukocyte recruitment to the colon during colitis

Changes in the flux of rolling leukocytes as well as the number of leukocytes adhering to the intestinal venules of the mice were assessed by intravital microscopy technique. All three models of colitis evoked a significant increase in the area under the curve for the flux of rolling leukocytes in wild-type mice (fig 4A,C,E). Interestingly, changes in leukocyte fluxes were not modified in PAR₂^−/− and PAR₂^+/+ mice after the induction of DSS colitis, but were reduced in PAR₂^−/− mice, for the other two models of colitis: TNBS and oxazolone (fig 4A,C,E). Leukocyte adherence was increased in wild-type mice after the induction of colitis (by either TNBS, DSS or oxazolone), compared to control non-inflamed mice, but was significantly reduced in PAR₂^−/− compared to PAR₂^+/+ mice (fig 4B,D,F). Overall, these data show that PAR₂ deficiency inhibits the steps necessary for the recruitment of leukocytes to the gut wall.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 Changes in flux of rolling leukocytes (A,C,E) and adherent leukocytes (B,D,F) in wild-type mice (PAR₂^+/+) and PAR₂-deficient mice (PAR₂^−/−). Mice were challenged or not (control non-inflamed), with three different models of colitis including DSS (A,B), TNBS (C,D), or oxazolone (E,F). Colitis was induced by using 2.5% DSS, 1 mg of TNBS in 40% ethanol or 1%/3% oxazolone (pre-sensitisation/challenge dose, respectively). The changes in leukocyte rolling/adherence were assessed by intravital microscopy 7 days after DSS/TNBS administration or 4 days after oxazolone challenge. *Significantly different from PAR₂^+/+ mice counterpart, p<0.05. †Significantly different from the control (non-inflamed) group. Values are shown as mean with the standard error of the mean. AUC, area under a curve; DSS, dextran sodium sulfate; PAR, protease-activated receptor; TNBS, trinitrobenzene sulfonic acid.

Differential expression of cyclooxygenases and adhesion molecules in the absence of PAR₂

The protein levels of inflammatory mediators such as cyclooxygenases (COXs) and adhesion molecules were measured using western blotting techniques in PAR₂^−/− and PAR₂^+/+ naive or after TNBS colitis (figs 5 and 6). In non-inflamed mouse tissues, the protein levels were significantly lower for COX-1 (fig 5A) and significantly higher for COX-2 (fig 5B) in PAR₂^−/− as compared to PAR₂^+/+ mice. Treatment with TNBS significantly increased the level of COX-1 expression in PAR₂^−/− mice but not in wild-type mice (fig 5A). For COX-2, the protein level was significantly elevated to a similar extent in both PAR₂^−/− mice and wild-type mice after TNBS treatment (fig 5B). Administration of TNBS also modulated the expression of various adhesion molecules (fig 6A–D). In non-inflamed tissues, the expression of adhesion molecules was not different between PAR₂^+/+ and PAR₂^−/− mice (fig 6A,B,D), with the exception of alpha-M where the expression was significantly lower in PAR₂^−/− compared with PAR₂^+/+ mice (fig 6C). TNBS administration significantly increased the expression levels of ICAM-1, alpha-M and alpha-4 in wild-type mice, whereas only alpha-4 was elevated in PAR₂^−/− mice. TNBS colitis failed to induce a significant increase in ICAM-1, VCAM-1 and alpha-M expression in PAR₂^−/− tissues (fig 6A,B,C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5 Differential expression of (A) cyclooxygenase-1 (COX-1) and (B) cyclooxygenase-2 (COX-2) measured by western blot in colonic tissues obtained from wild-type mice (PAR₂^+/+) and PAR₂-deficient mice (PAR₂^−/−) after TNBS-induced colitis or in naive (control non-inflamed) mice. The densitometry data of bands obtained by western blot for each protein is shown as a bar graph and is representative of a minimum of six mice. All densitometry readings are standardised to actin. *Significantly different from wild-type mice counterparts, p<0.05. †Significantly different from control (non-inflamed) group. Values are shown as mean with the standard error of the mean. PAR, protease-activated receptor; TNBS, trinitrobenzene sulfonic acid.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6 Protein expression of adhesion molecules ICAM (A), VCAM (B), alpha-M (C) and alpha-4 (D) measured by western blot analysis in colonic tissues obtained from wild-type mice (PAR₂^+/+) and PAR₂ deficient mice (PAR₂^−/−) after TNBS-induced colitis or in naive mice (control non-inflamed). The densitometry data of bands obtained by western blot for each protein is shown as a bar graph and is representative of a minimum of six samples. All densitometry readings are standardised to actin. Mice were injected with either vehicle (40% ethanol) or TNBS (2 mg in 40% ethanol) intracolonically and the level of protein expression was assessed 7 days later. *Significantly different from wild-type mice counterparts, p<0.05. †Significantly different from the control (non-inflamed) group. Values are shown as mean with the standard error of the mean. PAR, protease-activated receptor; TNBS, trinitrobenzene sulfonic acid.

DISCUSSION

The exact role of PAR₂ during intestinal inflammation is unclear due to the fact that previous studies have described both pro- and anti-inflammatory responses to PAR₂ agonists.3 5 Studies on the role of endogenous PAR₂ activity during inflammation have been difficult because of the lack of an effective and selective PAR₂ antagonist. In the present study, we demonstrate that PAR₂ deficiency gives protection from the development of inflammation induced by three different murine models of colitis (DSS, TNBS and oxazolone). The attenuated inflammatory responses in PAR₂^−/− mice were also associated with a reduction in leukocyte trafficking, suggesting that the mechanism by which PAR₂ deficiency is protective involves leukocyte recruitment interactions. This was confirmed by differential expression of adhesion molecules in the absence of PAR₂.

Since no single animal model can reproduce all the features of human inflammatory bowel disease, the use of several different mouse models is generally necessary to investigate the importance of one mediator in intestinal inflammatory pathologies. TNBS-induced colitis is a chronic inflammatory model that provokes transmural inflammation with a large predominance of T helper 1 (Th1) cytokines (interleukin 2 (IL2) and interferon γ (INFγ)), very similar to that which is observed in Crohn’s disease.15 The DSS-induced colitis presents multiple mucosal erosions associated with the release of both Th1 and Th2 cytokines (IL2, IL4), although the duration of this study (7 days) has been shown by other studies to produce mainly the Th1 cytokine profile.15 Finally, the oxazolone model shows mucosal erosions that are often limited to the mucosal surface, severe bowel wall oedema, and luminal exudates, which are very similar to lesions observed in ulcerative colitis. In addition, the oxazolone model presents a clear Th2 cytokine profile (upregulation of IL4 and IL10).15 The fact that inflammatory parameters were reduced in PAR₂^−/− mice in all three models of colitis not only indicates the endogenous pro-inflammatory role of PAR₂ during intestinal inflammation, but also that the effects of PAR₂ deficiency are independent of the Th cytokine profile. These results contrast with previous reports on another member of the PAR family, the thrombin receptor PAR₁. Activation of PAR₁ was pro-inflammatory in Th1- but not in Th2-induced colitis.4 7 It is important to point out that our study does not necessarily imply that PAR₂ activation, T cells and Th cytokines are unrelated. Indeed, PAR₂ is expressed on T cells and may play a significant role in priming T helper cell response by inducing dendritic cell maturation, as well as regulating lymphocyte adhesion and generating reactive oxygen species.16 17 A number of pieces of in vitro and in vivo evidence also indicate that PAR₂ agonists can induce the release of pro-inflammatory cytokines from various cell types.3 7 The protective effect of PAR₂-activating peptide during TNBS-induced colitis was also associated with changes in Th1 cytokine expression, although inhibition rather than induction of Th1 cytokines mRNA levels was observed after PAR₂ agonist treatment.5 Although PAR₂ activation may affect the expression profile of inflammatory cytokines, our study shows that PAR₂ deficiency protects against the development of inflammation independently of the Th cytokine profile expression.

PAR₂ is highly expressed in leukocytes and administration of PAR₂ agonists has been shown to induce migration of neutrophils and release of inflammatory mediators from eosinophils and lymphocytes.7 18^–20 Previous studies have indicated that PAR₂ is also highly expressed on the endothelium where the activators of PAR₂ can stimulate the synthesis of endothelin and exocytosis of adhesion molecules.21 22 Previous studies have shown that PAR₂ agonists can induce leukocyte rolling/adherence along the mesenteric venules, whereas leukocyte rolling was reduced in the cremaster muscles of PAR₂-deficient mice submitted to surgical trauma.13 23 Thus, PAR₂ activation may play an important role in one of the earliest event of the inflammatory reaction: the recruitment of inflammatory cells. Indeed, the results of this study indicate that endogenous PAR₂ is a key mediator in leukocyte trafficking during the development of colitis (fig 4). The different steps of leukocyte recruitment involve a series of complex interactions between leukocytes and endothelium, which are mediated by various adhesion molecules. It has been established that leukocyte rolling is mediated mainly through the selectins (E-, L- and P-selectins), whereas, leukocyte adherence is mediated mainly through the integrins (complex of alpha-L-beta-2) expressed on leukocytes interacting with intracellular adhesion molecule (ICAM) and vascular cell adhesion molecule (VCAM) expressed on the endothelium.24 In this study, we assessed the levels of ICAM-1, VCAM-1, alpha-M (which interacts with beta-2 integrin to form MAC-1) and alpha-4 (which interacts with beta integrin to form VLA-4), because these adhesion molecules have been shown to be upregulated upon administration of PAR₂-activating peptide in various in vitro experiments.25^–27 The expression of ICAM-1, VCAM-1 and alpha-M after TNBS treatment was significantly lower in PAR₂^−/− mice compared to wild-type mice, indicating that endogenous PAR₂ regulates the expression of adhesion molecules during colitis (fig 6). In the case of alpha-M, the reduced expression in PAR₂^−/− tissues could simply be due to the reduction in the number of infiltrating leukocytes expressing alpha-M. Interestingly, the induction of alpha-M but not alpha-4 was inhibited in PAR₂-deficient mice after TNBS challenge (fig 6). Alpha-M is readily expressed in leukocytes including neutrophils, monocytes, natural killer cells and T lymphocytes. Alpha-4, on the other hand, is readily expressed on most of leukocytes with the exception of neutrophils.28 29 Thus, reduction in neutrophil recruitment, as indicated by reduction in MPO activity in this study, may be responsible for the reduction in alpha-M expression in PAR₂^−/− mice. However, in non-inflamed tissues, all of the adhesion molecules being assessed were present at low levels and their expression was not different in PAR₂^−/− mice compared to PAR₂^+/+ mice, with the exception of alpha-M. The fact that expression of alpha-M in PAR₂^−/− mice was attenuated even in non-inflamed tissues in the absence of leukocyte recruitment may indicate that endogenous PAR₂ can directly regulate the expression of adhesion molecules on leukocytes. New agents for treating IBD try to target adhesion molecules (natalizumab, MLN-02 and ISIS-2302) as well as leukocytapherasis. Those treatments have produced significant therapeutic benefits.28 29 Our results suggest that PAR₂ expression and, potentially, activation is an important step upstream from adhesion molecule expression. PAR₂ could thus be considered as an additional molecular target to reduce inflammatory cell recruitment.

In contrast to the study by Fiorucci et al,5 our work shows that PAR₂ activation is pro-inflammatory rather anti-inflammatory. In that study, the authors showed that systemic and chronic (daily) treatments with PAR₂ agonists were protective against the development of colitis.5 A possible explanation for these discrepancies could be that local (colonic) activation of PAR₂ might be protective, while systemic activation could exert anti-inflammatory effects by causing hypotension and vasodilation. Another possible explanation is that chronic treatment with PAR₂ agonists might desensitise the normally occurring PAR₂-activation response, thereby limiting the pro-inflammatory aspects of PAR₂ activation. Finally, the pro- or anti-inflammatory effects of PAR₂ activation might depend on the cellular targets of PAR₂ agonists. For instance, PAR₂ activation on enteric nerves could be protective through the release of neuropeptides such as the calcitonin gene-related peptide,30 while PAR₂ activation on leukocytes or epithelial cells might send pro-inflammatory signals.13 31 In terms of drug development, targeting PAR₂ might thus have beneficial effects from the point of view of inflammatory cell recruitment/epithelial cell functions, but could also have adverse effects on activation of the enteric nervous system. Only the development of selective and specific PAR₂ antagonists or blocking PAR₂ antibodies and their use in in vivo models could answer those questions. However, it is interesting to note that PAR₂-deficient mice did not show specific phenotypes in non-pathological conditions. So far, no human data are available on the effects of PAR₂ activation in IBD; however, we have reported that increased proteolytic activity specific for an arginine site, similar to the PAR₂ cleavage site, is released from biopsies of patients with IBD.11 In addition, tissues from such patients over-express PAR₂.32 Taken together, these studies give weight to the hypothesis that PAR₂ could play a role in human disease as it does in mouse models of colitis.

As shown in previous studies,33 we report that the level of COX-2 but not COX-1 expression was upregulated after TNBS-induced colitis in wild-type mice. In contrast, in PAR₂^−/− mice, both COX-1 and COX-2 expression were significantly elevated after TNBS colitis and this was associated with a reduced inflammatory response (fig 5). It could be speculated that the two forms of COX, together, may exert anti-inflammatory effects during the development of colitis. Indeed, previous studies have indicated an active role for both COX-1 and COX-2 in the modulation of intestinal inflammation in many animal models of colitis.34 For example, administration of selective COX-1 or COX-2 inhibitors in TNBS- or DSS-induced colitis can exacerbate the severity of colitis and impair healing.34 Studies using mice deficient for either COX-1 or COX-2 also reported greater inflammatory reaction in these mice, indicating a possible protective effect exerted by COX-1 and COX-2 during intestinal inflammation.35 It is of particular note that, in non-inflamed tissues, the expression of COX-1 was significantly lower and COX-2 significantly higher in PAR₂^−/− mice. Hence, although PAR₂ is an important regulator of COX-1 during inflammation it is also involved in regulating expression of both COX-1 and COX-2 under naive conditions.

A role for PAR₁ in IBD has also been suggested by other studies,4 7 which raises the issue of potential cross-talk and redundancy between different PARs. However, the role of PAR₁ appeared not to be the same in colitis as the role we show here for PAR₂. Depending on the cytokine profile associated with colitis, PAR₁ appeared to be protective in Th2-type colitis, while it appeared to be pro-inflammatory in Th1-type colitis.4 7 In contrast, we show here that PAR₂ activation is involved in inflammatory cell recruitment independently of the cytokine profile. In addition, it is important to consider that PAR₁ and PAR₂ are not necessarily activated by the same enzymes. For instance, PAR₁ but not PAR₂ is activated by thrombin, while PAR₂, but not PAR₁ is activated by tryptase. Depending on the nature and, potentially, the quantities of the proteases released in IBD tissues, differential PARs might be activated and could exert different roles in the maintenance of chronic inflammation. Whether direct cross-talk between PAR₁ and PAR₂ in the same cell type of inflammatory nature is occurring or not has not yet been reported, but such an aspect should potentially be considered.

Overall, the results of our study are consistent with other studies where PAR₂^−/− mice have attenuated inflammatory responses in organs other than the gut: in the lungs36 or in joints.37 Nevertheless, the specific mechanisms by which PAR₂ deficiency leads to the attenuation in the inflammatory response were still unclear. In the present study, we have shown that PAR₂ deficiency was associated with a reduction in leukocyte trafficking, adhesion and protein levels of adhesion molecules (ICAM, VCAM, alpha-M). These results are in favour of a general pro-inflammatory role for endogenous PAR₂ in all types of chronic tissue inflammation, where PAR₂ activation would be essential to the recruitment of inflammatory cells. The present study further highlights PAR₂ as a potential target for the treatment of chronic inflammatory states, and for inflammatory bowel diseases in particular.