Carrageenan-induced acute inflammation in the mouse air pouch synovial model. Role of tumour necrosis factor

We used the mouse air pouch model of inflammation to study the interaction between cytokines, prostaglandin E2 (PGE2) and cell migration during the various phases of acute local inflammation induced by carrageenan. In serum, the levels of interleukin 1 (IL-1), interleukin 6 (IL-6), tumour necrosis factor (TNF), serum amiloid-A (SAA) and Fe++ were never different from controls, indicating that no systemic inflammatory changes were induced. Locally the exudate volume and the number of leukocytes recruited into the pouch increased progressively until 7 days after carrageenan. The same was true for PGE2 production. We could not measure IL-1 but the production of IL-6 and TNF reached a maximum after 5-24 h then quickly decreased. Anti-TNF antibodies inhibited cell migration by 50% 24 h after treatment. Pretreatment with interleukin 10 (IL-10) inhibited TNF production almost completely and cell migration by 60%. Carrageenan-induced inflammation was modulated by anti-inflammatory drugs. Pretreatment with dexamethasone (DEX) or indomethacin (INDO) inhibited cell migration and reduced the concentration of TNF in the exudate. Production of PGE2 or vascular permeability did not correlate with the number of cells in the pouch. Local TNF seems to play an important role in this model, particularly for leukocyte migration in the first phase of the inflammatory process. In conclusion, the air pouch seems to be a good model for studying the regulation of the early events of local inflammation, particularly the role of cytokines and cell migration.


Introduction
Many attempts have been made to correlate serum tumour necrosis factor (TNF) levels with the severity of different in ammatory diseases 1 but clinical studies 2 and experimental work 3 in various models of septic shock suggest that systemic levels of TNF are not a good predictor of the potential outcome of in ammation. However, in experiments of paw oedema induced by carrageenan injection, the local amount of TNF correlated with the swelling of the paw and it was therefore considered a good indication of the in ammatory status. 3 On the other hand, TNF seems to be an important mediator in the pathogenesis of rheumatoid arthritis. 4,5 Joint extracts from arthritic rats had signi cantly higher TNF levels (1054 147 pg g of tissue) than joint extracts of normal rats (110 42 pg g of tissue). 6 Anti-TNF antibodies are protective in animal models of arthritis 7 and clinical studies have shown an effect of anti-TNF antibodies in patients with rheumatoid arthritis. 8 We established an in vivo model of acute local in ammation in the mouse. This air pouch model of synovial in ammation, described by Sedgwick and coworkers, 9 consists of a subcutaneous injection of air on the back of rodents. This procedure induces the proliferation of cells that stratify on the surface of the cavity to form a structure similar to the synovia after 6 days. 10 Injection of carrageenan induces in ammation and the pouch serves as a reservoir of cells and mediators that can be easily measured in the uid that accumulates locally.
We investigated the production of TNF in relation to interleukin (IL)-1 and IL-6, which are reported to be regulated by TNF in synovial uid and in the serum of patients with rheumatoid arthritis, 8 and in relation to leukocyte migration. Because prostaglandin E 2 (PGE 2 ) suppresses various leukocyte functions, 11,12 dose-dependently regulating TNF production by macrophages, 13 we also studied PGE 2 production in the air pouch.
We characterized the experimental model in terms of cytokine production as well as PGE 2 production, cell migration and vascular permeability (measured as extravasation of uorescently labelled bovine serum albumin). The kinetics of all these parameters were investigated during the in ammatory response.
We also investigated whether a systemic acute phase response was induced. We measured serum TNF, IL-6 and IL-1 as well as serum amyloid A (SAA) and serum Fe as typical markers of the acute phase response. 14 We used the cyclooxygenase inhibitor indomethacin (INDO) to investigate the role of PGE 2 . We also used dexamethasone (DEX) and IL-10 potent inhibitors of the production of in ammatory cytokines that are protective in animal models of TNF-mediated endotoxic shock 15,16 and an anti-TNF antibody that neutralizes TNF in vivo and protects against endotoxic shock. 17 To evaluate the role of the polymorphonuclear leukocytes' (PMN) recruitment in the production of in ammatory mediators, we induced neutropenia by pretreating mice with 5-uorouracil (5-FU).

Animals
Male CD1 mice, 22± 25 g body weight (Charles River Italy, Calco, CO) were used for all experiments if not otherwise stated. The animals were housed at constant temperature (20 18 C) and relative humidity (60 10%) and provided with food and water ad libitum. All the procedures involving animals and their care were conducted in conformity with national and international laws and policies (EEC Council Directive 86609, OJ L358, 1, 12 December 1987; Italian Legislative Decree 116 92, Gazzetta Uf ciale della Repubblica Italiana n. 40, 18 February 1992; NIH Guide for the Care and the Use of Laboratory Animals, NIH publication no. 85± 23, 1985).

Air pouch model
After a period of adaptation the animals were anaesthetized with ether and 5 ml of air were injected under the skin on their back. After 3 days the pouches were reinjected with 3 ml of air. On day 6, 1 ml of 1%carrageenan (Sigma) in saline was injected into the pouch. The controls received 1 ml of saline. At different times after carrageenan the animals were anaesthetized and the pouches were washed with 1 ml of saline. The lavage uid was immediately cooled on ice and the volume was recorded, then 50 ml were used for cell count after staining with erythrosin. The remaining uid was centrifuged at 5000 rpm for 10 min at 48 C and supernatants were stored at 208 C until used for the measure of PGE 2 , TNF or IL-6.

Materials
IL-10 was a kind gift from Schering Plough, Italy. Dexamethasone phosphate was a kind gift from Laboratorio Farmacologico Milanese S.R.L. (Caronno P. Varese, Italy); indomethacin, watersoluble salt, was a commercial preparation from Chiesi Farmaceeutici (Parma, Italy). All drugs were given in saline. A rat anti-mouse TNF monoclonal antibody (V1q) was prepared and administered to mice as previously described. 18 The speci city and activity of this antibody has been described 19 and it protects mice against the lethality of LPS. 18,19 Mediator assays TNF was measured by a standard cytotoxicity assay using L929 cells. IL-1 was measured using a mouse ELISA kit purchased from Genzyme (Cambridge, MA). IL-6 was measured as hybridoma growth factor on 7TD1 cell line. PGE 2 was measured with a commercial ELISA kit (Amersham, Little Chalfont, UK). Serum amyloid-A was determined using an ELISA as previously described. 20

Measurement of plasma exudation
Plasma exudation in the pouch was measured using uorescein isothiocyanate-conjugated bovine serum albumin (FITC-BSA) (Sigma Chemical Co., St Louis, MO). At different times after carrageenan treatment, 0.6 mg mouse of uorescein isothiocyanate-conjugated bovine serum albumin dissolved in 0.2 ml of saline were injected in the tail vein of the animals; 30 min later the animals were anaesthetized with ether and the blood was collected. Fluorescence of the undiluted pouch uid and of serum, diluted 1:10 in saline, were measured with the excitation wavelength set at 490 nm and emission at 521 nm. Plasma exudation was calculated as the uorescence in the pouch as a percentage of that in serum.

Statistical analysis
Data were analysed with one-way analysis of variance coupled with Duncan's test for multiple comparisons or with Kruskal± Wallis nonparametric test coupled with Sach's test for multiple comparisons.

Results
Characterization of the air pouch carrageenan in¯ammation model Fig. 1 shows the kinetics of the volume of exudate and leukocyte recruitment during carrageenan-induced in ammation in the air pouch. The total number of leukocytes ( Fig. 1A) was very similar in control and treated groups until 5 h after treatment. From 24 h to 7 days there was massive recruitment of cells into the pouches of animals treated with carrageenan. The cells were mostly PMN until 24 h after treatment and changed to a prevalence of macrophages from 48 hours on (data not shown). After 24 h PMN amounted to 70± 80% of the total. Fig. 1B shows the volume of exudate recovered from the pouches after saline (control) or 1 ml of 1% carrageenan. In salinetreated animals the volume decreased due to the absorption of the saline injected at time 0. In carrageenan-treated animals the volume increased from 24 h to 7 days.
TNF and IL-6 concentrations ( Fig. 2A and B) measured in the pouch, were undetectable in control animals at all times. In carrageenantreated animals TNF and IL-6 both peaked 24 h after treatment, then declined very rapidly. Prostaglandin E 2 in pouch exudates (Fig. 3A) was never detectable in control animals but in carrageenan-treated animals there was a continuous increase from 5 h to 7 days. Plasma exudation measured as the percentage of FITC-BSA in the pouch compared with serum was considered as an indicator of the changes of vessel permeability induced by in ammation. Injection of carrageenan into the pouch signicantly increased plasma exudation after 30 min, 5 h and 24 h (Fig. 3B). IL-1b was never found in pouch exudate (detection limit < 50 pg ml). To detect any systemic in ammatory effects after carrageenan in the air pouch, we measured circulating levels of IL-1b, IL-6, TNF and the parameters associated with an acute phase response: acute phase SAA and serum iron. We did not nd hypoferraemia and TNF, IL-1, IL-6 and SAA were all below the limit of detection (data not shown), indicating that no systemic in ammatory effects were induced. To investigate whether the in ammation was con ned to the periphery because of the structure of the pouch or the type of stimulus, we injected 100 ng mouse of LPS into the pouch instead of carrageenan. In this case TNF, measured 90 min after treatment was increased both in pouch exudate (21399 9389 pg ml) and in serum (502 173 pg ml). Data are the mean SD of at least eight animals. P < 0 05, P < 0 01 according to Duncan's test for multiple comparisons.

Effect of anti-in¯ammatory drugs on the various parameters of in¯ammation
The effects of systemic treatment with the antiin ammatory drugs DEX or INDO are reported in Table 1. Both inhibited cell migration, the formation of exudate and the production of TNF, IL-6 and PGE 2 . However DEX had much more effect than INDO.

Effect of neutropenia
Neutropenia induced by pretreatment with 5-FU (Table 2), as expected reduced the number of leukocytes recruited into the pouch by carrageenan and the amount of TNF measured in the pouch. However the volume of exudate, plasma exudation and the concentration of PGE 2 were not different from the group treated with carrageenan alone.

Role of TNF
The protective effects of DEX and INDO, associated with an inhibitory effect on TNF production, suggest that TNF plays a role in this model of in ammation. To test this we used anti-TNF antibodies and IL-10, which inhibits TNF production. As shown in Table 3, anti-TNF antibodies (injected with carrageenan) reduced the induction of IL-6 in the exudate by 68%and cell recruitment by 44%. The dose of anti-TNF completely neutralized the levels of endogenous TNF induced by carrageenan ( Table 3).
Reduction of in ammation associated with a lowering of TNF levels was con rmed by treating the animals with IL-10 (1 mg mouse) administered into the pouch together with carrageenan or i.p. immediately before carrageenan. After 24 h the volume of exudate was not affected but TNF was inhibited by more than 90% in both experiments. Cell migration into the pouch was reduced by 67% and 58% respectively (data not shown).
Because mast cells were reported to be present in the lining of the pouch, we induced in ammation in a mast cell-de cient strain of mice to investigate the role of TNF in their granules. Table 4 reports a reduced in ammatory response 24 h after treatment. The inhibition of cell migration was particularly signi cant. One ml of sterile saline or 1 ml of 1% carrageenan in sterile saline were injected in 6 day-old pouches. 0.2% dexamethasone in drinking water was administered from 12 h before carrageenan. Two doses of indomethacin (2 mg kg) were injected intraperitoneally 30 min before carrageenan and 8 h later. Animals were killed 24 h after the irritant. Data are the mean SD of at least ® ve animals. P < 0 01, P < 0 05 according to Kruskal ± Wallis and Sachs's test for multiple comparisons.  Sterile saline (1 ml) or 1% carrageenan in sterile saline (1 ml) were injected in 6 day-old pouches. Anti-TNF antibodies (10 ml which diluted 1:12800 neutralizes 256 U of murine TNF) were injected into the pouches at time 0 together with carrageenan. The animals were killed 24 h after treatment. Data are the mean SD of at least eight animals in two different experiments. P < 0 05, P < 0 01 compared with the control group, P < 0 05, P < 0 01 compared with the carrageenan group according to Kruskal ± Wallis and Sachs's test for multiple comparisons.

Discussion
The objective of this study was to investigate the role of cytokines, particularly TNF, during the rst phase of acute in ammation and to assess the relationship between these mediators and the time course of the cell migration and PGE 2 production. Other commonly used models of local in ammation, such as the subcutaneous injection of turpentine 21 or the carrageenan paw oedema, 22 give information on the oedematous response, but do not allow easy measurement of the cellular or humoral effects locally. The advantage of the air pouch model lies in its structural similarities with the synovial lining tissue 10 and the fact that the in ammatory exudate can be easily sampled and quanti ed for its cellular and biochemical composition. In this experimental model only a local in ammatory response is induced, with none of the systemic acute-phase changes usually observed in other models of local in ammation, including increases of IL-1, TNF, SAA, or hypoferraemia. 23 The con nement of the in ammation to the site of injection is not due to the particular structure of the pouch because TNF in serum was measurable when LPS, even at a low dose, was injected into the pouch. One possible explanation for this effect is that carrageenan is a less potent stimulus than LPS; it might also diffuse less than LPS.
The amount of TNF measured in the pouch seems to be related to the number of migrating cells. After treatment with 5-FU, when no cells are present in the pouch, TNF is below the detection limit. In addition the peak of TNF, 24 h after carrageenan, correlates with the increase of cells in the pouch. However the kinetics of cell in ltration is delayed compared with that of TNF and, since TNF is not present in serum, a possible source during the rst phases might be the tissue of the pouch. In fact the lining of the pouch was reported to contain many mast cells, 24 the only cell type that stores TNF in the granules. 25 Therefore they might be the rst source of TNF after stimulation and the TNF released from mast cells was reported to account for 40%of the PMN migration into the peritoneal cavity after immunocomplex-induced peritonitis and Arthus reaction in rats. 25 Using a mast cellde cient strain of mice, carrageenan-induced cell migration was reduced 40%, seemingly in agreement with these data. However, the levels of TNF induced by carrageenan were comparable in normal and mast cell-de cient mice, indicating that these cells do not contribute signicantly to TNF production in this model (evaluated 24 h after carrageenan). The protective effects of treatment with anti-TNF antibodies or IL-10 on some parameters of carrageenaninduced in ammation strengthen the suggestion that TNF plays a role in the process.
The time course of the production of PGE 2 parallels that obtained by Sin et al. 26 although the absolute amounts are completely different. We measured the PGE 2 released spontaneously in the exudate while Sin et al. 26 measured it both in cells and exudate. Since the increase of PGE 2 parallels the increase of cells in the pouch PGE 2 might be produced and released by leukocytes recruited and activated by carrageenan. However, after treatment with 5-FU, which induces neutropenia and prevents the accumulation of cells, the amount of PGE 2 was not affected, so it might also be formed by the pouch tissue and not by the migrating cells. These data seem in agreement with the ones in the rat reported by Sedgwick and Lees 27 and by Simmons et al. 28 PGE 2 has been shown to down-regulate TNF production in vitro 12 and INDO, by breaking this negative feedback, potentiated LPS-induced TNF synthesis both in vitro 12,29 and in vivo. 30 We obtained contrasting results. Pretreatment with DEX or INDO reduced PGE 2 synthesis but also lowered TNF content. We measured TNF and PGE 2 after 24 h when the number of leukocytes in the pouch without INDO pretreatment was double that after INDO. Therefore if TNF is produced by leukocytes these data are not surprising.
Taken together the results indicate that TNF is a mediator of in ammation in the carrageenan-induced mouse air pouch model. One ml mouse of 1% carrageenan in sterile saline was injected in 6 day-old pouches of both experimental groups. The animals were killed 24 h later. FITC-BSA (0.6 mg mouse) was injected i.v. 30 min before the death. Data are the mean SD of at least six animals. P < 0 05 according to Student's t-test.