Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

In several diseases, the proinflammatory cytokine tumor necrosis factor- (TNF) forms a necessary element in the chain of pathophysiologic events leading to inflammation. Successful treatment with anti-TNF-antibodies in patients with Crohn's disease (van Dullemen et al., 1995; Stack et al., 1997; Targan et al., 1997; Present et al., 1999), with rheumatoid arthritis (Elliott et al., 1994), and with Jarisch-Herxheimer reaction (Fekade et al., 1996) illustrate anti-inflammatory strategies based on the specific blockade of TNF (Eigler et al. 1997). Among the agents known to inhibit TNF production rather than block its function, attention has focused on cAMP-elevating phosphodiesterase (PDE) inhibitors. The predominant PDE isoenzyme family in monocytes, a main source of TNF production, is the PDEs of type IV. Compared with the nonspecific PDE inhibitor pentoxifylline (Strieter et al., 1988), the specific type IV PDE inhibitor rolipram is a 500-fold more potent inhibitor of TNF synthesis in human mononuclear cells (Semmler et al., 1993).

Rolipram has initially been developed and studied in clinical trials as an antidepressant (Wachtel 1983). Recently, the potential therapeutic use of rolipram in TNF-dependent disease has been demonstrated in several animal models. Rolipram mitigated experimental autoimmune encephalomyelitis in rats (Sommer et al., 1995) and in nonhuman primates (Genain et al., 1995) and decreased clinical activity of experimental arthritis in rats (Nyman et al., 1997; Ross et al., 1997). In mice, rolipram decreased lipopolysaccharide-induced TNF plasma levels (Fischer et al., 1993) and protected from T-cell-mediated liver failure (Gantner et al., 1997).

In humans, the majority of inflammatory bowel disease occurs in two related, albeit clinically and histologically distinct disorders, ulcerative colitis and Crohn's disease. Both diseases are characterized by chronically relapsing inflammation of the bowel of unknown cause. Crohn's disease is characterized by a granulomatous, transmural inflammation of the bowel wall, predominantly in the distal ileum. In contrast, ulcerative colitis is defined by crypt abscesses and ulcerations limited to mucosa and submucosa, associated with a prominent inflammatory infiltrate. The mainstay of therapy for inflammatory bowel disease is aminosalicylates and topical and systemic glucocorticoids. Both provide therapeutic benefit and improve quality of life in many patients, but others suffer from recurrent disease despite systemic glucocorticoid therapy. More specific and more effective therapeutic agents with fewer side effects are needed for the permanent control of inflammatory bowel disease.

Several experimental models of inflammatory bowel disease have been described (Kim and Berstad, 1992; Dieleman et al., 1994; Elson et al., 1995). The dextran sulfate sodium (DSS) model of colitis has been recommended for preclinical testing of new pharmacologic compounds for therapy of chronic inflammatory bowel disease (Cooper et al., 1993; Elson et al., 1995). DSS-induced colitis has a number advantages, including its simplicity, the ability to induce inflammatory lesions, and the reproducibility in respect to both time course and severity among individual mice of a given inbred strain. As in Crohn's disease, macrophage activation and TNF production play a key role in DSS-induced colitis. Elevated levels of TNF have been found in the inflamed colons of DSS-treated mice (Dieleman et al., 1994).

Several studies pointed to a necessary mediator function of TNF, particularly in Crohn's disease (Targan et al., 1997; van Deventer, 1997; Present et al., 1999). TNF elevation was more pronounced in Crohn's disease than in ulcerative colitis both in plasma (Murch et al., 1991) and mucosal samples (Murch et al., 1993; Breese et al., 1994). In the present article, we extend our previous in vitro studies on TNF inhibition by rolipram (Semmler et al., 1993; Siegmund et al., 1997; Eigler et al., 1998) to demonstrate that rolipram attenuates the development of experimental colitis and improves recovery of animals with established colitis.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Animals and Induction of Colitis. Female BALB/c mice (~6 weeks of age, mean body weight 20 g) were purchased from Harlan Winkelmann GmbH (Borchen, Germany). Mice were kept under standard laboratory conditions at the animal facility at the Medizinische Klinik, Klinikum Innenstadt. Drinking water and food were provided ad libitum. Mice were sacrificed by cervical dislocation under isoflurane anesthesia (Forene; Abbott GmbH, Wiesbaden, Germany). All experiments were approved by the regional animal study committee and are in agreement with the guidelines for the proper use of animals in biomedical research. Animal handling and scoring of colitis were performed in a consequently blinded experimental design.

Clinical Activity Score. Colitis was quantified with a clinical score assessing weight loss, stool consistency, and bleeding (measured by guaiac reaction, hemoccult) as described previously (Cooper et al., 1993). No weight loss was counted as 0 points, weight loss of 1 to 5% as 1 point, 5 to 10% as 2 points, 10 to 20% as 3 points, and >20% as 4 points. For stool consistency, 0 points were given for well formed pellets, 2 points for pasty and semiformed stools that did not stick to the anus, and 4 points for liquid stools that did stick to the anus. Bleeding was scored 0 points for no blood in hemoccult, 2 points for positive hemoccult, and 4 points for gross bleeding. These scores were added and divided by 3, forming a total clinical score that ranged from 0.0 (healthy) to 4.0 (maximal activity of colitis).

Colon Length, Histologic Scoring, and Mean Cross-Sectional Area. Postmortem, the entire colon was removed from the cecum to the anus and placed without tension on cellulose. Colon length was measured as an indirect marker of inflammation.

Treatment with Rolipram. Rolipram (0.5 mg), kindly supplied by Schering AG (Berlin, Germany), was diluted in 1 ml of distilled water by heating the solution for 30 s at 60°C and then cooling it at room temperature for 2 min. This was repeated until rolipram was totally dissolved. The solution was then frozen into aliquots of 2 ml at 80°C. This proved to be the most reliable protocol to dissolve a relatively high concentration of rolipram for a low total injection volume of 200 µl. Rolipram (5 mg/kg b.wt. b.i.d.) was injected two times per day i.p. with a total injection volume of 200 µl each. Control mice were injected with 200 µl of 0.9% NaCl. To test the therapeutic efficacy of rolipram, two protocols were used: 1) in the concurrent treatment protocol (prevention of colitis), DSS was administered for up to 11 days with rolipram therapy starting the same day as DSS; and 2) in the delayed treatment protocol (treatment of established colitis), colitis was first induced by DSS administration from day 1 to day 7, rolipram therapy was then started at day 8, and was continued up to day 15.

Colon Cytokine Extraction. Strips (~4 cm) of colon from DSS-exposed mice with or without rolipram treatment were weighed, vigorously vortexed for 1 min in 100 µl of 0.01 M PBS (Boehringer Mannheim, Ingelheim, Germany), and centrifuged at 10,000g at 4°C for 15 min. TNF was quantified in the eluate with a commercial enzyme-linked immunosorbent assay (ELISA) kit (Endogen, Woburn, MA) according to the manufacturer's instructions. The lower limit of detection of the assay is 50 pg/ml.

Statistical Analysis. All data are expressed as means ± S.E. Statistical significance of differences between treatment and control groups was determined by the unpaired two-tailed Student's t test. Where applicable, P values were corrected by the Bonferroni method for three independent comparisons (clinical score, colon length, and histologic degree of inflammation). Differences were considered statistically significant for P < .050. Statistical analyses were performed with StatView 512 software (Abacus Concepts, Berkeley, CA).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Characteristics of Colitis. Treatment of BALB/c mice with 5% DSS in drinking water for 7 or 11 days resulted in clinical, gross, and histologic signs of colitis that resolved gradually when DSS administration was discontinued (Figs. 1-5). Mice produced loose stool or diarrhea, occult or gross rectal bleeding, and lost weight. After 11 days, the colon length of DSS-treated mice was 12.4 ± 0.3 cm (n = 5) compared with 17.4 ± 0.3 cm (n = 4; P < .001) in healthy controls. This has been described as a morphologic parameter of colon inflammation (Okayasu et al., 1990).

View larger version (127K): [in this window] [in a new window]   Fig. 1.   Histologic characteristics of DSS-induced colitis. A, normal colonic mucosa with regularly formed colonic folds covered by intact mucosa. Only occasional leukocytes are present in the lamina propria of the mucosa. B, a focal accumulation of lymphocytes, histiocytes, and fewer neutrophils compared with (A). Although the surface layer of the mucosa is still intact, there is marked crypt destruction by the inflammatory infiltrate. The inflammatory process is limited to the mucosa and submucosa. C, severe colonic inflammation, characterized by a dense inflammatory cell infiltrate in all colonic wall structures and widespread mucosal sloughing. Note the circumscription of the destructive process, immediately adjacent to normal appearing mucosa. Magnification of the images is 130-fold for (A) and (C), and 325-fold for (B).

View larger version (117K): [in this window] [in a new window]   Fig. 2.   Influence of rolipram on the histologic manifestation of DSS-induced colitis. Shown are representative cross sections of the transversal colon in control mice without DSS (A) and in DSS-treated mice treated with i.p. injections of 0.9% NaCl (B) or rolipram [10 mg/kg b.wt., (C)] at day 11. Note the regional mucosal destruction in DSS-treated mice (B), which is effectively reversed by rolipram (C). In rolipram-treated mice, there is still some degree of edema, but no cross-mucosal damage. Also note the difference in colon cross-sectional area, which is partially reversed by rolipram. Magnification of the images is 52-fold.

View larger version (9K): [in this window] [in a new window]   Fig. 3.   Time course of DSS-induced colitis at the histologic level. Mice were treated with 5% DSS in drinking water for 8 days. At the indicated time points, mice were sacrificed and sections of the ascending, transverse, and descending colon were examined and scored for cellular infiltration and tissue damage. The magnitude of the histologic score reflects the degree of colitis (0, no changes, to 6, severe tissue damage and cell infiltration). Data represent the mean ± S.E. histologic score (average of the scores of a ascending, transverse, and descending section) of two mice (days 1 to 13) or of seven mice (day 15). Mice without DSS treatment had no signs of colitis (histologic score 0).

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Accelerated remission of established colitis by rolipram. Mice were treated with DSS in drinking water for 7 days. Rolipram treatment (10 mg/kg b.wt. daily) was started at day 8 after discontinuation of DSS. The degree of colitis was quantified by the clinical score (assessing weight loss, stool consistency, and bleeding; see Materials and Methods). Mice with rolipram treatment () recovered earlier than mice without rolipram treatment (gray circles; P = .022 at day 15; unpaired t test). Control mice without DSS exposure injected with either rolipram or 0.9% NaCl developed no signs of colitis. Scores are depicted as means ± S.E. DSS groups, n = 14; control groups, n = 6).

View larger version (12K): [in this window] [in a new window]   Fig. 5.   Reduction of colon length in DSS-induced colitis with and without rolipram treatment. Mice were treated with DSS in drinking water for 7 days. Rolipram treatment (10 mg/kg b.wt. daily) was started at day 8 after discontinuation of DSS. At the indicated time points, two mice of both DSS groups and one mouse of both control groups (with or without rolipram, respectively) were sacrificed and the length of the colon was measured. At day 15, the remaining 7 mice of both DSS groups and the remaining 3 mice of both control groups were examined. DSS-induced colitis lead to shortening of the colon as a marker of inflammation. Rolipram treatment partially reversed this shortening (P < .001 at day 15; unpaired t test). Values are depicted as the mean length of the colon in each group. Error bars indicate S.E.M.

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Reduction of histologic signs of colon inflammation by rolipram. Mice were treated with 5% DSS in drinking water for 11 days and were given rolipram (10 mg/kg b.wt./day i.p.; n = 5) or 0.9% NaCl (n = 5). At day 11, mice were sacrificed and colon sections were stained and the histologic score (degree of inflammation: 0, no changes, to 6, extensive cell infiltration and tissue damage) was determined in a blinded fashion by a pathologist. In the rolipram-treated group, the histologic score was lower compared with mice given 0.9% NaCl (P = .020; unpaired t test). Control mice without administrating DSS developed no colitis. Scores are depicted as means ± S.E.

Prevention of Colitis with Rolipram. In the concurrent treatment protocol, we tested the effect of rolipram on the prevention of DSS-induced colitis. Mice were administered 5% DSS in their drinking water and were injected i.p. with rolipram or with 0.9% NaCl for a total of 11 days. This protocol was studied in two independent experimental series. The first series comprised 14 mice. Mice fed with DSS developed clinical signs of colitis expressed by an activity score >0.5 starting from day 4 (Fig. 7). Intraperotineal injection of rolipram in a dose of 10 mg/kg b.wt. daily did not retard onset of colitis during the first 6 days of DSS administration. After that, it significantly reduced the progression of colitis as expressed by a lower clinical activity score (1.1 ± 0.3; n = 5 in rolipram-treated mice compared with 2.4 ± 0.4; n = 5 in NaCl controls; P = .041; day 11) (Fig. 7). Each of the three clinical parameters assessed in the clinical score was beneficially influenced by rolipram (body weight, 19.5 g ± 1.5 g in rolipram-treated mice versus 18.0 ± 0.9 g in NaCl-treated controls; stool consistency score, 0.0 ± 0.0 versus 1.2 ± 0.5; rectal bleeding score, 0.8 ± 0.8 versus 2.4 ± 1.0). Control mice (without DSS) treated with 0.9% NaCl or rolipram i.p. developed no signs of colitis (Fig. 7). At day 11, all mice were sacrificed. Of the DSS-treated mice, those with rolipram therapy had a longer colon (15.4 ± 0.7 cm) than NaCl controls (12.4 ± 0.3 cm; P = .004). Thus, rolipram had partially reversed the colon shortening induced by DSS compared with mice with normal drinking water (17.4 ± 0.3 cm). This argued for a lower extent of inflammation, which was confirmed by histologic examination in the rolipram group (Fig. 2C) compared with the NaCl group (Fig. 2B). In DSS-treated mice, rolipram decreased the histologic score compared with NaCl-treated control mice (1.5 ± 0.6 versus 4.6 ± 0.5; P = .020; Fig. 6).

View larger version (16K): [in this window] [in a new window]   Fig. 7.   Mitigation of DSS-induced colitis by rolipram. Mice were treated with 5% DSS in drinking water for 11 days. The degree of colitis was quantified by the clinical score assessing weight loss, stool consistency, and bleeding (range from 0, healthy, to 4, maximal activity; see Materials and Methods). Rolipram treatment (10 mg/kg b. wt. b.i.d. i.p.; ; n = 5) beginning the same day as DSS administration reduced the development of colitis compared with mice 0.9% NaCl treatment (gray circles; n = 5; one mouse died on day 8; P = .041; unpaired t test). Control mice (n = 2) without DSS treatment but injection of rolipram (; n = 2) or of 0.9% NaCl (data not shown; n = 2) showed no clinical signs of colitis. Scores are depicted as means ± S.E.

View larger version (16K): [in this window] [in a new window]   Fig. 8.   Suppression of colonic TNF concentration by rolipram. Mice were treated with 5% DSS in drinking water for 11 days. At day 11, the colon was removed, weighed, vortexed for 1 min in 100 µl of PBS, and centrifuged at 10,000g for 15 min as described in Materials and Methods. TNF was quantified in the eluate by ELISA. Maximal TNF concentration could be measured in colonic tissue of DSS-treated mice (201 ± 39 pg/mg colonic tissue; n = 4) compared with rolipram-treated mice (115 ± 25 pg/mg colonic tissue; n = 7). In control mice colon, we observed the lowest concentrations, respectively (97 pg/mg colonic tissue; n = 2 in each control group).

Treatment of Established Colitis with Rolipram. To further evaluate the therapeutic value of rolipram, we studied the effect of rolipram on preexisting colitis (delayed treatment protocol). Colitis was induced by the administration of DSS for 7 days. On day 8, after discontinuing DSS administration, rolipram therapy was started. A total of 88 mice was included in two independent series of studies. The first series was designed to assess clinical score and postmortem morphologic parameters at defined time points during resolution of colitis. In the second series, clinical parameters were followed until complete resolution of clinical colitis in treatment and control groups.

View larger version (30K): [in this window] [in a new window]   Fig. 9.   Beneficial effect of rolipram on weight loss, stool consistency, and bleeding. Clinical parameters in two independent series of experiments with 54 mice (A) and 34 mice (B) examining the effect of rolipram in the model of established colitis. Mice were treated with 5% DSS for 7 days before starting therapy with rolipram (10 mg/kg b.wt./day) on day 8. Clinical signs of colitis are summarized for the final day of the first series [day 15 (A), time course of composite clinical score of this study shown in Fig. 8] and for the day of maximal difference between treatment groups in the second series [day 11 (B); in this series, no significant differences at day 15]. Loss of body weight, stool consistency, and bleeding were semiquantified by scores (see Materials and Methods). In the first series, one mouse in the DSS group without rolipram died at day 8 for unknown reasons. Data are shown as means ± S.E. Statistical evaluation was performed using the unpaired t test and P values were Bonferroni corrected.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Summary of Results. Type IV PDE inhibitors have been identified as a potential therapeutic principle for asthma (Turner et al., 1994) and multiple sclerosis (Genain et al., 1995; Sommer et al., 1995) because they are potent inhibitors of diverse leukocyte functions. In this study, we present evidence that the specific type IV PDE inhibitor rolipram abrogates experimental colitis in mice. A total of 138 female BALB/c mice was included in our studies. Administration of DSS in drinking water induced clinical and histologic signs of colitis. In agreement with published studies, we found reproducible and interindividually similar degrees of colitis in DSS-treated mice.

DSS Model. Several experimental models of inflammatory bowel disease have been described (for review, see Elson et al., 1995). The DSS model of colitis has been recommended for preclinical testing of new pharmacologic compounds for therapy of chronic inflammatory bowel disease (Cooper et al., 1993; Elson et al., 1995). DSS-induced colitis has a number of advantages, including its simplicity, the ability to induce both acute and chronic inflammatory lesions, the high degree of uniformity of the lesions, and the reproducibility in respect to both time course and severity among individual mice of a given inbred strain. This uniformity and reproducibility is not achieved in several other experimental models of inflammatory bowel disease (Elson et al., 1995).

Side Effects of Rolipram. In general, therapy with rolipram at a dose of 10 mg/kg b.wt./day was well tolerated by the mice. Immediately after injection of rolipram, mice showed reduced motility, but they returned to normal motoric behavior after a few minutes. These changes could not be attributed to the i.p. injection of fluid because changes in the behavior did not occur in control mice injected with 0.9% NaCl. In agreement with previous in vivo studies (Turner et al., 1994; Genain et al., 1995; Gantner et al., 1997; Nyman et al., 1997; Ross et al., 1997) with the same (10 mg/kg b.wt./day) or lower doses of rolipram, we observed no major side effects. For higher doses, side effects in nonhuman primates include vomiting, salivation, and mouth scratching, none of which was observed in the present study (Genain et al., 1995).

Cellular Effects of Rolipram. The anti-inflammatory activity of rolipram depends on its direct action on leukocytes. We and others have shown that rolipram strongly inhibits TNF production in monocytes and macrophages (Schade and Schudt, 1993; Semmler et al., 1993; Prabhakar et al., 1994; Seldon et al., 1995; Barnette et al., 1996). We have demonstrated that the synthesis of the anti-inflammatory cytokine interleukin (IL)-10 is enhanced by rolipram (Eigler et al., 1998) and that exogenous IL-10 acts synergistically with rolipram in decreasing TNF production (Siegmund et al., 1997). Furthermore, rolipram inhibits IL-2-mediated proliferation of primary T cells but not IL-2 production itself (Essayan et al. 1994) and inhibits -interferon synthesis of T cells (Essayan et al., 1994; Sommer et al., 1995). These anti-inflammatory characteristics of rolipram act in concert to effectively inhibit the inflammatory response in vivo (Turner et al., 1994; Genain et al., 1995; Sommer et al., 1995; Gantner et al., 1997).

Determination of Endpoints in Colitis Model. We quantified clinical activity with a scoring system that has been described to be a reliable marker of pathologic changes (Cooper et al., 1993). The clinical score was determined in a blinded fashion to exclude bias by the examining person. Shortening of the colon as a morphologic parameter for the degree of inflammation correlates well with pathologic changes (Okayasu et al., 1990). In our studies, the length of the colon proved to be an easily determined and consistent marker of colitis. Histologic examination was performed blinded and included the degree of infiltration by inflammatory cells in the mucosa and the degree of tissue damage. In our study, a histologic score calculated from these two markers was found to parallel clinical changes during induction of colitis.

TNF- and Colitis. There is evidence that TNF plays a central role in inflammatory bowel disease (for review, see van Deventer, 1997; Sandborn and Hanauer, 1999). The therapeutic benefit of TNF inhibition in Crohn's disease has been shown in clinical studies with chimeric anti-TNF antibodies (van Dullemen et al., 1995; Stack et al., 1997; Targan et al., 1997; Present et al., 1999). Although proving the principle of targeting TNF in inflammatory bowel disease, efficacy of anti-TNF antibody therapy may decrease with time because of the formation of anti-idiotype antibodies. Furthermore, the development of antinuclear antibodies has been observed with prolonged anti-TNF antibody therapy in patients with rheumatoid arthritis (Elliott et al., 1994).

PDE Inhibition and TNF-. Intracellular concentrations of cAMP increase either as a consequence of receptor-triggered adenylyl cyclase activation or by decreased activity of PDEs. Cyclic nucleotide PDEs have been classified into nine distinct families with several subgroups (Beavo, 1995). Because the predominant PDE family in monocytes, the main source of TNF, is PDE IV (Seldon et al., 1995; Souness et al., 1997), specific type IV PDE inhibitors such as rolipram have proven high potency in suppressing TNF synthesis. Activators of adenylyl cyclase, such as prostaglandin E₂ and prostacyclin analogs, synergize with PDE inhibitors both in increasing cAMP levels and in suppressing TNF synthesis (Sinha et al., 1995). One may speculate that PDE inhibitors given systemically may exert their maximal TNF-suppressing effect in tissue containing prostaglandin E₂ such as the inflamed mucosa of inflammatory bowel disease.

Rolipram In Vivo. In humans, the specific type IV PDE inhibitor rolipram has been extensively studied as an antidepressant. However, it has not been marketed because of the lack of an additional therapeutic benefit compared with established drugs. Several studies have revealed a beneficial effect of rolipram in inflammatory disease in vivo. In the rat model of experimental autoimmune encephalomyelitis, where TNF synthesis forms a central pathogenetic link, rolipram decreased disease activity (Sommer et al., 1995). In nonhuman primates, rolipram protected against autoimmune demyelinating disease even when administered after sensitization to central nervous system antigens (Genain et al., 1995). In rats, rolipram decreased clinical activity of experimental arthritis (Nyman et al., 1997; Ross et al., 1997). Additionally, LPS-induced TNF synthesis in mice could be suppressed by rolipram (Griswold et al., 1998). Rolipram reduced airway hyper-responsiveness in response to acute and chronic antigen exposure in monkeys (Turner et al., 1994). In this study, antigen-induced increases of TNF but not of IL-1 concentration were inhibited in bronchoalveolar lavage. This is in agreement with our in vitro findings of selective inhibition of TNF but not IL-1 synthesis by rolipram (Semmler et al., 1993).

Phosphodiesterase Inhibition and Inflammatory Bowel Disease. To date, specific type IV PDE inhibition has not been tested as a therapeutic strategy for inflammatory bowel disease. However, there are reports on the use of the nonspecific PDE inhibitor pentoxifylline for this indication. In the trinitrobenzene sulfonic acid model of colitis in rats, pentoxifylline treatment reduced the pathologic changes (Peterson and Davey, 1997). Yet, in a small clinical study with 16 patients with corticosteroid-dependent Crohn's disease, the administration of 400 mg of pentoxifylline administered four times a day did not improve clinical or histologic activity of disease (Bauditz et al., 1997). In vitro studies by the same research group revealed that a high concentration of pentoxifylline (IC₅₀ = 25 µg/ml) was necessary to inhibit TNF synthesis in organ cultures of inflamed mucosa (Reimund et al., 1997). Nonspecific PDE inhibitors are less selective than rolipram and require a 500-fold higher concentration for inhibition of TNF synthesis. In human mononuclear cells, the concentration that inhibits TNF synthesis by 50% (IC₅₀) is 70 µM for pentoxifylline and 130 nM for rolipram (Semmler et al., 1993). This limits the therapeutic use of compounds such as pentoxifylline as anti-inflammatory agents. In the present study, we could demonstrate the suppression of colonic TNF concentrations by rolipram close to TNF concentrations observed in control mice. This emphasizes the mediatory function of TNF in this model.