Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Unrestrained activation of the intestinal immune system appears to be responsible for sustained mucosal inflammation and the characteristic relapsing course of inflammatory bowel diseases (Podolsky, 1991). Interleukin-1 and TNF are two key cytokines that share a pivotal role in mucosal immunoinflammatory responses. Some studies have shown increased mucosal release of TNF in patients with inflammatory bowel disease (Breagger et al., 1992; Casellas et al., 1994). Furthermore, expression of TNF in the mucosa is increased in Crohn's disease (Breese et al., 1994). Hence, Van Dullemen et al. (1995) investigated the potential efficacy of an anti-TNF monoclonal antibody in the treatment of active Crohn's disease and reported a significant benefit in patients with steroid-refractory disease. Recently, controlled clinical trials have also shown a therapeutic effect of antibody neutralization of TNF in Crohn's disease (Stack et al., 1997; Targan et al., 1997).

TNF- is produced chiefly by activated macrophages and monocytes (Tracey and Cerami, 1993). This cytokine is involved in the activation of neutrophils (Shalaby et al., 1985), up-regulation of adhesion molecules (Gamble et al., 1985), induction of nitric oxide synthase (Thiemermann et al., 1993), cell-mediated immunity (Bromberg et al., 1992) and granuloma formation (Amiri et al., 1992), among other proinflammatory pathways (Tracey and Cerami, 1993). Most of these effects are mediated by a cascade of inflammatory substances that are induced or released by TNF, such as platelet activity factor, eicosanoids, nitric oxide, IL-1, IL-6 and IL-8. Therefore, inhibition of TNF overproduction down-regulates inflammation and is of potential therapeutic benefit. At the same time, however, TNF plays important roles in host defense against infection. In particular, TNF is essential for immunity against intracellular pathogens and also exerts antiviral effects (Havell, 1989; Rothe et al., 1993; Wong and Goeddel, 1986). Thus TNF blockade may have adverse consequences by disarming the organism against pathogens.

Increasing evidence incriminates enteric bacteria in the pathogenesis of intestinal inflammation. For instance, HLA-B27 transgenic rats spontaneously develop chronic colitis, but rats bred in a germ-free environment fail to develop the disease (Taurog et al., 1994). Likewise, intestinal inflammation in IL-10 knockout (Kühn et al., 1993) mice is dependent on the presence of normal, nonpathogenic intestinal microflora. Strong evidence is also provided by the fact that antibiotics mitigate intestinal inflammation in several animal models, including chronic granulomatous colitis by TNBS (Videla et al., 1994). As shown by our laboratory (García-Lafuente et al., 1997; Videla et al., 1997), bacteria of the common rat flora play a role in the induction of mucosal lesions in TNBS colitis. In this setting, TNF blockade might compromise mucosal defenses against luminal bacteria.

The aim of the current study was to test the efficacy and safety of anti-TNF therapy in experimental colitis induced by TNBS. We cultured colonic wall specimens for bacterial growth and measured native TNF protein synthesis in colonic tissue at various times after induction of colitis. The effect of antibody neutralization of TNF on colonic inflammation and tissue damage was evaluated.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Animals

Male Sprague-Dawley rats weighing 220 to 230 g were purchased from CERJ (Le Genest, France). The animals were maintained in a restricted-access room with controlled temperature (23°C) and light-dark cycle (12 h:12 h) and were housed in rack-mounted cages with a maximum of five rats per cage. Standard rodent chow pellets (Biocenter, Barcelona, Spain) and tap water were provided ad libitum. The study was approved by the local Research Committee (Comissio de Recerca, Hospital General Vall d'Hebron).

Experimental Design

Colonic wall bacteria in TNBS colitis. Twelve rats were included in this study. Chow pellets were withdrawn for 36 hr, and rats were kept on tap water with 20% sucrose and electrolytes. On day 0, the rats were lightly anesthetized with ether, and colitis (n = 9) was induced by intracolonic instillation of 1 ml of a solution containing 60 mg of 2,4,6-TNBS (Sigma Chemical Co., St. Louis, MO) in 20% ethanol (Merck, Darmstadt, Germany) using a rubber cannula (8 cm long, external diameter 2 mm) inserted through the rectum. Control rats (n = 3) received 1 ml of saline. Thereafter, chow pellets and tap water were provided ad libitum.

Native TNF protein synthesis in TNBS colitis. Thirty rats were included in this protocol. Colitis was induced in 20 rats by intracolonic instillation of TNBS, as described above, whereas 10 control rats received intracolonic saline. Rats were killed by cervical dislocation 1, 4, 10 or 18 days after the intracolonic challenge. The distal colon was removed, opened longitudinally and rinsed with normal saline. A 2-cm piece of distal colon was cut, washed thoroughly in saline, homogenized in 2 ml of phosphate-buffered saline (50 mM, pH 7) with 2 mmol/l of phenyl-methyl-sulfonyl fluoride (Sigma) using a Tissue Tearer (model 985-370, Biospec, Racine, WI) and stored at 20°C for later TNF assay.

Effect of treatment with anti-TNF antibody on TNBS colitis. Fifty-six rats were included in this protocol. Colitis was induced in all rats by intracolonic administration of TNBS, as described above. Two treatment schedules tested the effect of anti-TNF monoclonal antibody on chronic colitis. In the first schedule, anti-TNF therapy was started on day 4 after TNBS during the acute phase of TNBS colitis (early treatment), and in the second schedule, treatment was started on day 10 after TNBS during the chronic phase of colitis (delayed treatment). As shown by previous studies using this model, histological lesions on day 3 after TNBS instillation consist of extensive coagulative necrosis of the mucosa with epithelial exfoliation, severe submucosal edema and congestion with vascular thrombosis, and acute neutrophil infiltrate (Vilaseca et al., 1990). Chronic changes are observed from 1 to 3 weeks after TNBS (Morris et al., 1989; Vilaseca et al., 1990). Chronic lesions by TNBS are segmental and well circumscribed, consisting of mucosal ulcerations with granulation tissue at the base and mixed transmural inflammation with neutrophils, lymphocytes and macrophages. Small granulomas are observed in the submucosa and serosa. In severe lesions, transmural fibrosis causes stricture of the lumen. Uninvolved areas do not show mucosal inflammation.

Microbiological Studies

Homogenates from the colonic wall at appropriate dilution volumes were incubated under aerobic or anaerobic conditions. Aerobic media consisted of blood agar, blood agar with nalidixic acid, McConkey agar, CLED agar and Sabouraud agar. Anaerobic media consisted of lacquered blood agar enriched with hemin and vitamin K₁, blood agar with phenyl-ethanol and agar-aztreonam (20 µg/ml). Plates inoculated for obligate anaerobes were incubated in an anaerobic chamber for 48 to 72 hr at 37°C, and plates for aerobes in air at 37°C. After incubation, colonies were identified. Final counts of colonies in cultures of colonic homogenates were referred to a gram of tissue.

Analytical Methods

Eicosanoid concentration in dialysis samples was measured by specific radioimmunoassay for PGE₂, TXB₂ and LTB₄ without prior extraction and HPLC purification (Vilaseca et al., 1990). Tritiated standards were purchased from du Pont de Nemours (Dreiech, Germany), and antisera for PGE₂ and LTB₄ from Advanced Magnetic (Cambridge, MA).

For the TNF assay, colonic tissue was homogenized with the protease inhibitor phenyl-methyl-sulfonylfluoride (2 mmol/l). Afterwards, samples were centrifuged and the supernatants used for measurement of the concentration of TNF by a commercial ELISA method for rat TNF (Biosource, Camarillo, CA). Results are expressed as nanograms of TNF per gram of wet colonic tissue.

Assessment of Colonic Lesions

The macroscopic lesions were scored by two observers who were unaware of the treatment (JV and FG). A macroscopic score was obtained by summation of scores on severity of colonic adhesions to surrounding tissues, strictures, mucosal ulcerations and wall thickening (table 1). For the histological studies, samples were processed by routine techniques before embedding in paraffin. Sections were obtained from areas showing macroscopic damage, stained with hematoxylin and eosin and coded for blind examination by two pathologist (AS and GG). Both pathologists examined and scored all sections according to the presence of ulcerations, degree of inflammation, depth of the lesions and fibrosis (table 1).

Statistical Methods

Results are presented as mean and S.E.M. Overall statistical difference was determined by one-way analysis of variance, and post-test comparison between treatment and placebo matched group was performed by Student's t test. Event rates were compared by Fisher's exact test.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Colonic wall bacteria in TNBS colitis. Table 2 shows quantitative bacterial isolates in cultures of colonic wall homogenates from control rats and rats with TNBS colitis. All samples from control colons without macroscopic lesions were positive for Lactobacillus species. By contrast, samples from rats with colitis were positive for several bacterial species. As shown, isolates included aerobes (Gram-positive and Gram-negative) and obligate anaerobes that are commonly found among the predominant species of the rat microflora. Interestingly, a number of colonies of Staphylococcus, Streptococcus and Bacteroides were consistently grown in cultures of colonic lesions at day 4 after TNBS, but Lactobacillus were absent. On day 10, however, no growth of Staphylococcus species was observed, and Streptococcus and Bacteroides were found at lower counts than on day 4. Cultures were again positive for Lactobacillus, as in control samples. By day 18, no growth of Staphylococcus, Streptococcus or Bacteroides was detected.

Native TNF protein synthesis in TNBS colitis. Figure 1 shows TNF content in colonic homogenates from control rats and rats with colitis induced by TNBS. A significant increase in tissue TNF concentration was observed on day 1 after induction of colitis, but peak levels were reached on days 4 and 10 after TNBS. In colonic homogenates from rats killed on day 18 after induction of colitis, TNF concentration returned to base-line levels similar to those found in control rats.

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Content of TNF protein in colonic homogenates from control rats (open bars) and rats with colitis induced by TNBS (solid bars) (*P < .05 vs. controls).

Effect of treatment with anti-TNF antibody on TNBS colitis. In the first study, five rats from the placebo group and four from the anti-TNF group died immediately after induction of colitis and before the start of treatment. In addition, one rat from the placebo group and two rats from the anti-TNF group died during the follow-up, so that by the end of the study there were 10 surviving rats in each group. Both groups of rats manifested a slowing of the rate at which they gained weight after induction of colitis. By day 18, normal growth rates were resumed, and rats treated with the anti-TNF antibody were heavier (334 ± 11 g) than placebo-treated rats (288 ± 12 g).

View larger version (27K): [in this window] [in a new window]   Fig. 2.   Intracolonic release of PGE2, TXB2 and LTB4 on day 18 after induction of colitis in rats treated with placebo (solid bars) and in rats treated with anti-TNF monoclonal antibody (hatched bars). Early treatment was started on day 4 after induction of colitis and continued up to the end of the study. Delayed treatment was started on day 10 after induction of colitis and continued up to the end of the study (*P < .05 between placebo and anti-TNF rats).

View larger version (28K): [in this window] [in a new window]   Fig. 3.   Morphological scores of colonic lesions on day 18 after induction of colitis in the groups of rats described in figure 2. Macroscopic and histological scores were obtained according to the criteria stated in table 1. Solid bars represent data from placebo-treated rats, and hatched bars represent data from rats treated with anti-TNF monoclonal antibody (*P < .05 between placebo and anti-TNF rats).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The current study shows that in the early stages of TNBS-induced colitis in the rat, there is superinfection of the colonic lesions with enteric bacteria, including Streptococcus, Staphylococcus, Bacteroides, Clostridium and Gram-negative enterobacteria. Conversely, in chronic lesions only Gram-negative enterobacteria, clostridia and lactobacilli are isolated from the colonic wall. Histological evidence for bacterial invasion of inflammatory colonic lesions was previously shown in experimental colitis (Garcia-Lafuente et al., 1997) and human ulcerative colitis (Ohkusa et al., 1993). The present study suggests that infection of colonic lesions in TNBS-induced colitis is polymicrobial with mixed aerobic and anaerobic organisms. In acute lesions, we found streptococci and other bacterial species, such as Staphylococcus and Bacteroides, that possess destructive extracellular enzymes. Such histolytic enzymes include phospholipases, nucleases, proteases and collagenases (Bjornson, 1984; Macfarlane et al., 1988). In chronic lesions, however, infection by potentially histolytic species was much less common, and isolates consisted mainly of enterobacteria and lactobacilli. Enterobacteria (Spitz et al., 1994) and lactobacilli (Bernet et al., 1994) are adherent bacteria to the intestinal epithelium and may not directly induce tissue injury. Endotoxin lipopolysaccharide from enterobacteria exerts its biological effects largely by triggering the release of endogenous mediators of inflammation. Thus host responses, rather than intrinsic toxicity, account for most of the damage generated in tissues and organs by endotoxin. Inhibition of the inflammatory response by anticytokine strategies is clearly beneficial in endotoxin-induced shock (Beutler et al., 1985).

Our study shows a different effect of anti-TNF monoclonal antibody on colonic inflammation, depending on the time of dosing. Anti-TNF therapy effectively induced biochemical and histopathological remission of colonic inflammatory lesions when initiated beyond the acute phase of the disease. By contrast, anti-TNF therapy proved harmful when started during the early acute flare of the disease. Our data also indicate that native TNF production in inflamed colonic tissue is enhanced both in the acute phase and the chronic phases and that the outcome of anti-TNF therapy cannot be predicted by the tissue TNF levels. Our study therefore suggests that anti-TNF therapy may be effective and useful in the treatment of chronic intestinal inflammation, whereas acute mucosal lesions may not respond to this therapy.

TNF- prevents disseminated infection with massive bacteremia in several models (Havell, 1989). Evidence provided by experimental studies indicates that translocation of intestinal bacteria stimulates a systemic TNF response (Guo et al., 1995) and that anti-TNF therapy induces dissemination of intestinal bacteria in mice with an injured colonic mucosal barrier (Echtenacher et al., 1990). It was shown that s.c. infection by Staphylococcus aureus is aggravated by neutralizing antibodies to TNF (Vaudaux et al., 1992). A recent study reports that passive immunization against TNF induces a 4-fold increase in bacterial counts of Streptococcus pneumoniae in lung isolates from mice with pneumonia and impairs survival (Van der Poll et al., 1997). In our study, early treatment with anti-TNF antibody might boost overgrowth of bacteria with histolytic potential within the colonic wall as a consequence of TNF inhibition and thus expand the area of injury and inflammation. By contrast, delayed treatment would induce truly anti-inflammatory effects, because tissue-destroying bacteria do not appear to participate in chronic stages of colonic inflammation. We acknowledge, however, that further studies are needed to investigate the actual effect of anti-TNF therapy on bacterial superinfection of inflamed colon.

Other factors may contribute to the contrasting effects of anti-TNF therapy in acute and chronic colonic lesions. In our study, the placebo group received i.p. injections of normal saline. Likewise, trials on the efficacy and safety of anti-TNF therapy in patients do not use nonspecific immunoglobulins as a treatment for the placebo group but rather use normal saline (Reinhart et al., 1996), vehicle buffer solution (Van Hensbroek et al., 1996) or human serum albumin (Stack et al., 1997; Targan et al., 1997). Using this approach makes it possible to detect adverse effects due to immunogenic reactions to the exogenous immunoglobulin. Antigen-antibody complexes are known to direct or expand immuno-inflammatory responses either by initiating the complement cascade or via interaction with Fc receptors in macrophages, neutrophils and lymphocytes. Formation and deposition of such complexes by neutralization of native TNF with the monoclonal antibody at sites of inflammation could explain the inflammatory burst observed in our experiments when treatment was started in the acute phase of colitis. However, the antibody isotype is critical in determining immune complex-mediated inflammatory responses. Hence, immune complexes that contain murine immunoglobulines of the g1 isotype do not bind to Fc receptors or complement fraction 1q, and they are inactive. In fact, previous studies with the same antibody have shown that in vivo neutralization of native TNF in Sprague-Dawley rats did not induce immune complex-associated responses, as compared with the g2a isotype (Suitters et al., 1994). It is therefore unlikely that in our experiments, anti-TNF therapy exacerbated colitis because of immune complex-mediated reactions. However, immunogenic reactions to the antibody cannot be totally excluded. The fact that rats subjected to the early-treatment schedule received five doses of the antibody, whereas rats subjected to delayed treatment received only three doses, may have influenced the outcome. Nevertheless, such adverse effects were not significant in human studies with repeated doses of the antibody (Reinhart et al., 1996) or higher doses (20 mg/kg; Targan et al., 1997).

It is interesting to note that anti-TNF therapy was associated with increased body weight gain in rats subjected to the early-treatment schedule as compared with control rats, despite a higher release of inflammatory markers and more severe lesion scores in anti-TNF-treated rats than in controls. TNF has been implicated as humoral mediator of cachexia associated with infection and inflammation (Tracey and Cerami, 1993), and the effect on body weight gain may be due to blockade of the native cytokine. In our experiment using the delayed-treatment schedule, this effect was not observed. This group of rats received three doses of the anti-TNF antibody, whereas rats subjected to early treatment received five doses of the antibody.

Anticytokine strategies have previously been shown to be useful in experimental colitis. Beneficial effects of IL-1 receptor antagonists on intestinal inflammation were observed in rabbit immune complex colitis (Cominelli et al., 1990). The current study suggests that neutralization of TNF with monoclonal antibodies is an effective tool in the treatment of chronic intestinal inflammation. As we learned from our experimental model, caution must be exercised in the treatment during the acute destructive phase, because the balance between defense against bacterial aggression and organ inflammation may be altered inappropriately and precipitate a net unfavorable outcome.