Abstract

Inflammatory bowel disease (IBD) is thought to result from a dysregulated mucosal immune response to luminal microbial antigens, with T lymphocytes mediating the colonic pathology. Infection with Helicobacter spp has been reported to cause IBD in immunodeficient mice, some of which lack T lymphocytes. To further understand the role of T cells and microbial antigens in triggering IBD, we infected interleukin (IL)-10−/−, recombinase-activating gene (Rag)1−/−, T-cell receptor (TCR)-α−/−, TCR-β−/−, and wild-type mice with Helicobacter hepaticus orHelicobacter bilis and compared the histopathological IBD phenotype. IL-10−/− mice developed severe diffuse IBD with either H. bilis or H. hepaticus, whereas Rag1−/−, TCR-α−/−, TCR-β−/−, and wild-type mice showed different susceptibilities to Helicobacter spp infection. Proinflammatory cytokine mRNA expression was increased in the colons ofHelicobacter-infected IL-10−/− and TCR-α−/− mice with IBD. These results confirm and extend the role of Helicobacter as a useful tool for investigating microbial-induced IBD and show the importance, but not strict dependence, of T cells in the development of bacterial-induced IBD.

  • Helicobacter bilis
  • Helicobacter hepaticus
  • colitis
  • proinflammatory cytokines

the precise etiologyof the human idiopathic inflammatory bowel disorders ulcerative colitis and Crohn's disease is unknown. Proposed factors include infection with bacterial pathogens, disruption of the intestinal mucosal barrier, dysregulated immune responses to luminal antigens, and genetic susceptibility (2, 35), but their exact roles have not been defined. Within the past decade, numerous mouse models have emerged as tools to investigate mechanisms of initiating and perpetuating inflammatory bowel disease (IBD). Spontaneous intestinal inflammation has been observed in genetically altered mice with various immunologic defects. These include interleukin (IL)-2−/−, IL-10−/−, major histocompatibility complex (MHC) class II−/−, T-cell receptor (TCR) mutants, IL-7 transgenics, and Gαi-2 −/− mice (18,25, 32, 33, 44). Although the underlying genetic defect that confers susceptibility to IBD varies considerably in these mouse models, a common and central feature is that the development of IBD is dependent on the presence of intestinal luminal flora. For example, IL-2−/− mice maintained in a conventional facility develop IBD, but when rederived into specific pathogen-free (SPF) or germ-free conditions, intestinal lesions are delayed or prevented, respectively (33). Similarly, IL-10−/− mice spontaneously develop IBD when housed under conventional conditions (18), but there is no evidence of colitis when animals are maintained under germ-free conditions (37). In the CD4+ CD45RBhi adoptive transfer model, mice with reduced intestinal flora show less severe wasting disease (1), whereas mice treated with bacitracin and streptomycin are rescued from weight loss (27), again stressing the role that luminal bacteria play in initiating IBD. The adoptive transfer model has also shown that the colitis is mediated by a CD45RBhi subset of CD4+ T cells, (20, 27,28, 30), highlighting an important role for T cells in the disease process. Few studies have analyzed, in a controlled manner, the relative contribution of bacterial flora and T cell dysregulation in the development of IBD (4).

Infection with murine Helicobacter species has been implicated as a potential cause of “spontaneous” gastrointestinal inflammation mimicking IBD in various mouse strains raised in conventional facilities. Indeed, natural and experimental infections of certain strains of laboratory rodents with particular species ofHelicobacter have been shown to cause IBD. For example, immunodeficient mice with natural infections of H. hepaticusexhibit a proliferative typhlocolitis and proctitis (21, 42,43), and experimental inoculation of severe combined immunodeficiency mice with H. bilis has been reported to result in typhlitis and colitis (14, 39). Recently identified Helicobacter species have also been associated with IBD in IL-10−/− (11) and severe combined immunodeficiency mice (15). Although these data by no means suggest that Helicobacter is the only organism capable of initiating murine IBD, they do underscore the importance of this widely endemic bacteria in inducing gastrointestinal inflammation. Therefore, Helicobacter spp may be an excellent model organism with which to investigate the complex role of luminal bacteria in initiating experimental IBD.

Although the above-cited reports have solidified a link between experimental IBD and Helicobacter infection, important questions remain. Given the observation that variousHelicobacter species can cause disease, what is the relative efficiency of the different species to induce IBD? Is there identical or disparate pathology? In addition, given the importance of T lymphocytes in experimental IBD, how can one interpret the occurrence of Helicobacter-induced IBD in mice that lack T and B cells (41)? Therefore, the aims of this study were to determine1) whether H. hepaticus or H. biliswas more potent in producing bowel pathology, 2) whether the IBD induced by Helicobacter spp differs phenotypically in mice with a dysregulated cytokine network compared with mice deficient in specific immune cell subsets, and 3) whether proinflammatory cytokines are regulated during the disease process. To this end, a histopathological scoring system allowed us to quantitate the severity and regional extent of disease, and MHC class II expression and cytokine production were analyzed in colonic tissue. Here we report that IL-10−/− mice develop severe IBD after H. bilis or H. hepaticus infection. Furthermore, the IBD that develops is associated with increased MHC class II expression in the intestinal epithelium. None of the C57BL/10J or C57BL/6J immunocompetent mice infected with H. hepaticusor H. bilis showed any evidence of IBD. We also report that mice with absent [recombinase-activating gene (Rag)1−/−mice] or dramatically altered T cell populations [T cell receptor (TCR)-α−/− mice] show mild intestinal inflammation after Helicobacter infection. We analyzed cytokine mRNA expression associated with disease and show that T helper (Th)1-type cytokines are upregulated in the colons ofHelicobacter-infected mouse strains with IBD, regardless of whether the model has been previously associated with a Th1 or Th2 cytokine profile.

MATERIALS AND METHODS

Animals.

Three- to seven-week-old female C57BL/10J, C57BL/6J, IL-10−/− (C57BL/10J), IL-10−/− (C57BL/6J), Rag1−/− (C57BL/6J), TCR-α−/− (C57BL/6J), and TCR-β−/− (C57BL/6J) mice were obtained from Jackson Laboratories (Bar Harbor, ME). All mice were certified free ofHelicobacter spp by the vendor and retested in our animal facilities. Cohorts of mice were housed at the University of Washington (UW) or at Immunex Corporation (IMNX). Animals were housed in a SPF room in polycarbonate microisolator cages containing Bed-O'cob (The Andersons, Maumee, OH) and a nestlet. Mice were fed irradiated Picolab rodent diet 20 (PMI Nutrition International, Brentwood, MO) and autoclaved, acidified water. All supplies entering the animal rooms were autoclaved, and the room was maintained at 70–74°F, 45–55% humidity, with 28 air changes per hour and a 12:12-h light-dark cycle. To prevent cross contamination, uninfected and infected mice were housed in separate cubicles or separate rooms. At UW, mice were housed in separate cubicles, depending on infection status, within the same room, and cages were changed in “uninfected” or “infected” laminar flow changing stations. At IMNX, uninfected and infected animals were housed in separate rooms, each containing its own laminar flow changing station. Sentinel mice at the UW were tested quarterly for endo- and ectoparasites, mouse hepatitis virus, mouse parvovirus, and rotavirus and annually forMycoplasma pulmonis, pneumonia virus of mice, reovirus-3, Sendai virus, and Theiler's murine encephalomyelitis virus. Also, colon samples were screened quarterly for Citrobacter rodentium, non-lactose-fermenting Escherichia coli,Salmonella spp, Klebsiella spp, andClostridium spp (Phoenix Laboratories, Seattle, WA). At IMNX, random sentinel mice from all rooms were tested weekly for mouse hepatitis virus and monthly for 14 mouse viruses (Charles River Laboratories, Wilmington, MA). Biannually, sentinel mice from each room were sent to Charles River Laboratories, necropsied, and screened for several important murine viral, bacteriological, and parasitic pathogens and for Helicobacter spp. The UW Institutional Animal Care and Use Committee and the IMNX Animal Use and Care Committee approved all animal procedures.

Experimental design.

Before inoculation, mice were determined to be negative forHelicobacter spp by fecal PCR (see Clinical and histopathological findings associated with Helicobacter spp infection in IL-10−/− mice). In six separate experiments, 5–15 mice were given H. hepaticus orH. bilis, and 3–15 mice were given broth alone. Initially, fecal Helicobacter culture and species-specific fecal PCR were done on pooled cage samples taken every 2 wk until the end of the study. Later experiments used fecal PCR as the sole means to determine Helicobacter infection status. Between 3 and 36 wk postinfection (PI), uninfected and infected mice were euthanized with CO2 and necropsied. Tissues were taken for histopathology, immunohistochemistry, or cytokine analysis or combinations of the three. Mice were weighed weekly and monitored for weight loss, dehydration, and diarrhea. Additionally, fecal samples from both infected and uninfected mice were periodically tested by PCR for cross contamination with another Helicobacter spp in infected mice and to confirm the absence of Helicobacter infection in uninfected animals.

Inoculation.

Mice were inoculated by oral gavage with ∼2 × 107 H. hepaticus or H. bilis colony-forming units (CFU) in 0.2 ml of Brucella broth or withBrucella broth alone three times within a 9-day period. H. hepaticus was obtained from the American Type Culture Collection (ATCC 51448), and H. bilis was kindly provided by Dr. Lela Riley (University of Missouri, Columbia, MO). Briefly, organisms were streaked onto Brucella blood agar plates (Dept. of Microbiology, Media Laboratory, UW, Seattle, WA), grown in microaerobic conditions (90% N2, 5% H2, and 5% CO2) in vented jars (Oxoid), and kept at 37°C. Bacteria were harvested and inoculated into flasks containing 150 ml of Brucella broth supplemented with 5% FCS (Sigma Chemical, St. Louis, MO). The flasks were placed on a continuous shaker and incubated for 24–48 h at 37°C in microaerobic conditions. The organisms were centrifuged at 10,000 rpm at 4°C for 20 min. The resultant pellet was examined by Gram's stain and phase microscopy for purity, morphology, and motility (catalase, urease, and oxidase positive). The pellet was resuspended inBrucella broth, and optical density was adjusted to 1.0, as measured at 600 nm, for an estimated 108 CFU/ml (2 × 107 CFU/0.2 ml).

Helicobacter PCR and fecal culture.

Feces were analyzed for Helicobacter spp, H. hepaticus, and H. bilis, as described previously (12, 31, 38), with slight modifications. For generic Helicobacter PCR, 5 μl of unquantitated fecal DNA were used as a template, and for H. hepaticus and H. bilis PCR, 10 μl of unquantitated fecal DNA were used as a template for all samples. Primers, their sequences, and the PCR conditions used are summarized in Table1. To correlate fecal culture results and PCR status, fecal pellets were vortexed in PBS and incubated at room temperature for 30 min. Fecal slurries were then filtered through an 0.8-μm filter, and the filtrate was streaked onto Brucellablood agar plates containing trimethoprim, vancomycin, and polymyxin and incubated as described in Inoculation. Cultures were maintained for 2–3 wk before determination of negative growth was made.

View this table:
Table 1.

Conditions for PCR reactions

Pathology.

The colon was fixed in 10% buffered formalin with a “Swiss roll” technique (26). The liver, cecum, rectum, and mesenteric lymph nodes were also fixed in formalin. Fixed tissues were routinely processed, embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin and eosin. Tissue sections were coded to blind the pathologist (K. Waggie) to the strain and infection status of the animal. The cecum, proximal colon, middle colon, distal colon, and rectum from each mouse were scored on severity of mucosal epithelial changes, degree of inflammation, and extent of pathology (Table2). The segment score was derived by summing the severity scores [segment score = mucosal score + inflammation score + extent of segment affected in any manner (extent 1) + extent of segment affected at level 3 or 4 (extent 2); maximum segment score was 15]. The total score for each mouse was derived by summing the scores from the individual segments (maximum total score was 75). Subsequently, the mean segment and total scores were derived for each treatment group.

View this table:
Table 2.

Scoring system for histological evaluation of IBD lesions

MHC class II immunohistochemistry.

MHC class II expression was analyzed in H. bilis- andH. hepaticus-infected and uninfected IL-10−/−, Rag1−/−, and C57BL/10J mice. Briefly, 3- to 4-mm sections of proximal colon and distal colon were snap-frozen in optimal cutting temperature compound, and detection of MHC II expression was performed on 8-μm frozen sections. Sections were fixed with acetone and then incubated with a 1:200 dilution of biotinylated mouse anti-mouse I-Ab monoclonal antibody (PharMingen, San Diego, CA). Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 20 min followed by detection of signal with the biotin-streptavidin-peroxidase VECTASTAIN ABC Kit (Vector Laboratories, Burlingame, CA). The chromagen used was 3,3′-diaminobenzidine. Intensity of MHC II staining was based on the following scoring system: 0 = no expression, 1 = minimal expression, 2 = moderate expression, and 3 = maximal expression. As a negative control, colonic tissue from MHC class II−/− mice was stained for class II expression. All slides were scored independently by two investigators blinded to the strain and infection status of the animal.

Colonic RNA extraction and cytokine analysis by RT-PCR and RNase protection assay.

Qualitative analysis of cytokine mRNA expression in colons fromH. bilis-infected or uninfected IL-10−/−, Rag1−/−, C57BL/6J, TCR-α−/−, and TCR-β−/− mice was performed between 6 and 17.5 wk PI. Briefly, colonic fecal contents were emptied, and two to four colons were pooled and homogenized in guanidinium isothiocyanate buffer [4.5 M guanidinium isothiocyanate, 50 mM sodium citrate, and 0.5% (wt/vol) sodium sarcosyl] containing 2% 2-mercaptoethanol (Life Technologies, Rockville, MD). RNA was isolated over a cesium chloride cushion and subjected to phenol chloroform extraction as previously described (34). cDNA was prepared from 5 μg of isolated RNA with SuperScript II reverse transcriptase (Life Technologies, Rockville, MD) in the presence of 0.3 μg of random primers, 5× RT buffer, 0.1 M dithiothreitol, 10 mM deoxynucleotide triphosphate, and water. Interferon-γ and IL-4 messages were analyzed by RT-PCR with the mouse TH1/TH2 Switch Cytokine kit (BioSource International, Camarillo, CA). The following profile was used for amplification: 96°C for 1 min followed by 2 cycles of 96°C for 1 min and 59°C for 4 min, then 30 cycles of 94°C for 1 min and 59°C for 2.5 min, and a final cycle of 70°C for 10 min. As a negative control, cDNA was replaced with water, and the sample was subjected to the same PCR conditions. The PCR product and 1 μl of loading buffer were loaded on a 1.0-mm, 15-well, 10% TBE precast gel (Novex, San Diego, CA) and run at 126 V. The gel was stained with ethidium bromide and photographed. For RNase protection assay analysis, H. bilis-induced IL-1α, IL-1β, and IL-1 receptor antagonist (IL-1RA) levels were measured at selected time points in IL-10−/−, C57BL/6J, and Rag1−/− mice with the RiboQuant Multiprobe RNase Protection Assay kit mCK-2b (PharMingen). The assay was run according to the manufacturer's instructions, and the values obtained for each sample were normalized to glyceraldehyde-3-phosphate dehydrogenase.

Statistical analyses.

Differences between mean total pathology scores in IL-10−/− mice infected with either H. bilis orH. hepaticus were evaluated by two-way analysis of variance (Sigma Stat 2.0; Jandel, San Rafael, CA). Statistical significance was set at a P value <0.05.

RESULTS

Colonization with H. hepaticus or H. bilis.

Mice inoculated with H. hepaticus or H. bilisbecame fecal PCR positive 2–4 wk postinoculation and remained persistently positive for the duration of these experiments (12–36 wk PI). Consistent with other studies, PCR was a more sensitive and specific means to detect Helicobacter spp infection compared with culture (data not shown) (22). Therefore, PCR was the sole method used to confirm infection. Uninfected animals that received broth alone remained Helicobacter fecal PCR negative throughout these studies, as determined with Helicobactergenus primers. Periodic fecal PCR for the alternateHelicobacter spp with species-specific primers revealed that at no time were the mice cross contaminated.

Clinical and histopathological findings associated with Helicobacter spp infection in IL-10−/− mice.

An initial study (study 1) was performed to determine whether different species of Helicobacter induced similar or dissimilar forms of IBD after infection with either H. bilisor H. hepaticus in a well-characterized IBD model, the IL-10−/− mouse. Diarrhea was evident in theHelicobacter spp-infected IL-10−/− mice, but there were notable differences between the clinical course of H. bilis and H. hepaticus-infected animals. The earliest time to development of diarrhea differed depending on whether IL-10−/− mice were infected with H. hepaticus(1.5 wk) or H. bilis (3 wk). Despite the more rapid induction of diarrhea with H. hepaticus, weight gain inH. bilis-infected IL-10−/− mice was less than that seen in uninfected controls (Fig.1 A), whereas H. hepaticus-infected IL-10−/− mice showed normal growth curves (data not shown). It is important to note that there were no significant weight change differences in H. bilis-infected wild-type mice (Fig. 1 B). Diarrhea was not observed in Helicobacter spp-infected wild-type mice or in uninfected wild-type and IL-10−/− mice.

Fig. 1.

Helicobacter bilis infection induces altered weight gain in interleukin (IL)-10−/− mice. Change in body weight over time is expressed as percentage of original body weight before Helicobacter infection. Data are means ± SE of 9 mice/infected group and 3 mice/uninfected group unless otherwise stated. A: growth curve was altered in H. bilis-infected IL-10−/− mice. B: growth curves for Helicobacter-infected and uninfected wild-type mice were similar. aMice remaining after histopathological analysis at 3, 7, and 12 wk postinfection (PI). bOne animal in group was euthanized because of severe weight loss. Results are representative of 3 experiments.

Both H. bilis and H. hepaticus induced IBD in IL-10−/− mice (Fig.2 B and data not shown), whereas no IBD was observed in uninfected IL-10−/− mice (Fig. 2 A) orHelicobacter-infected and uninfected wild-type C57BL/10J mice (Fig. 2 C). Generally, both H. hepaticus andH. bilis produced chronic proliferative typhlocolitis and proctitis beginning at 3 wk PI. The mean total pathology scores are summarized in Table 3. AlthoughH. hepaticus produced its most severe lesions in the cecum, proximal colon, and rectum, H. bilis induced significant disease in all portions of the large bowel.

Fig. 2.

IL-10−/− mice infected with H. bilis orH. hepaticus developed severe inflammatory bowel disease (IBD) beginning 3 wk PI. However, H. bilis-induced disease may be more severe than that induced by H. hepaticus in middle and distal colon of IL-10−/− mice. A: distal (bottom) and proximal (top) colon of uninfected IL-10−/− mouse 12 wk after treatment.B: distal colon of IL-10−/− mouse infected with H. bilis 12 wk PI. Note severe mucosal inflammation and hyperplasia with crypt herniation into the submucosa. Similarly, at this time point, the distal colon of H. hepaticus-infected IL-10−/− mice showed moderate mucosal inflammation and epithelial hyperplasia. C: distal colon of C57BL/10J mouse infected with H. bilis 12 wk PI. There is no evidence of inflammation. Segments stained with hematoxylin and eosin; original magnification, ×10.

View this table:
Table 3.

Pathology scores in IL-10−/− mice infected with H. hepaticus or H. bilis

With H. hepaticus infection, the chronic colitis was characterized by varying degrees of mucosal thickening resulting from epithelial hyperplasia, with elongation and occasional branching of crypts (data not shown). Crypt penetration into the submucosa was observed in the rectum of two of three mice 12 wk PI (data not shown). An inflammatory cell infiltrate was present that consisted primarily of lymphocytes and macrophages with varying numbers of plasma cells and neutrophils. The infiltrate was most often confined to the lamina propria and submucosa but was occasionally transmural. Crypt abscesses and mucosal erosions were infrequently present. The typhlitis was characterized by severe mucosal hyperplasia and scattered erosions, goblet cell loss, and a predominantly lymphoplasmacytic infiltrate (Fig. 3, A and B).

Fig. 3.

IL-10−/− mice infected with H. hepaticus or H. bilis developed proliferative typhlitis and proctitis beginning 3 wk PI. A: cecum of uninfected IL-10−/− mouse 12 wk after treatment. B: cecum of IL-10−/− mouse infected with H. hepaticus12 wk PI. Note severe mucosal hyperplasia, inflammation, and focal erosion. Lumens of occasional crypts are dilated and contain debris. Ceca of H. bilis-infected mice at this time point had a similar appearance. C: rectum of uninfected IL-10−/− mouse 12 wk after treatment. D: rectum of IL-10−/− infected with H. bilis 12 wk PI. Note crypt branching and herniation. Similar lesions were present at 3 and 7 wk PI. Segments stained with hematoxylin and eosin; original magnification, ×10.

Microscopic changes in H. bilis-infected IL-10−/− mice were similar to those observed in H. hepaticus animals but tended to be more severe, especially in the middle and distal colon 7 and 12 wk PI (Table 3). Specifically, severe dilation of the submucosal lymphatics and submucosal edema were found at 3 and 7 wk PI. Transmural inflammation, crypt abscesses, and mucosal erosions were frequently observed at all time points, which differed from findings with H. hepaticus infection. Crypt penetration into the submucosa, observed at 12 wk PI in two of three mice, extended into the distal colon of H. bilis-infected mice (Fig.2 B), and the rectum was similarly affected (Fig. 3,C and D).

H. hepaticus- and H. bilis-induced IBD in IL-10−/− mice was associated with a marked increase in MHC class II expression in both the proximal and distal colon. The staining was evident on both epithelial cells and cells within the lamina propria (Fig.4 A), with scores ranging from 2 to 3 (Table 4). Uninfected IL-10−/− mice without colitis showed a minimal to moderate increase in MHC class II expression (Fig. 4 B; Table4). H. hepaticus- and H. bilis-infected and uninfected wild-type mice had minimal MHC class II expression that was primarily localized to the surface colonic epithelial cells (Fig.4 C). As a scoring reference, colonic tissue from MHC II−/− mice was stained and showed no MHC class II expression (Fig. 4 D).

Fig. 4.

Major histocompatibility complex (MHC) class II expression was elevated in both the proximal and distal colon ofH. hepaticus- and H. bilis-infected IL-10−/− mice. Sections were stained with 3,3′-diaminobenzidine and counterstained with methyl green. Shown are samples of distal colon (A–C) obtained from H. bilis-infected and uninfected mice 12 wk after treatment. Similar results were seen in the proximal colon. A: IL-10−/− mouse infected with H. bilis 12 wk PI. Note intense MHC class II staining of lamina propria and epithelial cells. Both surface and crypt epithelial cells are stained.B: uninfected IL-10−/− mouse 12 wk after treatment. Note decreased intensity of MHC class II staining, present primarily on surface epithelial cells and a few lamina propria cells.C: C57BL/10J mouse infected with H. bilis 12 wk PI. Decreased intensity of MHC class II staining is present primarily on surface epithelial cells and random lamina propria cells.D: colonic tissue section from an MHC class II−/− mouse. No expression of MHC class II can be seen. Original magnification, ×20 for all. Data are representative of 2 or 3 mice/infected group.

View this table:
Table 4.

MHC class II expression in proximal and distal colon of Helicobacter spp-infected and -uninfected mice

Our initial observations suggested that H. bilis infection in IL-10−/− mice may produce more extensive and severe IBD than H. hepaticus (Table 3). Therefore, in study 2, a larger group of IL-10−/− mice were infected with H. bilis or H. hepaticus to determine whether this difference was statistically significant. It was necessary to euthanize all mice at 4 wk PI because of weight loss observed in the IL-10−/− mice infected with H. bilis. This study, consisting of mixed ages of mice, showed no statistically significant differences between H. bilis- and H. hepaticus-induced IBD pathology (Fig.5, A and B). Mean total pathology scores were 43 and 44 for IL-10−/− mice infected with H. bilis or H. hepaticus, respectively (P = 0.605; Table 3). A low level of mild spontaneous IBD (mean total pathology score of 2) was observed in 4 of 15 uninfected IL-10−/− mice (data not shown), whereas all others remained disease free (Fig. 5 C). Interestingly, when mean total pathology scores from mice that were older (6–7 wk of age; data not shown) at the time of H. bilis or H. hepaticus infection were analyzed and compared, we found a trend towards significance (P = 0.076). Based on this observation, as well as the altered growth curves seen in H. bilis-infected compared with H. hepaticus-infected IL-10−/− mice, it is our impression that H. bilis-induced IBD in IL-10−/− mice tends to be more severe than H. hepaticus-induced disease, especially in mice that are older at the time of Helicobacter infection. Clearly, the data support the hypothesis that both species ofHelicobacter can induce severe IBD in this model.

Fig. 5.

IL-10−/− mice infected with H. bilis orH. hepaticus developed IBD but the disease was not significantly different in severity. A: distal colon of IL-10−/− mouse infected with H. bilis 4 wk PI. Note severe epithelial hyperplasia with a mixed inflammatory cell infiltrate into the lamina propria, extending somewhat into the submucosa. B: distal colon of IL-10−/− mouse infected with H. hepaticus 4 wk PI. Disease is similar to that seen in IL-10−/− mice infected with H. bilis, with severe epithelial hyperplasia and mucosal inflammation. However, rarely did the inflammatory cells infiltrate the submucosa. C: distal colon of uninfected IL-10−/− mouse 4 wk after treatment. This uninfected mouse shows no evidence of inflammation, although 4 of 15 uninfected IL-10−/− mice did show very mild spontaneous IBD. Segments stained with hematoxylin and eosin; original magnification, ×10.

Susceptibility of Rag1−/− mice to Helicobacter spp infection.

Rag1−/− mice infected with either H. hepaticusor H. bilis did not develop the clinical or typical histological features of IBD that were seen in the IL-10−/− mice in studies extending to 12 wk PI. Similarly, uninfected Rag1−/− mice showed no evidence of intestinal inflammation. However, one of three mice from the H. hepaticus and H. bilis groups developed an acute, focal, mild inflammation of the colon and/or cecum at the 3 or 12 wk time points, respectively. Although Helicobacterspp-infected Rag1−/− mice did not develop IBD, MHC class II expression was moderately increased (score of 2) in the proximal colon of H. hepaticus-infected Rag1−/− mice at 3 and 7 wk PI (Table 4) compared with uninfected controls. Similarly,H. bilis-infected Rag1−/− mice showed mild to moderate (score of 1.3–2.0) MHC class II staining in the proximal and distal colon at 3 and 7 wk PI. At all time points examined, uninfected Rag1−/− mice showed mild MHC class II staining intensity of the proximal and distal colon (score of 1.0–1.5; Table 4).

To determine whether long-term infection with Helicobactercould initiate IBD in Rag1−/− mice, both wild-type and Rag1−/− mice were infected with either H. hepaticus or H. bilis and their progress was followed for 30–36 wk. Significant morbidity occurred in our Rag1−/− mice as a result of Pneumocystis carinii infection and precipitated the end of the study at these time points. Approximately 9 mo after Helicobacterinfection, there were no clinical indications of IBD (diarrhea, rectal prolapse) in Rag1−/− or wild-type mice chronically infected with Helicobacter. This is in marked contrast to the rapid onset of disease in similarly infected IL-10−/−mice (1.5–3 wk PI). Additionally, no histological evidence of IBD was found in Helicobacter-infected and uninfected wild-type mice or uninfected Rag1−/− mice (Fig.6, A and C; data not shown). However, two of five Rag1−/− mice in both theH. hepaticus and H. bilis groups developed mild focal mucosal inflammation characterized by minimal to mild epithelial hyperplasia and low numbers of a predominantly neutrophilic infiltrate (Table 5; Fig. 6, B andD). Interestingly, lesions in H. bilis-infected Rag1−/− mice were found solely in the proximal colon, whereas those of H. hepaticus-infected Rag1−/−mice were found in the cecum (Fig. 6, B and D). It is important to note that typical IBD lesions such as those seen in similarly infected IL-10−/− mice did not develop inHelicobacter-infected Rag1−/− mice.

Fig. 6.

Recombinase-activating gene (Rag1−/−) mice infected long-term with either H. bilis or H. hepaticusdeveloped an acute mild form of IBD that differed from the severe chronic form of IBD seen in IL-10−/− mice infected withHelicobacter. A: proximal colon of uninfected Rag1−/− mouse 32 wk after treatment. B: proximal colon of Rag1−/− mouse infected with H. bilis 33 wk PI. Note the mild accumulations of (predominantly) neutrophils in the lamina propria and minimal epithelial hyperplasia.C: cecum of uninfected Rag1−/− mouse 33 wk after treatment. D: cecum of Rag1−/− mouse infected with H. hepaticus 30 wk PI. There are mild accumulations of inflammatory cells in the lamina propria, and there is minimal epithelial hyperplasia. Segments stained with hematoxylin and eosin; original magnification, ×20.

View this table:
Table 5.

Pathology scores in mice with absent or altered T cell subsets infected with H. hepaticus or H. bilis

Susceptibility of T cell-deficient mice to H. bilis infection.

Because we were unable to induce significant IBD withHelicobacter spp in a T cell-deficient mouse (Rag1−/−) but did obtain severe intestinal inflammation with Helicobacter in a model having T cells (IL-10−/−), we wanted to determine the importance of T cell populations in the initiation of IBD. Therefore, we infected wild-type, TCR-α−/−, and TCR-β−/− mice with H. bilis in the same manner previously described. We chose H. bilis for these experiments based on our clinical experience with this organism in eliciting profound rapid inflammation in the IL-10−/− mouse. Two separate experiments were performed, extending to 17.5 (study 1) and 29 (study 2) wk PI. Helicobacter-infected TCR-β−/− mice were only included in the initial 17.5-wk study. Of these two mutant strains, we were only able to induce IBD in TCR-α−/− mice in the time frame of these studies. The onset of disease in the TCR-α−/− mice was more delayed than that historically seen in the IL-10−/− mice infected under a similar protocol. Clinically, diarrhea was evident in the TCR-α−/− mice in these two studies beginning 10–12.5 wk PI. However, diarrhea was not evident in allHelicobacter-infected TCR-α−/− mice. This was much later than what we had observed in H. bilis-infected IL-10−/− mice (3 wk PI). Furthermore, growth curves for H. bilis-infected and uninfected TCR-α−/− mice were similar (data not shown).

At 17.5 wk PI, three of five H. bilis-infected TCR-α−/− mice had lesions in the cecum, colon, and rectum that were similar to but milder than lesions observed in the corresponding H. bilis-infected IL-10−/− mice (see Clinical and histopathological findings associated with Helicobacter spp infection in IL-10−/− mice). Pathology scores are summarized in Table 5. Although lesions in the TCR-α−/− mice were found in all areas of the large bowel, they were segmental within each area. Generally, the lesions were characterized by moderate crypt epithelial hyperplasia with inflammatory cell infiltrates limited to the mucosa and submucosa (Fig.7). One of four uninfected TCR-α−/− and one of four uninfected TCR-β−/− mice had mild colitis limited to a focal area in the middle colon. These focal lesions were similar to but milder than those in the H. bilis-infected TCR-α−/−mice and were probably the result of the non-Helicobacter-associated colitis reported to occur spontaneously in these strains (25).

Fig. 7.

T-cell receptor (TCR)-α−/− mice infected with H. bilis developed IBD that was milder than that seen in Helicobacter-infected IL-10−/− mice.A: distal colon of uninfected TCR-α−/− mouse 17.5 wk after treatment. B: distal colon of TCR-α−/− infected with H. bilis 17.5 wk PI. There is crypt hyperplasia and moderate mucosal inflammation. Segments mouse stained with hematoxylin and eosin; original magnification, ×10.

Study 2 was initiated to correlate histopathology and cytokine regulation during Helicobacter-induced IBD in TCR-α−/− mice (see Proinflammatory cytokine production in the colon of H. bilis-infected mice for cytokine results). In mice evaluated histologically at 29 wk PI, four of eightH. bilis-infected TCR-α−/− mice showed IBD lesions similar to those seen in our initial study (Table 5). Interestingly, the IBD in these diseased TCR-α−/− mice was completely subclinical, with no evidence of diarrhea or weight loss (data not shown). Again, we saw evidence of mild, focal, spontaneous IBD in uninfected TCR-α−/− mice (2 of 8).

Proinflammatory cytokine production in the colon of H. bilis-infected mice.

Proinflammatory cytokine expression correlated with the presence of intestinal inflammation in IL-10−/− and TCR-α−/− mice. With RT-PCR, IL-10−/− mice infected with H. bilis showed elevated levels of interferon-γ mRNA in the colon compared with uninfected IL-10−/− mice at all time points analyzed (Fig.8 A). Our results agree with previous studies that reported interferon-γ release on stimulation of mesenteric lymph node cells from H. hepaticus-infected IL-10−/− mice (19). Surprisingly, elevated levels of interferon-γ message were also seen in colonic tissue harvested from diseased H. bilis-infected TCR-α−/− mice compared with uninfected TCR-α−/− mice (Fig. 8 B). Although this model is recognized as a predominantly IL-4-driven model of IBD, we did not see elevated levels of IL-4 message in either H. bilis-infected or uninfected TCR-α−/− mice. In contrast, colonic tissue samples from H. bilis-infected Rag1−/−, TCR-β−/−, and wild-type mice (Fig. 8 A and data not shown) showed greatly reduced amounts of interferon-γ message compared withH. bilis-infected IL-10−/− mice with IBD. Using an RNase protection assay, we found that H. bilisupregulated expression of IL-1α (∼3-fold), IL-1β (∼15- to 26-fold), and IL-1RA (∼3-fold) in IL-10−/−mice with IBD (Fig. 9). As an indicator of dysregulation of proinflammatory cytokine activity, the IL-1RA-to-IL-1β ratio was determined for infected versus uninfected IL-10−/− mice. H. bilis-infected IL-10−/− mice had a lower ratio (0.626) than uninfected IL-10−/− mice (5.921). Expression of IL-1α, IL-1β, and IL-1RA in infected and uninfected wild-type and Rag1−/− mice was not significantly different (data not shown).

Fig. 8.

H. bilis induced increased mRNA expression of interferon (IFN)-γ in the colon of IL-10−/− and TCR-α−/− mice with IBD analyzed by RT-PCR. A: low levels of IFN-γ are seen in IL-10−/− uninfected, H. bilis-infected C57BL/6J, and H. bilis-infected and uninfected Rag1−/− mice. Additionally, low levels of IL-4 were seen in both H. bilis-infected and uninfected Rag1−/− mice. RNA samples (3–5 sets) were prepared from each infected and uninfected strain 6–17 wk PI. For each sample, the PCR reaction was performed 1–3 times. IL-10−/− mice were 6 wk PI, whereas the C57BL/6J and Rag1−/− mice were 17 wk PI. B: IFN-γ levels were significantly elevated in diseased H. bilis-infected TCR-α−/− mice compared with uninfected controls. Note the absence of IL-4 message in diseased TCR-α−/− mice. RNA samples were prepared from infected and uninfected TCR-α−/− mice 13 wk PI, and the PCR reaction was performed twice on each sample. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Fig. 9.

H. bilis induced increased mRNA expression of IL-1α (B), IL-1β (C), and IL-1 receptor antagonist (RA; D) in the colon of IL-10−/−mice as detected by RNase protection assay. No such elevations were seen in H. bilis-infected C57BL/6J mice (A–D). IL-10−/− and wild-type mice were 9 and 7 wk post-H. bilis infection, respectively. All values are normalized to GAPDH.

DISCUSSION

Previous studies (9, 18, 29, 33, 37) have shown that luminal bacteria are an important factor contributing to the development of IBD in mice and humans. Several laboratories have implicated infection with Helicobacter species in the development of mucosal inflammation and IBD in various mouse models (4, 6, 13, 14, 39). By comparing the effects of the less widely studied H. bilis with H. hepaticus in the IL-10−/− mouse, a well-established model of Th1-driven IBD (3, 7), we present data that both confirm and extend these results that suggest a role for Helicobacter in murine IBD. In addition, because of emerging literature suggesting that Helicobacter-induced IBD can occur in profoundly immunodeficient mice, we attempted to address the role of T cells in the pathogenesis of Helicobacter-induced IBD. Using the combination of a detailed histopathological scoring system, MHC class II immunohistochemistry, and cytokine analysis, we demonstrate herein that both H. bilis and H. hepaticus consistently produce severe IBD in IL-10−/− mice, whereas the ability of Helicobacter to induce intestinal inflammation in mice with absent T cells (Rag1−/− and TCR-β−/−mice) or significantly altered T cell subsets (TCR-α−/−mice) is attenuated. Furthermore, increases in MHC class II expression and colonic proinflammatory cytokine levels correlated with the presence of IBD in Helicobacter-infected IL-10−/− mice. Surprisingly, despite the fact that the TCR-α−/− mouse has been described as a “Th2” model of murine IBD (23, 24, 40), Helicobacterinfection of these mice resulted in colitis with elevated levels of interferon-γ without significant induction of IL-4. Taken together, the current studies strongly support the role of murineHelicobacter in the development of colitis and highlight the importance of, but not strict dependence on, T cells in the development of bacterially driven mucosal inflammation in mouse strains genetically susceptible to the development of IBD.

Numerous studies have investigated Helicobacter spp as model organisms in microbial-induced IBD. It is fair to say that the observations made in these studies have varied widely. For example, at one end of the spectrum, Kullberg et al. (19) have observed chronic colitis in SPF-reared IL-10−/− mice after infection with H. hepaticus. In an independent series of experiments, Chin et al. (6) described the ability ofH. hepaticus to induce significant mucosal inflammation in the cecum and colon of TCR-αβ mutant mice. In marked contrast, recent studies have reported that Helicobacter does not induce IBD in germ-free IL-10−/− mice or potentiate IBD in germ-free IL-10−/− mice that have been reconstituted with SPF flora (8, 37). It is important to note that the published studies have used different protocols in varied mouse facilities and, in so doing, have not resulted in a consensus regarding the role of Helicobacter as a potential pathogen in conventional mouse facilities or its utility as a model organism to study IBD. How do the data presented in this report fit into the context of these previous studies? Clearly, we have shown thatHelicobacter can be an important determinant in the induction of murine colitis in an SPF mouse facility. In our experience, IBD-prone strains such as IL-10−/− and TCR-α−/− mice rarely develop mucosal inflammation in aHelicobacter-free, SPF environment, whereas monoinfection with H. bilis or H. hepaticus induces a significantly greater incidence and severity of disease in these mice. A further conclusion from these studies is that the disease induced byHelicobacter varied significantly among the different IBD models. Foremost, both H. bilis and H. hepaticusreliably induced severe chronic IBD in the IL-10−/−mouse model of IBD. Disease was evident clinically 1.5–3 wk PI, and inflammation was evident from cecum to rectum. Intermediate in disease severity was the intestinal inflammation found in our TCR-α−/− mice infected with H. bilis. Lesions were similar to but milder than the inflammation seen in similarly infected IL-10−/− mice. Additionally, disease incidence in the Helicobacter-infected TCR-α−/− mice was less than that seen inHelicobacter-infected IL-10−/− mice. In contrast, minimal mucosal inflammation was found in Rag1−/− mice, which do not have T or B cells. This inflammation, when present, was characterized by minimal to mild focal epithelial hyperplasia and had a primarily neutrophilic infiltrate. Our observations support those of others that show Helicobactercan induce fulminant disease in certain immunodeficient strains of mice.

As mentioned, there are important differences among all these studies that may explain the varied results. In our opinion, the major variables include the unique bacterial flora present in geographically different mouse colonies and the contribution of genetic background to IBD susceptibility. Regarding the importance of mouse strain variation in the development of IBD, Berg et al. (3) have shown that IL-10−/− mice on a 129/SvEv or BALB/c background are inherently more susceptible to spontaneous IBD compared with IL-10−/− mice on a C57BL/6J background. Strain differences may explain why in our studies withHelicobacter-infected TCR-α−/− and TCR-β−/− mice, we saw less severe disease than that reported by Chin et al. (6). Specifically, our mice were on a C57BL/6J background, whereas those of Chin et al. were on a C57BL/6 × 129-Ola background. Moreover, it is nearly impossible to characterize the subtle yet potentially important differences in intestinal microbial flora between different colonies of mice. We and others (8) assume that in the complex ecosystem of the colonic microbiota, Helicobacter alone is not sufficient (even in a highly susceptible model like the IL-10−/−mouse) to induce mucosal inflammation. This may explain the findings that germ-free IL-10−/− mice infected with H. hepaticus do not develop IBD (8). Therefore, the profound inflammation seen in germ-free mice reconstituted with SPF flora, with or without Helicobacter (8, 37), does not discount the potential role of Helicobacter in murine colitis. Instead, in our opinion, it reinforces the hypothesis that multiple bacterial species can contribute to the establishment of IBD, particularly in a setting where mice without natural tolerance to luminal bacterial (as in germ-free mice not exposed to normal flora during development) are exposed to a wide range of bacterial flora.

Numerous studies have underlined a prominent role for T cells in the initiation and maintenance of IBD (3, 7, 17, 25, 28, 30). Therefore, it is somewhat surprising that several studies have suggested that Helicobacter can induce IBD in animal models devoid of T cells (6, 14, 15, 21, 39, 41). We tested this question directly by infecting Rag1−/− mice, which lack T and B cells, with H. bilis or H. hepaticus. In contrast to the consistent induction of severe IBD in IL-10−/− mice infected with Helicobacter spp, we observed mild acute inflammation in Rag1−/− mice. It is also important to note that the mucosal inflammation observed inHelicobacter-infected Rag1−/− mice was more focal than that seen in similarly infected IL-10−/− and TCR-α−/− mice and was particularly less consistent in its appearance. Still, the intestinal inflammation seen in Rag1−/− mice infected with Helicobactersuggests that the development of bacterial-driven mucosal inflammation is not strictly dependent on T cells. It is tempting to speculate that Helicobacter may trigger the innate immune system resulting in the induction of cytokines and/or other chemical mediators that may eventually produce inflammation. Whether it is through the modulation of intestinal barrier function and/or induction of proinflammatory cytokines via lipopolysaccharide-mediated activation of Toll-like receptors on dendritic cells, monocytes, or epithelial cells or via recently discovered virulence factors in certainHelicobacter spp (5, 45), this innate immune response may contribute to the establishment of mucosal inflammation and IBD.

In summary, our results in a variety of mouse models suggest thatHelicobacter infection plays an important role in the induction of IBD in SPF mice. In our experience, T cells appear to be required for the development of chronic, diffuse, and severe inflammation associated with Helicobacter infection. Still, the observations in Rag1−/− mice presented here, along with a recent report from another group (6), demonstrate that T cells are not strictly required for the induction of mucosal inflammation by Helicobacter. In addition, we show thatHelicobacter may harbor the ability to skew the cytokine responses toward a Th1 phenotype, as shown by the elevated levels of interferon-γ in the colons of Helicobacter-infected mice. Again, we suggest that Helicobacter is likely to be one of a group of bacteria that, in mice, can tip the balance toward mucosal inflammation. Whether there are unifying and/or overlapping features between these microbial triggers for IBD (predilection for skewing toward Th1 responses, for example) remains to be determined. However, a recent report has shown the ability of Citrobacter rodentium to elicit a Th1 cytokine phenotype associated with IBD, suggesting that Th1 responses may be a stereotypical response to certain luminal microbes (16). We believe organisms likeHelicobacter can serve as useful tools to study microbial-induced IBD and may be able to both assist in refining the hypotheses regarding the molecular pathogenesis of IBD and to test novel therapeutic interventions that may be relevant to the treatment of human IBD. Although there are currently no data to suggest that Helicobacter spp are implicated in human Crohn's disease or ulcerative colitis, we are intrigued by recent observations of novel Helicobacter species in both the cotton-top tamarin (36) and in humans with diarrhea (10).

Acknowledgments

We thank Craig Labenz for expert graphics assistance and Loida Torres for excellent animal care.

Footnotes

  • This work was supported, in part, by the Biomedical Research Training Grant for Veterinary Scientists T32-RR07019 (to A. Burich).

  • Address reprint requests to Dr. L. Maggio-Price, Dept. of Comparative Medicine, Box 357190, School of Medicine, Univ. of Washington, Seattle, WA 98195 (E-mail:lmprice{at}u.washington.edu).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

REFERENCES

View Abstract