Activation of inflammatory gene expression by the transcription factor NF-κB is a central pathway in many inflammatory disorders, including colitis. Increased NF-κB activity has been linked with development of colitis in humans and animal models, thus it was unexpected when NF-κB-deficient mice developed spontaneous typhlocolitis. To further characterize this finding, we induced typhlocolitis in rederived NF-κB-deficient mice using intragastric infection with Helicobacter hepaticus. At 6 wk postinfection (PI), severe colitis with increased type 1 cytokine expression was seen in infected mice that lacked the p50 subunit of NF-κB and were also heterozygous for the p65 subunit of NF-κB(p50−/−p65+/−). Mice lacking the p50 subunit alone (p50−/−) were less severely affected, and wild-type mice and p65+/− mice were unaffected. T cell development in NF-κB-deficient mice was normal. These data indicate that p50 and p65 subunits of NF-κB have an unexpected role in inhibiting the development of colitis.
Inflammatory bowel disease (IBD)3 is believed to result from an inappropriate mucosal immune response that leads to inflammation and intestinal damage (1). Although the etiopathogenesis of IBD in humans is unknown, the ability of microflora to induce mucosal inflammation in susceptible hosts may be important in the initiation of IBD (2, 3, 4, 5).
It has been suggested that the transcriptional regulator NF-κB may play a key role in regulating inflammatory gene expression in the colon. NF-κB is a family of five proteins related by the Rel homology domain that have been implicated as a central regulator of proinflammatory gene expression (6). These proteins are found as homo- or heterodimeric complexes in the cytoplasm of most cells, bound to members of the IκB family that inhibit nuclear translocation of NF-κB. Many inflammatory signals activate a set of IκB kinases which phosphorylate IκB molecules, leading to their targeted degradation (7). This process frees NF-κB complexes to translocate to the nucleus and influence transcription of key inflammatory cytokines including TNF-α, IL-1, and IL-12 (6).
Increased NF-κB activity and cytokine expression have been observed in macrophages from human Crohn’s disease patients and in several mouse models of colitis (8, 9). Increased NF-κB DNA-binding activity in macrophages from mice with colitis has been attributed predominantly to increased binding of the p50 and p65 subunits. Indeed, increased cytokine expression in mice with 2,4,6-trinitrobenzene sulfonic acid-induced colitis was inhibited by antisense oligonucleotides that inhibit p65 expression (10).These studies suggest a central role for NF-κB in mediating expression of proinflammatory cytokines, which is an important component of the pathology associated with IBD.
Recently, several mouse models lacking NF-κB family members have been developed (11, 12). Although mice lacking p50 are viable, mice lacking p65 die during embryogenesis. However, mice homozygously deficient for p50 (p50−/−) and also heterozygous for p65 (p50−/−p65+/−) are viable. We have been studying these mice and were surprised to find that both p50−/− and p50−/−p65+/− mice developed spontaneous typhlocolitis when housed in a virus Ab-free (VAF) colony. These mice were rederived into a helicobacter-free colony because some mouse models develop spontaneous enterocolitis that is reduced under defined microflora conditions and is inducible with targeted reinfection with intestinal microorganisms (2, 3, 4, 5). Mice were then infected with an enteric helicobacter to provide a reproducible model of colitis. In this study, we examine the development of typhlocolitis in rederived NF-κB-deficient mice experimentally infected with Helicobacter hepaticus.
Materials and Methods
Spontaneous typhlocolitis in mice from a VAF colony
All mice were housed in Association for the Assessment and Accreditation of Laboratory Animal Care approved facilities. Mice that developed spontaneous colitis were murine VAF, were naturally infected with opportunistic pathogens including Pasteurella pneumotropica and Helicobacter rodentium, but were not infected with H. hepaticus or Helicobacter bilis.
Rederivation of NF-κB mutant mice
For experimental infection, mice were rederived using uterine transfer of 3.5-day embryos of p50 and p65 mutants on a mixed (129 × C57BL/6) background. NF-κB mutant and wild-type (WT) mice were maintained under conditions free of known Helicobacter species.
H. hepaticus (type strain 3B1) was grown as described elsewhere (13). Cultures were examined by Gram stain and phase microscopy for contaminants and subcultured on blood agar to confirm purity. Bacteria were resuspended in Brucella broth with 30% glycerol at 108 bacteria/ml as confirmed by spectrophotometry (2). Experimental mice received 0.2 ml of fresh inoculum by gastric gavage every other day for three doses.
At age 6–8 wk, 64 helicobacter-free mice were dosed by gastric gavage with H. hepaticus suspended in broth and 40 were dosed with broth only. The infected and sham mice were housed in microisolator caging in different areas within the animal facility. These data were compiled from four replicate experiments using similar numbers of mice with comparable results for each group.
Confirmation of H. hepaticus infection
Cecal tissue was collected at necropsy and analyzed by PCR to confirm experimental infection using H. hepaticus-specific primers (14). Helicobacter-free status was confirmed in controls using PCR with Helicobacter genus-specific primers.
Formalin-fixed tissues were embedded in paraffin, cut at 5-μm sections, and stained with hematoxylin and eosin for assessment of morphology. A board-certified veterinary pathologist blinded to sample identity evaluated tissues. The cecal and colonic lesions were scored on the basis of size and frequency of hyperplastic and inflammatory lesions on a scale of 0–4, with ascending severity modified from Berg et al. (15).
Tissues were frozen in OCT in liquid nitrogen and cut into 5-μm sections using a Leica CM3050 cryostat. Slides were incubated with methanol and 3% hydrogen peroxide for 30 min, followed by 20% goat serum for 30 min and then washed with PBS. Primary Ab (rat anti-CD3 for T cells or rat anti-F4/80 for macrophages; Vector Laboratories, Burlingame, CA) or nonimmune rat serum was used at 1:10 for 60 min. For F4/80 staining, slides were incubated in HRP-conjugated goat anti-rat Ab at 1:25 for 30 min and diaminobenzidine was used as a color substrate. For anti-CD3 staining, slides were incubated with rabbit anti-rat Ab (1:100) for 60 min, followed by AP-conjugated goat anti-rabbit Ab at 1:500 for 60 min. Vector red was used as a color substrate.
Detection of cytokine mRNA expression in colon
One-centimeter segments of colon were harvested immediately upon euthanasia and snap frozen in liquid nitrogen. Frozen specimens were homogenized into Tri-reagent (Molecular Research Center, Cincinnati, OH) and RNA was prepared per the manufacturer’s instructions. RNase protection analyses were performed on 20 μg of total RNA using RiboQuant MultiProbe Template Sets (PharMingen, San Diego, CA). Intensities of the protected fragments were quantitated by phosphor imager analysis and normalized to internal controls.
Analysis of cecal and colonic lesion scores was performed using a Mann-Whitney U nonparametric test for categorical data. Data on RNase protection assays was analyzed using a two-tailed t test for nonequal sample size.
NF-κB-deficient mice spontaneously develop typhlocolitis
Spontaneous typhlocolitis was observed in a colony of VAF NF-κB-deficient mice at age >6 mo. Severity of disease was assessed histologically using a scale of 0–4 described in Materials and Methods. Although WT mice (n = 4) cohoused with NF-κB-deficient mice were clinically normal, p50−/−p65+/− mice had severe diarrhea, colonic perforation, peritonitis, and typhlocolitis (n = 9; cecum, p = 0.013; colon, p = 0.032). Mice lacking p50 (p50−/−) alone had a more mild disease (n = 9; cecum, p = 0.047; colon, p = 0.623). Mice heterozygous for p65 (p65+/−) did not have significant clinical or histological (n = 9; cecum, p = 0.215; colon, p = 0.361) evidence of typhlocolitis. These results suggested that the absence of certain subunits of NF-κB could predispose mice to the development of typhlocolitis.
Mice lacking NF-κB are highly sensitive to typhlocolitis induced by H. hepaticus
To determine whether typhlocolitis was experimentally reproducible using infection with H. hepaticus as previously described (2, 3, 4, 5), NF-κB-deficient mice were embryo transfer rederived into a helicobacter-free environment. Spontaneous typhlocolitis was not observed in the helicobacter-free NF-κB-deficient colony. Mutant and control mice were then experimentally infected with H. hepaticus. Two weeks after intragastric inoculation with H. hepaticus, p50−/−p65+/− mice developed severe diarrhea. Six weeks after H. hepaticus infection, approximately one-quarter of the p50−/− mice also developed diarrhea. All p65+/− and WT animals remained clinically well.
To evaluate experimental animals for the presence and severity of colitis, sections of cecum and colon were evaluated histologically. Representative sections from infected and uninfected WT, p50−/−, and p50−/−p65+/− are shown in Fig. 1. The severity of typhlocolitis was scored as described above. There was a correlation between genotype, helicobacter status, and pathology score (Table I). The typhlocolitis was significantly more severe in p50−/−p65+/− (n = 19; cecum, p < 0.00001; colon, p < 0.00001) and p50−/− (n = 15; cecum, p < 0.00001; colon, p = 0.0001) compared with infected WT controls. Furthermore, there was significantly more severe inflammation in p50−/−p65+/− (cecum, p = 0.0007; colon, p < 0.00001) compared with p50−/− mice, suggesting that the additional loss of a p65 allele exacerbated disease severity. There was minimal inflammation in uninfected mice of all genotypes.
As indicated by the typhlocolitis scores above, there was marked pathology in experimentally infected p50−/−p65+/− mice. At 6 wk PI, the ascending and transverse colon of infected p50−/−p65+/− mice were grossly enlarged and thickened (Fig. 2,a). Three (3/19) mice had colonic perforations indicated grossly by stricture and adhesions to other viscera. Histologically, there was extensive, severe chronic active transmural inflammation with marked distortion and replacement of the normal architecture (Fig. 2,b). Within these regions, the thickened mucosa was characterized by extensive ulcerations, elongated irregularly shaped glands, and crypt abscesses. The lamina propria (LP) was expanded by fibrosis with effacing infiltrate of granulocytes, macrophages, lymphocytes, and occasional microgranulomas. Lymphofollicular hyperplasia and plasmacytes were infrequent in these mice. The submucosa and serosa were frequently expanded and effaced by dense inflammatory infiltrates, granulation tissue, dilated lymphatics, and multifocal necrotizing vasculitis with thrombosis. Extensive areas of necrosis and edema within these layers resulted in complete loss of bowel wall integrity with extension of the lesions into adjacent mesenteric lymph nodes and mesenteric vessels. These dysplastic regions were interspersed with regions of mild to moderately severe colitis with mixed inflammatory infiltrates (Fig. 2,c). Although few macrophages or T lymphocytes were seen in the LP of infected WT mice, macrophages (F4/80+) were prevalent and T lymphocytes (CD3+) could be identified in scattered foci throughout the LP (Fig. 2, d–g) in infected NF-κB-deficient mice.
Affected tissues express high levels of inflammatory cytokines
To examine cytokine expression in NF-κB-deficient mice with helicobacter-induced typhlocolitis, total RNA was isolated from specimens of colon and inflammatory gene expression was analyzed by RNase protection analysis (Fig. 3). There were statistically significant increases in the expression of IFN-γ, IL-12p40, TNF-α, IL-1β, monocyte chemoattractant protein-1, IFN-γ-inducible protein-10, and macrophage-inflammatory protein-2 in H. hepaticus-infected p50−/−p65+/− and p50−/− mice compared with infected WT mice. Furthermore, there were significant increases in the expression of TGF-β and IL-1R antagonist (IL-1Ra) in infected p50−/−p65+/−, but not p50−/−, compared with infected WT mice. There were statistically significant increases in the expression of TNF-α, IL-1β, IFN-inducible protein-10, macrophage-inflammatory protein-2, and IL-1Ra in infected p50−/−p65+/− mice compared with infected p50−/−. Little inflammatory gene expression was observed in uninfected animals of any group.
The observation of spontaneous typhlocolitis in NF-κB mutant mice was rapidly reproducible using targeted infection with H. hepaticus. Typhlocolitis involved extensive epithelial damage and effacing inflammatory infiltrates. Increased expression of type 1 cytokines resembled findings in IL-10-deficient mice with colitis and in humans with Crohn’s disease (4, 16). Colitis was inducible using H. hepaticus in helicobacter-free mutant mice, as described in IL-10-deficient mice by Kullberg et al. (4), further supporting a role for intestinal microflora in the induction of IBD. Inability to induce colitis with H. hepaticus in some mouse models may be due to variations in methodologies, host genotype, or constitutive intestinal microflora (17).
Typhlocolitis in NF-κB-deficient mice was unexpected because NF-κB subunits p50 and p65 have been previously implicated as promoters of IBD (8, 9). Indeed, antisense p65 oligonucleotides have been proposed to abrogate the severity of 2,4,6-trinitrobenzene sulfonic acid-induced colitis (10). However, the findings of increased inflammatory gene expression in this study indicate robust inflammatory gene expression can occur in p50−/− or p50−/−p65+/− mice. Thus, it appears that mice lacking p50 exhibit a defect in a pathway that inhibits the development of colitis but this does not preclude the possibility that other NF-κB subunits could also play a reciprocal proinflammatory role.
The mechanism by which p50 inhibits inflammation is not clear. Unlike p65, p50 does not have intrinsic trans-activating potential. It has been suggested that p50 homodimers have the ability to directly inhibit gene expression. Also, p50 appears to suppress TNF-α expression after LPS challenge of peritoneal exudative macrophages (18). Increased expression of TNF-α or other inflammatory cytokines after exposure of NF-κB-deficient mice to H. hepaticus could lead to excessive inflammation and the development of typhlocolitis. However, it is not known whether increased inflammatory gene expression is an initiating factor in the development of disease. In contrast to a role as an inhibitor of inflammatory gene expression, p50 could be required to promote expression of a cytokine that inhibits inflammation such as IL-10, TGF-β, or IL-1Ra. However increased expression of IL-1Ra and TGF-β observed in the colons of NF-κB-deficient mice with colitis make it unlikely that p50 is necessary for regulation of these genes.
The p50-deficient mice used in this study also lack the p50 precursor protein p105. The C-terminal domain of p105 has homology to IκB family members and has been shown to function as an IκB in vivo (19). Therefore, the proinflammatory phenotype observed in p50-deficient mice could be related to defective p105 inhibitory function. Indeed, mice lacking the C-terminal IκB-like domain of p105 but retaining p50 exhibit lymphoid infiltration into skin and organs, although they have not been reported to develop colitis (20).
Defects in T cell development in the absence of p50 could predispose animals to the development of colitis (21, 22). However, we found normal distributions of CD4 and CD8 lymphocytes in the thymus and the spleen. Also, the percentage of CD25+CD4+CD8− T cells, which have been associated with repressive activity (23, 24)was normal in the thymus and the periphery (data not shown). Furthermore, H. hepaticus does not appear to induce generalized T cell activation, as the majority of CD4+ T cells in the mesenteric lymph node of infected mice were CD62LhighCD69low (data not shown). Thus, defects in the development of regulatory T cell populations do not appear to be responsible for the susceptibility of these mice to colitis, although function of regulatory T cells in NF-κB-deficient mice has not been examined.
In addition to certain regulatory T cell populations, B cells may also play a role in the development of IBD (25, 26). Although there are normal numbers of mature B cells in p50−/− and p50−/−p65+/− mice (data not shown), it has previously been shown that p50−/− mice lack marginal zone B cells (27) and p50-deficient B cells have severe defects in class switching and secretion of IgA in vivo and in vitro (11, 28). Indeed, lymphofollicular hyperplasia and plasmacytosis, prominent features of colitis in other murine models (15), were not features of typhlocolitis in the NF-κB-deficient mice (data not shown). These observations may represent a manifestation of defective B cell function in NF-κB mutants that is important in the etiology of colitis.
It is unclear how the loss of one of the two alleles of p65 exacerbated colitis in this model. Mice heterozygous for p65 alone do not develop colitis and, unlike p50, inhibitory activity has not been demonstrated for the p65 subunit. The loss of one allele of p65 in addition to both alleles of p50 could lead to a gene dosage effect that increases susceptibility to colitis, or the loss of one allele of p65 could dysregulate an independent process that is only required in the absence of p50.
The findings reported here suggest that NF-κB plays a critical role in suppressing the development of IBD and have led us to hypothesize that NF-κB activation initiates a feedback loop that inhibits the development of chronic inflammation in the colon. These observations could lead to important insights into the pathophysiology of IBD in humans and have implications for directed therapeutic strategies.
We thank Dr. Takeo Kosaka for assistance with immunohistochemistry, Dr. Melanie Ihrig for guidance with statistical analyses, Jennifer Cline for helping with necropsies, and Elaine Robbins for assistance with figures for this manuscript.
Funding support for this study was provided in part by National Institutes of Health Grants R01CA67529 and P01CA26731, and a Child Health Research Grant from the Charles H. Hood Foundation.
Abbreviations used in this paper: IBD, inflammatory bowel disease; VAF, virus Ab free; WT, wild type; PI, postinfection; LP, lamina propria; IL-1Ra, IL-1R antagonist.