Role of Bacterial Intimin in Colonic Hyperplasia and Inflammation

See allHide authors and affiliations

Science  23 Jul 1999:
Vol. 285, Issue 5427, pp. 588-591
DOI: 10.1126/science.285.5427.588


Enteropathogenic Escherichia coli (EPEC) cells adhere to gut epithelial cells through intimin α: the ligand for a bacterially derived epithelial transmembrane protein called the translocated intimin receptor. Citrobacter rodentium colonizes the mouse colon in a similar fashion and uses a different intimin: intimin β. Intimin α was found to costimulate submitogenic signals through the T cell receptor. Dead intimin β+ C. rodentium, intimin α–transfected C. rodentium or E. colistrain K12, and EPEC induced mucosal hyperplasia identical to that caused by C. rodentium live infection, as well as a massive T helper cell–type 1 immune response in the colonic mucosa. Mutation of cysteine-937 of intimin to alanine reduced costimulatory activity in vitro and prevented immunopathology in vivo. The mucosal changes elicited by C. rodentium were interferon-γ–dependent. Immunopathology induced by intimin enables the bacteria to promote conditions that are favorable for increased microbial colonization.

EnteropathogenicEscherichia coli (EPEC) is a major cause of infantile diarrhea in humans. The mouse homolog of EPEC, Citrobacter rodentium, causes colonic hyperplasia in young mice. EPEC and C. rodentium colonize the surface of epithelial cells in the intestine. They produce an attaching and effacing (AE) lesion on the epithelial cell lumenal surface, using a type III secretion system to secrete several proteins encoded by the chromosomal locus of enterocyte effacement (1). These proteins include EspA, EspB, and EspD and produce changes in host cell cytoskeleton and signaling pathways (2). An additional protein translocated into the epithelial cell is the translocated intimin receptor (Tir), which functions as the receptor for the bacterial outer membrane protein, intimin (3). EPEC and C. rodentium express homologous intimin molecules, intimin α and intimin β, respectively (4).

The mechanism by which EPEC causes diarrhea remains unknown, although colonization by EPEC results in increased epithelial permeability to macromolecules (5). Live C. rodentium expressing intimin β or C. rodentiumtransfected with EPEC intimin α infects the mouse colon and induces severe immunopathology involving mucosal thickening, a CD4+T cell infiltrate, and a pronounced T helper cell type 1 (TH1) mucosal immune response that is identical to that seen in mouse models of inflammatory bowel disease (IBD) (6). Intimin also binds to β1-integrins on T cells, the consequence of which is unknown (7).

We characterized the consequence of intimin binding to T cells in vitro by examining the effect of the eukaryotic cell–binding, 280–amino acid, COOH-terminal domain of intimin α (int280) (8) on submitogenic stimulation of T cells (9). When treated with both concanavalin A (Con A) and antibody to CD3 (anti-CD3), purified int280 (10) costimulated T cell proliferation in a concentration-dependent manner (Fig. 1A) and increased production of IL-2, IFN-γ, and IL-4 mRNA transcripts (Fig. 1B) (11). Int280 alone produced only minor proliferation [two times background counts per minute (cpm)] and no increase in cytokine transcripts (Fig. 1B). The mutation of Cys937 of int280 to alanine, which reduces integrin binding (7), also reduced costimulatory activity by 54 to 67% (12).

Figure 1

The effect of int280 on the proliferative response of splenocytes. (A) Proliferative response of splenocytes in the presence of int280 (1 to 25 μg/ml) with Con A stimulation (5 μg/ml) (open circle, dashed line), submitogenic anti-CD3 stimulation (145-2C11, 1 μg/ml) (open circle, solid line), or media only (triangle, solid line). (B) Cytokine mRNA transcripts in total RNA isolated from cultured cells cultured in media alone, Con A, or anti-CD3, with or without the addition of int280 at 25 μg/ml. Results are representative of two experiments. Background counts per minute were <1000; nd, not detected.

To answer the question of whether mucosal immunopathology in liveC. rodentium infection is a result of epithelial colonization or of intimin interactions with host immune cells, we intracolonically administered formalin-killed bacteria expressing different intimins to BALB/c mice. Wild-type C. rodentium expresses intimin β; strain DBS255 is aneae-mutant C. rodentium that does not express either intimin α or intimin β; strain DBS255 (pCVD438) expresses EPEC intimin α and no intimin β; and strain DBS255 (pCVD438CA) expresses the mutated form of intimin α in which the Cys937 is replaced by Ala, thus destabilizing the loop needed for integrin binding but not for Tir binding (13).

The epithelial barrier was transiently breached by the intracolonic administration of 50% ethanol (14). Mice were then given 4 × 108 of either killed wild-type intimin β expressing C. rodentium or intimin DBS255 and were killed 6 days later (14). Administration of the intimin β+killed bacteria recapitulated the immunopathology seen in the live infection (6), whereas intimin C. rodentium had no effect. Macroscopically, distal colons of mice given wild-type C. rodentium thickened (Fig. 2A) and were heavier than those of mice given DBS255 (Fig. 2B). Crypts were hyperplastic (Fig. 2C), and there was a predominantly CD4+ and CD3+ T cell and macrophage infiltrate in the lamina propria (15) (Fig. 2, D and E). The CD8 response varied between experiments but was always only a minor component of the infiltrate. Mice killed 14 days after administration still had elevated CD4+ and CD3+infiltrates (16). In the tissue, transcripts for TNF-α and IFN-γ and the epithelial growth factor keratinocyte growth factor (KGF) were increased (Fig. 2F). IL-4 transcripts were reduced by 80% (Fig. 2F). These changes are similar to those seen in murine models of IBD (17).

Figure 2

Intracolonic administration of formalin-killed C. rodentium to BALB/c mice. Mice were killed on day 6. (A) Distal colons of mice given either wild-type C. rodentium (intimin+) or DBS255C. rodentium (intimin). The intimin+ strain produced hypertrophy of the distal colon (bottom image). (B) Weight of the distal 4 cm of the colon. (C) Crypt length. (D) Numbers of CD3+, CD4+, and CD8+ cells in the lamina propria. (E) Immunohistochemistry for CD3+ T cells in the lamina propria of mice given intimin or wild-type (intimin+) strains as inFig. 2A. Positive cells are indicated by arrows (magnification, ×150). (F) mRNA transcripts for TNF-α, IFN-γ, IL-4, and KGF. In (B) through (D) and (F), all mice were given 50% ethanol followed by PBS (white bars), the DBS255 intimin strain of C. rodentium (black bars), or wild-type intimin β+ C. rodentium (hatched bars) [n = 5; asterisk indicates P < 0.05, Mann-Whitney U test with Bonferroni correction]. All data are shown as the mean ±1 SEM.

Mice receiving dead EPEC intimin α expressing C. rodentium (pCVD438) also showed mucosal hyperplasia, with a strong mucosal CD4+ cell response and an increase in TH1 mRNA transcripts. This effect was markedly reduced when the cysteine-mutated strain (pCVD438CA) was used, although there was a small increase in crypt length and in the number of infiltrating T cells as compared to the intimin control bacterium (Fig. 3, A through D). This is presumably because the CS mutation does not completely abolish intimin binding to β1 integrins. It is also difficult to extrapolate from int280 and int280CS costimulation of T cells in vitro to the in vivo effects of the molecules expressed on the bacterial surface. Mice receiving formalin-killed EPEC strain E2348/69 or a laboratory E. coliK12 strain transfected with EPEC intimin α also showed an increase in colon weight and crypt length (Fig. 3, A and B). The in vivo activity of the three different intimin-expressing bacteria suggests that intimin is responsible for inflammation and crypt hyperplasia.

Figure 3

Intracolonic administration of formalin-killed intimin C. rodentium DBS255, wild-type intimin β+ C. rodentium, intimin α–expressing DBS255(pCVD438), mutant intimin α strain (intαΔ) DBS255(pCVD438CA), intimin α+ EPEC E2348/69, or intimin α+ K12 E. coli to BALB/c mice. (A) Weight of the distal 4 cm of the colon. (B) Crypt length. (C) Numbers of CD3+ (black bars), CD4+ (white bars), and CD8+ (hatched bars) cells in the lamina propria. (D) mRNA transcripts for TNF-α, IFN-γ, IL-4, and KGF in total RNA isolated from colonic tissue of mice given the DBS255(pCVD438CA) intimin α mutant strain (intαΔ, black bars) or the DBS255(pCVD438) intimin α strain (white bars) [n = 5; asterisk indicates P< 0.05, Mann-Whitney U test with Bonferroni correction]. All data are shown as the mean ±1 SEM.

To directly assess the role of IFN-γ in the induction of the intimin-mediated T cell response and colonic hyperplasia, formalin-killed wild-type intimin β+ C. rodentium was given to mice that were genetically deficient in the IFN-γ receptor. No immunopathology was observed in these mice as compared to the severe changes seen in the 129Sv+/+controls. Neither colonic weight nor crypt hyperplasia was increased as compared to mice treated with the intimin strain (Fig. 4, A and B). A small increase in lamina propria CD4+ cells was seen in these mice (Fig. 4C), but mRNA transcripts for TNF-α, IFN-γ, and KGF were not significantly increased (Fig. 4D).

Figure 4

Intracolonic administration of formalin-killed C. rodentium to IFN-γR−/−mice and 129Sv+/+ wild-type control mice. All mice were killed on day 6. In IFN-γR−/− mice given intimin-expressing C. rodentium, there was no increase in (A) colonic weight (the weight of IFN-γR−/−colons was more than that of the 129Sv+/+ controls, because mice were slightly older; however, in many experiments we have observed that the weight of the mice makes no difference to the development of hyperplasia) or (B) crypt length. (C) Numbers of CD3+, CD4+, and CD8+ cells in the lamina propria of 129Sv+/+ and IFN-γR−/−mice. (D) mRNA transcripts for TNF-α, IFN-γ, IL-4, and KGF mRNA measured in total RNA isolated from colonic tissue. In all mice, 50% ethanol was administered, followed by the DBS255 intimin strain (black bars) or by wild-type C. rodentium (hatched bars) [n = 5; asterisk indicates P < 0.05, Mann Whitney U test with Bonferroni correction]. All data are shown as the mean ±1 SEM.

Thus, intimin is a bifunctional molecule, acting as a ligand for epithelial cell adhesion to Tir in the formation of the AE lesion and also driving mucosal TH1 responses and subsequent immunopathology. The binding activity of intimin to Tir and β1 integrins is restricted to the COOH-terminal 280 amino acids (18); this domain has two immunoglobulin-like regions and a C-type lectin-like module (10), so that there is the potential for interaction with several different receptors.

Although EPEC is traditionally considered a noninvasive pathogen, the high density of bacteria on the epithelial surface means that some bacteria must translocate into the lamina propria. Indeed, inC. rodentium infection in mice, bacteria can be seen in the lamina propria and submucosa and there are anecdotal reports of bacteria in the mucosa in EPEC infection (6). Thus intimin has the potential to interact with β1integrin–expressing lamina propria T cells (19). Int280 alone had very little direct effect on splenic lymphocytes in vitro (Fig. 1A) and is also not directly mitogenic for colonic lamina propria T cells in vitro (20), so it is unlikely that it directly activates lamina propria T cells. We were also unable to show that intimin could costimulate resident colonic lamina propria T cells (20). However, there are other plausible ways in which intimin can interact with T cells in the gut. The nonspecific recruitment of T cells from the blood into inflamed or infected intestine means that intimin may costimulate naı̈ve T cells responding to antigens from the gut lumen. Alternatively, because physiologically CD4+ T cells respond to gut antigens in the mucosal lymphoid follicles and then migrate to the lamina propria, intimin may be boosting TH1 responses at the inductive site of mucosal responses rather than at their effector sites in the lamina propria (20).

Dead intimin+ C. rodentium elicited a strong TH1 response in the mucosa, and this resulted in mucosal thickening and crypt cell hyperplasia, as seen in mouse models of IBD (17). Intimin-induced epithelial proliferation and mucosal thickening at later stages of infection after initial colonization would benefit EPEC and C. rodentium. An increased surface area in the crypts would provide a larger area for fresh colonization, and increased epithelial renewal would guarantee a continuing supply of uninfected cells. Increased shedding of colonized enterocytes would increase fecal shedding of the organism and transmission. The increased epithelial renewal in the intestine seen in T cell reactions in the gut has traditionally been considered to be controlled by the host. However, the results presented here strongly suggest that it is citrobacter or EPEC, using intimin, that is specifically driving a mucosal TH1 response and subsequent immunopathology. The possibility also exists of a bystander effect due to intimin, in that the TH1 response to other lumenal antigens such as cow's milk proteins or wheat may also be enhanced if antigen exposure in the gut occurs at the same time as EPEC infection.

Many microbial pathogens have evolved sophisticated ways of interacting with host cells and subverting host immune responses by secreting virulence proteins using type III secretion systems. These include, for example, Salmonella typhimurium and Yersinia pseudotuberculosis, which regulate host cell cytokine signaling pathways through the translocation of proteins into eukaryotic cells (21). Our data provide evidence that bacteria have evolved mechanisms to increase immune responses when this would benefit the pathogen.


View Abstract

Navigate This Article