Supported by the Ministère de la Recherche et de la Technologie (JE2526), INRA (USC-2018), and by grants from the Association F. Aupetit (AFA), Région Auvergne (Nouveau Chercheur), from European Commission through FP7 IBDase project, from ANR in the frame of ERA-NET PathoGenomics, and Grant 2010 from Infectiopôle Sud, Marseille, France.
Adherent-invasive Escherichia coli induce claudin-2 expression and barrier defect in CEABAC10 mice and Crohn's disease patients†
Article first published online: 17 JUN 2011
Copyright © 2011 Crohn's & Colitis Foundation of America, Inc.
Inflammatory Bowel Diseases
Volume 18, Issue 2, pages 294–304, February 2012
How to Cite
Denizot, J., Sivignon, A., Barreau, F., Darcha, C., Chan, H.F. C., Stanners, C. P., Hofman, P., Darfeuille-Michaud, A. and Barnich, N. (2012), Adherent-invasive Escherichia coli induce claudin-2 expression and barrier defect in CEABAC10 mice and Crohn's disease patients. Inflamm Bowel Dis, 18: 294–304. doi: 10.1002/ibd.21787
- Issue published online: 10 JAN 2012
- Article first published online: 17 JUN 2011
- Manuscript Accepted: 9 MAY 2011
- Manuscript Received: 2 MAY 2011
- inflammatory bowel disease;
- barrier function;
- type 1 pili;
- CEACAMs molecules;
Abnormal expression of CEACAM6 observed on the ileal epithelium in Crohn's disease (CD) patients allows adherent-invasive Escherichia coli (AIEC) to colonize gut mucosa. Since intestinal permeability is significantly increased in CD patients, we aimed at investigating whether and how AIEC alter barrier function.
Tissue microarray was performed on ileal biopsies from CD patients in quiescent and active phases. CEABAC10 or wildtype mice were orally challenged with 109 bacteria. Intestinal permeability was assessed by measuring 4 kDa dextran-FITC flux in serum, barrier integrity was analyzed using biotin tracer experiment, and claudin-2 protein immunostaining. Bacterial translocation was analyzed in Ussing chambers.
Pore-forming tight junction protein claudin-2 is strongly expressed in the ileum of 51% patients in quiescent phase and in 49% of the patients with active CD. Infection of CEABAC10 transgenic mice expressing human CEACAMs with AIEC, but not with nonpathogenic E. coli, led to a significant 3.0-fold increase in intestinal permeability and to disruption of mucosal integrity in a type 1 pili-dependent mechanism. This is consistent with the claudin-2 abnormal expression at the plasma membrane of intestinal epithelial cells observed in AIEC-infected CEABAC10 mice. AIEC bacteria were able to translocate through CEABAC10 intestinal mucosa.
These findings strongly support the hypothesis that AIEC type 1 pili-mediated interaction with CEACAM6 abnormally expressed in the quiescent phase of CD may disrupt intestinal barrier integrity before the onset of inflammation. Thus, therapeutic targeting claudin-2 induced by AIEC infection could be a new clinical strategy for preserving intestinal barrier function in CD patients. (Inflamm Bowel Dis 2011;)
Inflammatory bowel diseases (IBDs), including ulcerative colitis (UC) and Crohn's disease (CD), are immunologically mediated with genetic and environmental influences. Genetic factors include defective immunoregulation, mucosal integrity/repair, and bacterial killing. The abnormal inflammatory response observed in IBD involves an interplay between host susceptibility factors and the intestinal microbiota,1 (review2). Adherent-invasive E. coli (AIEC) are highly associated with ileal mucosa in CD patients.3 CD-associated AIEC adhered to the brush border of primary ileal enterocytes isolated from CD patients but not from controls without IBD via the recognition between type 1 pili expressed on the bacterial surface and carcinoembryonic antigen-related cell-adhesion molecule 6 (CEACAM6) abnormally expressed in the ileal epithelial cells of CD patients.4 CD-associated AIEC colonize and induce strong gut inflammation in CEABAC10 transgenic mice expressing human CEACAMs.5 Two independent studies have reported in vitro that AIEC decrease transepithelial resistance of intestinal epithelial cell (IEC) monolayer and induce disorganization of F-actin and mislocalization of ZO-1 and E-cadherin, suggesting an effect of bacterial infection on intestinal defective barrier function in CD patients.6, 7
The gut epithelial barrier separates the luminal contents from the mucosal immune system. Increased intestinal permeability has been linked to chronic mucosal inflammation and IBD.8 More recently, intestinal permeability was reported to be significantly increased in 36% of CD patients.9 Vogelsang10 recently suggested that environmental factors must be involved since 1) intestinal permeability can decrease over time even without ongoing therapy, 2) permeability is not increased in every patient with CD, and 3) permeability can be increased in spouses of CD patients. In parallel, bacterial translocation of E. coli, Enterococcus spp., Clostridium perfringens has been observed in the mesenteric nodes in 30%–50% of CD patients versus 5%–15% in healthy controls.11, 12 This can be explained by defects in the structure and function of apical junctional complexes (AJCs), which are involved in animal models of IBD and in IBD patients.13 Moreover, the epithelium in inflamed intestinal segments of CD patients is characterized by a decrease in tight junction strands, strand breaks, and alterations of tight junction protein composition. Barrier dysfunction precedes relapse of CD in asymptomatic patients, and studies demonstrate specific distribution pattern of the tight junction proteins in IBD patients.14, 15 These observations present novel insights into the physiopathology of CD by a previously unrecognized defect in intestinal barrier function in ileal CD. Therefore, in the present study we analyzed whether AIEC infection may alter intestinal barrier function in the context of CEACAM6 abnormal expression. Since CD patients have increased intestinal permeability and since claudin-2 overexpression has been shown to increase intestinal permeability, we hypothesized that AIEC colonization, involving CEACAM6 abnormal expression, may lead to the increase in claudin-2 expression altering intestinal permeability.
MATERIALS AND METHODS
Patients, Bacterial Strains, and Mice Model
All patients included in this study were hospitalized in the Department of Gastroenterology (Hôpital Archet II, Université de Nice Sophia, France) and provided signed consent for this study. The protocol was approved by the local ethics committee of the Hôpital Pasteur of Nice. Intestinal biopsies were obtained from macroscopically inflamed mucosa of the terminal ileum in 105 CD patients and from macroscopically noninflamed mucosa of the terminal ileum in 95 CD patients (quiescent phase of the disease). There were 115 men and 85 women, with a mean age of 44 years (range 19–64) and mean disease duration of 13 years (range 2–26). Patients were all French Caucasians. In addition, biopsies were taken from the ileum of 60 control patients consisting of individuals who had no significant pathological findings following endoscopic examination for changes in stool habits, abdominal pain, or cancer surveillance.
Ampicillin–erythromycin-resistant AIEC LF82 isolated from a CD patient were used.16 A nonpiliated insertion mutant 52D11 and the E. coli K-12 MG1655 rifampicin-resistant strain (laboratory stock) were also used in this study. Bacteria were cultured as described previously.5
All mice were housed in specific pathogen-free conditions in the animal care facility at the Université d'Auvergne (Clermont-Ferrand, France). FVB/N wildtype (WT) mice were purchased from Charles River Laboratories (Wilmington, MA) and CEABAC10 transgenic mice (heterozygote17) were maintained in our animal facilities. WT and CEABAC10 mice were coupled to obtain 50% WT mice and 50% CEABAC10 mice. Animals from the same generation were used for experimentation.
Tissue Microarray (TMA) Construction and Immunohistochemistry
Representative intestinal biopsies obtained for each individual in building TMAs were selected from hematoxylin and eosin (H&E)-stained sections. Briefly, TMA were constructed as previously described for digestive specimens.18, 19 The tissue cores were arrayed into a new paraffin block using a fine steel needle. Immunohistochemical methods were performed as previously described.18, 19 Anti-claudin 2 antibody (AbCAM), and monoclonal anti-CEACAM6 clone 9A6 (Genovac, Freiburg, Germany) were used. After immunostaining, slides were analyzed with an image analysis workstation (Spot Browser v. 7; Alphelys, Plaisie, France), as described previously.20
Mice infection, Clinical Assessment of Colitis, and Histological Evaluation of Colonic Damage
Seven to 9-week-old FVB/N WT or CEABAC10 female mice (body weight, 20–26 g) were pretreated by oral administration of the broad-spectrum antibiotic streptomycin (5 mg intragastric per mouse) to disrupt normal resident bacterial flora in the intestinal tract21, 22 and were orally challenged with 109 bacteria 24 hours later. For induction of experimental colitis, mice received 1% (w/v) dextran sulfate sodium (DSS, M.W. = 36,000–50,000, MP Biomedicals, Solon, OH) in drinking water for 6 days. The Disease Activity Index (DAI) was ascertained as defined in Table 1 and colonic histological damages were assessed as described in Table 2 after H&E staining.
|Body weight loss||0||no loss|
|1||1 to 5% loss of body weight|
|2||5 to 10% loss of body weight|
|3||10 to 20% loss of body weight|
|4||>20% loss of body weight|
|Stool consistency||0||normal feces|
|3||slimy diarrhea, little blood|
|4||severe watery diarrhea with blood|
|Blood in stool||0||no blood|
|2||presence of blood assessed by Hemoccult II test|
|Infiltration of inflammatory cells||0||rare inflammatory cells in the lamina propria|
|1||increased numbers of inflammatory cells, including neutrophils in the lamina propria|
|2||confluence of inflammatory cells extending into the submucosa|
|3||transmural extension of the inflammatory cell infiltrate|
|Infiltration of lamina propria by mononuclear cells||0 1 2||normal moderate Great|
|Infiltration of lamina propria by polynuclear cells||0 1 2||normal moderate great|
|Infiltration of epithelium by polynuclear cells||0 1 2 3||no infiltration surface inside the crypt cryptic abscess|
|Severity of epithelial damage||0 1 2 3||absence of mucosal damage lymphoepithelial lesions mucosal erosion/ulceration extensive mucosal damage and extension through deeper structures of the bowel wall|
|Surface of epithelial damage||0 1 2||normal focal wide|
Ussing Chamber Experiments and Fluorescence In Situ Hybridization (FISH)
Biopsies from WT or CEABAC10 proximal colon were placed in a chamber exposing 0.196 cm2 of tissue surface to 1.6 mL of circulating oxygenated Ringer solution at 37°C. Bacterial translocation was studied using ampicilin/erythromycin-resistant AIEC LF82, type 1 pili deficient mutant 52D11 or Rifampicyn-resistant E. coli K-12 MG1655 at a final concentration of 107 CFU/mL in the mucosal reservoir. Bacteria from the serosal compartment were numbered at 0, 3, and 5 hours after infection by plating on agar plate supplemented with appropriate antibiotics. For visualization of the interaction between E. coli bacteria and colonic mucosa used in Ussing chambers, 5-μm sections of paraffin-embedded colonic mucosa were subjected to FISH using Cy3-labeled probe EUB338.23 The slides were examined with a LSM 510 Meta confocal microscope.
In Vivo Intestinal Permeability Measurement and Tracer Experiments
In vivo intestinal permeability was measured using FITC-dextran 4 kDa (FD4, Sigma, St. Louis, MO). Mice were orally challenged with 15 mg FD4 diluted in phosphate-buffered saline (PBS) 5 hours before blood collection. Serum was collected by centrifugation (30 min, 5500g) and FITC concentration was determined by fluorescence measurement and compared with a standard curve of FD4 diluted in serum.
EZ-link Sulfo-NHS-Biotin (Pierce Chemical, Rockford, IL) was used as a small (443 Da) molecular tracer. Biotin was diluted to 2 mg/mL in PBS + 1 mM CaCl2 just prior to use. Animals were sacrificed and colons were removed and placed in PBS at 37°C. Biotin was slowly injected into the colon for 3.5 minutes. Colons were then snap-frozen and cut in optimal cutting temperature (OCT) in 5 μm slides. Tissue sections were incubated with a 1:500 dilution of Alexa488-conjugated streptavidin for 30 minutes and imaged by confocal microscopy using LSM 510 Meta. To test for endogenous biotin reactivity, biotin was not added to the tissue.
Immunofluorescence and Immunohistochemistry
Snap-frozen colon in isopentane were embedded in OCT medium and stored at −80°C. Eight micrometers of frozen colon were cut in a cryostat. Colonic mucosa were fixed in 1% paraformaldehyde (PFA) for 20 minutes, washed in PBS, and permeabilization was performed using 0.5× Triton X-100 in PBS for 20 minutes. Unspecific sites were blocked using PBS 5% foetal bovin serum (FBS) and 2% bovine serum albumin (BSA) for 1 hour. Rabbit anti-claudin-2 (AbCam) antibody was diluted in blocking buffer (1/150) and incubated overnight at 4°C. After PBS washes, tissues were incubated for 90 minutes with a donkey antirabbit Cy3-conjugated secondary antibody diluted in PBS-FBS 5% supplemented with Hoechst. Slides were mounted using Mountex-mounting medium (CellPath). Tissues were visualized using a confocal microscope Zeiss LSM 510 Meta (Carl Zeiss, Thornwood, NY).
Immunohistochemistry for human CEACAM6 expression in mice was performed on 5-μm deparaffinized sections using an automated single-staining procedure (Benchmark XT, Ventana Medical Systems, Roche Group, Tucson, AZ). Briefly, citrate-pretreated sections were immunostained for CEACAM6 (monoclonal mouse anti-CEACAM6, clone 9A6, 30 min incubation with a dilution of 1/50) and revealed with the Ultra View kit (Ventana).
For analysis of the significance of difference in claudin-2 and CEACAM6 immunostainings, assays were compared using Student's t-test. Values are expressed as the mean ± SEM of number of experiments. The association of claudin-2 expression with disease phases was made using chi squared analysis. For mice experiments, values are expressed as mean ± SEM or median. Statistical analyses were performed using GraphPad Prism 4.00 (GraphPad Software, San Diego, CA) software package for PC. Single comparisons were performed by unpaired Mann–Whitney test. P < 0.05 was considered as statistically significant.
Each patient who took part in the study provided a signed agreement, and the TMA protocol was approved by the local Ethics Committee of the University of Nice Sophia Antipolis, France. Animal protocols were approved by the Ethical Committee CEMEAA, France.
Pore-forming Tight Junction Claudin-2 Protein Expression Is Strongly Increased in the Ileal Epithelium of CD Patients
Intestinal biopsies were processed for claudin-2 immunostaining on ileal mucosa of 105 CD patients with macroscopically inflamed mucosa, of 95 CD patients with noninflamed mucosa, and of 60 controls using anti-claudin-2 antibody. In contrast to control ileal biopsies, immunohistochemistry showed a very strong staining of claudin-2 in ileal epithelium in the quiescent and acute phases of CD patients (Fig. 1A,a–c). In quiescent and active CD patients, claudin-2 immunostaining was observed in 48/95 (51%) and 51/105 (49%) individuals, respectively. In CD patients, claudin-2 was observed at the surface of the epithelium, mostly at the lateral plasma membrane. Quantification of claudin-2 immunostaining of ileal biopsies showed that the numbers of claudin-2-positive cells in patients in the quiescent or acute phase of CD were similar and significantly higher than those observed in controls (Fig. 1B). To determine whether a link exists between claudin-2 and CEACAM6 expression, we also analyzed the expression of CEACAM6 in CD ileal specimens. As previously reported,24 basal CEACAM6 expression was observed on the apical side of the epithelium in quiescent CD and was highly increased in the active CD patients included in the present study (Fig. 1A,d–f,B). Interestingly, strong staining of claudin-2 was observed together with that of CEACAM6 at the apical plasma membrane of the ileal epithelium of CD patients in quiescent phase.
Intestinal Expression of Human CEACAMs Alters Intestinal Homeostasis
We performed in vivo experiments in order to evaluate the effect of abnormal expression of human CEACAMs on gut barrier function. Intestinal permeability was assessed by measuring FITC-labeled 4 kDa dextran concentration in the serum 5 hours after intragastric administration. A significant 1.8-fold increase in fluorescence was measured in the serum of CEABAC10 mice compared to WT controls (Fig. 2A). A median of 0.75 μg/mL of FITC-dextran was detected in the serum of CEABAC10 mice compared to 0.41 μg/mL in the serum of WT mice (P = 0.003). However, histological examination of colonic mucosa from CEABAC10 mice showed no signs of inflammation. No epithelial damage and no infiltration of immune cells in the lamina propria of colonic mucosa were observed in the absence of infection (Fig. 2B). Abnormal expression of human CEACAMs in CEABAC10 mice and their involvement in intestinal permeability regulation in vivo without any clinical evidence of disease led us to predict that CEABAC10 mice would have enhanced sensitivity to induced colitis. Increased clinical symptoms of colitis were observed in CEABAC10 mice treated with DSS compared to WT mice. The DAI score of CEABAC10 mice treated with 1% of DSS was significantly higher at day 6 (5.2 ± 1.3) compared to WT mice treated with similar amounts of DSS (0.6 ± 0.3; P < 0.001) (Fig. 2C). Taken together, these data show that CEABAC10 mice did not develop spontaneous intestinal inflammation, but rather suggest a central role of human CEACAMs in the regulation of epithelial barrier function.
AIEC Gut Colonization Increases Intestinal Permeability in CEABAC10 Mice
We next assessed the impact of AIEC infection in the intestinal permeability of WT and CEABAC10 mice. Interestingly, at day 3 postinfection, which is a very early stage in the disease course, a significant 3.0-fold (2.26 μg/mL) (P < 0.001) increase in FITC-labeled 4 kDa dextran concentration was measured in the serum of AIEC LF82-infected CEABAC10 mice compared to noninfected CEABAC10 mice (0.75 μg/mL). In contrast, no increase in fluorescence in serum was detected in AIEC LF82-infected WT mice compared to noninfected WT mice (Fig. 3A). Of note, as observed after AIEC infection, a 6-day treatment with 1% DSS led to a 3.0-fold (2.24 μg/mL) increase in FITC-labeled 4 kDa dextran concentration measured in the serum of CEABAC10 mice, compared to untreated CEABAC10 mice (0.75 μg/mL) (n = 5) (Fig. 3A). Histological examinations by standard blinded histological scoring parameters of colonic tissues were performed to evaluate the degree of inflammation. The histological score was not significantly increased in AIEC LF82 infected-CEABAC10 mice 3 days postinfection and was similar to what was observed after 1% DSS treatment for 6 days in WT mice. In contrast, histological score was significantly increased in 1% DSS-treated CEABAC10 mice compared to 1% DSS-treated WT mice or nontreated CEABAC10 mice (Fig. 3B). Hemorrhagic walls with multiple ulcerations, mucosal edema, neutrophil infiltrations with transmural involvement, cryptic abscess formation, and presence of large erosion areas were observed only on histological colonic sections from 1% DSS-treated CEABAC10 mice. Thus, the increased 4 kDa dextran concentration measured in the serum of AIEC LF82-infected CEABAC10 mice is not a consequence of the heavy epithelial damage or inflammation as observed in 1% DSS-treated CEABAC10 mice. Interestingly, as observed with nonpathogenic E. coli MG1655, infection with the AIEC LF82 type 1 pili deficient mutant (52D11) did not increase FITC-dextran flux (Fig. 3C). These findings strongly suggest that type 1 pili-dependent interaction between AIEC bacteria and CEACAM molecules alters mucosa integrity, which precedes the development of inflammation.
AIEC LF82 Infection Induces Tight Junction Claudin-2 Protein in CEABAC10 Mice Intestine
To determine whether the integrity of the apical junction complex in the colon is functionally altered by AIEC infection, we used a biotin tracer molecule. In colon from LF82-infected CEABAC10 mice, biotin/streptavidin staining was observed within the intestinal mucosa and labeling was also noted within and between many colonic surface epithelial cells (Fig. 4A). In contrast, biotin/streptavidin staining was only detected on the surface of mucosa in WT-infected mice, as well as in noninfected WT and CEABAC10 mice. Integrity of the intestinal barrier depends on the pattern and localization of TJs proteins. Since we observed a strong increase in pore-forming claudin-2 expression in quiescent and acute phase in CD patients, we investigated whether AIEC infection could lead to an increase in claudin-2 in CEABAC10 mice. AIEC LF82 infection led to a strong expression of membrane associated pore-forming claudin-2 protein in the colonic mucosa of CEABAC10 mice infected with AIEC LF82 bacteria. In contrast, infection with E. coli K12 MG1655 increased claudin-2 expression, but labeling was not associated with cells plasma membrane (Fig. 4B). As control, claudin-2 protein was barely detected in uninfected WT and CEABAC10 mice colons. These findings indicate that AIEC are able to alter in vivo intestinal barrier function by modulating claudin-2 expression in the gut mucosa of mice expressing human CEACAMs.
Intestinal CEACAMs Expression Favors AIEC Translocation
We investigated whether increase in intestinal permeability, measured at the basal level or after AIEC-infection in CEABAC10 mice, could help AIEC LF82 bacteria to translocate across intestinal epithelium. Since immunostaining analysis indicated that expression of CEACAM6 was restricted to the epithelial surface of colonic mucosa (Fig. 5A), we mounted CEACAM6-expressing or WT proximal colons ex vivo in Ussing chambers to analyze bacterial translocation. AIEC LF82 were detected on the serosal side of CEACAM6-expressing proximal colon after 3 hours of infection (1956.7 cfu/mm2), and to a greater extent after 5 hours (55439.5 cfu/mm2) (Fig. 5B). In contrast, no bacteria were detected on the serosal side of WT colon. When the nonpathogenic E. coli K-12 MG1655 strain was used, no bacteria were detected on the serosal side of either WT or CEACAM6-expressing colon (Fig. 5C). FISH analysis visualized AIEC LF82 bacteria within the epithelium of CEACAM6-expressing colonic mucosa after 5 hours of infection in Ussing chambers (Fig. 5D), but not in WT mucosa. Of great interest, the ability of AIEC LF82 type 1 pili-deficient mutant (52D11) to translocate through colonic mucosa expressing CEACAMs was significantly impaired (Fig. 5E). Taken together, these results suggest the involvement of CEACAM molecules in AIEC translocation across gut mucosa in a type 1 pili-dependent manner.
Increased intestinal permeability has been observed in CD patients, and many studies have highlighted the major role played by tight junction proteins in intestinal homeostasis. Components of these TJ are down- or upregulated in CD ileal lesions.15, 25–31 Abnormal expression of CEACAM6 observed at the apical surface of the ileal epithelium in CD patients allows AIEC to colonize gut mucosa, leading to the development of inflammation.5, 24 We hypothesized that increased intestinal permeability in CD patients may results from the interaction between AIEC bacteria and CEACAM6 abnormally expressed on the ileal mucosa. Thus, we performed in vivo experiments to analyze whether CEACAMs overexpression and AIEC infection can result in abnormal intestinal permeability and disruption of barrier function.
To mimic abnormal expression of CEACAM6 observed in CD patients, we used the CEABAC10 transgenic mouse model expressing human CEACAMs.17 We showed that basal intestinal permeability and sensitivity to DSS-induced colitis were increased in CEABAC10 mice compared to WT mice. It was quite surprising that CEABAC10 mice had increased permeability without any clinical evidence of colitis. This can be explained by the observation that the colonic mucosa of CEABAC10 mice exhibit many abnormalities in epithelial appearance, such as hyperproliferation and differentiation blocking in colonic epithelium, which cause a decrease in goblet cell composition.32 We also hypothesized that CEACAMs molecules, which are intercellular adhesion molecules, may paradoxically weaken adhesion between adult colonocytes by interfering with other adhesion molecules situated between their lateral membranes. This could lead to a disruption of normal tissue architecture in the colonic epithelium.33 Our findings suggest a new role of CEACAMs in CD physiopathology, that of increasing basal intestinal permeability. In a subpopulation of CD patients, the primary defect may be abnormal intestinal epithelial barrier function leading to intestinal inflammation. Dysregulated epithelial barrier function precedes histologic evidence of ileitis in SAMP1/YitFc mice, a spontaneous model of IBD closely resembling CD in its histologic features and located at the terminal ileum.34 Given the high intestinal bacterial load in CD patients compared to controls,35 early barrier dysfunction gains special significance with regard to the pathogenesis of CD.
Immunohistological analysis of TMAs from CD patients and controls revealed a strong increase in claudin-2 expression in ileal biopsies of CD patients in quiescent phase compared to that in control biopsies. We observed that the level of claudin-2 abnormal expression was maximal and the same in the quiescent and active ileal mucosa of CD patients. In agreement with previous results, CEACAM6 was abnormally expressed in quiescent phase before the development of inflammation and significantly increased under inflammatory stimuli in the acute phase compared to the quiescent phase of CD.24 AIEC have been reported to be more prevalent in CD patients than in controls in studies from several countries.3, 7, 36–38 Since AIEC bacteria are able to colonize colonic mucosa of mice expressing human CEACAMs molecules, we hypothesized that one of the consequences of AIEC/CEACAM6 interaction in CD patients could be the induction of claudin-2 expression, leading to increased intestinal permeability preceding inflammation in predisposed patients abnormally expressing CEACAM6.9, 39, 40 Two independent studies reported in vitro that AIEC decreased the transepithelial resistance of the intestinal epithelial cell (IEC) monolayer and induced disorganization of F-actin and mislocalization of ZO-1 and E-cadherin.6, 7 Thus, we investigated the effect of AIEC infection on paracellular intestinal permeability in vivo using CEABAC10 mice model mimicking susceptibility of CD patients to AIEC infection. AIEC infection, but not type 1 pili AIEC negative mutant, significantly enhanced intestinal permeability only in CEABAC10 mice, suggesting that interactions between CEACAM molecule and type 1 pili are critical for AIEC to disrupt the intestinal barrier. Our previous studies showed AIEC LF82 translocation in CEABAC10 mice liver and spleen.5 In physiological conditions, immunostaining analysis of human CEACAM6 expression in CEABAC10 transgenic mice indicated that expression of CEACAM6 was restricted to the colonic mucosa (Fig. 5A), confirming the previously reported absence of ceacam6 mRNA in the ileal mucosa of CEABAC10 transgenic mice.17 In addition, in this mouse model challenged with AIEC bacteria, we previously observed an absence of ileal inflammation and colonization, which was explained by the lack of ileal CEACAM6 expression.5 We thus used this in vivo model of intestinal mucosa expressing human CEACAM6 to assess the consequence of the interaction between AIEC bacteria and human CEACAM6 in the colonic mucosa of transgenic mice. Here we report that AIEC LF82 was able to translocate only across a CEACAM-expressing proximal colon. This suggests that the interaction between CEACAMs molecule and AIEC bacteria on mucosal surface is essential in AIEC translocation. Interestingly, the increased intestinal permeability observed in AIEC LF82-infected CEABAC10 mice is similar to what was observed after 6 days of 1% DSS challenge. However, as observed at 3 days postinfection, increased intestinal permeability was not due to visible strong gut inflammation since a similar histological score was observed in 1% DSS-treated WT mice and AIEC-infected CEABAC10 mice and intestinal permeability was significantly different in the two conditions. Thus, the increased intestinal permeability in AIEC-infected CEABAC10 mice was probably due to alteration of the apical junctional complex functions. This is in good agreement with experiments using biotin tracer that have indicated a functionally deficient epithelial cell barrier in AIEC-infected CEABAC10 mice.
A specific distribution pattern of the tight junction proteins has been reported in IBD patients.15, 41 As observed in the quiescent phase in ileal mucosa of CD patients, AIEC infection in CEABAC10 mice led to a marked increase in pore-forming claudin-2 expression located at cell membrane in CEABAC10 colonic mucosa, before any signs of inflammation, which can enhance tight junction permeability for cation, thus providing a molecular basis for AIEC epithelial barrier disruption. Interestingly, E. coli K12 MG1655 is able to increase claudin-2 expression but its location remains cytoplasmic, suggesting a different mechanism in AIEC and nonpathogenic E. coli infection. Increased claudin-2 expression has been shown in an in vitro transfection system to increase epithelial permeability, possibly by forming lower-affinity interactions with other claudin isoforms on neighboring cells.42 In addition, it has recently been reported that EHEC infection of C57Bl/6J mice significantly increased claudin-2 expression and correlated with increased intestinal permeability.43 Of note, claudin-2 mRNA expression was more than 8-fold greater in SAMP than in WT ileum34, 44 and we observed that claudin-2 expression was much stronger in CD patients. The claudin-2 increased expression following AIEC infection in CEABAC10 mice may contribute to the defective barrier phenotype and a claudin-2-siRNA strategy could be used in vivo to target claudin-2 mRNA in intestinal epithelial cells expressing CEACAM6.
Increased intestinal permeability is associated with an increased relative risk of patients for relapses within 1 year.14, 45–50 Thus, understanding the mechanisms of barrier dysfunction and defects in permeability following AIEC infection has great potential for guiding the development of novel drugs for the treatment of IBD. Here we show that increased intestinal permeability induced by AIEC infection appears in early stages of the disease and before onset of strong gut inflammation. This is in good agreement with the observation that claudin-2 is strongly increased in quiescent phase in CD patients before any sign of inflammation.
In conclusion, findings from this study strongly demonstrate that preventing type 1 pili mediated AIEC/CEACAM6 interaction could prevent impairment of intestinal barrier function in CD patients. As AIEC/CEACAM6 interaction induces pore-forming claudin-2 expression in CEABAC10 mice, the development of new drugs targeting claudin-2 abnormal expression could be a new strategy to treat patients with abnormal ileal expression of CEACAM6 by preventing AIEC-induced disruption of mucosal integrity at an early stage of the disease.
We thank Dr. Abdelkrim Alloui for animal care (Animal facilities, Clermont-Ferrand, France); Ms. Céline Coelle for technical support for tissue microarrays and immunohistochemistry experiments; and Jeffrey Watts for native English editing.
- 50The mechanisms of prednisone inhibition of inflammation in Crohn's disease involve changes in intestinal permeability, mucosal TNFalpha production and nuclear factor kappa B expression. Aliment Pharmacol Ther. 2003; 18: 309–317., , , et al.