Intestinal Microbiota Shifts towards Elevated Commensal Escherichia coli Loads Abrogate Colonization Resistance against Campylobacter jejuni in Mice

Background The zoonotic pathogen Campylobacter jejuni is a leading cause of bacterial foodborne enterocolitis in humans worldwide. The understanding of immunopathology underlying human campylobacteriosis is hampered by the fact that mice display strong colonization resistance against the pathogen due to their host specific gut microbiota composition. Methodology/Principal Findings Since the microbiota composition changes significantly during intestinal inflammation we dissected factors contributing to colonization resistance against C. jejuni in murine ileitis, colitis and in infant mice. In contrast to healthy animals C. jejuni could stably colonize mice suffering from intestinal inflammation. Strikingly, in mice with Toxoplasma gondii-induced acute ileitis, C. jejuni disseminated to mesenteric lymphnodes, spleen, liver, kidney, and blood. In infant mice C. jejuni infection induced enterocolitis. Mice suffering from intestinal inflammation and C. jejuni susceptible infant mice displayed characteristical microbiota shifts dominated by increased numbers of commensal Escherichia coli. To further dissect the pivotal role of those distinct microbiota shifts in abrogating colonization resistance, we investigated C. jejuni infection in healthy adult mice in which the microbiota was artificially modified by feeding live commensal E. coli. Strikingly, in animals harboring supra-physiological intestinal E. coli loads, colonization resistance was significantly diminished and C. jejuni infection induced enterocolitis mimicking key features of human campylobacteriosis. Conclusion/Significance Murine colonization resistance against C. jejuni is abrogated by changes in the microbiota composition towards elevated E. coli loads during intestinal inflammation as well as in infant mice. Intestinal inflammation and microbiota shifts thus represent potential risk factors for C. jejuni infection. Corresponding interplays between C. jejuni and microbiota might occur in human campylobacteriosis. Murine models introduced here mimick key features of human campylobacteriosis and allow for further analysis of immunological and molecular mechanisms of C. jejuni – host interactions.


Introduction
Campylobacter (C.) jejuni continues to be the most frequently reported causative agent of bacterial gastroenteritis in humans worldwide [1,2]. The gram-negative bacteria form part of the commensal flora in many wild and domestic animals. Zoonotic transmission from livestock animals via consumption of contaminated meat products is the most common source for human C. jejuni infections [3,4,5]. Infected patients present a broad range of symptoms including abdominal pain, watery or bloody diarrhea, and fever. Histological examination of affected intestinal tissues commonly reveals crypt abscesses, ulcerations, and elevated numbers of immune cells in the colon in situ [6,7,8,9]. Although the infection is self-limiting in most cases, long-term sequelae such as reactive arthritis, gastrointestinal and neurological disorders have been reported in susceptible hosts [10,11]. C. jejuni infection correlates with reactivation of inflammatory bowel diseases (IBD), Miller-Fisher-Syndrome and Guillian-Barré-Syndrome [10,11]. The high prevalence of campylobacteriosis poses a significant health burden and accentuates the need for further unravelling the underlying pathogenesis of C. jejuni infections. The lack of understanding mainly stems from the scarcity of suitable experimental in vivo models of human infection [12]. Although chickens allow for the investigation of colonization and transmis-sion, intestinal disease manifestations such as enteritis and diarrhea are absent [12,13]. Murine models of C. jejuni infection have the disadvantage of sporadic colonization, absence of clinical signs of disease and/or lack of intestinal immunopathology [6,12]. This is in part due to the colonization resistance against C. jejuni exhibited by adult mice older than two months (commonly used in experimental settings) harboring a conventional intestinal microbiota [12]. Results from the 1980s demonstrated that colonization resistance was abrogated in isolator raised germfree (GF) mice. In these animals C. jejuni colonized the entire gastrointestinal (GI) tract and induced clinical signs of enterocolitis including granulocyte infiltrates, bloody diarrhea, and humoral immune responses [14,15,16]. More recent investigations showed that colonization resistance is mainly caused by the host-specific intestinal microbiota composition [17,18]. Previous work in our group revealed that colonization resistance of mice against C. jejuni is abrogated in gnotobiotic (GB) mice generated by quintuple antibiotic treatment as well as in GB mice replenished with human flora (hfa) [17]. In contrast to isolator raised germfree mice, however, GB and hfa animals display a fully established immune system. The finding that GB mice replenished with murine flora were completely resistant to C. jejuni proved that colonization resistance is caused by the host-specific microflora composition [17]. Most strikingly, C. jejuni infected GB and hfa mice displayed characteristic inflammatory responses seen in human campylobacteriosis [17]. The major role of the microflora composition in colonization resistance against C. jejuni was further supported by the recent finding that obesity caused by both feeding human ''Western style'' cafeteria diet (CAF) as well as leptin deficiency in ob/ob 2/2 mice rendered mice susceptible to C. jejuni infection [18]. Interestingly, quantitative analysis of main gut bacterial groups revealed 2-3 orders of magnitude higher loads of commensal E. coli in C. jejuni susceptible animals in all these models as a common feature as compared to respective resistant controls [17,18]. Recent studies on Salmonella enterica also point towards a pivotal role of microbiota composition in host susceptibility to enteropathogenic bacteria [19,20]. Mice displaying high commensal E. coli densities were more susceptible to S. enterica induced gut inflammation [19,20,21]. We have previously shown that murine models of acute small as well as large intestinal inflammation were associated with tremendous changes of the intestinal microbiota with decreasing species diversity and lactobacilli loads and an intestinal overgrowth with commensal E. coli and Bacteroides/Prevotella spp. [22,23,24,25]. We could further show that TLR-4-signalling of E. coli LPS initiated and perpetuated acute ileitis and colitis in mice [23,24]. Thus, these experimental models were mimicking immunopathological key features of human IBD Crohn's disease (CD) and ulcerative colitis (UC). The exact etiology of IBD remains unknown and the pathogenesis seems to be multifactorial, involving the complex interaction between the intestinal microflora, host immune and genetic factors as well as environmental stimuli [26,27], Gramnegative bacteria such as E. coli and Bacteroides spp. have been reported to accumulate at inflamed tissue sites in IBD patients with active intestinal inflammation, and further potentiate immunopathology by translocation via microlesions and ulcerations [28,29]. It is under current debate whether IBD patients might be more susceptible to intestinal pathogens such as Campylobacter or Salmonella spp. and whether acute bacterial enterocolitis is one of the factors inducing or exacerbating IBD in susceptible individuals [26,27,30]. Results from population-based studies point towards a potential role of enteric Campylobacter or Salmonella infections in IBD development [31,32,33].
In the present study, we investigated colonization resistance in mice displaying an altered gut microbiota composition under conditions of acute or chronic intestinal inflammation as well as in infant mice after weaning (lacking signs of intestinal inflammation) having all elevated intestinal E. coli loads in common. Finally we asked the question whether feeding healthy adult mice with live E. coli and thus artificially elevating the enterobacteria population within the commensal gut microbiota towards supra-physiological loads might overcome colonization resistance and renders the animals susceptible to C. jejuni. The results clearly demonstrate that colonization resistance is completely abrogated under conditions of elevated E. coli loads within the commensal intestinal microbiota. Following infection, C. jejuni induced pro-inflammatory immune responses mimicking key features of human campylobacteriosis. Taken together, the murine C. jejuni infection models presented here proved to be suitable for reproducible investigations of factors that predispose C. jejuni colonization and trigger the onset and progress of infection and immunopathology.

T. gondii-induced Acute Ileitis Abrogates Colonization Resistance of Mice Against C. jejuni Infection
Colonization resistance against C. jejuni exerted by mice harbouring a conventional intestinal microbiota can be overcome by modification of the gut microbiota composition. We therefore investigated C. jejuni infection capacities in mice with acute or chronic intestinal inflammation displaying distinct microbiota shifts towards higher intestinal enterobacteria loads. Three-months-old adult C57BL/6 mice harboring a conventional gut microbiota were infected perorally with T. gondii in order to induce acute ileitis. Four days thereafter, the time point when initial histopathological changes in the ileum mucosa can be observed, mice were orally subjected to 10 9 CFU C. jejuni ATCC 43431 by gavage. At day 7 following T. gondii infection, mice suffered from acute ileitis and were prone to death. Interestingly, by that time (i.e. 3 days following C. jejuni ATCC 43431 infection) C. jejuni ATCC 43431 could be cultured from the entire GI tract of diseased, but not healthy mice with high loads of 10 9 CFU/g in duodenum, ileum and colon (Fig. 1A). Notably, the E. coli concentrations in the ileum lumen had increased by more than 6 orders of magnitude up to more than 10 10 CFU/g during ileitis development (Fig. 1B). Live C. jejuni ATCC 43431 and commensal E. coli translocated into MLNs draining the small intestine (10060% and 78.667.1%, respectively), spleen (57.161.49% and 50.060.68%, respectively) and liver (57.160% and 42.960%, respectively) (Fig. 1C). Strikingly, in 92.960.68%, but only in 7.160.7% of mice with acute ileitis following T. gondii infection, live C. jejuni ATCC 43431 and E. coli, respectively, could be cultured from cardiac blood indicating C. jejuni bacteremia (Fig. 1C). A molecular survey of the ileum microbiota composition revealed that concentrations of total eubacteria, enterobacteria, enterococci and Bacteroides/ Prevotella spp. were increased in T. gondii-induced acute ileitis, whereas the lactobacilli population was lower as compared to healthy mice (Fig. 1D). Thus, during acute small intestinal inflammation accompanied by distinct intestinal microbiota shifts towards E. coli overgrowth in the GI tract colonization resistance against C. jejuni is abrogated with hazardous consequences for the host, as indicated by translocation and systemic spread of the pathogen. IL-10 Deficient Mice with Chronic Colitis are Susceptible to C. jejuni Colonization Next, we were interested whether colonization resistance against C. jejuni is also abrogated in mice with chronic intestinal inflammation accompanied with gut microbiota shifts towards higher commensal enterobacteria loads [34]. To address this question we infected 6-months-old IL-10 2/2 mice with chronic colitis harboring a conventional gut microbiota with C. jejuni ATCC 43431. Twelve days following oral infection, C. jejuni could be detected in the ileum and colon (in 55.2% and 69.0% of infected mice, respectively), but not in the stomach or duodenum of IL-10 2/2 mice with colitis, whereas healthy wildtype control animals remained free of C. jejuni ( Fig. 2A). Furthermore, the median C. jejuni loads in infected IL-10 2/2 mice were more than 4 orders of magnitude lower (10 5 CFU/g luminal content) when compared to mice with T. gondii-induced acute ileitis (Fig. 1A,  2A). Of note, uninfected IL-10 2/2 mice with colitis harbored 4 orders of magnitude higher E. coli counts in their large intestines when compared to healthy wildtype mice (Fig. 2B). Interestingly, live C. jejuni could be cultured only from less than 5% (3 out of 67) of MLNs derived from infected IL-10 2/2 mice, whereas no C. jejuni at all translocated into spleen, liver, kidney or cardiac blood at day 12 p.i. (not shown). A molecular survey of the main bacterial groups within the colonic lumen revealed that IL-10 2/2 mice with colitis harbored significantly higher numbers of enterobacteria, Bacteroides/Prevotella spp. and bifidobacteria, whereas lactobacilli, Clostridium leptum group and mouse intestinal bacteroidetes rRNA were lower as compared to wildtype controls, (Fig. 2C). Thus, colonization resistance against C. jejuni is abrogated by chronic large intestinal inflammation in IL-10 2/2 mice accompanied by distinct changes in colonic microbiota composition.

C. jejuni Infection of Infant Mice Harboring a Conventional Intestinal Microbiota
In earlier studies weaned mice harboring a conventional gut microbiota were shown to be susceptible to C. jejuni infection [35,36,37,38,39]. Very recently, we could demonstrate that compared to adult animals, infant mice harbored significantly higher concentrations of E. coli, Bacteroides/Prevotella spp. and Clostridium coccoides whereas lactobacilli, bifidobacteria and Clostridium leptum numbers were lower [40]. To further unravel immunopathological consequences of abrogated colonization resistance in infant mice in more detail, we infected 3-weeks-old animals right after weaning with C. jejuni strains ATCC 43431 or B2. Adult mice 3 months of age with a common commensal gut flora served as controls. Kinetic analyses of fecal pathogen burden revealed that the infant mice harbored higher C. jejuni ATCC 43431 and B2 loads until days 5 and 6 p.i., respectively, when compared to adult animals (Fig. 3A, B). Interestingly, C. jejuni strain B2, but not strain ATCC 43431, could stably infect infant mice until day 6 p.i., as indicated by 66.7% and 30.0% of culturepositive cases, respectively, with a median load of approximately 10 5 CFU C. jejuni B2/g feces (Fig. 3A, B). Feces samples taken from naïve infant mice contained significantly higher numbers of E. coli (median 10 6 CFU/g), as compared to uninfected adult animals (Fig. 3C). Thus, these data provide evidence that E. coli might be causatively involved in the abrogation of murine colonization resistance against C. jejuni.
The analysis of the clinical outcome revealed that acute campylobacteriosis was self-limiting in infant mice. A cumulative score addressing the clinical condition and occurrence of bloody diarrhea in infected animals revealed that either C. jejuni strain induced enterocolitis within 24 hours p.i. remaining on a comparable level until day 6 p.i. (Fig. 3D). C. jejuni infected infant mice were not severely compromised clinically but developed mild diarrhea with haemoccult-positive feces irrespective of the C. jejuni strain analyzed. Furthermore, no live C. jejuni could be cultured from MLNs, spleen, liver, kidney or cardiac blood at day 6 following infection with either strain (not shown).

Pro-inflammatory Immune Responses Following C. jejuni Infection in Infant Mice
Since human campylobacteriosis is accompanied by an influx of immune cells to the inflamed colon [6,7,8,9], we investigated numbers of apoptotic cells and immune cells within the colon mucosa of C. jejuni infected infant mice in situ by immunohistochemistry. Naive infant mice served as controls. At day 6 p.i., irrespective of the C. jejuni strain used, infected infant mice displayed a two-fold increase of apoptotic cells, paralleled by a significant increase of neutrophilic granulocytes and macrophages as compared to naive controls ( Fig. 4A-C). In addition, both C. jejuni strains induced multifold increases in T-lymphocytes, regulatory T-cells (Tregs) as well as B-lymphocytes in infected versus naïve infant mice (Fig. 4D-F). The immune cell responses were accompanied by increased expression of pro-inflammatory cytokines and mediators in the colon. The levels of nitric oxide, TNF-a, IL-6, and MCP-1 were significantly higher in colon biopsies derived from C. jejuni ATCC 43431 and B2 infected mice at day 6 p.i. as compared to naive controls ( Fig. 5A-D). For IFNc a significant increase was measured in C. jejuni B2 infected mice while C. jejuni ATCC 43431 infected mice displayed a trend towards elevated concentrations (Fig. 5E). Of note, following infection with either C. jejuni strain, adult mice harboring a conventional gut microbiota displayed neither clinical signs of enterocolitis nor relevant immune cell or pro-inflammatory responses in their colon (also refer to [17]). Taken together, C. jejuni infection of infant mice resulted in distinct pro-inflammatory responses and apoptotic changes in the colonic mucosa.
Treatment with Live E. coli Abrogates Murine Colonization Resistance Against C. jejuni In order to dissect whether colonization resistance is abrogated by shifts in the microbiota composition or by conditions of inflammation we increased intestinal E. coli loads artificially by treating healthy adult mice with live E. coli in the drinking water (10 8 CFU/mL) and investigated the susceptibility to C. jejuni infection. Analysis of E. coli in feces samples revealed that the intestinal E. coli loads increased by more than 3 orders of magnitude (from median 10 5 CFU/g to 10 8 CFU/g feces) and remained stable as long as the E. coli suspension was kept (Fig. 6). Within 24 hours after withdrawl of the E. coli suspension, the intestinal E. coli loads decreased back to basal levels (Fig. 6). Oral infection of E. coli-treated mice and non-treated controls with C. jejuni strains ATCC 43431 or B2 revealed that E. coli did significantly increase the colonization capacity of either strain. (Fig. 7A and B). C. jejuni B2 could be cultured from the entire GI tract of E. coli treated animals at day 12 p.i. with highest counts in distal parts such as ileum and colon (Fig. 7C). In non-treated control animals, however, pathogen burdens were significantly lower as compared to mice with elevated intestinal E. coli load (Fig. 7).

Pro-inflammatory Immune Responses in C. jejuni Infected Adult Mice with Elevated E. coli Loads
To further elucidate whether C. jejuni infection of adult mice with high intestinal E. coli loads was accompanied by proinflammatory immune responses, we analyzed apoptosis and inflammatory cells in the colon mucosa by in situ immunohistochemistry. Irrespective of the C. jejuni strain analyzed, infected mice displayed significantly higher numbers of apoptotic cells, neutrophils, macrophages, T-and B-lymphocytes as well as regulatory T-cells in the colon in situ at day 12 p.i. as compared to non-infected E. coli-treated control animals (Fig. 8).
Secretion of nitric oxide, IL-6, MCP-1, IFN-c and TNF-a in ex vivo colon biopsies were significantly increased at day 12 following infection with C. jejuni ATCC 43431 or B2 versus naïve animals with supra-physiological intestinal E. coli loads (Fig. 9). Thus, C. jejuni infection of adult mice treated with supraphysiological E. coli loads is accompanied by pro-inflammatory immunopathology of the colon. Of note, elevation of intestinal E. coli loads alone did not affect pro-inflammatory immune responses (data not shown). This was even true in our recent study when gnotobiotic mice generated by antibiotic treatment were replenished with a complex human microbiota [17].
Taken together, physiological (infancy) and pathological conditions (e.g. acute and chronic intestinal inflammation in mice and men) associated with shifts of the intestinal microbiota composition such as elevated commensal E. coli loads facilitate C. jejuni infection inducing or perpetuating immunopathology. Most strikingly, the physiological CR against C. jejuni can be abrogated by elevating a single bacterial commensal gram-negative species such as E. coli within the complex commensal intestinal microbiota towards supra-physiological levels. These results point towards a pivotal role of E. coli in facilitating C. jejuni infection.

Discussion
Investigations of the molecular mechanisms underlying C. jejuni induced immunpathology in vivo are hampered by colonization resistance of mice due to their host-specific intestinal microbiota composition. In the present study we investigated the susceptibility of an initially resistant host to C. jejuni during acute and chronic intestinal inflammation. In our acute T. gondii-induced ileitis model colonization resistance against C. jejuni was completely abolished and the pathogen colonized the entire GI tract with high counts. Moreover, C. jejuni and commensal E. coli translocated into MLNs, spleen, and liver. Strikingly, C. jejuni could be cultured from cardiac blood samples in .90% of infected animals displaying severe ileitis indicating bacteremia and subsequent sepsis. The impact of intestinal inflammation in overcoming colonization resistance was further supported by the finding that IL-10 2/2 mice suffering from chronic colitis were susceptible to C. jejuni infection. This is well in line with previous findings of Mansfield et al. showing that C. jejuni induced significant clinical signs of enterocolitis following infection of IL-10 2/2 mice [41]. However, in the IL-10 2/2 mice analyzed here the clinical signs of enterocolitis following C. jejuni infection remained rather subtle. The reason for this discrepancy in colitis severity is most likely attributed to differences in the genomic contents of the C. jejuni strains used.
The intestinal inflammation models exerting a compromised colonization resistance against C. jejuni had all in common significantly higher E. coli loads at inflamed tissue site, for instance, when compared to healthy controls with an intact colonization resistance, irrespective whether the inflammatory condition was acute or chronic, located in the small or large intestine. These findings are well in line with results from previous studies pointing towards a weakness of the physiological barrier during intestinal inflammation [19,20,21,42]. Bacteremia and sepsis display hazardous sequelae of C. jejuni infection in hosts with a compromized epithelial barrier function (e.g. Crohn's disease patients with acute episodes). A high prevalence of C. concisus was recently detected in paediatric CD as well as adult UC patients [43,44]. Although distinct from IBD, irritable bowel syndrome was described to occur as a potential post-infectious event following C. jejuni infection [45]. Taken together, these findings once more point towards a correlation between inflammation, intestinal microbiota and susceptibility to C. jejuni infection.
Given that colonization of the intestinal tract starts right after birth and complexity increases thereafter, the question arises whether infant mice already display a fully established colonization resistance. Neonatal mice had been used in Campylobacter research already three decades ago, investigating transmissibility from infant to dam, virulence factors, vaccines and the efficacy of bismuth subsalicylate in preventing the growth and colonization of the intestinal tract [35,36,37,39,46]. Remarkably, infection of 3-weeks-old mice right after weaning in our study revealed that C. jejuni B2, and to a less distinct extent strain ATCC 43431, could stably colonize the GI tract of mice. Moreover, infant mice displayed approximately 2-3 orders of magnitude higher E. coli counts in their feces as compared to 3-months-old adult animals. Thus, there is evidence that an altered microbiota towards higher enterobacteria numbers (such as E. coli) might facilitate C. jejuni infection. When assessing C. jejuni induced clinical pathology due to blood- positive stool samples and mild diarrhea in the first week following infection, the clinical outcomes were similar in infant mice infected with C. jejuni strains B2 or ATCC 43431. Both strains originate from humans with severe enteritis, are invasive in vitro and induced distinct pro-inflammatory immune responses within the colon mucosa of C. jejuni infected infant mice. The elevated intestinal E. coli loads in mice suffering from intestinal inflammation and in infant mice provide evidence that E. coli is essentially involved in abrogation of CR. This is further underlined by a substantially increased colonization capacity of C. jejuni in adult mice in which the commensal intestinal microbiota was artificially modified towards supra-physiological intestinal E. coli loads. This finding demonstrates again that CR against C. jejuni is mediated by the microbiota and that shifts in the flora represent a risk for C. jejuni infection.
We conclude that the murine models of C. jejuni infection presented here shed further lights into the mechanisms underlying CR against C. jejuni offering promising tools to further unravel the interplay between intestinal pathogens, gut microbiota and the innate and adaptive immune system.

Ethics Statement
Animal experiments were conducted according to the European Guidelines for animal welfare (2010/63/EU) with approval of the commission for animal experiments headed by the ''Landesamt für Gesundheit und Soziales'' (LaGeSo, Berlin, Germany; Registra-tion numbers: G0173/07 and G135/10). Animal welfare was monitored twice daily by assessment of clinical conditions.

Mice
Mice were bred and maintained in the facilities of the ''Forschungsinstitut für Experimentelle Medizin'' (FEM, Charité -Universitä tsmedizin, Berlin, Germany), under specific pathogen- Figure 9. Colonic pro-inflammatory cytokine responses of C. jejuni infected adult mice with increased E. coli loads. In conventionally colonized 3-months-old adult mice intestinal E. coli loads were raised up to 10 8 bacteria per gram feces by adding live E. coli into the drinking water (+ E. coli) as described in Methods. Then, mice were orally infected with C. jejuni strain ATCC 43431 (crossed circles) or B2 (solid circles). Concentrations (pg per mg colon) of (A) nitric oxide, (B) IL-6, (C) MCP-1, (D) IFN-c, and (E) TNF-a were determined in supernatants of ex vivo colon cultures at day 12 p.i. by Griess reaction or CBA-ELISA, respectively. Uninfected mice (Naïve, open circles) served as negative controls. Numbers of analyzed animals are given in parentheses. Means (black bars) and levels of significance (P-values) as compared to the uninfected (Naïve) group determined by Student's t-test are indicated. Data shown were pooled from at least three independent experiments. doi:10.1371/journal.pone.0035988.g009 free (SPF) conditions. C57BL/6 mice were used in the experiments. Age matched female mice 3 weeks (right after weaning; infant mice) and 3 months of age (adult mice) as well as 6 months old IL-10 2/2 mice (C57BL/6 background) with overt chronic colitis were used.

C. jejuni Infection of Mice
Mice were infected with approximately 10 9 viable CFU of C. jejuni strains ATCC 43431 or B2 by gavage in a total volume of 0.3 mL PBS on two consecutive days as described earlier in detail [17]. C. jejuni strain B2 was kindly provided by Prof. Dr. Uwe Groß, University of Göttingen, Germany.

Elevation of Intestinal E. coli Loads
In order to raise intestinal E. coli loads in 3-months-old adult mice harboring a conventional gut microbiota to supra-physiologic levels, live E. coli were isolated from naïve, healthy 3 months old C57BL/6 mice by culture on MacConkey media. This strain was shown to be a common commensal in healthy mice. PCR-based detection revealed that this strain does not contain known virulence factors of pathogenic E. coli such as stx 1 and 2, catA, hlyA, cspA, katP, and astA. This commensal E. coli strain was subcultivated on blood agar and added to the drinking water at a final concentration of 10 8 CFU per mL suspension starting at day 0 in naïve animals. The E. coli suspension was renewed daily. For kinetic experiments, the E. coli suspension was withdrawn after 8 days and replaced by regular tap water thereafter. For C. jejuni infection studies, the application of E. coli in drinking water was started 8 days prior the first oral C. jejuni infection and continued until the end of the experiment (12 days following the latest C. jejuni gavage).

Sampling Procedures, Determination of Colon Length and Histopathology
Mice were sacrificed by isofluran treatment (Abbott, Germany). From each mouse samples derived from the GI tract (stomach, duodenum, ileum, colon) were collected in parallel for histological, microbiological, immunobiological and molecular analyses. Cardiac blood and tissue samples from mesenteric lymphnodes, spleen, liver and the GI tract were asserved under sterile conditions. Histopathological changes were determined in colon samples immediately fixed in 5% formalin and embedded in paraffin. Sections (5 mm) were stained with respective antibodies for immunohistochemistry.

Cytokine Detection in Colon Culture Supernatants
Colon biopsies were cut longitudinally, washed in PBS and strips of 1 cm 2 placed in 24-flat-bottom well culture plates (Nunc, Wiesbaden, Germany) containing 500 ml serum-free RPMI 1640 medium supplemented with penicillin (100 U/ml) and streptomycin (100 ug/ml; PAA Laboratories). After 18 h at 37uC, culture supernatants were tested for TNF-a, IL-6, MCP-1 and IFN-c by the Mouse Inflammation Cytometric Bead Assay (CBA; BD Biosciences) on a BD FACSCanto II flow cytometer (BD Biosciences). Nitric oxide (NO) was determined by Griess-reaction as described earlier [22].

Analysis of the Intestinal Microflora
Cultural analyses, biochemical identification and molecular detection of luminal bacterial communities from stomach, duodenum, ileum, and colon as well as feces were performed as previously described [17,22,23,24].

Assessment of Bacterial Translocation Frequencies
For bacterial translocation experiments, mesenteric lymphnodes (MLNs) draining the small intestine, spleen, liver (1 cm 2 ) and cardiac blood (1 mL) were transferred into thioglycollate enrichment broths under sterile conditions and cultivated for 7 days. Turbid broths were streaked onto sheep blood, MacConkey as well as Karmali agar in order to detect translocated live E. coli and C. jejuni followed by biochemical species identification (API 20E and API Campy, respectively, Biomérieux). Given that bacterial growth in enrichment broths cannot be quantitated, the relative rates of translocating C. jejuni or E. coli into the respective extraintestinal compartment are expressed in % (number of broths culture-positive for either E. coli or C. jejuni multiplied with 100 and divided by the total number of broths inoculated with the respective organ or blood).

Statistical Analysis
Mean values, medians, standard deviations, and levels of significance were determined using appropriate tests as indicated (two-tailed Student's t-Test, Mann-Whitney-U Test). Two-sided probability (P) values #0.05 were considered significant. All experiments were repeated at least twice.