Pre-treatment with Bifidobacterium breve UCC2003 modulates Citrobacter rodentium-induced colonic inflammation and organ specificity

Citrobacter rodentium, which colonizes the gut mucosa via formation of attaching and effacing (A/E) lesions, causes transmissible colonic hyperplasia. The aim of this study was to evaluate whether prophylactic treatment with Bifidobacterium breve UCC2003 can improve the outcome of C. rodentium infection. Six-week-old albino C57BL/6 mice were pre-treated for 3 days with B. breve, challenged with bioluminescent C. rodentium and administered B. breve or PBS-C for 8 days post-infection; control mice were either administered B. breve and mock-infected with PBS, or mock-treated with PBS-C and mock-infected with PBS. C. rodentium colonization was monitored by bacterial enumeration from faeces and by a combination of both 2D bioluminescence imaging (BLI) and composite 3D diffuse light imaging tomography with µCT imaging (DLIT-µCT). At day 8 post-infection, colons were removed and assessed for crypt hyperplasia, histology by light microscopy, bacterial colonization by immunofluorescence, and A/E lesion formation by electron microscopy. Prophylactic administration of B. breve did not prevent C. rodentium colonization or A/E lesion formation. However, this treatment did alter C. rodentium distribution within the large intestine and significantly reduced colonic crypt hyperplasia at the peak of bacterial infection. These results show that B. breve could not competitively exclude C. rodentium, but reduced pathogen-induced colonic inflammation.


INTRODUCTION
Enterohaemorrhagic Escherichia coli (EHEC) is an extracellular zoonotic intestinal pathogen that produces Shiga toxin (Stx), and was responsible for over 1034 human infections in England andWales in 2009 (Health Protection Agency, 2011). Aside from causing acute gastrointestinal infections, EHEC can cause severe clinical disease syndromes such as haemorrhagic colitis and haemolytic uraemic syndrome (HUS) in humans . Treatment of EHEC infections with antibiotics such as trimethoprim/sulfamethoxazole and gentamicin has been demonstrated to increase Stx production and may increase the incidence of HUS (Dundas et al., 2005;McGannon et al., 2010;Panos et al., 2006). However, to date, no suitable therapy or intervention strategy exists for EHEC infections.
Probiotics have been demonstrated as an intervention strategy to prevent the colonization of humans and mice with pathogenic enteric micro-organisms in a process termed competitive exclusion (CE) (Bernet-Camard et al., 1997;Chen et al., 2005;Corr et al., 2007;Fanning et al., 2012;Preidis et al., 2011;Wu et al., 2008). CE is a process in which commensal bacterial species are used to outcompete invading pathogenic micro-organisms through a variety of mechanisms including the production of antimicrobial compounds (Bernet-Camard et al., 1997;Corr et al., 2007), competition for receptor sites on the gastrointestinal mucosa (Chen et al., 2007;Servin & Coconnier, 2003), competition for nutrients (Fuller, 1992) and interference with quorum-sensing signals (Medellin-Peña et al., 2007).
Probiotics are defined as live micro-organisms which, when administered in an adequate amount, provide a health benefit to the host (FAO, 2001). Historically, probiotics are from the genera Lactobacillus and Bifidobacterium (Felis & Dellaglio, 2007). Bifidobacteria are Gram-positive, obligate anaerobes that colonize the large intestines of humans and mice, and are among the most widely used probiotic bacteria (Picard et al., 2005). To date, significant understanding has been gained regarding the use of probiotics to prevent gastrointestinal infections (Huebner & Surawicz, 2006;Servin & Coconnier, 2003;Servin, 2004). However, the underlying mechanisms in the majority of these studies remain to be elucidated.
The mouse pathogen Citrobacter rodentium utilizes a type III secretion system (T3SS) to colonize the human intestinal mucosa via the formation of attaching and effacing (A/E) lesions, and is used as a small animal model of human infection with EHEC (Mundy et al., 2005). C. rodentium infection induces transmissible colitis and colonic epithelial cell hyperplasia, and results in a self-limiting disease in C57BL/6 mice which induces sterilizing immunity, preventing reinfection (Mundy et al., 2005).
The C. rodentium infection model has been widely adopted to study how probiotics can be used to treat gastrointestinal infections (Chen et al., 2005(Chen et al., , 2009D'Arienzo et al., 2006;Fanning et al., 2012;Gareau et al., 2010;Johnson-Henry et al., 2005;Jones & Knight, 2012;Rodrigues et al., 2012;Wu et al., 2008). Recently, Fanning et al. (2012) demonstrated that Bifidobacterium breve reduces C. rodentium colonization in a BALB/c infection model, and that this protective effect is dependent on the production of an extracellular polysaccharide. In addition, the probiotic yeast Saccharomyces boulardii has been demonstrated to reduce C. rodentium colonization by modulating T3SS expression (Wu et al., 2008). In contrast, pre-treatment of neonatal and adult mice with individual probiotic strains resuspended in PBS has been shown to reduce the intestinal inflammation associated with C. rodentium infection in a manner independent from reduced pathogen colonization (Chen et al., 2009;Gareau et al., 2010;Johnson-Henry et al., 2005;Wu et al., 2008).
Bioluminescence imaging (BLI) is widely used in infectious disease research to monitor the colonization of mice with pathogenic bacteria (Contag et al., 1995;Doyle et al., 2004;Hardy et al., 2004;Wiles et al., 2004) and to assess intervention strategies for bacterial infections including probiotics (Corr et al., 2007;Fanning et al., 2012). Recombinant micro-organisms expressing the bacterial luciferase operon luxCDABE from Photorhabdus luminescens can be detected non-invasively and monitored longitudinally during an infection through light production (Hardy et al., 2004;Wiles et al., 2004Wiles et al., , 2005. Importantly, as bioluminescence (BL) is an energy-dependent process, only live, metabolically active micro-organisms are detected (Szittner & Meighen, 1990). However, standard BLI is limited because it is not possible to determine the exact location of the BL foci in vivo; instead, localization of the BL signal is inferred from the surface of the animal where the signal is emitted, or through ex vivo analysis of the infected organs (Contag et al., 1995). In contrast, 3D BLI, known as diffuse light imaging tomography (DLIT), is performed by collecting BL images taken using different optical filters in the range of 500-620 nm for imaging of bacterial luciferase. The spectrally filtered BL is then used to reconstruct the BL source, location and intensity, resulting in a quantitative 3D reconstruction of the BL signal (Kuo et al., 2007). These BLI data can then be co-registered with a mCT scan of the entire mouse to give a detailed anatomical localization of the BL source in a technique known as DLIT-mCT.
In this study we used DLIT-mCT and BLI, combined with light and electron microscopy, to determine whether the prophylactic treatment of mice with B. breve UCC2003 could impact on C. rodentium infection.
B. breve UCC2003  was grown statically at 37 uC under anaerobic conditions (BBL GasPak EZ system) in MRS medium supplemented with 0.05 % cysteine HCL (MRS-C).
Mice. Pathogen-free female 18-20 g, 6-8-week-old albino C57Bl/6 mice were purchased from Charles River. All mice were housed in individually filtered cages with sterile bedding and with sterilized food and water ad libitum. All animal experiments were performed in accordance with the Caliper IACUC Ethical Review Committee. Two independent infection experiments were performed with six mice per group.
Daily treatment of mice with B. breve or PBS. Mice were inoculated for 3 days prior to C. rodentium infection and for a further 8 days post-infection (p.i.) by oral gavage with 200 ml of overnight B. breve UCC2003, which was resuspended in PBS supplemented with 0.05 % cysteine HCL (PBS-C) at a cell density of approximately 2610 9 c.f.u. (Cronin et al., 2010). Uninfected mice were gavaged with either B. breve UCC2003 or PBS-C as controls.
Oral infection of mice C. rodentium challenge. At 2 PM on day 10 post initiation of B. breve UCC2003 treatment, mice were inoculated by oral gavage with 200 ml of overnight LB-cultured C. rodentium, which was resuspended in PBS prior to infection at a cell density of approximately 5610 9 c.f.u., and uninfected mice were gavaged with PBS as a control. The numbers of viable bacteria in the inoculum were determined by serial dilution in PBS and spotting in triplicate onto LB agar supplemented with kanamycin (50 mg ml 21 ). Colonization was monitored by the collection of faeces from mice at day 7 p.i. and the numbers of viable bacteria per gram of faeces were enumerated. At day 8 p.i., the mice were euthanized by cervical dislocation, and colonic tissues were collected for microscopic analysis as outlined below.
Collection of samples, sample fixation and histopathology.
Segments of terminal colon from each mouse were collected postmortem from one experiment at day 8 p.i. Tissues were subsequently rinsed of their contents and fixed in 10 % buffered formalin for Imaging of C. rodentium in probiotic-treated mice rodentium infection from three representative mice per test condition at days 4 and 8 p.i. (c) Quantification of BLI (p s "1 cm "2 sr "1 ; p5photons) from all six mice per treatment condition at day 4 p.i. (d) Quantification of BLI (p s "1 cm "2 sr "1 ) from all six mice per test condition at day 8 p.i. microscopic examination. Additional colonic segments were fixed in 2.5 % glutaraldehyde for further electron microscopy analysis.
Measurement of crypt hyperplasia. Formalin-fixed tissues were then processed, paraffin-embedded, sectioned at 5 mm, and stained with haematoxylin and eosin (H&E) using standard techniques. H&E-stained tissues were evaluated for crypt hyperplasia microscopically without knowledge of the treatment condition used in the study, and the length of at least 20 well-oriented crypts from each section from all of the mice per treatment group (n56) was evaluated. H&E-stained tissues were imaged with an Axio Lab.A1 microscope (Carl Zeiss MicroImaging), and images were acquired using an AxioCam ERc 5s colour camera and computer processed using AxioVision (Carl Zeiss MicroImaging).
Histological damage score. Histological damage scoring was determined using criteria outlined by Wu et al. (2008). In brief, H&E tissue sections prepared as described above were assessed for the following damage and graded accordingly: severity of epithelial injury (0-3, from absent, to mild-superficial epithelial injury, and severe including multifocal erosions); the extent of inflammatory infiltrate (0-3, from absent to transmural); and goblet cell depletion (0-2, from absent to partial, complete). Five non-overlapping fields of view from one representative tissue section were graded from all of the six mice per treatment group and averaged to obtain a mean histological score.
Indirect immunofluorescence assay (IFA) on mouse colon sections. Indirect immunofluorescence was performed on formalinfixed paraffin-embedded (FFPE) sections and a rabbit polyclonal anti-Citrobacter antibody (gift from Simon Claire, Wellcome Trust Sanger Institute) was used to visualize C. rodentium. DNA from bacterial and intestinal epithelial cells was counterstained with Hoechst 33342. Sections were examined using an Axio Imager M1 microscope (Carl Zeiss MicroImaging), and images were acquired using an AxioCam MRm monochrome camera and computer processed using AxioVision (Carl Zeiss MicroImaging).
Electron microscopy. Additional murine colonic tissues infected as described above were processed for electron microscopy, as previously described (Girard et al., 2007). Samples for scanning electron microscopy (SEM) were examined without knowledge of the strain used, at an accelerating voltage of 25 kV using a JEOL JSM-5300 scanning electron microscope [JEOL (UK)]. Samples for transmission electron microscopy (TEM) were observed using a Tecnai 12 transmission electron microscope at an accelerating voltage of 120 kV (FEI). Digital pictures were taken using EMMENU V3.0 (TVIPS GmbH).
In vivo optical imaging of C. rodentium-infected mice 2D bioluminescent imaging. Whole-animal BLI was measured on days 4 and 8 p.i. using an IVIS Spectrum optical imaging system (Caliper). Mice were anaesthetized with isofluorane and each animal's abdominal region was depilated using depilating cream prior to BLI. Regions of interest were identified and quantified (photons s 21 cm 22 sr 21 ) (sr 5 Steradian) using Living Image 4.2 software (Caliper).
3D bioluminescent imaging. A representative mouse from each group was selected for 3D imaging based upon the 2D BLI data. Mice were anaesthetized using an XGI-8 Anesthesia System (Caliper) and subsequently transferred to a mouse imaging shuttle and humanly restrained using translucent tape. The mouse imaging shuttle was placed into an IVIS Spectrum and imaged using DLIT, with emission wavelengths ranging from 520 to 560 nm with photon binning of 8 and BL image acquisition times set to automatic for each filter set to maximize the signal to noise ratio. The 3D BL optical image was then reconstructed using Living Image 4.2, utilizing the multi-modality imaging tool, as described earlier (Kuo et al., 2007).
mCT imaging and Dicom file generation. Following DLIT, the mouse imaging shuttle was transferred to the Quantum FX mCT imager (Caliper). The mouse imaging shuttle was positioned so that the fiducial of the imaging shuttle lined up with the fiducial in the Quantum FX mCT. A whole-mouse mCT scan was performed in two stages using a 76 mm field of view (FOV) with a voxel size of 128 mm. The mCT scan was then automatically reconstructed into a Dicom file using RigakuSW (Caliper).
Co-registration of DLIT-mCT data and generation of a 3D reconstruction. Co-registration of the DLIT 3D BL images and the Dicom file containing the whole-mouse mCT scan was performed using Living Image 4.2 (Caliper). The fiducial on the mouse imaging shuttle was used to align the Dicom file with the DLIT 3D BL image to generate the final 3D DLIT-mCT reconstruction.
Generation of a temporal video of C. rodentium infection. The representative mice from each group, described above, were imaged daily using DLIT-mCT. The subsequent 3D DLIT-mCT reconstructions were 'stitched' together and suitably annotated using Windows Live Movie Maker (http://windows.microsoft.com/en-GB/windows7/products/features/ movie-maker) to generate a four-dimensional (4D) movie of the C. rodentium infection. The movies generated using this software were converted into .avi files using a file converter.
Statistical analysis. All results were presented as scatter plots with the mean values. A one-way analysis of variance (ANOVA) was performed with a Tukey's multiple comparison post-test using commercially available software (GraphPad 5, GraphPad software); a P value of ,0.05 was taken to be significant.

Treatment of mice with B. breve UCC2003 does not reduce C. rodentium colonization
To determine whether B. breve could competitively exclude C. rodentium, mice were pre-treated for 3 days, challenged with C. rodentium and administered B. breve daily for the duration of the infection (BB+CR). As controls, mice were also challenged with C. rodentium and administered PBS-C daily for the duration of the infection (PBS-C+CR), mockinfected with PBS and administered B. breve daily (BB+PBS), or mock-infected with PBS and mock-treated with PBS-C (PBS-C+PBS). B. breve treatment did not significantly reduce C. rodentium colonization when evaluated by bacterial enumeration (Fig. 1a) or BLI (Fig. 1b-d) when compared with the PBS-C+PBS-treated group. BLI of BB+CR mice did not significantly reduce (P.0.05) the total BL signal (Fig. 1c, d) when compared with the PBS-C+CR-treated control. Furthermore, in line with previous reports (Wiles et al., 2004), the spatial distribution of C. rodentium within infected mice quantified using BLI was heterogeneous between treatment groups (Fig. 1b-d), and irrespective of the treatment group demonstrated heavy C. rodentium colonization of the caecum at day 4 p.i. and the distal colon and rectum at day 8 p.i. (Fig. 1b). Interestingly, qualitative assessment of BLI of BB+CR mice demonstrated a shift in the BL signal towards the large intestine (Fig. 1b).
In addition to BLI, a representative mouse from each group was selected for DLIT-mCT and C. rodentium colonization was monitored daily up to day 8 p.i. The DLIT-mCT data were compiled to generate a 4D movie of the BB+CR or PBS-C+CR mice. Notably, C. rodentium appears to be randomly distributed within the small intestine between days 1 and 4 p.i. until day 5 p.i., where the BL foci concentrate in the caecum (Fig. 2, Videos S1 and S2 available with the online version of this paper), as described previously (Wiles et al., 2004). BB+CR or PBS-C+CR mice demonstrated similar DLIT-mCT profiles between days 3 and 5 p.i., and B. breve treatment did not affect this distinct caecal tropism observed at day 5 p.i. (Fig. 2, Videos S1 and S2). Strikingly, at day 6 p.i. in both the BB+CR and PBS-C+CR mice, C. rodentium appears to undergo a 'virulence switch' and subsequently colonizes the large intestine, which peaks at day 8 p.i. (Fig. 2, Videos S1 and S2). However, at day 8 p.i. in BB+CR mice, colonization was concentrated to the colon with a weak BL focus in the caecum, whereas in PBS-C+CR mice multiple bioluminescent foci were observed in the caecum, colon and rectum (Figs 1b and 2).
To determine whether B. breve treatment altered C. rodentium distribution within the colonic mucosa, FFPE sections of terminal colon taken during necropsy at the peak of bacterial infection (day 8 p.i.) were investigated by an indirect IFA. In line with previous reports (Crepin et al., 2010), C. rodentium colonized epithelial cells lining the colonic lumen and bacteria did not penetrate into the colonic crypts (Fig. 3b). BB+CR and PBS-C+CR mice demonstrated identical colonization by IFA (Fig. 3a, b).

Treatment of mice with B. breve does not reduce A/E lesion formation
We determined whether the mechanisms utilized by C. rodentium to colonize the intestinal mucosa had been altered by B. breve treatment. A/E lesion formation on sections of terminal colon taken at necropsy at the peak of bacterial infection (day 8 p.i.) were investigated qualitatively by SEM and TEM. Prophylactic treatment of mice with B. breve did not affect A/E lesion formation (Fig. 4a, e) when compared with the BB+PBS-and PBS-C+PBS-treated controls. Examination of the colonic mucosa of BB+PBS mice for A/ E lesions by SEM demonstrated no adverse pathology (Fig. 4c, g) and resembled the tissues from PBS-C+PBS-treated mice.
Prophylactic treatment of mice with B. breve significantly reduces C. rodentium-associated pathology Although probiotic treatment was unable to competitively exclude C. rodentium, other parameters of an infection Fig. 4. SEM (a,c,e,g) and TEM (b,d,f,h) of murine terminal colon taken at necropsy from mice treated with B. breve or PBS and subsequently infected with C. rodentium for 8 days. Mice were pre-treated for 3 days prior to C. rodentium infection and daily following infection with B. breve (a, b), PBS (c, d), or were B. breve-treated and PBS mock-infected (e, f) or PBS-C mocktreated and PBS mock-infected as a negative control (g, h). C. rodentium infection induces A/E lesion (arrowheads) formation irrespective of the probiotic treatment. MV, microvilli. such as tissue pathology play an important role in the disease severity and clinical symptoms. Histopathological analyses of sections of terminal colon taken at necropsy at the peak of bacterial infection (day 8 p.i.) were investigated by examination of H&E-stained tissues by light microscopy (Fig. 5a-j). Quantification of colonic crypt hyperplasia (CCH) demonstrated that B. breve significantly reduced colonic crypt length (P,0.001) when compared with the PBS-C+CR control group, and the difference between the arithmetic means of BB+CR and PBS-C+CR mice was 219.82±40.71 mm and 271.49±48.82 mm, respectively (Fig. 5i). B. breve treatment did not completely inhibit CCH, and BB+CR mice demonstrated significantly (P,0.001) more CCH than BB+PBS controls (Fig. 5i). Qualitative assessment of H&E-stained tissues using a histological damage score demonstrated a significant reduction (P, 0.01) in C. rodentium-associated pathology in BB+CR mice when compared with the PBS-C+CR control group (Fig.  5j). Importantly, there was a qualitative reduction in immune cell infiltration of the lamina propia in BB+CR mice when compared with PBS-C+CR mice (Fig. 5a-d), suggesting antiinflammatory activity of B. breve.

DISCUSSION
Probiotics, in particular members of the genera Lactobacillus and Bifidobacterium, are widely used to treat Fig. 5. Evaluation of CCH (black lines) in mice treated with B. breve and subsequently infected with C. rodentium for 8 days. Mice were pre-treated for 3 days prior to C. rodentium infection and daily following infection with B. breve (BB+CR; a, b) or PBS-C (PBS-C+CR; c, d). Control mice were treated with B. breve and PBS mock-infected (BB+PBS; e, f) or PBS-C mock-treated and PBS mock-infected (PBS-C+PBS; g, h). B. breve treatment significantly reduced C. rodentium-induced pathology and lymphocyte accumulation in the lamina propria (arrowheads). Bars: ¾10, 100 mm; ¾20, 65 mm. (i) Quantification of crypt hyperplasia following treatment of mice with B. breve or PBS. Treatment of mice with B. breve significantly (P,0.001, ***) reduced crypt hyperplasia when compared with PBS-treated mice. (j) Histological damage score of H&E-stained colonic sections demonstrating a significant reduction (P,0.01, **; P,0.001, ***) in C. rodentium-associated pathology following treatment with B. breve. gastrointestinal disease caused by enteric pathogens, especially in the developing world (Huebner & Surawicz, 2006;Picard et al., 2005;Preidis et al., 2011). The C. rodentium infection model is now routinely used to test the efficacy of putative probiotic strains and the mechanisms by which such strains confer protective effects upon the host (Chen et al., 2005(Chen et al., , 2009D'Arienzo et al., 2006;Fanning et al., 2012;Gareau et al., 2010;Johnson-Henry et al., 2005;Rodrigues et al., 2012;Wu et al., 2008). Our results demonstrate that the prophylactic administration of B. breve UCC2003 to mice did not prevent C. rodentium colonization or prevent A/E lesion formation, but significantly (P,0.001) reduced CCH and C. rodentium-associated pathology, including the infiltration of inflammatory cells into the colonic mucosa. Recently, a study by Jones & Knight (2012) demonstrated that Bacillus subtillus could reduce CCH caused by C. rodentium infection and that this immunomodulatory effect was independent of a reduction in pathogen colonization. In contrast, several groups have reported that the administration to mice of a variety of probiotic micro-organisms including Lactobacillus acidophilus, B. breve and S. boulardii in PBS could competitively exclude C. rodentium, largely through unknown mechanisms (Chen et al., 2005(Chen et al., , 2009Fanning et al., 2012;Wu et al., 2008). The mechanistic differences observed in these studies are likely due to a combination of the different probiotic micro-organisms, mouse strains and age of the animals used. Importantly, Fanning et al. (2012) demonstrated that B. breve could competitively exclude C. rodentium directly through the production of bacterial exopolysaccharide, presumably by competing with C. rodentium for host cell receptors on the caecal and colonic epithelium. That study used 6-8-week-old BALB/c mice, which demonstrated altered C. rodentium colonization dynamics that rapidly peaked by day 4 p.i. (Fanning et al., 2012), unlike the more widely used C57 BL/6 model, which peaks at day 8 p.i. (Rodrigues et al., 2012). Moreover, the authors of that study did not report data concerning CCH; however, future studies should focus on why we observe such dramatic differences in the CE effects of B. breve for C. rodentium between these two different mouse strains and how this impacts on the colonic inflammatory response.
In this study we generated the first 4D infection movie using DLIT-mCT and successfully used this technology to visualize how B. breve affects C. rodentium colonization up to the peak of bacterial infection (day 8 p.i.). The 4D movies facilitated the analysis of a large volume of data in a quick and easy to interpret format, and provided a unique insight into the development of C. rodentium infection. In particular, we observed a novel 'virulence switch' between days 5 and 6 p.i., whereby C. rodentium forms a focus of infection in the caecum at day 5, which then expands as the bacteria colonize the large intestine at day 6 p.i. It is tempting to speculate that C. rodentium is adapting to the caecal microenvironment, possibly by altering its metabolism, which allows the bacteria to multiply rapidly and colonize the large intestine.
Collectively, these data demonstrate that B. breve can modulate the murine colonic inflammatory response through unknown mechanisms. Further work is required to determine the underlying molecular mechanisms behind the reduction in C. rodentium-induced CCH and lymphocyte infiltration caused by the prophylactic administration of B. breve.