Lactobacillus reuteri HCM2 protects mice against Enterotoxigenic Escherichia coli through modulation of gut microbiota

Enterotoxigenic Escherichia coli (ETEC) is a leading cause of infectious diarrhea in children and postweaning piglets. ETEC infection results in induced pro-inflammatory responses in intestinal epithelial cells and dysbiosis of intestinal microbiota. Here, a Lactobacillus reuteri strain, HCM2, isolated from a healthy piglet showed a high survival rate in the harsh gastrointestinal tract environment and inhibited the growth of ETEC and its adherence to intestinal epithelial cells. Pre-supplementation with L. reuteri HCM2 for 14 days reduced the ETEC load in the jejunum of ETEC-infected mice and prevented the disruption of intestinal morphology by ETEC. The colonic microbiota of mice with or without HCM2 pre-supplementation were analyzed, and this analysis revealed that HCM2 could prevent dysbiosis caused by ETEC infection by stabilizing the relative abundance of dominant bacteria. These results indicate that L. reuteri HCM2 has the potential to attenuate the effect of ETEC on the colonic microbiota in infected mice.

. Survival of L. reuteri HCM2 in artificial gastric juice (pH 2.5) and artificial small intestine fluid after incubation for 0 min and 2 h. challenge, the crypt depth was significantly lower in HED3 mice than in ED3 mice (p < 0.05, Fig. 3H). Although the feed intake at D15 (Fig. 4E) and D16 (Fig. 4F) was lower in mice receiving L. reuteri HCM2 supplementation, there was no obvious difference in body weight between groups ( Fig. 4A-C). Levels of serum IgG and IgA, which reflect the system immune state 28 , were also determined and compared between the different groups. At D15, the HCM2 group had 2.9% and 7.67% higher serum IgG and IgA levels, respectively than the control group (Fig. 4G,H). At D16, the HED1 group had 38.93% and 35.63% higher levels of serum IgG and IgA, respectively, than the ED1 group (Fig. 4G,H). At D18, the HED3 group had 29.41% and 13.27% higher levels of serum IgG and IgA, respectively, than the ED3 group (Fig. 4G,H).
Colonic microbiota diversity and composition analysis. The colonic microbiota of mice in the six groups were analyzed by sequencing the bacterial 16 S rRNA V3 + V4 region. High-throughput pyrosequencing of the samples (n = 5) produced a total of 2 757 344 raw reads. The effective reads were clustered into OTUs based on a similarity threshold of 97%. The 2 004 194 clean tags remaining after removing low-quality sequences were clustered into a total of 2247 OTUs, which were present in at least five samples. The rarefaction and rank abundance curves showed that the total richness of the microbial community was sampled completely (Fig. S1A,B). The Chao1, OS (Observed species), ACE, Shannon, Simpson, and PD (phylogenetic distance whole tree) indexes The villus length and crypt depth were measured as indicated in the image in (E). The black arrows in G and I indicate jejunal tissues that were damaged by ETEC. Control: mice received a basal diet. HCM2: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks. ED1: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 1 day. HED1: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, and then received a basal diet for 1 day. ED3: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 3 days. HED3: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, and then received a basal diet for 3 days. were calculated (Table S1). Based on these indexes there was not a big difference in microbial structure between mice receiving the L. reuteri HCM2 pre-treatment and those that did not. The sequences across all samples were assigned to 27 phyla and 280 genera. Firmicutes and Bacteroidetes were the two predominant phyla in the mice gut microbiota, followed by Proteobacteria, Actinobacteria, and Saccharibacteria (Table 2). To investigate the gut bacteria in each group of mice, namely Control, The data are presented as mean ± SD with n = 5. *Indicates significance at p < 0.05. Control: mice received a basal diet. HCM2: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks. ED1: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received basal diet for 1 day. HED1: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, and then received a basal diet for 1 day. ED3: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 3 days. HED3: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, and then received a basal diet for 3 days.  The data are presented as mean ± SD with n = 5. * indicates significance at p < 0.05. Control: mice received a basal diet. HCM2: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks. ED1: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 1 day. HED1: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15 and then received a basal diet for 1 day. ED3: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 3 days. HED3: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, and then received a basal diet for 3 days.   The relative abundances (mean ± SD) of the genera Lactobacillus, Allprevotella, Alistipes, Bacteroides and Helicobacter were plotted against time for each treatment. * indicates significance at p < 0.05 for a single time point. Control: mice received a basal diet. HCM2: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks. ED1: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 1 day. HED1: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15 and then received a basal diet for 1 day. ED3: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 3 days. HED3: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, and then received a basal diet for 3 days. in the HED1 group, and then decreased to 1.2% in the HED3 group (Fig. 5B). As shown in Table 2, after ETEC challenge, the proportion of Enterobacteriaceae in mice without HCM2 supplementation increased from 0.003% in the Control group to 0.039% in the ED1 group and then to 0.042% in the ED3 group. This result demonstrates that ETEC challenge could increase the proportion of Enterobacteriaceae in the mouse colon. The proportion of Enterobacteriaceae in mice supplemented with HCM2 increased from 0.001% in the Control to 0.008% 1 day after ETEC challenge (HED1 group), but decreased to pre-infection proportion of 0.003% 3 days after ETEC challenge (HED3 group). This result demonstrates that although ETEC infection leads to an increased proportion of Enterobacteriaceae in the short term, pre-treatment with L. reuteri HCM2 could help Enterobacteriaceae recover to the pre-infection level in the gut. Taken together, these results indicate that pretreatment of mice with L. reuteri HCM2 for 14 days before ETEC challenge could alleviate the disruption of the bacterial community caused by ETEC infection and help the gut microbiota to recover to pre-infection levels.
The dynamic changes in the dominant genera in each group of mice were also analyzed. Lactobacillus was the predominant genera in the mouse gut microbiota and showed a similar dynamic trend as Firmicutes after ETEC challenge (Fig. 5C). This result suggests that Lactobacillus might be the main genus contributing to the changes in Firmicutes after ETEC challenge. The proportion of bacteria in the Helicobacter genus also increased with increasing time after infection in groups without L. reuteri HCM2 supplementation and fluctuated in groups with L. reuteri HCM2 supplementation, a trend similar to that observed for Proteobacteria. This result suggests that Helicobacter might be the main genus contributing to the changes in Proteobacteria (Fig. 5C). The relative abundance of Alloprevotella, which belongs to the phylum Bacteroidetes, increased from 0.1% in the HCM2 group to 6.6% in the HED1 group, and then decreased to 0.1% in the HED3 group (Fig. 5C). Two other dominant genera, Alistipes and Bacteroides, which also belong to the phylum Bacteroidetes, showed similar dynamic trends as the phylum Bacteroidetes (Fig. 5C).
We next compared the colonic microbiota of mice in groups ED3 and HED3 with those of mice in the Control group using the linear discriminant analysis (LDA) effect size (LEfSe) method. A cladogram representing the structure of the colonic microbiota and the predominant bacteria taxa is shown in Fig. S2. Compared with the Control group, mice in the HED3 group had a significantly higher abundance of Lactobacillus, while mice in the ED3 group had an increased abundance of Bacteroidia, Lachnospiraceae, Campylobacterales and Helicobacter (Fig. S2A,B). These results demonstrate that L. reuteri HCM2 could protect mice from ETEC challenge by modulating the gut microbiota, restoring the pre-infection structure and promoting high levels of probiotic bacteria.

Discussion
L. reuteri is a common species that inhabits the gastrointestinal tract of humans and many animals. Here we demonstrated that an L. reuteri HCM2 strain isolated from a healthy piglet could protect against ETEC infection in a mouse model by altering gut microbiota. The growth of ETEC was obviously inhibited by L. reuteri HCM2, and this might be caused by damage to the ETEC cell wall. As we did not observe any antimicrobial peptide production by L. reuteri HCM2, it seems that lactic acid might function as the antimicrobial agent inhibiting ETEC growth (data not shown). Competition for adhesion sites is one of the mechanisms by which probiotics fight against pathogen infection 29,30 . We found that L. reuteri HCM2 could survive in the harsh gastrointestinal tract environment and had the capacity to adhere to Caco-2 cells. Furthermore, a competition assay revealed that L. reuteri HCM2 was able to reduce the number of ETEC cells. These results demonstrate that L. reuteri HCM2 potentially protects against ETEC. We further observed that the ETEC load in both the jejunum tissues and contents of the mouse model was significantly decreased after L. reuteri HCM2 pre-supplementation. A similar phenomenon was also observed by Yang et al., who found that L. reuteri TMW1.656 could decrease the abundance of E. coli and K88 fimbriae in ETEC-challenged weanling pigs 31 .
Probiotics are important because they protect the host gastrointestinal micro-environment from invading pathogens 2 . In this study, ETEC infection led to inflammatory infiltration and loss of microvilli in jejunum tissues in mice not receiving L. reuteri HCM2 pre-supplementation (ED1 and ED3 groups). This is consistent with the finding of a previous study that ETEC infection promotes the expression of pro-inflammatory cytokines by activating the NF-kB and MAPK pathways, leading to the loss of microvilli in the jejunum 6 . However, inflammatory infiltration was not found in mice of the HED1 and HED3 groups, which were pre-supplemented with L. reuteri HCM2. The expression levels of IgA and IgG did not obviously increase in mice pre-supplemented with L. reuteri HCM2 compared with control mice. Gao et al. also reported that supplementation with L. plantarum did not obviously increase the expression levels of IgG in broiler chickens 20 , which supports our finding.
Several recent studies have noted that supplementation with L. reuteri improves the growth and feed efficiency of neonatal and growing pigs 8 . For example, Liu et al. reported that L. reuteri I5007 supplementation increased the average daily weight gain in formula-fed piglets 18 . Wang et al. found that pigs supplemented with L. fermentum I5007 had higher weight gain and feed intake than pigs without L. fermentum I5007 treatment 32 . In our study, mice supplemented with L. reuteri HCM2 took in less food, however, no significant difference in body weight was detected between control mice and mice supplemented with L. reuteri HCM2. L. reuteri HCM2 slightly decreased the feed conversion in HCM2 and HED1 mice (Fig. S3), which is consistent with the report that L. fermentum I5007 could slightly decrease feed conversion in piglets 32 . The decreased feed conversion in HCM2 and HED1 mice may due to the increased ratio of Firmicutes to Bacteroidetes (Fig. S4), which has been reported to be associated with improved energy-harvesting capacity 33 . Furthermore, L. reuteri HCM2 significantly increased the villus/crypt ratio in mice, which indicates that the absorption of nutrients might be increased by feeding mice L. reuteri HCM2.
The healthy colon is usually dominated by obligate anaerobes, while dysbiosis is often associated with a sustained increase in the abundance of facultative anaerobic Proteobacteria, which results in the disruption of anaerobiosis 34  dynamics of gut microbiota and found that the abundance of Proteobacteria gradually increased after ETEC infection (ED1 and ED3). By contrast, not much difference in Proteobacteria abundance was observed in the L. reuteri HCM2 pre-supplementation groups (HED1 and HED3). Higher levels of Enterobacteriaceae (belonging to Proteobacteria) were detected in mice without L. reuteri HCM2 pre-supplementation, and this may contribute to a thinner and more penetrable mucus layer, which increases the risk of chronic colitis 36 . ETEC are members of Enterobacteriaceae, and there was a significant difference in the relative abundance of Enterobacteriaceae between the Control and HCM2 groups. It is interesting that there was no big difference in the relative abundance of Enterobacteriaceae between the ED1 and HED1 and groups (p > 0.05). This might be due to the amount of ETEC challenge. However, after 3 days of recovery, the relative abundance of Enterobacteriaceae in ED3 was significantly higher than that in the group HED3 (p = 0.0397). It is clear that HCM2 pre-supplementation decreases the relative abundance of Enterobacteriaceae and may prevent the dysbiosis of gut microbiota caused by ETEC. It is noteworthy that the relative abundance of Helicobacter was increased by ETEC infection, while there was no obvious difference in the abundance of Helicobacter in the L. reuteri HCM2 pre-supplementation groups (HED1 and HED3) after ETEC infection. It has been reported that many cases of non-H.pylori Helicobacter (NHPH) infection in humans occur in immunocompromised patients, because NHPH can induce high levels of inducible nitric oxide synthase and the development of DNA double-stranded breaks, but the properties of Lactobacillus can prevent Helicobacter infection or its related pathologies 37 . In our study, L. reuteri HCM2 supplementation not only maintained the abundance of Helicobacter at a relatively low level, but also increased the relative abundance of Lactobacilli. It has been reported that the abundance of Helicobacteraceae in a rat model of colon cancer was decreased by administering a probiotic cocktail (Lactobacillus acidophilus, Bifidobacteria bifidum, and Bifidobacteria infantum), which promoted the growth of Lactobacilli and thus altered the gut microbiota 38 . We conclude that pre-supplementation with L. reuteri HCM2 may alter the gut microbiota and protect against ETEC and the increase in abundance of detrimental bacteria caused by ETEC.
We monitored the dynamic changes in dominant genera in each group of mice and intriguingly found that ETEC could reduce the relative abundance of Lactobacillus in ED1 and HED1 mice. The relative abundance of Lactobacillus in the ED3 group was lower than that of both the Control and ED1 groups, while in the HED3 group, the abundance of Lactobacillus was the same as that in the control, suggesting that L. reuteri HCM2 pre-supplementation prevented the reduction in Lactobacillus abundance. It was reported that lactobacilli abundance in the colon could be reduced by ETEC, while pre-administration of probiotics could increase the abundance of lactobacilli and enhance goblet cell function to ameliorate enteritis 35 . Lactobacilli, which are considered health promoting probiotics 39 , might promote defense against detrimental bacteria in the gut by creating an acidic environment (e.g., pH 4.5), synthesizing exopolysaccharides 5 , competitively excluding intestinal pathogens 40 , improving antioxidant activity 19 , or activating and enhancing local cell-mediated immunity against certain enteric pathogens 41 . So, stabilizing the relative abundance of Lactobacillus is important for maintaining the balance of gut microbes. The relative abundances of the dominant genera Alistipes, Bacteroides and Alloprevotella, which belong to the Bacteroidetes, were also higher after ETEC infection, but the abundances returned back to the pre-infections after 3 days of recovery (HED3). This also indicates that L. reuteri HCM2 pre-supplementation could stabilize the relative abundance of Bacteroidetes. In conclusion, L. reuteri HCM2 pre-supplementation may modulate the gut microbiota, preventing the dysbiosis of intestinal microbiota caused by ETEC.

Materials and Methods
Bacterial strains and cell lines. E. coli F4-producing strain W25K (ETEC; O149:K91, K88ac; LT, STb, EAST) 4,42 was used and cultured in LB medium. Lactobacillus reuteri HCM2 was isolated from a 6-week-old healthy piglet and was anaerobically cultured in MRS medium. All strains were sub-cultured twice prior to being used for experiments. The Caco-2 cell line was cultivated in complete Dulbecco's modified Eagle's minimal essential medium (DMEM, C12430500BT, purchased from Thermo Fisher Scientific (China) Co., Ltd.) supplemented with heat-inactivated fetal bovine serum (10% v/v) and 100 U/ml penicillin-Streptomycin. The medium was replaced every 2 days. Cells were grown at 37 °C, 5% CO 2 /95% air in T25 flasks.
Animals, feeding procedures, and infection. ICR 6,7,42 male mice (5 weeks of age) were purchased from SLACCAS (Shanghai Laboratory Animal Center). Only male mice were used to avoid differences in microbiota composition resulting from sex and maternal factors 43 . Animal experiments were approved by the Laboratory Animal Ethical Commission of the Chinese Academy of Sciences and performed according to its guidelines. In order to test the protective effect of L. reuteri HCM2 against ETEC in vivo, we developed a mouse model as described in a previous study 6,35 with minor modifications. The mice were housed in a pathogen-free mouse colony (temperature, 25 ± 2 °C; relative humidity, 45-60%; lighting cycle, 12 h/d). After acclimatization for three days, the mice were randomly divided into six groups (n = 5 for each group) as shown in Fig. 6. The Control group received a basal diet 44 ; the HCM2 group received a basal diet and a 200 μL suspension of 5 × 10 9 CFU/ ml L.reuteri HCM2 daily 45 for two consecutive weeks; the ED1 (basal diet + ETEC + Day 1) group and the ED3 (basal diet + ETEC + Day 3) group received a basal diet, were challenged with 100 μL of 10 9 CFU/ml ETEC 6,7 by intragastric administration at day 15, then received a basal diet for 1 and 3 days, respectively; the HED1 (HCM2 + ETEC + Day 1) group and the HED3 (HCM2+ ETEC+ Day 3) group received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 1 and 3 days, respectively. Mice were intragastrically administered L. reuteri HCM2 or ETEC via 12-gauge gavage needles (Bio-Medical Needles, Beijing Solarbio Science & Technology Co., Ltd.). Control and HCM2 mice were sacrificed at D15, ED1 and HED1 mice were sacrificed at D16, and ED3 and HED3 mice were sacrificed at D18. All mice were sacrificed at 10:00 am and samples were collected. Briefly, after cervical dislocation, blood was collected by removing the eyeball, then the jejunum, Antimicrobial activity assay. ETEC were cultured overnight in Shake-tubes and were standardized to OD 600 = 2. The supernatants of the L. reuteri HCM2 overnight cultures were filtered through a 0.22 μm filter, and 100 μl of CFS was placed into each well of an ETEC indicator strain agar plate. MRS medium was tested as a negative control. After incubation at 37 °C for 24 h, the diameters of the inhibition zones were measured. To further validate the capacity of L. reuteri HCM2 to antagonize ETEC, ETEC cells at a density of 1 × 10 6 CFUs/ml were seeded into the wells of a 96-well microtiter plate, and 0%, 5%, 10% or 20% (v/v) CSF of L. reuteri HCM2 was added to each well. The absorbance of ETEC at 600 nm was monitored every 2 h.
Scanning electron microscopy. SEM experiments were performed as described by Aiba et al. with slight modifications 46 . Briefly, a 10 8 CFUs/ml inoculum of ETEC was treated with 20% L. reuteri HCM2 CFS for 12 h at 37 °C, and a non-treated ETEC culture served as a control. After washing with fresh PBS, cells were fixed with 2.5% glutaraldehyde for 12 h. The bacteria were then washed with demineralized water 3 times, 6 min each time. After dehydration with ethanol (stepwise gradient of 50%, 70%, 85% 95% and 100%), the specimens were freeze-dried in a critical-point dryer (Czech FEI company) and coated with gold. The morphology of the cells was visualized by examining the specimens with a Hitachi cold field emission scanning electron microscope.
Characterization of the tolerance of L. reuteri HCM2. L. reuteri HCM2 cells from an overnight culture were washed twice with PBS (pH 7.2), and the concentration was adjusted to about 10 8 to 10 9 CFUs/ml. One hundred microliters of cells were transferred to 900 μl of artificial gastric juice (pH 2.5) 47 or artificial small intestine fluid (pH 6.8) and incubated anaerobically for 2 h at 37 °C. After recovery for 48 h, viable bacterial cells were counted by plating serial dilutions of the culture in PBS on MRS medium.
In vitro adhesion assay. Caco-2 cells were prepared on a Millicell EZ SLIDE 4-well and seeded at a concentration of 10 5 cells per well. Caco-2 cells were cultured for 15 days in cell culture medium to obtain confluence, and the medium was changed on alternate days. Caco-2 cells were washed with D-Hanks buffer solution and fresh culture medium without antibiotic solution was added 48 h before the adhesion assay 40 . The overnight L. reuteri HCM2 culture was washed with D-Hanks buffer solution and then re-suspended in fresh DMEM. The L. reuteri HCM2 suspension was diluted to 10 8 CFUs/ml and 1 ml was added into the Millicell EZ SLIDE 4-well. After co-culturing for 2 h, the Caco-2 monolayers were washed three times and fixed for 20 min at room temperature with 4% paraformaldehyde (w/v) fix solution. The Caco-2 monolayers were then washed five times with D-Hanks buffer solution, air-dried and Gram stained 48 . The images of L. reuteri HCM2 binding to Caco-2 cells were captured under microscope. The amount of L. reuteri HCM2 binding to Caco-2 cells was further quantified as described by Todoriki et al. with slight modifications 49 . Briefly, the Caco-2 cells were cultivated in 6-well plates until confluence was reached and then used for adhesion experiments. Caco-2 monolayers were washed twice with PBS, and 3 ml bacterial suspension (5 × 10 8 cells/ml) was added to each well of the tissue culture plate. The plates were incubated at 37 °C in 5% CO 2 /95% air. After 2 h of incubation, the monolayers were washed three times with PBS. Following the last wash, Caco-2 cell monolayers were covered with 3 ml distilled water and mechanically agitated by vigorous pipetting to suspend the Caco-2 cells and bacteria. Adherent bacteria were serial diluted by 10-fold and plated on MRS media. The number of L. reuteri HCM2 colonies were counted after anaerobic incubation for 48 h. Figure 6. A schematic of the experimental design. The gray boxes indicate that mice received a common diet, and blue boxes indicate that mice were treated with L. reuteri HCM2. The red arrow indicates when mice were challenged with ETEC. Control: mice received a basal diet. HCM2: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks. ED1: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 1 day. HED1: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15 and then received a basal diet for 1 day. ED3: mice received a basal diet, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, then received a basal diet for 3 days. HED3: mice received a basal diet and 10 9 CFUs L. reuteri HCM2 daily for two consecutive weeks, were challenged with 10 8 CFUs ETEC by intragastric administration at day 15, and then received a basal diet for 3 days. Competition and displacement assays. Caco-2 cells were cultivated in 6-well plates to 80% confluence and then used for adhesion experiments 40,50 . Suspensions of L. reuteri HCM2 and ETEC were prepared with DMEM medium. For the competition assays, bacterial cells were washed twice with PBS, and of 100 μl of L. reuteri HCM2 suspension (10 8 CFUs) and 100 μl of ETEC suspension (10 8 CFUs) were added to each well simultaneously. The cultures were incubated at 37 °C for 90 min, then the monolayers were washed twice with PBS and digested with 100 μl trypsin (0.25%) for 2 min. For displacement assays, 100 μl ETEC suspension (10 8 CFUs) was added to each well, and the samples were incubated for 45 min. Then 100 μl of L. reuteri HCM2 suspension (10 8 CFUs) was added, and the samples were incubated for another 45 min. The plates were incubated and digested as described above. For both assays, serial dilutions of the adherent bacteria were plated on MacConkey Agar containing 50 μg/ml streptomycin and incubated at 37 °C for 16 h. The number of colonies were then counted. The ETEC strain used in this study is resistant to streptomycin.

Morphological analyses.
Mouse jejunums were fixed with 4% paraformaldehyde-PBS overnight and then dehydrated and embedded in paraffin blocks. Sections (5 μm) from these blocks were deparaffinized, hydrated, and then stained with hematoxylin and eosin (H&E). At least three villus lengths and crypt depths per slide were measured using Image-Pro Plus software 6. Five mice were studied from each group. The data collectors were unaware of the treatment status of the examined slides. Illumina HiSeq sequencing and data processing. Total genomic DNA was extracted from the colonic contents of mice from the six groups using the Qiagen QIAamp DNA Stool Mini Kit. The V3-V4 hypervariable region of the bacteria 16 S rRNA gene was amplified using the primers 338 F 5′-ACTCCTACGGGAGGCAGCA-3′ and 806 R 5′-GGACTACHVGGGTWTCTAAT-3′. PCR products were mixed in equidensity ratios and purified with the Qiagen Gel Extraction Kit (Qiagen, Germany). Sequencing libraries were generated using the TruSeq ® DNA PCR-Free Sample Preparation Kit (Illumina, USA), and index codes were added. The library quality was assessed using the Qubit@ 2.0 Fluorometer (Thermo Scientific) and the Agilent Bioanalyzer 2100 system. Finally, the library was sequenced on the Illumina HiSeq2500 platform, and 250 bp paired-end reads were generated, assigned to samples based on their unique barcode and truncated by cutting off the barcode and primer sequences. Paired-end reads were merged using FLASH 51 , which was designed to merge paired-end reads when at least some of the reads overlap with the read generated from the opposite end of the same DNA fragment, and the trimmed sequences were called raw tags. The raw tags were filtered using the QIIME quality control process to obtain high-quality clean tags 51 . The tags were compared with the reference database (Gold database) using the UCHIME algorithm to detect chimeric sequences, which were later removed 52,53 . The effective tags were finally obtained.
Sequences sharing greater than 97% similarity were assigned to the same OTUs 54 . A representative sequence for each OTU was selected for further annotation. For each representative sequence, the GreenGene Database was used to obtain taxonomic information using the RDP classifier algorithm 55,56 . OTU abundance was normalized to the number of sequences in the sample with the fewest sequences. Subsequent analysis of alpha diversity and beta diversity were performed using the normalized data. Species diversity complexity of a sample was analyzed using six indices of alpha diversity, including Observed-species, Chao1, Shannon, Simpson, ACE and Good-coverage. All indices were calculated with QIIME (Version 1.7.0) and visualized using R software (Version 2.15.3). The differences in dominant bacterial communities between groups were determined based on LDA Effect Size.
Statistical analysis. Data shown are the means ± SD or SEM. Differences in the means between two groups were analyzed by performing an unpaired t test (Prism 7.0) if the data were normally distributed and the samples had equal variance, or by performing a non-parametric test (Mann-Whitney U test, Prism 7.0) if the data were not normally distributed. Means of more than two groups were analyzed by performing one-way ANOVA followed by the Dunnett multiple comparisons test (Prism 7.0) if the data were followed a Gaussian distribution and had equal variance, or by performing the Kruskal-Wallis test followed by Dunn's multiple comparisons test (Prism 7.0) if the data were not normally distributed. The Kolmogorov-Smirnov test (Prism 7.0) was used to determine if the data followed a Gaussian distribution. The homogeneity of variance test (SPSS 22.0) or the Brown-Forsythe test (Prism 7.0) was used to test for equal variance. Differences with p < 0.05 were considered significant.