B. fragilis strain ZY-312 was provided from Zhiyi Biological Technology Co., Ltd. (Guangzhou, Guangdong province, China; Deng et al., 2016; Wang et al., 2017). The strain was cultured in trypticase soya broth (TSB; Oxoid, Basingstoke, UK) supplemented with 5% fetal bovine serum (MP Biomedicals, USA) at 37°C for ~16 h in an anaerobic incubator (Bugbox, Ruskinn). Ten-fold serial dilutions of the overnight cultures were plated on trypticase soya agar (TSA; Oxoid, Basingstoke, UK) containing 5% fresh sheep blood to determine the initial bacterial concentrations [presented as colony forming units (cfu/ml)]. Wild-type V. parahaemolyticus strain rimd2210633 (Makino et al., 2003), an ampicillin-resistant clinical isolate originally isolated from a patient with diarrhea, was obtained from the Academy of Military Medical Science, Beijing, China. It was cultured in 3.5% NaCl Luria-Bertani broth (LB; Oxoid, Basingstoke, UK) at 37°C in a shaking incubator for ~8 h.

Bacterial cells were harvested in log phase (B. fragilis at 16 h, V. parahaemolyticus at 8 h) by centrifugation (4,000 rpm, 10 min), washed twice with sterile phosphate-buffered saline (PBS), and then diluted in PBS to OD₆₀₀ = 1.0 (corresponding to ~10⁹ cfu/ml). The bacterial cell density was adjusted to the target concentration with PBS where necessary. B. fragilis culture supernatant was filtered using a 0.22-μm filter membrane (Millex) to ensure all bacterial cells had been removed. A 1-ml aliquot of B. fragilis at a concentration of 10⁹ cfu/ml was lysed by sonication and then filtered through a 0.22-μm filter membrane. Aliquots (0.2 ml) of V. parahaemolyticus at a concentration of 10⁵ cfu/ml were then spread evenly on the surfaces of LB plates, and 3–4 Oxford cups were placed onto each plate. Next, 250-μl volumes of PBS, sterile TSB, filter-sterilized B. fragilis supernatant, B. fragilis lysate, and resuspended B. fragilis at a concentration of 10⁹ cfu/ml were individually added to the Oxford cups. The plates were then incubated at 4°C for 12 h and then 36°C for 12 h prior to observation.

Tumor-derived human colonic epithelial LoVo cells and tumor-derived mouse macrophage RAW 264.7 cells were obtained from the Academy of Military Medical Science. LoVo cells were cultured in Dulbecco's Modified Eagle Medium/nutrient mixture F12 (DMEM-H/F12) (1:1) (Gibco, USA) supplemented with 10% fetal bovine serum (MP Biomedicals, USA). RAW 264.7 cells were cultured in DMEM medium (Gibco) supplemented with 10% fetal bovine serum (MP Biomedicals). Both cell lines were cultured under a moist atmosphere in a 5% CO₂ incubator (MCO-18AIC SANYO, Japan) at 37°C. Cells were diluted to a density of 2 × 10⁵ cells/ml in their respective culture media. For the RTCA assays (RTCA iCELLigence, ACEA Biosciences, USA), 150 μl of culture media were added to each well of the RTCA plates, which were then inserted into the station to obtain baseline measurements. A 300-μl volume of cell suspension (2 × 10⁵ cells/ml) was added to each well, and plates were incubated in a 5% CO₂ incubator at 37°C for 12–14 h. Following incubation, the culture medium was aspirated from all wells. A 300-μl volume of fresh culture medium was added to all control and V. parahaemolyticus-only treatment wells, while 300 μl of culture medium containing B. fragilis [10⁸ cfu/ml; multiplicity of infection (MOI) = 500] were added to B. fragilis-only and B. fragilis + V. parahaemolyticus treatment wells. Plates were incubated for a further 3 h before 30 μl of culture medium containing V. parahaemolyticus (10⁸ cfu/ml; MOI = 50) were added to V. parahaemolyticus and B. fragilis + V. parahaemolyticus treatment wells. The bacteria and cells were then co-incubated for another 3 h. Cells were monitored throughout the experiment using the xCELLigence system according to the manufacturer's instructions. At the end of the experimental period, cell morphologies were observed and photographed using a light microscope. The experiments were repeated three times.

Plasmid pXEN-luxCDABE, was obtained from the Academy of Military Medical Science, Beijing, China. The plasmid encodes luciferase and substrate (Fatty Aldehydes). The luciferase and substrate, with the participation of oxygen and ATP, produce long chain fatty acids and release photons (450~490 nm) that can be detected. The plasmid also confers resistance to kanamycin. Wild-type V. parahaemolyticus was transformed with pXEN-luxCDABE by electroporation. 10⁹ cfu V. parahaemolyticus and 50 ng plasmid were added into 0.1 ml electroporation buffer (14% sucrose, 1 mM EDTA, 1 mM Hepes, pH 7.0) and placed in an ice-water mixture for 10 min before electroporation (25 μF, 200 Ω, 2.0 Kv). The resulting transformants were cultured on thiosulfate-citrate-bile salts-sucrose (TCBS) agar plates containing kanamycin (50 μg/ml), and then lux-expressing clones were selected using a NightOWL II LB983 imaging system (Berthold Technologies, Bad Wildbad, Germany). A lux-expressing clone was transferred to a fresh TCBS-kanamycin (50 μg/ml) agar plate (TCBS agar, LandBridge Beijing, China) for enrichment culture.

The wild-type and lux-expressing V. parahaemolyticus strains were cultured in 5 ml of LB broth and LB broth supplemented with kanamycin (50 μg/ml), respectively, at 37°C in a shaking incubator for ~8 h. Optical density at 600 nm was measured every 2 h, and the resulting growth curves were plotted.

LoVo cells were harvested and seeded at a density of 2 × 10⁴ cells/well into 96-well plates and then incubated overnight (12–14 h) at 37°C in a CO₂ incubator. Cells were then infected with either the lux-expressing or wild-type V. parahaemolyticus strain at a MOI of 50, and the plates were incubated for 4 h at 37°C in a cell incubator. An LDH Cytotoxicity Detection Kit (Roche, USA) was used to measure the LDH released from damaged cells according to the manufacturer's recommendations. The released LDH was detected using a spectrophotometer (Promega, Madison, WI, USA) at OD₆₀₀. Eight replicates were performed for each treatment, and the experiment was repeated three times.

LoVo cells were harvested and diluted to a density of 2 × 10⁵ cells/ml in fresh culture medium. A total of 150 μl of culture medium were added to each well of RTCA plates, which were then inserted into the station to determine baseline measurements. A 300-μl volume of LoVo cell suspension was added to each well, and plates were incubated in a moist atmosphere in a 5% CO₂ incubator at 37°C for 12–14 h. The culture medium was then aspirated from all wells, and a 300-μl volume of culture medium containing wild-type V. parahaemolyticus (1 × 10⁷ cfu/ml) was added to each of the V. parahaemolyticus treatment wells (MOI = 50). A 300-μl volume of culture medium containing lux-expressing V. parahaemolyticus (1 × 10⁷cfu/ml) was added to each of the lux-V. parahaemolyticus treatment wells (MOI = 50), and the plates were incubated for a further 12 h. Cells were monitored throughout the experimental period using the xCELLigence system.

All animal experiments were conducted using 6-week-old female BALB/c mice raised under specific pathogen-free conditions, provided by the Small Animal Breeding Center, Academy of Military Medical Sciences. All animal experiments were conducted in accordance with the Guidelines for the Welfare and Ethics of Laboratory Animals of China, and were approved by the Committee of the Welfare and Ethics of Laboratory Animals, the Academy of Military Medical Science, Beijing, China (NO. SCXK-2015-01070021). Mice were fasted for 4 h prior to gavage. Twenty-four hours prior to each experiment, mice were randomly divided into two groups (control and treatment groups), marked with a unique number by ear-tag, and then gavaged with 0.2 ml of streptomycin (100 mg/ml).

The lux-expressing V. parahaemolyticus strain was cultured in LB medium supplemented with kanamycin (50 μg/ml) for 8 h at 37°C with shaking. The cell concentration was adjusted to 3 × 10¹⁰ cfu/ml with PBS, and all mice were gavaged with 0.3 ml of bacterial suspension. At 3 h post-gavage, mice were anesthetized with 3% isoflurane for 10 min, and in vivo imaging was carried out using the NightOWL II LB983 imaging system. B. fragilis was cultured in TSB supplemented with 5% fetal bovine serum at 37°C for ~16 h in an anaerobic incubator. Once revived, mice in the treatment group were gavaged with 0.3 ml of B. fragilis cell suspension (3 × 10¹⁰ cfu/ml), while those in the control group were gavaged with 0.3 ml of PBS. At 10 h post-inoculation, in vivo imaging was performed for the two groups, and the gavage step was repeated. In vivo imaging was then performed every 4 h until the bioluminescence was completely undetectable in one group. Results were analyzed using IndiGO software (Berthold Technologies, Germany). Mice were imaged in the same order at each time point.

At the end of the experimental period, all mice were sacrificed with CO₂ and their intestinal tracts were harvested. The intestinal tracts were cut open and exposed to the air for 10 min, and then examined using bioluminescence imaging. In addition, stools from all animals were collected and weighed at 12 and 20 h post-inoculation, and again at the end of the experimental period. The number of V. parahaemolyticus contained in the stool samples was determined by plating a 10-fold serial dilution of the stool on TCBS agar (LandBridge Beijing, China) with and without kanamycin (50 μg/ml). V. parahaemolyticus appears as blackish green colonies on TCBS agar. Each group had 4 mice, and the experiment was repeated 3 times.

Statistical analyses were performed using SPSS software version 17.0. Differences in the amounts of released LDH between the control group, the lux-expressing strain, and the parental strain were determined using one-way analysis of variance (ANOVA). Repeated measurement data analysis of variance (RMANOVA) was used to test the difference between the average luminous intensity of the control and treatment groups. A probability value of p < 0.05 was considered statistically significant.

In the Oxford cup assay, active ingredients contained in the liquid will diffuse freely into the agar, resulting in a zone of inhibition around the cup if there is antibacterial activity. Assays are conducted at a low temperature to allow time for the diffusion to occur.

In our experiments, negative control strain V. parahaemolyticus grew both around and inside the Oxford cups. However, a zone of inhibition was observed around the cups containing B. fragilis culture. No zones of inhibition were observed around cups containing filtered B. fragilis lysate or resuspended bacteria (Figure 1A). To confirm these results, we cultured V. parahaemolyticus in LB broth with and without B. fragilis at a ratio of 1:1 by volume at 37°C anaerobically. After 8 h, the culture medium of the B. fragilis treatment remained clear, while the culture medium of the control was turbid (Figure 1B).

B. fragilis was added to LoVo or RAW 264.7 cells in an attempt to simulate the relationship between normal intestinal epithelial cells or macrophages and intestinal bacteria. The cells were then infected with V. parahaemolyticus to explore whether B. fragilis could protect the cells from damage caused by the pathogen. RTCA and light microscopy were used to detect toxic effects of V. parahaemolyticus on the cells.

Following incubation of LoVo or RAW 264.7 cells with B. fragilis (MOI = 500) for 8 h, the normal cell index (NCI) curves remained similar to those of the bacteria-free controls. However, the NCI curves of cells incubated with V. parahaemolyticus (MOI = 50) rapidly declined. The morphology of the cells infected with B. fragilis also remained very similar to that of the controls, while cells infected with V. parahaemolyticus became wrinkled and lysed indicating significant cytotoxicity. However, when the cells were infected with V. parahaemolyticus (MOI = 50) after being incubated with B. fragilis (MOI = 500) for 3 h, both the degree and the rate of decline of the NCI curves were reduced compared with the V. parahaemolyticus-only treated cells. In addition, the morphology of the B. fragilis pre-treated cells was much closer to that of untreated cells compared with the V. parahaemolyticus-treated cells (Figure 2).

Plasmid pXEN-luxCDABE was successfully transformed into V. parahaemolyticus (Figure 3A). To determine whether the plasmid affected the growth and toxicity of the bacterium, we compared the growth rates and cellular toxicity of the lux-expressing and parental strains. The growth curves showed that the growth rate of the lux-expressing strain was similar to that of the wild-type strain (Figure 3B). In addition, there was no significant difference (p = 0.831) in the amount of released LDH between the two strains (Figure 3C). The NCI curve of LoVo cells incubated with the lux-expressing strain was also similar to that of cells infected with the wild-type strain (Figure 3D).

Using in vivo imaging, we traced the real-time proliferation and displacement of lux-expressing V. parahaemolyticus in mice. Following gavage with lux-expressing V. parahaemolyticus at a dose of ~10¹⁰ cfu, the luminous region and luminous intensity at 3 h post-infection were similar for all mice (Figure 4A). Half of the mice were then treated with B. fragilis at 3 and 14 h post-infection with lux-expressing V. parahaemolyticus. At 14 h post-infection, the luminous intensity of the mice in the treatment group was obviously weaker than that of the control group, and continued to decrease with time. By 26 h post-infection, luminescence in the treatment group was almost below the limit of detection, while the control group still had significant, although markedly decreased, bioluminescence.

At 26 h post-infection, all mice were sacrificed and their intestinal tracts were removed for bioluminescence imaging. Bioluminescence was undetectable in the intestinal tracts of mice from the treatment group, but was still evident in the intestinal tracts of animals from the control group. There was no bioluminescence in the small intestine of control mice, with the luminous regions located in the colon, especially the cecum (Figure 4B). RMANOVA analysis of the average percent luminous intensity between the two groups revealed that the luminous intensity of the treatment group was significantly decreased (p = 0.01) compared with the control (Figure 4C). The difference in luminous intensity could reflect the difference in the number of bacteria, we inferred that treatment with B. fragilis decreased V. parahaemolyticus colonization of the mouse intestine.

The number of viable V. parahaemolyticus cells in stool samples collected from the two groups was determined using selective TCBS agar (with/without kanamycin, 50 μg/ml; Figure 5). During the experiment, some of the V. parahaemolyticus would lose the lux plasmid as well as the resistance to Kanamycin. Figure 5A showed the V. parahaemolyticus in the stool that still had the lux plasmid and Figure 5B showed all the V. parahaemolyticus in the stool. We found that the V. parahaemolyticus load in stools from the treatment group was significantly less than that of the control group, regardless of whether or not kanamycin was present in the medium.

The property of the Bacteroides fragilis strain ZY-312 we used in the experiments belongs to Guangzhou ZhiYi biotechnology Co. Ltd. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.