Study of the Ability of Bifidobacteria of Human Origin to Prevent and Treat Rotavirus Infection Using Colonic Cell and Mouse Models

Rotavirus is the leading cause of severe acute gastroenteritis among children worldwide. Despite effective vaccines, inexpensive alternatives such as probiotics are needed. The aim of this study was to assess the ability of probiotic candidate Bifidobacterium thermophilum RBL67 to inhibit rotavirus infection. Bacterial adhesion to intestinal cells and interference with viral attachment were evaluated in vitro. B. thermophilum RBL67 displayed adhesion indexes of 625 ± 84 and 1958 ± 318 on Caco-2 and HT-29 cells respectively and was comparable or superior to four other bifidobacteria, including B. longum ATCC 15707 and B. pseudolongum ATCC 25526 strains. Incubation of B. thermophilum RBL67 for 30 min before (exclusion) and simultaneously (competition) with human rotavirus strain Wa decreased virus attachment by 2.0 ± 0.1 and 1.5 ± 0.1 log10 (by 99.0% and 96.8% respectively). Displacement of virus already present was negligible. In CD-1 suckling mice fed B. thermophilum RBL67 challenged with simian rotavirus SA-11, pre-infection feeding with RBL 67 was more effective than post-infection feeding, reducing the duration of diarrhea, limiting epithelial lesions, reducing viral replication in the intestine, accelerating recovery, and stimulating the humoral specific IgG and IgM response, without inducing any adverse effect. B. thermophilum RBL67 had little effect on intestinal IgA titer. These results suggest that humoral immunoglobulin might provide protection against the virus and that B. thermophilum RBL67 has potential as a probiotic able to inhibit rotavirus infection and ultimately reduce its spread.


Introduction
Human rotavirus is the leading cause of severe dehydrating diarrhea in infants and young children worldwide, in both developed and developing countries. Peak incidence occurs in children competing human microbiota [39]. Its strong anti-Salmonella activity has been demonstrated in an in vitro continuous fermentation system simulating the juvenile intestinal ecosystem, along with its ability to rebalance the metabolic activity of gut microbiota after antibiotic treatment [38]. Its genome has been sequenced [40] and is available for genetic studies.
The aim of the present study was to evaluate the adhesiveness of B. thermophilum RBL67 to Caco-2 and HT-29 cells as well as its ability to interfere with rotavirus attachment. An in vivo study using suckling mice was also performed to assess its host-protection properties under intestinal conditions and its impact on the course of rotavirus infection, including diarrhea, virus replication, colon histology and the immune response.

Bacterial strains
Three strains of Bifidobacterium from the Research Network on Lactic Acid Bacteria (RBL Network) were used in this study, namely B. thermophilum RBL67, B. thermacidophilum RBL69 and B. thermacidophilum RBL70, isolated from stool samples obtained from infants [31,32] and preselected on the basis of resistance to gastrointestinal conditions. Two Bifidobacterium strains obtained from the American Type Culture Collection were used for comparison purposes: B. longum ATCC 15707 and B. pseudolongum ATCC 25526. All strains were cultured in MRS agar supplemented with 0.05% (wt/vol) L-cysteine-hydrochloride at 37°C under anaerobic conditions as described previously [41] and enumerated (in colony-forming units, cfu) on Beerens agar plates [42].

Cell cultures
All cell lines were cultured in Gibco media (Invitrogen, Burlington, Ontario, Canada) at 37°C under a 5% CO 2 atmosphere. Caco-2 cells (ATCC HTB-37) were grown routinely in Dulbecco's Modified Eagle medium as described previously [41]. HT-29 cells (ATCC HTB-38) were grown routinely in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 2 mmol/L L-glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin. Rhesus monkey kidney cell line MA-104 (ATCC CRL-2378.1) was routinely grown in Eagle's minimal essential medium supplemented with 10% FBS, 2 mmol/L L-glutamine, 1% nonessential amino acids, 1% HEPES buffer, 1.125 g/L sodium bicarbonate, 100 U/mL penicillin and 100 μg/mL streptomycin. For adhesion and inhibition assays, monolayers of human colon carcinoma cells (Caco-2 and HT-29) were seeded at a density of 10 4 cells per well in 24-well tissue culture plates (Falcon, Becton Dickinson and Co., Franklin Lakes, NJ). The culture medium was replaced every day, and the monolayer was used at post-confluence (10 6 cells/well) after 15 and 21 days for Caco-2 and HT-29 respectively. The medium was replaced with its antibiotic-free equivalent 18 h before performing the assays.

Viruses
Human rotavirus strain Wa (ATCC VR-2018) and simian rotavirus strain SA-11 (ATCC VR-1565) used for in vitro and in vivo assays, respectively, were propagated in MA-104 cells as described previously [6] with some modifications. Briefly, confluent cell monolayers were infected at an MOI of 1.5 with rotavirus pre-activated with 16 μg/mL of porcine trypsin (Sigma Chemical Co., St. Louis, MO) for 30 min at room temperature. After adsorption (1 h at 37°C), maintenance medium (MEM + 2% FBS) was added and cells were incubated for 48 h at 37°C with 5% CO 2 . The infected cultures were then frozen (-80°C) and thawed (37°C) three times, the contents centrifuged at 1,000 x g for 10 min and the supernatant dispensed into vials for storage at -80°C until use. Infectious viruses were quantified using the following cell based methods. Strain Wa titer was determined from a cell culture immunofluorescence assay [43] and expressed in focus-forming units (ffu) per mL. The titer of simian rotavirus strain SA-11 was determined by plaque assay [44] with some modifications and expressed in plaque-forming units (pfu) per mL. Briefly, confluent MA-104 cells in 6-well plates were washed with phosphate-buffered saline (PBS) and infected with 100 μL of pre-activated SA-11 at 37°C for 1 h in a 5% CO 2 atmosphere. Monolayers were then overlaid with MEM supplemented with 4% FBS, 1.2% agarose type II (Sigma) and 6.5 μg/mL trypsin. Plates were incubated for 3 days at 37°C (5% CO 2 ), fixed in 3.7% (v/v) formaldehyde and then stained with 0.1% (w/v) aqueous crystal violet solution to reveal lytic plaques resulting from the cytopathic effect.
In vitro adhesion of Bifidobacterium strains to Caco-2 and HT-29 cells Cells monolayers were washed twice with sterile PBS, drained, contacted with 250 μL of Bifidobacterium suspension (5×10 9 cfu/mL in PBS) per well, incubated for 60 min at 37°C under anaerobic conditions, then rinsed with PBS. Adherent bacteria were then collected in three washings of the monolayer with PBS after 15 min of trypsin-EDTA treatment at 37°C and enumerated by plating. Results are expressed as the adhesion index, which is the number of bacteria adhering per 100 cells.
In vitro antagonism against rotavirus attachment to Caco-2 and HT-29 cells Cell monolayers were washed twice with sterile PBS, drained, then contacted with 250 μL of pre-activated human rotavirus strain Wa suspension (5×10 6 ffu/mL) per well for 30 min after (exclusion), simultaneously with (competition) or 30 min before (displacement) the addition of 250 μL of B. thermophilum RBL67 suspension (5×10 9 cfu/mL) per well. Positive controls received PBS instead of B. thermophilum RBL67. Cells were then incubated for 90 min at 37°C under anaerobic conditions, rinsed with PBS, and attached virus was collected in three washings with PBS by repeated pipetting after 15 min of trypsin-EDTA treatment at 37°C. Attached virus was enumerated by cell culture immunofluorescence assay as described previously [43].

Mice
Timed pregnant CD-1 mice obtained from Charles River (St-Constant, Québec, Canada) were fed a standard laboratory rodent chow (Charles River Laboratories, Wilmington, MA), provided with water ad libitum, and kept in an animal room maintained at 22 ± 2°C with a 12 h light/dark cycle. Each dam and its litter were housed in a micro-isolator cage (Lab Products, Seaford, DE) on a cage ventilation rack (Lab Products) that provided a unidirectional flow of filtered air over the cage hoods. All dams were rotavirus antibody negative as measured by enzyme-linked immunosorbent assay (ELISA, described below). The experiments were performed with the approval of the animal care committee of Laval University "Comités de protection des animaux" (SIRUL number: 76543) and the mice were cared for and handled in conformity with the Canadian Guide for the Care and Use of Laboratory Animals [45]. No animal died during the study except for the euthanasia scheduled for the taking of biological samples.

Suckling mice treatment groups
Three experimental repetitions were done using pups from the litters of fifteen dams (10-14 pups per litter, total n = 189). Experiments were conducted on suckling mice because preliminary assays shown that adults mice are highly resistant to rotavirus infection and required either antibiotic pretreatment or a genetically modified mice model. Dams and their litters were assigned randomly to one of the five experimental groups (Table 1). B. thermophilum RBL67 was given orally at a daily dose of 1×10 9 cfu in 20 μL of PBS for 7 consecutive days, from the age of 3 days or 9 days depending on the experimental group. On day 9, simian rotavirus strain SA-11 not trypsin-activated was given orally at a single dose of 1×10 4 pfu in 20 μL of PBS with 10% blue food coloring (McCormick Co., London, Ontario, Canada) added as a tracer instead of human rotavirus Wa which is noninfectious to mice. Control animals received 20 μL of placebo (PBS with 10% blue food coloring). After each feeding and infection, the pups were returned to their mothers and allowed to suckle. Activity level, body weight and external signs of disease (diarrhea, dehydration) of mice were observed daily over the course of experiment. Mice were sacrificed under anesthesia by intraperitoneal injection of 0.01 mL/g of a ketamine/xylazine mixture (respectively 15 mg/mL and 1 mg/ml) for microbial and histological analysis and measurement of the immune response. All efforts were made to minimize animal suffering.

Enumeration of intestinal bifidobacteria and rotavirus
The intestines were removed aseptically from the sacrificed mice, weighed, placed in 1.0 mL of PBS with 5% antibiotic solution (for viral counts) or 0.5 g/L L-cysteine (for bacterial counts) and homogenized with an Ultra-Turrax (Janke and Kunkel, Staufen, Germany) on ice for 15 s at 13,500 rpm. For viral quantification, the suspension was then centrifuged at 13,500 rpm for 5 min to remove debris and frozen at -20°C until plaque assay. For bacterial quantification, the suspension was diluted immediately, plated on Beerens selective agar and incubated at 37°C for 72 h under anaerobic conditions. Isolates were then confirmed by Gram staining to be Grampositive rods and identified using API 50CH galleries (BioMérieux, Montréal, Québec, Canada) according to the manufacturer's instructions.

Histological analysis of the colon
Sections of colon were fixed in PBS containing 4% (w/v) paraformaldehyde (TAAB, Callera Park, Aldermaston Berks, England), embedded in paraffin, cut into 5 μm slices using a microtome (Model 2040, Reichert-Jung, Vienna, Austria) and stained with hematoxylin and eosin. A pathologist kept unaware of sample origin then examined the epithelium morphology using a light microscope (Leica Microsystems, Richmond Hill, Ontario, Canada). Images were acquired by a Hyper HAD camera (Sony Ltd., Willowdale, Ontario, Canada) and Image Matrox Inspector software 3.1 (Matrox Electronic Systems Ltd., Dorval, Québec, Canada). Quantification of the immune response in intestinal and serum samples by ELISA The level of intestinal rotavirus-specific IgA was determined by ELISA as described previously [46,47] with some modifications. Briefly, simian rotavirus strain SA-11 was fixed in a 96-well plate (8×10 5 pfu/mL in 2% formaldehyde-PBS solution). Diluted intestinal homogenate (500 mg/mL) was added, followed by peroxidase-labeled goat anti-mouse IgA and finally the O-phenylenediamine (OPD, Sigma) standard solution in the presence of H 2 O 2 . Optical density (OD) was measured at 450 nm on a Thermomax microplate reader (Molecular Devices, Opti-Ressources, Charny, Québec, Canada). Rotavirus-specific IgG and IgM in serum obtained by centrifuging (4,000 × g for 10 min at 4°C) cardiac blood from individual mice were determined using the same protocol except that peroxidase-conjugated goat anti-mouse (H+L, Kirkegaard and Perry Laboratories) was used. Intestinal fluid and serum from non-infected mice not fed Bifidobacterium were used as negative controls on each plate. Titers were calculated as log 10 (1/ Dc) where the cut-off dilution (Dc) was the dilution yielding twice the absorbance of the negative controls.

Statistical analyses
Adhesion indexes of Bifidobacterium strains and attachment of human rotavirus strain Wa to intestinal cells were compared among treatments using the one-way analysis of variance (ANOVA) general linear model followed by Tukey's HSD test. Body weight, simian rotavirus strain SA-11 titer, intestinal B. thermophilum RBL67 count and immunoglobulin concentration were compared among treatment groups using the one-way ANOVA general linear model followed by the Hsu-Dunnett test. Statistical significance was declared at P < 0.05. All statistical analyses were performed using JMP 1 software version 10.0 (SAS Institute Inc, Cary, NC).

B. thermophilum RBL67 decreased rotavirus attachment to intestinal cells in vitro
Inhibition of rotavirus attachment to intestinal cells by B. thermophilum RBL67 was evaluated using exclusion, competition and displacement assays (Fig 2). Preliminary 48-h trials indicated that the highest attachment to both intestinal cell lines was obtained after 1.5 h of contact time (S1 Fig). When rotavirus was added alone (positive control), 5.8 ± 0.1 and 5.7 ± 0.1 log 10 were attached to Caco-2 and HT-29 cell lines respectively. However, when rotavirus was added 30 min after the addition of B. thermophilum RBL67 (exclusion), the number of viruses attached dropped by two log cycles (to 3.8 ± 0.1). No significant difference between cell lines was observed (P > 0.05). In competition and displacement assays, rotavirus attachment was respectively 4.3 ± 0.1 log 10 and 5.6 ± 0.1 log 10 .  Intestinal concentration of Bifidobacterium strain RBL67 was not influenced by rotavirus infection The protective effect of B. thermophilum RBL67 observed in vitro was evaluated in suckling CD-1 mice distributed to five experimental groups (A to E, Table 1). Viable counts of Bifidobacterium in the intestinal contents were monitored for 72 h (Fig 3). In control mice (group A), bifidobacteria were detected at about 3.8 log 10 cfu/g, which remained stable throughout the experiment. Similar results (P > 0.05) were obtained with mice challenged with rotavirus (group C). In mice fed B. thermophilum RBL67 then challenged with rotavirus (group D), Bifidobacterium counts gradually decreased from 7.5 to 4.2 log 10 cfu/g and were no longer significantly different from control mice after 48 h. In contrast, counts increased from 4.2 to 6.6 log 10 cfu/g in mice that received B. thermophilum RBL67 for 7 days after challenge (group E) and were significantly higher than in control mice starting at 24 h.

Bifidobacterium strain RBL67 fed prior to challenge reduced diarrhea duration
Control mice (group A) did not develop diarrhea at any point. Mice challenged with the simian rotavirus strain SA-11 without receiving B. thermophilum RBL67 (group C) suffered from diarrhea beginning at 24-30 h post challenge and most intensely at 54 h. Reduced activity was observed during the 7 days post challenge. In contrast, mice that received B. thermophilum RBL67 for 7 days prior to challenge (group D) suffered from diarrhea later (starting at 48 h) and recovered earlier than did group C mice. The post challenge feeding with B. thermophilum RBL67 (group E) also delayed the onset of diarrhea slightly (30 h) without accelerating recovery. Body weight was significantly lower in groups C, D and E during the first 72 h post challenge (S2 Fig).

Pre-challenge feeding with B. thermophilum RBL67 limited intestinal replication of rotavirus
The course of rotavirus infection in suckling CD-1 mice was observed for 72 h by measuring strain SA-11 titer in the intestinal contents using ELISA (Fig 4). No rotavirus was detected in control mice (group A) or mice fed B. thermophilum RBL67 (group B). In mice not fed B. thermophilum RBL67 but challenged with strain SA-11 (group C), the titer increased from 2.9 log 10 initially, peaked to 5.2 log 10 after 24 h and decreased at 4.2 log 10 after 30 h, as described previously [49]. In mice that received B. thermophilum RBL67 for 7 days after challenge (group E), a 30 h delay in the increase from the initial titer (3.5 log 10 ) was noted. The titer then rose to 4.4 log 10 at 72 h. Pre-challenge feeding with B. thermophilum RBL67 (group D) appeared to control the infection, limiting the rotavirus titer to between 3.2 and 2.5 log 10 (significantly lower than in the other groups) for 72 h. Ingested Bifidobacterium strain RBL67 stimulated the humoral immune response in suckling mice

Pre-challenge feeding with B. thermophilum RBL67 limited intestinal lesions
The influence of B. thermophilum RBL67 on the immune response in suckling CD-1 mice was evaluated by measuring rotavirus-specific IgA in intestinal contents and IgG and IgM in serum (Fig 6). These immunoglobulins were undetectable in mice not challenged with the virus (groups A and B). Four days after challenge with rotavirus (group C), IgA reached 1 OD unit and remained stable until day 7. Ingestion of B. thermophilum RBL67 (group D) did not change this level significantly. Among group C mice, IgG and IgM reached 2.0 OD on day 7 and rose to 2.4 OD on day 14, and were even higher (2.9 OD) among group D mice (fed B. thermophilum RBL67 for 1 week prior to challenge) from days 7 through 14. Post-challenge feeding (group E) led to the highest level measured for IgG and IgM (3.7 OD), but not until day 14.

Discussion
Despite the effectiveness of vaccines against rotavirus, interest in alternative therapies such as probiotics remains keen, due particularly to the high cost of the vaccine and the lack of rotavirus vaccination programs in many developing countries [34, 50,51]. In the present study, the ability of the probiotic candidate B. thermophilum RBL67 to inhibit rotavirus infection was assessed in vitro using intestinal cell lines and then in vivo using suckling mice. Prior to in vivo study, the probiotic candidate should be screened for safety, resistance to gastric acidity and bile acid, adherence to human epithelial cells, and ability to reduce pathogen adhesion [19]. The long history of consumption of fermented milk and the growing body of knowledge on bifidobacteria taxonomy and physiology support the safety of the proposed use of bacteria such as B. thermophilum RBL67 [52]. This strain has been shown resistant to acidity [32] and should be resistant to bile since it possesses bile salt hydrolase [40]. We evaluated its adhesiveness (Fig 1) to two intestinal cell lines that present different functional characteristics, namely Caco-2 and HT-29 [53]. Its adhesion index on Caco-2 cells is similar to that of B. animalis subsp. lactis Bb12 [54] and B. pseudolongum ATCC 25526, which are considered highly adhesive [48]. The adhesion index of B. thermophilum RBL67 was also higher than that of Lb. rhamnosus GG (the most studied probiotic), estimated at 145 [55]. This ability might be associated in part with the presence of enolase and transaldolase genes, revealed by sequencing of the B. thermophilum RBL67 genome [40], which are involved in interaction processes with the host [56, 57].
An in vitro assay carried out to evaluate to what extent B. thermophilum RBL67 interfered with rotavirus attachment to intestinal cell lines revealed that this strain excludes rotavirus competitively. Among the three strategies tested, exclusion (contact with the cells for 30 min prior to challenge with the virus) and competition led to the greatest reductions in rotavirus attachment (Fig 2). The exclusion scenario simulates the presence of the probiotic in the intestinal lumen due to ingestion of bacteria as a supplement before infection. Rotavirus attachment was reduced by 2.0 ± 0.1 log 10 and 1.5 ± 0.1 log 10 in exclusion and competition assays respectively. To our knowledge, no such reduction has been reported previously in the literature. Reductions in the range of 9.0-49.8% (i.e. less than 1 log 10 ) have been observed previously for human rotavirus Wa attachment to MA-104 and HT-29 cells in the presence of 6 Bifidobacterium strains, whether in exclusion or competition assays [58], and in the range of 3.3-38% for attachment to Vero cells in the presence of 5 Bifidobacterium or 6 Lactobacillus strains [59]. In the present study, the third assay (displacement) simulating post-infection feeding led to negligible reduction of rotavirus attachment (0.2 ± 0.1 log 10 ), possibly explaining the poor health observed in group E mice.
Based on these results, an in vivo study was performed using suckling mice to evaluate the host-protective properties of B. thermophilum RBL67 and to determine whether or not daily intake could mitigate the risk of rotavirus infection. CD-1 suckling mice were chosen because of their susceptibility to rotavirus (induction of diarrhea and spread of the virus throughout the tissues in a reproducible manner) and they are well characterized, low cost, and easy to handle in large numbers [60, 61]. Ingestion of B. thermophilum RBL67 increased significantly the concentration of viable Bifidobacterium in the intestine, confirming that this probiotic candidate of fecal origin survived the various digestive stresses. The strain persisted in the intestine only for 2 days after daily consumption ceased (Fig 3). Although B. thermophilum RBL67 has been shown to grow in a simulated human juvenile intestinal environment [38], it did not likely persist in the intestinal microbiota of the suckling mice, since an indigenous population of Bifidobacterium was already present. Tannock, Munro [62] demonstrated that a stable population of lactobacilli prevented long-term colonization by Lb. rhamnosus DR20 ingested as a probiotic. In addition, B. thermophilum RBL67 did not induce any obvious adverse or toxic effect in suckling CD-1 mice.
Rotavirus induced diarrhea in unprotected mice (group C) over a period of 3 days, starting at 24-30 h post challenge. The onset of this symptom coincided with the increase of viral titer in the intestinal contents (Fig 4) and the appearance of lesions in the intestinal epithelium ( Fig  5). As expected [63,64], the infection triggered the production of rotavirus-specific intestinal IgA and humoral IgG and IgM. Daily ingestion of B. thermophilum RBL67 prior to infection (group D) had a moderate but clear protective effect, decreasing diarrhea and virus replication for at least 3 days. Such delays in the onset of diarrhea have been reported in association with other probiotic candidates such as B. longum [59] and Lb. rhamnosus GG [65]. In contrast, the published data concerning the control of virus shedding are discordant since this is likely strain-specific. B. longum [59], Lb. rhamnosus GG [65], Lb. acidophilus NCFM and Lb. reuteri ATCC 23272 [66] were reported to have no effect, unlike a combination of Lb. rhamnosus GG and B. animalis subsp. lactis Bb12, which controlled virus shedding until 4 days post challenge [67]. A significant initial reduction in rotavirus shedding has been noted following pre-challenge feeding with B. longum subsp. infantis CECT 7210, but this was not maintained beyond day 2 post challenge [58]. Daily ingestion of B. thermophilum RBL67 prior to challenge reduced damage to the intestinal epithelium, which recovered its normal architecture by 54 h post challenge. Some preservation of vacuolated enterocytes has been associated with Lb. rhamnosus GG, although normal architecture was still not restored 6 days post challenge [65]. Post-challenge feeding with B. thermophilum RBL67 (group E) provided less conclusive results than did pre-challenge feeding (group D), in which the increase in intestinal rotavirus titer and damage to the epithelium were limited to the 30-54 h period before returning to levels observed in control mice (group B). Nevertheless, these effects are similar to those observed with pre-challenge ingestion of B. longum subsp. infantis CECT 7210 [58].
Several mechanisms have been proposed to explain the efficacy of probiotics in the prevention and treatment of diarrheal diseases. The possible mechanisms include among others production of acidity, short-chain fatty acids and antimicrobial substances, normalization of perturbed microbiota, increased turnover of enterocytes, competitive exclusion of pathogens, improved barrier function, and stimulation of immune responses to pathogens [20,68]. The moderate protective effect associated with B. thermophilum RBL67 intake could be due partly to competitive exclusion of rotavirus, tight junction strengthening [39] and stimulation of the immune response. Indeed, recovery from infection is correlated primarily with rotavirus-specific antibody production [69]. We found no significant stimulation of IgA in response to ingestion of B. thermophilum RBL67. Regarding this immunoglobulin, conflicting results have been reported: some researchers have observed stimulation [70][71][72][73] while others did not [58]. However, ingestion of B. thermophilum RBL67 was associated with an obvious specific humoral response reflected in IgG and IgM (Fig 6). This was accelerated when B. thermophilum RBL67 was already present in the intestinal lumen at the time of challenge compared to when it was ingested post challenge. Others have shown that probiotic candidates Lb. acidophilus NCFM and Lb. reuteri ATCC 23272 increase total small intestinal IgM and IgG levels in rotavirus-infected gnotobiotic piglets [66]. These results suggest that the immunomodulatory effect of a probiotic is strain-specific [74]. Although IgA is thought to provide the most effective protection against rotavirus in the intestine, evidence suggests that serum IgG or IgM in sufficient quantities can reach the intestinal epithelium and provide additional protection [69,[75][76][77]. This study does not exclude other modalities of host immune response, in particular the effect on cytokine-secreting cells [78], which should be investigated in a future study.
In summary, the probiotic candidate Bifidobacterium thermophilum strain RBL67 reduced the duration of diarrhea, limited epithelial lesions, controlled rotavirus replication in the intestine, stimulated the humoral specific IgG and IgM response, and shortened the time of recovery from symptoms in CD-1 suckling mice. Furthermore, B. thermophilum RBL67 displayed high adhesiveness and competitiveness and hence interference with rotavirus attachment to intestinal cells in vitro. These functions might contribute to the mechanisms underlying the moderate protective effect of B. thermophilum RBL67 against rotavirus in this model of infection. In addition, B. thermophilum RBL67 meets several criteria for probiotic bacteria: presumed safety, resistance to gastric acidity and bile acids, adherence to human epithelial cells, ability to reduce virus attachment and to reduce the symptoms of infection in suckling mice, and its genome now fully sequenced for scientific scrutiny. Further studies are required in order to identify the mechanisms underlying the apparent beneficial effects of B. thermophilum RBL67 and to confirm our results in a more human-like model (such as piglets) before proceeding with human trials. If these beneficial effects are also observed in human, a prophylaxis use of B. thermophilum RBL67 coupled with supplementary or complementary treatment during the winter seasonal peaks of gastroenteritis could be envisaged in children.

S1 Fig. Attachment of human rotavirus strain Wa to intestinal cells: (A) Caco-2, (B) HT-29.
Monolayers were contacted with rotavirus suspended at a titer of 6.9 log 10 ffu/mL and attachment was measured directly by immunofluorescence assay. Error bars indicate the standard error of the mean. Different letters indicate significant difference between assay times (Tukey's HSD test, P < 0.05, n = 3).  Table 1

for details).
Error bars indicate the standard error of the mean. An asterisk indicates statistical significance compared to ◆ (Tukey's HSD test, P < 0.05, n = 189