J.R. Nicoli, Departamento de Microbiologia, Instituto de Ciências Biológicas, Universidade Federal de Minas Gerais, C.P. 486, 30161-970, Belo Horizonte, MG, Brazil (e-mail: firstname.lastname@example.org).
Aims: The effect of lactic acid bacteria on the immune system is well established under normal conditions and generally by in vivo determinations, but few data are available, in vivo, during an infectious challenge. The objective of this study was to obtain data on the putative protective role of bifidobacteria upon challenge with an intestinal pathogen.
Methods and Results: The effect of oral treatment with Bifidobacterium longum Bb46 on intragastric challenge with Salmonella Typhimurium was studied. Faecal bacterial levels were determined in gnotobiotic (GN) mice and mortality, histopathology (intestines, liver), immunoglobulin levels (IgM, IgG, IgG1, IgG2a) and cytokine production (IFN-γ, IL-10) were determined in conventional (CV) mice. Conventional mice received 0·1 ml probiotic milk (108 CFU) daily, 10 days before the oral pathogenic challenge (102 CFU). Then, probiotic treatment was continued until the end of the experiment. Probiotic treatment in germ-free mice consisted of a single dose at the beginning of the experiment. Control groups were treated with sterile skim milk and submitted to the same procedure. A higher survival (40%) was observed for probiotic-treated animals when compared with the control group (0%). This protective effect was confirmed by the histopathological and morphometric data. However, S. Typhimurium population levels in the faeces were similar among control and probiotic-treated groups. During the challenge with S. Typhimurium, a decrease in IFN-γ and IgG2a productions was observed in probiotic-treated mice.
Conclusions: The protective effect against the pathogenic challenge may be due to a reduced inflammatory response, mediated by the probiotic treatment, but not to a population antagonism.
Significance and Impact of the Study: Results suggest that dietary supplementation with B. longum could provide benefits against enteric infection.
The development of an immune response toward foreign antigens of microbial and viral origin provides a means through which potentially pathogenic organisms are recognized and controlled. There is an extensive body of literature addressing the possible health benefits associated with the consumption of probiotics, such as lactic acid bacteria (Streptococcus, Lactobacillus, Lactococcus) and bifidobacteria, which are known to exert a modulating effect on either humoral or cellular immune responses.
Lactobacilli have the longest history as probiotics and pertain to the predominant gastrointestinal microbiota of laboratory and farm animals (Tannock 1997). However, bifidobacteria colonize the human neonatal intestine soon after birth and inhabit the gastrointestinal tract throughout life, harbouring larger and more stable populations than lactobacilli (Kimura et al. 1998). Recently, the determination of Bifidobacterium longum genome sequence revealed several physiological traits that could explain the successful adaptation of these bacteria to the human colon in terms of metabolic and immunomodulatory activities, and adhesion ability (Schell et al. 2002). Additionally, B. longum is found in the digestive tracts of infant as well as of adult and elderly subjects and its oxygen tolerance can be considered as a technological advantage for biomass production when compared with more strict anaerobic species such as B. bifidum and B. adolescentis (He et al. 2001; Reuter 2001). For these reasons, bifidobacteria are considered as more adequate probiotics for prevention and/or treatment of human intestinal disorders than lactobacilli. Peptidoglycan produced by bifidobacteria induces immunopotentiation or sequential stimulation of innate immune responses in a wide variety of hosts (Sasaki et al. 1996). Some strains of bifidobacteria have been shown to enhance cellular immune responses, including phagocytosis, lymphocyte proliferation and cytokine production (Schiffrin et al. 1997) as well as humoral immune responses (Lee et al. 1993). This enhancement seems to correlate with an increased resistance to intestinal infection with Salmonella Typhimurium (Silva et al. 1999; Shu et al. 2000).
While the effect of these probiotics on the immune system is well documented and established under healthy conditions and generally by in vivo experiments, there are few data available on this influence during an in vivo experimental challenge with an infectious agent. The present study was conducted to investigate the effect of oral administration of B. longum on morbidity and mortality as well as on possible antagonistic relations in the gastrointestinal ecosystem and some immunological aspects in mice challenged with S. Typhimurium.
Material and methods
Germ-free and conventional 21-day-old Swiss/NIH mice (Taconic, Germantown, NY, USA) of both sexes were used in this study. Germ-free animals were housed in flexible plastic isolators (Standard Safety Company, McHenry, IL, USA) and handled according to established procedures (Pleasants 1974). Experiments with gnotobiotic mice were carried out in microisolators (UNO Roestvastaal B.V., Zevenar, The Netherlands). Conventional NIH mice were derived from the germ-free colony and only used after at least two generations following the conventionalisation. Water and commercial autoclavable diet (Nuvital, Curitiba, PR, Brazil) were sterilized by steam and administered ad libitum to all the animals. Conventional mice were maintained in a conventional animal house with controlled lighting (12 : 12 hours, light : dark). All experimental procedures were carried out according to the standards set forth in the ‘Guide for the Care and Use of Laboratory Animals’ of the National Research Council (1996).
Bifidobacterium longum Bb46 isolated from a lyophilized commercial product (DVS; Christian Hansen Laboratory, Horsholm, Denmark) was used to prepare the probiotic milk for the oral treatment of mice. Salmonella enterica serovar Typhimurium was isolated from a human being at the Fundação Ezequiel Dias (FUNED; Belo Horizonte, MG, Brazil). The bacteria were maintained at −86°C in medium containing 40% glycerol. The identity of each bacterium was regularly confirmed by Gram stain, and using API 20A and API 20E identification kits (BioMérieux, Marcy-L'Etoile, France), respectively.
Probiotic milk was prepared from 9·5% reconstituted skim milk, as recommended by the manufacturer (Christian Hansen Laboratory). A dose of 0·1 ml probiotic milk containing ca 109 colony forming units (CFU) ml−1 of B. longum was daily administrated by gavage to each conventional mouse (experimental group), 10 days before the challenge with S. Typhimurium, and then throughout the remaining experimental period. The same dose (108) was inoculated once in germ-free mice at the beginning of the experiment. Control groups were treated with 9·5% reconstituted skim milk, according to the same schedule as the probiotic-treated groups.
Salmonella Typhimurium was grown in brain–heart infusion (BHI) medium (Oxoid Ltd, Basingstoke, Hampshire, UK) at 37°C. Each mouse was orally challenged with 0·1 ml of bacterial suspension containing 102 viable cells, except for the determination of mortality for which a 105-CFU dose was used. Survival of conventional mice was monitored until the 28th day after challenge with S. Typhimurium.
Microbial counts in gnotobiotic mice
Freshly collected faeces from gnotobiotic mice were introduced into an anaerobic chamber (Forma Scientific Company, Marietta, OH, USA) containing an atmosphere of 85% N2, 10% H2 and 5% CO2, diluted 100-fold (w/v) in regenerated sterile buffered saline and homogenized by hand. Serial 10-fold dilutions were obtained and 0·1 ml was plated onto MacConkey agar (Oxoid Ltd) and de Mann, Rogosa and Sharpe (MRS) agar (Merck KGaA, Darmstadt, Germany) for Salmonella and Bifidobacterium counts, respectively. The Petri dishes containing MacConkey agar were withdrawn from the anaerobic chamber and cultured at 37°C during 24 h under aerobic conditions whereas MRS Petri dishes remained for 72 h at 37°C in the anaerobic chamber. After these incubation periods, colonies were counted.
Histopathologic and morphometric analysis in conventional mice
Tissue samples from intestines and liver from mice killed by ether inhalation were fixed in 10% formaldehyde and processed for paraffin embedding. The histopathological sections (5 μm) were stained with haematoxylin–eosin. The slides were coded and examined by a single pathologist, who was unaware of the experimental conditions of each group. The morphometry of the samples was performed using a JVC TK-1270/RGB microcamera and the KS 300 Software built in a Kontron/Carl Zeiss image analyser (Oberkohen, Germany). In the intestines, villous height was determined in five fields with an average of 12 measures. In the liver, the degree of inflammatory infiltration was considered as a damage index for hepatic tissue and inflammatory cells (mononuclear and polymorphonuclear cells) were counted for each 100 hepatocytes. Küpffer cells were also counted for 100 hepatocytes.
Cytokine quantitation in spleen cultures of conventional mice
Groups of mice were killed 10 days after B. longum treatment and 5 days after the pathogenic challenge, spleens were removed, macerated aseptically in RPMI-1640 (Sigma Chemical Co., St Louis, MO, USA) and centrifuged at 700 g for 10 min. Erythrocytes were lysed and spleen cells were washed twice. Cells were suspended in RPMI-1640 supplemented with 10% foetal bovine serum (Nutricell, Campinas, SP, Brazil), 0·05 mmβ-mercaptoethanol (Sigma), 10 mg ml−1 gentamycin sulphate and 3·2 mml-glutamine (Sigma). Cell number and viability were assessed by trypan blue dye exclusion on a Neubauer haematocytometer (Boeco, Hamburg, Germany) and the final cell suspension adjusted to 5 × 106 cells ml−1. Cells were cultured in 24-well tissue culture plates in the absence (unstimulated) and in the presence of 5 μg ml−1 concanavalin A (Sigma) or heat-killed B. longum (1 : 1). All cultures were performed in triplicates, and three animals were used for each time point. Cultures were incubated for 12, 24, 48 and 72 h at 37°C in 5% CO2. After incubation, supernatants were harvested and stored at −86°C for cytokine assay. Duoset kits for mouse IFN-γ and IL-10 (R&D Systems, Minneapolis, MN, USA) were used to determine cytokine levels in culture supernatants according to the manufacturer's instructions. Absorbance was read at 450 nm on a microplate reader (Bio-Rad Laboratories, Hercules, CA, USA).
Serum IgM, IgG, IgG1 and IgG2a were evaluated by capture ELISA after 10 days of B. longum or control treatment and 5 days after challenge with S. Typhimurium. Levels of IgM and IgG in the sera of mice were determined using goat anti-mouse IgM and IgG (Sigma) and horseradish peroxidase-conjugated goat anti-mouse IgM and IgG (Sigma). To detect IgG1 and IgG2a isotypes, biotinylated rat anti-mouse IgG1 and IgG2a antibodies (Southern Biotechnology Associates Inc., Birmingham, AL, USA) were used. Absorbance at 492 nm was determined with an ELISA plate reader (Bio-Rad). The concentrations of each immunoglobulin were determined using the respective purified mouse standard (Southern Biotechnology Associates).
The results are expressed as the average of at least three independent experiments. Statistical significance of all results was evaluated by the Mann–Whitney rank sum test for all data except survival, for which log rank test was used. The level of significance was set at P < 0·05.
Oral challenge with S. Typhimurium in conventional mice
Figure 1 shows that survival on day 28 after the oral challenge with S. Typhimurium was significantly higher (P = 0·05) in the probiotic-treated group (40%) when compared with the control group (0%).
Histopathological examination and morphometric data of organs from probiotic-treated (experimental) and control groups confirmed the survival data. The experimental animals had milder intestinal and liver lesions than controls. The intestinal lesions were markedly expressed as dilatation of the intestinal loop, villous hypotrophy, necrosis areas, mucosa erosion and oedema (Fig. 2). However, a higher reactional hyperplasia of the lymphoid component was observed in the intestinal mucosa of probiotic-treated mice. In the liver of challenged control animals, inflammatory infiltrate in multiple generalized foci of polymorphonuclear and mononuclear cells was noted, whereas these observations were practically absent in experimental mice. A computerized morphometric analysis of some parameters in liver and intestines, 6 days after the pathogenic challenge, is presented in Fig. 3. Villous height was higher (P = 0·011) in small intestine of experimental conventional mice when compared with the control group (Fig. 3a). In the liver, inflammatory cells number (mononuclear and polymorphonuclear cells) for each 100 hepatocytes was much higher (P = 0·029) in control than in experimental animals (Fig. 3b), but the numbers of Küpffer cells/100 hepatocytes were similar (P = 0·114) between the two groups (Fig. 3c).
Faecal bacterial counts in gnotobiotic mice
Figure 4 shows that in monoassociation, B. longum was well established in the digestive tract of the experimental gnotobiotic mice and the number of CFU reached a maximum of ca 1010 g−1 faeces. After the challenge with the pathogen, the B. longum faecal population decreased to levels ranging from 107 to 108 CFU g−1 faeces. The kinetics leading to the establishment of S. Typhimurium in the experimental and control groups are also shown in Fig. 4. In gnotobiotic group harbouring B. longum, the pathogenic bacteria became established at levels varying from 109 to 1010 CFU g−1 and remained at these high levels until death or killing. These levels were equivalent to those observed in gnotobiotic mice harbouring the pathogenic bacteria alone.
Cytokine production in spleen cell cultures of conventional mice
The production of IFN-γ (Fig. 5) and IL-10 (Fig. 6) in spleen cell cultures from control and experimental mice before and after the oral challenge with the pathogenic bacteria was investigated as a possible mechanism to explain the protective effect exerted by B. longum against S. Typhimurium. In all situations, higher levels of IFN-γ were observed in cultures from control animals stimulated by both concanavalin A (P = 0·024) or B. longum (P = 0·008) when compared with the experimental mice. For IL-10, there was no difference among experimental and control groups for either stimulation, as well as before and after the challenge.
Immunoglobulin production in sera of conventional mice
A tendency to higher production of total IgM (P = 0·057) and total IgG (P = 0·057) in the sera of mice treated with B. longum was observed only after the challenge with the pathogenic bacteria (Fig. 7). Figure 8b shows a decrease in IgG2a (P = 0·016) in the sera of mice treated with B. longum, after the challenge.
Several studies have indicated that physiological disorders and infectious diseases are frequently associated with a disturbance of human intestinal microbiota characterized by reduction in populations of bacteria which play an essential role in the development and well-being of the host (Berg 1996). For example, in infective and antibiotic-associated diarrhoea, constipation, Crohn's disease and gastritis, the faecal levels of bifidobacteria were found to be significantly decreased (Mitsuoka 1990). Probiotics, such as lactic acid bacteria and bifidobacteria, have been reported to be useful in the treatment of disturbed intestinal microbiota and diarrhoeal diseases (Marteau et al. 2001). Bifidobacteria are particularly attractive as potential probiotics for human beings as they constitute one of the predominant populations of the normal colonic microbiota and are extremely well adapted to this ecosystem. Bifidobacterium longum protected mice against the challenge with S. Typhimurium, as demonstrated by survival (Fig. 1), histopathological (Fig. 2), and morphometric (Fig. 3) data. Four mechanisms have been usually attributed to probiotics to explain their protective effects: antagonism by the production of substances which inhibit or kill the pathogen (Vandenbergh 1993), immunomodulation of the host (Perdigon et al. 1994; Neumann et al. 1998), competition with the pathogen for adhesion sites or nutritional resources (Bernet et al. 1994) and inhibition of the production or action of bacterial toxin (Corthier et al. 1985).
Inhibition of growth of several enteropathogenic bacteria by probiotics has been demonstrated but generally only in in vivo experiments (Liévin et al. 2000; Lee et al. 2003). Probiotics can survive in but not colonize the digestive tract of human beings and animals harbouring a complex intestinal microbiota and this is the reason why daily ingestion of these biotherapeutics is necessary to artificially maintain them in high and functional levels in the gastrointestinal tract. However, probiotic implantation is possible in germ-free animals using a single dose. For these reasons, the gnotobiotic animal provides in vivo simplified model that allows the observation of interrelationships between few microbial strains inoculated in the gastrointestinal ecosystem. Liévin et al. (2000) examined 14 human Bifidobacterium strains for antimicrobial activity against S. Typhimurium and showed that two among them expressed antagonistic activity in vivo. In germ-free mice, these two strains colonized the intestinal tract and protected mice against S. Typhimurium lethal infection. However, in this study, the faecal levels of S. Typhimurium were not determined and compared between germ-free and Bifidobacterium-monoassociated mice orally challenged with the pathogenic bacteria. For this reason, it was not possible to confirm if the protective effect observed in vivo was because of the antagonistic activity reported in vivo. The present study described quite similar initial experiments, using a commercial strain of B. longum in conventional mice challenged with S. Typhimurium. However, in addition to showing a protection against challenge with S. Typhimurium, we showed that the protection against the pathogenic bacteria observed in mice previously monoassociated with B. longum was not because of an antagonistic activity, as demonstrated by similar faecal population levels of S. Typhimurium among experimental and control gnotobiotic animals (Fig. 4). The Bifidobacterium used here also produced in vivo nonidentified diffusible antagonistic substance(s) against S. Typhimurium and other Gram-negative enteropathogenic bacteria (data not shown). These data showed the limitation of results obtained in in vivo experiments as well as the importance of the gnotobiotic animal model to obtain or dismiss the in vivo antagonistic effect. However, Asahara et al. (2001) showed that treatment of conventional mice with certain bifidobacteria (B. breve and B. pseudocatenulatum, 108 CFU per mouse) together with prebiotics prevents the antibiotic-induced disruption of colonization resistance to oral infection with S. Typhimurium and that the metabolic activity needed to produce organic acids and lower the intestinal pH is important in this anti-infectious property.
As an antagonistic activity was not observed in gnotobiotic mice to explain the protection exerted by B. longum against S. Typhimurium, the effect of the exposure to the bifidobacteria on some aspects of the immune response before and after the pathogenic challenge was investigated. It is well known that Küpffer cells represent one of the largest populations of tissue macrophages and, together with neutrophils, have been implicated as responsible for clearance of bacteria from the bloodstream (Hirakata et al. 1991). An increase in the number of Küpffer cells in the livers of Lactobacillus delbrueckii- and Saccharomyces boulardii-monoassociated mice compared with germ-free mice has been described and this increase was simultaneous with a higher clearance of pathogenic bacteria injected systemically (Neumann et al. 1998; Rodrigues et al. 2000). Numbers of Küpffer cells in the livers of B. longum-associated and germ-free mice were similar (Fig. 3c) and neither did the clearance capacity (data not shown).
Inflammation is a complex reaction of the immune system that involves the accumulation and activation of leukocytes and plasma proteins at sites of infection, toxin exposure, or cell injury. While inflammation exerts a protective function in controlling infections and promoting tissue repair, it can also cause tissue damage and disease. It has been reported that in response to Gram-negative bacteria, mice produce pro-inflammatory cytokines such as TNF-α, IL-12 and IFN-γ, which have been implicated in the resistance to enteric bacterial infections, such as by S. Typhimurium (Nauciel and Espinasse-Maes 1992). Various previous reports showed that, in vivo, bifidobacteria enhance production of pro-inflammatory cytokines by human and murine cell lines (Marin et al. 1997). Spleen cells from mice given B. lactis produced significantly higher amounts of IFN-γ in response to stimulation with concanavalin A than cells from control mice (Gill et al. 2000). A significant simultaneous increase in the phagocytic activity of peripheral blood leucocytes and peritoneal macrophages was also reported. These results contrast with an in vivo study (Nicaise et al. 1993), which has shown that implantation of B. bifidum in gnotobiotic mice did not increase production of pro-inflammatory cytokines above levels observed for macrophages from the germ-free animals. Additionally, more recent data showed that bifidobacteria from different origins induce cytokine secretion by a murine macrophage-like cell line in a strain- and species-specific manner. Some of the bacterial strains induced significantly more pro-inflammatory cytokine secretion whereas others down-regulated the inflammatory response (He et al. 2002). Peptidoglycans are major components of the cell wall and are known to induce pro-inflammatory cytokines. The different species of bifidobacteria have been found to vary greatly as to the induction of cytokine production. Furthermore, studies have indicated that the cell wall from different strains of the same species (B. adolescentis) caused different pro-inflammatory responses because of the differences in the peptidoglycan component (Zhang et al. 2001). On the contrary, showing an additional indirect influence, the specific immune response to Bacteroides thetaiotaomicron, another component of the predominant human faecal microbiota, was significantly lower in gnotobiotic rats simultaneously associated with B. adolescentis than in animals only monoassociated with Bact. thetaiotaomicron (Scharek et al. 2000). This effect was more pronounced in rats that had been associated sequentially. Succinctly, contact with bifidobacterium induces different patterns of immunoregulatory cytokines depending on the experimental conditions (in vivo or in vivo) and on the use of different bacterial species or even of different strains of the same species. The mechanistic basis for depressing cytokine expression is unclear, but the effect appears to be analogous to tolerance observed in human beings, experimental animals and cell cultures that are repeatedly exposed to microbial components (Mengozzi and Ghezzi 1993). Extended exposure to bacterial components from bifidobacteria may generate down-regulatory signals, thereby causing hyporesponsiveness. This modulation is supported by the observations that macrophage activity is increased after yoghurt consumption for 5 days, but it is reduced after 7 and 10 days (Perdigon et al. 1994). Thus, it might be speculated that suppression of some pro-inflammatory cytokines, such as IFN-γ, could actually attenuate the severity of infection. Treatment of IL-10 gene-deficient mice with a probiotic product based on bifidobacteria (VSL no. 3), resulted in normalization of colonic physiological function and barrier integrity in conjunction with a reduction in mucosal secretion of TNF-α and IFN-γ (Madsen et al. 2001).
In conclusion, the protective effect of the B. longum strain used in the present study against the pathogenic challenge is not caused by population antagonism, but may be due to a modulated inflammatory response. However, antagonistic and immunomodulatory mechanisms are not the only ones that could explain the protective effect of Bifidobacteria against experimental enteric infections. The competition for nutrients or adhesion sites and the modulation of toxin production or action must also be considered. As examples for the last hypothesis, some reports showed the inhibitory activities of bifidobacteria on Cl. difficile cytotoxin (Corthier et al. 1985), aflatoxin B1 (Oatley et al. 2000) and Vero cytotoxin (Kim et al. 2001). Studies investigating such other possible mechanisms are under way in our laboratory.
This study was supported by grants and fellowships from the Conselho Nacional do Desenvolvimento Científico e Tecnológico (CNPq), Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG) and Christian Hansen Laboratories (Valinhos, Brazil). The authors are grateful to Maria Gorete Barbosa Ribas for valuable technical help, and to Ronilda Maria de Paula (in memoriam), Maria Helena Alves de Oliveira and Antônio Mesquita Vaz for animal care.