Probiotics are live microbial feed supplements which beneficially affect the host animal by improving its intestinal microbial balance (
10,
11). The potential benefits that are claimed include improved nutrition and growth and prevention of various gastrointestinal disorders. Probiotic-containing products are available for human nutrition, as animal feed supplements, and also for aquaculture (
35,
36,
41,
43). In some countries probiotics are taken as prophylactic agents (for example, to prevent childhood diarrhea), while in southeast Asia they are also used as therapeutic agents (
25). Products containing endospores of members of the genus
Bacillus (in single doses of up to 10
9 spores/g or 10
9 spores/ml) are used commercially as probiotics, and they offer some advantages over the more common
Lactobacillus products in that they can be stored indefinitely in a desiccated form (
25). Originally, many commercial products were sold as products that carry
Bacillus subtilis spores, but recent studies have shown that most products are mislabeled and carry other
Bacillus species, including
Bacillus clausii,
Bacillus pumilus, and a variety of
Bacillus cereus strains (
13,
17). Product mislabeling raises a number of concerns about consumer confidence (
15), as well as attendant safety issues, since some of the organisms found were strains of
B. cereus, which is a major cause of gastrointestinal infections (
12).
Continued ingestion of large quantities of
Bacillus spores raises the question of what happens to the spores in the gastrointestinal tract (GIT). While no evidence of colonization has been found, it is possible that a spore can interact with the gut-associated lymphoid tissue (GALT). Recent studies have shown that orally ingested
B. subtilis spores are immunogenic and can disseminate to the Peyer's patches and mesenteric lymph nodes (MLN) (
5,
6). Additional work has provided compelling evidence that ingested
B. subtilis spores can germinate in the small intestine. This conclusion is based on three findings. First, when mice are given an oral inoculum, more spores are excreted than are ingested (
18). Second, vegetatively expressed mRNA is detected in the GIT by reverse transcription (RT)-PCR following administration of spores to mice (
2). Finally, systemic immunoglobulin G (IgG) responses are generated against vegetative
B. subtilis following administration of suspensions carrying only spores to mice (
5). Together, these studies show that spores may not be transient passengers in the gut or that if they are, they may still have an intimate interaction with the host cells or microflora that can enhance their potential probiotic effect.
The following three basic mechanisms have been proposed for how orally ingested nonindigenous bacteria can have a probiotic effect in a host: (i) immunomodulation (that is, stimulation of the GALT) (e.g., induction of cytokines), (ii) competitive exclusion of gastrointestinal pathogens (e.g., competition for adhesion sites), and (iii) secretion of antimicrobial compounds which suppress the growth of harmful bacteria (
10). Few studies have demonstrated a direct probiotic effect of
Bacillus spores, but preliminary studies with poultry have provided evidence that there is competitive exclusion of
Escherichia coli 078:K80 by
B. subtilis (
24) and a number of studies have demonstrated that
Vibrio harveyi in shrimp is suppressed by various
Bacillus spore formers (
34,
42). A recent study has described the characterization of an antibiotic produced by the
B. subtilis strain (
B. subtilis 3) found in the commercial product Biosporin, which has been shown to inhibit growth of
Helicobacter pylori (
31).
DISCUSSION
Previous studies with the laboratory strain PY79 have provided evidence that germination, growth, and resporulation occur in the GIT (
2,
18). This is based on the finding that when mice are inoculated orally, on some occasions more spores are excreted than are inoculated (
18). Probiotic strains may have a beneficial effect by persisting in the GIT, perhaps by association with the mucosa, and we addressed this possibility by examining the persistence of different probiotics in the mouse GIT (note that these experiments are different from those described previously [
18] since in this study we examined spore counts at distinct times rather than counts recovered in the total feces collected between time points). We found that all
B. cereus probiotics appeared to persist longer in the GIT and that after 15 days significant numbers were still present. In contrast, the
B. subtilis,
B. clausii, and
B. pumilus strains appeared to be cleared from the gut. This corresponds with studies showing that
B. cereus spores can efficiently adhere to human epithelial cells (Caco-2), an attribute related to the hydrophobicity of spores (
1). This feature, of course, may be important for germination and infection within the small intestine. Interestingly, the laboratory strain (PY79) was rapidly cleared from the GIT within just 5 days, and it is possible that as a frequently passaged laboratory strain this bacterium has lost one or more of its natural traits which would otherwise promote persistence within the GIT. We noted that the exosporium, which is the outermost layer of the spore, is absent from the laboratory strains of
B. subtilis (
16). Interestingly, the exosporium also seems to be absent from at least three newly discovered fecal isolates of
B. subtilis (Barbosa and Henriques, unpublished results) but is found in the other
Bacillus probiotic strains (
17). It should be interesting to determine whether the presence of this structure correlates with increased persistence of spores in the GIT. One of the
B. cereus strains, Biosubtyl
DL, has been shown to be facultatively anaerobic (
17). This strain did not appear to persist longer in the mouse GIT, so we assume that the ability of this strain to grow in anoxic conditions does not enable enhanced persistence.
One surprising result of this work was that spores of some of the probiotic strains were apparently sensitive to SGF and SIF. In the case of Subtyl this sensitivity was acute, and only a tiny fraction of the spores survived incubation in SGF and SIF. We do not believe, however, that the spores are themselves sensitive to acid; rather, we provide two explanations to account for these results. First, spore germination is activated by the low pH since acid activation of spore germination (as opposed to the more common heat-dependent activation) is known to promote germination of spores of
B. cereus (
8,
22) and rapid synchronized germination in the SGF and SIF could account for this rapid loss of viability. Second, Subtyl spores are intrinsically more sensitive to the extraction procedure which we used to prepare spores (most probably the high-pressure treatment with the French press), which makes them sensitive to the SGF. The second explanation is supported by the drop (50%) in viability of spores suspended only in water. If the former hypothesis is correct, however, and germination of Subtyl spores is acid sensitive, then in a natural environment physiological conditions (food composition and fluctuations in pH, etc.) should ensure that a greater number of spores survive, and our analysis of fecal counts did appear to show that in vivo Subtyl can escape cell death, which supports this explanation. Interestingly, in other work it has been shown that germination of spores of the laboratory strain
B. subtilis PY79 is partially inhibited in the presence of SIF (
5), so clearly conditions within the small intestine can have opposing effects on spores.
The most important finding of this work is that the three
B. cereus probiotics, Bactisubtil, Biosubtyl
DL, and Subtyl, produce enterotoxins. Both Biosubtyl
DL and Bactisubtil produce the Hbl enterotoxin, which is the primary virulence factor of
B. cereus food poisoning (
12,
23), while Biosubtyl
DL and Subtyl produce the Nhe enterotoxin. Although no commercial test is yet available to detect enterotoxin T, the presence of the structural gene of this toxin in Biosubtyl
DL and Bactisubtil suggests that the toxin could be produced under favorable conditions. One additional enterotoxin found in some
B. cereus strains, the 1.2-kDa emetic toxin (cereulide), was not tested for, primarily because strains producing this preformed toxin cause emesis at a low dose (total dose, 10
3 to 10
4 CFU/g), so it is not possible that an emetic strain is being used for human probiotics. The typical dose of spore probiotics is between 10
7 and 10
9 spores, and the total infective dose of pathogenic
B. cereus required to produce diarrhea is between 10
5 and 10
7 spores (
12). Food-poisoning
B. cereus spores adhere to the mucosal epithelium of the small intestine, germinate, and are able to produce enterotoxins that induce diarrhea (
12,
23). It seems probable, then, that the three
B. cereus probiotics tested here behave similarly. This appears to be a paradox, but there are several points that should be clarified. First, enterotoxins are not always produced, and the microenvironment (adhesion and competition with other commensal bacteria, food intake, luminal pH, etc.) within the GIT may affect enterotoxin production, as well as adhesion to the mucosa. In the case of Subtyl it is possible that germination of spores in the stomach and small intestine significantly reduces the infective dose, presumably explaining why food poisoning does not result. Although
B. cereus-based food poisoning is short-lived, an interesting and controversial concept is whether exposure to repeated doses of enterotoxin-producing
B. cereus strains actually immunizes and offers some level of protection (vaccination) from subsequent infection with an infectious food-poisoning strain of
B. cereus. If this is the case, protection would be afforded only against
B. cereus-induced food poisoning, so such a treatment cannot provide universal protection against other enteric infections that lead to diarrhea. Although this is not a rational reason for using the strains, the generation of enterotoxin-specific IgG could provide a mechanism that protects against subsequent
B. cereus food poisoning. Interestingly, the Bactisubtil
B. cereus strain is listed as IP 5832 and is apparently identical to the strain used in the animal feed additive labeled Paciflor. Paciflor was recently withdrawn from production because of an assessment by the Scientific Committee on Animal Nutrition of the European Commission. In this assessment the presence of both the Hbl and Nhe enterotoxins was demonstrated, and it was concluded that this was a risk to human health, primarily because of the risk of infection of humans in the slaughtering process (assessment by the Scientific Committee on Animal Nutrition on the Safety of the Product Paciflor for use as a feed additive, adopted 16 May 2003;
http://europa.eu.int/comm/food/fs/sc/index_en.html ). It is somewhat surprising that Bactisubtil is still listed as a product for human use by Aventis Pharma Portugal. In our work, we detected only the Hbl enterotoxin in Bactisubtil, suggesting that despite the identical strain designation (IP 5832) the strains most probably represent derivatives of a common ancestor and the loss of the Nhe enterotoxin can be attributed to spontaneous or deliberate inactivation of the
nheA gene.
Germination of the spore could allow production of antimicrobial agents, such as bacteriocin-like inhibitory substances, thereby contributing to the competitive exclusion of pathogens, and it is one factor that could support the probiotic effect. A number of
Bacillus species produce antimicrobial agents, and more than 80 different types have been reported (
25). These antimicrobial agents are active mostly against gram-positive bacteria, but some are active against gram-negative bacteria. Recently, an antibiotic compound isolated from a strain of
B. subtilis found in the probiotic Biosporin with activity against
H. pylori has been reported (
31). We show here that at least two probiotic strains, Biosubtyl
NT and Subtyl, produce antimicrobial agents (or bacteriocin-like inhibitory substances) that are active against other
Bacillus species. We note that certain
Bacillus species have been associated with infection of the GIT, but it is also possible that these agents are active against a broader group of species. In any case, production of the antimicrobial agents could be an element in the probiotic effect.
Immune stimulation as a mechanism for a probiotic effect is difficult to define, but this must result from induction of proinflammatory cytokines that increase phagocytosis (by macrophages or dendritic cells) and perhaps also stimulation of cytotoxic cells. We show here that when given orally to mice, all probiotic strains generate systemic IgG responses. This shows that spores are immunogenic and are not treated as a food. To generate humoral responses, spore antigens could interact with the GALT. As reported elsewhere, there is strong evidence that
B. subtilis spores enter the Peyer's patches and MLN, and presumably they do this by translocation across M cells (
6). In the case of Biosubtyl
NT the spore-specific IgG responses were almost 10-fold higher than the responses to the other strains, showing that this strain is particularly immunogenic. Analysis of cytokine expression in vivo showed that there is early production of IFN-γ and TNF-α in the secondary lymphoid organs and GALT following oral inoculation of mice with PY79 and Biosubtyl
NT spores. Using a coculture with macrophages in vitro, we failed to detect significant levels of TNF-α or IL-1α (IFN-γ was not tested), but clear production of IL-6 was observed; we have no explanation for these results. Interestingly, in a recent study (
29) performed with human monocytes (isolated from peripheral blood mononuclear cells) stimulated with
B. subtilis spores, significant levels of IL-1β and TNF-α were found to be produced. IFN-γ is an activator of cellular responses, particularly the Th1 response that, in turn, is responsible for stimulating phagocytosis. IFN-γ is also produced during inflammation (as opposed to a specific immune response), as is TNF-α, whose production by macrophages has been linked with chronic infections (
4,
26). These early responses suggest that there is an innate immune response and secretion of IFN-γ by peripheral blood mononuclear cells. Examination of macrophages cultured in vitro with PY79 and Biosubtyl
NT also showed that there was potent induction of the proinflammatory cytokine IL-6. Proinflammatory responses should not necessarily be considered a beneficial feature of a probiotic since these responses show been linked to autoimmune diseases, such as inflammatory bowel diseases, including ulcerative colitis and Crohn's disease (
37). Together, these inflammatory responses show the complexity of immunomodulatory responses that can result from oral consumption of bacterial spores. While in this study we examined cytokine responses elicited from the GALT, it should be emphasized that nonprofessional antigen-presenting cells (e.g., epithelial cells) could also play an important role in immunomodulation. The importance of these responses in a potential probiotic effect remains to be determined, but the responses may well play a role in increasing resistance to infection by recruitment and activation of immune and inflammatory cells (neutrophils and mast cells). Similarly, oral administration of various probiotic
Lactobacillus species has been shown to enhance the innate immune system and to enhance macrophage phagocytosis (
38), NK cell functions (
3), and production of macrophage lysosomal enzymes (
19).