Characterization of spore forming Bacilli isolated from the human gastrointestinal tract

Abstract

Aims:  To isolate and characterize spore-former bacteria able to colonize the human gastrointestinal tract (GIT).

Methods and Results:  A total of 25 spore-formers was isolated from faeces and ileal biopsies of healthy human volunteers and identified at the species level. Physiological analysis was performed to evaluate the ability of the various isolates to form biofilms, to swarm, to produce surfactants and molecules that have antimicrobial activity against selected pathogens. To assess the potential probiotic activity of the isolates, we tested the resistance of cells and spores to simulated gastric conditions, the ability to grow and sporulate in anaerobic conditions and the presence of toxin-encoding genes in their genome.

Conclusions:  Spore-formers belonging to various bacterial species have been isolated from the gut of healthy human volunteers. These strains appear to be well adapted to the intestinal environment and we propose them as potential probiotic strains for human use and as oral vaccine vehicles.

Significance and Impact of the Study:  To our knowledge this is the first detailed characterization of spore-forming Bacilli from the human GIT. Our data suggest that the isolated species do not transit, but rather colonize this specific habitat and propose them as probiotic strains for human use.

Introduction

Endospore-forming bacteria are Gram positive organisms belonging to various genera that, all together, include more than 200 species (Fritze 2004). These organisms are generally divided into two main groups of aerobic and anaerobic bacteria, with each group further subdivided into three genera: Bacillus, Sporosarcina, Sporolactobacillus and Clostridium, Desulfotomaculum, Sporomusa, respectively (Fritze 2004). However, few exceptions have been found and members of the aerobic Bacillus genus have been described as Gram-negatives (B. azotoformans, B. oleronius and B. horti) (Fritze 2004) or as capable of anaerobic metabolism (B. subtilis) (Nakano and Zuber 1998; Tam et al. 2006).

The common feature of spore-forming Bacilli is the ability to differentiate a peculiar cell form, the endospore (spore). Formation of the spore initiates when vegetative growth can no longer occur because of food shortage or other nonphysiological conditions in the environment. The spore is a quiescent cell form, characterized by several protective layers surrounding the dehydrated cytoplasm that contains the nucleoid (Henriques and Moran 2007). This structural organization makes the spores extremely resistant to external physical and chemical insults and able to survive almost indefinitely in the absence of water and nutrients. The exceptional longevity of the spore in the environment is the main reason for the ubiquitous distribution of these organisms, in particular, of the aerobic ones (Fritze 2004).

It is generally accepted that the primary reservoir of spore-forming Bacilli is the soil and the ability of spores to be dispersed in dust and water has been identified as the cause of their presence in almost every conceivable habitat. Several species of spore-formers are commonly found also in the gastrointestinal tract (GIT) of a variety of animals (Barbosa et al. 2005; Tam et al. 2006). Only few Bacillus species are pathogens of animals (B. cereus and B. anthracis) or insects (B. thuringiensis), while the majority of them are nonpathogenic. Their presence in the GIT has been considered as due to the ingestion of bacteria associated with soil, water, air or foods.

However, a new theory is now emerging in which spore-former species are thought to establish an endosymbiotic relationship with their host, being able to survive and proliferate within the GIT and specifically interact with immuno and intestinal cells (Hong et al. 2005). Recent work has shown that in a murine model ingested spores can safely cross the stomach barrier and germinate in the intestine (Casula and Cutting 2002). In the same experimental model it has been also shown that spores can perform a complete life cycle, with germination in the upper part of the intestine, vegetative growth and sporulation before being expelled in the faeces (Tam et al. 2006). Other studies have established that B. subtilis, in combination with Bacteroides fragilis, is able to induce the development of gut-associated lymphoid tissue (GALT) and preimmune antibody repertoire in rabbits (Rhee et al. 2004). This study also showed that sporulation, as opposed to vegetative cell growth, is essential for GALT development. An in vitro analysis has also shown that the Competence and Sporulation Factor (CSF) of B. subtilis, a five amino acid peptide secreted during exponential growth and acting as a quorum-sensing molecule for the induction of DNA uptake and sporulation, is able to induce heat-shock response in human enterocyte-like (Caco-2) cells (Fujiya et al. 2007).

In a rather empirical way, spores of several Bacillus species have been widely used as human and animal probiotics for decades. Some commercial products have proven to contain Bacillus species different from those declared on their label (Green et al. 1999; Hoa et al. 2000), some strains are of unknown origin, some are multidrug resistant and some even harbor toxin genes (Green et al. 1999). Moreover, little is known about how spores exert their beneficial action on humans and animals. An in vivo study with a murine infection model has shown that the oral administration of 1 × 10⁹ spores of B. subtilis one day before infection with 1·5 × 10³ CFU of the murine enteropathogen Citrobacter rodentium was able to drastically reduce the mortality rate and some signs of enteropathy but without affecting the animal immune-response to the pathogen (D’Arienzo et al. 2006).

All of the studies mentioned above have been performed with domesticated strains of B. subtilis. There is evidence that laboratory strains of B. subtilis differ from undomesticated strains, in several aspects including factors that are likely to affect their efficacy as probiotics (Branda et al. 2001; Earl et al. 2007, 2008). For these reasons, in this study, we aimed to isolate and identify aerobic spore-formers from the human GIT. Strains were characterized and tested for properties that would be beneficial to their survival in the gut and that could be desirable for probiosis. The collection of wild Bacilli of human origin described here will most likely provide a useful source of potential probiotics for human use, since it has been suggested that probiotic strains originate from the target animal microflora (Barbosa et al. 2005).

Materials and methods

Collection of ileal and faecal samples

Ileal biopsy samples were collected from eight adult human volunteers (M/F 5/3, mean age ± SD 45·0 ± 13) undergoing routine diagnostic colonoscopies. All patients recruited gave their informed consent to the study. The study was approved by the appropriate ethics committee and has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. The patients did not follow any special dietary regimen, and had not recently received any antibiotic or probiotic treatment. Samples were stored at −80°C in phosphate-buffered saline (PBS) containing 15% glycerol before subsequent analysis. Endoscopic appearance as well as histology of the ileum was normal in all patients. Faecal samples were collected from five healthy adult human volunteers that did not follow any special dietary regimen and who had not received any antibiotic treatment for at least 3 months.

Bacterial isolation and characterization

Ileal and faecal samples (20–40 and 50 mg/each, respectively) were heat-treated (80°C for 10 min) to kill all vegetative cells and individually placed on LB plates. After 36 h of incubation at 37°C, colonies were recovered and purified by streaking on fresh LB plates. Pure cultures were streaked on Difco sporulation medium (DSM), incubated at 37°C for 24–37 h and checked by light microscopy for the presence of spores.

Exponentially growing cells were used to extract chromosomal DNA as previously reported (Green et al. 1999). DNA coding for 16S RNA was PCR amplified by using chromosomal DNA as a template and oligonucleotides Ribo-For (5′-AGTTTGATCCTGGCTCAG-3′; annealing at position +9/+28) and Ribo-Rev (5′-CCTACGTATTACCGCGGC-3′ annealing at position +549/+531). Those two oligonucleotides were designed to amplify a 540 bp DNA fragment (MicroSeq 500 16S ribosomal DNA) previously indicated as sufficient for species identification (Woo et al. 2003). Amplified DNA was used to determine the nucleotide sequence (BMR Genomics, Padova, Italy) which was used for an online computer-assisted analysis of homology.

Exponentially growing cells of the various isolates were used for biochemical analysis by the use of API 50 CHL kit (Biomerieux) following the manufacturer’s instructions.

Unless otherwise specified, bacteria were grown in LB medium (for 1 l: 10 g Bacto-Tryptone, 5 g Bacto-yeast extract, 10 g NaCl, pH 7·0). Anaerobic conditions were obtained by incubating liquid and solid cultures in an anaerobic chamber (Oxoid).

Physiological analysis

Swarming motility was tested as previously reported (Connelly et al. 2004). Overnight cultures of all strains were spotted on LB or B (Julkowska et al. 2004) medium plates. LB and B plates were incubated at 37°C and 30°C, respectively, for 24–36 h. Surfactin production was assessed as described by Youssef et al. (2004) and by growing cells on B medium (Julkowska et al. 2004). To test biofilm production overnight cultures were used to inoculate liquid MSgg medium (100 mmol l⁻¹ MOPS pH 7·0, 0·5% glycerol, 0·5% glutamate, 5 mm potassium phosphate pH 7·0, 50 μg ml⁻¹ tryptophan, 50 mg ml⁻¹ phenylalanine, 2 mmol l⁻¹ MgCl₂, 0·7 mmol l⁻¹ CaCl₂, 50 μmol l⁻¹ FeCl₃, 50 μmol l⁻¹ MnCl₂, 2 μmol l⁻¹ thiamine, 1 μmol l⁻¹ ZnCl₂) (Branda et al. 2001) and cells grown at 37°C in static conditions for up to 48 h. Cells forming a solid layer at the liquid–air interface were considered as biofilm producers.

Resistance to GIT conditions was assessed as previously reported (Duc et al. 2004). Cells or spores were suspended in simulated gastric fluid [SGF: 1 mg of pepsin (porcine stomach mucosa; Sigma) per ml; pH 2·0] or small intestine fluid [SIF: 1 mg of pancreatin (porcine pancreas; Sigma) per ml and 0·2% bile salts (50% sodium cholate–50% sodium deoxycholate; Sigma); pH 7·4] and incubated at 37°C for 1 h. Samples were serially diluted and plated to determine the number of CFU per ml on LB agar plates. Resistance to antibiotics was assessed on plates by adding to LB plates the following antibiotics: neomycin (20 μg ml⁻¹), erythromycin (3 μg ml⁻¹), spectinomycin (200 μg ml⁻¹) or rifampicin (50 μg ml⁻¹). Production of antimicrobials was tested as previously reported (Baccigalupi et al. 2005).

Analysis of enterotoxins and virulence traits

Methods to detect putative B. cereus enterotoxin genes from Bacillus species by PCR amplification from chromosomal DNA have been reported previously (Duc et al. 2004). Primer sets were those described by Guinebretiere et al. (2002). Haemolysis was detected by streaking cells on horse blood (Oxoid) agar plates and 48-h incubation at 37°C.

Results

Isolation of spore-formers from human gut

Samples of faeces and ileal biopsies of healthy human volunteers, collected as described in ‘Materials and methods’, were heat-treated to kill all cells and incubated on a solid medium to allow germination and growth of heat-resistant spores. All recovered bacteria were purified and analysed for colony morphology and the presence of spores by light microscopy. From a total of eight ileal biopsies (20–40 mg/each from different individuals) and of five samples of faeces (50 mg/each from different individuals) 13 and 12 spore-formers were isolated, respectively.

Together with the spore-formers, other bacteria were also isolated, but only partially characterized and not used in the present study. Those organisms were either members of thermophilic species or mesophilic but probably part of abundant population not totally killed by the heat-treatment.

The 25 spore-formers isolated were characterized at the species level by analysis of the 16S rDNA sequence and biochemically by the use of API 50 CHL kit (Table 1). As shown in Table 1, mostly similar species of aerobic spore-formers were isolated from the two sources, whereas, due to the isolation procedure, we did not recover clones of anaerobic spore-formers. In addition, a B. thuringiensis clone was isolated from a faecal sample and B. megaterium and Paenibacillus chibensis isolates derived from ileal biopsies. For strain SF170, also of ileal origin, a species was not assigned since its 16S DNA sequence showed homology with an uncultured Bacillus (GenBank entry: AY493970). Because of the low number of samples and isolates, our results can not be taken as an indication that some species proliferate preferentially in one source or the other, but they do suggest that a very similar population of Bacilli can be found in both.

Table 1.   List of intestinal strains isolated

Strain	Species^*	Source	Accession number†
^*Species assignment was based on 16S rDNA sequence analysis and on the results of the API50 CHL kit. †Accession numbers of 16S rDNA sequences deposited to the EMBL nucleotide sequence database.
SF119	Bacillus pumilus	Feces	FM178952
SF120	Bacillus licheniformis	Feces	FM178953
SF147	Bacillus pumilus	Feces	FM178954
SF148	Bacillus subtilis	Feces	FM178955
SF149	Bacillus subtilis	Feces	FM178956
SF150	Bacillus clausii	Feces	FM178957
SF151	Bacillus subtilis	Feces	FM178958
SF152	Bacillus subtilis	Feces	FM178959
SF153	Bacillus subtilis	Feces	FM178960
SF154	Bacillus subtilis	Feces	FM178961
SF155	Bacillus subtilis	Feces	FM178962
SF168	Bacillus thuringiensis	Feces	FM178963
SF85	Bacillus pumilus	Ileum	FM178964
SF106	Bacillus subtilis	Ileum	FM178965
SFB2	Bacillus subtilis	Ileum	FM178966
SFB3	Bacillus subtilis	Ileum	FM178967
SF128	Bacillus subtilis	Ileum	FM178968
SF169	Bacillus licheniformis	Ileum	FM178969
SF170	Bacillus sp.	Ileum	FM178970
SF173	Bacillus megaterium	Ileum	FM178971
SF174	Bacillus clausii	Ileum	FM178972
SF185	Bacillus subtilis	Ileum	FM178973
SF186	Paenibacillus chibensis	Ileum	FM178974
SF188	Bacillus pumilus	Ileum	FM178975
SF195	Bacillus subtilis	Ileum	FM178976

Swarming motility and biofilm formation

As an initial characterization, all 25 isolates were tested for their ability to swarm and produce surfactin and biofilm (Table 2). Swarming is a typical movement of bacterial cells on a solid surface (Fig. 1) and, together with biofilm formation, is a property of Bacilli that is often lost (or much reduced) in laboratory strains. In confirmation of this, none of the strains in our lab collection was able to swarm or produce biofilm in control experiments (data not shown).

Table 2.   Physiological properties of the intestinal isolates

Strain	Swarming		Surfactin ring	Biofilm formation
	LB	B
SF119	X	X		X
SF120	X	X	X	X
SF147	X	X	X	X
SF148	X	X	X	X
SF149	X	X	X	X
SF150				X
SF151
SF152				X
SF153
SF154
SF155	X			X
SF168
SF85	X	X	X	X
SF106
SFB2
SFB3	X	X	X	X
SF128	X	X	X	X
SF169	X	X	X	X
SF170
SF173
SF174
SF185				X
SF186				X
SF188	X	X	X	X
SF195				X

Figure 1.  Examples of swarming motility on LB medium additioned of 0·7% agar. PY79 is a laboratory collection strain of Bacillus subtilis (Youngman et al. 1984) and does not show swarming motility. The other strains are three of the GIT isolates described in Table 1 and all show various types of swarming motility.

Studies in B. subtilis have shown that swarming is dependent on the presence of a flagellum in various physiological conditions and, only in a minimal medium, also on the ability of the strain to produce surfactin (Julkowska et al. 2005). For this reason we tested our strains for the ability to swarm in rich (LB) and synthetic (B; Julkowska et al. 2004) medium and also assayed their ability to produce surfactin.

The analysis of Table 2 indicates that all isolates belonging to the B. subtilis species behaved as previously reported for that species: those that were able to swarm in minimal medium also produced surfactin. This behaviour was not observed with SF119 (B. pumilus) and SF155 (B. subtilis) as they were able to swarm in minimal medium but did not produce surfactin. A possible explanation is that those two isolates produce a different surfactant that does not respond to the assay we used to detect surfactin (Materials and methods).

The majority of the strains (16 out of 25) formed biofilms. This is an interesting observation since biofilms have protective and adhesive properties and have been associated to a longer persistence of Bacilli in the GIT of animals (Huang et al. 2008).

The analysis of Table 2 indicates that all strains able of swarming mobility are also able to produce biofilms, whereas some of the biofilm-producers do not swarm, either in rich or minimal media. These results induced us to speculate that biofilm formation is essential but not sufficient for swarming motility. Additional studies will be needed to properly address this point.

Production of antimicrobial activity

All isolated Bacilli were then analysed for the production of antimicrobial molecules active against selected pathogens (Table 3). An exponential culture of each of the 25 isolates was used to ‘spot’ sterile LB plates. As previously reported (Baccigalupi et al. 2005), the spots were air-dried and used to overlay soft agar (0·7%) containing exponential cells of one of the indicator strains. Solidified plates were then incubated at the appropriate temperature for 18–24 h and the appearance of a growth–inhibition halo taken as an indication of the presence of an antimicrobial activity. As summarized in Table 3, most of the strains produced antimicrobial molecules, mainly active against the Gram-positive pathogens used in our study. In particular, most of the strains were active against Listeria monocytogenes. Only four strains, all isolated from fecal samples, were active against Salmonella enterica (SF148, SF149 and SF168) or Shigella sonnei (SF147).

Table 3.   Antimicrobial activities produced by the intestinal isolates

Strain	Bacillus cereus	Staphy-lococcus aureus	Listeria monocytogenes	Salmonella enterica	Shigella sonneii
SF119			X
SF120	X		X
SF147		X	X		X
SF148	X		X	X
SF149	X		X	X
SF150
SF151		X	X
SF152		X	X
SF153		X	X
SF154	X	X	X
SF155			X
SF168			X	X
SF85	X	X	X
SF106		X
SFB2		X	X
SFB3	X		X
SF128	X		X
SF169
SF170
SF173
SF174
SF185
SF186
SF188			X
SF195

Although the low number of isolates do not allow us to draw statistically significant conclusions, it is interesting to note that, while almost all fecal isolates (11 out of 12) produced antimicrobial molecules, less than 50% (6 out of 13) of the strains isolated from the ileal biopsies showed that property.

Resistance of spores and cells to simulated GIT conditions

We measured the survival of spore suspensions in simulated gastric fluid (SGF) and simulated intestinal fluid (SIF) as previously reported (Duc et al. 2004). Spores were prepared by the exhaustion method originally developed for B. subtilis (Nicholson and Setlow 1990). As previously reported for fecal-isolates of B. subtilis (Tam et al. 2006), we also noticed that some strains were faster to sporulate than other isolates, including a laboratory strain [PY79, a derivative of the 168 type strain (Youngman et al. 1984)].

Spores were then purified as previously described (Nicholson and Setlow 1990) and aliquots of 3–5 × 10⁸ spores suspended for 1 h in SGF and SIF. Almost identical numbers of cells were recovered on LB plates from treated and untreated spores of all 25 isolates and of the strain PY79, indicating an almost total resistance of spores to the condition used (data not shown).

We also measured the survival of vegetative cells to SGF and SIF. Exponentially growing cells (always between 1·0 × 10⁷ and 1·0 × 10⁸ CFU) were exposed for 1 h to either PBS or SGF or SIF, washed and plated on LB plates. While cells of most isolates and of the control strain PY79 were totally killed by both treatments (data not shown), cells of strains SF119 (B. pumilus) and SF128 (B. subtilis), showed only a minor reduction in CFUs (Table 4).

Table 4.   Survival to intestinal conditions

Strain	Initial CFU	CFU after 1 h in PBS	CFU after 1 h in SGF	CFU after 1 h in SIF
SIF, simulated gastric fluid; SIF, simulated intestinal fluid; n.d., not detectable.
PY79	9·5 × 10^7	8·8 × 10^7	n.d.	n.d.
SF119	2·0 × 10^7	1·8 × 10^7	1·7 × 10^6	2·8 × 10^6
SF128	5·3 × 10^7	5·4 × 10^7	8·0 ×10^6	6·0 × 10^6

We reasoned that such resistance could be due to a peculiar cell surface of the two strains, both biofilm-producers (Table 2), and that secreted molecules could make the surface proteins unaccessible for the lytic enzymes present in SGF and SIF. Although we did not analyse the two strains for the presence of an S-layer, it is well known that several Bacillus isolates produce such protective structure that surrounds the cell and reduces cell surface accessibility (Candela et al. 2005; Huber et al. 2005). We tested those two strains, as well as all other isolates, for sensitivity to some antibiotics. SF119 and SF128 were both sensitive to all antibiotics tested (not shown) suggesting that their resistance to SGF and SIF is not due to an impermeable surface. We do not have an explanation for the high level of resistance of cells of those two strains to simulated GIT conditions and can only speculate that it may reflect an adaptation of the cells to the environment they lived in.

The antibiotic inhibition test indicated that strains SFB3 and SF168 were resistant to erythromycin (3 μg ml⁻¹), strain SF195 was resistant to spectinomycin (200 μg ml⁻¹) and strain SFB2 was resistant to neomycin (20 μg ml⁻¹) and spectinomycin (200 μg ml⁻¹). All other strains were sensitive to the antibiotic tested (Material and methods).

Growth and sporulation in anaerobic conditions

Members of the Bacillus genus are aerobic bacteria, unable to grow anaerobically. However, there are exceptions and lab collections strains (derivatives of the 168 type strain of B. subtilis) are able to grow anaerobically when nitrate is provided as electron acceptor (Nakano and Zuber 1998). Tam et al. (2006) have reported that two wild isolates of B. subtilis can grow and sporulate on a solid sporulation-inducing medium (DSM) in anaerobic conditions, whereas the domesticated strain PY79 grew but sporulated at very low efficiency. In their experimental conditions, addition of nitrate to the medium did not improve significantly either growth or sporulation of the tested strains (Tam et al. 2006).

We analysed all 25 human isolates for their ability to grow in anaerobic conditions in a rich (LB) and a sporulation-inducing medium (DSM). As reported in Table 5, most of the isolates were able to grow anaerobically in DSM whereas only three of them grew anaerobically in rich LB medium. Of the 19 isolates that grew anaerobically on DSM plates, 13 were also able to sporulate in the absence of oxygen. This observation confirms the previous observation reported on B. subtilis isolates (Tam et al. 2006) and expands it to the other species present in our collection.

Table 5.   Growth and sporulation of the intestinal isolates in anaerobic conditions

Strain	Vegetative growth on LB	Vegetative growth on DSM	Sporulation
SF119
SF120	X	X	X
SF147
SF148		X	X
SF149		X	X
SF150		X	X
SF151		X
SF152		X
SF153
SF154		X
SF155
SF168		X	X
SF85
SF106		X	X
SFB2		X	X
SFB3		X	X
SF128	X	X	X
SF169	X	X	X
SF170
SF173		X	X
SF174		X
SF185		X	X
SF186		X
SF188		X	X
SF195		X

The analysis of Table 5 also reveals that three out of four B. pumilus isolated from the human GIT were strictly aerobic (SF119, SF147 and SF85) and that the B. subtilis isolates were mostly anaerobic since of the 13 isolated strains two were strictly aerobic (SF153 and SF155), four were able to grow but not to sporulate (SF151, SF152, SF154 and SF195) and seven were able to grow and sporulate (SF148, SF149, SF106, SFB2, SFB3, SF128 and SF185).

Presence of potential virulence factors

We used a PCR approach to evaluate the presence of known B. cereus enterotoxin genes in the chromosome of all isolates using, as a control, the B. cereus strain GN105 (Naclerio et al. 1993). This method has been applied previously to profile putative food-poisoning Bacillus strains (Duc et al. 2004; Guinebretiere et al. 2002; Phelps and McKillip 2002). Figure 2 reports the results obtained for the PCR amplification of two isolates (SF150 and SF188) and the control strain GN105. Those reported in Fig. 2 were the only two strains positive for the presence of known enterotoxins. All other isolates were negative in our PCR-based analysis allowing us to conclude that 23 out of 25 tested strains did not contain genes encoding known Bacillus toxins in their genome.

Figure 2.  Agarose gel electrophoresis of PCR amplification products. All selected oligonucleotide pairs amplified specific fragments of the expected size using chromosomal DNA of the Bacillus cereus strain GN105 (Naclerio et al. 1993) as a template. Only the oligonucleotide pair used to amplify part of the bceT gene amplified a fragment of the expected size using chromosomal DNA of strains SF150 or Sf188 as a template.

In vivo analysis showed that only strains SF128 produced α-haemolysis while four strains produced β-haemolysis (SF119, SF147, SF168 and SF188). All other strains did not produce haemolysis and, therefore, can be considered as γ-haemolytic.

Discussion

We used fecal samples and ileal biopsies of healthy human volunteers to retrieve 13 and 12 spore-forming isolates, respectively. Those bacteria were first characterized at the species level and then analysed for various physiological properties, some of which may be relevant for future use in probiotic preparations containing defined strains.

Some interesting conclusions can be drawn from this work. In agreement with Bacilli isolated from the soil (Branda et al. 2001; Earl et al. 2007), those described here, isolated from a seemingly peculiar environment such as the human gut, displayed swarming motility and biofilm formation. It is not yet known whether those Bacilli are able to swarm or form biofilm within the gut, but the observation that these properties have not been lost (as it has, instead, occurred in laboratory strains) allows us to hypothesize that they are important in the environment where these bacteria inhabit.

Biofilms have been proposed to have protective and adhesive roles for the bacteria producing them. These functions are potentially relevant for bacteria in the gut, and because of the protective environment of the biofilm, could enable survival in the intestinal conditions and adhere to mucus and epithelial cells more effectively than planktonic cells. In addition, within biofilms bacteria can respond to quorum-sensing molecules more easily than planktonic cells. Sporulation is known to be induced by quorum-sensing signals (e.g. the CSF of B. subtilis) and this has been shown to be essential for GALT development (Rhee et al. 2004).

Also the ability to grow and sporulate in an anaerobic environment appears as a common property of Bacilli isolated from the anaerobic gut. Rather than aerobes, Bacilli should be considered as facultative anaerobes, able to use oxygen or a different electron acceptor depending on the environmental conditions. The observation that some isolates were able to grow but not to sporulate in the anaerobic conditions obtained in the laboratory may suggest that different electron acceptors and/or different metabolic pathways are used during growth and sporulation.

While it is not surprising that all isolates survived the simulated GIT conditions in the spore form, it is striking that two isolates survived also in the vegetative cell form. The observation that the same two isolates were sensitive to common antibiotics allowed us to exclude that resistance to pepsin and pancreatin, present in SGF and SIF, respectively, was due to an unusual cell surface, impermeable to many external molecules. The mechanism of that resistance remains not known and further experiments will be needed to address this point.

Some features of the 25 human isolates, such as growth and sporulation in anaerobic conditions, cell survival to simulated GIT condition and biofilm formation, allow us to hypothesize that those strains are well adapted to the gut environment and potentially able to colonize that habitat.

Most of the isolates were sensitive to common antibiotics and did not contain genes encoding for known Bacillus toxins. Only two isolates, SF150 and SF188, contained a gene homologous to the bceT gene of B. cereus. However, bceT codes for enterotoxin T, a factor that has been shown not to contribute to food poisoning (Choma and Granum 2002). Although a proper safety assessment, with cytotoxicity and in vivo tests, is needed before these strains can be considered as probiotics, the preliminary data presented here are an encouraging starting point to identify Bacilli of human origin to be used as probiotics for human use.

An additional potential application of those strains is as oral vaccine vehicles. Spores of B. subtilis have been used to display heterologous antigens (Isticato et al. 2001; Mauriello et al. 2004). Recombinant spores, orally administered to mice, were able to induce a specific humoral (Duc et al. 2003) and cellular (Mauriello et al. 2007) response. The immune response induced by spores exposing a fragment of the tetanus toxin resulted protective in a challenge experiment, with immunized mice able to survive the injection of a lethal dose of the toxin (Duc et al. 2003). It has been proposed that part of the observed immune response is not due to antigens present on the spores orally administered to the animals but rather to the antigens produced inside the animal body when recombinant spores germinate and sporulate (Uyen et al. 2007). All those studies have been performed with a laboratory strain of B. subtilis, not producing biofilms and incapable of efficient sporulation in anaerobic conditions. It is then reasonable to hypothesize that natural Bacilli expressing heterologous antigens may perform better than lab strains. It is, then, clear that a more efficient biofilm-assisted adhesion of cells to the intestinal epithelium and a more efficient sporulation in anaerobic conditions of the gut would result in a more efficient expression of the antigens and, presumably, in a stronger immune response.

Acknowledgements

We thank Krzysztof Hinc for advise on swarming motility and surfactin production and Luciano Di Iorio for technical assistance. This work was supported by grants of the Italian Ministry of the University (MIUR-COFIN 2006) and of the European Union (seventh Framework no. 207948: Colorspore) to E.R.