New genome-based technologies enable us to study human microbial populations in depth, as well as to correlate the presence of certain bacteria in the human gut with health status [1]. Infact, dysbiosis in the intestinal microbiota is related with the emergence of several metabolic disorders [2] and diseases, including irritable bowel syndrome [3] and inflammatory bowel disease [4]. Among gastrointestinal populations, is one of the most widely studied probiotic genera, and research indicates that these microorganisms or their metabolites are related with the formation or establishment of an indigenous and well-balanced intestinal microbiota in newborn children and adults [5,6]. Furthermore, the intake of specifi c probiotic strains has been shown to reduce the incidence of acute infections and allergic disorders and to ameliorate infl ammatory conditions, including some inflammatory bowel diseases [7]. Considering these eff ects, one potential target for probiotic intervention could be Celiac Disease (CD), which is characterized by alterations in the composition of the gut microbiota, including reductions in Bifidobacterium numbers, as well as an aberrant infl ammatory response [8,9].

Celiac disease is one of the most common chronic diseases in the world, with a prevalence of 0.7 to 2.0 % in the general population, and 15 to 20% in first-degree relatives. It is an immune-mediated enteropathy, triggered by the intake of wheat-gluten proteins and related prolamins of rye and barley in genetically susceptible individuals [10]. In this disorder, gliadins (α, β, γ and ω types) are known to be the main toxic components of gluten. Typical symptoms of CD include body-weight loss, diarrhea and chronic constipation [10], while another symptom is iron-defi ciency anemia, found in 15% of children with CD [11].

The presence of gluten is common in most processed foods; therefore it is diffi cult to follow the life-long gluten-free diet strictly, which is currently the only available therapy for CD patients. Nonetheless, optimum histological recovery is not always achieved following this diet [12]. Alternative and adjuvant methods have been studied with probiotic-based therapy being one of the most promising approaches. In this context, the strain Bifi dobacterium longum ES1 (CECT 7347) has a demonstrated ability to reduce the toxicity and infl ammatory potential of gliadin-derived peptides In vitro [13,14] and to attenuate the aberrant immune response in a gliadin-induced enteropathy animal model, reducing infl ammatory cytokine production and CD4+ T-cell numbers [15].

Probiotics are defined as “live microorganisms which when administered in adequate amounts confer a health benefit on the host” according to a group of experts working under the umbrella of Food and Agriculture Organization (FAO) and the World Health Organization (WHO) [16,17]. Further to this general definition, the aforementioned working group and the European Union (EU)-funded Product Safety Enforcement Forum of Europe (EU-PROSAFE) project [18] outlined guidance criteria required for a strain to be considered as a probiotic, including demonstration of both efficacy and safety. In both guidelines, the unambiguous identification of the strain at genus and species level by phenotypic and molecular techniques, together with its deposit in an internationally recognized culture collection are essential. In addition, the probiotic strain should also be phenotypically characterized. The recommendations specifically include: i) determination of antibiotic resistance pattern; ii) assessment of certain metabolic activities (e.g., D-lactate production, bile salt deconjugation); iii) assessment of side-effects during human studies; iv) epidemiological surveillance of adverse incidents in consumers (post-market); v) toxin production if the strain under evaluation belongs to a species that is a known mammalian toxin producer; vi) determination of hemolytic activity if the strain under evaluation belongs to a species with known hemolytic potential and vii) assessment of lack of infectivity by a probiotic strain in immunocompromised animals [17]. The approaches recommended in these guidelines are currently being used to evaluate the safety assessment of potential probiotic strains [18-20]. In this context, new massive sequencing technologies enable analysis of the whole genome sequence of probiotic strains, thereby excluding possible risks associated with the unexpected presence of potentially harmful genes due to their encoding capacity (i.e., genes directly associated with antibiotic resistance and virulence) or their location with transferable elements [21].

Hence, the aim of this study was to evaluate in silico, In vitro and in vivo the safety of strain B. longum CECT 7347 following FAO/WHO recommendations and including whole-genome 454 platform-based sequence analysis, phenotypic characterization of antibiotic resistances and metabolic traits and an acute toxicity assay in a murine model.

Bacterial strains and growth conditions

The strain B. longum CECT 7347 and the commercial strain B. longum BB536 (Morinaga & Co. Ltd. Tokyo, Japan) were grown on de Man, Rogosa and Sharpe medium (MRS; Oxoid, Basingstoke, United Kingdom) supplemented with 0.05% (wt/vol) cysteine (Sigma Aldrich; St. Louis, MO; MRS-C), and incubated anaerobically at 37°C for 17 h.

Ex vivo toxicological evaluation

Sensitivity to antibiotics: The method established by the Clinical and Standards Institute to test broth dilution antimicrobial susceptibility was followed to determine strain sensitivity to antibiotics, expressed in terms of Minimum Inhibitory Concentration (MIC), in accordance with EFSA recommendations [22]. The MIC values of 20 antimicrobial agents were determined using the standardized LSM broth formulation, which consists of a mixture of IST broth (90% vol/vol, Oxoid) and MRS-C broth (10% vol/vol, Oxoid) adjusted to pH 6.7 [23]. The antibiotics were tested in a concentration range of 0.125-256 mg/L.

Lactic acid production: Lactic acid production was determined by using 24-h supernatant of each bacterial strain. B. longum CECT 7347 was grown anaerobically at 37°C in MRS-C medium. Supernatant was obtained by centrifugation at 12,000×g for 10 min. Lactic acid isomers were quantified using a commercial kit (D-lactic acid/L-lactic acid; Megazyme, Wicklow, Ireland) following the manufacturer’s instructions.

Bile salt deconjugation: A bile salt hydrolase activity assay was performed with glycocholate and taurocholate as substrates, following Kumar and co-workers technique [24].

Biogenic amine formation: The formation of biogenic amines (cadaverine, histamine, putrescine and tyramine) was determined in MRS-C free-cell supernatant of 24-h cultures at 37°C anaerobically. Amine formation was determined following the chromatographic method described by Eerola and co-workers [25].

Acute toxicity assay in mice

Animals: Thirty two BALB/cJ male mice, aged 6 to 8 weeks and bred at the animal house of the Institut Pasteur de Montevideo (Montevideo, Uruguay) under SPF (specific pathogen free) conditions were used. Mice were housed in Individually Ventilated Cages (IVC) with an environmentally controlled temperature of 19-21°C, relative humidity of 70% and dark-light cycles of 10-14 h. Animals were fed with standard mouse diet ad libitum and had free access to water during the experimental protocol. Protocols were approved by the “Comisión de Ética en el Uso de Animales” (CEUA) of the Institut Pasteur.

Experimental protocol: Animals were randomly distributed in four groups, eight animals per group. Group A: Immunocompetent Control Mice (ICC); Group B: Immunocompetent Treated Mice (ICT); Group C: Immunosuppressed Control Mice (ISC) and Group D: Immunosuppressed Treated Mice (IST).

Before being administered, cells were freeze-dried with skim milk as a cryoprotectant. After two weeks of acclimatization, 400 μL of bacterial suspension was administered daily by oral gavage to animals of ICT and IST groups, each mouse receiving approximately 5.4×10⁹ CFU. Skim milk (400 μL) was administered orally to the control groups (ICC and ISC). The experimental procedure lasted six days. Immunosuppressed mice were treated with 40 mg/kg cyclophosphamide i.e. daily, and the treatment started five days before the experimental protocol and lasted throughout the study.

Individual body weight was registered at the beginning and at the end of the trial. At the end of the experimental treatment, blood from each mouse was collected by submandibular vein bleeding in EDTA-containing tubes under sterile conditions. Immediately afterwards, mice were sacrificed by cervical dislocation then liver, spleen and mesenteric lymph nodes were removed in sterile conditions. Sections of intestine were excised for further histological examination.

Bacterial translocation: All liver, spleen, Mesenteric Lymph Nodes (MLN) and blood samples were collected under strict aseptic conditions to avoid any cross-contamination. Liver, spleen and mesenteric lymph nodes were homogenized separately in PBS, cultured in MRS-C agar medium (Oxoid) and incubated at 37°C for 72 h in anaerobic conditions (Anaerogen, Oxoid), to verify the presence of bacterial growth. After 72h of incubation, colony forming units were counted and results were expressed as incidence of bacterial translocation.

Histological observation: Duodenum-jejunum, proximal colon, distal ileum and distal colon were immediately excised for histological studies. Tissues were fixed in 10% (vol/vol) buffered formalin and then embedded in paraffin. Sections were cut into 5 μm pieces, stained with Hematoxylin and Eosin (HE) and the morphological parameters were examined under direct microscopy.

Whole genome sequencing

Bifidobacterium longum CECT 7347 was grown until the start of the stationary phase (approximately 109 cells). The cell culture was centrifuged for 1 min at 10,000 rpm and the pellet was used for extracting the total DNA using the “High Pure PCR” kit (Roche Diagnostics GmbH, Penzberg, Germany). We measured DNA quality and quantity in order to ensure the minimum standards for library construction.

A total amount of 15 μg of high quality DNA were used to construct an 8Kb Paired End (PE) library following the standard protocol developed by 454 LifeScience/Roche. The library was run in a GS FLX Titanium Series 454 LifeScience/Roche sequencer using the Titanium chemistry.

Bioinformatic analyses

Raw sequences obtained from the whole genome sequencing were stored in sff format. Calling qualities and sequence extraction from SFF files were carried out using sff tools version 2.6 provided by 454 LifeScience/Roche.

High quality sequences were assembled with the Newbler assembler version 2.6 with the parameters by default. Scaffolds and contigs that were not introduced into the scaffolding were used for genome annotation.

tRNAs and rRNA were predicted using tRNAscan-SE [26] and RNAmmer v 1.2 [27], respectively. Open Reading Frames (ORFs) were predicted using Glimmer v 3.02 [28-30]. We assigned functionality to all detected ORFs with a local alignment approach, as implemented in the tBLASTx [31] comparing each ORF against a database containing all sequences from the bacteria kingdom.

Once gene functionality had been associated to each ORF, we searched for virulence and antibiotic resistance genes comparing each detected ORF with two different databases through a BLAST search, sequences found at Virulence Factors of Pathogenic Bacteria (VFPB) (http://www.mgc.ac.cn/VFs/main.htm) [32] and sequences found at Antibiotic Resistance Genes Database (ARGD) (http://ardb.cbcb.umd.edu/) [33] for virulence and antibiotic resistance, respectively.

Statistical analysis

Results obtained were analyzed using Statgraphics plus 5.1 software (Manugistiscs, Rockville, USA). For acute toxicity studies in the murine model, statistical analysis was carried out with Paleontological Statistics 1.91 software package for education and data analysis (Past), with a statistical significance of p<0.05. In all cases, data were subjected to a one-way analysis of variance (ANOVA) using the strain as the variable. The least significant difference test was used for comparison of means. Results were expressed as mean ± Standard Deviation (SD).

Antibiotic resistance and metabolite production

MIC values obtained for 20 key antimicrobial agents are summarized in Table 1. Regarding EFSA breakpoints [23], erythromycin was the only antibiotic with MIC values slightly higher than values recommended by the EFSA (1 μg/mL vs 0.5 μg/mL).

Table 1: Sensitivities to antibiotics of the isolated strain CECT 7347 compared with commercial strain B. longum BB536. Data are expressed as means (n=3).

Bifidobacterium longum CECT 7347 produced higher levels of L-lactic acid (5.02 g/L) than D-lactic acid (0.63 g/L) in the supernatant. Regarding bile salt hydrolase assays, disrupted cells showed an activity of 0.35 U.I./mL in the case of taurocholate and 0.48 U.I./mL for glycocholate, whereas there was no activity in supernatants from 24-h cultures in either taurocholate or glycocholate. Cadaverine, histamine, putrescine and tyramine biogenic amines were not detected in the supernatant of B. longum CECT 7347 cultures.

In vivo toxicological assays in a murine model

Neither mortality nor clinical signs were observed throughout the study. Regarding body weight gain (Table 2), there were no statistically significant differences between the placebo and probiotic groups. In the case of immunosupressed mice, although all mice had normal food and water intake during the experimental procedure, general deterioration in body condition was observed, due to pharmacological treatment, without impairing body weight. Moreover, a decrease in spleen and MLN size was observed in IST and ISC mice dissection (data not shown). Bacterial translocation to the different organs of treated and control mice are shown in Table 2. No bacteremia was observed in any of the four groups. For those organs exhibiting bacterial growth, intestinal content contamination was detected and recorded during dissection. Mesenteric lymph nodes exhibited the highest bacterial growth, which was to be expected since they are located near the intestine, and may suffer crossed contamination with the normal microbiota.

Table 2: Results obtained in mice during the acute ingestion study with strain B. longum CECT 7347. Data are the means ± standard deviations of the means (n=8).

Regarding the results obtained in the histological analysis, the comparative observation between IC and IS mice showed an increase in average villous height in the duodenum and jejunum (Table 3). In the case of the ileum, the highest values were observed in non-treated immunocompetent mice (Table 3). Villi/cript ratio of the small intestine was constant in duodenum, jejunum and ileum for all groups (data not shown). As shown in Table 3, no significant difference in enterocyte height was found between ICT and ICC mice, but a slight increase was observed between IST and ISC mice. In the ileum, a score of 2.0 was obtained in all cases. The number of caliciform cells was reduced in ISC and IST mice. No significant differences were detected in the depth of crypts among all four groups.

The presence of mucosa-associated lymph tissue was detected in 75% of ICT and 84% of ICC mice, and 53% of IST and 75% of ISC mice. Particularly, lymphoid follicles showed significant differences between ICT and ICC mice in the distal colon section and also in IST and ISC mice in the jejunum section (P<0.05), with fewer lymphoid follicles in control groups (Table 3). Figure 1 shows a representative portion of the distal colon from ICT, ICC, IST and ISC mice.

Table 3: Histopathological results obtained in acute ingestion study of strain B. longum CECT 7347 in mice. Data are the means ± standard deviations of the means (n=8).

Figure 1: Representative HE-stained sections of the distal colon seen in ICT (A), ICC (B), IST (C) and ISC (D) mice. Bar=100 μm.

Whole genome sequencing

After the sequencing run a total of 212,612 paired end sequences with a mean of 351.83 nts were obtained, giving a total of 74.8 Mb of sequence. The assembly using Newbler gave 148 contigs where 130 were larger than 500 nts, the N50 of the contig assembly was 39,694 nts and the largest contig was 148,677 nts. Most of these contigs were ordered in two scaffolds, the N50 of the scaffolding and the largest scaffold was 2,391,528. Genome sequencing data were registered in EMBL under the project accession number CALH00000000.

Those contigs that could not to be included in the scaffolding were also used for further analysis. This combination of scaffolds and contigs resulted in an estimated genome size of 3.5 Mb, which is close to other strains of the same species. A total of 2,078 elements were detected, where 2,020 were ORFs (1,624 canonical and 396 non-canonical) and 58 were RNAs (3 rRNA and 55 tRNA).

Even though the blast search resulted in a significant e-value, none of the ORFs had an identity higher than 50% with that described for antibiotic resistance or virulence genes. This indicates that only domains in the ORF are similar to some of the putative virulence and antibiotic resistance genes, but no significant or highly identical genes to virulence and antibiotic resistance were detected.

Only a DOC (death-on-curing) family protein with an identity of 87% and a significant e-value was detected when searching for virulence factors. This protein is already present in B. longum JDM301 and ATCC 55813 strains, as well as virB4 (truncated type IV secretory pathway VirB4 component protein), and ABC transporters for antibiotic resistance genes [34].

In recent years, celiac disease prevalence has been increasing steadily. However, due to its idiosyncrasy, there is no therapy currently available for this disorder except the adherence to a strict gluten-free diet. Research into the use of probiotics as an adjuvant strategy to the gluten-free diet is underway in order to improve the quality of life of CD patients. The strain B. longum CECT 7347 has been demonstrated to ameliorate damage caused by CD, in both In vitro and in vivo animal models [13-15]. Once the efficiency of the strain has been demonstrated, given its potential in vivo functionality, a complete safety assessment is also required. Consequently, although strain CECT 7347 has been unambiguously ascribed to the species B. longum, which is included in the list of taxonomic units with QPS status [35] and considered safe, further safety assessments are recommended to ensure its harmlessness [17,23].

Following FAO/WHO recommendations, resistance to a broad range of antibiotics was assessed. Results obtained were close to results previously reported for other Bifidobacteria [36,37]. Only in the case of erythromycin were MIC values slightly higher than values recommended by EFSA [23]. In this respect, erythromycin resistance of certain Bifidobacterium species, including commercial strain B. animalis Bb12, has been reported previously [18,19,38]. Furthermore, and taking into account the whole genome sequence, neither erythromycin nor antibiotic-related resistance genes were found in plasmids or between insertion sequence elements.

The evaluation of FAO/WHO specific metabolite production (lactic acid isomers, products of BSH activity and biogenic amines) showed that D-lactate concentrations were very low in comparison with L-lactic isomer, and similar to levels previously reported in other Bifidobacteria [19] and Lactobacillus rhamnosus GG [39]. Production of L-lactic acid isomer by B. longum CECT 7347 was very similar to that produced by the commercial strain B. longum BB536 from Morinaga [20]. Regarding BSH activity, B. longum CECT 7347 displayed low activity in disrupted cells and always lower than the Morinaga strain. Concerning biogenic amines, none of those analyzed was detected in B. longum CECT 7347 culture supernatants.

Acute ingestion of the Bifidobacterial strain under study did not cause either mortality or morbidity in any group, even in the immunosuppressed group. Acute administration of strain B. longum CECT 7347 did not lead to significant Bifidobacterial organ translocation. Moreover, no biologically significant changes in histomorphology were found as a result of strain administration.

Although one putative virulence factor and one ABC transporter were found, the complete pathway in which they participate was not present; therefore, the mere presence of these isolated genes lacks biological relevance. Taking into account that the aforementioned genes have also been described in other strains of B. longum, and are similar to some phage proteins, misannotation of these proteins may have occurred.

Taking all these results into account, strain B. longum CECT 7347 can be potentially considered safe for human consumption, fulfilling one of the criteria required for probiotics according to FAO/WHO [16,17].

According to FAO/WHO criteria [16,17], standard methods for clinical evaluations are underway, in order to confirm that strain CECT 7347 is safe for human consumption (clinical trials Phase 1, safety study) and to determine whether the strain exerts health benefits (clinical trials Phase 2, efficacy study).

This study was supported by Biópolis SL. within the framework of the e-CENIT Project SENIFOOD from the Spanish Ministry of Science and Innovation.