Detection and characterization of Bifidobacterium crudilactis and B. mongoliense able to grow during the manufacturing process of French raw milk cheeses

Background The study of a production chain of raw milk cheeses (St Marcellin, Vercors area, France) led to the isolation of two Bifidobacterium populations: B. crudilactis and B. mongoliense, that were able to grow along the production chain. The aims of this study were to further detect and characterize these bacteria along the process and evaluate the ability of some strains to survive or grow in adverse conditions. Results Using PCR coupled with restriction fragment length polymorphism, B. crudilactis and B. mongoliense were detected in respectively 77% and 30% of St Marcellin cheeses from production chain after 21 days of ripening. They were present in more than half of all analyzed retail cheeses with counts going from 1.6 to 5 log cfu g-1 for B. crudilactis and 1.4 to 7 log cfu g-1 for B. mongoliense. Bifidobacterium mongoliense was sensitive to pH 2, with an observed decrease of at least 3 log for both studied strains (FR49/f/2 and FR41/2) after 1 h incubation. At pH 3, no significant decrease was observed. Good survival was observed for the same strains in presence of pancreatic juice with a decrease of less than one log. Survival of strain FR49/f/2 was better than FR41/2 with a decrease of 3 logarithms (in presence of 1% bile salts) and almost 2 logarithms (in presence of 0.5% bile salts). The genotypic analyses using total DNA-DNA hybridization, GC% content, 16S rRNA gene sequencing and multilocus sequencing analysis (MLSA) confirmed the classification of Bifidobacterium. crudilactis and B. mongoliense into two different clusters well separated from other bifidobacteria clusters. Conclusions According to the observed characteristics such as survival in adverse conditions and their ability to grow under 12°C during the manufacturing process of the cheeses, which has never been described for bifidobacteria and which is a very interesting technological asset, these B. crudilactis and B. mongoliense strains should be further investigated for a potential use in new food or in food supplements.


Background
The contamination of raw milk cheese production chains, using bifidobacteria as fecal indicators, was previously studied in two different cheese processes in France [1]. It has been demonstrated in that study that the species Bifidobacterium pseudolongum was an efficient indicator of fecal contamination from animal origin, since this species is predominant in cow feces and has never been isolated in human feces. In addition, as an indicator, B. pseudolongum was more sensitive than E. coli along the two different cheeses processes.
On the other side, bifidobacteria are also bacteria that are often desired in food products for their technological properties as thickening and stabilizing agents or for their health promoting effects. For these reasons, bifidobacteria and lactic acid bacteria are commonly added to fermented food such as fermented milk or yoghurt but also in food supplements. The positive effects encountered with bifidobacteria can be mitigated by the difficulty of some strains or species to resist or adapt to an industrial process. They prefer to grow in the absence of oxygen, which can be a challenge in industry. In addition, some strains are not able to survive the adverse conditions found in the stomach and duodenum making it difficult to exert any positive effect on the host.
The previous study of a production chain of L'Etoile du Vercors (St Marcellin cheese produced from raw cow's milk, Vercors area, France) showed an increase of total bifidobacteria along the process and led to the isolation of two other Bifidobacterium populations, B. crudilactis and B. mongoliense [2,3]. These populations were able to grow along the cheese production chain. The description of the species B. crudilactis was based on one of those populations [4]. The other population could be phylogenetically identified as B. mongoliense [5] that was recently described based on two isolates obtained from a traditional fermented mare's milk in Mongolia.
The aims of the study were (i) to further detect the two bifidobacteria populations along the L'Etoile du Vercors cheese production chain using 16S rDNA-Restriction Fragment Length Polymorphism, (ii) to evaluate the potential of B. mongoliense as a new food supplement or as a potential new probiotic by the study of several phenotypic characteristics such as sugar fermentation profile, growth temperature profile and aptitude to survive in adverse condition such as acidic environment or environment rich in bile salts and (iii) to perform a genotypic classification using DNA-DNA hybridization and MLSA (Multilocus Sequencing Analysis) to identify if those new populations belong to specific clusters.

Validation of the PCR-RFLP method on pure isolates of bifidobacteria
Sixty-four strains belonging to 13 different species were already tested in a previous study using PCR-RFLP [6]. Eight additional strains of Bifidobacterium crudilactis and four French isolates identified as B. mongoliense were analyzed in this part of the work ( Table 1). The RFLP patterns observed (i) with AluI was named V (5-95-152-206-285-311 bp), (ii) with TaqI were called IX (120-210-250-470 bp) or X (132-200-664 bp). The patterns combination V-X was attributed to B. crudilactis while the V-IX combination was observed for B. mongoliense.
Detection of Bifidobacterium crudilactis and B. mongoliense in the L'Etoile du Vercors cheese process and in commercialized cheeses -L'Etoile du Vercors cheese process ( Table 2).
Out of the 176 samples from the L'Etoile du Vercors cheese process (44 samples at each step of the process: A-raw milk, B-after addition of rennet, C-after removal from the mold, D-ripening) analyzed by PCR-RFLP, 108 (61%) were V-X type positive (B. crudilactis) and 31 (18%) were V-IX type (B. mongoliense) positive. Their average counts were respectively 2.27 ± 0.48 and 0.48 ± 0.7 log cfu ml -1 .
Bifidobacterium crudilactis was detected in 82% (step A) and 77% (steps B and D) of the samples, but in only 9% of the samples at step C. The species B. mongoliense was detected in only 18% of the total samples (31/176 samples). As observed for B. crudilactis, a very low percentage of samples were positive at step C (2%), while the highest percentage was found at stage D (30%).
The mean counts of B. crudilactis and of B. mongoliense were studied at each step of the production chain ( Table 2). The B. crudilactis species presented highly significant variations (p < 0.0005). A marked decrease at step C (0.30 log cfu ml -1 ) was followed by a high increase at step D (3.45 log cfu ml -1 ). The mean counts of B. mongoliense were low (less than 1 log cfu ml -1 or g -1 ) at the different steps of the production chain (A, 0.61; B, 0.46; C, 0.00; D, 0.86).
-Commercialized raw milk cheeses (Table 3). Thirteen commercialized raw milk cheeses were analyzed: five trade cheeses from the plant in Vercors and eight other trade raw milk cheeses with a manufacturing process similar to the one in Vercors (Table 3).
Bifidobacterium crudilactis was detected in four St Marcellin cheeses from the Vercors' plant (Day 40), with counts equal to 4 log cfu g -1 in two samples and 5 log cfu g -1 in the two others, but not in the Picodon cheese from this plant. B. mongoliense was detected in four cheeses out of five from this plant. This species was found in the Picodon cheese (7 log cfu g -1 ) and in In the eight other commercialized cheeses manufactured with a similar process to the one in Vercors, B. crudilactis was isolated from only three cheeses (mean count of 1.6 log cfu g -1 ) while B. mongoliense was isolated from five cheeses (mean count of 1.4 log cfu g -1 ), including one cheese made with pasteurized milk (2 log cfu g -1 ).
Phenotypic and genotypic characterization of Bifidobacterium crudilactis and B. mongoliense strains isolated from the cheeses The species Bifidobacterium crudilactis was described by [4] on the basis of 10 strains out of 141 isolates from raw milk and two raw milk cheese production chains. Of the 141 isolates, one hundred and thirty-seven were obtained from 34 raw milk cheeses produced in the same French raw milk cheese factory in the Vercors area, from April 2003 to February 2004 [4]. Another isolate was taken from a raw milk sample in another French raw milk cheese factory in the Courtenay area. The three remaining isolates came from three commercialized French raw milk cheeses (Picodon, Brie and Reblochon).
The description of Bifidobacterium mongoliense [5] was based on the study of two strains isolated from a traditional mare's milk product from Mongolia. All strains analyzed in this study are presented in Table 1. Using thirteen B. mongoliense strains isolated from French raw milk cheeses a more complete description of the species was made. Six of these strains were isolated from six raw milk cheeses at different steps of Vercors' plant (FR41/2 in raw milk; FR39/1 and FR47/2 after addition of rennet; FR66/e/1 after removal from the mold; FR49/f/2 during ripening at day 15; FR70/g/2 at day 21). Three strains (MarV1/5, MarV4/2 and MarV3/ 22) were obtained from three commercialized cheeses from the Vercors' factory (day 40). One strain (FR101/h/8) came from a raw milk cheese sample during ripening at day 45 in another French raw milk cheese factory in the Courtenay area, and three others (MarC1/13, MarF/3  Production steps: A, raw milk; B, after addition of rennet; C, after removal from the mold; D, ripening (D 21); n = number of cheese sample.
and PicD/1) from three commercialized French raw milk cheeses (Picodon and Saint-Marcellin). All cheeses containing B. crudilactis and B. mongoliense strains were made from raw cow milk, except for the two Picodon cheeses, which were made from raw goat milk. The strain FR41/2 was retained as representative of the B. mongoliense French isolates since it was phenotypically the closest to all other French isolates.

Phenotypic characterization
The major differential characteristics between the species Bifidobacterium crudilactis, B. mongoliense, and the type strain of B. psychraerophilum are presented in Table 4.
growth observed at pH 4.4 and no growth at pH 3.8 within 15 days). Similar growth characteristics were observed by [5] except that no growth of B. mongoliense YIT 10443 T occurred at 10°C, nor at 40°C.
Resistance to gastric, pancreatic juices and bile salts of Bifidobacterium mongoliense strains FR 49/F/2 and FR41/2 Gastric juice At pH 2, a decrease of at least 3 log was observed for both studied strains after 1 h of incubation. At pH 3, a good survival of both strains was observed, even after 5 h of incubation, with no significant decrease in bifidobacteria counts (Table 5).

Pancreatic juice
Both strains had a good survival in the presence of pancreatic juice (pH 8, Table 5). A slight decrease was observed for both of them. The strain FR49/f/2 presented a decrease of respectively 0.44, 0.46, 0.46 and 0.50 log after 2 h, 3 h, 4 h and 5 h. The strain FR41/2 presented a decrease of 0.41 log after 5 h.

Bile salts
In presence of a concentration of 1% bile salts, a decrease of 3.39 and 4.11 log was respectively observed for the strains FR49/f/2 and FR41/2 after 48 h. The strain FR49/f/ 2 was less sensitive to lower bile salts concentration with a decrease of 3.07 log and 1.76 log in presence of bile salts at a concentration of respectively 0.5% and 0.3%. The sensitivity of the strain FR41/2 to lower bile salts concentration was the same than in presence of 1% with a decrease of 3.93 log and 4.14 log with a concentration of bile salts of respectively 0.5% and 0.3% (Table 5).
The DNA G + C content of the representative strain FR41/2 was 60.5% (average of two experiments, SD = 0.71). The average DNA G + C content of six strains in the group was 61.1% (SD = 0.67), which is also the G + C content of the B. mongoliense type strain (YIT 10443 T ). Those values were higher than the G + C contents of B. psychraerophilum LMG 21775 T (55.7%) and B. crudilactis LMG 23609 T (56.4%), using the T m method. However, similar DNA G + C contents are found with the same method in the genus Bifidobacterium, for the species B. cuniculi, B. gallinarum, B. pseudolongum subsp. pseudolongum, B. pseudolongum subsp. globosum and B. scardovii [8].
DNA-DNA hybridizations were performed with DNAs of FR41/2, other French B. mongoliense isolates and the type strains of B. mongoliense, B. minimum and B. subtile, the closest phylogenetically related species ( Table 6). The DNA-DNA relatedness (%) between B. mongoliense DSM  (Table 6).

Strains clustering using a multilocus approach
Analysis of MLSA (Multilocus Sequencing Analysis) results of B. mongoliense FR41/2 and FR49/f/2 and B. crudilactis FR62/b/3 T , based on concatenated sequences of the housekeeping genes clpC, fusA, gyrB, ileS, purF, rplB and rpoB, revealed a cluster for these strains, that was separated from other Bifidobacterium strains from the MLST and NCBI database (Figure 1). An allelic profile was determined for the 3 strains (Table 7), according to the method of [9] and was compared to the allelic profile of previously studied strains [9]. The same sequence type was observed for the strains FR41/2 and FR49/f/2. There were different by only one nucleic acid for the allele purF and rplB.

Discussion
It appears from our results that the species Bifidobacterium crudilactis and B. mongoliense are commonly present in St-Marcellin and other cheeses having a similar manufacturing process such as Picodon, Camembert, Brie and Reblochon. B. crudilactis and B. mongoliense were detected in respectively 77% and 30% of cheeses from L'Etoile du Vercors plant after 21 days of ripening. In addition, they were present in more than half of analyzed retail cheeses. Their presence and their source are not explained yet but they probably contribute to organoleptic and technologic properties of those cheeses. During the L'Etoile du Vercors cheese process, a very low percentage of samples were positive at step C (after removal from the mold, day 2) with respectively 9% and 2% of positive samples for Bifidobacterium crudilactis and B. mongoliense. An explanation for the low percentage could be the presence of competitive flora at this stage associated with a low pH (4.35 at step C). However, during ripening (step D), the conditions seemed more favorable for those species with respectively 77% and 30% of positive samples for B. crudilactis and B. mongoliense.
The ripening temperature, 12°C, used in the Vercors factory enables these species to grow along the production chain. B. crudilactis was isolated at higher levels (mean count 3.45 log cfu g -1 at step D) than B. mongoliense in the Vercors' factory. These levels were still higher in the four commercialized St Marcellin cheeses (D40, mean count 4.5 log cfu g -1 ).
Bifidobacterium mongoliense was detected in only 30% of the samples during ripening of the Vercors' production chain (mean count 0.86 log cfu g -1 at step D) but was present at much higher levels in three of the commercialized St Marcellin cheeses (5 to 6 log cfu g -1 ) and in the Picodon cheese made with goat raw milk (7 log cfu g -1 ) from the Vercors area. These results suggest that both species, when present along the chain, continue to multiply in the cheeses after production, during storage. B.
mongoliense was detected in one cheese made from pasteurized milk. The most probable explanation is a cross contamination along the process or commercialization leading to the presence of this bacterial species. Interestingly, the other species, B. crudilactis was detected in thermized milk in our study (St Marcellin Cholet) and in thermized milk from Canada [10]. This suggests that B. crudilactis has the potential to resist to some heat treatment. Those properties should be better investigated, as they are important when these bacteria are incorporated in food products where heating steps are present.
The resistance to gastric and pancreatic juices, and bile salts of these bifidobacteria encourages further study for their potential to survive the transit through the human intestinal tract. The stomach normally produces around 2.5 l of HCl per day at a pH of 2.0 and a stomach pH can drop as low as 1.5 [11]. It is therefore important, when selecting probiotic strains candidates, to choose strains able to survive this barrier. Usually, acid resistance of probiotic strains is estimated for different pH values ranging from 1.5 to 5.0. Incubation time can vary from 1 to 360 minutes at 37°C, according to the fact that food usually remains in the stomach for about 90 minutes. In our study, a poor survival of strains FR49/f/2 and FR41/2 was observed in gastric juice at pH = 2, after 1 h of incubation. However, the survival of the strains was excellent at pH 3. Those results are quite encouraging knowing the limitation of the used method, namely HCl acidified media do not take into account the influence of diet or non-acidic gastric constituents on the microorganisms during their gastric transit making that test usually too severe. For example, the strain Lactobacillus GG cannot survive in static models, but is able to reach colon in high number in vivo [12,13]. The strains assessed in our study should be further tested in a dynamic in vitro model such as the SHIME or M-SHIME system [14,15] for confirmation of their survival through the transit. Pancreatic secretions are amongst the most aggressive fluids in the human body. In our case, using a commercial pancreatic solution at pH 8, a good resistance of both strains was observed even after 5 h incubation.
Physiological concentrations of bile salts are between 0.3 [16] and 0.5% [17]. This is why bile solutions containing bile salts at concentrations of 0.3, 0.5 and 1% were tested in our study. Usually, bacteria are exposed to bile salts from 3 to 96 h. Bile salts secretions in vivo are highly variable and difficult to predict. In our study, the strain FR 49/F/2 showed a better resistance than the FR 41/2 with 0.3% bile salts. In the other conditions, a decrease of approximately 3 logarithms was observed. It is important to note that bile salts solutions and animal bile extracts are more aggressive for bacteria than human bile [16]. According to the fact that human bile is not easily available, testing of bile salts is a good compromise even if the conditions tested are probably more aggressive than in reality. The best compromise would have been to use bovine bile extract, which is less toxic than bile salts, but still more toxic than human bile.
The results obtained by MLSA are in accordance with the other genomic data presented in this study and allow a temporarily classification of B. crudilactis and B. mongoliense in two different clusters, well separated from the other bifidobacteria clusters. If, in the future, additional genome sequences are available on public database such as NCBI or on Bifidobacterium MLST (Multilocus sequence typing) database (Paris Institut Pasteur; [9]), more closely related Bifidobacterium species such as B. bohemicum, B. minimum, and B. psychraerophilum could be included in this analysis as well. This method was already successfully used for species identification in several genera such as Vibrio or Bifidobacterium [18][19][20][21].
However, one of the next steps will be to continue the genetic analysis of the full genome sequences to better understand the evolution of those strains and be able to explain their specific phylogenetic characteristics.
Origin of the species B. crudilactis and B. mongoliense is not known. In a previous paper [1], B. pseudolongum was chosen as an efficient fecal indicator since its mean counts remained stable along both studied processes. In addition, B. pseudolongum defined a contamination from animal origin since this species is predominant in cow dung and has never been isolated in human feces. In the case of B. crudilactis and B. mongoliense, the fecal origin is not demonstrated.

Conclusion
The species B. crudilactis and B. mongoliense were detected in raw milk cheeses from the Vercors' area. The origin of these bacteria is not known yet, but their presence in the cheeses probably contributes to specific organoleptic and technologic characteristics of those cheeses. This study allowed to report that these bacteria are able to grow under 12°C during the manufacturing process of the cheeses, which has never been described for bifidobacteria, and which is a very interesting technological asset. In addition, the bacteria were able to resist to low pH and aggressive fluids such as pancreatic juice and biliar salts. Because of the special characteristics of these bifidobacteria, further investigations should be performed to better

Samples from St Marcellin process (Vercors plant)
The manufacturing process was already described previously. Briefly, milk was collected on farms and stored in tanks at the plant at 4°C as already described [22]. Next, the milk was prepared for maturation by addition of cream, starter and surface flora. Temperature was increased to 22°C. Animal rennet was added (Day 0). On the next day (Day 1), the following steps were successively performed: molding, a first manual turnover, a manual salting and a second turnover. During that day, pH decreased from 6.5 to 4.3 while temperature remained stable (22°C). On the second day, cheeses were removed from the molds and a new manual or mechanical salting was performed. Ripening was then carried out for 28 days. Temperature was 12°C from Day 8. During that stage, pH slowly increased from 4.35 (at the beginning of ripening), to 4.7 (Day 15), to 5.5 (Day 21), to more than 6 (Day 28). Forty-four raw milk cheeses at 4 different steps (176 samples) were analyzed at the following production steps: raw milk (Step A, Day 0), after addition of rennet (Step B, Day 0), after removal from the mold (Step C, Day 2) and during ripening (Step D, Day 21).
The enrichment medium was carried on as described previously [22] in supplemented Brain Heart Infusion (BHI, 37 g l -1 , Bio-Rad, Marnes-la-Coquette, France). One ml of milk or 1 g of raw milk cheese was transferred into a tube of 9 ml enrichment medium, and 1 ml of each of the ten fold appropriate sample dilutions in quarter-strength Ringer solution containing cystein chlorhydrate (0.3 g l -1 ) was also inoculated in tubes of enrichment medium in order to detect bifidobacteria in milk and raw milk cheese until the 10 -6 dilution to get semi-quantitative counts as described previously [1]. Estimated mean counts of bifidobacteria (log cfu g -1 ) were obtained after calculating the averages of the last positives dilutions for which a value of 1 cfu ml -1 were attributed. Tubes were incubated at 37°C for 72 h in aerobiosis, as the bacteria were able to grow in depth because of the presence of agar in the medium.
-DNA extraction and PCR-RFLP protocol. DNA for sequencing was extracted from culture broths obtained after the enrichment step as described previously [1] using the Wizard Genomic DNA purification kit (Promega, Madisson, WI, USA) with addition of lysozyme (10 mg/ml, Eurogentec, Seraing, Belgium).
The PCR method for the detection of the genus Bifido bacterium consisted of primers targeting the 16S rRNA gene followed by a digestion using 2 restriction enzymes AluI and TaqI (Roche; Basel, Switzerland) for species detection as described previously [6]. Twenty microliters of the PCR product was restricted by 1U of enzyme in 1X buffer at 37°C for 3 h with AluI and at 65°C for 3 h with TaqI in a total volume of 30 μl. Following the digestion, the products were analyzed by gel electrophoresis using 2.5% agarose gel. The profiles were analyzed using the Kodak 1D software (Thermolabsystems, Brussels, Belgium).

Phenotypic characteristics
The F6PPK test was assayed on isolates as described by [23], after cell disruption by sonication (Sonopuls homogenizer HD-70, Bandelin, Berlin, Germany) for 30 sec. Bifidobacterium crudilactis and French B. mongoliense strains were discriminated in a phenotypic numerical study using the unweighted pair group method with averages (UPGMA, [24]). Phenotypic tests were 49 carbohydrate fermentations, 30 enzymatic tests performed on API 50CH and Rapid ID32A kits (Bio-Mérieux-department API, La Balme les Grottes, France) according to the manufacturer's instructions. The ability of the strains to grow at 46°C was performed in Tryptone Phytone Yeast (TPY) broth within 48 h [4]. The levels of similarity between strains were calculated using UPGMA (Sneath et al. 1973), and the Jaccard index. The numerical analysis included type strains of Bifidobacterium species and Aeriscardovia aeriphila as described previously [24].
The range of temperature for the growth of Bifidobac terium crudilactis and French B. mongoliense strains was tested in TPY broth (in water bath) at 41°C, 41.5°C, 43°C, 44°C, 45°C, 46°C and 47°C for 8 days and at 4°C, 5°C, 6°C, 7°C, 8°C, 9°C, and 10°C for 14 days. All temperatures were controlled using a temperature standard probe certified by the French Committee for Accreditation (COFRAC). Growth of FR62/b/3 T (B. crudilactis) and of FR41/2 (B. mongoliense) were also tested on TPY agar at 39°C under aerobic and anaerobic conditions and at 39°C in TPY broth at different pH values (3.8, 4.4 and 4.7). All strains tested in this study are presented in Table 1.
Resistance to gastric, pancreatic juices and bile salts Tolerance to acidic conditions and gastric juice: To measure the resistance to gastric juice during digestion, the method of [11] was followed. Briefly, the strains FR49/ f/2 and FR41/2 were grown anaerobically in MRS for 36 h at 37°C. The pellets were collected by centrifugation at 4000 rpm and suspended in gastric juice solution (solution of 0.3% pepsin w/v (Sigma-Aldrich, Bornem, Belgium) and 0.5% NaCl w/v) adjusted to pH2 or pH 3. The solutions were incubated in anaerobic conditions during 5 h at 37°C. Plate counts were performed every hour using an automated spiral plater (Don Whitley Scientific LTD., Shipley, West Yorkshire, England). The results were compared to a blank composed of the corresponding strain in suspension in K 2 HPO 4 buffer solution (pH 6.5) instead of gastric juice solution.
Tolerance to pancreatic juice: The resistance to pancreatic juice was measured as described by [11]. Briefly, the strains FR49/f/2 and FR41/ 2 were grown anaerobically in MRS for 36 h at 37°C. The pellets were collected by centrifugation at 4000 rpm and suspended in pancreatic juice solution (solution of 1 g pancreatin (Sigma-Aldrich, Bornem, Belgium) in 1 l of solution containing 0.5% NaCl w/v solution) adjusted to pH 8 using 5 mM NaOH [11,25]. Every hour during 5 hours, the survival of bifidobacteria was estimated using plate counts on MRS agar medium using an automated spiral plater (Don Whitley Scientific LTD., Shipley, West Yorkshire, England). The results were compared to a blank composed of the corresponding strain in suspension in K 2 HPO 4 buffer solution (pH 6.5) instead of pancreatic juice solution.
Tolerance to bile salts: The resistance to bile salts was measured according to the method of [26]. Briefly, the strains FR49/f/2 and FR41/2 were inoculated in MRS supplemented with 0%, 0.3%, 0.5% and 1% of bile salts (LP0055, Oxoid). At T0 and after an incubation period of 48 h, plate counts were performed to evaluate the survival of both strains under the different bile salts concentrations. The results were compared to the strains incubated in the media containing 0% of bile salts.
The sequences were aligned with 16S rRNA gene sequences of cultured bacteria. Sequences producing significant alignments as determined by BLAST analysis were Bifidobacterium mongoliense, B. psychraerophilum, B. crudilactis and B. minimum (comparison based on an alignment of 1452, 1428 pb, 1384 pb and 1416 pb respectively).

Determination of DNA G + C content
The DNA G + C content were determined by the thermal melting point method (T m method, [29]) using a Cary 100 spectrophotometer connected to a Peltier temperature controller (Varian, France).

DNA-DNA hybridizations
DNA-DNA hybridizations between DNA of FR41/2 and the type strain of B. mongoliense DSM 21395 T were performed at BCCM/LMG. DNA for these hybridizations was extracted using the method described by [30], modified to a larger scale. The hybridizations were performed in the presence of 50% formamide at 47.5°C according to a modification [31,32] of the microplate method described by [33]. Reciprocal reactions (A x B and B x A) were performed, and the variation was within the limits described for this method. The DNA-DNA relatedness percentage reported for this pair of DNA is the mean of minimum 7 hybridizations.
DNA-DNA hybridizations between DNA of FR41/2 and the type strains of B. mongoliense, B. subtile and other French B. mongoliense isolates were performed with DNA extracted as described by [34], using achromopeptidase (5,000 U/500 mg bacteria, Sigma, France) and lysozyme (400,000 U/500 mg bacteria, MP Biomedicals, France) as lytic enzymes. This DNA was fragmented at approximately 100 MPa using an Aminco French pressure cell (Bioritech, France), and the degree of DNA-DNA relatedness was quantified from renaturation rates according to the method of [35], using a Cary 100 spectrophotometer connected to a Peltier temperature controller (Varian, France). The renaturation temperature was 25°C below the midpoint (T m ) i.e. 67.3°C, according to the G + C content of FR62/b/3 T . DNA-DNA relatedness values were calculated after 21 and 24 min of incubation. The first 3 min of renaturation, during which the temperature stabilised, were excluded from calculations. Reassociation percentages were 4 V m -(V a + V b ) / 2 (V a × V b ) ½ , where V a , V b and V m were the decreases in absorbance at 260 nm per min for organisms A and B and for an equal mixture of the two.
In order to obtain sufficient quantity of DNA for complete sequencing, the DNA extraction protocol of [36] was used. The library has been sequenced on a Roche FLX (1/8 plate) using Titanium chemistry (Roche Diagnostics, Vilvoorde, Belgium). For each gene, the alleles found in the genomes were compared to the alleles sequences obtained from the Bifidobacterium MLST (multilocus sequence typing) database (Paris Institut Pasteur) available for the probiotic Bifidobacterium species: B. bifidum, B. breve, B. longum subsp. animalis and B. longum subsp. longum [9] at http://www.pasteur.fr/ recherche/genopole/PF8/mlst/Bifidobacterium.html. This public database allows queries and downloads of allele sequences and allelic STs, provides several database tools and statistics and is also open to additional data on novel strains and species. The genomes of other Bifidobacterium species were also available on NCBI (B. angulatum DSM20098 (NZ_GG663536), B. catenulatum DSM 16992 (NZ_ABXY01000031), B. dentium ATCC 27678 (ABIX02000002) and B. gallicum DSM 20093 (NZ_ABXB03000001)). Alleles from each gene were selected using clustalW (http://www.ebi.ac.uk/) and concatenated. Next, the concatenated DNA sequences were aligned and analysed using the bioinformatics software "Geneious" (http://www.geneious.com/). Phylogenetic analysis was performed based on the concatenated sequences of the seven genes on Geneious (Figure 1).