The differentiation of Bifidobacterium species was performed with specific primers using the PCR technique, the amplified ribosomal DNA restriction analysis (ARDRA) technique based on reports on the sequence of the 16S rRNA gene and speciation based on a short region of the ldh gene. Four specific primer sets were developed for each of the Bifidobacterium species, B. animalis, B. infantis and B. longum. The use of the ARDRA method made it possible to discriminate between B. infantis, B. longum and B. animalis with the combination of BamHI, TaqI and Sau3AI restriction enzymes. The ldh gene sequences of 309–312 bp were determined for 19 Bifidobacterium strains. Alignment of these short regions of the ldh gene confirmed that it is possible to distinguish between B. longum and B. infantis but not between B. lactis and B. animalis.
The characterization of bifidobacteria is important in the food industry as manufacturing of some products requires a particular species. Accurate taxonomic identification of bifidobacteria can be difficult when conventional methods are used. Some information exists on the DNA analysis of bifidobacteria such as DNA–DNA hybridization, pulsed field gel electrophoresis (PFGE) and the randomly amplified polymorphic DNA profiles (RAPD) [1–4]. The 16S rRNA gene was also used for the systematic identification of bifidobacteria . Some papers demonstrated that distinguishing between Bifidobacterium longum and Bifidobacterium infantis and also between Bifidobacterium animalis and Bifidobacterium lactis is still confused using phenotypic characterization [1,2]. This differentiation is more difficult because the DNA encoding the 16S of the former human bifidobacteria strains is similar at 90–98% to B. animalis according to the 16S rRNA gene sequence data. In addition, DNA–DNA hybridization showed low similarity between B. lactis and the type strain of B. animalis. These results suggest that the taxonomic status of some species of bifidobacteria (B. longum/B. infantis and B. animalis/B. lactis) should be clarified.
The development of new approaches and techniques of molecular biology would make it possible to clarify in which species strains of bifidobacteria belong. Amplified ribosomal DNA restriction analysis (ARDRA) has an excellent potential for discrimination of organisms at the species level [6,7]. This technique is based on the amplification of the DNA sequence of a 16S rDNA region, followed by the digestion of PCR products with restriction enzymes [6,7]. A few authors used conserved sequences other than the 16S rRNA gene for the characterization of species such as the conserved gene encoding for the L-lactate dehydrogenase (ldh) studied in B. longum or the recA gene in human bifidobacteria [8–10]. Like the 16S rRNA, the recA gene is considered to be universally present in bacteria. A phylogenetic analysis of the short sequence recA products was used to accurately classify strains of bifidobacteria . Fructose-1,6-bisphosphate-dependent L-lactate dehydrogenase (LDH) is a key enzyme in lactic acid fermentation by most lactic acid bacteria. An evolutionary study based on LDH sequences separates the prokaryotic from eukaryotic enzymes except for the B. longum LDH which falls in a group with the eukaryotic enzymes and the conserved ldh gene, suggesting a common ancestor for bacterial and vertebrate genes .
The objective of this study was to distinguish between B. infantis and B. longum as well as B. animalis and B. lactis using a combination of molecular biological techniques based on reports of the sequence of the 16S rRNA and the ldh genes. Specific primers for identification of bifidobacteria by PCR and the ARDRA technique based on the 16S rRNA gene followed by the analysis of a short sequence region of the ldh gene were used for the speciation of bifidobacteria.
2Materials and methods
2.1Bacterial strains and cultivation
Strains of bifidobacteria were isolated from commercial preparations (dairy products and freeze-dried cultures) by the Food Research and Development Centre, Agriculture and Agri-Food Canada (FRDC, Table 1). Other bifidobacteria were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and the Deutsche Sammlung von Mikroorganismen (DSM, Germany). Strains (SRW-001, SRW-002, SRW-003, SRW-004) were isolated from commercial preparations by FRDC and used as unknown strains to control the alignment of short regions of the ldh gene as a method for differentiation and identification of bifidobacteria. Stock cultures were prepared in brain heart infusion supplemented with 10% glycerol and frozen at −40°C. Lactobacillus MRS broth (Difco Laboratories) supplemented with 0.05%L-cysteine–HCl was used to cultivate the frozen microorganisms and recovered strains were subcultured once. Active cultures were incubated for 18–24 h at 37°C in an anaerobic chamber (Anaerobic system, Forma Scientific) with a gas atmosphere of 5% CO2, 10% H2 and 85% N2.
Table 1. List of bifidobacteria and non-bifidobacterium strains and the results of PCR tests using specific primer sets
Genomic DNA was prepared according to Vincent et al.  from stationary-phase cultures in MRS broth containing L-cysteine–HCl (0.05%). The concentration of purified DNA was determined using mini-fluorometer TKO 100 (Hoefer Scientific) and capillary tubes.
2.3Identification by PCR with specific primers
The specific primers were chosen from highly conserved nucleotide sequences in the 16S rDNA region (Table 2) and reaction conditions varied for the different primers. The reaction volume used was 50 μl which contained 10 mM dNTP, 1.5 mM MgCl2 (Pharmacia LKB Biotechnology buffer), 5 U Taq DNA polymerase (Pharmacia LKB Biotechnology), 20 pmol μl−1 of each primer and 25 ng of DNA. PCR amplifications were performed with a Peltier Thermal Cycle (PTC-200, MJ Research) using the following cycle parameters: denaturation at 92°C for 30 s, primer annealing for 30 s and primer extension at 72°C for 1 min followed by a final extension at 72°C for 10 min. The program included a preincubation at 92°C for 5 min prior to the first cycle and the final extension step was followed by cooling at 4°C. The temperature of annealing was 50°C and the number of cycles was 35 for the primer set Pbi F1/Pbi R2. The temperatures of annealing for the primer sets Ban F2/Pbi R1, Bil F3/Inf R5 and Pbi F1/Lon R4 were 58, 55 and 56°C and the number of cycles was 30. The PCR products were run on 1% agarose gel (w/v) (Boehringer Mannheim Canada) in 1×TAE for 1 h at 250 V and made visible by ethidium bromide staining and UV transillumination.
Table 2. Bifidobacterium species and specific primer sets based on 16S rDNA sequences
aThe numbers correspond to numbers in the structure model of E. coli 16S rRNA (GenBank accession number J01859).
Sequence (5′ to 3′)
Product size (bp)
2.4Amplification and ARDRA
The optimization of each combination of primers was determined by PCR with a 50 μl total volume. Primers were Pbi F1 and Pbi R2 (Table 2). After cycling, bifidobacteria were differentiated by a restriction fragment length polymorphism analysis using the enzymes BamHI, Sau3AI and TaqI (Boehringer Mannheim Canada). Restriction digestion was carried out for 2 h at 37°C for BamHI and Sau3AI, and at 65°C for TaqI in 20 μl of incubation buffer (Boehringer Mannheim Canada) containing 5 U of restriction enzyme and 10 μl of PCR product. Reaction products (10 μl) were fractionated on 2.5% agarose gel in 0.5×TBE buffer for 1.5 h at 200 V. Gels were stained with ethidium bromide, made visible by UV transillumination and digitalized with the gel print 2000i system (Bio/Can Scientific Inc.). The images were analyzed with the software GelCompar (Molecular Analyst Software Fingerprinting Plus, Bio-Rad Laboratories). The background was subtracted by the rolling disk method and the normalized patterns as obtained with the respective enzyme were combined to generate a single pattern for each strain. The patterns were used to construct a dendrogram using the UPGMA (unweighted pair group method using arithmetic averages) clustering algorithm .
2.5Sequence analysis of the ldh gene from bifidobacteria
Primers used for the detection of ldh gene were based on the regions of the gene encoding LDH (EC 188.8.131.52) of B. longum which was cloned in Escherichia coli and sequenced by Minowa et al.  (between positions 914 and 1283; GenBank accession number M33585). The primers were synthesized by General Synthesis and Diagnostics (Canada). The optimization of the combination of primers was determined by PCR with a 50 μl total volume. The PCR reactions contained 10 mM dNTP, 1.5 mM MgCl2, 50 mM KCl, 10 mM Tris–HCl pH 8.3, 5 U Taq DNA polymerase (Pharmacia LKB Biotechnology), 20 pmol μl−1 of each primer and 25 ng of DNA. Primer sequences were as follows: forward primers (LDH F1 5′-TACATGCTCATCACCAACCCGGTCGAC-3′) and the reverse primer (LDH R1 5′-CATGCCGATGGCGTAGTTGGTGGCACCCTT-3′). Amplification of DNA was performed in a GeneAmp PCR system 2400 (Perkin Elmer) programmed for 9 min at 94°C for initial denaturation and 35 cycles of 30 s at 94°C for denaturation, 30 s at 62°C for annealing, 30 s at 72°C for extension, followed by 10 min at 72°C for a final extension. 10 μl of each PCR reaction mixture was run on 2% agarose gel (w/v) in 1×TAE for 1 h at 250 V and made visible by ethidium bromide staining and UV transillumination. The length of the PCR product was 370 bp, purified by QIAquick Gel Extraction Kit (Qiagen Inc.) and sequenced by Automatic DNA sequencer 373 XL Stretch option XL (Applied Biosystems, Perkin-Elmer). Multiple sequences were aligned using the CLUSTAL W program .
3.1Identification of Bifidobacterium spp. using genus-specific and species-specific primers
The four primer sets selected within the 16S rRNA gene (Table 2) were tested with genomic DNA of the pure cultures of 86 strains of bifidobacteria belonging to B. adolescentis, B. animalis, B. bifidum, B. breve, B. infantis, B. lactis and B. longum and four non-bifidobacterium species (Table 1). Specific amplification of Bifidobacterium DNA was achieved with primer Pbi F1 combined with primer Pbi R2 for all of the strains tested. The primer sets Bil F3/Inf R5, Pbi F1/Lon R4 and Ban F2/Pbi R1 were able to identify strains of B. infantis, B. longum and B. animalis, respectively. However, the species-specific primer set for B. animalis showed a positive PCR result for DNA of B. lactis. Finally, 50 commercial strains were identified as B. animalis (16) and B. longum (34) using the Ban F2/Pbi R1 and Pbi F1/Lon R4 primer sets, respectively.
3.2Identification by ARDRA
For 69 strains of bifidobacteria belonging to B. adolescentis, B. animalis, B. bifidum, B. breve, B. infantis, B. lactis and B. longum, the 16S rDNA was amplified using the genus-specific primer set (Pbi F1/Pbi R2) and restricted with the enzymes BamHI, TaqI and Sau3AI (Fig. 1). Cleavage by the restriction endonucleases revealed a variety of different DNA fragment patterns for six species of bifidobacteria. The clustering from these patterns consistently grouped restriction fragment patterns into two distinct subsets. The ARDRA patterns of known members of B. animalis/B. lactis, B. longum and B. infantis were typical for each group and distinct from those of B. breve, B. adolescentis and B. bifidum. The restriction pattern from the B. lactis strains was identical to that of the B. animalis strains. The ARDRA method confirmed the fact that B. longum was closely related to B. infantis but differentiation between these two species was obtained with the Sau3AI restriction enzyme. The B. infantis strains exhibited a lower band than the B. longum strains. The BamHI and Sau3AI restriction enzymes showed that B. longum RW-009, RW-019, RW-023, RW-024, RW-025 and RW-045 were very different from the other B. longum strains. This result is in agreement with the PFGE results published by Roy et al. . These six strains have the same bands as the B. animalis when we used the BamHI enzyme but they have a longer band than the other B. longum strains when analyzed with the Sau3AI enzyme. Hence, the correlation between ARDRA and the species-specific primer sets was very good.
3.3Speciation based on the ldh gene sequence
The ldh gene sequences of 370 bp were determined for 19 Bifidobacterium strains including four isolates of commercial origin. Comparison of the short region of the ldh gene was based on the sequences comprised between the two primers LDH F1 and LDH R1 which gave a sequence of 309–312 bp (Fig. 2). A sequence identity of 100% was found between the short region of the ldh gene of B. longum sequenced by Minowa et al.  and B. longum ATCC 15707T determined in the present study. Only one nucleotide was different for the ldh sequence of B. longum ATCC 15708 as compared with the other strains of B. longum. One isolate of commercial origin (SRW-003) exhibited an identical nucleotide sequence to B. longum ATCC 15707T, indicating that this strain belonged to B. longum.
Bifidobacterial strains of B. animalis, B. bifidum, B. breve, B. adolescentis and B. infantis possessed specific sequences of the ldh gene as compared to B. longum. In a comparison of the sequences of all of the strains tested, we found that B. lactis DSM 10140 (GenBank accession number AF261675) and the sequence of B. animalis ATCC 27536 (AF261674) were the same. The number of base differences between B. lactis and B. animalis ATCC 25527T (AF261673) was six and the number of base differences between B. lactis and B. longum was 55. The commercial isolate SRW-004 (AF261687) of B. animalis had an identical nucleotide sequence to B. lactis DSM 10140 and B. animalis ATCC 27536. Alignment of the ldh sequence of SRW-001 (AF261684) and SRW-002 (AF261685) showed very high identity with the reference strains of B. breve and B. bifidum, respectively.
The ldh sequences of strains of B. infantis were divergent from those of B. longum. In addition, strains of B. infantis fell into two groups. B. infantis ATCC 15702 (AF261680) and RW-27920 (AF261682) exhibited small differences with B. longum because the number of base differences was 15–18 whereas B. infantis ATCC 15697T (AF261679) and ATCC 25962 (AF261681) were very different. The PCR products of these two strains possessed specific sequences of the ldh gene highly divergent from those of B. longum and all of the other strains tested.
Fig. 2 shows that the proportion of the nucleotide variation in the third position for B. animalis ATCC 25527T as compared to that of B. longum ATCC 15707T was 42 versus seven and nine for positions 1 and 2, respectively. In comparison, the proportion of nucleotide variation in the third position for B. bifidum ATCC 29521T was 17 whereas it was four and six in positions 1 and 2, respectively. However, B. animalis ATCC 25527T and B. bifidum ATCC 29521T exhibited 34 and 14 silent mutations, respectively. The differences in the number of amino acids between B. longum ATCC 15707T and B. breve ATCC 15700T, B. adolescentis ATCC 15703T, B. animalis ATCC 25527T and B. bifidum ATCC 29521T were five, eight, 14 and nine, respectively. B. infantis ATCC 15702 and RW-27920 exhibited a low number of mutations in the ldh sequence whereas B. infantis ATCC 15697T and ATCC 25962 possessed 151 nucleotides that were different from those of B. longum ATCC 15707T. These differences resulted in 71 amino acids which were different from those of B. longum ATCC 15707T.
In the present study, we demonstrated that species-specific primers are useful for the differentiation of strains belonging to B. infantis, B. longum and B. animalis. It is well-known that B. infantis and B. longum are difficult to distinguish using phenotypic characterization and DNA–DNA hybridization  although their β-galactosidase electrophoretic patterns and genomic fingerprints are different [3,14]. Phylogenetic data obtained from both 16S rRNA sequence analysis and 16S–23S internal transcribed spacer sequence analysis demonstrated that B. longum ATCC 15707T and B. infantis ATCC 15697T are closely related .
Vincent et al.  reported that the RAPD profiles indicate that the 20 human strains isolated by Bahaka et al.  fell into the B. longum cluster which also included the reference strains of B. infantis. On the other hand, sequence information from a short fragment of the recA gene provided greater sensitivity than sequence information from the 16S rRNA gene for the differentiation of B. infantis. In addition, Matsuki et al.  noted that their newly developed primers based on the 16S rRNA gene distinguished B. longum and B. infantis, even though these taxa are closely related species. In the present study, the ARDRA method based on the 16S rRNA gene confirmed the fact that B. longum was closely related to B. infantis but differentiation between these two species was obtained with the Sau3AI restriction enzyme. In addition, the analysis of a short region of the ldh gene showed that it was possible to easily distinguish between B. infantis and B. longum.
The results of the ARDRA method indicate that the subcluster containing reference and commercial strains of B. animalis also included the new species B. lactis. It was found that commercially available industrial strains of B. animalis isolated from fermented milks were identical to the reference strain ATCC 27536 based on their genomic fingerprints using PFGE . Vincent et al.  also noted that isolates from commercial products were classified as belonging to the B. animalis group. However, according to their RAPD profiles, these isolates were more similar to the reference strains ATCC 27536, 27674 and 27673 than the ATCC type strain 25527. Recently, Meile et al.  described an isolate of Bifidobacterium sp. as a new species which was named B. lactis. These authors noted that B. animalis is closely related (98.6% similarity) to B. lactis according to the 16S rRNA sequence data. However, DNA–DNA hybridization showed only weak homology between B. lactis and the type strain of B. animalis (DSM 20104=ATCC 25527T). In the present study, the analysis of a short region of the ldh gene showed that strains of B. lactis and B. animalis ATCC 27536 were identical although it was possible to distinguish between B. lactis and B. animalis ATCC 25527T.
The combination of specific primers (PCR), ARDRA and analysis based on conserved genes such as ldh provided a distinction between bifidobacterial species. Indeed, the ARDRA method made it possible to characterize B. infantis, B. longum and B. animalis with the combination of BamHI, TaqI and Sau3AI restriction enzymes. Alignment of short regions of the ldh gene also confirmed that it is possible to distinguish between B. longum and B. infantis but not between B. lactis and B. animalis. This new approach showed that the ldh gene for the genus Bifidobacterium is well conserved and it is possible to use it for the identification and speciation of strains. Further work will be performed on the analysis of PCR-amplified ldh gene by denaturing gradient gel electrophoresis to identify isolates from probiotic products.
The authors wish to thank the Natural Sciences and Engineering Research Council of Canada (Ottawa, Ont., Canada) (Research Partnerships Program – Research Network on Lactic Acid Bacteria), Agriculture and Agri-Food Canada (Ottawa, Ont., Canada), Novalait Inc. (Quebec, Que., Canada), Dairy Farmers of Canada (Ottawa, Ont., Canada), and Institut Rosell Inc. (Montreal, Que., Canada) for financial support.