Identification of species belonging to the Bifidobacterium genus by PCR-RFLP analysis of a hsp60 gene fragment

Background Bifidobacterium represents one of the largest genus within the Actinobacteria, and includes at present 32 species. These species share a high sequence homology of 16S rDNA and several molecular techniques already applied to discriminate among them give ambiguous results. The slightly higher variability of the hsp60 gene sequences with respect to the 16S rRNA sequences offers better opportunities to design or develop molecular assays, allowing identification and differentiation of closely related species. hsp60 can be considered an excellent additional marker for inferring the taxonomy of the members of Bifidobacterium genus. Results This work illustrates a simple and cheap molecular tool for the identification of Bifidobacterium species. The hsp60 universal primers were used in a simple PCR procedure for the direct amplification of 590 bp of the hsp60 sequence. The in silico restriction analysis of bifidobacterial hsp60 partial sequences allowed the identification of a single endonuclease (HaeIII) able to provide different PCR-restriction fragment length polymorphism (RFLP) patterns in the Bifidobacterium spp. type strains evaluated. The electrophoretic analyses allowed to confirm the different RFLP patterns. Conclusions The developed PCR-RFLP technique resulted in efficient discrimination of the tested species and subspecies and allowed the construction of a dichotomous key in order to differentiate the most widely distributed Bifidobacterium species as well as the subspecies belonging to B. pseudolongum and B. animalis.


Background
Members of the genus Bifidobacterium are Grampositive, obligate anaerobic, non-motile, non-spore forming bacteria [1], and are the most important constituents of human and animal intestinal microbiota [2,3]. Recently, news species of bifidobacteria have been described [4][5][6] and now more than 30 species have been included in this genus.
Bifidobacterium spp. can be detected in various ecological environments, such as intestines of different vertebrates and invertebrates, dairy products, dental caries and sewage. Considering the increasing application of Bifidobacterium spp. as protective and probiotic cultures [7][8][9], and the fast enlargement of the genus, easy identification tools to discriminate new isolates are essential. Moreover, their correct taxonomic identification is of outmost importance for their use as probiotics [2]. Conventional identification and classification of Bifidobacterium species have been based on phenotypic and biochemical features, such as cell morphology, carbohydrate fermentation profiles, and polyacrylamide gel electrophoresis analysis of soluble cellular proteins [10]. In the last years several molecular techniques have been proposed in order to identify bifidobacteria. Most available bifidobacterial identification tools are based on 16S rRNA gene sequence analysis, such as ARDRA [11,12], DGGE [13] and PCR with the use of species-specific primers [14][15][16]. However, 16S rDNA of Bifidobacterium spp. has a high similarity, ranging from 87.7 to 99.5% and bifidobacterial closely related species (e.g. B. catenulatum and B. pseudocatenulatum) or subspecies (e.g. B. longum and B. animalis subspecies) even possess identical 16S rRNA gene sequences [17,18]. For this reason different molecular approaches have been tested based on repetitive genome sequences amplification, such as ERIC-PCR [19,20], BOX-PCR [21,22] or RAPD fingerprinting analysis [23]. These fingerprinting methods have the disadvantage of a low reproducibility, and they need

MB8 Chicken
Bifidobacterium pseudolongum subsp. globosum Ra27 Rabbit Bifidobacterium pseudolongum subsp. globosum VT366 Calf strict standardization of PCR conditions. The use of different polymerases, DNA/primer ratios or different annealing temperatures may lead to a discrepancy in the results obtained in different laboratories [24]. In recent years alternative molecular markers have been proposed for bifidobacteria identification (e.g. hsp60, recA, tuf, atpD, dnaK) and Ventura et al. [18] developed a multilocus approach, based on sequencing results, for the analysis of bifidobacteria evolution. The hsp60 gene, coding for a highly conserved 60 kDa heat-shock-protein (a chaperonin), has been evaluated for phylogenetic analysis in bifidobacteria by Jian et al. [25]. The sequence comparison of this gene has been already used for species identification and phylogenetic analysis of other genera (e.g. Staphylococcus, Lactobacillus) and enteric pathogens [26][27][28]. A chaperonin database (cpnDB) is available on line, collecting bacterial and eukaryotic sequences (http://www.cpndb.ca/ cpnDB/home.php) [29].  The purpose of this study is the development of a rapid, reproducible and easy-to-handle molecular tool for the identification of Bifidobacterium species isolated from various environments. The protocol is based on the restriction endonuclease analysis of the PCR-amplified hsp60 partial gene sequence (hsp60 PCR-RFLP) with the use of a single restriction enzyme and has been tested on the 30 most widely distributed Bifidobacterium species and subspecies. A diagnostic dichotomous key to speed up profile interpretation has also been proposed.

Bacterial strains and culture conditions
The type strains used to develop the technique are listed in Table 1, whereas the strains used to validate the method are reported in Table 2. The strains, belonging to BUSCoB (Bologna University Scardovi Collection of Bifidobacteria) collection, were isolated from faeces of human and animals and from sewage. Bacteria were maintained as frozen stocks at −80°C in the presence of skim milk as cryoprotective agent. Working cultures were prepared in TPY medium [1], grown anaerobically at 37°C and harvested at logarithmic phase.

In silico analysis
An in silico analysis was performed for the evaluation of a suitable restriction enzyme. Available hsp60 sequences had been retrieved from cpnDB database and GeneBank, thanks to the work of Jian et al. [25]. In silico digestion analysis was carried out on fragments amplified by universal primers H60F-H60R [30] using two on-line free software: webcutter 2.0 (http://rna.lundberg.gu.se/cutter2) and http://insilico.ehu.es/restriction softwares [31]. Blunt end, frequent cutter enzymes that recognize not degenerated sequences have been considered in order to find a suitable enzyme for all the species (e.g. RsaI, HaeIII, AluI, AccII). However in silico analysis had been performed also on sticky end enzymes (e.g. AatII, Sau3AI, PvuI).
DNA extraction from pure cultures 10 ml of culture were harvested and washed twice with TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.6), resuspended in 1 ml TE containing 15 mg lysozyme and incubated at 37°C overnight. Cells were lysed with 3 ml of lysis buffer (100 mM Tris-HCl, 400 mM NaCl, 2 mM EDTA, pH 8.2), 220 μl SDS (10% w/v) and 150 μl proteinase K (>600 mAU/ml, solution) and incubated for 2 hours in water bath at 60°C. One ml of saturated NaCl solution was added and the suspension was gently inverted twice. Pellets were harvested through centrifugation (5000 × g) at room temperature for 15 minutes. After the transfer of clean supernatants in new tubes, DNA was precipitated with 2.5 volumes of cold ethanol (95%) and resuspended in 300 μl of TE buffer [32]. MgCl 2 , 1 U Taq, 0.2 mM dNTP, final concentration) and 150 ng/μl DNA. The PCR cycle consisted of an initial denaturation of 5 min at 95°C followed by 35 cycles of denaturation (30s at 94°C), annealing (30s at 61°C) and extension (45 s at 72°C). The PCR was completed with a final elongation of 10 min at 72°C. The PCR amplification was performed with a PCR Verity 96-well thermal cycler (Applied Biosystems, Milan, Italy). After amplification, the product was visualized via agarose gel (1.3% w/v) in 1X TBE buffer and visualized with ethidium bromide under UV light. A 100 bp DNA ladder (Sigma-Aldrich) was used as a DNA molecular weight marker. Bands were excised from agarose gel (Additional file 1: Figure S1) and DNA was eluted with NucleoSpin W Gel and PCR Clean-up (Macherey-Nagel GmbH & Co. KG, Germany) in order to avoid possible non-specific amplifications. 3 μl of the eluted DNA was re-amplified in a 30 μl PCR reaction (see above). BSA was added to the reaction (5% v/v, Fermentas). The PCR products (2 μl) were checked for non-specific amplification on agarose gel. 20 μl (~6 μg) of PCR amplicons were digested with HaeIII enzyme. Restriction digestion was carried out for 2 h at 37°C in 30 μl  Precast gradient polyacrylamide gels (4-20%) (Lonza Group Ltd, Switzerland) were also used to obtain RFLP profiles, in order to have a comparison with agarose gels. The vertical electrophoresis apparatus used was P8DS™ Emperor Penguin (Owl, Thermo Scientific) with an adaptor for Lonza precast gels. The run was performed at 100 V in TBE 1X.

Diagnostic key
A dichotomous key was developed comparing in silico digestion results and the evaluation of visible bands with the use of ImageLab™ 2.0 software (Bio-Rad Laboratories, Inc.).

Results and discussion
In silico analysis The analysis and comparison of restriction profiles obtained with in silico digestion of bifidobacterial hsp60 sequences allowed the identification of a set of appropriate frequent-cutter endonucleases that recognize non degenerated sequences. The restriction enzyme HaeIII was found to give the clearest and most discriminatory profiles in theoretical PCR-RFLP patterns, discriminating the majority of Bifidobacterium type-strains tested (Table 3). Furthermore, the profiles of other strains, belonging to the investigated species, have been analyzed to confirm the conservation of RFLP profiles within species.
Amplification and restriction analysis of Bifidobacterium spp.
Theoretical restriction profiles have been confirmed in vitro on agarose gel. The obtained fragments ranged from 16 bp to 339 bp (Table 3). Fragments lower than 25 bp were not considered as they did not help in species discrimination and in addition they co-migrate with primers. Time course analysis of restricted samples showed the formation of a band of~200 bp in several species due to an over-digestion (data not shown) and this invalidated the RFLP profiles. For this reason the protocol has been optimized at 2 hours restriction time. Fragments greater than 360 bp were also not considered due to a possible incomplete digestion of such long fragments.
The obtained gels (Figures 1, 2, 3, 4 and 5) show species-specific profiles for all type-strains other than B. longum and B. thermacidophilum subspecies. This technique does not allow the identification of the subspecies belonging to these species, which displayed identical RFLP profiles. Matsuki et al. [14,17] proposed specific primers to differentiate the subspecies of the species B. longum, while B. thermacidophilum subsp. porcinum and B. thermacidophilum subsp. thermacidophilum can be differentiated according to Zhu et al. [33]. The proposed restriction analysis is efficient in discriminating very closely related species and subspecies as B. catenulatum/ B. pseudocatenulatum, B. pseudolongum subsp. pseudolongum/ B. pseudolongum subsp. globosum and B. animalis subsp. animalis/B. animalis. subsp. lactis.
The same method has been applied with the use of precast gradient polyacrylamide gels. The resolution was greater than that obtained on agarose gels, loading only 4 μl of the restriction reaction instead of the 30 μl used in horizontal electrophoresis. This may allow to reduce the volume of amplification reactions with a consequent reduction of costs.
The comparison between in silico digestion and the obtained gel profiles allowed to develop a dichotomous key ( Figure 6) for a faster interpretation of the restriction profiles.
Validation of PCR-RFLP analysis on bifidobacterial isolates 39 strains belonging to 12 different species/subspecies ( Table 2) have been investigated to validate the PCR- RFLP technique. Most of the strains tested were previously identified using biochemical tests and in some cases also molecular techniques (species-specific PCR, 16S rDNA sequencing). The obtained data confirmed a conservation of the profiles concerning the species and subspecies tested. Two figures are available as Additional files (Additional file 2: Figure S2: strains belonging to B. animalis subsp. lactis and B. animalis subsp. animalis. Additional file 3: Figure S3: strains belonging to B. longum subsp. longum, B. longum subsp. infantis, B. longum subsp. suis). About 95% of the strains confirmed the taxonomic identification previously assigned. Two strains, B1955 and Su864, previously classified as B. catenulatum and B. longum subsp. suis respectively, gave different profiles from those expected. The RFLP profiles of B1955 turned out to be the same of B. adolescentis ATCC 15703 (T), the dichotomous key confirmed the assignment to the B. adolescentis species. In addition, Su864 was identified as a B. breve strain. These results were also verified through a species-specific PCR [14].

Conclusions
In this work a PCR-RFLP based method to identify Bifidobacterium spp. was developed and tested on strains belonging to different species. The technique could efficiently differentiate all the 25 species of Bifidobacterium genus and the subspecies belonging to B. pseudolongum and B. animalis, with the support of an easy-to-handle dichotomous key. The technique turned out to be fast and easy, and presented a potential value for a rapid preliminary identification of bifidobacterial isolates.

Additional files
Additional file 1: Figure S1. Example of agarose gel electrophoresis of hsp60 amplicons from different bifidobacterial strains.

Competing interests
The authors declare that they have no competing interests. Figure 6 Dichotomous key to identify species of Bifidobacterium based upon HaeIII restriction digestion of~590 bp of the hsp60 gene.