ABSTRACT
Lactobacilli are lactic acid bacteria that are widespread in the environment, including the human diet and gastrointestinal tract. Some Lactobacillus strains are regarded as probiotics because they exhibit beneficial health effects on their host. In this study, the long-used probiotic strain Lactobacillus rhamnosus 35 was characterized at a molecular level and compared with seven reference strains from the Lactobacillus casei group. Analysis of rrn operon sequences confirmed that L. rhamnosus 35 indeed belongs to the L. rhamnosus species, and both temporal temperature gradient gel electrophoresis and ribotyping showed that it is closer to the probiotic strain L. rhamnosus ATCC 53103 (also known as L. rhamnosus GG) than to the species type strain. In addition, L. casei ATCC 334 gathered in a coherent cluster with L. paracasei type strains, unlike L. casei ATCC 393, which was closer to L. zeae; this is evidence of the lack of relatedness between the two L. casei strains. Further characterization of the eight strains by pulsed-field gel electrophoresis repetitive DNA element-based PCR identified distinct patterns for each strain, whereas two isolates of L. rhamnosus 35 sampled 40 years apart could not be distinguished. By subtractive hybridization using the L. rhamnosus GG genome as a driver, we were able to isolate five L. rhamnosus 35-specific sequences, including two phage-related ones. The primer pairs designed to amplify these five regions allowed us to develop rapid and highly specific PCR-based identification methods for the probiotic strain L. rhamnosus 35.
Members of the genus Lactobacillus are ubiquitous anaerobic, gram-positive, non-spore-forming rods or coccobacilli. In the environment, they are present in soil, in water, and on plants, particularly on decaying vegetation. In animals and humans, they are indigenous commensal inhabitants of the oral cavity, the gastrointestinal tract, and the female genital tract (26). Lactobacilli can also be isolated from a wide range of foods and beverages as spoilage species or, more often, as fermentative and biopreservative agents (37). They belong to the lactic acid bacteria (LAB), which are characterized by the ability to produce lactic acid by fermention of carbohydrates, resulting in acidification of the growth environment and contributing to growth inhibition of undesirable microorganisms (25). They have a predominant role in the fermentation (38) and flavor development (40) of many traditional dishes and drinks and are extensively used in the food industry as starters in fermented products, mainly dairy products such as cheese and yogurt.
In addition, some strains of Lactobacillus have beneficial health effects, in particular the protection of the gastrointestinal and genital tracts against pathogens (17), and have been marketed for many years as probiotics in pharmaceuticals or functional foods. Probiotics have been defined by the Food and Agriculture Organization, the World Health Organization, and the International Scientific Association for Probiotics and Prebiotics as “live microorganisms which, when administered in adequate amount, confer a health benefit on the host” (18, 33). Their long history of safe use in fermenting foods and their natural presence in the normal human microflora led to lactobacilli being considered as GRAS (generally regarded as safe) bacteria. As a result of the growing popularity of probiotics, numerous so-called probiotic products have been marketed, in most cases without proper trials, which has raised concern about their real safety and efficacy. The latter two properties are strain specific and thus cannot be extrapolated to a whole genus or species (19). Strain characterization is all the more difficult since traditional phenotypic identification methods are poorly reliable and often result in misidentifications with regard to LAB. Furthermore, the taxonomy of lactobacilli is constantly evolving along with molecular biology methods, and certain confusions arise. For instance, the status of the Lactobacillus casei group, which contains several probiotic bacteria, is not definite; the group comprises the species L. casei, L. paracasei, L. rhamnosus, and L. zeae, but most of the strains described as L. casei differ more from the species type strain (ATCC 393T) than from L. paracasei strains (9, 14).
The probiotic strain Lactobacillus rhamnosus 35, formerly called Lactobacillus casei rhamnosus 35, is an aerotolerant, homofermentative bacterium originally isolated from the gut flora of a healthy infant. It constitutes the active substance of different pharmaceutical products that have been successfully used in the treatment and prevention of diarrhea for more than four decades. This strain has properties fitting the definition of a probiotic: it is highly resistant to technological processes such as freeze-drying, is able to adhere to intestinal epithelial cells in vitro and to inhibit the growth and adherence of several pathogens (20), and can survive and persist within the gastrointestinal tract (10).
The aim of the present study was to characterize the probiotic bacterial strain L. rhamnosus 35 at a molecular level for phylogenetic purposes and to acquire an alternative to the biochemical culture-based identification methods currently used. The rrn operons of L. rhamnosus 35 and of seven reference strains belonging to the L. casei group were analyzed by sequencing, temporal temperature gradient gel electrophoresis (TTGE), and ribotyping, which validated the classification of L. rhamnosus 35 in the species L. rhamnosus and showed its close relatedness to the probiotic strain L. rhamnosus ATCC 53103 (also known as L. rhamnosus GG). Pulsed-field gel electrophoresis (PFGE) and repetitive DNA element-based PCR (rep-PCR), which are more discriminatory methods, were able to specifically identify each of the eight strains studied but could not distinguish two isolates of L. rhamnosus 35 sampled 40 years apart. Finally, five nucleotide sequences specific for L. rhamnosus 35 were revealed by performing subtractive hybridization between genomic DNAs from this strain and from L. rhamnosus GG, which offers the possibility of rapidly identifying this probiotic by strain-specific PCR.
MATERIALS AND METHODS
Bacterial strains and culture conditions. Lactobacillus rhamnosus 35 isolates conserved as lyophilizates since 2005 and 1966 were obtained from Laboratoires Lyocentre (Aurillac, France), with the batch freeze-dried in 2005 originating from successive subcultures of the lyophilizate from 1966. The reference Lactobacillus casei group strains L. casei ATCC 393T, L. casei ATCC 334, L. paracasei tolerans ATCC 25599T, L. paracasei paracasei ATCC 25302T, L. rhamnosus ATCC 7469T, L. rhamnosus ATCC 53103 (L. rhamnosus GG), and L. zeae ATCC 15820T were purchased from the collection of the Centre de Ressources Biologiques de l'Institut Pasteur (CRBIP). Lactobacilli were grown without agitation in De Man-Rogosa-Sharpe (MRS) medium (BD Difco, BD Diagnostics, Le Pont-De-Claix, France) at 37°C, except for L. paracasei toleransT, which was grown at 30°C in the presence of CO2 (bioMérieux, Marcy l'Etoile, France). Escherichia coli JM109 (ATCC 53323) was grown in Luria-Bertani (LB) medium (BD Difco) at 37°C under agitation, with addition of 50 μg/ml of kanamycin or 100 μg/ml of amoxicillin when required.
Preparation of DNA.Plasmid DNA was purified with the NucleoSpin plasmid kit as recommended by the manufacturer (Macherey-Nagel, Hoerdt, France). Routinely, total genomic DNA was extracted either with the NucleoSpin tissue kit (Macherey-Nagel) or by a method adapted from that described by Fouet and Sonenshein (21). Briefly, bacterial cells were harvested, washed twice with and resuspended in 100 mM Tris-HCl (pH 8)-100 mM EDTA-150 mM NaCl, and incubated in the presence of lysozyme (1 mg/ml) for 45 min at 37°C. One volume of 150 mM NaCl-10 mM EDTA (pH 8)-1.25% sodium dodecyl sulfate was then added to the samples before DNA was isolated from the cell debris by extracting once with an equal volume of phenol-chloroform and once with chloroform. DNA was precipitated by adding 1 volume of isopropanol, washed with 80% ethanol, and resuspended in water. For PFGE and ribotyping, genomic DNA was extracted in situ in agarose plugs, which were prepared according to a method adapted from those described by Tynkkynen et al. (44) and Yeung et al. (51). Briefly, bacteria were grown in MRS broth containing 1% glycine and 0.05% l-cysteine. Chloramphenicol was added at a final concentration of 100 μg/ml, and incubation was continued for 2 h. Cells were harvested and washed twice with 10 mM Tris-20 mM NaCl-50 mM EDTA (pH 7.2). The pellets were suspended in 0.5× Tris-borate-EDTA (TBE) buffer to an A650 of 1.2, warmed to 50°C, and mixed with an equal volume of 2% low-melting-point agarose (Invitrogen, Cergy Pontoise, France) in 0.5× TBE buffer at the same temperature. The suspension was allowed to solidify for 30 min at 4°C into plug molds. Cells included in the agarose blocks were lysed in situ with 10 ml of lysis solution (1 mg/ml lysozyme [Sigma-Aldrich, St-Quentin-Fallavier, France], 400 U mutanolysin [Sigma-Aldrich], 6 mM Tris-HCl, 1 M NaCl, 100 mM EDTA [pH 7.6], 0.2% sodium deoxycholate, 1% Na-lauroylsarcosine) at 37°C for 24 h and then with 5 ml of proteolysis solution (1 mg/ml proteinase K [Sigma-Aldrich], 250 mM EDTA [pH 8], 1% Na-lauroylsarcosine, 0.2% sodium deoxycholate) at 50°C for 48 h. The agarose blocks were washed under agitation once at 37°C for 1 h with 20 ml of 1 mM phenylmethylsulfonyl fluoride in 1× TE (10 mM Tris-HCl [pH 8], 1 mM EDTA), once at room temperature for 1 h with 20 ml of 1 mM phenylmethylsulfonyl fluoride in 1× TE, and three times at room temperature for 1 h with 20 ml of 1× TE. The plugs were stored at 4°C in 50 ml of fresh 1× TE until they were subsequently used for ribotyping and PFGE.
Sequencing.To sequence the L. rhamnosus 35 rrn operon, a genomic library was constructed by partially digesting L. rhamnosus 35 total DNA with Sau3AI and ligating the resulting 25- to 60-kb fragments into the pHC79 cosmid vector. In vitro packaging was carried out using Gigapack III XL packaging extract (Stratagene, Amsterdam Zuidoost, The Netherlands). The library was screened by colony hybridization with a DNA probe specific for the 16S rRNA gene of L. rhamnosus 35 (10). The probe was labeled by random priming using [α-32P]dCTP and High Prime (Roche Applied Science, Meylan, France). PstI fragments of a positive clone were then subcloned into pUC18, and the recombinant plasmid harboring the L. rhamnosus 35 16S rRNA gene served as a matrix for sequencing of the rrn operon using a chromosome walking strategy. The sequences were obtained from Genome Express (Meylan, France) and analyzed using the Basic Local Alignment Search Tool (BLAST, v2.2.14). To sequence the intergenic transcribed spacers (ITS), the 16S-23S ITS of L. rhamnosus 35 and the 23S-5S ITS of all the strains were amplified with Platinum High Fidelity Taq DNA polymerase (Invitrogen) using the primers lacto16-23 5′/3′ (5′-TAATCGCGGATCAGCACGC-3′/5′-ATTTCACGTGTTCCGCCGTA-3′) and lacto23-5 5′/3′ (5′-GTACCAGTTGTGCCGCCAGG-3′/5′-AGGCAGTTTCCCACCAACTA CT-3′), respectively. Thirty amplification cycles were performed, with denaturation for 30 s at 94°C, annealing for 30 s at 55°C (lacto16-23 5′/3′) or 61°C (lacto23-5 5′/3′), and extension at 68°C for 1 min (lacto16-23 5′/3′) or 30 s (lacto23-5 5′/3′). PCR products were cloned using the Qiagen PCR cloning kit (Qiagen, Courtaboeuf, France) and sequenced from recombinant vectors of E. coli JM109 transformants using the SP6 universal primer. Sequences were compared by Clustal-like multiple-sequence alignment (MAFFT-L-INS-i v6.500a), and dendrograms were generated by the unweighted-pair group method using average linkages (UPGMA). Sequences were considered to form a cluster if the branch length value was <0.2. The 16S-23S ITS sequences of L. rhamnosus 35 were compared with nucleotide databases (blastn) and tRNA gene sequences (tRNAscan-SE v.1.21).
TTGE.The V6-V7-V8 regions of the 16S rRNA gene were amplified by PCR using the primers GC-968f (5′-CGCCCGGGGCGCGCCCCGGGCGGGGCGGGGGCACGGGGGGAACGCGAAGAACCT-3′) and 1401r (5′-CGGTGTGTACAAGACCC-3′) according to the method of Lay et al. (29). Amplification was achieved using 2.5 U of HotStar Taq DNA polymerase (Qiagen), 2.5 mM of MgCl2, 200 μM of each deoxynucleoside triphosphate, and 0.4 μM of each primer. An initial denaturation step at 95°C for 15 min was followed by 30 cycles of denaturation for 1 min at 97°C, annealing for 1 min at 58°C, and elongation for 90 s at 72°C and a final extension at 72°C for 15 min. Electrophoresis was performed using the DCode universal mutation detection system (Bio-Rad, Hercules, CA). PCR products (300 ng) were loaded on an 8% polyacrylamide gel containing urea (7 M), Tris-acetate-EDTA (1.25×), N,N,N′,N′-tetramethylethylenediamine (0.06%), and ammonium persulfate (0.06%). After 15 min at 66°C and 20 V, electrophoresis was performed for 17 h at 68 V with a ramp rate of 0.2°C/h (temperature ranging from 66°C to 69.7°C).
Ribotyping.Two restriction enzymes, EcoRI and HindIII (Roche), were used separately to digest agarose-embedded DNA. Genomic DNAs were digested with 40 U of enzyme for 16 h at 37°C. Restriction fragments were separated by electrophoresis in a 0.8% agarose gel in 0.5× TBE buffer and transferred to a nylon membrane (Roche). The digoxigenin (DIG)-labeled 16S rRNA gene probe was obtained by PCR using the pair of primers 27f/1492r (49) with 200 μM each of dATP, dCTP, and dGTP, 190 μM of dTTP, and 10 μM of DIG-dUTP (Roche). Thirty-five amplification cycles (94°C for 1 min, 65°C for 1 min, and 72°C for 2 min) were performed with 1 U of Taq DNA polymerase (MP Biomedicals, Illkirch, France) using a mixture of DNAs from the different Lactobacillus strains as a template. Hybridization of the probe was detected with the DIG nucleic acid detection kit (Roche) according to the manufacturer's instructions.
PFGE.Macrorestriction of genomic DNA was performed with 40 U of the endonuclease NotI (New England Biolabs, Hitchin, United Kingdom) for 16 h at 37°C. PFGE was carried out with a CHEF-DR III apparatus (Bio-Rad) in a 1% agarose gel in 0.5× TBE at 10°C. The following running parameters were applied: voltage, 6 V/cm; angle, 120°; and pulse ramp, 20 to 1 s for 32 h. The gels were stained with ethidium bromide and photographed under UV light. A lambda ladder (successively larger concatemers of 48.5-kb DNA fragments) was used as a molecular size marker. PFGE patterns were visually compared according to the criteria of Tenover et al. (43) and further analyzed with GelCompar software (Applied Maths, St-Martens-Latem, Belgium), and UPGMA dendrograms were generated using the Dice or the Jaccard coefficient of similarity and a band position tolerance of 1%. Isolates were considered to be within a cluster if the Dice coefficient of similarity was >80%.
rep-PCR and rep-RT-PCR.Amplifications were performed with 2.5 U of Taq DNA polymerase (MP Biomedicals), 200 μM of each deoxynucleoside triphosphate, and 1 μM of each primer. An initial denaturation at 94°C for 5 min was followed by 35 cycles of denaturation for 30 s at 94°C, annealing for 1 min, and elongation at 72°C for 4 min and a final extension at 72°C for 7 min. Reactions were carried out with annealing at 40°C for the primer pair REP1R-I/REP2-I (5′-IIIICGICGICATCIGGC-3′/5′-IIICGNCGNCATCNGGC-3′), 50°C for the primer BOXA1R (5′-CTACGGCAAGGCGACGCTGACG-3′), and 45°C for the primer (GTG)5 (5′-GTGGTGGTGGTGGTG-3′) and with 4 mM, 2 mM, and 3 mM of MgCl2, respectively. Reproducibility was assessed by performing three different PCR runs using three independent DNA extracts for each strain. Samples were analyzed with BioNumerics software (Applied Maths) with Pearson's correlation coefficient and the UPGMA method. Isolates with >90% similarity were considered identical. For rep-reverse transcription-PCR (rep-RT-PCR), total RNA was extracted from 10 ml of overnight bacterial culture, as described by Gosink et al. (23), and treated with DNase I (Roche) to remove any contaminating genomic DNA. One microgram of RNA was reverse transcribed for 50 min at 42°C using 200 U of SuperScript II reverse transcriptase (Invitrogen). Reverse transcription was preceded or not by incubation of RNA with the primer(s) at 25°C for 10 min. Reactions were performed with 2.5 μM of BOXA1R, of (GTG)5, or of each of the primers REP1R-I and REP2-I. rep-PCR was performed with the same primer(s) as for the reverse transcription, using 1/10 of the resulting cDNA as a template. The reproducibility of the method was assessed by repeating the experiments with three independent L. rhamnosus 35 RNA extracts, from cultures in MRS broth, in MRS broth containing 1% glycine, and in sterile milk.
Subtractive hybridization.Suppression subtractive hybridization was carried out using the Clontech PCR-Select bacterial genome subtraction kit (Clontech-Takara Bio Europe, St-Germain-en-Laye, France) as recommended by the supplier. DNAs from L. rhamnosus 35 and L. rhamnosus GG were used, respectively, as tester and driver. Amplification products obtained by suppression subtractive hybridization were cloned into pDrive (Qiagen). The subtraction library thus constructed was screened using radiolabeled RsaI-digested L. rhamnosus GG DNA. Putative L. rhamnosus 35-specific fragments were sequenced from plasmid DNA of nonhybridizing clones using the SP6 vector primer. Primer pairs hyb-1 F/R, hyb-10 F/R, hyb-13 F/R, hyb-15 F/R, and hyb-21 F/R (Table 1) were designed to amplify each sequence and tested in 30 PCR cycles (30 s at 94°C, 30 s at 56°C, and 1 min at 72°C) with Taq DNA polymerase (MP Biomedicals) using DNA template from L. rhamnosus 35. Primer pair specificity was then assessed by 35 PCR cycles using template DNAs from other L. casei group strains and from L. acidophilus ATCC 4356, L. brevis ATCC 14869, Bacillus subtilis ATCC 633, Enterococcus faecalis ATCC 29212, Lactococcus lactis ATCC 19435, Staphylococcus aureus ATCC 25923, Staphylococcus epidermidis ATCC 12228, Streptococcus pneumoniae ATCC 49619 (purchased from CRBIP), Klebsiella pneumoniae LM21, L. acidophilus CH537, Clostridium perfringens CH106, Enterococcus faecium CH675, Lactococcus garvieae CH205, Listeria monocytogenes CH111, Listeria innocua CH112, Streptococcus pyogenes TPSb (laboratory strain collection), E. coli Nissle 1917, and L. plantarum WCSF1. Sequences were further analyzed using blastn and tblastx.
In silico analysis of L. rhamnosus 35-specific subtracted sequences
RESULTS
Phylogenetic analysis of L. rhamnosus 35 by examination of rrn operons.The 5,233 bp of the L. rhamnosus 35 rrn operon sequenced (GenBank accession number EU184020) shared 99% identity with partial rrn sequences of four L. rhamnosus strains, 98% identity with those of L. casei ATCC 334 and L. paracasei, 96% identity with that of L. acidophilus NCFM, 94% identity with those of L. johnsonii NCC 533 and L. plantarum WCSF1, and 93% identity with those of L. reuteri F275, L. sakei 23K, and L. salivarius UCC118. The L. rhamnosus 35 rrn sequence was found to share 90% or less identity with sequences from other lactobacilli (L. brevis, L. delbrueckii, L. murinus, and L. animalis), which is as high as identities found for other LAB genera such as Pediococcus, Lactococcus, Leuconostoc, Weisella, and Enterococcus, and even for Bacillus and Listeria strains. Further comparisons of the L. rhamnosus 35 16S rRNA gene sequence showed 99% identity with strains belonging to the species L. casei, L. rhamnosus, and L. zeae. Thus, L. rhamnosus 35 shares greater similarity with the L. casei group than with other bacterial species. However, these data did not provide a definite identification of the strain, since the four different species constituting the L. casei group could not be clearly distinguished from one another.
Three different nucleotide sequences were found when the L. rhamnosus 35 16S-23S ITS was sequenced (Fig. 1). Two 217-bp sequences, differing by only one base at nucleotide position 161, were called Lcr16-23 s and Lcr16-23 s′. Database comparison of these two sequences revealed 100% (Lcr16-23 s) and 99% (Lcr16-23 s′) identity with 16S-23S ITS from both L. rhamnosus and L. zeae strains. The third sequence, named Lcr16-23 l, was 430 bp long; it differed from Lcr16-23 s by eight nucleotides, and its extra 213-bp sequence contained mostly genes potentially encoding tRNAIle and tRNAAla, located tandemly from positions 87 to 163 and 195 to 267, respectively. The whole Lcr16-23 l sequence showed 100% identity with 16S-23S ITS sequences from L. rhamnosus strains and shared less than 96% identity with those of any other species. Using cloned PCR products as a matrix, three to six different 23S-5S ITS sequences were isolated and analyzed for each of the eight following strains: L. rhamnosus 35, L. casei ATCC 393T, L. casei ATCC 334, L. paracasei tolerans ATCC 25599T, L. paracasei paracasei ATCC 25302T, L. rhamnosus ATCC 7469T, L. rhamnosus GG, and L. zeae ATCC 15820T. These sequences were compared by multiple alignment, and a UPGMA dendrogam was constructed (Fig. 2). Three clusters emerged. L. zeaeT and L. casei ATCC 393T were gathered in one cluster, while a second cluster included the other L. casei strain, ATCC 334, together with the two L. paracasei strains. L. rhamnosus 35 composed a third cluster together with the probiotic L. rhamnosus GG and the L. rhamnosus type strain. Taken together, these results provide strong evidence for L. rhamnosus 35 belonging to the L. rhamnosus species and thus support the reclassification of this strain, formerly named L. casei subsp. rhamnosus, as L. rhamnosus. Furthermore, the L. casei type strain shares greater homology with L. zeaeT than with L. casei ATCC 334, which is closer to the L. paracasei strains included in this study.
Multiple-alignment comparison of the three L. rhamnosus 35 16S-23S ITS sequences. Sequences Lcr16-23 s and Lcr16-23 s′ are 217 bp long, and Lcr16-23 l consists of 430 bp. The 208 nucleotides which are identical in the three sequences are indicated by asterisks, gaps are represented by dashes, and single-nucleotide differences are boxed. tRNAIle- and tRNAAla-encoding genes (in bold) were found within the sequence Lcr16-23 l from position 87 to 163 and position 195 to 267, respectively.
Taxonomic dendrogram of L. casei group strains based on the 23S-5S nucleotide sequences. Three to six different 23S-5S ITS sequences were determined for each of the strains. The UPGMA dendrogram was generated by multiple-alignment comparison of these sequences. Lcr35, L. rhamnosus 35; L. GG, L. rhamnosus GG.
The TTGE patterns obtained by amplifying the V6-V7-V8 regions of the 16S rRNA genes from L. rhamnosus 35, L. rhamnosus GG, L. rhamnosus ATCC 7469T, L. casei ATCC 334, and the two L. paracasei strains were highly similar, despite the presence of an extra band in the L. rhamnosus type strain profile (Fig. 3). The L. casei ATCC 393T and L. zeaeT TTGE patterns exhibited a common band distinct from that in the other lactobacilli profiles, but they differed clearly in the position and intensity of their second bands. These data confirm that L. casei ATCC 334 has greater homology with L. paracasei strains than with the species type strain, which displayed a single TTGE pattern.
TTGE patterns of L. casei group strains. The V6-V7-V8 regions of the 16S rRNA-encoding gene were amplified and subjected to temperature gradient electrophoresis. MW, molecular weight marker; Lcr35, L. rhamnosus 35; L. GG, L. rhamnosus GG.
All the ribotype profiles obtained after EcoRI digestion of genomic DNAs from the nine Lactobacillus isolates differed from one another by at least one band, except for L. rhamnosus GG, L. rhamnosus ATCC 7469T, and the two isolates of L. rhamnosus 35, which showed the same pattern (Fig. 4A). Similar results were obtained with HindIII digestion, except for L. rhamnosus ATCC 7469T, whose profile was different by one band from those of L. rhamnosus 35 and L. rhamnosus GG (Fig. 4B). Regarding the other strains, and whatever the endonuclease used, the L. casei ATCC 334, L. paracasei tolerans ATCC 25599T, and L. paracasei paracasei ATCC 25302T ribotypes shared three bands and had only one band in common with L. casei ATCC 393T, whereas the L. zeae ATCC 15820T EcoRI ribotype had three common bands with that of L. casei ATCC 393T. These results indicate that L. casei ATCC 334 is closer to the L. paracasei type strains than to L. casei ATCC 393T, with the latter being more closely related to L. zeaeT. Moreover, they suggest that among the L. rhamnosus strains, L. rhamnosus 35 is more closely related to the probiotic L. rhamnosus GG than to the type strain.
Ribotype patterns of L. casei group strains. Ribotyping was performed using the endonuclease EcoRI (A) and the endonuclease HindIII (B). Lcr35, L. rhamnosus 35; L. GG, L. rhamnosus GG.
Strain-specific molecular fingerprinting.The PFGE patterns of L. rhamnosus 35, L. casei ATCC 393T, L. casei ATCC 334, L. paracasei tolerans ATCC 25599T, L. paracasei paracasei ATCC 25302T, L. rhamnosus ATCC 7469T, L. rhamnosus GG, and L. zeae ATCC 15820T were all distinct, as they differed from one another by more than seven bands (Fig. 5A). These different fingerprints shared less than 60% similarity when calculated with the Jaccard coefficient and less than 75% when calculated with the Dice coefficient (Fig. 5B). The results of PFGE pattern comparisons thus provide evidence that each of the strains used in this study is unique. In addition, the two L. rhamnosus 35 isolates sampled 40 years apart displayed identical NotI macrorestriction patterns.
PFGE analysis of L. casei group strains. The PFGE fingerprints (A) were obtained after NotI macrorestriction, and pattern similarity dendrograms (B) were constructed. λ, lambda molecular weight ladder; Lcr35, L. rhamnosus 35; L. GG, L. rhamnosus GG.
rep-PCRs performed using REP1R-I/REP2-I, BOXA1R, and (GTG)5 gave similar results whatever the primers used; each bacterial strain displayed a reproducible and specific pattern (Fig. 6). Although the L. casei ATCC 334 and L. paracasei paracasei ATCC 25302T REP1R-I/REP2-I patterns had a 92.35% correlation, their fingerprints were easily distinguishable visually, and the BOXA1R and (GTG)5 fingerprints of these two strains were correlated by only 85.74% and 89.94%, respectively. The rep-PCR patterns of all of the other strain shared less than 85% correlation. The fingerprints of the two L. rhamnosus 35 isolates were visually identical and showed Pearson's correlations of more than 96%. rep-RT-PCR performed with REP1R-I/REP2-I, BOXA1R, and (GTG)5 gave results similar to those of rep-PCR (Fig. 7); all the strains exhibited specific patterns whatever the primer set used. These results further confirm the singularity of these bacterial strains and the discriminative power of this method.
rep-PCR fingerprinting of L. casei group strains. The patterns were obtained using the primer pair REP1R-I/REP2-I (A), the primer BOXA1R (B), and the primer (GTG)5 (C). MW, molecular weight marker; Lcr35, L. rhamnosus 35; L. GG, L. rhamnosus GG.
rep-RT-PCR patterns of L. casei group strains. The patterns were obtained using the primer pair REP1R-I/REP2-I (A), the primer BOXA1R (B), and the primer (GTG)5 (C). MW, molecular weight marker; Lcr35, L. rhamnosus 35; L. GG, L. rhamnosus GG.
Identification of L. rhamnosus 35-specific nucleotide sequences.Suppression subtractive hybridization revealed five L. rhamnosus 35 DNA sequences different from the L. rhamnosus GG genome, and corresponding primer pairs were specifically designed (Table 1). The five regions were successfully amplified using genomic DNA extracted from either of the L. rhamnosus 35 isolates as the template, as well as from two L. rhamnosus 35 isolates grown under different conditions during the last decades (data not shown). Nucleotide database comparison of three of the L. rhamnosus 35-specific sequences, hyb-1, hyb-10, and hyb-13, revealed no significant similarity. However, their translated sequences showed homologies with already-described bacterial proteins (Table 1). Regarding the two other specific fragments, the nucleotide sequences of hyb-15 and hyb-21 were, respectively, 91% and 69% identical to different parts of the gene encoding the tape measure protein of L. casei bacteriophage A2 (accession number AJ251789). Combining the distal primers of these two pairs (hyb-15F/hyb-21R), a 2,495-bp fragment was amplified from the L. rhamnosus 35 genome and then sequenced. The deduced amino acid sequence of this fragment had 86% homology with the bacteriophage A2 tape measure protein, consistent with our previous results.
None of these five putative L. rhamnosus 35-specific sequences could be amplified using template DNAs extracted from strains belonging to different genera of gram-positive and gram-negative bacteria, including genera closely related to Lactobacillus such as Streptococcus, or from lactobacilli not belonging to the L. casei group. Moreover, no amplification was observed using either of the five primer pairs with DNA extracts from the seven L. casei group reference strains L. casei ATCC 393T and ATCC 334, L. paracasei tolerans ATCC 25599T, L. paracasei paracaseiT, L. rhamnosusT, L. rhamnosus GG, and L. zeaeT. Such results indicate that the five nucleotide sequences identified by subtractive hybridization are indeed highly specific for the strain L. rhamnosus 35.
DISCUSSION
Probiotics have aroused increasing interest in recent years, and their popularity has led to the emergence of numerous misidentified strains and mislabeled products on the market (8, 16, 24, 28, 42, 48). Such inaccuracies are principally due to the use of improper typing methods, delays in revising nomenclature, and the debatable taxonomic status of some Lactobacillus species (28). Over and above misinformation given to consumers, the problem raised by incorrect labeling is that the efficacy and innocuousness of probiotics are strain-specific properties. The correct identification and regular control of probiotic strains are thus essential to guarantee their identity and thereby their safety and proclaimed health benefits.
L. rhamnosus 35 has been used as a pharmaceutical probiotic for more than 40 years, and the absence of any related side or adverse effects together with the results of clinical trials suggest that it is innocuous (5, 45). To date, however, the strain has been identified by carbohydrate fermentation profiling, which takes several days and has poor taxonomic resolution and reliability (12, 41). In contrast, DNA-based techniques are useful tools for the accurate and reliable characterization of bacteria, since they are based on constant genetic properties of the strains (2, 31). In addition, most give quicker results than culture-based techniques. In order to accurately characterize L. rhamnosus 35 and to develop rapid identification tools for this strain, molecular typing methods were used to compare L. rhamnosus 35 with seven reference L. casei group strains.
Analyses of the sequence and distribution of the rRNA-encoding genes are widely used strategies for the classification and typing of bacteria (19, 34). In this study, the sequence of one complete rrn operon of L. rhamnosus 35 was determined and compared with databases. The resulting information was scarce, since the only Lactobacillus sequences available for BLAST comparison are either rare completely sequenced genomes or partial 16S or 16S-23S ITS sequences. Our data nonetheless provided supporting evidence that L. rhamnosus 35 belongs to the L. casei group, but they did not allow further taxonomic classification of this strain, since similarities of above 97%, which is the generally accepted limit, were found with DNA sequences from several species. Difficulties in identifying Lactobacillus species and strains by rRNA gene sequencing may be attributable to particularly high similarities between rRNA-encoding genes of close species and low phylogenetic coherence of this genus (34). Characterization at the species level by 16S rRNA gene sequencing is further hampered by the intraspecies polymorphism of these regions and by their intrastrain heterogeneity due to the presence of several rrn operons within the genome (7, 46). However, specific constant sequences at certain nucleotide positions, called sequence signatures, have been described to overcome these problems and discriminate L. casei group species from one another. The sequence determined for L. rhamnosus 35 matched L. rhamnosus 16S rRNA gene sequence signatures (30, 46) and the L. rhamnosus 16S-23S ITS sequence signature (15). Sequence signature analysis is therefore a convenient alternative to database comparison when the taxonomic resolution power of the latter is insufficient, as it is regarding the L. casei group.
The ITS sequences are a mosaic of highly conserved and hypervariable regions. Their great variability and specificity make them potentially more discriminating regions for phylogenetic purposes than rRNA-encoding genes (32), and they are commonly used for lactobacillus species identification (39) or intraspecific typing. Three distinct 16S-23S intergenic sequences were found in the L. rhamnosus 35 genome. The longest one included tandemly positioned tRNAIle and tRNAAla genes, an organization classically found in LAB and lactobacilli (13, 32). Database comparison of this sequence orientated identification toward the L. rhamnosus species. This result correlates with phenotypic identification results and with reclassification of L. rhamnosus 35, like other bacteria from the former L. casei subsp. rhamnosus (9), in the L. rhamnosus species. Analysis of the two shorter 16S-23S sequences did not discriminate L. rhamnosus from L. zeae. This underlines the need to sequence the diverse 16S-23S ITS coexisting in the genome to ensure good discrimination between closely related Lactobacillus strains, as already suggested for 16S rRNA gene-based phylogenetic analysis (7).
The sequences of the 23S-5S ITS have been described as more variable than those of the rRNA genes or even of the 16S-23S ITS, making them a yet more useful tool in taxonomic analyses and typing of closely related strains (6). However, these sequences have not been frequently studied, so only a few are available in databases, mainly in complete genome sequences. We therefore decided to sequence the 23S-5S ITS of both L. rhamnosus 35 and seven reference strains belonging to the L. casei group, and we were able to construct a taxonomic dendrogram by comparing these data. The dendrogram confirmed the results obtained by analysis of the 16S-23S ITS sequences; i.e., L. rhamnosus 35 grouped with the two reference L. rhamnosus strains studied. In addition, L. casei ATCC 334 appeared in the same cluster as L. paracasei tolerans ATCC 25599T and L. paracasei paracasei ATCC 25302T, while the species type strain L. casei ATCC 393T was closer to the L. zeae type strain. These results raise the question of the validity of the current taxonomy of the L. casei group, in which ATCC 334 and ATCC 393T are still classified in the same species in spite of evident discrepancies.
One of the problems encountered when studying rrn operons is the difficulty in directly sequencing PCR products, owing to the intrastrain multiplicity and sequence heterogeneity of these regions. TTGE is able to separate such PCR products, depending on their GC content, by regularly increasing the temperature during electrophoresis and thus exploits the polymorphism of the rRNA gene to generate specific patterns (46). TTGE of the 16S rRNA genes’ V6-V7-V8 fragments was able to discriminate between L. casei ATCC 393T, L. zeaeT, L. rhamnosusT, and the five other strains studied, i.e., L. rhamnosus 35, L. rhamnosus GG, L. casei ATCC 334, L. paracasei paracaseiT, and L. paracasei toleransT. The extra band displayed by L. rhamnosus ATCC 7469T suggests that L. rhamnosus 35 is closer to L. rhamnosus GG than to the species type strain. Ribotyping provided evidence that L. rhamnosus 35 harbors at least four rrn operons and belongs to the L. rhamnosus species. These data provided further information on the L. rhamnosus 35 phylogeny, corroborating that this probiotic strain is more closely related to the probiotic L. rhamnosus GG than to the L. rhamnosus type strain. Ribotyping also confirmed that L. casei ATCC 334 is closer to the L. paracasei strains than to the L. casei species type strain ATCC 393T. Strikingly, L. casei ATCC 393T displayed a TTGE profile distinct from that of L. zeaeT, while L. rhamnosus strains, L. paracasei strains, and L. casei ATCC 334 shared similar patterns. These findings are not in accordance with requests to reclassify L. casei ATCC 393T in the species L. zeae (11, 14) but emphasize the need to separate this strain and L. casei ATCC 334 into two different taxa.
Using PFGE, which is considered the gold standard of molecular genotyping methods, all Lactobacillus strains studied displayed single, distinct patterns after NotI macrorestriction, with identical patterns for the two L. rhamnosus 35 isolates. The UPGMA dendrogram generated by comparison of PFGE profiles did not correlate with the 23S-5S ITS-based dendrogram or with current taxonomic classification. Such findings are not surprising, since PFGE is appropriate for bacterial typing at the strain level but has too high a discriminatory potential to be suitable for species identification (51).
Unlike PFGE, rep-PCR not only is highly discriminative but also is rapid and reliable (47, 50). Despite the fact that it was first used mainly for pathogen subtyping in epidemiologic surveys, it has also proved to be a powerful identification tool with several types of LAB, including lactobacilli (1, 22, 28). This method was thus assessed using three different primer sets and gave results similar to those obtained with PFGE, confirming the accuracy of rep-PCR typing in strain identification. rep-RT-PCR, consisting of total RNA reverse transcription followed by rep-PCR with the same set of primers, could be more discriminative than PCR because the expression of mRNA can vary according to the environment or due to minor genetic rearrangements. The patterns obtained were distinct for each of the eight Lactobacillus strains tested, but the method had lower reproducibility. In addition, visual distinction between profiles of very close strains was less manifest than with rep-PCR fingerprints. In conclusion, rep-PCR was more rapid, was more reproducible, and had a higher resolution than rep-RT-PCR and is an even more convenient technique than PFGE for reliably determining the identity of lactobacilli at the strain level.
In order to devise a simple L. rhamnosus 35-specific identification assay, a subtractive hybridization was performed using L. rhamnosus 35 and L. rhamnosus GG as the tester and the driver, respectively. L. rhamnosus GG was chosen in accordance with our phylogenetic study, which showed it to be the closest strain to L. rhamnosus 35. Five L. rhamnosus 35-specific sequences were successfully identified with this method and were amplified from either of the L. rhamnosus 35 isolate genomes. This suggests that if these regions were acquired by genetic material transfer, the occurrence dates back to more than 40 years ago and hence that integration was stable. Database comparison of these fragments revealed two of them to be homologous with the sequence of the temperate bacteriophage A2. The presence of these phage-related sequences in the genome of L. rhamnosus 35 for more than four decades suggests the existence of a defective prophage. This sequence might also be part of a latent temperate phage or a remnant of an ancient virulent bacteriophage. Lysogeny and the presence of phage-related DNA in the bacterial chromosome are widespread in LAB such as streptococci (27) and lactobacilli (35, 36) and are recognized sources for genomic polymorphism of lactobacilli (4). The wide distribution and variation of phage sequences among the L. casei group led to their use as genetic markers, which not only allowed identification at the species level but actually gave the ability to design strain-specific PCR primers (3). In our study, no amplification of the L. rhamnosus 35 bacteriophage A2 homolog region was obtained when using DNA templates from a large variety of bacterial species, including L. rhamnosus, which confirms the usefulness of phage-related sequences as strain-specific markers.
In conclusion, PFGE, rep-PCR, and strain-specific PCR are valuable methods to determine L. rhamnosus 35 identity, as they were able to distinguish this strain from other closely related ones. In addition, no difference was found between two L. rhamnosus 35 clones isolated in 1966 and in 2005, even when using PFGE, which is the most discriminative molecular typing technique, confirming that this probiotic strain is currently the same as it was 40 years ago when first marketed. rep-PCR is the most useful technique to identify the different strains among the L. casei group, since it is less fastidious and time-consuming than PFGE. In addition, the identification of strain-specific sequences by subtractive hybridization led to the development of rapid routine L. rhamnosus 35-specific identity controls by PCR. Since the properties of probiotics are strain specific, the quality of products is closely linked to the degree of specificity of the identification techniques used. Hence, routine strain-specific PCR methods should be developed to a greater extent to guarantee the accurate and reliable identity of probiotics. If such molecular tools were used on a more regular basis, it is likely that there would be a reordering of the current taxonomic classification.
ACKNOWLEDGMENTS
Sophie Coudeyras is the recipient of a fellowship (CIFRE) from Laboratoires Lyocentre and Association Nationale de la Recherche Technique, Ministère de l'Enseignement Supérieur et de la Recherche.
We are grateful to Eric Oswald and Michiel Kleerebezem for having kindly provided the strains E. coli Nissle 1917 and L. plantarum WCSF1, respectively. We thank Lydia Gozzo for technical assistance. We also thank Corinne Teyssier and Josiane Campos for technical advice, Laure Diancourt and Sylvain Brisse for rep-PCR fingerprint software analysis, and Jean-Philippe Lavigne for PFGE pattern software analysis.
FOOTNOTES
- Received 8 October 2007.
- Accepted 27 February 2008.
- Copyright © 2008 American Society for Microbiology
