SGM Journals
ENVIRONMENTAL AND EVOLUTIONARY MICROBIOLOGY

Multilocus sequence typing reveals a novel subspeciation of Lactobacillus delbrueckii

Kana Tanigawa, Koichi Watanabe

  • Yakult Central Institute for Microbiological Research, 1796 Yaho, Kunitachi, Tokyo 186-8650, Japan
  • Correspondence
    Koichi Watanabe
    koichi-watanabe{at}yakult.co.jp

    • Microbiology 2011; 157(3):727–738 · https://doi.org/10.1099/mic.0.043240-0

    View at publisher PubMed

    Abstract

    Currently, the species Lactobacillus delbrueckii is divided into four subspecies, L. delbrueckii subsp. delbrueckii, L. delbrueckii subsp. bulgaricus, L. delbrueckii subsp. indicus and L. delbrueckii subsp. lactis. These classifications were based mainly on phenotypic identification methods and few studies have used genotypic identification methods. As a result, these subspecies have not yet been reliably delineated. In this study, the four subspecies of L. delbrueckii were discriminated by phenotype and by genotypic identification [amplified-fragment length polymorphism (AFLP) and multilocus sequence typing (MLST)] methods. The MLST method developed here was based on the analysis of seven housekeeping genes (fusA, gyrB, hsp60, ileS, pyrG, recA and recG). The MLST method had good discriminatory ability: the 41 strains of L. delbrueckii examined were divided into 34 sequence types, with 29 sequence types represented by only a single strain. The sequence types were divided into eight groups. These groups could be discriminated as representing different subspecies. The results of the AFLP and MLST analyses were consistent. The type strain of L. delbrueckii subsp. delbrueckii, YIT 0080T, was clearly discriminated from the other strains currently classified as members of this subspecies, which were located close to strains of L. delbrueckii subsp. lactis. The MLST scheme developed in this study should be a useful tool for the identification of strains of L. delbrueckii to the subspecies level.

    • The GenBank/EMBL/DDBJ accession numbers for the gene sequences reported in this paper are AB568647–AB568687 (fusA), AB568688–AB568728 (gyrB), AB568811–AB568851 (hsp60), AB568606–AB568646 (ileS), AB568729–AB568769 (pyrG), AB568770–AB568810 (recA) and AB568852–AB568892 (recG).

    • A supplementary figure is available with the online version of this paper.

    Edited by: D. A. Mills

    INTRODUCTION

    Lactobacillus delbrueckii is a species of lactic acid bacteria that is important to the production of many fermented foods of both plant and animal origin. In 1983, on the basis of differences in phenotypic features, this species was classified into three subspecies: L. delbrueckii subsp. delbrueckii, L. delbrueckii subsp. bulgaricus and L. delbrueckii subsp. lactis (Weiss et al., 1983). Lactobacillus delbrueckii subsp. bulgaricus and L. delbrueckii subsp. lactis are usually found in dairy products and L. delbrueckii subsp. delbrueckii is usually found in fermented vegetables. In 2005, L. delbrueckii subsp. indicus was described as a novel subspecies of L. delbrueckii (Dellaglio et al., 2005). Lactobacillus delbrueckii subsp. indicus was isolated from Indian dairy products.

    The identification of L. delbrueckii strains is essential both for basic research and for the food industry. These subspecies have been classified by using phenotypic identification methods, especially the sugar fermentation test. However, phenotypic tests often give ambiguous results (Tanigawa et al., 2010). Many researchers have been trying to discriminate between these subspecies by using molecular methods. These methods have included the use of randomly amplified polymorphic DNA PCR (RAPD-PCR) and pepIQ gene sequences to discriminate L. delbrueckii subsp. bulgaricus from the other subspecies of L. delbrueckii (Torriani et al., 1999); SDS-PAGE profiles of cell-wall proteins to separate L. delbrueckii subsp. delbrueckii and L. delbrueckii subsp. bulgaricus (Gatti et al., 2001); restriction fragment length polymorphism (RFLP) of protein-coding genes to characterize and discriminate the strains of L. delbrueckii subsp. lactis from L. delbrueckii subsp. bulgaricus (Giraffa et al., 2003); analyses of 16S rRNA sequence mutations and the expression of β-galactosidase and cell-wall-anchored protease; and characterization of the lactose operon locus and the sequence of the lacR gene, galactose metabolism, and the distribution of insertion sequences (IS-elements) in L. delbrueckii subsp. delbrueckii, L. delbrueckii subsp. lactis and L. delbrueckii subsp. bulgaricus (Germond et al., 2003). Unfortunately, none of these methods has proved to be useful for discriminating L. delbrueckii subsp. delbrueckii from the other three subspecies.

    16S rRNA gene sequencing is the technique most commonly used for identification at the species level. At the subspecies level, however, 16S rRNA gene sequence similarity is often very high, and these sequences therefore cannot be used for identification purposes (Stackebrandt & Ebers, 2006).

    The amplified-fragment length polymorphism (AFLP) method was used to identify L. delbrueckii subsp. indicus in 2005 (Dellaglio et al., 2005). The AFLP method is known to be a highly sensitive and reproducible technique for bacterial genotyping (Savelkoul et al., 1999). In this study, we selected AFLP to classify the subspecies of L. delbrueckii due to the success of this technique in identifying L. delbrueckii subsp. indicus and also because AFLP is regarded as a reliable nucleic acid fingerprinting method (Tindall et al., 2010).

    More recently, multilocus sequence typing (MLST) has also been used for discriminating bacterial strains. MLST is generally regarded as a bacterial identification method that, depending on the genomic properties or experimental design (such as the kinds of genes analysed), provides higher resolution than RAPD or AFLP methods. MLST uses automated DNA sequencing to characterize the alleles present at different housekeeping gene loci. Since it is based on nucleotide sequences, it is highly discriminatory and provides unambiguous results that are directly comparable among laboratories. This method was first described in 1998 (Maiden et al., 1998) and there are now a number of publicly accessible MLST databases, such as PubMLST (). The MLST method has been used in a number of important studies on the identification of lactobacilli: Lactobacillus casei (Cai et al., 2007; Diancourt et al., 2007), Lactobacillus plantarum (de las Rivas et al., 2006) and Lactobacillus sanfranciscensis (Picozzi et al., 2010). In these studies, strains could be classified with high resolution by using MLST based on the sequences of the housekeeping genes ftsZ, gyrB, g6pd, ddl, dnaE, gdh, metRS, mutL, mutS, nrdD, pgm, polA, purK, rpoB, recP and tkt4.

    The aims of this study were to: (i) further re-examine the subspecies composition of L. delbrueckii; (ii) develop an MLST method for L. delbrueckii; and (iii) compare the discriminatory powers of AFLP and MLST.

    METHODS

    Bacterial strains and growth conditions.

    A total of 41 bacterial strains that had been putatively assigned as L. delbrueckii subsp. bulgaricus, L. delbrueckii subsp. delbrueckii, L. delbrueckii subsp. indicus, L. delbrueckii subsp. lactis or L. delbrueckii sp., or derived from sunki (a traditional Japanese non-salted pickle) in Japan or from ‘stinky tofu’ brines in Taiwan, were obtained from the Culture Collection of the Yakult Central Institute (YIT; Tokyo, Japan) (Table 1⇓). All bacterial strains were grown in MRS broth (BD Difco) at 37 °C for 24 h.

    Table 1.

    Bacterial strains used in this study

    −, Unknown.

    Sugar fermentation tests.

    The ability of the strains to ferment N-acetylglucosamine (GlcNAc), glucose, lactose, maltose, sucrose and trehalose was confirmed by acid formation. First, a 4 ml sample of each bacterial culture (24 h culture at 37 °C) was pelleted by centrifugation at 2000 g for 20 min. The cell pellet was then suspended in 2 ml saline and 130 μl of this bacterial suspension was inoculated into 4 ml test medium (modified MRS broth, with the glucose replaced by 2 % of the test substrate), supplemented with 0.005 % chlorophenol red as the pH indicator, and then incubated at 37 °C for 24 or 96 h. Acid production in the test medium was determined by a change in colour of the medium from red (pH 6.8) to yellow (pH 5.4).

    Genomic DNA extraction.

    The genomic DNAs of the isolates were extracted by a previously described method (Watanabe et al., 2008). Briefly, 1 ml stationary-phase bacterial culture was harvested by centrifugation at 20 000 g for 3 min. The cell pellet was suspended in 250 μl extraction buffer (100 mM Tris/HCl, 40 mM EDTA, pH 9.0) and 500 μl benzyl chloride; 0.7 g of glass beads (0.1 mm in diameter) was added to the suspension and the mixture was shaken vigorously for 30 s with a FastPrep FP120 homogenizer (Bio 101) at a speed setting of 6.5 m s–1. Subsequently, 50 μl 10 % SDS was added to the suspension, which was then vortexed vigorously at 50 °C for 20 min in a MicroIncubator M-36 (Taitech). The mixture was cooled on ice for 15 min after the addition of 150 μl 3 M sodium acetate. After centrifugation of the mixture at 20 000 g for 15 min, the supernatant was collected and DNA was obtained by 2-propanol precipitation. Finally, the DNA was diluted to 10 μg ml–1 in TE buffer (10 mM Tris/HCl, 1 mM EDTA, pH 8.0).

    AFLP analysis.

    AFLP analysis for the discrimination of subspecies was performed by using an AFLP EcoRI Ligation/Amplification Module (Applied Biosystems) and an AFLP MseI Ligation/Amplification Module (Applied Biosystems). Total DNA was digested with EcoRI and MseI restriction enzymes, and the DNA fragments were ligated to the following double-stranded restriction-site-specific adaptors, EcoRI adaptors and MseI adaptors. For preselective and selective PCR amplification, the primers EcoRI-A and MseI-CA were used. PCR products were analysed on an ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems) in standard fragment analysis mode and with a genescan-500 LIZ size standard (Applied Biosystems). After electrophoresis, the AFLP patterns were analysed and extracted with GeneMapper software v.4.0 (Applied Biosystems). A threshold fluorescence value of 100 arbitrary units was used to eliminate background fluorescence and DNA fragments of between 51 and 500 bp were analysed. Bands positioned at the same length (DNA size) in different strains were assumed to be similar and were classified as the same alleles. Bands of different sizes were treated as independent loci with two alleles (present or absent). Data were exported in binary format, with ‘1’ for the presence of a band/peak and ‘0’ for its absence; the data were analysed phylogenetically with the mega4 program (Tamura et al., 2007) by using an unweighted pair-group method with arithmetic mean (UPGMA) (Sneath & Sokal, 1973) and Pearson's correlation coefficient, with a bootstrap analysis of 100 repetitions.

    DNA amplification and sequencing for MLST.

    MLST analyses were performed with seven housekeeping genes: fusA, gyrB, hsp60, ileS, pyrG, recA and recG. For the strains examined, parts of these genes were amplified by using the primers listed in Table 2⇓. These primers were designed by using the consensus sequences of the seven housekeeping genes of L. delbrueckii subsp. bulgaricus ATCC 11842T (GenBank accession no. CR954253) and L. delbrueckii subsp. bulgaricus ATCC BAA-365 (NC008529). The PCR amplification program consisted of an initial denaturation step at 94 °C for 2 min; 30 cycles of 94 °C for 20 s, 55 °C for 20 s, and 72 °C for 20 s; and a final extension step at 72 °C for 3 min. The PCR-amplified housekeeping genes were purified with a Wizard SV Gel and PCR Clean-Up System (Promega) and then sequenced with an ABI PRISM BigDye Terminator v.3.1 cycle sequence kit (Applied Biosystems) on an ABI 3130xl Genetic Analyzer (Applied Biosystems).

    Table 2.

    Genes and primers used for MLST

    Data analysis for MLST.

    For each MLST locus, an allele number was given to each distinct combination of alleles of the seven genes by using BioNumerics v.6.0 (Applied-Maths). Each allele number was assigned a sequence type (ST). The same ST was used for several strains when they shared the same allelic profiles. The mean G+C content of the DNA, the number of polymorphic sites and the nucleotide diversity were calculated by using DnaSP v.5.1 (Librado & Rozas, 2009). This software was also used to calculate the ratio of the number of synonymous changes per synynomous site (dS) to the number of non-synonymous changes per non-synonymous site (dN; dS/dN ratio) and to perform Tajima's D test.

    Phylogenetic analyses.

    Cluster analysis based on AFLP profiles was conducted by using UPGMA (Sneath & Sokal, 1973) in the mega4 program (Tamura, et al., 2007). Minimum spanning tree analysis was performed with BioNumerics v.6.0.

    RESULTS

    A total of 41 strains assigned as L. delbrueckii were used (Table 1⇑). Twenty-four of the strains had been previously classified into one of the four subspecies: four strains were identified as L. delbrueckii subsp. delbrueckii, nine strains as L. delbrueckii subsp. lactis, seven strains as L. delbrueckii subsp. bulgaricus and four strains as L. delbrueckii subsp. indicus. The other 17 strains were our own original strains, which were isolated mainly from sunki, a traditional Japanese pickle product, in Japan in 2004 (Watanabe et al., 2009) or from fermented stinky tofu brines in Taiwan in 2005 (Chao et al., 2008).

    Phenotypic identification

    Fermentation tests were performed with GlcNAc, glucose, lactose, maltose, sucrose and trehalose. The presence of chlorophenol red caused the colour of the medium to change from red to yellow as the pH changed from 6.8 to 5.4. All strains fermented glucose (data not shown); the other results of the fermentation tests are shown in Table 3⇓. The delineation of subspecies of L. delbrueckii based on the sugar fermentation profile has been reported previously (Dellaglio et al., 2005). The inability to metabolize lactose distinguishes L. delbrueckii subsp. delbrueckii from the other subspecies of L. delbrueckii. With the exception of L. delbrueckii subsp. delbrueckii, all strains that can ferment both maltose and trehalose are classified as L. delbrueckii subsp. lactis. In contrast, strains that can ferment GlcNAc but not sucrose are classified as L. delbrueckii subsp. bulgaricus. According to this definition, when the fermentation results from the 96 h cultures were used, only two of the four strains currently classified as L. delbrueckii subsp. delbrueckii showed the expected phenotypic characteristics of this subspecies and 12 of the unclassified isolates shared the same characteristics. Eight of the nine strains currently classified as L. delbrueckii subsp. lactis showed the phenotypic characteristics of this subspecies, two strains currently classified as L. delbrueckii subsp. delbrueckii had the phenotypic characteristics of L. delbrueckii subsp. lactis and four unclassified isolates shared the same characteristics. Five of the seven strains currently classified as L. delbrueckii subsp. bulgaricus showed the phenotypic characteristics expected for this subspecies and one strain of L. delbrueckii subsp. lactis had the same characteristics as L. delbrueckii subsp. bulgaricus. Three of the four strains currently classified as L. delbrueckii subsp. indicus had the phenotypic characteristics of this subspecies. The remaining three strains could not be identified as any known subspecies. Acid formation from lactose, maltose and GlcNAc was strain-dependent and the results varied depending on whether strains were cultured for 24 or 96 h. For example, strain YIT 0020 gave a negative result for fermentation of maltose after 24 h of culture but the result was positive after 96 h.

    Table 3.

    Typing of 41 strains by MLST, AFLP and sugar fermentation test results

    PT, Phenotypic tests, B, L. delbrueckii subsp. bulgaricus; D, L. delbrueckii subsp. delbrueckii; I, L. delbrueckii subsp. indicus; L, L. delbrueckii subsp. lactis; nd, not determined; G, group in minimum spanning tree of Fig. 2.

    AFLP analysis of 41 L. delbrueckii strains

    The AFLP method is useful for grouping bacterial strains according to their genetic relationships. The AFLP method was used to measure the extent of subspecies diversity, and a dendrogram was constructed based on the AFLP profiles of the strains examined (Fig. 1⇓). The 41 L. delbrueckii strains were separated into four major clusters (A, B, C and D). Cluster A consisted of three strains of L. delbrueckii subsp. delbrueckii, nine strains of L. delbrueckii subsp. lactis and 17 strains of L. delbrueckii sp. Cluster B comprised only a single strain (YIT 0080T), cluster C consisted of seven strains of L. delbrueckii subsp. bulgaricus and cluster D consisted of four strains of L. delbrueckii subsp. indicus. Clusters A, B, C and D were identified in accordance with the locations of the type strains as L. delbrueckii subsp. lactis, L. delbrueckii subsp. delbrueckii, L. delbrueckii subsp. bulgaricus and L. delbrueckii subsp. indicus, respectively.

    Figure image not available in archive
    Fig. 1.

    Dendrogram derived from the AFLP band pattern. Bootstrap values based on 100 replications are given at nodes. Bar, number of band mismatches.

    Cluster A was divided into four subclusters. Cluster A-1 included eight strains of L. delbrueckii subsp. lactis and one strain of L. delbrueckii sp.; cluster A-2 included one strain of L. delbrueckii subsp. delbrueckii, one strain of L. delbrueckii sp. and four strains of L. delbrueckii sp. isolated from sunki. Cluster A-3 included 12 strains of L. delbrueckii subsp. lactis and L. delbrueckii sp. and cluster A-4 included two strains of L. delbrueckii subsp. delbrueckii.

    Allelic and nucleotide diversity using MLST

    MLST analysis was used to examine the genetic diversity of L. delbrueckii and to discriminate subspecies. The sequences of the seven loci were determined for the 41 study strains. From the 41 L. delbrueckii strains, 34 different STs were obtained. Of these, 29 STs were assigned to single strains, three STs (ST8, ST10 and ST28) were assigned to two strains and two STs (ST7 and ST23) were assigned to three strains. Consensus sequence templates ranged in length from 380 bp (gyrB) to 515 bp (hsp60). Polymorphic sites are shown in Supplementary Fig. S1 (available in Microbiology Online). Non-synonymous substitutions per non-synonymous site were relatively rare compared with synonymous changes per synonymous site (Table 4⇓), indicating selection against amino acid changes and excluding strong positive selection on the observed allelic diversity, as is typically observed for housekeeping genes. Tajima's D values did not deviate significantly from zero, supporting a neutral selection of the alleles of the seven housekeeping genes. The DNA G+C content observed for all alleles of the seven genes ranged from 49.1 mol% (fusA) to 60.9 mol% (recA), which was a little higher than that of the complete genomes of L. delbrueckii subsp. bulgaricus ATCC 11842T and L. delbrueckii subsp. bulgaricus ATCC BAA-365 (49.0 mol%).

    Table 4.

    Genetic variability at L. delbrueckii loci

    dS, Number of synonymous changes per synonymous site; dN, number of non-synonymous changes per non-synonymous site.

    MLST for subspeciation of L. delbrueckii

    The number of alleles per locus ranged from 12 (fusA) to 22 (recG). By combining the seven gene loci, 34 STs were distinguished. To explore the relationships among the 41 strains, phylogenetic analysis based on the allelic profiles was performed by using the minimum spanning tree algorithm (Fig. 2⇓). When the distance between different STs was 6 or more allelic differences, the strains that stepped between these STs were bundled as different groups. The 41 strains of L. delbrueckii were thus divided into eight groups: group 1 (ST1), group 2 (ST2, ST3, ST4, ST5, ST6, ST7, ST8, ST17, ST18, ST23, ST24, ST25, ST26, ST29, ST31, ST32, ST34), Group 3 (ST9, ST10, ST12, ST13, ST14, ST15), Group 4 (ST16, ST20, ST21, ST22), Group 5 (ST19, ST28, ST33), Group 6 (ST30), Group 7 (ST11), and Group 8 (ST27).

    Figure image not available in archive
    Fig. 2.

    Minimum spanning tree of 41 strains of L. delbrueckii based on sequence type (ST). Each circle corresponds to the ST, and the circle size denotes the number of strains sharing the same ST. To illustrate the concordance of the MLST data with AFLP data, circles are coloured on the basis of AFLP cluster type: yellow, cluster A-1; pink, cluster A-2; blue, cluster A-3; green, cluster A-4; red, cluster B; purple, cluster C; sky blue, cluster D. Numbers between the circles indicate the number of allelic differences between the profiles. Coloured zones between some groups of circles indicate that these profiles belong to the same potential subspecies. The double line separates the strains into two groups on the basis of their ability to ferment lactose.

    We also performed phylogenetic analysis based on the position of mutations in the composite sequences that were constructed by concatenating the seven housekeeping genes (Fig. 3⇓). The 41 strains were separated into five clusters: cluster 1 consisted of three strains of L. delbrueckii subsp. delbrueckii, nine strains of L. delbrueckii subsp. lactis and 12 strains of L. delbrueckii sp. Cluster 2 comprised only a single strain (YIT 0080T) and cluster 3 consisted of five strains of L. delbrueckii sp. Cluster 4 consisted of seven strains of L. delbrueckii subsp. bulgaricus and cluster 5 consisted of four strains of L. delbrueckii subsp. indicus. Cluster 1 was divided into five subclusters: cluster 1a included two strains of L. delbrueckii subsp. delbrueckii, seven strains of L. delbrueckii subsp. lactis and two strains of L. delbrueckii sp.; cluster 1b included seven strains of L. delbrueckii sp.; cluster 1c included one strain of L. delbrueckii subsp. delbrueckii and two strains of L. delbrueckii subsp. lactis; cluster 1d included two strains of L. delbrueckii sp.; and cluster 1e comprised a single strain of L. delbrueckii sp. Cluster 3 was also divided into two subclusters: cluster 3a included four strains of L. delbrueckii sp. all derived from sunki; and cluster 3b comprised a single strain of L. delbrueckii sp.

    Figure image not available in archive
    Fig. 3.

    Dendrogram derived from the similarity of the seven housekeeping gene sequences. Bar, percentage sequence similarity.

    DISCUSSION

    Various methods for discriminating subspecies of L. delbrueckii have been proposed by many researchers. However, several research groups have recognized that the identifications based on phenotypic and genotypic methods do not match (Giraffa et al., 2004; Miteva et al., 2001). Furthermore, although we had attempted to identify L. delbrueckii subspecies by using single housekeeping genes in the past, the resulting subspeciation was of low resolution (data not shown).

    Lactobacillus delbrueckii subsp. indicus was first established as a novel subspecies in 2005 by using the AFLP method (Vos et al., 1995). The AFLP method is based on calculating the differences in the length of restriction-enzyme-digested genomic DNA, and it is known to be a reproducible method. However, the use of housekeeping gene sequences for the identification of bacteria is often more sensitive and more accurate. AFLP and MLST techniques target different aspects of the genome. MLST targets housekeeping genes, which evolve slowly, whereas AFLP targets large genome rearrangements. In addition, the principle of clustering in these methods is also different. AFLP compares the lengths of fragments of chromosomal DNA and MLST compares the base compositions of gene sequences. Therefore, we developed an MLST method based on seven housekeeping gene sequences (fusA, gyrB, hsp60, ileS, pyrG, recA and recG) to classify 41 strains of L. delbrueckii. The utility of MLST for the analysis of the genetic structure of bacterial populations is based mainly on the ability of housekeeping genes to have selectively neutral variability. Analysis of synonymous and non-synonymous changes in the allele sequences of a locus can be used to determine whether they are subject to positive selection; therefore, a dN/dS ratio >1 implies selection for amino acid changes. In our analysis of the divergences in the seven housekeeping genes, the dN/dS ratios were much lower than 1. The nucleotide diversity ranged from 0.0051 to 0.0096. By comparison, nucleotide diversity values obtained for Lactobacillus casei ranged from 0.0002 to 0.0076 (Diancourt et al., 2007) and nucleotide diversity values for Lactobacillus plantarum ranged from 0.0004 to 0.0072 (de las Rivas et al., 2006). The nucleotide diversity values of L. delbrueckii were a little higher than those of these two species.

    The three methods were compared, as shown in Table 3⇑. These subspecies were determined by the dendrogram based on AFLP (Fig. 1⇑) and the minimum spanning tree based on MLST (Fig. 2⇑) to be in accordance with the locations of the type strains. For the phenotypic tests, subspecies were determined by using previously reported sugar fermentation profiles (Dellaglio et al., 2005). In spite of the MLST and AFLP methods utilizing different aspects of the genome, subspecies identification by the two methods was in good concordance, with the exception of groups 5–8 in the MLST analysis. On the other hand, the phenotypic tests showed results for some strains that were discordant with the results from the other two methods. All the strains located in the four disparate groups (groups 5–8) by MLST were identified as L. delbrueckii subsp. lactis by AFLP and as L. delbrueckii subsp. delbrueckii by means of phenotypic tests. MLST analysis suggested that the strains in these four disparate groups (groups 5–8) did not belong to either L. delbrueckii subsp. delbrueckii or L. delbrueckii subsp. lactis. This finding indicated that these strains could represent a novel subspecies and that MLST has a higher resolution than AFLP for the subspeciation of L. delbrueckii.

    We confirmed that the results of the MLST method corresponded well to the results of the AFLP method with respect to subspecies identification of the 41 L. delbrueckii strains. By the AFLP method, the 41 strains of L. delbrueckii were divided into four clusters. The type strain of L. delbrueckii subsp. delbrueckii (YIT 0080T) was the single strain located in independent cluster B, whereas the other strains of L. delbrueckii subsp. delbrueckii were located in cluster A, which also included the type strain of L. delbrueckii subsp. lactis (YIT 0086T) and the other strains of L. delbrueckii subsp. lactis. By the MLST method, we also confirmed that strain YIT 0080T (group 1) was clearly separated from the other strains in this species (group 2). To the best of our knowledge, this is the first study to demonstrate the ability of the MLST technique to classify L. delbrueckii species and subspecies.

    MLST has a powerful ability to discriminate at strain level. This method is often used in the discrimination of strains by their origins. However, we could not find any association between a clonal complex (a cluster of linked STs; shown by groups in Fig. 3⇑) and the source of the strains included within it. The strains in our original culture collection of L. delbrueckii sp. derived from sunki and stinky tofu brines were not bundled in the same clonal complexes. However, four strains of the L. delbrueckii sp. isolated from sunki, strains YIT 11220, YIT 11221, YIT 11466 and YIT 11673, were bundled in the same clonal complex (group 5), which corresponded to subcluster A-2 in the AFLP dendrogram. Additionally, we found that there was a relationship between lactose utilization and the patterns of grouping in the minimum spanning tree. Group 2 was divided into two subgroups on the basis of the ability of the strains to utilize lactose: Group 2a, which comprised 13 lactose-fermenting strains and ten STs (ST2, ST3, ST4, ST5, ST6, ST7, ST8, ST17, ST18 and ST26), and Group 2b which comprised nine non-lactose-fermenting strains and seven STs (ST23, ST24, ST25, ST29, ST31, ST32 and ST34) (Table 3⇑).

    This study also confirmed that the phenotypic features based on sugar fermentation tests were not useful for the identification of subspecies because they varied with culture time and were sometimes ambiguous. In addition, phenotypic markers can often provide discordant results, as previously reported (Tanigawa et al., 2010). However, the phenotypic tests did reflect the strain's origin in some aspects. The 20 strains derived from milk products (seven strains of L. delbrueckii subsp. bulgaricus, nine strains of L. delbrueckii subsp. lactis and four strains of L. delbrueckii subsp. indicus), could all ferment lactose. Two strains of L. delbrueckii subsp. delbrueckii (YIT 12133, YIT 12134) and two L. delbrueckii sp. strains could also ferment lactose. In contrast, the L. delbrueckii sp. strains isolated from fermented vegetables, fermented brines for making stinky tofu, and human faeces were non-lactose-fermenting strains (except strains YIT 0057 and YIT 11501). These 15 non-lactose-fermenting strains were located in three clusters, cluster 1b, cluster 1d and cluster 3, in the MLST dendrogram (Fig. 3⇑) and in the disparate groups in the minimum spanning tree based on MLST (Fig. 2⇑), group 2b, group 5, group 6, group 7 and group 8. The two strains of L. delbrueckii subsp. delbrueckii were exceptions: strains YIT 0080T and YIT 11851 were located in cluster 2 and cluster 1c (Fig. 3⇑), and in group 1 and group 2b (Fig. 2⇑), respectively.

    The MLST method also had a higher resolution than the ALFP method with respect to the discrimination of the non-lactose-fermenting strains identified as L. delbrueckii subsp. lactis. By using the minimum spanning tree method, group 2 was divided into two subgroups: Group 2a consisting of the lactose-fermenting strains, which could be classified as L. delbrueckii subsp. lactis, and Group 2b consisting of the non-lactose-fermenting strains (Table 3⇑).

    From the evolutionary evidence revealed by these genotypic approaches, we confirmed that it was impossible to discriminate the strains of L. delbrueckii subsp. delbrueckii (except the type strain YIT 0080T) from the strains of L. delbrueckii subsp. lactis because of their close relationship. In addition, eight independent groups were observed by MLST, which could be discriminated at the subspecies level. The four L. delbrueckii sp. strains derived from sunki could represent a novel subspecies of L. delbrueckii.

    Therefore, we propose that L. delbrueckii subsp. delbrueckii YIT 0080T should be the sole strain from among the 41 strains examined here that should be classified as L. delbrueckii subsp. delbrueckii. The other strains of L. delbrueckii subsp. delbrueckii used in this study, YIT 11851 (=NCIMB 701744), YIT 12133 (=LMG 22235) and YIT 12134 (=LMG 22236), should be reclassified as strains of L. delbrueckii subsp. lactis.

    The MLST scheme proposed in this study could be a useful tool for subspeciation of strains in L. delbrueckii.

    References

    1. Cai, H., Rodríguez, B. T., Zhang, W., Broadbent, J. R. & Steele, J. L. (2007). Genotypic and phenotypic characterization of Lactobacillus casei strains isolated from different ecological niches suggests frequent recombination and niche specificity. Microbiology 153, 2655–2665.
    2. Chao, S.-H., Tomii, Y., Watanabe, K. & Tsai, Y.-C. (2008). Diversity of lactic acid bacteria in fermented brines used to make stinky tofu. Int J Food Microbiol 123, 134–141.
    3. de las Rivas, B., Marcobal, A. & Muñoz, R. (2006). Development of a multilocus sequence typing method for analysis of Lactobacillus plantarum strains. Microbiology 152, 85–93.
    4. Dellaglio, F., Felis, G. E., Castioni, A., Torriani, S. & Germond, J. E. (2005). Lactobacillus delbrueckii subsp. indicus subsp. nov., isolated from Indian dairy products. Int J Syst Evol Microbiol 55, 401–404.
    5. Diancourt, L., Passet, V., Chervaux, C., Garault, P., Smokvina, T. & Brisse, S. (2007). Multilocus sequence typing of Lactobacillus casei reveals a clonal population structure with low levels of homologous recombination. Appl Environ Microbiol 73, 6601–6611.
    6. Gatti, M., Fornasari, M. E. & Neviani, E. (2001). Differentiation of Lactobacillus delbrueckii subsp. bulgaricus and Lactobacillus delbrueckii subsp. lactis by SDS-PAGE of cell-wall proteins. Lett Appl Microbiol 32, 352–356.
    7. Germond, J. E., Lapierre, L., Delley, M., Mollet, B., Felis, G. E. & Dellaglio, F. (2003). Evolution of the bacterial species Lactobacillus delbrueckii: a partial genomic study with reflections on prokaryotic species concept. Mol Biol Evol 20, 93–104.
    8. Giraffa, G., Lazzi, C., Gatti, M., Rossetti, L., Mora, D. & Neviani, E. (2003). Molecular typing of Lactobacillus delbrueckii of dairy origin by PCR-RFLP of protein-coding genes. Int J Food Microbiol 82, 163–172.
    9. Giraffa, G., Andrighetto, C., Antonello, C., Gatti, M., Lazzi, C., Marcazzan, G., Lombardi, A. & Neviani, E. (2004). Genotypic and phenotypic diversity of Lactobacillus delbrueckii subsp. lactis strains of dairy origin. Int J Food Microbiol 91, 129–139.
    10. Librado, P. & Rozas, J. (2009). DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25, 1451–1452.
    11. Maiden, M. C., Bygraves, J. A., Feil, E., Morelli, G., Russell, J. E., Urwin, R., Zhang, Q., Zhou, J., Zurth, K. & other authors (1998). Multilocus sequence typing: a portable approach to the identification of clones within populations of pathogenic microorganisms. Proc Natl Acad Sci U S A 95, 3140–3145.
    12. Miteva, V., Boudakov, I., Ivanova-Stoyancheva, G., Marinova, B., Mitev, V. & Mengaud, J. (2001). Differentiation of Lactobacillus delbrueckii subspecies by ribotyping and amplified ribosomal DNA restriction analysis (ARDRA). J Appl Microbiol 90, 909–918.
    13. Picozzi, C., Bonacina, G., Vigentini, I. & Foschino, R. (2010). Genetic diversity in Italian Lactobacillus sanfranciscensis strains assessed by multilocus sequence typing and pulsed-field gel electrophoresis analyses. Microbiology 156, 2035–2045.
    14. Savelkoul, P. H. M., Aarts, H. J. M., de Haas, J., Dijkshoorn, L., Duim, B., Otsen, M., Rademaker, J. L., Schouls, L. & Lenstra, J. A. (1999). Amplified-fragment length polymorphism analysis: the state of an art. J Clin Microbiol 37, 3083–3091.
    15. Sneath, P. H. A. & Sokal, R. R. (1973). Numerical Taxonomy. San Francisco, CA. : W. H. Freeman & Company.
    16. Stackebrandt, E. & Ebers, J. (2006). Taxonomic parameters revisited: tarnished gold standards. Microbiol Today 33, 152–155.
    17. Tamura, K., Dudley, J., Nei, M. & Kumar, S. (2007). mega4: molecular evolutionary genetics analysis (mega) software version 4.0. Mol Biol Evol 24, 1596–1599.
    18. Tanigawa, K., Kawabata, H. & Watanabe, K. (2010). Identification and typing of Lactococcus lactis by matrix-assisted laser desorption ionization-time of flight mass spectrometry. Appl Environ Microbiol 76, 4055–4062.
    19. Tindall, B. J., Rosselló-Móra, R., Busse, H. J., Ludwig, W. & Kämpfer, P. (2010). Notes on the characterization of prokaryote strains for taxonomic purposes. Int J Syst Evol Microbiol 60, 249–266.
    20. Torriani, S., Zapparoli, G. & Dellaglio, F. (1999). Use of PCR-based methods for rapid differentiation of Lactobacillus delbrueckii subsp. bulgaricus and L. delbrueckii subsp. lactis. Appl Environ Microbiol 65, 4351–4356.
    21. Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J. P., Peleman, J. & other authors (1995). AFLP: a new technique for DNA fingerprinting. Nucleic Acids Res 23, 4407–4414.
    22. Watanabe, K., Fujimoto, J., Sasamoto, M., Dugersuren, J., Tumursuh, T. & Demberel, S. (2008). Diversity of lactic acid bacteria and yeasts in Airag and Tarag, traditional fermented milk products of Mongolia. World J Microbiol Biotechnol 24, 1313–1325.
    23. Watanabe, K., Fujimoto, J., Tomii, Y., Sasamoto, M., Makino, H., Kudo, Y. & Okada, S. (2009). Lactobacillus kisonensis sp. nov., Lactobacillus otakiensis sp. nov., Lactobacillus rapi sp. nov. and Lactobacillus sunkii sp. nov., heterofermentative species isolated from sunki, a traditional Japanese pickle. Int J Syst Evol Microbiol 59, 754–760.
    24. Weiss, N., Schillinger, U. & Kandler, O. (1983). Lactobacillus lactis, Lactobacillus leichmannii and Lactobacillus bulgaricus, subjective synonyms of Lactobacillus delbrueckii, and description of Lactobacillus delbrueckii subsp. lactis comb. nov. and Lactobacillus delbrueckii subsp. bulgaricus comb. nov. Syst Appl Microbiol 4, 552–557.