|Home | About | Journals | Submit | Contact Us | Français|
A large collection of Lactococcus lactis strains, including wild-type isolates and dairy starter cultures, were screened on the basis of their phenotype and the macrorestriction patterns produced from pulsed-field gel electrophoresis (PFGE) analysis of SmaI digests of genomic DNA. Three groups of dairy starter cultures, used for different purposes in the dairy industry, and a fourth group made up of strains isolated from the environment were selected for analysis of their chromosomal diversity using the endonuclease I-CeuI. Chromosome architecture was largely conserved with each strain having six copies of the rRNA genes, and the chromosome size of individual strains ranged between 2,240 and 2,688 kb. The origin of L. lactis strains showed the greatest correlation with chromosome size, and dairy strains, particularly those with the cremoris phenotype, had smaller chromosomes than wild-type strains. Overall, this study, coupled with analysis of the sequenced L. lactis genomes, provides evidence that defined strain dairy starter cultures have arisen from plant L. lactis strains. Adaptation of these strains to the dairy environment has involved loss of functions resulting in smaller chromosomes and acquisition of genes (usually plasmid associated) that facilitate growth in milk. We conclude that dairy starter cultures generally and the industrially used cremoris and diacetylactis phenotype strains in particular comprise a specialized group of L. lactis strains that have been selected to become an essential component of industrial processes and have evolved accordingly, so that they are no longer fit to survive outside the dairy environment.
Lactococcus lactis can be isolated from various environments but is predominantly studied because of its role as the main constituent of many industrial and artisanal starter cultures used for the manufacture of a vast range of fermented dairy products including fermented milks, sour cream, soft and hard cheeses, and lactic casein (Ward et al. 2002). For large-scale commercial production, the starter cultures used are commonly defined strains, which have been selected for their desirable properties especially in relation to acid production, flavor development, and bacteriophage resistance (Limsowtin et al. 1996).
The taxonomy of L. lactis has changed many times but is currently phenotypically based (Schleifer et al. 1985; van Hylckama Vlieg et al. 2006; Rademaker et al. 2007) and includes two subspecies (subsp. lactis and subsp. cremoris) and one biovar (subsp. lactis biovar diacetylactis). The lactis and cremoris phenotypes are differentiated on the basis of arginine utilization, growth temperature, and salt tolerance, whereas the biovar diacetylactis strains have the additional ability to metabolize citrate. Numerous studies including DNA–DNA hybridization, 16S rRNA, and gene sequence analysis have demonstrated the existence of two main genotypes. These two genotypic groups have also been called L. lactis subsp. lactis and L. lactis subsp. cremoris, but unfortunately the genotype and phenotype designations do not necessarily correspond, thus introducing a degree of confusion into the taxonomy of this species (Tailliez et al. 1998). An extensive study of 102 L. lactis isolates of dairy and plant origin using various genomic fingerprinting methods and multilocus sequence analysis has clearly demonstrated that two major lineages exist (Rademaker et al. 2007). One of these comprises those strains with a L. lactis subsp. cremoris genotype and includes strains with both lactis and cremoris phenotypes. The other comprises those strains with a L. lactis subsp. lactis genotype that includes strains with the lactis phenotype as well as biovar diacetylactis. Comparative genome hybridization (CGH) using 39 L. lactis strains of dairy or plant origin (Bayjanov et al. 2009) provides further evidence confirming the unusual taxonomic structure in this species. As a result, it is necessary to specify a genotype (cremoris or lactis) and a phenotype (cremoris, diacetylactis, or lactis) to adequately describe individual strains.
Strains that show both the subsp. cremoris genotype and phenotype cluster closely together and form a definite subgroup that shows limited diversity relative to the other L. lactis strains examined (Rademaker et al. 2007; Taïbi et al. 2010). These L. lactis subsp. cremoris strains are favored for use as defined strain starter cultures for Cheddar cheese production because they are less likely to cause bitterness and other flavor defects (Heap 1998). The citrate-metabolizing biovar diacetylactis strains contribute to the flavor and aroma profile of a range of fermented dairy products and are also a component of the starter blends used for lactic casein manufacture (Heap and Lawrence 1984). These strains have long been distinguished taxonomically (Kempler and McKay 1981), but with the description of the genus Lactococcus (Schleifer et al. 1985), they were incorporated into L. lactis subsp. lactis. Biovar diacetylactis dairy starter strains have been genotypically (Köhler et al. 1991; Beimfohr et al. 1997) and phenotypically (Bachmann et al. 2009) distinguished from other L. lactis cultures, suggesting that these cultures may also form a separate subgroup.
Both L. lactis subspecies have been isolated from a variety of environmental sources but are most commonly associated with fresh or fermented plant material or with milk and milk products. Strains that show the lactis subspecies genotype can be readily isolated from these environments, whereas isolations of cultures with the cremoris subspecies genotype are comparatively rare (Klijn et al. 1995; Salama et al. 1995). Attempts to isolate new cremoris or diacetylactis phenotype strains from environmental sources have met with little success as wild-type strains of both subspecies show the lactis phenotype (Klijn et al. 1995; Salama et al. 1995; Ward et al. 1998).
Because of its industrial relevance, L. lactis has become the best studied of the lactic acid bacteria and regarded as a model organism for this bacterial group, although most work has been focused on a small number of laboratory strains of dairy origin. Complete genome sequences have been published for four strains. These include the two plasmid-cured strains (IL1403 and MG1363) on which much of the detailed biochemical and genetic knowledge of L. lactis is based (Bolotin et al. 2001; Wegmann et al. 2007). Both IL1403 and MG1363 belong to L. lactis subsp. lactis phenotypically, but the parent strain of IL1403 (CNRZ157) has a citrate permease plasmid and is able to metabolize citrate placing it with L. lactis subsp. lactis biovar diacetylactis, whereas MG1363 has a lactis phenotype and a cremoris genotype. The third genome-sequenced strain (SK11) has been used as a cheese starter culture and belongs to the subgroup of strains with both the subsp. cremoris genotype and phenotype (Makarova et al. 2006). The fourth genome is from a L. lactis subsp. lactis strain of plant origin (KF147), and a partial sequence is also available for a second plant strain (KF282) (Siezen et al. 2008, 2010). Comparison of the genomes from plant and dairy isolates has highlighted the differences in gene content that can occur between individual strains in the same species (Siezen et al. 2008) and shows that sequencing of one representative genome does not give a complete picture of the genetic repertoire of a species. Attempts to describe this intraspecies diversity have led to the terms species genome (Lan and Reeves 2000) and pangenome (Medini et al. 2005) being defined to cover all the genes present in the characterized strains of a species. Under both definitions, the genome has a core of genes responsible for the basic aspects of the biology of the species and a set of auxiliary or dispensable genes that contribute to species diversity and may provide a selective advantage in certain environments. As a result, there is much to learn about diversity at the intraspecies level, and the aim of this work was to examine the chromosomal diversity of a large collection of L. lactis strains to provide a framework for future comparative genomic work with this industrially important bacterial species.
In initial screening, 558 L. lactis strains were examined by pulsed-field gel electrophoresis (PFGE) analysis of SmaI digests of genomic DNA. These included 289 strains with the cremoris phenotype and genotype (L. lactis subsp. cremoris), 197 strains with the lactis phenotype (L. lactis subsp. lactis), and 72 strains that were able to metabolize citrate (L. lactis subsp. lactis biovar diacetylactis). The cremoris phenotype strains and the citrate-metabolizing strains were all either used as or isolated from dairy starter cultures. The lactis phenotype strains included cultures with both cremoris and lactis genotypes and came from diverse origins. These included dairy starter cultures and individual strains isolated from raw milk, pasture, soil, plant material, the rumen, and insect gut. Some of these have been described elsewhere (table 1), and the strains isolated from plant sources were included in the L. lactis diversity study (Rademaker et al. 2007). Forty-eight of the lactis phenotype strains were nisin producers. Dairy starter strains were mainly from the culture collection of the Fonterra Research Center, Palmerston North, New Zealand, with additional cultures obtained from other culture collections or isolated from mixed strain dairy starters.
Lactococci were grown at 28 °C in M17 broth (Merck) (Terzaghi and Sandine 1975) supplemented with 0.5% w/v glucose for growth of the plasmid-free dairy strains (IL1403 and MG1363) and the wild-type strains. The stock cultures were maintained at −85 °C in M17 broth supplemented with 20% (v/v) glycerol. Tests for arginine and citrate metabolism were used to confirm the phenotype of the strains, and strain genotypes were determined using the polymerase chain reaction (PCR) primers and conditions described previously (Ward et al. 1998).
Cells were harvested from 1.5 ml of an overnight culture by centrifugation (10,000 × g, 10 min), washed twice with 1 M NaCl:10 mM Tris–Cl (pH 7.6), and 300 μl aliquots were mixed with an equal volume of 2% (w/v) low melt agarose (Bio-Rad Laboratories). Embedded cells were lysed by treatment with lysozyme (1 mg/ml in EC buffer, 6 mM Tris–Cl:1 M NaCl:100 mM ethylenediaminetetraacetic acid [EDTA]:1% [w/v] sarkosyl, pH 7.6) overnight at 37 °C and proteinase K (0.5 mg/ml in lysis buffer, 50 mM Tris–Cl:50 mM EDTA:1% [w/v] sarkosyl, pH 8.0) for 24 h at 50 °C. Agarose plugs containing intact genomic DNA were washed three times with Tris–EDTA buffer (10 mM Tris–Cl:1 mM EDTA, pH 8.0) before storage in 10 mM Tris–Cl:100 mM EDTA (pH 8.0) at 4 °C. DNA embedded in agarose was digested for 16 h with 1.0 U of ApaI, SmaI, or I-CeuI (New England Biolabs) in 100 μl of restriction enzyme buffer, loaded into wells of 1% (w/v) agarose gels (pulsed-field certified agarose, Bio-Rad), and run at 200 V for 20 h at 14 °C in 0.5× Tris–borate buffer (Sambrook et al. 1989) using a CHEF DR III PFGE apparatus and model 1000 mini chiller (Bio-Rad). Pulse times used were 1–30 s for ApaI or SmaI and 5–60 s for I-CeuI. To determine the size of the largest fragments from I-CeuI digests, gels were prepared from 0.8% chromosomal grade agarose (Bio-Rad) and run with the pulse time ramped between 150 and 400 s. Fragments smaller than 100 kb were also resolved and measured using a FIGE Mapper electrophoresis system (Bio-Rad). DNA was visualized by staining with ethidium bromide and the image captured using a Gel Doc 1000 system (Eastman Kodak). Partial digestion with I-CeuI was used to establish the rrn chromosomal skeleton as described by Liu et al. (1999). Genomic DNA prepared from L. lactis subsp. lactis IL1403 and phage lambda concatamers or Saccharomyces cerevisiae pulsed-field gel (PFG) markers (New England Biolabs) were used as size standards.
Plasmid DNA was isolated by the method of Anderson and McKay (1983) and the size of individual plasmid bands determined following electrophoresis in 0.7% agarose gels in Tris–acetate buffer (Sambrook et al. 1989) for 3 h at 4 V/cm and staining as above. Strains were also screened for the presence of large linear plasmids by running undigested genomic DNA in PFGs as described above.
To determine the relatedness of citrate-metabolizing strains based on the presence and chromosomal location of their prophage, the primers described by Chopin et al. (2001) were used in PCR reactions to amplify the chromosome–prophage junction regions from IL1403 for use as probes for PFGs. Hybridizations were done using the North2South Direct HRP Labeling and Detection Kit (Pierce) using the conditions recommended by the manufacturer. To determine the presence of plasmids encoding citrate permease (citP), plasmid gels were hybridized with a probe to the gene for citrate permease that was constructed by PCR amplification from strain LW1503 using primers described by Klijn et al. (1995). To determine if cultures, which failed to cut with SmaI, contained the ScrFI R/M system, PFGs were hybridized with a probe specific for the scrFIAM methylase, which was constructed by PCR amplification from strain UC503 using primers described by Szatmari et al. (2006).
Differences in average chromosome size among subspecific groups of L. lactis from each origin (dairy and wild type) were tested using a one-way analysis of variance and fitting a single factor comprising each genotype–phenotype origin combination. To test whether regions of the chromosome differ in degree of variability, we compared the variances in the lengths of different chromosomal regions based on I-CeuI fragments. The variance in the length of a given region is expected to increase linearly with fragment size assuming that the number of insertions, duplications, and deletions per fragment increase with fragment length. Therefore, the variances in the sizes of the I-CeuI fragments were standardized by dividing the variance of a given fragment by its average size before subjecting the values to pairwise F tests to test the equality of variances (Sokal and Rohlf 1981).
Genomic data for the four publicly available and completed L. lactis genome sequences (GenBank accession numbers AM406671, CP000425, AE005176, and CP001834) were downloaded from the National Center for Biotechnology Information (NCBI) Web site (http://www.ncbi.nlm.nih.gov). Genome alignments of chromosomal sequences were performed using Mauve software (Version 2.3.1) (Darling et al. 2004).
Phylogenetic relationships were determined using a supertree approach. Protein coding gene sets for the four L. lactis strains, three Streptococcus thermophilus strains (GenBank accession numbers CP000023, CP000024, and CP000419), five Lactobacillus species (GenBank accession numbers AL935263, CR936503, CR954253, CP000416, and CP000517), as well as two outgroups (Listeria species; GenBank accession numbers AL591824 and AL592022) were downloaded from the NCBI Web site. The S. thermophilus, L. delbrueckii subsp. bulgaricus and L. helveticus genomes were included for comparison because of their use as dairy starter cultures. For a full description of the supertree methodology used please refer to Fitzpatrick et al. (2006).
PFGE analysis of SmaI digests of genomic DNA from the 289 L. lactis subsp. cremoris strains showed that 230 (80%) could be linked into 12 groups of related strains. Representatives of these groups are included in table 1. Three (E8, HP, and SK11) of the four strains compared by Taïbi et al. (2010) were representative of groups of strains, whereas the fourth (Wg2) had a unique PFGE profile. The observation that many strains were related was not surprising because studies based on phage host range had previously indicated that a relatively small number of significantly different cremoris starter strains exist (Lawrence and Pearce 1972; Lawrence et al. 1978). The result is that many cremoris strains of diverse origin are unknowingly related and an example of this is shown in figure 1 where strains related to SK11 are compared. SK11 was isolated in New Zealand as a phage-resistant derivative of strain AM1 (DRI 1962), which had been obtained from Professor Collins from University of California Davis and originally named LT8 (DRI 1960). Two other cultures were introduced to New Zealand at the same time, AM2 (FC4) and AM3 (4B) (DRI 1960; Collins 1961), and all three had been isolated from commercial mixed strain cultures. AM1 and AM2 were regarded as slower starters and were found to make consistently good-flavored Cheddar cheese (Martley and Lawrence 1972). Figure 1 shows that AM1 and SK11 have the same PFGE SmaI digest pattern but that several other strains are similar. These include US3 and R6, which had both been in use as defined strain starters since the early 1950s (DRI 1951), strain 134 (originally described as a phage-resistant derivative of AM2, Limsowtin and Terzaghi 1976), and several isolates from mixed strain starter cultures.
The predicted PFGE pattern from the sequenced SK11 strain differs from that for SK11 shown in figure 1 but matches to that of strain 134. The major PFGE bands are very similar in these strains with the main difference being a deletion of ~20 kb from one of the largest PFGE bands (a 293-kb doublet which separates into 293- and 273-kb bands). When the 273-kb SmaI fragment from the SK11 genome is compared with the corresponding region from MG1363, the only major difference is the presence of a 19-kb prophage sequence (MG-1) in MG1363 (Ventura et al. 2007). A prophage (bIL310) is integrated in the same genomic region in IL1403, and there is strong homology and synteny between bIL310 and MG-1. Consequently, it is likely that a similar prophage has been lost from the sequenced SK11 strain.
DNA from 13 cremoris strains failed to cut with the enzyme SmaI, indicative of the presence of a restriction/modification system operating in these strains. Of the known lactococcal R/M systems, only the ScrFI methylase potentially blocks the SmaI recognition site (Szatmari et al. 2006). Unlike most lactococcal R/M systems, ScrFI is chromosomally encoded and a single 75-kb ApaI digest fragment of chromosomal DNA from these strains hybridized to a probe for the scrFIAM methylase gene. When the histories of these strains were examined, most were found to have been isolated from mixed strain starter cultures.
The PFGE patterns of SmaI-digested genomic DNA from 72 Cit+ strains showed that most strains had several bands in common, and representatives are shown in figure 2A. The relatedness of these strains is supported by the presence of prophage, and these were detected using probes to the chromosome–prophage junctions in strain IL1403 (Chopin et al. 2001). An example is shown in figure 2B where the left hand junction fragment between the IL1403 chromosome and the bIL309 prophage hybridizes strongly with all the Cit+ strains. Six prophages have been identified on the IL1403 chromosome (Bolotin et al. 2001), and junction fragments for five of these (bIL285, bIL309, bIL310, bIL311, and bIL312) could be detected in all strains. The sixth prophage (bIL286) was found only in IL1403 and CNRZ157 and may be less stable than the others as we isolated a derivative of CNRZ157, which was spontaneously cured of this prophage. In a study using minisatellite polymorphism to distinguish closely related L. lactis strains, a sequence from within the bIL286 prophage was used as a strain-specific minisatellite. This sequence was only found in the genome of IL1403 and not in nine other L. lactis strains (Quénée et al. 2005). The observation by McGrath et al. (2002) that the genomes of IL1403 and the citrate-utilizing strains IL409 (DRC1) and F7/2 contain identical prophage-encoded bacteriophage resistance genes is in agreement with our observation that these strains are closely related and harbor related prophage. The gene identified (sieIL409) showed 100% amino acid identity with orf2 of bIL309.
The plasmid-free strain IL1403 was originally derived from the citrate-utilizing strain L. lactis subsp. lactis biovar diacetylactis CNRZ157 (IL594) following protoplast-induced curing (Chopin et al. 1984; Bourel et al. 1996). Citrate utilization requires citrate permease to transport citrate into the cell and citrate lyase to initiate citrate breakdown (Drider et al. 2004). In lactococci, these activities are genetically separate with citrate permease being plasmid encoded on an 8-kb plasmid (Kempler and McKay 1981), whereas citrate lyase and other genes involved in citrate breakdown are chromosomal (Bolotin et al. 2001). The distribution of the chromosomal citrate-utilizing genes among different lactococcal strains is not known, but citrate lyase activity was not present in cell-free extracts of 24 dairy lactococcal strains (Harvey and Collins 1961), and the gene cluster mae-maeP-citRCDEFXG, which includes the genes required for the synthesis of active citrate lyase, is missing from the other sequenced L. lactis strains (Wegmann et al. 2007; Siezen et al. 2008).
PFGE was also used to compare 197 strains with the lactis phenotype (L. lactis subsp. lactis) made up of 110 dairy starter strains and 87 isolated from various sources. With the exception of a group of cultures used in lactic casein manufacture that have been described previously (Ward et al. 2004), the L. lactis subsp. lactis strains showed much more diversity than the other groups and the majority of strains gave unique PFGE patterns.
Two of the sequenced L. lactis strains have the lactis phenotype, MG1363 (Wegmann et al. 2007) and KF147 (Siezen et al. 2010). MG1363 was made plasmid free by UV treatment and protoplast-induced curing (Gasson 1983) and belongs to a group of related strains, which includes NCDO712, C2, ML3, LM0230, and 952 (Davies et al. 1981; Lucey et al. 1993; Le Bourgeois et al. 2000). All have a lactis phenotype but a cremoris genotype. NCDO712, C2, and MG1363 have a chromosomally integrated sex factor not found in other lactococcal strains. Strains belonging to this group have been widely used as conjugation recipients, and large DNA fragments are known to be capable of integration at several sites on the chromosome, but techniques developed for MG1363 have been difficult to transfer to dairy starter strains (Johansen 2003). KF147 is one of a group of L. lactis cultures isolated from minimally processed fruit and vegetable products (Kelly et al. 1998a) and has several novel properties not found in dairy starters.
Based on the SmaI PFGE patterns, 80 strains (table 1) were chosen as representative of the L. lactis species. These included 20 strains belonging to each of four groups chosen for comparison because of their origin or because of their use for different purposes in the dairy industry and also included the four strains whose genome sequence has been determined. These are 1) L. lactis subsp. cremoris dairy starter cultures, 2) L. lactis subsp. lactis biovar diacetylactis dairy starter cultures, 3) L. lactis subsp. lactis dairy starter cultures, and 4) wild-type L. lactis subsp. lactis strains. Based on PFGE patterns and strain history data, the cultures selected were believed to be unrelated to one another except for SK11, which is a bacteriophage-resistant derivative of L. lactis subsp. cremoris AM1, and IL1403 and MG1363, which are plasmid-free derivatives of L. lactis subsp. lactis biovar diacetylactis CNRZ157 and L. lactis subsp. lactis NCDO712, respectively.
Chromosomal mapping has shown that L. lactis has a circular chromosome (fig. 3A) with six ribosomal operons that are transcribed divergently from the origin of chromosomal replication (Davidson et al. 1996). Whereas it is expected that in any one strain, all six 16S rRNA copies will have the same nucleotide sequence, work by Pillidge et al. (2009) has highlighted an additional level of complexity. A small number of L. lactis subsp. cremoris strains, some of which show PFGE patterns similar to SK11, appear to be genotypic hybrids and have both cremoris-like and lactis-like 16S rRNA types in their genome. Some L. lactis subsp. cremoris strains contain plasmids with a lactis-like 16S rRNA pseudogene, and it is proposed that these chimeric strains are the result of homologous recombination between the pseudogene and the corresponding chromosome gene (Pillidge et al. 2009).
Because plasmid DNA contributes to the PFGE patterns resulting from SmaI digests (Ward et al. 1993), the homing endonuclease I-CeuI, which cuts only within the 23S rRNA gene (Liu et al. 1999), was used to produce a PFGE pattern based on chromosomal DNA alone. Macrorestriction patterns produced by PFGE of I-CeuI digests of genomic DNA (fig. 3B) provide information on chromosomal size, the number and position of rRNA operons, and an indication of chromosomal rearrangements or insertions and deletions. All the 80 L. lactis strains examined gave six fragments when their genomic DNA was digested with I-CeuI, indicating that the copy number of the rRNA genes is conserved in this species (table 1). These fragments were designated as Ce1–Ce6 following the nomenclature used by Le Bourgeois et al. (1992). Partial digests were used to determine the order of the I-CeuI fragments, and for most strains, the relative size and order of the various fragments are the same as in IL1403, suggesting that chromosomal structure is conserved in most cases. A minority of strains showed chromosomal rearrangements, and these were of two types typified by strains HP and FG2. From the 289 cremoris strains investigated, 21 had PFGE patterns similar to HP and 15 had patterns similar to FG2. The rearrangement in FG2 has been described previously during chromosome mapping studies (Davidson et al. 1996). Curiously, strains HP and FG2 both carry plasmids that specify the same uncommon type of cell envelope proteinase (lactocepin) linked to a partially deleted copy of abiB (Christensson et al. 2001). These two strains also cluster together and separate from strains SK11 and AM2 in the CGH study reported by Bayjanov et al. (2009), and both HP and Z8 (a culture with the same atypical PFGE pattern as FG2) were the only L. lactis strains shown to lack the busA operon in the osmolality study described by Obis et al. (2001). The L. lactis subsp. lactis biovar diacetylactis strain LW3079 isolated from a mixed strain starter culture has a rearrangement similar to that found in the HP-like strains. These chromosomal changes have no observable effect on cell growth, morphology, or phenotype.
With the availability of four different L. lactis genome sequences, it is possible to get an indication of the larger scale events that shape the lactococcal genome. Figure 4 compares the four genomes and shows that overall there is a high degree of conservation. A large inversion involving approximately half the chromosome has been described in MG1363 (Daveran-Mingot et al. 1998), although this occurs within the Ce1 fragment and does not result in a change to the I-CeuI pattern (fig. 4) or alteration in chromosomal symmetry. Major genome insertions in these strains are highlighted in figure 4 and are predominantly associated with prophages, the integration of plasmid genes, polysaccharide biosynthesis, or the ability to metabolize plant-derived carbohydrates. Prophages are an important feature, as phage resistance has been a major driver for strain selection programs for dairy cultures, and phage challenge has been a continual selective stress in their environment. Strain-specific genes of significance include the malate–citrate metabolism genes in IL1403 and the integrated sex factor in MG1363. KF147 also contains genes for nisin biosynthesis, although this strain does not produce nisin (Kelly et al. 1998a; Siezen et al. 2008), and for nonribosomal peptide and polyketide synthesis. The large insertion in the Ce3 fragment in KF147 (fig. 4) is a chromosomally integrated conjugative element that encodes the ability to metabolize alpha-galactosides such as melibiose and raffinose. Transfer of this element to a derivative of MG1363 and its integration at two different chromosomal sites have been described previously (Kelly et al. 1998b), and similar conjugative elements were found in several L. lactis strains of plant origin.
The availability of four complete genome sequences covering most genotype/phenotype combinations makes it possible to produce a phylogeny truly representative of the entire genome. Supertree methods (Fitzpatrick et al. 2006) were used to derive phylogenies from 1160 single gene families with the results shown in figure 5. This analysis strongly supports the conclusion from the study by Rademaker et al. (2007) that two main lineages exist in L. lactis. These correspond to the two genotypes. Strains with a L. lactis subsp. cremoris genotype include strains with both lactis (MG1363) and cremoris (SK11) phenotypes, whereas strains with the L. lactis subsp. lactis genotype includes strains with the both diacetylactis (CNRZ157, the parent strain of IL1403) and lactis (KF147) phenotypes. From the supertree results, it appears that a similar situation may exist in S. thermophilus, but this awaits further study.
The average chromosome size of the 80 strains of L. lactis was 2,483 kb, with the chromosomes of individual strains ranging in size from 2,242 to 2,688 kb. This variation in chromosomal length (~20% of the size of the smallest chromosome) is similar to that found in natural isolates of Escherichia coli (Bergthorsson and Ochman 1998). Table 2 shows the comparison between the mean chromosomal lengths of L. lactis strains from different genotype–phenotype origin combinations. The origin of the strains has the greatest influence on chromosome length with dairy strains having smaller chromosomes than the wild-type strains. This smaller chromosome size may be the result of a process of reductive genome evolution as a consequence of the adaptation to growth in milk. Among the dairy strains, those with the cremoris genotype and phenotype are significantly different and have the smallest chromosomes, whereas there is no significant difference between strains with the diacetylactis and lactis phenotypes. In this analysis, the five cremoris genotype/lactis phenotype dairy strains grouped with the wild-type strains. This could be related to the small sample size or could indicate that they are not as strongly adapted to the dairy environment and are closer to strains of plant origin. That MG1363 has much greater ability than either IL1403 or SK11 to grow on plant-derived carbohydrates (Wegmann et al. 2007) supports the latter option.
All the I-CeuI fragments show some degree of length variation (table 1), although, except for the strains that show chromosomal rearrangements, the sizes of the individual fragments do not overlap. To test whether some chromosomal regions are more variable than others, we compared the standardized variances for each fragment across strains in pairwise F tests. The four strains that showed major chromosomal rearrangements (HP, 2188, FG2, and LW3079) were not included so this comparison was based on 76 strains. There were no significant differences when most of the fragments were compared against each other; however, the largest fragment (Ce1) was significantly more variable (P ≤ 0.005), and it can be seen from figure 4 that the majority of chromosomal insertions are found in this region. Ce5 was significantly less variable (P ≤ 0.001) than the other fragments. A feature of the size variation in the I-CeuI fragments is a correlation between the two largest fragments (fig. 6A). This suggests that chromosomal insertions and deletions in these regions are not entirely independent events and that maintenance of chromosomal symmetry may be an important consideration.
There is a significant correlation between the I-CeuI fragment size and the standardized variance of each fragment (fig. 6B) with the two smallest fragments showing the greatest departure from this relationship. The smallest fragment (Ce6) shows greater variance than expected, whereas the variance in size of the second smallest fragment (Ce5) is less than for any of the other fragments. The size of Ce5 is strongly conserved (45–50 kb), and we have found only one strain (LW1509) that contains a large insertion of DNA in this region.
Because Ce6 exhibits larger variation than expected for a fragment of this size, we examined the genes present in this region from the four available L. lactis genome sequences (fig. 7). MG1363 and SK11 (both L. lactis subsp. cremoris genotype) show two common insertions relative to IL1403 and KF147. These are of genes of unknown function between purR and fusA and between rpsL and dacA. In IL1403, 5 of the 14 genes in this region are predicted to be highly expressed (Karlin et al. 2004). These are fusA (predicted to be the most highly expressed gene in the IL1403 genome) and four genes encoding ribosomal proteins (rpsI, rplM, rpsG, and rpsL). Curiously, in the SK11 genome sequence, the fusA homologue (LACR_2595) is identified as a pseudogene.
The main difference in the genome sequences for this region is a 15-kb insertion in SK11 between the ribosomal protein S9 gene (rpsI) and the 16S rRNA gene. This insertion contains two genes with high homology to type III restriction–modification systems found in other lactic acid bacteria, several hypothetical proteins, an integrase, and genes related to plasmid replication. It is probable that this insertion increases the phage resistance of SK11, and it may have been acquired by horizontal gene transfer, and subsequent chromosomal integration, of a small plasmid. It is apparent that many dairy starter strains have an insertion of similar size in the Ce6 region (table 1), and it will be of interest to determine if similar genes are found in other strains and whether they have an influence on cell growth rate given their location close to the chromosomal origin.
Many technologically important properties (lactose metabolism, lactocepin proteinase, citrate permease, and bacteriophage resistance) are plasmid encoded in L. lactis strains used as dairy cultures and can be exchanged between strains by conjugation and between replicons by insertion sequence elements. Examination of plasmid gels for 150 dairy starter cultures (90 L. lactis subsp. cremoris, 30 L. lactis subsp. lactis, and 30 L. lactis subsp. lactis biovar diacetylactis) gave an average of seven plasmids per strain (range 2–14) with up to 200 kb of plasmid DNA. By contrast, in plant strains, the average was <2 plasmids per strain (range 0–4). It was notable that small plasmids (<10 kb) were prevalent in dairy cultures but rare in the plant strains. Intact genomic DNA was run on PFGs but showed no evidence of large linear plasmids in these strains.
Plasmid profiles for 20 Cit+ strains are shown in figure 8A and show that these strains have acquired a range of different extrachromosomal elements. A plasmid-encoded citrate permease is required for citrate utilization in L. lactis, and to determine its location, a Southern blot of the plasmid gel was probed with a PCR-amplified product of the citP gene (fig. 8B). The citrate permease plasmid is conserved in size (8 kb) in most strains, but three strains (D6, LW3076, and LW1503) have an enlarged citrate plasmid of ~15 kb. We observed that the plasmid complement can be very unstable in some strains, and this spontaneous instability has previously been reported for CNRZ157 (Chopin et al. 1984).
Most of the L. lactis strains used in this study were dairy starter cultures and are representative of the cultures employed by the dairy industry since the concept of using defined strain starters was developed in the 1930s (Limsowtin et al. 1996). During this period, the best cultures were freely shared between laboratories, and this coupled with the repeated isolation of individual strains with particular characteristics from commercial mixed strain starters, and the development of bacteriophage-insensitive cultures has resulted in many closely related strains coexisting. Consequently, the relationship between strains is generally not known, although it has been speculated that the pool of strains with certain properties is small (Lawrence et al. 1978). Because of their differing histories of industrial use, even closely related strains may differ in some of their important characteristics and in their plasmid complement (Ward et al. 2004).
The aim of this work was to use PFGE to gain a measure of the chromosomal diversity present in a large collection of L. lactis cultures including both plant and wild-type strains, with emphasis on the L. lactis strains with the cremoris and diacetylactis phenotypes that are of particular importance to the dairy industry. The diacetylactis phenotype is unusual in that it depends on the inheritance of both plasmid and chromosomal components for citrate to be transported into the cell and metabolized. The chromosomal genes are not found in the other sequenced strains and show homology with the plasmid-encoded citrate genes from Leuconostoc and Weissella species isolated from the dairy environment. Therefore, it can be hypothesized that the diacetylactis strains used as defined strain dairy starters represent a single lineage within L. lactis in which both chromosomal and plasmid elements are maintained and essential for an industrially significant phenotype. It should be noted that this observation is restricted to the Cit+ strains used as defined strain dairy starter cultures. Wild-type L. lactis may differ in their citrate metabolism, and strains with a 23-kb plasmid-encoding genes that match those found in citrate-metabolizing leuconostocs have been isolated from Algerian dromedary's milk (Drici et al. 2010).
The cremoris phenotype is rather more complex to determine because it is measured as negative attributes including the inability to metabolize arginine and inability to grow at higher temperatures or at higher salt levels. From investigations of the genome sequences, it appears that these properties have arisen through the accumulation of mutations and that they are a response to the nutrient rich milk environment where certain gene functions are not longer required. The arginine deiminase–negative phenotype of SK11 correlates with a single base pair deletion (Wegmann et al. 2007), whereas various defects including the absence of the busA operon have been shown to influence salt tolerance (Obis et al. 2001). The taxonomy of L. lactis is currently based on phenotype, but as the genomic basis for these phenotypic differences is elucidated, a case could be made for review and use of genotypic data to define the two subspecies.
Comparison of our data with the analysis of the genome-sequenced L. lactis strains strongly supports a plant-associated origin for dairy starter strains. A significant proportion of the KF147 genome is devoted to genes involved in the degradation and metabolism of plant-derived carbohydrates, and there is clear evidence for remnants of these capabilities in the genomes of dairy strains, most notably MG1363. This reductive evolution can be seen as part of the adaptation to growth in milk and accounts for the smaller chromosome size in dairy starters as compared with wild-type strains. Similar observations have been made for Lactobacillus bulgaricus (Van de Guchte et al. 2006) and S. thermophilus (Hols et al. 2005) dairy cultures. We conclude that dairy starter cultures generally and especially strains with cremoris and diacetylactis phenotypes comprise a specialized group of L. lactis strains. This is in accordance with the results from the recent studies of L. lactis (Rademaker et al. 2007; Bayjanov et al. 2009) and fits with knowledge of the origin of industrially used dairy starter cultures. Consequently, the world's dairy industry is based on the same small group of good starter strains, and these cultures have transitioned from free-living organisms associated with plant material. We hypothesize that these specialized dairy starters are no longer fit to survive outside the dairy environment and have evolved to become essential components of industrial processes. Whereas wild-type strains may have genes that can be used to enhance the metabolism of dairy strains (van Hylckama Vlieg et al. 2006), it is unlikely that there is an environmental source for new dairy starter cultures similar to those currently in use.
This analysis together with other studies (Nomura et al. 2006; Rademaker et al. 2007; Siezen et al. 2008; Bachmann et al. 2009; Bayjanov et al. 2009; Liu et al. 2010) begins to illustrate the genomic and phenotypic diversity present within L. lactis. Currently, too few genome sequences are available to delineate sets of core and auxillary genes and describe the pangenome of L. lactis, but in the future, it will be interesting to examine strains from other environments in more detail and to further define the genes necessary to make a good dairy starter culture.
New Zealand Foundation for Research, Science, and Technology (Contract C10X0239).
We thank Zaneta Park for assistance with the statistical analysis and Dr David Fitzpatrick from the Genome Evolution Laboratory, National University of Ireland, Maynooth, Ireland, for assistance with the supertree phylogenetic analysis. We also wish to thank Howard Heap and the members of the Microbial Fermentation Unit, Fonterra, Palmerston North, for their support and advice on lactococcal dairy starter cultures.