Role of Hypermutability in the Evolution of the Genus Oenococcus

Abstract

Oenococcus oeni is an alcohol-tolerant, acidophilic lactic acid bacterium primarily responsible for malolactic fermentation in wine. A recent comparative genomic analysis of O. oeni PSU-1 with other sequenced lactic acid bacteria indicates that PSU-1 lacks the mismatch repair (MMR) genes mutS and mutL. Consistent with the lack of MMR, mutation rates for O. oeni PSU-1 and a second oenococcal species, O. kitaharae, were higher than those observed for neighboring taxa, Pediococcus pentosaceus and Leuconostoc mesenteroides. Sequence analysis of the rpoB mutations in rifampin-resistant strains from both oenococcal species revealed a high percentage of transition mutations, a result indicative of the lack of MMR. An analysis of common alleles in the two sequenced O. oeni strains, PSU-1 and BAA-1163, also revealed a significantly higher level of transition substitutions than were observed in other Lactobacillales species. These results suggest that the genus Oenococcus is hypermutable due to the loss of mutS and mutL, which occurred with the divergence away from the neighboring Leuconostoc branch. The hypermutable status of the genus Oenococcus explains the observed high level of allelic polymorphism among known O. oeni isolates and likely contributed to the unique adaptation of this genus to acidic and alcoholic environments.

Two species of lactic acid bacteria (LAB), Oenococcus oeni and the recently identified Oenococcus kitaharae, are described within the Oenococcus genus (12, 50). O. oeni, formerly Leuconostoc oenos (7), plays an important role in the elaboration of wine, where it is often added as a starter culture to carry out the malolactic conversion (28). Given the economic importance of this conversion, the taxonomic structure of this species has been studied in detail, the result of which indicates that the species is quite homogeneous (22, 44, 51, 59). Recently, however, multilocus sequence typing (MLST) of 18 strains revealed a high level of allelic diversity in O. oeni, which has a panmictic population structure where lines of clonal descent are difficult to define (5). Panmictic populations are often characterized by high levels of horizontal transfer and recombination among strains, and this was posited as a cause of the genomic diversity and evolution of O. oeni (5). On the basis of 16S rRNA analysis, Yang and Woese (56) proposed an accelerated evolution of the Leuconostoc group in general and of O. oeni in particular. This work was confirmed by recent phylogenetic comparisons of concatenated ribosomal and RNA polymerase subunits (29).

A recent comparative genomic analysis of O. oeni PSU-1 with other sequenced LAB indicated that O. oeni PSU-1 lacks the genes mutS and mutL (29, 30), which encode two key enzymes in the mismatch repair (MMR) pathway. The MMR pathway is an excision repair system that corrects many types of base pair mismatches (14, 24, 35, 36, 39). While there are some differences among MMR systems in different bacteria, the presence of mutS and mutL homologs is required. The MutS protein recognizes the DNA mismatch, and the MutL protein binds MutS, targeting the mutation for an excision (36). A new nonmethylated strand is then synthesized, and the mismatch is corrected (25). As expected, the correction of mismatches by MutS and MutL decreases the spontaneous mutation rate of a species. MMR is also involved in the regulation of interspecies recombination by preventing the incorporation of heteroduplex DNA (33, 43). Thus, a defect in the MMR system leads to an increase of the mutation frequency, as well as to an increase of the horizontal gene transfer among different strains (41).

This work demonstrates that the high mutation rate and compositional bias of spontaneous mutations in the two known species of Oenococcus, O. oeni and O. kitaharae, are consistent with a lack of MMR and likely contribute to the genetic variation and accelerated evolution of the genus. Given the small genomes, relatively limited metabolic capacity, and general functional similarities between oenococci and neighboring MMR-containing Leuconostoc species, the genus Oenococcus represents a unique opportunity for examining the role of MMR in bacterial evolution.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

O. oeni and O. kitaharae strains were grown in MLAB medium. This medium is constituted of MRS (Becton Dickinson, Sparks, MD) supplemented with 0.5% fructose, 0.1% malic acid, and 10% sterile tomato juice. Leuconostoc mesenteroides subsp. mesenteroides (herein termed L. mesenteroides) and Pediococcus pentosaceus were grown in MRS medium containing 0.5% glucose. All bacterial strains were grown at 30°C without shaking.

Isolation of spontaneous erythromycin- and rifampin-resistant mutants.

Spontaneous rifampin- and erythromycin-resistant mutants were obtained by growing 100 ml of cell culture until late logarithmic phase (optical density at 600 nm, 1), concentrating the culture 50-fold via centrifugation, and plating on selective media (MLAB or MRS containing an antibiotic) in triplicate. For each species, we used four times the MIC of antibiotic. MICs for rifampin were 50 μg/ml for O. oeni PSU-1, 5 μg/ml for O. kitaharae NRIC0645, 20 μg/ml for L. mesenteroides ATCC 8293, and 30 μg/ml for P. pentosaceus ATCC 25745. MICs for erythromycin were 30 μg/ml for O. oeni PSU-1 and O. kitaharae NRIC0645 and 3 μg/ml for L. mesenteroides ATCC 8293 and P. pentosaceus ATCC 25745. The original number of CFU per ml present in the concentrated culture was determined by performing serial dilutions and plating on media without antibiotics. Plates were incubated at 30°C until colonies started to appear (5 to 6 days for O. oeni and O. kitaharae strains and 3 to 4 days for L. mesenteroides and P. pentosaceus strains). Mutation frequencies were determined as the median number of mutants per average number of input CFU. The mutation frequencies of O. oeni PSU-1, O. kitaharae NRIC0645, L. mesenteroides ATCC 8293, and P. pentosaceus ATTC 25745, from a set of 10 cultures, were used to calculate the mutation rate (μ) by the method of Drake (9), using the formula μ = f ln Nμ, where f is the median mutation frequency and N is the average number of cells in the cultures.

Detection of mutations in the rpoB gene.

To determine the nature of mutations conferring resistance to the antibiotic rifampin, we examined the rpoB nucleotide substitutions (13). A 600-bp segment of the rpoB gene comprising the region most often associated with rifampin resistance (20) was amplified from O. oeni PSU-1, L. mesenteroides ATCC 8293, and P. pentosaceus ATCC 25745 by the use of the primers Omut1 (5′-GTCCGGTTGTTGCCGTAGTC) and Omut2 (5′-GAATGCGGAGCAACCAAAG), Lmut1 (5′-GTCCTGTTGTAGCCGCTGTC) and Lmut2 (5′-GAATGTGGATCAAGCAAAGG), and Pmut1 (5′-GTCCAGTAGTTGCTGCAACT) and Pmut2 (5′-GAATGTGGATCAACCAATG), respectively. The rpoB gene from O. kitaharae was amplified using the primers Ok1 (5′-SAAAACCTTCCGYATTGG) and Ok2 (5′-GCATCTTCGAAGTTATAWCCWTGCC).

Colony PCRs were carried out as previously described (42). PCRs were performed in a 25-μl amplification reaction mixture containing 20 mM Tris-HCl [pH 8.0], 50 mM KCl, 2.5 mM MgCl₂, 200 mM of each deoxynucleoside triphosphate, 1 μM of each primer, and 1 U of AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, CA). A Perkin-Elmer thermocycler was used with the following cycle parameters: 35 cycles of 94°C for denaturation, 1 min at 48°C for annealing, and 1 min at 72°C for polymerization. Amplicons were run on 1.0% agarose gels, stained with ethidium bromide, visualized under UV light, and photographed with a MultiImage light cabinet (Alpha Innotech Corporation, San Leandro, CA). PCR products were purified using the QIAquick purification kit (Qiagen, Valencia, CA). The sequences were aligned with the Vector NTI program (Invitrogen, Carlsbad, CA).

Bioinformatic analyses.

Genomic comparisons of O. oeni PSU-1 (GenBank accession number CP000411.1), L. mesenteroides ATCC 8293 (GenBank accession number CP000414.1), and P. pentosaceus ATCC 25745 (GenBank accession number CP000422.1) were performed using the Integrated Microbial Genomes program (31) developed by the DOE Joint Genome Institute. 16S rRNA-based dendrograms were generated using tools available on the Ribosomal Database Project website (4).

For the nucleotide substitution spectrum analysis, pairs of genomes of closely related Bacilli were downloaded from GenBank. For each pair of genomes, putative orthologs were identified using the symmetrical BLASTP hit scheme (49). Pairs of proteins were aligned using the MUSCLE program (10); protein pairs with more than 5% of their amino acid differences in the aligned region were discarded. Protein coding sequences were mapped to the corresponding amino acid alignments; third-codon positions belonging to fourfold-degenerate codons coding for the same amino acid in both genomes were extracted and concatenated across the pair of genomes. The resulting alignments of (mostly) neutrally evolving nucleotide sites were analyzed using the BASEML program of the PAML package (57) under the K80 substitution model to estimate the transition/transversion ratio. Raw numbers of observed transitions and transversions were compared to those in O. oeni by use of the χ² statistic to estimate the level of significance of the observed differences.

Nucleotide sequence accession number.

The partial sequence of the rpoB gene of O. kitaharae NRIC0645 has been submitted to the GenBank database under accession number EF417917.

RESULTS

Presence of genes involved in repair pathways.

A list of the various nucleotide repair pathways for three sequenced LAB, O. oeni, P. pentosaceus, and L. mesenteroides, is shown in Table S1 in the supplemental material. While all three genomes have homologs for most of the known repair genes, some differences are found. Homologs of the genes that encode deoxyribopyrimidine photolyase (phrB) and a DNA repair protein, RecT, are present only in the genome of L. mesenteroides. Moreover, O. oeni lacks some important mutator genes, such as the DNA helicase (recQ) gene and those involved in the MMR system, mutS and mutL. The O. oeni genome does have a mutS homolog belonging to the MutS2 subfamily, one of the two major subfamilies of MutS proteins. MutS2 group proteins are evolutionarily distinct from the MutS1 group involved in the MMR system (11). MutS2 proteins have been shown to recognize anomalous DNA structures, such as Holliday junctions or oxidative DNA damage, rather than DNA mismatches (47, 52). Others have proposed that MutS2 proteins function in chromosome segregation or crossing-over (11). Importantly, the mutSL operon, which is found in the majority of gram-positive bacteria (14, 40), is present in L. mesenteroides and P. pentosaceus but not in O. oeni.

Mutation rates.

To examine the possible impact of the absence of mutSL, we determined the rates for spontaneous mutation to rifampin and erythromycin resistance for O. oeni, O. kitaharae, and neighboring taxa, L. mesenteroides and P. pentosaceus. Notably, the spontaneous mutation rate for both antibiotics is significantly higher for O. oeni PSU-1 than those observed for the other species (Table 1). L. mesenteroides ATCC 8293 and P. pentosaceus ATCC 25745 possess a 100-fold-lower mutation rate than O. oeni PSU-1, suggesting that the lack of mutSL dramatically increases the spontaneous mutation rate in this strain. Interestingly, O. kitaharae exhibited a spontaneous mutation rate between those of mutSL-lacking O. oeni PSU-1 and mutSL-containing P. pentosaceus ATCC 25745 and L. mesenteroides ATCC 8293.

TABLE 1.

Spontaneous mutation rates for select LAB

Species and strain	Mutation ratea
	Rifampin	Erythromycin
Oenococcus oeni PSU-1	1.6 × 10−6	1.6 × 10−6
Oenococcus kitaharae NRIC0645	2.4 × 10−7	1.1 × 10−7
Leuconostoc mesenteroides subsp. mesenteroides ATCC 8293	1.5 × 10−8	4.2 × 10−9
Pediococcus pentosaceus ATCC 25745	5.5 × 10−8	4.7 × 10−9

^aThe spontaneous mutation rates for each antibiotic were determined by the method of Drake (9).

Nature of the rpoB mutations.

Previous work has shown that MMR systems more effectively correct transition substitutions than transversion substitutions (2, 13, 19, 45). As a consequence, host cells lacking the MMR system exhibit an increased level of transition substitutions relative to wild-type (WT) cells (36). In order to assess the mutational spectra in bacteria, others have examined the rpoB gene nucleotide changes occurring after selection for rifampin resistance (13, 23). This assay, while clearly documenting selected and nonrandom mutations, provides a workable assessment of mutation profiles that enables comparisons between closely related species.

To determine if transition substitutions occur more frequently in oenococci, we compared the types of nucleotide changes resulting from rpoB mutations conferring resistance to rifampin among O. oeni, O. kitaharae, L. mesenteroides, and P. pentosaceus. Sequence analysis of a portion of the rpoB gene fragment from 75 rifampin-resistant O. oeni clones revealed a total of 53 mutations within the analyzed region. The mutations are located at nine different rpoB sites or hot spots (Table 2). Two out of five of these base substitution types are transition mutations (GC↔ AT), with the predominant rpoB substitutions exhibiting GC→AT transition. In total, 44 of the 53 substitutions found in O. oeni PSU-1 were transitions (83%), while 9 (17%) of the rpoB substitutions were transversions. The predominant GC→AT transition substitutions in O. oeni differ from the predominant substitutions observed in other MMR-deficient species. Studies of the rifampin resistance mutations of an Escherichia coli mutS strain revealed a predominance of AT→GC substitutions in rpoB (45). All the O. oeni mutations resulted in a change of the corresponding amino acid, with the exceptions of the nucleotides at positions 1470 (R490) and 1830 (I610), which resulted in silent mutations. The characterization of 13 rpoB gene mutations from 25 O. kitaharae mutants revealed a similar predominance of transition substitutions (77%), with the most predominant substitution in O. kitaharae being a GC→AT transition.

TABLE 2.

Distribution of mutations in rpoB in select LAB^a

Substitution	O. oeni			L. mesenteroides			P. pentosaceus			O. kitaharae
	Occurrence	aa change	Site	Occurrence	aa change	Site	Occurrence	aa change	Site	Occurrence	aa change	Site
GC→TAb	6/53	D477Y	1429	24/36	V461L	1381	8/12	D477Y	1429	1/13	L43F	129
		H487N	1459		G469V	1406
		S492Y	1475
GC→ATc	35/53	D477N	1429	4/36	R491H	1472	4/12	H487Y	1459	9/13	S41L	122
		G483D	1448		S493L	1478
		H487Y	1459
		R490N	1469
		S492F	1475
		I610d	1830
GC→CGb	2/53	H487D	1459	1/36	H488D	1462
AT→GCc	9/53	H487R	1460	6/36	Q475G	1424				1/13	D26G	57
		L470S	1480		R490d	1470
					R501T	1503
AT→TAb	1/53	R490d	1470
AT→CGb										2/13	L40C	119

^aThe numbering system for the nucleotide bases of the rpoB gene originates with the ATG start codon. In the case of O. kitaharae, the numbering originates with the first base of the partial rpoB sequence (GenBank accession number EF417917). The percentages of transition mutations for O. oeni, L. mesenteroides, P. pentosaceus, and O. kitaharae are 83, 28, 33, and 77, respectively. aa, amino acid.

^bTransversion mutation.

^cTransition mutation.

^dThis change creates a silent mutation.

Nucleotide substitutions in L. mesenteroides and P. pentosaceus, species that contain mutSL, revealed dramatically different results. Of 36 rpoB gene mutations found in 25 L. mesenteroides mutants, only 28% were shown to be transition substitutions, while a similar analysis of 12 mutations identified among 25 P. pentosaceus mutants revealed that 33% were transition substitutions. In the L. mesenteroides rpoB gene, several hot spots were observed, including GC→TA changes at amino acid positions 461 and 469, GC→AT changes at amino acid positions 491 and 493, a GC→CG change at amino acid position 488, and AT→GC changes at amino acid positions 475, 490, and 501. In P. pentosaceus, only two hot spots are found, a GC→TA change at amino acid position 477 and a GC→AT change at amino acid position 487.

In bacteria, most rpoB mutations have been located in three regions, defined for E. coli as clusters I, II, and III (20). All the polymorphisms identified in this study are in cluster I, with the exception of the GC→AT change at amino acid 610 of RpoB in O. oeni. Even though it is a characteristic of some mutants in other bacterial species (23), no deletion mutations were found in any of the strains studied in this work.

Nature of nucleotide substitutions among members of the Lactobacillales.

Recently a draft sequence for a second O. oeni strain, BAA-1163, became publicly available (GenBank accession number AAUV00000000), enabling us to compare the transition/transversion ratio observed in O. oeni to those in other Bacilli. To this end, we performed a large-scale comparison of nucleotide substitutions between orthologs among other pairs of closely related genome sequences of Bacilli. As seen in Table 3, the ratio of transition substitutions to transversion substitutions for the two O. oeni strains is significantly higher than those for other Lactobacillales and Bacillales. These data confirm that the mutational spectra in O. oeni are clearly shifted, and the elevation of transition nucleotide substitutions is consistent with the lack of MMR in the species. This is also consistent with the results of the rpoB analysis (Table 2). A detailed comparison among 91 orthologs from O. oeni strain BAA-1163 and PSU-1, selected on the basis of a nucleotide identity of between 95 and 98%, is presented in Table S2 in the supplemental material.

TABLE 3.

Comparison of nucleotide substitution patterns among select Bacilli

Pair of genomes	No. of orthologsa	No. of sitesb	No. of transitionsc	No. of transversionsc	Transition/transversion ratiod	P (χ2)e
Oenococcus oeni ATCC BAA-1163 vs O. oeni PSU-1	1,175	143,263	2,490	675	7.51 ± 0.33	NA
Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842 vs L. delbrueckii subsp. bulgaricus ATCC BAA-365	867	126,000	777	365	4.28 ± 0.27	<0.00001
Streptococcus agalactiae 18RS21 vs S. agalactiae NEM316	1,045	129,326	967	361	5.40 ± 0.33	0.00002
Streptococcus thermophilus CNRZ1066 vs S. thermophilus LMD-9	994	137,970	1,180	633	3.76 ± 0.19	<0.00001
Streptococcus pneumoniae D39 vs S. pneumoniae TIGR4	1,192	159,725	2,058	1,108	3.76 ± 0.14	<0.00001
Streptococcus pyogenes M1 GAS vs S. pyogenes SSI-1	1,024	136,923	2,124	1,122	3.84 ± 0.14	<0.00001
Listeria monocytogenes 4b F2365 vs L. monocytogenes 4b H7858	2,224	298,580	2,000	1,247	3.23 ± 0.12	<0.00001
Bacillus anthracis Ames Ancestor vs B. thuringiensis serovar konkukian 97-27	2,071	251,082	5,600	2,881	3.97 ± 0.09	<0.00001
Staphylococcus aureus subsp. aureus COL vs S. aureus subsp. aureus JH1	1,685	179,897	2,127	1,777	2.47 ± 0.08	<0.00001

^aNumbers of orthologs whose amino acid differences across the aligned parts of proteins were less than 5% of the sequence.

^bNumbers of fourfold-degenerate codons encoding the same amino acid in both genomes.

^cNumbers of transitions or transversions observed in the third positions of the fourfold-degenerate codons.

^dValues are mean maximum likelihood estimates of the transition/transversion ratios ± standard deviations.

^eχ² test probability for comparison with O. oeni. NA, not applicable.

DISCUSSION

Oenococcus oeni is an acidophilic LAB commonly observed in fruit mash environments and often employed commercially to carry out the malolactic conversion in the production of wines. Yang and Woese (56) first noted that the phylogenetic position of O. oeni, as determined by 16S rRNA analysis, indicated a tachytelic, or fast-evolving, organism. This was confirmed by a recent phylogenetic analysis of concatenated ribosomal proteins or RNA polymerase subunits from sequenced LAB indicating that the Leuconostoc branch in general and O. oeni in particular exhibit an accelerated evolution compared to neighboring taxa (29, 30). The same comparative genomic analysis of the sequenced LAB also revealed that O. oeni had lost key enzymes involved in MMR, MutS and MutL, suggesting that a lack of MMR is one factor responsible for the accelerated evolution of the species (30).

Due to their importance in MMR, MutS and MutL are conserved across numerous taxa (11). However, the emergence of whole-genome sequences has revealed that several bacterial species lack mutS and mutL, including Helicobacter pylori, Campylobacter jejuni, Mycobacterium tuberculosis, Mycoplasma pneumoniae, and Mycoplasma genitalium, implying the absence of MMR in these species (11). The lack of MMR in H. pylori is evidenced by the high genetic diversity and high level of mutator strains among isolates (1, 53). The lack of MMR is postulated to contribute to observed genomic instability and variable expression in C. jejuni (18). Mutations in MMR-related genes have been implicated in the hypermutability of numerous clinical isolates, including E. coli (6), Salmonella enterica (27), Haemophilus influenzae (54), Staphylococcus aureus (41), Streptococcus pneumoniae (37), and Pseudomonas aeruginosa (39). Direct inactivation of mutS or mutL in E. coli (36), Bacillus anthracis (60), Staphylococcus aureus (40, 41), or Pseudomonas stutzeri (34) has been shown to increase spontaneous mutation frequencies from 10- to 1,000-fold.

In this work, we demonstrate that O. oeni PSU-1 exhibits a high spontaneous mutation rate compared to the rates observed for its phylogenetic neighbors L. mesenteroides ATCC 8293 and P. pentosaceus ATCC 25745, strains that contain mutS and mutL genes (29). Analysis of rpoB gene polymorphisms in rifampin resistance mutations of O. oeni PSU-1 indicates a higher level of transition mutations and a different mutational spectrum than were observed in L. mesenteroides or P. pentosaceus. This was additionally confirmed by a comparison of nucleotide polymorphisms between the sequence of PSU-1 and the draft sequence of O. oeni strain BAA-1163, which also demonstrated a preference for transition substitutions among similar alleles, similar to the results of rpoB mutation analysis. Moreover, a specific prevalence of transition substitutions is clearly evidenced among ortholog nucleotide sequences in paired O. oeni genomes compared to the substitutions in other genome sequences of Lactobacillales and select Bacilli, clearly indicating a different mutational spectrum within the species O. oeni, consistent with the lack of MMR.

A high mutation rate in O. oeni helps explain some discordant observations of the species. Many researchers have examined the diversity of O. oeni strains within and around wineries. Results obtained from the application of different techniques, such as the study of patterns of total soluble cell proteins (8), 16S and 23S sequence analyses (32), random amplified polymorphic DNA-PCR (59), and differential-display PCR (26), indicate that O. oeni is a homogeneous species. Recently, de las Rivas et al. (5) used MLST of five genes (gyrB, ddl, mleA, pgm, and recP) to examine the allelic diversity and population structures of various oenococcal isolates. This analysis was able to completely differentiate 18 strains, suggesting a higher level of genetic heterogeneity among oenococcal isolates. Those authors argued that the high level of diversity in O. oeni represents an example of a panmictic genetic population, a population in which recombination among constituents is so frequent that it randomizes sequences and generates linkage equilibrium. We propose that the lack of mutS and mutL in O. oeni, combined with the high mutation rate demonstrated here, provides an explanation for a high allelic diversity among strains, as seen from the MLST data. A similar situation exists in Helicobacter pylori, where the lack of mutH and mutL in the species is believed to contribute to high strain diversity, resulting in a panmictic population structure (48).

The absence of MMR in O. oeni may have also contributed to species diversity by creating a more favorable environment for recombination between different alleles among various strains. One role of MMR is to prevent recombination between similar, though not identical, alleles (46). As a result, the disruption of the genes involved in MMR in S. pneumoniae (3), P. aeruginosa (39), Acinetobacter sp. (58), E. coli (21), and S. aureus (41) has been shown to increase the recombination frequency up to 1,000-fold. The suppression of the MMR system in E. coli and Salmonella enterica serovar Typhimurium enabled recombination among these bacteria, two species which have ∼20% sequence divergence (43). An increased recombination rate in Oenococcus may also help to abate the increased mutational load generated due to the MMR-deficient status, since functional alleles could be more readily acquired via horizontal transfer.

Recently, Endo and Okada (12) identified a second species of Oenococcus, O. kitaharae, obtained from composting shochu residue in Japan. Shochu is a distilled alcoholic beverage produced predominately from fermented rice, potatoes, or barley. Like with O. oeni, the mutation rate of O. kitaharae was significantly higher than those of L. mesenteroides and P. pentosaceus, and analysis of rpoB substitutions in rifampin-resistant mutants indicated a preference for transition substitutions. Attempts in our lab to amplify mutS and mutL alleles by PCR using degenerate primers failed to reveal the presence of these genes (data not shown). While the genome sequence of O. kitaharae is yet unknown, the aggregate evidence suggests that a lack of MMR in O. kitaharae is due to a loss of mutS and mutL. Given the phylogenetic position of O. kitaharae, it is likely that this loss occurred just after, or coincident with, the divergence of an oenococcal ancestor away from the neighboring MMR system-containing Leuconostoc branch (Fig. 1). Interestingly, of the 50 strains of Lactobacillales for which public genome sequences are available, only the genus Oenococcus lacks mutSL (Fig. 1).

FIG. 1.

Presence or absence of mutSL genes in members of the Lactobacillales for which complete genome sequences are available. The dendrogram showing the genetic relatedness based on 16S rRNA gene sequences was constructed using Ribosomal Database Project tools (4). While more than one strain of some species have been sequenced (Streptococcus agalactiae, S. pneumoniae, S. mutans, S. thermophilus, S. pyogenes, and Lactobacillus delbrueckii subsp. bulgaricus), only a single representative is shown in the figure.

An increased ability to generate beneficial mutations, either through spontaneous errors in DNA replication or via a lowered stringency in recombination, has been proposed as a mechanism by which mutator strains out-compete nonmutators, particularly during adaptation to novel environments (17, 39). Therefore, one possible explanation for the loss of MMR in an ancestor of Oenococcus was that a high mutation rate generated beneficial mutations during adaptation to a restrictive environment. Both O. oeni and O. kitaharae are found solely in rather unique environments, wine and composting shochu distillate residue, respectively. By contrast, neighboring MMR system-containing LAB genera, such as pediococci, lactobacilli, and leuconostocs, are found in a variety of habitats, ranging from fermented vegetables, fruits, and meats to numerous niches on and/or within animals (55). Ironically, the lack of MMR within oenococci may also be one reason for the limited number of environments in which members of the genus are found. Giraud et al. (15) showed that a mutator status in E. coli enabled the rapid adaptation and colonization of the mouse gut at a rate greater than those of WT clones. However, these same isolates exhibited a decreased fitness when transferred into a second environment, due to the accumulation of mutations deleterious to growth in the new habitat.

Clearly, it is hard to reconcile the increased burden brought about by hypermutation with the eventual success of Oenococcus as a genus. Gong et al. (16) showed that cycling between a WT and a mutant mutL allele enabled Samonella serovar Typhimurium LT7 to diversify via beneficial recombination events and/or mutations, without the burden of accumulating too many negative mutations that would eventually arise from a prolonged MMR-deficient state. The situation is different in Oenococcus, since, unlike with Salmonella serovar Typhimurium, there is no template for the repair of the MMR-deficient phenotype through spontaneous mutation due to the complete absence of mutSL. Moreover, it remains to be determined if the MMR-deficient phenotype in Oenococcus can be rescued by mutSL from another species. If such a rescue is not viable, one might speculate that the extinction of the genus Oenococcus is an eventual outcome.

Recently, Nilsson et al. (38) showed that the rate of spontaneous DNA loss per generation is 50-fold higher in a mutS mutant of Salmonella enterica than in the WT, demonstrating that extensive genome reductions can occur within a rapid evolutionary time frame in MMR-deficient hosts. It is unclear if a similar genomic loss has occurred in O. oeni, primarily because the reduction of metabolic capacity via gene loss is a common theme in the evolution of the whole Lactobacillales clade (30) and the genome size of O. oeni is roughly similar to those of other LAB (∼1.8 Mb). Regardless, given the shared evolution, small genomes, and conserved presence of mutSL in the other members of the Lactobacillales, the genus Oenococcus offers an intriguing opportunity to examine how the lack of MMR influences speciation.