Multilocus Sequence Typing for Studying Genetic Relationships among Yersinia Species

    Department of Epidemiology and Preventive Medicine, University of Maryland School of Medicine, Baltimore, Maryland 21201
    University of Louvain, Faculty of Medicine, Microbiology Unit, B-1200 Brussels, Belgium

    ABSTRACT

    The intra- and interspecies genetic relationships of 58 strains representing all currently known species of the genus Yersinia were examined by multilocus sequence typing (MLST), using sequence data from 16S RNA, glnA, gyrB, recA, and Y-HSP60 loci. Yersinia aldovae, Y. bercovieri, Y. intermedia, Y. pestis, Y. pseudotuberculosis, Y. rohdei, and Y. ruckeri were genetically more homogeneous than were Y. enterocolitica, Y. frederiksenii, Y. kristensenii, and Y. mollaretii. The MLST data concerning the genetic relatedness within and among various species of Yersinia support the idea that Y. pestis and Y. pseudotuberculosis are two lineages within the same species rather than two distinct species. Y. ruckeri is the genetically most distant species within the genus. There was evidence of O-antigen switching and genetic recombination within and among various species of Yersinia. The genetic relatedness data obtained by MLST of the four housekeeping genes and 16S RNA agreed in most, but not all, instances. MLST was better suited for determining genetic relatedness among yersiniae than was 16S RNA analysis. Some strains of Y. frederiksenii and Y. kristensenii are genetically less related to other strains within those species, compared to strains of all other species within the genus. The taxonomic standing of these strains should be further examined because they may represent currently unrecognized Yersinia species.

    INTRODUCTION

    The genus Yersinia is a member of the gamma subdivision of Proteobacteria (60), and it is grouped in the family Enterobacteriaceae, based on biochemical tests and DNA-DNA similarity studies (10). Yersiniae have undergone extensive diversification during the course of their evolution, with one Yersinia species (Yersinia pestis) becoming the deadliest bacterium ever known in human history, and other species (e.g., Y. aldovae) diverging into completely nonpathogenic organisms (46, 55). Extensive research has been conducted to characterize 3 (Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica) of the 11 currently recognized species of Yersinia; however, the remaining 8 "Y. enterocolitica-like" species (Y. frederiksenii, Y. intermedia, Y. kristensenii, Y. bercovieri, Y. mollaretii, Y. rohdei, Y. ruckeri, and Y. aldovae) have been only moderately studied because they were not clearly recognized as human pathogens (55). This situation has resulted in a marked paucity of information pertaining to the phylogenetic interrelationships of all 11 species within the genus and to the mechanisms responsible for their serological and genetic divergences.

    Initial studies examining the relatedness among Yersinia species used serotyping and other phenotypic characteristics, such as biochemical properties (biotyping), susceptibility to antibiotics (13), and phage typing (7, 44). Brenner et al. (11) subsequently introduced DNA hybridization techniques to classify Y. enterocolitica-like species and to determine the genetic relatedness among them. However, the methodology was limited because it did not provide information required to determine evolutionary relationships among yersiniae, and it was prone to yield potentially misleading results influenced by the level of gene acquisition and loss (39). Therefore, several molecular typing methodologies, such as plasmid profile analysis (22, 50), restriction fragment length polymorphism of chromosomal DNA (9), ribotyping (40, 42), sequence analysis of the 16S RNA gene (31), pulsed-field gel electrophoresis (PFGE) (43), and variable-number tandem repeat analysis (2) were subsequently applied for typing of yersiniae during epidemiological investigations and for determining genetic relatedness among Yersinia strains. Most of these approaches have good or excellent discriminatory power (e.g., PFGE was proposed as the "gold standard" for typing of Yersinia strains [34]) and are well suited for short-term epidemiological studies. However, they are less suited for long-term epidemiological studies and for determining evolutionary traits of, and phylogenetic relationships among, various strains or species (18). A relatively recently developed approach called multilocus enzyme electrophoresis (MLEE) addressed some of the shortcomings of the above-mentioned methodologies by combining excellent discriminatory power with relative clonal stability (49). Several studies (12, 17, 25, 26) have utilized MLEE to characterize yersiniae. However, problems with band resolution, and the ability of environmental conditions to alter phenotypic expression of the enzyme under study, can adversely affect the reproducibility of MLEE typing and complicate data analysis.

    Nucleotide sequence-based approaches alleviate all of the above-mentioned drawbacks of MLEE. Indeed, sequencing of the entire genome of a bacterium, with subsequent in silico analysis of the data, constitutes the ultimate approach for delineating the genetic relatedness of various bacteria, and this approach is likely to become a common strategy during future studies of bacterial evolution and virulence traits (59). However, sequencing an entire bacterial genome is a major undertaking, and therefore it is usually limited to very few strains per species. Sequencing a single gene—usually the 16S RNA gene, which has been used for the molecular analysis of many bacterial pathogens, including Yersinia (31, 32)—also is not optimal. For example, interspecies recombinations in 16S RNA genes (initially thought to be very rare) have recently been inferred (52) to occur in at least some bacterial species, which underscores the importance of not focusing on a single gene during studies to determine the phylogeny/taxonomy of a bacterial species (39). In this context, using a technique that allows sequence-based analysis of multiple loci simultaneously will provide an ideal balance between sequence-based resolution and technical feasibility. Such an approach, called multilocus sequence typing (MLST), was developed by Maiden et al. (41) to characterize meningococci, and it has rapidly been applied to the study of many other bacterial species as well, including selected species of Yersinia (1). However, MLST data pertaining to genetic relationship, phylogenetic traits, and clustering of all known species within the genus Yersinia are not currently available, and the present study is the first attempt to use MLST to examine the intra- and interspecies relationships among all of the currently described yersiniae.

    MATERIALS AND METHODS

    Bacterial strains, growth conditions, and DNA isolation. The bacterial collection contained 57 Yersinia strains representing all of the Yersinia species except Y. pestis. Sequence data for the appropriate loci from Y. pestis strain KIM (NC-004088) (62) were obtained from GenBank (http://ncbi.nlm.nih.gov) and were analyzed with the rest of our Yersinia isolates (Table 1). The isolates included Y. aldovae (two strains), Y. bercovieri (nine strains), Y. enterocolitica (five strains), Y. frederiksenii (eight strains), Y. intermedia (six strains), Y. kristensenii (eight strains), Y mollaretii (eight strains), Y. rohdei (five strains), Y. ruckeri (four strains), and Y. pseudotuberculosis (two strains). Five of these strains were ATCC isolates, and the remaining strains were from G. Wauters' and A. Sulakvelidze's bacterial strain collections. Of the five Y. enterocolitica strains, strains 8081 and YE37 belonged to biogroup 1B, strain W22708 was in biogroup 2, and strains A1074 and 1476 belonged to biogroup 1A. All of the strains were grown (30°C, overnight) in Luria-Bertani (LB) broth. Bacterial genomic DNA required for PCR amplification was extracted with the PrepMan TM Ultra minikit (Applied Biosystems, Inc., Foster City, Calif.).

    MLST genes, primer design, and DNA sequencing. Five gene loci were selected for MLST analysis, including 16S RNA, Y-HSP60 (encoding a 60-kDa heat shock protein), glnA (glutamine synthetase), gyrB (DNA gyrase B subunit), and recA (DNA repair and recombination). The primers and PCR conditions for amplifying the 16S rRNA locus have been reported previously (29). Because of the limited sequence availability of some of the Yersinia species in GenBank, the primers for the remaining four loci were designed by using a two-step approach. The first set of primers was developed by using ClustalX (35) to align the corresponding Yersinia sequences contained in GenBank and then selecting conserved regions flanking potentially variable internal fragments of the targeted genes. The primers were examined for their ability to amplify the DNA in a limited number of isolates of various Yersinia species in our collection, and the amplified fragments were sequenced. A second set of primers (Table 2) was designed based on the sequences obtained, and efforts were made to select primers that would amplify the target loci in all strains of Yersinia species examined during this study. The PCR amplification conditions for the four housekeeping gene fragments (Y-HSP60, glnA, gyrB, and recA) were as follows: 94°C for 5 min, followed by 35 amplification cycles, each consisting of sequential incubation at 94°C (45 s), 51°C (45 s), and 72°C (1 min), followed by the final incubation at 72°C for 5 min. The amplified fragments were sequenced in both directions using the BigDye Terminator cycle sequencing kit (Applied Biosystems), and an ABI 3700 DNA analyzer was used to separate the labeled fragments by size.

    MLST data analysis. Reading of trace files and assembly of contigs were performed using the Phred (19, 20) and Phrap (available at http://www.washington.edu) programs, respectively. The contigs were viewed with Consed (24), the resulting DNA sequences were trimmed by removing low-quality nucleotide sequences from the ends, and the sequences were aligned by ClustalX (35). The START (sequence type analysis and recombinational tests) program (36) was used to determine the number of polymorphic sites, the ratio of synonymous and nonsynonymous nucleotide substitutions (dN/dS), and guanine-cytosine content. The method of split decomposition implemented in the Splitstree program (30) was applied to detect recombination between the strains and species. Splitsgraphs were computed from the sequences for each individual locus, and also from a "superlocus" consisting of 1,407 bp of concatenated recA, Y-HSP60, gyrB, and glnA sequences for each isolate (in order to obtain accurate alignments, the ends of some gene sequences were trimmed, which resulted in a total length of 1,407 bp for the "superlocus"). The genetic distances between the isolates were computed from the uncorrected distance matrices ("P") of pairwise differences (the total number of nucleotides that differed between the two sequences divided by the total number of nucleotides compared). The clustering of strains/species was determined with the PAUP program (Sinauer, version 3.1) using unrooted neighbor-joining trees based on maximum-likelihood criteria. Bootstrapping algorithms from Splitstree (30) were used to estimate the robustness of the phylogenetic analyses.

    Nucleotide sequence accession numbers. The DNA sequences of the 16S RNA, recA, Y-HSP60, gyrB, and glnA gene loci have been deposited in GenBank under accession numbers AY332802 though AY333069.

    RESULTS

    16S RNA sequence variations in yersiniae. The 16S RNA gene fragments of all the isolates were conserved: 55 (ca. 95%) of the 58 Yersinia strains we analyzed had pairwise distances of <0.062 (Fig. 1). The three exceptions were Y. frederiksenii WE 83/02, Y. intermedia WA 11/94, and Y. kristensenii WA 948, which had larger genetic distances, up to 0.132. The 16S RNA sequences of four species (Y. pestis, Y. pseudotuberculosis, Y. aldovae, and Y. ruckeri), did not vary; i.e., a single 16S RNA allele type was identified in each species (Table 1). The isolates of two other species, Y. intermedia and Y. rohdei, were very closely related to the other isolates of the same species; i.e., the distances between conspecific isolates were <0.008. Similarly, eight of nine Y. bercovieri isolates had maximal distances of <0.005, and the ninth isolate, ATCC 43970, grouped in a distinct cluster together with three Y. frederiksenii strains (Fig. 1). The two Y. pseudotuberculosis strains we examined were also very closely related, and they clustered, together with Y. pestis KIM, in a distinct genetic cluster (Fig. 1).

    The remaining four species (Y. enterocolitica, Y. frederiksenii, Y. kristensenii, and Y. mollaretii) appeared to be genetically more variable than the above-mentioned yersiniae. Two groups of Y. enterocolitica strains were identified, one containing three strains (WE 22708, A1074, and 1476) and one composed of two strains (8081 and YE 37). The two groups had within-group differences of <0.01 and between-group differences of ca. 0.025 (Fig. 1). Similarly, two groups were identified among the seven Y. frederiksenii strains we examined, one composed of four strains (ATCC 33641, WA 933, 7175, and WS 52/02) and one containing three strains (867, CNY 867, and WA 935). As noted above, the eighth Y. frederiksenii strain (WE 83/02) was a clear "outsider." The pairwise distances within the group containing four isolates were <0.009. Five isolates of Y. kristensenii (490, 105, WE 180/98, WS 45/98, and WE 414/93) were closely related and had pairwise distances of <0.009. However, Y. kristensenii WS 052 and ATCC 33638 were scattered across the 16S RNA maximum-likelihood tree (Fig. 1), and the distances to their nominal conspecifics ranged from 0.011 to 0.049. Y. kristensenii WA 948 was the most divergent strain among all of the Yersinia strains we analyzed, with the exception of the Y. ruckeri strains (Fig. 1). Six of the eight Y. mollaretii strains (including the ATCC 43969 type strain, also known as CNY 7263) were grouped together in a tight cluster (<0.003), and the remaining two strains, Y. mollaretii WS 43/94 and WA 309, appeared to be more closely related to Y. enterocolitica and Y. frederiksenii, respectively, than to the other strains of Y. mollaretii.

    Some of the Yersinia strains had identical 16S RNA sequences, e.g., Y. aldovae CNY 7112 and 35236 and Y. kristensenii WE 414/93, Y. intermedia WS 58/02 and Y. kristensenii 490, and Y. pestis KIM and Y. pseudotuberculosis CDC801-84 (Fig. 1). Eight strains did not cluster with their conspecifics: Y. frederiksenii WE 83/02, Y. intermedia WA 11/94, Y. kristensenii WA 948, Y. mollaretii WA 309 and WS 43/94, Y. bercovieri ATCC 43970, and Y. kristensenii WS 052 and ATCC 33638.

    Y-HSP60, glnA, recA, and gyrB sequence variations in Yersinia. All of the four loci were polymorphic: glnA, gyrB, recA, and Y-HSP60 had 109, 105, 92, and 74 polymorphic sites, respectively (Table 2). The calculated dN/dS ratios for those housekeeping genes ranged from 0.0052 for the recA locus to 0.0378 for the gyrB locus. The G+C contents of the loci ranged from 46.81% for the gyrB locus to 57.84% for the 16S RNA locus (Table 2).

    Two different allelic designations were assigned to each sequence, one based on the nucleotide sequence (NT) and the other on the translated amino acid sequence (AT). Thirty-six, 35, 37, and 32 NTs were identified for the glnA, gyrB, recA, and Y-HSP60 genes, respectively (Table 1). With the exception of Y. pestis and Y. pseudotuberculosis, identical NTs were observed only in strains within the same species, and strains possessing identical NTs for all of the loci were very rare (e.g., Y. bercovieri WA 17/96 and WS 19/96 had identical alleles for all four loci). One exception, where the strains of two different species had the same NTs, was Y. pestis KIM and the two Y. pseudotuberculosis isolates: these three strains shared identical ATs for each housekeeping gene locus examined. Translation of the nucleotide sequences into amino acid sequences resulted in fewer alleles (ATs) than NTs, which suggests that most of the nucleotide substitutions were silent (Table 1). For example, the 37 NTs for the recA locus translated into only 4 ATs. Overall, there were 9, 15, 4, and 14 ATs for glnA, gyrB, recA, and Y-HSP60, respectively.

    Some isolates of several Yersinia species shared identical ATs at one or more of the four housekeeping loci. As noted above, isolates of Y. pestis and Y. pseudotuberculosis shared ATs at all four loci. For glnA, Y. aldovae and Y. intermedia shared AT2, Y. enterocolitica and Y. kristensenii shared AT3, and Y. frederiksenii and Y. bercovieri shared AT3. For gyrB, Y. bercovieri and Y. mollaretii shared AT3, and Y. enterocolitica, Y. pseudotuberculosis, and Y. pestis shared AT8. For Y-HSP60, Y. pseudotuberculosis, Y. pestis, and Y. enterocolitica shared AT2, Y. intermedia and Y. frederiksenii shared AT4, Y. kristensenii and Y. frederiksenii shared AT1, and one strain each of Y. kristensenii and Y. mollaretii shared AT7, which was commonly found in the Y. bercovieri strains. The smallest number of ATs was observed for recA. The number of recA-based ATs identified was less than the number of currently known species of Yersinia; therefore, not surprisingly, the recA-based ATs were commonly shared across the species. For example, 28 isolates representing six Yersinia species possessed recA AT2, and 12 strains of Y. frederiksenii and Y. mollaretii had recA AT3. In contrast, with the exception of one possible match with Y. aldovae CNY 7112, Y. ruckeri's recA AT4 was not found in any other species. Moreover, the data obtained during our analysis of the other three housekeeping genes of Y. ruckeri indicated that their ATs also were unique to Y. ruckeri, i.e., they were not shared with any strains of the other Yersinia species.

    Genetic relationships among yersiniae. Inter- and intraspecies relationships among various yersiniae were examined by analyzing a 16S RNA-based maximum-likelihood tree (Fig. 1) and an unrooted phylogram (Fig. 2) constructed from the 1,407-bp concatenated sequences of all four housekeeping MLST loci. The 16S RNA sequences were not included in the concatenated locus, because multiple copies of the gene exist in various Yersinia strains (e.g., 6 and 7 copies in Y. pestis strains CO92 and KIM, respectively). The availability of the two trees enabled us to compare the genetic relatedness data obtained by 16S RNA sequencing with those obtained by MLST analysis of a concatenated tree containing four housekeeping gene loci.

    The concatenated tree was constructed based on the 39 isolates for which we had complete sequence data for all four loci (Fig. 2). The intraspecies genetic distances for seven species (Y. bercovieri, Y. enterocolitica, Y. intermedia, Y. mollaretii, Y. pseudotuberculosis, Y. rohdei, and Y. ruckeri) were <0.02. However, in two species (Y. kristensenii and Y. frederiksenii), all of the isolates did not cluster together. For example, four isolates of Y. kristensenii (WE 180/98, WS 45/98, 105, and ATCC 33638) diverged from each other by <0.01, while the other two isolates, WE 414/93 and WS 052, differed from the others by about 0.04 and 0.09, respectively (Fig. 2). The pairwise distances for four strains of Y. frederiksenii were <0.04, and they were >0.06 for the remaining isolates (Fig. 2). Only one Y. pestis isolate was analyzed; therefore, the genetic distances among strains of that species could not be determined. However, data obtained from the 16S RNA- and MLST-based trees (Fig. 1 and 2, respectively) indicated that the single Y. pestis strain we examined clustered tightly with the Y. pseudotuberculosis strains. Furthermore, as noted above, both species shared the same 16S RNA allele (NT25) and all four ATs (AT8, AT8, AT2, and AT2 for glnA, gyrB, recA, and Y-HSP60, respectively [Table 1]).

    The genetic distances between species ranged from 0.062 to 0.134, with the exception of Y. pestis and Y. pseudotuberculosis, for which it was <0.01. Other species that were closely related were Y. aldovae and Y. intermedia (pairwise genetic distances between isolates ranged from 0.066 to 0.068) and Y. bercovieri and Y. mollaretii (pairwise genetic distances ranged from 0.062 to 0.066). The Y. ruckeri isolates exhibited the largest genetic distances (0.116 to 0.134) from the other Yersinia species.

    Comparison of the 16S RNA and MLST data. The data obtained by 16S RNA analysis (Fig. 1) and MLST (Fig. 2) of 39 Yersinia strains placed 35 of the isolates in their nominal species-specific group. The four discordant strains were Y. kristensenii WS 052, Y. kristensenii WA 948, Y. frederiksenii WA 935, and Y. bercovieri ATCC 43970. Y. kristensenii WS 052 appeared to be more closely related to the Y. enterocolitica group (based on the 16S RNA analysis [Fig. 1]) and to the Y. enterocolitica, Y. frederiksenii, and Y. rohdei groups (based on the MLST analysis [Fig. 2]) than to other Y. kristensenii strains. In addition, according to the 16S RNA analysis (Fig. 1), Y. kristensenii WA 948 discordantly clustered with Y. intermedia WA 11/94—another "outsider" strain. The former strain also was observed, by MLST, to be discordantly but not tightly clustered with Y. bercovieri (Fig. 2). Based on the 16S RNA analysis (Fig. 1), Y. frederiksenii WA 935 was genetically distinct and not closely related to other Y. frederiksenii strains; however, it clustered with two other Y. frederiksenii strains that formed one of the two Y. frederiksenii clusters on the concatenated tree (Fig. 2). Similarly, based on the 16S RNA analysis (Fig. 1), Y. bercovieri ATCC 43970 did not cluster with the other Y. bercovieri isolates; however, it clustered tightly with four other Y. bercovieri strains on the MLST tree (Fig. 2).

    Evidence for O-antigen switching and genetic recombination in Yersinia. Several strains with identical or closely related genetic backgrounds had distinct O antigens; e.g., although Y. intermedia WS 11/01 and WE 24/02 were very closely related (Fig. 1 and 2), they had distinct O antigens (O:48 and O:14, respectively) (Table 1). Also, in a few cases, O antigens were shared by strains of species that were not closely related; e.g., Y. bercovieri W601 shared the O:8 antigen with the genetically distant Y. enterocolitica 8081 and YE 37 (Table 1; Fig. 1 and 2).

    The Splitstree analysis revealed parallelograms, for each individual locus and for the concatenated sequences, among strains within and among various Yersinia species. Parallelograms in the splitsgraph were accepted as evidence for parallel changes only if (i) the fit values were above 95% (indicating that most of the variation was accounted for in the analysis) and (ii) the bootstrap values on adjacent sides were greater than 90%. The criteria were never met for data sets containing the sequences for all the strains, presumably because the data from the genetically diverse species were too complex. However, the criteria were routinely met when small subsets of the sequences (i.e., simpler data sets) were analyzed. An example of such an analysis for a subset of Y. enterocolitica, Y. frederiksenii, Y. intermedia, Y. kristensenii, and Y. mollaretii strains is presented in Fig. 3. As shown in Fig. 3, the parallelograms often had adjacent sides with very high bootstrap values (e.g., all bootstrap values for the gyrB and recA loci had a maximum possible value of 100), and the fit values for all loci were at their maximum possible value of 100.

    DISCUSSION

    Genetic relatedness among Yersinia species. MLST is a relatively new approach that has been used to determine the genetic relatedness among strains of various bacterial species and, less commonly, among various species. The three most intensively studied, pathogenic Yersinia species (Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica) have been characterized by MLST (1). In that study, no sequence diversity was found in any of the six genes among the Y. pestis isolates examined, and the alleles identified were identical or nearly identical to alleles in Y. pseudotuberculosis. Based on these observations, the authors concluded that Y. pestis is an extremely homogeneous species that evolved from Y. pseudotuberculosis shortly before the first known pandemic of human plague. Other investigators have also suggested close genetic relatedness between these two species (1, 27, 45). In agreement with the above-cited reports, Y. pestis and Y. pseudotuberculosis were found to be very closely related in our study too, based on all criteria examined, including 16S RNA- and MLST-based clustering (Fig. 1 and 2, respectively) and their allelic profiles (both species shared the same 16S RNA allele and all four ATs [Table 1]).

    In addition to Y. pestis and Y. pseudotuberculosis, five Yersinia species (Y. aldovae, Y. bercovieri, Y. intermedia, Y. rohdei, and Y. ruckeri) were genetically fairly homogeneous. For example, based on the sequence analysis of the 16S RNA gene loci, all four Y. ruckeri strains and both of the Y. aldovae strains we examined were identical for each species, even though some of the strains were isolated in different years (e.g., Y. ruckeri WS 34/93 and WS 10/95 were isolated in Belgium in 1993 and 1995, respectively) or had different serotypes (e.g., the serotypes of Y. aldovae 35236 and CNY 7112 were O:17 and O:21, respectively). The Y. intermedia and Y. rohdei strains were also genetically homogeneous. The distances between conspecific isolates were <0.008, and some strains isolated ca. 6 years apart on different continents (e.g., Y. rohdei WA 9/90 and WA 14/96, which had the same serotype, O:76, were isolated in Germany in 1990 and in New Zealand in 1996, respectively) were identical based on their 16S RNA sequences. The two identical strains of Y. bercovieri, WA 17/96 and WS 19/96, were isolated from humans in 1996. They had identical alleles for all five loci (Table 1), and they shared the O:58,73 antigen; however, they were isolated from geographically distinct locations (New Zealand and Belgium, respectively).

    The genetic homogeneity of Y. aldovae, Y. intermedia, Y. rohdei, and Y. ruckeri has been suggested by other investigators (22, 48), and our observations are in agreement with those earlier reports. For example, our 16S RNA and MLST data regarding the genetic homogeneity of Y. ruckeri are in agreement with a previous report (22) in which Y. ruckeri strains isolated from a wide variety of sources were shown to be clonal, based on their plasmid profile and ribotyping pattern analysis, and one plasmid type and ribotype predominated. In addition, MLEE analysis (48) of Y. ruckeri strains indicated that they were primarily of one electrophoretic type. In contrast, Y. enterocolitica, Y. frederiksenii, Y. kristensenii, and Y. mollaretii were genetically more heterogeneous, which is also in agreement with previous reports (12, 17, 57).

    Comparison of 16S RNA analysis and MLST data regarding genetic relatedness among Yersinia species. In most, but not all, instances, the MLST and 16S RNA data correlated well with each other, and they agreed with the species' designations determined by biochemical tests. The MLST and 16S RNA data also correlated well in a few cases (e.g., Y. kristensenii WS 052 and WA 948) where the results of both analyses did not agree with the biochemical designation of the species; i.e., according to the 16S RNA and MLST analyses, the two strains mentioned above were clear outsiders in relation to other strains of the same species. Therefore, as discussed below, their taxonomic designations should be reexamined.

    In instances where the 16S RNA and MLST data did not correlate with each other, MLST was more consistent with the species' biochemical designations than was 16S RNA analysis. For example, with the exception of one isolate (Y. bercovieri strain ATCC 43970), all of the Y. bercovieri strains were either identical or very closely related, based on their 16S RNA sequence analysis (Fig. 1). The ATCC 43970 strain (also known as CNY 7506) was a clear outsider, and it appeared to be more closely related to other Yersinia species (e.g., Y. frederiksenii) than to the other Y. bercovieri strains we analyzed. Similarly, according to the 16S RNA data (Fig. 1), Y. kristensenii ATCC 33638 (also known as CIP 80-30) did not cluster with the other Y. kristensenii strains. Both ATCC strains (Y. bercovieri ATCC 43970 and Y. kristensenii ATCC 33638) are the type isolates of their respective species; therefore, our observation that they were genetically distinct (according to the 16S RNA analysis) from other strains of their respective species was puzzling. However, when five of the nine Y. bercovieri strains (including ATCC 43970) and seven of the eight Y. kristensenii strains (including ATCC 33638) were analyzed by MLST using concatenated trees based on Y-HSP60, glnA, gyrB, and recA loci, both type strains correctly clustered with isolates of their respective species (Fig. 2). Another example of MLST being better than 16S RNA sequencing for determining phylogenetic identity is Y. frederiksenii WA 935. According to 16S RNA analysis, that strain was genetically distinct from other Y. frederiksenii strains (Fig. 1), but it correctly clustered with other Y. frederiksenii strains on the concatenated MLST tree (Fig. 2).

    16S RNA sequencing has been and still is commonly used to identify various species, and its ability to yield reproducible and accurate information is commonly accepted. However, an increasing body of data suggests that 16S RNA does not always ideally identify/resolve strains on a species level. For example, a recent study (15) of Y. frederiksenii reported that gyrB was a better phylogenetic marker than that species' 16S RNA sequence. Also, as noted above, interspecies recombinations in 16S RNA genes, which initially were thought to be rare, have recently been suggested (52) to occur in at least some bacterial species, and such genetic exchanges may affect the determination of the phylogeny/taxonomy of bacterial species in which the changes have occurred (39). Our data support these observations, and they suggest that (i) although 16S RNA sequences are reasonably good phylogenetic markers for differentiating Yersinia species, genetic relatedness data obtained at least by the partial sequencing of this locus may not clearly distinguish between some members of this genus, and (ii) MLST is better suited for studying genetic relatedness among yersiniae than is 16S RNA-based analysis. On a more general level, our data support the idea that MLST can be an invaluable tool during continued refinement of the phylogenetic relationships among enteric bacteria (3).

    16S RNA analysis, MLST, and Yersinia taxonomy. In the genus Yersinia (as in other bacteria), species have been identified mainly according to certain biochemical tests. The introduction of DNA-based approaches has revolutionized bacterial taxonomy and has helped assign species designations to numerous bacteria, including yersiniae (11). 16S RNA analysis has been particularly useful in that regard, and it has become a molecular standard for identification of many bacterial species, including uncultivable bacteria (29). However, the question about the minimal percent difference in 16S RNA sequences needed to differentiate bacterial species has not yet been resolved, and the issue of whether a 2% or larger difference is required to assign strains to different species is still a matter of debate (6, 14). The results of our present analysis of various Yersinia species suggest that MLST can be useful in that regard and that MLST is better than 16S RNA analysis for determining the taxonomic designations of the yersiniae.

    The MLST data obtained during our study were based on four housekeeping gene loci that are generally considered to be conserved; therefore, a selection of conserved loci would be expected to represent the minimum differences between homologous loci in two genomes. Relatively few strains of the various Yersinia species were analyzed during our study, and therefore it is difficult to state with confidence whether there are "rules of thumb" for identifying the genetic distances between isolates of different Yersinia species. However, our data provide some preliminary insight into this issue. For example, based on our MLST data for seven Yersinia species (Y. bercovieri, Y. enterocolitica, Y. intermedia, Y. mollaretii, Y. pseudotuberculosis, Y. rohdei, and Y. ruckeri), the intraspecies genetic distances were <0.02, while the interspecies distances were >0.06. From these data, we observed a bimodal distribution in which intraspecies distances were in the first mode and interspecies distances were in the second mode. Thus, there is the potential to separate species using a bimodal distribution and a suitable statistic such as Giacomelli's dip statistic (23).

    Several strains could not be distinguished on a species level by the partial sequencing of the 16S RNA gene but were identified by MLST, which suggests that MLST, but not 16S RNA analysis, has sufficient resolving ability to unambiguously differentiate the isolates of the closely related Yersinia species. Additional studies are required to fine-tune and validate the cutoff values for interspecies and intraspecies genetic distances, and their results may have a profound impact on clarifying the taxonomic classification of various bacterial species (53). Indeed, MLST already has been proposed (3) as a useful tool for refining the taxonomy and phylogenetic relationships among enteric bacteria.

    We observed four exceptions to the above-mentioned bimodal distribution of the interspecies and intraspecies genetic distances between Yersinia isolates. The first two exceptions are Y. pestis and Y. pseudotuberculosis strains, which had genetic distances similar to those of intraspecies genetic distances. In addition, the same strains also (i) had the same 16S RNA allele (NT25), (ii) had identical ATs for all four housekeeping genes we analyzed (Table 1), and (iii) clustered tightly, as determined by 16S RNA sequencing (Fig. 1) and MLST (Fig. 2). The genomes of Y. pestis and Y. pseudotuberculosis are known to be very similar, based on the results of (i) DNA hybridization studies (8), (ii) 16S RNA analyses (56), and (iii) MLST studies (1), which has led to the hypothesis that the two species are actually two subgroups of the same species. Our results are in agreement with those previous reports, and they support the idea that Y. pestis and Y. pseudotuberculosis are two lineages within the same species rather than two distinct species. The third and fourth exceptions are Y. kristensenii and Y. frederiksenii, respectively. Some strains of those species (e.g., Y. kristensenii WA 948 and WS 052, and Y. frederiksenii WA 935) were separated from other strains of their respective species by genetic distances larger than those observed for other strains within the same species (i.e., the genetic distances were >0.06). Furthermore, on the maximum-likelihood trees, the same strains were not closely related to other strains of their respective species, based on the 16S RNA (Fig. 1) and concatenated MLST (Fig. 2) analyses. Y. frederiksenii has been shown (15-17, 33), by the results of comparative 16S RNA and gyrB sequence analyses, DNA hybridization studies, and MLEE analysis, to include strains that may represent a new subline within the genus Yersinia which is most closely related to Y. frederiksenii hybridization group 2 (unnamed genomospecies 2) and is biochemically indistinguishable from typical Y. frederiksenii (Y. frederiksenii genomospecies 1). Based on our 16S RNA sequencing and MLST data (Fig. 1 and 2), the Y. frederiksenii strains in our collection grouped in distinct clusters, which provides additional support for the idea that the species contains multiple subgroups, some of which may represent unknown species in Yersinia. Also, our MLST analyses (Fig. 2) identified at least one genetically atypical Y. frederiksenii isolate (WA 935) that was a clear outsider with regard to the other Y. frederiksenii isolates we examined.

    Dolina and Peduzzi (17) found that some of the Y. kristensenii strains in their collection were genetically more closely related to Y. mollaretii than to other strains of Y. kristensenii, as determined by MLEE analysis. Our MLST data obtained during our study of the atypical Y. kristensenii strains (e.g., Y. kristensenii WA 948 and WS 052) in our collection are consistent with their report. The number of Y. kristensenii and Y. frederiksenii strains analyzed during our study was relatively small, and we detected only a few atypical strains; thus, it is not clear how common those strains are in nature. However, our data support the idea that such strains do exist, and they emphasize the need for additional MLST studies to elucidate the prevalence of such strains and to delineate their taxonomic classification, since they may represent a new species within the genus Yersinia.

    Y. ruckeri (the etiologic agent of red mouth disease in fish) was a clear outsider with regard to all other yersiniae, based on (i) 16S RNA analysis (Fig. 1) and (ii) analysis of the concatenated sequences (Fig. 2). Moreover, the ATs of the four Y. ruckeri strains in our collection were unique to that species; i.e., AT7 (for glnA), AT15 and AT9 (for gyrB), AT4 (for recA), and AT11 (for Y-HSP60) were not found in any strains of any other Yersinia species we examined. The taxonomic status of Y. ruckeri has been controversial since its initial classification in 1978 (21), and data reported by several investigators (5, 17, 38, 48) suggest that Y. ruckeri should not be classified in the genus Yersinia. Our 16S RNA and MLST data suggest that Y. ruckeri is the genetically most distant species within the genus and that the taxonomic standing of Y. ruckeri may need to be reevaluated.

    Correlation between the genetic backgrounds of various Yersinia species and their O antigens. Many O antigens are shared among various species of Yersinia; e.g., Y. frederiksenii shares several O antigens with Y. enterocolitica (57), and many Y. kristensenii strains are agglutinable in the Wauters scheme (58). However, the genetic relatedness of Yersinia strains with common O antigens has never been rigorously examined, and the mechanisms responsible for this phenomenon are currently not well understood. During the present study we identified two Y. mollaretii strains (WA 134/92 and WE 302/92) and one Y. bercovieri strain (W601) that shared the O:3 and O:8 antigens (51, 61), respectively, with Y. enterocolitica (Table 1). Y. mollaretii strains with Y. enterocolitica O antigens have been described previously (4), and our data provide additional evidence for the existence of such strains. Genetically, the two Y. mollaretii strains with the O:3 antigen were closely related to one another, according to their 16S RNA sequences (Fig. 1) and their shared NTs (Table 1). However, both strains were genetically distinct from all of the Y. enterocolitica strains we examined (Table 1 and Fig. 1). Similarly, Y. bercovieri W601 was not closely related to the Y. enterocolitica strains, including the two strains (8081 and YE 371) with which it shared the O:8 antigen (Fig. 1 and 2). These observations suggest that O-antigen switching, which has been proposed (54) to have occurred among several other bacteria, may have also occurred between at least some Yersinia species.

    Evidence of intraspecies and interspecies genetic recombination in Yersinia. The START analysis of our MLST data revealed a noticeable difference in the G+C contents of the genes we analyzed (Table 2). The G+C content differed by as much as 11% among the various loci; e.g., the mean G+C contents for the gyrB and 16S RNA loci were 47% and 58%, respectively. The backbone G+C content of the Yersinia genome is estimated to range from 46 to 48% (5), and with the exception of the 16S RNA loci, which had a higher G+C content, the remaining loci we analyzed were within this typical G+C range for Yersinia. Furthermore, species-specific differences in the G+C content were not observed for any of the loci we examined; i.e., there was no apparent evidence that the loci were acquired from distinct ancestors.

    As an additional approach to detect recombinational changes, we used the Splitstree analysis, utilizing small subsets of randomly selected strains of various Yersinia species (simultaneously analyzing many strains overwhelmed the program, in agreement with what has been reported previously for other bacteria [37, 47]). The observation of parallelograms during Splitstree analysis indicates that recombination played a role in the evolution of the gene under examination, and we observed such parallelograms during our analyses (one example of such analyses for each of the four housekeeping genes is presented in Fig. 3). The bootstrapping values were greater than 90% (Fig. 3)—in many cases, the bootstrapping values were at their maximum of 100% and the fit values were the maximum possible value of 100—indicating robust results based on multiple genetic changes. We consider the possibility that these multiple changes resulted from multiple, independent, recurrent mutations to be very small, particularly since nearly all of the nucleotide changes failed to result in amino acid changes (Table 1). Thus, the parallel changes are likely to be the result of recombination rather than independent recurrent mutations. Also, the dN/dS ratios for all loci were <1 (Table 2), suggesting that there is no selective pressure on nucleotide changes in these genes.

    Parallelograms were also formed during our Splitstree analysis of the 16S RNA loci (data not shown) in strains within the same species and in strains of various Yersinia species. Interspecies recombinations in 16S RNA genes have recently been proposed (52) to occur in some non-Yersinia species, and our data suggest that the same phenomenon also may occur among the yersiniae. However, caution is required in interpreting these data, because a relatively small subset of strains was analyzed, and the number of nucleotide differences between the sequences is very small. Nevertheless, our data suggest that recombination in the 16S RNA genes of various Yersinia species is a possibility, and they indicate that additional studies in that direction are warranted. Such interspecies recombinations among the yersiniae could have important practical implications; e.g., they may complicate the development of 16S RNA-targeted tools to detect and identify various Yersinia species. These observations also support the idea (28) that gene trees based on a single gene (including conserved genes, such as those for 16S RNA) are not likely to permit accurate determination of the evolutionary history of the species. Therefore, analysis of several loci of the bacterial chromosome (e.g., by using MLST) is required for a rigorous understanding of the evolutionary history of yersiniae and for the development of advanced tools capable of accurately identifying and differentiating Yersinia species.

    ACKNOWLEDGMENTS

    This study was supported by a BTEP 10/ISTC G597 grant from the Department of Health and Human Services Biotechnology Engagement Program (to A.S.) and a 1R21 AI055660-01A1 grant from the National Institutes of Health (to A.S.).

    REFERENCES

    Achtman, M., K. Zurth, G. Morelli, G. Torrea, A. Guiyoule, and E. Carniel. 1999. Yersinia pestis, the cause of plague, is a recently emerged clone of Yersinia pseudotuberculosis. Proc. Natl. Acad. Sci. USA 96:14043-14048.

    Adair, D. M., P. L. Worsham, K. K. Hill, A. M. Klevytska, P. J. Jackson, A. M. Friedlander, and P. Keim. 2000. Diversity in a variable-number tandem repeat from Yersinia pestis. J. Clin. Microbiol. 38:1516-1519.

    Ahmad, S., W. G. Weisburg, and R. A. Jensen. 1990. Evolution of aromatic amino acid biosynthesis and application to the fine-tuned phylogenetic positioning of enteric bacteria. J. Bacteriol. 172:1051-1061.

    Aleksic, S. 1995. Occurrence of Y. enterocolitica antigens O:3, O:9 and O:8 in different Yersinia species, their corresponding H antigens and origin. Contrib. Microbiol. Immunol. 13:89-92.

    Aleksic, S., and J. Bockemühl. 1999. Yersinia and other Enterobacteriaceae, p. 483-496. In P. R. Murray, E. J. Baron, M. A. Pfaller, F. C. Tenover, and R. H. Yolken (ed.), Manual of clinical microbiology, 7th ed. American Society for Microbiology, Washington, D.C.

    Amann, R. I., W. Ludwig, and K. H. Schleifer. 1995. Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbiol. Rev. 59:143-169.

    Baker, P. M., and J. J. Farmer III. 1982. New bacteriophage typing system for Yersinia enterocolitica, Yersinia kristensenii, Yersinia frederiksenii, and Yersinia intermedia: correlation with serotyping, biotyping, and antibiotic susceptibility. J. Clin. Microbiol. 15:491-502.

    Bercovier, H., H. H. Mollaret, J. M. Alonso, J. Brault, G. R. Fanning, A. G. Steigerwalt, and D. J. Brenner. 1980. Intra- and interspecies relatedness of Yersinia pestis by DNA hybridization and its relationship to Yersinia pseudotuberculosis. Curr. Microbiol. 4:225-229.

    Blumberg, H. M., J. A. Kiehlbauch, and I. K. Wachsmuth. 1991. Molecular epidemiology of Yersinia enterocolitica O:3 infections: use of chromosomal DNA restriction fragment length polymorphisms of rRNA genes. J. Clin. Microbiol. 29:2368-2374.

    Bottone, E. J. 1997. Yersinia enterocolitica: the charisma continues. Clin. Microbiol. Rev. 10:257-276.

    Brenner, D. J., A. G. Steigerwalt, D. P. Falxo, R. E. Weaver, and G. R. Fanning. 1976. Characterization of Yersinia enterocolitica and Yersinia pseudotuberculosis by deoxyribonucleic acid hybridization and by biochemical reactions. Int. J. Syst. Bacteriol. 26:180-194.

    Caugant, D. A., S. Aleksic, H. H. Mollaret, R. K. Selander, and G. Kapperud. 1989. Clonal diversity and relationships among strains of Yersinia enterocolitica. J. Clin. Microbiol. 27:2678-2683.

    Cavalcanti, Y. V., N. C. Leal, and A. M. De Almeida. 2002. Typing of Yersinia pestis isolates from the state of Ceara, Brazil. Lett. Appl. Microbiol. 35:543-547.

    Dang, H., and C. R. Lovell. 2000. Bacterial primary colonization and early succession on surfaces in marine waters as determined by amplified rRNA gene restriction analysis and sequence analysis of 16S rRNA genes. Appl. Environ. Microbiol. 66:467-475.

    Demarta, A., S. De Respinis, M. Dolina, and R. Peduzzi. 2004. Molecular typing of Yersinia frederiksenii strains by means of 16S rDNA and gyrB gene sequence analyses. FEMS Microbiol. Lett. 238:423-428.

    Dolina, M., V. Gaia, and R. Peduzzi. 1995. Molecular typing of Yersinia frederiksenii strains by means of ribotyping and DNA-DNA hybridization. Contrib. Microbiol. Immunol. 13:140-144.

    Dolina, M., and R. Peduzzi. 1993. Population genetics of human, animal, and environmental Yersinia strains. Appl. Environ. Microbiol. 59:442-450.

    Enright, M. C., and B. G. Spratt. 1999. Multilocus sequence typing. Trends Microbiol. 7:482-487.

    Ewing, B., and P. Greene. 1998. Base-calling of automated sequencer traces using Phred. I. Accuracy assessment. Genome Res. 8:175-185.

    Ewing, B., and P. Greene. 1998. Base-calling of automated sequencer traces using Phred. II. Error probabilities. Genome Res. 8:186-194.

    Ewing, W. H., A. J. Ross, D. J. Brenner, and G. R. Fanning. 1978. Yersinia ruckeri sp. nov., the redmouth (RM) bacterium. Int. J. Syst. Bacteriol. 28:37-44.

    Garcia, J. A., L. Dominguez, J. L. Larsen, and K. Pedersen. 1998. Ribotyping and plasmid profiling of Yersinia ruckeri. J. Appl. Microbiol. 85:949-955.

    Giacomelli, F., J. Wiener, J. B. Kruskal, J. V. Pomeranz, and A. V. Loud. 1971. Subpopulations of blood lymphocytes demonstrated by quantitative cytochemistry. J. Histochem. Cytochem. 19:426-433.

    Gordon, D., C. Abajian, and P. Green. 1998. Consed: a graphical tool for sequence finishing. Genome Res. 8:195-202.

    Goullet, P., and B. Picard. 1988. Characterization of Yersinia enterocolitica, Y. intermedia, Y. aldovae, Y. frederiksenii, Y. kristensenii and Y. pseudotuberculosis by electrophoretic polymorphism of acid phosphatase, esterases, and glutamate and malate dehydrogenases. J. Gen. Microbiol. 134:317-325.

    Goullet, P., and B. Picard. 1984. Distinctive electrophoretic and isoelectric focusing patterns of esterases from Yersinia enterocolitica and Yersinia pseudotuberculosis. J. Gen. Microbiol. 130:1471-1480.

    Hinchliffe, S. J., K. E. Isherwood, R. A. Stabler, M. B. Prentice, A. Rakin, R. A. Nichols, P. C. Oyston, J. Hinds, R. W. Titball, and B. W. Wren. 2003. Application of DNA microarrays to study the evolutionary genomics of Yersinia pestis and Yersinia pseudotuberculosis. Genome Res. 13:2018-2029.

    Holmes, E. C., R. Urwin, and M. C. Maiden. 1999. The influence of recombination on the population structure and evolution of the human pathogen Neisseria meningitidis. Mol. Biol. Evol. 16:741-749.

    Hugenholtz, P., C. Pitulle, K. L. Hershberger, and N. R. Pace. 1998. Novel division level bacterial diversity in a Yellowstone hot spring. J. Bacteriol. 180:366-376.

    Huson, D. H. 1998. SplitsTree: analyzing and visualizing evolutionary data. Bioinformatics 14:68-73.

    Ibrahim, A. 1995. Genetic diversity among Yersinia enterocolitica strains as revealed by sequence analysis of the 16S rRNA gene. Contrib. Microbiol. Immunol. 13:277-280.

    Ibrahim, A., B. M. Goebel, W. Liesack, M. Griffiths, and E. Stackebrandt. 1993. The phylogeny of the genus Yersinia based on 16S rDNA sequences. FEMS Microbiol. Lett. 114:173-177.

    Ibrahim, A., W. Liesack, A. G. Steigerwalt, D. J. Brenner, E. Stackebrandt, and R. M. Robins-Browne. 1997. A cluster of atypical Yersinia strains with a distinctive 16S rRNA signature. FEMS Microbiol. Lett. 146:73-78.

    Iteman, I., A. Guiyoule, and E. Carniel. 1996. Comparison of three molecular methods for typing and subtyping pathogenic Yersinia enterocolitica strains. J. Med. Microbiol. 45:48-56.

    Jeanmougin, F., J. D. Thompson, M. Gouy, D. G. Higgins, and T. J. Gibson. 1998. Multiple sequence alignment with Clustal X. Trends Biochem. Sci. 23:403-405.

    Jolley, K. A., E. J. Feil, M. S. Chan, and M. C. Maiden. 2001. Sequence type analysis and recombinational tests (START). Bioinformatics 17:1230-1231.

    Kotetishvili, M., O. C. Stine, A. Kreger, J. G. Morris, Jr., and A. Sulakvelidze. 2002. Multilocus sequence typing for characterization of clinical and environmental Salmonella strains. J. Clin. Microbiol. 40:1626-1635.

    Kwaga, J., J. O. Iversen, and V. Misra. 1992. Detection of pathogenic Yersinia enterocolitica by polymerase chain reaction and digoxigenin-labeled polynucleotide probes. J. Clin. Microbiol. 30:2668-2673.

    Lan, R., and P. R. Reeves. 2001. When does a clone deserve a name A perspective on bacterial species based on population genetics. Trends Microbiol. 9:419-424.

    Lobato, M. J., E. Landeras, M. A. Gonzalez-Hevia, and M. C. Mendoza. 1998. Genetic heterogeneity of clinical strains of Yersinia enterocolitica traced by ribotyping and relationships between ribotypes, serotypes, and biotypes. J. Clin. Microbiol. 36:3297-3302.

    Maiden, M. C., J. A. Bygraves, E. Feil, G. Morelli, J. E. Russell, R. Urwin, Q. Zhang, J. Zhou, K. Zurth, D. A. Caugant, I. M. Feavers, M. Achtman, and B. G. Spratt. 1998. Multilocus sequence typing: a portable approach to the identification of clones within populations of pathogenic microorganisms. Proc. Natl. Acad. Sci. USA 95:3140-3145.

    Mendoza, M. C., R. Alzugaray, E. Landeras, and M. A. Gonzalez-Hevia. 1996. Discriminatory power and application of ribotyping of Yersinia enterocolitica O:3 in an epidemiological study. Eur. J. Clin. Microbiol. Infect. Dis. 15:220-226.

    Najdenski, H., I. Iteman, and E. Carniel. 1994. Efficient subtyping of pathogenic Yersinia enterocolitica strains by pulsed-field gel electrophoresis. J. Clin. Microbiol. 32:2913-2920.

    Nilehn, B. 1971. Some aspects on phage typing of Yersinia enterocolitica. Acta Pathol. Microbiol. Scand. B 79:446-447.

    Parkhill, J., B. W. Wren, N. R. Thomson, R. W. Titball, M. T. Holden, M. B. Prentice, M. Sebaihia, K. D. James, C. Churcher, K. L. Mungall, S. Baker, D. Basham, S. D. Bentley, K. Brooks, A. M. Cerdeno-Tarraga, T. Chillingworth, A. Cronin, R. M. Davies, P. Davis, G. Dougan, T. Feltwell, N. Hamlin, S. Holroyd, K. Jagels, A. V. Karlyshev, S. Leather, S. Moule, P. C. Oyston, M. Quail, K. Rutherford, M. Simmonds, J. Skelton, K. Stevens, S. Whitehead, and B. G. Barrell. 2001. Genome sequence of Yersinia pestis, the causative agent of plague. Nature 413:523-527.

    Perry, R. D., and J. D. Fetherston. 1997. Yersinia pestis—etiologic agent of plague. Clin. Microbiol. Rev. 10:35-66.

    Revazishvili, T., M. Kotetishvili, O. C. Stine, A. S. Kreger, J. G. Morris, Jr., and A. Sulakvelidze. 2004. Comparative analysis of multilocus sequence typing and pulsed-field gel electrophoresis for characterizing Listeria monocytogenes strains isolated from environmental and clinical sources. J. Clin. Microbiol. 42:276-285.

    Schill, W. B., S. R. Phelps, and S. W. Pyle. 1984. Multilocus electrophoretic assessment of the genetic structure and diversity of Yersinia ruckeri. Appl. Environ. Microbiol. 48:975-979.

    Selander, R. K., D. A. Caugant, H. Ochman, J. M. Musser, M. N. Gilmour, and T. S. Whittam. 1986. Methods of multilocus enzyme electrophoresis for bacterial population genetics and systematics. Appl. Environ. Microbiol. 51:873-884.

    Shubin, F. N., A. L. Gintsburg, V. M. Kitaev, N. V. Ianishevskii, and Z. G. Zenkova. 1989. Analysis of the plasmid composition of Yersinia pseudotuberculosis strains and its use for typing pseudotuberculosis pathogens. Mol. Gen. Mikrobiol. Virusol. 6:20-25.

    Skurnik, M., and P. Toivanen. 1993. Yersinia enterocolitica lipopolysaccharide: genetics and virulence. Trends Microbiol. 1:148-152.

    Smith, J. M., E. J. Feil, and N. H. Smith. 2000. Population structure and evolutionary dynamics of pathogenic bacteria. Bioessays 22:1115-1122.

    Stackebrandt, E., W. Frederiksen, G. M. Garrity, P. A. Grimont, P. Kampfer, M. C. Maiden, X. Nesme, R. Rossello-Mora, J. Swings, H. G. Truper, L. Vauterin, A. C. Ward, and W. B. Whitman. 2002. Report of the ad hoc committee for the re-evaluation of the species definition in bacteriology. Int. J. Syst. Evol. Microbiol. 52:1043-1047.

    Stroeher, U. H., and P. A. Manning. 1997. Vibrio cholerae serotype O139: swapping genes for surface polysaccharide biosynthesis. Trends Microbiol. 5:178-180.

    Sulakvelidze, A. 2000. Yersiniae other than Y. enterocolitica, Y. pseudotuberculosis, and Y. pestis: the ignored species. Microbes Infect. 2:497-513.

    Trebesius, K., D. Harmsen, A. Rakin, J. Schmelz, and J. Heesemann. 1998. Development of rRNA-targeted PCR and in situ hybridization with fluorescently labelled oligonucleotides for detection of Yersinia species. J. Clin. Microbiol. 36:2557-2564.

    Ursing, J., and S. Aleksic. 1995. Yersinia frederiksenii, a genotypically heterogeneous species with few differential characteristics. Contrib. Microbiol. Immunol. 13:112-116.

    Wauters, G., S. Aleksic, J. Charlier, and G. Schulze. 1991. Somatic and flagellar antigens of Yersinia enterocolitica and related species. Contrib. Microbiol. Immunol. 12:239-243.

    Weinstock, G. M. 2000. Genomics and bacterial pathogenesis. Emerg. Infect. Dis. 6:496-504.

    Woese, C. R., and G. E. Fox. 1977. Phylogenetic structure of the prokaryotic domain: the primary kingdoms. Proc. Natl. Acad. Sci. USA 74:5088-5090.

    Zhang, L., J. Radziejewska-Lebrecht, D. Krajewska-Pietrasik, P. Toivanen, and M. Skurnik. 1997. Molecular and chemical characterization of the lipopolysaccharide O-antigen and its role in the virulence of Yersinia enterocolitica serotype O:8. Mol. Microbiol. 23:63-76.

    Zhou, S., W. Deng, T. S. Anantharaman, A. Lim, E. T. Dimalanta, J. Wang, T. Wu, T. Chunhong, R. Creighton, A. Kile, E. Kvikstad, M. Bechner, G. Yen, A. Garic-Stankovic, J. Severin, D. Forrest, R. Runnheim, C. Churas, C. Lamers, N. T. Perna, V. Burland, F. R. Blattner, B. Mishra, and D. C. Schwartz. 2002. A whole-genome shotgun optical map of Yersinia pestis strain KIM. Appl. Environ. Microbiol. 68:6321-6331.