Previous Article | Next Article ![]()
Journal of Bacteriology, July 2006, p. 5319-5324, Vol. 188, No. 14
0021-9193/06/$08.00+0 doi:10.1128/JB.01662-05
Copyright © 2006, American Society for Microbiology. All Rights Reserved.
Mustafa Porsch-Özcürümez,2
Herbert Tomaso,2
Erwin Hofer,3
Wolf Splettstoesser,2,4
Ernst-Jürgen Finke,2
Helmut Tschäpe,1 and
Wolfgang Witte1
Robert Koch Institute, Wernigerode, Germany,1 Bundeswehr Institute of Microbiology, München, Germany,2 Austrian Agency for Health and Food Safety, Mödling, Austria,3 Department of Microbiology, University Hospital, Rostock, Germany4
Received 1 November 2005/ Accepted 28 April 2006
|
|
|---|
|
|
|---|
Four subspecies of F. tularensis are currently recognized (23). F. tularensis subsp. tularensis (also termed "type A") encompasses those isolates most virulent to humans. F. tularensis subsp. holarctica ("type B") is also highly infectious but rarely fatal in cases of human infection. Several strains of F. tularensis subsp. mediasiatica have been collected in central Asia, but little is known about their characteristics. F. tularensis subsp. novicida rarely causes disease in humans. It is the only subspecies of F. tularensis evidently not confined to the Northern Hemisphere, as it was recently reported in Australia (29).
Despite distinct variations in virulence and different geographic origins, isolates of F. tularensis display little genetic variation. Subspecies are closely related phylogenetically. The most efficient strain discrimination reported to date was achieved by PCR analyses of size polymorphisms of short-sequence tandem repeats in the genome of F. tularensis (variable-number tandem repeat analysis [VNTR]) (6, 7, 11, 12). A recent study revealed two genetically distinct clusters among isolates of F. tularensis subsp. tularensis and several clusters within F. tularensis subsp. holarctica (11). Analysis of the most-variable locus, a 9-bp repeat, provided 31 length variants (alleles) among 192 strains (11). In addition, pronounced linkage among alleles at different VNTR loci suggested a clonal population structure (11).
However, while highly variable short sequence repeats permit the currently most discriminatory typing procedure for Francisella tularensis, sequences from genes encoding proteins may be less prone to repeated mutations, in the absence of selection, and hence provide a more reliable phylogenetic signal (9, 14). In addition, typing results based on DNA sequencing are more portable between laboratories (24). In this study, we have investigated the genetic population structure of F. tularensis on the basis of nucleotide sequences from five housekeeping genes (tpiA, dnaA, mutS, prfB, and putA) and two genes encoding membrane proteins (tul4 and fopA) that had been exploited for PCR-based species identification in the past.
Sequence analyses. Bacterial isolates investigated in this study are listed in Table 1. Oligonucleotide primers for PCR were designed on the basis of genome sequences available from F. tularensis strains Schu S4 (subspecies tularensis; sequence accession no. AJ749949) and LVS (subspecies holarctica; preliminary sequence data available at http://bbrp.llnl.gov). Primer sequences are posted in Table S1 in the supplemental material. The seven loci investigated are located at physically distant positions on the bacterial chromosome as judged from the genome sequence from strain Schu S4 (Table 2). PCR and sequence analyses of both strands of the PCR products were performed by standard procedures. Sequences were deposited in the EMBL sequence database. Sequence divergence within F. tularensis (Table 2) was based on single-nucleotide polymorphisms rather than insertions or deletions. The ratio of mean nonsynonymous to synonymous substitutions per site (dN/dS), calculated with the software START (13), varied from 0.066 (dnaA) to 0.242 (tul4), indicating strong purifying selection (Table 2) acting on all loci investigated, including the genes encoding membrane proteins (tul4 and fopA). Sequences from locus fopA, however, were unusually uniform (Table 2). Different sequences at the five housekeeping loci resulted in 13 different allelic profiles or sequence types (ST1 to ST13) (Table 3). When the two loci encoding membrane proteins (tul4 and fopA) were included in the analysis, three additional sequence types were detected (ST14 to ST16).
|
View this table: [in a new window] |
TABLE 1. Bacterial strainsa
|
|
View this table: [in a new window] |
TABLE 2. Genomic positions of the seven loci and sequence variation within F. tularensis
|
|
View this table: [in a new window] |
TABLE 3. Allelic profiles
|
![]() View larger version (23K): [in a new window] |
FIG. 1. Phylogenetic trees based on gene sequences as indicated and calculated by using the maximum likelihood algorithm in the BIONUMERICS software package, version 4.0 (Applied Maths). One representative isolate from each sequence type (Table 3) was included in the computations. When available, gene sequences from Francisella philomiragia isolates (indicated by asterisks) were included as outgroups to enable bootstrap analyses of the deepest nodes within the F. tularensis clade. Bootstrap values from 100 resamplings are indicated when they are at least 90%. They are based on both the maximum-parsimony and neighbor-joining algorithms and were computed by using PHYLIP 3.65. Subspecies affiliations are indicated in the tree based on concatenated gene sequences. Scale bars indicate percent sequence similarity.
|
], 0.05). Reinvestigations allowing gamma-distributed evolution rate heterogeneity (four rate categories; shape parameter, 0.1) and altered transition/transversion ratios (range, 0.5 to 2.05) did not change these results for dnaA, prfB, and mutS, affirming recombination to be the underlying evolutionary cause for phylogenetic discordance rather than spatial variation of the evolution rate or selective pressure along the sequences (10). Predicted recombinational breakpoints in these sequences are indicated in Table 4. For tpiA, in contrast, a higher transition/transversion ratio suggested a locally increased negative, conservative, selective pressure (since transitions are more likely to be synonymous changes [10]). |
View this table: [in a new window] |
TABLE 4. Mosaic gene structures
|
The homoplasy test (implemented in the START software package [13]) was used to relate the frequency of observed synonymous homoplasies (that is, the same nucleotide changes in different branches of a maximum-parsimony tree) in the concatenated sequences from F. tularensis to the frequency expected under free recombination (18). Depending on the conservative consideration of potential very high or medium-high codon usage bias (18), respectively, this homoplasy ratio was calculated to be 0.09 to 0.10, which is low in comparison to several other bacterial species (26), suggesting that the overall contribution of recombination to the observed sequence variation is limited. However, the probability that the observed homoplasies could not be explained by mutations alone (and hence required recombination) was indicated to be 95% to 99% (18).
In conclusion, we have found several lines of evidence for limited horizontal genetic transfer and recombination in F. tularensis. No evidence was found, however, for recombination events in F. tularensis subsp. tularensis and F. tularensis subsp. holarctica. Rather, mosaic genes were discovered in F. tularensis subsp. novicida exclusively, and, in fact, omission of isolates F59, F60, and F62 from phylogenetic analyses eliminated the discrepancies among gene trees detected by incongruence length difference testing. These results are consistent with an underlying population in which recombination may be common and from which more-virulent clones that dominate medical interest and culture collections today have emerged and become widespread. Such a population structure previously has been termed "epidemic" (19). Consistently, based on gene content, F. tularensis subsp. novicida was recently concluded to be evolutionarily the oldest subspecies (27). Alternatively, recombination may be more frequent in, or restricted to, F. tularensis subsp. novicida. Our findings do not contradict the recently reported linkage disequilibrium among VNTR loci of this species, since only 2% of the isolates investigated in that study were affiliated with F. tularensis subsp. novicida (11). Currently, the genomes from several F. tularensis isolates are being sequenced (see www.genomesonline.org); those projects will provide abundant data for investigating the extent of recombination in these bacteria.
Phylogenetic diversity. Despite the detected signatures of recombinational events, the gene trees are far more similar than would be expected by chance alone (Fig. 1; also see Table S2 in the supplemental material). Hence, the extent of recombination in F. tularensis is not sufficient to completely obliterate the phylogenetic signal in the gene sequences. We therefore prefer to calculate phylogenetic relationships among the isolates on the basis of concatenated nucleotide sequences rather than allelic profiles, because the latter do not consider the extent of sequence difference (21, 24).
For F. tularensis subsp. holarctica and F. tularensis subsp. tularensis, results of phylogenetic analyses based on concatenated nucleotide sequences correlate very well with RD1 amplification fragment lengths and VNTR data, notably the amplification fragment sizes from the 16-bp repeat locus termed Ft-M10 (Fig. 1, Table 1); analyses of RD1 fragments and the short sequence repeats Ft-M3 and Ft-M10 were performed as described previously (3, 12). The sizes of both amplification fragments, RD1 and Ft-M10 (also termed SSTR16 or Ft-V2 [6, 11, 12]), have previously been reported to be conserved in each of these two subspecies (3, 6, 11, 12). Hence, we conclude that our sequence data confirm these two subspecies each to be monophyletic entities.
Isolates of F. tularensis subsp. holarctica display little genetic diversity, with the majority assigned to two genotypes (ST2 and ST6) and a few single-locus and double-locus variants of these. Generally, Ft-M3 fragment lengths from F. tularensis subsp. holarctica were found to be more variable than our allelic profiles. However, in several cases, isolates affiliated with different sequence types had Ft-M3 fragments of identical lengths. Remarkably, sequence types ST2 and ST6 differ at two housekeeping loci (tpiA and dnaA) and, on this basis, would not be considered particularly closely related (9) (Table 3). Even so, several isolates of both sequence types have Ft-M3 fragment lengths of 297 bp (for example, isolates F1 and F2) and 306 bp (for example, isolates F77 and F8 [Table 1]). We conclude that these identical Ft-M3 fragment lengths are homoplastic and provide misleading information about the isolates' phylogenetic relationships.
Isolates of F. tularensis subsp. tularensis form two separate phylogenetic clusters which likely correspond to the two distinct groups A.I (includes Schu S4) and A.II (includes ATCC 6223/F66) that were recently discovered on the basis of VNTR analyses (11). Definitive correlation with VNTR groups A.I and A.II cannot be confirmed at present, however, because the corresponding VNTR fragment lengths were not published in that previous report (11). One of the F. tularensis subsp. tularensis isolates, F87, has an Ft-M3 fragment of 306 bp which is not homologous to the same fragment in several sequence types (ST2, ST6, and ST14) among F. tularensis subsp. holarctica isolates. Nucleotide sequences of Ft-M3 fragments from the two subspecies differ by 5.5%.
Previously, restriction fragment length polymorphism analyses and genome comparisons through DNA microarray hybridization led to the conclusion that the subspecies F. tularensis subsp. holarctica and F. tularensis subsp. tularensis are highly homogeneous genetically (3, 28). While our data do not contradict these results, it should be considered that, even based on our limited data set, each of these subspecies appears to be more diverse than several highly uniform pathogens, including Mycobacterium tuberculosis, Bacillus anthracis, and Yersinia pestis (2, 21, 25). Hence, both subspecies of F. tularensis are likely to be significantly older than any of these recently emerged species (1).
Francisella tularensis subsp. novicida was indicated to be paraphyletic, as it does not include F. tularensis subsp. mediasiatica isolates F63 and F64, which descend from the same common ancestor (Fig. 1). Moreover, the three isolates assigned to F. tularensis subsp. mediasiatica did not form a single cluster, and F65 did not appear to be particularly related to F63 and F64. Instead, some affinity of F65 to F. tularensis subsp. tularensis may exist, which previously had been suggested for the entire subspecies (3, 11). Obviously, F. tularensis subsp. novicida and F. tularensis subsp. mediasiatica do not form phylogenetically coherent taxonomic entities and possibly should be reclassified. However, these taxonomic problems can be solved only when more isolates of these subspecies become available. Obviously, much of the diversity of F. tularensis extant in nature remains to be discovered.
Nucleotide sequence accession numbers. The nucleotide sequences determined in this study have been submitted to the EMBL database under accession no. AM 261086 to AM 261205.
Supplemental material for this article may be found at http://jb.asm.org/. ![]()
Present address: Robert Koch Institute, Berlin, Germany. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»