Figures
Abstract
A DNA marker that distinguishes plant associated bacteria at the species level and below was derived by comparing six sequenced genomes of Xanthomonas, a genus that contains many important phytopathogens. This DNA marker comprises a portion of the dnaA replication initiation factor (RIF). Unlike the rRNA genes, dnaA is a single copy gene in the vast majority of sequenced bacterial genomes, and amplification of RIF requires genus-specific primers. In silico analysis revealed that RIF has equal or greater ability to differentiate closely related species of Xanthomonas than the widely used ribosomal intergenic spacer region (ITS). Furthermore, in a set of 263 Xanthomonas, Ralstonia and Clavibacter strains, the RIF marker was directly sequenced in both directions with a success rate approximately 16% higher than that for ITS. RIF frameworks for Xanthomonas, Ralstonia and Clavibacter were constructed using 682 reference strains representing different species, subspecies, pathovars, races, hosts and geographic regions, and contain a total of 109 different RIF sequences. RIF sequences showed subspecific groupings but did not place strains of X. campestris or X. axonopodis into currently named pathovars nor R. solanacearum strains into their respective races, confirming previous conclusions that pathovar and race designations do not necessarily reflect genetic relationships. The RIF marker also was sequenced for 24 reference strains from three genera in the Enterobacteriaceae: Pectobacterium, Pantoea and Dickeya. RIF sequences of 70 previously uncharacterized strains of Ralstonia, Clavibacter, Pectobacterium and Dickeya matched, or were similar to, those of known reference strains, illustrating the utility of the frameworks to classify bacteria below the species level and rapidly match unknown isolates to reference strains. The RIF sequence frameworks are available at the online RIF database, RIFdb, and can be queried for diagnostic purposes with RIF sequences obtained from unknown strains in both chromatogram and FASTA format.
Citation: Schneider KL, Marrero G, Alvarez AM, Presting GG (2011) Classification of Plant Associated Bacteria Using RIF, a Computationally Derived DNA Marker. PLoS ONE 6(4): e18496. https://doi.org/10.1371/journal.pone.0018496
Editor: Stefan Bereswill, Charité-University Medicine Berlin, Germany
Received: October 19, 2010; Accepted: March 8, 2011; Published: April 21, 2011
Copyright: © 2011 Schneider et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by USDA grants 2008-34135-19370 and 2005-34135-15972. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Bacterial phytopathogens cause billions of dollars in crop losses annually (as summarized by Baker et al. [1]). Rapid and accurate identification of bacterial phytopathogens is essential for modern agriculture, as it permits informed decision making with respect to potentially costly but necessary control methods that include quarantine and the destruction of infected plant material in the field [2], [3]. The genus Xanthomonas contains some of the most important plant pathogenic bacteria that together infect over 392 hosts [4]. The bacterial genera Clavibacter, Ralstonia, Pseudomonas, Dickeya, Erwinia, Pantoea, Pectobacterium, and Xylella also contain species of important plant pathogens [5]–[12]. The phytopathogens X. oryzae pv. oryzae and R. solanacearum (Rs) race 3 biovar 2 are considered severe threats to US agriculture [3].
Bacterial taxonomic designations are derived from a diverse set of classification methods. Bacterial strains have been reclassified to different subspecies, species and even genera as diagnostic methods have evolved [6], [10]–[27]. For example, the historical designations of Xanthomonas pathovars and Ralstonia solanacearum races, based on the plant host and symptoms produced, are being replaced by modern DNA based methods [17] that have been used to reclassify some pathovars of X. axonopodis into different species and subspecies [15], [16] (e.g. X. axonopodis pv. vesicatoria type C was reclassified as X. perforans based on DNA-DNA hybridization). Likewise, 16S rDNA sequencing, DNA-DNA hybridization and other genetic and phenotypic traits have been used to reclassify some species of Pseudomonas, X. maltophilia, Erwinia chrysanthemi, E. carotovora and E. herbicola into species of Ralstonia, Stenotrophomonas maltophilia, Dickeya, Pectobacterium and Pantoea agglomerans, respectively [9]–[12], [28].
DNA-based species identification methods are robust and efficient, as evidenced by the large number of ongoing DNA barcoding efforts [29]. In DNA barcoding, sequences of a single DNA marker region are associated with reference strains that have been identified by taxonomists. The specific marker used is tailored to the range of target organisms. For animals, a region of the rapidly evolving mitochondrial cytochrome c oxidase (CO1) gene has been designated the DNA barcode region [30]. For the identification of algae, the universal plastid amplicon, a computationally derived region of the chloroplast genome [31] that can be amplified in photosynthetic organisms ranging from cyanobacteria to red, brown, golden and green algae as well as higher plants [32], has been proposed as a DNA barcode and proven to be practical in both biodiversity surveys and environmental sampling [33], [34].
The speed of PCR and the low cost of sequencing of the resulting amplicons have made bacterial identification based on DNA markers a viable method that is replacing earlier classification methods based on time-consuming biochemical or biological assays. The most commonly used markers are derived from ribosomal genes, in part because these can be amplified in the majority of species using universal primers [20]. 16S rDNA sequences have been determined for more than one million individual strains and environmental samples at http://rdp.cme.msu.edu [35]. Similarly, over 18,000 sequences of the internal transcribed spacer (ITS), which lies between the 16S and 23S rRNA genes, are available in GenBank [36]. Sequences from these two markers have been used to identify phytopathogens and study population diversity [37], [38], [39]. However, 16S rDNA sequences do not always resolve species within a genus, and even the more variable ITS sequences fail to resolve many genera below the species level [37]. Kang et al. [39] have shown that multiple copies of the 16S and 23S rRNA genes exist in over 80% (639 of 782 strains) of Gram+ and Gram− bacteria examined. Direct sequencing of rDNA amplicons from a genome containing different alleles (as observed for 415 of 782 strains by Kang et al. [39]) results in poor sequence quality. This is currently overcome by one of two time-consuming procedures, excising individual amplicons from agarose gels [40] if the alleles differ in length, or by cloning the amplicons.
DNA markers designed from housekeeping genes have been used singly or in combination to determine phylogenetic relationships of bacteria [21]–[27]. They generally require the use of genus-specific primers for amplification. Housekeeping genes are under stabilizing selection and are therefore expected to more accurately portray the genetic relationships among strains than genes that are under positive selection [41]. Genes under stabilizing selection may also be less prone to lateral transfer than other genes, such as those involved in pathogenicity. Nonetheless, even housekeeping genes may be transferred laterally as evidenced by recombination events within the gyrB gene of Vibrio species [42].
An ideal DNA marker for bacterial classification and identification is more variable than the rDNA regions, present in all target organisms as a single copy per genome, unlikely to undergo horizontal gene transfer, and amplifiable with universal primers. Attempts to design universal primers from sequences other than the rDNA regions have been unsuccessful ([43] and Supplemental Text S1), likely due to the great diversity and rapidly evolving nature of bacterial genomes.
We used six completely sequenced Xanthomonas genomes [44]–[47] representing four species to computationally identify a marker to classify closely related strains of bacteria. The replication initiation factor (RIF) marker, a region of the single-copy dnaA gene, was the best marker to distinguish closely related Xanthomonas strains. We sequenced RIF for a subset of 706 strains of six plant pathogenic genera from the Pacific Bacterial Collection (PBC) and the International Collection of Microorganisms from Plants (ICMP) at Landcare Research of New Zealand, representing a diverse range of hosts, geographic origins, subspecies, pathovars and races. These RIF frameworks can be queried online at http://genomics.hawaii.edu/RIFdb. Our results indicate that RIF is a suitable marker for the classification of strains from the six genera used in this study, complements other DNA markers in Xanthomonas MLSA studies, and may be expanded to other bacterial genera, the majority of which contain a single copy of the dnaA gene.
Results
Identification of RIF, a marker with subspecific resolution
A marker for the classification of bacteria was computationally identified using fifteen completely sequenced genomes representing six genera of plant pathogenic bacteria (asterisk in Table 1), including two Xanthomonas oryzae pv. oryzae strains (Xoo), two X. campestris pv. campestris strains (Xcc), one X. citri strain and one X. euvesicatoria strain (see Methods for details). In brief, 16,362 20+-mers (i.e. potential priming sites) were conserved between all six Xanthomonas genomes. Analysis of the Xcc strain 8004 and 33913 genomes revealed dnaA (Figure 1) as the sole gene that a) contains potential amplicons with more than one SNP (a total of five), b) has greater resolution than the ITS for the six Xanthomonas strains (Table 2) and c) is present as a single copy gene in all fifteen genomes. Two conserved 20+-mers that flank all SNPs were used to design primers and define the RIF marker. One SNP in the RIF marker also resolved the two Xoo strains with identical ITS sequences (Figure 2A). Primer pairs to amplify and sequence the RIF marker (Table 3) for Ralstonia and Clavibacter were designed within 60 nt of the Xanthomonas primers (Supplemental Figure S1).
The dnaA genes of X. campestris pv. campestris strains differ by 5 SNPs (pink highlight) and those of two X. oryzae pv. oryzae strains (green highlight) differ by 1 SNP. Conserved 20+-mers in all six Xanthomonas genomes are highlighted in blue. The only set of primers that flank all SNPs, have a melting temperature of 60±2°C and a G+C content of 50–60%, end in a G or C and have no complementary bases at the ends, are highlighted in yellow.
RIF resolves closely related strains of Xanthomonas that remain unresolved with ITS. The neighbor-joining cladograms for A) Xanthomonas, B) Clavibacter and C) Ralstonia using the ITS and RIF markers from fully sequenced strains in GenBank (Table 1). Bootstrap values >50% (shown at the node) are expressed as a percentage of 5,000 replicates. Red boxes contain two X. oryzae pv. oryzae strains resolved by one SNP between RIF sequences and zero SNPs between ITS sequences. Likewise, blue boxes contain two X. campestris pv. campestris strains resolved by five SNPs between RIF sequences and zero SNPs between ITS sequences. The RIF and ITS sequences of Xoo strains 10331 and PXO99 are identical. Xylella fastidiosa, Burkholderia mallei and Leifsonia xyli are included as outgroups.
The RIF marker is present as a single copy in the majority of bacterial genomes
The copy number of the dnaA gene was determined for 1,067 completed bacterial genomes. The dnaA gene is present once in 1,016 (95.2%) of 1,067 bacterial genomes, twice in 40 (3.7%) bacterial genomes and absent in eleven (1.0%) genomes (Supplemental Table S1). A single copy of the dnaA gene is present in all sequenced genomes of 366 genera, including all sequenced plant associated bacteria. Strains lacking the dnaA gene include seven genera of insect endosymbionts (Blattabacterium, Blockmannia, Carsonella, Hodgkinia, Riesia, Sulcia, and Wigglesworthia). All sequenced genomes of twelve genera contain two copies of the dnaA gene, including obligate intracellular animal pathogens (13 Chlamydia, 8 Chlamydophila, 1 Lawsonia, and 1 Parachlamydia), a rumen-associated bacterium (1 Fibrobacter), sulfate reducing bacteria (1 Desulfohalobium, 1 Desulfomicrobium and 7 Desulfovibrio), a metal reducing bacterium (1 Thermincola), a soil-associated heparin reducing bacterium (1 Pedobacter), a saltern crystallizer pond-associated bacterium (1 Salinibacter) and water-associated bacteria (2 Pirellula). Two genera, Mycobacterium and Mycoplasma, contain only one genome with two copies of dnaA, but nineteen and twenty strains, respectively, with a single copy of dnaA. Therefore, RIF may still be a useful marker for these two genera.
The RIF marker is unlikely to undergo horizontal gene transfer
Analysis of all available complete bacterial genomes revealed little evidence for horizontal transfer of dnaA between genera of bacteria or species of Xanthomonas. First, similar phylogenetic trees (data not shown) were constructed from the DnaA proteins and 16S rDNA regions of 1,016 genomes with a single copy of dnaA. Strains were placed in similar clades on both trees, consistent with a gene that has not undergone horizontal transfer between genera. Second, the phylogenetic tree constructed from the RIF marker of six Xanthomonas strains (Figure 2A) was similar to the dominant phylogenetic tree constructed from 229 Xanthomonas proteins present in all six sequenced strains [48]. These cladograms were representative of genetic regions that are unlikely to have been horizontally transferred between species of Xanthomonas. However, a subsequent analysis of 210 Rs strains for which we sequenced the dnaA, ITS and egl markers indicates that at least one of these may have undergone horizontal gene transfer (HGT) (see Rs results sections below).
Comparison of the RIF and ITS markers in sequenced reference genomes
RIF sequences of Xanthomonas and C. michiganensis reference strains contained more nucleotide polymorphisms than their corresponding ITS sequences. The average distance between eight strains of Xanthomonas representing four species (clade A in Figure 2A) was three times greater with RIF than with ITS (33.6 nt differences between seven unique RIF sequences vs. 11.3 nt nucleotide differences between five unique ITS sequences). A comparison of two sequenced subspecies of C. michiganensis (Figure 2B) revealed twice as many polymorphic nucleotides in the RIF marker as in the ITS marker (31 vs. 14 nucleotide differences).
The average distance between the six Ralstonia strains representing four species (clade A in Figure 2C) was similar with both RIF and ITS markers (50.5 and 54.8 nucleotide differences, respectively). Nevertheless, RIF separated R. solanacearum strains GMI1000 and UW551 from R. pickettii strain 12J (clades B and C in Figure 2C) while the ITS did not (clade B). ITS resulted in interspecies (Rs-Rp) distances (35 nt) similar to intraspecies (Rs-Rs) distances (26 nt) while RIF showed a clear separation between interspecies distances (42 nt) and intraspecies distances (18 nt).
For all three genera RIF yielded more consistent results with the same or greater number of different sequences than ITS.
Practicability of ITS and RIF markers as assessed by de novo sequencing of 263 PBC accessions
The RIF markers was amplified and directly sequenced in both directions using genus-specific primers (Table 3) and automatically base-called [49] and assembled [50] for 263 of the 840 Xanthomonas, Ralstonia and Clavibacter accessions of the PBC (Supplemental Table S2). Likewise, the ITS marker was amplified and directly sequenced in both directions using universal primers for the same set of strains. The length of the ITS sequences for the three genera ranged from 504–590 bp and that of the RIF sequences ranged from 654–700 bp. The majority of polymorphisms between RIF markers occurred in the wobble positions (data not shown). The success rate of high-quality direct sequence generation (i.e. overlapping sequence from both ends) for the 263 accessions was 1.22 times greater (Table 4) for RIF (248 sequences = 94%) than for ITS (205 sequences = 78%) and 16% of the accessions that yielded no ITS sequence produced RIF sequence (Supplemental Table S2). In addition, seven Xanthomonas and three Ralstonia accessions yielded a single RIF read in either the forward or reverse direction (Table 4).
The 200 strains for which both ITS and RIF sequences could be obtained were compared using unrooted trees (Supplemental Figures S2, S3, S4). RIF differentiated a greater number of Xanthomonas and R. solanacearum accessions than ITS (Table 4). However, the opposite was true for C. michiganensis subsp. michiganensis. RIF and ITS markers formed similar groupings for C. michiganensis (Supplemental Figure S4), but the RIF and ITS groupings were different for Xanthomonas and Ralstonia. Two X. campestris strains were placed outside of the X. campestris clade and one X. axonopodis strain was placed outside of the main X. axonopodis clade in the ITS tree, but not the RIF tree (asterisk in Supplemental Figure S2). For Ralstonia, strains grouped in one clade with the ITS marker were not grouped in the same clade with RIF (e.g., K0018/K0190 and K0024/K0157 in Supplemental Figure S3). In addition, although the Rs ITS clades were separated by a greater distance than the RIF clades, the sequence diversity within an ITS clade was less than that within a RIF clade (Supplemental Figure S3). R. pickettii was placed outside the Rs clade in the RIF but not the ITS tree. The different groupings observed for the RIF and ITS markers of some Rs strains indicate that one of the markers, most likely ITS, has undergone HGT. The higher sequencing success rate and more reliable groupings obtained with RIF encouraged us to exhaustively sample the sequence diversity present in the PBC and ICMP to build more complete RIF reference frameworks.
Building RIF frameworks from sequences of 682 characterized accessions of Xanthomonas, Ralstonia solanacearum and Clavibacter michiganensis
Only characterized strains in the PBC and ICMP were used as reference strains to construct RIF frameworks for the purpose of classifying unknowns. In addition to the 263 strains above we attempted to sequence RIF from an additional 452 characterized PBC strains (715 strains total) that had been collected over a span of 37 years, and 84 Xanthomonas strains from the ICMP (Table 5 and Supplemental Table S2). In total, the RIF marker was successfully sequenced for 682 (85.3%) of 799 characterized strains and yielded 109 different DNA barcodes for classifying isolates of Xanthomonas, R. solanacearum, and C. michiganensis with respect to their RIF genotype (Table 5).
Xanthomonas species.
Over 87.3% of the 361 Xanthomonas strains from the PBC yielded usable RIF sequence. The success rate for the ICMP strains was lower due to problems associated with shipping the DNA. In total, RIF sequences were obtained for 315 PBC and 43 ICMP strains (Table 5) representing 20 species and 29 pathovars (RIF sequence in Supplemental Table S2).
After automated alignment and trimming to 558 nt (see Methods), 82 different RIF sequences were obtained for Xanthomonas (Figure 3). These sequences formed 10 major clades in a neighbor-joining cladogram, each of which contained a dominant species: X. vasicola (B); X. oryzae (C); X. melonis (D); X. campestris (E); X. arboricola (G); X. fragariae (H); X. sp. from Dysoxylum (I); X. vesicatoria (J); X. (Stenotrophomonas) maltophilia (N and O) and X. albilineans (P). The species X. alfalfae, X. euvesicatoria, X. perforans, X. citri and X. fuscans, previously all classified as pathovars of X. axonopodis [13], grouped with the majority of X. axonopodis strains in the polyphyletic clade A. RIF sequences of X. axonopodis pv. dieffenbachiae were scattered throughout the RIF tree in clades A, E, G and N (red text in Figure 3). Clades F, G, K, M and N contained more than one species, but no two species shared the same RIF sequence. However, four RIF sequences in clades A and E (blue boxes in Figure 3) were shared by two or more species. Two X. campestris strains (clades F and G in Figure 3) were placed outside the main X. campestris clade E. Strains of the rice pathogen Xoo [3] yielded 5 different RIF sequences, all of which differed from the closely related X. oryzae pv. oryzicola by at least four nucleotides (Figure 3).
The rooted neighbor-joining cladogram was constructed from 358 characterized Xanthomonas strains from the PBC (Table 2), ten reference strains from GenBank (see Table 1 for strain names) and the outgroup Xylella fastidiosa. Xanthomonas abbreviations are given in the text box. Identical sequences are represented only once and the number of sequenced strains is indicated on each leaf. Bootstrap values of >50% (shown at the node) are expressed as a percentage of 5,000 replicates. Blue boxes contain strains of different species with identical RIF sequences. X. axonopodis pv. dieffenbachiae strains are found throughout the tree (red text).
Ralstonia solanacearum .
The RIF marker was successfully sequenced for 210 (87.8%) of 239 characterized strains of R. solanacearum representing four races, blood disease bacterium (BDB), and the atypical strain R. solanacearum UW433 (ACH0732). Seventeen different sequences were obtained after the alignment was trimmed to 617 nucleotides (Figure 4). In order to compare the sequencing efficiency and resolution of RIF to that of the egl marker, which has been used to classify Rs into sequevars [10], we attempted to generate egl sequence from 191 characterized Rs strains (Table 5). RIF was sequenced with a 12.6% higher success rate than egl. Strains for which both RIF sequence and an egl sequevar had been obtained were compared in a neighbor-joining tree (Figure 4). The egl sequevars did not group into a single clade in the RIF tree, e.g. egl sequevars 13–18 are present in both clades B and D. The different groupings of strains with the RIF and egl markers are in agreement with Fegan's observations for Rs strain ACH0732 grouping differently with the ITS, egl and polygalacturonase markers [18], and may again be indicative of HGT. Thus, although greater resolution was achieved with the egl marker (22 different sequences) than with the RIF marker (17 different sequences), the RIF marker, which is under stabilizing selection [41], should more accurately portray genetic relationships than the positively selected pathogenicity related egl gene.
The rooted neighbor-joining cladogram was constructed from 210 characterized Ralstonia strains from the PBC (Table 2), six reference strains from GenBank (see Table 1 for strain names)and the outgroup Burkholderia mallei. Identical sequences are represented only once and the number of sequenced strains is indicated on each leaf. Bootstrap values >50% (shown at the node) are expressed as a percentage of 5,000 replicates. All Rs strains (clade A) are clearly separated from other Ralstonia species. Rs1, Rs2, Rs3 and Rs4 are the four races of R. solanacearum, BDB is the blood disease bacterium and Rs† is R. solanacearum UW433 (ACH0732) which is an atypical race 1 biovar 2 strain [19]. Blue boxes contain strains of Rs races 1, 2, 4 and BDB that share an identical RIF sequence. In contrast, orange boxes contain only Rs3 strains showing that these RIF sequences are specific to race 3. Numbers in boxes next to tree leaves are egl sequevars obtained for those strains. Red text indicates Rs strains with no previous race designation.
Clavibacter michiganensis subspecies.
RIF sequences were generated for 114 of 115 Clavibacter michiganensis strains (99% success rate) including 111 strains belonging to C. michiganensis subsp. michiganensis that had been collected from every continent except Australia and Antarctica (Figure 5 and Table 5). Ten different sequences remained after the alignment was trimmed to 660 nt. The resulting neighbor joining tree contains three clades representing the three subspecies. Notably, there was no correlation between the RIF sequence of Cmm strains and their geographic origin, i.e. Cmm strains from distant geographic locations shared the same RIF sequence. Worldwide distribution of this seed-borne pathogen on infected tomato seed may account for this lack of correlation.
The rooted neighbor-joining cladogram was constructed from 114 characterized Clavibacter strains from the PBC (Table 2), two reference strains from GenBank (see Table 1 for strain names) and the outgroup Leifsonia xyli. Identical sequences are represented only once and the number of sequenced strains is indicated on each leaf. Bootstrap values >50% (shown at the node) are expressed as a percentage of 5,000 replicates. Red text indicates C. michiganensis strains with no previous subspecies designation. Cm† is a non-pathogenic strain isolated from tomato that is most similar to Cmi. Unknowns #1-10 were isolated from a recent bacterial canker outbreak of tomato (see text) and perfectly match two of the Cmm RIF reference sequences. Cmm: C. michiganensis subsp. michiganensis. Cmi: C. michiganensis subsp. insidiosus. Cms: C. michiganensis subsp. sepedonicus.
Expansion of RIF to three genera of the Enterobacteriaceae
To further evaluate the utility of RIF, we analyzed the plant pathogenic bacteria Dickeya, Pectobacterium and Pantoea. Initial Enterobacteriaceae RIF primer sets developed from sequenced strains of Dickeya (Dd_3937), Pectobacterium (Pa_1043) and Erwinia (NC_013971.1) were used for PCR amplification and sequencing (Table 3). New primers designed from the more recently sequenced Dickeya (Dd_703 and Dd_586) and Pectobacterium (Pcc_PC1 and Pw_Wpp163) genomes have improved amplification rates and genus specificity (data not shown).
In silico analysis of the RIF and ITS markers of the nine completely sequenced Dickeya, Pectobacterium and Erwinia genomes revealed 2–3 times more polymorphic nucleotides in RIF than ITS with Dickeya (80 vs. 40 nt) and Erwinia (58 vs. 21 nt) (Figure 6). These distances are similar to those observed for different species of Xanthomonas and illustrate the higher resolution of the RIF marker. In contrast, Pectobacterium strain PccPC1 was separated from other Pectobacterium accessions by longer branches in the ITS than the RIF tree (Figure 6). This is due to the numerous nucleotides affected by several indels in the ITS sequence of PccPC1 (27 nucleotides from multiple insertions and six nucleotides from a deletion relative to the two other Pectobacterium strains).
Sequences for Dickeya, Pectobacterium, Pantoea and Erwinia were extracted from fully sequenced strains in GenBank (Table 1). Bootstrap values >50% (shown at the node) are expressed as a percentage of 5,000 replicates. Note the longer branch lengths for the Erwinia and Dickeya species in the RIF tree.
The efficacy of the RIF and ITS markers was directly compared for 96 Enterobacteriaceae accessions (Table 6). RIF was easier to amplify and sequence in this family than ITS as illustrated by the higher success rate for the direct sequencing of RIF (53) as compared to ITS (35) from 64 strains of Dickeya (all verified to be Dickeya with the ADE primers [51]). A greater number of different sequences was obtained with RIF (12) than with ITS (11) for the same set of Dickeya strains (Supplemental Figure S5). We were able to obtain RIF but no ITS sequence for fourteen Pectobacterium strains. The difficulty in sequencing ITS was likely due to the presence of multiple rDNA copies that differ in sequence and length in these Enterobacteriaceae genomes. All Pantoea strains and the Pectobacterium strains K0509, K0574 and K0522 required a lower stringency annealing temperature (51°C vs 61°C) for RIF amplification (Table 6 and Supplemental Table S2).
Twenty-nine different RIF sequences were obtained for Enterobacteriaceae strains after the alignment was trimmed to 722 nt (Table 6 and Figure 7). Dickeya, Pectobacterium, Pantoea, Erwinia and Yersinia formed distinct clades in the neighbor joining tree (Figure 7). Notably, four D. dadantii strains were located in three separate sub-clades within the Dickeya clade (Figure 7). The sequenced strain P. wasabiae Wpp163 (formerly named P. carotovorum) and the potato strain P. carotovorum K0574 (classified based on bacteriological tests) differed from each other by only three nt. Similarly, the distance between several different P. atrosepticum strains in clades I and K (Figure 7) was greater (54.5 nt) than that between the species P. atrosepticum and P. carotovorum (50.8 nt and 48.6 nt, respectively, in clades J and K).
The rooted neighbor-joining cladogram was constructed from 74 Enterobacteriaceae strains, including 24 characterized strains of Dickeya, Pectobacterium and Pantoea from the PBC (Supplemental Table S2), eight reference strains from GenBank (see Table 1 for strain names) and the outgroup Yersinia pestis. Identical sequences are represented only once and the number of sequenced strains is indicated on each leaf. Bootstrap values >50% (shown at the node) are expressed as a percentage of 5,000 replicates. The true Erwinia species E. tasmaniensis separates from the renamed Erwinia species that branch into three different genera: Dickeya (clade A), Pectobacterium (clades G and H) and Pantoea (clade M). Red text indicates uncharacterized strains of Dickeya and green text indicates uncharacterized strains of Pectobacterium. K numbers are shown for the accessions discussed in the text (Supplemental Table S2).
Classifying uncharacterized strains based on RIF genotypes
Sequence frameworks were used to rapidly classify 70 uncharacterized cultures (11 Cm, 4 Pectobacterium, 46 Dickeya and 9 Rs). Ten putative Cmm strains from multiple outbreaks of tomato canker were classified within 48 hours of their arrival and matched two different cosmopolitan Cmm reference genotypes (Figure 5). A single Cm strain non-pathogenic on tomato (Cm† in Figure 5) from the PBC had a RIF genotype very similar to that of the C. michiganensis subsp. insidiosus reference strain.
Fifty strains of Pectobacterium and Dickeya from the PBC that could not be identified to species based on biochemical tests were placed in the RIF framework. Two strains of Pectobacterium sp. isolated from Aglaonema in Hawaii and two strains isolated from water grouped with the P. carotovorum clade near RIF genotypes from known reference strains (green text in Figure 7). The other forty-six strains grouped with distinct Dickeya reference strains (red text in Figure 7).
Nine Rs strains isolated from Guatemalan geraniums, and verified to be Rs by bacteriological and immunodiagnostic tests, grouped with Rs race 3 biovar 2 strains in clade C of the RIF tree (red text in Figure 4), which contains the select agent Rs_UW551. Thus, the RIF frameworks, even in their current form, have proven useful for classifying RIF sequences from uncharacterized isolates of four genera.
In silico classification of Xylella and Pseudomonas with RIF
The potential of RIF to classify bacteria from additional plant associated genera was further examined with an in silico analysis of Pseudomonas and Xylella strains available in GenBank. All but one of the completely sequenced strains contained a single copy of the dnaA gene and more than one copy of the ITS region (P. aeruginosa strain 39016 contains a single copy of both ITS and RIF). RIF primers for X. fastidiosa were developed from five strains (Table 3), primers for Pseudomonas were developed from twenty-three strains representing eight species. The five Xylella strains were classified into three RIF genotypes, whereas every Pseudomonas strain had a different RIF sequence.
Unexpected nucleotide distances between genera, species and subspecies using RIF
The reference frameworks constructed from 125 different RIF sequences (82 Xanthomonas, 17 Rs, 10 Cm, 8 Pectobacterium, 6 Dickeya and 2 Pantoea) allowed comparisons of the distances between genera, species and subspecies. In general, the distances (Supplemental Tables S3, S4, S5, S6, S7) as measured in nucleotide substitutions were greatest between genera and smallest between subspecies as would be expected. Exceptions included: (i) the genera Stenotrophomonas and Xanthomonas were separated by fewer nucleotide differences (70.5 nt) than species of Dickeya (90 nt) (Supplemental Tables S3 and S6); (ii) two species, X. pisi and X. bromi (asterisk in Supplemental Table S3), were separated by only 28 nucleotide differences (clade K in Figure 3) whereas subspecies of Clavibacter michiganensis were separated by 31 nucleotide differences; (iii) the strains within subclades of Stenotrophomonas maltophilia (clades N and O in Figure 3) differed by 50.1 nucleotides, whereas species of Ralstonia and species of Xanthomonas were separated by only 56 and 49.5 nucleotides, respectively. The inconsistent distances at different taxonomic levels may be due to the incorrect grouping of diverse strains to one taxon or the misclassification of some strains in the collection based on initial genetic and phenotypic analysis. Some of these inconsistencies may identify potential problems with the classification, identification or naming of these organisms.
Discussion
DnaA structure in relation to the RIF marker region
The DnaA protein is an essential core metabolic protein in bacteria known to regulate the initiation of chromosomal replication by binding DnaA boxes on the replicating chromosome [52]. The absence of the dnaA gene from eleven sequenced genomes of insect endosymbionts has been suggested to reflect the extreme dependency that these symbionts have with their host [53]. The DnaA protein contains several conserved domains, including a AAA+ superfamily domain near the N-terminal and a helix-turn-helix DNA binding domain at the C-terminal of the protein [52]. The primers were designed near, but not in, the AAA+ domain and on the edge of the helix-turn-helix DNA binding domain (Supplemental Figure S1). Thus, these conserved regions may function at the nucleotide rather than protein level, possibly in gene transcription or mRNA stability.
RIF can be used as an identification marker for the majority of bacteria
The fact that dnaA is a single copy gene in the vast majority of complete bacterial genomes examined and does not appear to frequently undergo horizontal gene transfer make it a valuable marker for bacterial strain classification. The ease with which we were able to produce sequence frameworks for six genera illustrates the practicability of this marker, even though RIF amplification required different primers for each genus. Since previous efforts to identify universal primers that amplify protein-coding regions were unsuccessful ([43] and Supplemental Text S1), the need for genus-specific primers was expected.
RIF can add resolution to multi-locus sequence analysis
The RIF framework for Xanthomonas was similar in structure to the tree produced for a Xanthomonas MLSA by Young et al. [25]. Although the larger amount of sequence data used in multi-locus analyses is expected to provide more resolution than any single marker, in this particular example five SNPs distinguished the RIF markers of Xcc strains 8004 and 33913 (Figure 1) while the four markers used in the MLSA were identical in the two strains (Supplemental Table S8). This suggests that RIF not only performs well as a single marker, but also that it should be considered for inclusion in multi-locus studies, such as those incorporated into PAMdb [54], an online MLSA database for classification of Xanthomonas, Pseudomonas, Ralstonia and Acidovorax. Indeed, a partial sequence of the dnaA gene (covering 52.5% of RIF) was included as one of eight loci in a recent MLSA of Rickettsia [55].
RIF provided greater resolution, higher sequencing success rate and more reliable strain groupings than ITS
Reference frameworks constructed with RIF can be used to classify strains of Clavibacter, Dickeya, Pantoea, Pectobacterium, Ralstonia, and Xanthomonas. Although RIF was originally derived by comparative genomics of Xanthomonas, it turned out to have better resolution than ITS for all genera except Clavibacter. The increased resolution may be attributed to the longer amplicon produced by RIF (654–700 nt) versus ITS (504–590 nt). In addition to providing higher resolution than ITS, RIF also had a higher sequencing success rate using the direct amplicon sequencing method. Also, RIF is part of a protein coding gene, which is more resistant to compensatory mutations, indels and inversions, thus yielding more consistent phylogenetic trees than those obtained with ribosomal DNA. For Ralstonia, RIF had a higher sequencing success rate than egl, but egl provided a greater number of unique sequences. However, egl is a pathogenicity related gene that may be prone to horizontal transfer.
RIF classification is consistent with other methods of classification
The cladograms constructed using the RIF marker were consistent with existing trees based on single or multiple markers (where available). RIF sequences supported previous genetic studies of Xanthomonas, the close relationship between Stenotrophomonas and Xanthomonas and the reclassification of Erwinia chrysanthemi, E. carotovora and E. herbicola to Dickeya, Pectobacterium and Pantoea, respectively [9]–[11], [13], [22], [23], [25]–[27].
RIF genotypes did not correspond to the current nomenclature for pathovars of X. campestris and X. axonopodis, nor to the race designations of R. solanacearum as expected based on previous observations [4], [5], [9], [11].
The polyphyletic clade A of Xanthomonas (Figure 3) contained six species, five of which have recently been renamed from X. axonopodis to X. alfalfae, X. euvesicatoria, X. perforans, X. citri and X. fuscans [15], [16]. These six species fell into subclades that agree with the groups of X. axonopodis observed by Vauterin et al. [13].
Classification of unknown Clavibacter strains isolated from a recent outbreak of bacterial canker based on RIF corresponded to the groupings obtained with rep-PCR fingerprinting analysis (unpublished): nine strains were identical while the tenth fell into a separate group with both methods.
The RIF database
In order to enable comparison of strains isolated worldwide with those characterized in this study, we have created an online database, RIFdb. For ease of use, the database can be queried with unprocessed single or paired chromatograms, or FASTA sequences, to search for the best reference strain match. Chromatograms are base-called, and paired chromatograms are assembled by RIFdb (see methods) prior to the search. Querying the database with a single read reduces the cost of identification if a user decides to sequence only one end. Queries are automatically aligned with existing sequences and visualized in a neighbor-joining tree using ArchaeopteryxE [56]. The RIF sequences in the database are automatically re-trimmed if a partial sequence is provided as a query, although excessive trimming will decrease the resolution of this marker.
Clearly the utility of the RIF database will increase as more sequences from as yet unrepresented accessions from around the world are added. Sequences of strains from diverse bacterial collections will increase global representation. Deposition of high quality chromatogram pairs along with key characteristics of strains will enable the expansion of RIFdb with strains from international collections and increase its utility to diagnosticians worldwide.
Conclusion
The RIF marker is suitable for the classification and meaningful grouping of strains from the six genera examined in this study. The RIF marker provides a greater sequencing success rate than the ITS in all six genera and a greater number of sequence barcodes than the ITS in three of four genera examined. Further expansion of RIF sequence frameworks with strains from other collections via the web accessible database will facilitate the classification of unknowns in the future. Finally, we show that the RIF marker system should be expandable to most bacterial genera, including Xylella and Pseudomonas.
Methods
Computational identification of suitable marker regions
The computational identification of a DNA barcode region that distinguishes closely related Xanthomonas strains was performed with completely sequenced Xanthomonas genomes (Table 1) as follows. First, SNPs were identified between the two Xanthomonas oryzae pv. oryzae genomes Xoo_10331 and Xoo_311018 using MUMmer [57] to match all conserved regions of 20 nucleotides or greater. A similar analysis was performed for two strains of X. campestris pv. campestris (Xcc_33913 and Xcc_8004). Nucleotides of Xoo_311018 and Xcc_33913 that had no MUMmer matches in the other strain of the same pathovar were masked with a Perl script. Second, single copy oligos in Xoo_311018 that were perfectly conserved in the five other Xanthomonas genomes (Xoo_10331, Xcc_33913, Xcc_8004, Xccit_306 and Xe_85-10) were identified using MUMmer [57] (-mum -l 20 -b) in sequential comparisons, and will be referred to as conserved 20+-mers. In this analysis, the regions conserved between Xoo_311018 and genome two were used as a query with genome three and so on. Third, a Perl script was used to identify regions in all six Xanthomonas genomes that a) were flanked by conserved 20+-mers, b) separated by >550 nucleotides, c) produced an amplicon of <1,000 nucleotides, and d) contained at least 20 masked out nucleotides (i.e. containing 2 or more SNPs) in Xoo_311018. The same analysis was performed using the masked genome of Xcc_33913. Fourth, another Perl script was used to align the extracted regions from the six Xanthomonas genomes using ClustalW [58] and output a distance matrix of nucleotide differences to confirm that orthologous regions from any two Xanthomonas strains contained one or more SNPs. Finally, a Perl script was used to extract every gene from the identified regions in Xcc_33913 and Xoo_311018, and confirm the presence of the gene in nine completely sequenced genomes of Ralstonia, Clavibacter, Pectobacterium, Erwinia and Dickeya (asterisk in Table 1) with a default BLAST [59] comparison. (No sequenced strains of Pantoea were available at the time of this study). The BLAST results were checked by hand to confirm the presence of the complete gene in all genomes of interest (asterisk in Table 1); only one gene (dnaA) was present in full in all genomes.
Copy number determination of bacterial dnaA genes
A set of 3,020 bacterial DnaA protein sequences was downloaded by querying the NCBI protein database (limited to bacterial organisms) with “dnaA” and “350∶1000” limited to gene and SLEN, respectively. Of these, only 1,188 protein sequences whose headers contained the terms “dnaA”, “chromosomal replication initiator” or “chromosomal replication initiation” were kept. Separately, all 1,159 completely sequenced bacterial and archael genomes were downloaded from NCBI (ftp://ftp.ncbi.nih.gov/genbank/genomes/bacteria) using a Perl script. The directory containing the genome “Escherichia coli strain RS218” was left out as it improperly contained Enterobacteria phage CUS-3. The subset of 1,067 bacterial genomes on the “List of prokaryotic names with standing in nomenclature” (http://www.bacterio.cict.fr) and the list of cyanobacterial genera (http://www.cyanodb.cz/valid_genera) were used in this analysis. Subsequently, a protein set was generated for each of the 1,067 bacterial genomes. Twenty-one proteins from these sequenced genomes that were annotated as “dnaA”, “chromosomal replication initiator” or “chromosomal replication initiation” but whose accession number was not in the DnaA protein set were added to the protein set for a total of 1,209 sequences. Another Perl script was used to identify copies of the DnaA protein in the genome of each strain using BLAST (-p tblastn -m8 –X 300 –e 1e-10) and the 1,209 DnaA proteins as a query. Any gene in the 1,067 bacterial genomes with a match to any protein query at >40% identity over >350 amino acids was counted as a DnaA protein. That query was then used to identify additional copies of DnaA in the genome. The dnaA gene of Yersinia pestis strain D106004 could not be identified in this way because of a frame-shift mutation (Supplemental Table S1), so bl2seq analysis using the dnaA gene from Y. pestis Angola was used. The dnaA gene of Lysinibacillus sphaericus strain C3 41a began at position 4,639,741 of 4,639,821 nt of the circular chromosome instead of position 1. Genomes containing more than one copy of the dnaA gene were compared with themselves using bl2seq to identify possible gene duplication or horizontal transfer events.
RIF primer design on the dnaA gene
The dnaA gene from the six sequenced Xanthomonas strains were first used for primer development. All sequences from Xanthomonas were aligned with ClustalW [58] and 18–20 nt primers were designed manually on conserved regions to produce the largest amplicon covering the greatest number of SNPs, shorter than 1,000 nucleotides. The primers were required to have a melting temperature of 60±2°C, 50–60% G+C content, end in a G or C and have no complementary bases at the ends. Only one of many possible regions covering all SNPs between the two Xcc and two Xoo strains was chosen. The gene previously extracted from Clavibacter, Dickeya, Pectobacterium, and Ralstonia was used to develop four more dnaA primer sets (Table 3). Primers for each genus had to be located within 60 nucleotides of the Xanthomonas primers and have a melting temperature of 60±2°C and 50–60% G+C content (Supplemental Figure S1). Primers were designed individually for Ralstonia, Clavibacter, Pectobacterium and Dickeya. Pectobacterium primers were used for Pantoea as no sequenced strains existed at the time (see DNA marker amplification below). Recently sequenced genomes allowed the development of primers with increased specificity to Pectobacterium and Dickeya using the dnaA gene from three Pectobacterium (Pa_1043, Pw_Wpp163 and Pcc_PC1) and three Dickeya strains (Dd_3937, Dz_1591 and Dd_703) (Table 3).
Bacterial culturing and DNA extraction
A subset of 840 bacterial clones from the PBC was chosen for DNA sequencing (Supplemental Table S2). Fourteen of the clones chosen were from duplicate characterized strains and were not counted in the total number of sequenced strains. The strains were plated on TZC medium to confirm that no contamination was present and transferred into sterile deionized water for genomic DNA isolation and to LB containing 15–20% glycerol for long-term storage at −80°C. DNA was isolated by adding 300 µl of water culture to 100 µl of a 40% mixture of deionized water-Chelex-100® resin (BioRad), and incubated at 60°C for 60 minutes [60]. DNA from ten suspected Clavibacter strains from a tomato canker was isolated using a mixture of water-Chelex-100® resin (BioRad). DNA from 84 Xanthomonas strains extracted with the REDExtract-N-Amp™ PCR ReadyMix™ (Sigma-Aldrich) were provided by John Young and Duck-Chul Park from the ICMP in New Zealand.
DNA marker amplification and sequencing
All markers were PCR amplified using an Eppendorf Mastercycler® Ep Gradient machine. All PCR reactions contained: 25 µl of JumpStart™ REDTaq® ReadyMix™ (Sigma-Aldrich) for high throughput PCR with 5 µl of 10 mM primer, 5 µl of template and 15 µl of deionized water. The cycling conditions included an initial denaturation step at 94°C for ten minutes, followed by 35 cycles of 94°C for 30 seconds, 61°C for one minute and 72°C for 30 seconds, followed by a ten minute extension at 72°C and a hold at 4°C. Pectobacterium and Pantoea strains that did not amplify with these settings were amplified with a 51°C annealing temperature using the Pectobacterium primers (RIF Pectobacterium amplification at 51°C in Supplemental Table S2). ITS marker amplifications for Xanthomonas, Clavibacter, Ralstonia, Dickeya, Pectobacterium and Pantoea were performed as described by Normand [20]. The presence of multiple rDNA operons of different sizes in the genus Dickeya necessitated manual isolation of the smaller ITS amplicon from an agarose gel using the band-stab method of Wilton [40], followed by 25 cycles of PCR. ADE marker amplification for Dickeya and egl marker amplification for R. solanacearum strains were performed as described by Nassar [51] and Fegan [19], respectively. All PCR products were separated on a 1.5% agarose gel to confirm quantity and quality, and DNA was purified and sequenced from all PCR attempts, even if no PCR product was visible on the gel. PCR products were purified using a Qiaquick 96 PCR Purification Kit (Qiagen), and the product was sequenced at the University of Hawaii sequencing facilities using forward and reverse primers.
DNA marker analysis
Chromatograms of sequenced amplicons were automatically converted to FASTA files using phred [49] with default settings and no trimming. Forward and reverse sequences were automatically assembled with phrap [50], set to a minimum match of 100 and a minimum quality score of 20. A Perl script was used to identify the longest stretch of high quality bases containing no more than five consecutive low quality bases from the ace file. Low quality bases were edited manually with BioEdit [61] and either retained, removed, or converted into degenerate bases. Assembled sequences spanning fewer than 550 high-quality bases and sequence reads that could not be assembled were removed. RIF sequences that did not match the expected strain annotation were re-isolated and re-sequenced. RIF sequences of each genus were aligned using MEGA [62] and trimmed to the same length. Distance matrices, and neighbor-joining (N-J) phylogenetic trees with bootstrapping scores, from 5000 replicates, were produced with MEGA using pairwise deletions and the number of nucleotide differences to indicate branch length. In and between group distances and average distances were computed with complete deletions using all RIF sequences from characterized strains in each respective clade of the genus-specific neighbor-joining tree. In and between group distances, average distances and N-J trees were produced using a single representative for each different sequence. Both the ITS and RIF markers were re-amplified and sequenced for strains that branched in unexpected positions to confirm their RIF sequence. N-J trees were rooted with the orthologous region obtained via BLAST [59] from a closely related sequenced strain in NCBI. Assembled high quality trimmed sequences for all accessions have been deposited in GenBank under accession numbers HM180945-HM181929 and HM469616-HM469894. The RIFdb database can be queried at http://genomics.hawaii.edu/RIFdb/.
Supporting Information
Figure S1.
Multiple sequence alignment of nucleotides 311 to 1311 of the dnaA genes of six genera. Primer regions are shown for Clavibacter, Xanthomonas, Ralstonia, Erwinia, Dickeya and Pectobacterium. Primer binding regions are shown in red with black background. The AAA+ domain (green) [52] and the C-terminal domain (pink) [52] are highlighted.
https://doi.org/10.1371/journal.pone.0018496.s001
(TIF)
Figure S2.
RIF distinguishes more Xanthomonas strains than ITS. Unrooted neighbor-joining trees for the RIF and ITS markers were constructed from eighty-four Xanthomonas strains from the PBC (Supplemental Table S2) and eight reference strains from GenBank (see Table 1 for strain names) . Identical sequences are represented only once and the number of sequenced strains is indicated on each leaf. Bootstrap values >50% (shown at the node) are expressed as a percentage of 5,000 replicates. Two X. campestris strains and one X. axonopodis strain localize to the appropriate clade with RIF but not ITS (red asterisk). Xc - X. campestris, Xa - X. axonopodis, Xe – X. euvesicatoria, Xcit – X. citri, Xo - X. oryzae, Xm – X. (Stenotrophomonas) maltophilia.
https://doi.org/10.1371/journal.pone.0018496.s002
(TIF)
Figure S3.
RIF sequences distinguish more Ralstonia strains than ITS. Unrooted neighbor-joining trees for the RIF and ITS markers were constructed from ninety-seven Ralstonia strains from the PBC (Supplemental Table S2) and three reference strains from GenBank (see Table 1 for strain names). Identical sequences are represented only once and the number of sequenced strains is indicated on each leaf. Bootstrap values >50% (shown on the node) are expressed as a percentage of 5,000 replicates. Rs strains grouped differently with the two markers, as illustrated by strains K0157, K0024, K0190 and K0018, which were re-sequenced and are marked with an asterisk. Although the average nucleotide difference between groups of Rs with the ITS marker is high, there is little sequence variation within each individual group (clades A, B and C), and fewer strains are resolved than with the RIF marker. Also, ITS sequence from Ralstonia pickettii strain 12J is placed within Rs clade B on the ITS tree, while RIF sequence from the same strain is placed outside Rs clade B on the RIF tree.
https://doi.org/10.1371/journal.pone.0018496.s003
(TIF)
Figure S4.
RIF sequences distinguish fewer strains of Clavibacter but produce a more robust tree. Unrooted neighbor-joining trees for the RIF and ITS markers were constructed from nineteen Clavibacter strains from the PBC (Supplemental Table S2) and two reference strains from GenBank (see Table 1 for strain names). Identical sequences are represented only once and the number of sequenced strains is indicated. Bootstrap values >50% (shown at the node) are expressed as a percentage of 5,000 replicates.
https://doi.org/10.1371/journal.pone.0018496.s004
(TIF)
Figure S5.
RIF sequences distinguish more Dickeya strains than ITS. Unrooted neighbor-joining trees for the RIF and ITS markers were constructed from twenty-nine Dickeya strains from the PBC (Supplemental Table S2) and three reference strains (with strain names) from GenBank (Supplemental Table S1). Identical sequences are represented only once and the number of sequenced strains is indicated. Bootstrap values >50% (shown at the node) are expressed as a percentage of 5,000 replicates.
https://doi.org/10.1371/journal.pone.0018496.s005
(TIF)
Table S1.
Copy number and location of the dnaA genes in 1,067 sequenced NCBI strains.
https://doi.org/10.1371/journal.pone.0018496.s006
(PDF)
Table S3.
X. albilineans is more distantly related to other Xanthomonas species than is Stenotrophomonas maltophilia .
https://doi.org/10.1371/journal.pone.0018496.s008
(PDF)
Table S4.
Average between species distances of the RIF marker from twenty-three different Ralstonia RIF sequences.
https://doi.org/10.1371/journal.pone.0018496.s009
(PDF)
Table S5.
Average between group distances of the RIF marker from ten different Pectobacterium RIF sequences.
https://doi.org/10.1371/journal.pone.0018496.s010
(PDF)
Table S6.
Average between group distances of the RIF marker from ten different Dickeya RIF sequences.
https://doi.org/10.1371/journal.pone.0018496.s011
(PDF)
Table S7.
Average between subspecies distances of the RIF sequence from eleven different Clavibacter michiganensis RIF sequences.
https://doi.org/10.1371/journal.pone.0018496.s012
(PDF)
Table S8.
In silico comparison of Xanthomonas with the RIF marker resolves one pair of closely related strains that is unresolved with four other housekeeping genes and the ITS.
https://doi.org/10.1371/journal.pone.0018496.s013
(PDF)
Text S1.
Computational derivation of a universal DNA marker from fifteen completely sequenced genomes was unsuccessful.
https://doi.org/10.1371/journal.pone.0018496.s014
(DOC)
Acknowledgments
We sincerely thank Duck-Chul Park and John Young for providing DNA extracts of key Xanthomonas accessions.
Author Contributions
Conceived and designed the experiments: KLS GGP. Performed the experiments: KLS GM. Analyzed the data: KLS GM AMA GGP. Contributed reagents/materials/analysis tools: AMA GGP. Wrote the paper: KLS GGP. Wrote Perlscripts for comparative genomics: KLS. Designed and implemented RIFdb: KLS.
References
- 1. Baker B, Zambryski P, Staskawicz B, Dinesh-Kumar SP (1997) Signaling in plant-microbe interactions. Science 276: 726–723.
- 2. Gottwald T, Hughes G, Graham JH, Sun X, Riley T (2001) The citrus canker epidemic in Florida: the scientific basis of regulatory eradication policy for an invasive species. Phytopathol 91: 30–34.
- 3. Hawks B (2005) Agricultural Bioterrorism Protection Act of 2002: Possession, use, and transfer of biological agents and toxins; final rule. Fed Regist 70: 13241–13292.
- 4.
Hayward AC (1993) The hosts of Xanthomonas. In: Swings JG, Civerolo EL, editors. Xanthomonas. London: Chapman and Hall. pp. 1–18.
- 5.
Denny TP (2006) Plant pathogenic Ralstonia species;. In: Gnanamanickam SS, editor. Plant-associated Bacteria. The Netherlands: Springer. pp. 573–644.
- 6.
Eichenlaub R, Gartemann KH, Burger A (2006) Clavibacter michiganensis, a group of gram-positive phytopathogenic bacteria;. In: Gnanamanickam SS, editor. Plant-associated Bacteria. The Netherlands: Springer. pp. 131–155; 195–351.
- 7. Simpson AJG, Reinach FC, Arruda P, Abreu FA, Acencio M (2000) The genome sequence of the plant pathogen Xylella fastidiosa. Nature 406: 151–159.
- 8. Bakker PAHM, Pieterse CMJ, van Loon LC (2007) Induced systemic resistance by fluorescent Pseudomonas spp. Phytopathol 97: 239–243.
- 9. Samson R, Legendre JB, Christen R, Fischer-Le Saux M, Achouak W, et al. (2005) Transfer of Pectobacterium chrysanthemi (Burkholder et al. 1953) Brenner et al. 1973 and Brenneria paradisiaca to the genus Dickeya gen. nov. as Dickeya chrysanthemi comb. nov. and Dickeya paradisiaca comb. nov. and delineation of four novel species, Dickeya dadantii sp. nov., Dickeya dianthicola sp. nov., Dickeya dieffenbachiae sp. nov. and Dickeya zeae sp. Int J Syst Evol Microbiol 55: 1415–1427.
- 10. Hauben L, Moore ER, Vauterin L, Steenackers M, Mergaert J, et al. (1998) Phylogenetic position of phytopathogens within the Enterobacteriaceae. Syst Appl Microbiol 21: 384–97.
- 11. Gavini F, Mergaert J, Beji A, Mielcarek C, Izard D, et al. (1989) Transfer of Enterobacter agglomerans (Beijerinck 1888) Ewing and Fife 1972 to Pantoea gen. nov. as Pantoea agglomerans comb. nov. and description of Pantoea dispersa sp. nov. Int J Syst Evol Microbiol 39: 337–345.
- 12.
Garrity GM, Bell JA, Lilburn T (2005) Phylum XIV. Proteobacteria phyl. nov. In: Brenner DJ, Krieger-Huber S, Stanley JT, editors. Bergey's manual of systematic bacteriology. New York: Springer. pp. 1–912.
- 13. Vauterin L, Hoste B, Kersters K, Swings J (1995) Reclassification of Xanthomonas. Int J Syst Bacteriol 45: 472–489.
- 14. Buddenhagen I, Kelman A (1964) Biological and Physiological Aspects of Bacterial Wilt Caused by Pseudomonas solanacearum. Annu Rev Phytopathol 2: 203–230.
- 15. Jones JB, Lacy GH, Bouzar H, Stall RE, Schaad NW (2004) Reclassification of the xanthomonads associated with bacterial spot disease of tomato and pepper. Syst Appl Microbiol 27: 755–762.
- 16. Schaad NW, Postnikova E, Lacy G, Sechler A, Agarkova I, et al. (2006) Emended classification of xanthomonad pathogens on citrus. Syst Appl Microbiol 29: 690–695.
- 17.
Young JM, Bull CT, De Boer SH, Dirrao G, Gardan L, et al. (2001) Committee on the Taxonomy of Plant Pathogenic Bacteria. International Standards for Naming Pathovars of Phytopathogenic Bacteria. http://www.isppweb.org/about_tpbb_naming.asp.
- 18.
Fegan M, Taghavi M, Sly LI, Hayward AC (1998) Phylogeny, diversity, and molecular diagnostics of Ralstonia solanacearum;. In: Prior P, Allen C, Elphinstone JG, editors. Bacterial Wilt Disease: Molecular and Ecological Aspects. Berlin: Springer-Verlag. pp. 19–33.
- 19.
Fegan M, Prior P (2005) How complex is the “Ralstonia solanacearum species complex”? In: Allen C, Prior P, Hayward AC, editors. Bacterial wilt disease and the Ralstonia solanacearum species complex. St. Paul, MN: APS Press. pp. 449–461.
- 20.
Normand P, Ponsonnet C, Nesme X, Neyra M, Simonet P (1996) ITS analysis of prokaryotes;. In: Akkermans DL, van Elsas JD, de Bruijn EI, editors. Molecular Microbial Ecology Manual. Dordrecht: Kluwer Academic Publishers. pp. 1–12.
- 21. Ma B, Hibbing ME, Kim H, Reedy RM, Yedidia I, et al. (2007) Host range and molecular phylogenies of the soft rot enterobacterial genera Pectobacterium and Dickeya. Phytopathol 97: 1150–1163.
- 22. Young JM, Park DC (2007) Relationships of plant pathogenic enterobacteria based on partial atpD, carA, and recA as individual and concatenated nucleotide and peptide sequences. Syst Appl Microbiol 30: 343–354.
- 23. Prior P, Fegan M (2005) Recent developments in the phylogeny and classification of Ralstonia solanacearum. Acta Hort 695: 127–136.
- 24. Castillo JA, Greenberg JT (2007) Evolutionary Dynamics of Ralstonia solanacearum. Appl Environ Microbiol 73: 1225–1238.
- 25. Young JM, Park DC, Shearman HM, Fargier E (2008) A multilocus sequence analysis of the genus Xanthomonas. Syst Appl Microbiol 31: 366–77.
- 26. Parkinson N, Aritua V, Heeney J, Cowie C, Bew J, et al. (2007) Phylogenetic analysis of Xanthomonas species by comparison of partial gyrase B gene sequences. Int J Syst Evol Microbiol 57: 2881–2887.
- 27. Parkinson N, Cowie C, Heeney J, Stead D (2009) Phylogenetic structure of Xanthomonas determined by comparison of gyrB sequences. Int J Syst Evol Microbiol 57: 2881–2887.
- 28. Yabuuchi E, Kosako Y, Yano I, Hotta H, Nishiuchi Y (1995) Transfer of two Burkholderia and an Alcaligenes species to Ralstonia gen. Nov.: Proposal of Ralstonia pickettii (Ralston, Palleroni and Doudoroff 1973) comb. Nov., Ralstonia solanacearum (Smith 1896) comb. Nov. and Ralstonia eutropha (Davis 1969) comb. Nov. Microbiol Immunol 39: 897–904.
- 29. Frézal L, Leblois R (2008) Four years of DNA barcoding: Current advances and prospects. Infect Genet and Evol 8: 727–736.
- 30. Hebert PDN, Cywinska A, Ball SL, DeWaard JR (2003) Biological identifications through DNA barcodes. Proc R Soc London Ser 270: 313–321.
- 31. Presting GG (2006) Identification of conserved regions in the plastid genome: implications for DNA barcoding and biological function. Can J Bot 84: 434–1443.
- 32. Sherwood AR, Presting GG (2007) Universal primers amplify a 23S rDNA plastid marker in eukaryotic algae and cyanobacteria. J Phycol 43: 605–608.
- 33. Wang N, Sherwood A, Kurihara A, Conklin K, Sauvage T, et al. (2009) The Hawaiian Algal Database: a laboratory LIMS and online resource for biodiversity data. BMC Plant Biol 9: 117.
- 34. Sherwood AR, Chan YL, Presting GG (2008) Application of universally amplifying plastid primers to environmental sampling of a stream periphyton community. Mol Ecol Resour 8: 1011–1014.
- 35. Cole JR, Wang Q, Cardenas E, Fish J, Chai B, et al. (2009) The Ribosomal Database Project: improved alignments and new tools for rRNA analysis. Nucleic Acids Res 37: D141–D145.
- 36. Benson DA, Boguski MS, Lipman DJ, Ostell J, Ouellette BF, et al. (1999) GenBank. Nucleic Acids Res 27: 12–17.
- 37. García-Martínez J, Acinas SG, Antón AI, Rodríguez-Valera F (1999) Use of the 16S–23S ribosomal genes spacer region in studies of prokaryotic diversity. J Microbiol Methods 36: 55–64.
- 38. Gurtler V, Stanisich VA (1996) New approaches to typing and identification of bacteria using the 16S–23S rDNA spacer region. Microbiol 142: 3–16.
- 39. Kang YJ, Cheng J, Mai LJ, Hu J, Piao Z (2010) Multiple copies of 16S rRNA gene affect the restriction patterns and DGGE profile revealed by analysis of genome database. Microbiol 79: 655–662.
- 40. Wilton SD, Lim L, Dye D, Laing N (1997) Bandstab: a PCR-based alternative to cloning PCR products. BioTechniques 22: 642–645.
- 41. Urwin R, Maiden MCJ (2003) Multi-locus sequence typing: a tool for global epidemiology. Trends Microbiol 11: 479–487.
- 42. Pascual J, Macián MC, Arahal DR, Garay E, Pujalte MJ (2010) Multilocus sequence analysis of the central clade of the genus Vibrio using the 16S rRNA, recA, pyrH, rpoD, gyrB, rctB and toxR genes. Int J Syst Evol Microbiol 60: 154–165.
- 43. Santos S, Ochman H (2004) Identification and phylogenetic sorting of bacterial lineages with universally conserved genes and proteins. Environ Microbiol 6: 754–759.
- 44. da Silva ACR, Ferro JA, Reinach FC, Farah CS, Furlan LR, et al. (2002) Comparison of the genomes of two Xanthomonas pathogens with differing host specificities. Nature 417: 459–463.
- 45. Qian W, Jia Y, Ren SX, He YQ, Feng JX, et al. (2005) Comparative and functional genomic analyses of the pathogenicity of phytopathogen Xanthomonas campestris pv. campestris. Genome Res 15: 757–767.
- 46. Lee BM, Park YJ, Park DS, Kang HW, Kim JG, et al. (2005) The genome sequence of Xanthomonas oryzae pathovar oryzae KACC10331, the bacterial blight pathogen of rice. Nucleic Acids Res 33: 577–86.
- 47. Thieme F, Koebnik R, Bekel T, Berger C, Boch J, et al. (2005) Insights into genome plasticity and pathogenicity of the plant pathogenic bacterium Xanthomonas campestris pv. vesicatoria revealed by the complete genome sequence. J Bacteriol 187: 7254–7266.
- 48.
Ewing A (2008) A study on the phylogenetics of gene transfer: from pathways to kingdoms. MS Thesis, University of Hawaii at Manoa, Honolulu, Hi.
- 49. Ewing B, Green P (1998a) Basecalling of automated sequencer traces using phred. II. Error probabilities. Genome Res 8: 186–194.
- 50. Ewing B, Hillier L, Wendl M, Green P (1998b) Basecalling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res 8: 175–185.
- 51. Nassar A, Darrasse A, Lemattre M, Kotoujansky A, Dervin C, et al. (1996) Characterization of Erwinia chrysanthemi by pectinolytic isozyme polymorphism and restriction fragment length polymorphism analysis of PCR-amplified fragments of pel genes. Appl Environ Microbiol 62: 2228–2235.
- 52. Kaguni J (2006) DnaA: Controlling the Initiation of Bacterial DNA Replication and More. Annu Rev Microbiol 60: 351–371.
- 53. Mackiewicz P, Zakrzewska-Czerwiñska J, Zawilak A, Dudek MR, Cebrat S (2004) Where does bacterial replication start? Rules for predicting the oriC region. NAR 32: 3781–3791.
- 54. Almeida N, Yan S, Cai R, Clarke CR, Morris CE, et al. (2010) PAMDB, a multilocus sequence typing and analysis database and website for plant-associated microbes. Phytopathology 100: 208–215.
- 55. Vitorino L, Chelo I, Bacellar F, Ze-Ze L (2007) Rickettsiae phylogeny: a multigenic approach. Microbiology 153: 160–168.
- 56. Zmasek CM, Eddy SR (2001) ATV: display and manipulation of annotated phylogenetic trees. Bioinformatics 17: 383–384.
- 57. Kurtz S, Phillippy A, Delcher AL, Smoot M, Shumway M, et al. (2004) Versatile and open software for comparing large genomes. Genome Biol 5: R12.
- 58. Chenna R, Sugawara H, Koike T, Lopez R, Gibson TJ, et al. (2003) Multiple sequence alignment with the Clustal series of programs. Nucleic Acids Res 31: 3497–3500.
- 59. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search tool. J Mol Biol 215: 403–410.
- 60. de Lamballerie X, Zandotti C, Vignoli C, Bollet C, de Micco P (1992) A one-step microbial DNA extraction method using “Chelex 100” suitable for gene amplification. Res Microbiol 143: 785–790.
- 61. Hall T (1999) BioEdit 7. Nucleic Acids Symp Ser 41: 95–98.
- 62. Kumar S, Tamura K, Nei M (2004) MEGA 3: Integrated Software for Molecular Evolutionary Genetics Analysis and Sequence Alignment. Brief Bioinform 5: 150–163.