The PXO99^A genome is a single circular chromosome of 5,240,075 bp with an overall GC content of 63.6%. It contains 5083 protein-coding genes, 2 ribosomal RNA operons, and 55 tRNAs (Table 1). The origin of replication was identified by similarity to other Xanthomonas genomes, by proximity to genes (dnaA, dnaN, and gyrB) often found near the origin on bacterial genomes, and by GC-skew analysis, which examines the excess of G versus C on the leading strand 18. A schematic representation of the genome is provided in Figure 1.

To assess the phylogenetic relationships among PXO99^A and related strains, we aligned the complete genome to the genomes of MAFF, and KACC, and strain BLS256 of X. oryzae pv. oryzicola, (GenBank Accession AAQN01000001), and generated a cladogram using Mauve 2.1.1 19. MAFF and KACC group together, but PXO99^A is clearly distinct and considerably more distant from MAFF and KACC than they are from one another (Additional file 1). The tree was confirmed by another tree built with all the sequenced Xanthomonas genomes and rooted with Xylella fastidiosa (Temecula strain) (data not shown).

Of the 5083 annotated protein coding genes in PXO99^A, 4910 have clear homologs in the MAFF strain. These genes map to just 4234 genes in MAFF (out of 4372 total), indicating a considerable expansion of some gene families. 194 of the shared genes are present in a 212 kb direct repeat near the replication terminus (see below). Of the remaining 173 PXO99-specific genes, 29 (including 18 tranposases) are missing from MAFF because they span breakpoints; i.e., a rearrangement, insertion, or deletion in MAFF has broken these genes into fragments. Fifty eight other PXO99^A genes only partially align to MAFF, including 29 transposases. Finally, 86 genes in PXO99^A are completely absent (based on sequence alignment) from MAFF.

Among the 138 annotated genes in the MAFF strain that are not present in PXO99^A, 20 are missing in PXO99^A because they span breakpoints, and 38 (including 12 transposases) are missing because they are truncated in PXO99^A. The remaining 80 genes in MAFF are entirely missing from PXO99^A.

Additional file 2 contains the lists of genes unique to PXO99^A and unique to MAFF. It is noteworthy that a majority of the genes unique to MAFF (64/80) are hypothetical proteins, which may represent annotation artifacts. These hypothetical genes have an average length of 182 bp, compared to 850 bp for an average gene. Of the 87 genes unique to PXO99^A, twenty are hypothetical while the remainder comprises genes similar to predicted genes in other strains and species.

All sequenced Xanthomonas genomes contain numerous IS elements, but the Xoo genomes contain the most diverse pool 20. Of the 19 known families of IS elements 21, eight families composed of 28 distinct elements appear in Xoo. MAFF and KACC have nearly identical numbers of IS elements (Table 1), while PXO99^A contains fewer elements overall, but more copies of ISXo8, IS1114/ISXoo4, and ISXo2.

Sequences unique to PXO99^A relative to MAFF include a 38,766 bp region (coordinates 4788763 – 4827529) that contains several predicted non-fimbrial adhesin genes (Figure 2). Of 20 genes at this locus, three (fhaB, fhaX and fhaB1) encode non-fimbrial adhesin related proteins and a fourth (fhaC) is predicted to help in transport of non-fimbrial adhesins. The fhaB gene, which encodes the longest protein (3527 aa) in PXO99^A, contains a hemagglutination activity domain and filamentous hemagglutinin repeats that are likely to serve in adhesion and autoaggregation. Two more genes (ORFs 2986 and 2987) are predicted to be involved in bacteriocin secretion while another (ORF 2973) encodes an ice-nucleation protein homolog. The locus also includes several IS elements and is flanked by direct repeats of ISXo5. These are in turn flanked by genes for a dual specificity phosphatase (DSP in Figure 2) and a DNA binding protein (DBP). In contrast, only one copy of the ISXo5 element is present between DSP and DBP in MAFF and KACC, indicating that the IsXo5 element was involved in the genomic rearrangement that led either to loss of the locus from MAFF and KACC or gain of the locus in PXO99^A. The former is likely the case because the arrangement in PXO99^A is present also in X. oryzae pv. oryzicola BLS256 (data not shown).

Specific primers were developed for the DSP and DBP genes that flank this locus as well as for the fhaB, fhaC, and fhaX genes. Using PXO99^A genomic DNA as a template, we amplified the expected PCR products for all five genes (data not shown). Using either MAFF 311018 or KACC 10331 genomic DNA as template, products of the expected size were obtained with primers specific to the DBP and DSP encoding genes, but no products were obtained with primers specific for fhaC, fhaB or fhaX. Also, a fragment of the expected size (~2.5 kb) was obtained via PCR with DBP- and DSP-specific primers using MAFF and KACC genomic DNA, but not with PXO99^A genomic DNA (data not shown). These results provide additional evidence that the non-fimbrial adhesin genes are indeed missing from the MAFF and KACC genomes. Based on PCR analysis using the above primers, the fhaC, fhaB and fhaX genes are also missing from the Indian Xoo strain BXO43, and in another Indian strain, BXO8, only fhaB appears to be present. However, all three genes were detected in the strain Nepal624 (data not shown), a result consistent with the close relationship, as established by DNA fingerprinting studies, between PXO99^A and Xoo strains from Nepal 22.

The PXO99^A strain contains a near-perfect tandem duplication of 212,087 bp. This unusually large repeat spans the intervals 2,502,622–2,714,708 and 2,714,709–2,926,795. The repeat is flanked by an insertion (1073 bp) of ISXo5 (Figure 5) at each end and between the two copies. Except for a single base difference in one IS copy, the two regions are 100% identical. Because the flanking ISXo5 is longer than a read, and because the repeat is much too long to be spanned by any pair of sequencing reads, the original assembly had collapsed these two repeats into a single region. Also, the positioning of the flanking short repeats meant that every sequence fit accurately into the collapsed assembly, with only the paired-end information indicating a problem. This collapse was discovered through the use of the Hawkeye assembly diagnostics tool, which identified a large set of mis-oriented paired-end sequences on either end of the collapsed version of the assembly 23. In order to provide additional validation of this duplication, we designed primers on either side of the unique junction where the two copies of the tandem repeat meet (see Additional file 2). We verified the presence of the junction by PCR amplification and re-sequencing of this region.

The 212 kb segment occurs once in the MAFF and KACC sequences. One question is whether the difficulty of assembling this region might mean that it is present in these strains, but undetected. Evidence that the duplication is indeed unique to PXO99^A is the sequence divergence (~0.3%) of PXO99^A from MAFF/KACC. This divergence implies that if the duplication had happened in a common ancestor, then the two distinct 212 kb regions, which would have existed since the divergence between strains, would be expected to have over 600 single-base differences. The fact that the copies have only one difference confirms that the large duplication in PXO99^A occurred much more recently than its divergence from MAFF and KACC.

A hallmark of the Xoo genome is the large number of transcription activator-like (TAL) type III effector genes, which are defined by their relatedness to the type members avrBs3 and pthA 242526. TAL effector genes are characterized in part by a region of 102 bp repeats, or more rarely 105 bp repeats, within the central coding portion 2728. Nineteen TAL effector genes were identified in the PXO99^A genome (Table 2 and Figure 3), including four previously associated with virulence and avirulence phenotypes and effector-specific gene expression in rice 162930. One of these, pthXo1, encodes the major virulence determinant for PXO99^A whose function is disrupted in rice by the recessive blight resistance gene xa13 2930. The TAL effector genes are located in nine loci distributed in the genome. Two loci consist of single genes, six consist of two genes oriented in the same direction, and one is a previously identified cluster of five genes all oriented in the same direction 16. Each of the genes within a cluster is preceded by a region of 990 bp that contains two or more short, predicted ORFs but is more likely non-coding DNA, suggesting that each gene has its own promoter, and that the clusters do not represent polycistronic operons. We have designated the genes numerically according to the locus in which they reside, sequentially from the origin of replication, and alphabetically, according to their position in that locus starting at the 5' end of the locus. Thus, the first TAL effector gene in the genome sequence, proximal to the origin, is tal1, the second (which is the second gene in a locus oriented toward the origin) tal2b, etc. The genes with known phenotypes are distributed in separate loci: tal1 is pthXo7, tal2b is pthXo1, tal5b is pthXo6, and tal9c is avrXa27. Among the genes, the number of repeat units varies from 12.5 (tal9d) to 26.5 (tal9c). None of the genes contains the rare 105 bp repeat. Gene pairs in loci 7 and 8 are identical copies in the 212 kb duplicated regions of the genome.

With the exception of the gene pairs within the 212 kb duplication, none of the genes share the same repeat region structure based on a comparison of the twelfth and thirteenth codons, which vary from repeat to repeat (Figure 4). Genes tal3a and tal3b each have two deletions of 43 and 15 codons in their 5' ends and are truncated in the 3' ends of their coding regions, so they are unlikely to produce functional effectors. The similarities in tal3a and tal3b indicate that one is a duplicate of the other. Gene tal6b has a frameshift mutation within the 5'-end of the coding region and is therefore also unlikely to be functional. Genes tal6b, tal7b, and tal8b share a novel eleven codon duplication (PERTSHRVADL-PERTSNRVADL) at their 3' ends.

Comparison of the TAL effector gene content and arrangement in PXO99^A with those in MAFF, using the scheme described above to name the MAFF genes (genes in the two strains are hereafter distinguished by subscript), indicates that the number of loci and relative positions are similar with the exception of the duplicated loci (7 and 8) in PXO99^A, as well as PXO99^A locus 4 and MAFF locus 3, which occupy unique relative positions in their respective genomes (Figure 3). MAFF loci 2 and 3 were considered by Ochiai et al. 9 as one locus, but we treat them as distinct based on the unusual distance (roughly 3 kb instead of the usual 990 bp) between the locus 3 gene and the closest locus 2 gene, and the presence of IS elements flanking locus 3. Despite the similarity in number and arrangement of the respective loci, only three PXO99^A TAL effector genes, all in PXO99^A locus 9, have counterparts in MAFF that are identical with respect to the number of repeats and the twelfth and thirteenth codons of the central repeat domain. The identical genes are tal9c_PXO99A and tal1c_MAFF, tal9d_PXO99A and tal1d_MAFF, and tal9e_PXO99A and tal3_MAFF. Genes tal9a_PXO99A and tal9b_PXO99A correspond in repeat number to tal1a_MAFF and tal1b_MAFF, respectively. The tal3_MAFF gene, which represents a break in the apparent overall synteny between PXO99^A locus 9 and MAFF locus 1, is flanked by IS elements. The tal9c_PXO99A gene is the avirulence determinant avrXa27, and its identity with tal1c_MAFF is consistent with the effectiveness of the corresponding host resistance gene Xa27 against both PXO99^A and MAFF, as well as a broad range of other strains 31. Two other PXO99^A TAL effector genes have counterparts in MAFF that are nearly identical with respect to the number of repeats and the predicted twelfth and thirteenth residues in each repeat: tal4_PXO99A has the same structure as tal4a_MAFF except for residue 12 in the fifteenth repeat, and tal6a_PXO99A has the same structure as tal5a_MAFF except for residues 12 and 13 in the fourteenth repeat. MAFF has two TAL effector genes, pthXo2 (tal8_MAFF) and avrXa7 (tal7d_MAFF), that are major virulence determinants 30. The pthXo2 gene occupies the same locus in MAFF that pthXo7 does in PXO99^A, while avrXa7 occupies the same locus as pthXo1, the major virulence determinant for PXO99^A. Some corresponding loci differ in their gene content. For example, locus 2 in PXO99^A consists of two genes but the corresponding locus in MAFF, locus 7, contains four. Absent from MAFF locus 5 is pthXo6, although previous evidence indicates that OsTFX1, a host gene expressed in a pthXo6-dependent manner, is induced upon infection with MAFF 32. Induction could be due to one of the other TAL effectors, or pthXo6 might have been misassembled in the MAFF sequence. Locus 6 in PXO99^A corresponds to locus 4 in MAFF, but locus 4 in MAFF contains the gene nearly identical to tal4_PXO99A in the uniquely positioned locus 4 of PXO99^A. The MAFF gene nearly identical to tal6a_PXO99A, is located in a corresponding neighboring locus, MAFF locus 5. Locus 3 in PXO99^A and 6 in MAFF contain two and one defective TAL effector genes, respectively. All three of these genes have identical repeat domains. Moreover, tal6_MAFF shares with the PXO99^A genes the 3' deletions of 43 and 15 codons discussed above, as well as a six-codon deletion in the repeat region (repeat 4 of tal6_MAFF, repeat 3 of tal3a_PXO99A, and repeat 4 of tal3b_PXO99A), indicating that these genes may represent a generally defunct locus in Xoo. The observed substitution of genes at conserved loci across the genomes, expansion or contraction of individual loci in a given strain, and divergence or degeneration of gene sequences at shared loci are presumably accomplished by the exchange of coding sequences through homologous recombination. Transposition of genes involving IS element-mediated recombination may also occur, as exemplified possibly by tal3_MAFF.

The PXO99^A strain of Xoo has experienced at least ten major rearrangements with respect to the MAFF strain, resulting in 29 distinct syntenic blocks, as shown in Figure 5. The majority of these rearrangements are symmetric about the origin of replication, as has been observed for many other bacterial rearrangements 33. Most of these rearrangements appear to be mediated by a diverse set of transposable elements. Some elements, such as ISXo5, ISXo8, and IS1389/ISXoo3, are responsible for multiple rearrangements. For example, ISXo5 occurs near each endpoint of both copies of the 212,087 bp tandem repeat (region E, Figure 5). Within each copy of the repeat there is a 116,872 bp inversion flanked by inverted copies of ISXo5. Only three major rearrangement events (H, I, and J in Figure 5) do not seem to be associated with IS elements.

The PXO99^A, MAFF, and KACC genomes each contain a CRISPR (clustered regularly interspersed short palindromic repeats) element. CRISPRs are identified by a set of Cas genes, followed by a leader sequence and then a variable number of alternating spacers and repeats; the elements here represent the Dvulg subtype 34. The repeats are identical, while the spacers represent foreign DNA that was laterally transferred from a bacteriophage or a plasmid 35. A growing body of evidence demonstrates that the spacers, acquired during phage infection, provide immune protection for the bacterium against the phage 36. Thus CRISPRs represent an inheritable immune system for bacteria.

Because the CRISPR region evolves very rapidly, it provides one of the most striking records of differentiation among PXO99^A, MAFF, and KACC. As shown in Figure 6, PXO99^A has the largest CRISPR region of the three strains, with 75 spacer elements. In contrast, MAFF and KACC contain just 48 and 59 spacers respectively, implying that PXO99^A has acquired a substantially greater resistance to phage infections than its cousins. Also worth noting is that the majority of the spacers are unique to each strain, attesting to the rapid evolution of these regions.

The alignment of the CRISPR spacers in the three Xoo strains (Figure 6) appears on first inspection to contradict the phylogenetic relationship of the strains, in that MAFF appears more distant from the other two strains. Spacers are inserted into a genome in chronological order, with new elements appearing next to the 188-bp leader sequence, which gives a clear picture of the shared history of these elements. Our alignment shows that all three strains share five of the oldest elements (S1–S5 in PXO99^A), but that all of the more recent elements in MAFF are unique to that strain. PXO99^A and KACC share the very oldest element, which has been lost in MAFF, as well as 10 additional older spacer elements in conserved order. These 10 spacers range from S6–S25 in PXO99^A and S8–S30 in MAFF (intervening elements are unique in each strain), indicating that these two strains diverged after the acquisition of spacer S25/S30. MAFF, in contrast, shares no spacers more recent than S5 with either of the other two strains. This appears to contradict whole-genome phylogenetic evidence and large-scale genome structure, both of which indicate that MAFF and KACC are much closer to one another than either is to PXO99^A. A likely alternative explanation, given the hypervariable nature of CRISPRs, is that MAFF lost these older spacers.

To validate the large-scale rearrangements between strains PXO99^A and MAFF, we obtained a library of 9 kb shotgun clones for MAFF and identified those clones that correspond to breakpoints shown in Figure 5. Two clones for each breakpoint were selected, except in one case where only one clone could be identified. These clones were end-sequenced and the ends compared to the MAFF genome. In addition, restriction enzyme analysis was performed for each of the shotgun clones.

In all cases, the analysis of the MAFF sequences confirmed that the MAFF genome is correctly assembled. Had there been any mis-assemblies, the clones would have shown significant length polymorphisms or would have mapped to inconsistent positions on the finished sequence. This evidence further strengthens the conclusion that breakpoints in the genome alignment between MAFF and PXO99^A represent genuine differences between the genomes. Because the MAFF and KACC strains have almost the same overall genome architecture, with very few rearrangements, we did not attempt separate verification of the KACC assembly.

Separately, we identified 18 significant insertions and deletions between MAFF and PXO99^A. We generated PCR primers to test for the presence or absence of each insertion, and amplified fragments from genomic DNA using both strains. In all cases the PCR tests verified the presence of the insertion in one strain and its absence in the other (data not shown).

GC-content frequently used for identifying regions of a genome with unusual composition, as might result from lateral gene transfer. PXO99^A has a GC-content of 63.6%, ranging from a high of 71.8% to a low of 41.6%. A more sensitive measure of unusual composition, used in many previous studies (e.g., 37) is based on trinucleotide composition. For this measure, we compute the X² statistic to compare the trinucleotide distribution in fixed-size windows to the overall trinucleotide distribution for the genome. Regions highlighted by this statistic are either caused by lateral gene transfer or else under very strong evolutionary constraints to maintain their atypical DNA composition. A plot of the X² statistic as well as GC-content across the genome is shown in Figure 7.

The figure shows multiple regions of highly unusual composition, which we then investigated further. The largest peak in the X² distribution, at position 918,000, is centered on a 424-aa protein (ORF04252) containing a lysin domain (often found in enzymes involved in bacterial cell wall degradation) but whose function is otherwise unknown. There is strong evidence that this gene has been laterally transferred via a bacteriophage: it is not found in any other Xanthomonads, and the closest matches are in Burkholderia, Campylobacter, and Shewanella, all very distantly related genera. Homologs in both B. pseudomallei K96243 38 and Erythobacter litoralis are annotated as acquired from bacteriophage, and a direct phage homolog occurs in Burkholderia phage phiE202. A phylogenetic tree of all homologs (data not shown) supports the conclusion that this gene was laterally transferred via a phage.

The second-highest peak in Figure 7 is in the midst of a broader region of unusual composition, extending from 3,540,900 to 3,571,800. This region contains a large prophage element with 41 phage-related genes, extending from ORF01364 (a phage portal protein, pbsx family) to ORF01326 (a site-specific recombinase, phage integrase family). PXO99^A contains a second, smaller prophage element spanning six genes from 2366221–2371236.

All 19 of the TAL effector genes show an unusual composition and correspond to peaks in Figure 7. Because the TAL effectors are adjacent to transposases, they too might have originated in another species, possibly as a single-copy gene that later expanded in number in Xoo or a progenitor. Conservation of the unusual composition in all members of the family might also reflect strong functional constraints.

A significant fraction of predicted genes in most bacterial genomes are annotated as hypothetical proteins. These open reading frames (ORFs) are predicted computationally, but because they lack sequence homology to other species, they cannot be assigned a name. An unknown number of these predicted genes are likely to be false predictions, and for most genomes there has been little basis for distinguishing true genes at the time of sequencing. For PXO99^A, we took advantage of the related MAFF and KACC genomes to improve upon the usual set of hypothetical predicted genes. Multiple sequence alignments among several closely-related species often reveal that the ORFs of hypothetical proteins are not maintained in sister species; i.e., they contain in-frame stop codons. Although it is possible that these interrupted ORFs are functional in only one of the species, a more parsimonious explanation is simply that the original gene prediction was wrong. This strategy has been used, for example, to identify several hundred incorrectly annotated genes in S. cerevisiae 39, using three related yeast genomes.

We aligned the DNA sequences for all 1273 hypothetical proteins in PXO99^A to the corresponding sequences in MAFF, KACC, X. axonopodis pv. citri, X. campestris pv. campestris, and X. campestris pv. vesicatoria. From these alignments, we identified all predicted PXO99^A genes with premature stop codons or crippling frameshift mutations in any other species. From these data, we identified 78 ORFs with multiple lines of evidence that they did not represent true genes; these predicted genes were deleted from the annotation.

Bacterial genomic DNA was randomly sheared by nebulization, end-repaired with consecutive BAL31 nuclease and T4 DNA polymerase treatments, and size-selected using gel electrophoresis on 1% low-melting-point agarose. After ligation to BstXI adapters, DNA was purified by three rounds of gel electrophoresis to remove excess adapters, and the fragments were ligated into the vector pHOS2 (a modified pBR322 vector) linearized with BstXI. The pHOS2 plasmid contains two BstXI cloning sites immediately flanked by sequencing primer binding sites. These features reduce the frequency of non-recombinant clones, and reduce the amount of vector sequences at the end of the reads. Two libraries with average insert size of 4.5 kb and 10 kb were constructed. The ligation reactions were electroporated into E. coli. Clones were plated onto large format (16 × 16 cm) diffusion plates prepared by layering 150 ml of fresh antibiotic-free agar onto a previously set 50-ml layer of agar containing antibiotic. Colonies were picked for template preparation, inoculated into 384-well blocks containing liquid media, and incubated overnight with shaking. High-purity plasmid DNA was prepared using the DNA purification robotic workstation custom-built by Thermo CRS (Thermo Fisher Scientific, Inc.) and based on the alkaline lysis miniprep 61 and isopropanol precipitation. DNA precipitate was washed with 70% ethanol, dried, and resuspended in 10 mM Tris HCl buffer containing a trace of blue dextran. The yield of plasmid DNA was approximately 600–800 ng per clone, providing sufficient DNA for at least four sequencing reactions per template. Sequencing was done using di-deoxy sequencing method 62. Two 384-well cycle-sequencing reaction plates were prepared from each plate of plasmid template DNA for opposite-end, paired-sequence reads. Sequencing reactions were completed using the Big Dye Terminator chemistry and standard M13 forward and reverse primers. Reaction mixtures, thermal cycling profiles, and electrophoresis conditions were optimized to reduce the volume of the Big Dye Terminator mix and to extend read lengths on the AB3730xl sequencers (Applied Biosystems). Sequencing reactions were set up by the Biomek FX pipetting workstations. Robots were used to aliquot and combine templates with reaction mixes consisting of deoxy- and fluorescently labeled dideoxynucleotides, DNA polymerase, sequencing primers, and reaction buffer in a 5 μl volume. After 30–40 consecutive cycles of amplification, reaction products were precipitated by isopropanol, dried at room temperature, resuspended in water, and transferred to an AB3730xl sequencer. 8,700 and 52,100 high-quality reads from the 4.5 kb and 10 kb insert libraries, respectively, were generated with an average trimmed sequence read length of 821 bp and a success rate of 93%. After initial assembly, gaps were closed by primer walking on plasmid templates, sequencing genomic PCR products that spanned the gaps, and by transposon insertion and sequencing of selected 10 kb shotgun clones.

Multiple rounds of assembly were performed, beginning with the shotgun reads and later including additional finishing reads. In the final assembly, 65,620 reads were trimmed to remove vector and low-quality sequence, and then assembled using Celera Assembler 63. The large (212 kb) tandem repeat was initially collapsed into one copy, which had twice the depth of coverage of the rest of the genome. This anomaly was detected and corrected to two copies after analysis aided by the Hawkeye assembly diagnosis software 23. Protein-coding genes were identified using Glimmer 3.0, which includes an algorithm to identify ribosome binding sites for each gene. Transcription terminators were predicted using TransTermHP 66 with parameter settings expected to yield over 90% accuracy. Transfer RNAs were identified with tRNAScanSE 67. Regions with neither Glimmer predictions nor RNA genes were searched in all six frames using blastx 68 to identify any missed proteins, and all annotations were manually curated as described previously 69, using the Manatee online annotation system 70. The origin and terminus of replication was determined using GC-skew analysis 18, which indicates an origin near position 50 kb and termini near 2,370 kb or 2,510 kb. The chromosome replication initiator gene dnaA, which is commonly found near the origin, is at position 45. Oligomer skew analysis 71, which identifies 8-mers preferentially located on the leading strand, indicates an origin at 4,895 kb (30 kb from the end of the genome) and a terminus at 2,381 kb, based on multiple 8-mers including CCCTGCCC and AGGACCAT. These 8-mers occur 328/376 and 218/248 times (over 87%) on the leading strand; for CCCTGCCC the likelihood that this occurred by chance is 3.6x10^-45. To determine genome rearrangements, the MUMmer/Nucmer suite of genome alignment programs 72 was used to align Xoo PXO99^A to the MAFF and KACC strains as well as to all other Xanthomonas genomes.

Genomic DNA was isolated from PXO99^A, BXO8, Nepal624, KACC10331 and MAFF311018 strains according to the procedure described by Leach et. al. 50. PCR was performed using a set of gene specific primers listed in Additional file 2.

The PXO99^A complete, annotated genome has been deposited in Genbank under accession number CP000967. The traces have been deposited in the NCBI Trace Archive 73 and the complete assembly is in the NCBI Assembly Archive 74.