Background

A diphtheria-like infectious disease caused by Corynebacterium ulcerans is increasing in clinical importance in developed countries and is now regarded as “diphtheria” in Europe [1, 2]. Infection with C. ulcerans occurs in a wide range of hosts, including cats, dogs, pigs, cows, and whales [3–9]. The first clearly documented case of zoonotic transmission involved a dog, as reported by Lartigue et al. [5]. This is in contrast to the causative agent of classical diphtheria, C. diphtheriae, whose host species is thought to be limited to humans [10]. Nevertheless, the two species share a common feature: upon lysogenization of tox-encoding bacteriophages, they become toxigenic and are able to produce the potent diphtheria toxin [1, 10]. This toxin is known to contribute to disease progression, occasionally leading to death. It is encoded by a single gene designated tox, situated inside prophages lysogenized in the bacterial genome of C. diphtheriae[11]. The prophages are capable of induction, by ultraviolet light or DNA-damaging agents such as mitomycin C, and yield β-, δ-, ω- and other functional bacteriophage particles [12]. Some types of bacteriophages can infect both C. diphtheriae and C. ulcerans[13–16]. Furthermore, the C. ulcerans tox gene is also encoded in a genome region surrounded by phage attachment (att) sites conserved between the two species [7, 16]. The nucleotide sequences of C. ulcerans tox genes were published by Sing et al. They showed some diversity in the genetic sequence among C. ulcerans strains, in contrast to the highly conserved C. diphtheriae tox gene [17, 18].

In 2003, the nucleotide sequence of the whole genome of C. diphtheriae strain NCTC13129 was reported [19]. The sequence information revealed some striking features of the bacterial genome, such as the presence of as many as 13 pathogenicity islands (PAIs) [19], uncommon among C. diphtheriae strains [20]. The presence of a tox-positive prophage flanked by the att regions was confirmed and supported the findings of previous reports [21]. Despite comparable clinical importance, the genomic sequence of toxigenic C. ulcerans has not yet been reported. In the present study, we determined the nucleotide sequence of the toxigenic C. ulcerans isolate 0102 genome, obtained in 2001 from the pharyngeal pseudomembrane of a 52-year-old woman presenting with a sore throat and fever. This was the first toxigenic C. ulcerans infection reported in Japan. This patient had been living with nearly 20 cats before the onset of illness [22]. Details of the bacteriological characteristics of the isolate have been described elsewhere [23]. Our analysis was especially directed towards the structure of the tox-positive prophage because of its unexpectedly novel structure.

Results

Genome sequence and genomic information for C. ulcerans 0102

To determine the complete genome sequence of C. ulcerans 0102, obtained short reads were assembled into five contigs by de novo assembly. Each gap was filled by direct PCR and sequencing. A circular chromosome sequence of C. ulcerans 0102 represents 2,579,188 bp, with a G + C content of 53.4% (Additional file 1) and corresponds to the predicted restriction fragment profiles obtained by PFGE analysis (Additional file 2). The chromosome possesses 2,349 coding sequences, 51 tRNA genes, and 4 rrn rRNA operons.

Comparative genome analysis of three pathogenic Corynebacterium spp

Pair-wise sequence alignment revealed a highly conserved synteny among pathogenic Corynebacterium spp. (C. pseudotuberculosis FRC41, C. ulcerans 0102, and C. diphtheriae NCTC 13129; Figure 1). No significant genome rearrangements, such as inversion or transposition events, were observed among the three species, in accordance with previous findings [24]. The sequence similarity suggests that the chromosomes of C. ulcerans 0102 and C. pseudotuberculosis FRC41 are highly similar compared with that of C. diphtheriae NCTC 13129 (Figure 1). Once again, this is in accordance with previous findings in other C. ulcerans strains [24]. Similarly, a neighbor-joining phylogenetic tree, based on the partial sequence of rpoB, indicates that C. ulcerans 0102 is closely related with C. pseudotuberculosis, but clearly distinguishable from the C. diphtheriae clade (Additional file 3). Three prophages, ΦCULC0102-I, -II, -III, were identified in C. ulcerans 0102. One of the prophages, ΦCULC0102-I, carries tox, the gene encoding the diphtheria toxin (Figure 1).

Figure 1
figure 1

Schematic genome comparison. C. ulcerans 0102 (middle) with C. pseudotuberculosis FRC41 (top) and C. diphtheriae NCTC 13129 (bottom) using a BLASTN homology search visualized by the ACT program. The red and blue bars between chromosomal DNA sequences represent individual nucleotide matches in the forward and reverse directions, respectively. BLASTN match scores less than 200 are not shown. A blue box and two yellow boxes represent a tox-positive prophage and other prophages on the chromosome of C. ulcerans 0102, respectively

Full size image

The tox-positive prophage of C. ulcerans 0102

The ΦCULC0102-I prophage of C. ulcerans 0102 is integrated into tRNAArg (CULC0102_t08) (Figure 2), suggesting that the integration site is identical to that in the C. diphtheriae NCTC 13129 corynephage. In contrast, the recently reported C. ulcerans 809 and C. pseudotuberculosis FRC41 genomes possess a phage-related integrase (intC) and a nitric oxide reductase (nor) gene, respectively, instead of a prophage (Figure 2). Putative attachment sequences were similar between both prophages carrying the tox genes (Additional file 4).

Figure 2
figure 2

Schematic representation and comparative analysis of tox -positive prophages and flanking regions. The tox-positive prophage and flanking regions of C. ulcerans 0102 and C. diphtheriae NCTC13129 are shown. The corresponding region of C. pseudotuberculosis FRC41 and C. ulcerans 809 is also shown. Boxes indicate individual coding regions with colors assigned to their functions. GenBank accession numbers are given in parentheses

Full size image

The two tox- positive prophages share the same structural features, with genes aligned in an ‘integrase - packaging - head - tail - lysis - toxin’ orientation (Figure 2). Pair-wise alignment of the prophages indicates a high similarity in the region encoding the putative integrase, the 3′-ends of CULC0102_0211 and CULC0102_0212, tox, and the attachment sites (Figure 2). The major phage machineries encoded in the internal phage region showed low similarity at the nucleotide and amino acid levels (less than 18%) between C. ulcerans 0102 and C. diphtheriae NCTC13129.

Discussion

Whole-genome sequencing has revealed that the C. ulcerans 0102 genome is composed of 2,579,188 bp with a G + C content of 53.4%. These values are similar to those recently reported for C. ulcerans strains 809 (2,502,095 bp, 53.3% G + C) and BR-AD22 (2,606,374 bp, 53.4% G + C) [24]. C. ulcerans 0102 shares many common features with the two previously reported strains, including 12 virulence factors. Strain 0102 is distinctive with respect to the features of prophages integrated in its genome. It possesses a unique tox-positive prophage, ΦCULC0102-I, in its chromosome (Figure 1 and Additional file 1). In the same position of the recently reported C. ulcerans 809 genome exists a remnant phage-related integrase (intC) gene [24] (Figure 2). The C. ulcerans 0102 prophage differs from the corresponding prophage in C. diphtheriae. Although the integrase and tox gene sequences of ΦCULC0102-I showed high similarity to those of the corynephage encoding tox in C. diphtheriae NCTC 13129, the major phage machinery genes in ΦCULC0102-I are distinct from those in other corynephages in C. diphtheriae (Figure 2). This suggests that C. ulcerans 0102 did not immediately acquire the C. diphtheriae tox-positive corynephage.

There are many possible explanations for the origins of these two prophages that are tox-positive but obviously different. One of the simplest explanations we can postulate is outlined in Figure 3. Generally, bacterial prophages are duplicated by excision from chromosomal DNA and subsequent concatenation at both ends of the att sites (Figure 3A). This duplication step indicates that two highly homologous regions, int and tox, could be in close proximity and adjacent to the att site concatenation. It could be speculated that homologous recombination between two prophages may facilitate the acquisition of the tox gene in C. ulcerans 0102 from an unknown tox- positive prophage (Figure 3B) [25]. Horizontal gene transfer is one of the major mechanisms of foreign gene acquisition by bacteria, as reviewed by Ochman et al. [26]. Liu et al. have demonstrated that horizontally transferred genes are often disabled and become pseudogenes. In these cases the genes are no longer beneficial to the recipients [27]. Non-toxigenic C. diphtheriae (CD450, CD119, CD448, and CD443 strains) carry tox pseudogenes that are relatively similar to the tox genes of C. ulcerans (Additional file 5), suggesting that horizontal gene transfer among Corynebacterium spp. might occur. Consistent with previous findings [7, 17, 18, 28], tthe tox gene in C. ulcerans 0102 is not identical to that of C. diphtheriae (Additional file 5); phylogenetic analysis of tox showed greater heterogeneity among C. ulcerans isolates than that for C. diphtheriae isolates (Additional file 5).

Figure 3
figure 3

Schema of the diphtheria toxin acquisition hypothesis. (A) Pair-wise comparison of regions with high similarity between C. ulcerans and C. diphtheriae. These structures of putative phages are constructed by connecting attachment sites. The plots above and below represent the GC content calculated with a window size of 500 bp. (B) Schematic representation of how diphtheria toxin has been acquired in C. ulcerans

Full size image

The C. diphtheriae tox gene is highly conserved among temporally and geographically diverse strains [29], therefore greater variation in tox genes from C. ulcerans isolates suggests that this strain might have acquired the tox gene before C. diphtheriae.

In a recent report, whole genome sequence analysis of non-toxigenic C. ulcerans 809 and BR-AD22 [24], the β-corynephage-like truncated integrases (CULC809_00176 and CULC22_00173) are located adjacent to the tRNAArg gene, similar to ΦCULC0102-I in C. ulcerans 0102 and C. diphtheriae. The tRNAArg gene (CULC0102_t08) appears to be a ‘hotspot’ for the acquisition of ΦCULC0102-I-like prophages by homologous integrase.

The whole genome sequences of C. ulcerans 809 and BR-AD22 contain possible virulence factors, such as corynebacterial protease (CP40), phospholipase D (Pld), neuraminidase (NanH), venom serine protease (Vsp1), trypsin-like serine protease (TspA), Rpf interacting protein (RpfI), cell wall-associated hydrolase (CwlH), and five surface-anchored proteins (SpaB–F) [24]. The SpaA-type pilin, encoded by the spaABC srtA gene cluster, is considered to play a crucial role in adhesion of C. diphtheriae[30]. The gene encoding the shaft protein of SpaA-type pilin (spaA) was absent in C. ulcerans 0102, a feature consistent with previous findings in C. ulcerans 809 and BR-AD2 [24]. As SpaB and SpaC proteins, which are assumed to be present in all three C. ulcerans strains, can contribute to host-cell adhesion in the absence of SpaA [30], this may imply a common mechanism of cell adhesion by C. ulcerans[24].

The C. ulcerans 809 strain was isolated from a patient with a rapid fatal pulmonary infection. The 809 strain-unique virulence factor (shiga toxin-like ribosome-binding protein, Rbp) is located adjacent to the truncated integrase (CULC809_00176) and corresponds to the integrase of ΦCULC0102-I. It appears that virulence factors have been acquired as a cassette gene in the ΦCULC0102-I-like prophage. It is intriguing to note that the 0102 strain does not carry the 809 strain-unique virulence factors (Rbp and the additional venom serine protease, Vsp2), but instead carries the tox gene on ΦCULC0102-I, which resulted in a diphtheria-like illness in a 52-year-old woman.

Isolates of C. ulcerans are generally obtained from a diverse range of animals, including humans. Isolation of a human pathogen C. diphtheriae from animals has been reported previously, although it is rare [31]. The tox gene might be frequently transmitted through common prophages with the aid of the highly homologous regions among Corynebacterium spp., including C. diphtheriae and C. ulcerans isolated from animal sources.

Conclusions

Toxigenic C. ulcerans is an emerging pathogen that can be transmitted from animals to humans [5]. In the host organism, as well as in C. diphtheriae, the tox gene [18] is encoded by prophages. Through genome sequencing, we have identified a novel structure in a tox-positive C. ulcerans prophage with no significant sequence homology to those in C. diphtheriae. This suggests distinct origins of the prophages and thus may also explain the difference in the primary structures of their tox genes. The tox-positive bacteriophages may increase the dissemination risk of toxigenic C. ulcerans isolates, therefore, C. ulcerans isolates from both human and animal sources should be investigated further to determine the level of variation.

Methods

This research was not carried out on humans. No experimental research on animals was carried out.

Bacterial strain

The toxigenic C. ulcerans isolate 0102 was obtained in 2001 as a human clinical isolate [22, 23].

Preparation of genomic DNA

Genomic DNA was isolated by conventional methods, using phenol extraction and ethanol precipitation from heat-killed bacterial cells propagated in brain-heart infusion liquid medium.

Short-read DNA sequencing using an Illumina Genome Analyzer IIx

DNA libraries of the ~600 bp insert length of C. ulcerans 0102 were prepared using a genomic DNA Sample Prep Kit (Illumina, San Diego, CA, USA). DNA clusters were generated on a slide using a Cluster Generation Kit (ver. 4) on an Illumina Cluster Station (Illumina), according to the manufacturer’s instructions. Sequencing runs for 80-mer short reads were performed using an Illumina Genome Analyzer IIx (GA IIx) and TruSeq SBS kit v5. Fluorescent images were analyzed using the Illumina base-calling pipeline RTA2.6/SCS2.8 to obtain FASTQ-formatted sequence data.

De novo assembly of short DNA reads and gap-closing

The 80-mer reads were assembled (parameters k64, n51, c32.1373) using ABySS-pe v1.2.0 [32]. Predicted gaps were amplified with a specific PCR primer pair, followed by Sanger DNA sequencing using a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA).

Validation of the complete genome sequence using short-read mapping and pulsed-field gel electrophoresis (PFGE)

To validate the genome sequence, 40–mer short reads were re-aligned with the sequence using Maq software (ver. 0.7.1) and the easyrun Perl-command [33]. Read alignment was inspected using the MapView graphical alignment viewer [34]. PFGE analysis was performed to validate the predicted restriction fragment profiles from the complete genome sequence, according to De Zoysa et al. [35]. Bacterial cells were lysed with lysozyme and protease [36], embedded in plugs, digested with the restriction endonuclease Sfi I (New England Biolabs, Ipswitch, MA, USA) and electrophoresed in a CHEF DRII apparatus (Bio-Rad, Hercules, CA, USA) at 11°C with a pulse time of 5–20 s for the first 20 h and 1–5 s for the following 18 h.

Annotation and pair-wise alignment analysis

Gene prediction from the complete sequence was performed using the NCBI Prokaryotic Genomes Automatic Annotation Pipeline (PGAAP; http://www.ncbi.nlm.nih.gov/genomes/static/pipeline.html). Several of the suggested errors were revised manually. Pseudogenes that were identified by PGAAP were checked using the read-mapping correction described above. Genomic information, such as nucleic acid variations and circular representation, was analyzed using IMC-GE software (Insilicobiology, Yokohama, Japan). A BLASTN homology search [37] was performed for the whole chromosome sequences of C. pseudotuberculosis FRC41 (accession no. NC_014329), C. ulcerans 0102, and C. diphtheriae NCTC 13129 (accession no. NC_002935). Aligned images of the homologous regions were visualized with the ACT program [38].

Phylogenetic analysis

Phylogenetic analyses of all nucleotide sequences were conducted using the neighbor-joining method with 1,000-times bootstrapping in ClustalW2 [39]. FigTree ver. 1.3.1 (http://tree.bio.ed.ac.uk/software/figtree/) software was used to display the generated tree.

Nucleotide sequence accession numbers

The complete chromosome sequence for the C. ulcerans 0102 strain has been deposited in the DNA Data Bank of Japan (DDBJ; accession no. AP012284).