Systematic BLAST analysis was carried out for each of the ORFs of V. parahaemolyticus RIMD2210633 compared with each of the ORFs from the genome sequences of V. cholerae N16961, V. vulnificus YJ016 and CMCP6, and V. fischeri ES114. This four-way BLAST analysis was used to construct genome BLAST atlases of chromosome 1 and 2 of the four Vibrio species with V. parahaemolyticus RIMD2210633 as a reference (Fig. 1). The four outer circles of solid color represent conserved proteins of the BLASTed genomes for both chromosome 1 and 2. The outer most circle represents the V. fischeri ES114 genome (purple circle), the next two circles represents V. vulnificus YJ016 and CMCP6 (navy and green circles), followed by the fourth circle (brown circle), which represents V. cholerae N16961. The innermost circles show DNA structure features, repeat sequences and base composition properties of the reference V. parahaemolyticus RIMD2210633 (Fig. 1). It is of interest to note that chromosome 1 shows a higher level of overall conservation among the species examined than chromosome 2 indicating that a lot of species specific genes lie on chromosome 2. There are approximately 44 gap regions (greater than 1 kb) on chromosome 1 and 29 gap regions (greater than 1 kb) on chromosome 2, common to all four outer circles and these gaps represent regions of the V. parahaemolyticus RIMD2210633 chromosomes that are unique being absent from all other isolates examined. Differences in these regions in their DNA structural features such as intrinsic curvature, stacking energy and position preference correlate with some of the gap regions and represent phages, integrons and genomic islands, that is signatures of foreign DNA acquired by horizontal transfer (Fig. 1).

Of the 73 gap regions, our analysis uncovered 24 regions greater than 10 kb that are present in RIMD2210633 and absent from the other species examined, that is the gap regions in the four outer circles in Figure 1 (Table 1). Of the 24 regions identified, 11 regions encoded an integrase or transposase, 9 regions had aberrant GC content (45 ± 3%), 7 regions of which had lower G+C content compared to the rest of the genome suggesting that these regions were acquired by horizontal gene transfer (Table 1). The 24 regions included 14 previously identified: lipopolysaccharide and capsule polysaccharide gene clusters, a class 1 integron, 2 f237 phage regions, 2 osmotic stress response gene clusters, 2 T3SSs, and the Vibrio parahaemolyticus island (VPaI) regions (Table 1) 202647. The 10 additional regions unique to V. parahaemolyticus RIMD2210633 included 2 regions encoded on chromosome 1 and 8 regions encoded on chromosome 2 (Table 1). On chromosome 1, region VP0081 to VP0092 encodes mainly hypothetical proteins of unknown function; VP0081 encodes a homologue of a hyper osmotic shock protection protein. Region VP1386 to VP1420 encodes hemo utilization/adhesion proteins, OmpA, a ClpA/B type protease, a BfdA homologue, a putative IcmF-related protein and related type VI secretion system (T6SS) proteins (VP1401 to VP1409), which is predicted to be involved in intracellular trafficking, secretion, and vesicular transport in other Gram-negative pathogens. On chromosome 2, region VPA0434 to VPA0458 encodes a large number of homologues of genes involved in degradation processes. Region VPA887 to VPA0914 encode proteins that show homology to phage f237 on chromosome 1. Within region VPA0950 to VPA0962 are homologues of biofilm associated proteins among others. Region VPA0989 to VPA0999 contains homologues of a number of peptidase, lipase and amylase genes, and region VPA1440 to VPA1444 encode a type I secretion system (Table 1). Region VPA1503 to VPA1513 contained a type 1 pilin homologue similar to Pap pilin identified in Burkholderia pseudomallie, Pseudomonas spp and Yersinia spp. Region VPA1559 to VPA1583 encodes a number of proteins with a possible role in antibiotic resistance and region VPA1652 to VPA1679 contains a ferric uptake system.

We also constructed genome BLAST atlases of all 28 genomes available for members of the family Vibrionaceae in the database, this included eight additional species of the Vibrionaceae family. V. parahaemolyticus RIMD2210633 as reference strain (inner most circle) versus V. parahaemolyticus AQ3810, V. cholerae 1587, AM-19226, MAK757, MO10, MZO-2, MZO-3, B33, NCTC8457, RC385, O395, V51, V52, 623–39, and 2740–80, V. harveyi ATCC BAA116, V. alginolyticus 12G01, Vibrio sp. Ex25, V. vulnificus CMCP6 and YJ016, V. splendidus 12B01, Vibrio sp. MED222, V. fischeri ES114, V. salmonicida LF1238, V. angustum S14, P. profundum SS9 and 3TCK for chromosome 1 (a) and chromosome 2 (b), in which each gap region can be zoomed in on to examine in detail (see Additional file 1 &2) 48.

Most of the 24 gap regions remained unique to V. parahaemolyticus, exceptions were noted (see Additional file 1 and 2). For example, on chromosome 1 region VP0081 to VP0092 is present in V. alginolyticus 12G01 and Vibrio sp. Ex25, the osmotolerance gene clusters are partially present in V. alginolyticus 12G01, V. harveyi ATCC BAA116 and Vibrio sp. Ex25, and the T3SS-1 and T6SS (VP1388 to VP1414) are present in V. alginolyticus, V. harveyi and Vibrio sp. Ex25 (see Additional file 1 and 2). On chromosome 2, region VPA0950 to VPA0962 was present in Vibrio sp. Ex25, homology to VPaI-7 within the T3SS-2 region (VPA1332 to VPA1355 and VPA1358 to VPA1370) is present in V. cholerae strains 1587, AM-19226, V51 and 623–39. Region VPA1440 to VPA1442 is present in Vibrio sp. Ex25. Region VPA1503 to VPA1521, which encodes a type I pilin is present in V. alginolyticus, V. harveyi and Vibrio sp. Ex25 (see Additional file 1 and 2).

The previously published V. parahaemolyticus genome sequence is from RIMD2210633 a tdh positive, trh and urease negative O3:K6 clinical isolate of the highly virulent clone recovered in Japan in 1996. In order to unravel events at the genome level that may have lead to the emergence and dissemination of the new highly virulent strain, we sequenced the genome of AQ3810, a tdh positive, trh and urease negative O3:K6 isolate recovered in Japan in 1983 for comparison. The complete genome sequence of AQ3810 is 5.8 Mb and 5509 proteins have been annotated within its genome compared with the 5.2 Mb genome of RIMD2210633, which has 4832 annotated proteins. There is extensive sequence homology between the two sequences, however genomic differences were noted. As had been found within other Vibrio species, the gene capture system, the -integron encoded in RIMD2210633 and AQ3810 do not share any significant sequence similarity.

We examined the genome of AQ3810 for the presence of the 24 regions identified as unique to RIMD2210633 from our species comparisons (Table 1). Of the 24 regions, 8 regions were absent from AQ3810, 5 genomic islands (VPa-1, VPaI-2, VPaI-4, VPaI-5 and VPaI-6), ORFs VP1386 to VP1420, which encodes T6SS, the class 1 integron, and phage f237 encoded on chromosome 1 (Table 1). These data confirm our previous result that VPaI-1, VPaI-4, VPaI-5 and VPaI-6 are unique to the new highly virulent strain recovered after 1995 26. For example, two of the missing regions, VPaI-1 and VPaI-4, integrate at a tRNA-met (VP0404.1) and tRNA-ser locus (VP2130.1), respectively in RIMD2210633, however, in AQ3810, both of these tRNA sites are empty (see Additional file 3). The other two missing regions VPaI-5 and VPaI-6 regions are located between core chromosomal ORFs VP2889 and VP2911, and VPA1252 and VPA1271, respectively in RIMD2210633, while in AQ3810, the homologues of these genes are contiguous indicating that these sites are empty (see Additional file 3).

Two VPaIs were rearranged. One, the VPaI-2 region (VP0634 to VP0643), is present at the tmRNA gene (ssrA) in RIMD2210633, a gene that encodes both tRNA and mRNA properties. In AQ3810, at this same locus, the first three genes of this region are present (VP0634 to VP0636), which encode homologues of a nitrilase/cyanide hydratase, OmpA and LysR, but the remaining genes are replaced by a novel region encoding an integrase (see Additional file 4). A second island region, VPaI-3 (VP1071 to VP1095) present at a second tRNA-ser locus (VP1070.1) in both RIMD2210633 and AQ3810 has 21 genes in common (ORFs VP1074 to VP1095) and 6 genes, two novel integrases and four hypothetical proteins, are adjacent to the tRNA-ser locus in AQ3810 (see Additional file 4).

Two additional regions named VPaI-8 and VPaI-9 were identified in AQ3810 (see Additional file 4). VPaI-8 is a 17 kb region located between homologues of VP3057 and VP3058 and contains ORFs A79_5175 to A79_5191, which encode a number of hypothetical proteins, homologues of SMF and KAP proteins, and two integrases separated by a single ORF (see Additional file 4). VPaI-9 is a 22 kb region integrated between homologues of VP0006 and VP0007. VPaI-9 encodes an integrase, an excisionase, a helicase and a type I restriction modification system.

ORFs VP1386 to VP1420 are absent from AQ3810. This regions encodes T6SS (ORFs VP1401 to VP1409) and a range of proteins that could be translocated by this system; hemo utilization/adhesion proteins, OmpA, a ClpA/B type protease, a BfdA homologue, a putative IcmF-related protein. This suggests the presence of an additional virulence mechanism in the highly virulent O3:K6 clone. T6SSs have been identified in a range of Gram-negative pathogens including pathogenic V. cholerae and in that species T6SS translocates a bacterial host protein into host cells that cross link actin 49.

Previously, we examined the distribution of VPaI-1, VPaI-4, VPaI-5 and VPaI-6 among a worldwide collection of V. parahaemolyticus isolates and found that these regions are unique to 24 isolates of the highly virulent O3:K6 clone 26. We determined the distribution of VPaI-2, VPaI-3 and VPaI-7 using primer pairs described in Table 2. Of the 42 V. parahaemolyticus isolates examined by PCR assays using two primer pairs encompassingVPaI-2, 27 isolates gave positive PCR bands. These isolates were recovered post-1995 and include the 24 isolates that were previously shown to contain VPaI-1, VPaI-4 to -6 (Fig. 2). VPaI-2 was also present in isolates UCMV586 and 1324, O8:K22 and O4:K6 isolates recovered after 1995, and ATCC43996, an O3:K4 clinical isolate from recovered in the UK in 1970 (Fig. 2). The presence of VPaI-2 in ATCC43996 indicates that this region was present in isolates before 1995, prior to its acquisition by the new highly virulent strain. VPaI-2 encodes an integrase, a resolvase, hypothetical proteins, a ribonuclease HI, an aminohydrolase, transcriptional regulators and a lipase.

To understand the evolutionary significance of the distribution of the VPaI regions among our collection of isolates, a phylogenetic frame work was constructed by multilocus sequence (MLS) analysis of an initial analysis of three housekeeping genes. MLS analysis was demonstrated in numerous studies to be a powerful method to both discriminate and determine the phylogenetic relationships among bacterial isolates including Vibrio species 535455. We found similar to others that V. parahaemolyticus isolates are highly related sharing substantial sequence similarity at the three loci we examined 3742 (Fig. 2). Within the 1854 bps examined among the 42 isolates, there were a total of 39 polymorphic sites of which 21 sites were phylogenetically informative and 12 sequence types (ST) were found. For the purposes of this study, a phylogenetic tree was constructed by the neighbor-joining method using Kimura 2-parameter, which clustered the strains into two closely related but distinct groups named A and B (Fig. 2). The first group contains all highly virulent isolates whereas group B is comprised of mainly environmental isolates recovered in Spain in the early 2000s. Within group A are 25 isolates that are identical at all three loci examined; these isolates include 22 O3:K6 isolates with worldwide distribution recovered pre-1995 and post-1995, and 3 O1 serogroup isolates (O1:K25 and O1:KUT) recovered post-1995. O3:K6 and O1:K25 isolates recovered post-1995, and O3:K6 isolates recovered pre-1995 shared identical sequence profiles (Fig. 2). These data support the hypothesis that O1:K25 and O1:KUT serogroups arose from the O3:K6 highly virulent strain by acquisition of novel O and K antigens similar to the emergence of the pathogenic V. cholerae O139 serogroup strain from an O1 El Tor isolate. Acquisition of novel O and K antigens would be evolutionary advantageous since it may play a role in host immune avoidance in V. parahaemolyticus infection. Clustering with these O3:K6, O1:K25, O1:KUT isolates are four identical O4 serogroup isolates and KE10464, a divergent O3:K6 pre-1995 (Fig. 2). Thus, it appears that acquisition of novel O antigens is frequent in this species and more recent data suggests this is an ongoing event 3. Also found within group A are several divergent O3:K6 pre-1995 and post-1995 isolates, and a single divergent O4:K8 post-1995, 1324 (Fig. 2). Group B consists of 5 isolates, with various serotype designations but all were recovered in Spain post-1995 (Fig. 2). Three of the strains were recovered from mollusks and sea sediment, and two strains 30824 and 428/00, which shared an identical ST, were from clinical sources. Overall, the phylogenetic tree constructed from concatenated sequences of three housekeeping genes indicates that pathogenic V. parahaemolyticus isolates are highly homologous as others have previously shown 42.

We mapped the distribution of each of the VPaI genomic islands onto the phylogenetic tree to elucidate the possible steps involved in the emergence of the globally distributed V. parahaemolyticus new highly virulent strain. We found that similar to VPaI-1, VPaI-4, VPaI-5 and VPaI-6, VPaI-2 and VPaI-3 are predominately present among the highly virulent isolates recovered after 1995 with only one exception noted for VPaI-2, strain ATCC43996 recovered in the UK in 1970, an O3:K4 serogroup (Fig. 2). The VPaI-3 region was present in two pre-1995 O3:K6 isolates, KE9967 and U5474 that have an identical sequence type, and in KE10462, which shows an identical sequence type to five additional pre-1995 and 16 post-1995 O3:K6 isolates. KE10462 also appears to have contained the VPaI-7 regions since it is partially present in this isolate (Fig. 2). KE10462 has been shown to be positive for group specific PCR (GS-PCR), which is based on the toxRS nucleotide sequence, that has previously been shown to differentiate post-1995 pandemic strains from non-pandemic and pre-1995 isolates 2639.

In conclusion, the most parsimonious scenario for the evolution of the new highly virulent O3:K6 clone suggests that a pre-1995 O3:K6 strain obtained regions VPaI-1 to VPI-7, and a T6SS encoded within ORFs VP1386–VP1420, this secretion systems along with T3SS-2 may explain the highly virulent nature of the O3:K6 virulent clone. It appears that V. parahaemolyticus isolates have the ability to acquire large regions of DNA and that this is an ongoing process among pathogenic isolates. For example, the O1 and K antigens, which are encoded in the same genomic region, are undergoing frequent change among closely related strains and this may be a mechanism to avoid the host immune system.

The possible origins of the V. parahaemolyticus variable regions appear to be quite diverse. Blast analysis of the VPaI-1 encoded proteins found 7 ORFs highly homologous to a 22 Kb island present in V. cholerae strain 623–39 at the same tRNA-met insertion site, whereas a similar analysis of VPaI-3 showed high sequence similarity to a region in V. harveyi HY01 (AIQ_705 to AIQ_762). VPaI-2 encoded several ORFs with high similarity to ORFs identified in Vibrio sp Ex25. Most of VPaI-5 showed homology to ORFs from Shewanella woodyi and Shewanella sp, and similarly several ORFs of VPaI-6 were homologous to a region in Shewanella sp ANA-3. The T3SS-2 region encoded on island VPaI-2 is most closely related to a T3SS recently identified in V. cholerae V51, a non-O1 serogroup isolate 56. Region VP1386 to VP1420, which encodes a T6SS as well as a Rhs element, showed extensive homology to a region in V. harveyi ATCC BAA-1116.

A total of 42 V. parahaemolyticus isolates were examined in this study as previously described 26. The 42 isolates were temporally (1970 to 2003) and geographically widespread (Asia, Europe and South America) and encompassed 10 different serotypes. All V. parahaemolyticus isolates were grown in Luria-Bertani broth (LB) supplemented with 3% NaCl and stored at -70°C in LB broth with 20% (v/v) glycerol.

We performed four-way BLAST analysis of V. parahaemolyticus RIMD2210633, an O3:K6 isolated in 1996, versus V. vulnificus YJ016, V. vulnificus CMCP6, V. cholerae N16961 and V. fischeri ES114 to identify regions that were unique to V. parahaemolyticus. Complete nucleotide sequences and annotations for the V. parahaemolyticus RIMD2210633, V. vulnificus YJ016, V. vulnificus CMCP6, V. cholerae N16961 and V. fischeri ES114 were retrieved and downloaded from NCBI 20575859. These were used to construct a genome atlas of the complete genome sequence of all five isolates. The genome atlas plot maps DNA structure features, repeats, and base composition properties of V. parahaemolyticus as well as each gene present in RIMD2210633, and their homologues in all four additional species oriented at the ori60. In addition, we constructed a zoomable genome atlas of the complete genome sequences of all 27 members of the family Vibrionaceae available in the database. This data can be interactively examined for chromosome 1 and for chromosome 2 on the web 48. We compared the genome of V. parahaemolyticus RIMD2210633 to the genome of V. parahaemolyticus AQ3810, an O3:K6 isolated in 1983, using the Artemis comparison tool (ACT) program 61.

Chromosomal DNA was extracted from each V. parahaemolyticus isolate using the G-nome DNA isolation kit from Bio 101. To determine the distribution of regions unique to V. parahaemolyticus among our collection of 41 isolates, PCR assays were performed. Primer pairs were designed to target within the regions of interest as well as flanking the regions (Table 2). PCR was performed in a 25 μl reaction mixture with the following cycles: an initial denaturation step at 96°C for 3 min followed by 30 cycles of denaturation at 94°C for 30s, 30s of primer pair annealing at the respective temperature, an extension step at 72°C for 1–4 min (depending on expected PCR product size). PCR primers to amplify three chromosomal housekeeping genes, gyrase subunit B (gyrB, VP0014), malate dehydrogenase (mdh, VP0325), and chaperonin (groEL-1, VP2851), were designed based on the sequence of V. parahaemolyticus RIMD2210633 (Table 1). The housekeeping genes were PCR amplified from chromosomal DNA isolated from all V. parahaemolyticus isolates and PCR products were purified using Jetquick PCR purification Kit (GENOMED). The mdh, gyrB and groEL-1 sequences were determined by MWG-Biotech based on the dye deoxy terminator method and the reaction products were separated and detected on an ABI PRISM 3100 genetic analyzer.

The multiple sequence alignment program ClustalW was used to align nucleotide sequences for each housekeeping gene 62. Rates of synonymous substitutions/synonymous site (K_S) were calculated by the methods of Nei and Gojobori and Nei and Lin 6364. To analyze the evolutionary relationships among V. parahaemolyticus isolates, the concatenated sequence of all three housekeeping genes was used to construct a Neighbour-Joining phylogenetic tree based on Kimura 2-parameter distance using MEGA-3 65.

The sequences of mdh, gyrB and groEL-1 were submitted to GenBank and given the accession numbers GenBank EU629305–EU629345.