Introduction

Species of the gammaproteobacterium Vibrio are well-known for their roles as pathogens in both terrestrial and aquatic environments1. However, not all species and strains of Vibrio are pathogenic2,3. For instance, Vibrio cholerae, a bacterium that causes the acute diarrhoeal disease cholera4,5,6 requires the acquisition of key virulence factors to become toxigenic. The two major virulence factors of V. cholerae, TCP (toxin co-regulated pilus) and an exotoxin called cholera toxin (CTX), are both introduced via horizontal gene transfer (HGT) and are encoded by a pathogenicity island7 and the genome of a single-stranded DNA lysogenic filamentous bacteriophage (or prophage) called CTXphi (CTXφ)6, respectively. The TCP receptor is a prophage receptor facilitating introduction of CTXφ prophage DNA into the cell6. In V. cholerae, CTXφ prophage transmits toxin genes among the bacteria that express the TCP prophage receptor. This process of inducing virulence in an otherwise harmless bacterium through prophage infection and integration of the prophage genome into the bacterial host genome is called lysogenic conversion8.

The health of coral reefs is declining on a global scale9,10. One of the causes of this deterioration is coral disease, the incidence and severity of which has increased worldwide over recent decades11,12,13. A number of coral disease studies have reported Vibrio species, V. coralliilyticus in particular, as the pathogen responsible for some instances of white syndrome (WS) disease14,15, as well as coral bleaching16,17. Coral bleaching is the loss of the algal endosymbiont, Symbiodinium, and/or its photosynthetic pigments from the coral tissue resulting in paling of the coral tissues. If these coral algal endosymbionts are not regained following bleaching, coral mortality is inevitable18. Bleaching tends to occur more gradually in comparison to the acute disease of WS that results in rapid and irreversible loss of tissue biomass (between 1 to 100 cm2 day−1)19. Tissue loss can result in white spots, bands or patches of exposed skeleton with a distinct lesion boundary between exposed skeleton and visibly healthy tissue17,19.

Pathogenicity of the coral-associated V. coralliilyticus in coral disease and bleaching has been attributed to increases in sea surface water temperatures14,20 and the immunocompetence of the coral host21. However, the underlying mechanisms are not understood and a possible role for prophages in driving virulence of coral pathogens, as is the case for V. cholerae, has not yet been considered. Inconsistent findings on the role of Vibrio in coral disease/bleaching events have been reported in the literature16,22,23. In the case of the coral pathogen V. coralliilyticus, identified as the causative pathogen of WS in the coral Pocillopora damicornis14,15, it was initially observed that V. coralliilyticus encodes for a zinc-metalloprotease that causes cellular damage in coral tissue17,24. However, a recent study did not detect differences in pathogenicity of a V. coralliilyticus strain (P1) between a wild-type and zinc-metalloprotease-deficient mutant in infection experiments25, indicating that factors other than zinc-metalloproteases (e.g. hemolysin activity; other proteases) produced by V. coralliilyticus strains may also be at play and contribute to virulence. Moreover, the possible role of a prophage as an alternative driver of virulence in V. coralliilyticus in these studies has not been addressed.

The aim of this study was to assess whether a prophage could potentially drive virulence of V. coralliilyticus by examining in silico five publically available V. coralliilyticus whole genome sequences20,25,26,27,28. Our focus was on the presence of prophage signatures and virulence factors in these genomes, to identify lysogenic conversion by a prophage. Our analysis demonstrates the presence of lysogenic prophages in V. coralliilyticus genomes that share similarities to prophages integrated in pathogenic V. cholerae strain genomes and supports our hypothesis.

Results

Recent studies have reported five whole genome sequences of V. coralliilyticus associated with corals found in different locations worldwide (Table 1). The first strain of V. coralliilyticus to be sequenced was isolated from diseased and bleached corals off the coast of Zanzibar (V. coralliilyticus ATCC-BAA-45014,20 referred to here as BAA450). Subsequently, whole genome sequences were published for a strain isolated from a Great Barrier Reef coral (V. coralliilyticus P115,25) and two strains isolated from corals located in Hawai’i (V. coralliilyticus OCN00828 and V. coralliilyticus OCN01427 ). Most recently, a fifth complete genome was sequenced from V. coralliilyticus isolated from oysters at a shellfish hatchery (V. coralliilyticus RE9826). We interrogated each genome for signatures of or complete prophage genomes.

Table 1 Description of the five whole genome sequences of Vibrio coralliillyticus (four isolated from corals and one from oysters) analysed in this study and implicated in diseases of corals and oysters.
Full size table

Prophages in V. coralliilyticus genomes

Our in silico analysis revealed the BAA450 genome contains an intact prophage genome that shares homology to the core region of the CTXφ genome (Figs 1 and 2) and encodes a number of genes that share close identity to those encoded by a filamentous prophage, VCYφ, isolated from an environmental V. cholerae population29 (Tables 2 and 3; Figs 2 and 3). The VCYφ prophage genome shares sequence similarity to the CTXφ prophage29 and like CTXφ, is a novel virus that has evolved through HGT events and recombination. A prophage genome encoding a number of genes homologous to those encoded by VCYφ was also detected in the V. coralliilyticus P1 strain but is absent from the OCN008 and OCN014 strains from Hawai’i and the RE98 genome (Tables 2 and 3; Figs 1 and 2).

Table 2 Details of prophages identified in five publically available Vibrio corallilyticus (VC) genomes.
Full size table
Table 3 Similarity values for ORFs in V. corallilyticus (VC) strains BAA450 and P1 genomes with closest matches to VCYφ phage genes.
Full size table
Figure 1
figure 1

(a) Pangenome of the five Vibrio coralliilyticus strains. OrthoMCL v2.0.9 was used to generate clusters of similar proteins, which are indicated by overlapping regions. The tick marks indicate size of the pangenome in millions of base pairs (Mbps) and key prophage signatures are highlighted and magnified 100x. (b) Clustering of Vibrio coralliilyticus proteins and Vibrio cholerae proteins with known Vibrio phage proteins. OrthoMCL v2.0.9 was used to detect similarity between proteins in the V. coralliilyticus genomes and phage genomes, which are indicated by the overlapping tracks. To identify related protein clusters a minimum similarity of 50% and an e-value of <10−5 was used. Overlapping lines denote the presence of a similar protein in each of the genomes. The genomes each have a unique colour and this is not related to similarity. The outer-most numbers in plot A only give a guide to the total size of the pan-genome; these do not represent locations within genomes. The orthoMCL program looks only for similar proteins with no regard to their location or synteny within different genomes and therefore, this plot should only be used to determine the presence or absence of particular proteins of interest.

Full size image
Figure 2
figure 2

Genome organisation of prophages found in the genomes of five Vibrio coralliilyticus strains isolated from corals and oysters.

BLAST values were determined using amino acid sequences with BLASTP. Linear ORF maps of the prophages were aligned based on their modular structures. ORFs or genes are represented by arrows oriented in the direction of transcription. Coloured arrows are annotated to indicate predicted function. Grey arrows represent unknown ORFs. The att (phage attachment sites), where identified, are indicated.

Full size image
Figure 3
figure 3

Amino-acid alignment of 7 of 9 ORFs encoded by the BAA450 prophage with 7 of 11 ORFs in the Vibrio phage VCY-phi (Accession No. NC_016162.1).

Conserved residues are shaded in red. Dashes indicate gaps.

Full size image

For the morphogenesis of the CTXφ prophage, three genes known as zot (zonal occludens toxin), ace (accessory cholera enterotoxin) and cep (core-encoded cepin) are required30. Both the zot and ace genes are identified as minor coat proteins of the CTXφ prophage but also exhibit toxic activity by encoding for toxins that play a direct role in disrupting the intestines of humans with cholera disease30. A rstB gene (Figs 1a and 2) encodes a DNA-binding protein essential for prophage formation and is identified in the CTXφ genome as playing a role in the site-specific integration of the prophage into the host chromosome31. Homologues of these genes are encoded by the genome of the Vibrio VCYφ –like prophage found in BAA450 and P1 (Table 3; Figs 1, 2, 3; Supplemental Figure 1). Identity scores for specific genes for these prophages in BAA450 and P1 are shown in Table 3. Sequence alignment of key genes in the BAA450 prophage genome with similar genes encoded by VCYφ genome is shown in Fig. 3. Genes in the BAA450 prophage that share close similarity with those in the VCYφ genome include a DNA replication initiation protein (identity = 66%; score = 559), a gene encoding a ssDNA-binding protein, similar to the rstB gene (identity = 53%; score = 103), a minor capsid protein gene similar to ace (identity = 29%; score = 158) and a gene encoding a Zot-like protein (identity = 61%; score = 578) (Table 3).

An intact prophage was detected in the OCN014 genome that shares similarity to a plasmid-like prophage VP88232 found in V. parahaemolyticus, a pathogen commonly associated with shellfish (Table 2; Figs 1b and 2). This prophage genome shares similarity to a small number of genes in the Pseudomonas phage CTXφ-like genome found in P1 (Fig. 2). PHAST33 analysis of the OCN008 genome detected one intact prophage that shares similarity with a V. cholerae prophage K13934 and also shares homology with the incomplete Kappa prophage located in the RE98 genome (Table 2; Figs 1b and 2). The phage K139 has been previously identified as associating with pathogenic V. cholerae strains but it is still unclear whether or how this prophage may increase pathogenic host fitness35. Strain RE98 has been identified as a particularly virulent pathogen responsible for high mortality rates in oyster shellfish hatcheries in the west coast of the US36. A second intact vB VpaM-like prophage was detected in the RE98 genome that shares homology to the Pseudomonas CTXφ-like prophage in P1 (Table 2: Figs 1b and 2). A heat map analysis indicates similarity between the prophage genomes identified in this study (see Supplementary Fig. S1 online).

Pathogenicity islands in the V. coralliilyticus genome

The synergistic effect of a number of genes, including toxicity genes introduced by CTXφ and a series of genes clustered in pathogenicity islands on the bacterial genome result in the evolution of V. cholerae7 to harmful epidemic and pandemic-causing strains. The pathogenicity islands (namely Vibrio Pathogenicity Island I and II, VPI-I/II) consist of virulence genes, a transposase gene, specific (att-like) attachment sites flanking each end of the island and an integrase with homology to a prophage integrase gene7. The TCP gene cluster induces virulent Vibrios to secrete the exotoxin cholera toxin (CTX) and is encoded on a pathogenicity island within the Vibrio chromosome7. These genes are present in BAA450 and P1, suggesting these elements may influence the virulence of BAA450 and P1. Both BAA450 and P1 genomes carry a Coralliilyticus Pathogenicity Island-I (CPI-I) that is located at a chromosomal site similar to the dif-like region that accommodates an insertion site for the prophages CTXφ and f237 of V. cholerae6 and V. parahaemolyticus37, respectively. The TCP is encoded for in CPI-1 of BAA450 and P1. While BAA450 has a similar genome arrangement to V. cholerae with the presence of a second pathogenicity island (CPI-II)20, the P1 genome does not contain a second pathogenicity island CPI-II. It is noteworthy that the CPI-II in BAA450 bears similarity to a second pathogenicity island found in the genome of the seventh pandemic strain of V. cholerae7. Further, of the five genomes analysed here, the BAA450, P1, OCN014 and RE98 genomes encode for a RTX (repeats in toxin) protein that is a harmful pore-forming toxin required for human host colonisation in incidences of infection by V. cholerae30. The RTX gene cluster in V. cholerae is physically located close to the CTXφ genome38. A significant upregulation of the RTX protein and a number of proteins located on the same pathogenicity island as the RTX protein occurs in BAA450 at increased temperatures of 27 °C and above, contributing to increased virulence at higher temperatures20.

Discussion

Prophage infections can transform ordinary bacteria to virulent pathogens, shown in a wide range of species including Vibrio cholerae6, Staphylococcus aureus, Clostridium botulinum, Corynebacterium diphtheriae, Streptococcus pyrogenes and Escherichia coli8,39. A landmark discovery in bacteriophage research revealed the role of lysogenic prophages in the conversion of the bacterium Vibrio cholerae from a non-pathogenic to a virulent strain, resulting in often devastating cholera outbreaks among human populations4,5,6. Our analysis shows that the same scenario may exist for the coral pathogen, Vibrio coralliilyticus, and future studies should rigorously address this previously overlooked possibility.

BAA450 was reported over a decade ago as causing bleaching and tissue necrosis in the coral species Pocillopora damicornis14,17. This strain was seen to induce bleaching at temperatures between 24–26 °C, but not at lower temperatures. An increase in protease activity resulting in the release of high amounts of toxic extracellular proteases was observed at temperatures between 27–29 °C, which resulted in mortality of the coral17. The presence of an intact prophage genome in the BAA450 genome and an incomplete prophage in P1 that share similarities to the genome of CTXφ prophage, responsible for toxigenic conversion of V. cholerae strains, suggests that similar prophage-driven virulence mechanisms could operate in both the human and coral system. Extracellular zinc-metalloproteases expressed by V. coralliilyticus strain P1 were previously reported to be the driver of tissue damage in WS infections17,24 but these enzymes were not the only drivers as demonstrated by subsequent mutant infection experiments25. These studies did not include an assessment of the presence of CTXφ prophage-derived toxins, such as ZOT and ACE proteins. We therefore encourage future experiments towards assessing the expression of these virulence factors in infections by V. coralliilyticus strains.

The OCN008 strain has been indicated as an aetiological agent in white syndrome of the coral species Montipora capitata28. Unlike strains BAA450 and P1, the pathogenesis of OCN008 is reported as being temperature-independent28 with only a relatively slower onset of infection observed at lower temperatures, resulting in tissue loss without bleaching. Previous comparative analysis of the genome sequences of BAA450, P1 and OCN008 reported approximately 84% similarity between the genomes of OCN008 and BAA450 and 89% similarity between the OCN008 and P1 genomes28. Genes upregulated at least 4-fold in BAA450 by a rise in temperature shared 97–100% sequence identity with genes in OCN00828. Why does infection by OCN008 appear to be independent of increasing temperature? The intact prophage genome detected in the OCN008 genome shares similarity to the Vibrio prophage K139 that is also present in the V. cholerae O1 O395 genome (Fig. 2). It is possible that the presence of this intact prophage in the OCN008 genome and the absence of a CTXφ-like prophage genome results in a different disease profile compared to those of the BAA450 and P1 temperature-dependent infections. OCN014 has been isolated from corals exhibiting symptoms of WS27. Further investigations are needed to link the presence of OCN014 to coral disease and we encourage the inclusion of virus characterisation in such experimental work.

V. cholerae is autochthonous to coastal and estuarine waters and rising seawater temperatures have been indicated as a major influence on V. cholerae pandemics40. Does the presence of a prophage similar to the CTXφ prophage in the BAA450 and P1 genomes explain these strain’s increased virulence with rising temperature14,17,41,42? It has been predicted, based on time-series studies, that as global temperatures rise, cholera outbreaks will increase40,43,44. Comparably, researchers have indicated the role of seasonal dynamics, microbial interactions and environmental factors driving the occurrence of disease caused by V. coralliilyticus and these are predicted to increase with ongoing environmental change16,41,45. For instance, a recent study reported V. coralliilyticus populations in P. damicornis increased by as much as four times with elevated seawater temperatures46. We postulate that as temperatures increase the BAA450 and P1 prophages will undergo increased active replication along with host genome replication and hence, their associated virulence factors (e.g. ZOT and ACE proteins, as well as TCP and RTX) will increase. With a changing climate, sea surface temperatures have steadily risen accompanied by expanding coral pathogen ranges47.

There have been greater than 30 coral diseases described41,48 but only a few diseases have been attributed to causative bacterial pathogens, namely Thalassotalea loyana (formerly Thalassomonas loyana) in white plague disease49, Vibrio shiloi in bleaching of the coral Oculina patagonica16,50 and V. coralliilyticus in coral bleaching and white syndrome disease14,15,17. Coral disease profiling is complex and suspected pathogens of disease have been identified in apparently healthy corals41,51. The aetiology and pathogenicity of V. coralliilyticus has generated much interest, but very little is understood about the dynamics of infection. The parallels between the Vibrio cholerae paradigm and an equivalent scenario in corals is highly relevant and comparable, as disease in both human and coral populations are influenced by environmental factors that are being altered through anthropogenic influences and changes in climatic conditions.

We encourage further experimental studies into the role of viruses in Vibrio infections in corals. Previous studies have described the presence of viruses associated with coral diseases52,53,54 but none have investigated the role of prophages and lysogenic conversion. As all V. coralliilyticus genomes sequenced to date have been isolated from corals with disease/bleaching symptoms, at present we cannot make comparisons and form conclusions on the presence of these prophages in genomes of V. coralliilyticus isolated from disease-free samples. We anticipate future sequencing efforts to include genomes of V. coralliilyticus from healthy corals, or other such sources, to enable an in-depth comparison. In this present study, we indicate possible signatures to target in V. coralliilyticus to screen for prophages. We envisage our in silico analysis to be a foundation and starting point for future experimental research on the role of lysogenic conversion in pathogenicity of Vibrio and the environmental conditions that potentially trigger virulence. Coral disease is not caused by a single factor but is the result of a complex interplay among biotic and abiotic factors and prophages are a potential central driver of virulence.

Methods

For interrogative and comparative analysis, whole genome sequences of five published V. coralliilyticus genomes were downloaded from the NCBI database, March 2015. The V. coralliilyticus strain names and their accession numbers are as follows and summarised in Table 1: V. coralliilyticus ATCC-BAA-45014,20 accession number ACZN00000000.1; V. coralliilyticus P115,25 accession number AEQS00000000.1; V. coralliilyticus OCN00828 accession number AVOO00000000.1; V. coralliilyticus OCN01427 accession numbers CP009264 and CP009265; and V. coralliilyticus RE9826 accession numbers CP009617 and CP009618 both OCN014 and RE98 genomes have been deposited under BioProject PRJNA224116. The five V. coralliilyticus genome sequences were curated and analysed using Ortho MCL v2.0.955, an analysis tool that identifies orthologous groups for use in comparative genomics and evolutionary processes. Three V. cholerae representative reference genomes were downloaded from NCBI: V. cholerae O1 2010EL-178656 GenBank accession number CP003069 and CP003070; V. cholerae O1 O39557 GenBank accession number CP000626 and CP000627; V. cholerae El Tor N1696158 GenBank accession number AE003852 and AE003853 and were also included in an OrthoMCL analysis. Comparative genomic analysis was visualised using the software tool Circos59 version 0.66. To detect integrated prophage genomes a PHAST (Prophage Search Tool, updated 01/05/2015)33 analysis was conducted on the five V. coralliilyticus genomes.

Additional Information

How to cite this article: Weynberg, K. D. et al. From cholera to corals: Viruses as drivers of virulence in a major coral bacterial pathogen. Sci. Rep. 5, 17889; doi: 10.1038/srep17889 (2015).