Here, we examined the ability of Vibrio isolates to cause mortality of C. gigas larvae under controlled experiments. Many of these isolates were collected from hatcheries where C. gigas larvae are produced. Comparative analysis of the genomes of these 51 isolates identified genes specifically associated with a pathogenic lifestyle from the large pan-genome. We found a strong phylogenetic signal among the strains that caused high mortalities; 15 out of 17 strains tested resulting in 50% or higher mortality rates were putatively identified as V. coralliilyticus strains, and one, V. sp. RE88, was found to be the closest outgroup to this species. Strain V. mediterranei 71105, was the lone pathogen not found within this clade. V. coralliilyticus has been previously implicated in disease outbreaks at a West Coast hatcheries [11]. Importantly, we did not have a V. coralliilyticus strain that did not cause mortalities in tests with C. gigas larvae, so although we attempted to control for phylogenetic signal with the phylolm package, we cannot definitively assert that the genes detected with this model are directly related to pathogenesis rather than just being a trait of the species serving some other ecological function. Still, in this discussion we highlight gene features, many of which are conserved in the V. coralliilyticus strains we tested, significantly correlated with mortalities, signifying that these genes may set apart pathogens from other vibrios ubiquitous in hatchery and marine environments.
A number of gene clusters negatively correlated with oyster survival were homologous to genes of the CPI-1 pathogenicity island of V. coralliilyticus ATCC BAA-450 [30]. These genes were encoded by the genomes of all V. coralliilyticus tested isolates, as well as strain Vibrio sp. RE88 and V. pectenicida 99-46-Y. Pathogenicity islands are a common way for bacteria to distribute large sets of virulence genes via horizontal gene transfer [31]. The most conspicuous virulence factor encoded by the CPI-1 is a putative T3SS. T3SSs are often used by pathogenic bacteria to invade or manipulate host cells. For example, the T3SS-2 of Vibrio parahaemolyticus facilitates host cell invasion [32]. The observation that the conserved T3SS gene, sctV, is more similar to non-Vibrio T3SS genes is consistent with a previous phylogenetic analysis [33] and suggests that the T3SS, and perhaps the surrounding the CPI-1 island, was acquired through horizontal gene transfer.
It is important to note that, although V. pectenicida 99-46-Y appears to encode a relatively complete CPI-1, it was not virulent in our testing. This allows for a comparison of the gene content of the CPI-1 from pathogens with that of the CPI-1 of V. pectenicida 99-46-Y in order to identify those genes that may be specific to oyster larvae pathogenesis. Genes located in CPI-1 of C. gigas larvae pathogens, but absent in the non-pathogenic V. pectenicida 99-46-Y, included those annotated as the vchA hemolysin (AXN30429.1) and another secreted lysin, vchB (AXN30428.1). These genes are highly similar to two homologs of V. vulnificus (vvhA [100% coverage and 76.86% identity], vvhB [100% coverage and 59.52% identity]), which are known to cause epithelial damage and contribute to intestinal growth of the bacterium [34]. Additionally, VchA of V. coralliilyticus, was shown to lyse eukaryotic cells, including erythrocytes [35]. Interestingly, though, culture supernatants from strains of recombinant V. cholerae expressing the vchA gene did not cause significant mortality to C. gigas larvae [18], indicating that although there appears to be a role for this protein in lysing eukaryotic cells, its role as an independent virulence factor in oysters is unclear.
As filter feeding organisms, oysters inherently accumulate large amounts of bacteria from the surrounding water, leading to many coincidental interactions with pathogens. Like other invertebrates, oysters rely on innate immunity and phagocytic cells (e.g., hemocytes in oysters) to protect against pathogenic microbes. Hemocytes are circulatory cells that develop by the early D-veliger larvae stage (17 hours post-fertilization) [36] and eliminate bacterial pathogens through specific binding mechanisms, phagocytosis, and subsequent digestion mechanisms [37–40]. Exposure of C. gigas larvae to pathogenic V. coralliilyticus is known to cause a decline in feeding rate, an activation of the immune response (e.g., hematopoiesis, activation of non-self recognition mechanisms, and production of antimicrobial peptides), and modulation of cell membrane composition [26]. Less is known, though, about the strategies that pathogens employ to escape the immune system of Pacific Oyster. Interestingly, V. coralliilyticus RE22 has been shown to suppress immune signalling pathways of Eastern Oyster (Crassostrea virginica) larvae [41]. Our work expands on these findings by identifying additional conserved mechanisms that Vibrio pathogens may utilize to circumvent immune responses by oyster larvae.
For instance, multiple genes of a ~ 30 Kb-long locus conserved within the genomes of the tested V. coralliilyticus isolates and V. sp. strain RE88 were found to be negatively correlated with larvae survival. This locus includes multiple reb genes and their nearby coding sequences, which are predicted to encode “R-bodies”. R-bodies are insoluble bacterial proteins that confer a phenotype known as the “killing trait” and that are involved in defense against eukaryotic grazers [42]. R-bodies of the intracellular symbiont, Caedibacter, switch between two stable conformations in response to a stimulus such as an extracellular pH change that occurs during phagocytosis. R-bodies of Caedibacter cells, which have been internalized within the lysosomes of a symbiont-free competitor paramecium, will then switch into a needle conformation and rupture the bacterial cell wall, whereupon the cytosolic contents will release unidentified toxins that induce paramecium death [42, 43]. Recombinant Reb proteins from V. nigripulchritudo, a shrimp pathogen, have been visualized by transmission electron micrographs [44], and the reb gene cluster of V. coralliilyticus are proposed to be derived from a horizontal gene transfer event [45], although no known role has been identified in this species. This region of genes appears to be highly conserved in V. coralliilyticus as it is also present in several V. coralliilyticus genomes not included in our study (data not shown). The high correlation of putative reb genes with pathogenicity make this region a target for further investigation to determine its role in virulence or survival within the host, whereby V. coralliilyticus infections may be facilitated by invading and/or modulating the cells of the oyster immune system.
A locus including gene AXN30831.1 was negatively correlated with high rates of larval survival and was annotated to degrade myo-inositol to acetyl-CoA. V. coralliilyticus strain BAA-450 is known to utilize myo-inositol as a carbon source [20]. Pathogens may be encountering inositol in the host environment. For example, host inositol promotes the growth and virulence of Legionella pneumophila within amoeba and host macrophages [46]. Phosphatidylinositols, membrane lipids with a myo-inositol sugar in the headgroup, are important in cell transport signaling and endocytosis and have been found to be important in C. gigas growth [47]. Furthermore, there is evidence that both phosphatidylinositol and soluble inositol phosphate signalling plays a role in immune response by bivalve hemocytes [39, 48, 49], and that C. gigas larvae challenged with Vibrio pathogens show increased in phosphatidyliniositol content [26]. These findings suggest that pathogenic vibrios may modulate oyster immunity by interfering with inositol phosphate signalling, and/or may be able to use host-derived myo-inositol as a carbon and energy source.
Multiple genes annotated to the COG category for carbohydrate utilization were negatively correlated with high larval survival. Two of these were putatively annotated as chitinase genes. One of these is a chiA homolog (AXN33250.1) that is uniquely encoded within V. coralliilyticus genomes, and is divergent from another chiA homolog found to be conserved among the genomes of all Vibrio isolates tested. A study by Lin et al. [50] noted that the former homolog was likely introduced into the V. coralliilyticus clade via horizontal gene transfer, while the more conserved homolog was hypothesized to be vertically inherited within the genus and under strong purifying selection. Another gene, AXN29974.1, with a partial N- acetylglucosamine-binding protein A domain was also negatively correlated with high larval survival. This domain functions in chitin-binding and in V. cholerae, is important for environmental and pathogen fitness, as it promotes biotic surface adherence and host colonization [51]. While the coding sequence of AXN29974.1 shares only 41% coverage and 56.67% identity with GbpA of V. cholerae, its conserved domain architecture is shared with many other V. coralliilyticus strains, as well as a homolog within the genome of V. bivalvicida (WP_054961628), another bivalve pathogen [52].
Genes for chitin binding and degradation are present in almost all Vibrio genomes [50, 53] and chitin metabolism is an important and highly conserved feature, with roles in survival with marine hosts [54], nutrient acquisition [55], and conjugation of extracellular DNA [56]. With regard to oyster hosts, chitin is one of several essential components of their shell matrix [57], and chitin synthase is highly expressed in early stages of C. gigas larval development [58], and in hemocytes and the mantle of adults [58, 59]. Interestingly, growth of V. coralliilyticus strain S2052 on chitin not only induces expression of chitin utilization genes, but also genes related to host colonization, pathogenesis (including reb-type genes), and natural competence [60]. Our observation of a correlation of these putative chitin utilization genes with pathogenicity supports a hypothesis that chitin is a key mediator of interactions between oysters and Vibrio pathogens. Furthermore, the production of chitin by hemocytes and on the oyster mantel indicates that these sites may be targets of Vibrio pathogens.
Multiple genes with putative functions for nutrient acquisition were found to be significantly correlated with pathogenicity. Several were proteases, which can aid in colonization of and persistence within a host, degradation of host biomass for energy and carbon acquisition, or the breakdown of accumulated waste [61]. One of these, a S8 serine peptidase (AXN34464.1) was highly similar to a protein (97.54% identity) found in the secretome of a V. coralliilyticus P1 vcpA mutant that is virulent towards Artemia and Symbiodinium [62]. Additional putative metabolic functions found to correlate negatively (Supporting Table 2) with larval survival included a conserved locus spanning three operons with genes coding for disaccharide transport, as well as the high-affinity pstSCAB phosphate transporter. Phosphate limitation may be specifically important in intracellular vesicles of phagocytic host cells, as has been indicated for Salmonella enterica within macrophages and epithelial cells [63]. These metabolism features correlated with pathogenicity may reveal strategies and adaptations for colonization of a host or other environments.
Genes vcpA, vcpR, toxR were not found to be significantly associated with pathogenicity, despite being known virulence factors of V. coralliilyticus [18, 27, 64]. These genes were generally found to be conserved among all genomes tested here, precluding detection of statistical relationships between their presence/absence and virulence. Additionally, multiple known virulence genes of V. splendidus pathogens shown to be important for virulence of species were not significantly correlated with our pathogenicity results. This includes the exported protein of unknown function R5-7 (AXN33180.1) [14, 65], present in 33 tested isolates, the MARTX toxin of V. splendidus (WP_108195411.1) [14], present in two tested isolates, and Type 6 Secretion Systems (T6SS) [14, 15, 65]. Apparent differences between previous results and ours may arise from minor, but important, differences in sequences of each locus, differences in methods applied for gene clustering, or divergent controls of transcription of conserved genes between pathogenic and non-pathogenic isolates. Host factors such as age (larvae versus adult C. gigas) or a host genotype factor [66] may also account for the divergent results. An additional caveat of our work is that gene clusters detected as in-paralogs, such as one of the T6SS clusters present in many genomes of pathogens, were not analyzed with the phylogenetic logistic regression precluding direct comparisons with previous results.
While only 19 homologous gene clusters were found to be positively correlated with survival of larvae, 508 homologs were identified to negatively correlate with high survival rates, indicating that Vibrio pathogenesis towards C. gigas larvae is a complex trait involving multiple adaptive features. The wide diversity of functions encoded by genes correlated with larval pathogenicity, included ones with more classical virulence functions such as proteases and T3SS-related functions, but also genes putatively encoding functions for colonization, nutrient acquisition, and those with unknown functions. Our results indicate that infection of oyster larvae by V. coralliilyticus may involve an intracellular stage facilitated by the T3SS and the Reb defense system. We therefore hypothesize that these pathogens interact with the immune system of larvae in a similar manner that is seen for V. coralliilyticus infection of the coral Pocillopora damicornis and of V. tasmaniensis with oysters, where infection proceeds through the formation of intracellular bacterial aggregates, followed by the repression of host innate immunity, host cell lysis and extensive tissue damage [15, 67, 68]. These findings point to the complexity of vibrosis outbreaks, whereby environment, microbial community structure, and genotypic and phenotypic features of Vibrio populations all have important roles in disease outcomes. The genetic features described here will be fruitful targets for future mechanistic studies of the oyster larvae pathogenesis.