Diversity and function of prevalent symbiotic marine bacteria in the genus Endozoicomonas

1It is increasingly recognized that eukaryotic organisms rely on bacterial associates to provide a diversity of functions, from supplying nutrients and essential amino acids, to protection from pathogenic microbes and degradation of toxins (McFall-Ngai et al. ). Despite only being recently described (e.g. Kurahashi and Yokota ), the bacterial genus Endozoicomonas (Gammaproteobacteria; Oceanospirillales) has been reported to associate with a large diversity of marine organisms, including cnidarians, poriferans, molluscs, annelids, tunicates, and fish (Jensen et al.; Morrow et al.; Forget and Juniper; Fiore et al.; Katharios et al. ). They are also globally distributed and have been found living symbiotically with organisms in all major oceans of the world (Neave et al. ). However, although they are ubiquitously distributed, the functional role of Endozoicomonas is unclear. Their suggested roles have ranged from a beneficial symbiont required for healthy host functioning to a pathogen that can rapidly cause host death (Bourne et al.; Katharios et al. ).

2This mini-review provides an update on the ever-expanding range of known Endozoicomonas hosts and global distributions and addresses recent advancements in understanding the genetic potential and possible functions of bacteria in the genus Endozoicomonas. We conclude with a discussion on strategies for uncovering new insights into the lifestyle of this cryptic and enigmatic genus, including emerging tools for the study of microbial-animal symbioses, and provide recommendations for future work.

3The genus Endozoicomonas was described less than a decade ago by Kurahashi and Yokota (Fig. 1), after they isolated an unknown gammaproteobacterial symbiont from the sea slug Elysia ornata. Physiological and phylogenetic analyses indicated that the creation of a new genus, Endozoicomonas (i.e. monad living inside an animal), was required (Kurahashi and Yokota ). After this initial description, references to Endozoicomonas remained relatively scarce in the literature until 2010, when Endozoicomonas montiporae (Yang et al. ) was described, and coral symbionts with similarity to Endozoicomonas were discovered (Kvennefors et al. ). Kvennefors et al. also noted that symbionts from earlier microbiome studies (before the creation of the genus) were closely related to Endozoicomonas. Examples include abundant gammaproteobacterial symbionts termed “PA1” from the coral Porites astreoides (Rohwer et al. ) and Gammaproteobacteria from the coral Pocillopora damicornis (Bourne and Munn ). These efforts to retroactively link past studies to the Endozoicomonas genus, in addition to the detection of Endozoicomonas in new hosts, resulted in a spike in citations of Endozoicomonas within the scientific literature. From 2013 onwards, more than 15 publications per year referred to the genus (Fig. 1). This relatively rapid rise in Endozoicomonas publications, and the novelty of the genus, has led to some taxonomic inconsistencies finding their way into journal articles and databases. For example, the Greengenes database (DeSantis et al. ) places the genus Endozoicomonas in an apparently newly created family called “Endozoicomonaceae”. However, despite not being explicitly stated, the type description of Endozoicomonas contained nucleotide information that placed the genus within the family Hahellaceae (Kurahashi and Yokota ) in the order Oceanospirillales and this taxonomy was maintained in subsequent type descriptions (Yang et al.; Nishijima et al. ). The incorrect term Endozoicomonaceae has appeared in a number of scientific publications (e.g. Dishaw et al.; Katharios et al.; Lawler et al. ) paying tribute to just how rapid this genus became scientifically popular. Confusion surrounding the initial naming of the genus has also produced inconsistencies in the scientific literature. For example, in March 2005 (prior to the Endozoicomonas description), a sequence named “Spongiobacter nickelotolerans” was submitted to GenBank (#AB205011), which is very similar (>97 % SSU rRNA identity) to the Endozoicomonas type strains. The associated taxonomic paper describing the isolation of S. nickelotolerans from a marine sponge, however, was not published and the name appears to have been abandoned (e.g. Pike et al.; McCauley et al. ). Nevertheless, several publications have referred to “Spongiobacter” or to the “Spongiobacter/Endozoicomonas” group (e.g. Costa et al.; La Rivière et al. ).

4The volume of genetic data available for Endozoicomonas bacteria in NCBI’s GenBank (Benson et al. ) has also rapidly increased (Fig. 1). Since the initial deposition of the first 1437 bps SSU rRNA sequence of E. elysicola by Kurahashi and Yokota, Endozoicomonas nucleotide information has steadily accumulated, reaching almost 100,000,000 bps by the start of 2016 (Fig. 1). Moreover, this number only takes into account data retrieved from GenBank; far more Endozoicomonas genetic information is available in other databases, such as NCBI’s Sequence Read Archive (SRA). Much of this rapid accumulation of genetic information can be attributed to the move towards whole genome sequencing, rather than marker gene sequencing. The first Endozoicomonas genome sequenced was E. elysicola in 2013 (Fig. 1) as part of the one thousand microbial genomes project (Kyrpides et al. ). The following year, Neave et al. released an updated version of the E. elysicola genome plus two new genomes, E. montiporae (Yang et al. ) and E. numazuensis (Nishijima et al. ). Since then, the E. montiporae genome has been re-sequenced (Ding et al. ), the genome of E. atrinae has been made available (Hyun et al. ), the genome of an undescribed pathogenic Endozoicomonas has been analysed (Katharios et al. ), and several other Endozoicomonas genome projects are underway (e.g. Appolinario et al. ). Most recently, Neave, Michell, Apprill and Voolstra (Endozoicomonas genomes reveals functional adaptation and plasticity in bacterial strains symbiotically associated with diverse marine hosts, Submitted) applied single-cell genomics and metagenomic binning to recruit four additional Endozoicomonas genomes from native coral host assemblages.

5Endozoicomonas symbionts have a global distribution in numerous marine hosts, from abyssal depths to warm photic zones (Fig. 2). They are most frequently detected as coral symbionts and occur throughout the global distribution of coral reefs, from the Great Barrier Reef in Australia (Bourne et al.; Lema et al. ), Papua New Guinea (Morrow et al. ), Indonesia and the Pacific (Yang et al.; Neave et al. ) to the Red Sea (Bayer et al.; Jessen et al.; Neave et al.; Ziegler et al. ), Indian Ocean (Neave et al. ), and the Caribbean (Morrow et al.; Rodriguez-Lanetty et al. ) (Fig. 2). Interestingly, different coral species appear to harbour specific Endozoicomonas genotypes (Neave et al. ). For example, the sympatric corals, Stylophora pistillata and Pocillopora verrucosa, each contain different Endozoicomonas genotypes, and these genotypes partition differently across large geographic scales (Neave et al. ). Moreover, patterns of symbiont specificity seem to co-align with differences in reproductive mode. For instance, the spawning coral P. verrucosa globally associates with the same Endozoicomonas symbionts, whereas the brooding coral S. pistillata harbours different Endozoicomonas genotypes in different regions (Neave et al. ).

6Although no quantitative methods have been applied to Endozoicomonas, trends in cell abundance have been inferred from SSU rRNA gene sequence abundances. These studies have linked the abundance of Endozoicomonas to the abundance of its coral host. For example, when the fungid coral Ctenactis echinata grew in its preferred Red Sea habitat, Endozoicomonas symbionts were more abundant than in habitats of degraded quality (Roder et al. ). Moreover, reduced abundances of the corals Acropora millepora and Porites cylindrica near carbon dioxide seeps in Papua New Guinea coincided with a 50 % reduction in Endozoicomonas symbionts (Morrow et al. ). Anthropogenic pollution can similarly decrease the abundance of Endozoicomonas bacteria. Near the large Red Sea city of Jeddah, the corals P. verrucosa and Acropora hemprichii contained a lower proportion of Endozoicomonas compared to corals further afield (Ziegler et al. ). In addition, bleaching of the coral A. millepora on the Great Barrier Reef induced a shift from Endozoicomonas-like symbionts to a Vibrio-dominated community (Bourne et al. ). Lesioned P. astreoides colonies also contained reduced Endozoicomonas sequence abundances, compared to non-lesioned colonies (Meyer et al. ). These studies suggest that Endozoicomonas bacteria are part of a healthy coral microbiome and reductions in their abundance may indicate unfavourable environmental conditions.

7Gorgonians, commonly known as sea fans, are closely related to corals and also have a microbiome frequently dominated by Endozoicomonas. In the Mediterranean, the main gorgonian species (Eunicella cavolini, E. singularis, E. verrucosa, Leptogorgia sarmentosa, and Paramuricea clavata) harbour a microbiome dominated by Endozoicomonas symbionts (van de Water et al.; Bayer et al.; La Rivière et al. ). Interestingly, patterns of host-specificity and possibly co-evolution between the gorgonian species and their specific Endozoicomonas genotype have also been observed (van de Water et al.; La Rivière et al. ). In addition, another octocoral species, the red coral Corallium rubrum, also harbour Endozoicomonas, although it is much less abundant compared to the other gorgonians and is dominated by Spirochaetales (van de Water et al. ). Outside of the Mediterranean, Endozoicomonas symbionts were found in the gorgonian E. verrucosa living off the south-west coast of England (Ransome et al. ), as well as in the Caribbean gorgonian Pseudopterogorgia elisabethae (Correa et al. ). Although corals and gorgonians dominate the Endozoicomonas literature, these symbionts have also been found in a range of other cnidarian species. Early work by Schuett et al. detected an Endozoicomonas strain with 98 % similarity to E. elysicola and observed bacterial cell aggregates in the tentacles of the sea anemone Metridium senile from Helgoland in the North Sea. Although these aggregates were not confirmed as Endozoicomonas, they used scanning electron microscopy to obtain detailed images of the bacterial cells forming large aggregates within host tissues. More recently, an Endozoicomonas strain with 99 % similarity to E. elysicola was found in another sea anemone, Nematostella vectensis, from a marsh of varying salinity conditions in Massachusetts, USA (Har et al. ). Endozoicomonas symbionts also dominated two jellyfish species, Mastigias cf. papua and Tripedalia cf. cystophora, in several Indonesian lakes (Cleary et al. ). These examples show that Endozoicomonas bacteria are symbiotic with a large diversity of cnidarian species and are often abundant and host species-specific, suggesting an important and ancient evolutionary relationship with lineage-specific evolution.

8Endozoicomonas are also known to associate with a wide range of other marine organisms (Fig. 2). A number of particularly interesting examples of the adaptability of Endozoicomonas come from deep-sea hydrothermal vent communities. Forget and Juniper collected the tubeworm, Ridgeia piscesae, from the Juan de Fuca Ridge in the north-east Pacific, which has high hydrogen sulphide concentrations (~40 μmol/L), temperatures up to 41.9 °C, and depths greater than 2000 m. Even in this extreme environment, R. piscesae contained abundant Endozoicomonas symbionts (Forget and Juniper ). Moreover, Endozoicomonas bacteria have been recovered from the gills of the hydrothermal vent snail Alviniconcha (Beinart et al. ), the gills of the deep-water bivalve Acesta excavata (Jensen et al. ), and from the tissues of deep-water corals (Meistertzheim et al. ). Endozoicomonas have also been associated with a number of sponge species, which are one of the oldest groups of metazoan invertebrates and often harbour a rich diversity of microbial symbionts (Rua et al. ). In the Caribbean, the giant barrel sponge (Xestospongia muta) not only contained Endozoicomonas bacteria, but evidence of their transcriptional activity was also detected (Fiore et al. ). Sponges in Brazil (Rua et al. ), Japan (Nishijima et al. ), and several other European seas also contain Endozoicomonas symbionts (Esteves et al.; Gardères et al. ). In addition, tunicates, which are basal chordates, have a microbiome that consistently contains a substantial proportion of Endozoicomonas, suggesting that these symbionts are core members of the tunicate microbiome (Dishaw et al. ). Although reports to date mostly associate Endozoicomonas with marine invertebrates, several examples of associations with fish have also emerged in the recent literature. In these cases, fish were kept in artificial aquaculture environments and Endozoicomonas bacteria are suspected to have caused disease (Mendoza et al.; Katharios et al. ). In both examples, Endozoicomonas formed cyst-like aggregations on the gills of the fish, resulting in epitheliocystis (Mendoza et al.; Katharios et al. ).

9The central picture that emerges is the remarkable ability of Endozoicomonas bacteria to adapt to a wide range of hosts and environments, from warm coral reefs to cold deep-sea mussels, and their apparent ability to transition from beneficial core microbiome members of corals and tunicates to disease-causing pathogens in fish.

10The proposed functions of Endozoicomonas can be summarized into three categories: nutrient acquisition and provision, structuring of the host microbiome, and roles in host health or disease (Table 1). Nutrient acquisition spans from nitrogen and carbon recycling (Nishijima et al.; Forget and Juniper; Correa et al.; Morrow et al. ), or methane and sulphur cycling (Bourne et al.; Forget and Juniper; Correa et al.; Dishaw et al.; Morrow et al. ), to the synthesis of amino acids and other essential molecules (Neave, Michell, Apprill and Voolstra, Endozoicomonas genomes reveals functional adaptation and plasticity in bacterial strains symbiotically associated with diverse marine hosts, Submitted). Bourne et al. found that the abundance of Endozoicomonas-related sequences (referred to as Oceanospirillales sp. 1, 3, 5, and 6) in invertebrate microbial communities correlated with the presence of photosymbionts, such as Symbiodinium algae in coral tissues. They suggested that the photosymbionts provide carbon and sulphur to the bacteria from the large quantities of dimethylsulfopropionate (DMSP) produced (Bourne et al.; Correa et al. ). On the other hand, Endozoicomonas bacteria are also found in hosts without photosymbionts (Bourne et al. ). In addition to nutrient cycling, Endozoicomonas-related members may also play a role in regulating bacterial colonization of the animal host via the production of bioactive secondary metabolites or probiotic mechanisms, such as competitive exclusion of pathogenic bacteria (Bayer et al.; Jessen et al.; Rua et al.; Morrow et al. ). Moreover, the loss of Endozoicomonas is often characteristic of corals with lesions, signs of disease, or if they are living in eutrophicated, warm, or acidic environments. Therefore, the abundance of Endozoicomonas seems to be linked with healthy colonies of diverse coral species (Morrow et al.; Bayer et al.; Roder et al.; Morrow et al.; Ziegler et al. ).

11The spatial location of Endozoicomonas bacteria within host tissues may also have functional implications. For example, Endozoicomonas frequently form aggregations in various host habitats (Bayer et al.; Mendoza et al.; Katharios et al.; Schreiber et al. ). In corals, fluorescent oligonucleotides have been designed for Endozoicomonas and used to hybridize to Endozoicomonas cells, confirming their residence within host tissues (Bayer et al. ), where they formed similar structures to cell-associated microbial aggregates or CAMAs as previously described using histological staining in corals (Work and Aeby ). More recently, using catalyzed reporter deposition fluorescence in situ hybridization (CARD-FISH) with this same probe, bacterial cells have been better resolved within the autofluorescent coral tissues and it was found that Endozoicomonas cells form dense aggregations and can reside within the tentacles of corals (Fig. 3; Neave et al. ). Aggregations filled with thousands of bacterial cells have also been found in anemone tentacles, and although SSU rRNA gene sequence data demonstrates that Endozoicomonas are in residence with these anemones, the tentacle-associated cells have not yet been confirmed as Endozoicomonas (Schuett et al. ). Another probe developed for Endozoicomonas also identified cells residing in extracellular aggregations in sea squirts (Schreiber et al. ). In fish, Endozoicomonas also form extremely dense aggregations containing thousands of individuals surrounded by a thin tightly enveloping membrane (Mendoza et al.; Katharios et al. ). These formations, particularly the membrane barrier, may provide protection from host immune cells or other host responses to bacterial infection. Functionally, the aggregations may act as centres of protein transformation and production that could be beneficial for the host. Moreover, several genotypes of Endozoicomonas are known to inhabit individual hosts (Neave et al. ), and the aggregations could be comprised of several complementary genotypes, or even different bacterial species, that work together to passage nutrients and proteins. The formation of aggregations also suggests that some form of cell-to-cell communication is required, such as quorum-sensing molecules (Waters and Bassler ). To date, all Endozoicomonas microscopy studies have found aggregations in host tissues, suggesting that these formations are an important part of Endozoicomonas function and colonization.

12The recent availability of Endozoicomonas whole genome sequences has significantly improved our understanding of their genetic potential, possible functional roles, and life cycle. These data have revealed that Endozoicomonas genomes are relatively large, ranging from 5.60 Mbs for E. montiporae (Neave et al. ) to 6.83 Mbs for E. atrinae (Hyun et al. ). The genomes support a correspondingly large number of protein-coding genes, suggesting that genome streamlining, which often occurs in symbiotic bacteria (e.g. Kwan et al. ), is not occurring or at least not predominantly. This may indicate that Endozoicomonas have periods during their lifecycle that require a full complement of genetic pathways, e.g. a free-living stage. In a recent study of the E. montiporae genome, Ding et al. found genes coding for an N-deglycosylation enzyme that was hypothesized to help with penetration of the mucus layer of their coral host. They also found genes that may be involved in initiating internalization and evasion of the host immune response and hypothesized that Endozoicomonas was a beneficial symbiont that could help increase the efficiency of host gluconeogenesis (Ding et al. ). Interestingly, a high proportion of repeat sequences was found in the E. montiporae genome (Ding et al. ), which is similar to what was reported for a pathogenic Endozoicomonas strain (Katharios et al. ). This suggests that repeat and insertion sequences may help Endozoicomonas strains to adapt to new hosts or to transition between mutualistic and parasitic lifestyles. Importantly, Neave, Michell, Apprill and Voolstra (Endozoicomonas genomes reveals functional adaptation and plasticity in bacterial strains symbiotically associated with diverse marine hosts, Submitted) comparatively analysed the genomes of E. elysicola, E. montiporae, E. numazuensis, and four newly sequenced Endozoicomonas strains from the Red Sea corals S. pistillata, P. verrucosa, and Acropora humilis and found a high proportion of transposable elements in the Endozoicomonas genomes, further implying that Endozoicomonas use these elements to rapidly evolve to new hosts or niches. In addition, the Endozoicomonas genomes were enriched for carbon sugar transport and protein secretion, suggesting that they contribute to carbohydrate cycling and delivery to their host organism (Neave, Michell, Apprill and Voolstra, Endozoicomonas genomes reveals functional adaptation and plasticity in bacterial strains symbiotically associated with diverse marine hosts, Submitted). The common denominator among Endozoicomonas genome projects is the high incidence of transposable elements incorporated into their genomes, possibly allowing for rapid adaptation. These genomic findings are corroborated by field evidence showing that Endozoicomonas are associated with numerous hosts in numerous marine habitats. The genomes also tend to show the ability to transport and transform protein products that may then be used by the host. Although these studies provide important advances, future studies may be improved by obtaining in situ gene expression data or other forms of functional information, in order to refine our understanding of the specific role of Endozoicomonas with a particular host.

13The functional role of Endozoicomonas bacteria within their many hosts is poorly understood. One fundamental but unknown aspect of many Endozoicomonas species is their genomic repertoire and metabolic potential. This question has traditionally been difficult to answer; however, new and emerging methods for determining bacterial function are likely to provide fascinating insights. Currently, many techniques that aim to discover bacterial function rely on the availability of cultured isolates. For example, bacterial cultures allow for efficient whole genome sequencing, for infectivity trials, and for traditional bacteriological tests of physiological and nutritional requirements. Although a number of Endozoicomonas species have been successfully cultured, many species are not readily amenable to isolation from host tissues (e.g. Katharios et al. ). In these cases, culture-independent methods can be used to obtain valuable data, such as whole genome sequences. For example, Katharios et al. micro-manipulated Endozoicomonas aggregates from sections of fish tissue, sequenced the DNA using high throughput technologies, and obtained a novel draft genome that provided valuable insights. Moreover, in cases where Endozoicomonas aggregates are not easily micro-manipulated, methods such as metagenomic binning and single-cell genomics can be used to acquire whole genome sequences. Recently, the genomes of Endozoicomonas associated with different species of Red Sea corals were obtained by differential coverage binning of coral metagenomes, as outlined by Albertsen et al., and by sequencing single-cell isolates (Neave, Michell, Apprill and Voolstra, Endozoicomonas genomes reveals functional adaptation and plasticity in bacterial strains symbiotically associated with diverse marine hosts, Submitted). With further improvements in metagenomic binning software (e.g. Imelfort et al.; Alneberg et al.; Eren et al. ) and in single-cell genomic methods, these may become useful for rapidly obtaining multiple genomes at a relatively small cost. A major improvement to these pipelines would be the addition of techniques that obtain gene expression data. One possible avenue that is already being developed is the use of RNA-Seq on isolated single cells (e.g. Wang and Navin ). Another is RNA subtraction technology, where total RNA is collected from host tissues, ribosomal RNA and eukaryotic messenger RNA are removed, and the remaining bacterial mRNA is sequenced (e.g. Stewart et al.; Daniels et al. ). When combined, these techniques could provide fascinating insights into the intricate associations of Endozoicomonas-animal symbioses.

14Although whole genome sequences and gene expression data are valuable, they cannot completely uncover the metabolic and physiological dynamics of Endozoicomonas bacteria in situ. For example, questions of how Endozoicomonas aggregates share metabolites with the host, or which Endozoicomonas genotypes are represented in aggregates, are difficult to answer. These questions may become tractable with the development of secondary ion mass spectrometry (SIMS), which has opened a new frontier in microbial ecology. While previously measurements of bacterial metabolism and physiology were restricted to community-based culture or whole-animal measurements, these novel technologies allow for the quantification of single-cell metabolisms at the spatial resolution of electron microscopy (Pernice and Levy ). Thereby, these technologies not only enable one to localize and quantify metabolic activity of selected tissues or microorganisms, but can also account for single-cell heterogeneity, a previously overlooked phenomenon, and of particular importance when hosts harbour multiple strains of Endozoicomonas (e.g. Neave et al. ). There are several SIMS technologies used to investigate metabolic functioning of bacteria in culture or within a host-symbiont framework. Nanoscale SIMS (NanoSIMS) is used to quantify stable isotope ratios at high spatial resolution. Coupled with pulse-chase isotope labelling, this technology can be used to detect the uptake, incorporation, and transfer of metabolites (Kopp et al. ). This technique may be particularly useful in detecting molecules that are consumed and produced by Endozoicomonas aggregations, providing data on molecule transformation within aggregations and allowing fascinating insights into potential benefits for the host. A recently developed modification of the traditional NanoSIMS approach is halogen in situ hybridization SIMS (HISH-SIMS) (Alonso et al. ), where specific probes are used to label targeted cells, which allows identification of the precise location of microbes within host tissue. This modification would permit Endozoicomonas aggregations to be interrogated for single-cell heterogeneity, including different genotypes or different bacterial species, and assess the relative contributions of these to aggregate functioning. Another technique called Time-of-Flight SIMS (ToF-SIMS) uses high-resolution mass spectrometry to identify molecules within a given cell or tissue (Colliver et al. ). This technology can be used to identify metabolic processes within microorganisms or the compounds they release, which would provide metabolic data at the level of individual Endozoicomonas cells and genotypes. While all of these technologies have proven to be powerful tools in understanding microbial function, they are even more powerful when combined. For instance, using Nano-, HISH- and ToF-SIMS on the same biological sample would make it possible to highlight specific Endozoicomonas genotypes in a host organism background and to characterize and quantify the nature of translocated metabolites, which should provide quantum leaps in our understanding of the function of host-associated Endozoicomonas bacteria.

15Endozoicomonas bacteria are rapidly being recognized as globally important marine symbionts. Few other bacterial genera are symbiotic with such a large diversity and distribution of host species, suggesting that the function and adaptability of Endozoicomonas bacteria are particularly exceptional. Indeed, several Endozoicomonas species have unusually flexible genomes and a great diversity of metabolic pathways. Studies on the exact functional roles of Endozoicomonas within their many hosts are only in their infancy, and many exciting and novel discoveries await. Moreover, with new technologies allowing for in situ genome sequencing, gene expression analysis, and the measurement of metabolites, future functional discoveries will be greatly accelerated.