INTRODUCTION
In recent years, the concept of the metaorganism or holobiont, which defines the associations formed by a host organism and its microbiome (
1–3), has become a cornerstone of biology (
4). Scleractinian corals are an excellent example of host-microbe associations, as they build reefs through close symbiotic interactions between the host modular animal, its endosymbiotic dinoflagellates (
Symbiodiniaceae), and an array of other microbial partners, including bacteria, archaea, and fungi (
3,
4). The bacterial taxa associated with corals can vary between coral species and geographical origin, though often there are patterns in the community structure that link microbial and coral taxa (
5,
6). Many original discoveries on the importance of coral-associated bacteria and their interactions with the coral host were made using culture-based methods (
7,
8). However, the majority of recent studies exploring the importance of coral-associated microbes have focused on the use of cultivation-independent approaches, based on 16S rRNA gene amplicon sequencing (
9) and, more recently, shotgun metagenomics (
10,
11). Such methods are central in identifying what bacteria are associated with corals and how their metabolic and functional potential contribute to holobiont health and response to environmental conditions (
9,
12–14). However, the bacterial metabolic pathways that interact with the host and respond to environmental changes are often best understood using culture-based approaches (
15). This is particularly relevant because metagenomic information gives insights into potential functional traits and other cellular traits only, and often environmental changes have pleiotropic effects on holobiont physiology that are impossible to grasp using metagenomics alone (
16–19).
Inherently, culture-based approaches retrieve only a small fraction of the total bacterial diversity within any given environment, a phenomenon known as the “great plate anomaly” (
20–22). Often however, it is not a case of being “unculturable” but of not yet knowing the (range of) conditions needed to culture specific microorganisms (
23). Cultivating host-associated microorganisms can be challenging, as their nutrient requirements and cross-feeding networks are often unknown (
24). In addition, many “environmental” microorganisms grow very slowly (in contrast to clinical isolates), and are not adapted to or capable of growing on commonly used nutrient-rich media, and are outcompeted by copiotrophic bacteria (
25,
26). To counter this, at least to some degree, recent studies have implemented novel and alternative culture-based methods to retrieve a higher proportion of the bacterial diversity present in any given sample (
24,
27,
28), and these approaches have also been applied to corals (
29–31).
Organismal, growth form, and tissue complexity create unique microenvironments that are thought to contribute to the high bacterial diversity often seen in corals (
32–34). The diverse coral bacteriome plays an integral role in the balance between health and disease of the coral holobiont (
35,
36) and represents a valuable source of biotechnological products (
37,
38). Disalvo (
39) was perhaps the first to isolate bacteria from coral in 1969, recovering strains from the skeletal regions of
Porites lobata, followed by Ducklow and Mitchell (
40) who reported on bacteria isolated from mucus of
Porites astreoides and two octocoral species 10 years later. Microbe-mediated diseases have also been well documented as driving declines in reef health, especially throughout the Caribbean for example (
41). This has fostered a great interest in understanding coral disease causative agents, stimulating cultivation efforts of coral-associated bacteria (
42–44). For example, Kushmaro et al. (
45) isolated a bacterium that caused bleaching of the coral
Oculina patagonica, and many subsequent studies have implicated vibrios in coral disease causation (
46–48)—although it should be noted that coral bleaching is not typically considered a disease and is ascribed to dysbiosis of the coral host and associated
Symbiodiniaceae (
49). Regardless, many of these studies focused on targeted isolation and conducted reinfection studies to satisfy Koch’s postulates, with varying success (reviewed in reference
50).
Counter to the notion of pathogenicity of certain bacteria, growing evidence underlines the key role secondary metabolites produced by (beneficial) bacteria have on host health (
35,
51–54). For instance, Ritchie (
55) was among the first to demonstrate that mucus-associated bacteria from healthy colonies inhibit the growth of potential pathogens. Subsequent studies revealed high antimicrobial activity among culturable coral-associated bacteria, with up to 25% of the isolates producing antimicrobial compounds (
56). Kuek et al. (
57) showed a strong link between observed antibiotic activity in well diffusion assays and existence of polyketide synthase (PKS) and/or nonribosomal peptide synthetase (NRPS) genes in the bacterial isolates. More recently, Raina et al. (
17) found that the antimicrobial compound tropodithietic acid (TDA) was produced by the coral-associated bacterium
Pseudovibrio sp. and subsequent studies found that
Pseudovibrio species harbor several biosynthetic gene clusters for the synthesis of bioactive compounds (
58,
59).
Bacterial isolates from corals represent an invaluable resource for assessing the virulence of potential pathogens, and for applying classical clinical approaches to elucidate disease etiology (
60). Beneficial traits that bacteria may provide to coral holobiont functioning can also be elucidated using pure bacterial cultures (
10,
18). Bacteria isolated from corals can also be used as probiotics to facilitate host health (
61,
62), and such approaches have been proposed to promote coral resilience in the face of environmental stress. For example, Rosado et al. (
53) showed that application of so-called “beneficial microorganisms for corals” (or BMCs) increases the resilience of the coral to temperature stress and pathogen challenge. However, despite the demonstrated importance of BMCs (
63), a centralized and curated collection of isolates obtained from corals and their associated genetic information does not currently exist. Moreover, many culture-based studies often focus on relatively few bacteria (targeted for pathogenic agents for example), meaning a large-scale comparison of which bacterial isolates can be cultured and their genetic information is currently missing. Here, we sought to centralize and curate the current cultured fraction of coral bacteria by combining published data with unpublished collections from around the world (
Fig. 1). Without doubt, some studies and culture collections will have been missed in this first compilation; however, our aim was to start building a resource that can be built upon. To highlight the importance of such a collection, we explore the relationships between the isolated bacteria, the host origin, and the media utilized for growth. Further, a total of 74 genomes of cultured coral bacteria, 36 of which are available in public databases and 38 of which are presented in this study for the first time, were investigated to infer potential genetic signatures that may facilitate a host-associated lifestyle. Finally, alternative ways and improvements for the isolation of bacterial groups not yet recovered from corals (including the specific targeting of obligate symbionts) are discussed. This study provides the most comprehensive synthesis of the cultured bacterial fraction of the coral holobiont thus far.
DISCUSSION
Here, we show that a taxonomically diverse array of bacteria can be isolated using a variety of medium and culture conditions. A total of 138 of these isolates (recruited from 52 studies) have been formally described, and at least 12 are putatively novel bacterial genera. It is promising that such extensive phylogenetic diversity can be captured from a limited number of culture media employed in the examined studies. Additional diversity is therefore likely to be captured through the implementation of alternative cultivation procedures that may improve our capacity to cultivate the “as-yet-uncultured” (
28). Testimony to this is the observation that most of the strains assigned to the phylum
Firmicutes in our meta-analysis were obtained almost exclusively from the various “custom media” utilized by different laboratories and blood agar alone, illustrating how diversification in cultivation design can widen the phylogenetic spectrum of the organisms isolated. In this regard, we anticipate that broader phylogenetic diversity will be gained within the culturable fraction if gradients in aerophilic conditions, temperature, and other physicochemical parameters are attempted along with innovative, less invasive techniques to extract microbial cells from the host matrix. The richness of bacterial phyla uncovered in this study corresponds to the phyla more often reported to dominate bacterial communities in corals by cultivation-independent studies (
12), namely,
Proteobacteria,
Bacteroidetes,
Actinobacteria, and
Firmicutes, yet how diversity at lower taxonomic ranks within each phylum is captured remains to be determined. Another exciting challenge ahead is the unveiling of host-microbe and microbe-microbe molecular interdependence networks (e.g., cross-kingdom signaling and cross-feeding cascades) (
79,
80). Such knowledge would likely enable laboratory captivation of so-far “unculturable” coral-specific or enriched lineages. Increasing the diversity of these coral-associated culturable bacteria will likely help in the identification of genomic features that could underpin the interaction with the host and its microbiome representing the foundation for experimental validation.
Although one of the initial aims of this study was to ascertain the percentage of culturable bacteria from a given coral species, it was deemed too speculative to report the findings due to variation in culture effort across the various studies. Indeed, this highlights the paucity of studies dedicated to determine exactly this, and there is a need for such mechanistic projects deploying multiple culture media and conditions to comprehensively sample bacterial associates from a single or a few host species. Collectively, studies aimed at capturing the culturable microbiome will extend our understanding of coral bacterial communities and their putative function in the coral holobiont. A catalog of cultures (as presented here and one which will hopefully be expanded) provides a means to increase our understanding of host-symbiote relationships. The ability to describe, understand, and culture specific symbionts from any given organism (like corals) also opens up the potential to utilize them as probiotics to restore degraded habitats (
53,
61). For example, specific traits found in certain coral-associated bacteria, such as the presence of the genes
nifH (nitrogenase),
nirK (nitrite reductase), or
dmdA (DMSP demethylase) involved in nitrogen and sulfur cycling, or those known to control pathogens, the enzymatic mitigation of reactive oxygen species (ROS) or other toxic compounds, may have roles in increasing coral health when the host is experiencing stress (
53,
63,
81,
82). Identifying these traits via molecular analyses and laboratory tests using cultured bacteria with defined coral hosts will allow for the more rapid administration of native bacteria with the potential to help rehabilitate damaged corals. In addition, such a resource increases the possibility of identifying novel compounds of biotechnological interest (
83). This seems particularly relevant in the case of coral-microbe symbioses, which are known to rank as one of the most prolific sources of bioactive molecules in the oceans (
38).
A search in public databases (National Center for Biotechnology Information [NCBI]) found that, despite the 1,045 cultured coral-associated bacterial sequences with full-length 16S rRNA gene sequences, only 36 had genomes available as of February 2020. Clearly, a systematic effort to disclose the genomic features of coral-associated bacteria is needed in order to better understand the holobiont ecology and identify potentially beneficial microbes. As part of this study, we were able to add a further 38 to this tally (see
Table S2 in the supplemental material). Even with this addition, the number of publicly available coral-associated bacterial genomes remains scant, and it is recognized that to more fully understand the roles of the cultivable fraction of coral bacteria, a thorough characterization of the species kept in culture, including genome sequencing, needs to be fostered alongside experimental biology and manipulative approaches. Moreover, a large collection of coral-associated genomes could also help to identify specific traits that are needed to thrive in the various niches within the hosts or point to those bacteria which offer a specific benefit to their host.
All of the available genomes were screened for an array of functions potentially important in establishing and maintaining interactions between bacterial symbionts and their marine invertebrate hosts. Overall, the
Endozoicomonas and
Pseudoalteromonas strains displayed high numbers of eukaryote-like protein-encoding genes important for host-symbiont recognition in well-studied systems such as marine sponges (
65,
84,
85). The strain with the second highest number of eukaryote-like repeat protein-related entries (1,208 CDSs, after
Endozoicomonas sp. G2_1 with 1,367 CDSs) was the octocoral associate
Aquimarina sp. strain EL33 (class
Flavobacteria). In the current culture collection, 15 additional
Aquimarina isolates are reported, from the scleractinian corals
Porites lutea,
Pocillopora acuta,
Stylophora pistillata,
Acropora millepora,
Acropora tenuis, and the octocoral
E. labiata. Retrieving the genomes from these candidates will allow us to explore these emerging patterns in greater detail. For example, a recent comparative genomics survey of host-associated and free-living
Aquimarina species revealed complex secondary metabolite biosynthesis and polycarbohydrate degradation capacities (
86), but further investigation into their mechanisms of interactions with corals is warranted.
Only eight
Endozoicomonas isolates (five of them type species) have so far been cultured from corals (according to our collated information). These are from the octocorals
Eunicea fusca and
Plexaura sp. and the scleractinian corals
Montipora aequituberculata,
Acropora cytherea,
Acropora hemprichii, and
Acropora sp. To date, only four of these (two from this study) have had their genomes sequenced (all from scleractinian corals) (
18,
87). This is surprising given that numerous studies found that this genus is highly abundant in the healthy coral holobiont and seems to decrease in abundance upon deteriorating environmental conditions (e.g., reviewed in references
35,
88, and
89). Future cultivation efforts should therefore be directed toward the
Endozoicomonadaceae family in order to increase the representation of their taxonomic and functional diversity in culture collections (
29). In this regard, this study finds evidence that supplementing culture media with DMSP is an approach worth investing in future attempts to cultivate coral-associated
Endozoicomonas, possible in combination with growth at lower temperatures (
29). The metabolic data obtained from the comparative analysis of these four strains can be used, for example, to drive the selection of specific nutrients and conditions required to culture this particular genus of coral symbionts. Furthermore, there are 55 cultured
Pseudoalteromonas strains in our collection which should also be explored regarding their symbiotic properties and their functional gene content (only 6 genomes currently available). Similar to
Endozoicomonas,
Pseudoalteromonas species are also frequent members of coral-associated microbiomes (
35). A number of
Pseudoalteromonas have been shown to display high antimicrobial activity, and many of these bacteria are isolated from coral mucus, lending support to the protective role the surface mucous layer has for the host and its importance in the coral holobiont’s defense—against bacterial coral pathogens in particular (
90). Indeed, five of the six
Pseudoalteromonas (where genomes are available) were shown to be effective BMCs when corals were challenged with the coral pathogen
Vibrio coralliilyticus (
53).
Having genomes available from the potential pathogens also allows for greater insight into coral biology, especially when interested in ascertaining pathogenicity-related traits (
91,
92). For example, from the 11
Vibrio species for which genomic data were available, we were able to show functional separation (based on Pfam profiles) of known pathogenic and nonpathogenic strains. This was further accompanied by a significantly higher abundance of CDSs encoding for the type VI secretion system, important for virulence in the pathogenic strains (
76). Prevalence of siderophore-encoding genes was also noted in the
Vibrionaceae strains, suggesting that these bacteria likely gain competitive advantages through efficient and extensive iron acquisition, which is a trait often seen in opportunistic and pathogenic bacteria (
93,
94). Hypothetically, the selection of beneficial microbes that are also good siderophore producers could add to the biological control of these pathogens. Indeed, two proposed BMC strains
Cobetia marina BMC6 and
Halomonas taenensis BMC7 harbor such siderophore clusters on their genomes and so did three of the four
Endozoicomonas strains. However, the five
Pseudoalteromonas BMC strains and the
Endozoicomonas montiporae CL-33 had low numbers of BGCs, possibly indicating a reduced investment into secondary metabolism. Indeed, the low number of BGCs in these
Pseudoalteromonas strains is in contrast to the established prevalence of biologically active compounds in many marine host-associated
Pseudoalteromonas strains (
95). In part, this may reflect a limitation of the software utilized to detect genes for all secondary metabolites, as genes for common metabolites (such as for the production of the antibiotic marinocin and those that produce tetrabromopyrrole coral larval settlement cues by
Pseudoalteromonas [
96,
97]) were not picked up. These bioinformatic limitations emphasize the importance of having bacterial cultures for the elucidation of the chemical ecology underpinning coral holobiont functioning.
Broader functional traits can also be ascertained from looking at the complete picture of isolates with annotated genomes. For example, 66% (49 out of 74) harbored the TauD gene, which is involved in taurine utilization (
98). Two proposed BMCs, the
Cobetia marina BMC7 and
Halomonas taeanensis BMC7, revealed the highest copy number of TauD CDSs (seven and eight, respectively), while others range between one and five TauD copies. Taurine is an organo-sulfur compound widely present in animal tissues, and recent research has shown that obligate symbionts of sponges have enriched copies of taurine catabolism genes and taurine transporters in comparison with free-living bacteria (
65,
72,
73). The widespread capability of the isolates studied here to potentially utilize host-derived taurine could guide the formulation of novel, taurine-containing cultivation media in the attempt to captivate coral symbionts, particularly from the important, yet underrepresented order
Oceanospirillales (TauD was consistently present in all
Oceanospirillales genomes [
N = 8] analyzed here). The ubiquitous occurrence of bacteriocin clusters among the genomes is another example of broad-scale trends which we have identified in our genome meta-analysis. These may confer the specific culturable symbionts with particular competitive capacities toward closely related taxa in highly dense microbiomes (
99,
100), as is commonly identified across corals and sponges. Moreover, the widespread presence of NRPS and beta-lactone clusters hints toward broad-spectrum antimicrobial and cytotoxic capabilities in multiple associates. It also corroborates the hypothesis that these marine metaorganisms are promising sources of novel bioactive compounds, representing targets for bioprospection (
38). Many strains also possess homoserine lactone-encoding BGCs indicative of sophisticated, cell-density-dependent chemical communication mechanisms. Antioxidant activities are likely conferred by the presence of aryl polyene BGCs in the genomes (
78,
101). These pigment type compounds, functionally related to carotenoids, characterized most of the proposed BMC strains. Furthermore, several coral-associated bacteria of different taxonomic origins are seemingly well equipped to handle osmotic stress as revealed by the occurrence of ectoine- and
N-acetylglutaminylglutamine amide (NAGGN)-encoding genes. Therefore, there is a need to continue the effort in culturing coral-associated bacteria to explore new biosynthetic potentials, both for bioprospecting purposes and for better understanding the chemical ecology of the metaorganism.
Identifying likely candidates for symbiosis is one challenge, but once the candidates are confirmed and characterized, the need to understand how the animal host establishes symbiosis and retains the relationship will also be critical. However, this is a two-way street. Current research in sponges has revealed that bacteria expressing the ankyrin genes avoid phagocytosis by sponge amoebocytes, thus becoming residents of the sponge microbiome by evading the host's immune system (
64,
70). Further, as ankyrin repeats are enriched in the microbial metagenomes of healthy corals (
10,
11), it is expected that commensal coral-associated bacteria also use this aspect of ankyrin genes to establish symbiosis. The evolutionary forces shaping the symbiosis are even trickier here, as bacteriophages encode for ankyrin biosynthesis in their genomes and might transfer this information across different community members (
70). As identified above with siderophore-encoding genes, similar patterns of symbiosis establishment and energy utilization may be adopted by both commensal and pathogenic bacteria.
To conclude, here we have highlighted that diverse coral-associated bacteria are already cultured, although these are often scattered across collections and rarely collated into one easily accessible location. Further, only a few of these have had their genomes sequenced. Despite the lack of genomes, we were able to identify a number of genetic features commonly encoded by these coral bacterial associates. These features include broad-spectrum antimicrobial, antioxidant, and cytotoxic compound production capabilities, high abundance of ankyrin repeat entries, tetratricopeptide, and WD40 repeats, and taurine degradation genes. That said, this can only be quantitatively assessed through comparison of metagenome profiles from corals versus other environments, such as sediments and seawater in a comprehensive fashion (several samples with replication, etc.). Such metagenome-based analyses should be complemented by (large-scale) marker gene surveys and/or visualization techniques to determine the nature and holobiont site of bacterial association, in particular since any metaorganism (configuration) is specific to a time and place and not static given the temporal (“fluidic”) nature of host-microbe interactions (
102). Even though the statistical power, with only part of the representative genomes available from cultures (as in this study), is limited, we exemplify here the importance of the cultured bacterial fraction of corals in hypothesis testing and applied microbiology.
We end by highlighting the importance and need for a global initiative to create an online catalog of genomic and physiological features of cultured coral-associated bacteria. Combining the use of these genomic insights with innovative culturing techniques (
37), aimed at improving the collection of coral-associated bacterial isolates, will see this field of coral biology move forward. Such an initiative should likely start with those microbes which have their complete genomes sequenced. This study pioneers the organization of such a global collection, as part of the efforts from the Beneficial Microbes for Marine Organisms network (BMMO), through a public invitation to researchers working in this field. As a result, we have here provided a list of cultured bacteria from corals that are currently available in public databases, plus isolates that were kept in collections from all the laboratories that responded to our invitation (
Table S1 and available now, open access via
http://isolates.reefgenomics.org). Now other researchers can access this virtual collection and/or contact specific laboratories for collaborations or solicitations of specific microbial strains.