INTRODUCTION
Corals harbor complex microbiomes that help sustain high rates of productivity and biomass in oligotrophic reef waters. The coral microbiome is composed of a diverse assemblage of microorganisms, including algae, other protists, bacteria, archaea, fungi, and viruses, and this consortium is collectively referred to as the holobiont (
1–3). Most attention has been dedicated towards studying the dynamics between endosymbiotic algae (generally
Symbiodinium) and corals because the photosynthate provided by these algae is fundamental for the metabolism, calcification, and overall growth of stony corals (
4,
5). In contrast, much less is known about the specific metabolic interactions between bacteria, archaea, and corals. For example, there is some evidence that these cells are capable of transforming and contributing to the cycling of essential and limited nutrients (
6–8), as well as producing antibiotics or other secondary metabolites required by the coral host for protection (
9,
10).
One of the key obstacles to understanding the functional contributions of prokaryotes to corals is the sheer diversity of microbes found in association with corals. In fact, sequencing-based studies have repeatedly described the taxonomic complexity of the coral microbiome (
11,
12). Studies have estimated that as many as 6,000 distinct small-subunit (SSU) rRNA gene ribotypes are associated with corals (
3,
11), spanning dozens of phyla and undescribed lineages (
12,
13). The high diversity and taxonomic complexity of coral-associated microbiomes provide considerable deterrents to identifying consistent microbial associates that might be biologically meaningful within the holobiont and possibly fulfill roles that are important to the health and functioning of corals. Recently, deep-sequencing studies of the coral microbiome have suggested several genera of bacteria that are indeed consistently or frequently detected with corals across their geographic distribution (
13,
14). Additionally, a modeling exercise applied to three coral microbiomes predicted that the consistent bacterial associates of corals are quite numerous, and even outnumber the more sporadic associates (
15).
In addition to utilizing deep sequencing to search for consistent microbial associates of corals, some of the complexity within the coral microbiome may be resolved if the coral colony is separated into discrete habitats (
16). Corals harbor microbial cells within their surface mucus layers as well as within their tissues and skeletons (
17,
18). In the past, the majority of coral microbial sequencing-based studies have either homogenized the entire coral (obtaining mucus, tissue, and skeletal material) (
11,
19) or airbrushed the specimen to separate the mucus and tissue from the skeleton (
3,
12). Both of these approaches result in the inclusion of microbes from all of the diverse coral habitats. Some efforts have been made to separate coral mucus, tissue, and skeleton. For example, several studies have utilized vacuum suction, syringes, and cotton swabs to collect mucus, so that only the mucus associates of corals are examined (
20,
21), although the syringe can introduce seawater microbes when used underwater (
20). While mucus separation is relatively straightforward, removing mucus and skeleton from the tissue in order to exclusively investigate tissue endosymbionts is more complicated (
20). Recently, a coral habitat differentiation approach was applied to corals; the coral was decalcified (dissolution of the skeleton), and the remaining intact tissue was used to describe endosymbionts (
13). This refinement in coral processing better positions investigators to address still outstanding questions about whether corals harbor consistent microbial associates within their tissues or endosymbionts and whether different microbially mediated functions occur in localized niches within the coral holobiont. Additionally, this approach also circumvents the common problems associated with visualizing microbial populations
in situ (
22).
The goal of this study was to test the hypothesis that tissue and mucus habitats of corals contain phylogenetically distinct microbial associates. Further, we hypothesized that if corals harbored specific mucus- or tissue-associated microbes that are important to coral functioning, they would be maintained as consistent associates over ecological reef gradients. To accomplish this, we separated the tissue and mucus habitats, as well as a holobiont fraction (containing biomass from both the tissue and mucus habitats as well as residual skeleton) from five common Caribbean corals that differ evolutionarily and ecologically across five distinct reef environments (
Fig. 1A). Specifically, we studied
Porites astreoides and
Porites porites within the long/complex evolutionary lineage of corals as well as
Montastrea cavernosa,
Orbicella faveolata, and
Diploria strigosa within the short/robust evolutionary lineage (
23).
P. astreoides is further differentiated from the other spawning corals because it uses a brooding reproductive strategy, and
P. porites is distinct because it grows with a branching morphology in comparison to the other mounding colonies included in this study. We then deeply sequenced partial SSU rRNA genes from the tissue, mucus, or holobiont bacteria and archaea to identify consistent members within each specific coral habitat. Our results reveal that corals do harbor distinct microbiomes that differ by coral habitat, including previously unrecognized microbes associated with coral mucus and tissues.
DISCUSSION
This study demonstrated that distinct tissue and mucus-associated microbes can be readily distinguished if the coral colony is separated into habitat fractions. This coral habitat differentiation approach led to the identification of previously unrecognized consistent microbial associates, including several specific mucus and tissue associates that have not been previously acknowledged in coral microbiome studies. One surprising outcome of this study is that the holobiont fractions of the
Porites corals contained a different assemblage of symbionts than the mucus and tissue fractions, which is an important consideration for studies using the holobiont approach to characterize coral microbiomes. We noted that skeletal slivers were consistently present within the holobiont biomass prior to DNA extraction, and these slivers were likely dislodged from the skeletal matrix during airbrushing of the samples. Further, the high recovery of
Curtobacterium and
Ralstonia sequences in these samples compared to the tissue and mucus fractions indicated that the airbrushing process recovered a reservoir of cells that were either not present or not detected in the tissue and mucus sample fractions. While there was consistency in the
P. astreoides holobiont sample recovering the same
Endozoicomonas OTUs also present in the tissue and mucus, it is possible that a deeper sequencing effort for the holobiont samples could have better demonstrated overlap with the tissue and mucus microbiomes of the other species. It should be noted that there were a few methodological inconsistencies in the treatment of the coral habitat samples that could have impacted the recovery of cells and have led to less than expected overlap between the tissue, mucus, and holobiont microbiomes. Due to the length of time necessary to decalcify tissue, tissue samples were preserved prior to decalcification, as conducted previously by a coral microbiome study (
13), which could have introduced preservation biases for some microbes. Decalcification was conducted with a weak acid which is recommended for other organisms for maintaining high DNA quality (
27), yet biases in the recovery of microbial community members are still possible. Additionally, an extended proteinase K digestion and added heat treatment were also applied to the decalcified tissues to aid in the retrieval of high-quality DNA (
28). The proteinase K treatment was different in the holobiont samples, and a head-to-head comparison of samples did not find that the differential treatments had a significant impact on the microbiome, but it is possible that the impact was subtler than we were able to detect. While it is possible that these collective differences did impart some biases on the results, the trends reported in this study are consistent with previous knowledge and expectations about where these microbial associates of corals might reside. For example,
Synechococcus, a common seawater bacterium, was found in the seawater and within the surface mucus layer of corals (
29).
One of the goals of this study was to provide descriptions of consistent microbial members of the coral holobiont that can then be targeted in functionally based investigations. Here we highlight and discuss the potential ecological or functional relevance of the consistent abundance-based taxa whose representation may be especially well suited for future studies. As such, “
Ca. Amoebophilus” bacteria were identified as a consistent abundance-based associate in the tissues of two species and were associated with the tissues of all Caribbean species examined. A previous study recovered highly related sequences from Caribbean corals (
11) (
Fig. 5), and the present study is the first to examine the phylogenetic placement of these sequences and confirm their position in a separate coral-specific monophyletic lineage most closely related to “
Ca. Amoebophilus.” “
Candidatus Amoebophilus asiaticus” is the first described species in this candidate genus and is an obligate intracellular symbiont of
Acanthamoeba, a freshwater amoeba that has the ability to vertically transmit symbionts across generations (
30). “
Ca. Amoebophilus” also forms a monophyletic group with symbionts of the tick
Ixodes scapularis and whitefly
Encarsia pergandiella (
30), and its genome has multiple eukaryotic domains, indicating mechanisms for a symbiotic lifestyle and host-cell interactions (
31). It is very possible that the coral-specific “
Ca. Amoebophilus” bacteria are also interacting with a protistan eukaryotic host, including
Symbiodinium spp., apicomplexans (
32,
33), or otherwise undescribed amoebae.
Members of the
Acidobacteria subgroup 10 TK85 lineage of
Holophagae were not previously recognized as tissue associates of tropical corals. Sequences belonging to the
Holophagae class have only otherwise been recovered from the skeleton and mucus of cold water corals (
34).
Acidobacteria are common associates of soil environments, but investigations into specific
Acidobacteria within the family
Holophagae have revealed this class to be ecologically diverse, including both marine isolates (
35) as well as plant symbionts (
36). Although they consistently associate within
M. cavernosa tissue, the specific role of the subgroup 10 lineage of
Holophagae may be difficult to decipher due to its relatively low sequence abundance in coral tissues.
Tumebacillus within the phylum
Firmicutes, emerged as a consistent mucus associate of the corals and was present in all species studied, which is surprising considering that these OTUs have not previously been identified in corals. Described members of this genus are spore-forming, associated with soils, Arctic permafrost, and decomposing algal scum, and are capable of utilizing a variety of carbon sources, including one strain that can oxidize sulfur to support growth (
37–40).
This is the first known report that identifies “
Ca. Actinomarina” as consistent members of a coral microbiome, and here they were found associated with
P. porites mucus as well as seawater. “
Ca. Actinomarina” bacteria are generally very small cells (volume of ~0.013 µm
3), and the genetic material has very low GC content (33%) (
34). In addition, “
Ca. Actinomarina” bacteria contain rhodopsin, suggesting that these cells rely on a photoheterotrophic lifestyle (
41). They are common inhabitants of surface seawater, residing at similar depths as picocyanobacteria (
41).
Sequences associated with the
Rhodobacteraceae family are commonly identified as members of the coral microbiome (reviewed in reference
42), including developing corals (
43,
44), and were found here to be widespread and abundant in tissue and mucus habitats. This family includes a metabolically and ecologically diverse group of organisms that frequently attach to phytoplankton surfaces and utilize exuded dissolved organic carbon (DOC) (
45).
Ruegeria, in particular, was identified as a consistent mucus associate of all species, and some members of this genus are able to assimilate dimethylsulfoniopropionate (DMSP) (
46), an abundant carbon source on corals (
47).
Ralstonia sequences were abundant in the holobiont fractions of both
Porites species corals, and their absence from the mucus and tissue fractions suggests that these cells may reside and proliferate within the coral skeleton. However, our finding differs from a recent study identifying
Ralstonia as symbionts of
Symbiodinium spp. within the tissues of Pacific corals (
13).
Ralstonia is a broad genus of symbiotic bacteria; phylotypes belonging to this genus are capable of denitrification (
48) and can be plant pathogens (
49) and could therefore serve diverse roles within corals.
This study also provided new evidence that several microbial symbionts reside in multiple coral habitats.
Endozoicomonas is recognized as a dominant member of the
P. astreoides microbiome (
21,
50), but to our knowledge, the present study is the first to identify
Endozoicomonas as both a mucus and tissue associate of any coral species, with the same OTU residing in both habitats of
P. astreoides. Cells have been localized within the epithelial tissue of
Stylophora pistillata tentacles (
14), and this habitat could facilitate transport or colonization of cells within the mucus.
Endozoicomonas genomes obtained from another coral species, a sponge, and a sea slug are large and include elements indicative of both a symbiotic and free-living stage (
51), thereby supporting a flexible lifestyle that may be able to switch between residing within tissue (endosymbiotic) and mucus (free-living) habitats. In about half of the
P. asteroides colonies examined,
Endozoicomonas was the dominant microbial member, and interestingly, these colonies were found only on two reefs. This observation may indicate reef-specific recruitment of
Endozoicomonas from parental colonies into brooded larvae or from other adult
P. astreoides on these reefs. Studies have associated the presence of
Endozoicomonas with healthy-appearing corals (
14,
21,
52–55), and these cells may play important roles in maintaining immunity or facilitating metabolic functioning of corals.
Endozoicomonas is clearly an important and globally ubiquitous symbiont, and the ecology behind its multihabitat residence within the coral, and why it was not a dominant tissue symbiont in all
P. astreoides colonies, requires further attention.
Cyanobacteria capable of fixing nitrogen are endosymbionts within some
M. cavernosa corals (
56), but here the non-nitrogen-fixing
Synechococcus cyanobacteria were identified only as consistent members of the seawater and the mucus and holobiont microbiomes rather than tissue. The abundance of
Synechococcus in the mucus microbiome was surprising, as they are typically associated with pelagic habitats, and photosynthesis in the coral is thought to be dominated by
Symbiodinium spp. However,
Synechococcus can be trapped in the coral mucus (
29). Additionally, a recent study found that
P. asteroides can graze on
Synechococcus cells (
57), and it is possible that entrapment of cells within the mucus may play a role in this process, further explaining their prevalence in the mucus microbiome of corals.
Conclusions.
The coral microbiome is a complex association of microorganisms, and elucidating specific ecological interactions between corals and their prokaryotic symbionts provide considerable challenges for investigators. The results presented here suggest that some prokaryotes are found only within specific coral habitats. Genomic, microscopic, or isotopic function-based investigations focused on these habitats may be able to resolve the dynamics and activities of coral microbial associates, as well as discover whether multiresidence symbionts like Endozoicomonas have distinct roles within the different habitats of a colony. These and other similar function-based investigations will provide considerable insight into the roles prokaryotes play in maintaining or disrupting the health of the coral holobiont.