Coral-Associated Bacterial Diversity Is Conserved across Two Deep-Sea Anthothela Species

Cold-water corals, similar to tropical corals, contain diverse and complex microbial assemblages. These bacteria provide essential biological functions within coral holobionts, facilitating increased nutrient utilization and production of antimicrobial compounds. To date, few cold-water octocoral species have been analyzed to explore the diversity and abundance of their microbial associates. For this study, 23 samples of the family Anthothelidae were collected from Norfolk (n = 12) and Baltimore Canyons (n = 11) from the western Atlantic in August 2012 and May 2013. Genetic testing found that these samples comprised two Anthothela species (Anthothela grandiflora and Anthothela sp.) and Alcyonium grandiflorum. DNA was extracted and sequenced with primers targeting the V4–V5 variable region of the 16S rRNA gene using 454 pyrosequencing with GS FLX Titanium chemistry. Results demonstrated that the coral host was the primary driver of bacterial community composition. Al. grandiflorum, dominated by Alteromonadales and Pirellulales had much higher species richness, and a distinct bacterial community compared to Anthothela samples. Anthothela species (A. grandiflora and Anthothela sp.) had very similar bacterial communities, dominated by Oceanospirillales and Spirochaetes. Additional analysis of core-conserved bacteria at 90% sample coverage revealed genus level conservation across Anthothela samples. This core included unclassified Oceanospirillales, Kiloniellales, Campylobacterales, and genus Spirochaeta. Members of this core were previously recognized for their functional capabilities in nitrogen cycling and suggest the possibility of a nearly complete nitrogen cycle within Anthothela species. Overall, many of the bacterial associates identified in this study have the potential to contribute to the acquisition and cycling of nutrients within the coral holobiont.


INTRODUCTION
Cold-water coral ecosystems contribute to vital biodiversity hotspots within the deep sea. Thriving in temperatures that range from 4 to 12 • C, these corals are globally distributed, occurring at depths between 50 and 4,000 m and inhabiting locations with strong currents and elevated topography (e.g., continental slopes and seamounts; Roberts and Hirshfield, 2004;White et al., 2005;Roberts et al., 2006). Similar to tropical reefs (Rohwer et al., 2002;Bourne and Munn, 2005), cold-water ecosystems provide a critical habitat for many organisms (Roberts et al., 2006) and are home to diverse and complex microbial assemblages (Penn et al., 2006;Neulinger et al., 2008Neulinger et al., , 2009Hansson et al., 2009;Kellogg et al., 2009;Galkiewicz et al., 2011;Gray et al., 2011;Bourne and Webster, 2013). Research addressing the microbial communities associated with coldwater corals has been limited due to the expense of sampling, which can be directly linked to the difficulty of sample retrieval at depth. While many of these corals have been identified since the 1800s, the first microbial study of cold-water corals was not published until 2006. This study assessed microbiota associated with dead and living samples of the scleractinian coral Lophelia pertusa (Yakimov et al., 2006). That same year, Penn et al. (2006) evaluated bacterial communities associated with a black coral and several bamboo corals in the Gulf of Alaska. These two studies were the first to describe the microbial communities associated with stony and soft cold-water coral species as well as demonstrate differentiation between these deepsea coral-associated communities and those of their surrounding environments (sediment and water column). Since then, studies have characterized the microbial diversity of additional coldwater corals: L. pertusa (Kellogg, 2008;Neulinger et al., 2008;Hansson et al., 2009;Kellogg et al., 2009;Schottner et al., 2009;Galkiewicz et al., 2011), Madrepora oculata (Hansson et al., 2009), and octocorals Paragorgia arborea, Plumarella superba, and Cryogorgia koolsae .
Previous studies have examined bacterial function within the coral host microbiomes. While some bacteria appear to play commensal or pathogenic roles (Nissimov et al., 2009;Shnit-Orland and Kushmaro, 2009;Bourne and Webster, 2013), many are not static in function, fluctuating with transitioning environmental conditions [e.g., increased microbial pathogenicity upon exposure to elevated thermal stressors (Bruno et al., 2007;Ben-Haim et al., 2003a)]. Because corals are dependent (in part) on their microbe-host interactions, examining the "core conserved" communities may reveal insights into the overall health of the coral host (Shade and Handelsman, 2012;Krediet et al., 2013). Many variables influence the bacterial presence within the coral holobiont including: specificity to host (genus; Littman et al., 2009, or species;Rohwer et al., 2002), association within a host niche (e.g., tissue vs. mucus; Bourne and Munn, 2005;Koren and Rosenberg, 2006;Sweet et al., 2010;Ainsworth et al., 2015), or fluctuations in the surrounding environment (Pantos et al., 2003;Reshef et al., 2006;Ainsworth and Hoegh-Guldberg, 2009). While bacterial communities may vary based on these parameters, conserved bacteria necessary for coral host health, defined as the "core" community, are consistently present. Because little is known about the bacterial functions within the cold-water coral holobiont, it is necessary to identify the core microbiota of each coral species (defined as those common in more than one sample; Shade and Handelsman, 2012).
Because cold-water corals are azooxanthellate (they do not acquire nutrients through photosynthesis), they obtain their nutrients from floating particulates and planktonic organisms and the microbial functional potential expressed within the coral holobiont (White et al., 2005;Zhang et al., 2015). Understanding the metabolic pathways and biochemical processing of nutrients such as nitrogen, carbon, and sulfur are essential in coral health and development . Based on phylogenetic inference and functional gene data, nitrogen-cycling microbes appear to be common members of tropical stony coral microbiomes (Wegley et al., 2007;Olson et al., 2009;Kimes et al., 2010;Lema et al., 2012Lema et al., , 2014Olson and Lesser, 2013;Pratte, 2013;Yang et al., 2013Yang et al., , 2015 and have been suggested to be species-specific (Wafar et al., 1990;Lema et al., 2012). The process of nitrogen fixation has been detected in a number of tropical stony corals (Williams et al., 1987;Shashar et al., 1994;Grover et al., 2014;Rädecker et al., 2014), as has nitrification (Wafar et al., 1990). Most nitrogen cycling studies have been conducted on stony corals, but there are a couple of recent studies that provide evidence of similar activities (nitrogen fixation, nitrite reduction, and ammonia oxidation) in soft corals (Yang et al., 2013;Bednarz et al., 2015). A recent study has proposed that nitrogen cycling might play a significant role in supplementing nutrition within the cold-water coral, L. pertusa (Middelburg et al., 2015). While it is evident that nitrogen availability is one of the driving factors in the proliferation and health of tropical coral hosts, little is known about its influence in the cold-water coral holobionts.
In an effort to further our understanding of cold-water octocorals and their microbial associates, this study evaluated three corals from the family Anthothelidae, initially targeting the species Anthothela grandiflora. Endemic to the Atlantic Ocean, A. grandiflora was first observed off the coast of Nova Scotia in the mid-1800s (Whiteaves, 1901), but to date no microbial analysis has been completed. For this study, samples from 23 individual colonies of gorgonian corals visually identified as Anthothela were collected from Baltimore and Norfolk canyons off the east coast of the United States in the Mid-Atlantic Bight. Overall, our objective was to provide the first characterization of the bacterial diversity associated with the cold-water octocoral genus Anthothela. Based on tropical coral research, we hypothesized that (1) bacterial communities would be conserved at the species level (Rohwer et al., 2002), (2) that different canyons might influence the bacterial community composition (Littman et al., 2009;Barott et al., 2011), and (3) that nitrogen-cycling microbes would be an important part of Anthothela spp. holobionts and constitute part of the core microbiome (Lema et al., 2012(Lema et al., , 2014Rädecker et al., 2015).

Sample Sites and Collections
In total, 23 individual Anthothelidae colonies were sampled during two research cruises conducted in August 2012 and May 2013. Site locations in the Mid-Atlantic Bight included Baltimore Canyon, which was sampled using the Kraken II remotely-operated vehicle (ROV; University of Connecticut) in 2012 and Norfolk Canyon sampled using the Jason II ROV (Woods Hole Oceanographic Institution) in 2013 (Figure 1). Environmental parameters were recorded for each sample site including location (latitude and longitude), depth, temperature, and salinity ( Table 1). Depths of sample collection ranged FIGURE 1 | Map of collection sites. Samples were collected from two submarine canyons, (A) Norfolk and (B) Baltimore, located off the Mid-Atlantic coast of the United States. Symbol shapes defined in the legend distinguish the coral samples based on genetic identification (e.g., Anthothela grandiflora, Anthothela sp., and Alcyonium grandiflorum).
from 401 to 704 m. Bottom types varied from rocky seafloor to rock cliffs and ledges. Common benthic fauna present at sample sites associated with coral colonies included adult galatheid squat lobsters (Eumunida picta) and cutthroat eels (Synaphobranchidae). Several scleractinians and octocorals were present in the vicinity of Anthothelidae colonies, including Desmophylum, Paragorgia, and Primnoa species.
Because some species of Anthothelidae can grow over other organisms, including dead coral branches and sponges, great care was taken to select sections of polyps from the tips of Highlighted samples (n = 16) were used in final analysis. * Samples listed as ND did not have genetic analysis completed.
Frontiers in Microbiology | www.frontiersin.org branches rather than the main stalk (Lumsden et al., 2007). This technique was employed to avoid accidental contamination of the sampled coral microbiome with that of the supporting organism. Branches were removed using the ROV's manipulator claw and each sample placed in an individual polyvinyl chloride (PVC) quiver. The quivers were cleaned before deployment using ethanol to remove any interior biofilms, filled with freshwater and sealed with a rubber stopper. This prevented contamination of the containers by the water column prior to sampling, during which the rubber stopper is removed and in situ seawater would replace the freshwater due to density differences. On the ship, coral samples were transferred from the ROV collection quivers to sterile 50 mL tubes containing the preservative RNAlater (Life Technologies, Grand Island, NY, USA). Samples were incubated overnight at 4 • C to allow the preservative to penetrate the coral tissues and then stored at −20 • C until processing.

Coral Genetic Identification
Of the 23 samples collected for microbiology, sufficient biomass remained in 19 samples for genetic analysis of the octocoral host ( Table 1). Octocorals were identified to genus initially using a key based on morphological characters (Bayer, 1981) and subsequently using a genetic barcode commonly applied to the Octocorallia: the mitochondrial mismatch repair gene homolog, mtMutS in combination with cox1 (McFadden et al., 2011). Total genomic DNA was extracted from octocoral tissue (whole polyps and adjacent coenenchyme) using a CTAB buffer and a single chloroform-only extraction method (Berntson et al., 2001;France, 2007), and quantified using the NanoDrop Lite Spectrophotometer (Thermo Scientific; USA). PCR amplifications were conducted using GoTaq polymerase (Promega) and included 5X GoTaq R Green buffer, 2 mM MgCl 2 , 0.4 µM dNTPs mix, 0.24 µM of each primer, 5 µg of bovine serum albumin (BSA, Sigma) and 1 U of GoTaq R polymerase, with the following cycle profiles: for mtMutS [amplicon 825 base pair (bp) length] 95 • C for 2 min, 35 cycles of 95 • C for 20 s, 50 • C for 30 s, 72 • C for 50 s and a final step of 72 • C for 5 min; for cox1 (amplicon 990 bp length) 94 • C for 3 min, 30 cycles of 94 • C for 20 s, 46 • C for 30 s, 72 • C for 60 s and a final extension step of 72 • C for 2 min.
All PCR amplified products were sent to Beckman Coulter Genomics (Beckman Coulter, Inc., Sykesville, MD, USA) for Sanger sequencing. Sequence traces were edited at the University of Louisiana, Lafayette using Sequencher version 4.6 (Gene Codes Corporation; USA) and then each sequence was queried against the Genbank nucleotide database using the BLASTn algorithm (NCBI, Altschul et al., 1990), and compared to unpublished data of colleagues working on similar taxa. Sequences were aligned to those from other octocoral taxa being worked on in the France lab or acquired from GenBank using Clustal W (Thompson et al., 1994) in the program BioEdit (Hall, 1998), and indels checked against the translated amino acid sequences to maintain reading frame. Aligned sequence datasets were then analyzed using the maximum likelihood method in RaxML 7.2.8 through the CIPRES web-portal (Stamatakis et al., 2008).

Nucleic Acid Extraction
Two polyps (∼50 mg) were removed from each coral sample using flame-sterilized forceps and dissecting shears. DNA was extracted using the MOBIO PowerPlant DNA Isolation Kit (MO BIO Laboratories; Carlsbad, CA, USA). Per Sunagawa et al. (2010), modifications to this protocol included the addition of lysozyme and extended incubation periods at room temperature 24 and 65 • C. Samples were then homogenized using 400 mg each of sterile 0.1 and 0.5 mm zirconia/silica beads (BioSpec Products; Bartlesville, OK, USA) in a Mini-BeadBeater-1 (Biospec Products; Sunagawa et al., 2010). The bacterial and universal primers 63F (5 CAGGCCTAACACATGCAAGTC3 ; IDT; Iowa City, IA, USA; Marchesi et al., 1998) and 1542R (5 AAGGAGGTGATCCAGCCGCA3 ; IDT; Pantos et al., 2003) were used to screen the samples to confirm amplification of the target 16S bacterial rRNA genes, rather than the possible amplification of coral 18S ribosomal rRNA genes by polymerase chain reaction (Galkiewicz and Kellogg, 2008). DNA concentrations from the extraction were quantified for each sample using a Quant-iT TM PicoGreen dsDNA Assay Kit (Invitrogen: Eugene, OR, USA) as outlined in the manufacturer's protocol and sent for sequencing.

Bioinformatics and Statistical Analysis
Analysis of the sequence data was conducted using the bioinformatics packages QIIME 1.5.0 on the Data Intensive Academic Grid (DIAG), a National Science Foundation funded MRI-R2 project #DBI-0959894, and QIIME 1.9.1 on the Amazon Elastic Compute Cloud (Amazon "EC2"; Caporaso et al., 2010). Our bioinformatics workflow and all resulting processed files are available online as a USGS data release (Kellogg and Lawler, 2015).
A total of 1,308,658 raw reads were generated from the 23 individual coral samples. Quality checks were performed using the split_libraries.py with the following parameters: sequence length (minimum sequence length of 200 bp and a maximum length of 700 bp), a minimum average quality score of 25, a maximum of one primer mismatch, and a maximum of a six homopolymer run (Kunin et al., 2010). SFF files were split into individual sample libraries based on the designated 10 bp identification barcode assigned to each sample during library preparation. The 889,914 sequences that passed the quality checks were then denoised using denoiser_preprocess.py, denoiser.py, and inflate_denoiser_results.py. This process was employed to reduce the number of erroneous operational taxonomic units (OTUs) and increase the accuracy of the sequence processing (Quince et al., 2011). Samples containing fewer than 10,000 sequences were removed prior to OTU selection to maximize the sequence data available. Furthermore, corals with no confirmed genetic identification were also removed at this stage, leaving a final total of 16 samples ( Table 1, highlighted). Moving forward, OTUs were selected using an open-reference method (pick_open_reference_otus.py), with a 97% similarity threshold (Rideout et al., 2014). This method clustered sequences from each sample against the Greengenes reference database release 13_8 (DeSantis et al., 2006). Sequences that were not matched during the reference comparison were reevaluated using the de novo reference method. Sequences were then aligned using usearch (Edgar, 2010), which included the removal of chimeras. Representative OTU sequences (defined as one representative from each OTU) were selected, assigned a taxonomic classification (uclust; Edgar, 2010), and used to create a phylogenetic tree (Price et al., 2010). Sequences were then filtered to remove absolute singletons (defined as an OTU present only once in the analysis). Sequences classified as chloroplasts and mitochondria were removed from the OTU table as were any sequences classified as Eukarya or Archaea. Samples were then rarified to the number of sequences present in the smallest sample (10,333) before further diversity analysis was completed. Analysis of the core diversity associated with the coral species was completed using compute_core_diversity.py.
Alpha and beta diversity calculations as well as relative abundance summaries were conducted using alpha_diversity.py, beta_diversity.py, and summarize_taxa_through_plots.py. Alpha diversity metrics included Chao index (Chao, 1984), Shannon diversity index (Shannon, 1948), and Simpson diversity index (Simpson, 1949) (Table 2). These indices were employed to assess the richness and evenness of the associated microbiota within each individual sample. To assess beta diversity (similarities or differences across samples), three matrices were used based on phylogenetic and taxonomic relationships between sequences. Weighted and unweighted unit fraction (UniFrac; Lozupone and Knight, 2005) measurements were recorded to evaluate the importance of the presence/absence of specific taxa within the samples (unweighted Unifrac) compared to the abundance of these taxa (weighted Unifrac; Fukuyama et al., 2012). Bray-Curtis was also assessed to evaluate differences between each sample based on the number of sequences per OTUs. To visualize beta diversity, principal coordinate analysis (PCoA) plots were prepared in R-Studio (R Development Core Team, 2014) using the previously described metrics. In addition, pairwise analysis of similarities (ANOSIM) was performed to further examine the statistical variation between sample groups (e.g., environmental parameters, location, or species diversity; Chapman and Underwood, 1999). A similarity percentage (SIMPER) was also used to determine the key contributing families responsible for the observed patterns. This statistical analysis was completed using PRIMER-E Ltd (Clarke and Warwick, 2001) and R-Studio. Figures for this study were produced in R-Studio (R Development Core Team, 2014) using the vegan (Oksanen et al., 2015) and gplots packages (Warnes et al., 2015).

RESULTS
The octocoral hosts were initially identified as A. grandiflora based on morphology (to genus) and distribution (to species). Comparisons to the GenBank nucleotide database and unpublished data suggested these represented three different species: A. grandiflora (the most commonly collected Anthothelidae in both canyons and 12 of the 23 samples analyzed in this study), Anthothela sp. (n = 5), and "Alcyonium" grandiflorum (the genus name is shown here in quotes to denote an ongoing systematic revision which will see Alcyonium grandiflorum transferred to a new genus in the family Anthothelidae; pers. comm., K. Moore, Commonwealth Science and Industrial Research Organisation, Australia; n = 2; Table 1). Mitochondrial DNA sequences revealed three haplotypes using both mtMutS (GenBank accession numbers: A. grandiflora, KU712085; Anthothela sp. KU712082; Al. grandiflorum, KU712083 and KU712084) and cox1 (GenBank accession numbers: A. grandiflora, KU712079; Anthothela sp. KU712080; Al. grandiflorum, KU712081). The Al. grandiflorum sequences are highly divergent from both Anthothela species (from A. grandiflora uncorrected p distance = 5% at mtMutS and p = 2% at cox1), but Anthothela sp. shows low sequence divergence from A. grandiflora (p distance < 0.2% at mtMutS and at cox1). Such a low level of interspecific divergence within a genus is not unusual for mitochondrial genes of octocorals (Pante and France, 2010;McFadden et al., 2011;Pante et al., 2015). Four of the 23 samples were not analyzed for coral genetics and were removed from our analysis (NF.01Q7, NF.02Q7, NF.16Q6, and RB.687Q4) although their bacterial sequence data are included in the SRA data file for completeness.
In addition, samples with fewer than 10,000 sequence reads (NF.02Q7, RB.686Q2, RB.687Q3, and RB.688Q2) were removed before primary analysis to increase rarefaction depth. As such, 16 samples were analyzed further, nine from Baltimore Canyon (one Anthothela sp. and eight A. grandiflora samples) and seven from Norfolk Canyon (one Al. grandiflorum, two as Anthothela sp., and four A. grandiflora; Table 1, Figure 1).

Alpha and Beta Measurements among Bacterial Diversity
Phylogenetic relationships samples (beta diversity) were compared using three primary diversity matrices (weighted UniFrac, unweighted UniFrac, and Bray-Curtis) and visualized using principal coordinates analysis (PCoA). Diversity of samples was first evaluated based on coral host. Here, samples within the genus Anthothela (A. grandiflora and Anthothela sp.) clustered separately from Al. grandiflorum, accounting for ∼57% of the statistical variation (Figure 2). Due to the small sample size (n = 1), Al. grandiflorum could not be included in the analysis of similarities (ANOSIM) to assess the correlation between bacterial diversity and coral species. Samples associated with Anthothela sp. and A. grandiflora were compared, revealing no significant difference between the two species (ANOSIM: R = 0.03, p = 0.27). Sample site was also assessed, indicating no significant correlation between coral-associated bacterial diversity and the canyon of origin (ANOSIM: R = −0.02, p = 0.45). The influence of depth of sample sites, water temperature, and salinity were also examined by PCoA, but were not correlated with bacterial diversity (results not shown). Similarity percentage analysis (SIMPER) was then used to examine representative bacterial taxa (family level) responsible for the differentiation between Anthothela samples and Al. grandiflorum. Overall, Anthothela bacterial communities had an average similarity of 63%. Considerable dissimilarity (∼72%) was observed between Al. grandiflorum and Anthothela microbiomes. Contributing families included Shewanellaceae (∼19%) and Pirellulaceae (∼11%) which were only present in Al. grandiflorum. Unclassified Oceanospirillales (16%), unclassified Spirochaetales (12%), Spirochaetaceae (10%), and Colwelliaceae (10%) also defined the differences seen between Anthothela samples and Al. grandiflorum.
To measure the bacterial diversity present within each individual sample, a series of alpha diversity indices were used ( Table 2). Shannon (Shannon, 1948), Simpson (Simpson, 1949), and Chao 1 (Chao, 1984) diversity indices account for evenness (defined as the abundance of species present) and richness (defined as the number of species or OTUs) as well as the total number of species observed. These measurements (Shannon = 5.54, Simpson = 0.87, and Chao 1 = 457.44) revealed greater species richness and evenness in the Al. grandiflorum sample compared to the rest of the Anthothela samples ( Table 2). In general, Shannon measurements were fairly consistent across Anthothela samples (average Shannon = 2.27) with increased diversity in two samples (RB.688Q4 = 2.68 and NF.18Q6 = 3.09). Similar trends were seen in Chao 1 and Simpson measurements. To visualize the diversity driving these patterns, bacterial communities were characterized at the phylum, order, and family levels for each coral host. Bacterial Community Composition Associated with Al. grandiflorum Proteobacteria dominated the Al. grandiflorum sample (RB.688Q3), accounting for ∼69% of the relative abundance with Planctomycetes representing the second most abundant at ∼17% (Figure 3A). Additional minor contributors to the bacterial diversity included Bacteroidetes (∼2%), Acidobacteria (∼1%), Actinobacteria (∼1%), Firmicutes (∼1%), and Verrucomicrobia (∼1%). Phyla representative of less than 1% of the relative abundance of the sample were labeled as "Other" (representing ∼5% of the sample's relative abundance); some of these included Chlamydiae, Deferribacteres, Lentisphaerae, and Nitrospira.

Bacterial Community Composition of Anthothela
In total, 15 coral samples were classified under the genus Anthothela, consisting of both A. grandiflora (n = 12) and an unknown Anthothela species (n = 3). At the phylum level, roughly half of the 15 samples were dominated by Proteobacteria (∼48% average relative abundance). The second most dominant bacterial group, Spirochaetes (Figure 3A), accounted for ∼42% average relative abundance of the Anthothela samples (∼43% in Anthothela sp. and ∼41% in A. grandiflora samples). Unlike the Al. grandiflorum sample, Proteobacteria and Spirochaetes were the only bacterial groups discernible at the phylum level. Phyla representative of less than 1% of the relative abundance in at least one sample were labeled as "Other" (∼10% of the total relative abundance). These included Chloroflexi, Lentisphaerae, and Nitrospira.
At the order level, all bacteria groups were classifiable with the exception of one, unclassified Spirochaetes ( Figure 3B). Bacterial diversity associated with Anthothela genus samples (A. grandiflora and Anthothela sp.) varied slightly with key communities including: Oceanospirillales (34%), unclassified Spirochaetes (24%), and Spirochaetales (17%). Other orders were observed at higher abundance in several of the Anthothela samples: Alteromonadales accounted for 26% of A. grandiflora sample NF.17Q7; Deltaproteobacteria Sva0853 were present in 6 of the 15 samples, ranging from 1 to 16% relative abundance. Kiloniellales were visibly represented in 10 samples, ranging from 1 to 34% relative abundance. Caulobacterales, with ∼36% relative abundance, were present in sample RB.686Q4. Bacterial groups observed in one or two of the Anthothela genus samples at lower relative abundance included: Rickettsiales present at ∼2% in three Norfolk Canyon samples (RB.686Q5, RB.687Q5, and RB.688Q5); Legionellales accounting for ∼2% of samples RB.688Q5 and NF.18Q7; and Vibrionales present at ∼2% of samples RB.687Q5 and NF.18Q6. Three of the bacterial groups present in Anthothela samples were not observed in Al. grandiflorum: Caulobacterales, Kiloniellales, and Rickettsiales.
Families and identifiable genera (present at greater than 1% relative abundance in at least one sample) were also assessed for Anthothela samples. Spirochaeta (order Spirochaetales, family Spirochaetaceae) were the most abundant bacteria (3-34% relative abundance) observed in 13 of the 15 samples. The family identified as Endozoicomonaceae accounted for ∼2% of two samples RB.687Q5 and RB.688Q1. Because we could not find this family defined in any taxonomic literature, sequences identified as Endozoicomonaceae were run through RDP Classifier (Wang et al., 2007) for further assessment. In Classifier, Endozoicomonaceae sequences were categorized as the family Hahellaceae. Lastly, Moritella sequences (order Alteromonadales, family Moritellaceae) were observed dominating A. grandiflora sample NF.17Q7, accounting for ∼26% relative abundance.

Core Microbiome
To evaluate the potential conserved-core diversity, samples were first assessed at the level of family Anthothelidae, i.e., across all 16 samples. One identifiable genus, Spirochaeta, was observed in every sample. Next, we evaluated the core diversity at the Anthothela genus level (n = 15). Assessment of Anthothela samples revealed no additional unique taxa, with Spirochaeta still the only shared taxon. From here, conserved bacteria were assessed at 90% sample coverage revealing four conserved communities: unclassified orders (Oceanospirillales, Kiloniellales, and Campylobacterales) and genus Spirochaeta (Figure 4). Individual species were examined at 100% sample coverage: A. grandiflora (n = 12) and Anthothela sp. (n = 3). A. grandiflora samples only shared the genus Spirochaeta, while Anthothela sp. included the genera Propionibacterium, Pseudoalteromonas as well as unclassified bacteria within Spirochaetes, Kiloniellales, Campylobacterales, Oceanospirillales, and Brachyspiraceae.

DISCUSSION
Relatively little is known about cold-water coral microbiomes in comparison to those of tropical corals. Prior to this study no microbial assessment had been completed on cold-water Anthothelidae corals. Because some tropical coral species have shown correlation between their bacterial composition and environmental parameters (e.g., geographic location, depth, ambient water-temperature, and surrounding organisms; Littman et al., 2009;Barott et al., 2011), similar relationships were anticipated within the host-microbe interactions of the cold-water corals collected during this study. However, bacterial composition of samples was not found to be significantly different based on canyon of origin (Figure 2). Beta diversity matrices indicated relationships between the taxa present and their abundance drove the diversity. Alpha diversity measurements also indicated higher bacterial diversity in the Al. grandiflorum sample ( Table 2). This as well as the ANOSIM results further supported the PCoA distribution showing the clustering of all Anthothela samples separate from Al. grandiflorum (RB.688Q3).
Biochemical properties of Colwelliaceae members include energy production through the decomposition of organic matter and nitrate reduction (Bowman, 2014). Species classified under the genus Shewanella (family Shewanellaceae) are generally Gram-negative, facultative anaerobes, recognized for their roles in nitrate and iron reduction (Hau and Gralnick, 2007;Coursolle and Gralnick, 2012;Kim et al., 2012;Satomi, 2014). Prevalent throughout the marine environment, members of this family have been isolated from numerous deep-sea and cold-water environments including the Marianas Trench  and the Arctic Ocean (Kim et al., 2012). Shewanella isolated from the mucus of tropical coral genus Favia demonstrated both antibacterial properties and antibiotic resistance (Shnit-Orland and Kushmaro, 2009;Shnit-Orland et al., 2010). These beneficial or mutualistic relationships are thought to contribute to the overall health of the host organism, providing protection from pathogenic or opportunistic bacteria.
Members of the order Rhodobacterales, specifically the genus Rhodobacter are thought to be among the most abundant, diverse, and metabolically influential bacteria within the marine environment (accounting for ∼25% of the total bacteria present in coastal and polar regions; Wagner-Dobler and Biebl, 2006). Rhodobacterales are found ubiquitously amongst coral reef systems, typically observed in surface waters, reef invertebrates, and their associated biofilms (Galkiewicz and Kellogg, 2008;Sunagawa et al., 2009;Sharp et al., 2012;Roder et al., 2014;La Rivière et al., 2015). Functionally these bacteria are diverse, capable of contributing to the reduction of trace metals and the production of antibiotic compounds (Brinkhoff et al., 2004). Other notable members also contribute to the global carbon and sulfur cycles through the oxidization of carbon monoxide and production of dimethylsulfide (Wagner-Dobler and Biebl, 2006). To date a few Rhodobacterales (particularly members of the family Rhodobacteraceae) have been identified in cold-water coral species: Al. digitatum (Alsmark et al., 2012) and L. pertusa (Neulinger et al., 2008;Kellogg et al., 2009). Desulfobacterales (class α-proteobacteria) are documented sulfate-reducing bacteria, functionally described as hydrogenotrophs (capable of using H 2 as an energy source within the metabolic pathway; Kuever et al., 2005;Kimes et al., 2010). These members appear to be common within deep-sea ecosystems, including microbial mats (Burow et al., 2014), sediments (Simister et al., 2015), seeps (Jaekel et al., 2013), and one unidentified cold-water coral (Simister et al., 2015). In contrast, Myxococcales (class α-proteobacteria) are commonly observed in terrestrial environments, with some present in marine ecosystems (though, notably less; Reichenbach and Dworkin, 1992). Several have been identified in coral species including Eunicella cavolini (Bayer et al., 2013a) and Mussismilia braziliensis (Garcia et al., 2013). Functionally these bacteria may contribute to the production of nutrients through sulfate reduction and the decomposition of organic matter (Reichenbach, 1999;Baker et al., 2015).

Proteobacteria in Anthothela Samples
Oceanospirillales was identified as one of the dominant bacterial groups in Anthothela samples, present in eight samples at over 20% relative abundance. Functional characteristics that have been associated with Oceanospirillales members include carbon fixation, sulfur oxidation, and biofilm production in the presence of trace metals, such as copper (Little et al., 1996;Swan et al., 2011;Speck and Donachie, 2012). These roles define both the acquisition of nutrients and potential attraction or inhibition of bacterial colonization within the host. In this study, Oceanospirillales were further classified to the family level, with identifiable groups consisting of unclassified Oceanospirillales and Endozoicomonaceae. At present time, it is unclear if Endozoicomonaceae is an accepted taxon, since those same sequences were identified by RDP Classifier as Hahellaceae with 99% confidence. With that in mind, members of the family Hahellaceae (specifically genus Endozoicomonas) are widespread in the marine environment. Numerous studies have assessed Endozoicomonas in tropical and temperate corals (Sunagawa et al., 2009(Sunagawa et al., , 2010Yang et al., 2010;Cardenas et al., 2012;Lee et al., 2012;Morrow et al., 2012;Apprill et al., 2013;Bayer et al., 2013a,b;Carlos et al., 2013;Correa et al., 2013;Jessen et al., 2013;La Rivière et al., 2013;Pike et al., 2013;Ransome et al., 2014;Roder et al., 2015), as well as sea slug (Kurahashi and Yokota, 2007) and sponge species (Nishijima et al., 2013;Rua et al., 2014). Because Endozoicomonas are both common and highly abundant in healthy tropical corals species, relationships between these bacteria and their hosts have been thoroughly examined. Functional characteristics include nitrate reduction, chemotactic activity, and production of antimicrobial compounds (Kurahashi and Yokota, 2007;Rua et al., 2014;Tout et al., 2015). While members of the family Hahellaceae are common in tropical and temperate environments, they do not appear to be common in deep-sea corals. Few studies have observed bacteria classified under Hahellaceae in deep (>100 m) cold-water corals (Hansson et al., 2009;Kellogg et al., 2009;Van Bleijswijk et al., 2015). The symbiotic relationships between Hahellaceae and zooxanthellae are thought to be one of the driving influences of their abundance and presence in tropical corals (Pantos et al., 2015). The lack of algal symbionts in deepsea corals may be why there are few Hahellaceae bacteria, if present at all, in these holobionts. In this study, members of the Hahellaceae (originally classified as Endozoicomonaceae) represent a small minority of the Oceanospirillales present (∼2% in A. grandiflora samples RB.687Q5 and RB.688Q1).
Similar to Oceanospirillales, members of the order Kiloniellales were present in multiple Anthothela samples at relatively high abundance. Kiloniellales bacteria have been observed in several tropical coral species (Sharp et al., 2012;Soffer et al., 2015) as well as mussels (Cleary et al., 2015), sponges (Cleary et al., 2013), and algae (Wiese et al., 2009). Soffer et al. (2015, identified members of the order Kiloniellales at higher abundances in healthy coral Orbicella annularis than in diseased colonies, suggesting association in a beneficial capacity. Functionally, these chemoheterotrophic bacteria have been found to utilize nitrates within the metabolic process through denitrification (Wiese et al., 2009;Imhoff and Wiese, 2014).
Several contributing Proteobacteria were identified in individual Anthothela samples including Alteromondales, Caulobacterales, and Rickettsiales. Genus Moritella (order Alteromondales, family Moritellaceae) was present in A. grandiflora sample NF.17Q7 accounting for ∼26% relative abundance. This bacterial group is specific to marine environments and generally classified as halophilic facultative anaerobes (Stanley et al., 2005;Urakawa, 2014). Moritella isolates have been collected from a wide variety of environments ranging from deep-sea sediments Nogi et al., 1998;Xu, 2003) to tropical corals (Rohwer et al., 2001;Bourne, 2005;Bourne and Munn, 2005). In cold-water coral species, Moritella sequences have been described in the scleractinian L. pertusa collected from the Gulf of Mexico (Kellogg, 2008). In the individual A. grandiflora sample RB.686Q4, members of the Caulobacterales were found at relatively high abundance (∼30%).
Recognized for their unique morphology, members of this order contain a stalk-like flagellum utilized for adhesion to adjacent surfaces, including but not limited to host organisms (Starr and Skerman, 1965). While these bacteria tend to exhibit parasitic tendencies, they have been described as facultative commensals, potentially contributing to the acquisition of nutrients through their roles in carbon cycling (Abraham et al., 1999). These free-living bacterial communities are often found throughout the water column and have been observed in several tropical corals including the gorgonian Pseudopterogorgia elisabethae (Correa et al., 2013) and acroporid species (A. granulosa, A. valida, and A. millepora;Littman et al., 2009;Ainsworth et al., 2015). While members of the group Caulobacterales have been identified in deep ocean waters (Eloe et al., 2011), no prior studies have observed Caulobacterales associated with cold-water corals.
Members of the order Rickettsiales have previously been described as opportunistic, pathogenic, and/or associated with diseased tropical corals (Peters et al., 1983;Casas et al., 2004;Garcia et al., 2013;Miller et al., 2014;Peters, 2014). While they are common in tropical corals, Rickettsiales have currently been identified in one cold-water octocoral species C. koolsae . In the present study, Rickettsiales were observed at low relative abundance (∼2%) in only three Norfolk Canyon samples (RB.686Q5, RB.687Q5, and RB.688Q5). Because their functional characteristics are generally driven by ambient environmental fluctuations (e.g., increases in temperature and light intensity; Peters, 2014;Gignoux-Wolfsohn and Vollmer, 2015), we speculate that the opportunistic tendencies exhibited in tropical environments differ from those in cold-water environments. In this case, additional research is necessary to definitively assess the potential pathogenicity and overall functionality of these bacteria within the cold-water ecosystem.

Spirochaetes
Spirochaetes were observed as one of the primary bacteria, dominating over half of the Anthothela samples. Spirochaetes are recognized as motile free-living, facultative/obligate anaerobes (Leschine et al., 2006). Functional characteristics displayed by members of this phylum include nitrogen and carbon fixation, as well as chemotactic responses to chemical stimulants (Greenberg and Canale-Parol, 1977;Kimes et al., 2010;Baker et al., 2015;Lilburn et al., 2015). Spirochaetes are commonly found in association with invertebrates at high abundance, including species of termites (Breznak, 2002), oligochaetes (Blazejak et al., 2005), sponges (Taylor et al., 2005), and tropical corals (Casas et al., 2004;Kimes et al., 2010Kimes et al., , 2013Closek et al., 2014). Previous studies using clone libraries have observed Spirochaetes in association with some cold-water corals (Penn et al., 2006;Kellogg et al., 2009;Gray et al., 2011), however, they have never been identified as a dominant member of the associated bacterial community. In this study, Spirochaetes were recognized as the prevailing phyla in roughly half of the total Anthothela samples (n = 15).
In Anthothela samples (A. grandiflora and Anthothela sp.), genus Spirochaeta (phylum Spirochaetes) continued to dominate the bacterial groups. New representative Spirochaeta sequences from this study were compared to those of environmental and invertebrate studies where Spirochaeta sequences had been observed. Sequences associated with Anthothela samples were most closely related to sequences isolated from deep-sea water (Accession KF758585, E-value of 7e −165 ) and microbial mats (Accession DQ218325, E-value of 1e −161 ). While sequences from Anthothela corals were not closely related to those of other coral species, presence of this bacterium across all 16 samples suggests conservation at the family level. Additionally, presence of this bacterium at such a high abundance, as observed in Anthothela samples, suggests a unique microbe-host interaction specific to that coral genus.

Core Microbiome
In this study we applied a stringent approach to examine "core" conserved communities, evaluating bacterial groups present at the family (Anthothelidae), Anthothela genus, and individual species (A. grandiflora and Anthothela sp.) levels. To begin, samples were evaluated at the family and genus levels at 100% sample coverage, revealing one core-conserved bacterium present across all samples, a member of the genus Spirochaeta. As previously described, members of the phylum Spirochaetes are common throughout coral species, both tropical and coldwater (Casas et al., 2004;Penn et al., 2006;Kellogg et al., 2009;Gray et al., 2011;Kimes et al., 2013;Closek et al., 2014). These free-living nonpathogenic anaerobes are known to be capable of carbon fixation and organic carbon degradation (Baker et al., 2015). In this study, Spirochaeta was found to be one of the dominant bacterial groups, accounting for roughly 16% of the total relative abundance, thus suggesting a potentially important role within the coral microbiome.
Samples were then assessed at 90% sample coverage, particularly looking for conserved bacteria specific to Anthothela samples (Figure 4). In addition to the genus Spirochaeta, OTUs classified under the phylum Proteobacteria (orders; Oceanospirillales, Kiloniellales, and Campylobacterales) were identified. Previous studies described members of the orders Oceanospirillales and Kiloniellales as beneficial bacteria, contributing to their host system through the production of biofilm and antibacterial properties, respectively (Little et al., 1996;Swan et al., 2011). Oceanospirillales members have also been recognized for their influence in nutrient dynamics within the coral holobiont through the formation of dissolved inorganic materials produced during carbon fixation and sulfur oxidation (Swan et al., 2011). In contrast to Oceanospirillales and Kiloniellales, members of the order Campylobacterales are most commonly known for their association with coral disease (Frias-Lopez et al., 2002;Sunagawa et al., 2009;Sweet and Bythell, 2012;Vezzulli et al., 2013;Gignoux-Wolfsohn and Vollmer, 2015), but are also present in healthy corals (Sharp et al., 2012). Functionally these bacteria are well-known for their metabolic influence in nitrogen cycling and communication through bacterial quorum sensing (Kern and Simon, 2009;Golz et al., 2012).
Microbial functional genes capable of completing, or nearly completing, the nitrogen cycle have been identified in Pacific (Wegley et al., 2007) and Atlantic (Kimes et al., 2010) tropical corals. A complete nitrogen cycle has been shown FIGURE 5 | Core bacterial groups potential roles within the nitrogen cycle. Each of the bacterial groups present within the core microbiome of Anthothela samples was previously recognized for their roles within the nitrogen cycle. This diagram illustrates a simplified overview of the bacterial groups with their possible functions. This figure was adapted from one presented in Wegley et al. (2007). via isotope incorporation in the cold-water coral L. pertusa, including ammonium production and assimilation, nitrification, denitrification, and nitrogen fixation (Middelburg et al., 2015), however, these roles have not been definitively associated to specific microbial symbionts of the coral yet. Based on previous descriptions of nitrogen-cycling abilities in the bacterial taxa that we observed in the conserved core of Anthothela spp., we have inferred a nearly complete nitrogen cycle (Figure 5). Members of the genus Spirochaeta, identified as the most conserved bacterium, were previously recognized for their roles in nitrogen fixation . During nitrogen fixation, nitrogen gas (N 2 ) is converted to readily available organic compounds that may then be taken in (Zehr and Kudela, 2011). Nitrogen fixation contributes up to 1/3 of the nitrogen requirement in L. pertusa (Middelburg et al., 2015), supplementing organic nitrogen obtained by capture feeding (Duineveld et al., 2012;Mueller et al., 2014). Several members of the order Campylobacterales were documented for their contributions through nitrate ammonification (Tiedje, 1988). This is the process by which nitrate is converted to ammonium, thereby recycling nitrogen back into the system (Tiedje, 1988;Simon, 2002;Rädecker et al., 2015). In the order Oceanospirillales, members have been described for their contribution as nitrate reducers, reducing nitrate to nitrite (Zehr and Kudela, 2011). Lastly, members of the order Kiloniellales, classified as chemoheterotrophic aerobic bacteria, have shown potential in the processing of molecular nitrogen through denitrification (Imhoff and Wiese, 2014). In this process, nitrates are reduced back into N 2 (dinitrogen) to be utilized by nitrogenfixing bacteria. While likely important for the functioning of the Anthothela coral holobiont, the rate of nitrogen fixation observed in L. pertusa was small compared to the rates occurring in the open ocean, suggesting that cold-water corals do not significantly impact nitrogen cycling in the deep sea (Middelburg et al., 2015). This is in line with research that ascribed a similarly minor role in the nitrogen budget of coral reef ecosystems to tropical corals .
Although the core microbiome of the Al. grandiflorum (n = 1) could not be evaluated, its bacterial community was assessed for possible functional characteristics associated with the biochemical processing of nitrogen. Similar to Anthothela samples, core members associated with Al. grandiflorum were found to belong to taxa previously documented for their metabolic abilities in nitrogen cycling. In addition to the potentially nitrogen-fixing Spirochaeta also seen in Anthothela spp., bacteria classified under families Shewanellaceae and Colwelliaceae (order Alteromonadales) have been acknowledged for their roles as nitrate reducers (Satomi, 2014;Lilburn et al., 2015). Classified as anaerobic ammonia-oxidizing bacteria, Pirellulales (phylum Planctomycetes) were thought to contribute through the removal of metabolic waste within the host microbiome (Mohamed et al., 2010). One of the end products of nitrate ammonification, ammonia, may be taken in by ammonia-oxidizing bacteria such as Pirellulales, resulting in the oxidization of ammonium and formation of nitrites (Zehr and Kudela, 2011). While members of the order Pirellulales and families Shewanellaceae and Colwelliaceae only show potential metabolic properties for a portion of the nitrogen cycle, other bacterial members may be present within Al. grandiflorum that complete the remaining functions. Further research using targeted functional genes (e.g., nifH, amoA), metagenomics, and transcriptomics are required to link these microbial roles definitively to specific bacterial community members.

CONCLUSION
Our study provides insight into the previously uncharacterized microbiome of cold-water octocorals classified under the family Anthothelidae. The two Anthothela species (A. grandiflora and Anthothela sp.) shared similar bacterial communities, in contrast to Al. grandiflorum which had a highly diverse microbiome distinct from the rest. The bacterial communities become more different as the coral taxa are more distantly related. Additionally, the bacterial composition revealed no differentiation due to geographic location (canyon of origin). Evaluations of the core microbiome at 90% revealed a conserved bacterial community associated with the Anthothela genus. Many of the core bacteria share potential metabolic functions associated with nutrient provision and properties aiding in the protection of the coral host. Overall health and proliferation of cold-water corals are dependent on capture feeding and the presence of microorganisms (Duineveld et al., 2004;Roberts et al., 2006). Members of the core microbiome of Anthothela samples were recognized for their potential roles in the uptake and remineralization of organic and inorganic material. More specifically, many of the bacterial groups present were previously documented for their roles in nitrogen cycling, including nitrogen fixation (Spirochaeta), nitrate ammonification (Campylobacterales), nitrate reduction (Oceanospirillales), and denitrification (Kiloniellales). Additionally, Al. grandiflorum also contained a microbial community potentially capable of various functions specific to nitrogen cycling: nitrogen fixation, nitrate reduction, and ammonia oxidation. Unfortunately, with a sample size of n = 1, this evaluation is only a partial representation of the overall bacterial and functional diversity present within this species. Further research is necessary to investigate the microbial-host interactions, specifically the functionality of these bacterial associates and their role within the cold-water coral holobionts.

AUTHOR CONTRIBUTIONS
SL performed the DNA extractions, analysis of results, and writing of the manuscript. CK planned the experimental design and participated in the manuscript writing/editing and data analysis. SF and RC completed coral genetics and provided detailed methods and results for this analysis. SR and SB planned the research cruises, acquired samples and multibeam data, and edited the manuscript.

FUNDING
Funding for this project was provided by the U.S. Geological Survey's Ecosystems Mission Area, Environments Program through the Outer Continental Shelf study on Mid-Atlantic Canyons. Funding was sponsored by the National Oceanographic Partnership Program and supplied by the Bureau of Ocean Energy Management (BOEM) contract number M10PC00100 (contracted to CSA Ocean Sciences, Inc.). Special thanks to G. Boland and R. Green (BOEM) for their coordination efforts, as well S. Viada (CSA) for support during the overall project. Additional funding for SL was provided by the George Lorton Endowed Fellowship through the University of South Florida, College of Marine Science as well as the Aylesworth Scholarship.

ACKNOWLEDGMENTS
The authors extend appreciation to A. Shade (Michigan State University) and the Explorations in Data Analyses for Metagenomic Advances in Microbial Ecology (EDAMAME) Workshop for critical lessons in workflow organization and bioinformatic tools. Additional thanks are extended to the captains and the crews of the NOAA ship Nancy Foster and the Kraken II ROV team as well as the NOAA ship Ronald H. Brown and Jason II ROV team. We thank K. Moore (Commonwealth Science and Industrial Research Organisation, Australia) for her insights on Anthothela taxonomy and for allowing us access to her pre-published data. Special thanks are extended to M.A. Gray (USGS) for his assistance in sample collection and advice during data analysis. M.P. Rhode (UNC-Wilmington) produced the multibeam maps of Baltimore and Norfolk canyons (Figure 1) and assisted at sea. B. Boynton (USGS) prepared final versions of the figures and M. Breitbart (USF) reviewed the manuscript. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.