Culture-Independent Characterization of Bacterial Communities Associated with the Cold-Water Coral Lophelia pertusa in the Northeastern Gulf of Mexico

Bacteria are recognized as an important part of the total biologyof shallow-water corals. Studies of shallow-water corals suggestthat associated bacteria may benefit the corals by cycling carbon,fixing nitrogen, chelating iron, and producing antibiotics thatprotect the coral from other microbes. Cold-water or deep-seacorals have a fundamentally different ecology due to their adaptationto cold, dark, high-pressure environments and as such have novelmicrobiota. The goal of this study was to characterize the microbialassociates of Lophelia pertusa in the northeastern Gulf of Mexico.This is the first study to collect the coral samples in individualinsulated containers and to preserve coral samples at depthin an effort to minimize thermal shock and evaluate the effectsof environmental gradients on the microbial diversity of samples.Molecular analysis of bacterial diversity showed a marked differencebetween the two study sites, Viosca Knoll 906/862 (VK906/862)and Viosca Knoll 826 (VK826). The bacterial communities fromVK826 were dominated by a variety of unknown mycoplasmal membersof the Tenericutes and Bacteroidetes, whereas the librariesfrom VK906/862 were dominated by members of the Proteobacteria.In addition to novel sequences, the 16S rRNA gene clone librariesrevealed many bacterial sequences in common between Gulf ofMexico Lophelia corals and Norwegian fjord Lophelia corals,as well as shallow-water corals. Two Lophelia-specific bacterialgroups were identified: a cluster of gammaproteobacteria relatedto sulfide-oxidizing gill symbionts of seep clams and a groupof Mycoplasma spp. The presence of these groups in both Gulfand Norwegian Lophelia corals indicates that in spite of thegeographic heterogeneity observed in Lophelia-associated bacterialcommunities, there are Lophelia-specific microbes.

INTRODUCTION

Top

INTRODUCTION

Cold-water and deep-sea corals have become a topic of interestdue to conservation concerns over the impacts of trawling, explorationfor oil and gas, and climate change (51, 52). Although the existenceof these corals has been known since the 1800s, our knowledgeof their distribution, ecology, and biology is limited due tothe technical difficulties of studying them. Lophelia pertusais a globally distributed cold-water scleractinian coral (53).In the Gulf of Mexico, Lophelia reefs occur primarily alongthe continental shelf break (300- to 500-m depth), providingan important complex habitat for a wide variety of fishes, crustaceans,and other invertebrates living below the photic zone (48).

The microbial ecology of cold-water corals in deep water isfundamentally different from that of shallow-water corals dueto the ambient environmental parameters (e.g., darkness, lowtemperature, and increased pressure) and the absence of symbioticzooxanthellae. A few studies have begun to address the microbialassociates of deep-sea corals, focusing on octocorals (9, 44)and on L. pertusa (27, 41, 42, 57, 72). To date, all the Lopheliastudies have been conducted on the eastern side of the Atlantic:the Mediterranean basin (72), Mingulay Bay, Scotland (27), andNorwegian fjords (41, 42, 57). These studies have confirmedthat the Lophelia-associated bacterial community is distinctfrom that of the surrounding seawater and sediments (27, 42,57, 72). A variety of community profile methods (automated rRNAintergenic spacer analysis, terminal restriction fragment lengthpolymorphism, and denaturing gradient gel electrophoresis [DGGE])were used to demonstrate differences between samples withina geographic area, suggesting that the Lophelia-associated microbialcommunity varies depending on regional environmental factors(27, 42, 57). Sequencing of 16S rRNA genes was done in onlytwo studies, and there was no overlap between their data (42,72). However, different methods of collection, extraction, amplification,and sequencing were employed, so the lack of commonality maybe due to methodology rather than biogeography.

Methodology is a concern, particularly the care with which samplesneed to be collected for microbial ecology studies. Deep-seacoral samples are typically collected by a trawl, net, or dredgeor by a submersible/remotely operated vehicle (ROV). With thesemethods, many corals may be combined in a single container,which is not acceptable for microbiological studies becausethe microbial community of one coral could contaminate thatof the other. Similarly, contact with sediment, other invertebrates,mobile fauna, or water masses between the collection point andthe surface could contaminate the coral samples. Unlike thecase with the northeastern Atlantic and Norwegian fjords, thetemperature and salinity gradients in the Gulf of Mexico duringthe warm months of the year can be considerable. In the caseof the Viosca Knoll sites, the bottom temperature was 8 to 11°C,compared to a surface temperature of $≥$ 30°C. Coral samplescollected in uninsulated containers in this area have been observedto be affected (e.g., polyps retracted and copious stress mucusproduction) compared to those in insulated containers. VioscaKnoll is also impacted by the Mississippi River plume. The surfacewaters at these sites were turbid and green and had a salinityof 30 practical salinity units (psu), but below the plume thewaters were clear and had a salinity of 35 psu. With this inmind, we designed a sampling container that would protect thecoral samples from dramatic changes in temperature and salinityby sealing them in individual insulated compartments (see Fig.S1 in the supplemental material). However, the question remainedwhether environmental gradients in light and pressure wouldhave an effect on the microbial diversity of the samples. Toaddress this question, each sample was collected in duplicate:one piece was sealed in a compartment alive, and a replicatepiece was sealed in another compartment and preserved at depthwith a fixative solution. Both sample types ("live" versus "fixed")were sealed and insulated, so temperature and salinity gradientsdid not affect them; live samples were subject to gradientsin light and pressure, while fixed samples were not.

The main objective of this study was to characterize the bacterialassociates of Lophelia pertusa from two sites in the northernGulf of Mexico. Comparing multiple individual colonies fromtwo geographic locations in the Gulf to each other and to bacterialdata from Lophelia samples on the eastern side of the Atlanticwill clarify whether Lophelia has a species-specific bacterialcommunity, as has been described for shallow-water corals (49,55). The results of this study will also better define the totalmicrobial diversity associated with this cold-water coral. Aspecialized sampling device (see Fig. S1 in the supplementalmaterial) was designed to minimize contamination and thermalshock and to allow the introduction of preservative at depthto determine if environmental gradients were affecting microbialdiversity during sampling.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Site descriptions.
Samples of Lophelia pertusa were collected by using a submersibleduring a cruise in the northern Gulf of Mexico: USGS-GM-2004-03,29 July 29 to 5 August 2004 (Table 1). The collection siteswere on the upper continental shelf within the Viosca KnollLease Block of the Minerals Management Service (Fig. 1). StationViosca Knoll 826 (VK826) was at a depth horizon of 500 m andfeatured prolific Lophelia thickets (13, 58). The corals hadthick, heavily calcified skeletons ("brachycephala" morphotype[12]), and there was localized seepage of hydrocarbons. Thewater temperature during collections was 8°C and typicallyranged from 7° to 9°C at this site (12). The averageoxygen saturation was 6.65 ± 0.02 ml/liter, and the averagepH was 7.79 ± 0.01. Station Viosca Knoll 906/862 (VK906/862)was at a depth horizon of 315 m and contained many small isolatedcolonies of Lophelia (13). At this site, the corals had less-calcified,fragile skeletons ("gracilis" morphotype [12]), and there wasno hydrocarbon seepage. Water temperature during collectionswas 11°C and ranged from 9° to 13°C at this siteduring 375 continuous days of measurement (12), occasionallyexceeding the hypothetical upper thermal limit of 12°C forLophelia (22, 23). The average oxygen saturation was 6.22 ±0.02 ml/liter, and the average pH was 7.86 ± 0.00. Salinitywas 35 psu at both locations. Although Lophelia is known tohave both red and white color morphs (42), only the white colormorph was observed at Viosca Knoll.

View this table:
[in this window]
[in a new window]

TABLE 1. Northern Gulf of Mexico Lophelia pertusa samples from which 16S rRNA gene sequences were obtained

View larger version (56K):
[in this window]
[in a new window]

FIG. 1. Locations of the two sampling sites in the northern Gulf of Mexico.

Sampling device.
None of the standard sample containers available on the Johnson-Sea-Linksubmersible were adequate for microbiological sampling. A customsampling device was designed and constructed specifically forthis study. The "Kellogg sampler" is an insulated containerthat has 10 separate sample compartments, each with a slidingdoor and o-ring seal (see Fig. S1 in the supplemental material).Syringes of preservative solution (20% dimethylsulfoxide, 0.25mM EDTA, saturated salt) (17, 59) are connected to five of thesample compartments by one-way valves. The preservative solutionis added at depth by pressing down on a plunger with the submersible'smanipulator arm. The Kellogg sampler was mounted on the frontof the Johnson-Sea-Link submersible and filled using the manipulatorarm or suction tube. The sealed separate compartments keep theindividual coral samples from becoming contaminated by contactwith other coral samples, sediment, or water during the returnto the surface. The insulated container keeps the corals atnear collection temperature. The ability to add fixative tohalf the samples at depth makes it possible to test the hypothesisthat coral-associated microbial communities shift during sampleretrieval in response to changes in light, pressure, or otheruncontrolled factors.

Sample collection.
Branches were removed from Lophelia colonies using either thesubmersible's manipulator claw or suction tube (with a screento prevent the coral from being sucked into the hose) and thenplaced in individual sample compartments in the Kellogg sampler.Coral samples were collected in duplicate from each colony;one sample was brought to the surface alive, and the other wasfixed at depth by the addition of a preservative solution. Metadataincluding time of collection, location, depth, temperature,and salinity were recorded (Table 1). The amount of time betweencollection of a coral sample and arrival on the deck of theship ranged from 45 min to 3 h. Within 15 min of arriving ondeck, Lophelia samples were aseptically transferred from theKellogg sampler compartments into sterile containers and broughtinto the ship's lab for immediate processing. The corals remainedin 7 to 10°C water or fixative until processed.

Previous studies of both shallow-water and deep-sea corals havefound that most of the coral-associated bacteria were novelspecies not present in the surrounding seawater (6, 24, 27,42, 43, 50, 54, 55, 72). Based on this knowledge, we have chosento concentrate our efforts solely on characterizing the Lophelia-associatedbacterial communities.

DNA extraction.
Flame-sterilized needle-nose pliers were used to snip off smallpieces of all Lophelia samples (live and preserved at depth)and place them in sterile aluminum weigh dishes. Each piecewas crushed using a sterile hammer, and the mixture of coralskeleton fragments, polyp tissue, and mucus was transferredto a 2-ml microcentrifuge tube. Microbial community DNA wasextracted from the Lophelia samples using the PowerSoil DNAextraction kit (MoBio Laboratories, Inc., Solana Beach, CA),following the manufacturer's protocol. All samples were processedin triplicate. DNA extracts were frozen (–20°C) fortransport back to the U.S. Geological Survey microbiology laboratoryin St. Petersburg, FL.

DG-DGGE.
Ten samples from site VK826 were compared using double-gradientDGGE (DG-DGGE) (Fig. 2). These 10 samples were actually pairsof samples from five Lophelia colonies: 4753K1/K2, K3/K4, K5/K6,K7/K8, and K9/K10. Odd-number samples were preserved at depth,and even-number samples were recovered without preservation.PCRs (50-µl total volume) were composed of 25 µlHotStarTaq master mix (Qiagen, Valencia, CA), 1 µl (10pmol) 1070F primer, 2 µl (20 pmol) 1392R-gc clamp primer,21 µl deionized water, and 1 µl template (a 1:1,000dilution of a 16S rRNA gene amplicon from 8F/1492R PCR). PCRproducts were resolved on a D-Code universal mutation detectionsystem (Bio-Rad, Hercules, CA) running a 6- to 12%-gradientpolyacrylamide gel containing a 40 to 80% denaturing gradientfor 4 h at 60°C and 109 V (14). The gel was stained withSYBR gold (Invitrogen, Carlsbad, CA) and visualized using aUV transilluminator. DNA bands were cut out of the gel usingsterile razor blades and extracted from the polyacrylamide usingQiaquick gel extraction kit (Qiagen, Valencia, CA). The primers1070F and 1392R were used to reamplify the DNA, which was thencloned as described above. Plasmid preps were conducted usinga MacConnell Mini-Prep 96 system (MacConnell Research, San Diego,CA). Cluster analysis was conducted using the BioNumerics softwareprogram (version 5.0; Applied Maths, Austin, TX).

View larger version (118K):
[in this window]
[in a new window]

FIG. 2. DG-DGGE gel showing bacterial community DNA "fingerprints" from five Lophelia communities, comparing DNA extracted from coral samples preserved at depth (odd numbers) to samples from the same colony brought up alive (even numbers). The gel image has been reversed (i.e., converted to a photo negative) to more clearly show the bands. M, marker; 1, 4753K1; 2, 4753K2; 3, 4753K3; 4, 4753K4; 5, 4753K5; 6, 4753K6; 7, 4753K7; 8, 4753K8; 9, 4753K9; 10, 4753K10. Circles highlight a prominent band in a preserved sample (band D) and its absence in the corresponding unpreserved sample. Band A was cloned and sequenced from lanes 2 to 5 and 7 to 9. Band B was cloned and sequenced from lanes 1, 4, and 6 to 9. Band C was cloned and sequenced from lanes 1 to 4, 7, and 9. Band D was cloned and sequenced from lane 1. Bands E, F, G, H, I, and J were cloned and sequenced from lane 10.

Clone libraries.
Libraries designated 4746K were from samples collected at siteVK906/862, and libraries designated 4753K were from samplescollected at site VK826 (Table 1). The primers 8F (5'-AGAGTTTGATCCTGGCTCAG)(19) and 1492R (5'-GGTTACCTTGTTACGACTT) (62) were used to amplifybacterial-community DNA from the Lophelia samples. PCRs (50-µltotal volume) were composed of 25 µl HotStarTaq mastermix (Qiagen, Valencia, CA), 1 µl (10 pmol) of each primer,13 µl of DI water, and 10 µl template. PCR productswere cloned into vectors using a Topo TA cloning kit (Invitrogen,Carlsbad, CA). Plasmid preps of transformed E. coli were conductedusing the Wizard Preps system (Promega, Madison, WI).

Sequencing.
Sequencing of PCR products from clones containing the correct-sizeinserts was done by using Northwoods DNA (Bemidji, MN). Partialsequences (500 to 755 bp) were obtained from the 5' end of theclone library sequences using the primer 8F. Clone library sequencesof less than 500 bp were discarded. Complete sequences of theDGGE bands were obtained using the primer 1070F (260 to 313bp).

Phylogenetic analysis.
The following software programs were used to process the rawsequence data: Phred (20) to base call and add quality scores;Greengenes (18) to trim poor-quality sections, and BLAST (1)to compare the sequences against the GenBank database. The sequenceswere dereplicated using the FastGroup II software program (73)and chimera checked with the Bellerophon program (30). Percentsimilarity was interpreted as follows: a 97 to 100% match toa GenBank entry was considered to be within the same species,a 93 to 96% match was considered to be within the same genus,and an 86 to 92% match was considered to be a related organism(61). However, note that there can be significant genetic andphysiological differences between two bacteria that have 99%similarity in their 16S rRNA genes (31), so all estimates ofdiversity are actually minimum approximations. The RibosomalDatabase Project (release 10, update 3) Classifier was employedto assign the 16S rRNA genes to appropriate taxonomic groups(67). Rarefaction analyses were conducted using FastGroup II(73). Alignments were constructed using ClustalX (66). Maximum-likelihoodtrees were created and bootstrapped using the PAUP softwarepackage (64).

Nucleotide sequence accession numbers.
The Lophelia clone library sequences are archived in GenBankunder accession numbers FJ041383 to FJ041890. Bands A to J fromthe DGGE analyses are archived under accession numbers FJ196875to FJ196884.

RESULTS

Top

RESULTS

Five Lophelia pertusa colonies (each) were sampled at VK906/862and VK826 (Fig. 1 and Table 1). Microbial communities from samplespreserved at depth were compared by DG-DGGE to those from thesame coral colony brought up alive to determine if environmentalfactors (e.g., light and pressure) were affecting the microbialdiversity of the samples. Clone libraries were prepared fromsix individual Lophelia colonies and used to describe the bacterialdiversity associated with this coral at both these Gulf of Mexicosites.

Preserving samples at depth.
Two samples each (one preserved at depth and one retrieved alive)from five Lophelia colonies at site VK826 were compared by DG-DGGE(Fig. 2). This method tends to detect the most abundant membersof a bacterial community (40) and therefore was used to screenfor major changes resulting from the type of sampling. Whilethe overall patterns are similar, there are a few obvious differencesbetween samples preserved at depth and those brought up withoutpreservation (i.e., samples that spent up to 3 h under conditionsof changing light and pressure) (Fig. 2). For example, bandD is present in a fixed sample (Fig. 2, lane 1) but is absentin the companion live sample (Fig. 2, lane 2), as indicatedin the figure. Lane 9 (fixed) looks similar to the other samples,but lane 10 (live sample from the same coral colony) has manyadditional bands (Fig. 2, bands E to J). The band patterns wereexamined by cluster analysis (see Fig. S2 in the supplementalmaterial). In all but one case (lanes 9 and 10), the samplesclustered in pairs based on the colony sampled rather than groupingby treatment (live versus fixed at depth).

Bacterial diversity.
Several of the prominent bands on the DG-DGGE gel (Fig. 2) wereexcised, the DNA extracted, and sequences obtained. The topband in the first doublet (Fig. 2, band A) was consistent acrosslanes 1 to 5 and 7 to 9 and was 100% identical to a gammaproteobacteriumclone (G02_CR02_full; GenBank accession no. AM911391) from thered color morph of Lophelia from a Norwegian fjord (42). Thelower band in the first doublet (Fig. 2, band B) also was consistentacross lanes 1 to 9 and was also 100% identical to a mycoplasmalsequence (D11_CW02_full; also known as "Candidatus Mycoplasmacorallicola"; GenBank accession no. AM911412) from a white NorwegianLophelia coral (41, 42). The next-closest similarity (89%) wasto Mycoplasma phocirhinis (35). All the bands in lane 10 (Fig.2, bands E to J) and one band in several other lanes (Fig. 2,band C) were nearly identical to each other and representedalphaproteobacteria in the family Rhodobacteraceae. These bandswere equally similar (95 to 96%) to that of clone 2WB_40 (GenBankaccession no. EU574666), which was recently isolated from amicrobial mat at a hydrothermal vent site (16) and to cloneSGUS450 (GenBank accession no. FJ202844) from a piece of thecoral Montastraea faveolata that had been kept in an aquariumfor 23 days (63). The conspicuous single band in lane 1 (Fig.2, band D) that is lost in the corresponding live sample belongedto a gammaproteobacterium with 99% sequence similarity to abacterial sequence (clone s1uc48; GenBank accession no. DQ416306)associated with the shallow-water coral Oculina patagonica inthe Mediterranean (36).

Seven clone libraries were screened from six individual coralcolonies (six samples preserved at depth and one of the coordinatinglive samples from that set). Three corals were from site VK906/862(library prefix 4746) and three from VK826 (library prefix 4753).Sequences of less than 500 bp were discarded, resulting in theanalysis of 508 total clones, comprised of 273 unique operationaltaxonomic units (OTUs) (based on a 97%-similarity cutoff). Manyof these 273 OTUs were combined into groups when the alignmentswere trimmed for consistency, so the number of OTUs visibleon the phylogenetic trees is reduced. Fifteen percent of thebacterial 16S clones were novel, having less than 97% similarityto previously described sequences. Fifty-two percent of theclones were unknown mycoplasmal members of the Tenericutes,33% were members of the Gammaproteobacteria, 9% were membersof the Alphaproteobacteria, and 3% were members of the Bacteroidetes(Fig. 3) (see Table S1 in the supplemental material). Interestingly,BLAST searches resulted in all 16 sequences classified as Bacteroidetesbeing most similar to species of mycoplasma. The remaining 3%consisted of members of the Betaproteobacteria (two clones),Deltaproteobacteria (one clone), Epsilonproteobacteria (twoclones), Cyanobacteria (one clone), Planctomycetes (one clone),Firmicutes (one clone), Lentisphaerae (three clones), and Spirochaetes(one clone) (see Table S1 in the supplemental material). Thethree clones classified as Lentisphaerae were 91 to 94% similarto Planctomyetes by BLAST search and clustered with Planctomycetesequences (Fig. 4).

View larger version (39K):
[in this window]
[in a new window]

FIG. 3. Composition of the clone libraries showing only taxa comprising at least 15% of sequences in at least one clone library. Libraries 4746K1 (74 clones), 4746K7 (43 clones), 4746K9 (52 clones), 4753K3 (50 clones), 4753K5 (87 clones), and 4753K9 (96 clones) are from samples preserved at depth. Library 4753K4 (106 clones) is from a nonpreserved sample of the same coral as library 4753K3.

View larger version (52K):
[in this window]
[in a new window]

FIG. 4. Maximum-likelihood tree of Lophelia-associated sequences (excluding the mycoplasma sequences) created with PAUP using the general time-reversible model with estimated base frequencies site specific by codon position distribution. Percentages of 1,000 bootstrap resamplings are shown at the nodes (values $≥$ 50% only). The tree was rooted using Methanococcus voltae U38488 as an outgroup (not shown). The scale bar denotes 0.02 nucleotide substitutions per alignment position. The gray box indicates coral-associated sequences most similar to those of thiotrophic symbionts. Anchor sequences labeled with an asterisk are coral associated. The tree is based on an alignment made with ClustalX, containing 534 characters per sequence.

Many of the clones had high similarity to previously describedsequences. Eighty-six percent of the clones had highest BLASTsimilarity matches to other coral-associated bacteria. Of those,79% were most similar to one of two sequences from NorwegianLophelia (42): either the thiotrophic symbiont sequence (G02_CR02_full;GenBank accession no. AM911391) (Fig. 4) or the mycoplasmalsequence (D11_CW02_full; also known as "Candidatus Mycoplasmacorallicola"; GenBank accession no. AM911412) (Fig. 5). The"thiotrophic cluster" (Fig. 4) also includes another clone exclusiveto red Lophelia corals from Norway (E12_CR02_full; GenBank accessionno. AM911378) and closely related shallow-water coral sequencesfrom Turbinaria mesenterina (S. E. Godwin, J. Borneman, E. Bentand L. Pereg-Gerk, unpublished). This coral-associated groupis similar to sulfur-oxidizing bacterial symbionts of methaneseep clam species (e.g., Calyptogena spp. and Bathymodiolusspecies) (Fig. 4). These sulfur oxidizers were the numericallydominant clone type in two libraries (4746K7 and 4746K9) andwere present in all the Lophelia bacterial 16S rRNA gene libraries.Of the remaining 7%, 32 clones were 92 to 95% similar to cloneFungiaD34clE09 from the Philippines (26), two clones were 95%similar to clone 1HP1-A14, which was associated with Turbinariamesenterina (Godwin et al., unpublished), and two clones showedsimilarity to sequences derived from Montastraea faveolata (clonesSHFS395 and SHFH617) (63). Although not included as "coral associated"in the strictest sense, there were also three clones that were97% similar to Ksed45, a sequence from sediments near a deep-watercoral reef in Norway (S. Jensen, J. D. Neufeld, N.-K. Birkeland,M. Hovland, and J. C. Murrell, unpublished data).

View larger version (23K):
[in this window]
[in a new window]

FIG. 5. Maximum-likelihood tree of the Lophelia-associated Mycoplasma sequences created with PAUP using the general time-reversible model with estimated base frequencies site specific by codon position distribution. Percentages of 1,000 bootstrap resamplings are shown at the nodes (values $≥$ 50% only). The tree was rooted using Clostridium innocuum M23732 as an outgroup (not shown). The scale bar denotes 0.02 nucleotide substitutions per alignment position. The tree is based on an alignment made with ClustalX, containing 533 characters per sequence.

The 14% of clones without sequence similarity to other coral-associatedsequences were an eclectic mix. There was a single clone with99% similarity to a bacterial symbiont of the pink sugarcanemealybug (21) and another clone with 99% similarity to fourSynechococcus clones from Chesapeake Bay (15). There were 30clones with 97 to 98% similarity to uncultured Colwellia strains(68, 69). A small number of clones were similar to other metazoan-associatedbacteria. Nine alphaproteobacterial clones were 90 to 92% similarto a squid symbiont (45), two epsilonproteobacterial cloneswere 92 to 93% similar to a bacterial sequence from deep-seashrimp guts (74), one gammaproteobacterial clone was 84% similarto a clone from a sponge (D. Sipkema, unpublished), and twogammaproteobacterial clones were 83% similar to a clone froma colonial ascidian in the Mediterranean (39). One Lophelia-associatedsequence belongs to the Spirochaetes, but it is less than 90%similar to anything in the current genetic database. Membersof the Spirochaetes have been detected previously in deep-seabamboo corals (44), but this is the first description of oneassociated with Lophelia. Two clones showed most sequence similarityto three putative archaeal sequences derived from a hydrothermalvent chimney (H. Oba, K. Inoue, S. Kato, and A. Yamagishi, unpublished).However, phylogenetic analyses place this entire group of sequencesfirmly in the Alphaproteobacteria (Fig. 4).

The rarefaction curve generated from the 508 bacterial clonesderived from all Gulf of Mexico Lophelia pertusa samples wascompared to a rarefaction curve based on 534 clones from deep-seabamboo corals (44) and was nearly identical (see Fig. S3 inthe supplemental material). Compared to rarefaction curves ofthe red and white color morphs of Lophelia corals sampled inNorway (42), Gulf of Mexico Lophelia displays greater diversity:79 ribotypes in the first 100 clones versus $~$ 25 (Norwegian white)and $~$ 40 (Norwegian red).

The clone libraries show a definite difference in coral-associatedbacterial communities between sites VK906/862 and VK826. Thethree libraries from VK906/862 (4746K1, 4746K7, and 4746K9)are dominated by the Proteobacteria, whereas the four librariesfrom VK826 (4753K3, 4753K4, 4753K5, and 4753K9) are shiftedto the mycoplasmal Tenericutes (Fig. 3, Fig. 5).

DISCUSSION

Top

DISCUSSION

Unlike collection of shallow-water corals, a situation in whichthere is only a minimal time lag or temperature difference betweencollection and preservation, sampling of cold-water corals inthe Gulf of Mexico presents some challenges. There was an approximately20°C difference between bottom and surface waters, and theMississippi River plume affects these sites, generating a lensof lower-salinity water at the surface. These factors were takeninto consideration, and sampling gear was designed to protectthe samples from these gradients. The question was still raisedas to whether other environmental gradients, such as light andpressure, could affect the coral-associated microbial communityduring the approximately 4 h of a submersible dive. The visualdifferences between DG-DGGE patterns of samples from the samecoral colony (one brought up alive and one preserved at depthimmediately after collection) (Fig. 2) suggested that theremight be some effect; however, cluster analysis (see Fig. S2in the supplemental material) indicated that in all but onecase (samples 4753K9 and 4753K10), the colony of origin wasa more important factor than preservation. This indicates thatin most cases, the bacterial communities are not being dramaticallyaltered by exposure to gradients in light and pressure. Thisis corroborated by Neulinger et al.'s terminal restriction fragmentlength polymorphism data (42); they collected samples with netsand found statistically significant differences between communitypatterns at stations 1 (depth, 54 m) and 2 (depth, 264 m) butnot between stations 1 and 3 (depth, 240 m), indicating thatthe differences were not due to depth (or the effects of collectingsamples at different depths). This is the first study of deep-seacoral microbial ecology that has preserved samples at depthto investigate this issue. Additionally, this is the only deep-seacoral study to use insulated, individually sealed containersto minimize cross-contamination and thermal shock. Previoussubmersible and ROV studies that have examined bacterial diversityassociated with deep-sea corals in general and Lophelia in particulardid not take these precautions (27, 42, 44, 72).

With only two geographic sites, it is impossible to determineif the bacterial community at one site is "normal" and the othersite "altered." Recent studies have revealed conserved microbialcommunities within species of shallow-water corals (54, 55)and sponges (28). However, Lophelia pertusa may have a moredynamic bacterial community across its wide geographic range.The similarity of "fingerprint" patterns among the five Gulfof Mexico Lophelia samples analyzed by DGGE (Fig. 2) and thesimilarity of dominant clone types in the clone libraries atVK906/862 versus VK826 (Fig. 3) indicate some conservation ofmicrobial communities within geographic sites. This corroboratesfindings from Lophelia corals sampled on the Mingulay reefsoff Scotland. DGGE patterns of coral-associated bacterial communitieswere different at different geographic sites but were more similarbetween proximal sites (27). However, there was concern thatsampling techniques may have influenced the degree of geographicvariability detected (27). There was no overlap between the12 bacterial 16S rRNA gene sequences from Mediterranean Lophelia(72) and the bacterial sequences from this study (Fig. 4 and5) or a recent study of Lophelia in a Norwegian fjord (42).While this is most likely due to methodological differences(e.g., different extraction, amplification, and sequencing protocols),it could indicate biogeographical differences in Lophelia-associatedbacterial communities. Ultimately, it will be valuable to havecomparable Lophelia samples from other areas in the Gulf ofMexico, such as Green Canyon off the coast of Louisiana (58)and the West Florida slope (48), to determine more clearly whatconstitutes the "typical" bacterial community of Lophelia inthe Gulf of Mexico and how much of an effect regional geographyhas on the structure of those communities.

There were definite similarities between the bacterial communitiesof Gulf of Mexico Lophelia and Norwegian Lophelia corals (42).Both were dominated by Proteobacteria (with strong representationof the family Rhodobacteraceae) and included mycoplasmas (Tenericutes),Bacteroidetes, and Planctomycetes) and a small number of cyanobacterialclones. However, the Norwegian samples had a greater representationof alphaproteobacteria whereas the Gulf samples had more gammaproteobacteria.The Norwegian bacterial communities also included Actinobacteria,Verrucomicrobia, and TM7, which were not detected in Gulf ofMexico samples. Interestingly, the Norwegian samples containedtwo color morphs of Lophelia, red and white, which had differentbacterial communities (42). The bacterial diversity of Gulfof Mexico Lophelia (only the white color morph was encounteredat the Viosca Knoll sites) is more like that of the Norwegianred color morph, which exhibited greater diversity than thewhite.

One key commonality is the abundance of members of the Tenericutesassociated with Lophelia corals at site VK826 (Fig. 5). Mycoplasmaspp. in the class Mollicutes have traditionally been listedunder the phylum Firmicutes, but the division Tenericutes (8)has been raised to a phylum to separate the wall-less Mollicutesfrom the gram-positive genera Bacillus, Clostridium, and Staphylococcuswithin the Firmicutes (2). Many mycoplasmas are pathogens thatparasitize animal host cells for nutrients they lack due totheir small genome size and limited biosynthetic capabilities(47, 56). Mycoplasmas have been detected in both bleached andhealthy Muricea elongata, a shallow-water gorgonian (L. K. Ranzer,P. F. Restrepo, and R. G. Kerr, unpublished). At least one novelMycoplasma sp. was associated with deep-sea bamboo corals (44).No Tenericutes were found in association with MediterraneanLophelia (72), but "Candidatus Mycoplasma corallicola" has beenfound in Norwegian Lophelia (41, 42). Fluorescent in situ hybridizationwas used to determine that "Candidatus Mycoplasma corallicola"in Norwegian Lophelia positions itself exclusively around nematocystbatteries, oriented toward the outside of the coral (41). Thepositioning of the mycoplasmas suggested that they were benefitingfrom "sloppy feeding" on the coral's part and that they werelikely to be harmless commensals. Phylogenetic comparison ofall coral-associated mycoplasmas demonstrates that they clustertogether (Fig. 5) and that "Candidatus Mycoplasma corallicola"shares 99% similarity with the largest group of mycoplasmalclones from Gulf of Mexico Lophelia (Fig. 5). The presence of"Candidatus Mycoplasma corallicola" in coral colonies on bothsides of the Atlantic indicates that this bacterium is a Lophelia-specificsymbiont.

The question then becomes, why are there not also mycoplasmasat VK906/862? Their absence at VK906/862 might be linked toa physical factor, e.g., temperature variability. Bacterialcommunities have been shown to differ between diseased/stressedcorals and healthy specimens (43), making it possible that thedramatic difference between the Lophelia-associated bacterialcommunities at VK906/862 and VK826 (Fig. 3) was due to thermalstress at the shallower site. Perhaps the difference in bacterialcommunities between the two sites was linked to a factor intrinsicto Lophelia, e.g., the corals at VK906/826 were more geneticallydifferentiated than other populations sampled in the Gulf, includingVK826 (C. L. Morrison, personal. communication), based on analysisof microsatellite markers. Differences in genetics, nutrition,or both could lead to biochemical or metabolic differences whichwould select for different microbial biota. Lophelia coloniesat VK826 were more heavily calcified than those at VK906/826,and based on sediment trap experiments, VK826 had higher numbersof plankton (12). The differences may also be influenced bythe limitations of clone libraries; Neulinger et al. found mycoplasmalsequences in the clone libraries for white Lophelia only (42)but showed by fluorescent in situ hybridization that mycoplasmaswere present in both white and red color morphs (41).

Molecular analyses of shallow-water corals have uncovered individualspecies-specific associations between corals and bacteria: analphaproteobacterium (PA1) has been found consistently associatedwith Porites astreoides, a closely related bacterium, PF1, withPorites furcata (55), and a Silicibacter sp. with Montastraeafranksi (54). Bacteria such as these are inferred to have aparticular symbiotic association with the coral. In additionto showing a mycoplasmal associate 99% similar to the one describedfor Norwegian Lophelia (Fig. 5), this study has also detecteda cluster of closely related gammaproteobacteria (Fig. 4) inall six clone libraries, dominant in two. This cluster is alsoclosely related (95 to 98% similarity) to a sequence (G02_CR02_full;GenBank accession no. AM911391) found in association with thered color morph of Lophelia in Norway (42), again indicatinga specific association that transcends geographic distance.Nearly one-quarter of the 16S rRNA gene clones associated withGulf Lophelia belong to this cluster, with similarity to thiotrophicsymbionts of seep-clam species (Fig. 4).

Lophelia-associated bacterial sequences also showed similarityto other types of symbionts, including bacterial associatesof an insect, squid, sponge, deep-sea shrimp, and sea squirt(ascidian). One of the bands excised from the DGGE (Fig. 2,band D) produced a sequence that was 99% similar to a bacterialsequence that had been extracted from a shallow-water coral,Oculina patagonica, in the Mediterranean (36). Clone matchesto Oculina patagonica have also been found in Lophelia fromNorwegian fjords (42). Coupled with the 7% of the clone librariesthat was most similar to tropical coral-associated sequences,this provides evidence that there may be scleractinian coral-specificbacterial associations, regardless of depth, corroborating similarfindings by Neulinger et al. (42).

It seems counterintuitive to find photosynthetic autotrophslike cyanobacteria associated with corals that live in totaldarkness. Synechococcus sequences were found associated withboth white and red color morphs of Norwegian Lophelia (42),and this study detected a Synechococcus sequence associatedwith Lophelia at VK826 (Fig. 5). Cyanobacteria are known tobe associated with photosynthetic corals and are believed tobenefit the corals through nitrogen fixation (37, 38, 60, 71).A group of globally distributed oceanic cyanobacteria has beenshown to lack the genes for oxygenic photosynthesis and carbonfixation (75). Therefore, it is possible that facultativelyheterotrophic species are associated with deep-sea corals inspite of the dark environment.

Three Lophelia-associated clones were classified into a recentlydescribed phylum of marine bacteria, Lentisphaerae (10). Taxonomically,this group is regarded as a sister phylum to Verrucomicrobia(46). Recent studies have discovered Lentisphaerae phylum clonesassociated with marine sponges (33, 65). This study is the firstreport of this bacterial phylum being associated with a coral.However, the three clones phylogenetically cluster with Planctomycetes(Fig. 4) and were most similar to a planctomycete clone derivedfrom salmon gills (7) and several coral-associated sequencesfrom healthy Montastraea faveolata (63).

Whereas spirochetes have been isolated from seawater and sedimentsin the vicinity of Lophelia (42), this is the first report ofspirochete-like bacteria associated with the coral. The singleclone (4746K1-41; GenBank accession no. FJ041427) is 87% similarto uncultured Spirochete clones but does cluster with a definedspirochete species (Fig. 4). The Spirochetes are typically free-livinganaerobes but are known to be symbionts of gutless worms commonlyfound in coral reef sediments (5). Further, these gutless wormshost multiple symbionts, mixing the Spirochetes with the thiotrophicGammaproteobacteria (5), as is the case with Lophelia.

Although archaea are known to be associated with shallow-watertropical corals (4, 32, 70), deep and cold waters (3, 11, 25,29), and marine sediment (34, 72), no archaea have yet beenfound associated with Lophelia pertusa (72). Two clones fromGulf of Mexico Lophelia were most similar to three sequencesclaimed to be archaeal clones from a hydrothermal vent. However,there is no associated publication to clarify why the submitterswere convinced the sequences were archaeal. When included ina phylogenetic tree, both the Lophelia sequences and the putativearchaeal sequences fall within the Alphaproteobacteria (Fig.4).

The majority of known coral species occur in deep water (52);however, there are very few publications about bacteria associatedwith deep-sea corals. This paper documents the first surveyof Lophelia-associated microbes in the Gulf of Mexico and isthe first deep-water study in which specimens were collectedin insulated, individual containers to minimize sample contaminationand thermal stress. The results revealed several novel bacterialassociates and two Lophelia-specific groups: a group of Mycoplasmaspp. and a cluster of symbiont-like thiotrophs.

ACKNOWLEDGMENTS

This work was supported by the USGS Outer Continental ShelfEcosystem Program and was sponsored and facilitated by the MineralsManagement Service.

We thank Z. Pratte for preparing the sequences for GenBank submissionand running some of the BLAST searches. M. Gray also providedphylogenetic advice and ran some of the BLAST searches. Thanksare also extended to the Harbor Branch Oceanographic Institution,the captain and crew of the R/V Seward Johnson, and especiallythe pilots and technicians of the submersible Johnson-Sea-Link,without whom this research could not have been conducted.

Any use of trade names is for descriptive purposes only anddoes not imply endorsement by the U.S. government.

FOOTNOTES

* Corresponding author. Mailing address: U.S. Geological Survey, 600 4th Street South, St. Petersburg, FL 33701. Phone: (727) 803-8747, ext. 3128. Fax: (727) 803-2031. E-mail: ckellogg{at}usgs.gov

Published ahead of print on 20 February 2009.

Supplemental material for this article may be found at http://aem.asm.org/.

REFERENCES

Top