Correspondence: Erwan G. Roussel, School of Earth and Ocean Sciences, Cardiff University, Main Building, Park Place, Cardiff CF10 3AT, Wales, UK. Tel.: +44 29 2087 4488; fax: +44 29 2087 4326; e-mail: email@example.com
The distribution of Archaea and methanogenic, methanotrophic and sulfate-reducing communities in three Atlantic ultramafic-hosted hydrothermal systems (Rainbow, Ashadze, Lost City) was compared using 16S rRNA gene and functional gene (mcrA, pmoA and dsrA) clone libraries. The overall archaeal community was diverse and heterogeneously distributed between the hydrothermal sites and the types of samples analyzed (seawater, hydrothermal fluid, chimney and sediment). The Lost City hydrothermal field, characterized by high alkaline warm fluids (pH>11; T<95 °C), harbored a singular archaeal diversity mostly composed of unaffiliated Methanosarcinales. The archaeal communities associated with the recently discovered Ashadze 1 site, one of the deepest active hydrothermal fields known (4100 m depth), showed significant differences between the two different vents analyzed and were characterized by putative extreme halophiles. Sequences related to the rarely detected Nanoarchaeota phylum and Methanopyrales order were also retrieved from the Rainbow and Ashadze hydrothermal fluids. However, the methanogenic Methanococcales was the most widely distributed hyper/thermophilic archaeal group among the hot and acidic ultramafic-hosted hydrothermal system environments. Most of the lineages detected are linked to methane and hydrogen cycling, suggesting that in ultramafic-hosted hydrothermal systems, large methanogenic and methanotrophic communities could be fuelled by hydrothermal fluids highly enriched in methane and hydrogen.
In the present study, we characterized the molecular genetic diversity, using 16S rRNA gene and functional genes of methanogens, methanotrophs and sulfate-reducers, associated with three ultramafic-hosted hydrothermal sites: Rainbow, Lost City and Ashadze. As these hydrothermal fluids are highly enriched in methane and hydrogen, these environments could harbor specific prokaryotic communities possibly associated with potential subsurface chemolithoautotrophic ecosystems. The aim of this study was to compare the microbial communities of these ultramafic-hosted hydrothermal sites using molecular genetic methods to correlate their phylogeny with ecological niches.
Materials and methods
Site location and sampling techniques
Fluid, chimney and sediment samples were collected during the scientific cruises EXOMAR (2005), SERPENTINE (2007) and MoMARDREAM-Naut (2007) conducted with the R.V. L'Atalante and Pourquoi pas? and using the ROV Victor 6000 and DSV Nautile. The three hydrothermal fields explored, Rainbow (36°13′N, 33°54′W; ∼2300 m depth), Lost City (30°07′N, 42°07′W; ∼750 m depth) and Ashadze 1 (12°58′N, 44°51′W; ∼4090 m depth), are all located along the MAR, although Lost City and Ashadze were further from the axis (Fig. 1a).
Fluid samples from Rainbow, Lost City and Ashadze were collected, respectively, from the thermitière chimney (36°13′76″N, 33°54′16″W; 2294 m depth; Fig. 1e) from a flange near the EXOMAR 11 Marker (30°07′43″N, 42°07′16″W; 748 m depth; Fig. 1c) and from two chimneys in the SE2 area at Ashadze 1 site (12°58′33″N, 44°51′78″W; 4097 m depth; Fig. 1f). Chimney samples were also retrieved from the two chimneys in the SE2 area of Ashadze 1 site where the fluids were collected previously. The sediment samples were retrieved from the Rainbow site close to the active hydrothermal area (36°13′76″N, 33°54′04″W; 2287 m depth, Fig. 1d) and from the immediate periphery of the Lost City site (30°07′57″N, 42°07′05″W; 752 m depth; Fig. 1b).
To describe the microbial communities from the surrounding seawater, the water column from the Rainbow site (36°13′76″N, 33°54′06″W; 2291 m depth) was also sampled. All fluid samples were collected using titanium syringes and analyzed as described elsewhere (Charlou et al., 2002). On board, the fluid samples were immediately removed aseptically from the titanium syringes and stored at −80 °C for molecular genetic analyses. The sediment cores (∼20 cm in length, 5 cm diameter) collected from the Rainbow site using a push-core device operated by the arm of the DSV Nautile were sectioned on board into three equal samples, designated as top, middle and bottom. The sediment surface sample from the Lost City site was collected using the PSDE system (Fig. 1b; C. Kato, unpublished data). The chimney fragments were collected in a biobox and sediment samples were stored aseptically at −80 °C for molecular genetic analyses.
DNA extractions and PCR amplification
To avoid contamination, all manipulations were carried out in a PCR cabinet (Biocap™ RNA/DNA, Erlab®), using Biopur® 1.5-mL Safe-Lock micro test tubes (Eppendorf™), Rnase/Dnase-Free Water (MP Biomedicals™) and UV-treated (>60 min) plasticware and pipettes.
DNA fluids were extracted from 50 mL of fluid left to thaw on ice before centrifugation (15 000 g for 60 min). The supernatant was carefully discarded and DNA was extracted from the pellet, following a modified FastDNA® Spin Kit for Soil (Bio101 Systems, MP Biomedicals™) protocol (Webster et al., 2003; Roussel et al., 2009a). DNA was also extracted from sediments and chimney fragments using the modified FastDNA® Spin Kit for Soil as described elsewhere (Roussel et al., 2009a).
All amplifications were performed using a GeneAmp PCR system 9700® (Applied Biosystems™). All PCR mixtures (50 μL) contained 5 μL of DNA template, 1 × Taq DNA polymerase buffer (MP Biomedicals™), 1 μL of dNTP (10 mM each of dATP, dCTP, dGTP and dTTP), 10 μM of each primer and 0.5 μL of Taq DNA polymerase (MP Biomedicals™). Negative controls were also carried out with DNA extractions performed without any sample. No PCR products were detected for any control. The inhibition of PCR amplification by soluble contaminants in the DNA extracts was also tested as described elsewhere (Juniper et al., 2001).
Archaeal 16S rRNA gene amplification was conducted by nested PCR with a combination of primers A8f (5′-CGG TTG ATC CTG CCG GA-3′) and A1492r (5′-GGC TAC CTT GTT ACG ACT T-3′) in the first round (Teske et al., 2002; Lepage et al., 2004) and A344f (5′-AYG GGG YGC ASC AGG SG-3′) and A915r (5′-GTG CTC CCC CGC CAA TTC CT-3′) in the second round (Stahl & Amann, 1991; Sørensen et al., 2004). The PCR cycles for the first round (A8f/A1492r) and the second round (A344f/A915r) were as described previously (Roussel et al., 2009a). To minimize PCR bias, five independent PCR products from the first round were pooled and purified (QIAquick PCR purification Kit; Qiagen™) and used as a template for the second round. This nested PCR was necessary to obtain visible PCR products on a 0.8% (w/v) agarose gel stained with ethidium bromide.
A portion of the mcrA gene was amplified using the ME primers (Hales et al., 1996) with the following reaction conditions as described elsewhere (Roussel et al., 2009a). A fragment of the pmoA gene was amplified using the pmoA189-mb661 primer couple (Holmes et al., 1995; Costello & Lidstrom, 1999) under the following reaction conditions: 1 cycle of 4 min at 92 °C, 35 cycles of 1 min at 92 °C, 1.5 min at 55 °C and 1 min at 72 °C and 1 cycle of 9 min at 72 °C. A portion of the dsrA gene was amplified using the DSR1F+ and DSR-R primers (Kondo et al., 2004) under the following reaction conditions: 1 cycle of 5 min at 94 °C, 35 cycles of 30 s at 94 °C, 30 s at 54 °C and 2.5 min at 72 °C and 1 cycle of 8 min at 72 °C. For all functional genes, two rounds with the previous reaction conditions were required to obtain visible amplification products. A 5-μL aliquot of three pooled PCR products of the primary amplification was used as a template for the second amplification round.
Co-migration denaturing gradient gel electrophoresis (CM-DGGE) analysis
To obtain the general archaeal 16S rRNA gene diversity associated with the hydrothermal environment and to compare it with the seawater diversity, a preliminary CM-DGGE analysis was performed as described elsewhere (Roussel et al., 2009b).
After the amplification of the nested PCR products, using two different fluorescent reverse-labelled (Cy3 or Cy5) primers from total DNA from either a hydrothermal sample or seawater, the products were pooled and loaded into the same lane. Archaeal 16S rRNA gene amplification was performed with the primers Saf-PARCH 519r, labelled with either Cy3 (hydrothermal samples) or Cy5 (seawater), following the touchdown PCR protocol as described previously (Nicol et al., 2003). All manipulations were performed in the dark. The PCR products were analyzed by DGGE using a DCode Universal Mutation Detection System® (BioRad™) on a 1-mm-thick (16 × 16 cm) 8% (w/v) polyacrylamide gel (acrylamide/bisacrylamide, 40%, 37.5 : 1, BioRad™) with a denaturant gradient between 30% and 70% prepared with 1 × TAE buffer (pH 8, 40 mM Tris Base, 20 mM acetic acid, 1 mM EDTA, MP Biomedicals™) and poured with a Gradient maker (Hoefer SG30®). Electrophoresis was carried in 1 × TAE buffer at 60 °C for 330 min at 200 V (initially at 80 V for 10 min). The gel was scanned using a Phospho fluorimager Typhoon 9400® (Amersham Biosciences™).
Cloning and sequencing
Fourteen 16S rRNA gene, one dsrA gene, four mcrA gene and eight pmoA gene clone libraries were constructed. To minimize PCR bias (Polz & Cavanaugh, 1998), five independent PCR products were pooled, purified (QIAquick PCR purification Kit; Qiagen™) and cloned into Escherichia coli (XL10-Gold; Stratagene™) using the pGEM-T Easy vector system I (Promega™) following the manufacturer's instructions. Positive transformants were screened by PCR amplification of the insert using the vector-specific M13 primers. Plasmid extraction, purification and sequencing of the insert were carried out by the sequencing Ouest-Genepole platform® of the Roscoff Marine Laboratory (France).
Phylogenetic analysis and statistical analyses
Chimeras (Cole et al., 2003) were excluded from the clone libraries and a total of 759 sequences (including those from the 16S rRNA gene and functional genes) were used for further phylogenetic analysis. The phylogenetic placement was carried out using the NCBI blast search program within GenBank (http://www.ncbi.nlm.nih.gov/blast) (Altschul et al., 1990). The 16S rRNA gene sequences (∼553 bases) were then edited in the bioedit 22.214.171.124 program (Hall, 1999) and aligned using clustalw (Thompson et al., 1994). The phylogenetic trees were constructed by the phylo_win program (http:// pbil.univ-lyon1.fr/) (Galtier et al., 1996) with the neighbor-joining method (Saitou & Nei, 1987) and Jukes and Cantor correction. The nonchimeric mcrA (∼0.76 kb), pmoA (∼0.51 kb) and dsrA (∼0.22 kb) sequences were translated into amino acids using bioedit and then aligned using clustalw, and the phylo_win program with the neighbor-joining algorithm, and PAM distance (Dayhoff et al., 1978) was then used for phylogenetic tree construction. For the entire phylogenetic reconstruction, the robustness of inferred topology was tested by bootstrap resampling (1000); values over 50% are shown on the trees. The richness from the clone libraries was estimated, with the rarefaction curves at 99%, 97% and 95% sequence identity levels, using the dotur program (Schloss & Handelsman, 2005). Operational taxonomic units (OTUs), using a 95% or a 97% sequence similarity, were generated using the son program (Schloss & Handelsman, 2006), and the percentage of coverage (Cx) of the clone libraries was calculated using Good's method (Good, 1953) as described by Singleton et al. (2001). Statistical estimators, the significance of population differentiation among clone libraries (FST) (Martin, 2002) and the exact tests of population genetic differentiation (Raymond & Rousset, 1995) were calculated using arlequin 3.11 (Excoffier et al., 2005).
Fifteen samples of fluids, chimney fragments and sediments were retrieved from three Atlantic ultramafic-hosted hydrothermal sites: Rainbow, Lost City and Ashadze (Fig. 1). The dilution of the hydrothermal fluid sample was estimated by pH measurements. Overall, the three sites had much higher hydrogen (<16 mM) and abiogenic methane (<2.5 mM) concentrations than the MAR basaltic-hosted hydrothermal sites (Charlou et al., 2010).
All the hydrothermal fluid samples from the Rainbow site were retrieved from the thermitière chimney group (Fig. 1e), except the PP27 swarm sample, which was obtained in close proximity to a shrimp swarm on the side of the PP27 chimney. The thermitière chimney group consisted of both diffuse and black smoker venting. The Rainbow sediment samples were retrieved close to the hydrothermal chimneys and were predominantly made of pelagic sediment (98% calcite) with a small amount of hematite, indicating a small hydrothermal contribution (Fig. 1d). For this study, the maximum temperature measured at Rainbow was 324 °C, and the less diluted hydrothermal fluid analyzed had a pH of 3.40 (Fig. 2b) and high concentrations of hydrogen (>10 mM), carbon dioxide (17 mM), iron (>17 mM) and methane (>1 mM) (Charlou et al., 2010).
The Lost City fluid samples were obtained from one of the hottest venting areas of this site, which was located above a flange (Fig. 1c). To date, Lost City is a unique off-axis hydrothermal site expulsing fluids with a high pH (∼11), as opposed to the other known ultramafic environments, which are acidic (Rainbow and Ashadze, pH∼3). The maximum temperature recorded at Lost City (93 °C) was lower than that for Rainbow and Ashadze. The less diluted hydrothermal fluid analyzed had a pH of 11.75 (Fig. 2a), and high concentrations of hydrogen (>7 mM) and methane (0.9 mM).
Ashadze, a hydrothermal field that was recently explored for the first time during the French–Russian Serpentine cruise (Fouquet et al., 2008), is one of the deepest active black smoker fields discovered so far (4100 m depth). Ashadze is characterized by an ultramafic rock environment (Charlou et al., 2007; Fouquet et al., 2007, 2008). Several groups of active 1–2 m high chimneys were observed at Ashadze 1 site. The fluid and chimney fragments were obtained from two different active chimneys in a unique group near the SE 2 marker (Fig. 1f). In this study, the maximum temperature measured at Ashadze was 353 °C. The less diluted hydrothermal fluid analyzed had a pH of 4.02 (Fig. 2b) and high concentrations of hydrogen (>10 mM), carbon dioxide (>2.5 mM), iron (7.3 mM) and methane (>0.80 mM).
Archaeal 16S rRNA gene analyses
All the 16S rRNA gene PCR products from the samples were screened by CM-DGGE before cloning in order to estimate the archaeal phylogenetic diversity of each hydrothermal sample and to compare it directly with the seawater diversity (Fig. 2a). Band pattern intensities from all Lost City samples, and from the less diluted hydrothermal fluids, were weaker than for all the other samples, suggesting a lower biomass and/or a high concentration of PCR inhibitors (Fig. 2a). The archaeal seawater CM-DGGE band pattern was different from all the hydrothermal fluid and chimney band patterns (Fig. 2a), suggesting low levels of seawater contamination. The band patterns from hydrothermal samples were mostly composed of DGGE fragments with higher melting points, a probable consequence of the higher GC content of the 16S rRNA gene. The high GC content of these 16S rRNA gene sequences indicates that the Archaea could be hyper/thermophiles (Kimura et al., 2006), as also suggested by the several putative hyper/thermophilic lineages detected in the clone libraries from hydrothermal fluids and chimneys (Archaeoglobales, Methanococcales, Thermococcales, Methanopyrales, Desulfurococcales, Nanoarchaeota, DHVE; Fig. 2b).
After the technical optimization and removal of soluble PCR inhibitors and in order to amplify sufficient PCR product for cloning, archaeal amplifiable DNA from all samples was retrieved by nested PCR. However, insufficient amplified PCR product was obtained for cloning from the less diluted fluid samples (pH 11.75) and from the chimney samples from Lost City. Fourteen different 16S rRNA gene clone libraries were constructed, representing a total of 610 sequences. The coverage values for the 16S rRNA gene clone libraries ranged from 68% to 97%, based on a 97% sequence similarity level (Fig. 2b). On the whole, rarefaction curves were asymptotic for all clone libraries, based on a 95% sequence similarity level, confirming that the sampling effort was sufficient (Supporting Information, Fig. S1).
The overall archaeal diversity analyzed was similar to previous studies (e.g. Brazelton et al., 2006; Flores et al., 2011) and distributed very heterogeneously between the sites (Lost City, Rainbow and Ashadze) and between the types of sample (seawater, hydrothermal fluid, chimney and sediment). The number of OTUs per clone library ranged from 5 to 19, based on a 95% genus level of phylotype differentiation (Schloss & Handelsman, 2004). The Shannon–Wiener index of diversity ranged between 0.63 and 3.04 (Fig. 2b). The archaeal diversity indices of all the samples were in the same range, except for the fluid associated with Lost City, which, as described previously (Schrenk et al., 2004), displayed the lowest detectable diversity (Fig. 2a and b). On average, the hydrothermal samples contained six different lineages, except for Lost City (Fig. 2a and b), which is also in agreement with most published studies on hydrothermal environments (e.g. Takai et al., 2001, 2004b; Nercessian et al., 2003; Schrenk et al., 2003, 2004; Kormas et al., 2006; Page et al., 2008; Nunoura et al., 2010). All the 16S rRNA gene sequences obtained from the clone libraries were assigned to 95 OTUs, based on a 95% sequence similarity level, forming a total of 21 different phylogenetic lineages (Figs 2b and 3a, b). Generally, 16S rRNA gene sequences were related to Euryarchaeota (51%), Crenarchaeota (48%) and Nanoarchaeota (1%). The 16S rRNA gene clone libraries obtained from hydrothermal samples (fluid and chimney) were dominated by sequences related to Euryarchaeota (69%), whereas the majority of sequences in the sediment (92%) and seawater samples (70%) were Crenarchaeota (Fig. 2b). Seven of the 15 Euryarchaeota lineages detected had at least one known cultured representative, and six of these seven had known thermophilic Archaea (Halobacteriales, DHVE2, Archaeoglobales, Methanococcales, Thermococcales, Methanopyrales). Although three of the five Crenarchaeota lineages detected had at least one cultured representative, only one was known to be thermopilic (Desulfurococcales). Moreover, marine group I (MG-I) Archaea had the highest intralineage diversity, with 25 OTUs based on a 95% genus level of phylotype differentiation.
Functional gene clone libraries
Diversity of the mcrA gene
The operon coding for the MCR-I, which includes the McrA subunit, is found in all known methanogens (Reeve et al., 1997). Four mcrA clone libraries were obtained from sediment, fluid and chimney samples from Ashadze and Rainbow sites. Although detected in previous studies (Kelley et al., 2005), no mcrA gene sequences were detected from Lost City samples. The diversity of the four mcrA libraries was limited to sequences related to the H2/CO2 methanogens Methanopyrales and Methanococcales orders (Fig. 4a), matching the 16S rRNA gene clone libraries (Fig. 3a). mcrA gene sequences affiliated to Methanopyrales were only detected at Rainbow. Moreover, the mcrA gene sequences from Rainbow and Ashadze matched the two groups of uncultured methanogenic Archaea previously retrieved from Rainbow (Nercessian et al., 2005).
Diversity of the dsrA gene
Sequences coding for the dsrA gene were only retrieved from Ashadze chimney 1 (Fig. 4b). dsrA gene sequences were previously detected in chimney samples from Lost City (Gerasimchuk et al., 2010) and in sediments from Rainbow (Nercessian et al., 2005). All dsrA gene sequences detected from Ashadze site were related to sequences from marine sediments and East-Pacific Rise hydrothermal vents, a probable consequence of a lack of dsrA gene sequences from the MAR in the databases. dsrA gene sequences were mainly affiliated to sequences from the Desulfobulbaceae family (Fig. 4b).
Diversity of the pmoA gene
The pmoA gene was the most widespread functional gene detected, as a PCR amplification was obtained on eight of the 15 samples tested (Figs 2b and 4c). The phylogeny of the pmoA gene is usually poorly resolved, the bacterial pmoA gene being distantly related to the ammonia monooxygenase subunit A (amoA) (Holmes et al., 1995; Nicol & Schleper, 2006), as revealed by the incongruence between tree topologies performed using different phylogenetic methods. However, two groups of pmoA sequences from Rainbow fluids and Ashadze chimney samples clustered (cluster pmoA 1 and cluster P-A) with sequences related to thermophilic methylotrophs (Inagaki et al., 2003; Hirayama et al., 2007) (Fig. 4c). Moreover, pmoA gene sequences from sediments from Rainbow grouped into two major clusters (cluster pmoA 2 and pmoA 3). Sequences from cluster pmoA 2 did not have any closely related sequences (Fig. 4c).
Community structures and distribution analyses
Although the seawater and the Rainbow sediment CM-DGGE band patterns were quite similar (Fig. 2a), all the sediment clone library community structures were indistinguishable from the combined communities and significantly different (P<0.001) from the seawater (Fig. 2b). Insignificant FST and P tests (P<0.001), based on an analysis at a 97% sequence similarity level, suggested that community structures from all the Rainbow hydrothermal fluids and Ashadze chimney 1 clone libraries were similar and indistinguishable from the combined communities (Fig. 2b). However, although the archaeal community structures from all the Rainbow hydrothermal fluids were also from similar lineage distributions, all the Ashadze chimney and fluid samples had significantly different population structures (P<0.001; Fig. 2a and b). The archaeal diversity of the other clone libraries was also significantly different from the seawater clone library (P<0.001), showing that the hydrothermal vent archaeal communities are probably adapted to their environment. According to pH measurements, the archaeal diversity in the hydrothermal fluids was always the most reduced in the less diluted fluids (Fig. 2b). A correlation was also observed between Methanococcales (P<0.001) and Thermococcales (P<0.05) lineages and the Rainbow fluids. Correlations were also shown between the MG-I lineage and the hydrothermal sediments (P<0.01) and between the unaffiliated Methanosarcinales cluster and the Lost City fluids (P<0.0001).
High diversity of putative chemolithoautotrophs
Overall, the analysis of the phylogenetic data showed a specific distribution of different putative metabolic processes over the different MAR ultramafic-hosted hydrothermal environments that were mainly dominated by putative chemolithoautotrophs.
Putative ammonia-oxidizing Crenarchaeota
MG-I was the most ubiquitous lineage found in the MAR ultramafic-hosted hydrothermal environments, as sequences related to the MG-I Archaea were detected in the majority of clone libraries (93%). Interestingly, the archaeal community structure of the seawater clone library was dominated by sequences related to MG-I (41%) and MG-II (48%), but was significantly different from all the other clone libraries (P<0.001). Takai et al. (2004c) congruently showed that the highest proportion of MG-I members in a hydrothermal environment from the Central Indian Ridge was found in the seawater adjacent to the hydrothermal emissions. MG-I sequences also dominated the sediment 16S rRNA gene clone libraries (≥86%), as commonly found in marine surface sediments (Inagaki et al., 2001; Teske & Sorensen, 2008; Roussel et al., 2009b). The highest diversity indices were also observed in these sediment samples, as a consequence of the very high intraphylum diversity observed within the MG-I. However, the MG-I 16S rRNA gene sequences from the sediment all clustered in different phylogenetic groups from the seawater MG-I, suggesting that specific MG-I communities could be associated with sedimentary environmental conditions (Roussel et al., 2009b). Moreover, the G+C content of the MG-I sequences ranged between 48% and 52%, in contrast to the high-GC content of the 16S rRNA gene sequences from hydrothermal fluids or chimneys, supporting the hypothesis that these MG-I Archaea are probably adapted to the cooler ecological niches of the hydrothermal environments (Kimura et al., 2006; Ehrhardt et al., 2007). Several studies also showed that specific phylogenetic groups of MG-I Archaea appear to be endemic to basaltic crust environments (Ehrhardt et al., 2007; Mason et al., 2007, 2009). Some of these specific subclades of MG-I Archaea could therefore be adapted to various environments, as they were also detected in aerobic and anaerobic basalt enrichment cultures and sediment slurries (Mason et al., 2007; Webster et al., 2010). Hence, as the MG-I Archaea were widespread in our hydrothermal fluid and chimney clone libraries, their presence could be the result of the mixing of ambient seawater with cool niche water in the rocks of the hydrothermal system.
The ultramafic hydrothermal fluids from Rainbow, Ashadze and Lost City were highly enriched in abiogenic methane and hydrogen as a result of serpentinization reactions between ultramafic rocks and seawater (Holm & Charlou, 2001; Charlou et al., 2002; Allen & Seyfried, 2004). Moreover, McCollom (2007) showed that ultramafic-hosted hydrothermal systems could theoretically provide twice as much chemical energy as comparable basaltic-hosted systems. More than half of the archaeal lineages detected from Rainbow, Ashadze and Lost City were related to known cultured species, two-thirds of which were involved in hydrogen- or methane-cycling processes, supporting the theory that these ecosystems could be mainly fuelled by hydrothermal fluids highly enriched in hydrogen and methane.
Methanogenesis was the most common putative hydrogen-oxidizing metabolism detected among ultramafic hydrothermal fluids and chimney samples. Indeed, among all the archaeal lineages in the hydrothermal samples, Methanococcales was the most widespread, as it was detected in a majority of the clone libraries (78%) obtained from hydrothermal fluids or chimney samples. Interestingly, all the sequences related to Methanococcales from Rainbow and Ashadze grouped with sequences previously detected at Rainbow (Nercessian et al., 2005), suggesting that these could be long-term stabilized populations in these chemically slow-evolving environments (Charlou et al., 2002). Methanococcales Archaea are strictly anaerobic autotrophic methanogens, using hydrogen and carbon dioxide or formate as energy sources (Whitman et al., 2001). Strains affiliated to Methanocaldococcus infernus were successfully cultured from hydrothermal chimney samples from Rainbow and Ashadze (C. Jeanthon & S. L'Haridon, pers. commun., respectively). Moreover, the methanogenic potential of Methanococcales at Rainbow and Ashadze was also confirmed by the detection of mcrA genes related to Methanothermococcus thermolithotrophicus (>97% similarity) and M. infernus (>97%). Hyperthermophilic or thermophilic members of the Methanococcales have been commonly cultured and detected using molecular tools from marine hydrothermal vent systems (e.g. Kelley et al., 2002; Nercessian et al., 2003; Schrenk et al., 2003; Takai et al., 2004a; Perner et al., 2007; Page et al., 2008). However, whereas Schrenk and colleagues reported that Methanococcales encompassed a low proportion (<5%) of the hydrothermal prokaryotic communities associated with the walls of a sulfide chimney, Takai and colleagues, on different hydrothermal sites, reported proportions of up to 76.5% (Schrenk et al., 2003; Takai et al., 2004a). These differences could be linked to environmental factors such as the high hydrogen production from these ultramafic systems, which fuel these communities.
Putative hyperthermopilic methanogens were also represented by Methanopyrales. Methanopyrales were rarely detected on the MAR by molecular methods (Flores et al., 2011), probably as a consequence of technical biases or of the restricted number of microbial studies of the MAR, although the first isolated member originates from a hydrothermal system north of Iceland (Kurr et al., 1991). Sequences related to Methanopyrales have rarely been detected elsewhere (Nercessian et al., 2003; Takai et al., 2004a; Ehrhardt et al., 2007; Page et al., 2008). However, in this study, 18 sequences related to Methanopyrus kandleri (>96% similarity) were detected from Rainbow and Ashadze fluids. Interestingly, Takai et al. (2008) also reported recently on an isolate related to M. kandleri capable of methanogenesis with H2/CO2 under elevated hydrostatic pressures and at 122 °C. As mcrA gene sequences related to M. kandleri (88% similarity) were also detected at Rainbow and as strains affiliated to Methanopyrales were successfully cultured from Rainbow and Ashadze (C. Jeanthon & S. L'Haridon, pers. commun., respectively), the Methanopyrales detected were probably capable of methanogenesis, supporting the hypothesis that these sites may harbor large methanogenic communities.
Archaeoglobales, another putative hydrogen-oxidizing archaeal lineage, was also found to be widespread among hydrothermal samples from Rainbow and Ashadze. 16S rRNA gene sequences closely related (>96% similarity) to members of genus Archaeoglobus, Geoglobus and Ferroglobus were retrieved from Rainbow and Ashadze. Interestingly, a new dissimilatory Fe(III)-reducing Archaeoglobaceae was also isolated from Ashadze and reported as growing autotrophically on hydrogen (Slobodkina et al., 2009). As several Archaeoglobales are also iron-cycling Archaea (e.g. Kashefi et al., 2002), the high concentrations of iron (>3 mM) released by acidic ultramafic-hosted hydrothermal environments could possibly fuel specific members of these Archaeoglobaceae communities.
Putative sulfur-cycling and methane-oxidizing communities
Members of the Archaeoglobales lineage also belong to the hyperthermophilic sulfate-reducing Archaea (Miroshnichenko & Bonch-Osmolovskaya, 2006). Interestingly, putative sulfur-cycling Archaea related to Thermococcales and Archaeoglobales lineages were detected in more than half of the clone libraries (56%) obtained from hydrothermal samples (fluid or chimney) and were always detected together, suggesting that they may require similar environmental conditions. However, contrary to all Rainbow hydrothermal fluid samples, Thermococcales and Archaeoglobales at Ashadze were only detected from the Ashadze chimney 1 samples, suggesting that the hydrothermal vents from Ashadze site did not share optimal conditions for putative sulfur-cycling microorganisms. Interestingly, the first obligate piezophilic hyperthermophilic microorganism, Pyrococcus CH1, was also recently isolated from the Ashadze chimney 1 (Zeng et al., 2009). Members of the Thermococcales order are mainly characterized as thermophilic to hyperthermophilic anaerobic heterotrophs that ferment peptides and sugars, and their growth can also be stimulated by sulfur reduction (Miroshnichenko & Bonch-Osmolovskaya, 2006; Zeng et al., 2009). However, some members of the Thermococcales were also able to grow on acetate-utilizing Fe(III) (Summit & Baross, 2001) or are capable of lithotrophic growth on carbon monoxide coupled with hydrogen production (Sokolova et al., 2004), thus matching the environmental conditions of ultramafic-hosted hydrothermal systems. It has also been suggested that Thermococcales and hyper/thermophilic members of the Methanococcales order could inhabit subseafloor ecosystems (Summit & Baross, 1998, 2001; Kelley et al., 2002; Takai et al., 2004a) and could be part of a hydrogen-driven subsurface lithoautotrophic microbial ecosystem (Nealson et al., 2005).
Putative methanotrophic ANME-2 sequences were detected within the methane-cycling communities associated with Rainbow, suggesting the occurrence of anaerobic methane oxidation communities associated with anoxic habitats below 90 °C (Kallmeyer & Boetius, 2004). Interestingly, dsrA gene sequences detected at Ashadze clustered with sequences detected previously in methane-rich hydrothermal systems and related to the Desulfobulbaceae family (Teske et al., 2002; Nercessian et al., 2005), indicating that these putative sulfate-reducing bacteria could be linked to these specific environmental conditions. Besides, some members of Desulfobulbaceae can live syntrophically with ANME-3 members (Niemann et al., 2006). However, although 16S rRNA gene sequences related to ANME-2 Archaea were detected, no ANME-3 Archaea were found. Nevertheless, the most widespread functional gene (pmoA) detected in ultramafic-hosted hydrothermal environments remained related to methanotrophic communities, a probable consequence of the high methane concentration prevailing in these ultramafic-hosted hydrothermal systems. Methanotrophic bacteria were also previously detected in Bathymodiolus species and among the gill chamber of Rimicaris exoculata at the Rainbow hydrothermal field (Duperron et al., 2006; Zbinden et al., 2008), suggesting that these symbionts could also be present in seawater. However, no known sequences related to symbionts were detected in the seawater, a possible consequence of low cell concentrations or of a technical bias. Moreover, the phylogenetic distribution of the pmoA gene was related to the habitat, suggesting that different methanotrophic communities were specifically adapted to different ecological niches (e.g. sediments and fluid/seawater mixing zones).
Specific distribution and ecological niches
The different environmental conditions (temperature, pH, hydrostatic pressure, metabolic substrates) at the different MAR ultramafic-hosted hydrothermal sites generate diverse microbial ecological niches (hydrothermal fluid, chimney, sediment and seawater) that appear strongly selective for specific communities.
Although the molecular techniques (PCR and cloning) used to build clone libraries are known to be inherently biased (Suzuki & Giovannoni, 1996; von Wintzingerode et al., 1997; Polz & Cavanaugh, 1998; Nocker et al., 2007), we assumed that the biases were equal for all samples, as they were analyzed under the same strict conditions (storage, DNA extraction, PCR amplification, cloning, sequencing) (von Wintzingerode et al., 1997). However, comparisons of population structures from other studies using different experimental conditions remain unreliable. For example, archaeal diversity from Rainbow chimneys as described by Flores et al. (2011) was much higher than that described by Voordeckers et al. (2008), suggesting either spatial and temporal heterogeneity or a technical bias. Interestingly, the archaeal diversity observed from the Rainbow fluids in the present study was similar to that from the chimneys analyzed using pyrosequencing reported by Flores et al. (2011). The present study also shows that some communities seemed to be site-specific and specifically adapted to different ecological niches (e.g. sediments and fluid/seawater mixing zones).
Sequences related to Nanoarchaeota, for example, were only detected in the Rainbow hydrothermal system, showing that the nanoarchaeal habitat extends to at least one of the deep hot marine hydrothermal systems of the MAR. The recently discovered novel Nanoarchaeota phylum has shown a wide distribution in high-temperature ecosystems (Hohn et al., 2002; Huber et al., 2002) and may represent pioneering communities in deep-sea hydrothermal vents (McCliment et al., 2006). Nanoarchaeota could also represent a fast-evolving euryarchaeal lineage related to Thermococcales (Brochier et al., 2005). Moreover, the nano-sized Nanoarchaeota were previously described to have a symbiotic relationship with Ignicoccus hospitalis, a member of the Desulfurococcales order isolated from the Kolbeinsey Ridge, north of Iceland (Paper et al., 2007). Interestingly, 16S rRNA gene sequences with 94% similarity to the hyperthermophilic chemolithoautotrophic sulfur- and hydrogen-utilizing I. hospitalis were also retrieved exclusively from the same Rainbow hydrothermal fluids, suggesting that a similar symbiotic relationship could occur between the Nanoarchaeota and specific Desulfurococcales from Rainbow.
Differences between the composition of archaeal communities associated with the two hydrothermal chimneys from Ashadze could probably be linked to environmental factors, as the Ashadze chimney 3 has a higher copper concentration than chimney 1 (J.L. Charlou, J.P. Donval & C. Konn, unpublished data). Moreover, 16S rRNA gene sequences related to Halobacteriales were only detected from Ashadze, which is to date the deepest known hydrothermal site. The highest similarity to a cultured relative was Natronomonas pharaonis (98%), an extremely halo-alkaliphilic archaeon. The occurrence of halotolerant prokaryotes in hydrothermal environments, growing at higher NaCl concentrations than most marine microorganisms, has been reported (Takai et al., 2001). Because of phase separation, it is known that venting of a condensed vapor phase with low salinity will generate a high-salinity phase at depth. This phase may be venting later or be trapped in the subsurface environments. In addition, some authors have suggested a double-diffusive hydrothermal system where brines are trapped in the deepest part of the system and only exchange heat with the upper convective system (Bischoff & Rosenbauer, 1989; Fouquet et al., 1993). If partially cooled, this deep high-salinity reservoir may constitute an extensive location for halotolerant prokaryotes. Hence, it was suggested that these communities could be associated with a subvent ecosystem as well as with hydrothermal chimneys (Kaye & Baross, 2000; Takai et al., 2001).
As described previously, the off-axis Lost City hydrothermal system is remarkable due to its geological, geochemical and biological settings (Kelley et al., 2005). The archaeal diversity associated with hot and very alkaline Lost City hydrothermal fluid was limited to unaffiliated Methanosarcinales and to MG-I sequences. The unaffiliated Methanosarcinales sequences detected matched the Lost City Methanosarcinales cluster (99% similarity) described by Schrenk and colleagues, suggesting that these Archaea were involved in methane-cycling processes (Schrenk et al., 2004; Boetius, 2005). However, no mcrA gene sequences were detected at Lost City, probably due to the low cell densities in the Lost City fluids (Brazelton et al., 2006). Members of the Lost City uncultured Methanosarcinales cluster are probably endemic communities associated with cooler (<95 °C) and very alkaline habitats, as they were not detected from any other hydrothermal sites. The occurrence of molecular genetic evidence in hot and very alkaline fluids also suggests that the Lost City Methanosarcinales have physiological potentials beyond the capacities of any known cultured isolates (Mesbah & Wiegel, 2008).
Specific ecological niches
To summarize, besides some specific Archaea that seemed endemic to some hydrothermal sites, the distribution of archaeal phylotypes and putative metabolic processes was linked to different microbial niches (seawater, sediments, macrofaunal communities, hydrothermal chimneys and fluids; Fig. 5). The cold and oxygenated seawater (<10 °C) overlaying the hydrothermal systems probably represents one of the largest microbial niches, and was characterized only by marine group lineages, some of which could be aerobic ammonia oxidizers. These psychrophilic seawater communities surrounding hydrothermal vents are most likely to benefit from high ammonium inputs from the chemolithotrophic primary producer associated with the hydrothermal structures. Another large ecological niche is probably represented by the cold (<10 °C) and porous sediments surrounding the hydrothermal systems, which may constitute a stable environment and a suitable substrate for the selection of specific seawater phylotypes (MG-I) and for colonization by specific psychrophilic unaffiliated Euryarchaeota and methylotrophic bacteria. The specific sedimentary microbial communities could be fuelled by the products from organic matter degradation, but also by methane seepage from these ultramafic systems. In contrast, the warm (∼15 °C) and relatively unstable mixing zones colonized by macrofaunal communities were probably the most metabolically active microbial niche, benefiting from oxidized seawater compounds and from reduced compounds from the hydrothermal system. Mixing zones between the adjacent ecological niches also occur as a result of the steep physico-chemical gradients characterizing these dynamic hydrothermal environments, therefore resulting in exchanging microbial communities. Mesophilic to thermophilic methane-oxidizing bacteria could dominate the moderate oxic habitats in the mixing environment, as revealed by the pmoA gene analyses. The detection of ANME-2 members suggests that moderate thermophilic (<90 °C) anaerobic methanotrophs could occur in probably restricted anoxic habitats, as a consequence of the very steep oxygen and temperature gradients. Methanotrophic archaeal communities fuelled by hydrogen and carbon dioxide probably dominate the more chemically reduced zones of this niche, which is closer to the hydrothermal chimney. The thermophilic communities composed of Methanococcales, Methanopyrales, Thermococcales, Archaeoglobales and Desulfurococcales were in all likelihood harbored by the hydrothermal chimneys and could be composed mainly of hydrogen-oxidizing members. Although hydrothermal fluids from ultramafic systems such as Rainbow do not have significant levels of hydrocarbons from biogenic origin, methanogenesis could still be the dominant archaeal metabolic process, as the high abiogenic methane concentration may mask the biogenic methane. Moreover, the hyper/thermophilic methanogen order of Methanococcales and Thermoccocales could be typical members of the hot anaerobic microbial ecosystem that could extend below the Rainbow hydrothermal system seafloor.
We thank Anne Godfroy, Françoise Gaill, Yves Fouquet and Georgy Cherkashov, chief scientists of Exomar, MoMARDREAM-Naut and SERPENTINE cruises. All crew members and the Scientific Party were crucial in this effort, especially the ROV Victor 6000 and DSV Nautile crews for the sampling efforts. This work was supported by the Université de Bretagne Occidentale, the Institut français de recherche pour l'exploitation de la mer, the Centre National de la Recherche Scientifique, the Région Bretagne and MoMARnet (Monitoring deep-sea floor hydrothermal environments on the MAR: a Marie Curie Research Training Network). We are also grateful to Ouest-Génopole for the use of their facilities. E.G.R. was supported by a grant from the Ministère de la Recherche.