Novel uncultured Epsilonproteobacteria dominate a filamentous sulphur mat from the 13°N hydrothermal vent field, East Pacific Rise


  • Hélène Moussard,

    1. Laboratoire de Microbiologie des Environnements Extrêmes, Centre National de la Recherche Scientifique, IFREMER, Université de Bretagne Occidentale, Institut Universitaire Européen de la Mer, Plouzané, France
    Search for more papers by this author
  • Erwan Corre,

    1. FR 2424, Station Biologique, Roscoff, France
    Search for more papers by this author
  • Marie-Anne Cambon-Bonavita,

    1. Laboratoire de Microbiologie des Environnements Extrêmes, Centre National de la Recherche Scientifique, IFREMER, Université de Bretagne Occidentale, Institut Universitaire Européen de la Mer, Plouzané, France
    Search for more papers by this author
  • Yves Fouquet,

    1. Département de Géosciences Marines, IFREMER Centre de Brest, Plouzané, France
    Search for more papers by this author
  • Christian Jeanthon

    1. Laboratoire de Microbiologie des Environnements Extrêmes, Centre National de la Recherche Scientifique, IFREMER, Université de Bretagne Occidentale, Institut Universitaire Européen de la Mer, Plouzané, France
    Search for more papers by this author

  • Editor: Michael Wagner

Correspondence: Christian Jeanthon, UMR 7144, Equipe Phytoplancton Océanique, Station Biologique, Place Georges Teissier, 29682 Roscoff Cedex, France. Tel.: +33 298 292 379; fax: +33 298 292 324; e-mail:


Rapid growth of microbial sulphur mats have repeatedly been observed during oceanographic cruises to various deep-sea hydrothermal vent sites. The microorganisms involved in the mat formation have not been phylogenetically characterized, although the production of morphologically similar sulphur filaments by a Arcobacter strain coastal marine has been documented. An in situ collector deployed for 5 days at the 13°N deep-sea hydrothermal vent site on the East Pacific Rise (EPR) was rapidly colonized by a filamentous microbial mat. Microscopic and chemical analyses revealed that the mat consisted of a network of microorganisms embedded in a mucous sulphur-rich matrix. Molecular surveys based on 16S rRNA gene and aclB genes placed all the environmental clone sequences within the Epsilonproteobacteria. Although few 16S rRNA gene sequences were affiliated with that of cultured organisms, the majority was related to uncultured representatives of the Arcobacter group (≤95% sequence similarity). A probe designed to target all of the identified lineages hybridized with more than 95% of the mat community. Simultaneous hybridizations with the latter probe and a probe specific to Arcobacter spp. confirmed the numerical dominance of Arcobacter-like bacteria. This study provides the first example of the prevalence and ecological significance of free-living Arcobacter at deep-sea hydrothermal vents.


At deep-sea hydrothermal vents, steep temperature and chemistry gradients created by the mixing of anoxic hydrothermal fluids with oxygenated seawater are known to offer a variety of habitats for the growth of diverse microorganisms. The microbial communities occupy both aerobic and anaerobic environments having in situ temperatures ranging from ambient seawater temperature to temperatures greater than 110°C (Karl, 1995). Cultivation of metabolically and phylogenetically diverse thermophilic and hyperthermophilic bacteria and archaea has been extensively reported from the hottest ecological niches of the hydrothermal environment (Jones et al., 1983; Harmsen et al., 1997a; Takai et al., 2001; Huber et al., 2002). In part, this is because black smokers and related sulphide structures are the most extensively studied vent environments. Microorganisms of other thermal classes have however been isolated from these heterogeneous and fluctuating habitats although most of the microbial diversity has only been characterized by molecular methods (Takai & Horikoshi, 1999; Jeanthon, 2000; Reysenbach et al., 2000; Teske et al., 2002; Nercessian et al., 2003).

Microbial mats growing on rocks, chimneys, sediments or animal surfaces which are exposed to vent waters represent another hydrothermal vent habitat with associated microbial communities (Karl, 1995). Most of the reports on microbial mats involved microscopic observations (Jannasch & Wirsen, 1981; Nelson et al., 1989; Taylor et al., 1999), or fatty acid (Hedrick et al., 1992) and phylogenetic analyses (Moyer et al., 1995; Longnecker & Reysenbachach, 2001; Alain et al., 2004). In certain locations, the mat-forming organisms were identified as large filamentous sulphur-oxidizing bacteria of the genera Beggiatoa and Thiothrix (Jannasch & Wirsen, 1981; Jannasch et al., 1989; Nelson et al., 1989). These bacteria are chemoautotrophic; they fix CO2 by oxidizing hydrogen sulphide present in hydrothermal fluids. Flocculent material entrained out of the seafloor in the venting fluids and rapid growth of white filamentous bacteria have also often been observed after recent eruptive events or at sites of extensive emission discharges at Loihi hydrothermal vents, Guaymas Basin, on the East Pacific Rise (EPR) and Juan de Fuca ridge (Nelson et al., 1991; Haymon et al., 1993; Embley et al., 1995, 2000; Moyer et al., 1995; Shank et al., 1998; Taylor et al., 1999). Subsequent studies of the emitted particles showed that they were composed of inorganic, sulphur-rich filaments with a low organic carbon content (Nelson et al., 1991). The associated microorganisms have never been isolated or phylogenetically characterized. However, the formation of filamentous sulphur of morphology nearly identical to the material discharged from diffuse-flow hydrothermal vents was documented for a sulphide-oxidizing bacterium isolated from shallow coastal marine waters (Taylor & Wirsen, 1997; Taylor et al., 1999). An extensive characterization revealed that this motile vibrioid autotrophic organism clustered with members of the genus Arcobacter within the Epsilonproteobacteria (Wirsen et al., 2002).

Molecular examination of mats attached to chimneys, animal surfaces and various sampling devices also revealed microbial communities dominated by a wide diversity of uncultured groups of Epsilonproteobacteria (Polz & Cavanaugh, 1995; Cary et al., 1997; Corre et al., 2001; Longnecker & Reysenbachach, 2001; Huber et al., 2003; Lopez-Garcia et al., 2003; Alain et al., 2004). Although there is generally no evidence available that they are also numerically abundant in situ, it is now recognized that the Epsilonproteobacteria are the predominant bacteria in the global deep-sea hydrothermal systems (Polz & Cavanaugh, 1995; Cary et al., 1997; Corre et al., 2001; Longnecker & Reysenbachach, 2001; Huber et al., 2003; Lopez-Garcia et al., 2003; Alain et al., 2004). Genes for key enzymes of the reductive tricarboxylic acid (rTCA) cycle, e.g. aclB– the β subunit of the ATP citrate lyase, were detected in cultured Epsilonproteobacteria and recovered in various hydrothermal vent habitats (Campbell et al., 2003; Campbell & Cary, 2004; Hügler et al., 2005; Takai et al., 2005). These reports suggested that autotrophic CO2 fixation via the rTCA cycle operate in diverse Epsilonproteobacteria and might sustain the predominant primary production in situ.

In this study we investigated the bacterial community from a white filamentous mat formed within a few days on a syntactic foam float associated with an in situ collector deployed at the 13°N hydrothermal site (EPR). The goal in this study was to determine the microbial compositions of the mat that was morphologically comparable to that observed and described by Taylor et al. (1999). On the basis of molecular phylogenetic surveys based on 16S rRNA and aclB genes and quantitative fluorescence in situ hybridization (FISH), we demonstrate the actual in situ dominance of yet uncultured Epsilonproteobacteria including the prevalence of Arcobacter relatives in the hydrothermal sulphur mat.

Material and methods

Hydrothermal site description

The hydrothermal sample was collected at the 13°N hydrothermal vent field on the EPR during the oceanographic cruise PHARE conducted with the research vessel ‘L'Atalante’ and the remoted operated vehicle (ROV) ‘Victor 6000’. The portion of the EPR situated at 13°N is about 300 km south of the Orozco fracture zone and 100 km north of a small transform fault located at 11°49′N. At 13°N, the dome-shaped ridge has a central rift valley varying between 200 m and more than 600 m in width and with a mean depth of about 2600 m (Hekinian & Fouquet, 1985). The volcanic activities at 13°N give rise to a wide range of hydrothermal venting structures from diffuse vents (up to 100°C) to black smokers with temperatures above 350°C. Our study focuses on the PP12 chimney of the Genesis hydrothermal site (12°48′66″ N, 103°56′44″ W). The chimney was colonized by dense populations of the polychaetous annelid Alvinella pompejana that builds tubes directly in contact with the sulphides. Few specimens of the tube-worm Riftia pachyptila were also observed.

Sample collection

An in situ collector (Nercessian et al., 2003), designed to concentrate the microorganisms discharged by hydrothermal emissions, was deployed on the PP12 chimney (Fig. 1). The in situ collector was brought to the deployment site into an insulated container and settled down by the arm of the ROV. Just before deployment of the in situ collector, the chimney's surface was scraped by the arm of the ROV during the sampling of Alvinella pompejana specimens. The mechanical removing of the overlying sulphide structure resulted in an increase of the vent emission. The collector was settled at the exact location where alvinellids were taken, a few centimeters above the vent emission (Fig. 1). Immediately after deployment and prior to removal, discrete temperature measurements were achieved at the bottom and at the top of the collector using the thermoprobe (Micrel, Hennebont, France) held by the teleoperated arm of the ROV. Video imagery of the in situ collector was acquired during successive dives. The collector was surveyed during the following Victor dives. At the fourth day of deployment, a thick and dense white biofilm was observed on the syntactic foam float attached to the collector (Fig. 1).

Figure 1.

 Photographs of the in situ collector deployed for 5 days at Genesis vent site. (a) Deposition of the collector near a Riftia pachyptila colony; the syntactic foam float (FL) is attached to the collector; white tubes of Alvinella pompejana (AL) are visible around the vent orifice (VE); bar, 10 cm (copyright Ifremer/PHARE 2002). (b) Recovery of the in situ collector after 5 days of incubation; the dense white biofilm formed on the syntactic foam float was scrapped once onboard.

After 5 days of exposure, the in situ collector was transferred into enclosed containers to minimize contamination from surrounding seawater during transportation to surface. Once onboard the ship, the microbial mat covering the marker was aseptically collected by scraping and immediately subsampled. Material for scanning electron microscopy (SEM) was fixed with 2.5% glutaraldehyde in 30 g L−1 sea salts (Sigma) and stored at 4°C. Aliquot for molecular and X-ray diffraction (XRD) analyses was stored in ethanol at −20°C. For FISH experiments, biomass was fixed for 3 h at 4°C with 4% paraformaldehyde in 3 × phosphate-buffered saline (PBS) (Sambrook et al., 1989).The fixation buffer was then replaced by a mixture of equal volumes of 3 × PBS and 100% ethanol and the fixed sample was stored at −20°C until further use.

Scanning electron microscopy coupled to energy-dispersive X-ray spectroscopy

Macroscopic mat filaments were deposited on a 10-μm-pore-size membrane filter (Millipore) and fixed overnight at 5°C in 2.5% formaldehyde. After salt was eliminated, the preparation was dehydrated in increasing ethanol concentrations (50%, 70%, 90% and 100%). Samples were then critical point dried and gold coated. Observation was carried out with a Philips (XL 30-LaB6) scanning electron microscope operated at 25 kV.

The elemental composition of the sample was analyzed by energy-dispersive X-ray spectroscopy (EDS) coupled to the scanning electron microscope.

X-ray diffraction

XRD analysis was accomplished with a Brucker D-500 goniometer coupled with the automated DACO-MP system for data acquisition. Mineral phases were identified on the diffraction spectra using the EVA program. Powder XRD scans were performed on dried crushed samples under the following conditions: Cu Kα radiation, 25 mA, 35 kV, and 5 s per step with a step size of 0.02°(2θ). The scan range for all samples was 5–70°. Background and Kalph2 lines were removed from the diagrams. The mineralogical composition of the sample was determined by comparing sample diffraction patterns to mineral standards provided by the JCPDS files.

Nucleic acid isolation and purification

The sample was centrifuged at 4000 g for 10 min. The pellet was suspended in 150 μL of homogenization buffer [50 mM Tris (pH 8), 25% sucrose]. Cells were lysed at room temperature with 1 mg mL−1 lysozyme for 10 min, followed by 10 min of incubation after addition of 1% SDS. Then, proteinase K (1 mg mL−1) were added to the lysis suspensions and the tubes were incubated at 50°C for 60 min. Nucleic acid extraction was performed with prewarmed (55°C) phenol–chloroform–isoamyl alcohol mixture as previously described (Teske et al., 2002). Nucleic acids were precipitated with NaCl 5 M (0.1 vol) and 100% ethanol (2 vol). After overnight incubation at −20°C, the precipitated DNA was centrifuged (10 000 g for 30 min) and washed with 75% ethanol. Nucleic acids was dried and suspended in 50 μL of sterile water.

PCR amplifications and cloning

Small-subunit rRNA genes were amplified by PCR using universal primer 1492R (5′-CGGTTACCTTGTTACGACTT-3′), Bacteria-specific primer 8F (5′-AGAGTTTGATYMTGGCT CAG-3′), and Archaea-specific primer 4F (5′-TCCGGTTGA TCCTGCCRG-3′). The final 25 μL PCR mixture consisted in 1 × buffer (Promega), 1.5 mM MgCl2, 0.25 mM of each dNTP, 0.2 μM of each primer, 1 U of Taq DNA Polymerase (Promega) and 2 μL of DNA. Contaminants in PCR reaction mixtures were avoided as previously described (Goldenberger & Altwegg, 1995). PCR cycles were performed on a Perkin Elmer 9700 thermal cycler. After a denaturation step at 95°C for 5 min, the following conditions were repeated for 30 cycles: denaturation at 95°C for 1 min, annealing at 53°C for 1 min, and chain extension at 72°C for 2 min. A final extension step of 8 min at 72°C was performed. Under these conditions, no PCR products were visible on 0.8% (w/v) agarose gels stained with ethidium bromide. To obtain visible PCR products, primary amplifications using primer pairs 8F-1492R and 4F-1492R were stopped at 15 cycles. Aliquots (1 μL) of the primary amplifications were then used as templates for secondary amplifications. Nested PCRs were performed using forward primers 50F (5′-AA CACATGCAAGTCGAACG-3′) and 63F (5′-CAGGCCTAAC ACATGCAAGTC-3′) specific for Bacteria and 341F (5′-CCT AYGGGGYGCASCAGGCG-3′) specific for Archaea with universal reverse primer 1407R (5′-GACGGGCGGTGWGT RCAA-3′). The PCR mixtures and conditions were as described before except that 30 cycles were applied.

PCR detection of the ATP citrate lyase beta subunit (aclB) was based on previously designed primers (Campbell et al., 2003). The cycling parameters for amplification of aclB genes were that described by Campbell & Cary (2004). Negative controls were performed without DNA template or PCR products.

PCR products were cloned using the TOPO XL PCR Cloning Kit (Invitrogen) into in Escherichia coli TOP 10 cells. Plasmids containing 16S rRNA genes were extracted using the QIAprep Miniprep kit (Qiagen).

Restriction fragment length polymorphism analysis, clone sequencing and phylogenetic analysis

Screening of clones containing 16S rRNA gene sequences was performed by restriction fragment length polymorphism (RFLP) analysis. The PCR products of the clone inserts amplified with primers M13F (5′-GTAAAACGACGGC CAG-3′) and M13R (5′-CAGGAAACAGCTATGAC-3′) were cut with the restriction endonuclease BstU1 (New England Biolabs) according to manufacturer instructions. The DNA fragments were separated by electrophoresis on 2% agarose gels run in 1 × Tris-acetate buffer (Sambrook et al., 1989).

One to three clones per RFLP pattern were sequenced with a 4200 IR2 Licor DNA sequencer, using the Thermo Sequenase Primer Cycle sequencing kit (Amersham Pharmacia Biotech) and the infrared-labeled primers M13F and M13R (MWG, Germany). DNA sequences were submitted to the CHECK_CHIMERA program of the Ribosomal Database Project (RDP) ( Sequences were subjected to BLAST searches within the GenBank database ( to determine 16S rRNA gene sequence similarities to cultured and not yet cultured organisms. Sequence alignments were performed using the clustalw program (Thompson et al., 1994) and refined manually using the seaview program (Galtier et al., 1996).

Phylogenetic trees were constructed using the PHYLO-WIN program (Galtier et al., 1996). Distance trees were constructed using the neighbor-joining algorithms (Saitou & Nei, 1987) with the Kimura two-parameters correction (Kimura, 1980) for nucleotides and PAM correction for proteins. Maximum-likelihood analyses on protein sequences were performed using the PHYML program (Kimura, 1980). The robustness of inferred topologies was evaluated using a bootstrap analysis (Felsentein, 1985). The sequences obtained in this study have been deposited in the GenBank/EMBL/DDBJ database with the following accession numbers: DQ071274DQ071290 (16S rRNA gene) and AM181241–AM181291 (aclB).

16S rRNA oligonucleotide probe design and optimization

The 16S rRNA oligonucleotide probe EPSI682 [S-*-Epsi-0682-a-A-19 according to (Alm et al., 1996)] specific to all Epsilonproteobacterial lineages identified in this study was designed using the PROBE_DESIGN function of the ARB software package ( The probe EPSI682 (5′-CGGATTTTACCCCTACACM-3′) was checked for specificity using the PROBE_MATCH function in the RDP (

Dot-blot hybridization was performed to estimate the specificity of the designed probe. The probe was labeled at its 5′ end with T4 polynucleotide kinase (phosphatase-free; Boehringer Mannheim Biochemicals, Meylan, France) and [γ-32P] ATP (Amersham, Les Ulis, France).

For determination of probe specificity, PCR-amplified 16S rRNA gene fragments from Alvinella pompejana epibiont clone APG13B (Haddad et al., 1995), and environmental clones VC1.2 (Corre et al., 2001) and VC2.2 (Corre, 2000) were used as targets. Total nucleic acids from Sulfurospirillum barnesii (DSM 10660T), Wolinella succinogenes (DSM 1740T), Arcobacter nitrofragilis (DSM 7299T), Campylobacter coli (DSM 4689T), C. jejuni (DSM 4688T), Thiomicrospira denitrificans (DSM 1251T), Helicobacter pylori (ATCC 43579), Zobellia galactanivorans (DSM 12802T), E. coli (DSM 3950) and Bacillus sp. strain SG9 (Marteinsson et al., 1996) were also used.

The probe specificity was determined as previously described (Harmsen et al., 1997b) with slight modifications. Denatured PCR-amplified 16S rRNA gene fragments and total nucleic acids from cultured strains were applied on nylon membrane (Hybond-N+, Amersham Biosciences, Orsay, France) using a dot-blot unit (SRC 96D, MinifoldI, Schleicher and Schuell, Dassel, Germany). The dissociation temperature (Td) was determined for the[γ-32P]ATP-labeled probe and hybridization was carried out as described by (Franks et al., 1998). In addition to the newly designed probe, the 16S rRNA-targeting oligonucleotide probe EUB338 (S-D-Bact-0338-a-A-18) specific for bacteria (Amann et al., 1990) was used as control in the dot-blot analysis. Hybridization responses of the radioactively labeled probe were detected with a Posphor Imager (Molecular Dynamics, Sunnyvale) and printed with Adobe (Seattle, USA) Photoshop 3.0.

Campylobacter jejuni (DSM 4688T), Arcobacter skirrowii (DSM 7302T), Nautilia lithotrophica (DSM 13520T) and Desulfurobacterium thermolithotrophum (DSM 11699T) were used for specificity studies in whole-cell hybridization experiments with the newly designed oligonucleotide probe ARC94 which targets the genus Arcobacter (Snaidr et al., 1997) and the bacterial probe EUB338. Fluorescein-, Cy-3- and rhodamine-labeled and unlabeled oligonucleotides were synthesized and purified by Eurogentec (Seraing, Belgium). Cells were fixed for 3 h in 4% paraformaldehyde in 1 × PBS (3 × PBS for marine strains) at 4°C, washed in 1 × (or 3 ×) PBS and stored at −20°C in 50% (v/v) ethanol-1 × PBS. FISH analysis was performed as previously described (Harmsen et al., 1997b). The optimal formamide concentration of probe EPSI682 was evaluated by varying the formamide concentration from 0% to 60%.

Fluorescence in situ hybridization on the hydrothermal mat sample

To quantify the amount of cells hybridizing with probes EPSI682, ARC94, EUB338 and ARCH945, 250 μL of mat sample fixed in 3 × PBS containing 4% paraformaldehyde were centrifuged 6 min at 3500 g. The pellet was resuspended in 50 μL of the hybridization buffer used above with 250 ng of each labeled probe used and incubated at 46°C for 3 h. Cells were then washed for 30 min with 5 mL of prewarmed (48°C) wash-buffer before being filtered onto 0.2-μm-pore-size Nucleopore polycarbonate filters (Costar, Dutscher, France). The cells were viewed and counted with an Olympus BX60 epifluorescence microscope equipped with a U-MWIB (for fluorescein) and U-MWG (for rhodamine) filter cubes (Olympus). A minimum of 25 cells per field, 15 fields per filter and a minimum of three filters per probe tested were counted.


Habitat characterization of the mat-forming organisms

In order to better understand microbial diversity, dynamics and colonization processes at vents, an in situ collector was deployed on the chimney PP12 at the Genesis site of the 13°N vent field (EPR). The deployment site was selected based on the maximum temperature observed in the venting fluid (<100°C). The collector was placed a few centimeters above the vent orifice within the zone of active mixing of diffuse flow and seawater (Fig. 1a). The temperatures recorded in the shimmering vent fluid near the collector ranged between 11 and 45°C, in the range of, or exceeding, temperatures recorded a few days earlier (Alain et al., 2004). No temperature measurements were performed near the float, but given that the distance between the collector and the float was about 15 cm, we can reasonably assume that average temperatures at which the biofilm formed were below 20°C. Chemical analyses had also previously shown that diffuse-flow fluids of this chimney contained high sulphide levels and negligible amounts of iron (Alain et al., 2004). At the fourth day of deployment, a thick and dense white biofilm visible on video imagery was observed on the syntactic foam float attached to the collector and at the collector's base (data not shown). The collector was recovered after 5 days of incubation (Fig. 1b). The in situ collector had been colonized by several specimens of juvenile and adult Alvinella worms in their freshly secreted organic tubes. The white mat that remained on the marker was collected for further detailed analyses. SEM analysis revealed that the mat consisted in a dense network of cells (mainly rods) embedded within a mucous matrix (Supplementary Fig. 1a and b). Cells were interconnected by the matrix throughout the biofilm.

The elemental microanalysis of the mat was investigated by EDS and XRD. The SEM-EDS analysis indicated that the network of microorganisms was coated with elemental sulphur. XRD diagrams confirmed the dominance of elemental sulphur (with 2θ peaks at about 23.1°, 25.9°, 27.8°, 28.7°, 28.95° and 31.4°) (data not shown). No traces of sulphide minerals, such as pyrite, were detected.

Molecular characterization of the mat-associated microbial communities

The genetic diversity of the microbial assemblages associated with the mat sample was analyzed by constructing 16S rRNA gene. Regardless of the PCR conditions used, no archaeal 16S rRNA genes could be amplified from the DNA extractions. 16S rRNA gene amplification products were obtained using two different bacterial primer pairs. A total of 78 clones containing full-length inserts (59 clones from the library obtained using 50F-1407R and 19 clones from the library generated with primers 63F-1407R) were categorized by RFLP. The clones were then grouped into distinct clusters on the basis of their restriction pattern. Verification of the RFLP clone family groupings was obtained by comparing full 16S rRNA gene sequences of up to three representative clones from each family. Sequences that were ≥97% similar to each other were grouped as a sequence type (Table 1). None of the sequences were identified as chimeras from RDP analysis. Homologous coverage values (Singleton et al., 2001) of the libraries generated using primers 50F-1407R (C‘50F’=93.2%) and 63F-1407R (C‘60F’=94.7%) indicated that a significant part of their 16S rRNA gene diversity was assessed.

Table 1.   Phylogenetic affiliations of bacterial type sequences detected in the hydrothermal mat sample
Sequence type*Other representativesClone numbers in librariesClosest cultivated strains (% identity)
  • *

    Sequence types were identified by RFLP analysis and phylogenetic analysis. All sequence types had generally different RFLP profiles. However, clones L50-WB20 and L63-WB1 displayed the same RFLP profile and were categorized in distinct sequence types on the basis of the phylogenetic analysis. Clones L50-WB6, L50-WB9 and L50-WB56 that had different RFLP types were categorized in one sequence type as their sequence similarity was ≥97%. For the same reason, L50-WB20 and L50-WB49 were assigned to a unique sequence type.

L50-WB6L50-WB9, L50-WB56, L63-WB2, L63-WB64815Arcobacter strain KT0913 (AF235110) (92%)
L50-WB53 1 Arcobacter strain KT0913 (AF235110) (93%)
L50-WB13L63WB1911Epsilonproteobacterial strain 49MY (AB091293) (96%)
L50-WB17 1 Epsilonproteobacterial strain 49MY (AB091293) (94%)
L50-WB20L50-WB495 Epsilonproteobacterial strain 49MY (AB091293) (95%)
L50-WB2 1 Nitratifractor salsuginis (AB175500) (93%)
L63-WB1L63-WB8, L63-WB16 3Sulfurovum lithotrophicum (AB091292) (96%)
L50-WB11L50-WB72 Sulfurimonas autotrophica (AB088431) (98%)

All clone sequence types were affiliated to the Epsilonproteobacteria (Fig. 2; Table 1). A majority of the clones (∼80%) had only 92% sequence identity with an uncharacterized Arcobacter strain isolated from marine surface water (Eilers et al., 2000) and was more closely related (95% similarity) with an environmental clone sequence (AY225610) recovered from a Mid-Atlantic ridge hydrothermal system (Lopez-Garcia et al., 2003). Five sequence types clustered within the group F defined by Corre et al. (2001) (Fig. 2). Three of them (L50-WB13, L50-WB17, and L50-WB20) had clones obtained from the hydrothermal environment as closest relatives (97–99% similarity) (Lopez-Garcia et al., 2002) (K.A. Kormas, M.K. Tivey, K. von Damm, and A. Teske, unpublished data). Their closest cultivated relative was a yet uncharacterized mesophilic microaerophile (strain 49MY) isolated from a yellow mat sample from the Okinawa Trough (Takai et al., 2003). The closest relatives of the two other clone sequences from group F (L-50WB2 and L63-WB1) were an environmental clone sequence obtained from tubes of A. pompejana (M.A. Cambon-Bonavita, V. Riou, K. Alain, V. Cueff, F. Lesongeur, G. Barbier, and J. Querellou, unpublished data) and Sulfurovum lithotrophicum, respectively. The latter organism, recently isolated from a deep-sea hydrothermal sediment core, is a mesophilic chemolithoautotroph capable of sulphur and thiosulphate oxidation and nitrate reduction (Inagaki et al., 2004).

Figure 2.

 Phylogenetic position of the epsilonproteobacterial sequences retrieved from the hydrothermal mat sample (in bold). The tree topology was obtained using the neighbour-joining method. Two ‘Gammaproteobacteria, Escherichia coli and Riftia pachyptila endosymbiont, and Bacillus subtilis were used as outgroup. The reference 16S rRNA gene sequences were obtained from GenBank; all environmental clone sequences presented in the tree have been retrieved from hydrothermal vents. The numbers at the nodes represent the bootstrap values (≥50%) obtained with 500 resamplings. Scale bar indicates the expected number of changes per sequence position. Environmental clone group F and group G are according to Corre et al. (2001) and Takai et al. (2003), respectively. The light grey and dark grey boxes encompass sequences targeted by probes EPSI682 and ARC94, respectively.

The remaining sequence type (L50-WB11) clustered into the family Helicobacteraceae (Corre et al., 2001) and had Sulfurimonas autotrophica as closest relative (98% similarity); this strain was recently isolated from hydrothermal vent sediment and is a microaerophilic chemolithoautotroph capable of growth on sulphur, thiosulphate and sulphide (Inagaki et al., 2003).

To get more insight into the ecophysiological characteristics of microbial components of the mat sample community, the presence and the diversity of aclB genes were evaluated. A total of 51 aclB clones were obtained and they showed high identity level between them (>95% on protein level). All were affiliated to the Epsilonproteobacteria (Fig. 3) and clustered with sequences of Nautilia strain Am-H, Lebetimonas acidophila, and environmental clones from alvinellid tubes (Inagaki et al., 2003) regardless the reconstruction methods used. Although Takai et al. (2005) recognized that phylogenies for the 16S rRNA and aclB genes are incongruent, no significant differences could be made between tree topologies when the 16S rRNA gene sequence dataset matched that of the available aclB sequences (cultured Epsilonproteobacteria) (Nye et al., 2006). This suggests that the aclB sequences retrieved from the mat sample were not that of the Arcobacter-like populations.

Figure 3.

 Phylogenetic position of translated alcB clone sequences retrieved from the hydrothermal mat sample (in bold). Other aclB sequences were obtained from GenBank. The tree topology was obtained using the maximum likelihood method. The numbers at the nodes represent the bootstrap values (≥50%) obtained respectively with 500 resamplings in neighbour-joining and 100 resampling in maximum likelihood method. Scale bar indicates the expected number of changes per amino acid position.

Design and specificity of probe EPSI682

The designed probe EPSI682 (5′-CGGATTTTACCCCTACACM-3′) matched perfectly all the sequences identified in the study (Fig. 2). The specificity of the probe encompassed the 16S rRNA gene sequences of environmental clones and cultured strains of families Campylobacteraceae, Helicobacteraceae and ‘Hydrogenimonaceae’. The probe EPSI682 also matched all the sequences affiliated to the group F defined by Corre et al. (2001), except that of the epibionts of hydrothermal invertebrates (one M : A mismatch at position 682; E. coli numbering). Members of the new family Nautiliaceae (Miroshnichenko et al., 2004) and of group G defined by Takai et al. (2003) contained a T : G mismatch at position 693.

The specificity of radioactively labeled probe EPSI682 was also ensured by optimization of experimental hybridization conditions. The Td value for probe EPSI682 experimentally measured at 46°C was used as the posthybridization washing temperature to test its specificity by dot-blot hybridization of nucleic acids from 27 target and nontarget species and environmental clones (Fig. 4). Confirming the in silico analysis, probe EPSI682 gave positive signals for environmental clones and strains belonging to families Helicobacteraceae and Campylobacteraceae. As expected, no hybridization signals were obtained for members of the family ‘Nautiliaceae’ (Fig. 4c, blots A4 and B4) and nontarget strains (Fig. 4c, line 9).

Figure 4.

 Dot blot analysis of the probe specificities. Hybridization responses of radioactively labeled probes to nucleic acids from 27 target and nontarget DNAs (left) of environmental clones (light grey background) and strains (dark grey background). The blots were hybridized with the following probes: BACT338 (middle) and EPSI682 (right). The layout of nucleic acid samples on the blot is the following: 1A, VC1.2 clone 01; 1B, VC1.2 clone 10; 1C, VC1.2 clone 42; 2A, VC1.2 clone 26; 2B, VC1.2 clone 32; 2C, VC1.2 clone D; 3A, VC1.2 clone 21; 3B, VC1.2 clone 57; 3C, VC1.2 clone 31; 4A, VC1.2 clone 69; 4B, VC2.2 clone 01; 4C, VC2.2 clone 02; 5A, VC2.2 clone 03; 5B, Alvinella epibiont APB13B; 5C, Campylobacter jejuni; 6A, Campylobacter coli; 6B, Geospirillum barnesii; 6C, Arcobacter nitrofigilis; 7A, Thiomicrospira denitrificans; 7B, Wolinella succinogenes; 7C, Helicobacter pylori; 8A, Escherichia coli; 8B, Zobellia galactanivorans; 8C, Bacillus sp. strain SG9).

The specificity of the fluorescently labeled probe EPSI682 was tested by whole-cell hybridization of paraformaldehyde-fixed cells of reference strains. The formamide concentration was optimized using mixtures of target (Arcobacter skirrowii and Campylobacter jejuni) and nontarget strains (Nautilia lithotrophica and Desulfurobacterium thermolithotrophum). The hybridizations were performed simultaneously with either bacterial probe EUB338 or Arcobacter-specific probe ARC94 (Snaidr et al., 1997). The probe ARC94 was also tested as a majority of the phylotypes retrieved in both libraries clustered within the Arcobacter group. The new designed probe showed the brightest fluorescence when hybridization buffers with 20% formamide were used. Only C. jejuni and A. skirrowii cells, specifically as target strains, had positive hybridization signals with probe EPSI682 (Fig. 5). The simultaneous hybridization with probes EPSI682 and ARC94 distinguished Arcobacter cells from that of C. jejuni (Fig. 5c).

Figure 5.

 Whole cell hybridizations of different mixtures of fixed cells hybridized simultaneously with fluorescein-labeled (green fluorescence) and a rhodamine- or Cy3-labeled probe (red fluorescence). Epifluorescence photomicrographs (double exposures) are shown at a magnification of × 1140 (scale bar, 10 μm). Cells that hybridized with both types of probes appeared yellowish or orange on the double-exposed film. (a) Cells of Campylobacter jejuni and Nautilia lithotrophica simultaneously hybridized with probes EPSI682 (green fluorescence) and EUB338 (red fluorescence). (b) Cells of Arcobacter skirrowii and Desulfurobacterium thermolithotrophum simultaneously hybridized with probes EPSI682 (red fluorescence) and EUB338 (green fluorescence). (c) Cells of A. skirrowii and C. jejuni simultaneously hybridized with probes EPSI682 (green fluorescence) and ARC94 (red fluorescence). (d) Cells of A. skirrowii and C. jejuni simultaneously hybridized with probe EUB338 (green fluorescence) and ARC94 (red fluorescence).

Quantification of cells in the mat sample

The probes EPSI682 and ARC94 were used simultaneously with bacterial specific EUB338 or the archaeal specific probe ARCH915 to identify and to quantify the microorganisms in the mat sample. No signals were obtained when paraformaldehyde-fixed cells were hybridized with probe ARCH915. This result and the failure to amplify archaeal 16S rRNA genes from the mat DNA suggest strongly that Archaea were absent, or are present in amounts undetectable by the currently applied methods.

FISH experiments (Fig. 6a and b) revealed that ∼95% of the bacterial cells reacted with probe EPSI682. Cells hybridizing with probe EPSI682 were morphologically diverse. Simultaneous hybridization with probes EPSI682 and ARC94 (Fig. 6c–e) showed that cells of the Arcobacter group represented around 60% of the cells detected by probe EPSI682. The cells that had positive hybridization signals with both probes had the same morphologies.

Figure 6.

 Whole-cell hybridizations of cells from the mat collected on the 13°N EPR with probes EPSI682 (green fluorescence) and EUB338 (red fluorescence) (a–b), with probes EPSI682 (green fluorescence) and probe ARC94 (red fluorescence) (c) and with probes ARC94 (red fluorescence) and EUB338 (green fluorescence) (d and e). Epifluorescence photomicrographs (double exposures) are shown at a magnification of × 1,140 (scale bar, 10 μm).


One of the objectives of the oceanographic cruise PHARE was to identify the microbial communities involved in the first steps of colonization processes on virgin surfaces exposed to the deep-sea hydrothermal vent fluids at the 13°N (Le Bris et al., 2002). Visible structures, macroscopically comparable to that observed at 9°N and Guaymas Basin hydrothermal sites (Taylor et al., 1999) formed in a few days during our colonization experiments. Our goal in this study was to examine the chemical composition of this material and the diversity of the microbial community involved in its formation. The microscopic and chemical analyses of the sample revealed that it consisted in a network of microorganisms embedded in a mucous sulphur-rich matrix. Given that hydrothermal chemical precipitates from the 13°N hydrothermal site do not generally contain such amounts of elemental sulphur (Y. Fouquet, unpublished results), we hypothesized that the sulphur formation had a microbial origin.

Our goal in this study was to determine the diversity of the microbial community involved in the mat formation. We applied the full-cycle rRNA approach, including the construction of 16S rRNA gene libraries and FISH experiments to identify the components of the mat community and to assess quantitatively the dominant microbial populations. The libraries were entirely composed of clones assigned to the Epsilonproteobacteria. These results were in line with many phylogenetic molecular surveys that showed that Epsilonproteobacteria can represent 40–98% of the clone libraries from hydrothermal environments sampled throughout the world's oceans (Haddad et al., 1995; Moyer et al., 1995; Polz & Cavanaugh, 1995; Cary et al., 1997; Reysenbach et al., 2000; Corre et al., 2001; Longnecker & Reysenbachach, 2001; Alain et al., 2002, 2004; Lopez-Garcia et al., 2002, 2003; Teske et al., 2002; Takai et al., 2003; Goffredi et al., 2004). However, contrary to previous studies that showed that the Epsilonproteobacteria of groups F and B (family ‘Hydrogenimonaceae’) were the most frequently retrieved phylotypes (Corre et al., 2001; Alain et al., 2002, 2004; Lopez-Garcia et al., 2002, 2003; Takai et al., 2003; Goffredi et al., 2004), the majority of our sequences grouped into a phylotype related to the genus Arcobacter. This finding was somewhat unusual as no Arcobacter isolates have been reported from deep-sea vents and Arcobacter-related phylotypes account generally for a minor fraction of the sequences obtained from this environment (Lopez-Garcia et al., 2002, 2003; Huber et al., 2003; Alain et al., 2004).

The genus Arcobacter is unique among the Epsilonproteobacteria as it contains validly published species isolated from both animal and environmental sources, including marine habitats (McClung et al., 1983; Vandamme et al., 1992; Mansfield & Forsythe, 2000; Donachie et al., 2005; Houf et al., 2005). Furthermore, a number of potentially novel species have been described from environments such as oil fields, marine waters and sediments (Eilers et al., 2000; Gevertz et al., 2000; Kaeberlein et al., 2002; Wirsen et al., 2002). Arcobacter species are generally known as common human and animal pathogens or free-living organisms that respond poorly in conventional biochemical and growth tests (On, 1996). Typical features of characterized species are optimal growth under microaerophilic conditions at room temperature (Vandamme et al., 1991). Given that the 16S rRNA gene sequence of the microorganisms dominant in the mat libraries shared ∼92% similarity with that of validly published Arcobacter species, we can only speculate about their ecophysiology. However, because XRD analyses identified elemental sulphur as a main component of the mat, it is tempting to suggest that our dominant group might be sulphur-oxidizing bacteria. Consistent with this hypothesis, the ability to oxydize sulphur compounds and to use oxygen as electron acceptor is a common trait in various members of the Epsilonproteobacteria isolated from deep-sea hydrothermal vents (Inagaki et al., 2003, 2004; Takai et al., 2003, 2004; Nakagawa et al., 2005). Although direct oxidation of hydrogen sulphide has been demonstrated for a restricted number of Epsilonproteobacteria isolated from hydrothermal vents (Inagaki et al., 2003; Takai et al., 2003), it seems reasonable to suggest that the mat dominant organisms might also share this metabolic capacity as ‘Candidatus Arcobacter sulfidicus’, which is the only described organism able to excrete sulphur filaments comparable to that of our mat, is a sulphide-oxidizer (Wirsen et al., 2002).

Given that autotrophic CO2 fixation through the rTCA cycle has been recently shown in Epsilonproteobacteria, the mat sample was analysed for the presence and diversity of the ATP citrate lyase gene. Although the recovered aclB environmental clones were also all affiliated to Epsilonproteobacteria, phylogenetic analyses showed that they did not correspond to that of Arcobacter-like populations. It could suggest that the main microbial components of the mat sample may use a different carbon fixation pathway. This hypothesis seems however unlikely as it has been demonstrated that the rTCA cycle operates in a significant number of representative strains from widespread phylogenetic subgroups and physiological types of Epsilonproteobacteria (Campbell et al. 2003; Hügler et al., 2005; Takai et al., 2005). Since the primers used in this study were designed using a limited number of aclB sequences, it is also possible that a range of aclB sequence types have escaped molecular detection. These hypotheses need to be evaluated in the future to clarify the main microbial pathways used in the hydrothermal sulphur mat formations.

The ecological dominance of the Epsilonproteobacteria in the hydrothermal environment has been mostly hypothesized on the basis of their distribution in 16S rRNA gene libraries. However, for a variety of reasons (copy number, bias during DNA extraction, PCR, cloning), the frequency of 16S rRNA gene clone appearance does not reflect the actual microbial community structure (Farrelly et al., 1995; von Wintzingerode et al., 1997). Several studies have reported that biases in PCR amplification could just explain the overwhelming abundance of Epsilonproteobacteria (Corre, 2000; Watanabe et al., 2002). Therefore, quantitative characterization was required to test if this dominance was also reflected in the makeup of the mat-associated populations in situ. Oligonucleotide probes specific to uncultured groups of Epsilonproteobacteria have been designed over recent years (Polz & Cavanaugh, 1995; Cary et al., 1997; Snaidr et al., 1997; Longnecker & Reysenbachach, 2001; Engel et al., 2003). However, some 16S rRNA gene sequences identified in this study were not targeted by the existing probes, justifying the development of a probe with a wider coverage. Although probe EPSI682 demonstrated the actual omnipresence of Epsilonproteobacteria, it is likely that nontargeted members of this lineage (Fig. 2) may also have contributed to the bacterial cells detected with the probe EUB338.

It is interesting to notice that striking differences exist between our results and that of Alain et al. (2004) although both colonization experiments took place on the same diffuse vent for the same time period. The very thin microbial veils that grew on the colonization device deployed by Alain et al. (2004) were macroscopically different from our thick sulphur mats. The absence of Archaea, or their very low representation, is one of the only common findings between both studies. This is probably due to the fact that biofilms had formed in environmental conditions (low temperatures, presence of oxygen) that are not compatible with the lifestyle of most known archaea (Alain et al., 2004). Although Epsilonproteobacteria also prevailed in the clone libraries derived from the veils, the dominant phylotypes belonged to group F and no Arcobacter-related sequences were retrieved. Several indications suggest that the two biofilm communities had settled under distinct environmental parameters (such as temperature, oxygen and sulphide concentrations). Just before the deployment of our collector, the mechanical removing of the overlying sulphide structure resulted in an increase of the vent emission (Fig. 1). This explains why the fluid temperatures measured with the ROV thermocouple and probably the sulphide concentrations to which our collector was exposed were generally higher than that recorded earlier (14–860 μM) (Alain et al., 2004). Consistent with this hypothesis, filamentous sulphur formation by ‘Candidatus Arcobacter sulfidicus’ required high sulphide concentrations (>400 μM) and its sulphide optimum was higher (1.5 mM) than those reported for other sulphur-oxidizers (Wirsen et al., 2002). The capacity to grow optimally and selectively at high sulphide concentrations may also explain how these bacteria can locally out-compete other sulphur-oxidizers as we observed in our sample.

This report further expands the known diversity of the Epsilonproteobacteria from the initial reports of the prevalence of this group at deep-sea vents. Given that the ecological dominance of these microorganisms had been hypothesized on the only basis of their distribution in 16S rRNA gene libraries, we provide here a first demonstration of the numerical abundance of these microorganisms in situ. We showed that bacteria taxonomically different from ‘Candidatus Arcobacter sulfidicus’ are probably involved in filamentous sulphur formation. This process can be ecologically significant in environments where opposing oxygen-sulphide gradients are available at the necessary concentrations and can contribute significantly to the overall organic matter production at hydrothermal vents. It would therefore be significant now to delineate the environmental parameters that govern this efficient sulphide oxidation and the extent of the diversity and distribution of the responsible organisms.


We thank Nadine Le Bris and Françoise Gaill, chief scientist and project leader of the Phare cruise for having inviting us on board. We thank the captain and crews of ‘L'Atalante’ and the pilots and support crew of the ROV ‘Victor’. We are grateful to Philippe Crassous (Ifremer, Brest) and Monique Briand for their valuable contribution in SEM experiments and figure drawings, respectively. We also thank the two anonymous reviewers for their constructive criticisms of an earlier draft of the manuscript. We thank Stéphanie Gouriou (Faculté de Médecine, Brest), Tristan Barbeyron (Station Biologique, Roscoff), Cathy Ragimbeau (CNEVA, Ploufragan), Jan Kuever (Max Plank Institute, Bremen), Barbara Campbell (University of Delaware, USA) for having provided us in cells or DNA of reference strains. HM was supported by a grant from the Ministère de la Recherche.