A white, filamentous microbial mat at the Milano mud volcano in the Eastern Mediterranean Sea was sampled during the Medinaut cruise of the R/V Nadir in 1998. The composition of the mat community was characterized using a combination of phylogenetic and lipid biomarker methods. The mat sample was filtered through 0.2 and 5-μm filters to coarsely separate unicellular and filamentous bacteria. Analyses of 16S rRNA gene sequences amplified from the total community DNA from these fractions showed that similar archaeal populations were present in both fractions. However, the bacterial populations in the fractions differed from one another, and were more diverse than the archaeal ones. Lipid analysis showed that bacteria were the dominant members of the mat microbial community and the relatively low δ13C carbon isotope values of bulk bacterial lipids suggested the occurrence of methane- and sulfide-based chemo(auto)trophy. Consistent with this, the bacterial populations in the fractions were related to Alpha-, Gamma- and Epsilonproteobacteria, most of which were chemoautotrophic bacteria that utilize hydrogen sulfide (or reduced sulfur compounds) and/or methane. The most common archaeal 16S rRNA gene sequences were related to those of previously identified Archaea capable of anaerobic methane oxidation. Although the filamentous organisms observed in the mat were not conclusively identified, our results indicated that the Eastern Mediterranean deep-sea microbial mat community might be sustained on a combination of methane- and sulfide-driven chemotrophy.
Previous studies have shown the presence of dense filamentous microbial mats in various marine environments [1–5]. The populations in these microbial mat communities are sustained through chemotrophy, in which energy is derived from the oxidation of chemical compounds by microorganisms, as light is not available as an energy source. The microorganisms present in these chemotrophy-based ecosystems are usually sulfide- or sulfur-oxidizing, methane-oxidizing and heterotrophic prokaryotes [6–9]. Some members of these communities have previously been characterized using culture-dependent and light- and electron-microscopy methods, and the presence of the filamentous sulfur-oxidizing bacteria Beggiatoa sp., Thioploca sp., Leucothrix, Thiotrix and Desmanthos has been confirmed [2,10–12]. Other studies performed, using culture-independent techniques, have shown that microbial mat communities in vent or cold seep environments can be quite diverse and include prokaryotes such as Epsilonproteobacteria[13,14] or even Archaea , that are responsible for the filamentous structure of the mats. These organisms are thought to participate in either sulfide- or methane-based chemoautotrophy. Most recently, Mills and co-workers investigated the microbial communities in mat-like structures in cold-seep environments in the Gulf of Mexico . This investigation showed a microbial mat community, including bacteria that could play a role in sulfide-based aerobic chemotrophy as well as Archaea, which might play a role in anaerobic methane oxidation.
In the present study, seven mud volcanoes (at 1600–2000 m depth) in the eastern Mediterranean Sea were explored with the Nautile submersible during the “MEDINAUT” cruise of the R/V Nadir in 1998. Their central parts actively seeped methane to the bottom waters and were covered by thick carbonate crust pavements. Microbial aggregates were visually discernible from the Nautile. White, filamentous material, tentatively assumed to contain Beggiatoa aggregates, was collected at the Milano mud volcano .
The aim of the current study was to identify the dominant members of the microbial community in these mats, using 16S ribosomal RNA (rRNA) gene-based methods and lipid biomarker analysis in combination with compound-specific carbon isotope measurements. The integrated use of these techniques enabled us to determine which microorganisms form the dominant members of the mat community, and indicate whether the community was based on sulfide- and/or methane-driven chemoautotrophy.
2Materials and methods
2.1Site description and sampling
Mat filaments were collected using a titanium vacuum bottle (Fig. 1: photograph showing the filaments being sampled with the titanium bottle), operated from the submersible Nautile and support vessel R/V Nadir. Samples were taken from the summit of the Milano mud volcano at 33°44.0′N, 24°46.7′E at a water depth of 1958 m in the Olympi field. Upon return of the submersible to the support vessel, filamentous mat material was transferred to two sterile 10-ml Greiner tubes, sealed, immediately frozen at −80 °C and kept frozen until analysis.
For the isolation of DNA, the mat samples were thawed on ice and mixed well, after which 10 ml was transferred from each tube to a 50-ml sterile Greiner tube. The tube containing the mat material was subjected to mild sonication for 3 min at 100 W in a sonication bath (Branson B1540) filled with water and melting ice. This sonication was repeated twice. Subsequently, the samples were separated into two fractions, by subsequent filtration through 5 and 0.2-μm pore size filters. We assumed that the 5-μm fraction consisted of larger microorganisms (possibly filaments), whereas the 0.2-μm fraction contained smaller prokaryotes, including free-living cells and filament-associated microorganisms that were dislodged during sonication. The two fractions obtained by filtration were designated Milano WF1 5μ and Milano WF2 0.2μ.
Total DNA was isolated using the Wizard Genomic DNA Isolation Kit (Promega Benelux B.V., Leiden, The Netherlands), following a procedure that was slightly modified from that recommended by the manufacturer. The filter for each of the two filter fractions was cut into small pieces using a sterile scalpel and placed in 2-ml Eppendorf tubes. To each tube, 355 μl of 0.1 M Tris/50 mM EDTA (pH 9.0), 40 μl of lysozyme solution (20 mg ml−1) and 15 μl of 20% sodium dodecyl sulfate (SDS) was added. The tubes were incubated for 1 h at 37 °C, shaken at 5 min intervals, and subsequently centrifuged at 16,000g for two min, after which the supernatant was collected for each filter fraction and placed on ice. To the remaining pellet, 600 μl of sterile Milli-Q water was added, followed by incubation at 80 °C for 5 min. After centrifugation at 16,000g for 2 min, the supernatants were recovered, pooled with the previous supernatant for each filter fraction and placed on ice. After this, DNA was extracted from the pellets and supernatants from each filter fraction, using the manufacturer’ s protocol. The extracted DNA was precipitated with 0.8 volume of iso-propanol, collected by centrifugation, and washed with ice-cold 70% ethanol. After centrifugation, the pellet was air-dried, suspended in 100 μl of 10 mM Tris (pH 8.0), and rehydrated overnight at 4 °C. As a final step, the DNA solution was cleaned with a WIZARD DNA cleanup kit (Promega Benelux B.V., Leiden, The Netherlands) and concentrated to a volume of 50 μl. The presence of of high molecular weight DNA was confirmed by gel electrophoresis and the concentration of DNA was measured spectrophotometrically.
2.3Amplification of 16S rRNA genes, cloning and sequencing
The 16S rRNA genes were amplified using primers specific for bacterial and archaeal 16S ribosomal RNA genes. Eubacterial 16S ribosomal RNA genes were amplified using the B8F (5′-AGAGTTTGATCMTGGCTCAG-3′) forward primer  and the universal U1406R (5′-ACGGGCGGTGTGTRC-3′) reverse primer . Archaeal 16S ribosomal RNA genes were amplified with the A2F (5′-TTCCGGTTGATCCYGCCGGA-3′) forward primer  in combination with the universal U1406R primer. For PCR, 1 μl of DNA template was used in 25-μl reactions that contained 10.2 mM Tris, 2.3 mM MgCl2, 50 mM KCl, 2% DMSO, 5 μg BSA, 0.2 mM of each dNTP, 0.2 μM of each primer and 0.5 U Taq DNA polymerase. Samples were amplified in a Perkin–Elmer GeneAmp PCR System 9700 (Perkin–Elmer Applied Biosystems Netherlands, Nieuwerkerk a/d IJsel, The Netherlands) using the following program: 95 °C for 5 min; 35 cycles of 94 °C for 1 min, 57.5 °C for 30 s, 72 °C for 4 min, with a final elongation step of 72 °C for 7 min.
PCR products were purified using QIA quick spin columns (Invitrogen BV, Groningen, The Netherlands) and were cloned in the pGEM-T easy vector system (Promega) using Escherichia coli JM109. Cloned inserts were amplified by PCR using the pGEM-T specific primers T7 (5′-TAATACGACTCACTATAGGG-3′) and SP6 (5′-GATTTAGGTGACACTATAG-3′). PCR mixtures were as described above, with the following PCR conditions: 94 °C for 5 min; 30 cycles of 94 °C for 1 min, 48 °C for 30 s, 72 °C for 4 min, with a final elongation step of 72 °C for 7 min. From each library, clones with inserts were selected and partial 16S rRNA gene sequences were determined using an ABI PRISM 3100 Genetic Analyzer (Perkin–Elmer Applied Biosystems, USA). Either the forward PCR primer or T7 vector primer was used in addition to a U515 (5′-GTGCCAGCMGCCGCGG-3′) forward primer with a Taq DyeDeoxy terminator sequencing kit (Applied Biosystems).
Chimeric sequences were detected by using the chimera detection utility at the Ribosomal Database Project homepage (http://rdp.cme.msu.edu/html/analyses.html). Nearest relatives in Genbank were identified using the basic local alignment search tool (BLAST) from the NCBI website (http://www.ncbi.nlm.nih.gov/blast). Selected sequences and their close relatives were aligned using the fast aligner utility of the ARB software package . Alignments were checked manually using the secondary structure of the 16S rRNA gene. 16S rRNA gene sequences showing more than 97% similarity with each other were considered to belong to the same phylotype. In addition, sequences were divided into phylogenetic groups that were consistent with taxonomically related phyla and orders. These groups were assigned by determining the taxonomic class of the nearest GenBank relative. Sequences that could not be linked to previously identified bacterial or archaeal taxonomic classes were listed as unclassified.
Evolutionary distances were calculated according to the Kimura two-parameter correction method , after which neighbor joining trees were constructed with 1000 bootstrap samplings using Treecon-W . The Shannon–Weaver indices of diversity were calculated for all samples on the basis of the phylotype distribution using the PAST program (http://folk.uio.no/ohammer/past/). Similarities between the two filter fractions on the basis of phylotypes and phylogenetic groups were determined using the Morisita–Horn index of similarity .
16S rRNA gene sequences determined in this study were deposited in Genbank under Accession Nos. AY592799–AY592933.
Filamentous mat material (not sonicated or filtered) was used for the analysis of lipids. A pellet of the filamentous mat material was prepared by centrifugation at 16,000g for 30 min. Lipids were ultrasonically extracted from the pellet using methanol, methanol/dichloromethane (1:1, v/v) and dichloromethane. An aliquot of the total lipid extract was methylated with diazomethane in diethyl ether, and filtered over a small SiO2 column with ethyl acetate as the eluent. Alcohols in the eluate were converted to their trimethylsilyl ether derivatives with bis(trimethylsilyl)trifluoracetamide (BSTFA) in pyridine. Individual compounds were analyzed by gas chromatography (GC) and GC–mass spectrometry (MS). Another aliquot of the extract was used for analysis of glycerol dialkyl glycerol tetraethers (GDGTs) by high performance liquid chromatography/atmospheric pressure chemical ionization mass spectrometry (HPLC–APCI-MS) as described elsewhere .
GC–MS was performed with a HP 5890 gas chromatograph (Hewlett–Packard, Palo Alto, CA, USA), equipped with a CP-Sil5 capillary column (25 m, 0.32 mm i.d., 0.12 μm film thickness) (Chrompack, Middelburg, The Netherlands) coupled to a VG Autospec Ultima Q mass spectrometer (VG, Manchester, UK). The column temperature was programmed from 70 °C to 130 °C at a rate of 20 °C/min and then to 320 °C (15 min isothermal) at 4 °C/min. Isotope-ratio-monitoring gas chromatography–mass spectrometry (irm-GC–MS) was performed using a Finnigan Delta-plus XL-irm-GC–MS system (Finnigan, Bremen, Germany), equipped with an on-column injector and fitted with a 25-m × 0.32 mm fused silica capillary column coated with a 0.12-μm film of CP-Sil5 (Chrompack, Middelburg, The Netherlands). Helium was used as the carrier gas and the oven was programmed as described for the GC analyses. Isotopic values were calculated by integrating the m/z 44, 45 and 46 ion currents of the peaks produced by combustion of the chromatographically separated compounds and those of CO2-spikes produced by admitting CO2 with a known 13C content at regular intervals into the mass spectrometer. Duplicate analyses were carried out and the results were averaged to obtain a mean value. Reported δ13C values are expressed against the Vienna Pee Dee Belemnite standard (VPDB), after correction for the addition of carbon during derivatization, and have an error of less than ±1.0‰.
3Results and discussion
To investigate the microbial community that constituted the white filamentous mat at the Milano mud volcano, the dominant morphotypes were studied with light microscopy, whereas the composition of the microbial community present was characterized using phylogenetic methods. Light microscopy revealed a microbial community that consisted of large filaments, as well as various coccoid- and rod-shaped organisms (see Fig. 1, insert). These results indicated that the microbial community in the mat was more diverse than expected when the mat was sampled. Although the majority of the cells appeared intact, freeze-thawing of the mat material diminished the quality of the sample in such a way that it was not suitable for scanning or transmission electron microscopy. As a result, the details of bacterial cell morphology could not be determined.
3.1Diversity and similarity
The diversity of the Milano mud volcano microbial mat community was studied by cloning and sequencing of 16S rRNA genes. Two filter fractions were obtained by filtration through 5 μm and a 0.2-μm pore size filters to distinguish between larger (filament- or particle-associated) and smaller prokaryotes (free-living microorganisms or prokaryotes liberated by sonication). Taking into account that freeze-thawing of the mat material may have introduced a bias in the relative abundances of the sequences obtained, the results of this study do not necessarily represent the in situ microbial community structure. However, given the amount of intact (filamentous) cells visible during microscopic examination, we assume that our approach identified the most abundant prokaryotes and the filaments, and distinguished filaments from smaller prokaryotes. From each filter fraction, about 45 bacterial 16S rRNA gene sequences were determined. In total, these 90 clones represented 52 distinct bacterial phylotypes and comprised nine bacterial phylogenetic clades (Table 1). A total of 31 and 14 archaeal 16S rRNA sequences were obtained for Milano WF1 5μ and Milano WF2 0.2μ, respectively. These represented eight archaeal phylotypes within three phylogenetic clades (Table 1). Rarefaction analysis of the distribution of phylotypes yielded asymptotic accumulation curves (data not shown), which indicated that the clone libraries represented the most abundant microbial populations in the microbial mat.
Table 1. Phylogeny and distribution of populations found in a deep-sea microbial mat at the Milano mud volcano (Eastern Mediterranean)
The 5-μm and 0.2-μm filter fractions contained distinctive bacterial phylotypes. Only four out of the 52 bacterial phylotypes were common to Milano WF1 5μ and Milano WF2 0.2μ. In contrast, most archaeal phylotypes were found in both samples (Table 1). This apparent difference in the distribution of the microbial populations of Milano WF1 5μ and Milano WF2 0.2μ was quantified by Morisita–Horn indices of similarity. The values showed that there was low to moderate similarity at the phylotype and phylogenetic group levels for the bacterial populations in the two fractions. For the archaeal populations these similarity values were relatively high (Table 1). The Shannon–Weaver diversity values were calculated to determine the level of diversity in the microbial mat community; these values were highest for bacterial phylotypes and relatively low for archaeal phylotypes (Table 1). Comparing both filter fractions, the overall diversity was highest in the 5-μm filter fraction (WF1, Table 1). These data show that the bacterial community of the Milano filamentous mat was more diverse than the archaeal one, with clear differences in the composition of the bacterial populations between filter fractions Milano WF1 5μ and Milano WF2 0.2μ (Table 1).
3.2Microbial mat community analysis and lipid composition
The relatedness of the most abundant sequences in the microbial mat from the Milano mud volcano and previously reported sequences is presented in Figs. 2 and 3. Neither the archaeal nor the bacterial 16S rRNA gene sequences showed high resemblance to sequences from cultivated species, which indicated that the main fraction of the prokaryotic community consisted of novel organisms.
The archaeal 16S rRNA gene sequences were all related to those of members of the Euryarchaeota (Fig. 2) and no sequences related to members of the Crenarchaeota were detected. Most archaeal 16S rRNA sequences from Milano WF2 0.2μ and all sequences from Milano WF1 5μ were distantly related (i.e., <90% sequence similarity) to those of cultured members of the Methanosarcinales, a phylum known to encompass methanogens. The majority of these Methanosarcinales-related sequences formed two distinct clades that were closely (>96% sequence similarity) affiliated with the ANME-2AB and ANME2-C sequences. These sequences have been linked to the anaerobic oxidation of methane (AOM) in previous studies [26–28]. Therefore, our results suggest that the Archaea in the microbial mat could play a role in the oxidation of methane. However, this tentative conclusion could not be fully substantiated by the results of the lipid analyses. The only archaeal lipid identified was GDGT without cyclopentane rings (i.e., GDGT-0). This lipid was found in relatively low amounts compared to bacterial lipids, indicating that Archaea were not abundant members of the microbial community. GDGT-0 is a membrane lipid that is only known to occur in quite disparate groups of Archaea; hyperthermophilic and mesophilic members of the Crenarchaeota, methanogens and euryarchaeota capable of AOM [29–34]. In view of the high numbers of archaeal 16S rRNA gene sequences from this study that were phylogenetically related to (euryarchaeal) ANME-2 sequences (Fig. 2), it would be plausible to attribute GDGT-0 to members of the Euryarchaeota capable of AOM. However, in other studies, the distribution of GDGTs in samples from sites with AOM is always characterized by an approximately equal abundance of three GDGTs: GDGT-0 and GDGTs containing one or two cyclopentane rings [30–34]. Recently, Blumenberg and co-authors have linked GDGTs to the ANME-1 group, whereas sn-2-hydroxyarchaeol and crocetane were most abundant in ANME-2 dominated carbonate reefs . Thus, the GDGT distribution in the microbial mat from the Milano mud volcano differs substantially from the pattern expected for Archaea involved in AOM. One way to determine if the GDGT-0 in the mat originated from Archaea capable of AOM would be to determine the carbon isotopic composition of the biphytane skeletons of GDGT-0. If the GDGT-0 originated from Archaea involved in AOM, a low δ13C value would be expected (cf. [30,31,34]). Unfortunately, the amount of material available was not sufficient to perform this assay. Therefore, direct evidence for the involvement of the ANME-2 related sequences in AOM is lacking.
The majority of the bacterial 16S rRNA gene sequences from Milano WF1 5μ and Milano WF2 0.2μ were affiliated with Alpha-, Gamma-, Delta- and Epsilonproteobacteria (Fig. 3). The most abundant deltaproteobacterial phylotypes in the mat were found in Milano WF2 0.2μ. These deltaproteobacterial sequences were related (>94% similarity) to sequences of known sulfate reducing bacteria (SRB), such as Desulfocapsa sulfexigens, Desulfovibrio sp. and Desulfosarcina sp. In both filter fractions, there were alpha- and epsilon-proteobacterial 16S rRNA gene sequences related to the sulfide-oxidizing bacteria Sulfitobacter sp. (>96% sequence similarity), Thiomicrospira sp. and Sulfurimonas autotrophica (93–95% sequence similarity), hydrocarbon seep bacteria (>96% sequence similarity), the aerobic methylotroph Methyloarcula terricola (<92% sequence similarity) and deep sea sulfide-based chemotrophic Alphaproteobacteria (>95% sequence similarity) [36–38]. In addition, sequences of Gammaproteobacteria related to sulfur oxidizing bacteria (>93–98% sequence similarity), aerobic methylotrophs (>93% sequence similarity) and methanotrophs (>95% sequence similarity) [37,39,40] were present in both filter fractions.
These analyses support the differences between the communities found in the two filter fractions, and suggest that the most abundant bacteria in the microbial communities consisted mainly of chemoautotrophs that use reduced sulfur compounds and/or methane, as well as sulfate-reducing prokaryotes. Despite the pitfalls associated with linking function and phylogeny, we can tentatively assume that functional properties are conserved among these phylogenetically related populations as shown previously [41–47]. Taking this assumption into account, we tentatively conclude that methane oxidation and the production and consumption of reduced sulfur compounds occurred in the Milano microbial mat. This conclusion was, to some extent, corroborated by the lipid analysis, which identified the C16:1, C16:0 and C18:1 fatty acids as the most abundant low molecular-weight lipids. Since these fatty acids are widespread in both the bacterial and eukaryotic domains of life, it was not possible to determine a link to specific bacteria. However, their predominance in the total low molecular-weight lipids provided supporting evidence that bacteria were the dominant members of the microbial communities in the microbial mat. Their stable carbon isotopic compositions showed low δ13C values of −51‰, −43‰ and −52‰, respectively, which do not directly suggest the existence of microbial communities involved in AOM, as more negative values would be expected . The δ13C values are too light for Rubisco autotrophy , but can be explained by assuming the assimilation of a carbon compound partially formed by oxidation of isotopically light (13C-depleted) methane. In addition, these δ13C values support chemo(auto)trophic processes such as aerobic methane oxidation and chemoautotrophic dissolved inorganic carbon (DIC) fixation (perhaps by sulfide-oxidizing bacteria) . The data from the lipid analysis therefore supported the occurrence of chemotrophic communities involved in methane and sulfide oxidation, as was deduced from the prokaryotic sequence data.
3.3Identification of the filamentous bacteria
While the data obtained demonstrate the existence of a diverse microbial community in the microbial mat, the identity of the filamentous organisms observed remains obscure. Individuals viewing the mat from the Nautile submersible in 1998 suggested that the filamentous organisms might be Beggiatoa sp. . Consistent with this presumption, our data indicate that organisms closely related to those previously shown to be present in a cold seep-associated mat  were also present in the Milano microbial mat. These sequences were affiliated with those of a large marine, sulfur-oxidizing Beggiatoa spp. isolated in the Bay of Concepcion, Chile (Teske et al., unpublished results), but constituted only about 5% of the total phylotypes found, as judged from their prevalence in the clone libraries. This could be misleading, because of possible biases in lysis efficiency, 16S rRNA copy number, PCR amplification and other factors [50–52]. Unfortunately, there was insufficient material available to identify the filamentous prokaryotes using fluorescent in situ hybridization techniques . Therefore, we can only speculate on the possible identity of other filamentous organisms, besides Beggiatoa-related sequences. We hypothesize that the relatively abundant 16S rRNA gene sequences in the Milano WF1 5μ filter fraction, which were absent in the Milano WF2 0.2μ sample, represented filamentous organisms. Likely candidates are the Gammaproteobacteria related to sulfide- or methane-oxidizing endosymbionts. Unfortunately, these endosymbionts have no cultured representatives and therefore, the morphology and identity of these “potential” filaments remains unresolved.
Our results show the existence of a diverse deep-sea microbial mat community in which filamentous organisms were not conclusively identified. However, Beggiatoa-related prokaryotic organisms could contribute to the filamentous structure and appearance of the deep-sea mat. Assuming that functional properties are conserved among phylogenetically related populations, we provide evidence for the notion that the deep-sea microbial mat community is based on a combination of methane- and sulfide-driven chemotrophy, both aerobic and anaerobic. This could provide valuable insights for future deep-sea investigations and cultivation attempts.
Samples of microbial filaments were obtained during the French–Dutch “MEDINAUT” expedition, an integrated geological, geochemical and biological study of mud volcanism and fluid seepage in the eastern Mediterranean Sea. We thank the officers and crew of the Nadir R/V and the Nautile submersible and the Medinaut and Medineth Scientific Party for their helpful co-operation during seagoing activities. We thank Maria Schneider and Mayee Wong of the University of Idaho (Moscow, USA) for their help and support during DNA sequencing, and Stephen Bent (University of Idaho, Moscow, USA) for help with Morisita–Horn similarity calculations. We thank two anonymous reviewers and Dick van Elsas (University of Groningen) for their useful comments. Financial support was provided by the Dutch funding organization, NWO-ALW (Project-Grant 809.63.013).