Community architectural distribution

To access the community architectural distribution, a DAPI (4', 6-diamidino-2-phenylindole) staining was applied on the samples S1 (before high-pressure incubation) and S2 (after 286 days high-pressure incubation). Based on DAPI staining cell counts, both single cells and aggregates were commonly observed in S1 and S2. The aggregates had different sizes ranging from 2 to 15 μm in diameter (Ø). In both S1 and S2 single cells were 1-2 orders more abundant than the aggregates (Figure 1A). Among all the aggregates, the ones with diameter from 2 to 5 μm were the most abundant ones (73.35 ± 2.63% in S1 and 73.28 ± 1.75% in S2). Few spherical aggregates bigger than 15 μm were observed in S1 or S2 (less then 4 × 10⁴ aggregates/ml slurry). For some aggregates we observed that it was dividing into two smaller spherical aggregates in both S1 and S2 (data not shown). This was also reported in another enrichment from a semi-continuous bioreactor operated under 1.4 MPa methane pressure [9]. It is an indication that these large aggregates may have reached a "critical size" during growth, which then may disintegrate into smaller aggregates for further growth.

Cell aggregates accounted for the major part of the biovolume (Figure 1B). The middle size aggregates (Ø = 6, 7, 8, 9, 10 μm) contributed for about half of the total biovolume (52.73 ± 9.04% in S1 and 47.02 ± 8.67% in S2). Although the big size aggregates (Ø = 11, 12, 13, 14, 15 μm) had very low concentrations (2.22 ± 0.74 *10⁵/ml slurry in S1 and 4.93 ± 1.56 *10⁵/ml slurry as shown in Figure 1A), they also contributed for large part of the biovolume (26.67 ± 7.83% in S1 and 33.34 ± 8.54% in S2).

Enrichment of total biomass

The total biovolume concentration increased from (1.28 ± 0.06)*10⁹ μm³/ml slurry in S1 to (4.49 ± 0.51)*10⁹ μm³/ml slurry in S2 (Figure 1B). Since the reactor volume was fixed and the biomass washing out during reactor operation was negligible [11], the total biomass inside the reactor increased 2.5 times within 286 days. This reactor system was the first system that was able to accumulate total biomass while maintaining high SR-AOM activity--0.5 mmol sulfide production per day while the reactor was operated at batch mode under 8 MPa methane pressure [11]. In the systems previously reported by other authors, either only specific groups but not the total biomass was quantified [16] or there was major loss of biomass due to sampling and decay [9, 10].

The biovolume data was converted into cell dry weight for a comparison with VSS (Volatile Suspended Solids) data. Taken the same assumption as described by Nauhaus et al. [9], there was about 0.2 g cell dry weight/ml biovolume in the sediment sample with SR-AOM activity. Therefore the biomass concentration in the high-pressure bioreactor increased from 0.3 (g cell dry weight/l slurry) in S1 to 0.9 (g cell dry weight/l slurry) in S2. However, this value was one order lower compared to the 8 g/l of VSS (based on weight difference between drying sample at 105°C and at 650°C) as reported by Zhang et al. [11]. One possibility is that the assumption 0.2 g cell dry weight/ml biovolume was based on analysis of two strains of small marine microorganism [9, 17], which could be not representative of the cells enriched in the reactor. Another possibility would be the extracellular polymeric substances (EPS) contributed large part of VSS. For example, for granular microbial aggregates enriched in an OLAND (oxygen-limited autotrophic nitrification-denitrification) reactor, as much as 50-80% of the space occupied by bacteria was constituted of EPS [18]. For the deep-sea sediment, the presence of EPS has been reported both from in situ sediment and in vitro enrichments at different locations [9, 19]. However whether the production of EPS was stimulated during high-pressure incubations and what was the mechanism behind still needs to be further investigated.

Community structure

The CARD-FISH result showed that a large part of biomass in S1 and S2, especially single cells, did not belong to ANME or SRB. There was growth of other unknown microbes within a mixed community of ANME/SRB. Therefore a clone library analysis was performed on S2 to approach to the complete archaeal and bacterial communities. Archaeal community had extremely low diversity, where ANME-2a and MBG-D (marine benthic group D) were the only two groups of archaea detected. ANME-2a was the dominant, which accounted for 88% of the archaeal community (Figure 2). No 16S rRNA gene from ANME-3 was detected. The absence of ANME-3 in the archaeal clone library was contradictory to CARD-FISH result. The size of the clone library was not large enough to detect the rare ANME-3 or the hybridization experiment may have led to mis-hybridization, thus giving false positive signal. Dissimilar from archaeal community, the bacterial community was highly diverse (Figure 3). Gammaproteobacteria (43%) were the most dominant followed by the Deltaproteobacteria (17%), which includes the SRB. Among total bacteria population in S2, 8% was belonging to SEEP-SRB1a subgroup of Deltaproteobacteria, which were found to be specifically associated with ANME-2a in other enrichments mediating SR-AOM process [20]. Most of the Gammaproteobacteria found in the community were closely related to Methylophaga sp. and Methylobacter sp., which are known to use reduced one-carbon compounds, such as methane, methanol or dimethylsulphide [21]. The presence of such bacteria in our anaerobic reactor is intriguing since methane and sulphate were the only electron donor and acceptor supplied. The presence and even production of sulphide (sulphide concentration increased up to 0.5 mM everyday in the reactor) was an indication of anaerobic condition inside the reactor. However we cannot exclude the possibility of a limited amount of dissolved oxygen in the reactor influent, which could explain the presence of aerobic. Further tests need to show if these Gammaproteobacteria are playing an important active role in the reactor.

Enrichment of ANME-2 and SRB

CARD-FISH results showed that percentages of ANME-2 and SRB biovolume increased from 13.4 ± 4.2% and 22.7 ± 5.3% in S1 to 50.4 ± 15.9% and 60.6 ± 5.5% in S2 (Table 2). By combining with the total biovolume data from DAPI staining (Figure 1B), the biovolume of ANME-2 in S1 was:

(1.28*10⁹ μm³/ml slurry) * 13.4% = 1.7*10⁸ μm³/ml slurry

The biovolume of ANME-2 in S2 was:

(4.49*10⁹ μm³/ml slurry) * 50.4% = 2.3*10⁹ μm³/ml slurry

Therefore after 286 days incubation, the ANME-2 population increased for 12.5 times. Following the same method of calculation, the SRB population increased for 8.4 times after 286 days incubation in this high-pressure bioreactor. The populations of ANME-2 and SRB both increased faster than the total biomass, which indicated that ANME-2 and SRB were selectively enriched in the system. This selective enrichment of ANME-2 and SRB was another proof that the incubation condition inside this high-pressure bioreactor was favourable for SR-AOM community.

Incubation condition

In a previous study, the sediment sample originally from Captain Arutyunov Mud Volcano (Gulf of Cadiz, North East Atlantic) was diluted 12 times with artificial sea water medium and incubated in a continuous high-pressure bioreactor at 15°C [11]. This bioreactor system was a simulator for cold seep ecosystems, where sulphate and high-pressure methane were supplied. Because the high apparent affinity for methane (37 mM) in SR-AOM reaction and low dissolubility of methane in seawater (1.3 mM at 15°C at ambient pressure), it is necessary to supply high pressure methane to obtain high concentration of dissolved methane which can be directly used by microorganisms for high in vitro SR-AOM activity [11]. During this research, the reactor was operated in a fed-batch mode or a continuous mode. When it was in fed-batch mode, the methane pressures were switched between 1, 4.5 and 8 MPa. When it was in continuous mode, the methane pressure was either 1 or 8 MPa and the flow rate was 0.1 ml/min (HRT 100 hours). The SR-AOM activities under different operational conditions have been described previously [11]. To take a slurry sample, the incubation vessel was open under a nitrogen atmosphere and manually stirred to make the slurry sample homogeneous. The slurry samples before (S1) and after (S2) 286 days incubation were fixed in 4% formaldehyde and stored at 4°C for cell staining. Additional slurry from S2 was stored at -20°C for DNA extraction and clone library analysis.

Cell and aggregates quantification

To assess the number and the size of cells and aggregates, DAPI (4', 6-diamidino-2-phenylindole) staining was performed on S1 (after 2000 times dilution) and S2 (after 5600 times dilution). Subsequently, the samples were filtrated onto a circular GTTP polycarbonate filter (0.2 μm, Millipore, Germany) with a diameter of 2.5 cm. The number of cells (or aggregates) was quantified under a microscope (Zeiss, Carl Zeiss Microimaging GmbH, Germany) at 1,000 times magnification. The diameter of a single cell was assumed as 0.45 μm, which is the average size of ANME and SRB cells [3, 9]. The diameters of the aggregates were measured according to a reference scale bar built in the eyepiece of the microscope. The biovolume was calculated assuming that both cells and aggregates have spherical shapes. For each sample, 4 individual staining were applied. For each staining 50 fields of view were counted for calculation.

Cell and aggregates identification

In order to evaluate which type of ANME and SRB were present and enriched in the reactor, catalyzed reporter deposition fluorescence in situ hybridization (CARD-FISH) was applied on S1 and S2. The slurry samples were embedded onto GTTP filters. The filters were incubated in methanol with 0.15% H₂O₂ for 30 min at room temperature before washed with water and ethanol and dried. For each sample, 2 filters were prepared. One was incubated in lysozyme solution (10 mg/ml in 0.05 M EDTA, pH 8.0; 0.1 M Tris-HCl, pH 8.0) for 15 min at 37°C to achieve permeablilization of bacterial cells, and another one was incubated in Proteinase K solution (15 μg/ml in MilliQ water) for 3 min at room temperature to achieve permeabilization of achaeal cells. Afterwards the filters were cut into 4 pieces. Each piece was for hybridization with one probe (Table 1). The hybridization was performed according to the protocol previously described [23]. After hybridization, the filter was stained with DAPI to target all cells present on the filter.

During CARD-FISH, a few steps of washing the filter may cause the loss of cells and aggregates. It was assumed that all types of cells or aggregates were washed out in the same ratio. Therefore the percentage of ANME or SRB among the total cells did not change after washing. For each hybridization, cells and aggregates in 50 fields of view were analyzed under microscope. For each field, both probe staining and DAPI staining were counted to quantify the concentration of ANME-1 (or ANME-2, ANME-3 and SRB) among total biomass.

For a more detailed investigation on the microbial community, the archaeal and bacterial 16S rRNA gene clone libraries were performed on S2 according to protocol previously described [24, 25] with the primers listed in Table 1. For archaeal library, 56 clones were obtained while 50 clones were randomly picked for sequencing. For bacterial library, 110 clones were obtained while 100 clones were picked for sequencing. The sequences were compared with their best match in NCBI to classify their phylogenetic group (Additional file 1, Table S1). To calculate the percentage of each phylogenetic group into total archaeal/bacterial community, the number of clones within one phylogenetic group was divided by the number of sequenced clones within archaeal/bacterial library. All the sequences described in the paper have been deposited in the databases of GenBank, under accession numbers HQ405602 to HQ405741. The archaeal hpylogenetic tree has been constructed by Maximum Likelihood based on 212 sequences using the online RaxML tool at http://phylobench.vital-it.ch/raxml-bb/index.php[26]. The bacterial phylogenetic tree has been constructed using the online maximum likelihood tool at http://www.atgc-montpellier.fr/phyml/[27].