Growth of Anaerobic Methane-Oxidizing Archaea and Sulfate-Reducing Bacteria in a High-Pressure Membrane Capsule Bioreactor

Abstract

Communities of anaerobic methane-oxidizing archaea (ANME) and sulfate-reducing bacteria (SRB) grow slowly, which limits the ability to perform physiological studies. High methane partial pressure was previously successfully applied to stimulate growth, but it is not clear how different ANME subtypes and associated SRB are affected by it. Here, we report on the growth of ANME-SRB in a membrane capsule bioreactor inoculated with Eckernförde Bay sediment that combines high-pressure incubation (10.1 MPa methane) and thorough mixing (100 rpm) with complete cell retention by a 0.2-μm-pore-size membrane. The results were compared to previously obtained data from an ambient-pressure (0.101 MPa methane) bioreactor inoculated with the same sediment. The rates of oxidation of labeled methane were not higher at 10.1 MPa, likely because measurements were done at ambient pressure. The subtype ANME-2a/b was abundant in both reactors, but subtype ANME-2c was enriched only at 10.1 MPa. SRB at 10.1 MPa mainly belonged to the SEEP-SRB2 and Eel-1 groups and the Desulfuromonadales and not to the typically found SEEP-SRB1 group. The increase of ANME-2a/b occurred in parallel with the increase of SEEP-SRB2, which was previously found to be associated only with ANME-2c. Our results imply that the syntrophic association is flexible and that methane pressure and sulfide concentration influence the growth of different ANME-SRB consortia. We also studied the effect of elevated methane pressure on methane production and oxidation by a mixture of methanogenic and sulfate-reducing sludge. Here, methane oxidation rates decreased and were not coupled to sulfide production, indicating trace methane oxidation during net methanogenesis and not anaerobic methane oxidation, even at a high methane partial pressure.

INTRODUCTION

Anaerobic oxidation of methane (AOM) coupled to sulfate reduction (SR) is a process influenced by the CH₄ partial pressure. The SR rate of sediment from Hydrate Ridge was significantly higher at an elevated CH₄ partial pressure (1, 2). At between 0 and 0.15 MPa, there was a positive linear correlation between the CH₄ partial pressure and the AOM and SR rates of an anaerobic methanotrophic enrichment obtained from Eckernförde Bay sediment (3). The rate of methane-dependent sulfide production by microbial mats from the Black Sea increased 10- to 15-fold after the methane partial pressure was increased from 0.2 to 10.0 MPa (4). The affinity constant for methane (K_m) of anaerobic methanotrophs from Gulf of Cádiz sediment is about 37 mM, which is equivalent to 3 MPa CH₄ (5). Because of the more negative Gibbs free energy change (ΔG) at elevated CH₄ partial pressures, the growth of anaerobic methanotrophs might be faster when the CH₄ partial pressure is increased (see Fig. S1 in the supplemental material). Bioreactor studies with high methane pressure have been performed (4, 5), but it is not clear how the different subtypes of communities of anaerobic methane-oxidizing archaea (ANME) and associated sulfate-reducing bacteria (SRB) are affected by the methane pressure. This information would contribute to an understanding of the process of AOM coupled to SR and would help in further attempts to cultivate the responsible organisms.

In this study, we investigated the effect of the CH₄ partial pressure on methane oxidation and methane production rates in Eckernförde Bay sediment from the Baltic Sea. We also studied the effect of long-term incubation (240 days) under high methane pressure (10.1 MPa CH₄) on the activity of this sediment (reactor HP-1). These results, together with the results of microbial community analysis, were compared with data from a bioreactor at ambient pressure (reactor AP) (6, 7) inoculated with the same sediment used to inoculate reactor HP-1 and with the original Eckernförde Bay sediment (EB). We also investigated the effect of the CH₄ partial pressure on methane oxidation and methane production rates in mixed methanogenic and sulfate-reducing granular sludge in both short- and long-term incubations (reactor HP-2). This was done to evaluate the capacity of methanogenic and sulfate-reducing communities to perform methane oxidation under favorable conditions. A summary of the experimental setup is given in Fig. 1.

FIG 1.

Schematic representation of the different experiments performed in this study. Experiment 0 represents the study of reactor AP published previously (6, 7) and the original Eckernförde Bay sediment, samples of which were stored and analyzed in this study. Experiments 1 and 2 were fully conducted in this study. The piston graphic was modified from reference 11 with the permission of Oxford University Press [© 2010 Federation of European Microbiological Societies. Published by Blackwell Publishing Ltd. All rights reserved.].

MATERIALS AND METHODS

Origin of inocula.

The samples of the Eckernförde Bay sediment used for the initial activity assays and to inoculate reactor HP-1 were taken from station B (water depth, 28 m; position, 54°31′15″N, 10°01′28″E) in Eckernförde Bay (Baltic Sea) during a cruise of the German research vessel Littorina in June 2005. This sampling site has been described by Treude et al. (8). Sediment samples were taken with a small multicore sampler based on the construction described previously (9). The cores had a length of 50 cm and reached 30 to 40 cm into the sediment bed. Immediately after sampling, the content of the cores was mixed in a large bottle, which was made anoxic by replacing the headspace by anoxic artificial seawater. Back in the laboratory, the sediment was homogenized and transferred into 1-liter bottles in an anoxic chamber. The 1-liter bottles were closed with butyl rubber stoppers, and the headspace was replaced by CH₄ (0.15 MPa).

The mixed sludge used for the initial activity assays and to inoculate reactor HP-2 was sampled at two full-scale mesophilic upflow anaerobic sludge blanket reactors: a methanogenic reactor treating wastewater from paper mills (Industriewater Eerbeek, Eerbeek, the Netherlands, June 2005) and a sulfate-reducing reactor fed with ethanol (Emmtec, Emmen, the Netherlands, May 2006). The two sludge types were crushed by pressing them sequentially through needles with diameters of 1.2, 0.8, and 0.5 mm, mixed, and transferred into anaerobic bottles.

The bottles with sediment and sludge were stored in the dark at 4°C until the experiments were started.

Medium preparation.

The basal marine medium used for the incubations with Eckernförde Bay sediment was made as described previously (7). The basal freshwater medium used for the incubations with mixed sludge was made as described by Meulepas et al. (11). Both media were minimal media and did not contain any carbon source or any electron acceptor other than sulfate. The media were boiled, cooled down under a nitrogen (N₂) flow, and transferred into stock bottles with an N₂ headspace until use. The final pH of the media was 7.2. The phosphate provided a buffering capacity to maintain a neutral pH value.

Effect of CH₄ partial pressure on initial activity.

The effect of the CH₄ partial pressure on the CH₄ oxidation and methane production rate of both the Eckernförde Bay sediment and the mixed sludge was assessed in triplicate incubations with 0.02 g volatile suspended solids (g_VSS) at atmospheric (0.101 MPa) and elevated (10.1 MPa) methane pressure (Fig. 1, experiment 1). These tests were performed in glass tubes (18 ml), which were sealed with a butyl rubber stopper, capped at one side, and equipped with a piston at the opposite side (De Glasinstrumentenmakerij, Wageningen, the Netherlands) (11). The glass tubes were filled with sediment or mixed sludge and filled with 9 ml marine medium or freshwater medium, respectively. Then, the tubes were closed and flushed with N₂. After removing the N₂ gas with a syringe and needle, 3 ml ¹³CH₄ (purity, 5.5) was added. The glass tubes were incubated statically at 20°C in a nonpressurized incubator or in a 2.0-liter pressure vessel (Parr, Moline, IL, USA) filled with 1.8 liters water. The vessel was pressurized with N₂ gas. The pH, liquid volume, gas volume, and gas composition in the tubes were measured weekly. To do so, the pressure vessel had to be depressurized. Both pressurization and depressurization were done gradually over a period of 2 h.

Effect of long-term high-pressure incubation.

Two high-pressure vessels (Parr, Moline, IL, USA) were controlled at 20 ± 1°C and equipped with a stirrer controlled at 100 rpm (Fig. 1, experiment 2). One vessel was filled with 1.8 liters marine medium and inoculated with 25 membrane capsules, each containing 0.038 ± 0.003 g_VSS Eckernförde Bay sediment (reactor HP-1). The other vessel was filled with 0.5 liter freshwater medium and inoculated with 25 membrane capsules, each containing 0.072 ± 0.006 g_VSS mixed sludge (reactor HP-2). The membrane capsules were cylindrically shaped, 14 mm in diameter, and 20 mm long and had a membrane surface area of 840 mm². The polysulfone membranes (Triqua BV, Wageningen, the Netherlands) had a pore size of 0.2 μm to retain microorganisms. The filled capsules were slightly lighter than water, which made them float when the stirrer was turned off. During inoculation, the lid of the vessel was removed in an anaerobic glove box containing 90% N₂ and 10% H₂. Afterwards, the high-pressure vessel was connected to a bottle with pressurized CH₄ (purity, 5.5). The vessel was flushed with approximately 10 liters CH₄ (the gas entered the vessel at the bottom to remove any dissolved gas) and subsequently slowly pressurized to 10.1 MPa. At four time points (60, 110, 160, and 240 days), the pressure was gradually released and the vessel was opened in an anaerobic glove box to replace the medium and to sample two membrane capsules per reactor. Subsequently, the vessel was closed, flushed, and pressurized again with CH₄ gas as described above. The high-pressure vessels were equipped with sampling ports for liquid-phase sampling just before depressurization for sulfide determination. For activity determination, the sampled membrane capsules were incubated in 25-ml serum bottles at ambient pressure, closed with butyl rubber stoppers, and filled with 20 ml medium. The 5-ml headspace was filled with pure ¹³C-labeled CH₄ (0.13 MPa). The serum bottles were incubated at 20°C in orbital shakers (100 rpm). For about 30 days, the pH, liquid and gas volumes, pressure, gas composition, and sulfide concentration in the serum bottles were measured weekly. After these assays, the two membrane capsules per sampling point were frozen at −20°C for subsequent extraction of DNA for molecular analysis. At the last sampling point at 240 days, only one membrane capsule was taken.

Geochemical analyses.

Total dissolved sulfide species (H₂S, HS⁻, and S²⁻) were measured photometrically using a standard kit (LCK 653) and a photo spectrometer (Xion 500), both of which were from Hach Lange (Dusseldorf, Germany).

Gas composition was measured on a gas chromatograph (GC)-mass spectrometer (MS) from Interscience (Breda, The Netherlands). The system was composed of a TRACE GC equipped with a GS-GasPro column (30 m by 0.32 mm; J&W Scientific, Folsom, CA) and an ion-trap MS. Helium was the carrier gas at a flow rate of 1.7 ml min⁻¹. The column temperature was 30°C. The fractions of ¹³CH₄, ¹²CH₄, ¹³CO₂, and ¹²CO₂ were derived from the mass spectrum as described previously (12), with the retention times in the gas chromatogram being 1.6 min for CH₄ and 1.8 min for CO₂.

The pressure in the bottles and tubes was determined using a portable membrane pressure unit (WAL 0–0.4 MPa Absolute; WalMess- und Regelsysteme, Oldenburg, Germany).

The pH was checked by means of pH paper (Macherey-Nagel, Dűren, Germany).

Calculations.

For an explanation on the methods used for the calculation of total ¹³CO₂, ¹²CO₂, ¹³CH₄, and ¹²CH₄, see information in the supplemental material and Table S1.

DNA extraction.

DNA was extracted from the membrane capsules using a Fast DNA kit for soil (MP Biomedicals, OH, USA), according to the manufacturer's protocol, with two 45-s bead beating steps using a FastPrep instrument (MP Biomedicals, OH, USA). In parallel, DNA was extracted from stored samples from reactor AP and from the original Eckernförde Bay sediment (EB) (Fig. 1, experiment 0).

Clone library construction.

DNA extracted at the last sampling point at 240 days was used for clone library construction. To amplify almost full-length bacterial 16S rRNA genes for cloning, primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-GYTACCTTGTTACGACTT-3′) (13) were used. The archaeal 16S rRNA genes were amplified using primer A109f (ACKGCTCAGTAACACGT) (14) and universal reverse primer 1492R. PCR amplification was done with a GoTaq polymerase kit (Promega, Madison, WI, USA) and a G-Storm cycler (G-Storm, Essex, United Kingdom) with a predenaturing step of 2 min at 95°C, followed by 35 cycles of 95°C for 30 s, 52°C for 40 s, and 72°C for 1.5 min. Lastly, a postelongation step of 5 min at 72°C was done. PCR products were pooled and purified using a PCR Clean & Concentrator kit (Zymo Research Corporation, Irvine, CA, USA), ligated into the pGEM-T Easy plasmid vector (pGEM-T Easy vector system I; Promega, Madison, WI, USA), and transformed into Escherichia coli XL1-Blue competent cells (Stratagene/Agilent Technologies Santa Clara, CA, USA). Both ligation and transformation were performed according to the manufacturer's instructions.

DGGE analysis.

DNA extracted from the membrane capsules at every sampling point was used for denaturing gradient gel electrophoresis (DGGE) analysis, as was DNA extracted from reactor AP and from EB. The V3 region of the archaeal 16S rRNA sequences was amplified with primers GC-ARC344f (5′-ACGGGGYGCAGCAGGCGCGA-3′) and ARC519r (5′-GWATTACCGCGGCKGCTG-3′) (15) using the GoTaq polymerase kit (Promega, Madison, WI, USA). PCRs were performed in the G-Storm cycler (G-Storm, Essex, United Kingdom) with a predenaturing step of 5 min at 94°C, followed by 10 cycles of 94°C for 10 s, 61°C for 10 s (−0.5°C/cycle), and 72°C for 40 s; 25 cycles of 94°C for 10 s, 56°C for 20 s, and 72°C for 40 s; and a postelongation step of 30 min at 72°C. Bacterial 16S rRNA V6-V8 regions were amplified using Phire Hot Start II polymerase (catalog number F-122L; Thermo Scientific, Waltham, MA, USA) with the DGGE primer pair F-968-GC (5′-AACGCGAAGAACCTTAC-3′) and R-1401 (5′-CGGTGTGTACAAGACCC-3′) (16). Bacterial amplicons were produced with the G-Storm cycler (G-Storm, Essex, United Kingdom) using a predenaturing step of 30 s at 98°C, followed by 35 cycles of 98°C for 10 s, 56°C for 10 s, and 72°C for 30 s and a postelongation step of 1 min at 72°C. Forward primers had a GC clamp of 40 bp attached to the 5′ end, as used by Yu et al. (15). DGGE analysis was performed as previously described (17, 18) in a Dcode system (Bio-Rad, Germany) at 60°C for 16 h with a denaturing gradient of 30 to 60% for bacterial profiles and a denaturing gradient of 40 to 60% for archaeal profiles, as recommended elsewhere (15).

To clarify which of the most intense DGGE bands corresponded to an operational taxonomic unit (OTU) found in the clone library, clones were subjected to PCR-DGGE after cell lysis, using the same primer pairs that were used for the previous DGGE profiling. One clone of every OTU was loaded on a DGGE gel parallel to the last sample (obtained at 240 days) from reactor HP-1. Clones whose bands corresponded to the bands of the DGGE pattern of reactor HP-1 were annotated as such using BioNumerics software (version 4.61; Applied Maths NV, Belgium).

Phylogenetic analysis.

For the archaeal and bacterial clone library, 75 and 82 white colonies, respectively, that were picked were sent to GATC Biotech (Constance, Germany) for sequencing with the primer pair SP6 (5′-ATTTAGGTGACACTATAGAA-3′) and T7 (5′-TAATACGACTCACTATAGGG-3′). All reverse and forward sequenced overlapping reads were trimmed of vector and bad-quality sequences and were assembled into contiguous reads using DNA Baser software (Heracle BioSoft S.R.L., Pitesti, Romania). After assembly, possible chimeras were removed using the Greengenes Bellerophon chimera check program (http://greengenes.lbl.gov) (19). Whole 16S rRNA sequences were checked with the BLASTN program (20). Sequences were aligned using the SINA online alignment tool (version 1.2.11) (21). After merging of aligned sequences with the Silva SSU Ref database (release 111) (22), phylogenetic trees were constructed using the ARB software package (version 5.3-org-8209) (23). Phylogenetic trees were calculated by use of the ARB neighbor-joining algorithm.

qPCR.

DNA extracted from the membrane capsules at every sampling point was used for quantitative real-time PCR (qPCR) analysis, as was DNA from reactor AP and from EB. The DNA concentration was determined with a Qubit (version 2.0) fluorometer (Thermo Fisher Scientific, MA, USA). Amplifications were done in triplicate in a Bio-Rad CFX96 system (Bio-Rad Laboratories, Hercules, CA, USA) in a final volume of 25 μl using iTaq Universal SYBR green supermix (Bio-Rad Laboratories, Hercules, CA, USA), 5 ng of template DNA, and primers with concentrations and annealing temperatures optimal for the highest efficiency and specificity (see Table S2 in the supplemental material), all according to the manufacturer's recommendations. New primer sets were designed using the ARB software package (version 5.3-org-8209) (23). Triplicate standard curves were obtained with 10-fold serial dilutions that ranged from 2 × 10⁵ to 2 × 10⁻² copies per μl of plasmids containing 16S rRNA archaeal inserts of ANME-2a/b and ANME-2c and bacterial inserts of SEEP-SRB2 and the Eel-1 group. The efficiency of the reactions was up to 100%, and the R² values of the standard curves were up to 0.999. All primers used were extensively tested for specificity with cloned archaeal inserts of ANME-1, ANME-2a/b, ANME-2c, Methanococcoides, and Methanosarcinales and bacterial inserts of SEEP-SRB1, SEEP-SRB2, the Eel-1 group, Desulfuromonadales, Desulfosarcina, and Myxococcales and with genomic DNA of Methanosarcina mazei TMA (DSM-9195) and Desulfovibrio sp. strain G11 (DSM-7057). PCR conditions consisted of a predenaturing step for 5 min at 95°C, followed by 5 touchdown cycles of 95°C for 30 s, annealing at 60°C for 30 s with a decrement per cycle to reach the optimized annealing temperature (temperatures are shown in Table S2 in the supplemental material), and extension at 72°C (times are shown in Table S2 in the supplemental material). This was followed by 40 cycles of denaturing at 95°C for 15 s, 30 s of annealing, and extension at 72°C. The PCR products were checked for specificity by a melting curve analysis (72 to 95°C) after each amplification step and gel electrophoresis. Quantification of specific archaeal and bacterial groups was expressed as the total number of 16S rRNA gene copies per g_VSS extracted from the capsules.

Nucleotide sequence accession numbers.

The nucleotide sequence data reported here are available in the DDBJ/EMBL/GenBank databases under accession numbers HF922229 to HF922386.

RESULTS

Effect of CH₄ partial pressure on initial activity.

The results of the initial activity experiment (Fig. 1, experiment 1) are shown in Table 1, which presents the effect of an elevated ¹³CH₄ partial pressure on the oxidation of ¹³CH₄ to ¹³CO₂ and the ¹²CH₄ production of Eckernförde Bay sediment and mixed sludge. In both incubations with Eckernförde Bay sediment and incubations with mixed sludge, we observed ¹²CH₄ production and ¹³CO₂ production. Since no carbon source other than ¹³CH₄ was added, the ¹²CH₄ must have been produced from endogenous organic matter. At 0.101 MPa CH₄, both Eckernförde Bay sediment and mixed sludge showed ¹³CO₂ production during net methanogenesis. At 10.1 MPa, the Eckernförde Bay sediment showed no methane production, and rates of ¹³CH₄ oxidation to ¹³CO₂ were 4 times higher than those at 0.101 MPa. The oxidation of ¹³CH₄ to ¹³CO₂ by the mixed sludge was approximately 2 times higher at 10.1 MPa CH₄ than at 0.1 MPa CH₄, but net methane production was still shown.

TABLE 1.

¹³CO₂ and ¹²CH₄ production rates by EB and mixed sludge at 0.101 and 10.1 MPa ¹³CH₄ in initial activity experiment

Molecule produced	Production ratea (μmol gVSS−1 day−1)
	EB		Mixed sludge
	0.101 MPa 13CH4	10.1 MPa 13CH4	0.101 MPa 13CH4	10.1 MPa 13CH4
13CO2	5.8 ± 0.3	20.9 ± 4.5	8.6 ± 0.9	16.3 ± 6.2
12CH4	8.5 ± 1.4	0.0 ± 0.1	47.1 ± 1.9	36.6 ± 7.3

^aStandard deviations are for biological triplicates of 0.02 g VSS inoculum per glass tube.

Effect of long-term high-pressure incubation.

The long-term effects of an elevated methane partial pressure were tested in reactors with either Eckernförde Bay sediment or mixed sludge (Fig. 1, experiment 2). At 10.1 MPa CH₄, the methane oxidation rate in reactor HP-1 increased from 0.006 mmol g_VSS⁻¹ day⁻¹ to 0.024 mmol g_VSS⁻¹ day⁻¹ during the 240-day incubation (Fig. 2A; see also Table S3 in the supplemental material). The ¹²CO₂ production rate, on the other hand, decreased, likely because the available endogenous organic matter was depleted. After 240 days, the rate of ¹³CO₂ production was higher than the rate of endogenous ¹²CO₂ production. Initially, the SR rate by reactor HP-1 also decreased, but from day 110 onwards the SR rate was correlated to the methane oxidation rate. During long-term incubation of the mixed sludge, methane oxidation and sulfide production in reactor HP-2 did not increase, nor were they coupled during the 160-day incubation at 10.1 MPa CH₄. The total CO₂ and sulfide production rates decreased during the reactor run (Fig. 2B; see also Table S3 in the supplemental material).

FIG 2.

¹³CO₂ (○), ¹²CO₂ (Δ),¹²CH₄ (×), and sulfide (□) production rates derived from ambient-pressure activity measurements with ¹³CH₄ of capsules sampled from reactor HP-1 (A) and reactor HP-2 (B) after different periods of incubation with 10.1 MPa ¹²CH₄ at 20°C. Error bars represent standard deviations from independent measurements.

Microbial community of Eckernförde Bay sediment reactor.

An archaeal clone library of a sample taken from reactor HP-1 at 240 days of incubation showed that the total of 75 sequences were dominated by different clades of ANME (Fig. 3; see also Table S4 in the supplemental material). The highest percentage of ANME clones belonged to the ANME-2a/b group (56% of all sequences), followed by ANME-2c (19%) and ANME-1b (4%). Other clones with a relatively high frequency in the clone library cluster were miscellaneous crenarchaeotal (MCG) group 15 (MCG-15) (9%) and marine benthic group D (MBG-D) (8%). Archaeal DGGE profiling of membrane capsule DNA from reactor HP-1 at all sampling points was done to see initial community changes. Afterwards, PCR-DGGE of cloned inserts with known composition revealed that the bands belonging to ANME-2a/b and ANME-2c increased in intensity (see Fig. S2 in the supplemental material). The results of qPCR analysis of the same samples with 16S rRNA primers specific for ANME-2a/b, ANME-2c, and total Archaea are shown in Fig. 4. A significant increase (two-tailed t test with unequal variance, P < 0.05) of both ANME-2a/b and ANME-2c 16S rRNA gene copy numbers at 110 days of incubation was observed, confirming the initial DGGE results. The increase of ANME continued throughout the reactor run and coincided with an increase of AOM and SR rates (Fig. 2A; see also Table S3 in the supplemental material). The ANME-2a/b clade comprised a major fraction of the total Archaea, whereas the ANME-2c abundance was much lower during reactor operation (Fig. 4). However, ANME-2c 16S rRNA gene copy numbers showed a higher rate of increase than ANME-2a/b gene copy numbers at between 160 and 240 days of incubation.

FIG 3.

Phylogenetic tree of 16S rRNA gene sequences from an archaeal clone library constructed with a sample taken from reactor HP-1 at 240 days of incubation. The tree was constructed with the ARB neighbor-joining method with terminal filtering and the Jukes-Cantor correction using almost-full-length 16S rRNA sequences. Clones detected in this study are indicated in bold. The numbers in parentheses indicate the number of sequences of each phylotype found. Closed circles, bootstrap values of >70% (1,000 replicates). The scale bar represents the percentage of changes per nucleotide position.

FIG 4.

Absolute 16S rRNA gene abundance of ANME-2a/b and total Archaea (A) and ANME-2c (B) in reactor HP-1 sampled in duplicate at 60, 110, and 160 days of incubation (samples A and B) and once at 240 days of incubation (sample A). Results were compared to those for the reactor AP and EB inocula. Standard deviations from triplicate analyses are shown.

At 240 days of incubation, a bacterial clone library of 82 sequences from reactor HP-1 showed a high bacterial diversity (Fig. 5; see also Table S4 in the supplemental material). All but two sequences within the clone library showed 97% or less similarity to known cultivated members. From the Deltaproteobacteria, the most common phylotypes recovered belonged to the methane seep-associated Eel-1 (6% of all sequences) and Eel-2 (13%) clades, as described by Orphan et al. (24), of which the Eel-2 clade clusters within the SEEP-SRB2 group. We also found sequences that are affiliated with the order Desulfuromonadales (7%). Members of the Desulfobacteriaceae were the least abundant, and only 2% belonged to the Desulfosarcinales/Desulfococcus cluster SEEP-SRB1. Some sequences found belonged to the Myxococcales group. The remaining bacterial phylotypes were very diverse, and many groups were also found previously in sediments and reactor systems with AOM activity. Some were represented by only one phylotype derived from the clone library (see Table S4 in the supplemental material).

FIG 5.

Phylogenetic tree of 16S rRNA gene sequences from a bacterial clone library constructed with a sample taken from reactor HP-1 at 240 days of incubation. The tree shows only the canonical sulfate-reducing bacterial phylotypes found. The tree was constructed with the ARB neighbor-joining method with terminal filtering and Jukes-Cantor correction using almost-full-length 16S rRNA sequences. Clones detected in this study are indicated in bold. The numbers in parentheses indicate the number of sequences of each phylotype found. Closed circles, bootstrap values of >70% (1,000 replicates). The tree outgroup, Clostridium, was removed after tree construction. The scale bar represents the percentage of changes per nucleotide position.

The results of qPCR analysis of membrane capsule DNA from reactor HP-1 with 16S rRNA primers specific for total Bacteria, primers specific for SEEP-SRB2, and the newly designed primers specific for Eel-1 at all sampling points are shown in Fig. 6. An 8-fold increase of SEEP-SRB2 16S rRNA gene copy numbers was observed at 160 days of incubation, and Eel-1 16S rRNA gene copy numbers increased 4-fold. The abundance of Eel-1 decreased slightly in parallel with that of total Bacteria after 160 days of incubation, whereas the abundance of SEEP-SRB2 continued to slightly increase. This resulted in a relative increase of SEEP-SRB2 throughout the reactor run, whereas Eel-1 remained present at a constant level of 2.5% of total Bacteria. From the qPCR results, we also calculated the ratios of ANME-2a/b and ANME-2c copy numbers over Eel-1 and SEEP-SRB2 copy numbers. We observed that only ANME-2a/b and SEEP-SRB2 were detected in a constant ratio of about 1:2 throughout the time of reactor operation and in EB (Fig. 6C). Many more ANME-2a/b copies than SEEP-SRB2 copies were detected in reactor AP. The Eel-1 copies did not show a constant ratio with any ANME subtype. We could not analyze Desulfuromonadales within the reactor, as we were not able to design primers specific for this clade.

FIG 6.

Absolute 16S rRNA gene abundance of the SEEP-SRB2 and Eel-1 groups (A) and total Bacteria (B) with standard deviations from triplicate analyses and the ratio of ANME-2a/b and SEEP-SRB2 (C) with combined standard deviations calculated as described previously (63). Reactor HP-1 was sampled in duplicate at 60, 110, and 160 days of incubation (samples A and B) and once at 240 days of incubation (sample A), and the results were compared to those for the reactor AP and EB inocula.

Microbial community of mixed sludge reactor.

Analysis of the microbial community in mixed sludge reactor HP-2 was restricted to archaeal and bacterial DGGE analysis (see Fig. S2 and S3 in the supplemental material), as no increase in methane oxidation was observed. For both the archaeal and the bacterial DGGE profiles, we did not see any community changes during the reactor run.

DISCUSSION

Activity of Eckernförde Bay sediment.

Our initial activity experiments showed that the Eckernförde Bay sediment performed trace methane oxidation (TMO) during net methanogenesis at 0.101 MPa CH₄ and net anaerobic oxidation of methane (AOM) at 10.1 MPa CH₄ without methane production (Table 1). Because the ¹³CO₂ production rate was also 4 times higher at 10.1 MPa CH₄ than at 0.101 MPa CH₄, we expect that the AOM activity of Eckernförde Bay sediment is stimulated by the higher methane partial pressure, although the sediment originates from relatively shallow waters of 28 m in depth (8). The AOM activity in reactor HP-1 did not, however, increase faster than the reported AOM activity of the same Eckernförde Bay sediment in reactor AP at 0.101 MPa CH₄. In reactor HP-1, the AOM rate increased from 0.006 to 0.025 mmol g_VSS⁻¹ day⁻¹ over 240 days (Fig. 2A; see also Table S3 in the supplemental material), and in reactor AP, the AOM rate increased from 0.003 to 0.55 mmol g_VSS⁻¹ day⁻¹ in 842 days (7).

Despite the good mixing of reactor HP-1, the increase of the AOM rate could have been limited by the larger diffusion distances. In reactor HP-1, the biomass was present in membrane capsules with a diameter of 14 mm, whereas reactor AP was a membrane bioreactor (MBR), where the biomass was present as 0.1-mm flocks that were directly in contact with the bioreactor medium (7). In reactor HP-1 at day 240, the average methane flux though the membranes was 0.11 μmol cm⁻² day⁻¹ (which is equal to 0.025 mmol g_VSS⁻¹ day⁻¹ × 0.038 g_VSS/8.8 cm²). At this flux, the change in the CH₄ concentration (Δ[CH₄])/Δx, where Δx is the difference in distance, is 16 mM cm⁻¹, according to Fick's first law of diffusion [CH₄ flux = −ØD_methane(Δ[CH₄]/Δx)], where ØD_methane is the average porosity [88%] multiplied by the molecular diffusion coefficient of methane) (see Table S1 in the supplemental material). At 10.1 MPa CH₄ and 20°C, the CH₄ concentration in the bulk liquid was approximately 152 mM. The average CH₄ concentration near the microorganisms was therefore only marginally lower than that in the bulk liquid and cannot explain the slow activity increase.

A more plausible explanation for the slow activity increase could be related to the method of measuring the activity of the high-pressure reactor samples. Sampled membrane capsules were incubated in 25-ml serum bottles at ambient pressure, using 0.13 MPa pure ¹³C-labeled CH₄ (Fig. 1, experiment 2). Activity measurement at ambient pressure previously showed decreased AOM activity compared to the AOM activity determined at high pressure (25), but also, the microorganisms could have adapted to the higher pressure and would be less active when incubated at ambient pressure, as shown for true piezophiles (26). Indeed, the doubling times calculated from the exponential increase in the AOM rate in both reactors were 3.8 months (R² = 0.98, n = 12) for reactor AP and 3.9 months (R² = 0.90, n = 15) for reactor HP-1. The doubling times calculated from qPCR analysis were 1.5 months for ANME-2a/b, 1.1 months for ANME-2c, and 1.4 months for SEEP-SRB2. This indicates that high methane partial pressure had a positive effect on the AOM-mediating microorganisms which was not reflected in AOM activity measurements.

A less likely explanation could be that reactor HP-1 was operated in the fed-batch mode. Here, sulfide and bicarbonate accumulated until the medium was replaced. Sulfide levels during the first (days 0 to 60) and the last (days 160 to 240) incubation periods reached 2.7 mM (Table 2). This could have been limiting the overall activity of the AOM-mediating microorganisms, as 2.4 ± 0.1 mM sulfide was found to completely inhibit AOM and SR in reactor AP (7). In reactor AP, sulfide levels were below 1.5 mM in the first 800 days of the reactor run, reaching only 1.9 mM in the last 7-day period.

TABLE 2.

Sulfide concentration in reactor HP-1 inoculated with EB

Time (days)	Sulfide concn (mM)
0	0
60	2.7
110	1.5
160	2.1
240	2.7

Microbial community of Eckernförde Bay sediment reactor.

An increase in the copy numbers of the 16S rRNA gene of ANME-2c archaea was observed only in reactor HP-1, a high-pressure reactor. In reactor AP, an ambient-pressure reactor, only ANME-2a/b was present (6), which was verified by DGGE and qPCR (Fig. 4). ANME-2a/b also showed growth at high pressure, indicating that both phylotypes could grow at high methane partial pressure. Previous studies showed the predominance of ANME-2c archaea at high methane partial pressure (27) and in the interior of hydrates (28) and showed a transition from ANME-2a/b to ANME-2c sequence abundance with increasing sediment depth and sulfide concentration (29). Also, ANME-2a/b archaea seem to exist in sediments with little or no free sulfide (30). Because ANME-2c archaea were not present in reactor AP at atmospheric pressure and a lower sulfide concentration, it is likely that these methanotrophs do not grow at low methane pressure and that they have a higher sulfide tolerance. This could have resulted in higher growth rates for ANME-2c than for ANME-2a/b. Indeed, ANME-2c showed faster growth than ANME-2a/b at the end of the run of reactor HP-1 (Fig. 4), and the doubling time of 1.1 months for ANME-2c was shorter than the doubling time of 1.5 months for ANME-2a/b in the exponential phase. An eventual predominance of ANME-2c in reactor HP-1 after a prolonged incubation is therefore plausible. ANME-1b archaea were the least abundant methanotrophs in both AOM-SR reactors, which could be explained by the continuous high sulfate and low sulfide concentrations that seem to preferentially select for ANME-2 archaea. Several studies showed a dominance of ANME-1 archaea in sulfate-depleted environments (31), together with elevated sulfide levels (30), and it was suggested that ANME-1 could perform AOM independently of sulfate-reducing bacteria (32–34) or even perform methanogenesis (35).

Archaeal DGGE bands that were intense throughout the incubation of reactor HP-1 belong to MCG-15 and MBG-D (see Fig. S2 in the supplemental material). MBG-D represented 8% of our clone library sequences, have been found in many cold marine (deep-sea) sediments (24, 36, 37), and were consistently found in bioreactors (7, 38). These archaea are related to the sulfur-reducing order Thermoplasmatales and appear to include methanogens named “Methanoplasmatales” (39). The miscellaneous crenarchaeota that were present until the end of the reactor run are abundant in marine deep-subsurface sediments (40). One hypothesis is that MCG archaea are heterotrophic anaerobes (41), and carbon isotopic signatures and polar lipid analysis also indicated an organic carbon metabolism in sediments dominated by MCG sequences (42). Recently, it was found with single-cell genomic sequencing that the MCG and MBG-D archaea could play a role in protein degradation (43). The batch mode of operation of our reactor implies a long retention time of products of endogenous activity that could function as potential new substrates. This may have led to less selective enrichment and could explain the richness in archaeal diversity in our reactor.

Deltaproteobacteria of the Eel-1 group and the SEEP-SRB2 clade were present during the run of reactor HP-1, as qPCR and clone library results showed. Eel-1 members are closely related to the marine sulfate reducer Desulfobacterium anilini (44). Most members of the SEEP-SRB2 clade are related to Dissulfuribacter thermophilus (92% similarity) and Desulfobulbus propionicus DSM 2032 (89% similarity), both of which are sulfur-disproportionating bacteria (45, 46). Sequences related to sequences of the Desulfuromonadales were as abundant as the Eel-1 group in the clone library and clustered closely to sequences of the Pelobacter genus. Pelobacter is distinguished from Desulfuromonas species by being able to ferment specific hydrocarbons and being unable to reduce Fe(III) and/or elemental sulfur (47). Both the SEEP-SRB2 and the Eel-1 groups had increased in 16S rRNA gene copy numbers at 160 days, but the Eel-1 group decreased in abundance with reactor time, in parallel with the decrease in the abundance of total Bacteria (Fig. 6). The Eel-1 group was previously hypothesized to be directly or indirectly involved in AOM in situ (24). We found, however, that only the growth of ANME-2a/b coincided with the growth of SEEP-SRB2 with a stable ratio of about 1:2 (Fig. 6C), excluding the possibility of at least the direct involvement of Eel-1 members in AOM. This finding, together with the observed similar doubling times, could indicate that ANME-2a/b is growing in consortia with SEEP-SRB2, which, to our knowledge, has not been shown before. ANME-2c archaea could have been paired with the other most abundant Desulfuromonadales. This SRB group was previously found in AOM-mediating enrichments (27) and in cold-seep sediment (29, 48). However, as with the Eel-1 group, abundance is not an indication for involvement in AOM-SR. It could be that ANME-2c is actually forming consortia with SEEP-SRB2 as well, but a strong correlation was not found because ANME-2c copy numbers were very low at the start of the reactor run and increased the most at between 160 and 240 days of incubation. A stronger correlation between ANME-2c and SEEP-SRB2 may have been found if the reactor would have been monitored longer.

Only 2% of the sequences in the clone library of reactor HP-1 belonged to the SEEP-SRB1 branch. In previous research on different AOM sediments, cloning results showed a co-occurrence of ANME-2 archaea and SEEP-SRB1 bacteria. In contrast, when ANME-1 archaea were present, the Eel-1 and SEEP-SRB2 groups seemed to be more abundant (Table 3). With microscopy techniques, other researchers recently found ANME-2c to be associated with SEEP-SRB2 (49) or other ANME-2 partners, such as Desulfobulbus sp.-related SRB (50, 51) and unidentified bacteria (32). Other ANME types besides ANME-2 were also found to aggregate with SEEP-SRB1 (52, 53). Recently, a novel bacterial partner named HotSeep-1 was found in thermophilic AOM (54), and ANME-1a was even found at 90°C in the absence of SRB (55).

TABLE 3.

Overview of archaeal and bacterial 16S rRNA genes detected in different studies on AOM-mediating marine sediments^a

Sampling site	Reference	16S rRNA gene detected
		Archaea		Bacteria
		ANME-1	ANME-2	SEEP-SRB1	SEEP-SRB2	SEEP-SRB3	SEEP-SRB4	Eel-1	Eel-3
Eel River Basin at depth of:
0–4 cm	24	++	+	+	+	+		+	+
20–22 cm	24	++	+	+	+	−		+	−
4–7 cm	24	+	++	+	−	+		−	−
Hydrate Ridge, Beggiatoa mat	52, 64	+	++	++	+	+	+	−
Santa Barbara, CA (depth, 139 cm)	65	+		+	−			++
Gulf of Mexico (depth, 15–18 cm)	66	+			++			+
Guaymas Basin
Core A	67	+	−	−	+	+		−
Core C	67	−	+	+	−		+	−
Santa Barbara, CA (depth, 13–16 cm)	24	−	+	+	−	−		−	+
HR	27	−	+	++	−			−
Isis	27	−	+	++	−			−
Eckernförde Bay, reactor AP	6	−	+	++	+			+
Eckernförde Bay, reactor HP	This study	+	++	+	++	−	−	+	−

^aSymbols and abbreviations: +, presence of 16S rRNA gene sequences; ++, dominance of 16S rRNA gene sequences; −, absence of 16S rRNA gene sequences; HR, Hydrate Ridge; Isis, Isis mud volcano.

Our findings clearly indicate that the syntrophic relationship between different types of ANME and SRB is flexible and dependent on environmental factors. It was suggested before that syntrophy in AOM depends on the metabolism or ecological niche of the SRB (49, 51), and nitrate was suggested to be the basis for niche differentiation between some groups of SRB (56). Uncultivated SRB belonging to SEEP-SRB2 are dominating seep habitats and are believed to be able to use nonmethane hydrocarbons (49, 56). We observed the growth of SEEP-SRB2 in reactor HP-1, indicating that this clade is indeed involved in AOM and does not need other nonmethane hydrocarbons for growth. More likely, environmental parameters such as methane partial pressure and sulfide concentration play a key role in the growth of SEEP-SRB2 and ANME-2c. This could explain the lack of ANME-2c and SEEP-SRB2 in reactor AP at ambient methane pressure and low sulfide levels and the lack of SEEP-SRB1 in reactor HP-1 at high pressure and increased sulfide levels. Further studies are, however, needed to clarify which environmental parameters are crucial and which mechanism underlies the syntrophic interaction between ANME and SRB. A continuous-flow bioreactor which mimics in situ conditions with little disturbance already showed differential growth dynamics between ANME-1 and ANME-2 populations dependent on alterations in pore water flow rates (57). Similar studies where only the methane partial pressure or sulfide concentration is the varying factor could also give more insight into the differential growth and activity of ANME-2a/b and ANME-2c phylotypes and the associated SRB.

Activity and microbial community of mixed sludge.

Our initial activity experiments showed that mixed sludge performs TMO during net methanogenesis with both 0.101 MPa CH₄ and 10.1 MPa CH₄ (Table 1). Whereas reactor HP-1 showed increasing AOM activity during long-term incubation, reactor HP-2 did not. The total amounts of CO₂ and sulfide production decreased during the reactor run as endogenous substrates became depleted. Microbial analysis was restricted to DGGE profiling, which did not show major community changes, as was observed in the HP reactor performing net AOM (see Fig. S2 and S3 in the supplemental material). This demonstrates that even with 10.1 MPa CH₄, the anaerobic community in granular sludge was not able to utilize the available energy for AOM coupled to SR during 160 days of incubation or that it does not have the metabolic flexibility to do so. This is in agreement with previous findings that granular sludge mediates TMO during net methanogenesis (11, 58), which results in much higher ¹³CO₂ production rates from ¹³CH₄ than from the reported carbon back flux (59). In contrast, Eckernförde Bay sediment showed a clear uncoupling between methane oxidation and endogenous methanogenic activity and a coupling of ¹³CO₂ and sulfide production after 110 days of incubation. The production of ¹²CO₂ dropped to about 37 μmol g⁻¹ day⁻¹ when AOM started to occur and kept on decreasing, whereas the production of ¹²CO₂ in the sludge reactor never reached less than 90 μmol g⁻¹ day⁻¹ during the 160 days of the reactor run. According to Hoehler et al. in 1994, the hydrogen concentration must be low enough for AOM to occur (60). Assuming that ¹²CO₂ production coincides with hydrogen production from organic matter degradation in anoxic sludge (61), then the hydrogen concentration was probably low enough in the Eckernförde Bay sediment reactor at 110 days but too high in the mixed sludge reactor. If we would have allowed ¹²CO₂ production in the sludge reactor to drop to as low as 37 μmol g⁻¹ day⁻¹, it may have allowed AOM to occur. It was shown recently that in anaerobic digestion of a diverse mixture of samples, the chemical oxygen demand also drastically drops in the first 150 days of reactor incubation and reaches steady state at about 160 days (62). Long-term incubation is therefore indispensable to distinguish between labeled-methane oxidation during net methanogenesis (TMO) or net AOM.