Introduction

Anaerobic oxidization of methane (AOM) coupled to sulfate reduction is a key process that effectively controls the methane emission from anoxic marine waters and sediments to the oxygenated ocean (Reeburgh, 1976, 2007). The microorganisms thought to be mediating AOM have been designated as anaerobic methane-oxidizing archaea (ANME), close phylogenetic relatives of the methanogenic archaea. These microorganisms consist of three established groups (ANME-1, ANME-2 and ANME-3) that can form syntrophic aggregates with sulfate-reducing bacteria (SRB) or exist as single cells (Hinrichs et al., 1999; Boetius et al., 2000; Orphan et al., 2001; Knittel and Boetius, 2009). AOM has been hypothesized to be operated via a reversal of the methanogenesis reaction (Hoehler et al., 1994), and this hypothesis has been partly supported by metagenome-based analyses of communities dominated by ANME-1 (Hallam et al., 2004; Meyerdierks et al., 2010). However, the evidence remains inconclusive because one of the genes in canonical CO2-dependent methanogenesis, the N5,N10-methylene-tetrahydromethanopterin (methylene-H4MPT) reductase (Mer, which catalyzes the reaction from methylene-H4MPT to methyl-H4MPT) gene mer, has not been found in the metagenomes of ANME-1. Meanwhile, very limited genomic information is available for ANME-2 or ANME-3 (Hallam et al., 2004; Pernthaler et al., 2008; Stokke et al., 2012). Consequently, it remains unresolved if and how the AOM pathway differs among the various ANME clades.

The available ANME metagenomes do not provide solid information about electron transport and energy conservation in reverse methanogenesis. A novel type of sulfate reduction has been recently recognized as electron sink to methane oxidation in an ANME-2 enrichment by generating the energy for ANME-2 with zero-valent sulfur being an intermediate then disproportionated by SRB cells (Milucka et al., 2012). However, the intracellular electron transporting within ANME remains to be unveiled. AOM via the ‘reverse methanogenesis hypothesis’ requires the presence of genes for electron transport from the primary electron acceptors CoM-S-S-CoB (generated in the heterodisulfide reductase reaction), coenzyme F420 (generated in the methylene-H4MPT reductase and methylene-H4MPT dehydrogenase reactions) and ferredoxin (generated in the formylmethanofuran dehydrogenase reaction) to a terminal electron acceptor that has been proposed to be protons (Hoehler et al., 1994; Thauer, 2011). All known methanogenic archaea require H2 for CO2 reduction. In the case of reverse methanogenesis with H2 as a product, the presence of hydrogenases would be required. However, these hydrogenases have not been confirmed in the ANME-1 metagenome (Meyerdierks et al., 2010).

Therefore, central mechanistic details on AOM remain unresolved such as whether ‘reverse methanogenesis’ properly describes the AOM mechanism, and how electrons are transferred within ANME. In this study, we try to address these gaps in knowledge through an integrated analysis of single-aggregate ANME-2a genome and its gene expression, to reveal the pathways for methane oxidation and energy transportation in ANME-2a.

Materials and methods

Origin of sample

Sediment samples from Capt Aryutinov Mud Volcano (coordinates: 35:39.700/07:20.012, Gulf of Cadiz, Atlantic Ocean, 1200 m water depth) were collected during a Maria S. Merian 1/3 cruise on 30 April 2006. After the activation of AOM activity at ambient pressure, a total volume of 600 ml diluted sediment slurry (1:12 w:w) was incubated into a special designed continuous high-pressure bioreactor under 8 MPa methane pressure at 15 °C for over 1 year. A highly enriched AOM culture was obtained, of which an approximate 0.5 mmol day−1 methane oxidation-dependent sulfate reduction rate has been measured (Zhang et al., 2010, 2011). A well-mixed slurry sample (∼120 ml) of this active enrichment was taken out and immediately fixed with RNAlater (Sigma-Aldrich, Munich, Germany) and stored at −80 °C before mRNA extraction for metatranscriptome analysis. From the same enrichment, a slurry sample was taken out directly without fixation for single-aggregate extraction and genome analysis.

RNA isolation and complementary DNA synthesis

RNA was isolated using the RNA isolation kit (Omega Bio-Tek, Doraville, GA, USA) following the user manual. DNA contamination was ruled out according to PCR results of the RNA sample where it was used as template and amplified with archaeal and bacterial 16S ribosomal RNA (rRNA) gene universal primers (Arch21F and Arch958R for archaea; Eubac27F and Eubac 1492R for bacteria; see Zhang et al., 2011). The mRNA fraction was enriched by the enzymatic digestion of rRNA molecules (mRNA-ONLY Prokaryotic mRNA Isolation kit, Epicentre Biotechnologies, Madison, WI, USA) followed by subtractive hybridization of rRNA with capture oligonucleotides (Ambion MICROBExpress kit, Life Technologies, Gaithersburg, MD, USA). The mRNA isolates were amplified (MessageAmp II-Bacteria kit, Ambion, Life Technologies) and reversely transcribed into complementary DNA and then directly sequenced using the Illumina (Shenzhen, China) GAIIx platform for metatranscriptome analysis.

Sample preparation and single-aggregate extraction

Separation of cells from AOM enrichment was conducted using Percoll discontinuous density gradients (1.080, 1.069 and 1.060 g ml−1) method. Enrichment was firstly sonicated on ice at an output power of 165 W for 30 s with 5 s pulse, and then overlaid on the top of the gradients to centrifuge at 1000 g for 2 h. The total supernatant from gradients was filtered through a 3-μm pore-size polycarbonate filter that was subsequently washed twice with phosphate-buffered saline to remove any trace Percoll and afterwards suspended into 1 ml phosphate-buffered saline solution. From the cell suspension, 500 μl was transferred onto an adhesion slide and incubated inside a moist chamber. After 10 min of sedimentation, most of the cells and aggregates were fixed onto the surface of the slide. In the light field of microscope, aggregates were captured with a capillary glass micropipette (with an inner diameter of 8 μm, prepared with Micropipette Puller system P-2000, Sutter Instrument Company, Los Angeles, CA, USA) using a XenoWorks Microinjection System (Sutter Instrument Company). The captured aggregates were transferred into sterile 0.2 ml tubes that were pre-filled with 3.5 μl of phosphate-buffered saline. Each tube contained one aggregate. These aggregate samples were immediately stored at −70 °C for further characterization.

Whole-genome amplification

Whole-genome multiple displacement amplification (MDA) on single aggregate was conducted by REPLI-g Mini kit reagents (Qiagen, Hilden, Germany) following the manufacturer’s protocol. DNA (500–800 ng) was generated after each MDA reaction and the first-round MDA product was used for 16S rRNA gene characterization. Forward primer Arch21F (Delong, 1992) and three reverse primers ANME-2-538 (Treude et al., 2005), Arch915 (Stahl and Amann, 1991) and Arch958 (Delong, 1992) were applied for archaeal 16S rRNA gene amplification. The bacterial 16S rRNA gene fragments were amplified with the primer set of Bac27F and Bac1492R. The PCR was initiated with a denaturing step at 94 °C for 2 min. A total of 35 cycles of 30 s, 94 °C; 30 s, 55 °C; and 30 s, 72 °C (60 s for bacteria) were used for the PCR, and the reaction was terminated after 7 min of elongation. A second-round MDA was performed to meet the requirements of Illumina genome sequencing. During the second-round MDA, five parallel running of amplification reaction was performed for each sample to minimize stochastic bias. For each reaction, 5 μl of purified first-round MDA products was used as template. Amplified DNA products from all five reactions were pooled together and purified using QiaAmp DNA mini kit (Qiagen) following the manufacturer’s protocol.

Genome sequencing, assembly, annotation and taxonomic assignment

Sequencing libraries (500 bp and 2 kbp) of the single-aggregate samples (second-round MDA products) were sequenced using the Illumina GAIIx 2 × 90 bp pair-end technology. Assembly of the Illumina reads was performed with SOAPdenovo 1.05 software (Li et al., 2009). Gene prediction was carried out by using Glimmer3 software (Delcher et al., 2007). Ribosomal RNA genes were detected by using the rRNA prediction algorithm (Huang et al., 2009). For each predicted open reading frame, functional information was retrieved from the National Center for Biotechnology Information (NCBI) non-redundant (NR) database using blastx (Altschul et al., 1997) with an E-value cutoff of <10−5. Those sequences that had reliable hits from NR database were searched against the Kyoto Encyclopedia of Genes and Genomes (KEGG; Ogata et al., 1999) database as well as the database for Clusters of Orthologous Groups of proteins database (COG; Tatusov et al., 2000) (E-value cutoff of <10−5) for functional annotation. To analyze the taxonomic contents, blastx results of all predicted gene features were visualized in MEGAN (Huson et al., 2007). Each predicted gene feature in the single aggregate genome was assigned to a certain taxon when at least 75% of the BLAST hits of this query were from that specific taxon. Raw sequence data of the metatranscriptome and the single-aggregate genome have been submitted to the NCBI Sequence Read Archive under accessions SRX359351 and SRX359142, respectively. The assembled genome sequences of M25 have been incorporated into the Integrated Microbial Genomes (IMG) system with the ER submission ID 20381.

Phylogenetic tree construction

For phylogenetic analyses, multiple sequence alignments were conducted with CLUSTALW2 (Larkin et al., 2007), and all phylogenetic trees were constructed by PhyML (Guindon and Gascuel, 2003). The LG (La Guscuel) substitution model was used for all the amino acid sequence-based phylogenies. Bootstrap values were obtained from 1000 replicates.

Estimation of complete genome size of ANME-2a

Genome size of ANME-2a was estimated based on a conserved single copy gene (CSCG) analysis (Woyke et al., 2009). Here, genes from a single aggregate (where the only detected 16S rRNA gene belonged to ANME-2a) were used as a pool for CSCG detection. To identify relevant CSCGs for ANME-2a, 16 finished Methanomicrobia genomes (Supplementary Table S1) available in July 2012 at the IMG site (Markowitz et al., 2012) were analyzed. COG distribution for the 16 analyzed genomes were retrieved from IMG and used for CSCG identification. The number of CSCGs (designated as Cn when shared by n genomes) was plotted against the number of genomes (n) and a power function fit was applied to the data. A regression curve was drawn to predict the number of CSCGs (Cn+1) remaining after adding one more genome (n+1). Gn+1 presented the number of CSCGs on the assembly of the single-aggregate genome. The expected complete genome size of ANME-2 was estimated as described previously (Woyke et al., 2009): GS=Co × AS/RCSCG, where GS is the expected complete genome size; AS is the size of current genome assembly; RCSCG is the recovery of CSCGs (RCSCG=Gn+1/Cn); Co is the correction coefficient to compensate for the expected lower number of CSCGs shared by n+1 genomes relative to n genomes (Co=Cn+1/Cn).

Metatranscriptome sequencing, assembly and annotation

All metatranscriptomic reads were first aligned to a ribosome RNA database SILVA (Pruesse et al., 2007) using blastn (E-value <10−10). Only those non-rRNA reads were included in further analysis and assembled using SOAP alignment software (Li et al., 2009). Open reading frames were predicted with MetaGene Annotator (Noguchi et al., 2008) and compared with the NCBI NR database using blastx with an E-value cutoff of <10−5. Those sequences that had reliable hits from NR database were compared with the KEGG and COG databases (with E-value <10−5) for functional annotation.

Estimation of gene expression in single-aggregate genome from metatranscriptomic transcripts

Metatranscriptomic reads were assigned to single-aggregate genes by blastx with E-value cutoff 10−10 and >90% identity, and then subject to manual inspection to ensure exact mappings between single-aggregate genes and their assigned metatranscriptomic reads. The expression level of each transcript in the transcriptome was quantified by RPKM (reads per Kb per million reads) to measure the read density, which reflected the molar concentration of a transcript in the starting sample by normalizing for RNA length and for the total read number in the measurement (Mortazavi et al., 2008). This approach facilitated comparisons of transcript levels both within and between samples. The expression levels of single-aggregate genes were roughly estimated by the RPKM values of the mapped gene transcripts in the metatranscriptome.

Results and discussion

Genomic and transcriptomic data analysis

The AOM enrichment investigated in this study contains ANME-2a as the dominant archaeal group (Zhang et al., 2011), mainly in the form of aggregates (Supplementary Figure S1). A few cell aggregates were captured from the enrichment, with their DNA isolated, amplified and sequenced as described in the Materials and methods. One captured single aggregate named M25 was found to have ANME-2a as the sole archaeal group with no bacterial partners: no bacterial 16S rRNA gene fragments were amplified, and only ANME-2a 16S rRNA gene was retrieved by archaeal 16S rRNA gene amplification (for details see Materials and methods). M25 was then subjected to in-depth genomic sequencing. Initially, 11 111 112 reads with a length of 90 bp were generated for this single-aggregate genome M25, and resulted in 3.64 Mbp of assembly (Supplementary Table S2). No bacterial 16S rRNA gene fragments were detected from the M25 assembly, and only one 16S rRNA gene sequence assigned to ANME-2a was found (with its phylogeny displayed in Supplementary Figure S2). Meanwhile, as demonstrated in the phylogeny of the methyl-coenzyme M reductase subunit A (McrA, a key enzyme for methane production and oxidation), mcrA retrieved from the M25 genome was also assigned to the ANME-2a subgroup (Supplementary Figure S3). The genomic evidences (the annotated 16S rRNA gene and mcrA analysis, the absence of bacterial marker genes for sulfate reduction) together with the direct 16S rRNA gene amplification confirmed that M25 contained only ANME-2a without bacterial partners. The genome size for our ANME-2a is estimated to be 3.96 Mbp, with a recovery of ∼90% by the current M25 assembly (Supplementary Figure S4, see conserved archaeal single copy gene analysis in Materials and methods). Based on the results of blastx (see taxonomic assignment in Materials and methods), all the genes discussed in this study (as listed in Table 1 and Supplementary Tables S3 and S4) were assigned to Methanosarcinales, where ANME-2a belongs. At present, our understanding on these yet uncultivated ANME-2 genome is limited to a few studies: 11 ANME-2a originated fosmids, totaling ∼367 Kb sequence data (Hallam et al., 2004); short sequence read on ANME-2c of 99 bp each (∼380 Kb in total) (Pernthaler et al., 2008); and a recently near completed genome for ANME-2d (with a genome size of 3.2 Mbp) capable of methane oxidization coupled to nitrate reduction (Haroon et al., 2013). The genomic data on ANME-2a reported here will improve our understanding on the molecular mechanisms of AOM substantially.

Table 1 Identification of methanogenesis-associated genes in the ANME-2a enrichment sample
Full size table

The metatranscriptome of the enrichment contained 1.50E+07 sequence reads, 2262.9 Mb of sequences, and 43.3% of which were non-rRNA thus used for further analysis. Approximately 45% of all non-rRNA transcripts in the metatranscriptome were of archaeal origin (determined by MEGAN+blastx results), and the majority (57%) of these archaeal sequences showed highest similarities to Methanosarcinales. The above data sets were exploited to a detailed investigation on the methane metabolizing pathway in ANME-2a (Table 1 and Supplementary Tables S3 and S4).

Methane-oxidizing pathway

As shown in Table 1, all genes that encode enzymes responsible for the seven central steps in the methanogenic pathway have been identified in the single-aggregate M25 genome and have corresponding transcripts in the metatranscriptome. Our previous activity tests on the same enrichment have proven that the sulfate reduction is dependent on methane oxidation and no methane production was observed when methane was eliminated from the medium (Zhang et al., 2010). ANME-2a is demonstrated to have all the required genes for a complete methane-oxidizing pathway from CH4 to CO2 following reversed-methanogenesis hypothesis, and all these genes were actively expressed during the cultivation (Figure 1 and Table 1). Although we cannot completely rule out the possibility of methane productions by ANME-2a under certain conditions (such as the in situ environmental conditions), under the controlled incubation condition where the enrichment was generated, methane oxidation instead of methane production is solely observed.

Figure 1
figure 1

The proposed methane-oxidizing pathway and energy-converting mechanisms in ANME-2a. Only positive gene identifications were displayed (in boxes). The carbon flow was demonstrated with black arrows; the electron flow was indicated with grey arrows. Detailed information of methanogenesis-associated genes is displayed in Table 1, and the names of genes involved in electron transport are displayed in Supplementary Table S3.

Full size image

From our M25 assembly, two complete mer genes were identified with 49% sequence identity and designated as mer-1 and mer-2. Phylogenetic analysis of Mer clearly showed a vertical transfer of mer gene sequences within different orders of methanogenic archaea (Figure 2). The gene mer-1 clustered closely with that from ANME-2d (Haroon et al., 2013), classified within the Methanosarcinales, to which ANME-2 belongs, whereas mer-2 substantially differed from mer genes of all known methanogenic archaea. The phylogenetic position of mer-2 is still unclear, it may come from an unknown archaeon through horizontal gene transfer. The mer-1 gene was adjacent to fpoF (the gene encoding F420-phenazine oxidoreductase subunit F). This type of gene organization is conserved in Methanosarcina species (Baumer et al., 2000; Kulkarni et al., 2009) (Supplementary Figure S5). The finding of canonic mer is suggesting that the conversion of methyl-H4MPT to methylene-H4MPT in ANME-2 is the reversal of the corresponding reaction operating forward in methanogenesis; the same reversal reaction was not identified in ANME-1 (Hallam et al., 2004; Meyerdierks et al., 2010). The mer-2 gene was adjacent to one of two copies of mtr (the gene encoding tetrahydromethanopterin S-methyltransferase), which was designated as mtr-2 (Supplementary Figure S6). The whole sequence identity between two mtr sets is ∼40%, with the highest identity (45%) found between mtrA and the lowest (27%) between mtrC. Considering the low sequence similarity between the two sets of mtr genes, it is unlikely that one set of the mtr genes is an artifact resulting from MDA. The Mtr methyltransferase complex MtrA-H catalyzes methyl transfer from N5-methyltetrahydromethanopterin (or N5-methyltetrahydrosarcinapterin) to coenzyme M in methanogenic archaea (Thauer, 1998). The presence of two sets of mer and mtr in this ANME-2a is intriguing and, to our knowledge, there is no report of an organism possessing two mtr complex sets. It is still a question of whether these two sets of mer and mtr come from a single organism, as M25 contained multiple cells that may have undertaken extensive genomic differentiation. Nevertheless, the data presented here demonstrate that ANME-2a archaeal group harbors two sets of mer and mtr genes. The expression levels of mer-1 and mtr-1 were nearly one order of magnitude higher than the expression levels of mer-2 and mtr-2, respectively, as evaluated by the gene expression RPKM values (Table 1), suggesting a major role of mer-1 and mtr-1 in methane oxidation. It remains unclear whether mer-2 and mtr-2 take part in the same enzymatic reactions as mer-1 and mtr-1 or have other functions and why ANME-2a maintains an additional set of these genes.

Figure 2
figure 2

The maximum likelihood tree showing the deduced amino acid sequences of mer-1 and mer-2 genes to the selected reference sequences. A total of 25 full-length amino acid sequences of mer were aligned. Bootstrap values were based on 1000 replicates and shown at the nodes. Mer-1 and Mer-2 from M25 identified in the single-aggregate genome were highlighted, with their detailed information displayed in Table 1.

Full size image

H2-independent energy-converting mechanisms

No canonical hydrogenase was found in either genomic or transcriptomic data, and thus ANME-2a from our enrichment is most probably performing AOM without H2 as the intracellular electron carrier or intercellular electron shuttle to SRB. Instead, all of the components in the energy-conserving electron transport pathways of non-H2-utilizing methanogenic archaea, in particular genes for the membrane-associated Coenzyme B–Coenzyme M heterodisulfide reductase HdrDE, genes for CytC (cytochrome c) and genes for the Rnf (Rhodobacter nitrogen fixation) complex, which are missing in the H2-utilizing species, are all identified and expressed in ANME-2a (Supplementary Table S3). Therefore, versatile electron transfer pathways independent of H2 are postulated within ANME-2a cells. Redox components F420/ F420H2 (the oxidized/reduced forms of a 5′deazaflavin derivative with high concentrations in methanogenic archaea), Fdox/Fdred (the oxidized/reduced forms of ferredoxin) and MP/MPH2 (the oxidized/reduced forms of membrane-soluble methanophenazine (MP)) can be recycled to couple methane oxidation with adenosine triphosphate (ATP) synthesis through multiple pathways (Figure 1). For example, the oxidation of F420H2 can be completed via the F420H2:heterodisulfide oxidoreductase system while reducing CoM-S-S-CoB, comprising membrane-bound HdrDE and Fpo, as commonly found in Methanosarcina (Baumer et al., 2000; Deppenmeier and Mueller, 2008; Kulkarni et al., 2009). Similarly, Fdred/Fdox is cycled via a Fdred:heterodisulfide oxidoreductase system including CytC, the Rnf complex and HdrDE that results in a sodium gradient that is exchanged for a proton gradient (Ferry and Lessner, 2008; Wang et al., 2011). This is in agreement with the previous studies that the electron transport pathways in non-H2-utilizing marine methanogenic species of the Methanosarcina genus are fundamentally distinct from those in H2-utilizing freshwater species (Guss et al., 2009). For example, the non-H2-utilizing, marine methanogen Methanosarcina acetivorans replaces the Ech hydrogenase with the Rnf complex to generate a transmembrane ion gradient for ATP synthesis, which has been interpreted as an adaptation strategy of this species to the marine environment (Wang et al., 2011). In addition, cytoplasmic HdrABC, another type of heterodisulfide reductase, has been detected with multiple copies (Supplementary Table S3). From the phylogenetic analysis, our ANME-2a-originated hdrA genes were placed into clades with methyltrophic or/and acetoclastic methanogenic archaea but not H2-dependent methanogenic archaea (Supplementary Figure S7) (Buan and Metcalf, 2010). We note here that the identified electron transport components HdrDE and the Rnf complex in ANME-2a have not been reported in ANME-1 metagenomes (Hallam et al., 2004; Meyerdierks et al., 2010), indicating that the methane-oxidizing archaea ANME-1 and ANME-2 may have evolved distinct electron transport pathways or strategies, which warrant detailed investigations in the future.

Acetate-metabolizing possibility

Interestingly, Acd (ADP-forming acetyl-CoA synthetase), a recently identified enzyme in archaea that catalyzes the conversion of acetyl-CoA to acetate coupled with the conversion of ADP to ATP (acetyl-CoA+ADP+Pi→acetate+ATP+CoA) (Brasen et al., 2008), was detected in the single-aggregate genome and the metatranscriptome (Supplementary Table S4). This finding supports the possible integration of acetate to acetyl-CoA at the cost of ATP before entering reverse methanogenesis and/or biomass synthesis. As a corollary, if acetate from acetyl-CoA is excreted, it could be utilized by SRB. The potential for ANME-2a to metabolize acetate is further supported by its possession of an Rnf complex in the energy-converting system that has only been found in non-H2-utilizing acetoclastic Methanosarcinas species (Ferry and Lessner, 2008; Wang et al., 2011). The utilization of acetate as an intermediate contradicts results obtained from in vitro experiments (Nauhaus et al., 2005), but is consistent with extreme 13C depletion in acetate with δ13C values approaching −90‰ in some methane seep sediments (Heuer et al., 2006). The possibility that the ANME-2a from our enrichment utilizes other methylated C1 compounds such as methylamines or methylsulfides for AOM (Moran et al., 2008) is not supported by this study (Supplementary Table S4).

Ecological and evolutionary implications

Although we cannot entirely rule out the potential presence of hydrogenase because of incompleteness of our current ANME-2a genome, the results from this study suggest that the interspecies hydrogen transfer does not play a major role in forming syntrophic consortia of ANME-2a and bacteria in cold seep ecosystems. It is commonly assumed that interspecies electron transfer is a key process in shaping syntrophic communities, especially in methanogenic, anaerobic methanotrophic, sulfate-reducing and even subsurface ecosystems (Nealson et al., 2005; Stams and Plugge, 2009). The utilization of H+/H2 as the electron shuttle is facilitated by the relatively easy transfer of protons and the lacking requirement of enzymes with complex active centers (Vignais and Billoud, 2007). On the other hand, the low midpoint redox potential of this redox couple (E°=−414 mV, which is lower than those of NAD+/NADH, FADH/FADH2, Fdox/Fdred) causes an energetic problem unless hydrogen is continuously produced and removed by two distinct organisms. A recent publication described a mechanism in which zero-valent sulfur produced via sulfate reduction by ANME-2 served as intermediate for an ANME/SRB syntrophic community and was disproportionated by SRB (Milucka et al., 2012). It needs to be tested in the future whether sulfur/acetate rather than H2 serves as electron shuttle between ANME-2a and its SRB partner.

This report demonstrates that ANME-2a oxidizes methane to CO2 with genes from versatile electron transport pathways that do not likely involve H2, and that ANME-2a possesses the potential ability to produce and utilize acetate. It is shown that ANME-2a in our enrichment clearly shares similarities with members of the Methanosarcinales, in particular with non-H2-utilizing acetoclastic Methanosarcina species. Both ANME-2 and Methanosarcina contain cytochrome and MP and they perform complete reverse or forward methanogenesis respectively, while generating ATP without hydrogenase. These features directly reveal a close evolutionary relationship between ANME-2 and the Methanosarcinales and reflect the natural habitat of ANME-2. As the only order of methanogenic archaea with cytochromes, Methanosarcinales requires a higher threshold of H2 partial pressure to produce methane from H2 and CO2 compared with methanogenic archaea without cytochromes. Therefore, some species from Methanosarcinales, including the ancestors of ANME-2, may have lost the ability to metabolize H2, whereas most species can grow on acetate and C1 compounds (Bonin and Boone, 2006). In addition, compared with methanogenic archaea from fresh water, marine isolates tend to forego H2-dependent electron transport pathways to outcompete with other organisms because of the easy loss of freely diffusible H2 gas within marine environments (Guss et al., 2005). Moreover, the possession of cytochromes constrains Methanosarcinales to habitats colder than 60 °C, that is, a temperature above which their growth is energetically unfavorable (Conrad and Wetter, 1990; Kotsyurbenko et al., 2001). This is consistent with the finding that ANME-2 preferentially inhabits in low-temperature settings, whereas ANME-1 is widely distributed even in high-temperature environments such as hydrothermal vents (Holler et al., 2011; Rossel et al., 2011; Biddle et al., 2012). Furthermore, genes involved in a broad substrate utilization spectrum and versatile energy-converting mechanisms are expressed in ANME-2, highlighting the flexibility of ANME-2 to adjust to fluctuating environments with intermittently high redox potentials, such as sporadically oxygenated shallow marine sediments (Knittel et al., 2005; Lazar et al., 2011; Rossel et al., 2011).

Our study of an ANME-2a enrichment demonstrated that its methane-oxidizing and electron-transporting pathways are distinct from ANME-1 and advanced our understanding on the globally relevant AOM biogeochemical process. These results have important implications for the evolution, physiology and niche differentiation of different ANME clades.