Reverse Methanogenesis: Testing the Hypothesis with Environmental Genomics

See allHide authors and affiliations

Science  03 Sep 2004:
Vol. 305, Issue 5689, pp. 1457-1462
DOI: 10.1126/science.1100025


Microbial methane consumption in anoxic sediments significantly impacts the global environment by reducing the flux of greenhouse gases from ocean to atmosphere. Despite its significance, the biological mechanisms controlling anaerobic methane oxidation are not well characterized. One current model suggests that relatives of methane-producing Archaea developed the capacity to reverse methanogenesis and thereby to consume methane to produce cellular carbon and energy. We report here a test of the “reverse-methanogenesis” hypothesis by genomic analyses of methane-oxidizing Archaea from deep-sea sediments. Our results show that nearly all genes typically associated with methane production are present in one specific group of archaeal methanotrophs. These genome-based observations support previous hypotheses and provide an informed foundation for metabolic modeling of anaerobic methane oxidation.

Anaerobic oxidation of methane (AOM) in marine sediments has been estimated to consume more than 70 billion kilograms of methane annually (1). Analyses of pore waters from methane-oxidizing sediments along continental margins have mapped extensive zones of sulfate and methane depletion, which define the geographic and geochemical boundary conditions for AOM (24). Combined geochemical and biological evidence indicate that microbial consortia, largely composed of archaea and sulfate-reducing bacteria (SRB), can couple methane oxidation to sulfate reduction (5, 6). Current models suggest that methane is converted by methanotrophic archaea to carbon dioxide and reduced by-products (possibly including molecular hydrogen), which are subsequently consumed by sulfate-reducing bacteria (6). In anoxic deep-sea sediments, AOM catalyzes the formation of authigenic carbonates with highly depleted 13C content, thereby providing an enduring geochemical signature for past and present methane oxidation (79). Microbial mediation of AOM significantly influences both local and global biological and biogeochemical processes. The process reduces methane flux to the water column, stimulates subsurface microbial metabolism, and also supports vigorous deep-sea chemolithotrophic communities that derive energy from one of its by-products, hydrogen sulfide.

Although no archaeal methanotrophs have yet been isolated in pure culture, phylogenetic, isotopic, and biochemical analyses indicate that several different methanogen-related archaeal groups are involved in AOM (1013). Two groups of putative anaerobic methane-oxidizing Archaea (ANME-1 and ANME-2) (10, 11) and several SRB groups typically occur together in methane-rich marine sediments, although environmental surveys and incubation studies have identified distinct population structures and distributions associated with specific habitats (10, 1417). The extent to which ANME and SRB groups cooperate in AOM is uncertain, but specific physical associations between them have been observed (11, 15).

To better define the process of AOM, we used environmental genomic techniques (18) to analyze methane-oxidizing archaeal populations found in deep-sea methane seeps. Our samples originated from a 6- to 9-cm-deep sediment pushcore interval (PC45) obtained from the Eel River Basin off the Mendocino California coastline (19). Previous geochemical and chemotaxonomic analyses of the sampling site determined that ANME and SRB groups represent active and abundant members of a microbial community associated with AOM in Eel River sediments (14, 15, 17). Microbial cells, including ANME-1, ANME-2, and associated SRB, were enriched from the sediment using density centrifugation and size selection (19). High-molecular-weight DNA purified from this cell enrichment was used to construct several 3000– to 4000–base pair (3- to 4-kbp) insert whole-genome shotgun (WGS) libraries, and one 32- to 45-kbp insert fosmid library (19). A total of 111.3 million base pairs (Mbp) of DNA sequence generated from 224,736 reads (averaging 732 bp per read) was derived from the WGS libraries. Paired-end sequencing of the fosmid library generated 4.6 Mbp of DNA sequence from 7104 reads (averaging 700 bp per read). Then, 191 fosmids encompassing 7.4 Mbp of DNA were selected for subcloning and sequencing on the basis of the results of pairedend sequencing, small-subunit ribosomal RNA (SSU rRNA), and functional gene screening (19). Fosmid-sequencing efforts focused on archaeal clones to maximize large-insert coverage depth of ANME genomes.

The cell purification procedure used in sample preparation was intended to reduce the complexity in the original, diverse sediment-associated microbial community and to enrich for the AOM microbial consortia. Based on SSU rRNA gene representation in both the WGS and fosmid libraries, the microbial community structure was dominated by ANME-1, ANME-2, and SRB groups (Fig. 1). The dominance of ANME-1 in the purified cell population is also supported by the distribution and types of methane-oxidizing Archaea (MOA)–specific methyl coenzyme M reductase (MCR) subunit A (mcrA) genes present in the library (figs. S1 and S2) (17). This enrichment of ANME cells and genomic DNA facilitated detailed genomic analyses of this population subset.

Fig. 1.

Taxonomic distribution of SSU rRNA sequences identified in whole-genome shotgun sequencing (n = 114) and fosmid DNA libraries (n = 18).

ANME-1 and ANME-2 SSU rRNA and mcrA genes encoded on large genomic fragments formed distinct groups and subdivisions, each showing specific substitutions, transpositions, and indels. However, over the length of each fragment, gene content and operon organization were highly conserved within any given subdivision (Fig. 2 and fig. S1). In several instances, gene content was shared among fosmids from different groups containing SSU rRNA or mcrA, or between fosmids containing SSU rRNA and mcrA (Fig. 2). On average, the G+C content of ANME-1 fosmids containing SSU rRNA and mcrA was 45.1%, compared with 51.1% for those of ANME-2 (table S2). Subgroups of ANME-1 and ANME-2 fosmids containing SSU rRNA or mcrA harbored additional methanogenesis-associated genes (Figs. 2 and 3; table S2), providing linkage information used in determining the origin of related sequences in the WGS and fosmid libraries.

Fig. 2.

Comparison of ANME-1 and ANME-2 fosmids containing SSU rRNA (A) or mcrA (B) based on predicted gene content and order. Genes shared in common among or between fosmids are connected by shaded boxes. ANME-1 mcrBGA subunits are separated from the mcrC component by a spacer region varying between ∼1 and 14 kbp from the predicted mcrB start site. In every case, the mcrC component is linked to a tandem duplication of the atw locus, an adenosine 5′-triphosphate (ATP)–binding protein associated with activation of the MCR holoenzyme in vitro (26).

Fig. 3.

Determination of ANME-1 or ANME-2 fosmid identity based on G+C content and depth of WGS coverage. Bin I bounded by red ellipse corresponds to ANME-1. Bin II bounded by orange ellipse corresponds to ANME-2. Fosmids containing SSU rRNA (filled circle), mcr (filled square), and mch (open circle) genes are highlighted in red (ANME-1) or orange (ANME-2). Read depth corresponds to the total number of WGS nucleotides aligning to a given fosmid, divided by the length (in bps) of that fosmid (19).

Surveys of the environmental libraries revealed the presence and relative abundance of many genes encoding enzymes typically associated with the methanogenic pathway (Table 1, Fig. 4, and fig. S2) (1921). With the exception of step 5, encoded by F420-dependent N5, N10-methenyltetrahydromethanopterin (methylene-H4MPT) reductase (mer), components of all enzymatic steps (steps 1 to 4 and steps 6 to 7, Table 1) were represented in both WGS and fosmid library data sets (Table 1 and table S2). Four sequences encoding mer were encountered only in the WGS data set, but these probably reflect a low-level presence of bona fide methanogens in the sample. This observation is consistent with previous SSU rRNA surveys in Eel River sediments, where a few acetoclastic methanogen sequences occurred together with ANME-1 and/or ANME-2 ribotypes (14). Although no bona fide methanogen SSU rRNAs were identified in either the WGS or fosmid library sequences, several mcr subunit sequences affiliated with the Methanosarcinales lineage were identified (fig. S2). These are readily distinguished from ANME gene fragments by both their phylogeny and WGS coverage.

Fig. 4.

Hypothetical model for reverse methanogenesis in ANME-1. (A) A combined pathway for methanogenesis. Gene identifications are shown in black. (B) A reconstructed pathway for ANME-1 based solely on predicted gene content of identified ANME-1 fosmids. Positive gene identifications are shown in red. Negative gene identifications are shown in gray.

Table 1.

Identification of methanogenesis-associated genes in Eel River sediment genomic DNA libraries. Step numbers indicate points in the H4MPT–dependent methanogenic pathway. Total number of gene ids among libraries: Identification based on tblastN results constrained to expectation cut-off >E – 10. Positive identifications (ids) are indicated by numbers, and negative ids are indicated by (–) and underlined gene name and locus.

Step Gene name Total no. of gene ids among libraries
Locus Shotgun Fosmid ends Completed fosmids
1 Formylmethanofuran dehydrogenase, subunit A fmdA 55 12 4
subunit B fmdB 69 6 4
subunit C fmdC 50 1 4
subunit D fmdD 23 1 2
subunit E fmdE 26 2 2
subunit F fmdF 9View inline 2 -
subunit G fmdG - - -
subunit H fmdH - - -
2 Formylmethanofuran-tetrahydromethanopterin formyltransferase ftr 67 7 3
3 N5,N10-methenyltetrahydromethanopterin cyclohydrolase mch 29 1 5
4 F420-dependent methylenetetrahydromethanopterin dehydrogenase mtd 25 2 2
H2-forming N5,N10-methylene-tetrahydromethanopterin cyclohydrolase hmd - - -
5 Coenzyme F420-dependent N5,N10-methenyltetrahydromethanopterin reductase mer View inline 4 - -
6 N5-methyltetrahydromethanopterin-coenzyme M methyltransferase, subunit A mtrA 9 - 8
subunit B mtrB 3 1 3
subunit C mtrC 10 1 3
subunit D mtrD 10 2 3
subunit E mtrE 11 - 3
subunit F mtrF - - -
subunit G mtrG 7 - 4
subunit H mtrH 39 4 6
7 Methyl coenzyme M reductase, subunit α mcrA 45 3 11
subunit β mcrB 28 9 11
protein C mcrC 33 1 13
protein D mcrD 2 1 3
subunit γ mcrG 31 1 11
Heterodisulfide reductase, subunit A hdrA 319View inline 22 4
subunit B hdrB 80 3 8
subunit C hdrC 45 4 3
subunit D hdrD 67View inline 4 -
subunit E hdrE View inline 7 - -
CO dehydrogenase/acetyl-CoA synthase, subunit α cdhA 74 8 8
subunit β cdhC 52 1 3
subunit γ cdhD 38 3 1
subunit δ cdhE 56 3 4
subunit ϵ cdhB 16 - 4
ADP-forming acetyl-CoA synthetase acd 139View inline 9 4
  • View inline* Includes related iron-sulfur [Fe-S] proteins.

  • View inline Includes related coenzyme A-binding proteins.

  • View inline Identified only in shotgun.

  • Fosmid sequences were compared on the basis of their G+C content and WGS coverage (19). This approach was chosen on the basis of the clear G+C bias between ANME-1 and ANME-2, as well as the apparent high representation of ANME-1 genomic DNA (Fig. 1) in the WGS and fosmid libraries. Two bin distributions, I and II, were evident from using this approach (Fig. 3 and table S2). The depth of WGS coverage for bin I ranged between 0.3 and 7.9 × and between 0.4 and 1.4 × for bin II (Fig. 3). All ANME-1 fosmids containing SSU rRNA or mcrA mapped to bin I, and all ANME-2 fosmids containing SSU rRNA or mcrA mapped to bin II (Fig. 3 and table S2). Independent phylogenetic and linkage analyses clearly identified a total of 16 ANME-1 fosmids, all of which grouped in bin I (Fig. 3). Similarly, all five fosmids that could be unambiguously identified as ANME-2 grouped into bin II (Fig. 3). Assembly of binned fosmids generated 13 unique scaffolds from within bin I, and one from within bin II, with no cross-assembly between the bins (22). Together, these data provide strong support for the assignment of fosmids encoding methanogenesis-associated genes to ANME-1 or ANME-2 groups, according to their bin distribution. Specific identification of many ANME-1–derived genome fragments provided the framework necessary for modeling a presumptive pathway for methane consumption within this group.

    The available data strongly suggest that the ANME-1 group contains all steps in the canonical seven-step methanogenic pathway with the exception of step 5, encoded by mer (Table 1 and Fig. 4). Although this gene is required for methanogenesis from CO2 and one carbon (C1) compounds including methanol and methylamines, loss of mer activity in ANME-1 could promote AOM by increasing the activation barrier for conversion of methylene-H4MPT to methyl-H4MPT. Given this observation, identification of methanofuran (MF)/H4MPT-dependent C1 transfer enzymes mediating steps 1 to 4 of methanogenesis in ANME-1 sequences is intriguing. It is possible that methylene-H4MPT derived from reduced CO2 becomes a substrate for assimilatory metabolism via the serine cycle (23). Alternatively, the C1 transfer module in ANME-1 could play a role in formaldehyde detoxification, analogous to the properties of other methylotrophic (24) or nonmethylotrophic bacteria (25).

    An F420-dependent quinone oxidoreductase (fqo) and numerous iron-sulfur cluster proteins were identified among the ANME-1 sequences (fig. S4, table S2). Moreover, electron input modules encoded by coenzyme F420-reducing hydrogenase subunit B (frhB) were identified on 12 ANME-1 fosmids containing methanogenesis-associated genes (table S2), which suggested possible coupled expression and functioning of these enzymes, as well as the generation of a proton motive force derived from reduced F420 or ferredoxin. Given these observations, the “unfavorable” thermodynamics of methane activation in AOM might be overcome by metabolic coupling to the energy conservation reactions driven by the F420-dependent respiratory chain. In ANME-1, fqo is most similar in operon structure and gene sequence to homologous genes in Archaeoglobus fulgidus (fig. S3), which indicates that ANME-1 contains genomic features of both sulfate-reducing and methanogenic Archaea.

    The identification of most of the genes associated with methanogenesis in the ANME-1 group (and to a lesser extent, ANME-2) lends strong support to the reverse-methanogenesis hypothesis. The presence of genes that typify methane production in methanotrophic Archaea renders some of the classical molecular biomarkers of methanogenesis somewhat ambiguous. At the same time, these data provide new insight into the evolution, ecological roles, and diversity of methane-cycling Archaea and their unique metabolic machinery. The data also facilitate a more mechanistic biological understanding of the environmentally significant biogeochemical process of methane oxidation in anoxic marine habitats.

    Supporting Online Material

    Materials and Methods

    Figs. S1 to S3

    Tables S1 and S2

    References and Notes

    References and Notes

    View Abstract

    Navigate This Article