More Than 200 Genes Required for Methane Formation from H2 and CO2 and Energy Conservation Are Present in Methanothermobacter marburgensis and Methanothermobacter thermautotrophicus

The hydrogenotrophic methanogens Methanothermobacter marburgensis and Methanothermobacter thermautotrophicus can easily be mass cultured. They have therefore been used almost exclusively to study the biochemistry of methanogenesis from H2 and CO2, and the genomes of these two model organisms have been sequenced. The close relationship of the two organisms is reflected in their genomic architecture and coding potential. Within the 1,607 protein coding sequences (CDS) in common, we identified approximately 200 CDS required for the synthesis of the enzymes, coenzymes, and prosthetic groups involved in CO2 reduction to methane and in coupling this process with the phosphorylation of ADP. Approximately 20 additional genes, such as those for the biosynthesis of F430 and methanofuran and for the posttranslational modifications of the two methyl-coenzyme M reductases, remain to be identified.


Introduction
In 1972, Zeikus and Wolfe [1] isolated Methanothermobacter thermautotrophicus (DSM 1053) (formerly Methanobacterium thermoautotrophicum strain ΔH) from sludge from the anaerobic sewage digestion plant in Urbana, Illinois, USA. This thermophile grew on H 2 and CO 2 as sole energy source (reaction 1) and CO 2 as sole carbon source with doubling times of less than 5 h and to very high cell concentrations (1.5 g cells (dry mass) per L). For the first time, it became possible to obtain sufficient cell mass of a hydrogenotrophic methanogen for the purification of enzymes and coenzymes involved in CO 2 reduction to methane. In 1978, Fuchs et al. [2] reported the isolation of Methanothermobacter marburgensis (DSM 2133) (formerly Methanobacterium thermoautotrophicum strain Marburg) from anaerobic sewage sludge in Marburg, Germany.
The Marburg strain grew on H 2 and CO 2 even faster (doubling time less than 2 h) and to even higher cell concentrations (3 g cells (dry mass) per L) than the ΔH strain and was, therefore, subsequently used in Marburg and elsewhere for the study of methanogenesis. Most of what is presently known about the biochemistry of CO 2 reduction to methane with H 2 was worked out with either M. thermautotrophicus [3] or M. marburgensis [4][5][6] 4H 2 + CO 2 −→ CH 4 + 2H 2 O, ΔG • = −131 kJ mol −1 (1) (ÅG • calculated for H 2 , CO 2 , and CH 4 in the gas phase).
The genome of M. thermautotrophicus (NC 000916) was one of the first genomes of Archaea to be sequenced [7] and that of M. marburgensis (NC 014408/CP001710) has just

The Phenotypes of M. marburgensis and M. thermautotrophicus
M. marburgensis differs from M. thermautotrophicus not only in the growth rate and the final cell density reached but also in the composition of the pseudomurein sacculus (galactosamine instead of glucosamine), in the size of the subunit O of DNA-dependent RNA polymerase (120 kDa instead of 96 kDa), and in the membrane-associated ATPase activity (<0.1 μmol min −1 mg −1 versus 1.4 μmol min −1 mg −1 ) [19]. Unlike M. thermautotrophicus, M. marburgensis contains a 4439 bp circular multicopy plasmid (pME2001 = pMTBMA4; NC 014409) [20,21]. M. marburgensis is specifically infected and lysed by the phage ΨM1 [22,23], whereas M. thermautotrophicus is specifically infected by the phage ΦF1 [24]. Prophage sequences have not been found in either genome sequence; however, such a sequence has been identified in the genome sequence of the closely related Methanothermobacter wolfeii [25].
The two Methanothermobacter species have in common a growth temperature optimum near 65 • C and the ability to grow on H 2 and CO 2 as carbon and energy source, NH 3 as nitrogen source, and H 2 S or sulfite but not sulfate as sulfur sources. Methanothermobacter species all require Na + , K + , Fe 2+ , Co 2+ , Ni 2+ , Zn 2+ , MoO 2− 4 and/or WO 2− 4 , and possibly Ca 2+ for growth [26,27]. The sodium requirement is in the millimolar range. Their growth is not stimulated by the addition of organic compounds although formic acid [28], acetic acid [29], propionic acid [30], pyruvate [31], isobutyric acid, isovaleric acid, phenylacetic acid, phydroxyphenylacetic acid, indoleacetic acid [32], succinic acid [33], δ-aminolevulinic acid [34], methionine [35], guanine [36], and biotin [37] can be assimilated. The ability of the two Methanothermobacter species to assimilate formate, however, does not mean that they can grow on it as energy source, which is an ability of the related strains M. thermautotrophicus strain Z-245 and Methanothermobacter wolfeii [17,38].
Like all members of the Methanobacteriales, M. marburgensis and M. thermautotrophicus are not motile, do not conjugate, and are devoid of heme proteins and methanophenazine and, as mentioned above, selenocysteine proteins. Accordingly, their genomes lack CDS for these functions, with a few exceptions. Each of their genomes contains CDS predicted to encode a homolog of selenophosphate synthase (SelD) (MTBMA c04350; MTH1864) and of selenocysteine synthase (SelA) (MTBMA c04850; MTH1914), which catalyzes the formation of Sec-tRNA Sec from Ser-tRNA Ser using selenophosphate as selenium donor. The SelA homolog appears to be restricted to these two Methanothermobacter species, whereas a CDS for the SelD homolog has been found in every methanogen genome sequenced to date. In nonselenoprotein-containing methanogens, SelD may function in the synthesis of selenouridine-modified tRNAs and/or of selenium-dependent molybdenum hydroxylases, which some methanogens could contain [13,39,40].
Methanothermobacter species are not only found in anaerobic sewage sludge but also in anoxic freshwater sediments [17]. In these anoxic environments, the temperature is usually below 20 • C and thus well below the observed temperature growth range of thermophiles. The origin of Methanothermobacter species in such mesophilic and psychrophilic habitats is uncertain. It is possible that the thermophiles originated from thermophilic anoxic environments, such as nonmarine hot springs [42], but when this could have occurred is unknown. The kinetics and energetics of growth only of M. marburgensis have been analyzed in detail. The methanogen grows at 65 • C with doubling times of 1.6 h when optimally gassed with 80% H 2 and 20% CO 2 at 10 5 Pa [27]. The apparent K m determined with growing cultures is 20% H 2 and 10% CO 2 [27]. From the apparent K m , it can be calculated that the doubling time increases to over 100 days when the H 2 partial pressure is below 10 Pa, as in the anaerobic sewage sludge from which the organism was first isolated. From the digester dilution rates, a doubling time of at least 30 days is predicted. At 10 Pa, the free energy change associated with reaction 1 is only −40 kJ mol −1 , which can support the synthesis of less than one ATP from ADP and inorganic phosphate.

Comparison of the M. marburgensis and M. thermautotrophicus Genomes
To compare the genomes of M. marburgensis and M. thermautotrophicus, we used a two-step approach. In the first step, we used a bidirectional search method that identifies the most similar protein and RNA (tRNA, rRNA, and ribozyme) encoding sequences in the two genomes and allows a sequence to be used only once in the comparison. Each pair of CDS identified in this way was kept. Neither the second-nor third-best hits nor CDS homologs within the same organism were considered. Therefore, if a sequence is not common to the two organisms, this does not mean that there are no paralogs or orthologs of this sequence in the two organisms. Two sequences with a basic local alignment search tool (BLAST) expectation value (E-value) in the NCBI database less than 10 −8 were considered to be of a common origin. A cutoff at lower E-values, for example, at an E-value of 10 −25 , would have resulted in 40 fewer common CDS, and among these would have been several CDS for proteins with a known function in both organisms, for example, for four ribosomal proteins. In the second step, we aligned fulllength sequences using optimal global alignment [43]. Pairs of proteins with full-length alignments with ≥10% identity at the amino acid level were considered as putative orthologous proteins. Using this method and cutoffs, the genomes of M. marburgensis and M. thermautotrophicus were found to have 1,607 CDS in common, 411 CDS not in common, 39 RNA-coding sequences in common and 1 RNA-coding sequence not in common ( Table 1). The two genomes show a high degree of synteny ( Figure 1).

CDS in Common
Of the 1,607 CDS common to M. marburgensis and M. thermautotrophicus, some encode proteins with identical or almost identical sequences, and others encode proteins with only a low level of sequence identity, which reflects either large differences in sequence divergence or orthologous gene replacements. Only approximately 57% of the deduced amino acid sequences of the common CDS have Evalues <10 −100 and corresponding optimal global similarityalignment scores >89.2%; approximately 28% have E-values between 10 −100 and 10 −50 and similarity-alignment scores between 89.2% and 78.3%; approximately 21% have Evalues between 10 −50 and 10 −25 and similarity-alignment scores between 78.3% and 50.0%; 3.7% have E-values between 10 −25 and 10 −8 (cutoff) and similarity-alignment scores between 50% and 10%; only 3.7% have a similarityalignment score of 100. These results indicate that many of the proteins in the two organisms have undergone extensive mutations without having lost their function or that these proteins have no essential function and could, therefore, accumulate mutations extensively. Approximately 30% of the CDS in common encode conserved hypothetical proteins.

CDS for Membrane Proteins and Protein Export.
Approximately 330 of the 1607 CDS in common are predicted to form at least one transmembrane helix indicating their location in the cytoplasmic membrane. Most of the other CDS appear to encode for cytoplasmic proteins. Only very few CDS appear to have a "periplasmic" location. Both genomes lack CDS for a Tat (twin arginine translocation) system involved in the export of proteins with prosthetic groups such as iron-sulfur centers that can only be assembled in the cytoplasm. The lack of a Tat system appears to be a general property of all methanogens lacking cytochromes. Therefore, the two Methanothermobacter species probably do not contain redox-active proteins whose active sites face outwards. This is an issue since there are reports suggesting that one member of the Methanobacteriales, Methanobacterium palustrae, can pick up electrons from the surface of electrodes and use these electrons for the reduction of CO 2 to methane [44,45]. Interesting in this respect is the finding that M. marburgensis and M. thermautotrophicus contain a complete Sec protein export system (Table 1). In principle, the methanogens could, therefore, produce electron-conducting fimbriae (nanowires) [46][47][48] that transfer electrons from the electrode to a cytoplasmic electron acceptor. There is evidence that M. thermautotrophicus can form fimbriae with which the organism may attach to H 2 -forming bacteria [49]. Whether these fimbriae can function as nano-wires is not known, and it is also not known whether M. palustrae has fimbriae when it picks up electrons from electrode surfaces. The other methanogen-specific CDS are for a predicted molybdenum-iron protein (NflD) homologous to NifD, for a radical-S-adenosylmethionine (SAM) protein homologous to NifB, for a homolog of selenophosphate synthetase (SelD), for a methyltransferase related protein (MtxX), for a peptidyl-prolyl cis-trans isomerase-related protein, for a predicted UDP-N-acetylmuramyl pentapeptide synthase, for a predicted DNA-binding protein, for a predicted metalbinding transcription factor, for a predicted phosphomannomutase and for 12 conserved hypothetical proteins.

Methanogen
One of the methanogen-specific CDS, namely, mcrA, is used as a specific marker for methanogenic archaea and anaerobic archaea that contain methyl-coenzyme M reductase and oxidize methane [50].

Methanothermobacter-and Methanobacteriales-Specific
CDS. Of the CDS in M. marburgensis, 177 have a counterpart only in M. thermautotrophicus; 140 of these are for hypothetical proteins. Ninety-one CDS have a counterpart only in M. thermautotrophicus, Methanobrevibacter smithii, Methanobrevibacter ruminantium, and Methanosphaera stadtmanae, all members of the order of Methanobacteriales. Of these 91 CDS, 67 are for hypothetical proteins. We expect that the Methanothermobacter-specific and the Methanobacteriales-specific CDS are for anabolic (biosynthesis) rather than for catabolic (energy metabolism) functions. An exception is MTBMA c06120, which is one of three CDS predicted to encode coenzyme F 390 synthetase in both Methanothermobacter species. This enzyme catalyzes the conversion of coenzyme F 420 to a redox-inactive form, which stops methanogenesis from H 2 and CO 2 [52].

CDS Not in Common
The genome of M. marburgensis also contains 145 CDS not present in M. thermautotrophicus (Supplementary Table 2), and the genome of M. thermautotrophicus also contains 266 CDS not present in M. marburgensis (Supplementary Table  3). These CDS not in common are dispersed around the two genomes, but many are concentrated at four genome areas ( Figure 1). Their origin was traced back to gene-splitting events (frame shifts caused by single-base insertion/deletion; 15%), gene-deletion events (30%), gene-duplication events (30%), and lateral gene-transfer events (24%). (The percent values given are for M. marburgensis; for the method of determination, see the supplement.) Of the CDS not in common and with an annotated function, 18 CDS in M. thermautotrophicus and 1 CDS in M. marburgensis are predicted to encode Cas proteins, that is, proteins associated with clustered regularly interspaced short palindromic repeats (CRISPR). CRISPR loci encode small RNAs and are, therefore, described in the following subsection. Many of the CDS not in common are predicted to be involved in cell surface sugar biosynthesis (11 CDS in M. marburgensis and 23 CDS in M. thermautotrophicus). One CDS specific for M. thermautotrophicus is for a fimbria protein (MTH60) [49]. This protein shows low sequence similarity to two CDS in each methanogen (MTBMA c07820 and MTBMA c07830; MTH382 and MTH383), which are predicted to encode exported proteins. Whether these proteins also form fimbriae is not known. Only the genome of M. marburgensis has 2 CDS for a putative transposase of the IS630 family (MTBMA c01240 and MTBMA c01250) and 15 IS-like elements. The two CDS for the transposase are preceded and followed by palindromic sequences [53].

RNA-Coding Sequences
The genome of M. marburgensis harbors 40 tRNA-coding sequences, whereas that of M. thermautotrophicus harbors only 39 tRNA-coding sequences. The extra tRNA in M. marburgensis is for serine, for which there are 5 tRNAs in M. marburgensis and 4 in M. thermautotrophicus. The fifth tRNA-Ser coding sequence lies next to that of another tRNAs for serine with the same anticodon. Therefore, the sequence is most likely the result of a gene-duplication event.
In both methanogens, three of the tRNA-encoding sequences, specifically those for tRNA-Trp, tRNA-Met, and tRNA-Pro, carry an intron. Accordingly, the genomes of the two methanogens also encode a tRNA-splicing endonuclease (MTBMA c07000; MTH250).
In the genome of M. marburgensis, there is only one CRISPR locus with 36 repeats, located after MTBMA c02230. In the genome of M. thermautotrophicus, there are three CRISPR loci with a total number of 175 repeats (http://genoweb1.irisa.fr/Serveur-GPO/outils/repeats-Analysis/DOMAIN/indexDOMAIN.php). CRISPR loci encode small CRISPR RNAs (crRNAs) that contain a full spacerflanked by partial repeat sequences. Together with genes encoding Cas (CRISPR-associated) proteins (see above), they protect bacteria and archaea from invasion by phage and plasmid DNA through a genetic interference pathway [54][55][56]. Interestingly, spacer sequences from the CRISPR region of locus 2 from M. thermautotrophicus match to nucleotide sequences found in phage ΨM1 of M. marburgensis and in phage ΨM100 of M. wolfeii [57]; this is in agreement with the observation that M. thermautotrophicus is not lysed by these two phages.

Genes Involved in CH 4 Formation from CO 2 and H 2 and in Energy Conservation
Approximately 90 of the annotated CDS present in both M. marburgensis and M. thermautotrophicus, including those for the methyltransferase MtrA-H, the energy-converting hydrogenases EhaA-T and EhbA-Q, and the A 1 A 0 ATP synthase AhaA-IK, encode proteins directly involved in CO 2 reduction to methane with H 2 and in energy conservation [6]. Another 80 CDS are required for the synthesis of the coenzymes and prosthetic groups, and more than 30 are predicted to have a function in the translocation of ions other than sodium. Their function within energy metabolism is shown in  Table 2. Other CDS remaining to be identified are also listed in Table 2, with numbers in parentheses, such as those for coenzyme F 430 biosynthesis and those for posttranslational modifications in the two methylcoenzyme M reductases. Some of the approximately 200 CDS also have an anabolic function, such as those for the energy-converting hydrogenases EhaA-T and EhbA-Q (ferredoxin reduction, for example, for CO 2 reduction to CO) and for the enzymes involved in the reduction of CO 2 with H 2 to methyl-tetrahydromethanopterin (methyl-H 4 MPT) (autotrophic CO 2 fixation). The methyl group of methyl-H 4 MPT is transferred to CO in a coenzyme-A-dependent reaction, yielding acetyl-CoA, from which most cell compounds are synthesized [58].

Genes for Enzymes Catalyzing H 2 Activation
In the genomes of M. marburgensis and M. thermautotrophicus, CDS for five different hydrogenases are found [5] ( Figure 2, Table 2). Three are mainly involved in CO 2 reduction to methane with H 2 (Figure 2), and two are mainly involved in autotrophic CO 2 fixation. The three mainly catabolic hydrogenases are (a) the cytoplasmic "methylviologen"-reducing [NiFe]-hydrogenase (MvhADG) associated with the heterodisulfide reductase (HdrABC) for the coupled reduction of ferredoxin (Fd) and the heterodisulfide CoM-S-S-CoB with H 2 , (b) the cytoplasmic coenzyme F 420 -reducing [NiFe]-hydrogenase (FrhABG), and (c) the cytoplasmic [Fe]-hydrogenase (Hmd), which substitutes for FrhABG under nickel-limiting growth conditions. The mainly anabolic hydrogenases are the two membraneassociated energy-converting [NiFe]-hydrogenases EhaA-T and EhbA-Q for the reduction of ferredoxin with H 2 .
Under physiological standard conditions [H 2 partial pressure (pH 2 ) =10 5 Pa; pH 7], the H + /H 2 couple has a redox potential E 0 of −414 mV. However, under in vivo conditions (pH 2 ≈10 Pa; pH 7) (see Section 1), the E of the H + /H 2 couple is only −300 mV. The E 0 of the electron acceptors of the hydrogenases are less than −400 mV for ferredoxin (EhaA-T; EhbA-Q), −360 mV for F 420 (FrhABG), −140 mV for CoM-S-S-CoB (MvhADG/HdrABC), and −380 mV for methenyltetrahydromethanopterin (methenyl-H 4 MPT + ) (Hmd) [6]. Under in vivo conditions, the E of the Fd ox /Fd red couple is probably −500 mV, since this is the redox potential of most ferredoxin-dependent reactions in methanogens [6]. Therefore, the reduction of ferredoxin with H 2 requires energy, that of F 420 and of methenyl-H 4 MPT + operates near equilibrium [59], and that of CoM-S-S-CoB is exergonic enough to be coupled with energy conservation.
9.1. MvhADG. The cytoplasmic [NiFe]-hydrogenase (MvhADG) is frequently referred to as methyl-viologenreducing [NiFe]-hydrogenase or F 420 -nonreducing hydrogenase. MvhADG is associated with the cytoplasmic heterodisulfide reductase HdrABC, with which the hydrogenase forms a tight complex. The complex catalyzes the CoM-S-S-CoB-dependent reduction of ferredoxin with H 2 and the ferredoxin-dependent reduction of CoM-S-S-CoB with H 2 . The stoichiometry of ferredoxin (Fd) and CoM-S-S-CoB reduction with H 2 has been determined to be: 2H 2 + Fd ox + CoM-S-S-CoB = Fd 2− red + CoM-SH + CoB-SH + 2H + [60]. Apparently, the MvhADG/HdrABC complex couples the endergonic reduction of ferredoxin with H 2 to the exergonic reduction of CoM-S-S-CoB with H 2 , and it has been proposed that the coupling proceeds via flavinbased electron bifurcation [6]. The reduced ferredoxin generated in the MvhADG/HdrABC-catalyzed reaction is required for the reduction of CO 2 to formylmethanofuran (E 0 = −500 mV) [61]. Evidence was recently provided that MvhADG/HdrABC and formylmethanofuran dehydrogenase form a super complex in the cytoplasm of Methanococcus maripaludis [62].
In M. marburgensis and in M. thermautotrophicus, the genes encoding MvhADG/HdrABC are organized in three nonadjacent transcription units (mvhDGAB, hdrA, and hdrBC). The mvhDGAB operon lies directly downstream of the mtrBDGA operon, which encodes isoenzyme II of methyl-coenzyme M reductase, and there is evidence that the two operons can be cotranscribed [63,64].
mvhA and mvhG encode the large and small hydrogenase subunits, respectively; mvhD encodes a [2Fe2S] clustercontaining subunit; and mvhB encodes a 12[4Fe4S] polyferredoxin, which is probably the ferredoxin reduced by the MvhADG/HdrABC complex ( Figure 2). HdrB harbors the active site for CoM-S-S-CoB reduction and contains zinc and an unusual [4Fe4S] cluster [65]. HdrC harbors two [4Fe4S] clusters, and HdrA contains four [4Fe4S] clusters and FAD. HdrA is considered to be the site of electron bifurcation. Interestingly, HdrA is one of the most highly conserved proteins in all methanogenic archaea and is also found in other archaea and bacteria, which indicates an electron-bifurcating function in these organisms within a different context. Interesting in this respect is that in most methanogens, the CDS for HdrA is located separate from the CDS for HdrBC [66] consistent with HdrA being used not only in combination with HdrBC.
Many members of the Methanomicrobiales lack the genes for the subunits MvhAG. It has been proposed that in these hydrogenotrophic methanogens, FrhAG (see below) rather than MvhAG forms a functional complex with MvhD/HdrABC [5,67]. The finding that in most methanogens the CDS for MvhADG are located separate from those for HdrA and HdrBC [66] is consistent with HdrABC being used not only in combination with MvhADG for which there is genetic evidence in Methanococcus maripaludis [62].

FrhABG. This cytoplasmic [NiFe]
-hydrogenase catalyzes the reversible reduction of coenzyme F 420 with H 2 . The FrhABG complex aggregates to form a complex with a molecular mass of >900 kDa. Upon ultracentrifugation of cell extracts, the F 420 -reducing hydrogenase is recovered in the membrane fraction, which is why it was long believed   Hmd harbors a novel iron-guanylylpyridinol (FeGP) cofactor covalently bound to the homodimeric enzyme only via the thiol/thiolate group of a cysteine residue. The enzyme and the cofactor are not found in most members of the Methanomicrobiales. They appear to be absent in Methanosarcinales and Methanocellales [5].

EhaA-T and EhbA-Q.
Both membrane-associated [NiFe]-hydrogenases belong to the group of energyconverting hydrogenases that catalyze the reduction of ferredoxin with H 2 driven by the proton-motive force or the sodium-ion-motive force [5,72]. ehaO/ehbN encodes the large subunit harboring a [NiFe] center, and ehaN/ehbM encodes the small subunit characteristic of all [NiFe] hydrogenases. Most of the other eha and ehb genes encode membrane proteins. In M. marburgensis, both enzyme complexes are considered to be sodium-ion dependent and to have a function in providing the cells with reduced ferredoxin mainly for anabolic reactions, such as the reduction of CO 2 to CO (E 0 = −520 mV) and of acetyl-CoA plus CO 2 to pyruvate (E 0 = −500 mV) [58]. But the two enzymes are probably also required in CO 2 reduction with H 2 to methane if or when in the MvhADG/HdrABCcatalyzed reaction less ferredoxin is reduced than required for CO 2 reduction to formylmethanofuran (E 0 = −500 mV) (see above). Accordingly, deletion of the ehb genes in M. maripaludis reveals a function of Ehb in autotrophic CO 2 fixation: the mutant is an acetate auxotroph. Deletion of the eha genes was not possible [73,74]. In the genomes of M. marburgensis, the genes encoding EhaA-T and EhbA-Q are organized in the transcription units ehaA-T and ehbA-Q, and their transcription is differentially regulated [75].
Four subunits encoded by EhaA-T and EhbA-Q show sequence similarities to the core subunits of the NADH:ubiquinone oxidoreductase (NuoA-N) and the formate hydrogen lyase from E. coli. This is why two of these subunits in M. thermautotrophicus were annotated in 1997 as NADH dehydrogenase (EhaH/EhbO and EhaJ/EhbF) and two as formate hydrogen lyase (EhaN/EhbM and EhaO/EhbN) [76]. At that time, the subunits EhaS and EhaT were annotated as formylmethanofuran:H 4 MPT formyltransferase and ribokinase, respectively. However, these CDS have been shown to be cotranscribed with the ehaA-R genes, making a function in CO 2 reduction to methane (formyltransferase) or sugar activation (ribokinase) very unlikely [75].
Another CDS within the ehaA-T operon of M. thermautotrophicus, ehaP2, encodes a 6[4Fe4S] polyferredoxin that is not found in M. marburgensis. In both species, the ehaA-T transcription unit is followed by a gene for another 6[4Fe4S] polyferredoxin that could be the electron acceptor used by the hydrogenase.
Many members of the Methanomicrobiales lack the genes for EhaA-T and EhbA-Q. Instead these hydrogenotrophic methanogens contain genes for energy-converting hydrogenases different from those found in the other three orders of hydrogenotrophic methanogens [5,67].

Genes for Enzymes Catalyzing CO 2
Reduction to Methane  Table 2). The CDS for a second Ftr in M. thermautotrophicus [7] turned out to be the subunit EhaT of the energy-converting EhaA-T complex (see above).
10.1. FwdA-DFGH and FwdA/FmdBCE. These two cytoplasmic enzymes catalyze the reduction of CO 2 to formylmethanofuran with reduced ferredoxin. FwdA-DFGH is a tungsten enzyme. FmdBCE is a molybdenum enzyme. In both enzymes, the transition metal is coordinated by two molybdopterin molecules [77]. Interestingly, the tungsten and the molybdenum enzymes share the subunit FwdA, which is synthesized constitutively. In contrast, the molybdenum-dependent enzyme is only synthesized when molybdate is present in the growth medium [78,79].  [80]. Crystal structures of all four enzymes are available [81]. A CDS (MTBMA c06530; MTH204) in the genomes of the two Methanothermobacter species encodes a putative 5formyltetrahydrofolate cycloligase present also in the genome of Methanopyrus kandleri. The function of the enzyme is unclear, since tetrahydrofolate, a structural and functional analog of H 4 MPT, has not been found in these methanogens [82]. Therefore, the CDS might encode a 5-formyl-H 4 MPT cyclohydrolase with a yet unknown function in reduction of CO 2 to methane.

MtrA-H.
Of the enzymes involved in CO 2 reduction to methane, only the MtrA-H complex is a membrane enzyme. It is a cobalamin-dependent enzyme with the corrinoid bound to MtrA. The membrane complex couples the exergonic methyl transfer from methyl-H 4 MPT to coenzyme M (ΔG • = −30 kJ/mol −1 ) with the endergonic translocation Archaea of sodium ions [83]. The sodium-ion-motive force thus generated is used to drive the phosphorylation of ADP via the A 1 A 0 ATP synthase present in all methanogens (see below).
A CDS in the genomes of M. marburgensis and M. thermautotrophicus encodes the methyltransferase MtxX [84] (MTBMA c06800 and MTH231), which is also present in the genomes of all other methanogens (Supplementary Table  1 Table 1).
Methyl-coenzyme M reductases are only active when their prosthetic group F 430 is in the Ni(I) oxidation state. To render the enzyme from the inactive Ni(II) state to the active Ni(I) state by reduction, several activating enzymes, reduced ferredoxin, and ATP are required. One of the enzymes, component A2 (AtwA), which has an ATP-binding cassette, has been identified [86]. In the genomes of M. marburgensis and M. thermautotrophicus, two and three CDS, respectively, for AtwA are found (Supplementary Table 1).

Genes Involved in Coupling of Methanogenesis with ADP Phosphorylation via the Sodium-Ion-Motive Force
Methanogenesis from CO 2 and H 2 is dependent on sodium ions, which are required for coupling methanogenesis with ADP phosphorylation (Figure 2,  [83,88]. The ATP synthase shows a conserved Na + -binding motif [89], and it is generally assumed that four sodium ions are required for the phosphorylation of one ADP [90]. Figure 2 shows the proposed reduction of ferredoxin with H 2 via Eha or Ehb, driven by the sodium-ion-motive force with a Na + to e − stoichiometry of 1; however, this has not yet been established [72]. The sodium/proton antiporter is most likely there for pH homeostasis [91].

Genes for the Synthesis of Prosthetic Groups of Methanogenic Enzymes
Many of the enzymes catalyzing the reactions involved in CO 2 reduction to methane with H 2 contain prosthetic groups that have to be synthesized (Table 2, Figure 2 Both M. marburgensis and M. thermautotrophicus contain a carAB transcription unit. The encoded proteins are probably involved in the synthesis of carbamoyl phosphate from glutamine, bicarbonate, and 2 ATP. A second carB gene is probably for the synthesis of carbamoyl phosphate from ammonium, bicarbonate, and 2 ATP. Notably, carbamoylphosphate is not only required in methanogens for the synthesis of the active site of [NiFe] hydrogenases but also for the first committed step in pyrimidine and arginine biosynthesis.

FeGP Cofactor. The biosynthesis of the FeGP cofactor (prosthetic group of the [Fe] hydrogenase Hmd)
has not yet been elucidated. In silico analysis indicates that seven genes co-occurring with the hmd gene are involved. In M. marburgensis and M. thermautotrophicus, six of the hmd cooccurring genes (hcgA-F) form a transcription unit directly upstream of the hmd gene. The gene hcgG (MTBMA c15200; MTH1137) is located five CDS downstream of the hmd gene in M. marburgensis and four CDS downstream in M. thermautotrophicus [5].
The gene hcgA (MTBMA c15270; MTH1143) is predicted to encode a protein with a sequence similar to that of the radical-SAM protein BioB, which is involved in sulfur insertion in biotin biosynthesis [92]. However, HcgA lacks the N-terminal signature motif CX 3 CX 2 C or CX 4 CX 2 C, which is characteristic for the radical-SAM protein superfamily that coordinates a [4Fe4S]-cluster essential for radical formation. Instead, HcgA universally harbors a unique CX 5 CX 2 C motif [92]. The functions of the hcgB-G genes remain to be established [5]. The genes encoding Hmd and HcgA-G are also found in Methanobrevibacter smithii, Methanobrevibacter ruminantium, all members of the Methanococcales, Methanopyrus kandleri, and one member of the Methanomicrobiales (Methanocorpusculum labreanum) [5].
In the genomes of the two Methanothermobacter species, two genes homologous to hmd are found. The two encoded proteins, HmdII and HmdIII, show only low sequence identity (<20%) to [Fe]

Iron-Sulfur
Clusters. M. marburgensis and M. thermautotrophicus contain many iron-sulfur proteins. Amongst these are the hydrogenases, formylmethanofuran dehydrogenases, heterodisulfide reductase, and the ferredoxins involved in CO 2 reduction with H 2 to methane ( Figure 2). Accordingly, the iron requirement for growth of the two methanogens is very high [94]. How iron-sulfur clusters are assembled in methanogens is still a mystery. In bacteria, two independent systems, Suf and Isc, have this function [95]. In the two Methanothermobacter species, only CDS for a cysteine desulfurase homolog (IscS/SufS), a SufB/SufD homolog (persulfide acceptor), and a SufC homolog (ABC-type ATPase) were found. Additionally, in the genome of M. thermautotrophicus, there is a homolog of the bacterial apbC/eukaryotic NBP35 gene that encodes an iron-sulfur cluster transfer protein [96,97]. However, this ApbC homolog is not found in M. marburgensis (Table 2). In Methanococcus maripaludis, which lacks genes for cysteine desulfurase, cysteine has been shown not to be the sulfur source for the biosynthesis of iron-sulfur clusters and methionine [98].

Molybdopterin.
The biosynthesis of molybdopterin appears to proceed as in bacteria, starting from GTP [99]. CDS for MoaABCE and MoeAB (molybdopterin cofactor biosynthesis proteins) and for MobAB (molybdopterin-guanine dinucleotide biosynthesis proteins) are found ( Table 2).
13.6. Cofactor F 430 . The synthesis of the nickel tetrapyrrole is predicted to branch off the cobalamin pathway at the intermediate precorrin-2 (dihydrosirohydrochlorin), where also the biosynthesis of siroheme (prosthetic group of assimilatory sulfite reductase) branches off (CysG1 = MTBMA c06180; MTH167) [107]. Only one of probably six intermediates has been identified [108], and the enzymes involved are not yet known. Two chelatases structurally related to the cobalt chelatase CobNS [109] could be nickel chelatases that catalyze the incorporation of Ni 2+ into precorrin-2 or a precorrin-2 product (MTBMA c10550-10570, 09440; MTH673, 556). It has been proposed that proteins structurally related to the nitrogenase [Fe]-protein NifH and the [MoFe]-protein NifDK could have a function in pyrrole ring reduction involved in the synthesis of F 430 from precorrin-2 [110]. The proposal is based on the finding that homologs of NifD and NifH are involved in protochlorophyllide reduction in phototrophs [111] and that nifDand nifH-like genes (nflD and nflH) (MTBMA c01050 and 10230; MTH1522 and 643) are present in all methanogens, also in those that lack nif genes.
A small number of CDS for conserved hypothetical proteins are found in every genome of methanogenic archaea (Supplementary Table 1) and in the meta-genome of methanotrophic archaea [112] but are not found in any other organism. Among these methanogen-specific CDS, to which also NflD belongs, could be some that function in F 430 biosynthesis because F 430 has not been found outside methanogenic archaea and the phylogenetically closely related methanotrophic archaea [113,114].
The two Methanothermobacter species have CDS for formate dehydrogenase even though they cannot grow on formate. They require the enzyme for CO 2 reduction to formate, which in turn is required for the synthesis of purines and as an electron donor for anaerobic ribonucleotide reductase (class III) [82]. Methanogens without cytochromes incorporate formate into C 2 of purines in an ATPdependent reaction with formyl phosphate as intermediate, as catalyzed by 5-formaminoimidazole-4-carboxamide-1 β-d-ribofuranosyl 5 -monophosphate synthetase (PurP) (MTBMA c15790; MTH1201) [128]. CO 2 reduction to formate is their only means of generating formate. Accordingly, formate-dehydrogenase-negative mutants of M. marburgensis require formate for growth on H 2 and CO 2 [28]. M. marburgensis and M. thermautotrophicus contain CDS for a second formate dehydrogenase subunit FdhA and a formate dehydrogenase accessory protein FdhD, but they lack a CDS for a formate carrier (FdhC) [129,130], which is present in methanogens that can grow on formate.
14.2. Methanofuran. The pathway for the biosynthesis of methanofuran and the responsible genes have yet to be identified. A clear structural element in all known methanofurans is tyramine, likely produced by the decarboxylation of ltyrosine [131]. In M. marburgensis and M. thermoautotrophicus, the decarboxylation is catalyzed by MfnA.  [132]. The biosynthesis of the nonpterin portion involves at least nine steps, the first being catalyzed by ribofuranosylaminobenzene 5 -phosphate synthase [133].
14.4. Coenzyme M. All but one (ComF) of the CDS required for coenzyme M synthesis (ComA-F) have been identified in the genomes of M. marburgensis and M. thermautotrophicus. Biosynthesis starts from phosphoenol pyruvate with sulfolactic acid, sulfopyruvic acid, and sulfacetaldehyde as intermediates. ComA catalyzes the Michael addition of sulfite to phosphoenolpyruvate. ComB is a Mg 2+dependent acid phosphatase specific for 2-hydroxycarboxylic acid monophosphate esters. ComC catalyzes the oxidation of the (R)-sulfolactate intermediate to form sulfopyruvate, which is decarboxylated to produce sulfoacetaldehyde via ComDE. The CDS for ComE is one of the methanogenspecific genes (Supplementary Table 1). The final postulated enzyme in CoM biosynthesis, ComF, which has not yet been identified in any organism, catalyzes the reductive thiolation of sulfoacetaldehyde to coenzyme M, a reaction which most likely does not proceed spontaneously. The absence of comA, comB, and comC in the genomes of Methanosarcina spp. and members of the Methanomicrobiales implies that these methanogens synthesize sulfopyruvate by a different route [116].
14.5. Coenzyme B. The biosynthesis of coenzyme B starts from acetyl-CoA and 2-oxoglutarate and proceeds via 2-oxoadipate, 2-oxopimelate, 2-oxosuberate, and suberate semialdehyde as intermediates. CDS for homologs of (R)citrate synthase [134], aconitase, and isocitrate dehydrogenase have been found. There is only one synthase for the three synthase reactions, one isomerase for the three isomerization reactions, and one dehydrogenase for the three dehydrogenation reactions. The three enzymes are homologs of isopropylmalate synthase (LeuA), isopropylmalate isomerase (LeuC/D), and isopropylmalate dehydrogenase (LeuB), respectively, [116,135] for which there are two annotated gene copies in the genomes of M. marburgensis and of M. thermautotrophicus.

Genes for Transport of Ions Required for Growth
The growth of methanogens is dependent not only on sodium ions (see above), but also on nickel, cobalt, iron, magnesium, and potassium cations and on molybdate or tungstate and phosphate anions [26,27]. Growth is probably also dependent on zinc and calcium cations present as trace contaminations in the growth media. All these ions, all of which are required for the synthesis of enzymes, prosthetic groups, and coenzymes, must be taken up from the growth medium (Table 2, Figure 2). With respect to Fe 2+ , Co 2+ , Ni 2+ , and Zn 2+ uptake, it has to be considered that M. marburgensis and M. thermautotrophicus thrive in habitats, where the H 2 S/HS − concentrations are generally high and the pH is near 7. The transition metal ions in such habitats are mostly present as sulfides, and therefore, the concentrations of the free ions are very low (<10 −8 M), with the lowest being that of free zinc ions. The solubility product constants are 4.5 × 10 −24 for ZnS, 2 × 10 −21 for NiS, 4 × 10 −21 for CoS, and 6 × 10 −16 for FeS [136]. MTH1708-1714) involved in autotrophic CO 2 fixation. The ABC transporter involved has not yet been identified. There appears to be no close homolog to the NikA-E nickel transport system in E. coli. In the genome of the two Methanothermobacter species, there are two sets of CDS (Cbi1 and Cbi2) predicted to encode a Co 2+ ABC transporter (see below), one of which (CbiM1N1O1Q1) has been proposed to be a Ni 2+ ABC transporter [102,137]. But the Ni 2+ transporter could also be encoded by one of the five sets of CDS for ABC transport systems without an annotated function present in the genomes of M. marburgensis ( hydrogenase are thought to be taken up by the ATP-driven FeoAB transport system encoded by feoAB [138].

Zinc.
Of the proteins involved in CO 2 reduction with H 2 to methane only the subunit B of heterodisulfide reductase contains zinc [65]. But zinc ions are also required for RNA polymerase and other biosynthetic enzymes. The gene cluster for the putative high-affinity Zn 2+ ABC transporter ZnuABC/ZupT in M. marburgensis and M. thermautotrophicus lies next to an open reading frame for the nickel-responsive transcriptional regulator NikR homolog (MTBMA c09830; MTH603). Therefore, the NikR homolog might in reality be a zinc-responsive regulator [139]. NikR Archaea from E. coli also binds zinc ions, but without a conformational change response [140].
15.5. Magnesium. Magnesium ions are required in ATP-and ADP-dependent reactions, because synthetases and kinases generally use complexes of ATP and ADP with Mg 2+ as substrates and products. Mg 2+ is predicted to be taken up by the MgtE system [141].
15.6. Calcium. The crystal structure of Mch from Methanopyrus kandleri revealed the presence of a structural calcium ion [80]. Methane formation in cell suspensions of M. thermautotrophicus is stimulated by Ca 2+ [142]. These findings indicate a function of Ca 2+ in methanogenesis. A membrane-associated Ca 2+ ATPase has been identified via bioinformatic methods [143]. Available evidence indicates that Ca 2+ uptake is inhibited by Ni 2+ and Co 2+ [142]. If a Ca 2+ uptake system is present, it must be a high-affinity uptake system, since media for the growth of M. marburgensis do not have to be supplemented with calcium salts for the methanogen to grow optimally [27]. The contaminating calcium ion concentration in the media has been determined to be 0.5 μM [142]. 15.7. Potassium. Potassium ions are not directly involved in methanogenesis from CO 2 and H 2 O, but most of the methanogenic enzymes function optimally only at high K + concentrations. In growing M. marburgensis cells, the intracellular K + concentrations have been determined to be above 0.5 M [144]. The potassium ions are most probably taken up by the low-affinity TrkAH system [145], for which CDS in the genomes of the two Methanothermobacter species have been found.

Transcriptional Regulation and Posttranslational Modifications
Up to here, the regulation and posttranslational modifications of enzymes involved in CO 2 reduction with H 2 to methane have only been mentioned. They are, in the following, dealt with in more detail.  [151]. When M. marburgensis grows under nickellimiting conditions, transcription of the mtd and hmd genes is upregulated and that of the frhADBG genes is downregulated [151]. The genome of M. marburgensis has three CDS and the genome of M. thermautotrophicus has two CDS predicted to encode a nickel-responsive transcriptional regulator (NikR) [139]. In bacteria, NikR regulates transcription of genes involved in the synthesis of nickel enzymes and nickel transport [152]. The putative presence of several nickel responsive regulators might reflect the methanogens' use of an unusually high number of different nickel proteins and their growth in habitats where the nickel concentration is sometimes growth limiting [5].  Table 2) is synthesized as a preprotein, from which a C-terminal sequence has to be clipped off after the DPCxxCxxH/R motif involved in [NiFe] center coordination. This is the last step in [NiFe] center synthesis [68]. Therefore, genes for four hydrogenase-specific endopeptidases should be present. In M. marburgensis and M. thermautotrophicus, only the endopeptidase encoded by frhD (MTBMA c16850; MTH1299) in the frhADGB transcription unit could be unambiguously identified. In addition to frhD in the frhADGB operon, a second frhD gene (MTBMA c11320; MTH737) (homologous to hycI in E. coli) and several other genes for metalloproteases outside the transcription units for the four [NiFe] hydrogenases are found.

Methyl-Coenzyme M Reductases.
In the structure of the two methyl-coenzyme M reductase isoenzymes (Mcr and Mrt), a thioglycine, a C 2 -methyl alanine, a C 5 -methyl arginine, an N-methyl histidine, and an S-methyl cysteine are found in the α-chain [160]. The methyl groups are posttranslationally introduced from S-adenosylmethionine (SAM) [161,162]. The formation of C 2 -methyl alanine and C 5 -methyl arginine involves a C-methylation and is, therefore, predicted to involve radical SAM enzymes [163]. The formation of N-methyl histidine and S-methyl cysteine is predicted to involve SAM dependent methyltransferases. The genomes of M. marburgensis and M. thermautotrophicus encode at least 14 radical-SAM enzymes and more than 15 SAM-dependent methyltransferases; a function for most of these has not yet been assigned. SAM is synthesized by an archaeal-type SAM synthetase [164].

Conclusions
We identified approximately 200 CDS in M. marburgensis and M. thermautotrophicus that encode proteins directly or indirectly involved in CO 2 reduction to methane with H 2 and in coupling this process with energy conservation (Figure 2). More than 50 of these CDS are concentrated in the genome region between MTBMA c16880 and MTBMA c15110 and between MTH1302 and MTH1128, whereas the others are scattered all over the genome. Approximately 90 CDS are for membrane-associated protein complexes, of which only the MtrA-H complex has been purified. Crystal structures of many of the cytoplasmic enzymes catalyzing CO 2 reduction to methane have been determined [81,93,123]. However, of the biosynthetic proteins involved in coenzyme and prosthetic group synthesis, only a few have been characterized, and approximately 20 have not yet even been identified. The lack of a genetic system for the two Methanothermobacter species presently allows their identification only by reverse genetics.
The comparison of the genomes of two Methanothermobacter species has revealed that they are pretty much the same in all core catabolic and anabolic reactions. Some of the 1,607 CDS in common encode proteins with identical or almost identical sequence; others, however, encode proteins with only low sequence identity. The differences in phenotype observed, such as differences in growth rate and in ATPase activity, can, therefore, easily be the result of these sequence differences, since even point (single nucleotide) mutations can result in a change in the phenotype. But it is also likely that some of the differences, such as in cell wall sugar composition and susceptibility to phage infection, lie hidden within sequences that the two organisms do not have in common. Of these, those for the synthesis of cell surface polysaccharides and for IS-like elements and CRISPR are most apparent.
Our comparison of the genomes provides a roadmap for defining the majority of functional components responsible for the methanogenic phenotype in M. marburgensis and M. thermautotrophicus and a template for metabolic pathway reconstruction and gene discovery in comparisons of clonal populations or meta-genomes. Most of the 200 genes that have a direct or indirect function in CO 2 reduction with H 2 to methane are also found in the Methanococcales and Methanopyrales, whereas some are lacking in the Methanomicrobiales. Thus, many members of the Methanomicrobiales lack the genes for the heterodisulfide-reductase (HdrABC-) associated hydrogenase subunits MvhAG and many contain genes for energy-converting hydrogenases different from those found in the other three orders of hydrogenotrophic methanogens [5,67]. Interestingly, the cytochrome-containing Methanocellales appear to be more similar with respect to their core catabolic genes to Archaea the Methanobacteriales than to the cytochrome-containing Methanosarcinales.
It was first thought that most of the reactions, coenzymes, and prosthetic groups involved in CO 2 reduction with H 2 to CH 4 in M. thermautotrophicus and M. marburgensis would be unique to methanogens. Only much later was it discovered that "methanogenic" genes are also present in other archaea and in bacteria. For example, the sulfate-reducing Archaeoglobus fulgidus uses many enzymes and coenzymes in anaerobic lactic acid oxidation to 3 CO 2 that are also used by methanogenic archaea in CO 2 reduction to methane [165], and Desulfobacterium autotrophicum contains gene clusters for the heterodisulfide reductase HdrABC [166]. Another example is the methylotrophic bacteria, which use methanogenic enzymes and coenzymes in their energy metabolism [167]. Finally, [NiFe] hydrogenases, which were discovered first in M. marburgensis [168], are also found, for example, in E. coli [68,169]. M. marburgensis and M. thermautotrophicus are therefore not only model organisms for the study of methanogenesis from H 2 and CO 2 but also for the study of H 2 and C 1 metabolism in general.