Highlighting the Unique Roles of Radical S-Adenosylmethionine Enzymes in Methanogenic Archaea

Radical S-adenosylmethionine (SAM) enzymes comprise one of the largest known enzyme superfamilies, with over 700,000 estimated members (1). These enzymes use SAM and a [4Fe-4S] cluster to catalyze complex reactions on inert substrates in various biochemical pathways, including cofactor biosynthesis, secondary metabolite biosynthesis, and nucleic acid modification (2, 3). Radical SAM enzymes are generally defined by the presence of a CX₃CX₂C motif that coordinates a [4Fe-4S] cluster, with the fourth iron in the cluster being ligated to the amino and carboxylate groups of SAM (Fig. 1) (4, 5). Although the majority of radical SAM enzymes (>90%) contain the canonical CX₃CX₂C motif, several noncanonical cysteine motifs have also been described in enzymes carrying out radical SAM chemistry (6). In a typical radical SAM reaction, the reduced [4Fe-4S]¹⁺ cluster donates an electron to the sulfonium group of SAM, resulting in homolytic bond cleavage to generate methionine and a highly reactive 5′-deoxyadenosyl radical (5′-dAdo·) (3, 7, 8). The 5′-dAdo· abstracts a hydrogen atom from an otherwise unreactive site to generate 5′-deoxyadenosine (5′-dAdoH) and a substrate radical that can undergo further chemistry (Fig. 1). Established reactions catalyzed by radical SAM enzymes include radical generation on a protein substrate, isomerization, cyclization, carbon backbone rearrangement, sulfur insertion, and methylation (3).

Most radical SAM enzymes are found encoded in bacterial genomes, with the next largest proportion coming from archaea (6). Eukaryotes generally have the fewest radical SAM superfamily members. For example, the human genome encodes eight radical SAM enzymes, all of which have established functions (9–11). Given the extensive suite of natural products that are produced by different bacterial organisms, many radical SAM enzymes in various bacteria are involved in secondary metabolite biosynthesis. Conversely, although secondary metabolites and their biosynthetic pathways have been identified in some archaea (12, 13), most archaeal organisms do not exhibit the potential for robust secondary metabolism, and thus, their radical SAM enzymes are more likely to be involved in essential primary metabolism.

Methanogenic archaea (“methanogens”) are diverse and widespread microbes that are strictly dependent upon methanogenesis, a form of anaerobic respiration that reduces simple carbon compounds to generate methane as an end product. These organisms produce nearly 1 billion tons of methane each year, accounting for about 70% of global methane emissions (14, 15). Given the role of methane as a potent greenhouse gas and as a potential renewable energy source (16, 17), investigating essential enzymes in methanogens could lead to the development of methane mitigation strategies and could have implications for bioenergy applications. Interestingly, methanogens contain an abundance of putative radical SAM enzymes compared to other archaea, eukaryotes, and many bacteria. However, currently, only about half of the putative radical SAM enzymes in methanogens have established or likely functions based on experimental evidence or high sequence similarity to well-characterized radical SAM enzymes (see Tables S1 and S2 in the supplemental material for lists of putative radical SAM enzymes in two model methanogens, Methanococcus maripaludis S2 and Methanosarcina acetivorans C2A). Thus, these organisms represent an underexplored domain for uncovering new functions and mechanisms of radical SAM enzymes.

Some of the radical SAM enzymes with known functions in methanogens are well-characterized and widespread enzymes that are additionally found in bacteria and/or eukaryotes, while others catalyze less common reactions that are unique to only methanogens or a smaller subset of organisms including methanogens (Table 1). This minireview focuses on the latter group of methanogenic radical SAM enzymes to summarize the current state of the field regarding the diverse functions of radical SAM enzymes in methanogens, especially highlighting recent exciting work in this area as well as discussing remaining questions that require further research.

One of the defining features of archaea that distinguishes them from bacteria and eukaryotes is the unusual chemistry of their membrane lipids. As opposed to membranes composed of fatty acids with ester linkages to glycerol backbones, archaea contain isoprenoidal membranes with ether linkages to glycerol backbones. There are two major types of isoprenoid ether lipids: glycerol diphytanyl diethers (archaeol) and glycerol dibiphytanyl glycerol tetraethers (GDGTs) (Fig. 2A) (18). Similar to traditional phospholipid-based membranes, archaeol lipids form a bilayer membrane structure, whereas GDGTs, which consist of two diether lipids fused at their hydrophobic ends (Fig. 2A), result in a membrane-spanning structure to generate monolayer membranes. Different archaea have various proportions of archaeol and GDGTs in their membranes, where the latter are especially prevalent in halophilic and thermophilic organisms (18, 19). The higher rigidity and lower permeability of GDGT lipids likely provide an advantage in extreme environments. Supporting this idea, growth experiments have shown that archaea increase their GDGT/archaeol ratio when cultured at elevated temperatures (20, 21). In some nonmethanogenic archaea, GDGTs are further modified through the introduction of cyclopentane or cyclohexane rings at the C-7 and C-3 positions of each chain (18). Additional GDGT modifications include the addition of hydroxyl or methyl moieties, alteration of the polar head groups, and cross-linking of the two biphytanyl chains. These modifications influence the physicochemical properties of the lipids and thus serve as a mechanism for adaptation to various environmental conditions (18).

The enzyme(s) and mechanism involved in the key step of GDGT biosynthesis, the diether-to-tetraether conversion (Fig. 2A), remained a mystery until just this year (22). Through bioinformatic analyses, a putative radical SAM enzyme was identified in the genomes of tetraether lipid-containing archaea, including the model methanogen Methanosarcina acetivorans, which was a likely candidate to catalyze the required C-C bond formation. The heterologous expression of the proposed gene MA_1486 in Methanococcus maripaludis, a non-GDGT-containing methanogen, resulted in the production of GDGT, thus indicating that the MA_1486 gene product is sufficient for GDGT formation, likely using archaeol precursors (22). Thus, MA_1486 was defined as a tetraether synthase (Tes) (Fig. 2A). The M. acetivorans genome encodes another Tes homolog (MA_1114) that, when heterologously expressed in M. maripaludis, does not result in GDGT production. However, the authors noted the production of trace amounts of macrocyclic archaeol (Fig. 2A) in the strain expressing MA_1114. Macrocyclic archaeol was also observed in cells expressing the confirmed Tes (MA_1486 or Maeo_0574) and was proposed to be a side product that occurs during heterologous expression (22). Interestingly, macrocyclic archaeol has been identified in only a select few archaea, all of which are thermophilic methanogens from hydrothermal vents (23–25).

Although not present in methanogenic lipids, the cyclopentane rings in GDGTs found primarily in the phyla Crenarchaeota and Thaumarchaeota are also installed by radical SAM enzymes (26). None of these radical SAM enzymes involved in archaeal lipid biosynthesis have been biochemically characterized, so the mechanism and direct substrates utilized for C-C bond formation in tetraether lipid synthesis and the formation of cyclopentane rings in archaeal lipids remain unclear. Now that the identities of these enzymes have been revealed, in vitro experiments can be pursued to reveal the mechanistic details of these intriguing reactions.

Pyrrolysine is the 22nd genetically encoded amino acid, which was initially described in the active site of the monomethylamine methyltransferase MtmB from the methanogen Methanosarcina barkeri (27). Pyrrolysine is encoded by amber (UAG) codons, which were first identified in the genes encoding the three methyltransferases required for methanogenesis from methylamines (28). Since then, pyrrolysine has been found in other enzymes, including the tRNA^His guanylyltransferases (Thg1) (29) and certain transposases (30) in Methanosarcina, and has also been identified in some bacteria (31). Five pyl genes are necessary for pyrrolysine biosynthesis and its incorporation into proteins (32). The pylT gene encodes tRNA^Pyl, and pylS encodes pyrrolysyl-tRNA synthetase. The other three genes are required for the biosynthesis of pyrrolysine from 2 molecules of lysine. The first step is carried out by PylB, a radical SAM enzyme that catalyzes a carbon skeleton rearrangement of l-lysine to produce (3R)-3-methyl-d-ornithine (Fig. 2B) (33). The crystal structure of PylB expressed and purified from Escherichia coli revealed both SAM and the product bound to the active site, thus providing strong support for the function of PylB in (3R)-3-methyl-d-ornithine production (34). Despite structural information, an in vitro demonstration of PylB catalytic activity has not yet been reported. The reaction is proposed to occur analogously to that for B₁₂-dependent glutamate mutase (32, 34, 35). Thus, the radical SAM-generated 5′-dAdo· would abstract a hydrogen atom from C-4 of lysine, leading to C-C bond cleavage that results in a glycyl radical intermediate and 4-aminobutene. The glycyl radical can then recombine with the former C-4 position of the substrate to form a new C-C bond followed by reduction and protonation to yield the final (3R)-3-methyl-d-ornithine product (32). Future in vitro studies are necessary to confirm PylB activity and provide experimental evidence for the proposed mechanism.

Tetrahydromethanopterin (H₄MPT) (Fig. 2C) is a modified folate cofactor found in methanogens that, like tetrahydrofolate (H₄F), functions to carry and transfer C₁ groups of various oxidation states (36). H₄MPT is used for several steps in the methanogenesis pathway and, in many methanogens, is also used instead of H₄F for canonical folate-dependent reactions in amino acid and nucleic acid biosynthesis. One interesting structural feature that distinguishes H₄MPT from H₄F is the presence of methyl groups at the C-7 and C-9 positions (Fig. 2C). The unreactive nature of the sites of methylation led to the proposal that radical SAM chemistry may be involved in their installation (37). Genomic analyses revealed that in some methanogens, a radical SAM-encoding gene is found near the gene encoding β-ribofuranosylaminobenzene 5′-phosphate synthase, a key enzyme required for H₄MPT biosynthesis (38, 39). Thus, the corresponding radical SAM-encoding gene from Methanocaldococcus jannaschii, MJ0619, was used for heterologous-expression experiments in E. coli to gain insight into the potential role of this gene product in H₄MPT biosynthesis (37). When MJ0619 was expressed in E. coli, various methylated pterins were detected in the cell extracts by liquid chromatography-mass spectrometry (LC-MS), consistent with this enzyme catalyzing methylation at both the C-7 and C-9 positions of a pterin substrate (Fig. 2C). Since E. coli does not normally produce methylated pterins, this suggested that MJ0619 catalyzes the methylation reactions in H₄MPT biosynthesis. Additionally, because 6-hydroxyethyl-7-methylpterin was identified in cell extracts containing MJ0619, but the corresponding dimethyl-folate was not identified, it was concluded that the likely substrate for MJ0619 is 6-hydroxmethyl(dihydro)pterin, a common early intermediate in both H₄F and H₄MPT biosynthesis (37).

Surprisingly, isotope feeding studies with E. coli expressing MJ0619 revealed that the enzyme likely does not use SAM as a methyl group donor since none of the deuterium atoms from CD₃-Met were incorporated into the C-7 pterin methyl group added by MJ0619 (37). This is different from all other characterized radical SAM methylases, which use 2 molecules of SAM, one as the C₁ donor for the methylation reaction and one as the source of 5′-dAdo· for radical SAM chemistry (40). Interestingly, further feeding studies with CD₃-acetate led to the proposal that MJ0619 uses N⁵-N¹⁰-methylene-tetrahydrofolate (CH₂H₄F) as a C₁ source for the methylation of pterins when it is expressed in E. coli (37). This is analogous to thymidylate synthase, where CH₂H₄F serves first as a methylene donor and then as a hydride donor to generate the final methyl group for the biosynthesis of dTMP from dUMP (41, 42). Confirmation of the function of MJ0619 and its homologs as well as details of the unique mechanism used for methylation during H₄MPT biosynthesis will require in vitro enzymatic studies, which have yet to be reported.

Methyl-coenzyme M reductase (MCR) catalyzes the final rate-determining step of methanogenesis. MCR is a dimer of heterotrimers with an α₂β₂γ₂ configuration, harboring two active sites containing the F₄₃₀ nickel tetrapyrrole prosthetic group. When the first crystal structure of MCR from Methanothermobacter marburgensis was determined, five unusual posttranslational modifications (PTMs) were revealed in the α subunit near the active site: thioglycine, N¹-methylhistidine, S-methylcysteine, 5-(S)-methylarginine, and 2-(S)-methylglutamine (43). Except for N¹-methylhistidine, the genes/enzymes required to install each of these PTMs have now been identified (44–47), including the identification and biochemical characterization of the radical SAM methylases responsible for the synthesis of 5-(S)-methylarginine (Fig. 2D) (44, 48–50) and 2-(S)-methylglutamine (Fig. 2E) (47). Upon searching the genome neighborhood of MCR operons for potential genes encoding enzymes involved in installing PTMs, a radical SAM-encoding gene annotated as “methanogenesis marker 10” (mmp10) was identified. To elucidate mmp10’s function, it was deleted in M. acetivorans, and the resulting MCR PTMs were analyzed by mass spectrometry (44). The authors found that the α subunit from the mmp10 deletion strain contained an unmodified arginine residue at position 285, thus demonstrating that Mmp10 is required for generating the 5-(S)-methylarginine PTM found in methanogenic MCRs. In terms of the physiological impacts of the unique PTM, it was observed that the lack of methylated arginine impaired the ability of M. acetivorans to grow under oxidative or thermal stress conditions. Furthermore, the purified variant MCR was shown to have a decreased melting temperature (T_m) (74.6°C versus 82.6°C for the wild type), further indicating that the 5-(S)-methylarginine impacts the thermal stability of MCR from M. acetivorans (44). Interestingly, the methylated arginine seems to have a more substantial impact on M. maripaludis MCR, where a deletion strain lacking this modification showed a highly impaired growth rate and the rate of methanogenesis was only about half the rate of the wild type (49). The in vitro enzymatic activities of MCR variants lacking specific PTMs have not yet been reported; thus, any specific role that 5-(S)-methylarginine plays in catalysis remains unclear. Interestingly, the crystal structure of MCR from M. acetivorans lacking the 5-(S)-methylarginine and other unique PTMs was virtually identical to the wild-type MCR structure (45).

Biochemical characterization of Mmp10 revealed that the enzyme is a cobalamin-dependent radical SAM methylase (48) despite the lack of an N-terminal cobalamin binding domain found in other cobalamin-dependent radical SAM methylases (40, 51). Further experiments demonstrated that cobalamin functions as an intermediate methyl group carrier, where SAM is used as the initial methyl group donor to generate methylcobalamin and S-adenosylhomocysteine (SAH) in a nucleophilic substitution reaction with cob(I)alamin (48). Thus, Mmp10 is proposed to employ a mechanism similar to those of other cobalamin-dependent methylases (51), where one SAM is used in nucleophilic chemistry to generate methylcobalamin and the other SAM is used in radical SAM chemistry to generate a substrate radical at the site of methylation that then reacts with methylcobalamin to generate the methylated product and cob(II)alamin. Cob(II)alamin can then be reduced to cob(I)alamin with titanium(III) citrate in vitro to allow remethylation with SAM, followed by subsequent turnovers (48). Recently, the crystal structure of Mmp10 was determined, which supports the proposed mechanism and provides further details of how the enzyme controls the two different reactivities of SAM (50). Interestingly, the radical SAM [4Fe-4S] cluster is coordinated by 3 cysteine residues as well as a strictly conserved tyrosine residue, which excludes SAM binding to the “unique” iron that is normally a defining feature of radical SAM enzymes. It is proposed that this places SAM in a position to react with cob(I)alamin to generate the methylcobalamin intermediate (50). Upon binding another SAM as well as the protein substrate, the active site then appears to reorganize, displacing tyrosine from the cluster, thus allowing coordination with SAM and subsequent radical SAM cleavage to produce 5′-dAdo· that initiates the radical methylation reaction. The structure of Mmp10 also revealed a mononuclear iron site coordinated by 4 cysteine residues. Mutagenesis studies indicated that the intact iron site is required for methylation activity; however, its detailed function in Mmp10 catalysis remains unknown (50).

Like C-5 the arginine methylation, it was proposed that a radical SAM enzyme would be required to catalyze the methylation of the α-carbon of a glutamine residue to generate the 2-(S)-methylglutamine (Fig. 2E) found in the active sites of some MCRs (52), and the identity of this enzyme was recently confirmed (47). Bioinformatic investigations identified a radical SAM enzyme-encoding gene located nearby the mcr gene cluster and mmp10 in some methanogens (44). This gene encodes a predicted N-terminal cobalamin binding domain, as observed in other characterized cobalamin-dependent radical SAM methylases outside Mmp10. The enzyme from Methanoculleus thermophilus was heterologously expressed and purified from E. coli, followed by biochemical characterization (47). The reconstituted protein binds a single [4Fe-4S] cluster as well as cobalamin in a base-off, His-off state. Assays with a 28-mer peptide substrate resulted in the C-2 methylation of the expected glutamine residue and produced 5′-dAdoH and SAH in equal molar ratios. Additionally, the cobalamin cofactor was demonstrated to cycle between cob(I)alamin, methylcob(III)alamin, and cob(II)alamin, consistent with the established mechanism for cobalamin-dependent radical SAM methylases (51). The physiological relevance of the 2-(S)-methylglutamine PTM of MCR is not yet known, but this can now be explored through knockout studies since the identity of the gene required for the modification has been confirmed. Notably, the 2-(S)-methylglutamine modification exhibits a more sporadic distribution compared to the conserved 5-(S)-methylarginine modification in various MCRs for which the PTMs have been elucidated (53).

Heme is an essential prosthetic group for several enzymes involved in a range of fundamental biological processes in most organisms. In methanogens, heme is found in cytochrome-containing methanogens of the order Methanosarcinales, where cytochromes play a key role in energy metabolism (54). In all eukaryotes and most bacteria, heme is synthesized in a well-established canonical pathway (55). In the later steps of heme biosynthesis, two oxidative decarboxylation reactions are required to convert two propionate side chains of coproporphyrinogen III to two vinyl groups in the protoporphyrinogen IX product. In most bacteria, this reaction is catalyzed by oxygen-independent coproporphyrinogen III oxidase, a well-characterized radical SAM enzyme known as HemN that contains two simultaneously bound SAM molecules in the active site (56). Recent mechanistic studies have elucidated the roles of each SAM: the canonical SAM bound to the [4Fe-4S] cluster, SAM1, is the source of 5′-dAdo·, which abstracts a hydrogen atom from the methyl group of SAM2 to generate a SAM-based methylene radical. The latter species is uniquely proposed to be responsible for the key hydrogen atom abstraction from the β-carbon of a propionate side chain to initiate the vinyl conversion reaction (57).

Archaea utilize an alternative heme biosynthesis pathway where, among other differences in later steps of the pathway, the installation of the two vinyl side chains occurs in the last reaction involving the conversion of iron coproporphyrin III to heme b catalyzed by the heme synthase radical SAM enzyme AhbD (Fig. 2F) (58–60). Interestingly, AhbD from M. acetivorans and HemN from E. coli share only 17% identity, and although they catalyze the same reactions, they use different substrates: a porphyrinogen in the case of HemN and an iron-porphyrin for AhbD. In contrast to HemN, there is no evidence for AhbD binding more than one SAM. Furthermore, in addition to the radical SAM cluster, AhbD contains an auxiliary [4Fe-4S], which is coordinated by a cysteine-rich motif in the C terminus (61). In vitro enzymatic and mutagenesis studies on AhbD from M. barkeri indicated that the auxiliary cluster was essential for AhbD heme synthase activity but was not required for SAM cleavage and not directly involved in substrate binding (62). Based on electrochemical studies, the role of the auxiliary cluster was proposed as an electron acceptor necessary for the decarboxylation reaction (62). Thus, the proposed mechanism involves traditional radical SAM chemistry to generate 5′-dAdo· that, as opposed to HemN, abstracts an H atom directly from the propionate side chain of the substrate to initiate the decarboxylation reaction. The single electron resulting from the decarboxylation reaction is proposed to first reduce the heme iron to Fe(II) and then be transferred to the auxiliary cluster, which finally donates the electron to an external acceptor to reset the enzyme for another turnover (62).

The alternative heme biosynthesis pathway contains one additional radical SAM enzyme called AhbC, which catalyzes the penultimate step of the alternative pathway to remove two acetate side chains at C-2 and C-7 of 12,18-didecarboxysiroheme to produce the iron-coproporphyrinogen III substrate for AhbD (Fig. 2F) (58, 60). In M. acetivorans, AhbC has 31% identity and 52% similarity to AhbD and also contains two putative [4Fe-4S] cluster binding motifs, one of which is the canonical radical SAM motif CX₃CX₂C and the other of which is in the C terminus. This enzyme has not yet been biochemically characterized, and thus, the mechanistic details of the reaction remain unclear.

Cofactor F₄₂₀ (Fig. 2G) is a deazaflavin hydride carrier found in methanogens and select bacteria. This cofactor was first discovered in mycobacteria (63) and was subsequently structurally characterized from a methanogen (64, 65). The redox-active core ring system of F₄₂₀, 7,8-didemethyl-8-hydroxy-5-deazariboflavin, with the 5′-ribityl group, is known as F₀, while the complete F₄₂₀ contains an added phosphate as well as a bridging lactyl moiety followed by a γ-linked polyglutamate tail (Fig. 2G). Compared to the isoalloxazine ring of flavin cofactors, the deazaisoalloxazine ring of F₀ has a decreased reduction potential and can participate in two-electron chemistry only, as opposed to the characteristic one- or two-electron chemistry catalyzed by flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) (66). Bacteria containing F₄₂₀ use the cofactor in various redox reactions catalyzed by enzymes that often share homology to flavin-utilizing enzymes (67). Methanogens employ F₄₂₀ for reduction steps in the methanogenesis pathway as well as for other hydride transfer reactions in various pathways.

F₀ synthase is a radical SAM enzyme that carries out the key deazaflavin-forming step in F₄₂₀ biosynthesis to catalyze the oxidative coupling of l-tyrosine to 5-amino-6-ribitylamino-2,4[1H,3H]-pyrimidinedione (ribityl diaminouracil) (Fig. 2G), a common intermediate in the riboflavin biosynthetic pathway. In mycobacteria and other actinobacteria, F₀ synthase, encoded by fbiC, is a single bifunctional enzyme with two radical SAM motifs, while in archaea and cyanobacteria, F₀ synthase is comprised of two independent radical SAM enzymes known as CofG and CofH, which closely resemble the N- and C-terminal halves of FbiC, respectively (68, 69). In vitro studies demonstrated that CofH catalyzes the first step to produce a stable intermediate, followed by the action of CofG to produce the final F₀ molecule (70). Further detailed mechanistic experiments revealed that the first part of the CofH reaction likely proceeds similarly to those of other radical SAM tyrosine lyases such as HydG and ThiH to produce a p-hydroxylbenzyl radical (71), which can then undergo addition to the double bond of diaminouracil, followed by oxidation to produce an intermediate that is the substrate for CofG-catalyzed cyclization and oxidation to produce the final F₀ (71). No crystal structures have been reported for any F₀ synthase and would be important to obtain in the future to provide structural evidence for the mechanistic details of the two radical SAM reactions necessary for the synthesis of the unique deazaflavin core.

Found in some bacteria and archaea, nitrogenase plays an essential role in the global nitrogen cycle by reducing molecular nitrogen (N₂) to bioavailable ammonia (NH₃). The active site where N₂ reduction occurs consists of a complex metallocluster [Mo-7Fe-9S-C-(R)-homocitrate], referred to as FeMo-co, which can be described as a [4Fe-3S] cluster combined with a [Mo-3Fe-3S] cluster, connected by three sulfides and one carbide and partially ligated by (R)-homocitrate (Fig. 2H) (72). Molybdenum nitrogenases are the most common and most well characterized, but vanadium- and iron-nitrogenases also exist, where the above-mentioned metal replaces Mo in the unique metallocluster (73).

NifB is a radical SAM enzyme that plays a key role in FeMo-co biogenesis by converting the “K-cluster,” a pair of 4Fe-4S clusters, into the “L-cluster,” containing the bridging carbide carbon and the 9th sulfur atom that closely resembles the Fe-S core of FeMo-co (Fig. 2H) (74–77). Many details of NifB catalysis have been elucidated via in vitro mechanistic studies, and crystal structures have recently been reported (78, 79). Similar to many other radical SAM methylases (40), NifB uses SAM for two distinct reactivities: as a methyl group donor in a nucleophilic substitution reaction to generate SAH as a by-product and as a source of 5′-dAdo· in the radical SAM reaction to produce a substrate radical and 5′-dAdoH as a by-product. Thus, SAM is first used to methylate a sulfide atom in one of the [4Fe-4S] clusters that make up the K-cluster. Next, 5′-dAdo· produced at the radical SAM [4e-4S] cluster abstracts a hydrogen atom from the newly added methyl group to generate a methylene radical (80). Further processing of this species, using either an additional round(s) of radical SAM chemistry or recently proposed deprotonation facilitated by a His residue (81), finally results in the bridging SAM-derived carbide present in FeMo-co. This process is accompanied by the coupling and rearrangement of the two [4Fe-4S] modules of the K-cluster to generate the L-cluster (74) (Fig. 2H). A major remaining question regarding the NifB reaction is the mechanism involved in inserting the 9th sulfur. Sulfite has been shown to be the sulfur source in vitro (82), but the NifB-facilitated chemistry that would be involved in this process remains unclear.

Hydrogenases are abundant across various organisms in all domains of life, where they are responsible for the reversible conversion of H₂ to protons and electrons. These metalloenzymes are often categorized based on the transition metals present in their active site cofactors: [NiFe], [FeFe], or [Fe]. Methanogens utilize five different types of hydrogenases, four of which are [NiFe] hydrogenases and one of which is an [Fe]-hydrogenase that is found only in methanogens without cytochromes (83). The latter is employed under nickel-limiting conditions to replace the function of the F₄₂₀-reducing [NiFe] hydrogenase required for the reduction of methenyltetrahydromethanopterin to methylenetetrahydromethanopterin during hydrogenotrophic methanogenesis (84–86). Additionally, when nickel is limiting, the [Fe]-hydrogenase is also required for the production of reduced coenzyme F₄₂₀, which is involved in several essential processes in methanogens (83).

The [Fe]-hydrogenase utilizes the unique iron-guanylylpyridinol (FeGP) cofactor (Fig. 2I), whose biosynthetic pathway is still not completely resolved. This biosynthesis utilizes genes located in the hcg gene cluster containing seven genes (hcgA–G) that cooccur with the [Fe]-hydrogenase-encoding genes. Recent in vitro studies confirmed the roles of HcgC and HcgB in the production of guanylylpyridinol (GP) in the early stages of FeGP cofactor biosynthesis (87). Furthermore, enzymatic and structural studies demonstrated that HcgE and HcgF catalyze the formation of a thioester-activated acyl group using a two-stage mechanism where HcgE catalyzes the adenylylation of the carboxy group of GP followed by a transesterification reaction to generate a Cys-S-GP thioester on the HcgF protein (88). This is presumably the activated substrate that is then converted to the final FeGP cofactor. HcgA is a radical SAM enzyme present in the hcg gene cluster that contains a noncanonical CX₅CX₂C motif. Recombinant HcgA has been studied in vitro to demonstrate that the enzyme contains a radical SAM-characteristic [4Fe-4S] cluster and carries out uncoupled SAM cleavage to produce 5′-dAdoH in the presence of dithionite (89). Due to the sequence similarity of HcgA to the well-characterized radical SAM enzyme HydG, which synthesizes the CN and CO ligands of the H-cluster in [FeFe]-hydrogenase (90), it has been proposed that HcgA similarly produces the CO ligands for the FeGP cofactor (89). However, in vivo isotope labeling studies indicated that the CO ligands of the FeGP cofactor are derived from CO₂, and no incorporation from pyruvate was observed (91), whereas the CO ligands of the H-cluster are derived from l-tyrosine via dehydroglycine (92, 93). These results indicate that the synthesis of CO ligands for the FeGP cofactor occurs by a different pathway/mechanism compared to the H-cluster, and thus, many open questions remain surrounding the role of HcgA in FeGP cofactor biosynthesis. Future work on HcgA should be facilitated by the recent development of an in vitro system to study the later steps of FeGP cofactor biosynthesis using precursors isolated from Δhcg mutated strains (87).

tRNAs are highly modified in all organisms, with up to 10 to 20% of nucleosides modified in a given tRNA and over 170 different types of modifications that have been described (94, 95). The distribution of tRNA modifications can be highly variable, where some are found at the same position in all tRNAs throughout all domains of life and others are found in only one or a few tRNAs in one domain of life (96, 97). These modifications ensure accurate decoding and aminoacylation and stabilize the tertiary structure of tRNAs. Modifications can be simple (e.g., methylation or isomerization) or complex, requiring multiple enzymatic steps. Several radical SAM enzymes in methanogens are involved in tRNA modifications (Table 1).

Archaeosine (7-formamidino-7-deazaguanine [G⁺]) (Fig. 2J) is a structurally complex modified nucleoside in archaea, where it is found in the D-loop at position 15 of virtually all archaeal tRNAs (98). As a 7-deazaguanosine nucleoside, archaeosine is related to queuosine, the latter of which is ubiquitous throughout eukaryotic and bacterial organisms. The biosynthesis of both of these modified nucleosides involves the common intermediate 7-cyano-7-deazaguanine (preQ₀) (Fig. 2J). The biosynthesis of preQ₀ from GTP involves the action of four or five enzymes, one of which is QueE, a ubiquitous radical SAM enzyme that catalyzes the key heterocyclic rearrangement reaction to convert 6-carboxy-tetrahydropterin to 7-carboxy-7-deazaguanine, expelling a nitrogen as ⁺NH₄ (99). In archaea, preQ₀ serves as a substrate for tRNA-guanine transglycosylase, which catalyzes the exchange of guanine at position 15 for preQ₀ (100). In most Euryarchaeota, including methanogens, preQ₀-tRNA is then converted to G⁺-tRNA by archaeosine synthase (ArcS), of which the enzyme from M. jannaschii has been investigated in vitro (101). While it was originally thought that ArcS alone synthesizes G⁺ from PreQ₀, it was recently shown that some Euryarchaeota, including M. acetivorans, utilize a complex consisting of ArcS and the radical SAM enzyme for archaeosine formation (RaSEA) (102). Here, ArcS first transfers lysine to preQ₀, where the ε-amino group of lysine is added to the cyano group of preQ₀ (Fig. 2J). RaSEA is then proposed to abstract a hydrogen atom from the ε-carbon of the lysine moiety, which activates the molecule for C-N bond cleavage to eventually form G⁺ and 1-piperidine-6-carboxylic acid as a by-product (102). Virtually all sequenced Euryarchaeota contain homologs of both ArcS and RaSEA; thus, this may be the main pathway for G⁺ synthesis in these organisms (102). In the future, further biochemical and structural studies of ArcS/RaSEA should be carried out to elucidate the mechanistic details of the reaction and to clarify the dependence of ArcS on a radical SAM partner since in vitro experiments with ArcS from M. jannaschii showed that the enzyme alone can produce G⁺ from preQ₀ (101).

Wyosine derivatives are fluorescent modified nucleosides found in eukaryotes and archaea but not bacteria. They are located at position 37 of phenylalanine-specific tRNA and consist of an imidazopurine (tricyclic) core structure. The key ring formation step in the biosynthesis of these wyosine derivatives is catalyzed by wybutosine 4-demethylwyosine synthase (Tyw1), also referred to as Taw1 in archaea (103). This radical SAM enzyme forms the tricyclic ring of 4-demethylwyosine from N¹-methylguanosine and pyruvate (Fig. 2K) (104), which is an intermediate in the biosynthesis of several imidazopurine-containing nucleosides, including wyosine and wybutosine and the archaeon-specific isowyosine and 7-methylwyosine (105–107). The Tyw1 reaction is initiated with hydrogen atom abstraction from the methyl group of N¹-methylguanosine, which then reacts with pyruvate, followed by decarboxylation and transamination to finally form the tricyclic product (108). Interestingly, eukaryotic Tyw1 contains an N-terminal flavodoxin domain used for reducing the radical SAM [4Fe-4S] cluster to the catalytic +1 state; however, this domain is absent in the archaeal enzyme (109). A crystal structure of Tyw1 from M. jannaschii revealed that pyruvate is bound in the active site via Schiff base formation with a conserved lysine residue (110). Additionally, the structure showed that Tyw1 contains an auxiliary [4Fe-4S] cluster that is partially coordinated by the lysine nitrogen and carboxylate oxygen of the Schiff base intermediate. Thus, the auxiliary cluster is proposed to serve as an electron sink to facilitate the decarboxylation of pyruvate (110).

Residue 37, which is 3′ adjacent to the anticodon, is a frequently modified position in most tRNAs. The modified nucleosides found exclusively in this position include N⁶-isopentenyladenosine (i⁶A), N⁶-threonylcarbamoyladenosine (t⁶A), and N⁶-hydroxynorvalylcarbamoyladenosine (hn⁶A) (95, 97). Depending on the sequence features of the specific tRNA, all three of these can be further modified via the addition of a methylthio group at the 2′ position to generate ms²i⁶A, ms²t⁶A, and ms²hn⁶A, respectively. In methanogens, the threonyl-containing (t⁶A and ms²t⁶A) and hydroxynorvalyl-containing (hn⁶A and ms²hn⁶A) modifications have been identified in various organisms, but the isopentenyl-containing modifications are absent (105, 107, 111–113). The methylthiolation reaction is catalyzed by a radical SAM methylthiotransferase (MTTase): MiaB for the ms²i⁶A modification and MtaB for the ms²t⁶A modification (Fig. 2L). The bacterial MiaB MTTase has been well characterized with detailed mechanistic and structural information available. MTTases contain two iron-sulfur clusters: the radical SAM cluster and an auxiliary cluster that is the source of the sulfur atom that is eventually transferred to the tRNA substrate (114–117). The first stage of MTTase catalysis involves a nucleophilic substitution reaction to transfer the methyl group of SAM to a sulfur species associated with the auxiliary cluster that is proposed to be [3Fe-4S]⁰ (115). In the second stage, radical SAM-dependent chemistry facilitates the transfer of the resulting methylthio group to C-2 of i⁶A to produce ms²i⁶A (114–117). MtaB is expected to utilize analogous chemistry for the methylthiolation of t⁶A; however, although the enzyme from Bacillus subtilis has been purified and its [4Fe-4S] clusters have been characterized (118), the in vitro catalytic activity has not yet been investigated, and no MtaB structural information is available. Additionally, although MtaB homologs are proposed to be responsible for the methylthiolation of hn⁶A to produce ms²hn⁶A (Fig. 2L) (119), this has yet to be experimentally confirmed.

5-Carboxymethyluridine (cm⁵U) is an intermediate modified nucleoside that leads to a variety of uridine modifications at the wobble position of tRNA in primarily eukaryotes and archaea (120). The synthesis of this modification is carried out by Elp3 (Fig. 2M), a radical SAM enzyme present in the elongator complex, a large protein complex involved in cm⁵U synthesis and previously implicated in histone acetylation. Enzymatic studies of Elp3 from the methanogen Methanocaldococcus infernus confirmed that the enzyme is solely responsible for the cm⁵U modification in vitro and utilizes radical SAM-characteristic chemistry (121). Crystal structures of a rare bacterial Elp3 (122) as well as M. infernus Elp3 (123) revealed a radical SAM domain that is highly similar to that of RlmN (124), the radical SAM methylase that modifies specific tRNA and rRNA. Elp3 also contains a C-terminal lysine acetyltransferase (KAT) domain similar to those of the GCN5-like acetyltransferases (125). Based on the available biochemical and structural data, the following mechanistic features have been proposed: (i) tRNA substrate binding initiates a conformational change that leads to acetyl-CoA binding in the KAT domain (123); (ii) acetyl-CoA hydrolysis occurs, and the released acetate is transferred to the radical SAM domain (123); (iii) radical SAM chemistry leads to hydrogen atom abstraction from the methyl moiety of acetate, which then adds to C-5 of the uridine residue to form a C-C bond (121); and (iv) the loss of an electron followed by general base-catalyzed proton abstraction results in the final cm⁵U product (121). Further detailed mechanistic studies are required to confirm the proposed mechanism, especially with respect to the chemistry occurring in the radical SAM domain. Additionally, the specific roles of the additional proteins present in the elongator complex need to be elucidated.

Radical SAM enzymes in methanogens catalyze several unique reactions compared to those found in other organisms and often utilize especially intriguing catalytic mechanisms, even among radical SAM superfamily members. Based on the presence of the canonical CX₃CX₂C motif, the model methanogens M. maripaludis S2 and M. acetivorans C2A have 32 and 58 putative radical SAM proteins (see Tables S1 and S2 in the supplemental material), respectively, with only about half of these having established functions based on experimental evidence or high sequence similarity to well-characterized radical SAM enzymes. Thus, many questions remain regarding the physiological functions and catalytic versatility of radical SAM enzymes in methanogens. Characterizing these enzymes will further expand the catalytic repertoire of the radical SAM superfamily and aid in defining previously unknown aspects of primary metabolism in methanogens. Furthermore, understanding the functions and mechanisms of unique radical SAM enzymes in methanogens could inform methane mitigation strategies as well as bioengineering efforts for biofuel applications.

We thank Anne Brown and Amanda Sharp (Virginia Tech Department of Biochemistry) for bioinformatics assistance to identify radical SAM domain-containing proteins in various methanogenic genomes.

Research on methanogenic radical SAM enzymes in the Allen Lab is funded by the National Science Foundation (CHE-2105598).