Physiology, Biochemistry, and Applications of F₄₂₀- and F_o-Dependent Redox Reactions

SUMMARY

5-Deazaflavin cofactors enhance the metabolic flexibility of microorganisms by catalyzing a wide range of challenging enzymatic redox reactions. While structurally similar to riboflavin, 5-deazaflavins have distinctive and biologically useful electrochemical and photochemical properties as a result of the substitution of N-5 of the isoalloxazine ring for a carbon. 8-Hydroxy-5-deazaflavin (F_o) appears to be used for a single function: as a light-harvesting chromophore for DNA photolyases across the three domains of life. In contrast, its oligoglutamyl derivative F₄₂₀ is a taxonomically restricted but functionally versatile cofactor that facilitates many low-potential two-electron redox reactions. It serves as an essential catabolic cofactor in methanogenic, sulfate-reducing, and likely methanotrophic archaea. It also transforms a wide range of exogenous substrates and endogenous metabolites in aerobic actinobacteria, for example mycobacteria and streptomycetes. In this review, we discuss the physiological roles of F₄₂₀ in microorganisms and the biochemistry of the various oxidoreductases that mediate these roles. Particular focus is placed on the central roles of F₄₂₀ in methanogenic archaea in processes such as substrate oxidation, C₁ pathways, respiration, and oxygen detoxification. We also describe how two F₄₂₀-dependent oxidoreductase superfamilies mediate many environmentally and medically important reactions in bacteria, including biosynthesis of tetracycline and pyrrolobenzodiazepine antibiotics by streptomycetes, activation of the prodrugs pretomanid and delamanid by Mycobacterium tuberculosis, and degradation of environmental contaminants such as picrate, aflatoxin, and malachite green. The biosynthesis pathways of F_o and F₄₂₀ are also detailed. We conclude by considering opportunities to exploit deazaflavin-dependent processes in tuberculosis treatment, methane mitigation, bioremediation, and industrial biocatalysis.

1. INTRODUCTION

Flavin- and deazaflavin-dependent enzymes mediate a wide range of redox reactions in biological systems (1, 2). Flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) are versatile flavin cofactors that are central to metabolism across the three domains of life. Some organisms also synthesize and utilize 5-deazaflavin compounds (3, 4), in which a carbon atom substitutes for the N-5 atom of the isoalloxazine ring. Two such compounds are relevant to biological systems, namely, 7,8-didemethyl-8-hydroxy-5-deazariboflavin (F_o) and its lactyl oligoglutamate phosphodiester derivative (F₄₂₀) (Fig. 1) (5, 6). While structurally similar to flavins, these compounds have markedly different physicochemical properties (6–9): they serve as obligate two-electron hydride carriers, have low standard redox potentials (−340 mV), and have blue-shifted intrinsic fluorescence. As elaborated upon in section 2, the chemical properties and biological functions of F₄₂₀ are in fact more similar to nicotinamides (i.e., NAD, NADP) than flavins, leading to its description as a “nicotinamide in a flavin's clothing” (7, 10).

FIG 1.

Structures of riboflavin, F_o, and F₄₂₀.

F_o and F₄₂₀ have entirely distinct physiological roles. F_o is distributed across the three domains of life (Bacteria, Archaea, and Eukarya), but it appears to serve only one function: as a light-harvesting antenna in some DNA photolyases that repair pyrimidine dimers following exposure to UV light. As a result, F_o can be considered a chromophore rather than a cofactor; while it can substitute for F₄₂₀ in vitro (11–13), it does not appear to have any redox roles in living cells. The biosynthesis, distribution, and photochemistry of this chromophore are covered in section 2. In contrast to F_o, F₄₂₀ has a very limited taxonomic distribution and has been chemically identified in only two phyla thus far (Euryarchaeota and Actinobacteria). However, this cofactor has diverse catalytic roles in such organisms and mediates many of the challenging redox transformations necessary for their catabolic, detoxification, and biosynthetic pathways. F₄₂₀ appears to have been selected for such processes due to its unique electrochemical properties compared to other flavins, namely, its two-electron reactivity and low redox potential. By maintaining a pool of hydride transfer redox cofactors separate from NAD(P), cells may also be able to better control the flux of specific redox reactions. The roles and enzymology of the reactions catalyzed by F₄₂₀ are discussed in sections 3 and 4 of this review.

Nine years after the discovery of methanogenesis (14), Cheeseman et al. formally identified F₄₂₀ in 1972 (5) in Wolfe's laboratory. They demonstrated that the compound was responsible for the characteristic 420-nm absorbance and blue-green fluorescence of oxidized lysates of Methanobacterium bryantii (5). The compound, thereafter named factor 420 (abbreviated F₄₂₀; sometimes called coenzyme F₄₂₀ or cofactor F₄₂₀), was shown to be a redox-active 5-deazaflavin derivative (6) that is present at levels up to 400 mg/kg in methanogens (15). It was demonstrated that F₄₂₀ facilitated multiple central metabolic redox reactions in methanogens, including oxidation of energy sources (H₂ and formate) (16, 17) and reduction of cofactors (NADP and tetrahydromethanopterin) (16, 18). Later, it was realized that F₄₂₀ is also synthesized by sulfate-reducing archaea (19), halophilic archaea (20), and likely methanotrophic archaea (21). As a result of more than 5 decades of study, scientists developed a rich understanding of the physiology and biochemistry of F₄₂₀ in the methanogenic and sulfate-reducing archaea (22), as summarized in section 3.

However, our understanding of the roles of F₄₂₀ in bacteria remains in its infancy. While cofactors with properties corresponding to F₄₂₀ were isolated in mycobacteria and streptomycetes in 1960 (23, 24), it was not until decades later that the cofactor was formally identified in these genera (25–27). As discussed throughout section 4, F₄₂₀ is implicated in the catabolic, biosynthetic, and detoxification pathways of both saprophytic actinobacteria (28–30) and their pathogenic descendants (e.g., Mycobacterium tuberculosis) (31, 32). Interest in F₄₂₀ metabolism has surged following the discovery that the recently clinically approved antimycobacterial prodrug delamanid is activated by a specific F₄₂₀H₂-dependent reductase (33–36). However, the physiological and pharmacological roles of F₄₂₀ are still poorly understood in actinobacteria, and the majority of the predicted F₄₂₀-dependent enzymes in such organisms remain functionally unannotated (30, 37). There is also genomic evidence that F₄₂₀ might be more widely distributed than previously thought, with potential roles in Chloroflexi, Alphaproteobacteria, and Betaproteobacteria inhabiting aerated soil ecosystems (30, 37). This review concludes by considering the diverse implications and potential environmental, medical, and industrial applications of deazaflavin compounds (section 5).

2. 5-DEAZAFLAVIN COMPOUNDS

2.1. Properties

The structure of F_o (7,8-didemethyl-8-hydroxy-5-deazariboflavin; also sometimes referred to as 8-HDF, F₀, and FO) is similar to that of riboflavin (Fig. 1). However, its physical and chemical properties are modulated by three substitutions in its isoalloxazine rings (38): N-5 is substituted for a carbon, C-7 and C-8 are demethylated, and C-7 is hydroxylated (6). F₄₂₀ is a derivative of F_o; the ribityl side chain forms a phosphoester bond, with a lactate moiety forming the phosphodiester and linking to an oligoglutamate chain (6). While the substitutions that distinguish 5-deazaflavins from flavins may seem superficial, pioneering work by Walsh has shown that they profoundly influence the physicochemical properties of these molecules (7, 8, 39, 43). Several years prior to their discovery in biology, chemically synthesized 5-deazaflavins (3, 4, 39) were used as probes to study the flavin-dependent reactions (40–43), revealing distinct electrochemical and photochemical properties from their flavin counterparts (44). Upon the discovery of 5-deazaflavins in biological systems (5, 6), it was realized that the electrochemical properties of these compounds are central to the role of F₄₂₀ as a redox cofactor (6), while the photochemical properties are exploited by F_o as an antenna chromophore for DNA photolyases (45). Three features define the roles of 5-deazaflavins in biology.

(i) Two-electron carrier.

Whereas flavins can serve as one or two electron carriers, 5-deazaflavins are obligate two-electron (hydride) carriers (44, 46). This is because flavins are stable as semiquinones (both neutral and anionic), whereas 5-deazaflavins are not. The nitrogen atom in position 5 is required for an unpaired electron to efficiently delocalize through the isoalloxazine ring; indeed, radicals of pyrazine groups (of flavins) are much lower energy than those of pyridine groups (of 5-deazaflavins) (7, 43). Reflecting this reactivity, F₄₂₀-dependent enzymes mediate diverse hydride transfer reactions that transform C=C and CC bonds (28, 29, 47, 48), alcohol and imine groups (49, 50), and certain inorganic compounds (51, 52). Furthermore, due to the substitution, 5-deazaflavins do not readily undergo single-electron reactions. Thus, unlike flavins, reduced 5-deazaflavins are relatively stable against air oxidation with a half-life on the order of hours instead of seconds for flavins (39, 44). This autooxidation in air has also been reported to be influenced by other factors such as stimulation from ambient light (8, 44) and, in the case of F₄₂₀ and F_o, the addition of the 8-hydroxy group that results in the formation of a delocalized paraquinoid anion upon deprotonation of the oxidized species at pH above 6 (8). The low electrophilic reactivity of this anion results in a slower disproportionation/self-exchange reaction between F₄₂₀ and F₄₂₀H₂ (8). Similarly, 5-deazaflavins also exhibit reduced reactivity with reducing agents that act primarily as single-electron donors (e.g., dithionite) (6, 8, 39).

(ii) Strong reductant.

As a result of the substitution of N-5 to C-5, 5-deazariboflavin has a much lower standard redox potential (−310 mV) than riboflavin (−210 mV), FAD (−220 mV), or FMN (−190 mV) (7, 53). Due to the electron-withdrawing groups added to the isoalloxazine ring, F_o and F₄₂₀ are even stronger reductants (−340 mV) than 5-deazariboflavin and thus some of the lowest-potential redox cofactors in biology (8, 9). This redox potential may be modulated under physiological conditions; for example, it will be −380 mV at standard temperature in hydrogenotrophic methanogens that maintain a 10:1 ratio of oxidized to reduced F₄₂₀ (9). This redox potential places F₄₂₀ at the center of the redox biology of methanogens (Table 1); the compound is capable of being reduced by exogenous fuels (H₂ and formate) and reoxidized by key cofactors (NADP and tetrahydromethanopterin derivatives) in an energetically efficient manner (7, 8, 53). Bacteria likewise appear to tightly couple substrate oxidation (glucose-6-phosphate and NADPH) to F₄₂₀ reduction, presumably to enhance catalytic efficiency (Table 1). Partly due to its low redox potential, the F₄₂₀H₂ produced is capable of reducing a wide range of organic compounds otherwise recalcitrant to activation as discussed in section 4 (28, 54, 55). Recent work also indicates that F₄₂₀ may be utilized in aerobic bacteria in hypoxic and anoxic environments, potentially substituting for high-potential nicotinamide cofactors (NAD and NADP) (−320 mV) (30, 32, 56).

TABLE 1.

List of standard redox potentials for key F₄₂₀-linked redox reactions^a

Substrateb	Reaction	E0′ (mV)	Reference
Ferredoxin	Fd + 2 e− → Fd2−	−500 to −400	487
CO2/formate	CO2 + 2 e− + H+ → HCO2−	−420	487
H+/H2	2 H+ + 2 e− → H2	−410	487
Methenyl/methylene H4MPT	CHH4MPT + 2 e− + H+ → CH2=H4MPT	−390	301
F420	F420 + 2 e− + 2 H+ → F420H2	−340	8
6PGL/G6P	6-Phosphogluconolactone + 2 e− + 2 H+ → Glucose-6-phosphate	−330	488
Methylene/methyl H4MPT	CH2=H4MPT + 2 e− + H+ → CH3-H4MPT	−320	301
NAD(P)+	NAD(P)+ + 2 e− + H+ → F420H2	−320	487
Acetone/propan-2-ol	Acetone + 2 e− + 2 H+ → Propan-2-ol	−290	53
FAD	FAD + 2 e− + 2 H+ → FADH2	−220	53
Riboflavin	Riboflavinox + 2 e− + 2 H+ → Riboflavinred	−210	53
FMN	FMN + 2 e− + 2 H+ → FMNH2	−190	53
Methanophenazine	Mphenox + 2 e− + 2 H+ → Mphenred	−170	489
Heterodisulfide	CoM-S-S-CoB + 2 e− + 2 H+ → CoM-SH + CoB-SH	−140	487
Sulfite/sulfide	SO3− + 6 H+ + 6 e− → S− + 3 H2O	−120	490
Menaquinone	Menaquinone + 2 e− + 2 H+ → Menaquinol	−70	53
O2/H2O	O2 + 4 H+ + 4 e− → 2 H2O	+820	53

^aThis list of standard redox potentials (E₀′) demonstrates that the electrochemical properties of F₄₂₀ enable the cofactor to mediate a wide range of oxidation and reduction reactions in biological systems, especially methanogenic archaea. In whole cells, physiological redox potentials can differ considerably due to the mass action ratios of substrates/products and differences in physical conditions (487). Potentials were determined under standard conditions (25°C, 1 atm, pH 7.0) against the standard hydrogen electrode.

^b6PGL, 6-phosphogluconolactone; Mphen_ox and Mphen_red, oxidized and reduced methanophenazine, respectively.

(iii) Intrinsic fluorophore.

Like flavins, 5-deazaflavins are intrinsically fluorescent compounds. The delocalized charge on the isoalloxazine ring undergoes π → π* transitions upon exposure to UV-visible light. In its oxidized state, the absorbance spectrum of F₄₂₀ peaks at 420 nm, and the emission spectrum peaks at 470 nm (6) (Fig. 2). These peaks are pH dependent with a shift in the absorbance peak to 375 nm at lower pH along with reduced intensity (6). The reduced species F₄₂₀H₂ loses the absorbance peak at 420 nm for a new peak at 320 nm with a lower molar absorption coefficient (6) (Fig. 2). Due to the substitution of C-5 to N-5, the visible absorption spectra and fluorescence emission spectra of 5-deazaflavins are blue-shifted by about 50 nm compared to flavins (6, 44). As a result, light captured by 5-deazaflavins can be efficiently transferred to flavins through Förster resonance energy transfer (FRET). As elaborated below, this is central to the mechanism of the F_o-utilizing DNA photolyases (57, 58). The autofluorescence of F₄₂₀ has also been used for detecting methanogens (59–66) and mycobacteria (67, 68).

FIG 2.

UV-visible absorption spectra of F₄₂₀ (blue) and F₄₂₀H₂ (red). Adapted from reference 31.

2.2. Chromophore F_o

2.2.1. Biosynthesis

Despite its structural similarity to riboflavin, the biosynthetic pathway for F_o and other 5-deazaflavins diverges at an early step in the pathways leading to the synthesis of flavin cofactors (Fig. 3). The deazaflavin and flavin biosynthetic pathways both proceed from the pyrimidine ribityldiaminouracil (5-amino-6-ribitylamino-2,4[1H,3H]-pyrimidinedione). In the flavin pathway, this substrate is condensed with 3,4-dihydroxy-2-butanone 4-phosphate to make a lumazine derivative (6,7-dimethyl-8-ribityllumazine) (69); two of these molecules subsequently condense to regenerate 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione with concomitant production of riboflavin (69). In the deazaflavin pathway, ribityldiaminouracil is instead condensed with the amino acid tyrosine (not 4-hydroxyphenylpyruvate as previously proposed [70]) leading to formation of F_o (71). The enzyme responsible for this condensation step, F_o synthase, is encoded by two polypeptides in archaea (CofG and CofH) (70) and a two-domain fusion protein (FbiC) in bacteria and eukaryotes (72). Each subunit/domain contains a radical S-adenosylmethionine (radical SAM) catalytic site (71, 73). A recent mechanistic study demonstrated that formation of the complex heterocycle depends on the coordinated action of the two radical SAM active sites, each of which abstract a hydrogen atom from the tyrosine (73).

FIG 3.

Summary of flavin and deazaflavin biosynthesis pathways.

2.2.2. Distribution

F_o serves as an antennal chromophore in DNA photolyases in a range of organisms across the three domains of life. Auxiliary to the catalytic chromophore FADH⁻, F_o captures light more effectively than FADH⁻ owing to its longer wavelength absorption maximum and higher molar absorption coefficient (74). This is particularly important under low-light conditions, during which F_o enhances the efficiency of DNA repair by orders of magnitude (75). F_o-utilizing photolyases have been identified in multiple bacteria (e.g., Synechococcus elongatus, Streptomyces griseus) (76–80), archaea (e.g., Methanothermobacter marburgensis, Methanosarcina mazei, Halobacterium halobium) (81–83), and unicellular eukaryotes (e.g., Acutodesmus obliquus, Chlamydomonas reinhardtii, Ostreococcus tauri) (84–86). Genes encoding probable F_o synthases (CofG/CofH or FbiC) are consistently present in the genomes of such microorganisms. The question of whether F_o is utilized in higher eukaryotes is more controversial. Structural and chemical studies have demonstrated that F_o binds tightly to, and enhances the efficiency of, the two photolyases of the higher eukaryote Drosophila melanogaster (85, 87). Catalytically active and nucleus-targeted F_o-utilizing DNA photolyases are also known to be produced by insect baculoviruses (88–93). However, it is perplexing how such photolyases could utilize F_o in vivo, given that the genomes of higher eukaryotes lack F_o synthase-encoding genes (94). One explanation is that the dispensable F_o-binding domain of such enzymes is an evolutionary remnant, although it is also plausible that these organisms carry genes that encode components of a novel F_o biosynthesis pathway or acquire F_o from microbial endosymbionts and baculoviruses (85); in contrast to the highly anionic cofactors F₄₂₀, FMN, and FAD, F_o is uncharged and hence can readily diffuse through cell membranes (95–97). While F_o-utilizing DNA photolyases are widespread, they are hardly universal: photolyases of many species use different antennal chromophores or lack them altogether (75, 98), while eutherian lineages appear to have lost the capacity for light-driven DNA repair (99).

2.2.3. Enzymology

Enzymes of the DNA photolyase superfamily use the energy of blue light (350 to 450 nm) to facilitate the reductive cleavage of DNA pyrimidine dimers formed by far UV irradiation (200 to 300 nm). Distinct, but related, photolyases cleave cyclobutane pyrimidine dimers (CPD photolyases) and pyrimidine-pyrimidone photoproducts (6-4 photolyases) (75, 98). All DNA photolyases use the twice-reduced flavin FADH⁻ as the catalytic chromophore. Most photolyases also use an antennal chromophore to optimize light capture, namely, methenyltetrahydrofolate or the flavin/deazaflavin compounds F_o, FMN, or FAD (100–103). Crystal structures reveal that the F_o-utilizing CPD photolyase (76, 77) from Synechococcus elongatus (Protein Data Bank [PDB] identifiers [IDs] 1QNF, 1TEZ, and 1OWL) (57, 104–106) is a single-subunit enzyme containing an N-terminal α/β domain and a C-terminal α-helical domain. Both chromophores are deeply buried, with F_o located in a cleft between the domains and FADH⁻ embedded in the α-helical domain (Fig. 4) (57, 106). The ∼17-Å distance between the chromophores enables efficient FRET while potentially preventing competitive electron transfer reactions between the cofactors (74).

FIG 4.

Structure and function of the F_o-utilizing DNA photolyases. (a) Crystal structure of the F_o-utilizing CPD photolyase of Synechococcus elongatus (PDB ID 1TEZ) (106). (b) Catalytic cycle of the enzyme. FRET is an acronym for Förster resonance energy transfer. The blue asterisk after FADH⁻ indicates that the molecule is in the excited state.

The catalytic cycle of F_o-utilizing CPD photolyases has been elucidated through extensive spectroscopic and structural studies on the S. elongatus photolyase (Fig. 4). In the light-independent initial reaction, the enzyme recognizes and binds to damaged duplex DNA on the basis of its bent orientation (106). The antennal chromophore F_o thereafter captures a photon of blue light with an absorbance peak at 437 nm (red-shifted due to the strong interaction of the chromophore with the protein) (77). Femtosecond-scale spectroscopic studies show that F_o then transfers the energy to FADH⁻ through FRET (107). The excited catalytic chromophore (FADH⁻*) thereafter transfers an electron to the pyrimidine dimer, leading to its cleavage, and back-electron transfer restores the catalytic chromophore to an active form ready for a second catalytic cycle (108, 109). As reviewed in detail elsewhere (75, 110), similar reaction cycles facilitate light capture by other antennal chromophores and cleavage of pyrimidine-pyrimidone dimers. F_o-dependent photolyases are generally more efficient than methenyltetrahydrofolate-dependent ones, and the quantum yields of the energy transfer and electron transfer steps have been shown to be at near-unity (58, 107).

2.3. Cofactor F₄₂₀

2.3.1. Biosynthesis

The chemical structure of F₄₂₀, a lactyloligoglutamyl phosphodiester of F_o, was inferred from spectroscopic analysis of its degradation products (6) and validated by chemical synthesis (111–113) (Fig. 1). Reflecting its modular molecular structure, F₄₂₀ is synthesized from several precursors: F_o, lactate, the amino acid glutamate, and the nucleotide GTP (97, 114, 115). Through a combination of biochemical and genetic studies in methanogens and mycobacteria, the majority of the steps in the F₄₂₀ biosynthetic pathway have been resolved (Fig. 5).

FIG 5.

Summary of the F₄₂₀ biosynthesis pathway from F_o.

There are two major steps in the conversion of F_o to F₄₂₀. In the first, the lactate-derived intermediate l-lactyl-2-diphospho-5′-guanosine (LPPG) is condensed with F_o (116) to form the phosphodiester F₄₂₀-0 (i.e., F₄₂₀ containing no glutamate side chain). This reaction is catalyzed by a 2-phospho-l-lactate transferase (named CofD in archaea and FbiA in actinobacteria) (117, 118). The structure of this enzyme (PDB ID 3C3D) demonstrates that the deazaflavin ring of F_o interacts with a hydrophobic pocket and two water molecules, while the nucleotide moiety of LPPG is accommodated in a Rossmann fold domain with a Mg²⁺ ion. It is proposed that, following conformational changes initiated by substrate binding, the condensation proceeds following the abstraction of a proton from the terminal hydroxyl group of F_o by the β-phosphate of LPPG (119).

Thereafter, the nonribosomal peptide synthase F₄₂₀:γ-l-glutamyl ligase (CofE/FbiB) catalyzes the GTP-dependent addition of an oligoglutamate tail (118, 120–122). l-Glutamate residues are added via γ-linkages to F₄₂₀-0 (F₄₂₀-0 + glutamate + GTP → F₄₂₀-1 + GDP + P_i) and glutamated derivatives thereof (F₄₂₀-n + glutamate + GTP → F₄₂₀-n + 1 + GDP + P_i) in a sequential manner. The X-ray crystallographic structure of the enzyme from Archaeoglobus fulgidus (PDB ID 2PHN) demonstrates that it forms a butterfly-like homodimer that accommodates GTP and Mn²⁺ at the dimer interface. It is proposed that the cofactor is activated by phosphorylation (at the terminal hydroxyl group of the lactate moiety of F₄₂₀ and the terminal glutamate of F₄₂₀-n derivatives), and the resultant acyl-phosphate is subject to nucleophilic attack by the amino group of the incoming glutamate residue (123). The number of glutamate residues on F₄₂₀ is highly species specific, ranging from two or three in methanogens without cytochromes (124), four or five in methanogens with cytochromes (124), and five to seven in mycobacteria (125). The physiological significance and biochemical basis for these differences is not yet understood. In some archaea, a terminal α-linked glutamate residue (126, 127) is also added by γ-F₄₂₀-2:α-l-glutamate ligase (CofF) (128), an enzyme of the ATP-grasp superfamily.

The pathway that leads to the production of LPPG from the precursor l-lactate has only been partially resolved. Detailed studies on Methanocaldococcus jannaschii indicate that lactate is exclusively synthesized from l-lactaldehyde (129, 130); lactaldehyde is generated from the reduction of methylglyoxal or the aldol cleavage of fuculose-1-phosphate and is in turn oxidized to lactate by the NAD⁺-dependent l-lactaldehyde dehydrogenase (CofA) (130). Though unconfirmed, it is assumed that lactate (synthesized from glycolytic pyruvate by l-lactate dehydrogenase) is also the precursor for LPPG in bacteria. It has been shown in methanogens that lactate can be phosphorylated to form 2-phospho-l-lactate in a GTP-dependent manner (116); however, the enzyme responsible (to be named CofB) has remained elusive in the 15 years since the reaction was discovered. Finally, the 2-phospho-l-lactate is converted to LPPG by the GTP-dependent enzyme 2-phospho-l-lactate guanylyltransferase (CofC) (PDB ID 2I5E) (116, 131). Homologous enzymes are required for F₄₂₀ production in mycobacteria (132).

2.3.2. Distribution

F₄₂₀ has a more restricted taxonomic distribution than F_o and the ubiquitous redox cofactors FAD, FMN, and NAD(P). The cofactor has been identified in a single phylum each of bacteria and archaea using analytical chemistry methods. Among the archaea, F₄₂₀ is thought to be distributed in all methanogens, a group of strictly anaerobic methane-producing archaea (5, 15). In these organisms, F₄₂₀ serves as a central catabolic cofactor and is also central to two of the three main methanogenesis pathways. While present in low levels in some methanogens (e.g., Methanosarcinales), it is present at concentrations between 100 to 400 mg per kg in many hydrogenotrophs (15, 61, 62). F₄₂₀ has also been identified in several nonmethanogenic euryarchaeota, including three species of the sulfate-reducing genus Archaeoglobus (19, 133–135) and seven species of the photosynthetic genera Halobacteria and Halococcus (20, 136). The cofactor is also proposed to be central to the metabolism of the various lineages of the anaerobic methanotrophic archaea (ANME) (21, 137). Comparative genomics indicate that the genes required for F₄₂₀ biosynthesis are also distributed in the Thaumarchaeota, Aigarchaeota, Geoarchaeota, Bathyarchaeota, and Lokiarchaeota (138–142). The absorbance spectra of single cells of the ammonia- and cyanate-oxidizing thaumarchaeon Nitrososphaera gargensis are also consistent with the presence of F₄₂₀ (143, 144). It is unclear whether F₄₂₀ is produced by Crenarchaeota; while the cofactor was reported to be present at low levels in representatives of the Sulfolobus and Thermoplasma (20), the genomes of these organisms suggest that in fact they lack the capacity to synthesize this deazaflavin by any currently understood biosynthetic mechanism.

It is assumed that F₄₂₀ has a more restricted distribution among bacteria. The cofactor has been identified in representatives of the actinobacterial genera Mycobacterium (23, 27, 125, 145), Streptomyces (25, 27, 29, 146), Nocardia (27, 145), and Nocardioides (54). Most of these representatives are saprophytic soil bacteria that adopt a heterotrophic, aerobic lifestyle. The cofactor has also been reported in several mycobacterial pathogens, namely, the major obligate pathogens Mycobacterium tuberculosis, Mycobacterium bovis, and Mycobacterium leprae, as well as several opportunistic species (145). Comparative genomic analyses show that the genes involved in F₄₂₀ biosynthesis and utilization are also found in representatives of the Chloroflexi, Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria (30, 37), which constitute some of the most dominant taxa in aerated soil ecosystems (147). The occasional references to F₄₂₀-dependent processes in Cyanobacteria are erroneous; these have emerged from authors misattributing F_o-dependent processes to F₄₂₀ (72, 148) or relying on incorrect automated sequence predictions (149). Indeed, F₄₂₀ has yet to be chemically identified in any species outside the phyla Euryarchaeota and Actinobacteria.

2.3.3. Enzymology

In most archaea and some actinobacteria, F₄₂₀ is reduced through coupled steps in central catabolic pathways (Table 2). Methanogens are able to oxidize their substrates for growth using F₄₂₀, i.e., H₂ (via the F₄₂₀-reducing hydrogenase [Frh]) (150), formate (via the F₄₂₀-reducing formate dehydrogenase [Ffd]) (17), or secondary alcohols (via the F₄₂₀-reducing secondary alcohol dehydrogenase [Adf]) (49). This facilitates the entry of electrons into the CO₂-reducing pathway of methanogenesis and generates F₄₂₀H₂ to drive cellular redox reactions (151). Note that, contrary to historical reports (152, 153), carbon monoxide dehydrogenase, pyruvate dehydrogenase, and α-ketoglutarate dehydrogenase of methanogens are not F₄₂₀ dependent in methanogens (151, 154). Mycobacteria also reduce F₄₂₀ via their central catabolic pathways by using the F₄₂₀-reducing glucose-6-phosphate dehydrogenase (Fgd), one of two entry points to the reductive pentose phosphate pathway. However, pathways also exist to reduce F₄₂₀ using other redox cofactors depending on external and internal redox states, i.e., NADP (via the F₄₂₀-NADP oxidoreductase [Fno]) in many actinomycetes (12, 155) and tetrahydromethanopterin (via methylene tetrahydromethanopterin dehydrogenase [methylene-H₄MPT dehydrogenase {Mtd} and methylene-H₄MPT reductase {Mer}]) in methylotrophic methanogens (156, 157). As emphasized by the central placement of F₄₂₀ in the redox ladder of Table 1, many of these reactions are physiologically reversible. The physiology and biochemistry of the F₄₂₀-reducing dehydrogenases is discussed in detail in sections 3.2 and 4.2.

TABLE 2.

Activity, role, and distribution of F₄₂₀-dependent oxidoreductases^a

Oxidoreductase and domain	Physiological roleb	Taxonomic distributionc	Description	EC no.	PDB ID	Reference(s)
F420-reducing dehydrogenases
Archaea
Frh: F420-reducing hydrogenase	Hydrogenotrophic methanogenesis. Couples oxidation of H2 to reduction of F420. May be physiologically reversible.	All orders of methanogens	Section 3.2.1	1.12.98.1	4OMF, 4CI0, 3ZFS	11, 16, 150, 219, 224, 227
Ffd: F420-reducing formate dehydrogenase	Formatotrophic methanogenesis. Couples oxidation of formate to reduction of F420. May be part of the electron-bifurcating complex.	Methanobacteriales, Methanococcales, Methanopyrales, Methanomicrobiales, Methanocellales	Section 3.2.2	1.2.99.9		17, 185, 190, 242, 259
Adf: F420-reducing secondary alcohol dehydrogenase	Growth on secondary alcohols. Couples oxidation of secondary alcohols (e.g., isopropanol) to reduction of F420.	Methanomicrobiales, Methanocellales	Section 3.2.3	1.1.98.5	1RHC	49, 271, 272
Bacteria
Fno: F420-reducing NADPH dehydrogenase	Exchanges electrons between NADP and F420. F420 reduction important in bacteria, as F420 is the secondary cofactor.	Many Actinomycetes (e.g., Streptomyces, Rhodococcus, Nocardia, Nocardioides), Αlpha- or Betaproteobacteria?	Section 4.2.1	1.5.1.40		12, 155
Fgd: F420-reducing glucose-6-phosphate dehydrogenase	Heterotrophic growth. Couples oxidation of glucose-6-phosphate to reduction of F420 via the pentose phosphate pathway.	Many Actinomycetes (e.g., Mycobacterium, Actinoplanes, Microbacterium, Amycolatopsis), Chloroflexi?	Section 4.2.2	1.1.98.2	3B4Y	163
fHMAD: F420-reducing hydroxymycolic acid dehydrogenase	Cell wall biosynthesis. Catalyzes F420-dependent oxidation of hydroxymycolic acids to ketomycolic acids.	Few Mycobacterium (primarily pathogenic species)	Section 4.2.3			364, 365
F420H2-dependent reductases
Archaea
Mtd: F420-reducing methylene-H4MPT dehydrogenase	Reduces CHH4MPT to CH2=H4MPT with F420H2 in methanogenesis. Reaction physiologically reversible.	All orders of methanogens, Archaeoglobales, ANME	Section 3.3.1	1.5.98.1	1QV9, 1U6I, 3IQF, 3IQE	18, 47, 166, 275, 284, 285
Mer: F420H2-dependent methylene-H4MPT reductase	Reduces CH2=H4MPT to CH3-H4MPT with F420H2 in methanogenesis. Reaction physiologically reversible.	All orders of methanogens, Archaeoglobales, ANME, Halobacteriales	Section 3.3.1	1.5.98.2	1F07, 1EZW, 1Z69	48, 159, 166, 284
Fpo: F420H2-dependent methanophenazine reductase	Proton-translocating primary dehydrogenase in respiratory chain transferring electrons from F420H2 to heterodisulfide	Methanosarcinales	Section 3.3.2	1.1.98.4		162, 304–306, 327
Fqo: F420H2-dependent quinone reductase	Proton-translocating primary dehydrogenase in respiratory chain transferring electrons from F420H2 to sulfate	Archaeoglobales, ANME	Section 3.3.2	1.1.98.4		21, 198–200
Fpr: F420H2-dependent oxidase	Detoxifies O2 by mediating the four-electron reduction of O2 to H2O with F420H2	Methanobacteriales, Methanococcales, Methanomicrobiales, Methanocellales	Section 3.3.3	1.5.3.22	2OHH, 2OHI, 2OHJ	161, 192
Fsr: F420H2-dependent sulfite reductase	Detoxifies sulfite by mediating the six-electron reduction of sulfite to sulfide with F420H2. Also enables use of sulfite as an S source.	Methanobacteriales, Methanococcales	Section 3.3.4	1.8.98.3		51, 191
Fno: F420H2-dependent NADP reductase	Exchanges electrons between NADP and F420. NADP reduction important in archaea, as NADP is the secondary cofactor.	All orders of methanogens, Archaeoglobales, ANME	Section 3.3.5	1.5.1.40	1JAY, 1JAX	16, 22, 160, 201
Bacteria
Ddn: F420H2-dependent nitroreductases	May serve to detoxify redox cycling agents and other exogenous compounds. Also catalyze nitroimidazole activation.	Most Actinomycetes (e.g., Mycobacterium, Streptomyces, Rhodococcus), Chloroflexi?, Methanosarcinales?	Section 4.3.1		3H96, 4Y9I, 3R5R, 3R57	28, 30, 32, 164
Fbr: F420H2-dependent biliverdin reductases	Reduce the heme degradation product biliverdin to bilirubin. May also reduce mycobillins. FDOR-B3 and -B4 family.	Most Actinomycetes (e.g., Mycobacterium, Streptomyces, Rhodococcus), Chloroflexi?, Halobacteriales?	Section 4.3.1		2ASF, 4QVB, 1W9A	30, 165, 418, 419
Fts: F420H2-dependent tetracycline synthases	Reduce dehydrotetracyclines during streptomycete antibiotic synthesis. Role in mycobacteria unknown. FDOR-B1 family.	Most Actinomycetes (e.g., Mycobacterium, Streptomyces, Rhodococcus), Chloroflexi?, Halobacteriales?	Section 4.3.1		3F7E, 1RFE	28–30
Other F420H2-dependent flavin/deazaflavin oxidoreductases	Activities of A2-A4, B1, B2, B5, B6, AA1-AA5 families unknown. AA1s may be fatty acid saturases.	Most Actinomycetes (e.g., Mycobacterium, Streptomyces, Rhodococcus), Chloroflexi?, Halobacteriales?	Section 4.3.1		4ZKY	30, 55
Fht: F420H2-dependent picrate reductases	Reduces 2,4,6-trinitrophenol (picrate) for use as a C and N source through hydride transfer to the nitroaromatic ring	Few Actinomycetes (Rhodococcus, Nocardia, Nocardioides)	Section 4.3.2			54, 155
Fps: F420H2-dependent tetrahydropyrrole synthases	Reduces 4-propylidene-3,4-dihydropyrrole-2-carboxylate during biosynthesis of pyrrolobenzodiazepines antibiotics	Few Actinomycetes (Streptomyces, Streptosporangium)	Section 4.3.2			50, 389
Other F420H2-dependent luciferase-like hydride transferases	Unknown. Likely to have diverse roles in endogenous and exogenous redox metabolism of organic compounds.	Most Actinomycetes (e.g., Mycobacterium, Streptomyces, Rhodococcus)	Section 4.3.2

^aFor more information about the enzymes, see the sections in the text where the enzymes are described, Enzyme Commission (EC) entries, Protein Data Bank structures, and key primary references.

^bNote that several of F₄₂₀-dependent reactions are physiologically reversible, including those catalyzed by Fno, Mtd, Mer, and possibly Frh. Fno is primarily an F₄₂₀H₂-dependent NADP reductase in methanogens and a F₄₂₀-reducing NADPH dehydrogenase in bacteria; the enzyme appears to be similar in archaea and bacteria but is used in a different physiological context.

^cEuryarchaeota are listed by order, namely, six methanogenic orders (Methanobacteriales, Methanococcales, Methanopyrales, Methanomicrobiales, Methanocellales, and Methanosarcinales) and two nonmethanogenic orders (Archaeoglobales and Halobacteriales). The various lineages of the uncultured anaerobic methanotrophic archaea are denoted as ANME. Actinobacteria are listed by genus (Mycobacterium, Streptomyces, Rhodococcus, Nocardia, Nocardioides, Streptosporangium, Microbacterium, Actinoplanes, and Amycolatopsis).

The physiological roles of F₄₂₀ are primarily elicited by the coupling of the oxidation of F₄₂₀H₂ to the reduction of other compounds (Table 2). In methanogens, F₄₂₀H₂ oxidation sustains a wide range of processes. F₄₂₀H₂ is used to reduce one-carbon units bound to tetrahydromethanopterin, the central one-carbon carrier in methanogenesis pathways (158), and NADP, the central cofactor for anabolic processes (16). This depends on the aforementioned reactions catalyzed by Mtd (47), Mer (159), and Fno (160). The cofactor can additionally be used to detoxify O₂ (via F₄₂₀H₂-dependent oxidase [Fpr]) (161), mobilize sulfite (via F₄₂₀H₂-dependent sulfite reductase [Fsr]) (51), and in methanogens with cytochromes, reduce methanophenazine for respiratory energy conservation (via F₄₂₀H₂-dependent methanophenazine reductase [Fpo]) (162). F₄₂₀H₂ can be used to reduce diverse organic compounds in actinomycetes, including endogenous metabolites (e.g., quinones, porphyrins, fatty acids) (30, 32) and exogenous compounds (e.g., tetracyclines, picrate, aflatoxins) (28, 29, 54). These activities depend on two diverse superfamilies distinguished by their split β-barrel (flavin/deazaflavin oxidoreductases [FDORs]) (30, 37) or TIM barrel (luciferase-like hydride transferases [LLHTs]) protein folds (Fig. 6) (37). The F₄₂₀H₂-dependent reductase enzymes are discussed in more detail in sections 3.3 and 4.3.

FIG 6.

Structures of F₄₂₀-binding protein domains. (a) TIM barrel fold of Fgd (PDB ID 3B4Y [163]), (b) structure of Frh subunit B (PDB IDs 4OMF [150] and 3ZFS [167]), (c) split β-barrel fold of Ddn (PDB ID 3R5R [164]), (d) novel protein fold of Mtd (PDB ID 3IQE [166]), (e) Rossmann fold of Fno (PDB ID 1JAY [160]) and (f) the interface between β-lactamase and flavodoxin folds in Fpr (PDB ID 2OHJ [161]). Where available, the F₄₂₀ molecule is shown in green, and key residues at the F₄₂₀-binding site are highlighted in cyan. In panels b, d, and f, the positions of the FAD (orange), NADP (purple), and FMN (yellow) molecules required for F₄₂₀ binding are also shown.

The majority of F₄₂₀- and F₄₂₀H₂-binding proteins bind the cofactor within either TIM barrel (Adf, Mer, F₄₂₀-reducing hydroxymycolic acid dehydrogenase [fHMAD], Fgd, and other LLHTs) (48, 49, 163), FrhB-like (Frh, Fpo, Ffd, and Fsr) (150), or split β-barrel (FDORs) (28, 30, 164, 165) folds (Fig. 6). Exceptions to this are the structures of Mtd (novel Mtd-like fold) (166), Fno (Rossmann fold) (160), and Fpr (interface of β-lactamase and flavodoxin folds) (52). Of these known F₄₂₀ binding architectures, the F₄₂₀-binding TIM barrel and split β-barrel proteins share structural homology with related FMN- and FAD-binding proteins (30, 48). In contrast, the Mtd-like and FrhB-like folds have been found only in F₄₂₀- or F₄₂₀H₂-dependent proteins (150, 166). All of the proteins carry out hydride transfer on the Si-face of F₄₂₀ (48, 49, 150, 160, 163, 166, 167), with the exception of the FDORs that catalyze the reaction on the Re-face (28, 165). These proteins are adapted for F₄₂₀ binding by the presence of a positively charged channel or region that associates with the phospholactate and polyglutamate chain. In FDORs, LLHTs, Mtd, and Fno, hydrogen bonding interactions at the pyrimidine and hydroxyl of the deazaflavin moiety anchor the cofactor, along with hydrophobic interactions to the Re-face (Si-face for FDORs) that is not involved in the enzyme reaction (28, 48, 49, 150, 160, 161, 163–166). In FrhB, Fno, and Fpr, stability is also provided by aromatic interactions with the enzyme-bound FAD, NADP, or FMN (52, 150, 160).

3. F₄₂₀ IN METHANOGENS AND OTHER ARCHAEA

3.1. Physiological Roles

3.1.1. Methanogens

F₄₂₀ is a catabolic redox cofactor in both methanogenic and nonmethanogenic archaea. Methanogens are microorganisms that produce methane as the end product of their anaerobic pathways of energy generation (168). These organisms encompass at least six phylogenetically distinct, metabolically diverse orders of the archaeal phylum Euryarchaeota: Methanobacteriales, Methanococcales, Methanopyrales, Methanomicrobiales, Methanocellales, and Methanosarcinales (169–173). F₄₂₀ is synthesized in all of these orders, where it serves as a redox cofactor in both methanogenesis pathways and wider cellular processes (5, 15). In fact, the characteristic fluorescence of many methanogens is due to the presence of this cofactor (5, 59, 60).

Methanogens can generate methane through three major routes, the CO₂-reducing, methylotrophic, and aceticlastic pathways (174–177) (Fig. 7). In the CO₂-reducing pathway, CO₂ is progressively reduced to methane using exogenously derived electrons (151, 168, 178). This pathway sustains hydrogenotrophic growth using H₂-derived electrons (16), formatotrophic growth using formate-derived electrons (17), and in some organisms, growth on secondary alcohols (179). In the methylotrophic pathway, the methyl groups of methanol, methylated amines, and methylated sulfides are converted into CH₄ (reductive route) and CO₂ (oxidative route), with the oxidative reactions occurring through a reverse arm of the CO₂-reducing pathway (157, 175, 180). In the aceticlastic pathway, acetate is fermented to methane (through reduction of the methyl group) and CO₂ (through oxidation of the carboxy group) (175, 180, 181). Most methanogens are capable of hydrogenotrophic growth, with cytochrome-containing methanogens (i.e., the Methanosarcinales) primarily respiring H₂ and the other five orders conserving energy through electron-bifurcating pathways (182, 183). Formatotrophic growth is also widespread (17, 184, 185), but it does not occur in the Methanosarcinales (186). In contrast, only a few taxa are capable of methylotrophic growth (the family Methanosarcinaceae and genus Methanosphaera) (176, 187) and aceticlastic growth (the families Methanosarcinaceae and Methanosaetaceae) (176, 188). These pathways are nevertheless quantitatively important, with the aceticlastic pathway responsible for up to two-thirds of global net methane production. The biochemistry, physiology, and ecology of methanogenesis will be discussed further only in the context of F₄₂₀ metabolism; readers requiring further background on this topic are referred to several excellent reviews (151, 154, 168, 178, 182, 189).

FIG 7.

CO₂-reducing (green), methylotrophic (pink), and aceticlastic (blue) pathways of methanogenesis. The routes for energy generation from H₂/CO₂, formate, secondary alcohols, methanol, and acetate are shown. Processes common to all pathways are shown in black, and dashed arrows in gray show alternative pathways. F₄₂₀-dependent oxidoreductases are highlighted in red and catalyze both forward and reverse reactions, except for FpoF which is known to catalyze only F₄₂₀H₂ reoxidation. Abbreviations: Fd_red/ox, reduced/oxidized ferredoxin; MF, methanofuran; H₄MPT, tetrahydromethanopterin; H₄SPT, tetrahydrosarcinapterin; H-SCoM, 2-mercaptoethanesulfonate (reduced coenzyme M); CoBS-H, N-7-mercaptoheptanoylthreonine phosphate (reduced coenzyme B); MPh/MPhH₂, reduced/oxidized methanophenazine.

F₄₂₀ is central to the CO₂-reducing and methylotrophic pathways of methanogenesis. Dedicated F₄₂₀-dependent hydrogenases/dehydrogenases oxidize H₂ (Frh) (17, 150), formate (Ffd) (17, 190), and secondary alcohols (Adf) (49, 179) for entry into the CO₂-reducing pathway. F₄₂₀ also serves as the redox cofactor for the Mtd and Mer reactions, which mediate the fourth and fifth steps of the CO₂-reducing pathway, reducing methenyl-tetrahydromethanopterin (methenyl-H₄MPT) to methyl-H₄MPT with F₄₂₀H₂ (47, 159). They operate in the reverse direction in the methylotrophic pathway, oxidizing methyl-H₄MPT to methenyl-H₄MPT. However, F₄₂₀ is not involved in the aceticlastic pathway, which depends on a largely distinct set of enzymes (175, 181). In addition to mediating methanogenesis, dedicated F₄₂₀-dependent enzymes mediate a wide array of other cellular reactions in methanogens, including reduction of NADP for biosynthetic pathways (Fno) (22), mobilization of sulfite as a sulfur source (Fsr) (51, 191), and detoxification of atmospheric O₂ (Fpr) (161, 192). Methanogens with cytochromes can use F₄₂₀H₂ generated through the methylotrophic pathway as an input to the respiratory chain using the proton-translocating F₄₂₀H₂-reducing methanophenazine reductase (Fpo) (162, 193). Interestingly, F₄₂₀ is still present in acetate-grown Methanosarcina (194) and the obligately aceticlastic genus Methanosaeta (195, 196), reinforcing the idea that the cofactor has been selected for roles well beyond methanogenesis. On the basis of metagenomic studies, it was recently reported that members of the newly defined phylum Bathyarchaeota may also be F₄₂₀-dependent methylotrophic methanogens (141).

3.1.2. Sulfate-reducing archaea

F₄₂₀ is also known to be synthesized by two orders of nonmethanogenic archaea, the Archaeoglobales and Halobacteriales (20, 136). Archaeoglobi are primarily heterotrophic, sulfate-reducing thermophiles that inhabit deep-sea vents (19), whereas Halobacteria are primarily phototrophic, facultatively aerobic halophiles that dominate hypersaline waters (197). While the two orders have very different metabolisms, both to methanogens and to each other, they are closely phylogenetically related to the Methanomicrobiales, Methanosarcinales, and Methanocellales (169, 171, 172). It is likely that F₄₂₀ was synthesized in the common ancestor of each of these five orders prior to their metabolic divergence. While little is known about the role of F₄₂₀ in Halobacteria (20, 136), a range of biochemical studies indicate that F₄₂₀H₂ is a central catabolic electron donor in Archaeoglobus fulgidus (133). F₄₂₀H₂ donates electrons to the sulfate-reducing respiratory chain via the proton-translocating F₄₂₀H₂-dependent quinonereductase (Fqo) (198–200). Additionally, the F₄₂₀H₂-dependent NADP reductase (Fno) is proposed to generate NADPH for various biosynthetic pathways (160, 201). F₄₂₀ appears to be reduced through distinct routes depending on whether the growth substrate is H₂/CO₂ or lactate. It is well-established that, during the anaerobic oxidation of lactate to CO₂, F₄₂₀ can be reduced by Mtd and Mer (133, 200, 202). Given that the organism lacks Frh, it remains to be resolved how A. fulgidus generates F₄₂₀H₂ during hydrogenotrophic growth (203); possible routes include electron transfer from reduced ferredoxin (Fd_red) (via a hypothetical complex), quinols (via reverse electron transfer), or NADPH (via Npo) (135, 200).

3.1.3. Methanotrophic archaea

There is strong evidence that F₄₂₀ is also central to the metabolism of anaerobic methanotrophic archaea (ANME). In contrast to methanogens, these archaea consume, rather than produce, methane and use the electrons liberated from methane to drive sulfate- and nitrate-reducing respiratory chains (204–207). While these organisms have yet to be successfully cultured, they are of enormous ecological and geochemical significance; it is predicted that 90% of the methane produced by methanogens in marine sediments is immediately recycled by ANME (189, 208, 209). Extensive studies of microbial ecology have demonstrated that these organisms are closely related to two orders of methanogens (Methanosarcinales and Methanomicrobiales), and form at least three major phylogenetic clades (ANME-1, ANME-2, and ANME-3) (210, 211). A range of genomic and biochemical evidence suggests that these archaea predominantly grow through a reverse methanogenesis pathway (similar to the methylotrophic pathway; Fig. 7), through which F₄₂₀H₂ is generated via the Mer and Mtd reactions (137, 212–215). The F₄₂₀H₂ that is produced from this pathway is proposed to be reoxidized by the proton-translocating Fqo complex, with sulfate or nitrate serving as the terminal respiratory electron acceptor (21, 215). This proposal was recently supported by a metagenomic/metatranscriptomic study that showed that the nitrate-reducing methanotroph Methanoperedens nitroreducens (part of the ANME-2 lineage) expresses a complete reverse methanogenesis pathway, along with all the F₄₂₀ biosynthesis genes and a putative Fqo complex (137). Environmental sequencing has also inferred a role for F₄₂₀ in other ANME lineages (21, 215, 216). Also consistent with the presence of F₄₂₀, ANME, like methanogens, are autofluorescent under UV light (217, 218).

3.2. F₄₂₀-Reducing Dehydrogenases

3.2.1. Frh: F₄₂₀-reducing hydrogenase

The F₄₂₀-reducing hydrogenase directly couples H₂ to F₄₂₀ reduction (9, 11, 219, 220). The enzyme is encoded by genes in all classes of methanogens (183) and is the preferred route to F₄₂₀ reduction during hydrogenotrophic methanogenesis (Fig. 7) (221–223). This hydrogenase is essential for growth on H₂/CO₂ in Methanosarcina barkeri (Ms. barkeri) (224), but it appears to be dispensable in methanogens with genes that encode an alternative pathway for F₄₂₀ reduction such as Methanococcus maripaludis (Mc. maripaludis) (225). The enzyme complex, encoded by the transcriptional subunit frhADGB (222), is a product of the association of an F₄₂₀ reductase subunit of the F₄₂₀-binding protein family (functionally analogous to F₄₂₀ reductase domains of Fsr, Fpo, and Ffd) with a H₂-oxidizing [NiFe]-hydrogenase of the group 3a family (226). Structural characterization of this complex from Methanothermobacter marburgensis through cryo-electron microscopy (167, 227) and X-ray crystallography (150) revealed a large dodecameric complex of heterotrimers (FrhABG), arranged as a shell with a solvent-filled core (Fig. 8). Each heterotrimeric protomer (FrhABG) contains a [NiFe]-hydrogenase large subunit (FrhA; matured by the endopeptidase FrhD), a [NiFe]-hydrogenase small subunit (FrhG), and an F₄₂₀ reductase subunit (FrhB). While the complex is located in the cytoplasm (150), it is often purified from the membrane fraction due to its high molecular mass of 1.2 MDa (228–230).

FIG 8.

(a) Structure of the dodecameric complex of Frh (PDB ID 4OMF [150]), where a single protomer (identified with darker shades) contains three subunits: FrhA (green), FrhB (blue), and FrhG (pink). (b) Electron transfer route from H₂ to F₄₂₀ within the Frh subunits during hydrogenotrophic methanogenesis.

During the H₂-dependent reduction of F₄₂₀, H₂ binding and oxidation occur at the buried [NiFe] center of FrhA, which is facilitated by a hydrophobic channel that extends from the [NiFe] center to the outer surface of the enzyme complex (150). On the basis of structural and spectroscopic studies, it is proposed that H₂ is heterolytically cleaved in a mechanism similar to other [NiFe]-hydrogenases (150, 219, 231–233). As with other [NiFe]-hydrogenases, the protons generated are relayed from the [NiFe] center to the outer surface of the complex, where they are released to the bulk solvent near a covalently bound FAD molecule on the FrhB subunit of a neighboring protomer (150). Concomitantly, electrons from the H₂ cleavage reaction are individually transferred via four [4Fe4S] clusters (three on FrhG and one on FrhB) to the FAD molecule bound to FrhB of the same protomer, generating FADH₂ (Fig. 8). The terminal step involves hydride transfer from FADH₂ to F₄₂₀, which binds reversibly at a solvent-accessible pocket on FrhB, with the 5-deazaflavin rings (Si-face) next to the isoalloxazine of the FAD cofactor (Si-face) (150, 234). Kinetic and structural data suggest that hydride transfer to F₄₂₀ occurs rapidly and is rate limited by diffusion, rather than conformational change (227, 235). The remarkable oligomerization of the complex does not appear to influence the reaction kinetics of the hydrogenase and instead may serve to protect metal centers from redox-active compounds of the cytosol (150).

It has been proposed that Frh is physiologically active in both the forward and reverse directions (224, 225). While Frh primarily sustains H₂-mediated F₄₂₀ reduction during hydrogenotrophic growth, it may mediate F₄₂₀H₂-mediated H₂ production during methylotrophic methanogenesis and formate-dependent growth (224, 225). This is consistent with the observations of severe defects of Ms. barkeri Δfrh mutants during growth on methanol and on H₂ production during formate-dependent growth of Mc. maripaludis (225). While F₄₂₀ reduction is more thermodynamically favorable (E₀ F₄₂₀ = −340 mV; E₀ H₂ = −410 mV), the reverse reaction may occur when F₄₂₀H₂ accumulates and H₂ partial pressure [pH₂] is low. This is supported by biochemical data that Frh purified from Methanobacterium formicum can sustain a moderate rate of F₄₂₀H₂-mediated H₂ evolution (230). However, genetic dissection experiments will be required to definitively confirm whether Frh-mediated H₂ evolution can occur in vivo at physiologically relevant rates.

Several variants of Frh can be encoded by genes in the same genome. Many methanogens carry genes that encode both a selenium-containing F₄₂₀-reducing hydrogenase (Fru) in addition to a selenium-free one (Frh) (183, 221, 222). Studies on the purified [NiFeSe]-hydrogenase from Methanococcus voltae suggest that the selenium-containing isozymes are faster acting and more oxygen tolerant than the selenium-free variant (236, 237). Hence, transcription of Fru over Frh occurs in selenium-containing conditions in this organism (221, 238, 239). In addition, variants of Frh were recently found to be encoded by genes of several non-F₄₂₀-producing species of the archaeal order Thermococci and the bacterial family Desulfobacteriaceae (183). Biochemical and sequence analyses indicate that these enzymes cannot reduce F₄₂₀ and instead couple to another electron acceptor, such as a flavin (240); these enzymes and their F₄₂₀-reducing relatives are capable of reducing FAD and FMN in vitro (16, 240).

3.2.2. Ffd: F₄₂₀-reducing formate dehydrogenase

Many hydrogenotrophic methanogens can also grow using formate as the sole electron donor, including species from the genera Methanococcus (241, 242), Methanobacterium (243), and Methanospirillum (184). This process is especially ecologically significant, given that formate produced by fermentative bacteria can be consumed by methanogens through interspecies transfer (244). It is well established that formatotrophic growth is linked to F₄₂₀ metabolism (17) and that it depends on F₄₂₀-reducing formate dehydrogenases (called Ffd or Fdh) (242). Although Ffd has not been structurally characterized, biochemical studies on the enzyme from Methanobacterium formicicum (Mb. formicicum) have revealed its core architecture. Ffd is a membrane-bound heterodimeric enzyme containing several redox centers (190, 245, 246). The large subunit is homologous with the structurally characterized bacterial formate dehydrogenases (247), and it is predicted to contain a molybdopterin guanine nucleotide cofactor (MGD) (248–252) and a [4Fe4S] center (190). The small subunit is unique to methanogenic archaea and is predicted to contain two [4Fe4S] clusters (190), an FAD cofactor (190, 253, 254), and an F₄₂₀-binding site that is homologous to FrhB (150). It has been proposed that formate is oxidized at the molybdopterin center and that electrons are shuttled via the FeS clusters to the electron gate FAD and finally to F₄₂₀ (254) (Fig. 9). Like most other F₄₂₀-dependent enzymes (255), hydride transfer to C-5 of F₄₂₀ is Si-face stereospecific (254).

FIG 9.

Proposed architecture of Ffd and electron transfer route from formate to F₄₂₀.

Two pathways that facilitate formate-dependent methanogenesis have been elucidated (Fig. 7). In the first pathway, it has been proposed that electrons derived from formate are funneled through the hydrogenotrophic pathway, with F₄₂₀H₂ and H₂ serving as intermediates (225, 256). First, formate is disproportionated through the combined activity of Ffd (formate + F₄₂₀ → CO₂ + F₄₂₀H₂) and Frh (F₄₂₀H₂ → F₄₂₀ + H₂) (257). Subsequently, the H₂ and CO₂ produced are converted to methane through the hydrogenotrophic pathway (225). More recently, it was proposed that Ffd can form an electron-bifurcating complex with heterodisulfide reductase; in this model, the oxidation of formate simultaneously drives the exergonic reduction of heterodisulfide and endergonic reduction of ferredoxin (258, 259). This pathway is supported through analysis of protein-protein interactions, which indicate that Ffd forms a membrane-bound supercomplex with a heterodisulfide reductase (Hdr) and a hydrogenase subunit (VhuD) (259, 260). Genetic dissection studies likewise show that Ffd but not Frh is essential for formatotrophic growth of Mc. maripaludis (259, 261–263). In fact, a suppressor mutant of Mc. maripaludis sustains formatotrophic growth when all of its seven hydrogenases are deleted (261). Costa et al. proposed that, in addition to providing electrons to Hdr, Ffd must also provide F₄₂₀H₂ to sustain the central reactions catalyzed by Mer and Mtd in the methanogenesis pathway (259).

As with Frh, methanogens have evolved selenium-free and selenium-containing variants of the Ffd. Whereas Mb. formicicum carries a gene that encodes a single Ffd, Methanococcus vannielii carries genes that encode both selenium-free and selenium-containing variants of the Ffd (185, 264). Selenium supplementation markedly stimulates formate-driven growth of the organism, suggesting that the selenocysteine-containing Ffd may be the more efficient variant (265). In contrast, both Ffd variants in Mc. maripaludis are selenoproteins (266); hence, the organism requires the presence of selenium to grow on formate (267, 268). Genetic dissection has demonstrated that each homolog confers a competitive growth advantage, with single mutants impaired and double mutants unviable for formatotrophic growth (242). Interestingly, while some Methanosarcina species carry genes that encode Ffd homologs (269), methanogens with cytochromes cannot sustain formate-dependent growth. Thauer et al. rationalize that the high H₂ threshold of these organisms compared to other methanogens means that they would not be able to competitively oxidize H₂ produced from formate metabolism (182). An alternative explanation is that they lack the electron-bifurcating systems required to efficiently couple formate oxidation to growth (259).

3.2.3. Adf: F₄₂₀-reducing secondary alcohol dehydrogenase

Some methanogens are capable of low-yield growth using alcohols as electron donors. Whereas methanogens oxidize primary alcohols (e.g., ethanol) using standard NADP-reducing alcohol dehydrogenases (Adh) (22, 270), some can also metabolize secondary and cyclic alcohols using a phylogenetically unrelated class of F₄₂₀-dependent secondary alcohol dehydrogenases (Adf) (271, 272). The enzymes that mediate this are sparsely distributed, encoded by genes on just six sequenced methanogens in the NCBI database, all of the class Methanomicrobia. The F₄₂₀H₂ generated from the reduction of secondary alcohols (e.g., isopropanol, butan-2-ol) to ketones (e.g., acetone, butanone) is, in turn, used to sustain the CO₂-reducing pathway of methanogenesis and other cellular reductive processes (271, 272). Adf belongs to the bacterial luciferase superfamily (TIM barrel protein fold), which also includes other F₄₂₀-dependent enzymes Fgd (163), Mer (48), and Fht (155). As with other enzymes of the luciferase superfamily, crystallographic analysis shows that Adf from Methanoculleus thermophilicus is dimeric, containing a nonprolyl cis peptide bond toward the Re-face of F₄₂₀ that keeps the 5-deazaflavin rings in a bent “butterfly” conformation (49). The structure contains the inactive F₄₂₀-acetone adduct (Fig. 10) (thought to form due to acetone accumulation in the presence of oxidized F₄₂₀ in a reductive environment); small secondary alcohol substrates, such as isopropanol, bind in the same pocket in the active enzyme (49). Hydride transfer occurs on the Si-face of the cofactor, facilitated by the abstraction of a proton from the alcohol by a catalytic histidine residue and the stabilization of the alcoholate anion transition state by nearby tryptophan and glutamate residues (49, 273).

FIG 10.

Structure of the active site of Adf. (a) Cartoon representation of the protein (PDB ID 1RHC [49]) showing the bound F₄₂₀-acetone adduct. (b) Proposed mechanism of isopropanol reduction to acetone (49). R₁ is the ribitylphospholactyl-oligoglutamate chain of F₄₂₀.

3.3. F₄₂₀H₂-Dependent Reductases

3.3.1. Mtd: F₄₂₀-reducing methylene-H₄MPT dehydrogenase/Mer: F₄₂₀H₂-dependent methylene-H₄MPT reductase

In all methanogenesis pathways, tetrahydromethanopterin (H₄MPT) serves as the carrier of one-carbon (1C) units (158, 274). 1C units can be conjugated to H₄MPT in various oxidation states, including formyl (CHO-H₄MPT), methenyl (CHH₄MPT), methylene (CH₂=H₄MPT), and methyl (CH₃-H₄MPT). In hydrogenotrophic and formatotrophic methanogenesis, CO₂ is activated through three F₄₂₀-independent initial steps (Fig. 7). The resultant methenyl-H₄MPT adduct is reduced to methylene-H₄MPT and methyl-H₄MPT via two successive F₄₂₀-dependent steps. The first is catalyzed by the F₄₂₀-reducing methylene-H₄MPT dehydrogenase (Mtd; CHH₄MPT⁺ + F₄₂₀H₂ → CH₂=H₄MPT + F₄₂₀ + H⁺) (18, 275–278). The second is catalyzed by the F₄₂₀H₂-dependent methylene-H₄MPT reductase (Mer; CH₂=H₄MPT + F₄₂₀H₂ → CH₃-H₄MPT + F₄₂₀) (279–284). Reflecting the standard redox potentials of F₄₂₀, methylene-H₄MPT, and methenyl-H₄MPT (Table 1), these reactions are physiologically reversible. Hence, Mer and Mtd can also be used to oxidize CH₃-H₄MPT to CHH₄MPT⁺ with the concomitant reduction of two mole equivalents of F₄₂₀ (156, 157). This is particularly important in the oxidative arm of the methylotrophic methanogenesis pathway, which generates reducing agents (F₄₂₀H₂, Fd_red) through the oxidation of CH₃-S-CoM (coenzyme M) to CO₂ (Fig. 7) (157).

A succession of crystal structures of Mtd and Mer have revealed much about their architectures and mechanisms. The structure of Mtd from Methanopyrus kandleri revealed a unique protein fold compared to other F₄₂₀-binding proteins (47, 166, 285, 286). Whereas most F₄₂₀-binding proteins adopt bacterial luciferase-like (TIM barrel) (163), FDOR-like (split β-barrel) (30), or FdrB-like (novel αβ fold) (150) protein folds, Mtd folds into a unique tertiary structure (47, 166) (Fig. 6). Each protein chain of the homohexameric complex of Mtd (a trimer of dimers) contains an αβ domain, a smaller helical bundle domain, and a C-terminal sheet segment (47). Methenyl-H₄MPT and F₄₂₀H₂ bind opposite each other at the active site, which is located between the two domains and capped by the loop segment of the adjacent chain (Fig. 11) (166). The reaction is catalyzed through a ternary complex mechanism (276, 284), wherein hydride transfer occurs between C-14a of methylene-H₄MPT (Re-face stereospecific) and C-5 of F₄₂₀H₂ (Si-face stereospecific) (166, 287–289). Crystal structures of Mer homologs have been solved from three organisms, Methanoplanus kandleri (159), Methanothermobacter marburgensis (159), and Methanosarcina barkeri (48). As a member of the bacterial luciferase superfamily, Mer contains a characteristic TIM barrel fold and a nonprolyl cis-peptide bond close to the F₄₂₀-binding site (48, 159). Modeling studies indicate that methylene-H₄MPT and F₄₂₀H₂ are likely to bind opposite each other to form a ternary complex like in Mtd (48), enabling direct hydride transfer in a stereospecific manner (289) (Fig. 11).

FIG 11.

Structure and mechanism of F₄₂₀H₂-dependent hydride transfers to one-carbon compounds conjugated to tetrahydromethanopterin. (a) Structure of Mtd (PDB ID 3IQE [166]) as a ternary complex with F₄₂₀ (green) and methenyl-H₄MPT⁺ (pink). The large αβ-domain of a single subunit is shown in purple, and the helical bundle domain is shown in cyan. The secondary subunit in the dimer is shown in gray. (b) Mechanism of hydride transfer between F₄₂₀H₂ (Si-face) and methenyl-H₄MPT⁺ (Re-face) leading to methylene-H₄MPT production (166). (c) Structure of Mer (PDB ID 1Z69 [48]) as a ternary complex with F₄₂₀ (green) and polyethylene glycol (blue) occupying the methylene-H₄MPT-binding site. (d) Inferred mechanism of hydride transfer between F₄₂₀H₂ (Si-face) and methylene-H₄MPT (Re-face) leading to methyl-H₄MPT production (166).

In four of the methanogenic orders, the fourth step in the CO₂ reduction pathway can be effected using H₂ instead of F₄₂₀ (Fig. 7) (183). The methylene-H₄MPT hydrogenase (Mth; also known as the [Fe]-hydrogenase, the H₂-forming methylenetetrahydromethanopterin dehydrogenase, and Hmd) directly reduces methenyl-H₄MPT to methylene-H₄MPT using H₂ (CHH₄MPT⁺ + H₂ → CH₂=H₄MPT + H⁺) (290–292). Several transcriptome analyses have indicated that, while the F₄₂₀-dependent route is constitutive, the H₂-dependent route predominates at high H₂ partial pressures (pH₂) that induce rapid growth (293–295). Consistently, Mtd mutants of Methanobacter thermoautotrophicus are unable to grow at low pH₂ (296). Methanogens can also reduce F₄₂₀ using H₂ through the combined action of Mth (CHH₄MPT⁺ + H₂ → CH₂=H₄MPT + H⁺) and Mtd (CH₂=H₄MPT + F₄₂₀ + H⁺ → CHH₄MPT⁺ + F₄₂₀H₂) (the net reaction is H₂ + F₄₂₀ → F₄₂₀H₂) (225, 297). Hendrickson and Leigh demonstrated through genetic dissection in Mc. maripaludis that this Mth-Mtd cycle can fully compensate for Frh during hydrogenotrophic growth; the pathways could be eliminated separately, but not together (225). Transcriptional and biochemical studies on Methanothermobacter marburgensis (Mt. marburgensis) have suggested that the Mth-Mtd cycle is particularly important during nickel-limiting conditions when the F₄₂₀-reducing [NiFe]-hydrogenase cannot be synthesized (297, 298).

Homologs of Mtd and Mer are also present in sulfate-reducing archaea (299, 300). Archaeoglobus fulgidus converts lactate to three molecules of carbon dioxide using an Mtd/Mer-facilitated 1C pathway similar to methylotrophic methanogenesis (133, 300). The F₄₂₀H₂ produced by Mtd and Mer can be subsequently respired through a sulfate-reducing electron transport chain (200). It has also been proposed that these enzymes operate during the reverse methanogenesis pathway of anaerobic methanotrophic archaea (ANME). In support of this, genes encoding homologs of Mtd and Mer have been found in some reconstructed ANME metagenomes (21, 137, 215). Heterologously expressed Mtd from an ANME-1 archaeon catalyzed the same reaction as Mtd from methanogens, with similar catalytic specificity and cofactor dependence (214). In addition to F₄₂₀-dependent enzymes, NAD(P)-dependent methylenetetrahydromethanopterin dehydrogenases have been characterized that have central roles in the formaldehyde assimilation pathways of aerobic methylotrophic bacteria (301, 302).

3.3.2. Fpo: F₄₂₀H₂-dependent methanophenazine reductase/Fqo: F₄₂₀H₂-dependent quinone reductase

The single order of methanogens containing cytochromes, i.e., the Methanosarcinales, can translocate protons by coupling the oxidation of F₄₂₀H₂ to the reduction of heterodisulfide (CoM-S-S-CoB). It was initially thought that this activity was mediated by a single hypothetical enzyme complex, the F₄₂₀H₂:heterodisulfide oxidoreductase (303). However, it is now appreciated that this system is in fact formed from two respiratory complexes (304–306), the F₄₂₀H₂-dependent methanophenazine reductase (Fpo) (162) and the methanophenazine-dependent heterodisulfide reductase (Hdr) (307), which are linked by the redox-active membrane-diffusible cofactor methanophenazine (305, 308–310) (Fig. 10). Constituting the primary dehydrogenase, Fpo is a respiratory proton pump exclusive to the order Methanosarcinales (162). Serving as the terminal reductase, Hdr is anchored to the membrane by a b-type cytochrome (307, 311, 312). Together, these enzymes translocate four protons (two each through Fpo and Hdr) per molecule of F₄₂₀H₂ that is oxidized (303). In contrast, the Hdr-linked complexes of methanogens without cytochromes are primarily cytosolic and do not serve a respiratory role (182).

The complete Fpo complex has been purified from only a single species, Methanosarcina mazei (Ms. mazei) (162, 313–315). The complex is very similar to bacterial NADH:ubiquinone oxidoreductase I (Nuo; also known as complex I) in both overall subunit composition and amino acid sequence (316, 317). The Fpo complex is formed of 13 subunits that associate into a hydrophilic portion (FpoFBCDIO) and a transmembrane portion (FpoAHJKNML) (162, 318). The hydrophilic electron input (FpoF) and electron output (FpoBCDI) modules catalyze electron transfer from F₄₂₀H₂ to methanophenazine and are largely conserved with Nuo. However, there are several key differences: an F₄₂₀H₂-oxidizing subunit (FpoF) replaces the NADH-oxidizing module (NuoEFG), the phenazine-reducing subunit (FpoD) has a modified binding pocket compared to its quinone-reducing equivalent (NuoD), and a subunit of unknown function (FpoO) is present. The remaining hydrophobic portion of Fpo is embedded in the membrane, consisting of the proton-translocating E-channels (FpoAJKH) and Mrp antiporter-like channels (FpoNML) that are homologous to those in Nuo (316, 317, 319). Unlike Nuo, which pumps four protons per two input electrons, the Fpo complex is thought to translocate two protons per molecule of F₄₂₀H₂ (162). On the basis of the structure of bacterial Nuo (319, 320), a basic model for the mechanism of Fpo has be proposed (Fig. 12): electrons are transferred from F₄₂₀H₂ to methanophenazine, methanophenazine reduction propagates conformational changes to the E-channel and in turn the antiporter module, and two protons are subsequently translocated through half-channels via conserved lysine and glutamate residues.

FIG 12.

Model of respiration in Methanosarcina mazei using F₄₂₀H₂ as an electron donor and heterodisulfide as an electron acceptor. In this system, the primary dehydrogenase is the proton-translocating F₄₂₀H₂-dependent methanophenazine reductase (Fpo) and the terminal reductase is methanophenazine-dependent heterodisulfide reductase (Hdr). Arrangement of Fpo subunits and the proposed electron and proton transfer pathways are inferred from the homology of the system to bacterial complex I (Nuo) (317, 319, 320). Gray lines show the propagation of conformational change in the E-channel (FpoAJKH) and antiporter (FpoNML) modules upon electron transfer to methanophenazine (MPh/MPhH₂), and dashed arrows show possible routes for proton transfer based on structural analysis of complex I. The protein topology of Hdr is not shown in detail.

During methylotrophic methanogenesis, it is proposed that the F₄₂₀H₂ formed serves as the major respiratory electron donor (Fig. 7). In this pathway, one-carbon compounds (e.g., methanol, methylamine) are activated to produce methyl-coenzyme M (methyl-S-CoM) and thereafter converted to CO₂ or methane; the oxidative branch yields F₄₂₀H₂ via the Mer and Mtd reactions, while the reductive branch generates proton motive force by coupling F₄₂₀H₂ oxidation to heterodisulfide reduction (318, 321). Consistently, trimethylamine-cultured Δfpo mutants of Ms. mazei are severely compromised in growth and methane formation compared to the wild-type strain (193). Surprisingly, these findings do not extend to Ms. barkeri; in this organism, Fpo appears to be dispensable for methylotrophic growth, whereas Frh is essential (224). On this basis, Kulkarni et al. (224) in Metcalf's laboratory have proposed that H₂ is an intermediate during methylotrophic growth wherein electrons from the F₄₂₀H₂ produced by Mer and Mtd may be used to drive H₂ production by Frh. The H₂ produced is in turn reoxidized via a hydrogenase (Vhu) that can reduce methanophenazine to facilitate heterodisulfide reduction by Hdr, thereby bypassing the need for Fpo (224). Frh activity is consistently 10-fold higher in Ms. barkeri than in Ms. mazei; hence, Frh may be able to fully substitute or compensate for loss of Fpo activity only in the former organism (193). Fpo is also likely to be dominant during methylotrophic growth in Methanosarcina acetivorans, which exhibits low levels of hydrogenase expression and activity (322, 323).

Beyond methylotrophic methanogenesis, several other roles have been proposed for the Fpo system. For example, the proton gradient generated by Fpo is thought to contribute to ATP synthesis during hydrogenotrophic methanogenesis, while H₂ oxidation can be coupled to methanophenazine reduction directly (via the methanophenazine-reducing hydrogenase), F₄₂₀ is also sometimes preferentially used as an intermediate (through the combined activities of Frh and Fpo) (182, 224). There is also evidence that Fpo contributes to the growth of Methanosarcina barkeri on carbon monoxide (324). More recent work also suggests that FpoF may sometimes function as a cytosolic enzyme independently of the other membrane-bound Fpo components in certain methanogens (193, 325). Consistently, the fpoF gene is genomically separated from the rest of the fpo operon in several Methanosarcina species (269, 326), and the protein is expressed at high levels in the cytosolic fraction of Ms. mazei cells (193). FpoF from Ms. mazei can slowly, but specifically, catalyze electron transfer from Fd_red to F₄₂₀ (Fig. 7), which may help to maintain redox balance among methanogenic cofactors (193). Interestingly, members of the genus Methanosaeta (part of the order Methanosarcinales) contain a variant of Fpo (fpoABCDHIJKLMNO) that lacks the F₄₂₀H₂-oxidizing subunit FpoF and instead may be dependent on another reducing agent, e.g., Fd_red (196, 327).

A related multimeric membrane-bound proton-translocating complex is also present in some nonmethanogenic archaea (198). The enzyme appears to serve as an F₄₂₀H₂-dependent menaquinone reductase (Fqo) during sulfate respiration of Archaeoglobi (198, 199). Transcriptome analysis has shown that Fqo is constitutively expressed at high levels in Archaeoglobus fulgidus together with the other respiratory chain components (200). The enzyme is composed of 11 subunits that assemble in a manner similar to Fpo in methanogenic archaea, but it likely reduces menaquinone rather than methanophenazine via the FqoD subunit (199). Homologous enzymes are also encoded by some ANME archaea (e.g., Methanoperedens nitroreducens) and are proposed to input electrons derived from methane oxidation into sulfate- and nitrate-reducing respiratory chains (21, 215, 328).

3.3.3. Fpr: F₄₂₀H₂-dependent oxidase

Among the more recently discovered F₄₂₀-binding proteins, the physiological role of the F₄₂₀H₂-dependent oxidases (Fpr/FprA) is to catalyze the four-electron reduction of dioxygen (O₂) to water (H₂O) in methanogens (161, 192). In contrast to terminal oxidases, these enzymes are not linked to respiratory chains and instead appear to have evolved to detoxify O₂. Encoded in the genomes of five of the six presently recognized orders of methanogens (173), the F₄₂₀H₂ oxidases are part of the flavodiiron protein family, which have been implicated in O₂ and/or NO detoxification in microorganisms across all three domains of life. The methanogen enzymes share particularly high sequence identity (∼40%) to the reductases in the anaerobic bacteria Moorella thermoacetica and Desulfovibrio vulgaris (52, 329), but they use F₄₂₀H₂ rather than an additional rubredoxin domain containing FMNH₂ as the reductant. Fpr has been correlated with the ability of methanogens such as Methanobrevibacter arboriphilus and Methanothermobacter marburgensis (Mt. marburgensis) to efficiently scavenge micromolar concentrations of O₂ in their environment (192). Although yet to be confirmed through genetic dissection, it has been hypothesized that such enzymes are responsible for the surprising and potentially ecologically significant aerotolerance of many members of the methanogens (all obligate anaerobes) (192, 330, 331). Some methanogens carry genes that encode multiple isozymes (e.g., Mt. marburgensis encodes three FprA homologs) (332), though it has yet to be resolved whether they are differentially regulated and kinetically distinct.

X-ray crystal structures of Fpr from Methanothermobacter marburgensis have been determined. They reveal that each monomer of this homotetrameric enzyme binds a diiron center, an FMN cofactor, and a solvent-diffusible F₄₂₀H₂ molecule (161, 192, 255). The enzyme forms a functional homodimer, with the diiron center of one subunit associating with the FMN cofactor of another (161). The structure of this enzyme has been solved in three conformational states (reduced-active, oxidized-active, and oxidized-inactive states) by altering the oxygen exposure of the protein crystals prior to data collection (161). This has enabled the elucidation of the probable catalytic mechanism for this protein (Fig. 13). Dioxygen binding occurs at the reduced-active state [Fe(II)Fe(II)FMNH₂], where the F₄₂₀H₂-binding site adjacent to FMN is in a “closed” conformation. The oxygen molecule forms a peroxo intermediate that bridges the diiron center and is reduced to release two water molecules through a diferric transition state. This forms the oxidized-active state of the enzyme [Fe(III)Fe(III)FMN], inducing conformational changes to “open” the F₄₂₀H₂-binding site. Two subsequent F₄₂₀H₂ molecules can then bind in a Si-Si conformation adjacent to the oxidized FMN, reducing both the diiron center and FMN to regenerate the reduced-active state. The enzyme also adopts a third oxidized-inactive state where the iron ion closest to FMN is displaced. An additional iron ion is also present, which locks the F₄₂₀H₂-binding site in the “open” state, preventing oxygen binding. This is hypothesized to occur in the presence of excess oxygen to prevent loss of reducing power (161).

FIG 13.

Summary of F₄₂₀H₂-dependent oxygen detoxification by Fpr. The mechanism was inferred based on the three crystallographic states of the active site (161): (a) the reduced-active state where the F₄₂₀H₂-binding site adjacent to FMN is “closed” by a loop with bulky aromatic residues (PDB ID 2OHI), (b) the oxidized-active state where the F₄₂₀H₂-binding site is “open” due to conformational changes in the loop (PDB ID 2OHH), and (c) the oxidized-inactive state where one iron atom in the diiron center is displaced and an additional third iron is present locking the loop in the “open” state (PDB ID 2OHJ). Fe(III) is shown in green, Fe(II) is shown in brown, water molecules at the predicted dioxygen-binding site are red, and FMN is in yellow. (d) Catalytic mechanism of Fpr. The five steps are shown as follows. (i) Transient binding of dioxygen to the reduced-active state; (ii) oxidation of the diiron center with the release of a water molecule; (iii) oxidation of FMN to release the second water molecule; (iv) reduction of the diiron center via FMN at the oxidized-active state, which binds F₄₂₀H₂ as an electron donor; (v) reduction of FMN by a second F₄₂₀H₂ molecule.

3.3.4. Fsr: F₄₂₀H₂-dependent sulfite reductase

The F₄₂₀H₂-dependent sulfite reductase (Fsr) catalyzes the six-electron reduction of sulfite to sulfide (51). Discovered by Johnson and Mukhopadhyay, the enzyme appears to have a dual role in methanogens: detoxification of sulfite and growth on sulfite as the sole sulfur source (51, 191). While sulfite is generally inhibitory for growth of methanogens (e.g., Methanococcus maripaludis) (191, 333), diverse species are able to utilize it as a sole sulfur source (e.g., Methanocaldococcus jannaschii) (51, 334, 335). Mc. maripaludis can be rendered sulfite tolerant through recombinant expression of Mc. jannaschii Fsr (191). Fsr purified from Mc. jannaschii rapidly catalyzes sulfite reduction using F₄₂₀H₂ (51). The single-subunit enzyme appears to have arisen through the fusion of an F₄₂₀H₂-binding protein with a sulfite reductase (336, 337): the N-terminal domain is homologous to the FhrB-like domains of other F₄₂₀H₂ dehydrogenases, while the C-terminal domain is similar to siroheme-containing dissimilatory sulfite reductases (51). It is therefore proposed that, as in Frh, Ffd, and Fpo (Fig. 8, 9, and 12), F₄₂₀H₂ is oxidized at the N-terminal domain and electrons are funneled to the C-terminal domain via a possible flavin, an [4Fe4S] cluster, and siroheme, where the sulfite is subsequently reduced (337). The enzyme appears to be sensitive to oxygen, but it can be reactivated by cellular thioredoxins (338). Other than Fsr, some methanogens can also mobilize sulfite using the P₅₉₀-type sulfite reductases (339), the physiological role of which are still incompletely resolved (336).

3.3.5. Fno: F₄₂₀H₂-dependent NADP reductase

In most cases, the catabolic pathways of methanogens reduce F₄₂₀ and ferredoxin, but not nicotinamides. In order to generate NADPH for biosynthetic processes, methanogens instead transfer electrons from F₄₂₀H₂ to NADP (151). This process depends on F₄₂₀H₂:NADP oxidoreductase (Fno), a physiologically reversible enzyme that primarily acts as an F₄₂₀H₂-dependent NADP reductase in methanogens and an F₄₂₀-reducing NADPH dehydrogenase in bacteria. Fno is present in all six orders of methanogens and can reduce NADP using electrons derived from F₄₂₀H₂ during hydrogenotrophic, formatotrophic, and methylotrophic growth (16, 152, 340). An exception is those methanogens that grow on primary alcohols (e.g., Methanoculleus thermophilicus), which instead use an NADP-reducing primary alcohol dehydrogenase (272); in such organisms, Fno serves as an F₄₂₀-reducing NADPH dehydrogenase that generates sufficient F₄₂₀H₂ to drive the fourth and fifth steps in the CO₂-reducing pathway of methanogenesis (22). In contrast, methanogens that harbor an F₄₂₀-reducing secondary alcohol dehydrogenase use Fno in the typical NADP-reducing direction (22). Homologous enzymes also appear to bridge catabolic and anabolic processes in Archaeoglobi (201) and Halobacteria (136).

One of the best-understood F₄₂₀-dependent enzymes, Fno has been purified and characterized from methanogens of the genera Methanococcus (341, 342), Methanothermobacter (343, 344), Methanosphaera (345), and Methanogenium (22). The structure of Fno from Archaeoglobus fulgidus complexed with F₄₂₀ and NADP gives direct structural insight into its hydride transfer mechanism (Fig. 14). The single-subunit enzyme contains a small C-terminal domain and an N-terminal domain characteristic of a dinucleotide-binding Rossmann fold. The nicotinamide and deazaflavin moieties of the cofactors are bound roughly parallel to each other (Si-face to Si-face) in a hydrophobic pocket between the domains (160). The aromatic groups are laterally shifted relative to each other, such that the C-4 atom of NADP is positioned exactly above the C-5 atom of F₄₂₀ (201, 346) to allow for hydride transfer at an optimal distance of 3.1 Å. The affinity of F₄₂₀ for Fno increases in the presence of NADP, suggesting that NADP binding facilitates F₄₂₀ binding (160). Consistently, structural comparison between apo- and holoenzymes indicates that NADP binding facilitates a conformational change that induces F₄₂₀H₂ binding and generates a catalytically active ternary complex (160).

FIG 14.

Structure and catalytic mechanism of Fno. (a) Structure of the active site of Fno (PDB ID 1JAY [160]), showing F₄₂₀ and NADP positioned for electron transfer. (b) Hydride transfer mechanism from the Si-face of F₄₂₀ to the Si-face of NADP⁺ (160). R₁ is the ribitylphospholactyl-oligoglutamate chain of F₄₂₀, and R₂ is 2-phosphoadenosine 5-diphosphate.

3.4. Cofactor F₃₉₀

Two purinated derivatives of F₄₂₀ are formed in methanogens under certain conditions, and these two derivatives of F₄₂₀ are collectively referred to as F₃₉₀ (347, 348). F₃₉₀-A and F₃₉₀-G are formed when F₄₂₀ forms a phosphodiester linkage with AMP and GMP, respectively, via the 8-hydroxy group of the 5-deazaflavin ring (348, 349). Seemingly exclusive to methanogens, F₃₉₀ has been identified in genera as diverse as Methanothermobacter (347, 350), Methanobacterium (351), Methanobrevibacter (330), and Methanosarcina (352). Owing to their electron-donating groups, F₃₉₀ compounds have a higher standard redox potential (−320 mV) than F₄₂₀ (−340 mV) and hence may be ideal for sensing or catalytic roles under oxidizing conditions (353). These derivatives are synthesized when methanogens are exposed to oxygen and are hydrolyzed back to F₄₂₀ and AMP/GMP upon reestablishment of anaerobiosis (349). Production depends on an ATP/GTP-dependent F₃₉₀ synthetase of the adenylate-forming superfamily (354–356), while a hydrolase mediates the AMP/GMP-forming hydrolysis reaction (356, 357). As F₃₉₀ synthesis appears to be sensitive to both redox state and oxygenation levels (296, 355), it has been proposed that the cofactor derivative is part of a redox-sensing system that regulates metabolic activity of methanogens. It has been consistently demonstrated that F₃₉₀ synthetase transcription and F₃₉₀ cellular expression levels are correlated with the availability of reductant in Methanobacterium thermoautotrophicum (358, 359). However, no genetic or phenotypic studies have been performed to resolve its physiological role. In fact, there has been an almost complete absence of literature on this molecule over the last 2 decades.

4. F₄₂₀ IN MYCOBACTERIA AND OTHER BACTERIA

4.1. Physiological Roles

4.1.1. Mycobacteria

Relatively little is known about the roles of F₄₂₀ in bacteria. The cofactor has been experimentally shown to be synthesized in only one bacterial phylum thus far, Actinobacteria, where it has mainly been studied for its roles in secondary, rather than primary, metabolism. Nevertheless, a number of recent phenotypic and biochemical studies have shed light on the endogenous roles of F₄₂₀ in mycobacteria, an actinobacterial genus of major medical and environmental significance (360, 361). F₄₂₀ is synthesized and reduced by all members of the genus Mycobacterium, including saprophytes (e.g., M. smegmatis, M. fortuitum), opportunistic pathogens (e.g., M. avium, M. kansasii), and the causative agents of tuberculosis (M. tuberculosis complex) and leprosy (M. leprae) (20, 125, 145). The observation that F₄₂₀ is synthesized even in M. leprae, rendered an unculturable, host-dependent organism through massive genome decay (362), suggests that it has an evolutionarily conserved central role in mycobacterial metabolism. In contrast to methanogens, F₄₂₀ is not essential for the viability of mycobacteria under ideal conditions: F₄₂₀ biosynthesis (fbiC) and reduction (fgd) genes have been successfully deleted or disrupted in M. smegmatis (28, 31, 132, 363), M. tuberculosis (32, 35), and M. bovis (72). However, there is a range of evidence that F₄₂₀ contributes to the notorious ability of mycobacteria to persist in deprived and challenging environments (56). Mycobacteria that are unable to synthesize F₄₂₀ are unable to survive oxygen deprivation, oxidative stress, nitrosative stress, or antibiotic treatment (Fig. 15) (31, 32, 363).

FIG 15.

Pleiotropic phenotypes associated with loss of function of F₄₂₀ in mycobacteria. Relevant reactions in the F₄₂₀ biosynthesis and utilization pathway are shown in gray. Hollow arrows show observed chemical and phenotypic effects due to loss-of-function mutations in specific enzymes in the pathway. Q, quinone; QH₂, dihydroquinone; HQ^•, semiquinone.

Several F₄₂₀-dependent enzymes have been functionally annotated in mycobacteria. Pathogenic mycobacteria such as M. tuberculosis encode F₄₂₀-reducing hydroxymycolic acid dehydrogenases (fHMAD) that oxidize hydroxymycolic acids to ketomycolic acids in the cell wall (364, 365). These mycolic acid derivatives appear to influence the integrity and permeability of the cell envelope, which renders them less sensitive to cytotoxic agents such as antibiotics (366–368). Preliminary data indicate that a subgroup of the flavin/deazaflavin oxidoreductase superfamily (FDOR-AAs) may also be involved in fatty acid modification (30). Other members of this superfamily (FDOR-Bs) reduce the degradation products formed during heme oxygenation (30): biliverdin (produced by host heme oxygenase-1 and mycobacterial HugZ in the CO-generating pathway) and possibly mycobilin (produced by mycobacterial MhuD in the CO bypass pathway) (369–371). Our biochemical studies have shown that M. smegmatis carries a gene that encodes a conserved F₄₂₀H₂-dependent biliverdin reductase that rapidly reduces biliverdin to bilirubin (30), a potent antioxidant (372, 373).

There is also evidence that F₄₂₀ contributes to an oxidative stress response system in mycobacteria. The survival rate of ΔfbiC strains of M. tuberculosis is 100- to 1,000-fold lower than wild-type cells following challenge with redox cycling agents (i.e., menadione, plumbagin) and antibiotics (i.e., isoniazid, clofazimine) (32). Δfgd strains of M. smegmatis are similarly impaired (363). One explanation is that mycobacteria store electrons as glucose-6-phosphate (G6P) and mobilize them using Fgd (F₄₂₀-dependent glucose-6-phosphate dehydrogenase) in response to oxidative stress; G6P levels in M. smegmatis are consistently approximately 100-fold higher than those of E. coli during preferential growth conditions, but the levels become depleted following challenge with redox cycling agents (363). F₄₂₀H₂-derived electrons may be used in endogenous redox processes to prevent or reverse damage from reactive oxygen species. For example, it has been proposed that a subgroup of the flavin/deazaflavin oxidoreductase superfamily (FDOR-As) are F₄₂₀H₂-dependent menaquinone reductases that maintain the respiratory chain in a reduced state in response to oxidative stress (32). Several previous reports have demonstrated that the mycobacterial respiratory chain can be remodeled in response to environmental changes (374, 375), and the ability of F₄₂₀H₂ to serve as a respiratory electron donor has already been demonstrated for respiratory archaea (162, 199). However, this hypothesis has yet to be supported with data on phenotypes or energy, and it remains unclear whether purified FDOR-As are capable of reducing menaquinone (30, 32). There is also evidence that mycobacteria instead use electrons liberated from G6P by Fgd to directly detoxify exogenous agents (363). Two independent studies have demonstrated that FDOR-As rapidly reduce menadione and plumbagin using F₄₂₀H₂ (30, 32), and it is also plausible that these highly promiscuous proteins (28, 55) can directly detoxify certain antibiotics too. However, genetic studies have yet to definitively link FDORs to antibiotic resistance and oxidative stress responses.

The potentially related role of F₄₂₀ in nitrosative stress resistance is also perplexing. M. tuberculosis transposon mutants of fbiC are hypersusceptible to acidified nitrite (376); this was shown through an in vitro screen designed to simulate the environment of an activated macrophage, in which inducible nitric oxide synthase (iNOS)-derived NO is oxidized to NO₂⁻, acidified into HNO₂, and dismutated into NO and NO₂ (377), which have antimycobacterial properties (378). One study showed that NO₂ is rapidly nonenzymatically reduced to NO by F₄₂₀H₂ under aerobic conditions (31). However, it is likely that F₄₂₀-dependent enzymatic mechanisms also contribute to nitrosative stress resistance, either through direct detoxification or indirect mechanisms. Indeed, it is possible that F₄₂₀ may confer protection against cytotoxic agents in multiple ways: enhancing physical barriers through cell wall synthesis, direct detoxification by reducing exogenous agents, and maintaining redox balance through endogenous metabolism. Given the diverse roles of F₄₂₀ in mycobacterial metabolism and the pleiotropic phenotypes associated with the cofactor's absence, it seems likely that F₄₂₀ is required for latent tuberculosis infection in vivo, though this has yet to be definitively confirmed. In line with this, M. tuberculosis strains incapable of synthesizing ketomycolic acids are attenuated in macrophages and mice (366–368). One study surprisingly indicated that transposon mutants of fbiC are viable in vivo in the murine model of acute infection (379), though it is unclear whether such mutants would be capable of establishing a chronic infection.

4.1.2. Streptomycetes

It is well established that F₄₂₀ is required for the synthesis of tetracycline antibiotics, a group of broad-spectrum aromatic polyketide antibiotics produced by streptomycetes (380). As far back as 1960, McCormick et al. isolated a hydride-transferring cofactor mediating tetracycline biosynthesis (24, 381–383), now known to be F₄₂₀ (29, 122, 384). A combination of genetic and biochemical studies have since shown that an F₄₂₀H₂-dependent reductase (OxyR) catalyzes the final step of oxytetracycline biosynthesis (385), namely, reduction of the C-5a=C-11a double bond of dehydrooxytetracycline (Fig. 16) (29). This enzyme can also perform the equivalent reaction for tetracycline. Closely related enzymes are involved in the same step during biosynthesis of chlorotetracycline (CtcR) and dactylocyclinone (DaCO4) in Streptomyces aureofaciens and Streptomyces rimosus (29, 386). Encoded by the oxytetracycline (oxy), chlorotetracycline (ctc), and dactylocyclinone (dac) gene clusters (29), these enzymes are members of the flavin/deazaflavin oxidoreductase (FDOR) superfamily (30) and utilize F₄₂₀H₂ reduced through the action of Fno (387).

FIG 16.

Reactions catalyzed by F₄₂₀H₂-dependent reductases in the biosynthesis pathways of tetracyclines (26), lincosamides (146, 388, 389), and aminoglycosides (394).

F₄₂₀ is also required for the synthesis of lincosamide antibiotics by Streptomyces lincolnensis strains (146, 388, 389), including lincomycin, the precursor of the clinical semisynthetic antibiotic clindamycin (390). On the basis of the accumulation of 4-propylidene-3,4-dihydropyrrole-2-carboxylic acid by strains unable to biosynthesize F₄₂₀, it is proposed that an F₄₂₀H₂-dependent reductase catalyzes the reduction of the imine moiety of the dihydropyrrole to tetrapyrrole (Fig. 16) (389, 391). The biosynthesis of other pyrrolobenzodiazepine antibiotics (392) are facilitated by equivalent F₄₂₀H₂-dependent imine reduction steps, namely, tomaymycin (Streptomyces achromogenes) (50), sibiromycin (Streptosporangium sibiricum) (393), kasugamycin (Streptomyces kasugaensis) (394), and anthramycin (Streptomyces rifuineus) (395). However, biochemical studies have yet to definitively identify which enzymes are responsible for these reactions. The strongest candidates are the putative F₄₂₀H₂-dependent luciferase-like hydride transferases (LLHTs) encoded in the sequenced antibiotic synthesis gene clusters (50, 391, 393–395) of each of these organisms. Because all research thus far has focused on the roles of F₄₂₀ in the secondary metabolism of streptomycetes, little is known about the roles of this cofactor in central metabolism of this genus; it is likely that streptomycetes use F₄₂₀ to support some important metabolic pathways, as they carry genes that encode homologs of mycobacterial enzymes such as the F₄₂₀H₂-dependent biliverdin reductase (30).

4.1.3. Other actinobacteria

It is established that F₄₂₀ is synthesized in multiple other actinobacterial genera, including Rhodococcus, Nocardia, and Nocardioides (27, 54, 145). However, all studies of such genera have focused on the roles of F₄₂₀ in exogenous substrate reduction, and very little is known about the endogenous roles of F₄₂₀-dependent processes. The richest literature is on the degradation of picrate (2,4,6-trinitrophenol) and related compounds (e.g., 2,4-dinitrophenol, 2,4-dinitroanisole) (396, 397). A number of actinobacteria, including Rhodococcus opacus and Nocardioides simplex, are able to mobilize picrate as their sole carbon and nitrogen source (396, 398). This depends on reductive activation of these particularly electron-deficient aromatic compounds using two F₄₂₀H₂-dependent hydride transferases (hydride transferase I [HTI] and hydride transferase II [HTII]) (section 4.3.2) (155). Fno supplies the reductant for this process and is expressed from the same operon as the hydride transferases (54, 155, 396). While polynitroaromatic compounds are anthropogenic, actinobacteria may have evolved the capacity to degrade them from preexisting pathways that metabolize naturally occurring nitroaromatic compounds (e.g., chloramphenicol) (399, 400). It has also been demonstrated that F₄₂₀H₂-dependent oxidoreductases of the flavin/deazaflavin oxidoreductase superfamily have broad substrate specificity; enzymes purified from genera as diverse as Mycobacterium, Frankia, Nocardia, and Rhodococcus are capable of reducing coumarin natural products (28, 55). F₄₂₀ may also contribute to the well-reported abilities of soil actinomycetes to biodegrade a wide variety of other polycyclic aromatic hydrocarbons (401). While the physiological advantage of this promiscuity is unclear, it might provide actinobacteria an adaptive or selective advantage to consume or detoxify the wide range of natural products in their respective environments (402, 403).

4.2. F₄₂₀-Reducing Dehydrogenases

4.2.1. Fno: F₄₂₀-reducing NADPH dehydrogenase

Fno is the only redox-active F₄₂₀-dependent protein proven to be conserved between archaea and bacteria. Whereas Fno primarily serves to reduce NADP in methanogens (F₄₂₀H₂-dependent NADP reductases), its homologs generally act in the reverse direction to reduce F₄₂₀ in bacteria (F₄₂₀-reducing NADPH dehydrogenases) (219); this reflects that, whereas F₄₂₀ is a central catabolic cofactor in methanogens, it is of secondary importance to NADP in the central metabolism of most bacteria (168). While Fno has yet to be structurally characterized in actinobacteria, the enzyme is expected to have a similar structure and mechanism: sequence comparisons and biochemical studies (12) indicate that the overall architecture and cofactor-binding sites are conserved with the archaeal enzyme (section 3.3.5) (160). The F₄₂₀H₂ generated by Fno in bacteria is used for various reductive processes, for example, biosynthesis of tetracycline antibiotics by Streptomyces (387) and the mobilization of picrate by Rhodococcus and Nocardioides species (54, 155).

4.2.2. Fgd: F₄₂₀-reducing glucose-6-phosphate dehydrogenase

While Fno appears to be the enzyme primarily responsible for F₄₂₀ reduction in most actinobacteria, it is replaced by the F₄₂₀-reducing glucose-6-phosphate dehydrogenase in several genera, including Mycobacterium (Table 2). This enzyme directly links central carbon catabolism in actinobacteria to F₄₂₀ reduction (glucose-6-phosphate + F₄₂₀ → 6-phosphogluconolactone + F₄₂₀H₂) (163, 404). First identified in the soil bacterium M. smegmatis (148, 404), Fgd has since been identified in multiple other environmental actinobacteria and the obligate pathogens M. tuberculosis and M. leprae (145). Fgd is either the sole or main source of F₄₂₀H₂ in mycobacteria; neither ΔfbiC and Δfgd strains are capable of activating exogenous substrates through F₄₂₀H₂-dependent reactions in M. tuberculosis (33, 35) and M. smegmatis (28, 363). Fgd therefore appears to have evolved principally as a mechanism to generate F₄₂₀H₂. As elaborated above, there is also evidence that glucose-6-phosphate serves as an electron store in mycobacteria that is mobilized by Fgd in response to oxidative stress (32, 363). The role of Fgd in generating flux through the pentose phosphate pathway appears to be supplementary, given that most mycobacteria also encode conventional NADP-dependent glucose-6-phosphate dehydrogenases (145). An interesting exception may be M. leprae, as genome analysis and biochemical studies indicate that it employs F₄₂₀, but not NADP, for G6P oxidation (145, 362, 405).

The F₄₂₀-reducing and NADP-reducing glucose-6-phosphate dehydrogenases are not phylogenetically related (148). Fgd is a member of the bacterial luciferase family (163) with a similar TIM barrel structure and catalytic mechanism reminiscent of Adf (49) and Mer (159). The cofactor is accommodated in the active site, with the isoalloxazine rings innermost and the oligoglutamate tail extending into the solvent (Fig. 6), where the isoalloxazine is in a bent butterfly-like conformation due to steric interactions with the protein backbone, including the nonprolyl cis-peptide bond behind its Re-face (163). The glucose-6-phosphate has been modeled to bind in a positively charged pocket adjacent to the Si-face of the deazaflavin (163), similar to what was observed in the ternary complex of the related Adf (Fig. 10). Hydride transfer is thought to occur similarly to Adf (Fig. 17) and is mediated by conserved histidine, tryptophan, and glutamate residues (49, 163): proton abstraction is initiated by the histidine, tryptophan stabilizes the resulting anion transition state, and glutamate is likely to serve as the proton donor for N-2 of the deazaflavin for F₄₂₀H₂ formation (49, 163).

FIG 17.

Proposed catalytic mechanism of Fgd (163). F₄₂₀ is reduced to F₄₂₀H₂, and glucose-6-phosphate is oxidized to 6-phosphogluconate.

4.2.3. fHMAD: F₄₂₀-reducing hydroxymycolic acid dehydrogenase

The F₄₂₀-reducing hydroxymycolic acid dehydrogenase (fHMAD) is responsible for oxidizing hydroxymycolic acids to ketomycolic acids during cell wall biosynthesis (365). A member of the bacterial luciferase family, the enzyme shares 36% sequence identity with Fgd (364). However, in contrast to its original annotation, the enzyme cannot oxidize glucose-6-phosphate (364) and is specific for hydroxymycolic acids (365). The enzyme is translocated through the cell membrane by the Tat pathway and is anchored to the outside of the cell membrane, where it can function in cell wall modification (364). Reflecting the taxonomic distribution of fHMAD (364, 365), ketomycolic acids are distributed in pathogenic mycobacteria (e.g., M. tuberculosis complex, M. avium complex) but are absent from most soil species (e.g., M. smegmatis) (406). Ketomycolic acids appear to be critical for the virulence of M. tuberculosis; strains lacking oxygenated mycolic acids have profoundly altered envelope permeability, are hypersusceptible to antibiotics, and are attenuated in macrophages and mice (366–368). Consistent with the synthesis of ketomycolic acids in response to stress, the gene encoding fHMAD is under the control of the alternative sigma factor SigF in M. tuberculosis (407, 408). It was recently confirmed that fHMAD is inhibited by the nitroimidazopyran prodrug pretomanid (PA-824) (365); this interaction may be responsible for the altered mycolic acid composition of pretomanid-treated cells and may contribute to the mode of action of this next-generation bactericidal agent (33).

4.3. F₄₂₀H₂-Dependent Reductases

4.3.1. FDORs: flavin/deazaflavin oxidoreductase superfamily

F₄₂₀H₂-dependent reductases elicit the physiological roles of F₄₂₀ in actinobacteria. They are split into two superfamilies, the flavin/deazaflavin oxidoreductases (FDORs) (30) and the luciferase-like hydride transferases (LLHTs; section 4.3.2) (37). FDORs are small (∼150-residue) enzymes that accommodate a cofactor-binding channel and substrate-binding pocket into a split β-barrel fold (30). This superfamily is highly diverse in terms of catalytic activity (reductases, oxidases, and oxygenases), cofactor specificity (F₄₂₀, FMN, FAD, and heme), and substrate range (30, 409). We have shown that they have diversified into two major families, FDOR-As and FDOR-Bs, that share less than 30% sequence similarity but share the same protein fold (28, 30) (Fig. 18). Proteins from the FDOR-A family are exclusively F₄₂₀-binding proteins (28, 35, 55, 164, 410) restricted to the phyla Actinobacteria and Chloroflexi (28, 30, 37). In contrast, FDOR-B proteins are widely distributed, including in bacteria that do not synthesize F₄₂₀. They include the ubiquitous FMN-dependent pyridoxine/pyridoxamine 5′-phosphate oxidases (PnPOx) involved in vitamin B₆ biosynthesis (411–413), heme oxygenases (HugZ) involved in heme catabolism (414–416), and several groups of uncharacterized FAD-binding proteins (30, 417). Actinobacteria and Chloroflexi also carry genes that encode multiple F₄₂₀H₂-dependent reductases of the FDOR-B family, which are broadly divided into six subgroups (28, 30, 165, 418, 419). Structural and sequence analyses demonstrate that conserved motifs define cofactor specificity (30); in the case of F₄₂₀H₂-dependent reductases, deazaflavin binding is stabilized by a large hydrophobic groove complementary to the isoalloxazine ring and a positively charged groove that interacts with the oligoglutamate tail (28, 30, 164, 165). Interestingly, unlike all other F₄₂₀-binding proteins characterized thus far, the most likely substrate-binding pocket of the F₄₂₀-binding FDORs appears to be toward the Re-face of the cofactor (30, 164, 165), similar to the FMN-dependent members of the superfamily (420).

FIG 18.

Representative crystal structures of FDOR-A (monomers) and FDOR-B (dimers) proteins. (a) Structures of the quinone-reducing FDOR-A1 proteins MSMEG_2027 (PDB ID 4Y9I [30]) and MSMEG_3356 (PDB ID 3H96 [28]) overlaid with the complex of rv3547 with F₄₂₀ (PDB ID 3R5R [164]) with menadione docked into the active site. (b) Overlay of solved structures of F₄₂₀H₂-dependent FDOR-B proteins. These proteins include the FDOR-B1 proteins rv2991 (PDB ID 1RFE) and MSMEG_3380 (PDB ID 3F7E [28]), FDOR-B2 protein MSMEG_6526 (PDB ID 4ZKY [30]), FDOR-B3 protein rv1155 complexed with F₄₂₀ (PDB ID 4QVB [165]) and FDOR-B4 protein rv2074 (PBD ID 2ASF [419]). Biliverdin, a substrate of FDORs B3 and B4, is docked at the active site (30).

In mycobacteria, there is a multiplicity of F₄₂₀H₂-dependent reductases of the FDOR family: 30 in M. smegmatis, 15 in M. tuberculosis, and 3 in M. leprae (30). As most of these enzymes remain to be functionally annotated, the reasons behind the extreme expansion and diversification of this superfamily remain unclear. The highly diverse architecture of the substrate-binding sites of these proteins, concurrent with a high degree of conservation of the F₄₂₀-binding site, suggests that they have evolved to catalyze the F₄₂₀H₂-dependent reduction of a variety of substrates (30). Some subgroups (e.g., FDOR-B2s, FDOR-B4s) are tightly phylogenetically clustered and have probably been constrained for a specific function (30). In contrast, representatives of the manifold subgroup FDOR-A1 exhibit broad and overlapping substrate specificities (28, 30). Such enzymes are capable of reducing a wide range of exogenous substrates, including coumarin natural products such as fungus-derived aflatoxins and plant-derived furanocoumarins (e.g., angelicin, methoxsalen) (28, 55). They also show potent activity against redox cycling agents such as menadione and plumbagin (30, 32). The physiological role of these enzymes may therefore be to detoxify a wide range of oxidizing agents in their environment using electrons channeled from G6P. The absence of such detoxification systems may contribute to the profound sensitivity of ΔfbiC and Δfgd mutants to redox cycling agents and antibiotics, as discussed in section 4.1.1 (32, 363). Consistent with a role in detoxification or biodegradation, there is some evidence from expression studies (28, 30, 164) and proteome analyses (421, 422) that these enzymes are bound to the membrane through their N termini. Another enzyme of this class, rv3547 (Ddn; deazaflavin-dependent nitroreductase) has also attracted much attention for its role in the activation of nitroimidazole prodrugs (e.g., pretomanid, delamanid) by M. tuberculosis (section 5.1) (34, 35, 164).

The endogenous roles of the FDOR-type F₄₂₀H₂-dependent reductases in mycobacteria are presently being resolved. We have shown that a structurally characterized (419) subgroup of this family (FDOR-B4s) are efficient F₄₂₀H₂-dependent biliverdin reductases (30). They convert the heme degradation product biliverdin—produced by HugZ in environmental mycobacteria (30) and host heme oxygenase 1 (HO1) (369) during tuberculosis infection—to bilirubin via hydride transfer to C-10 (30). This may be advantageous for survival of oxidative stress, given that bilirubin is a potent antioxidant that can compensate for 10,000-fold excess in peroxide radicals (372, 373). A recent study showed that addition of bilirubin enhanced the survival of Mycobacterium abscessus in HO1-inhibited macrophages, possibly via modulation of intracellular reactive oxygen species (ROS) levels (423). These proteins may also reduce mycobilins (30), the product of the CO bypass pathway of heme oxygenation by mycobacterial MhuD (370). This FDOR group is only the second family of biliverdin reductases to be identified; a previously characterized family of mammalian and cyanobacterial biliverdin reductases employs nicotinamides as an electron source (424, 425). We also observed low-level biliverdin reductase activity in the structurally related FDOR-B3s (30). However, their low catalytic efficiency and suboptimal active site structure for biliverdin binding suggests that this promiscuous activity may result from a common evolutionary origin to the FDOR-B4s; FDOR-B3 enzymes are therefore likely to have a different, currently unidentified physiological substrate (30).

Among other FDORs, there is preliminary evidence that FDOR-AAs are F₄₂₀H₂-dependent fatty acid reductases; these membrane-bound enzymes may contribute to cell wall modification and host invasion, although their substrate specificity has yet to be defined (30). While it has been proposed that FDOR-A proteins are F₄₂₀H₂-dependent menaquinone reductases (32), to date, activity has been observed only with nonphysiological quinones (e.g., menadione), rather than with menaquinone (30, 32); hence, it is unclear whether these enzymes have primarily evolved to input electrons into the respiratory chain or instead detoxify exogenous redox cycling agents (section 4.1.1). Finally, it was recently shown that the F₄₂₀H₂-dependent step in the biosynthesis of antibiotics of the tetracycline, oxotetracycline, and chlortetracycline classes (122, 381, 382, 384) is mediated by enzymes of the FDOR-B1 subgroup in streptomycetes (section 4.1.2) (29).

4.3.2. LLHTs: luciferase-like hydride transferase superfamily

Luciferase-like hydride transferases (LLHTs) are another diverse superfamily of flavin/deazaflavin enzymes. These enzymes were previously defined as luciferase-like monooxygenases (LLMs), but this is inappropriate given that their reaction mechanisms are O₂ independent. Like the FDORs, members of this superfamily vary in their cofactor preferences (F₄₂₀, FMN, FAD) and catalytic activities (oxidases, reductases, oxygenases) (163, 426–428). F₄₂₀-binding LLHTs can be distinguished by a conserved glycine residue that binds the phosphate group without steric hindrance, which is not conserved in the FMN-binding proteins of this family (48). The best-characterized F₄₂₀-dependent LLHTs are the three aforementioned dehydrogenases: F₄₂₀-reducing methylene-H₄MPT dehydrogenase (Mtd), F₄₂₀-reducing glucose-6-phosphate dehydrogenase (Fgd), and F₄₂₀-reducing hydroxymycolic acid dehydrogenase (fHMAD). However, comparative genome analysis indicates that there are numerous other F₄₂₀-dependent LLHTs in actinomycetes, the majority probably serving as reductases (37). These have been implicated in a variety of roles, ranging from pyrrolobenzodiazepene antibiotic synthesis in streptomycetes (50, 393, 394, 429) to cell wall metabolism in mycobacteria (37) and exogenous substrate mobilization by rhodococci (155). A bioinformatics analysis predicted that there are some 45 F₄₂₀-binding LLHTs in M. smegmatis and 17 in M. tuberculosis, though this has yet to be validated experimentally (37). In contrast to the FDOR superfamily (30), to date, no comprehensive analysis of the phylogeny, structure, and function of these enzymes has been performed.

The best-characterized F₄₂₀H₂-dependent reductases of this superfamily are the hydride transferases involved in the biodegradation of the explosive picrate and related compounds (54, 155). In Rhodococcus opacus, two LLHTs known as hydride transferase I (HTI) and hydride transferase II (HTII) catalyze the reduction of picrate into hydride-Meisenheimer and dihydride-Meisenheimer complexes (430–432). Subsequent tautomerization, nitrite elimination, reduction, and hydrolysis steps lead to the production of 4,6-dinitrohexanoate, which can then be oxidatively degraded (432). The complete pathway involved is shown in Fig. 19. This pathway enables such organisms to grow using picrate and related compounds as the sole carbon and nitrogen sources (396, 398). The genes encoding the hydride transferases are organized in an operon together with genes encoding other enzymes in the pathway, including Fno which supplies reductant to the pathway (155, 433). Consistent with these genes having a physiological role in the biodegradation of nitroaromatic compounds, the repressor NpdR usually silences these genes, but it is inactivated in the presence of nitroaromatics (434).

FIG 19.

F₄₂₀-dependent degradation of picrate. (a) Genetic determinants of picrate degradation in Rhodococcus opacus. F₄₂₀-utilizing oxidoreductases are highlighted in gray, namely, two luciferase-like hydride transferases (HTI and HTII) and the F₄₂₀-reducing NADPH dehydrogenase (Fno) (155). Translation of the operon is silenced by the transcription factor NpdR, which is inactivated in the presence of nitroaromatic compounds (434). (b) Mechanism of picrate and 2,4-dinitrophenolate mobilization by Rhodococcus opacus. Hydride transfer from F₄₂₀H₂ is catalyzed by HTI and HTII, while F₄₂₀H₂ is regenerated by the F₄₂₀-reducing NADPH dehydrogenase Fno. The combined action of these enzymes generate hydride-Meisenheimer complex (compound 1 [shown as boldface 1 in the figure]) and dihydride-Meisenheimer complex (compound 2) of picrate and hydride-Meisenheimer complex (compound 3) and dihydride-Meisenheimer complex (compound 4) of 2,4-dinitrophenolate (394, 422).

The hydride transferases that mediate these reactions share approximately 30% amino acid sequence identity with Mtd of methanogens (435). The results of comparative genomics suggest that homologs of these proteins are exclusively encoded by the genera Nocardioides, Rhodococcus, and Nocardia among presently sequenced organisms. Empirical studies consistently indicate that equivalent enzymatic pathways can degrade nitroaromatic compounds in five additional Rhodococcus species (398, 436–438) and three Nocardioides species (54, 396, 397, 432, 439). Beyond picrate and 2,4-dinitrophenol, LLHTs are involved in the biodegradation of other nitroaromatic compounds. We recently reported a Nocardioides strain that is able to mineralize 2,4-dinitroanisole (DNAN) through an initial O-demethylation step (catalyzed by a novel hydrolase) followed by degradation of the resultant 2,4-dinitrophenol by LLHTs (397). 2,4,6-Trinitrotoluene (TNT) can also be initially reduced to an equivalent hydride-Meisenheimer complex in Rhodococcus and Mycobacterium strains (440, 441); however, this is unproductive, as the complex cannot be further metabolized to yield carbon or nitrogen sources (441).

5. APPLICATIONS AND IMPLICATIONS

5.1. Tuberculosis Treatment

Globally, tuberculosis (TB) is the most significant bacterial disease in terms of morbidity and mortality, infecting approximately 2 billion individuals and causing approximately 1.5 million deaths in 2013 (442). The standard treatment for tuberculosis relies on a 6-month, four-drug combination therapy (isoniazid, rifampin, pyrazinamide, and ethambutol) (443). There are major issues with this therapy: high cost per patient, poor compliance and management, growing worldwide drug resistance, and extensive drug-drug interactions (444). These problems are a reflection of the extraordinary biology of M. tuberculosis, which can transition between chronic and latent infection states that can evade the immune system and resist drug treatment (56), necessitating potent drug regimens to eliminate all tubercle bacilli from infected patients. There is thus an urgent need to develop new antimycobacterials to supplement or replace the current first-line drugs. F₄₂₀ is implicated in the abilities of M. tuberculosis to maintain nonreplicating persistent states and resist antibiotic treatment, oxidative stress, and nitrosative stress (section 4.1). Hence, there may be particular promise in developing small-molecule inhibitors of F₄₂₀ biosynthesis and enzymatic pathways in order to target persistent mycobacteria. The pleiotropic importance of F₄₂₀ in M. tuberculosis (31, 32), combined with its absence from human cells and commensal microflora, suggest that a specific inhibitor would be highly potent while having few off-target effects. Such an inhibitor is likely to have a synergistic effect if used with existing drug regimens (with the exception of nitroimidazole prodrugs that require F₄₂₀H₂ for activation [33, 34]), given that strains unable to synthesize F₄₂₀ are hypersusceptible to first-line and second-line antimycobacterials (32, 363). There are opportunities to use our knowledge of the F₄₂₀ biosynthesis pathways for fragment-based drug screening and structure-based drug design (445), although no significant progress has been reported in this area thus far. The F₄₂₀ system might also be exploited for the treatment of other serious mycobacterial diseases (145), for example those caused by M. bovis, M. ulcerans, M. marinum, and M. leprae (360).

However, there may be even more promise in exploiting the F₄₂₀ system to activate prodrugs. Delamanid (OPC-67683; approved for multidrug-resistant TB [MDR-TB]) (446), pretomanid (PA-824; phase III clinical trials) (33), and the next-generation TBA-354 (phase I clinical trials) (447, 448) are recently developed nitroimidazole prodrugs that are activated by hydride transfer from F₄₂₀H₂ (Fig. 20). These compounds have been shown to inhibit M. tuberculosis growth at submicromolar levels and exhibit no cross-resistance with current clinical drugs in vitro due to their novel mode of action (33, 34, 446, 449–451). In particular, delamanid shows great promise in the treatment of multi- and extensively drug-resistant TB (MDR-TB and XDR-TB, respectively) (452–454), while combination therapies that incorporate pretomanid exhibited highly promising 14-day bactericidal activity with minimal side effects (455, 456). The mechanism of activation of these prodrugs has been studied primarily with pretomanid (Fig. 20). A member of the FDOR-A1 family (28, 30), rv3547 (deazaflavin-dependent nitroreductase [Ddn]), mediates the hydride transfer from F₄₂₀H₂ to the nitroimidazole (35, 164, 457). Hydride addition leads to the formation of an unstable intermediate, which decomposes into three primary metabolites (predominantly a des-nitro compound) (33–35). During the decomposition, the nitro group is eliminated, resulting in accumulation of reactive nitrogen species (nitric oxide, nitrous acid) in a dose-dependent manner (34, 458). Transcriptome profiling indicates that the prodrug has a dual bactericidal mode of action as a result of the products formed (33, 459): the primary decomposition products prevent mycolic acid biosynthesis (possibly by inhibiting fHMAD [365, 446]), while reactive nitrogen species (RNS) release causes respiratory poisoning (34). Other mycobacteria are thought to be resistant to pretomanid because they either lack homologs of the activating enzyme rv3547 (i.e., M. leprae) (460) or carry genes that encode homologous enzymes with mutations in the nitroimidazole-binding site (e.g., M. smegmatis) (30, 164).

FIG 20.

Reductive activation and mode of action of the prodrug pretomanid by the F₄₂₀H₂-dependent reductase rv3547 (FDOR-A1) (33, 34, 365).

There are, however, concerns that M. tuberculosis will rapidly develop resistance against nitroimidazoles (35, 461). Point mutations in Ddn may be able to prevent pretomanid activation without inhibiting the protein's native quinone reductase activity (30, 32). Likewise, loss of function of rv3547, fbiC, or fgd result in cross-resistance to delamanid and pretomanid (458). In the clinic, it was recently reported that an XRD-TB patient rapidly acquired delamanid resistance through loss of function of the F₄₂₀ system (462). Interestingly, the original lead nitroimidazole compound for combating M. tuberculosis, CGI-17341 (now abandoned due to safety concerns) (463), depends on the presence of F₄₂₀ but not Ddn for antimicrobial activity (458). As CGI-17341 lacks the hydrophobic tail and phenyloxazole residues of delamanid and pretomanid, it is likely to be activated by a wider range of FDORs (458). It may therefore be possible to develop next-generation nitroimidazoles that are broadly activated by FDORs and hence will have more promising antimicrobial resistance profiles.

5.2. Methane Mitigation

Methane is the second most important anthropogenic greenhouse gas and contributes to about 20% of total anthropogenic climate forcing. Approximately 70% of methane emissions result from the activity of methanogens, the abundance of which has increased as a result of ruminant animal farming, rice paddy agriculture, and solid and liquid waste production (464). As a dominant catabolic cofactor in methanogens, as well as a central mediator in hydrogenotrophic, formatotrophic, and methylotrophic methanogenesis, F₄₂₀ facilitates these emissions. One strategy targeted at reducing methane emissions from ruminant animals and rice paddy fields is to administer methanogen inhibitors (465–467). Economical methanogenesis inhibitors may be particularly attractive in livestock agriculture, as they may simultaneously reduce greenhouse gas emissions while enhancing ruminant productivity (468). Highlighting the potential in this area, a recent study demonstrated that administration of the methyl-CoM reductase inhibitor 3-nitrooxypropanol to dairy cattle feed decreased methane production and increased body weight gain (468, 469). Other highly promising targets for methane mitigation include the F₄₂₀ biosynthesis enzymes CofG/CofH and oxidoreductase Mer, given their presence and predicted essentiality in all methanogens, including obligately aceticlastic species (196). Given that these targets are absent from host cells and other ruminal microbiota (where ANME are not competitive), specific inhibitors are likely to have minimal off-target effects.

5.3. Bioremediation

Many F₄₂₀H₂-dependent reductases have broad substrate specificity and can reductively degrade diverse xenobiotic compounds. For example, mycobacterial flavin/deazaflavin oxidoreductases can degrade coumarin derivatives (28, 55), while rhodococcal luciferase-like hydride transferases can reduce polynitroaromatic compounds (438, 440). While the physiological advantage conferred by this promiscuity has not been fully resolved, it does provide a basis for the exploitation of F₄₂₀ in bioremediation applications (470). It may be possible to deploy F₄₂₀-dependent organisms to remediate lands and waters contaminated with toxins and explosives. The most significant environmental contaminants that may be remediated through F₄₂₀-dependent processes are picrate, aflatoxins, and dyes such as malachite green (Fig. 21). Among the most carcinogenic and hepatotoxic compounds known, aflatoxins are a group of mycotoxins produced by Aspergillus flavus and Aspergillus parasiticus that contaminate crops in tropical climates (471, 472). As coumarin derivatives, difurocoumarocyclopentenones (aflatoxins B1 and B2) and difurocoumarolactones (aflatoxins G1 and G2) can be efficiently degraded by mycobacterial F₄₂₀H₂-dependent reductases (28, 409). Rhodococcus erythropolis and Nocardia corynebacterioides can also degrade aflatoxin, possibly through using homologous enzymes (472–474). Environmental mycobacteria are also capable of decolorizing and detoxifying malachite green in an F₄₂₀-dependent manner (132, 475); while once extensively used as an antiparasitic in aquaculture, this compound has since become regulated against due to its toxicological properties (476).

FIG 21.

Chemical structures of xenobiotics reduced by actinobacterial F₄₂₀H₂-dependent reductases of the FDOR and LLHT superfamilies. The structures shown are of carcinogenic aflatoxins (aflatoxin G₁ [AFG₁], AFG₂, AFGB₁, and AFGB₂) (28), nitroaromatic explosives (picrate, 2,4-dinitrophenol, 2,4,6-trinitrotoluene, and 2,4-dinitroanisole) (396), and the toxin malachite green (132).

Picrate and related nitroaromatic compounds are highly toxic explosives that extensively contaminate soils in current and former explosive manufacturing, processing, and storage facilities (477). Luciferase-like hydride transferases from certain actinomycetes can initiate mineralization of such compounds (438, 440). In the case of 2,4,6-trinitrotoluene (TNT), hydride transfer from LLHTs lead to the formation of dead-end products that cannot be further degraded (440, 441). However, multiple strains of Rhodococcus, Nocardia, and Nocardioides can completely mineralize picrate, 2,4-dinitrophenol (DNP), and 2,4-dinitroanisole (DNAP) as the sole carbon and nitrogen sources (396, 397). Administration of such bacteria to nitroaromatic-contaminated sites may be a cheaper and faster alternative to traditional physical remediation methods (477, 478). Consistently, there are reports of Rhodococcus sp. strain NJUST16 being used to biodegrade picrate from contaminated soils (437). As with bioremediation of aflatoxins and malachite green, administration of live bacteria is a more promising option than cell-free enzymatic systems, because F₄₂₀ must be enzymatically reduced before it is utilized by F₄₂₀H₂-dependent reductases.

5.4. Industrial Biocatalysis

F₄₂₀ may also prove a useful addition to the toolboxes of synthetic chemists. F₄₂₀-dependent processes already provide essential steps in some industrial processes, for example in the synthesis of some of the oldest-known antibiotic classes (29, 381), and there is considerable potential to expand the role of F₄₂₀-dependent enzymes as catalysts for synthetic chemistry. F₄₂₀H₂-dependent reductases of the FDOR and LLHT superfamilies can catalyze the stereospecific reduction of enones (28, 55, 291, 409) and imines (50, 388, 393) in diverse heterocycles. The broad substrate range of these enzymes may be particularly useful for catalyzing hydride addition to nonnatural compounds in a potentially stereospecific manner (479, 480). Such enzymes may be particularly useful in whole-cell biosynthetic cascades if coexpressed with cofactor recycling systems. A promising precedent in this regard is provided by the use of old yellow enzymes (OYEs) for the asymmetric reduction of enone moieties in yeast and bacteria (481). OYEs are mechanistically predisposed to trans-hydrogenation, whereby a hydride is delivered to the substrate from the cofactor and a proton is delivered to the opposite face of the substrate from an active site tyrosine (482). As F₄₂₀H₂-dependent reductases deliver hydrides from the cofactor, it is likely that they will provide access to cis-hydrogenation of enones for biocatalytic processes (including in vivo). Asymmetric imine reduction by enzymes is a promising area for development (483), not least because of the prominence of chiral amines in modern synthetic chemistry: ∼40% of pharmaceuticals and ∼20% of agrochemicals contain at least one chiral amine (484). However, the toolbox of enzymes available for use in such applications is still small and incomplete; there are few enzymes that will reduce a prochiral imine in a linear molecule, for example (483). The capacity of F₄₂₀-dependent enzymes to catalyze such imine reductions has, as yet, been explored only superficially (470).

A significant barrier to industrial application of F₄₂₀-dependent enzymes in biocatalytic applications is the commercial unavailability of F₄₂₀. While total chemical synthesis has been achieved (485), the most efficient and affordable way to obtain the cofactor is presently through extraction from F₄₂₀ producers. Most laboratory-scale preparations of the cofactor currently rely on Mycobacterium smegmatis, a safe “fast”-growing aerobic bacterium that synthesizes micromolar quantities of F₄₂₀ during fermenter growth (96). Bashiri et al. (486) were able to enhance F₄₂₀ production in this organism by overexpressing the fbiABC genes in trans and inducing F₄₂₀ production in a rich autoinduction medium. F₄₂₀ can subsequently be purified from lysed cells by anion-exchange chromatography, followed by hydrophobic-interaction chromatography (96, 486). In the long-term, it would be preferable to metabolically engineer large-scale recombinant F₄₂₀ production in Escherichia coli; however, this depends on the identification of the elusive enzyme responsible for production of 2-phospho-l-lactate (470). The capacity to produce F₄₂₀ in heterologous organisms that do not naturally produce or use the cofactor also raises some interesting possibilities for synthetic biology. “Exotic” cofactors may enable wholly orthogonal synthetic pathways for chemical production in an organism, essentially divorcing the pathway from the central metabolic and regulatory background of the production organism.

6. CONCLUDING REMARKS

On first inspection, it seems surprising that 5-deazaflavins are involved in such disparate processes; very little seems to unify methanogenesis, tetracycline biosynthesis, and DNA photoreactivation other than this class of compounds. Underlying the selection of 5-deazaflavins across biology, however, are the unique properties conferred by the N-5 (flavin) to C-5 (deazaflavin) substitution. The photochemical properties of 5-deazaflavins are crucial for the role of F_o in light capturing and FRET. The electrochemical properties of F₄₂₀ place it at the center of methanogenic redox metabolism and provide actinobacteria with a way of catalyzing low-potential hydride transfer reactions in their primary and secondary metabolism. The enzymes that synthesize 5-deazaflavins share conserved sequences and folds, suggesting that they were either present in the last universal common ancestor or were laterally transferred between archaea and bacteria. However, oxidoreductases appear to have evolved the capacity to utilize F₄₂₀ on multiple occasions from related nicotinamide- or flavin-dependent proteins. Three types of F₄₂₀-binding sites are nevertheless conserved throughout biology, namely, those in FrhB-like, TIM barrel, and split β-barrel folds. Many F₄₂₀-dependent enzymes have a modular nature—as particularly evident in Frh, Fpo, and Fsr—suggesting that F₄₂₀ is versatile enough to be accommodated in a wide range of redox enzyme systems.

For the future, there are numerous opportunities to both explore and exploit F₄₂₀. While we have a relatively rich understanding of the physiology and biochemistry of F₄₂₀ in methanogenesis, there are still conundrums to solve, for example in relation to the structurally unresolved Ffd, Fpo, and Fsr enzymes. Our understanding of the roles of F₄₂₀ in actinobacteria is much less sophisticated, and there are multiple important questions to resolve. For example, why is F₄₂₀ required for mycobacterial persistence and antibiotic resistance? Why do mycobacteria encode such a multiplicity of FDORs and LLHTs? What are the primary roles of F₄₂₀ in the metabolism of streptomycetes and rhodococci? Looking at the bigger picture, it is still poorly understood how F₄₂₀ biosynthesis pathways have evolved and why F₄₂₀ is distributed in relatively few phyla. However, the finding that F₄₂₀ is likely to be synthesized by ANME, Chloroflexi, and Proteobacteria indicates that the cofactor may be more important in oxic and anoxic communities than previously anticipated. There is also an urgent need to understand the role of F₄₂₀ at the ecosystem level, particularly in relation to how F₄₂₀-dependent biodegradation processes influence the community structuring and chemical composition of soils. Fueled by the recent approval of delamanid for treatment of multidrug-resistant tuberculosis, there is also room to explore the application of F₄₂₀ for medical, environmental, and industrial purposes. Half a century since their discovery by the Wolfe laboratory, 5-deazaflavins continue to surprise biologists and chemists alike.