Discovery of Multiple Modified F₄₃₀ Coenzymes in Methanogens and Anaerobic Methanotrophic Archaea Suggests Possible New Roles for F₄₃₀ in Nature

Abstract

Methane is a potent greenhouse gas that is generated and consumed in anaerobic environments through the energy metabolism of methanogens and anaerobic methanotrophic archaea (ANME), respectively. Coenzyme F₄₃₀ is essential for methanogenesis, and a structural variant of F₄₃₀, 17²-methylthio-F₄₃₀ (F₄₃₀-2), is found in ANME and is presumably essential for the anaerobic oxidation of methane. Here we use liquid chromatography–high-resolution mass spectrometry to identify several new structural variants of F₄₃₀ in the cell extracts of selected methanogens and ANME. Methanocaldococcus jannaschii and Methanococcus maripaludis contain an F₄₃₀ variant (denoted F₄₃₀-3) that has an M⁺ of 1,009.2781. This mass increase of 103.9913 over that of F₄₃₀ corresponds to C₃H₄O₂S and is consistent with the addition of a 3-mercaptopropionate moiety bound as a thioether followed by a cyclization. The UV absorbance spectrum of F₄₃₀-3 was different from that of F₄₃₀ and instead matched that of an F₄₃₀ derivative where the 17³ keto moiety had been reduced. This is the first report of a modified F₄₃₀ in methanogens. In a search for F₄₃₀-2 and F₄₃₀-3 in other methanogens and ANME, we have identified a total of nine modified F₄₃₀ structures. One of these compounds may be an abiotic oxidative product of F₄₃₀, but the others represent naturally modified versions of F₄₃₀. This work indicates that F₄₃₀-related molecules have additional functions in nature and will inspire further research to determine the biochemical role(s) of these variants and the pathways involved in their biosynthesis.

INTRODUCTION

Methanogens are a diverse group of anaerobic archaeal organisms with an energy metabolism dependent on one-carbon biochemistry to reduce CO₂, CO, formate, methyl group-containing compounds, and/or acetate to generate methane (see Fig. S1 in the supplemental material) (1). Several specialized coenzymes are required for this process. One such molecule, coenzyme F₄₃₀ (Fig. 1), is a nickel-containing hydroporphinoid which is chemically most closely related to vitamin B₁₂ and siroheme. It was originally characterized as a prosthetic group of methyl coenzyme M (CoM) reductase (MCR), the terminal key enzyme in methanogenesis (2). MCR catalyzes the reduction of the methyl group of methyl-CoM to generate methane in the final step of methanogenesis. The mechanism and role of F₄₃₀ in this reaction are still unclear, but the Ni(I) center of F₄₃₀ is thought to initiate methane formation by one of two mechanisms. The first involves an organometallic methyl-Ni(III) intermediate (3, 4), and the second involves the formation of a methyl radical (5, 6). The structure of F₄₃₀ was determined on the basis of biosynthetic incorporation experiments, chemical stabilities, and nuclear magnetic resonance (NMR) spectroscopy (7–9).

FIG 1.

Structures of coenzyme F₄₃₀ and F₄₃₀-2 and the proposed structure of F₄₃₀-3 described in this work.

The anaerobic oxidation of methane (AOM) is a microbially mediated process which consumes approximately 90% of the methane produced in marine sediments and, hence, is of importance for the global budget of this greenhouse gas (10). The organisms carrying out this difficult biochemical reaction consist of anaerobic methanotrophic archaea (ANME) and sulfate-reducing bacteria found near methane seeps in the ocean, which together convert methane and sulfate to carbonate and hydrogen sulfide (11, 12). ANME can be classified into at least three groups which are phylogenetically related to methanogens (13–15). Group 1 ANME (ANME-1) are likely most closely related to the Methanomicrobiales, while ANME-2 and ANME-3 are affiliated with the Methanosarcinales. The chemically challenging activation of methane by ANME occurs via a reversal of the MCR-catalyzed step in methanogenesis (16). Metagenomic data have shown that ANME possess homologs of MCR as well as most genes normally associated with methanogenic archaea, indicating that these organisms oxidize methane using a pathway which is essentially methanogenesis in reverse (15, 17). A modified form of F₄₃₀, F₄₃₀-2 (Fig. 1), was identified in samples collected from cold methane seeps in the Black Sea enriched with ANME-1 (18). The structure of this modified F₄₃₀ was determined by mass spectrometry and NMR spectroscopy and was shown to contain a methylthio group at the 17² position (19).

Here, we describe the presence of a new modified F₄₃₀ in two Methanococcales organisms. This is the first report of a modified F₄₃₀ occurring in methanogens. Based on high-resolution mass spectral data coupled with traditional biochemical methods, we have assigned the structure shown in Fig. 1 (F₄₃₀-3). This discovery led us to expand our search for additional modified F₄₃₀ molecules in cell extracts from three Methanococcales species as well as extracts from samples containing ANME. Multiple F₄₃₀ variants were identified in each of the samples that we analyzed. One of the compounds is likely an abiotic oxidative product of F₄₃₀, but the others could not have been produced during the extraction procedure and therefore must be biologically relevant.

MATERIALS AND METHODS

Chemicals.

All chemicals and reagents were purchased from Sigma-Aldrich.

Source of methanogenic cells.

Methanocaldococcus jannaschii (JAL-1, DSM 2661) was grown on an inorganic salts medium as previously described (20), and cell pellets were stored at −80°C. Methanococcus maripaludis (S2, DSM 2067 [JJ]) was grown on mineral medium containing 10 mM sodium acetate, 2% Casamino Acids, and 1% vitamin mixture (McCA) as previously described (21) and was supplied by William B. Whitman. Methanococcus vannielii (DSM 1224 [SB]) was grown on a defined salts medium containing formate (22).

Source of samples of ANME.

Black Sea mat material was sampled from methane-derived microbial chimneys during RVPoseidon cruise 317-2 in August 2004 using the submersible Jago (44°46′N, 31°60′E). Microbial methane oxidation was maintained by a methane headspace and repeatedly exchanging the marine sulfate reducer medium at 4°C. The thermophilic AOM enrichment was derived from the Guaymas Basin (Gulf of California, Mexico; 27°00.437′N, 111°24.548′W). Samples were obtained during the RV Atlantis cruise AT15-56 in November and December 2009 with the submersible Alvin. Cored material was incubated at 50°C with marine seawater medium and a methane headspace. The samples were repeatedly diluted, and medium was exchanged when sulfide concentrations exceeded 12 mM until sediment-free cultures were retrieved. Samples were taken from the living cultures within the last year and pooled. Samples are dominated by ANME-1 and HotSeep-1 partner bacteria. The sample Caldera derives from methane-rich sediments of the Nile Deep Sea Fan (32°07′N, 28°12′E) and was sampled in 2003 during the RV l'Atalante (expedition NAUTINIL). The sample Hydrate Ridge was retrieved from seeps in the Cascadia Margin (northeast Pacific; 44°34.2′N, 125°08.7′W). Cored material was incubated at 20°C (Caldera) or 12°C (Hydrate Ridge) with marine seawater medium and a methane headspace, and due to proliferation and repeated dilution, the material has been sediment free for several years. All samples possessed a methane-dependent sulfide production of approximately 100 to 250 μmol liter⁻¹ sulfide per day before sampling.

Preparation of methanogen extracts.

Frozen cell pellets (1 g [wet weight]) were added to a sample tube and suspended in 5 ml water. Six milliliters of methanol was added to this suspension, and the sample was shaken to obtain an even suspension of material. The tube was then sealed, and the sample was heated with shaking for 15 min at 100°C. This heating ensured that all proteins were denatured so as to release all bound cofactors. After cooling, the samples were centrifuged and the pellets were reextracted in the same manner a second time. The resulting combined extracts were evaporated to 1 ml with a stream of nitrogen, and large molecules were removed from the sample using an Amicon Ultra 0.5-ml centrifugal filter with a molecular mass cutoff of 3 kDa. Filtrate was evaporated to 0.1 ml prior to analysis.

Preparation of extracts of ANME.

The dried lyophilized cells (100 mg) were suspended in 1 ml of water and mixed thoroughly. One and a half milliliters of methanol was added, and the sample was heated at 100°C for 5 min. The sample was then centrifuged, and the resulting extract was separated from the pellet. The pellet was reextracted by the same procedure. The resulting combined extracts were evaporated to 1 ml with a stream of nitrogen, and large molecules were removed from the sample using an Amicon Ultra 0.5-ml centrifugal filter with a molecular mass cutoff of 3 kDa. Filtrate was evaporated to 0.1 ml prior to analysis.

Liquid chromatography–high-resolution mass spectrometry (LC–HR-MS) analysis of cell extracts.

The cell extracts were analyzed directly on a Waters SYNAPT G2-S high-definition mass spectrometer connected to a Waters Acquity UPLC I-class system with an Acquity UPLC BEH C₁₈ column (2.1 mm by 75 mm; particle size, 1.7 μm; Waters). Solvent A was water with 0.1% formic acid, and solvent B was acetonitrile. The flow rate was 0.2 ml/min, and gradient elution was employed in the following manner (time [min], percent solvent B): (0.01, 5), (6, 15), (21, 35), and (23, 65). Ten microliters of sample was injected. The mass spectral data were collected in high-resolution MSe continuum mode (nonselective MS/MS acquisition mode). A lock spray scan (function 3) was collected every 20 s for calibration, and the lock spray analyte used was leucine-enkephalin. Parameters were a 2.8-kV capillary voltage, a 125°C source temperature, a 350°C desolvation temperature, a 35-V sampling cone, 50-liter/h cone gas flow, a 500-liter/h desolvation gas flow, and a 6-liter/h nebulizer gas flow. The collision energies for the low-energy scans (function 1) were 4 V and 2 V in the trap region and the transfer region, respectively. Collision energies for the high-energy scans (function 2) were ramped from 25 to 45 V in the trap region and 2 V in the transfer region. Data were analyzed using MassLynx software (Waters).

Isolation of F₄₃₀-3 from M. jannaschii.

A portion of the cell extract from M. jannaschii was purified by chromatographic separation using a Shimadzu Prominence ultrafast LC system equipped with a photodiode array detector (PDA) and a fluorescence detector. A strong anion-exchange column (8.0 by 75 mm; particle size, 12 μm; Shodex QA-825) was used at a flow rate of 1 ml/min, and the elution profile consisted of a 5-min water wash followed by a 40-min linear gradient to 2 M ammonium bicarbonate. The injection volume was 0.1 ml, and 1-ml fractions were collected. In this chromatographic system, F₄₃₀-3 eluted in a fraction (fraction 19) after the F₄₃₀ isomers, which were observed in a single fraction (fraction 16).

Liquid chromatography-UV/visible light spectrophotometry-mass spectrometry (LC-UV-MS) analysis of metabolites.

In order to simultaneously obtain the absorbance spectra and mass spectral data for coenzyme F₄₃₀ derivatives, analyses were performed using an Agilent 1200 Series liquid chromatograph with a Zorbax Eclipse XDB-C₁₈ column (4.6 by 50 mm; particle size, 1.8 μm; Agilent) equipped with a PDA detector interfaced with an AB Sciex 3200 Q TRAP electrospray ionization mass spectrometer. Fractions from the Shodex anion-exchange purification were evaporated 3 times with a stream of nitrogen gas to remove ammonium bicarbonate and resuspended in 0.1 ml water for LC-UV-MS analysis. Solvent A was water with 0.1% formic acid, and solvent B was methanol with 0.1% formic acid. The flow rate was 0.5 ml/min, and gradient elution was employed in the following manner: (time [min], percent solvent B): (0.01, 5), (10, 65), and (15, 65). Ten microliters of sample was injected. The mass spectral data were collected in positive mode, and electrospray ionization was employed at 4,500 V and a temperature of 400°C. The curtain gas was set at 35, and gas source 1 and gas source 2 were 60 and 50, respectively. Analyst software (MDS SCIEX; Applied Biosystems) was used for system operation and data processing.

Preparation of the methyl esters of F₄₃₀ and F₄₃₀-3.

The M. jannaschii cell extract (50 μl) was placed in a vial, and the water was removed by evaporation. The sample was dissolved in 200 μl of methanol and evaporated a second time to remove any remaining water. The sample was then placed in 200 μl of methanol with ∼0.2 mg of p-toluene sulfonic acid, and the solution was stored at room temperature in the dark overnight. The methanol was removed by evaporation, 100 μl of 0.1 M NaClO₄ was added, and the sample was extracted with chloroform (2 times with 300 μl each time). The combined extracts were dried by passing through a small amount of cotton, the chloroform was evaporated, and the extract was dissolved in 60 μl of 40% methanol for LC-MS analysis.

RESULTS

High-resolution mass spectral analysis of M. jannaschii extracts.

Analysis of M. jannaschii cell extracts showed five isomers of coenzyme F₄₃₀ eluting at 7.79, 8.08, 8.35, 8.46, and 8.76 min (Fig. 2A, middle). The relative ratios of the peaks were 5, 80, 100, 7, and 95, respectively, on the basis of the absorbance at 430 nm and the intensity of their M⁺ ions. The interconversion of F₄₃₀ to different, more thermodynamically stable isomers has been observed previously (23–25) and has been studied by empirical force field analysis (26). Each peak showed the same molecular ion M⁺ at 905.2867 (calculated mass = 905.2862 for C₄₂H₅₁N₆NiO₁₃⁺) and confirmed the presence of F₄₃₀ in the sample, as was expected. Each peak had the same isotopic signatures (Fig. 2B) with the following exact masses (and isotopomeric intensities): 905.2867 (100), 906.2927 (37), 907.2961 (40), 908.2865 (15), 909.2875 (7), and 910.2891 (2). The M⁺ ion is specifically observed as a result of the net single positive charge from the presence of Ni(II). The data are the same as those previously reported by others (27) and confirmed the presence of Ni in the molecule. The increase in the intensity of the M⁺ + 2 ion is indicative of Ni⁵⁸ and Ni⁶⁰ isotopes (Ni⁵⁸ = 57.9353 [68.27%], Ni⁶⁰ = 59.9308 [26.22%]). The doubly charged MH⁺² at 453.1464 was also observed for each of the isomers, further supporting the identity of coenzyme F₄₃₀. We have previously observed the same ions and isomers for F₄₃₀ isolated from Methanobacterium thermoautotrophicum ΔH (unpublished results).

FIG 2.

High-resolution mass spectral analysis of M. jannaschii cell extract. (A) Total ion chromatogram of M. jannaschii cell extract (top); extracted ion chromatogram of M⁺ of 905.286 for coenzyme F₄₃₀ showing isomers eluting at 7.79, 8.08, 8.35, 8.46, and 8.76 min (middle); extracted ion chromatogram of M⁺ of 1009.287 for F₄₃₀-3 (bottom). The peaks at 6.26, 7.37, and 7.52 min represent the different isomers of F₄₃₀-3. (B) Molecular ion isotope peaks for F₄₃₀. (C) Molecular ion isotope peaks for F₄₃₀-3.

In addition to the expected F₄₃₀ peaks, three peaks eluting at 6.26, 7.37, and 7.52 min at relative ratios of 6, 37, and 100 were also observed. These peaks contained an F₄₃₀ derivative that will be referred to as F₄₃₀-3. The most intense of these peaks, the one eluting at 7.52 min, was similar in intensity to the most intense F₄₃₀ peak (Fig. 2A, bottom). Each peak had the same molecular ion M⁺ at 1,009.2781 (calculated exact mass = 1,009.2794 for C₄₅H₅₅N₆NiO₁₅S⁺). Each peak also had the same isotopic signatures (Fig. 2C) with the following exact masses (and relative isotopomeric intensities): 1,009.2781 (100), 1,010.2847 (43), 1,011.2780 (46), 1,012.2786 (23), 1,013.2728 (11), and 1,014.2744 (4). The doubly charged MH⁺² at 505.1452 was also observed with the following isotope peaks (and relative isotopomeric intensities): 505.1452 (100), 505.6462 (51), 506.1425 (60), 506.6469 (25), 507.1407 (13), and 507.6426 (5). The calculated ratios for the six isotope peaks with nominal masses of 1,009 to 1,014 are 100, 52.9, 59.8, 28.4, 15.9, and 5.9, which is consistent with the observed isotope peaks. These same peaks for F₄₃₀-3 were also observed in negative ion mode at the same elution positions. The major ion observed was the (M⁺ − 2H)⁻ at 1,007.2595 (calculated exact mass = 1,007.2649 for C₄₅H₅₃N₆NiO₁₅S⁻) for each of the isomers and the (M⁺ -3H)⁻² at 503.1265 (calculated exact mass = 503.1288 for C₄₅H₅₂N₆NiO₁₅S²⁻) for the doubly negatively charged molecule.

Characterization of the coenzyme F₄₃₀ variant from M. jannaschii.

Coenzymes F₄₃₀ and F₄₃₀-3 from the M. jannaschii cell extract were partially purified by strong anion-exchange chromatography. F₄₃₀-3 eluted after F₄₃₀ in the chromatographic system. In order to obtain absorbance data simultaneously with mass spectral data, the fractions containing F₄₃₀ and F₄₃₀-3 were assayed by high-pressure liquid chromatography interfaced with a photodiode array detector and a mass spectrometer (LC-UV-MS). The UV absorbance spectrum of F₄₃₀ was as expected (Fig. 3, top), while the UV spectrum of the F₄₃₀-3 was surprisingly quite different (Fig. 3, bottom) and did not have a 430-nm absorbance band. The major isomer of F₄₃₀-3 showed absorbances at 224, 281, ∼330, and ∼365 nm, and the minor isomer showed absorbances at 209, 285, ∼317, and ∼373 nm. Each absorbance peak corresponded to the peaks in the F₄₃₀-3 1,009 M⁺ ion elution profile. Neither of these UV spectra is like any of the spectra for the eight different tetrahydro- and hexahydrocorphinoid Ni(II) complexes used to model the F₄₃₀ structure (28–30). The spectra are not consistent with a change in the α-axial ligands to the Ni(III), since such ligation is reported to affect the absorbance spectrum to only a small extent (31). The spectrum that we observed for F₄₃₀-3 was, however, comparable to the UV spectrum of the F₃₃₀ produced by the reduction of the exocyclic keto group at the 17³ position of the tetrahydrocorphin with borohydride (32). This reduction produces a 100-nm shift in the absorbance and a decrease in the absorbance intensity at the higher wavelength, similar to what we observed. We propose that the same hydroxyl is present in F₄₃₀-3, and the elimination of this π bond decreases the overall π-bond order. Reduction of the keto group, however, would increase the mass of the molecule by 2 mass units. Since this was not the case, another site of desaturation not conjugated to the absorbing system must be in the structure. We propose that this results from the introduction of the tetrahydrothiophene ring in F₄₃₀-3 as opposed to the presence of a linear 3-mercaptopropionate moiety.

FIG 3.

UV-visible spectra of F₄₃₀ (top) and F₄₃₀-3 (bottom) isolated from M. jannaschii.

The UV data, retention times, and mass spectral data all indicate that the F₄₃₀-3 from M. jannaschii is the structure depicted in Fig. 1. The difference in the masses between the M⁺ for F₄₃₀ at 905.2867 and the modified F₄₃₀-3 at 1,009.2781 is 103.9914, corresponding to C₃H₄O₂S (calculated exact mass = 103.9932). This represents the addition of either 2- or 3-mercaptopropionate to the F₄₃₀ structure. The addition of a carboxylic acid to the F₄₃₀ structure is consistent with the increased polarity of F₄₃₀-3, as evidenced by the retention time compared to that of F₄₃₀ on reverse-phase and anion-exchange columns. Confirmation of this addition was obtained by preparing a methyl ester derivative, which was found to contain six methyl groups with an M⁺ of 1,093 [1,009 + (6 × 14)]. This is 118 mass units larger than that observed for the pentamethylated F₄₃₀ M⁺ of 975 [905 + (5 × 14)], consistent with the addition of one more methyl group to F₄₃₀-3 compared to the number of methyl groups on F₄₃₀. On the basis of the observed exact mass of F₄₃₀-3, the mercaptopropionate moiety would have to be added to the F₄₃₀ structure not as an amide, ester, or thioester but as an insertion of the sulfur into a C—H bond. A precedent for this can be found in the structure of the first F₄₃₀-2 variant shown in Fig. 1, which was isolated from ANME-1 (19). In this structure, a CH₃S group is added to the 17² carbon. We propose that this is the case here for addition of the mercaptopropionate modification. The addition of such a group at 17², however, would not explain the change in the UV absorbance spectrum that was observed, as described above. Instead, this can be resolved by an intramolecular Claisen condensation of the introduced group with the carbonyl group at the 17³ position. This would produce the desired mass, and an alcohol in place of a carbonyl at the 17³ position would alter the UV to be consistent with what we observed (32). Additional support for the F₄₃₀-3 structure is the observation of a M − 104 fragment in the high-energy collision-induced dissociation (CID) mass spectral analysis, which corresponds to the loss of a thiirane-2-carboxylic acid moiety (Fig. 4). An equivalent fragment was not seen in the F₄₃₀ high-energy CID spectrum.

FIG 4.

Observed molecular ion and CID fragment ions for F₄₃₀-3. Analogous fragments were observed for F₄₃₀, except for the M⁺ − 104 fragment.

Identification of additional F₄₃₀ variants in methanogens and ANME.

Having found and characterized F₄₃₀-3, we then searched for the presence of this molecule in two other methanogens, Methanococcus maripaludis and Methanococcus vannielii, as well as in samples containing archaea of the ANME type. The samples containing a mixture of ANME-1 and ANME-2 were derived from two specific enrichments from the Black Sea (33). The ANME-1 sample was from the Guaymas Basin (Gulf of California). To date, enrichments from the Guaymas Basin are the only cultures of ANME-1 that proliferate under laboratory conditions (34). The samples of ANME-2 were derived from enrichments obtained from Hydrate Ridge (35) and Caldera (Mediterranean seep).

F₄₃₀-3 was present in M. maripaludis but absent from the other organisms that we analyzed (Table 1). In the process of our analysis, we consistently observed a wide assortment of additional compounds that contained Ni on the basis of the clear isotopic signature and had masses within 50 m/z units of the mass observed for F₄₃₀. When the UV data could be obtained, most of these additional molecules had absorbance spectra consistent with those for the F₄₃₀ core structure. The results of our search for additional F₄₃₀ variants are reported in an abbreviated format in Table 1 and with further details in Tables S1 to S5 in the supplemental material. We observed a total of 10 F₄₃₀-related coenzymes, including the canonical F₄₃₀ and the previously described F₄₃₀-2 coenzyme. The proposed structures (Fig. 5) were deduced from the high-resolution mass measurements and the assumption that they are all derived biosynthetically from F₄₃₀. Since F₄₃₀ has a wide distribution in deep sea sediments (27) and was detected in all the samples and since the new F₄₃₀ structures can be explained by biochemically feasible alterations of the original F₄₃₀ structure, it is very likely that all of these variants are in fact derived from F₄₃₀.

TABLE 1.

Occurrence of coenzyme F₄₃₀ and its variants in methanogens and ANME^a

Sample	Occurrence of the following variants:
	F430	F430-2	F430-3	F430-4	F430-5	F430-6	F430-7	F430-8	F430-9	F430-10
M. jannaschii	+++	−	+++	−	−	−	−	−	−	+
M. maripaludis	+++	−	+++	+++	++	−	−	++	−	−
M. vannielii	+++	−	−	++	+	−	−	−	−	−
Black Sea ANME-1 and ANME-2	+++	+++	−	+++	++	−	−	+	−	−
Guaymas Basin ANME-1	+++	+++	−	−	−	+	++	−	+	−
Caldera ANME-2	+++	+++	−	+++	++	−	−	−	−	−
Hydrate Ridge ANME-2	+++	+++	−	+++	++	−	−	−	−	−

^aThe corresponding structures are shown in Fig. 5. +, ++, and +++, the relative amount of each coenzyme detected; −, coenzyme not detected.

FIG 5.

Proposed structures for the F₄₃₀ variants based on mass spectral data and how they are chemically related to F₄₃₀. The red portions indicate the proposed structural modifications.

Coenzyme F₄₃₀-2 was not detected in the three methanogens but was detected in all ANME samples. 12,13-Didehydro-F₄₃₀ (Fig. 5, F₄₃₀-4) was detected in all samples except the M. jannaschii and Guaymas Basin samples. This molecule has been reported and is considered to be an oxidative degradation product of F₄₃₀ (24). However, since the samples in this study were all extracted by the same method yet this compound was not detected in all the samples, F₄₃₀-4 may not simply be produced by an air oxidation of F₄₃₀ but could be a functioning coenzyme in some of these organisms. Additional support for the possibility that F₄₃₀-4 is biologically relevant is that we did not detect a didehydro form of F₄₃₀-2 or F₄₃₀-3, even though they should also be susceptible to air oxidation, just as was observed with F₄₃₀.

An intriguing F₄₃₀ derivative that was detected in some samples of both methanogens and ANME has a mass of 859.2836, consistent with oxidative decarboxylation of one of the propionic acid side chains of F₄₃₀ to generate vinyl-F₄₃₀ (Fig. 5, F₄₃₀-5). There are three possible positions for the vinyl modification, which we cannot yet distinguish. A derivative of F₄₃₀-2 containing a vinyl side chain was also observed in the sample of ANME-1 from Guaymas Basin (Fig. 5, F₄₃₀-6). These F₄₃₀ derivatives with a vinyl group can be compared to heme, which contains two propionic acid-derived vinyl groups.

Another possible modified form of F₄₃₀-2 was detected in samples containing ANME-1. The mass spectral data for this molecule, designated F₄₃₀-7 (Fig. 5), is consistent with the presence of a methyl sulfoxide in place of the methyl thiol. We also detected various oxidized F₄₃₀s with the addition of either an extra keto group (F₄₃₀-8), a hydroxyl group (F₄₃₀-9), or two hydroxyl groups (F₄₃₀-10). We cannot yet confirm the position of these modifications; however, we propose that the modifications are present at the 17² or 17³ position.

DISCUSSION

Here we report the identification of several previously unknown modified coenzyme F₄₃₀ molecules in methanogens and ANME. Our ability to identify these compounds was based on two primary factors. First, Ni-containing F₄₃₀ molecules are readily identified in a complex mixture by mass spectrometry due to the unique isotopic signature and exact mass of Ni. Second, the Ni acts as a built-in single positive charge which allows the detection of molecular ions for the different F₄₃₀ molecules using low-energy electrospray ionization. The presence of this single positive charge greatly increases the sensitivity of detection. These two factors, coupled with high-resolution mass measurements and fragmentation information, have allowed us to propose structures to the compounds and possible biosynthetic pathways for their production. Although this work represents a special case of metabolite identification, it demonstrates the power of current analytical methods to obtain likely chemical structures with only trace amounts of metabolites.

Since the distribution of the modified F₄₃₀ coenzymes varies considerably among the organisms analyzed, while the canonical coenzyme F₄₃₀ is present in all the samples, it is very unlikely that the new F₄₃₀ variants result from abiotic transformations of F₄₃₀. Additionally, the chemistry required for all transformations, except the didehydro derivative (F₄₃₀-4), could not occur spontaneously. The presence of these modified coenzymes in these organisms indicates that they must be biologically relevant and brings up two important questions. First, what is the purpose of these modifications, and second, how are they generated biosynthetically?

The previously described F₄₃₀-2 was originally characterized from samples collected from the Black Sea (18). This modified F₄₃₀ was shown to function in an MCR homolog from ANME that presumably functions in vivo to oxidize methane instead of generate methane (16). This distinct functional role of F₄₃₀-2 may be the reason for the presence of the 17² methyl thiol group; however, the role of the modification has not been elucidated. A recent study suggested that F₄₃₀-2 was associated only with ANME-1 and not ANME-2 (36). However, we observed F₄₃₀-2 in both the Hydrate Ridge and Caldera samples, which contain only ANME-2. This indicates that F₄₃₀-2 and not F₄₃₀ is exclusively used by the MCR homolog to perform the first step in AOM in ANME-1 and ANME-2. Additionally, our studies indicate that the presence of F₄₃₀-2 can possibly be used as a marker for both groups of ANME.

Since the only known function of F₄₃₀ is as a coenzyme for MCR, the presence of multiple modified F₄₃₀ coenzymes is unexpected and suggests that F₄₃₀ may be involved in many other reactions. Another possibility is that the modifications of F₄₃₀ modulate the activity of MCR under various environmental conditions. We are currently carrying out studies to determine the protein that binds F₄₃₀-3 in M. jannaschii and M. maripaludis. Our current results indicate that F₄₃₀-3 is not associated with MCR. When the total cellular proteins present in M. jannaschii were fractionated via anion-exchange chromatography, F₄₃₀, but not F₄₃₀-3, was identified in the fraction containing MCR. F₄₃₀-3 was instead present in a fraction containing other proteins.

F₄₃₀-5 and F₄₃₀-6, which contain a vinyl side chain in place of a carboxylate, are reminiscent of heme. The vinyl modification allows the heme to bind to hydrophobic protein cores and, more importantly, act as a site for its covalent attachment to the protein through cysteine residues (37, 38). We are presently performing studies to determine whether this modification allows the covalent attachment of F₄₃₀ to MCR and/or other F₄₃₀-dependent enzymes that have not yet been discovered.

Model reactions (39) indicate that the Ni in F₄₃₀ can function in a manner analogous to Co in vitamin B₁₂-related coenzymes found in the methanogens (40). This indicates that enzymes with these modified F₄₃₀s as coenzymes have the potential to function in a wide assortment of different reactions, as seen with vitamin B₁₂ biochemistry (41). Thus, F₄₃₀ variants may be involved in methyl group transfer to CoM, as seen in the methanol:CoM-methyltransferase (42), or they may be involved in the formation or breakdown of acetate, as seen in acetyl coenzyme A synthase/decarboxylase (43). The possibilities are vast and will likely reveal new coenzyme F₄₃₀-dependent chemistry.

The biosyntheses of F₄₃₀-2 and F₄₃₀-3 likely involve similar chemistries. Thio methyl group biosynthesis is well-known in many different molecules. In the biosynthesis of caldariellaquinone in Sulfolobus spp. (44), both the sulfur and the methyl groups are derived from methionine but not as an intact subunit (45). The enzyme(s) that catalyzes this reaction has not yet been described. Other examples involve enzymes belonging to the radical S-adenosyl-l-methionine (SAM) superfamily termed methylthiotransferases (MTTases). Radical SAM enzymes use a reduced [Fe-4S]⁺¹ cluster (cluster I) and SAM to generate a 5′-deoxyadenosyl radical (Ado-CH₂·) (46). Ado-CH₂· then abstracts a hydrogen atom from a generally unreactive site to activate a substrate for further chemistry. MTTases, exemplified by MiaB and RimO, catalyze the generation of an —SCH₃ group in place of a hydrogen on either tRNA or a ribosomal protein substrate, respectively, resulting in the formation of a methyl thioether (47). The source of the sulfur for this reaction is still unclear, but it has been proposed to arise from a sulfide coordinated by an auxiliary [4Fe-4S] cluster (cluster II) (48, 49). The methyl group is derived from one of the two SAM molecules required for the reaction (47). A series of reactions analogous to the chemistry performed by radical SAM MTTases provides an excellent mechanism for the biosynthesis of F₄₃₀-2, as shown in Fig. 6 (48). In the first part of the reaction, the methyl group of SAM is transferred to a sulfur coordinated by the [4Fe-4S] cluster II. A radical is then formed at the 17² position of F₄₃₀ via hydrogen atom abstraction with typical radical SAM chemistry. The substrate radical attacks the sulfur of the methylated sulfur atom to generate 17²-methylthio-F₄₃₀.

FIG 6.

Possible pathway for the biosynthesis of F₄₃₀-2. SAH, S-adenosylhomocysteine.

Biosynthesis of F₄₃₀-3 likely also involves radical SAM-dependent chemistry and may also require two molecules of SAM, as proposed for F₄₃₀-2 biosynthesis. In our first proposed biosynthetic scheme (Fig. 7A), sulfide bound by the [4Fe-4S] cluster II attacks the C-4 carbon of SAM, resulting in the transfer of the aminocarboxypropyl (ACP) portion of SAM and yielding a homocysteine bound to the [4Fe-4S] cluster II. This use of the ACP portion of SAM through the exploitation of the electrophilicity of the C-4 carbon has precedent in N-acylhomoserine lactone production and in tRNA modification (50). Next, a radical SAM-generated 17² radical attacks the sulfur of the homocysteine to yield an F₄₃₀ intermediate containing a homocysteine in place of the thiomethyl group seen for F₄₃₀-2. The amino acid could then be converted to the keto acid through a transamination reaction followed by an oxidative decarboxylation and an intramolecular Claisen condensation to produce F₄₃₀-3 (Fig. 7A). Alternatively, sulfur insertion may be the first step in the biosynthesis of F₄₃₀-3 (not shown here), catalyzed by an enzyme similar to the radical SAM sulfur insertion enzymes biotin synthase and lipoyl synthase (51, 52). If this were the case, the newly added sulfide at the 17² position would react with SAM and the remaining biosynthetic scheme in Fig. 7A would proceed and would be catalyzed by additional enzymes. A more straightforward mechanism in which the 17² radical reacts directly with 3-mercaptopropionate coordinated to the [4Fe-4S] cluster II followed by a Claisen condensation to produce the final F₄₃₀-3 is also possible (Fig. 7B). 3-Mercaptopropionate is highly prevalent in anoxic marine environments and is synthesized from methionine, homocysteine, or acrylate and dimethyl sulfide (53, 54). If any of these proposed mechanisms is occurring, the enzyme catalyzing this reaction is likely to harbor two [4Fe-4S] clusters, one for radical SAM chemistry (cluster I) and the other to bind the sulfur-containing substrate or provide the sulfur for the reaction (cluster II).

FIG 7.

Possible pathways for the biosynthesis of F₄₃₀-3.

F₄₃₀-5 and F₄₃₀-6, which contain a vinyl side chain, are likely derived from the oxidative decarboxylation of the propionic acid side chain via a mechanism similar to that for the well-characterized radical SAM enzyme anaerobic coproporphyrinogen III oxidase (HemN) (55).

The identification of these modified F₄₃₀s in methanogens and ANME opens up exciting new areas for future investigations. Coenzyme F₄₃₀ may be involved in several biochemical reactions that await characterization, and the reactions/enzymes required for generation of the modifications will reveal unique biochemistry. Further work will also be necessary to analyze a variety of other methanogens for the presence of these and other modified F₄₃₀ coenzymes. Importantly, this work further strengthens the biochemical similarity between methanogens and ANME.