RamA, a protein required for reductive activation of corrinoid-dependent methylamine methyltransferase reactions in methanogenic Archaea.

RAMA, A PROTEIN REQUIRED FOR REDUCTIVE ACTIVATION OF CORRINOID-DEPENDENT METHYLAMINE METHYLTRANSFERASE REACTIONS IN METHANOGENIC ARCHAEA Tsuneo Ferguson, Jitesh A. Soares Tanja Lienard, Gerhard Gottschalk, and Joseph A. Krzycki From the Department of Microbiology, Ohio State University, 484 West 12 Avenue, Columbus OH 43210; Göttingen Genomics Laboratory, Institute of Microbiology and Genetics of the Georg-August-University Göttingen, Grisebachstraße 8, D-37077 Göttingen, Germany; and The Ohio State Biochemistry Program, Ohio State University Running Head: Reductive activation of corrinoid-dependent methyltransferases Address correspondence to Joseph Krzycki, Department of Microbiology, Ohio State University, 484 West 12 Avenue, Columbus OH 43210, Tel. 614-292-1578; Fax 614-292-8120 E-Mail: Krzycki.1@osu.edu.


Archaeal
methane formation from methylamines is initiated by distinct methyltransferases with specificity for monomethylamine, dimethylamine or trimethylamine.
Each methylamine methyltransferase methylates a cognate corrinoid protein, which is subsequently demethylated by a second methyltransferase to form methyl-Coenzyme M, the direct methane precursor.
Methylation of the corrinoid protein requires reduction of the central cobalt to the highly reducing and nucleophilic Co(I) state. RamA, a 60 kDa monomeric iron-sulfur protein, was isolated from Methanosarcina barkeri and is required for in vitro ATPdependent reductive activation of methylamine:CoM methyltransfer from all three methylamines. In the absence of the methyltransferases, highly purified RamA was shown to mediate the ATP-dependent reductive activation of Co(II) corrinoid to the Co(I) state for the monomethylamine corrinoid protein, MtmC. The ramA gene is located near a cluster of genes required for monomethylamine methyltransferase activity, including MtbA, the methylamine specific CoM methylase and the pyl operon required for co-translational insertion of pyrrolysine into the active site of methylamine methyltransferases. RamA possesses a C-terminal ferredoxin-like domain capable of binding two tetranuclear iron-sulfur proteins. Mutliple ramA homologs were identified in genomes of methanogenic Archaea, often encoded near methyltrophic methyltransferase genes. RamA homologs are also encoded in a diverse selection of bacterial genomes, often located near genes for corrinoid-dependent methyltransferases. These results suggest that RamA mediates reductive activation of corrinoid proteins, and that it is the first functional archetype of COG3894, a family of redox proteins of unknown function.
Most methanogenic Archaea are capable of producing methane only from carbon dioxide. The Methanosarcinaceae are a notable exception as representatives are capable of methylotrophic methanogenesis from methylated amines, methylated thiols, or methanol. Methanogenesis from these substrates requires methylation of 2-mercaptoethanesulfonic acid (coenzyme M or CoM) that is subsequently used by methylreductase to generate methane and a mixed disulfide whose reduction leads to energy conservation (1)(2)(3)(4).
CoM methylation with methanol requires the methyltransferase MtaB and the corrinoid protein MtaC, which is then demethylated by another methylcobamide:CoM methyltransferase, MtaA (13-15). The methylation of CoM with methylated thiols such as dimethylsulfide in M. barkeri is catalyzed by a corrinoid protein that is methylated by dimethylsulfide and demethylated by CoM; but in this case an associated CoM methylase carries out both methylation reactions (16).
In Bacteria, analogous methyltransferase systems relying on small corrinoid proteins are used to achieve methylation of tetrahydrofolate (THF). In Methylobacterium spp., CmuA, a single methyltransferase with a corrinoid-binding domain, along with a separate pterin methylase, effect the methylation of THF with chloromethane (17,18). In Acetobacterium dehalogenans and Moorella thermoacetica various three-component systems exist for specific demethylation of different phenylmethylethers, such as vanillate (19) and veratrol (20), again for the methylation of THF. Sequencing of the genes encoding the corrinoid proteins central to the archaeal and bacterial methylotrophic pathways revealed they are close homologs. Furthermore, genes predicted to encode such corrinoid proteins and pterin methyltransferases are widespread in bacterial genomes, often without demonstrated metabolic function. All of these corrinoid proteins are similar to the well-characterized cobalamin-binding domain of methionine synthase (21,22).
In contrast, the TMA, DMA, MMA, and methanol methyltransferases are not homologous proteins. The methylamine methyltransferases do share the common distinction of having in-frame amber codons (6,8) within their encoding genes that corresponds to the genetically encoded amino acid pyrrolysine (23)(24)(25). Pyrrolysine has been proposed to act in presenting a methylammonium adduct to the central cobalt ion of the corrinoid protein for methyl transfer (3,23,26). However, nucleophilic attack on a methyl donor requires the central cobalt ion of a corrinioid cofactor is in the nucleophilic Co(I) state, rather than the inactive Co(II) state (27). Subsequent demethylation of the methyl-Co(III) corrinoid cofactor regenerates the nucleophilic Co(I) cofactor. The Co(I)/Co(II) in the cobalamin binding domain of methionine synthase has an E m value of -525 mV at pH 7.5 (28). It is likely to be similarly low in the homologous methyltrophic corrinoid proteins. These low redox potentials make the corrinoid cofactor subject to adventitious oxidation to the inactive Co(II) state (Fig 1).
During isolation, these corrinoid proteins are usually recovered in a mixture of Co(II) or hydroxy-Co(III) states. For in vitro studies, chemical reduction can maintain the corrinoid protein in the active Co(I) form. The methanol:CoM or the phenylmethylether:tetrahydrofolate methyltransferase systems can be activated in vitro by the addition of Ti(III) alone as an artificial reductant (14,19). In contrast, activation of the methylamine corrinoid proteins further requires the addition of methyl viologen as a redox mediator. Ti(III) reduces methyl viologen to the extremely low potential neutral species. In vitro activation with these agents does not require ATP (5,7,9). Cellular mechanisms also exist to achieve the reductive activation of corrinoid cofactors in methyltransferase systems. Activation of human methionine synthase involves reduction of the co(II)balamin by methionine synthase reductase (29), while the E. coli enzyme requires flavodoxin (30). The endergonic reduction is coupled with the exergonic methylation of the corrinoid with S-adenosylmethionine (27). A cellular activation system exists in cellular extracts of A. dehalogenans which can activate the veratrol:THF 3-component system and catalyse the direct reduction of the veratrolspecific corrinoid protein to the Co(I) state, however, the activating protein has not been purified (31).
For the methanogen methylamine and methanol methyltransferase systems, an activation process is readily detectable in cell extracts that is ATP and hydrogen dependent (32,33). Daas et al. examined the activation of the methanol methyltransferase system in M. barkeri, and purified in low yield a protein termed MAP which in the presence of a preparation of hydrogenase and uncharacterized proteins was required for ATP-dependent reductive activation of methanol:CoM methyl transfer (34,35). MAP was found to be heterodimeric protein without a UV-visible detectable prosthetic group. Unfortunately, no protein sequence has been reported for MAP, leaving the identity of the gene in question. The same MAP protein was also suggested to activate methylamine:CoM methyl transfer, but this suggestion was based on results with crude protein fractions containing many cellular proteins other than MAP (36).
Here we report of the identification and purification to near-homogeneity of RamA (reductive activation of methyltransfer, amines) a protein mediating activation of methylamine:CoM methyl transfer in a highly purified system (Fig.  1). Quite unlike MAP, which was reported to lack prosthetic groups, RamA is an iron-sulfur protein that can catalyze reduction of a corrinoid protein such as MtmC to the Co(I) state in an ATPdependent reaction (Fig. 1). Peptide mapping of RamA allowed identification of the gene encoding RamA and its homologs in the genomes of Methanosarcina spp. RamA belongs to COG3894, a group of uncharacterized metal-binding proteins found in a number of genomes. RamA thus provides a functional example for a family of proteins widespread among Bacteria and Archaea whose physiological role had been largely unknown.

EXPERIMENTAL PROCEDURES
Cell cultures and preparation of extracts----M. barkeri MS (DSM 800) was cultured under anaerobic conditions in a phosphate buffered medium supplemented with 80 mM MMA as described previously (37). Cell extracts were prepared as previously described (37).
Materials---Gases were purchased from Linde Specialty Gases (Columbus, Ohio) and passed through an R3-11 catalyst (Chemical Dynamics corp., South Plainfield, New Jersey) to remove trace O 2 contamination. DE-52 cellulose was purchased from Whatman LabSales (Hillsboro, Oregon) while Mono Q 10/10 and Phenyl-Sepharose HP cartridge were obtained from Pharmacia Biotech (Uppsala, Sweden). UNO Q-2 and Q1 and Bio-Scale CHT2-I hydroxyapatite and reagents for electrophoresis were purchased from Bio-Rad Laboratories (Hercules, California). Titanium (III) chloride (10% acidic solution), all other chromatographic materials, and all other reagents were purchased from Sigma-Aldrich Chemical Co., Inc. (St. Louis, MO).
Purification of methyltransferases and corrinoid proteins involved in MMA:CoM methyl transfer---MtmB, MtmC, and MtbA were either purified independently (5,37); or, MtmB and MtmC were co-purified aerobically from MMAgrown cells (23) in a purification scheme that also yielded MtbA. Cell extracts were applied to a DE-52 column (5 x 10 cm) equilibrated in 50 mM Tris HCl, pH 8, then eluted with a 2 l gradient of 50 to 500 mM NaCl in 50 mM Tris, pH 8. MtmB, MtmC, and MtbA copurified from the column in a total of 240 ml between 260 and 310 mM NaCl. This sample was then loaded onto a Q-Sepharose column (15 cm X 5 cm) equilibrated with 50 mM MOPS, pH 6.5, then eluted with a 1100 ml gradient with 100 to 500 mM NaCl the same buffer. MtbA eluted first in a 170 ml volume between 295 mM and 360 mM NaCl, followed by MtmC and MtmB, which eluted together in a 100 ml volume between 360 mM and 395 mM NaCl. This sample was then applied to a 400 ml (2.5 X 80 cm) Sephacryl S-100 column equilibrated in 50 mM MOPS, pH 7.0. MtmB and MtmC co-eluted after 200 mls in 20 ml total volume. MtbA was also subjected to this same column and eluted in 20 ml volume after 275 mls had eluted from the column. MtmB and MtmC were co-purified to near homogeneity with a Mono Q 10/10 chromatography column equilibrated in 50 mM MOPS, pH 6.5, then eluted using a 160 ml linear gradient from 100 mM NaCl to 500 mM NaCl in 50 mM MOPS, pH 7.0. The MtmB-MtmC fraction was then made anaerobic by flush evacuation, followed by pre-incubation with 5 mM titanium (III)-citrate prepared as described by Seefeldt and Ensign (38). Samples were then loaded onto a Sephacryl-100 gel filtration column (80 x 2.5 cm) which was preequilibrated in 50 mM MOPS, pH 7.0 and eluted with the same buffer.
MtmB and MtmC respectively eluted after 200 ml and 260 ml of elution with the same buffer. The MtbA fraction was further purified on hydroxyapatite as described in (39).
The components for the TMA:CoM (MttB and MttC) and for the DMA:CoM (MtbB and MttC) methyl transfer reactions were isolated as previously described (7,9).
Purification of RamA---MMA:CoM methylation reactions as catalyzed by MtmB, MtmC, and MtbA were used to locate peaks of RamA activity during column chromatography. Activity assays used to isolate RamA were conducted using stoppered 2.2 ml glass serum vials containing an atmosphere of H 2 . A typical 4 reaction contained 20 µg of purified MtmB, 10 µg of purified MtmC, 10 µg of purified MtbA, varying amounts of fractionated cell protein, 2 mM CoM, 100 mM MMA, 10 mM ATP, 20 mM MgCl 2 , 1.5 mM Ti(III)-citrate, and 1 mM bromoethanesulfonate in a total volume of 125 µl. Reactions lacking MtmB, MtmC, or MtbA served as controls that the column fractions contained an activating protein for the reaction, rather than one of these reaction components. After incubation at 37 °C, the remaining free thiol of unmethylated CoM was quantitated by removing 10 µl of reaction mixture and adding to 90 µl of 0.5 mM Ellman's reagent as described previously (37). This method has been previously been found to be an accurate measure of the MMA-dependent methylation of CoM in both whole and fractionated extracts of M. barkeri (5,37).
RamA was purified from MMA grown cellfree extract in a Coy anaerobic chamber containing 98% N 2 and 2% H 2 (Grass Lake, MI). The column matrices and buffers were made anaerobic by repeated cycles of evacuation followed by flushing with N 2 . Purification was initiated by applying 1040 ml (21.84 g total protein) of cell free extract onto a 5 x 50 cm DE-52 cellulose column equilibrated in 50 mM Tris, pH 8.0. A 4500 ml linear gradient of 50 mM to 500 mM NaCl in 50 mM Tris, pH 8.0, was applied to the column at 4 ml/min. RamA activity eluted in 680 ml (3.74 g total protein) between 280 mM and 300 mM NaCl. The pooled active fractions were concentrated, diluted 4.5-fold with 50 mM MOPS, pH 6.5, then loaded onto an 80 X 5 cm Q-Sepharose column equilibrated in 50 mM MOPS, pH 6.5. A 2400 ml linear gradient from 50 mM to 500 mM NaCl in the same buffer was applied at 4 ml/min. RamA activity eluted in 400 ml (800 mg total protein) between 320 mM and 400 mM NaCl. The RamA active pool was concentrated to 30 ml and two 15 ml samples were loaded onto an 80 X 2.5 cm Sepharose CL-6B column equilibrated in 50 mM MOPS, pH 7.0, containing 100 mM NaCl. RamA was eluted in 50 ml (670 mg total protein) after 360 ml of isocratic elution at 1 ml/min with the same buffer. The RamA activity was adjusted to pH 7.5 with 500 mM MOPS, pH 7.5 then loaded onto a 20 X 5 cm DEAE-Sepharose column equilibrated in 50 mM MOPS, pH 7.5 with 150 mM NaCl. A 2000 ml linear gradient from 150 mM to 450 mM NaCl in 50 mM MOPS, pH 7.5 was applied at 3 ml/min. RamA activity eluted between 245 mM and 308 mM NaCl. The fractions from 245 mM to 285 mM were chosen to minimize MtmC contamination that co-eluted with the later portion of the RamA activity. The collected fractions were concentrated from 260 ml (220 mg total protein) to 50 ml, then diluted with 100 ml of 50 mM MOPS, pH 6.5. Three aliquots were then successively purified on a Mono-Q HR 10/10 column. For each aliquot, a 120 ml gradient of 50 mM to 500 mM NaCl in 50 mM MOPS, pH 6.5 was applied at 2 ml/min. Three peaks of RamA activity eluted from each Mono-Q run between 195 mM and 220 mM, 270 mM and 275 mM, and 290 mM and 298 mM NaCl. The first peak in each case contained the highest levels of activity with the second and third peaks containing relatively little total activity (<10% in either peak). The major peak of RamA activity from each of the three Mono-Q runs were pooled (17 ml, 60 mg of total protein), then combined, and adjusted to pH 8.0 with 500 mM Tris-HCl. This RamA active fraction was loaded onto a BioRad Uno anion exchange column (two BioRad Uno Q-2 and two BioRad Uno Q-1 columns were connected in series, equally 6.6 ml of total matrix) equilibrated in 50 mM Tris, pH 8.0. After a 120 ml linear gradient from 50 mM to 300 mM NaCl in 50 mM Tris, pH 8.0 at 1.5 ml/min, RamA activity eluted as a single peak between 160 mM and 220 mM NaCl in 28 ml (33 mg protein). The RamA active fraction was divided into three aliquots and directly applied to a 1 ml BioRad CHT2-I hydroxyapatite column. The column was washed with 5 column volumes of 50 mM MOPS, pH 7.0 containing 5 mM potassium phosphate followed by a 30 ml gradient from 5 mM to 80 mM phosphate in 50 mM MOPS, pH 7.0 at 1 ml/min. RamA activity eluted in 12 ml (9 mg of total protein) between 40 mM and 53 mM phosphate. The RamA fractions from each of the three column runs were pooled and adjusted to 600 mM (NH 4 ) 2 SO 4 . This fraction was then loaded onto a Pharmacia Phenylsepharose HP cartridge (1 ml) equilibrated in 50 mM MOPS, pH 7.0 with 600 mM (NH 4 ) 2 SO 4. A 25 ml gradient from 600 mM to 0 mM ammonium sulfate and RamA eluted in a total of 6 ml between 240 mM and 320 mM ammonium sulfate.
ATP-dependent RamA activation of methylamine:CoM methyl transfer activity---Unless otherwise stated, the activity of purified RamA independent activation of MMA, DMA, or TMA methyltransferase reactions was performed by omitting RamA and MgATP and replacing them with 1 mM methylviologen as previously described (5,7,9). In each assay, CoM methylation was monitored by measuring the loss of the free thiol of CoM with Ellman's reagent, as described above.
The amino-terminus of RamA was determined by Edman degradation from 300 pmol of RamA in a liquid sample by the Protein Structure Lab at the University of California at Davis. RamA (10 µg) subjected to 12.5% denaturing polyacrylamide gel electrophoresis, yielded a single 60 kDa polypeptide on detection with Coomassie staining. The sequence of the polypeptide was determined as described previously (42) by digestion with chymotrypsin and trypsin followed by LC-MS/MS. UV-Visible spectroscopy of RamA was performed in anaerobic sealed vials under an atmosphere of 98 % N 2 and 2 % H 2 using a Beckman Instruments DU-70 Spectrophotomer using a 100 µl cuvette. Total Fe bound by RamA was determined by the o-phenanthroline method (43). Acid-labile sulfide was measured as described by Beinert (44) using purified Clostridium pasteuranium ferredoxin as a standard (kind gift of Richard Bare). Under these conditions Co(II)-MtmC was formed initially, and reduction to Co(I)-MtmC could be initiated by the addition of 0.12 mM RamA. The total reaction volume was 113 ml. The activity of RamA in the activation of MtmC and MtbC was followed by monitoring the increase in the Co(I) absorbance peak at 386 nm and a corresponding decrease in Co(II) absorbance at 475 nm. The Δε values at 386 nm and 475 nm were determined by the relating the differences in absorption at the two wavelengths of the respective absorption maxima to the concentration of MtmC.

RESULTS
Isolation of RamA-In previous work with the reconstituted MMA:CoM methyltransferase reaction, activity was completely dependent upon Ti(III)citrate and methyl viologen (6). Methyl viologen was postulated to act as a redox mediator which facilitated reduction of MtmC to the Co(I) state required for subsequent nucleophilic attack on the methyl group of MMA.
Reductive activation of the methyltransferase reaction with Ti(III)citrate and 6 methyl viologen is independent of ATP. However, soluble cell extracts of M. barkeri contained an activating protein or proteins which allowed methyl viologen-independent, but ATP-dependent, reductive activation of MMA:CoM methyltransferase activity by purified MtmB, MtmC, and MtbA with Ti(III)citrate. The activating activity present in the extract was estimated by titration, and it was found that under conditions where the activating fraction was rate limiting, the MtmB/MtmC/MtbA proteins in the assays were activated by 0.38 nmol CoM consumed/min for each mg of extract protein added to the reaction. The MMA:CoM methyltransferase activity in the added crude fraction itself was negligible, and activity in this assay was further dependent upon the addition of the three purified methyltransferase components, in addition to ATP.
In order to isolate the activating protein or proteins, the extract was fractionated by column chromatography, and fractions tested for the ability to activate MMA:CoM methyl transfer dependent by MtmB, MtmC, and MtbA in the presence of Ti(III)citrate and ATP. As described in Experimental Procedures, eight successive columns were required to achieve a preparation of a single 60 kDa polypeptide that was designated as RamA, and which was nearly homogeneous as judged by SDS PAGE and staining with Coomassie Brilliant Blue (Fig. 2). In the final preparation 3.8 mg RamA were obtained from 21.8 g of soluble protein in the starting cell extract. A complete duplication of the purification procedure yielded similar amounts and quality of the purified RamA protein.
Minimal requirements for RamA activated MMA:CoM methyl group transfer--We tested the requirements for the purified protein and components of the MMA:CoM methyltransferase reaction as reconstituted with RamA (Fig. 3). During purification of RamA, the column fractions were analyzed under a hydrogen atmosphere to allow for the possibility that hydrogenase might be required for activation. However, the reaction could take place under nitrogen in the absence of hydrogen (Fig. 3, panel A), but only if Ti(III)citrate was maintained. The activity was completely dependent on RamA addition, and under conditions where RamA was limiting, a rate of 56 nmol CoM consumed /min•mg RamA was observed, corresponding to 2.5% recovery of the initial activation activity in the starting material.
We found that, unlike activation with Ti(III)citrate and methyl viologen, activation with Ti(III) and RamA was absolutely dependent upon the addition of ATP. In addition, the reaction required the further addition of purified MtmB, MtmC, and MtmB, thereby indicating that purified RamA fraction did not contain detectable amounts of these active proteins. Previous studies conducted with the MAP protein required the addition of a primary DEAE fractionation of cell extracts containing many uncharacterized proteins for activation activity (34)(35)(36). In contrast, no such fraction was required for the RamA, MtmB, MtmC, and MtbA catalyzed MMA:CoM methyl transfer activity using the proteins at the level of purity shown in Fig. 2.
RamA affects both lag times and rates during assay of MMA:CoM methyl transfer-If RamA functioned as an activation protein, we posited that lag times to the steady-state rate would depend upon the amount of RamA added to reactions. Furthermore, if the activated protein was unstable, the steady-state pool of active protein (and thus the steady-state methyltransferase rate) would depend also on RamA concentration. Therefore, increasing amounts of RamA were added to reaction vials containing 1000 pmol MtmB, 500 pmol each of MtmC and MtbA, in addition to Ti(III)citrate and ATP (Fig. 4, panel A). RamA concentration affected the reaction rate under steady state conditions. Observed maximum steady-state reaction rates were reached when 25 pmol was added to the reaction, or 20 times less than the total amount of MtmC in the reaction. Even as little as 1 pmol RamA was capable of supporting a low rate of MMA:CoM methyltransferase activity (Fig. 4, panel B). With increasing RamA, lag times were observed to decrease until reaching a point they were not detectable, which occurred upon addition of 85 pmol of RamA to the reaction containing 1000 pmol of MtmB and 500 pmol each of MtmC and MtbA (Fig. 4, panel C).
RamA activates the TMA-and DMAdependent CoM methyltransfer reactions-The corrinoid proteins required for the TMA-, DMA-, and MMA-dependent CoM methyltransferase reactions are homologous, and additionally, are all demethylated by the same protein, MtbA, during CoM methylation (5)(6)(7)(8)(9). Like the MMA:CoM methyltransferase reaction, the 7 DMA-and TMA-dependent CoM methyltransferases required activation by Ti(III)citrate and methyl viologen. As observed with the MMA:CoM methyltransferase proteins, RamA and ATP could activate either DMA-or TMA-dependent CoM methylation with the purified methyltransferases specific for each reaction (Table I). The rates achieved using RamA and ATP were similar to those achieved with the ATP-independent chemical activation using methyl viologen as a redox mediator.
RamA mediates ATP-dependent reduction of MtmC-The effect of RamA concentration on MMA:CoM reaction rates and lag times is consistent with a catalytic role in activating the reaction components. That RamA and ATP replaced the need for a redox-mediating dye in activation of MMA:CoM methyltransfer suggested a role in reducing MtmC to the active Co(I) state. Therefore, purified RamA was tested for the ability to mediate ATP-dependent reduction of MtmC with electrons provided by Ti(III)citrate (Fig. 5). The redox state of MtmC was monitored by the characteristic UV-Visible spectra of the bound corrinoid cofactor. Ti(III)citrate alone was capable of reducing MtmC from the Co(III) to the Co(II) state, however, Co(I)-MtmC was not observed even after prolonged incubation with Ti(III)citrate (data not shown). Addition of ATP without RamA did not cause further reduction of the corrinoid protein, but upon addition of purified RamA reduction of Co(II)-MtmC occurred, as signified by decrease of the absorbance peak at 475 nm (Fig. 5). Coincident with loss of the Co(II) signature peak was formation of an intense absorbance band at 386 nm, typical of the Co(I) form of the bound cofactor. We determined the extinction coefficients at 386 and 475 nm for both the initial Co(II) state and the final Co(I) state after the reaction reached saturation. These agreed well with previously determined coefficients for other small methyltrophic corrinoid proteins (31), and the Δε were then used to estimate the rate of loss of the Co(II) substrate and formation of Co(I) product. Either extinction coefficient allowed estimation of the turnover rate as 1 min -1 at 23°C. This number is comparable with the apparent activation rate of 3 min -1 found at 37°C with the same RamA:MtmC ratio during activation of the MMA:CoM methyl transfer reaction (Fig 4).
Reduction of MtmC to the Co(I) state was not observed if ATP was not added to reactions that contained RamA. However, upon addition of ATP, reduction of Co(II)-MtmC to the Co(I) form occurred, as evidenced by the increase in absorbance at 386 nm (Fig. 6).
RamA is a monomeric iron-sulfur protein--The native molecular mass determinations of two different purified RamA preparations averaged 71 kDa when determined by gel filtration chromatography. As this is very near the mass of the denatured enzyme in SDS page, RamA appears to be predominantly a monomer in solution. A UV-Visible spectrum of the protein revealed a broad absorbance from 350 to 450 nm, indicative of the presence of iron-sulfur clusters (Fig. 7). No increase was observed in the as-isolated sample upon addition of potassium ferricyanide, suggesting the clusters were fully oxidized in the as-isolated form (not shown). Addition of dithionite to the protein resulted in bleaching in this region, indicating that these clusters were redox active and could be reduced in the absence of ATP. Analysis of purified RamA revealed 8.0 mol acid-labile sulfide and 6.5 mol Fe per mol monomer, suggesting the presence of two Fe 4 S 4 clusters or one Fe 4 S 4 cluster with an additional Fe 3 S 4 cluster.
RamA is encoded by a gene with a Cterminal ferredoxin-like domain-The Nterminal amino acid sequence of RamA was determined by Edman degradation to be MYGIALDLGTSGFRTQLIDLETKETLKTVIT MGHPLPGGN, and was used to identify possible ramA genes in sequenced Methanosarcina spp. genomes (Supplemental Fig. S1, Supplemental Table S1). Initially, the M. mazei genomic sequence (45) was examined, which led to identification of MM1440 (genbank NP_633464), whose predicted product differed from the N-terminus of the isolated protein by only 2 out of 40 residues. Examination of the M. barkeri Fusaro genome (46) similarly revealed another gene (Mbar_A0840, genbank AAZ69816) encoding a protein whose Nterminus was 95% identical to that determined for the isolated M. barkeri MS RamA protein.
To further confirm the identity of the RamA gene, we PCR amplified, cloned, and sequenced the gene directly from M. barkeri MS (also known as strain DSM800), the strain from which it was isolated (Genbank accession number FJ477063). Degenerate PCR primers were designed from the first eight residues of the Nterminal sequence of the isolated RamA protein 8 and the final seven residues from the M. barkeri Fusaro ramA sequence and used to obtain the M. barkeri MS ramA gene from genomic DNA. The N-terminus predicted for the sequenced PCR product was in complete agreement with all 40 residues determined for the N-terminal sequence for RamA as isolated from M. barkeri MS (Fig.  S1). The sequence predicts a product whose molecular mass is 59,046 Da. The C-terminus of the protein was found to be similar to Fe 4 S 4 ferredoxins (Fig S1). Two sequences which might binding tetranuclear iron-sulfur clusters were identified, one was identical to that found in many ferredoxins (CX 2 CX 2 CX 3 CP) whereas the second had an insertion of two additional residues between the first and second cysteine residues (CX 4 CX 2 CX 3 CP).
We confirmed that the sequenced M. barkeri MS ramA gene encoded the isolated RamA protein by MS-MS sequencing of peptides from chymotryptic and tryptic digests of the highly purified RamA protein (Tables S2 and S3). Peptides were found that covered 78% of the predicted product of the ramA gene (Fig. S2), and these peptides exactly matched the predicted sequence of the ramA gene product.
The M. barkeri MS ramA gene has multiple homologs in every Methanosarcinaceae genome whose sequence is available (Supplementary Table  1). Genes were identified in M. mazei (MM1440), M. barkeri Fusaro (Mbar_A0840), and M. acetivorans (MA0150) genomes whose predicted gene products are 93 to 97% similar to that produced by the entire length of ramA from M. barkeri MS. Fig. 8 shows the local gene organization surrounding the closest ramA homolog from M. barkeri Fusaro, whose product is overall 97% identical to the predicted M. barkeri MS RamA protein sequence. In M. barkeri Fusaro, the ramA gene is found sandwiched by a cluster of genes required for conversion of MMA to methyl-CoM. This cluster of genes includes those encoding the monomethylamine methyltransferase (mtmB), the MMA corrinoid protein (mtmC) and the methylamine specific cobamide:CoM methyltransferase (mtbA). Furthermore, nearby are also the pyl genes required for the biosynthesis and genetic encoding of pyrrolysine. Very similar gene arrangements surround the closest ramA homologs found in each of the other Methanosarcina spp. genomes that have been sequenced. Methanococcoides burtonii, a close relative of Methanosarcina spp., also had its most similar ramA homolog found adjacent to an ORF predicted to encode a methylamine methyltransferase, in this case it is the mttB gene encoding trimethylamine methyltransferase (Table S1).
Aside from the RamA homologs encoded near the methylamine gene clusters, another set of RamA homologs (designated here "RamM") were found that were approximately 95% similar to one another, yet approximately 65% similar to the actual RamA genes (Supplemental Table I).
The "RamM" genes were found in each Methanosarcina spp. genome adjacent to genes specific for methanol:CoM methylation. RamA homologs are widely distributed in genomes near corrinoid proteins or their associated methyltransferases. A wide number of genomes possess ramA homologs. Indeed, the methanosarcinal ram genes are part of a recognized group, COG3894, which is annotated as uncharacterized metal binding proteins (http:@www.ncbi.nlm.nih.gov/COG, (49)). We surveyed this group, as well as ran BLAST searches against completed archaeal and bacterial genomes for RamA homologs, using the database maintained by the National Center for Biotechnology Information. We then examined the adjacent genes for homologies and predicted function. Many genomes were found that possessed ramA homologs near genes encoding small corrinoid proteins or their associated methyltransferases (Table S4). The corrinoid proteins were most often homologous to MtmC, although in a few instances the adjoining genes encoded subunits of the corrinoid-iron sulfur proteins of acetyl-CoA 9 synthase/decarbonylase (50,51). The methyltransferases found encoded near RamA genes were often similar to the family of proteins comprised of methylcobamide:CoM methyltransferases, such as MtbA and CmuA. Many representatives in the gram-positive group Firmicutes and gram-negative alphaproteobacteria of ramA homologs fit this trend. Only one or two examples were found in the bacterial Chloroflexi and the Fibrobacter/Acidobacteria groups. Besides the Methanosarcinales, ramA homologs with associated corrinoid related genes were found in the crenarchaeaote Staphylothermus, and the euryarchaeaote Archaeoglobis fulgidis. Examples of ramA homologs were found in some genomes which were not apparently near corrinoid-related genes, however, these were far less prevalent than those near corrinoid proteins or their associated methyltransferases.
Alignments of most of the above ramA homologs found outside the Methanosarcinales revealed that only the first 400 residues of the methanosarcinal RamA protein are conserved ( Fig.  S3 and Table S4). The C-terminal region containing the tetranuclear iron-sulfur clusters (circa 130 residues) are not present in most examples found outside of Methanosarcina spp. and close relatives. However, the predicted bacterial and crenarchaeal proteins had instead a N-terminal domain with homology to cd00207, a family of proteins binding Fe 2 S 2 clusters (Fig. S3). Alignments of several of these proteins revealed that only the conserved cysteine residues for a single iron-sulfur cluster could be detected in the N-terminal domain (CX 5 CX 2 CX 16-44 C). All bacterial instances were found to possess only genes encoding homologs of RamA bearing the Fe 2 S 2 ferredoxin signature in the N-terminal domain.
Several crenarchaeaota and euryarchaetoae genomes have ram gene homologs whose predicted products have N-termini bearing an Fe 2 S 2 binding signature. Only in Methanosarcina spp., Methanosphaera stadtmannii, and Methanobrevibacter smithii genomes were ramA homologs found whose products are predicted to possess a C-terminal tetranuclear iron-sulfur domain.
Distant relationship of RamA to sugar kinases. A psi-BLAST search with residues 1-450 M. barkeri ramA gene product (that is, excluding the C-terminal ferredoxin-like domain) revealed a relationship to xylulose kinase from a number of different sources, a member of the FGGY family of carbohydrate kinases. These proteins share a conserved fold that binds and hydrolyses ATP (52). The best alignments following 14 th iterations of psi-BLAST possessed an expect value of <10e-124 and aligned nearly the entire length of the ramA product. The most highly conserved region between xylulose kinase and RamA were the first 25 residues (Fig. S4), which included a highly conserved aspartate residue (D-7 in the ramA gene product) that is part of the MgATP binding pocket (52).

DISCUSSION
RamA is a protein which mediates activation of methylamine:CoM methyl transfer reactions when reconstituted with highly purified proteins, and which has characteristics unlike any other such activation protein previously described. RamA replaces an otherwise obligate in vitro requirement in this system for a reduced low-potential redox dye, but only by the introduction of a new dependence of the reaction on ATP. Highly purified preparations of RamA increasingly shortened the lag time to steady state CoM methylation from MMA with purified MtmB, MtmC, and MtbA. These results were indicative of a catalytic role for RamA in the activation of the methyltransferase reaction. In keeping with this expectation, RamA mediates the reduction of purified MtmC from the Co(II) to the Co(I) form in reactions containing those two proteins at high levels of purity (Fig. 2) recognize multiple corrinoid proteins involved in the methylamine:CoM methyltransfer pathways, yet discriminate against others has precedent in the methylamine:CoM methyltransferase pathways, as MtbA, a methylcobamide:CoM methylase, will utilize the TMA, DMA, and MMA corrinoid proteins as substrates, but not the methanol corrinoid protein MtaC (10).
The gene encoding RamA reveals traits consistent with the protein's function in the reductive activation of methylamine corrinoid proteins. Methanogens have been noted to cluster genes required for methylamine metabolism on their genomes (6,53). The deduced primary sequence of the M. barkeri MS RamA protein is nearly identical to a gene within the methylamine cluster in the M. barkeri Fusaro, M. mazei, and M. acetivorans genome. It is notable that flanking each side of the ramA gene in these genomes are genes encoding proteins acting in all methylamine:CoM methyltransferase pathways. These genes include mtbA, encoding the CoM methylase that can be used in common with either the TMA, DMA, or MMA corrinoid proteins; and the pyl gene cluster, which is essential for biosynthesis and co-translational insertion of pyrrolysine into the TMA, DMA, and MMA methyltransferases (54). The deduced primary sequence of RamA also supports its involvement in the ATP-dependent reduction of corrinoid proteins. RamA possesses an N-terminal motif associated with the binding of MgATP found in xylulose kinase and associated proteins. Furthermore, RamA carries a redox-active ferredoxin-like domain at its C-terminus. The presence of two cysteine clusters within this domain that can both bind a tetranuclear ironsulfur cluster supports the interpretation that 2 Fe 4 S 4 clusters are present in the protein. However, the somewhat lower than expected iron bound to the isolated protein (6.5 rather than 8 Fe/monomer) may indicate one cluster could be present as a Fe 3 S 4 center. Further study will be required to resolve this issue.
All published genomes of the Methanosarcinacea have multiple copies of genes homologous to ramA. As in M. barkeri Fusaro, the ramA homolog most identical to M. barkeri MS ramA was found in the monomethylamine methyltransferase gene cluster in M. acetivorans and M. mazei. In Mc. burtonii the closest ramA homolog is near mttB, the TMA methyltransferase gene. In contrast, ramA homologs with less similarity to M. barkeri ramA are found near other methylotrophic corrinoid proteins (Table  S1). These observations suggest the possibility that these proteins might be isozymes that also act as corrinoid protein reductases, but with primary specificity for other methyltransferase systems. During isolation of RamA, some minor peaks of activation activity were observed to separate from the major peak of RamA activity. This may be due RamA complexes with methylamine corrinoid proteins, however, it is possible these minor activity peaks could represent the gene products of ramA homologs whose primary partner is another methyltransferase system, but which have crossreactivity with the methylamine corrinoid proteins.
The characteristics of the purified RamA protein suggest it is distinct from the MAP protein that was isolated from M. barkeri growing on methanol and which was suggested to be involved in the activation of the methanol:CoM methyltransferase (34,35). Unfortunately, a gene encoding MAP was never identified due to the limiting amounts of the isolated protein, so a direct comparison of the sequences of the two proteins is not possible. MAP and RamA were reported to have similar native masses, but MAP was reported to be a heterodimer of 60-kDa and 30-kDa subunits. We could not find a circa 30-kDa subunit in any of our purified RamA preparations (Fig. 2), although it may be present at concentrations that were far sub-stoichiometric to the RamA polypeptide and so undetectable in our Coomassie-stained SDS gels. Occasionally, a 31 kDa contaminant was present in our MtmC preparations (see, for example, Fig. 2), but this was lacking in many preparations of MtmC employed in this study (not shown), for example, those in which it was reduced upon the addition of purified RamA and ATP.
Most significantly, MAP lacked any chromogenic prosthetic group when examined by UV-Vis spectroscopy (35), suggesting the absence of the iron-sulfur clusters that are readily apparent in the UV-Vis spectrum of RamA. The amount of MAP isolated (40 µg) may have been simply too low to allow assessment of the iron-sulfur cluster feature in the spectrum, but it also cannot be excluded that MAP is a distinct protein from RamA protein. In this regard, it is notable that one of the ramA homologs in the sequenced genomes of the Methanosarcinaceace is found encoded adjacent to the methanol methyltransferase genes. This "RamM" homolog also has signatures for a Cterminal ferredoxin sequence. Although not diagnostic of function, this genomic context is suggestive that this homolog could act with the methanol corrinoid proteins, and that previous work with MAP might have simply missed the iron-sulfur clusters due to the low amount finally isolated as a pure protein. Unfortunately, the activities observed with MAP are not useful in resolving the question of the identity of RamA (or "RamM") and MAP. MAP activation with any methyltransferase could not observed except with the addition of portions of a primary DEAE fractionation of cell extract, indeed most experiments with MAP were conducted with only primary DEAE fractions from cell extracts that contained MAP (34). Even under this condition, MAP could not achieve the direct reduction of Co(II) to Co(I) corrinoid protein, but rather appeared to cause conversion of base-on Co(II) corrinoid protein to the base-off form.
Under the conditions of our in vitro assay, we employed Ti(III)citrate as an electron donor, which has a redox potential of -750 mV. In principle this low potential should not require additional energy input (such as ATP hydrolysis) for reduction of the corrinoid protein, even with a redox potential for the Co(II)/Co(I) corrinoid couple lower than -500 mV. However, as Ti(III)citrate cannot reduce the corrinoid protein without a redox dye as mediator, this suggests the role of RamA is essentially to replace that mediator, but also that RamA function obligately requires ATP. There are three possible general scenarios by which RamA might operate which could require ATP, and which can be tested in future experiments. Firstly, it is possible that RamA binds the corrinoid protein and in an ATPdependent manner which, while it does not cause a change in RamA redox potential, causes a conformation change of MtmC which allows its reduction by Ti(III)citrate (or a low-potential cellular reductant). That a low potential donor would be an obligate requirement for activation seems unlikely, since hydrogen is sufficient for reduction of the corrinoid protein to the Co(I) form in cell extracts supplemented with ATP. An alternative hypothesis is RamA may bind and/or hydrolyse ATP in such a way as to lower the redox potential of the reduced iron-sulfur clusters in the C-terminal ferredoxin domain, allowing these clusters to directly reduce MtmC. Such a scenario could involve RamA autophosphorylation, and/or a conformational change that exposes the clusters for electron transfer. This would allow higher potential electron donors in the cell to fuel the lower potential reduction of the corrinoid protein. This would be generally analogous to the strategy employed by benzoyl-CoA reductase, or 2hydroxyglutaryl-CoA dehydrase activating enzyme in carrying out low potential reductions (55)(56)(57). Finally, RamA may bind the corrinoid protein and in an ATP-dependent fashion raise the Co(II)/Co(I) potential in order to allow reduction of the corrinoid protein with a higher potential redox center, such as the C-terminal ferredoxin-like domain of RamA. Of these three hypothesis, the latter two seem most likely, as they would permit an ATP-dependent activation of the methyltransferase system under in vivo conditions where the redox potential could not achieve direct reduction of the corrinoid protein.
The widespread nature of genes encoding RamA homologs suggests that many members of COG3894 protein family has found utility among both bacteria and archaea. Our results with the methylamine methyltransferase system lead us to suggest that many of these proteins function in corrinoid-dependent methyltransferase activation, presumably by mediating the ATP-dependent reduction of corrinoid proteins. A large number of ramA homologs are found adjacent to genes homologous to those for methylotrophic corrinoid proteins and/or corrinoid/cobalamindependent methyltransferases. As many prokaryotes cluster genes used in similar physiological tasks, this common genomic context for ramA-like genes near corrinoiddependent methyltransferase genes is consistent with a role in the activation of such methyltransferases. It is striking that many of the bacterial RamA homologs are predicted to possess an N-terminal domain binding a single dinuclear iron-sulfur cluster, rather than the Cterminal dual iron-sulfur clusters found in methanosarcinal RamA. This suggests that the function of the protein family inevitably requires electron transfer, but that a single electron center is sufficient. This may be in keeping with a function of many members of this protein family in one-electron reduction, i.e. the formation of 12 Co(I) corrinoid cofactors from the inactive Co(II) forms and the subsequent activation of the methyltransferase reactions in which they participate.
14 † These authors contributed equally to this work.
The abbreviations used are: MMA, monomethylamine; DMA, dimethylamine; TMA, trimethylamine; and CoM, coenzyme M or 2-mercaptoethanesulfonic acid.               Table S1; and Mbar_A3583, a Ram homolog found near the methylthiol:CoM methyltransferase genes in M. barkeri Fusaro (RamS in Table S1). The cysteines that form the ferredoxin binding signatures for Fe 4 S 4 clusters are indicated in yellow. Non-identical residues are in black, identical residues in 2 or more sequences are in red, and conserved substitutions in 2 or more sequences are in blue. Figure S2: Peptides identified by LC MS-MS analysis of trypsin or chymotrypsin digestion of RamA. The sequence predicted from the ramA gene from M. barkeri MS is shown. Those peptides from trypsin (red) or chymotrypsin (highlighted in yellow) digests are indicated. Sequence that was identified in both digests are both highlighted and colored red. Peptides identified from each digest are also listed in Tables S2 and S3. The combined peptides obtained from the RamA protein covered 78% of the predicted product of ramA. The sequence match between identified RamA peptides and the predicted ramA product was 100%. Identical residues are shaded red, conserved residues are in blue, and non-conserved residues are black. (See Table  S2 for Genbank accession numbers of non-methanosarcinal proteins).  29 Table S1. RamA homologs encoded in genomes of Methanosarcinaceae family members. The genes are identified with the locus numbers assigned in the original genomic sequencing projects cited in the paper. Each class of ram homolog was designated as either RamA, RamM, or RamK by binning most similar sequences. The similarities of homologs to either RamA or RamK are noted in the table above, while the RamK proteins averaged 92% similarity to each other. Notably, RamA, Ram K, and RamM were adjacent to genes of similar physiological roles. The RamS homologs were assigned this designation based on adjacency to two genes encoding putative methylthiol:CoM methyltransferase. RamX is near a CoM methylase gene homolog. However, the RamS and RamX genes were only 62% similar to each other and did not bin with RamM, RamA, or RamK homologs. All noted similarities are based on total percent positive scoring residues during BLAST alignments using the Blosum62 matrix.   Table S3. Predicted and observed m/z values from LC-MS/MS analysis of RamA digested by chymotrypsin. The protein had been modified with iodoacetamide prior to digestion which led to the identification of carbamidomethylated cysteines (C (CAM) ) . Analysis of the b/y ions for sequence determination was carried out using PEAKS from Bioinformatics Solutions.