A New Type of Sulfite Reductase, a Novel Coenzyme F420-dependent Enzyme, from the Methanarchaeon Methanocaldococcus jannaschii*

Methanocaldococcus jannaschii is a hypertheromphilic, strictly hydrogenotrophic, methanogenic archaeon of ancient lineage isolated from a deep-sea hydrothermal vent. It requires sulfide for growth. Sulfite is inhibitory to the methanogens. Yet, we observed that M. jannaschii grows and produces methane with sulfite as the sole sulfur source. We found that in this organism sulfite induces a novel, highly active, coenzyme F420-dependent sulfite reductase (Fsr) with a cell extract specific activity of 0.57 μmol sulfite reduced min-1 mg-1 protein. The cellular level of Fsr protein is comparable to that of methyl-coenzyme M reductase, an enzyme essential for methanogenesis and a possible target for sulfite. Purified Fsr reduces sulfite to sulfide using reduced F420 (H2F420) as the electron source (Km: sulfite, 12 μm; H2F420, 21 μm). Therefore, Fsr provides M. jannaschii an anabolic ability and protection from sulfite toxicity. The N-terminal half of the 70-kDa Fsr polypeptide represents a H2F420 dehydrogenase and the C-terminal half a dissimilatory-type siroheme sulfite reductase, and Fsr catalyzes the corresponding partial reactions. Previously described sulfite reductases use nicotinamides and cytochromes as electron carriers. Therefore, this is the first report of a coenzyme F420-dependent sulfite reductase. Fsr homologs were found only in Methanopyrus kandleri and Methanothermobacter thermautotrophicus, two strictly hydrogenotrophic thermophilic methanogens. fsr is the likely ancestor of H2F420 dehydrogenases, which serve as electron input units for membrane-based energy transduction systems of certain late evolving archaea, and dissimilatory sulfite reductases of bacteria and archaea. fsr could also have arisen from lateral gene transfer and gene fusion events.

involved monitoring the oxidation of reduced F 420 (H 2 F 420 ) with sodium sulfite at 400 nm. The anaerobic assay method has been described previously (17). The reaction rate was calculated from an extinction coefficient of 25 mM Ϫ1 cm Ϫ1 for F 420 at 400 nm (18). For each standard assay a 1-ml reaction mixture containing the following components was used: 50 mM potassium phosphate buffer, pH 7, 40 M reduced F 420 (H 2 F 420 ), and 1.5 mM sodium sulfite. The assay was initiated by the addition of enzyme. For pH studies, the potassium phosphate buffer was replaced with constant ionic strength buffers (14). H 2 F 420 was generated by chemical reduction of F 420 , which was purified from M. thermautotrophicus (17,19), in water with NaBH 4 (20). Unreacted NaBH 4 was titrated with HCl. When methylviologen (MVϩ) replaced sulfite as an electron acceptor, the assay mixture also contained 1.44 mM metronidazole (20). In this assay the oxidation of H 2 F 420 could be followed at 400 nm without an interference from reduced methylviologen (MV 0 ), because the latter was continuously re-oxidized by metronidazole (20). For determining the rate of oxidation of MV 0 with sulfite, H 2 F 420 in the standard assay was replaced with MV 0 , and the reaction was followed at 560 nm (18); MV 0 was generated by reducing MVϩ with titanium citrate (21).
Assay for Sulfide Produced in Fsr Reaction-A 160-l standard Fsr reaction mixture containing H 2 F 420 , sodium sulfite, and 0.2 g of purified enzyme was used. To avoid interference from reactants and side products, the sulfide produced in the reaction was recovered from the reaction mixture by use of a micro gas diffusion cell prior to assay (22). The resulting Na 2 S was quantified by the methylene blue method (23).
To determine the amount of F 420 generated by the reaction, an aliquot of the assay mixture was diluted in potassium phosphate buffer, pH 7, and the absorbance of this solution at 400 nm was determined. From this absorbance data and by use of the extinction coefficient for F 420 , as indicated in the preceding section, the amount of the oxidized coenzyme formed was calculated.
Purification of F 420 -dependent Sulfite Reductase-All steps, except the centrifugation of ammonium sulfate treated extract (see below), were performed inside an anaerobic chamber containing a gas atmosphere of N 2 and H 2 (95:5, v/v) and maintained at 25°C. M. jannaschii cells (1.8 g of wet weight) grown with sulfite were lysed in 20 ml of 25 mM potassium phosphate buffer, pH 7.0 (buffer A), via osmotic shock and subsequent three passages through a 25-gauge needle. The lysate was centrifuged at 1,230 ϫ g and 25°C for 20 min to collect a supernatant. The pellet was resuspended in 15 ml of buffer A, and the suspension was centrifuged to obtain additional supernatant. The two supernatants were combined, and the pool was fractionated by precipitation with ammonium sulfate at 30 and 60% saturation on ice. At each step a tube containing the treated extract was sealed inside the chamber, and then it was centrifuged outside the chamber at 10,000 ϫ g. After centrifugation, the tube was transferred inside the chamber where the supernatant was collected. Further fractionation steps involved chromatography; the (NH 4 ) 2 SO 4 and NaCl solutions used at these steps contained 25 mM potassium phosphate buffer, pH 7.0. The column fractions with Fsr activity were examined via SDS-PAGE for purity. The volume, protein content, and activity of the enzyme preparation recovered at each of these steps are given in TABLE ONE. First, the 60% supernatant was loaded onto a 1.5-cm ϫ 6-cm Phenyl-Sepharose column (Amersham Biosciences, bed volume, 10.6 ml) that was pre-equilibrated with 1.5 M (NH 4 ) 2 SO 4 . The bed was then washed with 20 ml of equilibration solution and eluted with a 75-ml reversed gradient of 1.5 to 0 M (NH 4 ) 2 SO 4 . The column fractions with 379 -205 mM (NH 4 ) 2 SO 4 contained Fsr activity, of which those with a minimum number of non-Fsr polypeptides were pooled. This pool was diluted with buffer A to lower the (NH 4 ) 2 SO 4 concentration to 25 mM. The diluted preparation was loaded onto a 0.7-cm ϫ 10.5-cm coenzyme F 420 -Sepharose column (20,24) (bed volume, 4 ml) that was pre-equilibrated with buffer A. The bed was washed with 10 ml of buffer A, and then an 80-ml gradient of 0 to 0.5 M NaCl was applied to the column. Fsr eluted at a NaCl concentration of ϳ260 mM and the active fractions with a minimum number of non-Fsr polypeptides were pooled. The pooled product was diluted with buffer A to a final NaCl concentration of ϳ10 mM. The diluted enzyme was loaded onto a 0.7-cm ϫ 5.5-cm QAE-Sephadex column (Amersham Biosciences, bed volume, 2.1 ml) that was equilibrated with buffer A. The column was washed with 10 ml of buffer A, and the enzyme was eluted at a NaCl concentration of ϳ325 mM under a 80-ml gradient of 0.0 to 1.0 M NaCl. The active fractions were pooled. This preparation was homogeneous.
Sequence and Phylogenetic Analysis-The ClustalW program was used for the comparative analysis of amino acid sequences of proteins (25). For drawing phylogenetic inferences, aligned protein sequences were analyzed by using the neighbor-joining bootstrap analysis (26) in the ClustalX program (27). The phylogenetic tree was developed using NJ plot (28).

RESULTS
Growth of M. jannaschii with Sulfite as the Sulfur Source and Expression of a Novel Sulfite Reductase-M. jannaschii used sulfite as the sole sulfur source for growth; the growth and methane formation patterns were similar to that seen with a culture grown with the same level of sulfide (Fig. 1). A medium without an added sulfur source did not support growth. When grown with sulfite, the organism expressed a 70 kDa a The activity was assayed in a standard reaction mixture of the following composition: 50 mM potassium phosphate buffer, pH 7, 40 M reduced F 420 (H 2 F 420 ), and 1.5 mM sodium sulfite. b The reason for the apparently low value for the fold of purification: The cell extract contained a very high level of Fsr ( Fig. 2A). Therefore, only a moderate fold of purification provided a homogeneous preparation of the enzyme. c The reason for the poor yield at each stage of fractionation beyond the ammonium sulfate precipitation: The column fractions with Fsr activity were examined for purity via SDS-PAGE, and only those fractions with a minimum number of non-Fsr polypeptides were pooled. This strategy provided a homogeneous preparation of the enzyme but resulted in a substantial loss of Fsr. polypeptide ( Fig. 2A). As judged from the SDS-PAGE data for cell lysates, this polypeptide was absent in cells that received sulfide. Mass spectrometry with an unseparated tryptic digest and database searches matched several peptide fragments of the 70-kDa band with ORF MJ0870 (9) ( Fig. 3 and supplemental Table 1S); the matches covered 25% of the identified ORF. MJ0870 has been annotated as the ␤ subunit of a coenzyme F 420 -reducing hydrogenase (FrhB) (9). However, from a comparative sequence analysis (25), it was apparent that this 620-residue polypeptide had two distinct domains. As shown below, the N-terminal half comprised of 311 residues represented a reduced coenzyme F 420 (H 2 F 420 ) dehydrogenase, and the C-terminal half (296 residues) was homologous to siroheme-dependent dissimilatory sulfite reductases. Thus, the 70-kDa polypeptide had the potential of using F 420 as electron carrier for reducing sulfite into sulfide. Accordingly, we called this enzyme a coenzyme F 420 -dependent sulfite reductase or Fsr. We examined M. jannaschii cell extracts for Fsr activity, which was measured in terms of the oxidation of H 2 F 420 with sulfite as the electron acceptor. Extracts of cells grown with sulfite catalyzed this reaction at a specific rate of 1.3-1.7 mol of H 2 F 420 oxidized min Ϫ1 mg Ϫ1 protein. This activity was absent in cells grown with sulfide.
Purification and Molecular Properties of Fsr-The Fsr activity in cell extracts was highly oxygen sensitive. About 92% activity was lost if the cell extract was exposed to air for 60 min. Dithiothreitol and 2-mercapotoethanol neither protected the enzyme from inactivation nor restored activity in an air-inactivated enzyme preparation. Therefore, all purifications steps were conducted under strictly anaerobic conditions. TABLE ONE provides a summary of the results from a purifica-tion experiment. At the ammonium sulfate precipitation step, 67.5% of cell extract Fsr activity was recovered in the 60% saturated supernatant. From this supernatant the enzyme was purified to homogeneity via phenyl-Sepharose, coenzyme F 420 -Sepharose, and QAE-Sephadex chromatography steps. In a non-denaturation polyacrylamide gel electrophoresis experiment the final enzyme preparation showed a single FIGURE 1. Growth and methane formation in M. jannaschii cultures with sulfide or sulfite as sole sulfur source. Methanogenesis data represent accumulated methane. Each experiment was done in duplicate, and both sets are shown. The bottles were re-pressurized to 3 ϫ 10 5 Pa with H 2 -CO 2 (80:20, v/v) prior to each sampling. At each of the times indicated, 50 -100 l of headspace gas and 3 ml of culture sample were removed for methane assay and optical density (600 nm) measurement. The concentration of dissolved sulfide was Ͻ1 mM, and most of the added sulfide existed as H 2 S in the headspace gas, because the pH of the medium was 6.0 and the pK a for H 2 S/HS Ϫ is 7.04 (69). Ϫ2 ) or sulfide (S Ϫ2 ). A lysate was prepared by boiling cells in a solution containing 62.5 mM Tris-HCl buffer, pH 6.8, 10% glycerol, 5% 2-mercaptoethanol. The gels were stained with Coomassie Blue. The overexpressed band in the SO 3 Ϫ2 lane is for MJ0870 or Fsr. The arrows with labels A, B, and C point to the suggested locations of ␣, ␤, and ␥ subunits of methylcoenzyme M reductase. B, non-denaturing PAGE for homogenous Fsr. C, SDS-PAGE for homogenous Fsr. Identities from mass spectrometry data: 2XFsr, linked dimer; Fsr, one polypeptide; FsrЈ, degradation product.  Fig. 2A) was digested with trypsin. The resulting tryptic peptide fragment mixture was analyzed via MALDI-TOF mass spectrometry. A search of the NCBI protein data base matched 17 of the mass spectrometry-derived mass values to the regions of ORF MJ0870 that are shown in red; the corresponding tryptic fragments are shown underlined. The regions of MJ0870 that were not represented in the mass spectrometry data are shown in black. See "Materials and Methods " for the experimental details and Table 1S for an analysis of the mass spectrometry data. *, cysteines modified with iodoacetamide to form carbamidomethyl cysteine. band (Fig. 2B). In a SDS-PAGE gel, the same preparation exhibited three bands (Fig. 2C). Mass spectrometric analyses (13,14) showed that each of these bands corresponded to MJ0870 (data not shown). Therefore, we concluded that the 70-kDa band in Fig. 2C was due to intact MJ0870 and the 42-and 140-kDa bands were, respectively, for a degradation product and a covalently linked dimer that were probably created during sample preparation. Accordingly, the enzyme obtained from the QAE-Sephadex chromatography was homogeneous. The fractionation scheme used in this work provided a final yield of 2.6% (TABLE ONE). This apparently low value was the result of an aggressive approach that was necessary for generating a homogeneous Fsr preparation. Our early efforts for obtaining a homogeneous Fsr preparation were not successful. Therefore, in further attempts at each step of purification the column fractions with Fsr activity were examined for purity via SDS-PAGE. Only those fractions with a minimum number of non-Fsr polypeptides were pooled. This strategy provided a homogeneous preparation of the enzyme, but resulted in a substantial loss of Fsr protein and consequently a poor overall yield. The fold of purification was 14 (TABLE  ONE). This value is justified by the fact that the cell extract contained a very high level of Fsr ( Fig. 2A). Therefore, only a moderate fold of purification provided a homogeneous preparation of the enzyme.
From the gel filtration data the hydrodynamic radius and the apparent native molecular mass of the enzyme were estimated to be 69.5 Å and 350 kDa, respectively. The UV-visible spectrum of Fsr in as isolated form exhibited peaks at 280, 395, and 590 nm (Fig. 4).
Catalytic Properties of Fsr-The purified enzyme oxidized H 2 F 420 with sodium sulfite. The optimum pH and temperature for this activity were 7.0 and Ͼ95°C, respectively. Activities at temperatures above 95°C were not determined due to technical limitations. From the straight line section of the Arrhenius plot (25-75°C), a value of 45.9 kJ/mol was obtained for the activation energy. Within a sulfite concentration range of 1.4 -300 M and with a fixed H 2 F 420 concentration of 40 M, the apparent K m value SO 3 Ϫ2 was 12.2 Ϯ 1 M, and the corresponding apparent V m value was 16   The results of a sulfide production assay showed that from 0.366 mol of H 2 F 420 and 3.66 mol of sulfite, Fsr produced 0.362 Ϯ 0.003 mol of H 2 F 420 and 0.088 Ϯ 0.004 mol of sulfide; these values represent averages of data from three independent assays. Sodium salts of thiosulfate and sulfate did not serve as the electron acceptor for H 2 F 420 oxidation by Fsr. Also, Fsr could not use NADH and NADPH for the reduction of sulfite.
Homologs of MJ0870-The ORF MTH280 of Methanothermobacter thermautotrophicus (10,29) and ORF MK0799 of Methanopyrus kandleri (10,29) were found to be homologous to MJ0870 or Fsr (Fig. 5). None of these MJ0870 homologs have been studied experimentally. MTH280 has been annotated as FrhB (10) and MK0799 as a protein with FrhB and nitrite reductase characteristics (29), respectively. Simi-   Fig. 5A). These centers are absent in the F 420 -interacting sub-units of the F 420 -reducing hydrogenases of the methanogens (FrhB and FruB; Fig. 5A) and are located at the C terminus of the homologous subunit (␤ or FdhB) of F 420 -depependent formate dehydrogenase (the segment not shown). Four fully conserved Cys residues of MJ0780 (Cys 160 , Cys 201 , Cys 256 , and Cys 259 ; marked with black bullets in Fig. 5A) that do not lie close to each other were also found conserved in all F 420 -interacting polypeptides shown in Fig. 5A. A great number of non-Cys residues of MJ0870 were conserved in FrhB, FruB, and FdhB (Fig.  5A). As judged from an analysis by use of 3D-PSSM, a web-based program for protein fold recognition (www.sbg.bio.ic.ac.uk/ϳ3dpssm) (34), none of these residues were the part of a known motif. Fsr did not possess a recognizable coenzyme F 420 -binding sequence feature (35,36). All of the above-described features of MJ0870 were present in its homologs, MTH280 and MK0799 (Fig. 5A).
Sequence Features of the C-terminal Half of Fsr-We use Fsr-C as an abbreviation for the C-terminal half of Fsr (residues 325-620 in MJ0870). Fsr-C and the corresponding regions of MTH280 and MK0799 were compared with the hemoprotein subunit (SirHP) of Escherichia coli assimilatory sulfite reductases or ASR, the A and B subunits (DsrA and DsrB) of archaeal or bacterial dissimilatory sulfite reductases AF-DsrB, A. fulgidus ␣ and ␤ subunits of dissimilatory sulfite reductase (ORFs AF0423 and AF0424); DV-DsrA and DV-DsrB, D. vulgaris ␣ and ␤ subunits of dissimilatory sulfite reductase (accession numbers AAA70107 and AAA70108); ST-AsrC, S. enterica hemoprotein subunit of small size anaerobic sulfite reductase (accession number CAD02753). Black and teal bullets, conserved Cys and Pro in MJ0870, respectively; gray bullets, Cys and Pro conserved elsewhere. "ϩ" represents sulfite binding Arg residues. The following colors were used to show sequence identity (shading) and functional conservation (colored letters): red, D and E; blue, Arg, Lys, and His; yellow, Phe, Tyr, Trp, and His; gray, Ile, Leu, and Val; olive, Ala, Gly, Ser, and Thr; green, Gln and Asn; purple, Met; teal, Pro; and white, Cys. or DSR, and the siroheme containing subunit (AsrC) of a small sulfite reductase that is expressed in Salmonella enterica under anaerobic growth conditions (Fig. 5B) (33,(37)(38)(39). As explained in the following section, S. enterica AsrC is a dissimilatory sulfite reductase (DSR). In an ASR or DSR, three sequence elements, designated H1, H2, and H3, are thought to house a siroheme-[Fe 4 -S 4 ]-cluster and bind sulfite (37). These elements were conserved in the C termini of the MJ0870 and its homologs (Fig. 5B). The Arg 355 and Arg 423 of MJ0870 (marked by the "ϩ" in Fig. 5B) corresponded to the sulfite binding Arg residues of sulfite reductases (37). The siroheme-[Fe 4 -S 4 ]-binding CX 5 CX n CX 3 C sequence is fully conserved in SirHP, DsrA, and AsrC and only partially conserved in a DsrB (Fig. 5B). This element was found fully conserved in MJ0870, MTH280, and MK0799 (Cys 428 -Cys 434 -Cys 468 -Cys 472 in MJ0870, Fig. 5B). However, in overall sequence features, the C-terminal regions of the archaeal proteins were more similar to DSRs (Fig. 5B) and not to the E. coli Sir HP (data not shown). The Pro 488 -Cys 495 -Cys 498 -Cys 501 -Cys 505 and Cys 524 -Cys 527 -Cys 530 -Cys 534 -Pro 535 of MJ0870 have the potential of forming two [Fe 4 -S 4 ] clusters, and both of these were found conserved in AsrC, MTH280, and MK0799. The first of these clusters was absent in both DsrAs and DsrBs, and the second was fully conserved in the DsrBs but partially in DsrAs (Fig. 5B).
Phylogenetic Analyses-Fsr-N and the N-terminal halves of MTH280 and MK0799 were closely related to FqoF (Fig. 7A). The closest relative of the Fsr-C and its homologs was S. enterica AsrC (Fig. 7B); a close relationship was also seen with the subunits of dissimilatory sulfite reductases, DsrA and DsrB. AsrC is a dissimilatory enzyme (39,40); the abbreviation Asr means anaerobic sulfite reduction and not assimilatory sulfite reductase (39). Therefore, Fsr-C belonged to the dissimilatory sulfite reductase group (AsrC, DsrA, and DsrB).

DISCUSSION
Sulfide is an essential nutrient for M. jannaschii (8), and sulfite is inhibitory to the methanogens (1). Yet we found that this methanogenic archaeon was able to use sulfite as a sole sulfur source and did not show a significant inhibition of methanogenesis by this oxyanion (Fig. 1). As discussed below, this ability was due to a novel enzyme that was expressed on demand. We also discuss the physiological, ecological, and evolutionary implications of our findings.
A Novel Sulfite Reductase-Sulfite induced the expression of a sulfite reductase in M. jannaschii ( Fig. 2A) that was encoded by orf MJ0870. The subunit size of this enzyme was 70 kDa (Fig. 2C). It reduced sulfite into sulfide and thereby allowed M. jannaschii to grow with sulfite as sole sulfur source. Interestingly, this sulfite reductase used reduced coenzyme F 420 (H 2 F 420 ) as the electron source. Previously described sulfite reductases use nicotinamides and cytochromes as electron carriers (Fig. 8, A-C) (37). F 420 is a naturally occurring 5-deazaflavin that was originally discovered in the methanogenic archaea (41). It is an obligatory 2-electron or hydride-transferring coenzyme (41,42). The midpoint potential values presented below (42,43) (44)). The overall reaction leading to the reduction of sulfite with H 2 F 420 is exergonic (Reactions 1-3).
The use of sulfite as a sulfur source has been observed previously with two methanogens, M. thermolithotrophicus and M. thermautotrophicus (6). However, the enzymatic basis for this observation remains unknown. A small siroheme sulfite reductase (size subunit, 23 kDa) has been isolated from Methanosarcina barkeri (45), but the physiological electron donor and the in vivo role for this enzyme are yet to be identified. It is not known if M. barkeri can use sulfite as sole sulfur source. Therefore, this is the first report of a coenzyme F 420 -dependent sulfite reductase that we called Fsr.
We also determined that Fsr reduced sulfite to sulfide with H 2 F 420 as the electron donor. The assay was performed in the presence of a 30-fold excess of sulfite. This calculation was based on the moles of electrons that were required for converting all of the supplied sulfite to sulfide. Under these conditions, for every 3 mol of H 2 F 420 oxidized 0.73 mol of sulfide was generated. If Fsr carried out only complete reduction of sulfite, producing sulfide as the sole product, this ratio would have been 3:1 (see the reactions above). The observed deviation could either be due to an experimental error or the production of partially reduced sulfur compounds in addition to sulfide. Production of partially reduced products, such as trithionate and thiosulfate, has been observed with certain sulfite reductases (46). However, the production of thiosulfate as an intermediate of sulfite reduction by Fsr is an unlikely possibility, because Fsr did not oxidize H 2 F 420 with thiosulfate. We will address this issue in our future work.
Interestingly, the rates of the two Fsr partial reactions, oxidation of H 2 F 420 and reduction of sulfite, as measured by use of methylviologen as the electron carrier, were about the same. However, these rates were three to five times higher than that for the overall reaction. The reason for this difference is currently unknown.
Our data did not allow us to describe the quaternary structure of Fsr. A dissimilatory sulfite reductase is composed of two DsrA and two DsrB subunits (Fig. 8C) (37). In E. coli sulfite reductase, the siroheme component or SirHP is a tetramer and the flavoprotein component (SirFP) is octameric (Fig. 8A (37)). Our native mass data seem to indicate that that Fsr was composed of five subunits. However, a gel filtration data-derived value is less reliable (47). A final conclusion on the quaternary structure of Fsr must await an accurate determination of the native molecular mass for Fsr by use of a more appropriate method such as analytical ultracentrifugation (47).
A Detoxification Role for Fsr-Fsr was induced with sulfite. The cellular level of the Fsr activity was relatively high for an anabolic enzyme. Taken together these observations indicated that Fsr was poised to protect M. jannaschii from the toxic effect of sulfite. Remarkably, the methane production rate in a culture with sulfite was comparable to that for a culture with the same level of sulfide (Fig. 1). Therefore, M. jannaschii avoided the inhibition of its methyl-coenzyme M reductase by sulfite. In a methanogen, methyl-coenzyme M reductase is expressed to a level as high as 30% of the total cell protein (48, 49). M. jannaschii has the potential of expressing two methyl-coenzyme M reductase isoenzymes  (9). Accordingly, the 61-, 48-, and 30-kDa bands in Fig. 2A most likely corresponded to the methyl-coenzyme M reductase subunits. The intensities of these bands showed that in cells grown with sulfite, the methyl-coenzyme M reductase subunits and Fsr were expressed at comparable levels. This observation is consistent with a detoxification role for Fsr. However, this hypothesized link between a need to protect the methyl-coenzyme M reductase and the induction of Fsr needs to be justified by the K i value of the enzyme for sulfite. Although, the inhibition of this enzyme with sulfite has been reported (3,4), the corresponding K i value is yet to be determined.
A H 2 F 420 Dehydrogenase Domain in Fsr-Fsr polypeptide contained 620 amino acid residues. The N-terminal half of this polypeptide (residues 1-311; called Fsr-N) possessed the sequence features of FqoF and FpoF (Fig. 5A) (31,32) and exhibited a close phylogenetic relationship to FqoF (Fig. 7A). In A. fulgidus FqoF, a H 2 F 420 dehydrogenase, introduces electrons into a membrane-based electron transport system (Fig. 6A). The process involves transfer of electrons from H 2 F 420 to 1 electron carrying Fe-S clusters by using protein-bound flavin as a 1-electron/2electron switch (31). FpoF performs a similar task in M. mazei (Fig. 6A) (32). Therefore, we propose that Fsr-N is the H 2 F 420 dehydrogenase domain of Fsr. Our data showed that Fsr had a H 2 F 420 dehydrogenase activity. In addition to performing sulfite-dependent H 2 F 420 oxidation, Fsr was able to transfer electrons from H 2 F 420 to methylviologen, a one electron-restricted carrier. FqoF and FpoF also catalyze the latter reaction (31,50). Because FqoF and FpoF are flavoproteins (31,50), Fsr was expected to carry flavin, but, the spectrum in Fig. 4 does not show a clear indication for a protein-bound flavin in Fsr. It is possible that Fsr lost a major amount of the bound flavin and the 300 -550 nm region in the spectrum represented a superimposition of the spectra of siroheme and a small amount of flavin (Fig. 4); a major loss of protein-bound flavins during the purification has been observed with both FqoF and FpoF (31,50). However, this explanation was not supported by our data on the effect of added flavin coenzymes on the activity of homogeneous Fsr. Either the presence of FAD, FMN, or FAD plus FMN in the assay mixture or incubation of purified Fsr with these compounds did not increase the specific activity of the enzyme. Rather such treatments inhibited the enzyme slightly. A detailed assessment on the role of flavins in Fsr reaction, if any, will come from our future work on the prosthetic groups of Fsr.
A Dissimilatory Sulfite Reductase-type Domain in Fsr-Although our observation described Fsr as an anabolic enzyme, in primary structure this protein was unrelated to the assimilatory sulfite reductase from E. coli (Fig. 7B). Rather, the C-terminal half of Fsr (residues 325-620) or Fsr-C was closely related to the siroheme-containing dissimilatory sulfite reductases (DSR) of the archaea and bacteria (Figs. 5B and 7B). We hypothesize that Fsr-C carried siroheme, where the electrons derived from the 1-electron restricted iron-sulfur centers of the H 2 F 420 dehydrogenase domain or Fsr-N were used for the reduction of sulfite to sulfide. Our experimental data showed that Fsr contained the elements of a DSR. First, the 280, 395, and 590 nm peaks in the UV-visible spectrum of Fsr were typical of a sulfite reductase hemeprotein (Fig. 4) (51). This spectrum did not show a 714 nm peak, which is a signature for a high spin ferric complex of isobacteriochlorin that is found in the assimilatory enzymes from E. coli and Desulfovibrio vulgaris (51,52). Second, Fsr was able to utilize MV 0 , a one-electron donor, for the reduction of sulfite into sulfide. This reaction has been observed with the bacterial and archaeal dissimilatory sulfite reductases (53,54).
Fsr, a Novel Enzyme with a Unique Chimeric Structure-As shown above, the homologs of two halves of Fsr have been encountered previously as independent entities (Figs. 5 (A and B), 6A, and 8C). However, Fsr presents the first example where the structural and functional attributes of these units have been brought together (Figs. 5, 6B, and 8D). Although, assimilatory sulfite reductase from E. coli was found to be structurally unrelated to Fsr, these two enzymes share some of the functional characteristics (Figs. 8, A and D). In the E. coli enzyme, a flavoprotein subunit (SirFP) derives electrons from NADPH, a 2e-restricted donor, and passes those via FAD, FMN, and [4Fe-4S] centers to the siroheme of a hemoprotein subunit (Sir-HP), the site of sulfite reduction (Fig. 8A) (37). This role of NADPH is similar to that of H 2 F 420 in the Fsr reaction, but the electron transfer route between H 2 F 420 and siroheme in Fsr is currently unknown.
Ecological Relevance and a Possible Energy Production Role for Fsr-Our findings are consistent with the following environmental data and deductions on the hydrothermal vents. As mentioned above, both M. jannaschii and M. kandleri are inhabitants of the deep-sea hydrothermal vents (8,55). Cold seawater that permeates through the chimney wall brings the temperature of the nutrient rich vent fluid down from 350°C to a level where life can exist (56,57). This process also brings oxygen into the vent. Sulfide, that is present at a high level in the vent fluid (56,57), reacts with this oxygen and helps to establish anaerobic conditions in the cooled zones, thereby providing conditions conducible for the growth of a methanogen. However, this reaction has the potential of producing sulfite. A lack of free oxygen in the vent water and a decrease of H 2 S concentration with temperature within the vents have been thought to indicate that sulfur species with oxidation states between Ϫ2 and ϩ6 (representing sulfide and sulfate) are present in the vent fluid (57); sulfate is the predominant oxidation product. The sulfite concentration in vent water is expected to be rather low, and this condition is consistent with the observed low K m value of Fsr for sulfite. It should be noted that the literature on the deep-sea hydrothermal vent does not provide a value for the sulfite level in vent water. We have considered the possibility for Fsr providing an alternate non-methanogenic energy production route to the vent methanogens. The existence of a sulfite-based energy metabolism in a hydrothermal vent has been reported previously. Archaeoglobus veneficus, which was isolated from the walls of hydrothermal vents, derives energy from sulfite oxidation (58). It is a close relative of the late evolving methanogens such as the Methanosarcina species. However, thus far we have been unable to grow M. jannaschii with H 2 and sodium sulfite as the energy production substrates in a MES-NaOH-buffered medium containing sodium acetate as the sole carbon source.
Evolutionary Implications-We propose two equally possible hypotheses on the evolution of Fsr: 1) Fsr is the ancestor of FqoF or FpoF and Dsr. 2) Fsr originated from a fusion of laterally transferred dsrA and fqoF genes. The fusion could have preceded the lateral transfer. Because Dsr and FqoF/FpoF have been studied in the context of the evolution of metabolism (59,60), the above two possibilities are discussed here. M. jannaschii, M. kandleri, and M. thermautotrophicus are strictly hydrogenotrophic autotrophs (8,55,61), a characteristic that fits the conditions of early earth (62). Their positions in the 16 S rRNA-based tree of life are closer to the root than that of Methanosarcina and A. fulgidus (63). M. jannaschii, M. kandleri, and M. thermautotrophicus carry Fsr homologs (Fig. 5), but they do not possess a complete sulfate reduction pathway or an Fqo/Fpo-type membrane energy transduction system (9,10,29,32). A. fulgidus does not contain Fsr. Instead it carries homologs of both Fsr domains as separate polypeptides (33). This archaeon employs FqoF, an Fsr-N homolog, in energy transduction (Fig. 6A) (31) and DsrA and DsrB, the Fsr-C homologs, in a dissimilatory sulfate reduction pathway (Fig. 8C) (33); DsrA and DsrB are also used by sulfate-reducing bacteria (60). Methanosarcina species possess FpoF (Fsr-N homolog) (Fig. 6A) (32) and express a small size (23 kDa) siroheme sulfite reductase (45). Based on these pieces of information it could be hypothesized that fsr is an ancestral gene. The 5Ј-half of fsr gave rise to the input domain (fqoF/fpoF) of the membrane-based energy transduction systems in certain late evolving archaea and the 3Ј-half yielded bacterial and archaeal dissimilatory siroheme sulfite reductases (dsrA and dsrB). The catabolic type highly active nature of Fsr fits this hypothesis. One could imagine that fsr provided a selective advantage to an ancestral methanogen for surviving sulfite exposure when oxygen appeared on earth. Recent reports suggest that the development of a fully oxic atmosphere followed a protracted oxygenation period (64 - 66), where a small supply of oxygen was quickly and fully sequestered in a process that could have generated sulfite, an incomplete oxidation product of sulfide. At a later time fsr allowed the development of the sulfate reduction pathway within a methanogen. The M. jannaschii genome bears more signs of this possibility. The ORFS MJ0066 and MJ0973 show some sequence similarities to a 3Ј-phosphoadenosine-5Јphosphosulfate reductase and sulfate adenylyltransferase or ATP sulfurylase (9), which participate in the reduction of sulfate to sulfite in certain bacteria and archaea (67). On the other hand, the limited distribution of fsr and a wide distribution of dsr could question the hypothesized ancestral nature of fsr. Because M. jannaschii and M. kandleri are hydrothermal vent-associated organisms (8,55), a search for additional fsr homologs in the vent environment might shed some light on the evolution of fsr. The generation of fsr from a fusion of fqoF or fpoF and dsr is a clear possibility, because the latter two genes have been found to co-exist in a cell (31,33). Nevertheless, our results bring new thoughts to the study of the evolution of early metabolisms on earth, especially in the field where DsrA and DsrB sequences of cultured and uncultured microorganisms and sulfur isotope records have been extensively analyzed in search of the evolutionary origin of energy metabolisms of extant prokaryotes (40,60,68).