Biosynthesis of 4-Thiouridine in tRNA in the Methanogenic Archaeon Methanococcus maripaludis *

Background: Bacterial ThiI catalyzes 4-thiouridine biosynthesis by using a rhodanese-like domain for sulfur transfer. Results: ThiI in methanogenic archaea employs a conserved CXXC motif to generate persulfide and disulfide intermediates for sulfur transfer. Conclusion: Methanogens possess a unique sulfur relay strategy. Significance: Sulfur metabolism in methanogens is a model for the evolution of sulfur metabolism in the anaerobic sulfide-rich environments common on ancient earth. 4-Thiouridine (s4U) is a conserved modified nucleotide at position 8 of bacterial and archaeal tRNAs and plays a role in protecting cells from near-UV killing. Escherichia coli employs the following two enzymes for its synthesis: the cysteine desulfurase IscS, which forms a Cys persulfide enzyme adduct from free Cys; and ThiI, which adenylates U8 and transfers sulfur from IscS to form s4U. The C-terminal rhodanese-like domain (RLD) of ThiI is responsible for the sulfurtransferase activity. The mechanism of s4U biosynthesis in archaea is not known as many archaea lack cysteine desulfurase and an RLD of the putative ThiI. Using the methanogenic archaeon Methanococcus maripaludis, we show that deletion of ThiI (MMP1354) abolished the biosynthesis of s4U but not of thiamine. MMP1354 complements an Escherichia coli ΔthiI mutant for s4U formation, indicating that MMP1354 is sufficient for sulfur incorporation into s4U. In the absence of an RLD, MMP1354 uses Cys265 and Cys268 located in the PP-loop pyrophosphatase domain to generate persulfide and disulfide intermediates for sulfur transfer. In vitro assays suggest that S2− is a physiologically relevant sulfur donor for s4U formation catalyzed by MMP1354 (Km for Na2S is ∼1 mm). Thus, methanogenic archaea developed a strategy for sulfur incorporation into s4U that differs from bacteria; this may be an adaptation to life in sulfide-rich environments.

modification was found in about 70% of the bulk tRNAs (2), and its level in some tRNA species varied depending on the growth rate (3). A Salmonella typhimurium mutant that lacked s 4 U was more sensitive to broadband near-UV killing, suggesting that this modification protects the cells from near-UV light (4). In response to near-UV irradiation, s 4 U in tRNAs serves as a photosensor ( max of s 4 U is 334 nm) by cross-linking with the nearby cytosine at position 13 (5)(6)(7). These cross-linked tRNAs are poor substrates for aminoacylation (6,8), leading to the accumulation of uncharged tRNAs (9). This effect mimics amino acid starvation and triggers the stringent response through the synthesis of ppGpp and ApppGpp, which inhibit cell growth and induce the expression of specific proteins that enhance cell survival after the stress (2,4,7,10).
Two enzymes in bacteria are required for the biosynthesis of s 4 U, the cysteine desulfurase IscS and ThiI (11)(12)(13)(14)(15). IscS is a pyridoxal 5Ј-phosphate-dependent enzyme that liberates sulfur from free cysteine, resulting in a cysteine persulfide in its active site and free alanine (16 -18). This IscS persulfide then donates sulfur for the biosynthesis of s 4 U, other thiolated nucleosides in tRNAs, and sulfur-containing cofactors (e.g. Fe-S clusters, thiamine, and molybdopterin) (14, 19 -23). For the synthesis of s 4 U, the IscS persulfide serves as a S 0 donor to generate a persulfide on ThiI, which catalyzes the adenylation of tRNA U8. This ThiI persulfide serves as the proximal sulfur donor for the thiolation of U8.
The E. coli ThiI contains four domains as follows: the N-terminal ferredoxin-like domain (NFLD); the thiouridine synthases, RNA methyltransferase and pseudouridine synthases domain (THUMP domain); the PP-loop pyrophosphatase domain (PP-loop domain); and the C-terminal rhodaneselike domain (RLD). The NFLD and THUMP domains bind tRNA by recognizing the acceptor-stem region (24 -26); the PP-loop domain adenylylates the 4-carbonyl group of tRNA U8 at the expense of ATP (27,28); and the RLD transfers sulfur (29,30). During the sulfur transfer, the catalytically essential Cys 456 located in the RLD receives sulfur from IscS to form the ThiI persulfide (29 -31). After donation of the terminal sulfur of the ThiI persulfide to tRNA U8, Cys 456 forms a disulfide bond with Cys 344 located in the PP-loop domain (30,32,33). Presumably, the Cys 456 -Cys 344 disulfide needs to be reduced to complete the enzymatic cycle. In E. coli and Salmonella enterica, the RLD of ThiI also participates in the sulfur transfer for thiamine biosynthesis (34 -36). However, ThiI homologs in many other bacteria lack the C-terminal RLD and are not necessary for thiamine biosynthesis (37)(38)(39). A recent report demonstrated that the Bacillus subtillis ThiI, which lacks the RLD, cooperates with a specialized cysteine desulfurase NifZ to synthesize s 4 U (37). However, the participation of this ThiI in sulfur transfer is still unclear.
The presence of s 4 U in archaeal tRNAs has been demonstrated in Thermoproteus neutrophilus (40) and a number of methanogenic archaea, including Methanococcus vannielii, Methanococcus maripaludis, Methanothermococcus thermolithotrophicus, Methanocaldococcus igneus, Methanocaldococcus jannaschii, and Methanococcoides burtonii (41)(42)(43). However, some fundamental questions remain about the mechanism of sulfur transfer to generate s 4 U in archaea. First, the sulfur donor is not known. Many methanogens and the sulfur-dependent hyperthermophilic archaea do not encode a recognizable cysteine desulfurase in their genomes (44). The methanogens also use a tRNA-dependent pathway for cysteine biosynthesis (45) and produce a much smaller pool of free cysteine in comparison with E. coli (46). These archaea use sulfide instead of cysteine as the sulfur source for Fe-S cluster and methionine biosynthesis (46). These results suggest that cysteine may also not serve as a sulfur source for s 4 U biosynthesis. Second, although most archaea have ThiI homologs, many homologs do not have the C-terminal RLD. Therefore, it is not known if ThiI participates in sulfur transfer. In this study, we used M. maripaludis, an obligately anaerobic methane-producing archaeon, as the model organism to study the biosynthesis of s 4 U in archaea. The M. maripaludis ThiI (MMP1354), which lacks RLD, is essential for s 4 U but not for thiamine biosynthesis. Furthermore, a conserved CXXC motif located in the PP-loop domain is essential for both in vitro and in vivo activities of s 4 U formation. These two cysteines form persulfide and disulfide intermediates during sulfur transfer. Finally, sulfide is a sufficient sulfur donor in vitro. Based upon these results, a model of s 4 U biosynthesis in methanogenic archaea is proposed.

EXPERIMENTAL PROCEDURES
Media and Culture Conditions of M. maripaludis-M. maripaludis was grown in 28-ml aluminum sealed tubes with 275 kilopascals of H 2 /CO 2 (4:1, v/v) at 37°C in 5 ml of McNA (minimal medium ϩ 10 mM sodium acetate, reduced with 3 mM L-cysteine), McNACoM (McNA reduced with 3 mM coenzyme M instead of cysteine), or McC (McNA ϩ 0.2% (w/v) casamino acids ϩ 0.2% (w/v) yeast extract) medium as described previously (47). The 100-ml cultures were grown in 1-liter bottles pressurized to 138 kilopascals with H 2 /CO 2 (4:1, v/v). Antibi-otics were not included when comparing the growth of the wild-type and mutants. The inocula were 0.1 ml of cultures (ϳ10 7 cells) grown in McNA or McNACoM medium. Puromycin (2.5 g/ml) or neomycin (500 g/ml in plates and 1 mg/ml in broth) was added when needed. Before inoculation, 3 mM of sodium sulfide was added as the sulfur source. When grown with elemental sulfur as the sulfur source, 0.1 g of S 0 was added to 5 ml of medium before autoclaving.
Mutagenesis of mmp1354 in M. maripaludis-The replacement of the thiI gene (mmp1354) with a puromycin resistance cassette was made by transformation of the wild-type M. maripaludis strain S2 with pIJA03-mmp1354, which was constructed from the integration vector pIJA03 as described (48 -50). The transformants were plated on McC medium plus puromycin. Puromycin-resistant isolates were restreaked on the same medium, and isolated colonies were then transferred to broth cultures containing 5 ml of McC medium plus puromycin. The genotype of the ⌬mmp1354 mutant (S620) was confirmed by Southern hybridization (data not shown).
For complementation of S620, the mmp1354 gene was cloned into the shuttle vector pMEV2 (49), and the resulting plasmid, pMEV2-mmp1354, was transformed into S620. The transformants were plated on McC medium plus puromycin and neomycin. The complemented strain expressing the wildtype MMP1354 was named S624. The mutations of MMP1354 (including C78A, C265A, C268A, and C348A), were constructed using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA).
The resulting plasmids were individually transformed into the E. coli SG13009[pREP4] strain (Qiagen) for expression of recombinant proteins. The transformed cells were grown in 1 liter of Luria-Bertani (LB) medium supplemented with 100 g/ml ampicillin and 25 g/ml kanamycin at 37°C with shaking until they reached an absorbance at 600 nm of 0.6 -0.8. Isopropyl ␤-D-1-thiogalactopyranoside was added to a final concentration of 1 mM to induce overnight production of recombinant proteins at 25°C. For anaerobic protein purification, the harvested E. coli cells were transferred into the anaerobic chamber (atmosphere of 95% N 2 and 5% H 2 ) and resuspended in 20 ml of buffer A (20 mM sodium phosphate, 0.5 M NaCl, 20 mM imidazole (pH 7.4)). The cells were disrupted by addition of 2 ml of BugBuster (Novagen). RNA and DNA were digested with 10 l of benzonase (Sigma) by incubation at room temperature for 30 min. The cell lysate was then centrifuged at 100,000 ϫ g for 30 min at 4°C. The supernatant was applied to 1 ml of TALON metal affinity resin (Clontech) equilibrated with buffer A. Proteins bound to the column were eluted with 10 ml of buffer B (20 mM sodium phosphate, 0.5 M NaCl, 200 mM imidazole (pH 7.4)). The elution fractions of the desired proteins were analyzed by SDS-PAGE, dialyzed against buffer C (50 mM of HEPES-NaOH, 150 mM KCl, 10 mM MgCl 2 , 40% (v/v) glycerol (pH 7.0)), concentrated with a 30-kDa cutoff centrifugal filter (Millipore), and stored at Ϫ80°C until use. Protein concentrations were determined with the bicinchoninic acid assay (51). No spectroscopic features indicative of Fe-S clusters were present.

4-Thiouridine Biosynthesis in Methanococcus maripaludis
Complementation of the E. coli ⌬thiI Mutant with mmp1354-The E. coli ⌬thiI mutant strain JW0413 (from the Keio collection (52)) was transformed with the plasmid pQE2-mmp1354. The transformants were grown in LB or M9 medium supplemented with 1 mM thiamine in the presence of 100 g/ml ampicillin and 25 g/ml kanamycin. The expression of MMP1354 in the complemented strain (JW0413-MMP1354) was induced with 1 mM isopropyl ␤-D-1-thiogalactopyranoside and confirmed by Western blotting with the monoclonal mouse anti-His tag antibody (Sigma).
Isolation, Digestion, and Nucleoside Analysis of total tRNAs-The total tRNAs were isolated from M. maripaludis by a modification of the method of Gupta (53). M. maripaludis strains were grown in 200 ml of McC medium to an absorbance of ϳ1.0 at 600 nm. The cells were collected by centrifugation at 10,000 ϫ g for 30 min at 4°C and suspended in 0.5 ml of 10 mM Tris-HCl buffer (pH 7.7). Then the cell suspension was mixed with 0.5 ml of phenol and centrifuged at 16,000 ϫ g for 10 min at 4°C. The aqueous phase was transferred to a new microcentrifuge tube and mixed with 0.1 volume of cold 20% (w/v) potassium acetate (pH 5.0). The nucleic acids were precipitated with 2 volumes of ethanol at Ϫ80°C for 30 min and collected by centrifugation. Low molecular weight nucleic acids were extracted twice from the pellet with 0.1 ml of cold 1 M NaCl, precipitated again with 2 volumes of ethanol at Ϫ80°C for 30 min, and collected by centrifugation. The pellet was dissolved in 0.1 ml of the buffer containing 10 mM Tris-HCl (pH 7.5) and 5 mM MgCl 2 and incubated with 10 units of RNase-free DNase (New England Biolabs) for 30 min at 37°C. The solution was then extracted with 1 ml of chloroform, and the tRNAs were precipitated with 0.1 volume of cold 20% (w/v) potassium acetate (pH 5.0) and 2 volumes of ethanol at Ϫ80°C for 30 min and collected by centrifugation. The pellet was dissolved in 0.4 ml of 0.5 M Tris-HCl buffer (pH 8.8) and incubated for 1 h at 37°C to deacylate aminoacyl-tRNAs. The tRNAs were precipitated again as described above. The pellet was dissolved in 0.5 ml of 0.3 M sodium acetate (pH 7.0) on ice, followed by dropwise addition of 0.27 ml of cold isopropyl alcohol. The mixture was raised to room temperature and centrifuged at 16,000 ϫ g for 10 min at room temperature. The tRNAs in the supernatant were precipitated with dropwise addition of 0.98 volume of isopropyl alcohol and collected by centrifugation at room temperature. The tRNA preparations were rinsed with ethanol and dried under vacuum.
The tRNAs were digested by the method of Mesbah et al. (54). About 70 -250 g of tRNAs were dissolved in 70 l of HPLC water, heated in a boiling water bath for 2 min, and placed immediately on ice. Then 5 l of 0.3 M sodium acetate (pH 5.1), 5 l of 20 mM zinc sulfate, and 1 unit of S1 nuclease (Sigma) were added to each sample. The samples were incubated at 37°C for 2 h. The nucleic acids were then dephosphor-ylated with 1 unit of alkaline phosphatase (Sigma) in 10 mM glycine-HCl (pH 8.0 -8.5) at 37°C for 2 h.
The digested nucleic acids were analyzed with a Waters 2695 separation HPLC system. For each tRNA preparation, 90 l of sample (20 g of tRNAs) were loaded onto an Altima C-18 reversed phase column and eluted with the buffer containing 20 mM triethylamine phosphate and 7.5% methanol (pH 5.1) at 30°C. The nucleosides were monitored at 260 nm, and s 4 U ( max at 334 nm) was monitored at 330 nm.
Sulfur Transfer Assay with Radioactive Sulfur-The procedure was carried out in the anaerobic chamber with an atmosphere of 95% N 2 and 5% H 2 . The maltose-binding proteintagged-IscS (5 M The reaction was stopped by addition of nonreducing SDS loading dye. The protein mixture was then separated by SDS-PAGE, and the radioactivity retained on the gel was followed by autoradiography. Identification of the Persulfide Modification on MMP1354 by Mass Spectrometry-The anaerobically purified recombinant MMP1354 (50 M) in the 0.1 M ammonium bicarbonate/formic acid buffer (pH 7.2) was digested with 20 g/ml trypsin (Promega) at 37°C overnight. The digestion was stopped with 1% (v/v) formic acid, and the samples were dried in vacuum. The peptides obtained from the trypsin digestion were analyzed with LC-MS/MS as described (56).
Assay of in Vitro Formation of s 4 U-The M. jannaschii tRNA Cys substrate was synthesized by in vitro T7-RNA polymerase run-off transcription as described (57). Before use, the tRNA transcripts were folded by heating at 80°C for 5 min, cooling down slowly to 45°C, and adding 5 mM MgCl 2 before placing on ice. The M. jannaschii substrate was used because the M. maripaludis tRNA Cys transcripts failed to fold properly (57).
The in vitro formation of s 4 U was performed anaerobically. The tRNA Cys transcripts (20 M) were incubated with MMP1354 (50 nM) at 37°C for 10 -120 min in the buffer containing 50 mM HEPES-NaOH (pH 7.0), 150 mM KCl, 10 mM MgCl 2 , and 2 mM ATP in the presence of 5 mM sodium sulfide, sodium thiosulfate, sodium thiophosphate, or L-cysteine as the sulfur donor. The reaction was stopped by addition of an equal volume of the formamide loading dye. The tRNA thiolation was then analyzed with the [(N-acryloylamino)phenyl]mercuric chloride (APM)-retardation gel (58). The thiolated and unmodified tRNA Cys (2.5 g per lane) were separated by electrophoresis in 12% urea-polyacrylamide gels supplemented with 9 g/ml APM. The tRNAs were visualized by staining with 0.1% (w/v) toluidine in 40% (v/v) methanol and 1% (v/v) acetic acid. OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44

Phylogenetic Distribution and Conserved Cys Residues of
Archaeal ThiI-ThiI homologs are conserved in most archaea with completely sequenced genomes except some members of the Thaumarchaeota (Cenarchaeum symbiosum A, Nitrosopumilus spp., and Candidatus "Nitrosoarchaeum limnia"), the haloarchaea (Haloquadratum walsbyi and Halorubrum lacusprofundi), and the Sulfolobales (Metallosphaera spp.). The NFLD, the THUMP domain, and the PP-loop domain are present in most archaeal ThiI homologs. However, the C-terminal RLD, which contains the essential Cys 456 for generating the persulfide enzyme adduct in E. coli, is only present in homologs from the Thermoplasmatales (belonging to the Euryarchaeota phylum) and the Thermoproteales (belonging to the Crenarchaeota phylum) (Fig. 1). Interestingly, ThiI from methanogenic archaea and some other archaea (Aciduliprofundum boonei, Archaeoglobus profundus, Staphylothermus spp., Thermosphaera aggregans, and Candidatus "Korarchaeum cryptofilum") have three conserved Cys residues located in the PPloop domain (Fig. 1). Two of them are arranged in a CXXC motif, and the third one is aligned in the sequence at an equivalent position as Cys 344 in the E. coli ThiI, which forms a disulfide bond with Cys 456 after sulfur transfer to tRNA U8 (30,32,33).
MMP1354 Is Required for s 4 U Formation but Not for Thiamine Biosynthesis in M. maripaludis-The physiological function of the M. maripaludis ThiI (MMP1354) was investigated by construction and characterization of a ⌬mmp1354 mutant (S620). This mutant had no growth defects in minimal medium in the absence of thiamine, suggesting that MMP1354, unlike the enteric bacterial ThiI (34 -36), is dispensable for thiamine biosynthesis. The mutant also grew similarly as the wild-type in the presence or absence of cysteine (supplemental Fig. S1), suggesting that MMP1354 is not an essential enzyme under the tested growth conditions.
The involvement of MMP1354 in s 4 U biosynthesis was examined by reverse phase HPLC analysis of the nucleosides derived from total tRNA digestion. In wild-type cells cultivated on rich medium, s 4 U constituted ϳ1.0 Ϯ 0.2% (mol %) of the total tRNA nucleosides, suggesting that ϳ75% tRNAs contained this modification. This level was close to that observed in E. coli (s 4 U is present in ϳ70% tRNAs) (2). The tRNAs isolated from the ⌬mmp1354 mutant (strain S620) cells lacked detectable amounts of s 4 U (Fig. 2), indicating that MMP1354 was essential for s 4 U biosynthesis in vivo. Complementation of S620 with MMP1354 expressed from a shuttle vector (strain S624) restored production of s 4 U (Fig. 2), indicating that the lack of s 4 U production in S620 was not caused by a polar genetic effect.
MMP1354 Complements an E. coli ⌬thiI Mutant for s 4 U Formation-To determine whether archaeal ThiI without the RLD is functional in E. coli, a complementation test of an E. coli ⌬thiI mutant strain (JW0413) with mmp1354 was performed. The JW0413 mutant requires thiamine for growth and is unable to form s 4 U. Upon complementation, the strain expressing MMP1354 (JW0413-MMP1354) still required thiamine for growth (data not shown), suggesting that MMP1354 was unable to transfer sulfur for thiamine biosynthesis in E. coli. This observation agreed with the proposal that the RLD of ThiI is necessary and sufficient for thiamine biosynthesis (35). In contrast, strain JW0413-MMP1354 contained s 4 U in its tRNAs (Fig. 3), suggesting that MMP1354 without RLD was sufficient for s 4 (29 -33). Because both Cys 265 and Cys 268 were important for MMP1354 activity, three methods were applied to investigate whether MMP1354 also forms a persulfide intermediate.

4-Thiouridine Biosynthesis in Methanococcus maripaludis
First, the amount of persulfide in the recombinant MMP1354 expressed in E. coli was measured with a fluorescent labeling assay (17). The protein was purified anaerobically in the absence of reducing agents; the free sulfhydryl groups of the purified protein were labeled with I-AEDANS, a fluorescent derivative of iodoacetamide (55); after removal of unreacted fluorescent dye by centrifugal filtration, the protein was treated with 100-fold excess of DTT. If persulfide is present in the protein, the fluorophore would be released from the protein into solution upon reduction with DTT. In three independent treatments of 1 nmol of wild-type protein, 87 Ϯ 2 pmol of the fluorophore were released (Table 2). This suggests that ϳ9% of the purified protein contained persulfide if only one Cys residue forms persulfide per protein molecule. The C78A variant had similar amounts of persulfide as the wild-type protein, whereas the C265A, C268A, and C348A variants contained no detectable amount of persulfide (Table 2). This result suggests that Cys 265 , Cys 268 , and Cys 348 are essential for the generation or stabilization of the persulfide in MMP1354.
Second, the generation of persulfide by MMP1354 was followed by a sulfur transfer assay with [ 35 S]Cys as the sulfur source (31). Anaerobically purified recombinant MMP1354 was incubated with [ 35 S]Cys in the absence or presence of the maltose-binding protein-tagged E. coli IscS. The protein(s) was then subjected to SDS-PAGE analysis under nonreducing conditions, and the sulfur transfer from [ 35 S]Cys to protein(s) was monitored by autoradiography. MMP1354 was not labeled with [ 35 S]Cys alone, demonstrating the absence of cysteine desulfurase activity (lane 2 of Fig. 4). However, when wild-type MMP1354 together with IscS were incubated with [ 35 S]Cys, both proteins were radiolabeled (lane 3 of Fig. 4), indicating that MMP1354 can accept 35 S from IscS. Addition of the reducing agent ␤-mercaptoethanol (1%, v/v) to the protein mixture removed the radiolabel (lane 10 of Fig. 4), suggesting that 35 S was attached to both proteins as persulfide. Although an IscS homolog is not encoded in M. maripaludis, this experiment suggested that a persulfide can be generated on MMP1354 with a persulfide sulfur donor. Alteration of Cys 265 , Cys 268 , or Cys 348 to Ala abolished the 35 S labeling of MMP1354 (lanes 4 -9 of Fig.  4), which further confirmed that these three cysteines are essential for the generation or stabilization of the persulfide under in vitro conditions. Third, the location of persulfide in MMP1354 was analyzed with mass spectrometry. The anaerobically purified recombinant protein was digested with trypsin, and the resulting peptides were then analyzed by LC-MS/MS (Fig. 5). For the peptide 261 DKYTCLYCK 269 , the precursor ions can be observed with Ϫ2 Da (mass of Ϫ2H) and ϩ30 Da (mass of Ϫ2H ϩ 1S) shift. Analysis of the MS/MS spectra of the Ϫ2 Da shifted precursor ion matched the expected amino acid sequence when the the-    OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 oretical masses were corrected with an intra-peptide disulfide between Cys 265 and Cys 268 (Fig. 5A). The fragmentation of the ϩ30 Da shifted precursor ion matched the modification as a trisulfide linkage between Cys 265 and Cys 268 (Fig. 5B). This fragmentation pattern was similar to that observed for the M. jannaschii O-phosphoseryl-tRNA:Cys-tRNA synthase, which contained a trisulfide linkage in a CXXC motif, presumably resulting from an oxidation (loss of dihydrogen) of a Cys persulfide and a Cys thiol or from a reaction between two persulfide groups to expel a bisulfide (56). The trisulfide linkage was possibly formed either in vivo or during the protein purification and the mass spectrometry processes due to the instability of persulfide. No modification was observed for the peptide containing Cys 348 . Overall, the mass spectrometry analysis of MMP1354 suggests that Cys 265 and Cys 268 easily form a disulfide bond and one of them carries a sulfane sulfur in a proportion of the protein.

4-Thiouridine Biosynthesis in Methanococcus maripaludis
In Vitro Formation of s 4 U by MMP1354 with Sulfide as the Sulfur Donor-The formation of s 4 U by MMP1354 with different sulfur donors was tested using M. jannaschii tRNA Cys as a substrate. As shown by APM-retardation gel analysis, only Na 2 S resulted in significant thiolation of tRNA Cys (Fig. 6A). The K m for Na 2 S was 1.0 Ϯ 0.2 mM, which was within the range of intracellular free sulfide concentration in methanococci of 1-3 mM (59). This result suggests that MMP1354 has a much higher affinity for sulfide than E. coli ThiI (K m Ͼ20 mM) (26), and sulfide is a physiologically relevant sulfur donor for s 4 U biosynthesis in M. maripaludis. Furthermore, unlike the E. coli ThiI that requires exogenous reductant for multiple turnovers (32), the addition of DTT inhibited the s 4 U formation by MMP1354 (Fig.   6B). This result suggests that an oxidized form of MMP1354 (presumably with a disulfide) is important to initiate the catalysis and the sulfur donor is a S 2Ϫ equivalent. Alteration of Cys 265 , Cys 268 , or Cys 348 to Ala abolished tRNA thiolation (Fig.  6C), suggesting that these three cysteines are required for the in vitro activity of MMP1354.

DISCUSSION
Proposed Model of s 4 U Formation in M. maripaludis-The data presented here demonstrate that MMP1354 requires two conserved Cys residues (Cys 265 and Cys 268 ), which are arranged in a CXXC motif in the PP-loop domain, for both in vitro and in vivo activities to generate s 4 U in tRNAs. Furthermore, both cysteines are essential for the formation of a persulfide enzyme adduct and also readily form a disulfide linkage, as identified by mass spectrometry. The recruitment of the CXXC motif for sulfur transfer resembles the sulfur relay mechanism of O-phosphoseryl-tRNA:Cys-tRNA synthase, which catalyzes the conversion of tRNA-bound O-phosphoserine to cysteine in methanogenic archaea (56). Based upon these findings, a model for catalysis by MMP1354 is proposed in Fig. 7. First, a sulfur donor (equivalent of S 2Ϫ ) attacks the Cys 265 -Cys 268 disulfide linkage to generate a persulfide on either Cys 265 or Cys 268 , leaving the other Cys as a free thiol. Then, a thiolate derived from deprotonation of the free thiol attacks the bridging sulfur of the persulfide to liberate the terminal sulfur. After donation of the sulfur (in the Ϫ2 oxidation state) to form s 4 U, the disulfide (with both sulfurs in the Ϫ1 oxidation state) is consequently regenerated. No exogenous electron donor or acceptor is required for the catalytic cycle.
The in vivo and in vitro characterizations of MMP1354 yielded apparently contradictory results on the importance of Cys 348 , which is at an equivalent position as Cys 344 in E. coli ThiI. In vivo, the alteration of Cys 348 to Ala had a moderate effect on the s 4 U level, suggesting an important but not essential role. However, in vitro the C348A variant protein had no detectable activity for s 4 Fig. S2A). In addition, Cys 268 is adjacent to a highly conserved Lys or Arg residue (Arg 265 of B. anthracis ThiI), which is in contact with the phosphate group of AMP (24). Therefore, these two catalytic cysteines are presumably located near the adenylated form of U8. Second, Cys 265 and Cys 268 are at the N terminus of an ␣-helix, which is a conserved location of a CXXC motif in the thiol:disulfide oxidoreductase superfamily (60). In this superfamily, the CXXC motif is essential for the catalysis of     OCTOBER S2B). These observations are in agreement with the proposal that Cys 265 and Cys 268 form persulfide and disulfide intermediates for sulfur transfer to the C4 atom of U8.

4-Thiouridine Biosynthesis in Methanococcus maripaludis
Comparisons of Bacterial and Archaeal ThiI-In summary, MMP1354 is clearly different from bacterial ThiI. First, the E. coli ThiI is a dual-function enzyme required for both s 4 U and thiamine biosynthesis. MMP1354 is necessary for the biosynthesis of s 4 U but not of thiamine. This result is consistent with the recent reports that the C-terminal RLD of bacterial ThiI, and B. subtillis ThiI are not required for thiamine biosynthesis (37). Given the conserved function of ThiI for tRNA modification in bacteria and archaea, tRNA modification is likely the ancestral function, and the protein was subsequently recruited for thiamine biosynthesis in certain bacteria. This conclusion is supported by the observation that several genes encoding the bacterial pathway for thiamine biosynthesis, e.g. the sulfur donor protein ThiS (36), cannot be readily recognized in methanococcal genomes. Thus, methanococci apparently have a different route of sulfur transfer during thiamine biosynthesis. The thiamine levels in methanogens are about 5-fold lower than that in E. coli (63), but it is unclear whether the low levels of thiamine in methanogens are correlated with a unique sulfur transfer route. Second, ThiI in different organisms has acquired distinct strategies for sulfur transfer. ThiI in E. coli and S. enterica uses a Cys residue in the RLD to generate a persulfide for sulfur transfer from IscS to U8; ThiI in B. subtilis cooperates with a specialized cysteine desulfurase NifZ to transfer sulfur presumably through a persulfide (37), although the catalytic Cys has not been identified; and ThiI in methanogens and some other archaea uses a conserved CXXC motif in the PP-loop domain to generate a persulfide for sulfur transfer. Consistent with the importance of the CXXC motif in archaeal ThiI, s 4 U is not observed in the tRNAs from Haloferax volcanii (64) and Solfolobus solfataricus (65), which possess ThiI homologs lacking the CXXC motif. Given the similar sulfur transfer mechanisms of methanogen ThiI and O-phosphoseryl-tRNA:Cys-tRNA synthase, the utilization of a CXXC motif to generate a persulfide is possibly a common scheme of sulfur chemistry in methanogens. Third, MMP1354 has much higher affinity for sulfide than E. coli ThiI. The K m value of MMP1354 for Na 2 S is ϳ1 mM and is within the range of the intracellular level of free sulfide, suggesting that sulfide is a physiologically relevant sulfur donor. Methanococci are well adapted to live in sulfide-rich environments as they lack many targets of sulfide toxicity (46). They also use sulfide instead of free cysteine as the sulfur source for Fe-S cluster biosynthesis (46), which suggests that sulfide plays an important role in sulfur traffic in methanogens. Therefore, it is likely that sulfide replaces cysteine and cysteine desulfurase to generate persulfide on some sulfur carrier proteins. Presumably, methanogens have evolved explicit mechanisms to control the specificities of sulfur incorporation mediated with sulfide and persulfide under anaerobic conditions. These processes may provide a paradigm for studying ancient sulfur metabolism on the early anoxic earth.