Redundant Synthesis of Cysteinyl-tRNACys in Methanosarcina mazei*

A subset of methanogenic archaea synthesize the cysteinyl-tRNACys (Cys-tRNACys) needed for protein synthesis using both a canonical cysteinyl-tRNA synthetase (CysRS) as well as a set of two enzymes that operate via a separate indirect pathway. In the indirect route, phosphoseryl-tRNACys (Sep-tRNACys) is first synthesized by phosphoseryl-tRNA synthetase (SepRS), and this misacylated intermediate is then converted to Cys-tRNACys by Sep-tRNA:Cys-tRNA synthase (SepCysS) via a pyridoxal phosphate-dependent mechanism. Here, we explore the function of all three enzymes in the mesophilic methanogen Methanosarcina mazei. The genome of M. mazei also features three distinct tRNACys isoacceptors, further indicating the unusual and complex nature of Cys-tRNACys synthesis in this organism. Comparative aminoacylation kinetics by M. mazei CysRS and SepRS reveals that each enzyme prefers a distinct tRNACys isoacceptor or pair of isoacceptors. Recognition determinants distinguishing the tRNAs are shown to reside in the globular core of the molecule. Both enzymes also require the S-adenosylmethione-dependent formation of m1G37 in the anticodon loop for efficient aminoacylation. We further report a new, highly sensitive assay to measure the activity of SepCysS under anaerobic conditions. With this approach, we demonstrate that SepCysS functions as a multiple-turnover catalyst with kinetic behavior similar to bacterial selenocysteine synthase and the archaeal/eukaryotic SepSecS enzyme. Together, these data suggest that both metabolic routes and all three tRNACys species in M. mazei play important roles in the cellular physiology of the organism.

Methanosarcina mazei strain Gö1 is a methanogenic archaeon capable of utilizing a broad spectrum of growth substrates, including H 2 /CO 2 , acetate, methanol, and methylamines (1). The capacity of M. mazei to adapt to changing environmental conditions is a hallmark of the Methanosarcina. Organisms belonging to this genus coexist with fermenting bacteria in freshwater and marine sediments, where their capacity to metabolically transform bacterially produced acetate into methane forms a closed carbon cycle that helps to determine atmospheric levels of this greenhouse gas (2). In some marine environments, the Methanosarcina also flourish together with methanotrophs and sulfate-reducing bacteria, where they participate in a microbial community that generates very high sulfate reduction rates via anaerobic methane oxidation (3). Comparative genomic studies of M. mazei, Methanosarcina barkeri, and Methanosarcina acetivorans C2A, the three Methanosarcina for which complete sequences are available, show that the 4.1-megabase M. mazei genome is the smallest so far known in the genus and may approximate the ancestral state (1,4,5). Of the 3371 M. mazei open reading frames identified, 1043 find their closest homolog in the bacterial domain (1), suggesting that horizontal gene transfer has played an important evolutionary role in the development of the organism.
The distinguishing methanogenesis pathway, present in all methanogens, produces methane as a catabolic product while also generating an electrochemical gradient that is harnessed to synthesize ATP (5,6). Methanogens also possess many subsidiary metabolic activities that function to directly support methane production, including those involved in cofactor biosynthesis and in energy transduction through proton and sodium gradients (6). Interestingly, genomic data reveal a further correlation; all but two methanogens for which full sequence information is available also possess a unique two-step pathway by which cysteine is incorporated into proteins (7). In this pathway, tRNA Cys is first aminoacylated with the nonstandard amino acid phosphoserine by phosphoseryl-tRNA synthetase (SepRS). 2 The misacylated Sep-tRNA Cys is then converted to Cys-tRNA Cys through the action of the PLP-dependent enzyme SepCysS. For some methanogens, which lack a canonical cysteinyl-tRNA synthetase (CysRS), this indirect route is the sole pathway to synthesis of Cys-tRNA Cys . However, other methanogens, including all three Methanosarcina for which genome sequences are available, possess both pathways. Comparative phylogenetic studies have shown that both the direct CysRS route and the indirect two-step pathway were in existence at the time of the last universal common ancestor (8). The coincident phylogenetic distributions of SepRS/SepCysS and the enzymes of methanogenesis also suggest the existence of a linkage between primary energy production and protein synthesis in contemporary methanogens, although the nature of and biological rationale for such a linkage have so far been obscure (8).
The only nonmethanogen known to possess the two-step Cys-tRNA Cys biosynthetic pathway is the hyperthermophilic sulfate-reducing archaeon Archaeoglobus fulgidus, a closely related archaeon that retains nearly all of the methanogenesis enzymes and cofactors needed for the hydrogenotrophic pathway, with the exception of the essential methyl-CoM reductase (9).
Recently, it was reported that SepRS and CysRS from Methanococcus maripaludis share major identity elements for heterologous aminoacylation of an unmodified, in vitro transcript corresponding to the single tRNA Cys species present in the M. jannaschii genome (10). Consistent with other studies in which a common tRNA is recognized by distinct tRNA synthetases, these data were interpreted to support a general hypothesis that aminoacyl-tRNA synthetases evolved in the context of preestablished tRNA identity. Other work on the M. jannaschii pathway has shown that SepRS and SepCysS form a stable binary complex that assists in conversion of the intermediate Sep-tRNA Cys to Cys-tRNA Cys , a complex reminiscent of the transamidosome particle that facilitates prokaryotic tRNA-dependent asparagine biosynthesis (11,12).
Here, we sought to address distinct questions pertaining to the contemporary function of apparently redundant Cys-tRNA Cys synthesis pathways in M. mazei. Both M. mazei and M. barkeri possess genes encoding three distinct isoacceptors of tRNA Cys , a feature not known in other bacteria or archaea and suggestive that the machinery for incorporation of cysteine into proteins in these organisms may interact with other cellular components (1,4,8). By examining aminoacylation of the three wild-type as well as a set of hybrid tRNA Cys species as substrates for either CysRS or the combined SepRS/SepCysS pathway, we show that each enzyme prefers a distinct tRNA Cys isoacceptor or pair of isoacceptors and that the nucleotides determining CysRS versus SepRS selectivity among tRNA Cys species are located in the globular core of the tRNA. Through the development and application of a new anaerobic assay for the quantitative analysis of SepCysS kinetics, we also provide evidence that the tRNA Cys isoacceptor preferences of SepRS and Sep-CysS are identical. This supports the notion that tRNA Cys acceptors functioning in the two-step pathway could be efficiently shuttled between these two enzymes. Together, these findings suggest that both pathways may play an important role in the cellular physiology. We also show that the aminoacylation efficiency of M. mazei CysRS is very significantly enhanced by the addition of a methyl group to the N1 position of G37 by the S-adenosylmethione-dependent Trm5 enzyme, consistent with a similar dependence found for SepRS (11,13). Since S-adenosylmethione is synthesized from methionine (14), the tRNA Cys methylation requirement for aminoacylation by both CysRS and SepRS links protein synthesis with sulfur metabolism in this organism.

EXPERIMENTAL PROCEDURES
Expression, Purification, and Quaternary Structure Determination of M. mazei CysRS and SepCysS-M. mazei genomic DNA was purchased from the American Type Culture Collection and used as template for PCR amplification of the CysRS and SepCysS open reading frames. For SepCysS, the resulting DNA fragments were digested with NdeI and XhoI and inserted into pet22bϩ for expression of His-tagged proteins in Escherichia coli Rosetta2(DE3) pLysS cells. Cultures were grown aerobically at 37°C in LB medium supplemented with 100 g/ml ampicillin, 34 g/ml chloramphenicol, and 0.01% pyridoxine. When the cultures reached an A 600 of 0.5, expression of Histagged protein was induced by adding isopropyl-␤-D-thiogalactoside to a final concentration of 0.6 mM. Cultures were then grown for a further 5 h at 37°C prior to harvesting.
Cells expressing SepCysS were resuspended in a buffer containing 50 mM NaH 2 PO 4 (pH 8.0), 20 mM NaCl, 3 mM DTT, 20 M PLP, and 10% glycerol and were disrupted by a French press. Under aerobic conditions, the clarified lysate was applied to a 30-ml DEAE column preequilibrated with 50 mM NaH 2 PO 4 (pH 8.0), 20 mM NaCl, 3 mM DTT, and 20 M PLP. SepCysS was eluted with a gradient from 20 -500 mM NaCl, and the pooled fractions were dialyzed into a solution containing 50 mM NaH 2 PO 4 (pH 8.0), 150 mM NaCl, 5 mM ␤ME, and 20 M PLP. The enzyme was then applied to a Superdex 200 gel filtration column preequilibrated with 50 mM NaH 2 PO 4 (pH 8.0), 150 mM NaCl, 5 mM ␤ME, and 20 M PLP. Fractions containing SepCysS were then directly applied to a Ni 2ϩ -nitrilotriacetic acid column equilibrated with 50 mM NaH 2 PO 4 (pH 8.0), 300 mM NaCl, 5 mM ␤ME, and 20 M PLP. After washing of the column in this buffer, the enzyme was then eluted by the addition of 150 mM imidazole. The enzyme was then dialyzed into a solution containing 10 mM NaH 2 PO 4 (pH 7.4), 20 mM NaCl, 2 mM MgCl 2 , 1 mM ␤ME, 100 M PLP, and 50% glycerol and stored at Ϫ20°C. His-tagged SepCysS was recovered at better than 95% purity as judged by SDS-polyacrylamide gel electrophoresis, at a yield of 2 mg of protein/liter of culture. The PLP content of SepCysS was verified using a colorimetric assay in which the titration of PLP released upon boiling is monitored (15).
Construction of the recombinant DNA vector, induction of enzyme expression, and purification of M. mazei CysRS were accomplished by procedures identical to those described for M. mazei SepRS (13). The yield of highly purified CysRS was ϳ5 mg of protein/liter of culture.
The oligomeric structures of CysRS and SepCysS were determined by analytic size exclusion chromatography, using conditions and molecular mass markers identical to those reported in the accompanying paper (13). The 473-amino acid M. mazei CysRS enzyme (calculated M r ϭ 53,861) exhibits 40.2% sequence identity with the monomeric E. coli CysRS and migrates as a monomer, as expected. M. mazei SepCysS belongs to the PLP-dependent superfamily of enzymes, which typically function as dimers with one molecule of PLP bound in each active site (16). As expected, the 386-amino acid enzyme (predicted M r ϭ 42,324 Da) elutes at a position corresponding to a molecular mass of 84 kDa, consistent with dimer formation (13).
Steady-state Aminoacylation Kinetics-SepRS plateau aminoacylation levels and assays to determine steady-state kinetic parameters were carried out using tRNA 32 P-labeled at the 3Ј-internucleotide linkage, using the methodology reported in the accompanying paper (13,17). Measurements of aminoacylation by CysRS with all tRNA Cys species were performed with a conventional assay that tracks incorporation of 35 S-labeled cysteine into tRNA, as described (18). Saturation for tRNA was not observed in any CysRS reaction, limiting the analysis to determination of k cat /K m . k cat /K m was derived by fitting initial velocities to the following linear relation, where (v/S) is the slope of the linear fit, and [E] t is the enzyme concentration. Errors are reported in Tables 1-3 as the mean and S.D. values of at least two measurements.
SepCysS Activity Assay-SepCysS was assayed anaerobically under an atmosphere of 95% N 2 , 5% H 2 /CO 2 in an anaerobic chamber (Coy Laboratories). All enzymes, buffers, and substrates entering the chamber were first deoxygenated by five cycles of vacuuming followed by purging with argon (19). Anoxic solutions were verified by testing larger volumes with a BBL anaerobic test strip. Anoxic solutions were kept in sealed glass vials and stored at Ϫ20°C until needed.
The Sep-tRNA Cys substrate needed for the assay of SepCysS was generated in one of two ways. A 2-fold excess of SepRS, usually 30 -40 M, was incubated with 10 -20 M tRNA Cys in a buffer containing 50 mM Tris (pH 7.5), 20 mM KCl, 10 mM MgCl 2 , 5 mM ATP, 3 mM phosphoserine, and 5 mM DTT for 8 min. 32 P-labeled tRNA sample mixes were prepared as described above for aminoacylation assays. Aliquots were removed for evaluation of plateau aminoacylation levels. After phosphoserylation, the reactions were diluted 2-fold in water, and concentrated sodium acetate (pH 5.0) was then added to a final concentration of 200 mM to acidify the reaction and prevent tRNA deacylation. The reaction was then extracted with phenol/chloroform/isoamyl alcohol, and the phosphoserylated tRNA was recovered by ethanol precipitation. tRNA was then resuspended in the anaerobic chamber in an anoxic solution consisting of 20 mM sodium cacodylate (pH 6.0), 5 mM MgCl 2 . Aminoacylation levels were requantified by removing a small aliquot and digesting with P1 nuclease followed by TLC separation as described above. The aminoacylated tRNA was then stored in a sealed glass vial at Ϫ20°C until needed.
Alternatively, Sep-tRNA Cys was generated in the same reaction as used for the SepCysS assay, by first performing the aminoacylation reaction with SepRS under anaerobic conditions. Excess SepCysS and sulfur donor are then added to initiate the SepCysS activity assay. In this approach, 20 M 32 P-labeled tRNA Cys was first heated to 85°C and slowly cooled under anaerobic conditions in the presence of 2 mM MgCl 2 . Sep-tRNA Cys was then synthesized in a buffer consisting of 50 mM Tris (pH 7.5), 20 mM KCl, 10 mM MgCl 2 , 5 mM ATP, 3 mM phosphoserine, and 5 mM DTT. The phosphoserylation reaction was initiated by adding SepRS to 20 -30 M and incubating for 8 min at 37°C. SepCysS was preincubated at 37°C in a buffer containing 10 mM NaH 2 PO 4 (pH 7.4), 20 mM NaCl, 2 mM MgCl 2 , 1 mM ␤ME, 100 M PLP, and 50% glycerol and was then added to concentrations of 0.05-40 M, followed immediately by sodium sulfide to a concentration of 5 mM. Typical concentrations of Sep-tRNA Cys used in the second reaction, after dilution with reaction components, were 5-10 M. In reactions where Sep-tRNA Cys was generated outside of the anaerobic chamber, it was first preincubated in the chamber at 37°C with 50 mM Tris (pH 7.5), 20 mM KCl, 10 mM MgCl 2 , and 5 mM DTT for several minutes before the addition of SepCysS and sodium sulfide. Anhydrous sodium sulfide was purchased from Aldrich (product 407410) and was opened and stored in an anaerobic environment. The pH of cysteine and sodium sulfide stock solutions was adjusted to neutrality.
Sep-tRNA Cys , Cys-tRNA Cys , an intermediate-tRNA Cys (each as 3Ј-aminoacylated A76), and nonreacted substrate (as unmodified AMP) were separated by TLC on polyethyleneimine-cellulose plates, which were run in 1 M acetic acid with adjustment of the pH to 3.5 using concentrated ammonium hydroxide. These conditions for TLC separation were reoptimized for visualization of the Cys-AMP product and differ from those employed to characterize SepRS reactions alone (13). Quantitation was done by phosphorimaging analysis, as described previously (17). Time courses from 0.5 to 60 min were fit to linear or sigmoidal equations. Results using either method of Sep-tRNA Cys preparation were identical.
tRNA Binding to SepCysS-Equilibrium fluorescence titrations were performed at ambient temperature with 0.25 M M. mazei SepCysS in 20 mM Tris-HCl (pH 7.5), 50 mM NaCl, and 5 mM ␤ME. Tryptophan fluorescence was excited at 295 nm, and the emission was monitored from 300 to 400 nm. An emission wavelength of 350 nm was used to quantitate binding after correction for dilution and for intrinsic quenching due to RNA absorbance at 295 nm, according to the formula, F c ϭ F obs ϫ antilog((A 295 ϩ A 350 )/2) (20). Control solutions of bovine serum albumin or of tryptophan showed no fluorescence response to tRNA. tRNA was titrated in the concentration range of 0.05-10 M. The K d values were derived by fitting the data to a quadratic binding equation using Scientist software.

Differential Utilization of tRNA Cys Isoacceptors by SepRS and
CysRS-We chose to study Cys-tRNA Cys formation in M. mazei because genetic techniques have been developed in this organism (21) and because its biosynthetic apparatus is expanded to its known limit in M. mazei and M. barkeri, suggesting the possibility that the macromolecules involved may be regulated in unusual ways or may perform physiological roles unrelated to protein synthesis. Both of these Methanosarcina possess all three enzymes corresponding to the two distinct pathways, CysRS, SepRS, and SepCysS, as well as three tRNA Cys isoacceptors ( Fig. 1). In contrast, M. acetivorans, the other Methanosarcina for which the genome sequence is available, possesses all three enzymes but only one tRNA Cys acceptor that reads both the UGU and UGC codons. Among the other sequenced methanogens, M. maripaludis and Methanospirillum hungatei possess biosynthetic apparati with identical components to M. acetivorans, whereas Methanothermobacter thermautotrophicus, M. jannaschii, Methanopyrus kandleri, and Methanococcoides burtonii possess only the two-step indirect pathway with a single tRNA Cys isoacceptor. Only the human intestinal archaea Methanosphaera stadtmanae and Methanobrevibacter smithii provide examples of methanogens possessing just the canonical CysRS with one tRNA Cys acceptor and lacking the SepRS/SepCysS pathway (22,23).
Among all sequenced bacteria and archaea, the presence of three tRNA Cys isoacceptors is found only in M. mazei and M. barkeri. Indeed, analysis of genomic data shows that only 10 bacterial or archaeal species possess even two tRNA Cys isoacceptors. In addition to the GCA anticodon, all three M. mazei isoacceptors possess the U73 discriminator base typical of tRNA Cys species; these four nucleotides form a key part of the "identity set" for recognition by both CysRS and SepRS (10,24). tRNA 2 Cys and tRNA 3 Cys are of canonical structure and are highly similar to each other, being distinguished only by the base pair at position 7-66 at the bottom of the acceptor stem (Fig. 1). However, tRNA 1 Cys possesses several unusual features: the presence of a rare purine at position 33 (found in only four other known tRNAs, two of which are also tRNA Cys acceptors from methanogens), a GU wobble pair at position 51-63 of the T-stem, and an A65-C49 mismatch at the inside of the T-stem (found in only one other known tRNA, the homologous tRNA 1 Cys isoacceptor from M. barkeri). tRNA 1 Cys also differs from the other two M. mazei tRNA Cys isoacceptors in featuring A57 in the T-loop as well as sequence differences at positions 20, 21, 44, 46, and 47 in the D-loop and variable loops. The last five differences probably produce distinct base triples and/or stacking interactions in the augmented D-stem region of the globular core. Despite the unusual structure, analysis of the genomic DNA surrounding tRNA 1 Cys reveals the presence of both the A-box and B-box transcriptional initiation motifs (25). Therefore, this unusual tRNA may well be expressed in vivo.
Resequencing of the genomic DNA in this region also revealed no errors in the deposited sequence (data not shown).
We sought to explore the biological rationale for the existence of redundancy in the machinery for Cys-tRNA Cys synthesis by determining the efficiency by which each of the three tRNAs is processed by the two pathways. The three tRNAs were synthesized by in vitro transcription, were each methylated at position m1 G37 using recombinant M. jannaschii Trm5, and were purified from unmethylated transcripts (see "Experimental Procedures" and Refs. 11 and 13). SepRS aminoacylation  Table 1 is as follows: tRNA Cys ⌬2, introduction of A33 from tRNA 1 Cys into tRNA 3 Cys ; tRNA Cys ⌬3, introduction of A57 from tRNA 1 Cys into tRNA 3 Cys ; tRNA Cys ⌬5, introduction of both C49 and U51 from tRNA 1 Cys into tRNA 3 Cys ; tRNA Cys ⌬4, introduction of the five D-loop and variable loop nucleotides U20, C21, U44, A46, and G47 from tRNA 1 Cys into tRNA 3 Cys . kinetics was measured as described (13) (Fig. 1A), whereas CysRS aminoacylation kinetics was measured using a filter binding assay in which [ 35 S]cysteine is attached to tRNA (18). Preliminary plateau aminoacylation assays showed that SepRS aminoacylates each transcript to ϳ1300 pmol/OD (78 -85% aminoacylation), whereas CysRS aminoacylates the transcripts to lower levels (33-52% aminoacylation) ( Table 1). As observed for SepRS (11,13), plateau levels for aminoacylation of all three tRNA Cys species by M. mazei CysRS were very low in the absence of methylation by Trm5 (Fig. 2). Steady-state parameters for phosphoserylation and cysteinylation of the three isoacceptors revealed that each enzyme possesses distinct preferences. SepRS preferentially aminoacylates tRNA 2 Cys and tRNA 3 Cys by a factor of 4 -5-fold as compared with tRNA 1 Cys , with the selectivity manifested almost entirely in k cat (Table 1). In contrast, CysRS aminoacylates tRNA 1 Cys 5-9-fold more efficiently than the other isoacceptors. We also find that SepRS is the better functioning enzyme for all isoacceptors; phosphoserylation is favored over cysteinylation by 140 -220fold for tRNA 2 Cys and tRNA 3 Cys and by 9.5-fold for tRNA 1 Cys . The k cat of 0.5-0.6 s Ϫ1 toward preferred tRNAs by SepRS is comparable with many other tRNA synthetases, including the M. jannaschii SepRS (11), and is ϳ5-fold greater than reported for phosphoserylation of fully unmodified M. jannaschii tRNA Cys by M. maripaludis SepRS (10). The primary defect in aminoacylation by M. mazei CysRS is the inability to observe tRNA saturation at concentrations up to 100 M, limiting analyses to the determination of k cat /K m . However, it is also notable that the unusual structure of tRNA 1 Cys that apparently produces antideterminants for phosphoserylation nonetheless operates to provide positive recognition features for cysteinylation that are not present in tRNA 2 Cys or tRNA 3 Cys . This suggests the operation of an evolutionary selection by which the parallel direct pathway is retained by the organism.
To examine the structural basis for the tRNA Cys isoacceptor preferences of SepRS and CysRS, we constructed a series of tRNA mutants in which structural elements of tRNA 1 Cys were systematically transplanted into tRNA 3 Cys and evaluated the effectiveness of the mutants as CysRS and SepRS substrates by steady-state aminoacylation ( Fig. 1 and Table  1). As observed for the three wild-type isoacceptors, all of the tRNA mutants are aminoacylated by SepRS more efficiently than by CysRS. A33 is not a distinguishing determinant, because introduction of this nucleotide into tRNA 3 Cys does not significantly alter the relative preferences of SepRS and CysRS for that substrate (tRNA Cys ⌬2; Table 1). (In a separate experiment, mutation of A33 to U33 in the tRNA 1 Cys framework also does not shift enzyme preferences (tRNA Cys ⌬1; Table 1)). Incorporation of A57 into tRNA 3 Cys (tRNA Cys ⌬3), however, both decreases the efficiency of phosphoserylation and increases the efficiency of cysteinylation; the SepRS/ CysRS preference of this mutant is decreased from about 140-to 40-fold. Incorporation into tRNA 3 Cys of the tRNA 1 Cys nucleotides in the augmented D-stem/variable loop region (C20U/U21C/A44U/C46A/A47G) has qualitatively similar but more marked effects, so that the SepRS/CysRS preference is decreased to 20-fold (tRNA Cys ⌬4; Table 1). Finally, combining the tRNA Cys ⌬3 and tRNA Cys ⌬4 substitutions generates a mutant that fully recapitulates the SepRS/CysRS preferences of tRNA 1 Cys (tRNA Cys ⌬6; Table 1). Therefore,  A57 together with the five nucleotides U20/C21/U44/A46/ G47 represents the distinguishing identity set (Fig. 1B). Interestingly, the unusual T-stem base pairs in tRNA 1 Cys appear not to play a role in modulating the SepRS/CysRS preference. Independent introduction of these two base pairs into tRNA 3 Cys produced a mutant that could not be efficiently aminoacylated by either enzyme (tRNA Cys ⌬5; Table 1). This suggests that other structural elements in tRNA 1 Cys are required to maintain the structural stability of its unusual core domain.
These data are consistent with studies of other SepRS enzymes. Work in the M. maripaludis system indicates that the G6:C67 base pair, C20, G37, A47, and A59 are putative tRNA Cys identity elements for SepRS (10). tRNA 2 Cys and tRNA 3 Cys of M. mazei each possess all six of these nucleotides, whereas tRNA 1 Cys is missing C20 and A47. Further, the crystal structure of the A. fulgidus SepRS⅐tRNA Cys complex shows that U33 does not contact the enzyme (26), consistent with the ineffectiveness of this nucleotide in modulating the SepRS/CysRS preferences. The importance of T-loop, D-loop, and variable loop nucleotides shown in this work, together with similar findings from studies of the M. maripaludis enzyme, suggest that SepRS makes important interactions with the globular tRNA core. Because the A. fulgidus SepRS crystallized as an unproductive complex in which interactions are made only with the tRNA anticodon (26), further structural analyses will be required to provide a reliable basis for these observations.
The tRNA Cys K m values of 6 -8 M measured toward SepRS are some 10-fold higher than typically observed for highly efficient cognate aminoacylation by tRNA synthetases, and the equivalent values toward CysRS are more than 100-fold higher. It is notable that the addition of the m1 G37 modification, which substantially improves both initial velocities of aminoacylation and aminoacylation plateau values for both enzymes (11, 13) ( Fig. 2), still produces such weak apparent complementarity in the enzyme-tRNA interfaces. It appears quite possible, therefore, that further base or ribose modification may be required for optimal Cys-tRNA Cys formation by SepRS, CysRS, or both enzymes. A positive tRNA modification requirement for efficient aminoacylation has been observed previously in only a small number of cases, primarily involving the wobble nucleotide at the 5Ј-position of the anticodon in the isoleucyl, glutamyl, and glutaminyl systems (27)(28)(29)(30)(31). The only previous observation of a positive tRNA methylation requirement for aminoacylation is the need for 1-methyladenosine at position 9 of nematode mitochondrial tRNAs lacking the T-arm (32). In these cases, however, the methyl group clearly plays a compensating structural role in repairing the tRNA core and may not be directly important for interaction with the tRNA synthetase. It is also of interest to note that the m1 G37 modification functions as a negative determinant to ensure aminoacylation specificity of yeast tRNA Asp ; when G37 is unmodified in that tRNA, it is significantly misacylated by arginyl-tRNA synthetase (33). We know of no tRNA-synthetase system, however, in which modification at two separate nucleotides has been shown to be required for full aminoacylation efficiency in vitro.
Embedded metabolic linkages among macromolecules responsible for synthesis of Cys-tRNA Cys , for methanogen-esis, and for other aspects of sulfur metabolism could help to explain why the presence of the canonical CysRS has not resulted in elimination of the SepRS/SepCysS pathway in methanogens and the related A. fulgidus, as it has in all other archaea (8). The unusually high number of Fe-S cluster proteins in methanogens may provide one rationale for the existence of unique linkages in these organisms (34). We suggest that the S-adenosylmethione-dependent methylation requirement for Cys-tRNA Cys synthesis shown here also points to such linkages, since S-adenosylmethione is synthesized intracellularly from methionine (14). Further, free cysteine is known to be a source of metabolic sulfur for biosynthesis in bacteria (35). Hence, the finding that deletion of the SepRS gene in M. maripaludis leads to cysteine auxotrophy suggested a clear connection between the SepRS/SepCysS pathway and cysteine biosynthesis (7). Finally, the finding that SepCysS remains essential in the context of this SepRS deletion (where Cys-tRNA Cys is provided by CysRS), suggests a further independent role for that enzyme (36).
No correlation is evident between the pathway of cysteine biosynthesis and the number of tRNA Cys isoacceptors among the known Methanosarcina. M. acetivorans, which retains only duplicate copies of a single tRNA Cys gene that is highly similar to tRNA 2 Cys and tRNA 3 Cys of M. mazei, possesses both the bacterial and mammalian enzymes needed for cysteine biosynthesis. However, M. barkeri, which similarly possesses redundant cysteine biosynthesis pathways, contains all three tRNAs. By contrast, M. mazei lacks clear homologs for either the O-acetylserine sulfhydrylase or the cystathionine ␤-synthase enzymes needed for the bacterial and mammalian pathways, respectively (37). Cysteine may be synthesized in M. mazei entirely by means of deacylating tRNA Cys ; alternatively, the organism might use a weakly related O-phosphoserine sulfhydrylase to synthesize cysteine from phosphoserine and sulfide ions (38,39); this enzyme exists in some archaea and provides an alternative to the bacterial pathway from O-acetylserine.
Despite the inconclusive results from these genome comparisons, it is of interest to note that the presence of A33 in the unusual tRNA 1 Cys species may render it a less efficient substrate for the ribosome, because U33 is required for the anticodon loop to adopt the conserved U-turn structure. Several studies have indicated that tRNAs lacking the highly conserved U33 bind the ribosome with lower efficiency (40,41), although it has also been shown that U33 is not essential for tRNAs to perform in eukaryotic in vitro suppression assays (42). The dedication of tRNA 1 Cys to a function other than protein synthesis would suggest a further linkage between Cys-tRNA Cys synthesis and other aspects of metabolism. Although tRNA 1 Cys is clearly an unusual tRNA, its selective aminoacylation by CysRS, capacity to be methylated by Trm5, conservation in M. barkeri, and retention of transcription activation signals all suggest a true functional role.
Sulfur Donor and tRNA Cys Isoacceptor Substrate Preference of SepCysS-Rigorous characterization of the enzymatic activity of SepCysS, which catalyzes the conversion of Sep-tRNA Cys to Cys-tRNA Cys , is challenging because of the need to prepare the misacylated substrate and because the assay requires anaerobic conditions. To first characterize the ligand-bound state of the M. mazei enzyme purified from recombinant E. coli cells, we monitored the UV-visible absorbance of an extensively dialyzed sample. Distinct peaks were observed at 230, 280, and 415 nm; the latter peak corresponds to the lysine-bound ketoenamine form of the cofactor (Fig. 3A) (43). Colorimetric analysis of SepCysS reveals that the molar ratio of PLP to enzyme subunits is 1.2:1, indicating that both active sites bind the cofactor (data not shown) (13). Consistent with this analysis, the crystal structure of A. fulgidus SepCysS showed PLP bound in a covalent aldimine linkage to Lys 209 in both subunits (44).
We next considered how best to quantitate SepCysS activity. The only reported functional assay monitors the PLP-dependent transformation of 14 C-labeled phosphoserine attached to tRNA Cys into 14 C-labeled cysteine (7). Because this assay is insensitive and requires the use of commercially obtained radiolabel at prohibitive cost, we sought to develop an alternative that would be cost-effective, highly sensitive, and hence more suitable for detailed investigation of enzymatic properties. We found that the assay reported here and elsewhere for SepRS function, which generates 32 P-labeled Sep-tRNA Cys with the radiolabel positioned at the 3Ј-internucleotide linkage of the tRNA, can be extended to monitor the conversion of Sep-tRNA Cys to Cys-tRNA Cys (11,13,45). After the SepCysS reaction using 32 P-labeled Sep-tRNA Cys as substrate, the tRNA is hydrolyzed to its constituent nucleotides, and Sep-Ap* (substrate) is separated from Cys-Ap* (product) by thin layer chromatography (Fig. 3B). Separate SepRS and CysRS reactions are used as controls to identify the observed spots. The assay allows for straightforward determination of the fraction of substrate tRNA that is phosphoserylated, to control for the possibility that the input Sep-tRNA Cys may become deacylated in the processing steps between reactions.
We generated the Sep-tRNA Cys substrate in two ways. First, tRNA Cys was phosphoserylated to 30 -85% plateau levels in an aerobic environment at 37°C and then isolated by acidic phenol/chloroform extraction and ethanol precipitation. The Sep-tRNA Cys was then reconstituted in an anaerobic environment, in a buffer consisting of 10 mM sodium cacodylate (pH 6.0) and 2 mM MgCl 2 . Alternatively, tRNA Cys transcripts were lyophilized, transported into the anaerobic environment, and subjected to consecutive SepRS and SepCysS reactions without recovering the Sep-tRNA Cys intermediate. SepCysS converted identical proportions of substrate to product with either method of substrate preparation. Consequently, most experiments were performed using the latter approach, since it obviated the need to isolate Sep-tRNA Cys and provided continuous assay capability requiring only supplementation of the SepRS reaction with PLP-bound enzyme and sulfur donor. We also verified that both the proportion of overall substrate conversion to product by SepCysS and the rates of conversion were independent of the initial plateau level of phosphoserylation, in the range of 30 -85%. Many SepCysS reactions were performed using tRNA Cys at 40 -50% phosphoserylation levels, solely for reasons of economy and efficiency in generating the Sep-tRNA Cys . All SepCysS reactions were performed under strictly anoxic conditions, using reagents made anoxic through cycles of vacuum followed by purging with oxygen-free argon gas.
All reactions also utilized tRNA Cys transcripts modified to carry the m1 G37 modification. Whether the added methyl group influences the activity of SepCysS is not known, because in the absence of methylation, the levels of Sep-tRNA Cys that can be synthesized by SepRS are too low to permit measurement of sulfide transfer.
Initial characterization of SepCysS was performed under conditions of molar enzyme excess over tRNA. Previously, sodium sulfide was used as the sulfur donor in SepCysS reactions (7). To explore the sulfur donor requirement more thoroughly, we carried out reactions using Sep-tRNA 2 Cys as substrate, in the presence of 5 mM sodium sulfide, 5 mM thiophosphate, or 5 mM cysteine. Higher concentrations of any of these reagents did not increase the overall conversion to product. Each compound functioned as a sulfur donor, with sodium sulfide (72% end point conversion of Sep-tRNA Cys to Cys-tRNA Cys ) providing higher activity than either thiophosphate or cysteine (52-55% conversion; Table 2). Only sodium sulfide yields the correct Cys-tRNA Cys product; the capacity of the enzyme to utilize the other donors demonstrates the presence of some structural adaptability in the active site.
To evaluate whether SepCysS exhibits substrate preference among the three M. mazei tRNA Cys isoacceptors, a 5 M concentration of each Sep-tRNA Cys species was incubated with 30 M SepCysS and 5 mM sodium sulfide, and the reactions were allowed to proceed to completion over 1 h. The percentage conversion to Cys-tRNA Cys was reproducibly higher for tRNA 2 Cys and tRNA 3 Cys (72-75%) as compared with tRNA 1 Cys (53%) ( Table 3). By this measure, then, SepCysS exhibits the same tRNA substrate preference observed for SepRS ( Table 1). Measurement of the SepCysS substrate conversion efficiencies for tRNA mutants tRNA Cys ⌬2 and tRNA Cys ⌬4 further confirmed the SepCysS preference. tRNA Cys ⌬2, which is strongly preferred by SepRS, is also a good substrate for SepCysS, whereas tRNA Cys ⌬4, which is shifted substantially toward CysRS preference, is a poorer SepCysS substrate (Tables 1 and  3). Part of the SepCysS preference appears to arise from differences in binding affinity to the tRNA body; measurements of equilibrium dissociation constants for SepCysS interaction with each native unacylated isoacceptor, by fluorescence quenching (20), showed that K d for tRNA 2 Cys and tRNA 3 Cys are 2-3-fold lower than K d for tRNA 1 Cys (Table 3).
To further characterize the properties of M. mazei SepCysS, we investigated the kinetics of converting Sep-tRNA Cys to Cys-tRNA Cys . Under steady-state conditions at 1-50 nM Sep-CysS, we observed no reaction when Sep-tRNA Cys substrate levels were maintained below 1 M. However, product formation was detected at 50 nM SepCysS and 5 M Sep-tRNA Cys . At this substrate concentration, reaction velocities then were found to increase with enzyme concentration (Fig. 3C). It is likely that this behavior arises from dissociation of the SepCysS dimer into monomers, with K d for subunit association in the range of 10 Ϫ7 M. This interpretation is consistent with the crystal structure of A. fulgidus SepCysS, which shows that each active site is composed of amino acids from both subunits, suggesting that monomers do not retain activity (44). No Sep-tRNA Cys saturation was observed under steady-state conditions of substrate excess ([S]/[E] Ͼ 20) and with Sep-tRNA Cys concentrations up to 15 M. However, the initial velocities allow estimation of k cat /K m at 0.009 -0.014 min Ϫ1 M Ϫ1 , with no significant differences measured among the three isoacceptors ( Table 3). Levels of product formation substantially exceed the amount of enzyme present, demonstrating unambiguously that SepCysS is able to function as a multiple-turnover catalyst.
The higher substrate conversion efficiency and binding affinities for tRNA 2 Cys and tRNA 3 Cys by SepCysS matches the preference of SepRS and suggests that these two species may be ded-   icated to the indirect pathway. This may represent an adaptation that serves to maximize the efficiency of Cys-tRNA Cys synthesis in vivo. However, the SepCysS selectivity is not large and is also not detected at the level of k cat /K m . Further, our work in assay development showed that the presence of SepRS in the SepCysS reactions does not affect either the rate or plateau level of Sep-tRNA Cys conversion under these experimental conditions. Nevertheless, work on the homologous M. jannaschii pathway showed that SepRS and SepCysS from that organism do form a binary complex that binds tRNA Cys and that catalyzes the SepRS reaction with 2-fold higher k cat than SepRS alone (11). Thus, a modest enhancement in substrate throughput may be a common characteristic of the two-step indirect pathway in methanogens. In addition, a further important role for the formation of the SepRS-SepCysS-tRNA Cys ternary complex is to sequester the misacylated Sep-tRNA Cys from EF-1␣, thus preventing misincorporation of phosphoserine into cellular proteins (11).
Mechanism of SepCysS-The kinetic experiments performed in the presence of a molar excess of enzyme showed that the required period of incubation to achieve maximal conversion of Sep-tRNA Cys to Cys-tRNA Cys by SepCysS is quite long (Fig.  3B). However, 10 -40-min time courses were required for maximal substrate conversion by eukarytic selenocysteine synthase, another PLP-dependent enzyme that acts on aminoacylated tRNA (46). The k cat /K m for the M. mazei SepCysS reaction with bisulfide as sulfur donor is also very similar to that reported for the analogous conversion of Ser-tRNA Sec to Sec-tRNA Sec by bacterial selenocysteine synthase when thiophosphate is employed as sulfur donor in that enzyme (47).
To further examine the kinetics of SepCysS, we measured substrate conversion while systematically varying the concentration of enzyme. At 5 M Sep-tRNA Cys and at SepCysS concentrations above 200 nM, product formation exhibits an upward slope after reactions proceed for ϳ30 min (Fig. 4A). Further, at equimolar E/S ratios and under conditions of enzyme molar excess, the reaction progress curve is sigmoidal. For example, at 10 M Sep-tRNA Cys and 15-30 M SepCysS, a distinct lag phase is followed by a rapid increase in product formation, with maximal levels reached at 50 -60 min. During the slow lag phase, an intermediate that migrates farthest on the TLC plate is observed to first accumulate and then to disappear as Cys-tRNA Cys formation begins to rapidly increase (Figs. 3B and 4B). It appears that formation of this intermediate is rate-limiting under single-turnover conditions. The distinct lag phase suggests a rate-limiting step associated with the mechanistic step in which the exogenous sulfur reacts (Fig. 5). Further, the measured k cat /K m for conversion of Sep-tRNA Cys to Cys-tRNA Cys is ϳ500-fold lower than k cat /K m for phosphoserylation of tRNA 2 Cys and tRNA 3 Cys by SepRS (Tables 1  and 3). Thus, the rates by which the substrate is processed by each enzyme appear to be greatly mismatched; this would produce a highly inefficient pathway if present in vivo. Both of these observations suggest that the sulfur donors employed in these in vitro experiments are not likely to be used in vivo, so that a major experimental challenge associated with elucidating the true pathway of sulfur delivery to the active site remains.
By analogy with the mechanism proposed for the PLP-dependent conversion of seryl-tRNA Sec to selenocysteyl-tRNA Sec by bacterial selenocysteine synthase (48), we suggest that binding of Sep-tRNA Cys to SepCysS is followed by formation of a Schiff base between the ␣-amino group of the phosphoserine residue and the formyl group of pyridoxal 5-phosphate, resulting in elimination of the phosphate group and formation of an aminoacrylyl-tRNA Cys , which would correspond to the inter- mediate observed on TLC as aminoacrylyl-A76 (Figs. 3B, 4B, and 5). In our in vitro reactions, HS Ϫ or another donor then adds slowly to the double bond, in the rate-limiting step, to generate product (Fig. 5).
Use of the natural donor selenophosphate provides a 330fold rate enhancement to selenocysteine synthase as compared with thiophosphate (47); such an improvement, if conferred to SepCysS via use of the proper sulfur donor, would suffice to nearly match the catalytic efficiency of SepRS (Tables 1 and 3). It is most likely that the true sulfur donor to tRNA is a persulfide group formed from one or more of the three conserved cysteines in the active site (Cys 51 , Cys 54 , and Cys 260 in the M. mazei enzyme) (44). The persulfide group provides sulfur for both cofactor and nucleotide biosynthesis in a wide variety of biological systems (49) and could presumably function to increase the reaction rate here by properly positioning the sulfur for transfer in relation to the bound Sep-tRNA Cys . Interestingly, SepCysS itself is strongly homologous to the cysteine desulfurases CsdB, NifS, and IscS, suggesting that it may function as both a sulfur recipient to form Cys-tRNA Cys and a sulfur donor in other, as yet unknown, reactions (8,49). In this regard, the essentiality of the enzyme in the context of a cell deleted for SepRS may be relevant (36). The identity of the enzyme that may convert one of the SepCysS active site cysteines to a persulfide is also unknown at this time; by analogy to pathways such as the sulfuration of tRNA, it is possible that the route of sulfur delivery to SepCysS involves transfer through multiple proteins (50). After the sulfur transfer, the resulting enzyme-bound disulfide would then be reduced by reaction with another cysteine residue, releasing PLP and Cys-tRNA Cys (Fig. 5). This mechanism predicts essentiality for at least two of the three conserved active site cysteines. The sulfur relay system would then regenerate persulfided SepCysS for the next catalytic cycle.
Formation of Cys-tRNA Cys by SepCysS probably also has many elements in common with the formation of selenocysteyl-tRNA Sec in archaea and eukaryotes. In these domains, serylation of tRNA [Ser]Sec by SerRS is followed by a phosphorylation reaction leading to formation of a phosphoserylated tRNA: Sep-tRNA [Ser]Sec . This intermediate is then converted to Sec-tRNA Sec by another PLP-dependent enzyme designated alternatively as selenocysteine synthase (46,51) or SepSecS (52). This enzyme uses selenophosphate as a selenide donor and is weakly homologous to SepCysS (46). Kinetic analysis of the selenocysteine synthase reaction has shown that, as proposed for SepCysS and demonstrated for bacterial selenocysteine synthase (48), generation of Sec-tRNA [ (46). Interestingly, selenocysteine synthase can also utilize phosphoseryl-tRNA [Ser]Sec as a substrate (46), suggesting that comparative studies among all three enzymes may be a productive approach to acquiring detailed insight into the mechanisms of these remarkable RNA-dependent amino acid modification reactions.
Perspectives-The selectivities of CysRS and SepRS for distinct tRNA Cys isoacceptors clearly suggest that both synthetic routes and all three tRNAs have physiological significance. The detailed nature of how the tRNAs are partitioned between path-ways in vivo, however, and the extent to which one or more of them may play other metabolic roles remain as open questions for future research. In many simpler tRNA-synthetase systems, in vitro reconstitution by expression of a single enzyme, together with use of unmodified tRNA transcripts, suffices to recapitulate most biochemical properties of interest. By contrast, the Cys-tRNA Cys biosynthetic system in M. mazei illustrates how the full reconstitution of activities is likely to require multiple tRNA modifications as well as essential cofactor proteins. Identification of additional tRNA Cys -modifying enzymes and of the enzyme that delivers sulfur to SepCysS may suffice for a definitive understanding of tRNA selectivities in CysRS, SepRS, and SepCysS. Such identifications may provide clues that, together with genetic, biochemical, and bioinformatic studies, could elucidate novel metabolic relationships connecting the protein synthesis machinery with sulfur and energy metabolism in methanogens. The crucial role of these archaebacteria in the biogeochemical processes determining global climate change offers strong impetus for these experiments.