Differential modes of transfer RNASer recognition in Methanosarcina barkeri.

Two dissimilar seryl-transfer RNA (tRNA) synthetases (SerRSs) exist in Methanosarcina barkeri, one of bacterial type and the other resembling SerRSs present only in some methanogenic archaea. To investigate the requirements of these enzymes for tRNASer recognition, serylation of variant transcripts of M. barkeri tRNASer was kinetically analyzed in vitro with pure enzyme preparations. Characteristically for the serine system, the length of the variable arm was shown to be crucial for both enzymes, as was the identity of the discriminator base (G73). Moreover, a novel determinant for the specific tRNASer recognition was identified as the anticodon stem base pair G30:C40; its contribution to the efficiency of serylation was remarkable for both SerRSs. However, despite these similarities, the two SerRSs do not possess a uniform mode of tRNASer recognition, and additional determinants are necessary for serylation specificity by the methanogenic enzyme. In particular, the methanogenic SerRS relies on G1:C72 identity and on the number of unpaired nucleotides at the base of the variable stem for tRNASer recognition, unlike its bacterial type counterpart. We propose that such a distinction between the two enzymes in tRNASer identity determinants reflects their evolutionary pathways, hence attesting to their diversity.

To maintain translational accuracy, aminoacyl-transfer RNA (tRNA) 1 synthetases are highly selective toward their amino acid and tRNA substrates. In the process of tRNA recognition, the cognate and non-cognate substrates are discriminated according to characteristic nucleotides in certain positions of the tRNA, specific for the tRNA/synthetase system. 2 These recognition (identity) elements are commonly located in the tRNA anticodon, the acceptor stem, and the discriminator base at position 73 (1)(2)(3)(4)(5). However, as some systems show little or no conservation among anticodons of their tRNA isoaccep-tors (the six-codon families Arg, Leu, and Ser), additional parts of the tRNA are necessary for specific recognition (6 -13).
Identity elements required for serylation have been studied in a number of organisms, providing insights into tRNA Ser recognition in the different domains of life (6, 14 -26). Contrary to the identity requirements of the majority of aminoacyl-tRNA synthetases, Escherichia coli seryl-tRNA synthetase (SerRS) was found to recognize neither the anticodon nor the discriminator base of tRNA Ser (27). Instead, the length of the variable arm and the characteristic tRNA Ser tertiary structure were shown to be crucial for serylation. In vitro studies and footprinting experiments using yeast SerRS and tRNA Ser (19,22,23) revealed the discriminator base to be unimportant and, while the variable arm functions as the major identity element, discrimination in yeast, in contrast to E. coli, is believed to be more sequence-and less structure-dependent (14,22). Although G73 serves only as an antideterminant in bacteria (9) and lower eukaryotes (14,22), it is an essential identity requirement for human tRNA Ser (13,25). Furthermore, recent results show a dual mode of recognition of two unusual tRNA Ser isoacceptors in mammalian mitochondria, with the T-loop as the main identity target (29 -32). Taken together, the variety of tRNA Ser identity determinants suggests that mechanisms of SerRS recognition may have diverged during evolution.
Inspection of the available protein sequences reveals two major types of SerRSs (33): a "standard" enzyme found in most organisms and a highly diverged SerRS confined to some members of the methanogenic archaea (34). Compared with the standard bacterial type SerRSs, the methanogenic enzymes are characterized by insertions in their N-terminal part and by a notable gap in the motif 2 loop (34). Because crystallographic studies on E. coli and Thermus thermophilus tRNA⅐SerRS complexes show both direct involvement of the N-terminal coiled coil in the variable arm recognition and interaction of the motif 2 loop with the major groove of tRNA Ser (7,35), such distinction in the sequences raises questions as to how methanogenic SerRSs recognize their tRNA substrates.
Our interest in tRNA Ser identity focused on Methanosarcina barkeri, whose genome contains two SerRS genes of different origin: its bacterial type SerRS is related to the SerRSs found in some Gram-positive bacteria (36), whereas its methanogenic homolog clusters with SerRSs from Methanothermobacter thermautotrophicus, Methanopyrus kandleri, Methanocaldococcus jannaschii, and Methanococcus maripaludis. As tRNA Ser determinants have not been established for a methanogenic or any archaeal SerRS, it was of interest to define the nature of tRNA Ser recognition in a representative of this group of organisms. Moreover, the coexistence of two distantly related SerRS enzymes in M. barkeri provides a favorable system for evaluation of the evolutionary aspects of tRNA discrimination. tRNA Cloning and Preparation-tRNA Ser genes were identified in the M. barkeri genomic DNA sequence at the U.S. Department of Energy Joint Genome Institute using tRNAScan-S.E. (www.geneticwustl.edu/ eddy). Wild-type and mutant tRNA genes were constructed from the synthetic oligomers carrying the tRNA gene under the T7 promoter sequence. The genes were ligated into the BamHI/HindIII restriction sites of pUC18. Following plasmid maxi preparation, the DNA template was prepared by digestion with NsiI or BstNI, phenol/chloroform extraction, and ethanol precipitation. 50 g of template was incubated for 1 h at 37°C in a 0.5-ml reaction mixture that included 40 mM Tris-HCl, pH 8.1, 44 mM MgCl 2, 0.1% Triton, 2 mM spermidine, 10 mM dithiothreitol, 4 mM each of the nucleoside triphosphates, 10 mM GMP or AMP, and 10 g of pure T7 RNA polymerase. (Because of the low yield, some of the transcription reactions were proportionally increased 10-fold.) Reactions were stopped by phenol/chloroform extraction, precipitated with ethanol, and resuspended in gel loading buffer (8 M urea, 20% sucrose, 0.1% bromphenol blue, 0.1% xylene cyanol). Transcripts were purified on a denaturing polyacrylamide gel (12% acrylamide:bisacrylamide (19:1), 8 M urea, 89 mM Tris borate, pH 8.3, 2 mM EDTA) and extracted in buffer containing 1 M sodium acetate, pH 6, 1 mM MgCl 2 , and 0.1% SDS. Finally, the tRNA samples were desalted on NAP-5 columns and refolded (5 min at 95°C followed by a gradual reduction of temperature; 5 mM MgCl 2 was added to the refolding mixture at 55°C). tRNA concentration was determined spectrophotometrically at 260 nm.

Materials-Oligonucleotides
Enzyme Cloning and Preparation-Methanogenic and bacterial type SerRS genes were identified in the preliminary M. barkeri Fusaro genomic DNA sequence using JGI BLAST analysis (www.jgi.doe.gov). The DNA sequences were amplified by PCR using Expand High Fidelity polymerase and cloned into the pET15b vector for expression of Nterminal His 6 -tagged proteins. Purification on nickel affinity columns was performed as published (37).
Aminoacylation Assay-To determine the amount of the active enzyme in the preparation, active site titration was performed. The assay is based on quantifying the amount of the complex between the enzyme and radioactively labeled amino acid adenylate retained on the nitrocellulose filters (38). The reaction was performed in 0.1 ml of 0.5ϫ EAP buffer (50 mM Tris-HCl, pH 7. The enzyme concentrations were experimentally determined for each tRNA transcript in order to obtain linear velocities (4 -17 nM for the wild-type transcripts and 4 -170 nM for the mutant transcripts). Radioactive aminoacyl-tRNA synthesized after 2-8 min was quantified as described previously (37). Experimental kinetic data are based on three measurements. The kinetic constants were derived from Hanes-Woolf plots using five different tRNA concentrations. In exceptional cases where K m values were too high to be determined, k cat /K m values were obtained from the slope of the plot of initial rates against substrate concentrations (39). For ATP kinetics, 30 or 50 M of tRNA SerGCU were used for the bacterial type SerRS or the methanogenic enzyme, respectively. Both enzymes were used at a final concentration of 150 nM as determined by linear velocity.

M. barkeri Possesses Two Functional SerRS Enzymes-Both
the bacterial type and the methanogenic type M. barkeri SerRS showed efficient serylation activity with similar affinities for serine (or ATP) (K m values of 25 Ϯ 4 (13.6 Ϯ 1.6) and 34 Ϯ 4 (13.8 Ϯ 1.5) M for the methanogenic and bacterial type SerRS, respectively) determined in the aminoacylation reaction. Furthermore, both enzymes successfully aminoacylated transcripts of all M. barkeri tRNA Ser isoacceptors (tRNA SerGCU , tRNA SerCGA , tRNA SerGGA ; Fig. 1), although with different efficiencies (Table I). tRNA SerCGA and tRNA SerGGA isoacceptors were similar in their kinetic properties toward the methanogenic SerRS. The same is true for the bacterial type enzyme; however, whereas the methanogenic enzyme comparatively showed a notable decrease in the efficiency of tRNA SerGCU aminoacylation, this isoacceptor seemed to be a preferred substrate for the bacterial type SerRS.
Identity Determinants of tRNA Ser -To elucidate tRNA Ser identity determinants for the two M. barkeri SerRS enzymes, mutant tRNA SerCGA species were produced (Fig. 2) based on the conservation of the nucleotides among tRNA Ser isoacceptors and on the known identity requirements of bacterial and eukaryotic serine systems. Kinetic analysis of their serylation efficiency revealed that a number of mutations remarkably affected the relative k cat /K m values for both enzymes (Table II). Contributions of the specific positions and structural elements in different tRNA domains are discussed below.
The importance of the discriminator base and of nucleotides from the first three base pairs in the acceptor stem has been shown in E. coli by in vivo identity conversion of a leucine to a serine suppressor tRNA (6) and emphasized by work on tRNA Ser minihelices (18). The nucleotides in this part of the M. barkeri tRNA Ser species are most conserved, which suggests their involvement in specific tRNA Ser recognition. Accordingly, tRNA Ser Recognition in M. barkeri mutational analysis shows a dramatic drop of the aminoacylation efficiencies for variants of G73 for both enzymes, as well as for the first base pair position in the case of the methanogenic SerRS (mutants 1-4). Inability of these mutants to be efficiently serylated results primarily from the decrease in their k cat values. However, the contribution of the acceptor stem is limited to its upper part, as mutations of the conserved base pairs C2:G71, C3:G70 and G6:C67 (mutants 5-9) either retained the wild-type activity or resulted in a slight stimulation of serylation.
The D-arm and the D-loop have been identified as elements contributing to tRNA Ser identity. In addition to the changes in the acceptor stem and the discriminator base, the in vivo identity conversion from bacterial tRNA Leu to tRNA Ser required the change from G11:C24 to C:G (6). This base pair, however, is not conserved among tRNA Ser isoacceptors in M. barkeri. Furthermore, bacterial tRNAs Ser share a specific D-loop feature, marked by a deletion at position 17 and a double insertion of the 20a and 20b nucleotides. The significance of this structural element is manifested through its involvement in the tertiary interactions with the variable arm, whose orientation is thus defined (7). On the other hand, deletion of position 17 is rather infrequent in archaeal serine tRNAs, as are the insertions at positions 20a and 20b in the D-loop. tRNA Ser isoacceptors from M. barkeri do not share a uniform structure: although tRNA SerCGA and tRNA SerGGA possess deletion 17, tRNA SerGCU contains a double insertion at the same position. When a CC insertion was introduced into tRNA SerCGA , the resulting variant showed a somewhat reduced serylation efficiency (mutant 12) as did the mutation of A9 (mutant 10). Possibly, these alterations affected the tertiary structure of tRNA Ser , because the D-loop interacts with the T-loop and because position 9 partic-ipates in formation of base triplets in yeast tRNA Phe , tRNA Asp , and E. coli tRNA Ser (7). A similar effect was observed for mutant 11, although the relative contributions of its kinetic parameters to serylation efficiency appear to differ between the two enzymes. These results imply that conservation of the G10:C25 base pair among M. barkeri and a significant number of other archaeal tRNAs Ser is not coincidental.
The anticodon was expectedly shown not to contribute to serine identity (mutant 15), in accordance with the published results of other serine systems (27). However, mutation of G30:C40 in the anticodon stem resulted in a significant loss of serylation activity (mutants 13 and 14) for both enzymes with notable K m effect. Although the upper part of the anticodon stem has been suggested to possess discriminatory function by footprinting experiments in the yeast system (base pair A27: U43) (19), none of the anticodon stem nucleotides have directly been identified as recognition determinants in a mutational study. Support of such a notion comes from the fact that high conservation of nucleotides in this region of bacterial tRNAs Ser cannot be detected; contrary to that, the G30:C40 base pair is absolutely conserved among archaeal tRNA Ser species. This position has been identified as an identity determinant for human phenylalanyl-tRNA synthetase (40). Our preliminary results have indicated that the anticodon arm of tRNA Ser also participates in specific recognition by M. maripaludis SerRS. 3 tRNAs with a variable arm of at least 11 nucleotides (type 2 tRNAs) are restricted to three families: tRNA Tyr , tRNA Leu , and tRNA Ser . Regardless of the fact that the variable arm of tRNA Ser differs in both length and sequence within isoaccep-3 I. Gruic-Sovulj, personal communication. tRNA Ser Recognition in M. barkeri tors, it makes the largest contribution to serine identity, as confirmed by different experimental approaches (9,17,41). Despite direct contacts that have been observed between T. thermophilus SerRS and the variable arm stem (7), sequence alterations of the variable arm only insignificantly affected the serylation efficiency (27). It was therefore concluded that the variable arm of tRNA Ser is not recognized sequence specifically, but its length was shown to be most critical for efficient serylation. Likewise, our results show that recognition of tRNA Ser in M. barkeri essentially depends on the length of its variable arm. Deletions of both single base pair (Ce12:Ge22, mutant 20) and double base pair (Ue11:Ae21 and Ce12:Ge22, mutant 21) largely reduced the ability of the tRNA Ser transcript to be serylated. In contrast, enlargement of the variable arm stem was permitted (mutant 22), and although it moderately reduced the aminoacylation efficiency by the methanogenic SerRS, it somewhat stimulated serylation by the bacterial type enzyme. As many SerRSs serylate tRNA Sec species, whose variable arm lengths often exceed those of tRNAs Ser (42), enlarged variable arms of tRNAs Ser are not unexpected as reasonably good substrates. Similarly, changes of length and identity of the nucleotides in the variable arm loop had negligible effect on serylation (mutants 16 -18), in accordance with the published results (27) and absence of the sequence conservation. In contrast, the only base pair in the variable arm shared between the isoacceptors (Ce17:Ge27) seems to be important either structurally or specifically, as its mutation shows a moderate but notable effect on the serylation efficiency of both enzymes (mutant 19).
In addition to its length, the orientation of the variable arm was determined to be essential for the specific recognition. In bacteria, each of the type 2 tRNAs has a unique orientation of the variable arm that is defined by the number of unpaired nucleotides between the 3Ј-base of the variable stem and the nucleotide at position 48 (none in tRNA Ser , one in tRNA Leu , and two in tRNA Tyr ). Whereas deletion of one base pair in the leucine system caused only a small decrease in leucylation (9), elimination of the two unpaired U residues at the 3Ј-base of the variable arm in E. coli tRNA Tyr led to a 150-fold reduction in the aminoacylation efficiency (14). Unlike E. coli, M. barkeri tRNA Ser species possess one unpaired nucleotide at the 3Ј-base of the variable arm, tRNAs Leu possess two, while a long variable arm is not present in archaeal tRNAs Tyr (43). To assess whether a tRNA Leu -like structure of the variable stem base would affect tRNA Ser recognition, G46 was replaced by a double insertion of U (mutant 23); indeed, decrease in serylation efficiency was observed, although more conspicuously for the methanogenic SerRS. Moreover, mutation of C48 reduced relative affinity of the tRNA Ser for the methanogenic enzyme without decrease of its ability to be serylated by the bacterial type SerRS (mutant 24). Finally, mutation of the G49:C65 pair moderately reduced serylation efficiency for both enzymes by increasing their K m values (mutant 25).
Transplantation of Serine Recognition Elements to tRNA Leu -As a means of affirming the entirety of the identity set determined for a given system, transplantation into a non-cognate tRNA is commonly performed. In studies dealing with serine identity, tRNA Leu is frequently chosen as a host tRNA (23,27). Conversion of leucine to serine identity in E. coli requires alterations of C2:G71 to G:C and G59 to A, changes in the D-loop and in the pairing pattern in the variable arm, and introduction of the tertiary base pair G15:C48 (27); the same type of conversion in yeast is achieved by a single nucleotide insertion to the long variable arm (23,27). Our results show that adaptation of M. barkeri tRNA Leu to a serine-accepting variant requires several changes: enlargement of the variable arm, reduction in the number of unpaired nucleotides at the 3Ј-base of the variable stem, mutation of A49:U65 to G:C, and change of the discriminator base from A to G (Fig. 3). Serylation of the tRNA Leu variant containing these modifications was  (Table III, variant ML3), but its efficiency was further improved in the case of the bacterial type SerRS by an additional mutation (change of C22 to U, variant ML1), which allowed formation of a D-stem consisting of four base pairs, as in tRNA Ser . Evidently, the serylation potential of variant ML1 is comparable with that of the wild-type tRNA SerCGA transcript, which substantiates the completeness of the determined tRNA Ser identity set. Our results also show that omission of some of the transplantation elements from the serine-chargeable tRNA Leu variants results in either complete loss of serylation or in much less efficient serylation (variants ML2, ML4, and ML5).

DISCUSSION
Two SerRS Enzymes in M. barkeri-Among the Methanosarcinaceae only M. barkeri contains two SerRS enzymes, whereas the other members of this genus possess only the bacterial type SerRS. Because M. barkeri has a canonical cysteinyl-tRNA synthetase, neither SerRS is needed to participate in Cys-tRNA formation as was suggested for some methanogens (34,44).
tRNA Ser Identity in M. barkeri-Specificity of tRNA aminoacylation is maintained both by structural and kinetic discrimination of tRNA molecules in the network of aminoacyl-tRNA synthetases. Evidently, tRNA recognition is contextually dependent and identity requirements for a given system are influenced by characteristics of other isoaccepting systems with substantially similar tRNA substrates. It is widely assumed that tRNA Leu , because of its long variable arm, presents a possible candidate for misacylation by serine, and discrimination between tRNA Ser and tRNA Leu has received great attention (13,25,45). It has been suggested that the discriminator base G73 in E. coli tRNA Ser serves as an anti-determinant for leucyl-tRNA synthetase and tyrosyl-tRNA synthetase, where it is most disfavored (27). Furthermore, other parts of the acceptor stem are believed to possess similar functions, because G2:C71 in tRNA Ser was proposed to contribute to the rejection of the C2:G71-bearing tRNA Leu (27). The M. barkeri SerRS enzymes, however, work in a different milieu, as both tRNA Ser and tRNA Leu possess the C2:G71 pair and their possible discrimination at this position is thus thwarted. However, the base pair C3:G70 is conserved among M. barkeri tRNA Ser isoacceptors, and, as all M. barkeri tRNA Leu isoacceptors also have the G3:C70 base pair, the discriminatory role of this position was tested. However, none of the changes reduced serylation efficiency (Table II, mutants 6 -8). Considering these observations, it may have been advantageous for the M. barkeri SerRSs to recognize the discriminator base, as our mutational study has confirmed.
Furthermore, discrimination against tRNA Leu and tRNA Tyr in bacteria is achieved by recognition of the variable arm orientation, specific for each of the type 2 tRNAs. The recent crystal structure of T. thermophilus TyrRS complexed to tRNA Tyr shows that orientations of the variable arms in tRNA Ser and tRNA Tyr differ by ϳ50°because of the number of unpaired bases preceding nucleotide 48 (46). The specific direction of the variable arm in tRNA Ser is defined by stacking of the base G20b with the first base pair in the variable stem. This insertion, although present in bacterial tRNAs Ser , is absent in most archaeal and cytoplasmic eukaryotic serine isoacceptors, which, together with the disparity between bacterial and archaeal tRNAs Ser in the number of unpaired bases at the 3Ј-end of the variable arm, suggests that details of the core packing of tRNA Ser may differ between bacteria and archaea. This assumption is additionally supported by the fact that bacterial SerRSs have to discriminate between three tRNAs containing a long variable arm, whereas lack of this structural element in archaeal tRNA Tyr obviates the need for its rejection. However, our results imply that some elements of tertiary structure are pertinent to the specific tRNA Ser recognition by both methanogenic and bacterial type SerRS.
In general, the two M. barkeri SerRS enzymes recognize their tRNAs in a similar fashion: both strongly depend on G73 identity, on the length of the variable arm, and on the G30:C40 pair in the anticodon stem for efficient serylation. However, some aspects of their identity requirements are remarkably different; whereas the methanogenic SerRS relies on G1:C72 identity, the bacterial type enzyme does not (mutant 4). Likewise, change in the number of unpaired nucleotides at the base of the variable stem dramatically affects serylation of the methanogenic, but not the bacterial, type SerRS (mutant 23). Such notable distinction could possibly be correlated to the evolutionary divergence of the two enzymes (discussed below).
Considering the structural aspects of tRNA Ser recognition, we may infer that the N-terminal domain in the methanogenic SerRS participates in tRNA Ser variable arm recognition, analogous to the helical arm of the bacterial SerRS counterparts. Conservation of the long variable arm in the methanogenic tRNA Ser species and the effects of variable stem shortening on the serylation activity of the methanogenic enzyme (Table III,  tRNA Ser Recognition in M. barkeri of the bacterial and methanogenic sequences within the motif 2 loop and the lack of structural data. However, sequence alignments may suggest participation of methanogenic M. barkeri SerRS Glu-338 in discriminator base recognition. As to the bacterial enzyme, its motif 2 loop shows a great degree of similarity with the loop of T. thermophilus SerRS; therefore, it may be proposed that Glu-230 of the M. barkeri SerRS is involved in G73 recognition analogous to Glu-260 in the T. thermophilus structure (47). Unfortunately, the available data are insufficient to comment on anticodon stem recognition.
Evolutionary Aspects of tRNA Ser Recognition-As patterns of tRNA recognition reflect variations in both tRNA and their corresponding aminoacyl-tRNA synthetases, their inspection is informative with respect to evolutionary strategies for preserving tRNA identity. Given the similarity of the tRNA Ser identity requirements for the two dissimilar SerRS enzymes, it can be inferred that a specific mode of tRNA Ser recognition may have been defined before methanogenic and bacterial type SerRSs diverged from the common ancestor. A previously published phylogenetic analysis on SerRSs implies that these two forms may be unrelated and is suggestive of horizontal gene transfer events (28,33). If so, it is conceivable that Methanosarcinaceae received a bacterial type of SerRS from a Gram-positive bacterium and subsequently lost the original, presumably methanogenic, SerRS enzyme. In this respect, M. barkeri might represent a transient state with two functional SerRS enzymes. Moreover, M. barkeri is also remarkable in that it possesses two unrelated LysRSs of different classes (36). Our data do not suggest functional complementarity of the two SerRSs, and it is plausible that one of the enzymes is dispensable. On the other hand, one of the SerRS enzymes may possibly perform a cellular function unrelated to direct aminoacylation of tRNA Ser .
Regarding tRNA recognition, we can hypothesize adaptation of the bacterial SerRS to the methanogenic environment: differences in recognition of the pairing patterns of the variable arm and absence of recognition of positions 2:71 and 3:70 by the two SerRSs (mutants 22 and 5-8) may be indicative of tRNAdriven alteration of the identity requirements. Supposing that M. barkeri acquired the bacterial type of SerRS from a bacterium with identity requirements similar to the ones defined for the E. coli system, it would have been challenged to recognize the archaeal tRNA substrate in a highly specific manner. The most conspicuous differences between archaeal and bacterial tRNAs Ser are located in the acceptor stem, where the G2:C71 base pair is absolutely conserved in bacteria, whereas the majority of archaeal tRNAs Ser contain a C:G pair at the same position. Furthermore, A:U is the most common base pair at position 3:70 in bacterial tRNAs Ser , unlike C3:G70 in archaea. In that respect, and given the way the acceptor stem of tRNA Ser is identified in E. coli (27), the bacterial type SerRS may have modified its identity requirements and adopted G73, pertinent to both archaeal and bacterial tRNAs Ser , as one of the major elements of serine identity.