Trypanosoma Seryl-tRNA Synthetase Is a Metazoan-like Enzyme with High Affinity for tRNASec*

Trypanosomatids are important human pathogens that form a basal branch of eukaryotes. Their evolutionary history is still unclear as are many aspects of their molecular biology. Here we characterize essential components required for the incorporation of serine and selenocysteine into the proteome of Trypanosoma. First, the biological function of a putative Trypanosoma seryl-tRNA synthetase was characterized in vivo. Secondly, the molecular recognition by Trypanosoma seryl-tRNA synthetase of its cognate tRNAs was dissected in vitro. The cellular distribution of tRNASec was studied, and the catalytic constants of its aminoacylation were determined. These were found to be markedly different from those reported in other organisms, indicating that this reaction is particularly efficient in trypanosomatids. Our functional data were analyzed in the context of a new phylogenetic analysis of eukaryotic seryl-tRNA synthetases that includes Trypanosoma and Leishmania sequences. Our results show that trypanosomatid seryl-tRNA synthetases are functionally and evolutionarily more closely related to their metazoan homologous enzymes than to other eukaryotic enzymes. This conclusion is supported by sequence synapomorphies that clearly connect metazoan and trypanosomatid seryl-tRNA synthetases.

Trypanosomatids are important human pathogens that form a basal branch of eukaryotes. Their evolutionary history is still unclear as are many aspects of their molecular biology. Here we characterize essential components required for the incorporation of serine and selenocysteine into the proteome of Trypanosoma. First, the biological function of a putative Trypanosoma seryl-tRNA synthetase was characterized in vivo. Secondly, the molecular recognition by Trypanosoma seryl-tRNA synthetase of its cognate tRNAs was dissected in vitro. The cellular distribution of tRNA Sec was studied, and the catalytic constants of its aminoacylation were determined. These were found to be markedly different from those reported in other organisms, indicating that this reaction is particularly efficient in trypanosomatids. Our functional data were analyzed in the context of a new phylogenetic analysis of eukaryotic seryl-tRNA synthetases that includes Trypanosoma and Leishmania sequences. Our results show that trypanosomatid seryl-tRNA synthetases are functionally and evolutionarily more closely related to their metazoan homologous enzymes than to other eukaryotic enzymes. This conclusion is supported by sequence synapomorphies that clearly connect metazoan and trypanosomatid seryl-tRNA synthetases.
The formation of aminoacyl-tRNA, catalyzed by aminoacyl-tRNA synthetases, is a crucial step in maintaining the fidelity of protein biosynthesis. To avoid misacylation of non-cognate tRNAs, each synthetase recognizes identity elements idiosyncratic to their cognate substrates. tRNA identity elements can be unique and universally distributed, as in the case of the G3:U70 base pair presented by all tRNA Ala (1,2). On the other hand, the evolutionary constraints imposed by the necessity of translation fidelity are not rigid, and the set of recognition elements for many tRNAs has changed during evolution (1,3). Thus, the comparison between sets of tRNA recognition ele-ments in extant species could be used to estimate their evolutionary or biological relatedness.
Here we study the evolution of the serylation reaction of tRNA Ser and tRNA Sec by characterizing this activity in trypanosomatids, a group of protozoa normally considered to form a basal group within eukaryotes (4). Intriguingly, the analysis of the evolutionary history of Trypanosoma seryl-tRNA synthetases (SerRSs) 3 and the determination of the identity elements used by this enzyme to recognize tRNA Ser show that Trypanosoma SerRSs are closer relatives of metazoan SerRSs than plant, fungi, or other protozoan enzymes.
SerRSs are class II aminoacyl-tRNA synthetases that contain the characteristic active site domain of this family of enzymes (5,6). In addition to this domain, the structures of most SerRSs include a coiled-coil domain at the amino end of their sequences and, in the case of eukaryotic enzymes, a C-terminal extension that plays a modest role on protein stability and amino acid recognition (7,8). The N-terminal coiled-coil domain is essential for the recognition of the elbow and, especially, the long variable loop of tRNA Ser (9,10).
Studies of the recognition modes between SerRS and tRNA Ser from different kingdoms of life have shown that some, but not all, tRNA Ser identity determinants have been conserved during evolution (for a review, see Ref. 11) and that different recognition modes exist. Thus, in bacteria and Saccharomyces cerevisiae, the specific charging of tRNA Ser depends on the recognition of some base pairs in the acceptor stem but not of the discriminator base at position 73 (12)(13)(14). In these cases, the sequence of the variable loop is also crucial. The crystallographic structures of Thermus thermophilus SerRS complexed with tRNA Ser , confirm these observations (15,16).
A second type of recognition mode, seen in archaeal and human SerRS, is characterized by the crucial role of the discriminator base (17)(18)(19). In these cases, the sequence of the variable loop is not important, but its size and orientation are fundamental for the interaction with the enzyme (20,21). The main common features between the bacterial and eukaryotic recognition strategies are the importance of the variable loop and the dispensable character of the anticodon stem-loop.
Finally, a third type of tRNA Ser recognition is seen in the SerRSs of methanogenic archaea, which use an idiosyncratic N-terminal domain to recognize the acceptor stem and the variable loop of their tRNA substrates (22). These unusual enzymes are also sensitive to the discriminator base position of their cognate tRNAs (23).
SerRS also aminoacylates tRNA Sec with serine as the first step for the incorporation of selenocysteine into proteins. Components of the selenocysteine insertion machinery have been identified by computational methods in species of the three branches of the tree of life. Recently, some of these components have also been identified in trypanosomatids, and the existence of a tRNA Sec in these organisms has been verified (24). The Trypanosoma selenoproteome appears to contain three selenoproteins, and Trypanosoma brucei is highly sensitive to the action of specific selenoprotein inhibitors (25). Thus, the characterization of the selenoprotein incorporation machinery of Trypanosoma may offer new insights into an essential pathway for these parasites and offer new strategies for drug development.
The serylation reaction of tRNA Sec has been characterized in detail in Escherichia coli and Homo sapiens (26,27), but the system has not been characterized in protozoans or other basal eukaryotes. In H. sapiens and E. coli, tRNA Sec transcripts have been described as being, respectively, 10-and 100-fold less efficient substrates for SerRS than tRNA Ser . It has been proposed that this difference is caused by the unique structure of tRNA Sec (27).
Here we report the first characterization of the serylation of tRNA Sec in trypanosomatids. We confirm the expression of tRNA Sec in T. brucei and show that, unlike most trypanosomal tRNAs, it is exclusively localized in the cytosol. Furthermore, we characterize its aminoacylation with serine by Trypanosoma cruzi SerRS. We show that the aminoacylation constants of tRNA Sec in Trypanosoma differ substantially from those reported in other organisms, suggesting that kinetic differences in the serylation activity of tRNA Sec may be species-specific and that regulatory strategies may exist based on the efficiency of serine-tRNA Sec synthesis.

EXPERIMENTAL PROCEDURES
Materials-Oligonucleotides were synthesized by Sigma Genosys. L-[ 3 H]serine and HisTrap nickel columns were from Amersham Biosciences. Restriction enzymes were from New England Biolabs, Pfu Ultra DNA polymerase was from Stratagene, and vector pQE-70 was from Qiagen. Novablue cells were from Novagen. T. cruzi genomic DNA was a gift from Dr. P. Bonay (Universidad Autónoma de Madrid, Spain).
RNAi-mediated Ablation of SerRS-RNAi-mediated ablation of the T. brucei SerRS was performed using stem-loop constructs containing the puromycin resistance gene as described The blots were probed for the T. brucei tRNA Ser and tRNA Ile . The latter serves as a control that is not affected by the RNAi. The RNA fractions were resolved on long acid urea gels that allow us to separate aminoacylated (aa) from deacylated (dea) tRNAs. The relative amounts of deacylated tRNA Ser and tRNA Ile are indicated at the bottom. For each lane, the sum of aminoacylated and deacylated tRNA was set to 100%. C, the same as B, but analysis was done for the tRNA Sec . The x refers to an unidentified band probably corresponding to phosphoserine. Right panel, the gel was processed for Northern hybridization and probed for the tRNA Sec , the cytosolspecific initiator tRNA Met and the imported tRNA Ile . B, tRNA Sec is expressed in both life cycle stages. Total RNA, 5 and 7 g each, of procyclic (Proc.) and bloodstream (Blood) stage T. brucei was analyzed by Northern hybridization as in A. The relative expression levels (set to be 1 for procyclic RNA) of the tRNA Sec when compared with the expression of tRNAIle are indicated at the bottom. C, expression of tRNA Sec is not induced in medium containing selenium. Five g of total RNA isolated from procyclic T. brucei grown in the absence (0) or the presence of 0.005 and 0.5 g/ml added Na 2 SeO 3 were analyzed by Northern hybridization as in A and B. The relative expression levels are indicated. (28). As an insert, we used a 498-bp fragment (nucleotides 1-498) of the T. brucei SerS gene. Transfection of T. brucei (strain 29-13), selection with antibiotics, cloning, and induction with tetracycline were done as described (29). To analyze the in vivo charging levels of the tRNA Ser and the tRNA Sec , we isolated total RNA from uninduced and induced cells by using the acid guanidinium isothiocyanate procedure (30). The tRNAs remain aminoacylated during this procedure due to the low pH employed by the method. Subsequently, the RNA samples were analyzed on 50-cm-long acid urea polyacrylamide gels as described (31), which can resolve aminoacylated from deacylated tRNAs. The gels were analyzed by Northern hybridization (32). The following 32 P 5Ј-end labeled oligonucleotides were used as probes: 5Ј-TGGCGTCAC-CAGCAGGATTC-3Ј (for the tRNA Ser CGA ) and 5Ј-ACCAGCT-GAGCTCATCGTGGC-3Ј (for tRNA Sec ).
tRNA Localization Studies-To determine the intracellular localization of tRNA Sec , we prepared mitochondria free of cytosolic RNAs by digitonin extraction and subsequent RNase A digestion from procyclic T. brucei as described (32). RNAs from total cells or isolated mitochondria were purified as indicated above (30). To test whether tRNA Sec concentrations were different in different life stages of T. brucei, total RNA of procyclic (strain 427) and bloodstream (strain AnTat1.1, a gift from Dr. Seebeck, University of Bern, Switzerland) stages of T. brucei was purified. To test whether expression of tRNA Sec is induced in medium containing selenium or under oxidative stress, total RNA was isolated from procyclic T. brucei grown in the absence (0) or the presence of 0.005 and 0.5 g/ml added Na 2 SeO 3 or in the absence or presence of 10 and 25 mol of added H 2 O 2 (data not shown). For all RNA fractions, the relative expression level of the tRNA Sec was analyzed by Northern hybridization using the tRNA Sec -specific oligonucleotide indicated above. The initiator tRNA Met and tRNA Ile were detected using the oligonucleotide 5Ј-GTTGGTTTCGATCCAACG-3Ј and 5Ј-TGCTCCCGGCGGGTTCGAA-3Ј, respectively.
tRNA Preparation-Constructions containing a T7 promoter followed by the gene encoding wild type or mutated tRNAs were assembled using six DNA oligonucleotides that were first annealed, and then ligated, between HindIII and BamHI restriction sites of plasmid pUC19. In vitro transcription using T7 RNA polymerase was performed according to standard protocols (33). Transcripts were separated on denaturing PAGE, and full-length tRNAs were eluted from gel using an electroelution apparatus (Schleicher & Schüll) and refolded (2 min at 90°C followed by a gradual reduction of temperature in presence of 2.5 mM MgCl 2 ). Finally, aminoacylation plateaus were used to calculate the concentration of active molecules for each tRNA preparation.
Enzyme Cloning and Purification-The 1.4-kbp intron-less gene coding for T. cruzi SerRS (Tc00.1047053511163.10) was amplified by PCR from genomic DNA (strain MC) using Pfu Ultra DNA polymerase and cloned in the vector pQE-70 for bacterial expression of a C-terminal His 6 -tagged protein. The correct sequence of the gene was checked by sequencing it entirely. Novablue E. coli cells (Novagen) transformed with this construction were grown at 21°C up to A 700 nm ϭ 0.3. Protein expression was then induced with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside during 12 h. Purification on nickel affinity columns was performed using standard procedures.

JOURNAL OF BIOLOGICAL CHEMISTRY 38219
Ci/mol) and varying concentrations of tRNA transcripts (1-80 M). The reaction was initiated by the addition of pure enzyme, and samples of 20 l were spotted onto Whatman No. 3MM discs at varying time intervals (usually 2 min). Radioactivity was measured by liquid scintillation. Enzyme concentration was experimentally determined for each tRNA to obtain linear velocities. Kinetic constants were obtained from Lineweaver-Burk plots using a minimum of two independent measurements and five tRNA concentrations.
Phylogenetic and Sequence Analyses-The sequences of seryl-tRNA synthetases used were pulled from GenBank TM , either directly or by BLAST searches of available genomic sequences (34). The total set of sequences was aligned with CLUSTALX (35). All sequences were initially included in our analysis but were later culled to focus our analysis on the eukaryotic clade that contains the Trypanosoma sequences. Coiled-coil predictions were done using the programs PAIR-COIL (36), COILS (37), and MULTI-COIL (38). Phylogenetic distributions were calculated by parsimony, distance and maximum likelihood methods using PHYLIP 3.63 package (39). Maximum parsimony analysis was done using PROTPARS. The neighbor-joining method was applied using the programs NEIGHBOR and PRO-TDIST, with the Dayhoff 120 substitution matrix. Maximum likelihood phylogenies were calculated with the program PROML (39) using the JTT substitution matrix. The programs SEQBOOT and CONSENSE (39) were used to estimate the confidence limits of branching points from 1000 bootstrap replicates in the parsimony and distance calculations, whereas 100 bootstrap replicates were used for the maximum likelihood trees. Several combinations of sequences were used in our analyses to test the robustness of the trees obtained. All the groups of sequences used produced exactly the same tree topology reported here.

RESULTS
RNAi-mediated Ablation of the Trypanosomal SerRS-To confirm that the putative T. brucei SerS gene encodes the functional enzyme in charge of serylation of tRNA Ser and tRNA Sec , we established a stable transgenic cell line that allows tetracycline-inducible RNAi-mediated ablation of the protein. This cell line stops growing after induction of RNAi, confirming that the putative SerRS is essential for normal growth of the parasite (Fig. 1A). To determine the biochemical phenotype of the RNAi cell line, we isolated total RNA from untreated cells and from cells grown in the presence of tetracycline. Subsequently, the RNAs were resolved on long acid urea polyacrylamide gels (31), which, in combination with Northern analysis, allow us to determine the ratio of charged to uncharged tRNA Ser and tRNA Sec , respectively. The results in Fig. 1, B and C, show that ablation of the putative SerRS results in selective accumulation of both uncharged tRNA Ser and uncharged tRNA Sec , indicating that the SerRS identified in our study is the enzyme responsible for in vivo serylation of both of these tRNAs.
Localization of the tRNA Sec -Unlike in other eukaryotes, most tRNAs in trypanosomatids have a dual localization. Approximately 95% are found in the cytosol and function in cytosolic translation; however, a small fraction (ϳ5%) are imported into the mitochondrion and function in organellar protein synthesis (32). Consequently, trypanosomal tRNAs are always encoded in nucleus and never on the mitochondrial DNA. T. brucei encodes a single selC gene, coding for the tRNA Sec (25). To confirm the expression of this tRNA and to determine its intracellular localization, we carried out Northern blot analyses using total and mitochondrial RNA fractions from procyclic T. brucei. The result in Fig. 2A shows that the tRNA Sec is only detected in the cytosol but not in the mitochondrion. Exclusive cytosolic localization is exceptional in trypanosomatids, the only cytosol-specific tRNA known to date being the initiator tRNA Met (32).
tRNA Sec Expression Is Not Influenced by the Life Cycle Stage, Selenium, or H 2 O 2 -In the next series of Northern blots, we tested whether the relative amount of tRNA Sec when compared with other tRNAs was dependent on the life cycle stage or sensitive to the concentration of selenium or H 2 O 2 in the culture medium. Fig. 2B shows that trypanosomal tRNA Sec is expressed to very similar levels in both the procyclic and the bloodstream forms of T. brucei. This suggests that selenocysteine-containing proteins play a role throughout the life cycle of T. brucei. In

FIGURE 5. Sequence alignment of the region of SerRS sequences that contains the synapomorphy that links metazoan and trypanosomatid enzymes.
A, the alignment was performed using CLUSTAL X (35). T. cruzi and T. brucei SerRS sequences are in bold. Helix 1 and helix 2 indicate the long helices that form the coiled-coil arm found in T. thermophilus SerRS. The insertion that is present in metazoan and trypanosomatid sequences is boxed and labeled accordingly. B, crystallographic structure of one monomer of T. thermophilus SerRS (16). The apparent insertion point of the metazoan and trypanosomatid synapomorphy is marked by an arrow and labeled accordingly.

TABLE 1 Serylation parameters of tRNASec, tRNA Ser , and its variants in presence of SerRS
The rate of aminoacylation of the wild type tRNA Ser is 9.56 s Ϫ1 . For better readability, the K cat of each mutant is expressed as a percentage of the K cat value of the wild type tRNA Ser (i.e. a rate of 9.56 s Ϫ1 corresponds to a K cat of 100%). The overall efficiency of serylation has been determined as the ratio between the K m and the K cat . To enhance the readability, the efficiency of the wild type tRNA Ser has been set at 1. Values higher than 1 denote tRNA variants that are less efficiently aminoacylated and vice versa. Numbers in parentheses refer to the mutations detailed on Fig.  4, B and C. contrast to reports in vertebrates (40), the expression of the tRNA Sec in procyclic T. brucei is not affected by the addition of selenium to the growth medium (Fig. 2C). A similar result was obtained when the parasites were grown in the presence of H 2 O 2 (data not shown).

Characterization of the Serylation Reaction of tRNA Ser and tRNA Sec by T. cruzi SerRS-
The cloning of SerRS gene was performed directly from genomic DNA. The 54-kDa protein was expressed in E. coli and purified to homogeneity by affinity chromatography. Gel filtration experiments with the purified enzyme confirmed its ability to form dimers (data not shown), suggesting that T. cruzi SerRS is a classical class II aminoacyl-tRNA synthetase (5).
Transcripts of Trypanosoma tRNA Ser and tRNA Sec were used in aminoacylation assays, and both were shown to be efficiently aminoacylated by SerRS (Fig. 3A). The kinetic constants for the serylation of both tRNAs by SerRS were then determined (Fig. 3B). The K m for tRNA Ser of T. cruzi SerRS was calculated at 3.2 M, a value comparable with those found in other systems but 8-fold higher than that calculated for tRNA Sec with the same enzyme. On the other hand, the K cat was essentially identical to that calculated for tRNA Sec . As a result, the K cat /K m value for Trypanosoma tRNA Ser is 7-fold lower than that of T. cruzi tRNA Sec . Interestingly, these values are markedly different from those reported in other species. Thus, in the human system (26), the serylation of tRNA Sec is 10-fold less efficient than that for tRNA Ser , and this difference grows to 100-fold in favor of tRNA Ser in E. coli (27) (Fig. 3C).
Identity Determinants in the Acceptor Stem of tRNA Ser -To understand how T. cruzi SerRS recognizes its cognate tRNAs, mutants of tRNA Ser CGA were produced, and their ability to be charged with serine was measured in vitro in the presence of purified SerRS. Fourteen tRNA variants were generated, covering nearly 80% of the primary structure. The design of the mutations was dictated both by the sequence conservation among T. cruzi tRNA Ser and tRNA Sec sequences (Fig. 4A) and by the distribution of identity elements in homologous tRNA Ser sets. A set of modifications targeted individual nucleotides or base pairs in the acceptor stem (variants 1-9) (Fig. 4B), whereas a second set of changes introduced large deletions or sequence swaps of discrete tRNA domains (variants 10 -14) (Fig. 4C).
The change of the discriminator base G73 for pyrimidines had a dramatic effect on the charging of tRNA Ser (mutants 1 and 2) (Table 1), indicating that the enzyme strongly recognizes this position of the acceptor stem. This sensitivity to the discriminator base sequence is reminiscent of the recognition mechanisms described for H. sapiens and archaeal SerRSs (17)(18)(19).
The first base pair (1:72) was found to be the most sensitive position in the acceptor stem of tRNA Ser . Indeed, the simple inversion of the G1:C72 base pair (mutant 4) completely abolished the aminoacylation of this substrate by the enzyme. The introduction of a G1:U72 base pair in the tRNA Ser scaffold had no effect on the ability of this tRNA to be charged by SerRS (mutant 5), indicating that G1 may be contributing important contacts to the recognition by SerRS (Table 1). It should be noted that both the discriminator base G73 and the G1 base found to be essential for recognition of tRNA Ser by Trypanosoma SerRS are conserved in the sequences of tRNA Sec of the same species.
Additional mutations in other regions of the acceptor stem had no significant effect on the overall recognition of tRNA Ser by T. cruzi SerRS (mutants 6 -9). Overall, our acceptor stem mutations affect mostly K cat values, suggesting a defect in the reactive positioning of the acceptor extremity rather than a loss in binding energy.
Identity Determinants Beyond the Acceptor Stem of tRNA Ser -Next, we wanted to investigate the importance of the pseudoknot (D arm, T arm, and variable loop) and anticodon domains of T. cruzi tRNA Ser for its recognition by SerRS. The three domains of the tRNA that form the pseudoknot were individually flipped (mutants 11-13) to vary their sequences without modifying the local secondary structure or the strength of their internal base pairs (Fig. 4C). None of the three corresponding mutants showed any catalytic defect, suggesting that the recognition mechanism for the elbow of the tRNA is based exclusively on interactions with the sugar-phosphate backbone (Table 1).
To further test this hypothesis, the entire variable loop was excised and then replaced by the most commonly found class I variable loop (mutant 14) (Fig. 4C). This replacement had a dramatic effect on aminoacylation, causing a Ͼ8000-fold drop in K cat /K m . This mutation affected both catalytic and affinity constants. Our combined data indicate that this domain is recognized in a sequence-unspecific manner, provides binding energy, and acts as a guide for the reactive positioning of the tRNA (Table 1).
Finally, the entire anticodon loop was deleted and replaced by a stretch of four uridines (mutant 10) (Fig. 4C). This linker was used because it is the least likely to stabilize unanticipated structures in this region of the tRNA (13). This deletion mutant had only a modest effect on the velocity of the aminoacylation reaction. This is in agreement with other studies that have shown that the anticodon domain of eukaryotic tRNA Ser contributes poorly to the serine identity (for a review, see Ref. 11) ( Table 1).
Phylogeny of Trypanosoma SerRS-The known genomes of Trypanosoma code for only one SerRS gene. The corresponding protein is likely used both in the cytoplasm and in the mitochondria, but no conventional targeting sequence can be identified in their sequences. The protein is 477 amino acids long, and it contains an N-terminal domain predicted to form a coiled coil domain (37), as seen in the available structural data for homologous enzymes (16,41).
Multiple sequence alignments readily show the presence of a large synapomorphy that clusters trypanosomatid SerRSs with FIGURE 6. Phylogenetic analysis of trypanosomal SerRS. A, maximum likelihood tree of archaeal, bacterial, and eukaryotic seryl-tRNA synthetase sequences. The overall architecture of this tree was obtained with all methods used, but bootstrap support was weak (not shown). B, consensus tree of eukaryotic seryl-tRNA synthetase sequences. The association between trypanosomatid and metazoan SerRSs is highlighted. Numbers correspond to percentages of bootstrap support for each node from the parsimony, distance, and maximum likelihood analyses. DECEMBER 15, 2006 • VOLUME 281 • NUMBER 50 the rest of metazoan sequences (Fig. 5A). This synapomorphy is an insertion of 20 amino acids located at the center of the N-terminal coiled-coil motif of this group of SerRSs (Fig. 5B). Coiledcoil prediction of this group of sequences suggest that in metazoans and trypanosomatids, this region may extend beyond the length seen in the structures solved so far (data not shown).

SerRS Is Metazoan-like Enzyme with High Affinity for tRNA Sec
Our phylogenetic analysis of SerRS sequences is the first one of this enzyme to include kinetoplastid sequences. When sequences from all life domains were used, our results consistently agreed with previous analyses (42)(43)(44)(45), supporting the conclusion that the overall evolution of this enzyme conforms to the canonical phylogenetic tree derived from ribosomal RNA sequences (Fig. 6A) (42,44). Although we were unable to obtain significant bootstrap support for the central branching points of the general tree, the trypanosomatid sequences strongly associated with metazoan sequences with strong bootstrap support, both in general trees and in those limited to the major eukaryotic lineages (Fig. 6, A and B). It should be stressed here that the region corresponding to the synapomorphy that also links Trypanosoma sequences with metazoan ones was not used in our phylogenetic analyses.

DISCUSSION
The constraints acting over tRNA recognition are strictly intra-specific and, for each species, constitute a complex set of recognition and rejection elements that ensures faithful translation (46). Possibly, this complex set of positive and negative identity elements contributes to the stability of the genetic code and limits its size. Amino acid fidelity can be controlled by editing domains, but the incorporation of new amino acids also requires an increase in complexity of the tRNA recognition problem, which may be impossible to assume without increasing the rate of aminoacylation errors (3).
To study the evolution of these sets of recognition elements, we decided to characterize the recognition of tRNA Ser and tRNA Sec by Trypanosoma SerRS. In doing so, we were expecting to extract information about the evolution of this recognition mechanism in the basal part of the eukaryotic phylogenetic tree. For convenience, we chose T. brucei as a model for the in vivo studies, whereas the in vitro studies were performed with T. cruzi proteins and tRNAs. The overall identity between T. cruzi and T. brucei SerRSs is 80%, and the sequences of their tRNA Ser are essentially identical.
We have shown that Trypanosoma SerRS recognizes its cognate tRNAs using a combination of structural signals in the variable loop and the sequence information of the discriminator base and the first base pair of the acceptor stem. The discriminator base G73 and the G1 base are essential for recognition by Trypanosoma SerRS. These bases are conserved among all serine tRNA isoacceptors in Trypanosoma, and strongly influence the velocity of the serylation reaction.
Previous studies on tRNA Ser recognition in the three kingdoms of life show that acceptor stem recognition by SerRS fluctuates between the first four base pairs and the discriminator base of the acceptor stem (17,47,48). In the case of Trypanosoma, this recognition has shifted toward the CCA end of the molecule, and the discriminator base and the first base pair constitute the region recognized by SerRS.
We have established that the presence of a long variable loop is a prerequisite to tRNA Ser recognition but that this mechanism is not sequence-specific. Moreover, the complete deletion of tRNA Ser anticodon domain does not affect its serylation by SerRS, confirming that the entire stem and anticodon loop is ignored by the enzyme during the recognition process. This feature is common to all serine systems studied so far with the only exception of Methanosarcina barkeri SerRS, which displays a strong interaction with the G30:C40 base pair in the anticodon stem (47).
In summary, our data suggest that the tRNA specificity of Trypanosoma SerRS relies on a two-pronged mechanism based on interactions with the G73 discriminator base and the G1 base on one side and the sequence-independent recognition of the variable loop on the other. Interestingly, this recognition mechanism functionally associates trypanosomatid SerRSs with their metazoan homologs and separates them from other eukaryotic enzymes such as S. cerevisiae SerRS. These functional relationships are in agreement with the phylogenetic connections that we report here and the presence of clear sequence synapomorphies that link these enzymes evolutionarily.
A possible explanation to this functional convergence would be a lateral gene transfer event between trypanosomatids and metazoans that could be favored by the parasitic nature of the former. A second possibility, more in agreement with polyphiletic views of eukaryote evolution (4), would be that the recognition solution displayed by metazoans and trypanosomatids is ancient, having persisted in these two groups of organisms since their radiation at the base of the eukaryotic evolutionary tree. Our phylogenetic studies do not allow us to discard either possibility.
From our studies, it is clear that tRNA Sec is not imported into the mitochondria of T. brucei. Some of us have recently shown that tRNA localization signals in this species are confined to two nucleotide pairs in the T-stem (49). In the cytosol-specific initiator tRNA Met , these positions include the anti-determinants that, based on experiments with vertebrate initiator tRNA Met (50), are predicted to prevent interaction with translation elongation factor 1a (eEF-1a). The T-stem sequence of the cytosolic tRNA Sec is different from all other T. brucei tRNAs including the cytosolic initiator tRNA Met . However, a feature that is shared between the two cytosol-specific tRNAs in T. brucei is that neither interacts with eEF-1a. Instead, the initiator tRNA Met interacts with initiation factor 2, and the tRNA Sec interacts with the specialized elongation factor SelB/ EFsec. Thus, although the structural determinants that prevent interaction with EF-1a are different for both tRNAs, the mechanism for cytosolic localization might in both cases be achieved by exclusion of EF-1a binding. In other words, all tRNAs that interact with eEF-1a might be imported into mitochondria, and the ones that do not remain in the cytosol.
Another interesting feature of trypanosomatid SerRSs is their apparently high affinity for tRNA Sec . Indeed, the K cat /K m value for the aminoacylation of Trypanosoma tRNA Ser (which are comparable with values established for other species) is 7-fold lower than that found for T. cruzi tRNA Sec , a result that is in direct contrast with the values reported for the human sys-tem (26) (where the tRNA Ser is 10-fold more efficiently charged than tRNA Sec ) or in E. coli (27) (100-fold difference in favor of tRNA Ser ) (Fig. 3B).
It should be noted here that all the studies where these values are reported (including this one) were performed with transcript tRNAs, thus excluding the potential effect of base modifications in either substrate. However, there seems to be a clear difference between the relative aminoacylation of tRNA Sec and tRNA Ser between trypanosomatids and other species. This may be explained by a high requirement for selenoprotein synthesis in these species. Our attempts to increase the levels of tRNA Sec in T. brucei by the presence of oxidative stress or different growth conditions were unsuccessful, but these results are not necessarily contradictory with a high level of selenocysteine used by these organisms.
Indeed, our results may be related to a recently reported analysis of the Trypanosoma proteome (25) that revealed a novel multidomain selenoprotein named SelTryp. This, together with the high sensitivity displayed by T. brucei to the selenoprotein inhibitor auranofin (25), may indicate that these organisms have evolved new uses for selenoproteins that require a particularly efficient tRNA Sec serylation reaction.