Atypical archaeal tRNA pyrrolysine transcript behaves towards EF-Tu as a typical elongator tRNA

Abstract

The newly discovered tRNA^Pyl is involved in specific incorporation of pyrrolysine in the active site of methylamine methyltransferases in the archaeon Methanosarcina barkeri. In solution probing experiments, a transcript derived from tRNA^Pyl displays a secondary fold slightly different from the canonical cloverleaf and interestingly similar to that of bovine mitochondrial tRNA^Ser(uga). Aminoacylation of tRNA^Pyl transcript by a typical class II synthetase, LysRS from yeast, was possible when its amber anticodon CUA was mutated into a lysine UUU anticodon. Hydrolysis protection assays show that lysylated tRNA^Pyl can be recognized by bacterial elongation factor. This indicates that no antideterminant sequence is present in the body of the tRNA^Pyl transcript to prevent it from interacting with EF-Tu, in contrast with the otherwise functionally similar tRNA^Sec that mediates selenocysteine incorporation.

INTRODUCTION

In the methanogenic archaeal Methanosarcina species, methylamine methyltransferases are essential enzymes involved in methanogenesis. All known methylamine methyltransferase genes possess an in-frame amber (UAG) codon that does not stop the translation during protein synthesis (1,2); instead, a modified lysine residue, called pyrrolysine, is incorporated at these positions in the active site of the proteins (3). The mechanism by which this occurs is not yet understood, but there is a strong analogy with the selenocysteine system (4). The Methanosarcina barkeri genome sequence reveals the presence of the pylT gene, which codes for the suppressor tRNA^Pyl, and of three LysRS genes, one typical of class I, one of class II and the third one (pylS) encoding a LysRS-like protein (5). It was first proposed that, after lysylation of tRNA^Pyl by the pylS product (PylS), lysyl-tRNA^Pyl is converted into pyrrolysyl-tRNA^Pyl. However, it was recently shown that class II LysRS catalyses tRNA^Pyl aminoacylation in a ternary complex with class I LysRS (6), albeit the protein PylS is proposed to catalyse the modification step that produces pyrrolysyl-tRNA^Pyl.

To shed light on the pyrrolysine incorporation pathway, one can question whether tRNA^Pyl bears more resemblance to a typical elongator tRNA or to tRNA^Sec. The latter, after serylation by canonical SerRS and conversion into selenocysteyl-tRNA^Sec, binds a specialized elongation factor, the SELB protein (7), rather than EF-Tu (8) to reach the ribosomal A-site. The presence of an antideterminant box of three base pairs in the acceptor branch of bacterial tRNA^Sec distinguishes it from other tRNAs and is responsible for the rejection by EF-Tu (9).

The present study establishes structural features of tRNA^Pyl in solution and the ability of tRNA^Pyl transcripts to be charged by a canonical class II LysRS as well as its possible interaction with a canonical elongation factor. These functional peculiarities of tRNA^Pyl, which differentiate it from tRNA^Sec, will be discussed.

MATERIALS AND METHODS

Materials

T7 RNA polymerase (10) and EF-Tu-GDP from Thermus thermophilus HB8 (11) were purified from overproducing strains as described. LysRS from yeast was cloned in fusion with 6 C-terminal histidine residues in plasmid pQE70 (Qiagen), overexpressed and purified as described (12). Oligonucleotides were from Genset, radioactive l-[¹⁴C] lysine (318 mCi/mmol) was from Amersham, and phosphoenolpyruvate, pyruvate kinase and the ribonucleotides were from Boehringer. Nucleases S1, V1, T2 and T1 were from Pharmacia, USB, Invitrogen and IRL New Zealand, respectively, and lead acetate was from Merck.

Preparation of transcript and tRNA

Transcripts of tRNA^Pyl from M.barkeri were obtained by in vitro transcription of synthetic genes by an established procedure (13). Yeast tRNA^Lys transcript was prepared as described (14). Purification of the transcripts was performed by electrophoresis on preparative denaturing 12% polyacrylamide gels and elution from gel slices using a Schleicher and Schuell Biotrap apparatus (Dassel, Germany). An enriched yeast tRNA^Lys fraction was a gift from G. Keith (IBMC, Strasbourg).

Ultraviolet melting measurements

Melting curves of M.barkeri tRNA^Pyl and yeast tRNA^Lys were recorded on a Uvikon V799 spectrophotometer (Kontron Instruments). The temperature was raised from 15 to 75°C at a rate of 0.5°C/min using a Peltier system (Biotek). For measurements, tRNAs were diluted to a concentration of 0.32 µM in a buffer containing 50 mM sodium cacodylate pH 7.4, 50 mM KCl and 5 mM MgCl₂. Sodium cacodylate was used because it shows only slight pH variation as a function of temperature. It was verified by gel electrophoresis under denaturing conditions that the transcripts were not degraded after the melting experiments.

Structural mapping procedures

Labelling of the 5′-end of the tRNA^Pyl transcript was performed as described previously (15). RNA was digested with various nucleases for 10 min at 20°C (for V1, S1, T1, T2) or 5 min at 50°C (for V1, T1, T2) in buffer T (100 mM KCl, 5 mM MgCl₂ and 50 mM HEPES–NaOH pH 7.5). For digestion with S1 nuclease, 1 mM ZnCl₂ was added. Reaction mixtures (10 µl) contained 50 000 c.p.m. of labelled transcript supplemented with 1 µg of unlabelled tRNA and 9 × 10^–4 U RNase V1, 4.7 U RNase S1, 0.48 U RNase T1 or 0.12 U RNase T2. Reactions were stopped by immediate cooling on ice and addition of 20 µl 0.6 M NaOAc, 4 mM EDTA and 0.1 µg/µl total tRNA, followed by phenol extraction and ethanol precipitation of the tRNA.

Lead mapping was performed in buffer T at two temperatures. A freshly prepared Pb(Oac)₂ solution in H₂O was used at 1, 3 and 10 mM final concentrations. Samples (10 µl) were incubated for 5 min at 20°C or 1 min at 50°C. Reactions were stopped by cooling on ice, adding EDTA to a final concentration of 33 mM and ethanol precipitation.

For assignment of cleavage positions, alkaline degradations were performed by incubation of labelled transcript for 4 min at 80°C in 50 mM NaHCO₃ pH 9.0. Guanine ladders were generated by RNase T1 digestion under denaturing conditions (16). Controls without probes were run in parallel on each gel.

Aminoacylation assays

Before aminoacylation, transcripts were renatured by heating for 2 min at 80°C followed by slow cooling to room temperature. Lysylation of tRNA^Pyl transcript and yeast tRNA^Lys was performed in 100 mM HEPES–NaOH pH 7.4, 30 mM KCl, 20 mM MgCl₂, 2 mM ATP, 10 mM DTE, 16 µM L-[¹⁴C]lysine and 5.9 µM yeast LysRS at 37°C. Kinetic parameters for tRNA^Pyl and tRNA^Lys were determined in the presence of 400 nM and 1 nM yeast LysRS, respectively. Transcript concentrations covered the range between 0.6 and 2.4 µM. Kinetic constants (K_m, k_cat) were derived from Lineweaver–Burk plots.

Hydrolysis protection assays

These experiments were conducted essentially as described before (9,17). Preparative lysylations of tRNA^Pyl(uuu) and tRNA^Lys transcripts were performed under the conditions given above. After incubation at 37°C, reactions were stopped by phenol extraction in the presence of sodium acetate at pH 4.5. Transcripts were ethanol precipitated, washed and dried. Prior to that treatment, an aliquot of each sample was spotted on a 3MM Whatman filter and 5% trichloroacetic acid (TCA) precipitated to measure the amount of lysylated tRNAs. For the experiments, ∼70 pmol of each aminoacylated transcript (taking into account the lysylation level) were dissolved in buffer H (50 mM Tris–HCl pH 7.5, 50 mM KCl, 10 mM MgCl₂) and incubated with 5- or 10-fold excess of EF-TuGTP in a volume of 100 µl. Controls were carried out in parallel with EF-TuGTP replaced by either EF-TuGDP or buffer H. After formation of the ternary complexes (10 min incubation on ice), samples were further incubated at 37°C to allow hydrolysis of the chemically unstable ester bond to proceed. Aliquots (20 µl) were removed after 0, 30, 60 and 90 min and spotted onto 3MM Whatman filters. After TCA precipitation, the residual radioactivity was measured by liquid scintillation counting.

RESULTS AND DISCUSSION

Structural peculiarities of tRNA^Pyl

The sequence of tRNA^Pyl that was tentatively folded into a cloverleaf shows a number of peculiarities suggesting deviations from the canonical fold (5) (Fig. 1A). They concern the lengths of the anticodon arm and D-loop, the size of the linker between acceptor and D-arms and the number of nucleotides forming the variable region. Deviations also include the absence of the conserved G₁₉ and C₅₆ residues in the D- and T-loops, suggesting an atypical interaction between these two loops. Given these features, tRNA^Pyl shares only 40% identity with conventional lysine-accepting tRNAs from M.barkeri (Fig. 1A) and its sequence does not contain a full lysine identity set, namely residue U₃₆ present in all known lysine-accepting tRNAs. Notice, however, that the central anticodon position is occupied by a U residue, which contributes to the identity of class IIb aaRSs in Escherichia coli (19).

Figure 1.

Cloverleaf structure of (A) M.barkeri tRNA^Pyl and (B) bovine mitochondrial tRNA^Ser(UGA). Numbering is according to Sprinzl et al. (18). Missing residues compared with canonical tRNAs are indicated by empty dots and non-canonical primary features are boxed. Consensus nucleotides in M.barkeri lysine-accepting tRNAs are in grey, whereas those conserved among all tRNAs are circled.

The actual fold of tRNA^Pyl was searched in silico and by solution probing. The M-fold program (20) yielded various structures, including a cloverleaf. This opens the possibility that several tRNA^Pyl transcript conformers coexist in solution, as observed for human mitochondrial tRNA^Lys. In the latter tRNA, the structural heterogeneity could be ascribed to the absence of post-transcriptional modifications that hinder formation of the correct fold (21). For tRNA^Pyl transcripts, the coexistence of different conformers was sought by melting experiments (Fig. 2), but no evidence in support of this hypothesis was found. Indeed, the melting profile of the transcript presents rather well-defined transitions corresponding likely to the opening of the tertiary structure at 36°C and especially of a unique secondary structure at 65°C. This behaviour is also observed for yeast tRNA^Lys transcript (Fig. 2) with, however, a 5°C download shift in the melting temperature of the cloverleaf. This difference is likely due to another distribution of G–C pairs with a strong continuous G–C stack in the acceptor stem of tRNA^Pyl.

Figure 2.

Melting profile of M.barkeri tRNA^Pyl and yeast tRNA^Lys transcripts. First derivatives of the absorption profiles obtained at 260 nm with M.barkeri tRNA^Pyl (black circles) and yeast tRNA^Lys (open circles) transcripts are displayed.

A cloverleaf fold mimicking mitochondrial tRNA^Ser

An overall view of the single- and double-stranded regions of tRNA^Pyl came from structure mapping using lead as the structural probe sensitive to single-stranded regions. Experiments were performed in the presence of increasing amounts of the probe at 20 and 50°C, corresponding to the presence and absence of tertiary structure based on the UV melting profile (Fig. 2). The cleavage patterns show strong cuts in the anticodon loop and variable region and weaker ones in the D- and T-loops (Fig. 3A). At 50°C, and even with a low lead concentration, all positions of the tRNA are cut, except those in the T-arm which consists entirely of G–C pairs. In agreement with the melting experiments, these data indicate an opening of the tertiary structure as well as of the acceptor, D and anticodon stems. Additional data were obtained by probing with RNases (Fig. 3). RNase V1 cleavages agree with the existence of double-stranded acceptor, D and anticodon stems. Nuclease probing with RNase S1, T1 and T2 also supports the existence of single-stranded residues belonging to the anticodon loop, discriminator position G73 and variable region. Altogether, lead and nuclease probing confirms the cloverleaf fold depicted in Figure 1A, as previously suggested (5). A slightly different fold with an opening of the anticodon stem at base pair 27–43 was recently proposed (6), but the RNase V1 cuts in this region favour a continuous stacked anticodon branch (Fig. 3).

Figure 3.

Probing of tRNA^Pyl transcript. (A) Autoradiogram of a 12% denaturing gel of probing experiments on 5′-labelled M.barkeri tRNA^Pyl(UUU) transcript. Experiments using Pb²⁺ were conducted at 20 and 50°C (left part). Final concentrations of Pb²⁺ are indicated. Experiments with RNases T1, T2, V1 and S1 were conducted at 20°C (right part). Control incubations (C) were run in parallel. Lane –S1 checks the effect of ZnCl₂ present in nuclease S1 buffer, lane AH represents an alkaline ladder and lane L represents a denaturing RNase T1 ladder. Numbering defines the position of G residues. T-, anticodon and D-loops and the variable domain are indicated. (B) Summary of the probing experiments displayed on the tRNA^Pyl cloverleaf. Lines at the 3′- and 5′-ends indicate untested regions. Nucleotides cut by nuclease S1, T1 and T2 are indicated by triangles. Cuts generated by RNase V1 are emphasized by white letters on grey or black backgrounds corresponding to weak and strong cleavages, respectively. Yellow areas correspond to nucleotides cut by probes specific of single-stranded regions and by Pb²⁺ at low temperature. Blue areas correspond to nucleotides cut by probes specific of double-stranded regions and by Pb²⁺ at high temperature. Regions cut by both types of enzymatic probes are in green. Regions resistant to Pb²⁺ cuts are boxed. Strokes indicate positions of spontaneous degradations.

In contrast with canonical native elongator tRNAs (22,23), but in accordance with other in vitro transcripts (24), cuts by both double- and single-strand specific probes are visible in the D-loop of the tRNA^Pyl transcript (Fig. 3). This indicates a relaxed tertiary structure of tRNA^Pyl, which probably corresponds to an overall L-shaped conformation. Whether post-transcriptional modifications in tRNA^Pyl stabilize this conformation remains unknown.

tRNA^Pyl strikingly resembles bovine mitochondrial tRNA^Ser(UGA) (Fig. 1) which has been shown to adopt an L-shaped architecture by chemical probing and NMR (25,26). Among the five peculiarities emphasized by boxes in tRNA^Pyl (Fig. 1A), four exist in mitochondrial tRNA^Ser (Fig. 1B), namely an anticodon stem of six base pairs, a small variable region, a one-nucleotide linker between acceptor and D-arms and a reduced D-loop.

Yeast LysRS aminoacylates tRNA^Pyl with UUU anticodon

An unsolved question in the pyrrolysine system is whether a canonical elongation factor can recognize charged tRNA^Pyl or if a specialized protein is required to carry it to the ribosome. To shed light on this question, we have searched for a possible interaction between Lys-tRNA^Pyl and EF-Tu from T.thermophilus. The prerequisite to this study is the preparation of an aminoacylated tRNA^Pyl molecule. For that, we attempted tRNA^Pyl aminoacylation by yeast class II LysRS. However, because of the peculiar structure of tRNA^Pyl and also due to the presence of the CUA amber anticodon lacking U36, a major lysine identity element in E.coli (27), lysylation of wild-type tRNA^Pyl seemed unlikely. This assumption was experimentally verified since no lysylation of the wild-type tRNA could be detected (Table 1). Therefore a tRNA^Pyl mutant with a lysine UUU anticodon was produced and assayed for lysylation. As anticipated, this tRNA^Pyl variant was a substrate for yeast LysRS, but with kinetic parameters very different from those measured with homologous tRNA^Lys (Table 1). Despite its reduced catalytic efficiency, tRNA^Pyl(UUU) can be charged to a plateau level of 12%, dependent on the yeast LysRS concentration. Incomplete charging is not due to a high proportion of inactive molecules, but reflects an equilibrium between the aminoacylation reaction and the enzymatic or chemical deacylation reactions that catalyse hydrolysis of ester bond between lysine and tRNA^Pyl. Since the lysylation rate is low, the equilibrium leads to incomplete charging as predicted by the plateau theory (28,29).

Table 1. Kinetic parameters for lysylation by yeast LysRS of yeast tRNA^Lys and M.barkeri tRNA^Pyl transcripts.

	KM (nM)	kcat (min–1)	kcat/KM (min–1 nM–1)
Yeast tRNALys	12	230	19.1
M.barkeri tRNAPyl(CUA)	ND	ND	ND
M.barkeri tRNAPyl(UUU)	800	0.018	2,25.10–5

ND, no lysylation detected.

In the context of lysylation, it is interesting that the wild-type tRNA^Pyl(CUA) is unable to inhibit tRNA^Pyl(UUU) lysylation (data not shown), indicating the importance of the UUU triplet interacting with the anticodon binding domain of yeast LysRS. This observation agrees with the recent finding that class IIb LysRS from M.barkeri is unable alone to lysylate tRNA^Pyl(CUA) but needs a helper protein, the class I LysRS (6). This latter protein likely anchors the CUA-containing anticodon loop in the tripartite complex.

Lysylated tRNA^Pyl transcript interacts with EF-Tu

Biochemical data (30) and crystal structures of two ternary complexes (31,32) indicate that elongator tRNAs interact with thermophilic elongation factor Tu via their aminoacylated 3′-end, phosphorylated 5′-end and ribose phosphate backbone of their aminoacyl acceptor branch (10 first base pairs). Modified bases as well as the whole anticodon branch are not necessary for this interaction (30,33,34). These properties justify use of the tRNA^Pyl(UUU) transcript to model the interaction between tRNA^Pyl and EF-Tu.

The potential of lysylated tRNA^Pyl to form a ternary complex with T.thermophilus EF-Tu was thus tested by hydrolysis protection assays. The method is based on the fact that the activated factor protects the labile amino acid ester bond from spontaneous base-catalysed hydrolysis when bound to aminoacyl-tRNA (17). Thermophilic EF-Tu (P07157) was used in these assays because it is similar to the archaeal factor (sequence information on M.barkeri genome was retrieved from DDBJ/EMBL/GenBank accession number DOE_ 49338). Phylogenic comparison of EF-Tu sequences shows that conserved amino acids are essentially present in M.barkeri EF-Tu, including those in contact with the acceptor stem of tRNA (32). Pyrrolysine attached to tRNA would not be expected to interfere with complex formation with EF-Tu since many other unnatural amino acids of similar size can be incorporated into proteins by the canonical translational machinery (35).

Deacylation of Lys-tRNA^Pyl occurs in the absence of EF-Tu with a half-life of 29 min, whereas in the presence of a 10-fold (or 5-fold) excess of activated protein it is increased to 310 min (or 155 min) (Fig. 4). These variations in deacylation rates compare well with those obtained with other elongator tRNA transcripts, in particular with yeast tRNA^Lys transcript (data not shown). A control experiment in the presence of the inactive form of EF-Tu (EF-TuGDP) shows a deacylation rate similar to that obtained in the absence of the protein with a half-life of 28 min (Fig. 4). Taken together, these results demonstrate the ability of aminoacylated tRNA^Pyl to bind standard elongation factor, suggesting the absence of antideterminant residues in the tRNA^Pyl framework that would hinder its recognition by EF-Tu. However, as discussed by LaRiviere et al. (36) and Asahara et al. (37), discrimination could also occur via specific thermodynamic contribution of the amino acid.

Figure 4.

Interaction of EF-TuGTP with Lys-tRNA^Pyl transcript. Rates of hydrolysis of the lysine residue from tRNA^Pyl(UUU) transcript in the presence of 7 or 3.5 µM EF-TuGTP (black and open triangles, respectively), EF-TuGDP (black circles) and absence of protein (black squares). The concentrations were 0.7 µM for RNA. When used, EF-TuGDP was at 7 µM. Times required to hydrolyse 50% of the labile lysyl-ester bound (t_1/2) are indicated.

Functional implications

The above conclusions lead to unforeseen consequences if they apply to native tRNA^Pyl. Indeed, recognition of lysylated tRNA^Pyl by EF-Tu would imply its participation in protein synthesis and the incorporation of lysine at amber stop codons. The alternative would be that the lysine bound to the tRNA by the M.barkeri lysylation system (6) would have to be sequestered and converted into pyrrolysine in a concerted reaction after charging before interaction with EF-Tu.

In plants and fungi, exclusion of initiator tRNA^Met from the elongation process is due to the presence of a large 2′-phosphoribosyl modification at purine 64 in the T-stem (38,39). Since our study uses unmodified tRNA^Pyl, it is not excluded that a modification in native tRNA^Pyl at a position important for EF-Tu recognition could prevent ternary complex formation. If so, participation of a specialized elongation factor might be possible. These open questions remain to be answered by complementary experiments, in particular to determine whether modified tRNA^Pyl is able to interact with EF-Tu.

How far is the parallel between selenocysteine and pyrrolysine systems?

A parallel between selenocysteine and pyrrolysine incorporation into proteins was pointed out after discovery of the 22nd amino acid (4,40) because of similarities at various steps of these two translation pathways. Both use tRNAs that are first charged by standard amino acids (serine or lysine) and are then modified in a second step into non-standard derivatives (selenocysteine or pyrrolysine). The modified amino acids are then incorporated into the active site of specific proteins in response to stop codons.

The parallel is imperfect, however, since selenocysteine is incorporated in response to the UGA opal codon and pyrrolysine in response to the UAG amber codon. Further, the selenocysteine pathway is widespread in all kingdoms of life while that of pyrrolysine is restricted to certain archaea and the eubacterium Desulfitobacterium hafniense (5). The two systems differ also in the overall structure and size of the tRNAs, large for tRNA^Sec and small for tRNA^Pyl. Another remarkable difference is the inability of tRNA^Sec transcript to bind EF-Tu (8,41) while the tRNA^Pyl transcript, as shown in this work, can interact with this factor. Finally, we notice the puzzling resemblance of tRNA^Pyl with a mitochondrial tRNA^Ser species. Whether this mimicry is fortuitous or has evolutionary significance remains to be investigated.