Taurine-containing Uridine Modifications in tRNA Anticodons Are Required to Decipher Non-universal Genetic Codes in Ascidian Mitochondria

Abstract

Variations in the genetic code are found frequently in mitochondrial decoding systems. Four non-universal genetic codes are employed in ascidian mitochondria: AUA for Met, UGA for Trp, and AGA/AGG(AGR) for Gly. To clarify the decoding mechanism for the non-universal genetic codes, we isolated and analyzed mitochondrial tRNAs for Trp, Met, and Gly from an ascidian, Halocynthia roretzi. Mass spectrometric analysis identified 5-taurinomethyluridine (τm⁵U) at the anticodon wobble positions of tRNA^Met(AUR), tRNA^Trp(UGR), and tRNA^Gly(AGR), suggesting that τm⁵U plays a critical role in the accurate deciphering of all four non-universal codes by preventing the misreading of pyrimidine-ending near-cognate codons (NNY) in their respective family boxes. Acquisition of the wobble modification appears to be a prerequisite for the genetic code alteration.

Introduction

Non-universal codons are known to be used in the mitochondrial (mt)³ genetic code (1–6). DNA sequencing of protein-coding genes in various metazoan mtDNAs followed by comparison of the deduced amino acid sequences with those of other species revealed amino acid assignment deviations for the codons UGA, AUA, AGR, and AAA. The termination codon UGA in the universal genetic code specifies Trp in the mitochondria of all species investigated so far (1, 2). The AUA codon for Ile changes to specify Met in most metazoa except echinoderms, planaria, and coelenterates (2–4). AGR codons are used for Ser in most invertebrates, for Gly in urochordates, and for termination in vertebrates (7–10). The AAA codon specifies Asn in echinoderms (7) and platyhelminths (11).

Our group previously reported that the mitochondrial decoding system of ascidian Halocynthia roretzi employs four non-universal genetic codes: UGA for Trp, AUA for Met and AGR for Gly (Table 1) (9, 12). Sequence analysis of H. roretzi mtDNA identified three tDNAs responsible for these non-universal codons. Each of these tDNAs contains thymidine (T) at the first (wobble) position of their anticodons (Table 1). Because an unmodified U at the wobble position can decode any of the four bases at the third letter of the codon, according to the mitochondrial four-way wobble rule (5, 13), the uridines of these tRNAs at the wobble position must be modified in order to decipher their respective non-universal genetic codes accurately. To elucidate the molecular mechanism of the decoding system, it is necessary to determine the chemical structure of the anticodon wobble base in H. roretzi mt tRNAs. We previously reported 5-carboxymethylaminomethyluridine at the wobble position of mt tRNA^Met(AUR) from H. roretzi (14). However, the nucleotide identification was inaccurate because it was analyzed by conventional two-dimensional TLC, and the modification was determined based on a comparison of its rate of flow value with that of known modified nucleotides (15). In addition, we also reported that H. roretzi mt tRNA^Gly(AGR) has an unidentified modified U at the wobble position (16). We describe here the chemical structures of the wobble bases in mt tRNAs for non-universal genetic codes in the H. roretzi mitochondrial decoding system and discuss evolutionary insights into the role of wobble modification in genetic code alteration.

TABLE 1.

Condon table of the ascidian mitochondrial genetic code

Non-universal genetic codes are shown in boldface on a gray background: AUA for Met, UGA for Trp, and AG(A/G) for Gly. The anticodon sequence of each tRNA gene is shown in parentheses.

EXPERIMENTAL PROCEDURES

Isolation of mt tRNAs

Two hundred grams of the commercially available edible muscle of an ascidian, H. roretzi, was minced, and the tRNA fraction was obtained by phenol extraction as described previously (17). Two thousand A₂₆₀ units of crude tRNA were separated by the DEAE-cellulose column chromatography. Five hundred and eighty A₂₆₀ units of tRNA were isolated from 900 A₂₆₀ units of crude tRNA by removing contaminating polysaccharides with ISOGEN-LS (Wako Chemicals) followed by rinsing with chloroform. For isolating individual mt tRNAs, the following 5′-EC amino-modified DNA probes (Sigma-Aldrich) were designed using “Racess” software (18) and used for reciprocal circulating chromatography (19): 5′-TGGTGCATCAAAGGGACCAACC-CTAACTTA-3′ for tRNA^Gly(AGR), 5′-CCCACAAACTTTTTCATAGTTTACTGTTGT-3′ for tRNA^Met(AUR), 5′-TGGCAGGAAATATATATAATCATTCTATATAACTACT-3′ for tRNA^Trp(UGR), 5′-TGGTACTACGTCCCTAAAGGATCCCATGAT-3′ for tRNA^Gly(GGN), and 5′-AAAGAAATAAACTTTCTGGTTTATATTAACCTATATTACC-3′ for tRNA^Met(AUG). The tDNA sequences were obtained from H. roretzi mtDNA (accession no. AB024528) (12).

The DNA probes were covalently immobilized on NHS-activated Sepharose 4 Fast Flow (GE Healthcare) and packed in the respective tip columns for reciprocal circulating chromatography. Reciprocal circulating chromatography was performed with five-tip columns under the following conditions: the binding of tRNA to the tip columns was performed at 66 °C with binding buffer (1.2 m NaCl, 30 mm HEPES-KOH (pH 7.5), 15 mm EDTA, and 1 mm DTT). Washing was performed at 40 °C with washing buffer (20 mm NaCl, 0.5 mm HEPES-KOH (pH 7.5), 0.25 mm EDTA, and 0.5 mm DTT). The elution of individual tRNAs from tip columns was performed at 68 °C with washing buffer.

Mass Spectrometric Analysis of tRNAs

Three hundred femtomol of each purified tRNA was digested with RNase T₁ and then analyzed by capillary liquid chromatography/nanoelectrospray ionization mass spectrometry as described (20). A linear ion trap-orbitrap hybrid mass spectrometer (LTQ Orbitrap XL, Thermo Fisher Scientific) equipped with a custom-made nanospray ion source and a splitless nano HPLC system (DiNa, KYA Technologies) was employed in this study. Modified bases were assigned by comparing observed m/z values of RNase T₁-digested fragments to calculated ones and allocated based on their sequences by the interpretation of collision-induced dissociation spectra assisted by evolutionary conservation of each modified base in tRNA. Mono- or dinucleotide fragments passed through the LC and were not detected. The percent frequencies of modifications were calculated by the ratio of the mass chromatogram peak heights for RNA fragments containing modified or unmodified bases.

RESULTS

Five species of mt tRNAs, including three tRNAs for non-universal genetic codes and their isoacceptors, were successfully isolated homogeneously by reciprocal circulating chromatography from H. roretzi muscle (supplemental Fig. 1). The yields of tRNA^Gly(AGR), tRNA^Met(AUR), tRNA^Trp(UGR), tRNA^Gly(GGN), and tRNA^Met(AUG) were 2.15, 1.52, 3,54, 1.24, and 1.09 μg, respectively.

The isolated tRNAs were digested by RNase T₁ and subjected to LC/MS to analyze the chemical structures of the modifications. RNA fragments from each tRNA were well separated by LC due to their length, base composition and sequence (Fig. 1A). Singly or multiply charged negative ions of RNA fragments were detected with high mass resolution (30,000). Modified bases in the fragments could be identified from the exact molecular mass of each RNA fragment, which was calculated by deconvoluting the observed m/z values from the mass spectra (supplemental Table 1). Each fragment was further analyzed by an MS/MS experiment using collision-induced dissociation to determine the positions of detected modifications within the tRNA sequence (Fig. 1B). The modifications can be unambiguously positioned in the sequence by the assignment of y- and c-series product ions (21). As shown in the mass chromatograms of the anticodon-containing fragments, the three tRNA species for AGR, AUR, and UGA commonly possess 5-taurinomethyluridine (τm⁵U) at their wobble positions (Figs. 1, A and B, and 2C). Nucleoside analysis was also carried out by LC/MS to confirm the presence of τm⁵U in tRNA^Gly(AGR) (22). The anticodon sequences of tRNA^Gly(AGR), tRNA^Met(AUR), and tRNA^Trp(UGR) were found to be τm⁵UCU, τm⁵UAU, and τm⁵UCA, respectively (Fig. 2A). The wobble base of mt tRNA^Met(AUR), which was previously misidentified as 5-carboxymethylaminomethyluridine (14), has therefore now been corrected. Full sequences of the tRNAs with other modifications are shown in Fig. 2A. In contrast, the anticodons of isoacceptor tRNAs for Gly(GGN) and Met(AUG) were not modified (Fig. 2B).

FIGURE 1.

Mass spectrometric analysis of H. roretzi mt tRNAs. A, base peak chromatograms (BPC) of RNase T₁-digested fragments are shown for tRNA^Gly(AGR) (top), tRNA^Met(AUR) (middle), and tRNA^Trp(UGR) (bottom). Mass chromatograms for the anticodon-containing fragments of each tRNA are also indicated. Sequences and molecular masses of the RNA fragments numbered on base peak chromatograms are listed in supplemental Table 1. The anticodons are underlined. B, collision-induced dissociation spectrum of the anticodon-containing fragment derived from tRNA^Gly(AGR). The triply charged negative ion (m/z 1200.8) was employed as a precursor ion for collision-induced dissociation. Product ions of the c- and y-series (21) are indicated on the spectrum and assigned to the sequence.

FIGURE 2.

Clover leaf structures including modifications of H. roretzi mt tRNAs. Secondary structures of H. roretzi mt tRNAs responsible for the non-universal (A) and universal (B) genetic codes. Abbreviations for modified nucleosides are as follows: m¹G, 1-methylguanosine; m²G, N²-methylguanosine; m¹A, 1-methyladenosine; i⁶A, N⁶-isopentenyladenosine; ms²i⁶A, 2-methylthio-N⁶-isopentenyladenosine; Ψ, pseudouridine; τm⁵U, 5-taurinomethyluridine; and s²U, 2-thiouridine. The numbering system of tRNAs is based on the tRNA database (tRNAdb 2009) (47). Pseudouridines in tRNAs^Gly and tRNA^Met(AUR) were determined previously (14, 16, 22). Variations of the wobble bases are indicated in parenthesis (A). Other partial modifications were also found: N²-methylguanosine 10 (7%), ms²i⁶A37 (56%) and i⁶A37 (44%) in tRNA^Trp(UGR); m¹A9 (88%) in tRNA^Gly(AGR); and 1-methylguanosine 37 (97%) in tRNA^Met(AUG). At position 63 in tRNA^Gly(GGN), 24% of A63 is replaced with G63, probably due to heteroplasmy or a polymorphism in the mtDNA. The unpaired top bases and consecutive GCs are boxed in tRNA^Met(AUG).

The relative frequency of τm⁵U modifications in each tRNA can be calculated based on the signal intensity of the RNA fragments. The wobble bases of tRNA^Gly(AGR) and tRNA^Trp(UGR) were almost completely modified to τm⁵U (Fig. 2A). However, the wobble base was modified to τm⁵U in 76% of tRNA^Met(AUR) molecules and was modified to its 2-thio derivative, 5-taurinomethyl-2-thiouridine (τm⁵s²U), in 20% of tRNA^Met(AUR) molecules (Fig. 2A). The remaining molecules have a 2-thiouridine or an unmodified U as the wobble base (Fig. 2A).

The two methylguanosines 1-methylguanosine and N²-methylguanosine cannot be distinguished from each other by their molecular mass. Because N²-methylguanosine is cleavable by RNase T₁, whereas 1-methylguanosine is not, the methylguanosines at position 6 in tRNA^Gly(AGR) as well as position 10 in tRNA^Met(AUR) and tRNA^Trp(UGR) can be identified as N²-methylguanosine (Fig. 2A and supplemental Table 1). Intriguingly, the 1-methylguanosine at position 37 in tRNA^Met(AUG) was cleaved, leaving 2′,3′-cyclic phosphate at the 3′ terminus of the cleaved fragment (supplemental Table 1). Pseudouridine is a mass-silent modification that cannot be identified only by mass measurement because it has the same molecular mass as U. For tRNA^Gly(AGR) (16), tRNA^Gly(GGN) (16) and tRNA^Met(AUR) (14), positions for pseudouridine were determined biochemically by the post-labeling method (15).

DISCUSSION

These experiments have identified τm⁵(s²)U at the wobble positions of ascidian mt tRNAs. τm⁵(s²)U was first identified in human and bovine mt tRNAs (23, 24), and has subsequently been found in fish.⁴ Thus, τm⁵(s²)U appears to be conserved in vertebrate mt tRNAs (23). τm⁵(s²)U was not found in yeast and Caenorhabditis elegans mt tRNAs (25–27), which instead used cmnm⁵(s²)U as the wobble modification. Due to the chemical similarity between 5-carboxymethylaminomethyluridine and τm⁵U, both nucleotides are thought to be synthesized by a similar biosynthetic pathway, in which glycine is replaced with taurine in the latter case (23, 27). Regarding the evolutionary aspect of xm⁵s²U-type modifications, the identification of τm⁵U in ascidian mt tRNAs in this study provides a key piece of information suggesting the phylogenetic cross-road at which 5-carboxymethylaminomethyluridine was replaced with τm⁵U. Specifically, τm⁵U seems to have emerged at least in protochordate mt tRNAs.

Because the non-universal genetic codes AUA, AGR, and UGA are all purine-ending codons (NNR), it is unsurprising that the three tRNAs for these codons commonly bear τm⁵U at the wobble positions (Fig. 3). In general, tRNAs responsible for decoding NNR codons frequently possess the xm⁵(s²)U-type modifications, including τm⁵(s²)U, at their wobble positions (13, 28), where they prevent the misreading of NNY near-cognate codons (28). Furthermore, it has been shown that the τm⁵U modification in human mt tRNA^Leu(UUR) stabilizes the τm⁵U:G wobble pairing at the ribosomal A site during decoding (29). A structural study of the ribosomal 30S subunit containing codon-anticodon pairing at the A-site provided a mechanistic explanation for the manner in which the 5-taurinomethyl modification enables efficient decoding of G-ending codons (30). In the human mitochondrial decoding system, τm⁵(s²)U is found in five tRNAs for Leu(UUR), Trp, Lys, Glu, and Gln (23, 24, 31, 32). Human homologs of the enzymes GTPBP3 and MTO1 were predicted to be responsible for synthesizing τm⁵U (33, 34). The substrate specificity of GTPBP3/MTO1 homologs in H. roretzi might be altered to enable them to recognize the three mt tRNA species carrying the non-universal codons. Acquisition of the wobble modification by modulating the substrate specificity of the tRNA-modifying enzyme seems to be critical for genetic code alterations in the urochordate lineage.

FIGURE 3.

Decoding system of the non-universal codons in H. roretzi mitochondria. Non-universal AGR and universal GGN codons are decoded by tRNA^Gly with the τm⁵UCU and UCC anticodons, respectively. AUR codons for Met in elongation are decoded by tRNA^Met with the τm⁵(s²)UAU anticodon. The AUG codon for initiation can be recognized by tRNA^Met with the CAU anticodon. UGR codons for Trp are recognized by tRNA^Trp with the τm⁵UCA anticodon. Non-universal codons are boxed.

Another intriguing feature in the ascidian mitochondrial decoding system is the presence of two tRNA^Met species bearing τm⁵UAU and CAU anticodons (Fig. 2AB). Two mt tRNA^Met species are also encoded in the mtDNAs of porifers (35) and bivalves (Mollusca) (36). Except for these species, however, a single tRNA^Met is encoded in animal mtDNAs in general (37). Mammalian mt tRNA^Met has 5-formylcytidine at the wobble position (38), which enables the decoding of AUA in addition to AUG through 5-formylcytidine: A base pairing (39–41). 5-Formylcytidine is a cytidine modification that allows the CAU anticodon to expand its decoding capability and recognize the AUA codon. This single tRNA^Met can be charged with Met and then formylated by mitochondrial transformylase (42, 43). Then, the formylmethionyl(fMet)-tRNA^Met is recognized by mitochondrial initiation factor 2 and used as an initiator tRNA (44, 45). As Met-tRNA^Met is not fully formylated by mitochondrial transformylase in mitochondria, Met-tRNA^Met is also used as an elongator tRNA. In fact, mammalian mt tRNA^Met possesses characteristic structural features conserved in bacterial initiator tRNA^Met, including an unpaired top base pair in the acceptor stem, which is recognized by mitochondrial transformylase, and consecutive GC pairs in the anticodon stem for targeting to the P-site of the ribosome (46). As shown in Fig. 2, A and B, H. roretzi mt tRNA^Met(AUG) has an unpaired top base pair (A1:A72) in the acceptor stem, whereas tRNA^Met(AUR) contains a G1:U72 wobble pair. The consecutive GC pairs are partially conserved in tRNA^Met(AUG). These findings suggest that, in ascidian mitochondria, tRNA^Met(AUG) and tRNA^Met(AUR) act as the initiator and the elongator tRNAs, respectively (Fig. 3). Further studies will be necessary to clarify the functional and physiological significance of this issue.