Biogenesis and growth phase-dependent alteration of 5-methoxycarbonylmethoxyuridine in tRNA anticodons

Abstract

Post-transcriptional modifications at the anticodon first (wobble) position of tRNA play critical roles in precise decoding of genetic codes. 5-carboxymethoxyuridine (cmo⁵U) and its methyl ester derivative 5-methoxycarbonylmethoxyuridine (mcmo⁵U) are modified nucleosides found at the anticodon wobble position in several tRNAs from Gram-negative bacteria. cmo⁵U and mcmo⁵U facilitate non-Watson–Crick base pairing with guanosine and pyrimidines at the third positions of codons, thereby expanding decoding capabilities. By mass spectrometric analyses of individual tRNAs and a shotgun approach of total RNA from Escherichia coli, we identified mcmo⁵U as a major modification in tRNA^Ala1, tRNA^Ser1, tRNA^Pro3 and tRNA^Thr4; by contrast, cmo⁵U was present primarily in tRNA^Leu3 and tRNA^Val1. In addition, we discovered 5-methoxycarbonylmethoxy-2′-O-methyluridine (mcmo⁵Um) as a novel but minor modification in tRNA^Ser1. Terminal methylation frequency of mcmo⁵U in tRNA^Pro3 was low (≈30%) in the early log phase of cell growth, gradually increased as growth proceeded and reached nearly 100% in late log and stationary phases. We identified CmoM (previously known as SmtA), an AdoMet-dependent methyltransferase that methylates cmo⁵U to form mcmo⁵U. A luciferase reporter assay based on a +1 frameshift construct revealed that terminal methylation of mcmo⁵U contributes to the decoding ability of tRNA^Ala1.

INTRODUCTION

RNA molecules are frequently modified post-transcriptionally on their nucleobase and ribose moieties. These modifications carry qualitative information embedded in RNA molecules associated with various biological processes. To date, more than 100 species of modifications have been identified in various RNAs from all domains of life (1), the majority of which are found in tRNAs. tRNA modifications play critical roles in decoding properties and stabilization of tertiary structure (2,3). In particular, a wide variety of modifications occur at the first (wobble) position of the anticodon. These modifications play pivotal roles in modulating codon recognition and ensuring accurate translation of the genetic code (4).

In the classical wobble hypothesis proposed by Crick (5), uridine (U34) at the wobble position pairs with A and G at the third letter of the codon. In the decoding system of Mycoplasma species and mitochondria (6–9), however, U34 can recognize any of the four bases in a family box due to its conformational flexibility, a phenomenon termed ‘four-way wobbling’ (4), although efficiency of this type of wobbling strongly depends on the second and third letters of codons. To restrict decoding capability, tRNAs responsible for two codon sets ending in a purine (NNR) often contain 5-methyl-(2-thio)uridine derivatives [xm⁵(s²)U] at the wobble position (3,4), including 5-carboxymethylaminomethyl-(2-thio)uridine [cmnm⁵(s²)U] and 5-methylaminomethyl-(2-thio)uridine [mnm⁵(s²)U] in bacterial tRNAs, 5-methoxycarbonylmethyl-(2-thio)uridine [mcm⁵(s²)U] and its derivatives in eukaryotic cytoplasmic tRNAs and 5-taurinomethyl-(2-thio)uridine [τm⁵(s²)U] in mitochondrial tRNAs. Due to the conformational rigidity of xm⁵s²U modifications, which are largely fixed in the C3′-endo ribose pucker conformation (10), xm⁵s²U prefers to base-pair with A and G, thus preventing misreading of near-cognate codons ending in pyrimidine (NNY) (2,11). In addition, C5-substituent of xm⁵U modification plays a critical role in stabilizing U-G wobble pairing at the A-site of ribosome (12,13).

By contrast, to expand decoding capacity in most bacterial species, 5-hydroxyuridine derivatives (xo⁵U) are present at the wobble position of tRNAs responsible for recognizing family boxes. 5-carboxymethoxyuridine (cmo⁵U, also called uridine-5-oxy acetic acid) (Figure 1A) is present in tRNAs from Gram-negative bacteria including Escherichia coli and Salmonella enterica (3,4,14). cmo⁵U was first reported as a minor nucleoside at the wobble position of E. coli tRNA^Val1 (15,16). The chemical structure of cmo⁵U was determined in 1970 (17). Subsequently, cmo⁵U was also found in E. coli tRNA^Ser1 (18), tRNA^Ala1 (19) and tRNA^Leu3 (20). In addition, cmo⁵U is present in tRNA^Pro3 from S. enterica (serovar Typhimurium)(21). The presence of the methyl ester derivative of cmo⁵U, 5-methoxycarbonylmethoxyuridine (mcmo⁵U, also called uridine-5-oxy acetic acid methyl ester) (Figure 1A) was predicted by a series of in vitro studies (22–24). In Gram-positive bacteria, 5-methoxyuridine (mo⁵U) is present in tRNA^Thr from Bacillus subtilis (25). Collectively, these findings suggested that cmo⁵U or mcmo⁵U are present at the anticodon wobble position in E. coli tRNA^Val1, tRNA^Ala1, tRNA^Ser1, tRNA^Thr4, tRNA^Pro3 and tRNA^Leu3 (14,20). However, the exact frequency of these modifications in each tRNA has not been determined.

Figure 1.

5-carboxymethoxyuridine (cmo⁵U) and E. coli tRNAs. (A) Chemical structures of 5-carboxymethoxyuridine (cmo⁵U, left), 5-methoxycarbonylmethoxyuridine (mcmo⁵U, center) and 5-methoxycarbonylmethoxy-2′-O-methyluridine (mcmo⁵Um, right). (B) Secondary structures of E. coli tRNA^Ala1 (left) and tRNA^Ser1 (right) with post-transcriptional modifications: 4-thiouridine (s⁴U), 2′-O-methylguanosine (Gm), dihydrouridine (D), 2′-O-methylcytidine (Cm), 2′-O-methyluridine (Um), 5-methoxycarbonylmethoxyuridine (mcmo⁵U), 5-methoxycarbonylmethoxy-2′-O-methyluridine (mcmo⁵Um), 2-methylthio-N⁶-isopentenyladenosine (ms²i⁶A), 7-methylguanosine (m⁷G), 5-methyluridine (m⁵U) and pseudouridine (Ψ). The position numbers of the residues (gray letters) are displayed according to the nucleotide numbering system (64). Pairs of gray triangles indicate the positions of cleavage by RNase T₁ that generate RNA fragments containing the wobble positions. The sequence of E. coli tRNA^Ala1 is the same as that of tRNA^Ala1B (65).

cmo⁵U and mcmo⁵U enable non-Watson–Crick base pairing with guanosines and pyrimidines at the third positions of codons, thereby expanding decoding capability. In vitro studies revealed that cmo⁵U at wobble position allows tRNAs to recognize codons ending in G or U (18,26–29). In addition, tRNAs with cmo⁵U are able to read codons ending in C in the family boxes for Ala, Pro and Val in vivo (30–32). Solution structure of cmo⁵U nucleoside analyzed by NMR prefers to adopt the C2′-endo ribose pucker conformation, providing a mechanistic insight into base pairing between cmo⁵U and pyrimidines (10). Such base pairing at the ribosomal A-site was visualized in a crystal structure of the 30S ribosome in complex with an anticodon stem-loop (ASL) bearing cmo⁵U and its cognate codon (33). cmo⁵U forms an intramolecular hydrogen bond that pre-structures the anticodon loop, enabling cmo⁵U to pair with all four bases at the wobble position in the mRNA. Unexpectedly, cmo⁵U adopts the C3′-endo form and pairs with G at the third letter of codon by the standard Watson–Crick geometry, rather than classical U-G wobble geometry, indicating that the enol form of the uracil base is involved in this base pairing interaction (33).

A pathway consisting of multiple enzymatic reactions has been proposed for biogenesis of cmo⁵U and mcmo⁵U (see Figure 7). Björk and colleagues made a primary contribution to characterizing this pathway (3,14). cmo⁵U biogenesis requires chorismate, an end product of the shikimate pathway, which is involved in biosynthesis of aromatic amino acids and vitamins (34,35). First, U34 is hydroxylated to form 5-hydroxyuridine (ho⁵U) via an unknown pathway. One carbon atom of the 5-carboxymethoxy-group in cmo⁵U is derived from methionine, indicating the involvement of AdoMet methyltransferase in this step (35). Two genes, cmoA and cmoB, both of which have AdoMet binding motifs, are responsible for formation of cmo⁵U (31). Subsequent structural analyses revealed that CmoA contains a novel derivative of AdoMet, S-adenosyl-S-carboxymethyl-l-homocysteine (SCM-SAH or Cx-SAM) (36,37). Kim and Almo's group successfully reconstituted cmo⁵U formation in tRNA in vitro using recombinant CmoA and CmoB in the presence of AdoMet and prephenate (a derivative of chorismate) (37). In this process, CmoA first synthesizes SCM-SAH from AdoMet and prephenate. Next, CmoB employs SCM-SAH as a substrate to transfer its carboxymethyl-group to the 5-hydroxy group of ho⁵U, thereby generating cmo⁵U on tRNA (see Figure 8). In some tRNAs, cmo⁵U is further methylated to form mcmo⁵U. Biochemical studies in S. enterica cell lysate revealed that cmo⁵U in tRNA^Ser1 and tRNA^Ala1 is further methylated by an unidentified AdoMet-dependent methyltransferase, whereas cmo⁵U in tRNA^Val1 is not (22). Initially, the cmo⁵U methyltransferase was predicted to be encoded by supK (23); however, supK was subsequently identified as prfB which encodes release factor 2 (RF2) (38). CmoA was also speculated to use its second function to methylate cmo⁵U to generate mcmo⁵U (31). However, it is now clear that CmoA is responsible for catalyzing SCM-SAH formation (37). Thus, the gene responsible for the cmo⁵U methyltransferase remains to be identified.

Figure 7.

Terminal methylation of mcmo⁵U contributes to GCG decoding. (A) Schematic depiction of the dual-luciferase reporter constructs based on the RF2 recoding system. SD, Shine-Dalgarno sequence. Renilla and firefly luciferases were fused with a linker containing the +1 frameshift signal of the RF2 recoding site. The frameshift target site was replaced with a GCG codon for tRNA^Ala1, a UCG codon for tRNA^Ser1or GG for zero frame (used as a control). (B) Relative pausing activity at the frameshift site with GCG (left), UCG (middle) or zero frame (right) was calculated based on relative Fluc activity normalized to Rluc activity in wild-type, ΔcmoM and ΔcmoB strains. Data are presented as means ± SD (n = 4). *, P < 0.01 versus control (Student's t-test).

Figure 8.

Biosynthesis of mcmo⁵U. In E. coli, U34 of six tRNAs responsible for decoding NCN codons is modified to ho⁵U34 by an unknown pathway, and subsequently modified to cmo⁵U in a reaction catalyzed by CmoM. CmoB uses SCM-SAH as a substrate and transfers its carboxymethyl group to ho⁵U34 on tRNAs. SCM-SAH is synthesized from prephenate and AdoMet catalyzed by CmoA. Four tRNAs (Ala1, Ser1, Pro3 and Thr4) are further modified to mcmo⁵U by CmoM, using AdoMet as a substrate. mcmo⁵U frequency is altered by growth phase only in tRNA^Pro3. In a minor pathway, mcmo⁵U34 in tRNA^Ser1 is further methylated by TrmL to yield mcmo⁵Um34. Alternatively, cmo⁵U34 could be first converted to cmo⁵Um34, then to mcmo⁵Um34.

In this study, we isolated individual tRNAs from E. coli and analyzed the modification status of each tRNA by mass spectrometry. We found that mcmo⁵U was present as a major modification in tRNA^Ala1, tRNA^Ser1, tRNA^Pro3 and tRNA^Thr4, whereas cmo⁵U is primarily present in tRNA^Leu3 and tRNA^Val1. In addition, we discovered 5-methoxycarbonylmethoxy-2′-O-methyluridine (mcmo⁵Um) (Figure 1A) as a novel derivative of mcmo⁵U in tRNA^Ser1. Frequency of terminal methylation of mcmo⁵U in tRNA^Pro3 was dependent on growth phase. Moreover, we identified the cmo⁵U methyltransferase, which we named CmoM, that methylates cmo⁵U to form mcmo⁵U in the presence of AdoMet. This terminal methylation of mcmo⁵U contributes to the decoding ability of tRNA^Ala1.

MATERIALS AND METHODS

Strains and plasmid construction

The E. coli deletion strains ΔcmoB::Km^r, ΔtrmL::Km^r, ΔcmoM (ΔsmtA)::Km^r, and their parental strain BW25113, were obtained from the National BioResource Project (NBRP), National Institute of Genetics, Japan (39). All strains were cultured in LB media at 37°C.

To generate a vector for expression of recombinant CmoM with an N-terminal His6-tag, cmoM was polymerase chain reaction (PCR)-amplified from the BW25113 genome and inserted into the NdeI/NotI site of pET-28a (Novagen) to yield pET-cmoM-N-His6. To complement ΔcmoM, cmoM with its 5′ flanking region (including the promoter) was PCR-amplified and cloned into the low-copy plasmid pMW118 (Nippon Gene) to yield pMW−cmoM (psmtA). Point mutations were introduced in pMW-cmoM by QuikChange^TM site-directed mutagenesis (Agilent Technologies). For dual-luciferase reporters, the fusion gene of Renilla and firefly luciferases was PCR-amplified from pQE-Luc(+1) (40) using primers containing NcoI and XhoI sites, and then inserted into the corresponding site of pBAD/Myc-His (Invitrogen) to yield pBAD-RFLuc. Subsequently, the RF2 recoding site (32,41), including an SD-like sequence and GCG as a test codon, was introduced by PCR into the linker region between the two luciferases to yield pBAD-RFLucGCG. Two variants in which the test codon was replaced by UCG (pBAD-RFLucUCG) and GG (pBAD-RFLucGG) were generated by QuikChange^TM site-directed mutagenesis. All constructs used in this study were verified by Sanger sequencing. The primers used in this study are listed in Supplementary Table S1.

RNA extraction and tRNA isolation

Total RNA from each E. coli strain was extracted by phenol in acidic condition (42). Individual tRNAs were isolated by reciprocal circulating chromatography, as described previously (42,43). The 5′-terminal ethylcarbamate amino-modified DNA probes used in this method are listed in Supplementary Table S1.

Mass spectrometry of tRNA modifications

For RNA fragment analysis, isolated tRNA (1.25 pmol) was digested at 37°C for 1 h in 12.5 μl of a solution containing 20 mM NH₄OAc (pH 5.3) and 125 U RNase T₁. The digested RNA was mixed with 12.5 μl of 0.1 M triethylamine-acetate (pH 7.0), and 10 μl of the digest was analyzed by capillary liquid chromatography (LC) coupled to nano electrospray (ESI)/mass spectrometry (MS) on a linear ion trap-Orbitrap hybrid mass spectrometer (LTQ Orbitrap XL; Thermo Fisher Scientific) as described (42,44).

Total nucleosides were analyzed basically as described (44,45). Isolated tRNA^Ser1 (110 pmol) was digested at 37°C for 3 h in 15 μl of a solution containing 20 mM trimethylamine-HCl (TMA-HCl) (pH 7.0), 0.05 U of nuclease P1 and 0.1 U of BAP. The digested RNA (100 pmol) was adjusted as 50 μl of 90% acetonitrile and subjected to HILIC/ESI-MS using Q Exactive^TM Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Fisher Scientific) equipped with a Dionex UltiMate^TM 3000 LC System (Thermo Fisher Scientific) using a ZIC-cHILIC column (3 μm, 2.1 × 150 mm, Merck Millipore). For ribonucleome analysis, a reverse genetic approach combined with RNA-MS (44), nucleosides of total RNAs obtained from knockout strains were subjected to RPC-ESI/MS using LCQ Advantage ion-trap mass spectrometer (Thermo Fisher Scientific) equipped with an HP1100 liquid chromatography system (Agilent Technologies) using an Inertsil ODS-3 column (2.1 × 250 mm, GL science).

Shotgun analysis of tRNA fragments by RNA-MS

Total RNA of each strain cultured in 2–20 ml of LB medium was extracted with TRIzol^TM (Life Technologies). Then, 50–250 μg of total RNA was dissolved in 800 μl of 3 M NH₄OAc (pH 5.3), mixed with 640 μl (0.8 vol) of isopropanol at room temperature and centrifuged at 15 000 rpm for 10 min to precipitate long RNAs including rRNAs. The supernatant was collected and precipitated with ethanol. Depletion of rRNA was verified by denaturing PAGE. The resultant small RNA fraction (50 ng) was digested for 1 h at 37°C with RNase T₁ (125 U) in 20 mM NH₄OAc (pH 5.3), and the digested RNA (20 ng) was subjected to capillary LC coupled to nanoESI/MS as described above.

Mutation study of CmoM by genetic complementation

The E. coli ΔcmoM strain was transformed with psmtA, its mutant derivatives or an empty vector (pMW118), and the functional importance of the mutations was assessed by monitoring restoration of mcmo⁵U in the transformants. Transformants were cultured overnight at 37°C in 2 ml of LB medium containing 100 μg/ml ampicillin. Total RNA was subjected to the shotgun analysis as described above to detect the anticodon-containing fragment of tRNA^Pro3. Modification frequency was determined by calculating the intensity ratio of mass chromatograms between UmUcmo⁵UGp (m/z 683.567, z = 2) and UmUmcmo⁵UGp (m/z 690.575, z = 2).

Preparation of recombinant protein

E. coli BL21 (DE3) transformed with pET-cmoM-N-His6 was cultured at 37°C to an OD₆₀₀ of ≈0.7, supplemented with 0.1 mM IPTG and cultured at 37°C for an additional 4 h. Cells were harvested and disrupted by sonication in lysis buffer consisting of 50 mM HEPES-KOH (pH 7.5), 100 mM KCl, 10 mM MgCl₂, 1 mM dithiothreitol and 0.2 mM PMSF. Purification of recombinant protein was performed basically as described (42,46) using a HisTrap column (GE Healthcare) with a linear gradient of imidazole (25–500 mM). The purified protein was dialyzed in lysis buffer, added with glycerol to f.c. 30% and stored at –20°C.

In vitro reconstitution of mcmo⁵U

tRNA^Ser1, a substrate for CmoM, was isolated from the ΔcmoM strain. In vitro methylation was performed for 1 h at 37°C in a 10 μl reaction mixture containing 50 mM HEPES-KOH (pH 7.5), 100 mM KCl, 10 mM MgCl₂, 7 mM 2-mercaptoethanol, 1 mM AdoMet, 10 pmol of tRNA^Ser1 and 1 pmol of recombinant CmoM. After the reaction, tRNA was extracted with acidic phenol and chloroform, followed by ethanol precipitation. The tRNAs were digested with RNase T₁ and subjected to LC/MS as described above.

Luciferase reporter assay

The luciferase reporter assay was performed essentially as described (40). E. coli wild-type, ΔcmoM and ΔcmoB strains were transformed with the series of dual-luciferase reporters described above. Each transformant was pre-cultivated at 37°C in 2 ml LB medium containing 100 μg/ml ampicillin overnight. The preculture (1%) was inoculated to 2 ml of LB medium containing 100 μg/ml ampicillin and 100 μM arabinose to induce expression of the reporter. When the OD₆₀₀ reached 0.3–0.7, 1 ml aliguot was centrifuged, and the pelleted cells were resuspended in 200 μl of lysis buffer [10 mM HEPES-KOH (pH 7.5), 100 mM NaCl, 10 mM MgCl₂, 7 mM 2-mercaptoethanol, 400 μg/ml lysozyme]. Cell lysates were prepared by the freeze-thaw method (47) and cleared by centrifugation. The reporter assay was carried out on 5 μl of lysate using GloMax^TM 96 Microplate Luminometer (Promega) with the Dual-Luciferase^TM Reporter Assay System (Promega). The Fluc luminescence signal was normalized against the Rluc luminescence signal.

RESULTS

Modification status of individual tRNAs analyzed by mass spectrometry

Because the methyl ester of mcmo⁵U is easily hydrolyzed during preparation and handling of tRNAs (22), little information is available regarding the presence of mcmo⁵U in individual tRNAs. To determine which tRNA species contain this modification, we employed reciprocal circulating chromatography (RCC) (43) to isolate six tRNA species (tRNA^Ala1, tRNA^Leu3, tRNA^Pro3, tRNA^Ser1, tRNA^Thr4 and tRNA^Val1) (Figure 1B) predicted to contain cmo⁵U or mcmo⁵U from E. coli cells harvested at stationary phase. Each individual tRNA was digested with RNase T₁ and subjected to capillary liquid chromatography (LC)/nanoelectrospray ionization mass spectrometry (MS) to precisely analyze its post-transcriptional modifications (44). The RNA fragment containing the wobble modification was detected as multiply-charged negative ions (Figure 2A and Supplementary Figure S1) and further probed by collision-induced dissociation (CID) to determine the position of each modification (Supplementary Figure S2). Frequencies of the wobble modifications in these six tRNA species were calculated from the peak areas of mass chromatograms (Figure 2B). The wobble modifications of tRNA^Ala1, tRNA^Pro3, tRNA^Ser1 and tRNA^Thr4 consisted of ≈80% mcmo⁵U and ≈20% cmo⁵U. By contrast, over 90% of tRNA^Leu3 and tRNA^Val1 molecules contained cmo⁵U, whereas the remaining 10% contained mcmo⁵U or unmodified U. In tRNA^Ala1, we observed partial modification (≈11%) of 2′-O-methyluridine (Um) at position 32 (Figure 2A), which was confirmed by CID analysis (Supplementary Figure S2A).

Figure 2.

Mass spectrometric analysis of individual tRNAs isolated from stationary-phase E. coli. (A) Mass chromatograms of RNase T₁-digested fragments containing cmo⁵U and its derivatives from tRNA^Ala1 (left panels) and tRNA^Ser1 (right panels) isolated from stationary-phase E. coli. Top and middle panels: extracted-ion chromatograms (XIC) for doubly-charged negative ions of cmo⁵U34-containing fragments (CUUcmo⁵UGp of tRNA^Ala1, MW 1660.18, m/z 829.08; UCmUcmo⁵UGp of tRNA^Ser1, MW 1674.19, m/z 836.09) and mcmo⁵U34-containing fragments (CUUmcmo⁵UGp of tRNA^Ala1, MW 1674.19, m/z 836.09; UCmUmcmo⁵UGp of tRNA^Ser1, MW 1688.21, m/z 843.10), respectively. Bottom left and bottom right panels: XICs for doubly-charged negative ions of Um32/mcmo⁵U34-containing fragment of tRNA^Ala1 (CUmUmcmo⁵UGp, MW 1688.21, m/z 843.10) and mcmo⁵Um34-containing fragment of tRNA^Ser1 (UCmUmcmo⁵UmGp, MW 1702.22, m/z 850.10), respectively. The peaks marked with asterisks represent Um32/cmo⁵U-containing fragment (CUmUcmo⁵UGp, MW 1674.19, m/z 836.09) in tRNA^Ala1 and cmo⁵Um-containing fragment (UCmUcmo⁵UmGp, MW 1688.21, m/z 843.10) in tRNA^Ser1. The RNA fragments containing unmodified C32 are also present in tRNA^Ser1, but they are not described here due to high frequency of Cm32 in tRNA^Ser1 isolated from stationary-phase E. coli. (B) Modification frequencies of cmo⁵U and its derivatives at position 34 in six species of tRNAs isolated from stationary-phase E. coli. Relative composition of each modification was calculated from the peak area ratio of mass chromatograms of RNase T₁-digested fragments containing mcmo⁵Um34 (red), mcmo⁵U34 (green), cmo⁵U34 (blue) or U (gray). (C) Collision-induced dissociation (CID) spectrum of a fragment of E. coli tRNA^Ser1 containing mcmo⁵Um34. The doubly-charged negative ion of the RNase T₁-digested fragment containing mcmo⁵Um34 (m/z 850.10) was used as the precursor ion for CID. The product ions were assigned as described previously (66). Sequences of parent ion and assigned product ions are described on the upper left side of this panel. (D) Nucleoside analysis of E. coli tRNA^Ser1. Top panel: UV chromatogram at 254 nm of the four major nucleosides (C, U, G and A). Second panel: mass chromatograms of the protonated nucleosides (MH⁺) on the base peak chromatogram (BPC). Third to bottom panels: XICs for cmo⁵U (m/z 319.08), mcmo⁵U (m/z 333.09) and mcmo⁵Um (m/z 347.11), respectively. The peak marked with an asterisk represents unspecific peak. (E) CID spectrum of mcmo⁵Um nucleoside. The protonated mcmo⁵Um (MH⁺, m/z 347.11) was used as the precursor ion for CID. The N-glycoside bond cleaved to generate the base-related ion (BH₂⁺) and other product ions are assigned on the chemical structure.

Discovery of mcmo⁵Um as a novel modification

At the wobble position of tRNA^Ser1, we found a novel minor modification with a molecular mass 14 Da larger than that of mcmo⁵U (Figure 2A). CID analysis of this fragment (Figure 2C) revealed a specific product ion (a-B4) lacking the 5-methoxycarbonylmethoxyuracil-base (mcmo⁵U-base), strongly suggesting that the 2′ hydroxyl group of the ribose was methylated. To confirm this result, we analyzed total nucleosides of tRNA^Ser1 and detected a proton-adduct (MH⁺) of this minor nucleoside (m/z 347) (Figure 2D) which we then probed by CID analysis (Figure 2E). We clearly detected a base-related fragment (BH₂⁺) with m/z 201, which also appears in mcmo⁵U, confirming that the ribose portion is methylated. We also verified the absence of this modification in a knockout strain of trmL, which encodes a 2′-O-methyltransferase that targets position 34 (see Figure 4C). Taken together, we conclude that the novel modification found in tRNA^Ser1 is 5-methoxycarbonylmethoxy-2′-O-methyluridine (mcmo⁵Um) (Figure 1A,B). About 4.2% of tRNA^Ser1 molecules contain this modification (Figure 2A,B). Given that 2′-O-methylation stabilizes C3′-endo ribose pucker conformation (48), mcmo⁵Um is likely to allow tRNA^Ser1 to recognize UCG codon more efficiently than mcmo⁵U.

Figure 4.

Reverse genetic approach identified a gene responsible for terminal methylation of mcmo⁵U. (A) Nucleoside analyzes by LC/MS using reverse phase chromatography of total RNA from wild-type (left panels), ΔsmtA (middle panels) and ΔsmtA rescued with psmtA (right panels). Top panels: UV trace at 254 nm. Second and bottom panels: XICs for cmo⁵U (m/z 319) and mcmo⁵U (m/z 333), respectively. Intensity of each peak was normalized to that of cyclic t⁶A (m/z 395). (B) Mass chromatograms of RNase T₁-digested fragments containing cmo⁵U and its derivatives from tRNA^Ala1 isolated from wild-type (left panels) and ΔsmtA (right panels) strains. Top, middle and bottom panels: XICs for doubly-charged negative ions of the cmo⁵U34-containing fragments (CUUcmo⁵UGp, m/z 829.08), the mcmo⁵U34-containing fragments (CUUmcmo⁵UGp, m/z 836.09) and the Um32/mcmo⁵U34-containing fragment (CUmUmcmo⁵UGp, m/z 843.10), respectively. The peak marked with an asterisk represent the Um32/cmo⁵U-containing fragment (CUmUcmo⁵UGp, m/z 836.09). (C) Mass chromatograms of RNase T₁-digested fragments containing cmo⁵U and its derivatives from tRNA^Ser1 isolated from wild-type (left panels), ΔsmtA (middle panels) and ΔtrmL (right panels) strains. Top, middle and bottom panels: XICs for doubly-charged negative ions of the cmo⁵U34-containing fragments (UCmUcmo⁵UGp, m/z 836.09), the mcmo⁵U34-containing fragments (UCmUmcmo⁵UGp, m/z 843.10) and the mcmo⁵Um34-containing fragment (UCmUmcmo⁵UmGp, m/z 850.10), respectively.

Frequency of mcmo⁵U in cellular tRNAs estimated by shotgun approach

Although mcmo⁵U was detected in the four tRNAs with a frequency of ≈80% (Figure 2B), it is possible that the methyl ester of mcmo⁵U was partially hydrolyzed during isolation of individual tRNAs by RCC. To exclude the effects of the inevitable hydrolysis of mcmo⁵U during tRNA isolation, we analyzed the wobble modifications in total tRNAs using a shotgun approach. In order to profile the huge number of RNA fragments derived from all species of tRNAs, we prepared crude E. coli tRNA fractions from different growth phases, digested them with RNase T₁ and subjected the digested material to LC/MS (Figure 3A). Among the four tRNAs bearing mcmo⁵U, we clearly detected an anticodon-containing fragment of tRNA^Pro3 and tRNA^Thr4 containing mcmo⁵U or cmo⁵U, each of which has a unique molecular mass and does not overlap with other fragments. Judging from the peak area of mass chromatograms for UmUmcmo⁵UGp (m/z 690.575) and UmUcmo⁵UGp (m/z 683.567) in tRNA^Pro3 isolated from cells in stationary phase, mcmo⁵U was present in tRNA^Pro3 with a frequency of 98% (Figure 3B). Similarly, tRNA^Thr4 isolated from cells in stationary phase contained nearly 100% mcmo⁵U (Figure 3B). These results suggest that about 20% of the cmo⁵U detected in tRNA^Pro3 and tRNA^Thr4 (Figure 2B) was generated by artificial hydrolysis of mcmo⁵U during tRNA isolation. Given that mcmo⁵U was present in the other two isolated tRNAs (Ala1 and Ser1) at a frequency of ≈80%, nearly 100% mcmo⁵U should be present in all four tRNAs in stationary-phase E. coli.

Figure 3.

Shotgun analysis of tRNA fragments and growth phase-dependent alteration of mcmo⁵U. (A) Shotgun analysis of total tRNA digested by RNase T₁. Top panel: base peak chromatogram (BPC) of RNase T₁-digested total tRNA isolated from early log phase (2 h) E. coli. Second panel: XICs for doubly-charged negative ions of the cmo⁵U34-containing fragment (UmUcmo⁵UGp, m/z 683.57) and the mcmo⁵U34-containing fragment (UmUmcmo⁵UGp, m/z 690.57) from tRNA^Pro3. Bottom panel: XICs for doubly-charged negative ions of the cmo⁵U34-containing fragment (ACmUcmo⁵UGp, m/z 847.60) and the mcmo⁵U34-containing fragment (ACmUmcmo⁵UGp, m/z 854.61) from tRNA^Thr4. (B) mcmo⁵U level is altered by growth phase. tRNA^Pro3 (filled circle and rectangle) or tRNA^Thr4 (open circle and rectangle) were calculated from the peak area ratio of XICs for the fragments containing mcmo⁵U34 and cmo⁵U34 at different cultivation times. Cell growth (green line) was monitored by absorbance at 610 nm.

Growth phase-dependent alteration of mcmo⁵U in tRNA^Pro3

To investigate the effect of growth phase on mcmo⁵U, we prepared total tRNA from E. coli cells harvested at various growth phases. The total tRNA was digested with RNase T₁, and analyzed by LC/MS to monitor the RNA fragment containing mcmo⁵U or cmo⁵U derived from tRNA^Pro3 (Figure 3A). Based on the ratio of the two fragment peaks, we calculated the frequency of mcmo⁵U over the course of E. coli cultivation (Figure 3B and Supplementary Figure S3A). In early log phase (1–2 h after inoculation), mcmo⁵U was present in tRNA^Pro3 with a frequency of ≈30%, whereas cmo⁵U was present in the remaining molecules. The mcmo⁵U frequency gradually increased as growth proceeded, reaching nearly 100% in late log and stationary phase. By contrast, mcmo⁵U frequency in tRNA^Thr4 was consistently high (>99%) in all growth phases (Figure 3B and Supplementary Figure S3B).

The shotgun analysis failed to detect specific fragments bearing mcmo⁵U or cmo⁵U derived from tRNA^Ala1 and tRNA^Ser1, because the molecular masses of these fragments overlapped with those of other fragments. To investigate the modification status of these tRNAs in log phase, we isolated four tRNAs containing mcmo⁵U from E. coli cells harvested in mid-log phase (OD₆₀₀ = 0.4). Mass spectrometric analysis revealed that both tRNA^Ala1 and tRNA^Ser1 contained mcmo⁵U with frequencies of 82% and 84%, respectively (Supplementary Figure S4). Because 20% of mcmo⁵U is hydrolyzed and converted to cmo⁵U during isolation of tRNAs by RCC (Figure 2B), we concluded that both tRNA^Ala1 and tRNA^Ser1 were fully modified with mcmo⁵U in mid-log phase. Meanwhile, tRNA^Pro3 and tRNA^Thr4 contained mcmo⁵U at frequencies of 34% and 80%, respectively (Supplementary Figure S4). These results are consistent with those observed in the shotgun analysis (Figure 3B). Based on these findings, we conclude that mcmo⁵U content in tRNA^Pro3 is dependent on growth phase, whereas the other three tRNAs are fully modified with mcmo⁵U during all phases of cell growth.

Identification of a gene responsible for terminal methylation of mcmo⁵U

To identify the gene encoding the AdoMet-dependent methyltransferase that methylates cmo⁵U to form mcmo⁵U, we conducted a genome-wide screen to identify genes responsible for RNA modifications. The method we employed, ribonucleome analysis, uses a reverse genetic approach combined with RNA-MS (44). This approach has been used to successfully identify many genes responsible for tRNA/rRNA modifications among uncharacterized genes in E. coli (40,46,49–52) and Saccharomyces cerevisiae (53–56). Screening of E. coli knockout strains revealed that mcmo⁵U (m/z 333) was completely absent in the ΔsmtA strain; instead, the level of cmo⁵U (m/z 319) was higher than that in the wild-type (Figure 4A). When smtA was introduced on a plasmid (psmtA) into the ΔsmtA strain, formation of mcmo⁵U (m/z 333) was restored (Figure 4A). These data indicate that SmtA is a methyltransferase that produces mcmo⁵U from cmo⁵U.

Next, we isolated tRNA^Ala1 and tRNA^Ser1 from the ΔsmtA strain, and analyzed their wobble modifications by LC/MS (Figure 4B,C). As expected, mcmo⁵U-containing fragments were completely converted to cmo⁵U-containing fragments in both strains.

As mentioned above, we discovered mcmo⁵Um (Figure 1A) as a minor modification in tRNA^Ser1. We confirmed the absence of mcmo⁵Um in both ΔsmtA and ΔtrmL (Figure 4C). trmL encodes a 2′-O-methyltransferase responsible for 2′-O-methylation of cmnm⁵Um34 of tRNA^Leu4 and Cm34 of tRNA^Leu5 (57). Therefore, we conclude that mcmo⁵Um in tRNA^Ser1 is generated by 2′-O-methylation of mcmo⁵U by TrmL.

In vitro reconstitution of mcmo⁵U mediated by CmoM

To determine whether SmtA actually has methyltransferase activity, we conducted in vitro reconstitution of mcmo⁵U formation by recombinant SmtA. In this experiment, tRNA^Ser1 bearing cmo⁵U was isolated from the ΔsmtA strain and used as a substrate. We clearly detected mcmo⁵U in the tRNA^Ser1 only in the presence of both recombinant SmtA and AdoMet (Figure 5A). CID analysis of the anticodon-containing fragment confirmed the methylation occurred at the wobble position (Figure 5B). This result demonstrated that SmtA is an AdoMet-dependent methyltransferase that transfers a methyl group to cmo⁵U34 of tRNAs to form mcmo⁵U34. Based on the enzymatic activity, we renamed this gene CmoM (cmo⁵U methyltransferase).

Figure 5.

In vitro reconstitution of cmo⁵U methylation by recombinant CmoM. (A) E. coli tRNA^Ser1 bearing cmo⁵U isolated from the ΔcmoM strain was incubated in the presence or absence of recombinant CmoM with or without AdoMet. Top and bottom panels: XICs for doubly-charged negative ions of the cmo⁵U34-containing fragments (UCmUcmo⁵UGp, m/z 836.09) and the mcmo⁵U34-containing fragments (UCmUmcmo⁵UGp, m/z 843.10), respectively. (B) A CID spectrum of RNase T₁-digested fragment of E. coli tRNA^Ser1 incubated in the presence of recombinant CmoM with AdoMet. The doubly-charged negative ion of the mcmo⁵U34-containing fragment (UCmUmcmo⁵UGp, m/z 843.10) was used as the precursor ion for CID. The product ions were assigned according to the literature (66). Sequences of parent ion and assigned product ions are described upper left side in this panel.

Characterization of CmoM

CmoM belongs to the Class I AdoMet-dependent methyltransferase (MTase) family, whose members contain a Rossmann-fold as a characteristic structural motif (Figure 6A) (58–60). The high-resolution crystal structure of E. coli CmoM (SmtA) (PDB ID:4HTF) revealed that CmoM forms a homodimer, and that each subunit contains one molecule each of AdoMet, sulfate, acetate and 2-mercaptoethanol as ligands (Figure 6C and Supplementary Figure S5). Based on this structure, we designed eight mutant cmoM constructs bearing single amino-acid alterations (Figure 6A). The mutated residues, which are conserved in cmoM homologs, are located at the catalytic site where AdoMet is bound (Figure 6C). The ΔcmoM strain was transformed with each of the mutant constructs, and total RNA extracted from each construct was digested with RNase T₁ and subjected to LC/MS to detect the RNA fragment of tRNA^Pro3 bearing mcmo⁵U or cmo⁵U (Figure 6B). The positive and negative controls behaved as expected: mcmo⁵U was fully restored by wild-type cmoM, whereas no mcmo⁵U was formed in cells transfected with an empty vector. Little or no mcmo⁵U was observed in the ΔcmoM strain introduced by the mutant constructs R26A, D73A, W124A, Y150A, R209A, D213A and R246A, whereas mcmo⁵U formation was partially restored by the Y247A mutant. The data indicate that the highly conserved residues in the catalytic center are essential for normal methyltransferase activity.

Figure 6.

Characterization of CmoM. (A) Sequence alignment of CmoM homologs from six γ-proteobacteria, Escherichia coli (NP_415441.1), Salmonella enterica (NP_455477.1), Aeromonas hydrophila (YP_856903.1), Vibrio cholerae (NP_231353.1), Pseudoalteromonas atlantica (ABG40695.1), Pseudomonas aeruginosa (NP_253478.1), and two Actinobacteria, Modestobacter marinus (WP_014741474.1), Streptomyces coelicolor (WP_011028138.1). Identical or similar residues are shaded in black or gray, respectively. Red triangles indicate residues that are essential (filled) or non-essential (open) for generic complementation. Motifs I to VI are conserved in Class I AdoMet-dependent methyltransferases. (B) XICs for doubly-charged negative ions of the cmo⁵U34-containing fragment (black lines, UmUcmo⁵UGp, m/z 683.57) and the mcmo⁵U34-containing fragment (red lines, UmUmcmo⁵UGp, m/z 690.57) from tRNA^Pro3 in the ΔcmoM strain rescued by plasmid-encoded wild-type cmoM or its mutant derivatives. The peak marked with an asterisk represents unspecific peak. (C) Close-up view of the AdoMet-binding site in the crystal structure of E. coli CmoM (PDB ID: 4HTF) containing ligands, AdoMet, acetate and sulfate. Predicted hydrogen bonds between ligands and CmoM are indicated by red dotted lines.

Terminal methylation of mcmo⁵U contributes to the decoding process

To investigate the functional role of the terminal methylation of mcmo⁵U, we constructed dual-luciferase reporters based on the RF2 recoding system (32,41). Renilla luciferase (Rluc) was fused to firefly luciferase (Fluc) by a linker sequence bearing the +1 frameshift signal of the RF2 recoding site (Figure 7A). The original UGA codon at the frameshift site was replaced with GCG or UCG to examine the decoding abilities of tRNA^Ala1 and tRNA^Ser1, respectively. The GCG codon is exclusively deciphered by tRNA^Ala1 (mcmo⁵UGC), whereas the UCG codon is redundantly recognized by tRNA^Ser1(mcmo⁵UGA) and tRNA^Ser2(CGA) (Supplementary Figure S6) (14). We also prepared a control reporter construct lacking the +1 frameshift site (zero frame) (Figure 7A). Each of these reporters was introduced into wild-type (WT), ΔcmoM and ΔcmoB strains. The decoding ability of the test codon at the frameshift site is reflected by the +1 frameshift activity. Because the +1 frameshift activity in this system is promoted by the ‘hungry’ A-site, the ability of the cognate tRNA to decode the test codons can be estimated indirectly. The +1 frameshift activity was calculated from the Fluc signal against the Rluc signal (F/R value). No difference in the frameshift activity of the zero frame construct, used as a negative control, was observed in the three strains (Figure 7B). In the ΔcmoB strain, in which all mcmo⁵U/cmo⁵U should be converted to ho⁵U, we observed clear stimulation of +1 frameshift activity at both GCG and UCG codons (Figure 7B), indicating that the carboxymethyl group of cmo⁵U contributes to decoding of G-ending codons. This result is consistent with the previous reports (32,61). On the other hand, in the ΔcmoM strain, we detected a slight but significant stimulation of frameshift activity at the GCG codon (Figure 7B), but not at UCG. This observation indicates that the terminal methyl group of mcmo⁵U is partially involved in GCG decoding by tRNA^Ala1. The absence of any reduction in UCG decoding in ΔcmoM strain can be explained by the fact that this codon is redundantly deciphered by tRNA^Ser1 and tRNA^Ser2; tRNA^Ser2 might have compensated for the reduced decoding ability of hypomodified tRNA^Ser1 in the ΔcmoM strain.

DISCUSSION

It is difficult to estimate the exact frequency of an RNA modification with unstable chemical structure, such as an ester group, in individual tRNAs, because such modifications are easily hydrolyzed during tRNA isolation (42). Although the presence of mcmo⁵U in cellular tRNAs was reported previously, the exact frequency of this modification in individual tRNAs remained unknown (14,22). Using the shotgun approach, we showed here that mcmo⁵U is present in nearly 100% of tRNA^Pro3 and tRNA^Thr4 molecules isolated from stationary-phase E. coli (Figure 3B). This method can be applied to analysis of other RNA modifications with unstable chemical structures from various sources, including 5-methoxycarbonylmethyluridine (mcm⁵U), wybutosine (yW), cyclic N⁶-threonylcarbamoyladenosine (ct⁶A), glutamylqueuosine (GluQ) and their derivatives. In addition, we used RCC to estimate the fraction of mcmo⁵U hydrolyzed during tRNA isolation. Judging from the mcmo⁵U frequency (≈80%) in the isolated tRNAs, we concluded that ≈20% of mcmo⁵U was converted to cmo⁵U during tRNA isolation (Figure 2B), indicating that all four tRNAs (tRNA^Ala1, tRNA^Ser1, tRNA^Pro3 and tRNA^Thr4) are fully modified with mcmo⁵U in stationary-phase E. coli. The four tRNAs containing mcmo⁵U all have G35 at the second letter of the anticodon, and therefore specify NCN codons, implying that CmoM preferentially recognizes G35. However, because mcmo⁵U was present (albeit at a low frequency, <10%) in tRNA^Leu3 and tRNA^Val1, which do not have G35 (Figure 2B), this residue is not an essential determinant for CmoM.

By applying the shotgun approach to total tRNA, we observed growth phase-dependent alteration of mcmo⁵U in tRNA^Pro3 (Figure 3B). In all phases of cell growth, cmo⁵U was fully incorporated into this tRNA. In early log phase, cmo⁵U of tRNA^Pro3 was partially modified by CmoM to yield mcmo⁵U with a frequency of 30%. As growth proceeded, the level of mcmo⁵U gradually increased, and the level of cmo⁵U concomitantly decreased. At late log and stationary phases, tRNA^Pro3 was fully modified with mcmo⁵U. According to the GEO profile database (ID: 35525121 and 35543521) (62), the steady-state level of cmoM mRNA is temporarily elevated in early log phase, and is expressed at a constant in late log and stationary phases. Therefore, we speculate that hypomodification of mcmo⁵U in tRNA^Pro3 might be due to slow methylation by CmoM that fails to catch up with fast production of tRNA in early log phase; as growth rate decreases, mcmo⁵U accumulates gradually. By contrary, tRNA^Thr4 was fully modified with mcmo⁵U in all growth phases (Figure 3B). In addition, a high level of mcmo⁵U was also found in tRNA^Ala1, tRNA^Ser1 as well as tRNA^Thr4 isolated from log phase E. coli. These results imply that cmo⁵U34 is a better substrate for CmoM in these three tRNAs than in tRNA^Pro3. The growth phase dependency of mcmo⁵U in tRNA^Pro3 might be involved in regulatory decoding of CCN codons in growth phase-specific gene expression. Further studies will be necessary to examine this speculation.

Because ΔcmoM exhibited no obvious growth phenotype (data not shown)(63), the functional role of the terminal methyl group of mcmo⁵U may be limited. To characterize this modification, we employed a reporter assay based on the RF2 recoding system to estimate the ability to decode GCG and UCG codons in the presence or absence of cmoM (Figure 7B). The +1 frameshift activity at a GCG codon increased specifically in the absence of cmoM, indicating that mcmo⁵U facilitates decoding of GCG by tRNA^Ala1. However, we observed no change in UCG decoding in the absence of cmoM, because the UCG codon is redundantly recognized by the other isoacceptor. Similarly, it is difficult to assess the decoding ability of other NCN codons in the absence of cmoM, because NCA codons are recognized exclusively by Watson–Crick base pairs, and other NCN codons are redundantly deciphered by two isoacceptors (14). However, in light of our findings, it is reasonable to assume that the terminal methyl group of mcmo⁵U contributes to NCN decoding in general.

According to the crystal structure of the 30S ribosomal subunit in complex with ASL of tRNA^Val bearing the cmo⁵UAC anticodon and its cognate codons (33), the carboxylate of cmo⁵U forms a hydrogen bond with the N6 amino group of A35 (the second letter of the anticodon). This interaction is one component of the intramolecular hydrogen bonding network that pre-structures the anticodon loop, so that cmo⁵U can pair with all four bases at the third letters of codons. We showed here that mcmo⁵U is primarily present in tRNAs with anticodons containing G35. When cmo⁵U is present in these tRNAs at the ribosome A-site, the carboxylate of cmo⁵U cannot form a hydrogen bond with G35. To make the matter worse, cmo⁵U might be destabilized due to electrostatic repulsion between the carboxylate and the O6 carbonyl oxygen of G35, both of which are negatively charged. The terminal methyl group of mcmo⁵U neutralizes the negative charge of cmo⁵U carboxylate, suggesting that mcmo⁵U is involved in stabilizing the wobble base in the anticodons containing G35.

The crystal structure of CmoM also reveals AdoMet and other ligands bound to the positively-charged surface, which might be involved in tRNA recognition (Supplementary Figure S5). Seven residues essential for mcmo⁵U formation, R26, D73, W124, Y150, R209, D213 and R246, which we identified in this study, reside near the ligand-binding site on the positively charged surface of CmoM (Figure 6C). R26 and D73 play a critical role in positioning AdoMet by forming hydrogen bonds. D73 is a conserved carboxylate in motif II of AdoMet MTase (Figure 6A) (58,59). W124 may participate in AdoMet binding via a stacking interaction with the adenine base of AdoMet. Given that W124 is in close proximity to the methyl group of AdoMet, it might be involved in the interaction with the cmo⁵U base of tRNA and thereby facilitate the cmo⁵U methylation. R26 forms a network of hydrogen bonds with two other essential residues, R209 and Y247, along with a sulfate. R209 extends the network to D213, which interacts with R246. R246 and Y150 to form the binding site for an acetate. The functional roles of the sulfate and the acetate bound to the catalytic site remain unknown, but these ligands may act as mimics for the phosphate group of tRNA bound to CmoM.

Phylogenetic analysis revealed that cmoM is present in γ-proteobacteria, actinobacteria and a few species in other bacterial clades (Figure 6A). Because actinobacteria doesn't have homologs of cmoB, cmo⁵U is not predicted to be present in this organism, indicating that actinobacterial counterpart is not a functional homolog of cmoM. Consistent with this, two essential residues, W124 and Y150, in E. coli CmoM are not conserved in there organisms (Figure 6A). Thus, CmoM and mcmo⁵U are mainly distributed in γ-proteobacteria. Presence of cmoM homologs in Spirochaeta cellobiosiphila (Spirochaeta), Zetaproteobacteria bacterium (ζ-proteobacteria) and Paenibacillus sophorae (Firmicutes) indicates horizontal gene transfer of cmoM from γ-proteobacteria to these species.

In the biogenesis of cmo⁵U and mcmo⁵U (Figure 8), six tRNA species (tRNA^Ala1, tRNA^Leu3, tRNA^Pro3, tRNA^Ser1, tRNA^Thr4 and tRNA^Val1) first undergo hydroxylation at C5 of the uracil base in U34 to yield ho⁵U34; the enzyme and substrate involved in this process are unknown. Next, ho⁵U34 is further modified by CmoB using SCM-SAH as a substrate to yield cmo⁵U34. SCM-SAH is generated from AdoMet and prephenate in a reaction catalyzed by CmoA. For four tRNA species (tRNA^Ala1, tRNA^Pro3, tRNA^Ser1 and tRNA^Thr4), cmo⁵U34 is methylated by CmoM in the presence of AdoMet to yield mcmo⁵U34. Only in tRNA^Ser1, small portion of mcmo⁵U34 is further methylated by TrmL to yield mcmo⁵Um34. Alternatively, cmo⁵U34 could be first converted to cmo⁵Um34, then to mcmo⁵Um34. Growth phase-dependent alteration of mcmo⁵U34 takes place in tRNA^Pro3, implying a possible mechanism of translational control mediated by the regulatory decoding efficiency of CCN codons.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

Grants-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science, Sports, and Culture of Japan [to T.S.]; JSPS Fellowship for Japanese Junior Scientists [to Y.S. and S.K.]. Funding for open access charge: Grants-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Science, Sports, and Culture of Japan [to T.S.].

Conflict of interest statement. None declared.