Identification of determinants for tRNA substrate recognition by Escherichia coli C/U34 2′-O-methyltransferase

Abstract

Post-transcriptional modifications bring chemical diversity to tRNAs, especially at positions 34 and 37 of the anticodon stem-loop (ASL). TrmL is the prokaryotic methyltransferase that catalyzes the transfer of the methyl group from S-adenosyl-L-methionine to the wobble base of tRNA^Leu_CAA and tRNA^Leu_UAA isoacceptors. This Cm34/Um34 modification affects codon-anticodon interactions and is essential for translational fidelity. TrmL-catalyzed 2′-O-methylation requires its homodimerization; however, understanding of the tRNA recognition mechanism by TrmL remains elusive. In the current study, by measuring tRNA methylation by TrmL and performing kinetic analysis of tRNA mutants, we found that TrmL exhibits a fine-tuned tRNA substrate recognition mechanism. Anticodon stem-loop minihelices with an extension of 2 base pairs are the minimal substrate for EcTrmL methylation. A35 is a key residue for TrmL recognition, while A36-A37-A38 are important either via direct interaction with TrmL or due to the necessity for prior isopentenylation (i⁶) at A37. In addition, TrmL only methylates pyrimidines but not purine residues at the wobble position, and the 2′-O-methylation relies on prior N⁶-isopentenyladenosine modification at position 37.

Introduction

Transfer RNAs (tRNAs) act as adaptors by linking nucleotide sequences and amino acids through codon-anticodon pairing.¹ tRNAs function only following post-transcriptional modifications during tRNA maturation.² Modifications to tRNAs are the most extensive of all classes of RNA molecules.³ tRNA post-transcriptional modifications are universally found in bacteria, archaea and eukarya.⁴ More than 90 modifications have been identified in tRNA nucleotides,^5,6 with the majority occurring in the main body and the anticodon stem loop (ASL), especially at positions 34 and 37.^2,7 Modifications in the anticodon region have been shown to be the most effective to restrict the motional dynamics of tRNAs.⁸ Many of these modifications contribute to efficiency and fidelity of protein synthesis, and are further involved in the cellular stress response, the immune response and even human diseases such as cancers, neurological disorders, diabetes, and many mitochondrial disorders.^9-14 A notable example is the Trm9-mediated mcm⁵U modification of the tRNA wobble base U34 in Saccharomyces cerevisiae, which enhances codon-specific translation elongation and modulates expression levels of critical damage response proteins during the cellular DNA damage response.¹⁵

2′-O-methylation of wobble nucleotide 34 in the 2 Escherichia coli isoacceptors tRNA^Leu_CAA (EctRNA^Leu_CAA) and tRNA^Leu_UAA (EctRNA^Leu_UAA) is incompletely characterized.¹⁶ The 2′-O-methylation of pyrimidine nucleotides at position 34 can stabilize the C3′-endo form of the 3′-nucleotidyl unit to confer local conformation rigidity.¹⁷ Deficient 2′-O-methylation in E. coli results in reduced efficiency of codon-wobble base interactions and impacts the recovery of cells from the stationary phase.¹⁶ The methyltransferase introducing the 2′-O-methyl group onto the 2 tRNA isoacceptors has been identified as a SPOUT (SpoU-TrmD) class member, TrmL.¹⁶ The SPOUT superfamily is a class of S-adenosyl-L-methionine (SAM)-dependent methyltransferases (MTases) with a characteristic α/β knot structure.^18,19 TrmL, which consists only of a SPOUT catalytic domain without the usual RNA binding extension domain, is the simplest MTase in the SPOUT family. Interestingly, TrmL can independently catalyze the transfer of the methyl group from the methyl donor SAM to the tRNA^Leu_CAA and tRNA^Leu_UAA isoacceptors without the aid of any other tRNA binding proteins in a direct methyltransfer assay using recombinant TrmL protein.²⁰

The tRNA^Leu_CAA and tRNA^Leu_UAA isoacceptors are the only 2 RNA substrates of TrmL.^16,20 TrmL functions as a homodimer using basic amino acid surface residues of TrmL for tRNA recognition.²⁰ In prokaryotes, 3 other SPOUT tRNA MTases (TrmH, TrmD and TrmJ) exist; these SPOUT MTases have far more tRNA substrates than TrmL. TrmH catalyzes the 2′-O-methylation of tRNA G18.^21-24 Substitution at G18G19 causes deficiencies in methyl transfer activity, and the oxygen 6 atom of G18 is a key recognition element for TrmH.^25-27 TrmD and TrmJ act on the tRNA anticodon loop. TrmD is a tRNA (m¹G37) MTase, ^28-32 containing an extension sequence for tRNA binding, in contrast to TrmL.²⁰ While G36G37 is the substrate determinant of TrmD, TrmD requires only an anticodon stem-loop structure with an extended stem.^28,31,33-37 TrmJ is the tRNA Um32/Cm32 methyltransferase.^38,39 E. coli TrmJ (EcTrmJ) requires full-length tRNA molecules as substrates.³⁹ The elements of tRNA required for recognition and methylation by TrmL have not yet been identified. The mechanism by which TrmL localizes to the anticodon domain of the large L-shaped structure and distinguishes C/U in the wobble position from the other nucleotides is an intriguing question. In this study, we identified the essential nucleotides in tRNA^Leu required for recognition by TrmL. In addition, TrmL activity is sensitive to the modified base at position 37, supporting the previous observation that a defect in synthesis of ms²i⁶A37 at position 37, led to the loss of 2′-O-methylation at C/U 34 at tRNA^Leu_CAA and tRNA^Leu_UAA isoacceptors.¹⁶ The isopentenyl modification at position 37 of transcribed tRNA^Leu is sufficient to recruit TrmL for methylation of nucleotides C34/U34.

Results

i⁶A37 is sufficient to restore 2′-O-methylation at C34 of EctRNA^Leu_CAA transcripts in vitro

It has previously been shown that in vitro tRNA transcripts lacking natural post-transcriptional modifications are not modified by EcTrmL. ²⁰ This suggests that one or more modifications that occur during the tRNA maturation process are essential for guiding EcTrmL methylation activity. Earlier studies showed that the ms²i⁶ modification, in which MiaA catalyzes the transfer of 5-carbon dimethylallyl to the adenosine at position 37 in the first step, was the key modification essential for EcTrmL activity. ¹⁶ We overexpressed and purified EctRNA^Leu_CAA lacking the ms²i⁶ modification in a MiaA gene deletion strain (dMiaA-tRNA^Leu_CAA). Then, the ms²i⁶A37 deficient EctRNA^Leu_CAA was tested in a tRNA methyl transfer assay. As anticipated from previous studies, ¹⁶ the ms²i⁶A37 deficient EctRNA^Leu_CAA was not methylated at all by EcTrmL (Fig. 1A); however, when the unmodified A37 base was modified in vitro by recombinant EcMiaA, it became a substrate of EcTrmL (Fig. 1A). In vitro synthesized EctRNA^Leu_CAA transcript, modified by recombinant MiaA protein (i⁶A-Ts-tRNA^Leu_CAA) was also used in the tRNA methyl transfer assay. The modified transcript (i⁶A-Ts-tRNA^Leu_CAA) could be equally well methylated to modified tRNA^Leu_CAA overexpressed from MiaA gene knockout strain (i⁶A-dMiaA-tRNA^Leu_CAA) by EcTrmL (Fig. 1B), suggesting that i⁶A modification at position 37 is essential for guiding EcTrmL methylation activity, consistent with the previous report that ms²i⁶A at position 37 is a prerequisite for TrmL activity. ¹⁶

Figure 1.

i⁶A37 is essential for TrmL recognition. (A) The capacity of dMiaA-EctRNA^Leu_CAA and i⁶A-dMiaA-EctRNA^Leu_CAA to be methylated by EcTrmL. dTrmL-tRNA^Leu_CAA was purified from a TrmL gene knockout strain and used as a positive control, and dMiaA-EctRNA^Leu_CAA was purified from a MiaA gene knockout strain. (B) The capacity of i⁶A-Ts-EctRNA^Leu_CAA to be methylated by EcTrmL. Ts-EctRNA^Leu_CAA was transcribed by T7 RNA polymerase in vitro and used as a negative control. (C) The binding affinities of i⁶A-Ts-EctRNA^Leu_CAA to EcTrmL as determined by electrophoretic mobility shift assay. The results are the average of 3 independent experiments with standard deviations indicated.

We next analyzed the binding affinities of EctRNA^Leu_CAA with EcTrmL using an electrophoretic mobility shift assay (Fig. 1C). Three tRNAs, including native EctRNA^Leu_CAA extracted from strain JW3581-1 and EctRNA^Leu_CAA transcript with or without the i⁶A37 modification, were incubated with increasing concentrations of EcTrmL (0.75–10 μM). For native EctRNA^Leu_CAA, the shift representing the EcTrmL-tRNA complex (tRNA bound 1) was initially observed at an enzyme concentration of 0.75 μM. A supershift was observed above 3.0 μM EcTrmL, representing a larger molecular mass complex or aggregate (tRNA bound 2). Transcripts of EctRNA^Leu_CAA without any natural post-transcriptional modifications showed no detectable binding to EcTrmL in the range of enzyme concentration tested. The assay with i⁶A37 modified EctRNA^Leu_CAA transcripts showed a decreased binding affinity to EcTrmL when compared with native EctRNA^Leu_CAA. The shift was first observed at 1.5 μM enzyme, and a supershift was observed above 3.0 μM EcTrmL. This assay showed that although the affinity of i⁶A-Ts-tRNA^Leu_CAA to EcTrmL was lower than for WT EctRNA^Leu_CAA, the i⁶A modification had a consequent effect on the binding strength of tRNA transcripts for EcTrmL.

Construction of EctRNA^Leu_CAA mutants

SPOUT MTase TrmL functions as homodimer by using basic surface amino acid residues to recognize its tRNA substrates. Only tRNA^Leu_CAA and tRNA^Leu_UAA isoacceptors are substrates for TrmL, however, nucleotides that determine the “TrmL identity” have not yet been determined. ¹⁶ We designed a systematic approach to replace nucleotides in EctRNA^Leu_CAA. Figure 2A shows the secondary structure of EctRNA^Leu_CAA and EctRNA^Leu_UAA. The two anticodon loops are conserved except for nucleotide 34, which are C in EctRNA^Leu_CAA and U in EctRNA^Leu_UAA. The anticodon stem and variable loop do not show sequence conservation, especially within the variable loop region where EctRNA^Leu_UAA has a longer stem than EctRNA^Leu_CAA. Based on these variations, we performed site-directed mutagenesis in EctRNA^Leu_CAA as follows: deletions within the whole variable arm, which is consistent with most EctRNAs with the exception of EctRNA^Ser and EctRNA^Tyr (designated VLAD); deletion of the U-A base pair at position 46-47F in the variable arm (VLSD); and a guanine nucleotide insertion into the middle of variable loop between positions 47B and 47C (VLI). In the anticodon stem, we substituted the G-C base pair at position 30-40 to A-U (ASL30), and the adjacent A-U at position 31-39 to G-C (ASL31). In the anticodon loop, we substituted each nucleotide from positions 32 to 38 by changing purines to pyrimidines or vice versa, resulting in EctRNA^Leu_CAA-U32C/A/G, EctRNA^Leu_CAA-U33A, EctRNA^Leu_CAA-A35G/U/C, EctRNA^Leu_CAA-A36C, EctRNA^Leu_CAA-A37C, and EctRNA^Leu_CAA-A38C (Fig. 2B). Most of these tRNAs were overexpressed and purified by phenol extraction followed by DEAE-Sepharose chromatography and C18 reversed-phase HPLC as described below (see Materials and Methods). The tRNA mutants which could not be overexpressed in the TrmL gene deletion strain were transcribed by T7 RNA polymerase and modified by EcMiaA at position 37 in vitro. As a control, wild type EctRNA^Leu_CAA was purified in the TrmL gene deletion strain (dTrmL-EctRNA^Leu_CAA) or transcribed by T7 RNA polymerase in vitro (Ts-EctRNA^Leu_CAA).

Figure 2.

Overview of the EctRNA^Leu_CAA mutations examined in this study. (A) Secondary structures of EctRNA^Leu_CAA and EctRNA^Leu_UAA. (B) Cloverleaf structure summarizing the EctRNA^Leu_CAA mutations. The variants were created using site-directed mutagenesis. Cyan shows the variable region, pink shows the anticodon region, and bright yellow highlights the mutations. The arrows identify the locations of mutations; ψ, pseudouridine; D, dihydrouridine; Cm, 2′-O-methylcytidine; Um, 2′-O-methyluridine;ms²i⁶A, 2-methylthio-N⁶-isopentenyladenosine; –, hydrogen bond between the standard bases; ·, hydrogen bond between the wobble base; △, deletion; ▽, insertion.

The variable arm of tRNA is not a recognition element for EcTrmL

EctRNA^Leus belong to the small family of class II tRNAs which have more than 10 nucleotides in the variable arm. The long variable arm protrudes on one side of the tRNA and may affect binding or recognition by modification enzymes, or give information on how EcTrmL approaches the wobble position of EctRNA. To address this possibility, we examined the effect of drastic mutations in the variable arm and loop of EctRNA^Leu_CAA (mutants VLSD, VLAD and VLI, Fig. 2B). The methyl transfer assay showed that the 3 tRNA mutants were all methylated as well as the WT EctRNA^Leu_CAA (dTrmL-EctRNA^Leu_CAA) (Fig. 3A). The steady-state kinetic constants shown in Table 1 indicated that deletions of one base pair in the variable arm (VLSD), deletion of the entire variable arm (VLAD) or nucleotide insertions (VLI) had no significant effect on the methylation parameters, indicating that EcTrmL is not sensitive to either the sequence or the length of tRNA variable loop.

Figure 3.

Essential recognition residues within the variable loop and anticodon stem loop of tRNA. (A) The capacity of EctRNA^Leu_CAA with mutations in the variable loop to be methylated by EcTrmL. VLSD, deletion of the U-A base pair at position 46-47F in the variable arm; VLI, a guanine nucleotide insertion into the middle of variable loop between positions 47B and 47C; VLAD, deletion of the whole variable arm. (B) The capacity of EctRNA^Leu_CAA with mutations in the anticodon stem to be methylated by EcTrmL. ASL30, substitution of the conserved G-C base pair at position 30-40 to A-U in the anticodon stem; ASL31, substitution of A-U at position 31-39 to G-C in the anticodon stem. (C-F) The capacity of EctRNA^Leu_CAA with mutations in the anticodon loop to be methylated by EcTrmL. All mutants were performed based on the sequence of EctRNA^Leu_CAA. dTrmL-U32C/A/G, -U33A, -A35U, -A37C and -A38C represent mutations at relevant positions overexpressed from a TrmL gene knockout strain. i⁶A-Ts-A35G/C represent mutations at position 35 which were made by in vitro transcription, and followed with i⁶A37 modification by MiaA in vitro. The negative control of the capacity of transcribed Ts-EctRNA^Leu_CAA to be methylated by EcTrmL was not shown in order to make the panels clearer. The results are the average of 3 independent experiments with standard deviations indicated.

Table 1.

Kinetic parameters of EcTrmL methylation of various tRNA substrates

tRNAs	apparent Km (μM)	apparent kcat (min−1)	kcat/Km
tRNALeuCAAa	3.39 ± 0.33	0.44 ± 0.07	0.13
tRNALeuCAA-C34A	ND	ND	ND
tRNALeuCAA-C34G	ND	ND	ND
tRNALeuUAA-U34A	ND	ND	ND
tRNALeuUAA-U34G	ND	ND	ND
tRNALeuCAA -VLI	3.12 ± 0.30	0.71 ± 0.10	0.23
tRNALeuCAA -VLSD	2.74 ± 0.39	0.75 ± 0.04	0.27
tRNALeuCAA -VLAD	3.35 ± 0.31	0.63 ± 0.09	0.19
tRNALeuCAA -ASL30	4.07 ± 0.26	0.85 ± 0.11	0.21
tRNALeuCAA -ASL31	3.39 ± 0.18	1.18 ± 0.02	0.35
tRNALeuCAA -U32C	3.54±0.19	0.99±0.14	0.28
tRNALeuCAA -U32A	5.37±0.27	1.20±0.16	0.22
tRNALeuCAA -U32G	ND	ND	ND
tRNALeuCAA -U33A	4.46 ± 0.26	0.40 ± 0.05	0.09
tRNALeuCAA -A35U	ND	ND	ND
tRNALeuCAA -A37C	ND	ND	ND
tRNALeuCAA -A38C	ND	ND	ND
tRNAPheGAA-G34C	1.82 ± 0.22	0.58 ± 0.05	0.32
tRNAPheGAA-G34U	2.26 ± 0.15	0.50 ± 0.12	0.22

^aValues from Liu et al.²²

The results are the average of three independent experiments with standard deviations indicated.

The anticodon stem of tRNA does not contain recognition elements for EcTrmL

Sequence comparison of the 2 EctRNA^Leu isoacceptors showed that the fourth and fifth base pairs in the anticodon stem of the 2 EctRNA^Leu isoacceptors are conserved (Fig. 2A). We performed G30A:C40U (ASL30) and A31G:U39C substitutions (ASL31) in EctRNA^Leu_CAA (Fig. 2B). The efficiency of methylation of the 2 mutants by TrmL was similar to WT EctRNA^Leu_CAA (Fig. 3B) and the steady-state kinetic constants were similar to the native enzyme parameters suggesting that the 2 conserved base pairs in the tRNA anticodon stem are not important for EcTrmL recognition.

Essential recognition residues within the anticodon loop

Except for the wobble nucleotide that is the target of TrmL methylation, the other nucleotides of the anticodon loop are strictly conserved in EctRNA^Leu_CAA and EctRNA^Leu_UAA (Fig. 2A). Ten mutations were performed in EctRNA^Leu_CAA. Three mutations at position 32, resulting in EctRNA^Leu_CAA-U32C/A/G (Fig. 2B), were overexpressed in the E. coli JW3581-1 strain. The methylation assays of dTrmL-tRNA^Leu_CAA-U32C/A showed results similar to those for dTrmL-tRNA^Leu_CAA, whereas dTrmL-tRNA^Leu_CAA-U32G was a poor substrate for EcTrmL (Fig. 3C). The apparent K_m values and k_cat values for EctRNA^Leu_CAA-U32C and EctRNA^Leu_CAA-U32A were similar to those for WT EctRNA^Leu_CAA (Table 1). These results suggest that residue U32 is not crucial for EcTrmL recognition. U33 is at the position adjacent to the EcTrmL catalytic site, so we performed a EctRNA^Leu_CAA-U33A mutation (Fig. 2B) and overexpressed this tRNA in E. coli JW3581-1. The efficiency of methylation of dTrmL-tRNA^Leu_CAA-U33A by TrmL was similar to that of WT EctRNA^Leu_CAA (Fig. 3D). The apparent K_m value of EcTrmL for EctRNA^Leu_CAA-U33A (4.46 μM) is approximately the same as for WT EctRNA^Leu_CAA (4.12 μM) (Table 1), and the apparent k_cat values are also similar (0.40 s⁻¹ and 0.55 s⁻¹, respectively) (Table 1). These results suggest that residue U33 is not a crucial residue for EcTrmL recognition either. A35 is the other residue adjacent to the wobble base position, and previous work showed that TrmL catalyzed modification was significantly reduced on introduction of an EctRNA^Leu_CAA-A35U mutation into the test system in vivo.¹⁶ Therefore, 3 mutations were performed at position 35, resulting in EctRNA^Leu_CAA-A35U/G/C (Fig. 2B). The total levels of isopentenylation by EcMiaA in vitro for transcribed tRNA^Leu_CAA-A35U, -A35C and -A35G are the same (Fig. S1). Only EctRNA^Leu_CAA-A35U could be overexpressed in E. coli strain JW3581-1 (dTrmL-tRNA^Leu_CAA-A35U). EctRNA^Leu_CAA-A35G and -A35C were transcribed by T7 RNA polymerase and modified by EcMiaA at position 37 in vitro (i⁶A-Ts-tRNA^Leu_CAA-A35G/C). None of these 3 mutants could be recognized by TrmL, suggesting that A35 is the key residue for EcTrmL recognition (Fig. 3E). Previous work showed that the A36-A37-A38 motif in the ASL of tRNA was strictly recognized by EcMiaA0⁴⁰ so we checked whether these 3 residues are also recognized by EcTrmL. Mutations at positions 37 and 38 resulting in EctRNA^Leu_CAA-A37C and -A38C (Fig. 2B), could be overexpressed in E. coli JW3581-1 (dTrmL-tRNA^Leu_CAA-A37C and -A38C), but neither of these 2 mutants could be methylated by EcTrmL (Fig. 3F). A mutant at position 36, resulting in EctRNA^Leu_CAA-A36C (Fig. 2B), however, could not be overexpressed in E. coli JW3581-1, and a DMAPP transfer assay showed that transcribed EctRNA^Leu_CAA-A36C could not be recognized by EcMiaA in vitro. Therefore, we could not gain i⁶A-Ts-tRNA^Leu_CAA-A36C for further methyl transfer assay.

Alteration of the main tRNA body does not affect recognition by EcTrmL

EctRNA^Phe_GAA belongs to the class I tRNAs, which have a short variable loop of 4-5 nucleotides. ⁴¹ Sequence alignment of EctRNA^Phe_GAA, EctRNA^Leu_CAA and EctRNA^Leu_UAA showed that the sequences in the anticodon loop were almost identical between the 3 tRNAs, with the only difference at the wobble position, which is G, C, and U in EctRNA^Phe_GAA, EctRNA^Leu_CAA and EctRNA^Leu_UAA, respectively. However, the sequences of the 3 anticodon stems are completely different. The main tRNA body, including the amino acid acceptor stem, the variable loop region, the D loop region and the TψC loop region of EctRNA^Phe_GAA also differs from that of EctRNA^Leu_CAA or EctRNA^Leu_UAA. EctRNA^Phe_GAA has a conserved A36-A37-A38 motif and possesses an i⁶A modification introduced by MiaA at position 37. ⁴⁰ Accordingly, we mutated G34 of EctRNA^Phe_GAA to C or U in order to mimic the anticodon loops of EctRNA^Leu_CAA and EcRNA^Leu_UAA, respectively (Fig. 4A). Methylation assays of EctRNA^Phe_GAA-G34C and EctRNA^Phe_GAA-G34U showed results similar to those for EctRNA^Leu_CAA. These findings indicate that the sequences of the anticodon stem and variable loop are not essential for TrmL recognition (Fig. 4B). In the methylation assay, the apparent K_m value for EctRNA^Phe_GAA-G34C was 1.82 μM and the apparent k_cat was 0.58 s⁻¹; for EctRNA^Phe_GAA-G34U the values were 2.26 μM and 0.50 s⁻¹, respectively. These values are similar to those of WT EctRNA^Leu_CAA. Sequence alignment of EctRNA^Ser_CGA and EctRNA^Leu_CAA showed that positions 32 and 35 of the anticodon loop were different with each other, while other positions including 33, 34, 36, 37 and 38 were the same. And the main tRNA body, including the amino acid acceptor stem, the variable loop region, the D loop region and the TψC loop region of EctRNA^Ser_CGA also differs from that of EctRNA^Leu_CAA (Fig. 4C). We mutated G35 of EctRNA^Ser_CGA to A in order to mimic the anticodon loop of EctRNA^Leu_CAA (Fig. 4C). The total levels of isopentenylation of the plateau by EcMiaA for transcribed tRNA^Ser_CGA and tRNA^Ser_CGA-G35A are the same (Fig. S2). Further methylation assay showed that i⁶A37 modified EctRNA^Ser_CGA-G35A but not EctRNA^Ser_CGA, was the substrate of EcTrmL (Fig. 4D). Taken together, these results suggest that the identity of anticodon loop only, but no other parts of the tRNA, is crucial for tRNA recognition by EcTrmL.

Figure 4.

Alteration of the main tRNA body does not affect recognition by EcTrmL. (A) Secondary structure of EctRNA^Phe_GAA highlighting the mutations at position 34 (EctRNA^Phe_GAA-G34C and EctRNA^Phe_GAA-G34U). (B) The capacity of dTrmL-tRNA^Phe_GAA-G34C and dTrmL-tRNA^Phe_GAA-G34U to be methylated by EcTrmL. (C) Secondary structure of EctRNA^Ser_CGA highlighting the mutation at position 35 (EctRNA^Ser_CGA-G35A). (D) The capacity of i⁶A-Ts-tRNA^Ser_CGA and i⁶A-Ts-tRNA^Ser_CGA-G35A to be methylated by EcTrmL. The results are the average of 3 independent experiments with standard deviations indicated.

Anticodon stem minihelices are substrates of EcTrmL

In general, tRNA MTases can be divided into 2 groups based on their sensitivity to structural elements in the tRNA mole-cule. ⁴² The first group can only modify nucleosides using well-folded, full-length tRNA molecules as substrates, while the second group can efficiently modify truncated tRNA fragments.⁴² We examined whether the L-shaped structure of tRNA or the anticodon stem loop alone was sufficient for EcTrmL recognition. We constructed several EctRNA minisubstrates derived from EctRNA^Leu_CAA (Fig. 5A). The minisubstrates were synthesized by in vitro transcription; they all lacked tRNA specific domains including the D-loop, the variable loop and the TψC regions, and only retained the tRNA anticodon stem-loop domain (Fig. 5A). For example, ASL-5 represents the tRNA anticodon stem-loop domain with 5 base pairs in the stem, and ASL-7 retained the 5 base pairs of the tRNA anticodon stem directly fused to 2 base pairs of the acceptor stem, and so on. The Acc-ASL mutant retained the anticodon stem-loop fused to the entire acceptor stem with a few nucleotides linking the 2 domains, which originally existed in the D-loop region and the TψC region. All the transcripts were modified by recombinant EcMiaA at position 37 before they were tested as substrates in the tRNA methyl transfer assay. The effective concentrations of these transcribed minisubstrates or minihelices were determined by tRNA isopentenylation assay through detection of ³H labeled DMAPP to ensure that these i⁶A37 modified tRNA minihelices were used in the same effective concentration in later methyl transfer assays (see Materials and Methods).

Figure 5.

Truncated tRNA minihelices could be recognized by EcTrmL. (A) Secondary structure of the minihelices of the ASL of EctRNA^Leu_CAA (ASL5/7/9/11 and Acc-ASL). (B) and (C) show the capacity of the minihelices to be methylated by EcTrmL. The results are the average of 3 independent experiments with standard deviations indicated.

The methylation capacities of the i⁶A dimethylallylated minisubstrates revealed that i⁶A-ASL-5 was a poor substrate, but the addition of extra base pairs in i⁶A-ASL-7, i⁶A-ASL-9, i⁶A-ASL-11 and i⁶A-Acc-ASL improved the methylation levels (Fig. 5B,C); these minihelices were methylated to levels comparable to i⁶A modified EctRNA^Leu_CAA transcript. Together, these results showed that an anticodon stem-loop minihelix extension of 2 base pairs is the minimal substrate requirement for EcTrmL methylation.

Only pyrimidines at the wobble position are recognized by EcTrmL

EcTrmL could independently catalyze 2′-O-methylation at C34 and U34 in the EctRNA^Leu isoacceptors. Previous in vivo study showed that TrmL could not methylate tRNA^Leu_CAA-C34A in E. coli.¹⁶ To determine whether EcTrmL could recognize a purine nucleotide at position 34, we mutated C34 or U34 in the tRNA^Leu isoacceptors to A34 or G34, resulting in EctRNA^Leu_CAA-C34A/G or EctRNA^Leu_UAA-U34A/G, respectively (Fig. 2A). Mutants C34A and U34A were purified following in vivo overexpression as described above, however, mutants C34G and U34G could not be overexpressed and were transcribed by T7 RNA polymerase before being modified by EcMiaA at position 37 in vitro. The methyl transfer assay by recombinant EcTrmL showed no detectable activity when using EctRNA^Leu_CAA or EctRNA^Leu_UAA with purine substitutions at position 34, whether the i⁶A37 modification was present or not (Fig. 6A,B). This finding shows that only pyrimidine nucleotides at position 34 are substrates of EcTrmL.

Figure 6.

Only pyrimidines at the wobble position are recognized by TrmL. (A) and (B) The capacity of dTrmL-EctRNA^Leu_CAA-C34A, dTrmL-EctRNA^Leu_UAA-U34A, i⁶A-Ts-EctRNA^Leu_CAA-C34G and i⁶A-Ts-EctRNA^Leu_UAA-U34G to be methylated by EcTrmL. dTrmL-EctRNA^Leu_CAA-C34A and dTrmL-EctRNA^Leu_UAA-U34A were overexpressed from a TrmL gene knockout strain, i⁶A-Ts-EctRNA^Leu_CAA-C34G and i⁶A-Ts-EctRNA^Leu_UAA-U34G were made by in vitro transcription by T7 RNA polymerase followed with i⁶A37 modification by MiaA in vitro. The results are the average of 3 independent experiments with standard deviations indicated.

Discussion

The i⁶A37 modification has been found to be widespread and conserved. ⁴³ Defects in this modification were recently found in human pathogenic mutations.⁴⁴ In bacteria, the ms²i⁶A37 modification is carried out in 2 successive steps by MiaA and MiaB.⁴⁵ It has been shown that the ms²i⁶A37 modification occurs earlier than other modifications in the anticodon stem-loop and a tRNA helical stem-loop containing an A36-A37-A38 motif are determinants for MiaA recognition.⁴⁰ In E. coli, both tRNA^Leu_CAA and tRNA^Leu_UAA isoacceptors contain the ms²i⁶A37 modification. Interestingly, in vitro transcribed tRNA^Leu_CAA and tRNA^Leu_UAA without modifications are not substrates of TrmL methylase, suggesting a sequential process starting with the ms²i⁶A37 modification. ²⁰ We show here that synthetic transcripts of tRNA^Leu_CAA and tRNA^Leu_UAA with the first i⁶A37 modification, catalyzed by recombinant MiaA, are substrates of TrmL. Similarly, tRNA^Leu purified from an E. coli MiaA-knockout strain was an effective substrate of TrmL after in vitro introduction of i⁶A37 modification by recombinant MiaA. These data demonstrate that only the i⁶A37 modification is a strict prerequisite for TrmL-catalyzed methylation and these results are consistent with a previous study showing that Yibk (TrmL) is inactive with tRNA substrates deprived of the fully modified ms²i⁶A37 base.¹⁶ Altogether, the results confirm that the modifications of the wobble base and the purine at position 37 occur in chronological order in vivo.

tRNA^Leu_CAA and tRNA^Leu_UAA are the only 2 RNA substrates of TrmL. The mechanism by which TrmL distinguishes its targets from all other tRNAs, which all have similar L-shaped tertiary structure, is unclear. Our data indicated that only pyrimidine nucleotides at the wobble position could be methylated by TrmL, which is consistent with previous results showing that TrmL could not methylate EctRNA^Leu_CAA-C34A.¹⁶ EctRNA^Phe_GAA is a type I tRNA and naturally does not contain a 2′-O-methylation modification at position 34. However, with the exception of position 34, which is a G, the 6 other nucleotides of the anticodon loop of EctRNA^Phe_GAA are identical to EctRNA^Leu_CAA and EctRNA^Leu_UAA, although the remaining components of the tRNA body are different. Surprisingly, EctRNA^Phe_GAA-G34C and EctRNA^Phe_GAA-G34U mutants were methylated by EcTrmL as efficiently as EctRNA^Leu_CAA.

Our results also showed that TrmL could accommodate many changes in the tRNA^Leu_CAA structure. For instance, changes in the variable loop of EctRNA^Leu_CAA do not affect TrmL activity. We further showed that several residues of the anticodon loop, but not the anticodon stem, were crucial for TrmL activity. Mutations of U32 and U33 preserved EcTrmL methylation activity. The nucleotide at position 32 in E. coli tRNAs is generally a semi-conserved cytidine or uridine residue,⁴⁶ and U33 is invariant in all tRNAs.^40,47 Therefore, that TrmL does not use these conserved nucleotides at positions 32 and 33 as recognition determinants seems reasonable. A35 functions as an identity element for EcTrmL recognition, which is consistent with the results that TrmL could not methylate tRNA^Leu_CAA-A35U in E. coli.¹⁶ The tRNA mutant A36C could not be overexpressed in E. coli JW3581-1 in vivo, neither its transcript could be recognized by EcMiaA in vitro. The tRNA mutants A37C and A38C resulted in loss of the i⁶A37 modification and therefore of recognition by TrmL. Because A36-A37-A38 motif is strictly recognized by MiaA, any change of it could result in loss of the i⁶A37 modification. Therefore, the reason for these mutants could not be recognized by TrmL may be the loss of i⁶A37 modification. However, it is also probable that some nucleotide(s) in the A36-A37-A38 motif is/are the recognition determinant(s) for TrmL through direct interactions. Mutant EctRNA^Ser_CGA-G35A could not be overexpressed in the TrmL gene deletion strain. So we transcribed it by T7 RNA polymerase and modified by EcMiaA at position 37 in vitro. Our results show that EctRNA^Ser_CGA-G35A turns EctRNA^Ser_CGA into a substrate for TrmL. Although this mutant could be recognized by TrmL, the efficiency of EcTrmL to methylate EctRNA^Ser_CGA-G35A is rather low compared with i⁶A37 modified EctRNA^Leu_CAA transcripts. Because main body of tRNA, including anticodon stem, amino acid acceptor stem, the variable loop region, the D loop region and the TψC loop region of EctRNA^Ser_CGA differ from that of EctRNA^Leu_CAA, the reason for EctRNA^Ser_CGA-G35A being a relatively poor substrate of TrmL might be due to slightly local conformation change and/or some steric hindrance. Taken together, these results suggest that the tRNA anticodon loop, but not any other part of the tRNA, as well as the i⁶A37 modification, are crucial for TrmL recognition. In a last assay to further test the critical role of the anticodon loop of tRNA, we showed that EcTrmL could modify tRNA minihelices exhibiting the anticodon stem-loop structure (ASL) and the i⁶A37 modification catalyzed by EcMiaA.

In summary, these results suggest that EcTrmL recognizes its substrate tRNAs based on (i) the stem-loop structure of ASL extended by 2 base pairs, (ii) the correct identity elements of the anticodon loop, (iii) a pyrimidine at the wobble position, and (iv) a preexisting i⁶A37 modification (Fig. 7). Thus, it seems that TrmL may stretch out many hands and arms, to identify those recognition determinants at the same time (Fig. 7). This recognition pattern is very different from the other 3 SPOUT tRNA MTases (TrmH, TrmD and TrmJ) in prokaryotes. TrmH is one of the methyltransferases which catalyzes 2′-O-methylation at position 18 of tRNA,^21-24 and TrmH exhibits different recognition of its tRNA substrates than TrmL. TrmHs can be divided into 2 subclasses; one can recognize all tRNA species, whereas the other, including EcTrmH, can only modify a subset of tRNA species.²⁷ Early study suggested that G18G19 and the D-arm structure of the tRNA are essential requirements for recognition by Thermus thermophilus TrmH, ²¹ but further work showed that T. thermophilus TrmH recognized G18 with some flexibility and the oxygen 6 of G18 was the crucial determinant for TrmH recognition.²⁵ TrmD is a tRNA (m¹G37) MTase.^28-32 It belongs to the first group of tRNA modifying enzymes, which don't need the L-shaped structure of tRNA for substrate recognition.⁴² TrmD can recognize and methylate short RNA structures such as the stem-loop structure of ASL, similar behavior to TrmL.^{28,31,33,36,48} G36-G37 is the known recognition determinant for TrmD.^34,35,37 EcTrmJ catalyzes 2′-O-methylation at position 32 of tRNA.³⁸ EcTrmJ needs the full-length tRNA and correct identity elements within the D stem and loop region, which is totally different from TrmL.³⁹

Figure 7.

A scheme to summarize the determinant elements of tRNA recognized by TrmL. Hands and arms show that TrmL is recognizing the ASL including the stem-loop structure, the correct identity elements, a pyrimidine at the wobble position and a preexisting i⁶A37 modification.

Our data showed that 2′-O-methylation at the wobble position by TrmL requires a previous i⁶A37 modification. It is, however, unclear how the i⁶A37 modification influences tRNA recognition by TrmL. A direct interaction could occur between TrmL and the bulky isopentenyl group added by MiaA at A37, which could act as a recognition determinant. Alternatively, the isopentenyl group could change the structure of the anticodon loop region, which would affect the interaction with TrmL. Indeed, it was reported that the i⁶A37 modification changes the structural conformation of E. coli tRNA^Phe_GAA ASL, as observed in NMR structures.⁴⁹ The unmodified ASL molecule adopts a stem-loop conformation composed of 7 base-pairs and a compact 3 nucleotide loop, which is obviously different from the classical loop observed in fully modified tRNA which contains a U-turn motif at position 33.⁴⁹ Therefore, the conformational change of ASL resulting from the i⁶A37 modification might also be important for recognition by TrmL.

Methylation is one of the most common and ubiquitous RNA modifications, and it is present in many different types of cellular RNAs, including rRNA, mRNA, tRNA, microRNA and other small RNAs. Recently, RNA demethylases were identified that were found to be involved in fundamental regulatory roles in many important life processes, including germline development, cellular signaling, and circadian rhythm control.^50,51 Intriguing questions regarding how these RNA MTases or demethylases recognize or bind to their RNA substrates remain, since unlike DNA binding domains that share common structural motifs like zinc fingers, leucine zippers etc., little is known about RNA binding motifs. Our study of how the SPOUT MTase precisely recognizes its RNA substrates sheds light on RNA substrate recognition.

Material and Methods

Materials

[Methyl-³H] SAM (78.0 Ci/mmol) was purchased from PerkinElmer Inc. (Waltham, MA, USA); [³H] DMAPP (20.0 Ci/mmol) was obtained from BIOTREND Chemicals (USA); SAM was purchased from NEW ENGLAND Biolabs Inc..; Dimethylallyl diphosphate (DMAPP), dithiothreitol (DTT), NTPs, 5′-GMP, pyrophosphate, Tris-base, β-mercaptoethanol (β-Me), MgCl₂, NaCl and KCl were purchased from Sigma-Aldrich Co. LLC. (St. Louis, MO, USA); Isopropyl β-D-thiogalactoside (IPTG) was obtained from AMRESCO (OH, USA); Nitrocellulose membranes (0.22 μm) were purchased from Merck Millipore (Darmstadt, Germany); Primers for PCR were synthesized by BioSune (Shanghai, China); Nickel-nitrilotriacetic (Ni-NTA) Superflow was purchased from Qiagen Inc.. (Germany); KOD-plus mutagenesis kit, Pyrobest DNA polymerase and the dNTP mixture were obtained from Takara (Japan); T4 polynucleotide kinase, T4 DNA ligase, Ribonuclease inhibitor and all restriction endonucleases were obtained from Fermentas/Thermo Scientific (MA, USA); The E. coli JW3581-1 and JW4129-2, strains in which the gene of TrmL and MiaA were deleted respectively, were purchased from the E. coli genetic stock center (Yale University, New Haven, CT, USA); PrepHT 300SB-C18 was purchased from Agilent Technologies (CA, USA).

Preparation of tRNAs and mutagenesis

The genes encoding EctRNAs including EctRNA^Leu_CAA, EctRNA^Leu_UAA and EctRNA^Phe_GAA were amplified from the genome of E. coli MG1655. Site-directed mutagenesis of EctRNAs was performed using the KOD-plus mutagenesis kit as described previously. ⁵² The 2′-O-methylation deficient tRNAs were overexpressed in E. coli JW3581-1, and the isopentenylation deficient tRNA was overexpressed in E. coli JW4129-2. These in vivo overexpressed tRNAs were purified by DEAE-Sepharose chromatography, urea denaturing PAGE and C18 reversed-phase HPLC chromatography as described previously.²⁰ Unmodified tRNAs were made by in vitro transcription by T7 RNA polymerase according to previous protocols.⁵³ The tRNA minihelices were synthesized by in vitro transcription of DNA templates that were made by direct primer annealing.³¹ The A37-dimethylallyl-modified (i⁶A37) tRNAs were made using recombinant E. coli MiaA (EcMiaA) and dimethylallyl diphosphate (DMAPP) in the tRNA isopentenylation assay. The tRNA concentration was determined by UV absorbance at 260 nm, and the molar absorption coefficient was calculated according to the sequence of each tRNA.

tRNA methyl transfer assay

The gene encoding E. coli TrmL (EcTrmL) was inserted into vector pET30b and expressed in E. coli BL21 (DE3). The protein purification and the tRNA methylation assay were performed as described previously.²⁰ For measurement of methyl transfer capability, 7 μM wild-type, mutant EctRNA^Leu and i⁶A37 modified tRNA minihelices were used as substrates. The reaction was performed under the same conditions, including 0.5 μM TrmL, as previously described.²⁰ At time intervals ranging between 5 and 20 min, aliquots of 10 μL were removed, absorbed on paper discs and precipitated in trichloroacetic acid. The kinetic parameters of the methylation reactions of EctRNAs were determined by using a range of 0.5–20 μM tRNA, 100 μM SAM and 0.2 μM EcTrmL.

tRNA isopentenylation assay

The gene encoding EcMiaA was inserted into the pET30b vector with an N-terminal His tag and expressed in E. coli BL21 (DE3). The protein was purified following a previously described protocol used for His-tagged EcTrmL.²⁰ The isopentenylation reaction contained 50 mM Tris-HCl (pH 7.5), 3.5 mM MgCl₂, 5 mM DTT, 1 mg/mL BSA, 7 μM EctRNA, 50 μM DMAPP and 1 μM EcMiaA. The reaction was incubated at 37°C for 20 min before quenching by phenol extraction. To measure the effective concentrations of transcribed minihelices, reactions were performed under identical conditions at 37°C, ³H labeled DMAPP was used, and aliquots of 5 μL were removed, absorbed on paper discs and precipitated in trichloroacetic acid, at time intervals ranging from 5 to 60 min. After the precise effective concentration was confirmed, these transcribed minihelices were modified by EcMiaA using non isotopic labeled DMAPP, in the same conditions as for the previous reaction, for 20 min, before quenching by phenol extraction.

Electrophoretic mobility shift assay

Electrophoretic mobility shift assay (EMSA) was carried out as previously described.²⁰ Then, 80 nM tRNA with a range of 0.75–10 μM EcTrmL was incubated in a 30 μL reaction volume at 37°C for 20 min. After incubation, each sample was loaded onto a 6% polyacrylamide native gel immediately after adding loading buffer. The gel was stained with ethidium bromide.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors gratefully acknowledge Xian Xia in their institute for his excellent technical assistance. We also thank Dr. Gilbert Eriani at Institut de Biologie Moléculaire et Cellulaire (IBMC) for carefully reading the manuscripts and important discussion.

Funding

This work was supported by the Natural Science Foundation of China [grant numbers 31270775, 31130064], and the National Key Basic Research Foundation of China [grant number 2012CB911000]. The Natural Science Foundation of China [grant number 31130064] helps cover open access publication fees.

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.