Identification of the enzymes responsible for m2,2G and acp3U formation on cytosolic tRNA from insects and plants.

Posttranscriptional modification of tRNA is critical for efficient protein translation and proper cell growth, and defects in tRNA modifications are often associated with human disease. Although most of the enzymes required for eukaryotic tRNA modifications are known, many of these enzymes have not been identified and characterized in several model multicellular eukaryotes. Here, we present two related approaches to identify the genes required for tRNA modifications in multicellular organisms using primer extension assays with fluorescent oligonucleotides. To demonstrate the utility of these approaches we first use expression of exogenous genes in yeast to experimentally identify two TRM1 orthologs capable of forming N2,N2-dimethylguanosine (m2,2G) on residue 26 of cytosolic tRNA in the model plant Arabidopsis thaliana. We also show that a predicted catalytic aspartate residue is required for function in each of the proteins. We next use RNA interference in cultured Drosophila melanogaster cells to identify the gene required for m2,2G26 formation on cytosolic tRNA. Additionally, using these approaches we experimentally identify D. melanogaster gene CG10050 as the corresponding ortholog of human DTWD2, which encodes the protein required for formation of 3-amino-3-propylcarboxyuridine (acp3U) on residue 20a of cytosolic tRNA. We further show that A. thaliana gene AT2G41750 can form acp3U20b on an A. thaliana tRNA expressed in yeast cells, and that the aspartate and tryptophan residues in the DXTW motif of this protein are required for modification activity. These results demonstrate that these approaches can be used to study tRNA modification enzymes.

Introduction

Posttranscriptional modification of tRNA is required for efficient and accurate protein translation, and tRNA from all organisms is extensively modified [1]. In the model yeast Saccharomyces cerevisiae, defects in certain tRNA modifications can lead to temperature sensitivity, impaired cell growth, or even lethality [2]. Mutations in genes encoding enzymes responsible for tRNA modifications in humans, with their corresponding loss of modifications, cause diseases including intellectual disability [3], mitochondrial disorders [4-8], and familial disautonomia [9-11]. Additionally, genes encoding tRNA modification enzymes have been associated with diseases such as cancer and metabolic disorders [12, 13]. tRNA modifications have also been shown to be involved in response to stress and other environmental stimuli [14-17]. The study of tRNA modifications in multicellular model organisms such as the plant Arabidopsis thaliana and the insect Drosophila melanogaster has also recently added insight into the roles of modifications in development and disease [18-23]. Due to the roles of tRNA modifications in varied processes, identifying and characterizing the enzymes that form them is of importance in our understanding of gene expression and human health.

The roles that many post-transcriptional modifications play in tRNA function and stability have been identified [24, 25]. In general, modifications in and around the anticodon loop are critical for translational fidelity by ensuring proper tRNA charging [26-28], reading frame maintenance [29-31], and decoding of the RNA [32]. Modifications in the body of the tRNA are generally important for proper folding of the tRNA and for tRNA stability [33-36]. For example, eukaryotic tRNAs lacking certain body modifications undergo degradation by the 5’ to 3’ exonucleases Rat1 and Xrn1 through a process known as rapid tRNA decay (RTD) [33, 37]. This process is exacerbated by high temperature, primarily due to instability of the acceptor and T-stems [38-40]. Some of these modifications in the body of the tRNA enhance stability by preventing unwanted intramolecular base-paring and/or by modulating base-pairing. Thus, the N2,N2-dimethylguanosine (m2,2G) modification found at position 26 of most G26:A44 base-pair containing eukaryotic tRNAs likely modulates non-canonical pairing with A44, and blocks alternative tRNA conformations by blocking base-pairing of G26 with other residues [41-44]. The m2,2G26 modification is formed on both cytosolic and mitochondrial tRNA by Trm1 in yeast [45, 46] and by its homolog TRMT1 in humans [47, 48]. Yeast TRM1 and human TRMT1 are encoded in the nucleus, but once expressed, localize to both the nucleus and the mitochondria [47, 49]. Defects in m2,2G26 cause temperature sensitivity in yeast [50], and intellectual disability in humans [51-54]. Additionally, correct pre-tRNA folding in the yeast Schizosaccharomyces pombe was shown to require m2,2G26 modification by Trm1, and/or La RNA chaperone activity [55].

Most eukaryotic tRNA modifications and the enzymes that form them have been identified experimentally, and both appear to be conserved throughout eukaryotes [1]. For example, of the 25 modifications known to occur on Saccharomyces cerevisiae cytoplasmic tRNA, 19 have been found on human cytoplasmic tRNA [1, 56]. To date, genes encoding the majority of the proteins required for all 25 of these modifications in S. cerevisiae have been identified [2]. Most of these genes were identified using one of two general approaches. In the first approach, yeast homologs were predicted by sequence homology to known tRNA modification enzymes, often from Escherichia coli, followed by modification analysis of tRNA from yeast knockout strains. This method was used to identify most of the yeast pseudouridine synthases [57-61], which convert uridine to pseudouridine on yeast tRNA, as well as most of the tRNA methyltransferases [62-68]. In the second approach, several other tRNA modification enzymes were discovered using proteomic libraries to detect in vitro enzymatic activity towards a tRNA substrate [69]. Yeast members of the tRNA dihydrouridine synthase family [70, 71], and three other tRNA methyltransferases [72-74] were discovered using this approach. After their discovery in yeast, many human tRNA modification enzymes have been identified in humans by sequence homology and direct experimental analysis of cells deficient for candidate genes [75].

Humans and other multicellular eukaryotes also contain tRNA modifications that are not found in yeast. For example, some plant and animal tRNAs contain the 3-amino-3-propylcarboxyuridine (acp3U) at residues 20, 20a, or 20b on cytoplasmic tRNAs [1, 76]. For a description of tRNA numbering, please see reference [77]. The human enzymes responsible for acp3U20 and acp3U20a were recently identified as DTWD1 and DTWD2, respectively [76]. Lack of these modifications in tandem resulted in slow growth of cultured human cells, suggesting a role for these modifications in translation [76]. This modification is also found at residue 47 of bacterial, mitochondrial, and plastid tRNAs [1]. In bacteria, this modification increases thermal stability [76]. The E. coli protein YfiP (now named TuaA or TapT) was recently identified as the enzyme that forms acp3U47 on bacterial tRNA [76, 78], and lack of this modification results in a motility defect and genome instability in cells [76].

In this report, we describe two different approaches that we use to experimentally identify the plant and insect orthologs of the Trm1 family of enzymes required for the m2,2G26 modification of cytosolic tRNA. The first approach involves co-expression of an exogenous tRNA and a candidate gene in yeast, followed by detection of the modification by fluorescent primer extension (Fig 1A). Using this approach we identify two A. thaliana genes that encode enzymes capable of forming m2,2G26 on cytosolic plant tRNA. In the second approach, we use RNA interference (RNAi) and primer extension analysis to identify the m2,2G26 enzyme in the model insect D. melanogaster (Fig 1B). We then use these approaches to identify orthologs of the newly discovered DTWD2 family of proteins that are required for the acp3U20a tRNA modification of cytosolic tRNA in D. melanogaster, and that can form the acp3U20b modification on an A. thaliana tRNA.

Fig 1

Schematic of approaches to identify tRNA modification enzymes.

(A) Expression of tRNA and candidate genes in S. cerevisiae. An exogenous tRNA known or expected to receive the modification of interest, and a candidate modification gene are co-expressed in a yeast strain lacking the modification. RNA is extracted from the yeast and analyzed by primer extension with a fluorescent oligonucleotide specific for the tRNA to determine if the modification is present. (B) Analysis of D. melanogaster tRNA in dsRNA-treated S2R+ cells. S2R+ cells are treated with dsRNA complementary to the gene of interest, and tRNA is analyzed using fluorescent primer extension to detect the modification.

Materials and methods

Yeast strains and plasmids

Yeast strains used in this study are listed in Table 1. The BY4742 trm1Δ::kanMX strain was purchased from Dharmacon. The BY4742 trm1Δ::kanMX met22Δ∷ble strain (yMG2129-4) was constructed by standard methods [37]. Plasmids used in this study are listed in Table 2. The A. thaliana TRM1a (AT3G02320), TRM1b (AT5G15810), TRM1c (AT3G56330), DTWD2A (AT2G41750), and DTWD2B (AT5G54880) yeast expression plasmids were constructed by PCR amplification of genes from an A. thaliana cDNA library [79] with primers that added PacI and EagI cut sites, followed by digestion of the product with PacI and EagI and ligation into vector BG2596 [80]. Variants of A. thaliana genes were generated by performing site-directed mutagenesis on plasmids containing wild type genes using the Q5 site-directed mutagenesis kit (Promega) following manufacturer’s instructions. The A. thaliana tRNALeu(AAG)-G37C plasmid was generated by PCR amplification of the tRNA gene from A. thaliana genomic DNA with primers that added XhoI and BglII sites, followed by digestion of the product with XhoI and BglII, and ligation into plasmid pMG24A [56] to replace the S. cerevisiae tRNALeu(UAA) gene, yielding plasmid pMG669B. A G37C mutation was then added by performing site-directed mutagenesis on plasmid pMG669B using Q5 site-directed mutagenesis to yield A. thaliana tRNALeu(AAG)-G37C (pMG671B).

Table 1

Yeast strains used in this study.

Strain	Genotype	Source
BY4742	MATα his3-Δ1 leu2Δ0 met15-Δ0 ura3-Δ0	Open Biosystems
BY4742 trm1Δ	BY4742, trm1Δ::kanMX	Dharmacon
yMG2129-4	BY4742, trm1Δ::kanMX, mett22Δ::bleR	This study

Table 2

Plasmids used in this study.

Plasmid	Parent	Description	Source
pBG2596		2μ URA3 PGAL110 LIC	(Quartley et al. 2009)
pMG662B	pBG2596	2μ URA3 PGAL10 A. thaliana TRM1a	This study
pMG729E	pBG2596	2μ URA3 PGAL10 A. thaliana TRM1a-D200A	This study
pMG717A	pBG2596	2μ URA3 PGAL10 A. thaliana TRM1b	This study
pMG730E	pBG2596	2μ URA3 PGAL10 A. thaliana TRM1b-D200A	This study
pMG718A	pBG2596	2μ URA3 PGAL10 A. thaliana TRM1c	This study
pMG735B	pBG2596	2μ URA3 PGAL10 A. thaliana TRM1c-D172A	This study
pMG24a		2μ LEU2 S. cerevisiae tRNALeu(UAA)	(Guy et al. 2012)
pMG669B	pMG24a	2μ LEU2 A. thaliana tRNALeu(AAG)	This study
pMG671B	pMG669B	2μ LEU2 A. thaliana tRNALeu(AAG)-G37C	This study
pMG725A	pBG2596	2μ URA3 PGAL10 A. thaliana DTWD2A	This study
pMG726A	pBG2596	2μ URA3 PGAL10 A. thaliana DTWD2B	This study
pMG731C	pBG2596	2μ URA3 PGAL10 A. thaliana DTWD2A-D133A	This study
pMG732E	pBG2596	2μ URA3 PGAL10 A. thaliana DTWD2A-W136A	This study

RNA isolation

Yeast strains were grown in YPD or selective synthetic media to late log phase. For primer extension, qRT-PCR, and tRNA purification, bulk low molecular weight RNA was extracted using hot phenol as previously described [72]. RNA extraction from D. melanogaster S2R+ cells was performed using TRIzol (Thermo Fisher) according to manufacturer’s instructions.

Primer extension assays

For primer extension, 10–20 μmol 5’ Tye665-labeled oligonucleotides (Integrated DNA Technologies) was annealed to 2–10 μg RNA, heated to 95°C and slow cooled to 37°C. Labeled oligonucleotides are listed in Table 3. The entire reaction was then incubated with 1 mM dNTPs and 1.89 U of Avian Myeloblastosis Virus (AMV) reverse transcriptase. Reactions were incubated at 37°C overnight, and analyzed by 15% PAGE with 7M Urea. The gel was placed between overhead projector sheets and visualized using a Typhoon 9200 scanner with a 620 BP30 Cy5 emission filter at high sensitivity and PMT between 600–800. Sequencing reactions were annealed similarly and extended overnight with the addition of 0.1mM ddNTP’s. Primer MPG1772 was used for detection of m2,2G26 on A. thaliana tRNALeu(AAG)-G37C, and MPG1587 was used for detection of acp3U20b on the same tRNA.

Table 3

Tye665-labeled oligonucleotides used for primer extension.

Primer	Target tRNA	Sequence	Target Nucleotides
MPG951	S. cerevisiae tRNATyr	CGAACGCCCGATCTCAAGATT	55–37
MPG1012	S. cerevisiae tRNATrp	GTGAAACGGACAGGAATTGAACCTG	74–50
MPG1587	A. thaliana tRNALeu(AAG)-G37C	CTTTCGGACCAGAAGCTTAATCT	e5-29
MPG1772	A. thaliana tRNALeu(AAG)-G37C	CCTTTCGGACCAGAAGCTTAA	47–33
MPG1045	D. melanogaster tRNATyr	GAACCAGCGACCTATGGATCTACAG	56–32
MPG1126	D. melanogaster tRNAVal(CAC)	GAACCGGGGACCTTGTGCGTGTG	56–34

Northern blot analysis

Bulk RNA was analyzed by PAGE and transferred to a nitrocellulose membrane as previously described [33]. For detection by Northern blot, S. cerevisiae tRNATrp was detected with 5’ Tye665-labeled probe MPG1012 and A. thaliana tRNALeu(AAG)-G37C was detected with MPG1587 (Table 3).

Quantitative Real Time PCR (qRT-PCR)

RNA was treated with RQ1 RNase-Free DNase (Promega) and then reverse transcribed using a Verso cDNA Kit (Thermo Scientific) using a 3:1 (v/v) mix of random hexamers and anchored oligo-dT primers. DNA was then PCR amplified using DyNAmo HS SYBR™ Green qPCR Kit (Thermo Scientific) master mix and primers specific to indicated genes. Primers used for qRT-PCR are listed in Table 4. In yeast, RNA was normalized to ACT1, and in D. melanogaster, RNA was normalized to Act42A.

Table 4

Oligonucleotides used for qRT-PCR.

Primer	Organism	Target	Sequence
MPG1598	S. cerevisiae	ACT1	GAAATGCAAACCGCTGCTCA
MPG1599	S. cerevisiae	ACT1	TACCGGCAGATTCCAAACCC
MPG1756	S. cerevisiae	TUB1	CCAAGGGCTATTTACGTGGA
MPG1757	S. cerevisiae	TUB1	GGTGTAATGGCCTCTTGCAT
MPG1617	A. thaliana	TRM1a	GGGACTCATGTGAATCCGCT
MPG1618	A. thaliana	TRM1a	CGAACCAATGTGACGCGAAA
MPG1652	A. thaliana	TRM1b	CCACGTCGGTTCGCTTAGTA
MPG1653	A. thaliana	TRM1b	CGACCTGCCCTTATCTTGGG
MPG1654	A. thaliana	TRM1c	AATGAGATTGGGCTGCGGAT
MPG1655	A. thaliana	TRM1c	CATGTAGCTTCCCACGGTGA
MPG1266	D. melanogaster	TRM1	CGGAAGAAGGTCAAGGAACA
MPG1267	D. melanogaster	TRM1	GATCCGAAGTCCATCCTCATATC
MPG1635	D. melanogaster	ACT42a	CAAGAGTACGACGAGTCGGG
MPG1636	D. melanogaster	ACT42a	TTCGATGAGGAACGACCACG
MPG1706	D. melanogaster	DTWD2A	GTCCGTCATCAAACCGGACT
MPG1707	D. melanogaster	DTWD2A	GCATGGTAGGTGTGCTTTGC
MPG1708	A. thaliana	DTWD2A	TGGTCAAAGCTAGCGAGGTG
MPG1709	A. thaliana	DTWD2A	CTCCACTCACTGATGCGTCT
MPG1710	A. thaliana	DTWD2B	CAAAGAAGTACGACGGCAGC
MPG1711	A. thaliana	DTWD2B	TTCACCGATCTCCTTCATCGC

tRNA purification

tRNALeu(AAG)-G37C was purified from yeast total RNA extracts using a 5’-biotinylated oligonucleotide with the sequence 5’-TGGTGTTGACAGTGGGATTTGAACCC-3’, which is complementary to nucleotides 76–51 in the tRNA, as previously described (72). Briefly, biotinylated oligonucleotide bound to streptavidin magnetic particles was incubated with yeast RNA to capture tRNALeu(AAG)-G37C. Bound tRNA was then washed extensively to remove non-specific tRNA. Eluted tRNA was quantified using A260 prior to LC-MS/MS analysis.

LC-MS/MS analysis of nucleosides

tRNA (2–4 μg) was digested to single nucleosides as previously described using nuclease P1, snake venom phosphodiesterase, and bacterial alkaline phosphatase [81, 82]. The lyophilized samples were reconstituted in mobile phase A. Reversed phase chromatography was carried out with a high-strength silica column (Acquity UPLC HSS TS, 1.8 μm, 1.0 mm X 100 mm, Waters, Milford, MA) on an ultra-high-performance liquid chromatography (UHPLC) system (Vanquish Flex Quaternary, Thermo Fisher Scientific, San Jose, CA). Mobile phase A contained 5.3 mM ammonium acetate in water, pH 5.3, and mobile phase B contained a mixture of acetonitrile/water (40:60) with 5.3 mM ammonium acetate.

The gradient program consisted of: 0% B (from 0 to 7.6 min), 2% B at 15.7 min, 3% B at 19.2 min, 5% B at 25.7 min, 25% B at 29.5 min, 50% B at 32.3 min, 75% B at 36.4 min (hold for 0.2 min), 99% B at 39.6 min (hold for 7.2 min), then returning to 0% B at 46.9 min. After that, a re-equilibration step at 0% B for 18.1 min was employed prior to the next injection. A flow rate of 100 μL min-1 was used. The column temperature was set at 30°C.

An Orbitrap FusionTM LumosTM Tribrid mass spectrometer (Thermo Fisher Scientific) interfaced with a heated electrospray (H-ESI) source was used for the identification and relative quantification of m2,2G. All instrument parameters were as described previously [83], except that the collision-induced dissociation (CID) energy was set at 40%, and higher-energy collisional dissociation (HCD) energy was set at 80 arbitrary units.

QuantivaTM triple quadrupole mass spectrometer interfaced with an H-ESI source (Thermo Fisher Scientific) was used for the identification and relative quantification of acp3U. Instrument settings included detection in positive polarity with an H-ESI electrospray voltage of 3.8 kV, Ion Transfer Tube temperature at 290°C, vaporizer temperature of 100°C, sheath gas, auxiliary gas and sweep gas at 35, 10, 0 arbitrary units respectively. Collision gas pressure was 1.5 mTorr. Q1 and Q3 Resolution (FWHM) was set at 0.7 Da. Selected reaction monitoring (SRM) transitions for acp3U were m/z 346–214, 346–197, 346–168 with collision energy (CE) of 20 V, RF lens was 58 V. SRM transition for U was m/z 245–113 with collision energy (CE) of 10 V and RF lens 30 V.

Data were analyzed using Qual browser of Xcalibur 3.0.

dsRNA generation

dsRNA for RNAi was generated by PCR amplification of the desired gene from D. melanogaster genomic DNA with primers containing 5’ T7 promoter sites. Primers used for dsRNA generation are listed in Table 5. Regions for dsRNA were selected using the Drosophila RNAi Screening Center (DRSC) website to minimize off-target effects. The PCR product was analyzed by agarose gel, purified using a QIAquick gel extraction kit under low salt conditions (Qiagen), and 0.2 μg was used as a template for in vitro transcription using a T7 Megascript kit and purified according to manufacturer’s instructions (Thermo Fisher).

Table 5

Oligonucleotides used for D. melanogaster dsRNA generation.

Primer	Target	Sequence
MPG932	TRM1	GAATTAATACGACTCACTATAGGGAGATCTTTCCATGGTCCAAGAGG
MPG933	TRM1	GAATTAATACGACTCACTATAGGGAGAGATAATGCCGATTTGCGACT
MPG736	TRM7	GAATTAATACGACTCACTATAGGGAGAGAGTGCCTTCAAGTTGCTCC
MPG737	TRM7	GAATTAATACGACTCACTATAGGGAGAGCATCTGGGAGCTCAATAGC
MPG1698	DTWD2A	GAATTAATACGACTCACTATAGGGAGAACACCCTGCAGAGGAGAAGA
MPG1699	DTWD2A	GAATTAATACGACTCACTATAGGGAGAAGTCACTTATACCCACGGCG
MPG1700	DTWD2A	GAATTAATACGACTCACTATAGGGAGATCTACAAAGGCAAGCGGTTT
MPG1701	DTWD2A	GAATTAATACGACTCACTATAGGGAGAAGGCGGTGTTTCTAAGCAGA

D. melanogaster S2R+ cell culture and RNAi treatment

D. melanogaster S2R+ cells [84] were grown in Schneider’s medium (Gibco) + 10% fetal bovine serum (FBS) (Gibco) at 27°C in culture flasks. RNAi was performed in 6-well plates as described on the DRSC website, essentially as previously described [85]. Briefly, 2 X 106 cells were plated to each well of a 6-well plate in 1 mL of serum free media, 10 μg of dsRNA was added to each well, and cells were incubated at room temperature for 30 minutes. After incubation, 3 mL of media + 10% FBS was added to each well. Cells were harvested after 3 days, or for multiple RNAi treatments, medium was removed, each well was resuspended in 1 mL of serum-free medium, and 250 μL of cells were plated in new wells already containing 750 μL serum-free medium. Cells were then treated with dsRNA as previously described and harvested or passaged again after 3 days.

Results and discussion

Primer extension assays using fluorescent oligonucleotides sensitively detect m2,2G26 on tRNA from yeast cells

We first determined whether fluorescently-labeled oligonucleotides could be used to sensitively detect tRNA modifications in yeast by primer extension [78, 86]. Thus, we used a Tye665 5’-end labeled oligonucleotide specific to Saccharomyces cerevisiae tRNATyr (Fig 2A) to perform primer extension on RNA from wild type and trm1Δ cells, to detect the m2,2G26 modification. In a primer extension assay, the presence of a base-pair blocking modification such as m2,2G prohibits incorporation of a nucleotide complementary to the modified nucleotide by reverse transcriptase. Thus, presence of m2,2G26 results in a stop at residue 27 of the tRNA when analyzed by polyacrylamide gel electrophoresis (PAGE). We mixed wild type cells and trm1Δ cells in different ratios, extracted small molecular weight RNA, and then performed a primer extension assay specific to tRNATyr. PAGE of the reactions showed a robust block at position 27 on tRNATyr from wild type cells (Fig 2B), whereas a block at this position was not observed on tRNATyr from trm1Δ cells. Instead, a larger band representing a primer extension block closer to the 5’ end of tRNATyr was observed, which corresponds to the string of dihydrouridines found in the D-loop of the tRNA (Fig 2A and 2B). Detection of m2,2G by this method appears to be sensitive, because a primer extension block corresponding to m2,2G26 could be easily detected in tRNA from wild-type cells mixed with trm1Δ cells at a ratio of 1/10, with detection at a ratio of 1/20 also possible (Fig 2B). Our ability to detect the m2,2G26 modification in wild-type cells from a background of trm1Δ cells indicates that this technique is sensitive enough to detect cells with a tRNA modification from a pool of cells lacking the modifications, which could be useful in the development of a screen to identify genes required for tRNA modifications. This sensitivity also further suggests that in many instances fluorescence could be a viable alternative to traditional primer extension [78, 86], which uses 32P. 32P has exquisite sensitivity, and is critical for the study of tRNA modifications. However, the use of radioactivity has several drawbacks, which include expense, safety concerns requiring extensive additional training of personnel, and a relatively short window in which the radioactivity can be used, due to the short half-life of 32P [86].

Fig 2

Detection of m2,2G26 by fluorescent primer extension in yeast cells.

(A) Schematic of S. cerevisiae tRNATyr. Location of primer binding is shown in blue. Selected nucleotides are numbered in red. (B) Detection of m2,2G26 by fluorescent primer extension in yeast cells. Left, bulk RNA was extracted from indicated strains and analyzed by primer extension to yeast tRNATyr. Selected nucleotides from the sequencing reactions are provided on the left of the gel. Right, bulk RNA was extracted from culture of indicated individual strains or culture from mixtures of strains in indicated ratios, and then analyzed by primer extension to yeast tRNATyr.

Identification of A. thaliana TRM1 enzymes using yeast cells and primer extension

We next sought to identify tRNA modification genes from multicellular organisms by expressing the genes in yeast and using fluorescence-based primer extension to detect modification of the tRNA. This approach could provide some advantage over purely in vitro assays, because subcellular trafficking and localization of the protein in yeast could yield insights into the tRNA specificity of the enzyme. However, it is also possible that exogenously expressed proteins could be targeted incorrectly in yeast, thus confounding results. Nonetheless, this system provides a convenient method to rapidly determine if an enzyme has the ability to modify a given tRNA. To demonstrate the feasibility of this approach, we determined which predicted TRM1 enzymes can form the m2,2G26 modification on Arabidopsis thaliana tRNA. Interest in the study of tRNA modifications in A. thaliana has increased, with recent reports showing that tRNA modifications are involved in development [20, 87–89], disease resistance [18, 90], and stress response [23, 91, 92].

There are three known m2G/m2,2G26 methyltransferases in eukaryotes, which include Trm1, Trm11, and Trm14 [45, 68, 93]. All three of these eukaryotic enzymes have a similar Rossman-fold methyltransferase domain, but unlike Trm11 and Trm14 proteins, Trm1 proteins lack a THUMP domain which is involved in tRNA binding [94]. Three predicted TRM1 candidate genes have been identified in A. thaliana based on sequence homology [23] (S1 Fig), and the lack of a THUMP domain. To determine if these Trm1 proteins form m2,2G26, we co-expressed an A. thaliana tRNA with A. thaliana TRM1 candidate genes in S. cerevisiae, and then detected m2,2G26 by fluorescent primer extension. To monitor m2,2G modification upon expression of A. thaliana candidate genes, we generated a high copy yeast expression plasmid encoding A. thaliana tRNALeu(AAG) with a G37C mutation (tRNALeu(AAG)-G37C). The modification profile of cytosolic A. thaliana tRNALeu(AAG) has not been determined, but cytosolic tRNALeu(AAG) from the plant Lupinus luteus contains a m2,2G26 modification [95], and A. thaliana tRNALeu(AAG) has a G at position 26 [1], suggesting that the A. thaliana tRNA should also receive the m2,2G26 modification in the presence of a TRM1 enzyme. Moreover, because L. luteus tRNALeu(AAG) contains a 1-methylguanosine (m1G) modification at residue 37, and A. thaliana tRNALeu(AAG) contains a G37 [1], we added a G37C mutation to A. thaliana tRNALeu(AAG) to ensure that G37 on the A. thaliana tRNA would not be modified to m1G by yeast Trm5 [62] (Fig 3A). m1G blocks base-pairing, and would therefore likely interfere with primer binding or primer extension in the assay.

Fig 3

Identification of TRM1 genes from A. thaliana in yeast cells.

(A) Schematic of A. thaliana tRNALeu(AAG)-G37C. Location of primer binding is shown in blue. G37C mutation is boxed. Predicted location of m2,2G26 (cyan) and acp3U20b (magenta) modifications are circled. Selected nucleotides are numbered in red. (B) A. thaliana tRNALeu(AAG)-G37C is expressed in yeast cells. RNA was extracted from yeast expressing the indicated plasmid (vertical label) and analyzed by Northern blot to the indicated probe (horizontal label) as described in Materials and methods. (C) TRM1a or TRM1b expression results in a primer extension block consistent with m2,2G26 on A. thaliana tRNA. Bulk RNA was extracted from a trm1Δ yeast strain expressing the indicated plasmids and then analyzed by primer extension to A. thaliana tRNALeu(AAG)-G37C.

We transformed yeast trm1Δ cells with the plasmid expressing cytosolic A. thaliana tRNALeu(AAG)-G37C and with a high copy plasmid expressing A. thaliana gene TRM1a (AT3G02320) [23] under control of the P promoter, and then detected the m2,2G26 modification on tRNALeu(AAG)-G37C using a primer extension assay. Cells were grown in selective media containing galactose, and were grown at room temperature to decrease the likelihood that the tRNALeu(AAG)-G37C would be degraded by RTD [33, 40]. Northern blot analysis using a fluorescent oligonucleotide specific to A. thaliana tRNALeu(AAG)-G37C tRNA demonstrated that the tRNA was readily expressed from the plasmid (Fig 3B). Two bands of differing mobilities specific to tRNALeu(AAG)-G37C were detected. Expression of TRM1a mRNA in cells transformed with a TRM1a expression plasmid was confirmed by quantitative real-time PCR (qRT-PCR) (Table 6). Primer extension specific to A. thaliana tRNALeu(AAG)-G37C resulted in a primer extension block consistent with the m2,2G26 modification in trm1Δ cells when TRM1a was expressed (Fig 3C), suggesting that TRM1a is a Trm1 ortholog in A. thaliana.

Table 6

Relative mRNA levels of A. thaliana TRM1 homologs expressed in yeast.

A. thaliana gene	target mRNA	relative levelsa
TRM1a	TRM1a	1.51 ± 0.34
TRM1a-D200A	TRM1a	8.76 ± 2.41
TRM1b	TRM1b	68.6 ± 35.2
TRM1b-D200A	TRM1b	32.4 ± 9.98
TRM1c	TRM1c	12.86 ± 5.11

aRelative to S. cerevisiae tubulin (TUB1), after both genes were normalized to S. cerevisiae actin (ACT1). Values are from three independent growths.

Because there are two other predicted TRM1 genes in A. thaliana [23], we also tested the ability of TRM1b (AT5G15810) and TRM1c (AT3G56330) to form m2,2G on A. thaliana tRNALeu(AAG)-G37C from high copy expression plasmids in yeast. Expression of A. thaliana TRM1b resulted in a primer extension block on A. thaliana tRNALeu(AAG)-G37C consistent with the presence of m2,2G26, whereas expression of A. thaliana TRM1c did not result in a block within the limits of our detection (Fig 3C). The lack of m2,2G formation on tRNA from cells expressing TRM1c was not due to lack of mRNA expression (Table 6). This data suggests that TRM1b is also a Trm1 ortholog. It is possible that TRM1c is responsible for the m2,2G26 modification found on several plastid tRNAs [1], because it is predicted to contain a chloroplast transit peptide and is predicted to localize to this cellular organelle [96, 97]. We cannot rule out the possibility that expression of TRM1c failed to generate a detectable primer extension block due to lack of protein expression and/or incorrect subcellular localization of the protein resulting from expression of this exogenous plant gene in yeast.

To further determine if the primer extension block observed upon the expression of TRM1b is due to Trm1 activity, we mutated a predicted active site aspartate residue [98] in Trm1a and in TRM1b (S1 Fig), expressed the variants in yeast, and measured modification activity by primer extension assay. Cells expressing a TRM1a or TRM1b variant containing an alanine substitution of the predicted active site aspartate (D200A) failed to generate a primer extension block (Fig 3C), even though the genes were expressed (Table 6). These data further support the conclusion that TRM1a and TRM1b are A. thaliana Trm1 orthologs.

To verify that the primer extension block observed on A. thaliana tRNALeu(AAG)-G37C at position 27 was due to the m2,2G26 modification, we purified tRNALeu(AAG)-G37C in the presence or absence of TRM1a or TRM1b, and analyzed nucleotide content by liquid chromatography followed by mass spectrometry (LC-MS). To this end, tRNALeu(AAG)-G37C was expressed in the presence or absence of A. thaliana TRM1 genes in trm1Δ mett22Δ yeast, purified using a biotinylated nucleotide specific to the tRNA, and then digested to nucleosides. The met22Δ mutation was included in these cells to inhibit the rapid tRNA decay pathway [37], which would be expected to reduce degradation of tRNALeu(AAG)-G37C and increase tRNA yields. Nucleosides were separated by reversed phase HPLC and identified using a high-resolution, accurate-mass (HRAM) orbitrap mass spectrometer [83]. The m2,2G modification on tRNALeu(AAG)-G37C purified from cells expressing TRM1a or TRM1b was readily detected using this method, whereas levels of m2,2G on the same tRNA purified from cells expressing a vector were near background (S2 Fig). We therefore conclude that A. thaliana TRM1a and TRM1b form m2,2G on A. thaliana tRNALeu(AAG).

Our finding that A. thaliana has two enzymes capable of forming cytosolic m2,2G26 is different than what has been observed in yeast and humans, which each have been shown to only encode one enzyme responsible for formation of cytosolic m2,2G26 [45, 47]. Although humans encode the TRMT1L protein, which is related to TRMT1 by sequence, it does not form m2,2G26 on cytosolic tRNA [47]. Human TRMT1L and A. thaliana TRM1c share only 22% sequence identity, whereas human TRMTL1 shares 27% identity with A. thaliana TRM1a, and A. thaliana TRM1c shares 27% identity with human TRMT1. This sequence data strongly suggests that TRMTL and TRM1c do not have similar functions. Homologs to A. thaliana TRM1c are found in a wide array of plant species with amino acid sequence identities ranging from 37% in the green algae Micromonas pusilla to 68% in Carica papaya (S3 Fig). Furthermore, as mentioned previously, TRM1c contains a predicted chloroplast transit peptide. Thus, it is highly likely that TRM1c is responsible for m2,2G26 modification of plastid tRNA.

Identification of the D. melanogaster TRM1 enzyme in cultured cells using RNA interference and primer extension

In a separate approach to identify tRNA modification enzymes in a multicellular eukaryote, we used RNA interference (RNAi) coupled with fluorescent primer extension to identify the D. melanogaster TRM1 enzyme. Robust RNAi can be achieved by treating cultured S2R+ cells with double stranded RNA (dsRNA) specific to the gene of interest generated directly from a PCR product [85, 99–101]. We tested the effect of RNAi to the predicted D. melanogaster TRM1 gene (CG6388), followed by primer extension to detect m2,2G on tRNATyr, which receives the modification (Fig 4A) [102]. We treated S2R+ cells with dsRNA to TRM1 twice over the course of 6 days, harvested cells, extracted RNA, and used fluorescent primer extension specific to tRNATyr to detect the presence of m2,2G. Quantification of TRM1 gene expression at day 6 by qRT-PCR revealed that TRM1 mRNA levels were significantly reduced compared to untreated cells (Fig 4B), verifying that dsRNA treatment resulted in a significant reduction in gene expression. Primer extension specific to tRNATyr from cells treated with no dsRNA or with dsRNA to D. melanogaster TRM7/FTSJ1 [99, 103] resulted in the presence of two primer extension blocks, the darker, bottom one of which was consistent with the size expected for an m2,2G modification on residue 26 (Fig 4C). In contrast, primer extension analysis of tRNATyr from cells treated with dsRNA to TRM1 showed a marked decrease in the intensity of the bottom band of the doublet, with a corresponding appearance of a band of lesser mobility, consistent with the presence of the acp3U modification found at position 20 of the tRNA [102] (Fig 4C, S4 Fig). No obvious difference in cell growth or morphology was observed between cells treated with either TRM1 or TRM7 dsRNA and untreated cells. The top band of the doublet is consistent with a pause at residue C25, so although D. melanogaster tRNATyr has G residues at positions 26 and 27, it does not appear that the doublet is due to m2,2G modification of both residues 26 and 27, as has been observed for human tRNATyr [104] and for tRNACys from the eubacterium Aquifex aeolicus [105].

Fig 4

CG6388 encodes the TRM1 enzyme in D. melanogaster.

(A) Schematic of D. melanogaster tRNATyr. Location of primer binding is shown in blue. Selected nucleotides are numbered in red. (B) TRM1 mRNA levels significantly decrease in dsRNA-treated S2R+ cells. Cells were treated with dsRNA to TRM1 as indicated and mRNA levels were measured by qRT-PCR. RNA levels are expressed relative to actin (ACT42a). (C) Knockdown of CG6388 by RNAi results in loss of a primer extension block consistent with m2,2G26. S2R+ cells were treated twice over 6 days with dsRNA to indicated gene as described in Materials and methods. After harvest of cells, RNA was extracted and primer extension to tRNATyr was performed. (D) Time course of CG6388 knockdown by RNAi. S2R+ cells were treated as indicated, RNA was extracted, and primer extension to tRNATyr was performed.

To further optimize RNAi for knockdown of tRNA modifications in S2R+ cells, we varied the number of dsRNA treatments over the course of 8 and 11 days, and harvested cells at different time points. We found that three dsRNA treatments of cells over the course of 8 days was sufficient to knock out nearly all detectable levels of the primer extension block corresponding to m2,2G (Fig 4D). Thus, our results demonstrate that CG6388 encodes the TRM1 enzyme responsible for the m2,2G26 modification in D. melanogaster, because tRNATyr in D. melanogaster has previously been shown to have an m2,2G26 modification [102], and because of the high sequence homology between this D. melanogaster protein and other bona fide Trm1 proteins. These results also suggest that this system could be used to rapidly screen candidate genes for novel tRNA modifications in D. melanogaster.

DTWD2 homologs are responsible for the acp3U20a modification in flies and plants

DTWD proteins were recently identified as the enzymes responsible for acp3U modifications on eukaryotic tRNA [76]. DTWD2 was shown to be responsible for the acp3U20a tRNA modification in human cells, and homologs of these genes in other organisms were also identified by sequence [76]. We used our fluorescent primer extension approach to determine if DTWD2 genes are responsible for the acp3U20a modification found on tRNA from D. melanogaster and for the acp3U20b modification found on A. thaliana [1]. To this end, we silenced CG10050, which is the predicted DTWD2 homolog in D. melanogaster [76] (Fig 5A), and determined acp3U20a modification levels on tRNAVal(CAC) from these cells by primer extension (Fig 5B). A robust primer extension block on tRNAVal(CAC) of a size consistent with that expected for acp3U20a was observed on tRNA from untreated cells or cells treated with TRM1 dsRNA (Fig 5C). In contrast, treatment of S2R+ cells with either of two different dsRNA constructs directed to D. melanogaster DTWD2 resulted in a significant loss of the primer extension block (Fig 5C). No obvious difference in cell growth or morphology was observed between S2R+ cells treated with either DTWD2 dsRNA construct and untreated cells. These results strongly suggest that D. melanogaster DTWD2 forms acp3U20a in D. melanogaster cells.

Fig 5

CG10050 encodes the DTWD2A enzyme in D. melanogaster.

(A) Sequence alignment of DTWD2 proteins. Green box represents the DTW domain, and arrows denote predicted active site residues. (B) Schematic of D. melanogaster tRNAVal(CAC). Location of primer binding is shown in blue. Selected nucleotides are numbered in red. (C) Knockdown of CG10050 by RNAi results in loss of a primer extension block consistent with acp3U20b on tRNA. S2R+ cells were treated three times over 8 days with dsRNA to indicated genes. After harvest of cells, RNA was extracted and primer extension to tRNAVal(CAC) was performed.

We next sought to identify the A. thaliana enzyme responsible for acp3U20b formation on tRNA. In A. thaliana, DTWD2A (AT2G41750) and DTWD2B (AT5G54880) are the putative homologs of human DTWD2 [76] (Fig 5A). We therefore expressed each gene in trm1Δ met22Δ yeast cells with A. thaliana tRNALeu(AAG)-G37C, which is expected to receive the acp3U20b modification, because tRNALeu(AAG) from L. luteus contains acp3U20b [95]. We used yeast cells containing the trm1Δ mutation to ensure that yeast Trm1 would not form the base-pair blocking m2,2G26 modification, and cells containing the met22Δ mutation to inhibit the rapid tRNA decay pathway [37]. We extracted RNA from the yeast cells and detected the formation of acp3U20b by fluorescent primer extension. We found that expression of DTWD2A resulted in a primer extension block on tRNALeu(AAG)-G37C when both the gene and the tRNA were expressed in yeast, consistent with the size expected for the acp3U20b modification (Fig 6). This primer extension block was absent in yeast cells expressing only the tRNA. The block was also not present within the limits of our detection in cells expressing DTWD2B and the tRNA (Fig 6, Table 7).

Fig 6

A. thaliana DTW2A expression results in a primer extension block consistent with acp3U20b on A. thaliana tRNA.

Bulk RNA was extracted from a trm1Δ met22Δ yeast strain expressing the indicated plasmids and then analyzed by primer extension to A. thaliana tRNALeu(AAG)-G37C.

Table 7

Relative mRNA levels of A. thaliana DTWD2 genes expressed in yeast.

A. thaliana gene	target mRNA	relative levelsa
DTW2A	DTW2A	25.38 ± 2.05
DTWD2A-D133A	DTWD2A	12.12 ± 5.19
DTWD2A-W136A	DTWD2A	17.23 ± 7.35
DTW2B	DTW2B	39.23 ± 7.35

aRelative to S. cerevisiae tubulin (TUB1), after both genes were normalized to S. cerevisiae actin (ACT1). Values are from three independent growths.

To verify that the primer extension block at position 21 on tRNALeu(AAG)-G37C was due to the acp3U20b modification, we performed LC-MS/MS on nucleosides from purified tRNA. tRNA nucleosides from trm1Δ mett22Δ cells expressing tRNALeu(AAG)-G37C with or without DTWD2A were analyzed by reverse phased LC coupled with a triple quadrupole mass spectrometer. Selected reaction monitoring (SRM) was used to monitor the mass transitions for acp3U and U. We readily detected acp3U on tRNALeu(AAG)-G37C purified from cells expressing DTWD2A, whereas levels of the modification on tRNA from cells expressing the vector control were near background levels (S5 Fig). To further verify that the observed LC peak was indeed acp3U, the nucleoside digest from the DTWD2A expressing cells was compared to a total E. coli tRNA nucleoside digest. E. coli contains several tRNAs with the acp3U modification [1]. The chromatogram and MS/MS spectra of acp3U from purified tRNALeu(AAG)-G37C match the retention time and fragment ratio of acp3U from total E. coli tRNA (S6 and S7 Figs). Thus, we conclude that the primer extension block formed on tRNALeu(AAG)-G37C in the presence of DTWD2A is due to acp3U.

These data suggest that DTWD2A forms acp3U20b on plant tRNA, and are the first demonstration of an enzyme capable of forming an acp3U modification at position 20b on a tRNA. This result is somewhat surprising, because A. thaliana DTWD2A is more closely related to human DTWD2 than is A. thaliana DTWD2B [76]. Whether DTWD2A forms the modification in plant cells, and whether it can also form acp3U20a is not clear. Additionally, we note that it is possible that DTWD2B is also capable of forming acp3U20b, but that the protein is not expressed well in yeast, or that it is incorrectly targeted in yeast cells.

DTWD2 belongs to the TDD superfamily of proteins, and is localized to the nucleus and the cytoplasm [78, 106, 107]. This protein superfamily contains E. coli TuaA/TapT, as well as Tsr3, which is required to add the acp group to the 1-methyl-3-(3-amino-3-carboxypropyl)-pseudouridine modification (m1acp3Ψ) found at residue U1191 on the yeast 18s rRNA and residue U1248 on human 18s rRNA. [78, 107]. TDD superfamily proteins contain a DTW domain with an E/DXS/TW motif [78, 106] (Fig 5A). In previously studied TDD proteins, the Asp/Glu and Trp residues of the motif are found in the active site and are required for catalysis. Indeed, Asp137 of this motif in E. coli TuaA/TapT is required for activity, and is likely the catalytic base, while Trp140 is required for activity and S-adenosyl methionine (SAM) cofactor binding [78, 107]. To determine if the corresponding Asp and Trp residues of the motif are required for the activity of a eukaryotic DTWD2 protein (Fig 5A), we generated D133A and W136A variants of A. thaliana DTW2a and tested their modification activity. We found that coexpression of either the DTWD2A-D133A or DTWD2A-W136A variants in yeast with tRNALeu(AAG)-G37C did not result in the formation of a primer extension block consistent with acp3U20b (Fig 6, Table 7), whereas expression of wild type DTWD2A resulted in formation of the primer extension block. These results further suggest that DTWD2A is responsible for the acp3U20b modification found on A. thaliana tRNA. Because an in vitro assay for eukaryotic DTWD2 has not been reported, these results are the first experimental evidence that the Asp and Trp residues of the E/DXS/TW motif of a eukaryotic acp3U tRNA modification enzyme are required for activity. Whether A. thaliana DTWD2B and/or DTWD2A are required for the acp3U20a modification of cytosolic tRNA, and whether the A. thaliana TapT-like enzyme is required for acp3U47 modification on plastid tRNA remains to be determined [76].

Sequence alignment of TRM1 proteins.

Green arrow denotes location of predicted active site aspartate residue.

(PDF)

Click here for additional data file.

A. thaliana TRM1a and TRM1b form m2,2G on A. thaliana tRNALeu(AAG).

Overlay of extracted ion chromatograms showing the abundance of m2,2G (m/z 312.131 ± 5 ppm) on A. thaliana tRNALeu(AAG)-G37C expressed in trm1Δ mutant yeast expressing TRM1a (green), TRM1b (red), or a vector control (blue).

(PDF)

Click here for additional data file.

Sequence alignment of TRM1c homologs in various plants species.

(PDF)

Click here for additional data file.

Knockdown of CG6388 by RNAi results in loss of a primer extension block consistent with m2,2G26 and the appearance of a block consistent with acp3U20.

S2R+ cells were treated twice over 6 days with dsRNA to indicated gene as described in Materials and methods. After harvest of cells, RNA was extracted and primer extension to tRNATyr was performed. Sequencing was performed on cells treated with dsRNA to CG6388 to determine the location of the new primer extension block which appeared upon loss of the m2,2G26 block.

(PDF)

Click here for additional data file.

A. thaliana DTWD2A is necessary for acp3U modification on A. thaliana tRNALeu(AAG).

Chromatogram overlays of acp3U (top) and uridine (bottom) from A. thaliana tRNALeu(AAG)-G37C expressed in yeast cells expressing DTWD2A (red) or a vector control (blue). Equal amounts of tRNA from indicated samples were analyzed by nucleoside LC-MS/MS. The abundance of acp3U (top; mass transition 346 → 214, 346 → 197, 346 → 168) and uridine (bottom; mass transition 245 → 113) are shown.

(PDF)

Click here for additional data file.

Comparison of acp3U chromatograms from E. coli tRNA nucleosides and A. thaliana tRNALeu(AAG) nucleosides.

Chromatograms of acp3U (mass transition 346 → 214, 346 → 197, 346 → 168) from E. coli tRNA nucleosides (black) and A. thaliana tRNALeu(AAG)-G37C nucleosides from cells expressing DTWD2A (red) are shown.

(PDF)

Click here for additional data file.

Comparison of MS/MS of acp3U in E. coli tRNA nucleosides and A. thaliana tRNALeu(AAG) nucleosides.

MS/MS of acp3U (mass transition 346 → 214, 346 → 197, 346 → 168) from E. coli tRNA nucleosides (top) and A. thaliana tRNALeu(AAG)-G37C nucleosides from cells expressing DTWD2A (bottom) are shown.

(PDF)

Click here for additional data file.

(PDF)

Click here for additional data file.