Identification of human tRNA:m⁵C methyltransferase catalysing intron-dependent m⁵C formation in the first position of the anticodon of the ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$

Abstract

We identified a human orthologue of tRNA:m⁵C methyltransferase from Saccharomyces cerevisiae, which has been previously shown to catalyse the specific modification of C₃₄ in the intron-containing yeast ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$. Using transcripts of intron-less and intron-containing human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ genes as substrates, we have shown that m⁵C₃₄ is introduced only in the intron-containing tRNA precursors when the substrates were incubated in the HeLa extract. m⁵C₃₄ formation depends on the nucleotide sequence surrounding the wobble cytidine and on the structure of the prolongated anticodon stem. Expression of the human Trm4 (hTrm4) cDNA in yeast partially complements the lack of the endogenous Trm4p enzyme. The yeast extract prepared from the strain deprived of the endogenous TRM4 gene and transformed with hTrm4 cDNA exhibits the same activity and substrate specificity toward human pre-tRNA^Leu transcripts as the HeLa extract. The hTrm4 MTase has a much narrower specificity against the yeast substrates than its yeast orthologue: human enzyme is not able to form m⁵C at positions 48 and 49 of human and yeast tRNA precursors. To our knowledge, this is the first report showing intron-dependent methylation of human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ and identification of human gene encoding tRNA methylase responsible for this reaction.

INTRODUCTION

Maturation of eukaryotic cytoplasmic tRNAs is a multistage process that includes the processing of 5′ and 3′ ends, intron splicing in the case of intron-containing pre-tRNAs, transport from the nucleus to the cytoplasm and numerous nucleoside modifications that take place both in the nucleus and in the cytoplasm. Three classes of intron-containing tRNA genes are present in the human nuclear genome: ${\text{tRNA}}_{(\text{G}\Psi \text{A})}^{\text{Tyr}}$ (8 genes with intron lengths ranging from 16 to 21 bp), ${\text{tRNA}}_{{(\text{m}}^{\text{5}}\text{CAA})}^{\text{Leu}}$ (5 genes with intron lengths ranging from 22 to 25 bp) and ${\text{tRNA}}_{\left(\text{UCU}\right)}^{\text{Arg}}$ (1 gene containing a 15 bp intron) (Genomic tRNA database, http://lowelab.ucsc.edu/GtRNAdb). In all cases, introns are located one nucleotide downstream from the anticodon, which is a typical feature of nuclear intron-containing tRNA genes.

In the case of ${\text{tRNA}}_{(\text{G}\Psi \text{A})}^{\text{Tyr}}$ genes, it has been shown that the introduction of pseudouridine in the middle position of the anticodon (Ψ₃₅) is intron dependent in yeast, plants, animals and humans (1–3). Moreover, it has been demonstrated that the introduction of Ψ₃₅ strictly depends on the nucleotide sequence surrounding the U₃₅ to be modified and depends rather on the length of the intron than on its structure (4). Transfer ${\text{RNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ genes contain introns in yeast and vertebrates (5,6). However, in plants, there are no introns in these genes (7). The first position of the yeast anticodon sequence—cytosine—in ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ is methylated to 5-methylcytosine (m⁵C)₃₄. In yeast, this methylation is intron dependent and its introduction strictly depends on the intron structure (5).

The first functional evidence of the importance of Ψ₃₅ in tRNA^Tyr and m⁵C₃₄ in tRNA^Leu came from the Abelson laboratory (1,5). The construction of mutant yeast tRNA^Tyr and tRNA^Leu genes without introns resulted in the production of mature tRNA molecules without appropriate modified bases. When the suppressor activity of a tRNA^Tyr product derived from intron-containing tRNA (SUP 6 coding for ochre suppressor tRNA) with the ochre suppressor tRNA transcribed from the gene without intron was compared, it turned out that tRNA^Tyr derived from the mutated gene without intron exhibited a strong reduction in the suppressor activity. In an analogous experiment, a mutant, intron-less yeast tRNA^Leu gene encoding amber suppressor tRNA (SUP 53) also gave a product with a weak suppressor activity when compared with the intron-containing amber tRNA^Leu suppressor gene product. In both cases, the decreased suppressor activity correlated with the absence of Ψ₃₅ and m⁵C₃₄, respectively. Similarly, Ψ₃₅ is required in the anticodon (GΨA) of cytoplasmic tRNA^Tyr from Nicotiana rustica for UAG and UAA suppression in the translation of the tobacco mosaic virus (TMV) RNA (8). Summarizing the afore-mentioned experiments, it is clear that both Ψ₃₅ and m⁵C₃₄ are necessary to stabilize anticodon–codon pairing leading to the correct decoding of mRNA. In this paper, we show that intron in the human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ is indispensable for the C5-methylation of cytosine in the first position of the anticodon. We also show that the modification of C₃₄ depends on the nucleotide sequence surrounding the position to be modified and on the structure of intron-containing prolongated anticodon stem.

Enzymes that introduce Ψ₃₅ and m⁵C₃₄ in the intron-containing pre-tRNA^Tyr and pre-tRNA^Leu, respectively, have been identified in yeast (9,10). However, our knowledge about the corresponding enzymes in multicellular eukaryotes remains obscure. The first human m⁵C methylase Dnmt2 has been reported very recently and is responsible for the modification of C₃₈ in tRNA^Asp in mice, Drosophila melanogaster and Arabidopsis thaliana. This MTase was found to be localized in cytoplasm and acts at the level of a mature tRNA molecule (11). Here, we present the characterization of a human gene encoding a methyltransferase (MTase) that is involved in the formation of m⁵C₃₄ in ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$. The product of the gene, hTrm4, acts at the level of the intron-containing tRNA precursor and is localized in the nucleoplasm and nucleolus. During the preparation of this paper, it was found that this MTase is a novel downstream Myc-target which mediates Myc-induced cell proliferation and growth. Therefore, the characterization of hTrm4 substrate specificity can be essential since this MTase is a potential target for cancer therapies (12).

MATERIALS AND METHODS

Enzymes and reagents

The restriction enzymes, T4 DNA polymerase and T4 DNA ligase were from Roche, Fermentas and Promega. RNase T₂ and SAM (S-adenosyl-l-methionine) were obtained from Sigma. [α-³²P]NTPs were from Amersham/Pharmacia Biotech. Taq polymerase, T7 RNA polymerase, Calf Intestine Alkaline Phosphatase (CIAP), dNTPs and NTPs were from Fermentas. P1 nuclease was obtained from Roche. Marathon-Ready cDNA library was obtained from the Clontech, TAKARA BIO company. DNA sequencing was carried out using fmol DNA Cycle Sequencing System from Promega and CEQ DTCS Quick Start Kit from Beckman Coulter. Oligonucleotides for amplification and mutagenesis were synthesized by oligo.pl. Cellulose plates for TLC were purchased from Merck. Kits for plasmid isolation and for DNA isolation from agarose were obtained from A&A Biotechnology. The QuickChange Site-Directed Mutagenesis Kit was from Stratagene. All other reagents were from Amersham/Pharmacia Biotech, Fermentas and Sigma.

Bacterial, yeast and HeLa strains and plasmids

Escherichia coli TG1, CJ236 and JM109 were used as hosts for the propagation of pUC19 plasmid and M13mp19 phage derivatives, respectively. The recombinant plasmids used in this study were: pHLIVS2 carrying the human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ gene and pYLIVS carrying the yeast ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ gene, respectively. Both tRNA genes were inserted under T7 RNA polymerase promoter and contained the MvaI restriction site to linearize plasmids and terminate the in vitro transcription. Human HeLa cells were used for nuclear cell-free extract preparations. Saccharomyces cerevisiae haploid strain BY4742 (wild type) and its deletion derivative with the open reading frame YBL024w (corresponding to the TRM4 gene) replaced by kanMX4, a kanamycin resistance gene mediating G418 resistance, were obtained from EUROSCARF. Yeast strains were used for transformation and tRNA:m⁵C methyltransferase activity assays. pYES2 (Invitrogen) vector was utilized for protein expression in yeast.

Construction of tRNA^Leu gene mutants

The recombinant plasmid pHLIVS2 harbouring human tRNA^Leu gene that contains the 22 bp intron was mutagenized using the Kunkel method (13) or QuickChange Site-Directed Mutagenesis Kit according to the manufacturer's instructions. The introduced mutations were confirmed by DNA sequencing. The mini-helix consisting of the prolongated anticodon stem of ${\text{tRNA}}_{\text{CAA}}^{\text{Leu}}$ gene was synthesized according to the Milligan's method (14) with modifications described in ref. (15) using two DNA oligonucleotides: one (58 nt long) including the 17 nt long sequence of T7 RNA polymerase promoter followed by the coding sequence of prolongated anticodon stem containing 22 nt long intron and the other with the sequence complementary to the T7 RNA promoter.

T7 RNA transcription of ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ gene and its mutants

Plasmids carrying the yeast tRNA^Leu gene and the human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ gene or its mutants (with the exception of mutant pHLIVS2 C₃₅) were linearized by digestion with MvaI before transcription. Plasmid pHLIVS2 C₃₅ contains a mutation that introduces a new MvaI restriction site into the tRNA^Leu mutant gene and was linearized with BamHI. In vitro transcription was carried out in Fermentas T7 RNA polymerase buffer, 500 μM each of CTP, GTP, UTP, 12 μM ATP, 50 μCi of [α-³²P]ATP, 1 μg of linearized plasmid and 20 U of T7 RNA polymerase in total volume of 20 μl at 37°C for 1.5 h. Radiolabelled ATP was added in the case of the wild-type gene and all the mutant transcription reactions with the exception of pHLIVS2 C₃₅, pHLIVS2 G₃₅ and pHLIVS2 C₃₅ where [α-³²P]CTP, [α-³²P]GTP or [α-³²P]UTP were added, respectively. For the visualization of m⁵C localized at position 48/49 in tRNAs plasmids carrying the yeast ${\text{tRNA}}_{\left(\text{GUC}\right)}^{\text{Asp}}$, ${\text{tRNA}}_{\left(\text{AGA}\right)}^{\text{Ser}}$ and the human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ genes were transcribed using [α-³²P]GTP. Transcripts were purified on 10% denaturating polyacrylamide gels, identified by autoradiography, excised from the gels and eluted with 0.3 M sodium acetate (pH 5.2), 0.5 M EDTA, 0.1% (w/w) SDS. After phenol/chloroform extraction and ethanol precipitation, the RNA was resuspended in water. The average amount of the recovered labelled tRNA transcripts varied between 1 and 2 pmol per reaction. The transcription of mini-helix was performed as described in (14,15).

In vitro 5-methylcytosine formation assays in HeLa and yeast extract

A nuclear cell-free extract was prepared from HeLa cells according to Dignam et al. (16). Yeast cells used for protein extract preparations were grown as follows: single yeast colony was picked from YPD agar plate and inoculated into 15 ml SD-U synthetic medium containing 0.67% nitrogen base, 2% glucose, amino acids and no uracil. The culture was grown overnight at 30°C with shaking. The OD₆₀₀ of the overnight culture was determined and the amount of culture necessary to obtain an OD₆₀₀ of 0.4 in 50 ml of induction medium was calculated and taken for further experiments. Cells were pelleted at 1500 g for 5 min at 4°C, resuspended in 2 ml of induction medium (SD-U medium containing 2% galactose instead of 2% glucose), inoculated into 50 ml of the induction medium and grown overnight at 30°C with shaking. Yeast S-10 extract was prepared by the following procedure: 5 ml of cell suspension were centrifuged, collected, washed twice in 500 μl of distilled water and resuspended in twice their volume of lysis buffer [50 mM Tris–HCl, pH 7.5, 100 mM KCl, 0.1 mM EDTA, 10 mM MgCl₂, 10% glycerol, 2% Protease Inhibitor Cocktail for yeast (Sigma)]. One volume of glass beads was added. The mixture was frozen in liquid nitrogen and thawed at room temperature. The suspension was vigorously vortexed 5–7 times for 1 min with 5 min pause between each vortexing when the suspension was kept on ice. The resulting homogenate was centrifuged at 12 000 g for 20 min to remove cell debris. The resulting supernatant was aliquoted, quickly frozen in liquid nitrogen and kept frozen at −70°C for later use.

Quantification of m⁵C₃₄ in ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$

About 3000 c.p.m. of ³²P-labelled RNAs, obtained as described above, were digested with 0.5 U of RNase T₂ for 5 h at 37°C in 10 μl of 5 mM ammonium acetate, pH 4.6. The labelled nucleotides were identified by a two-dimensional (2D) chromatographic analysis on cellulose TLC plates as described by Grosjean et al. (17). The efficiency of m⁵C₃₄ formation was measured by cutting out the labelled spots from the TLC plates and counting the radioactivity by the liquid scintillation technique. Otherwise, radioactivity in the spots was evaluated by ImageQuant 5.1 software after exposure to a PhosphorImager screen (Typhoon, Molecular Dynamics, USA) and scanning. Taking into account the amount of label in each of the nucleotide spots (AMP, UMP, GMP, CMP and/or m⁵CMP) and knowing the relative number of labelled nucleotides per RNA substrate, the number of moles of m⁵C per mole of pre-tRNA was easily calculated.

Cloning of yeast and human TRM4 cDNA

Yeast Trm4 protein was used as a query in searching for sequences of homologous proteins in the human expressed sequence tags (ESTs) database using tBLASTn (18). For the localization of the gene in human genome and the determination of the chromosomal localization, BLAT software was used.

hTrm4 cDNA was amplified from the human cDNA library using following primers: 5′-ATGGTGGTCAACCATGATGCCTCCAGCATAC-3′ and 5′-TCACCGGGGTGGATGGACCCCCGCCGG-3′. PCR was performed in two step ‘touchdown’ manner, with the following conditions: 30 s at 96°C, followed by 25 cycles of 5 s at 96°C and 4 min at 72°C. The PCR product of expected size was excised from the agarose gel and eluted. The cDNA product was re-amplified with primers harbouring sequences for XbaI and EcoRI restriction enzymes and cloned into the pUC-19 vector (giving hTRM4/pUC construct) and sequenced. Yeast genomic DNA was isolated using QIAGEN Genomic DNA kit according to manufacturer's instructions. Primers used for the amplification of yeast Trm4p cDNA were: 5′-AAAGGATCCACCATGGCTAGAAGAAAGAATTTCAA-3′ (Kozak translation initiation sequence is marked in italics, start codon is underlined and the BamHI restriction site is in boldface) and 5′-AAAGAATTCTCAATTAGCAGCGCTAGGAG-3′ (the stop codon is underlined and EcoRI restriction sequence is in boldface). After amplification, a product of expected size was excised and eluted from the EtBr/agarose gel. Yeast and human TRM4 cDNA were cloned into pYES2 yeast expression vector (constructs TRM4/pYES2 and hTRM4/pYES2, respectively). After each cloning procedure, the correctness of insert-vector ligation was assessed by sequencing or restriction digestion.

Yeast cells transformation

Yeast strain ΔYBL024w was made competent by lithium acetate/PEG 3350 treatment according to procedures described in the User Manual for pYES2 vector system (Invitrogen). Competent cells were then transformed by purified pYES2 plasmid or its derivative constructs carrying the sequence corresponding to the yeast TRM4 gene or human TRM4 cDNA, respectively. Transformed yeast cells were grown on a synthetic selective medium as required to maintain plasmids.

Preparation of low-molecular weight RNA from S.cerevisiae

Yeast RNA was extracted as described in (19). RNA products were purified on 8% denaturing polyacrylamide gel. After EtBr staining, bands corresponding to approximate tRNA size (80–110 nt) were eluted from the gel, ethanol precipitated, carefully washed with ethanol/water (70:30), resuspended in DEPC-treated water and kept frozen at −70°C for later use.

HPLC chromatography

Low-molecular RNA (10–20 μg) was digested by P1 nuclease according to the manufacturer's manual (Roche). Nucleotides were dephosphorylated by CIAP phosphatase. The volume of the reaction was reduced to 10 μl by vacuum centrifugation. HPLC separation was performed on the reversed phase SUPELCOSIL™ LC-318 HPLC Column, 25 cm × 4.6 mm (Supelco) attached to ÄKTAexplorer 10S (Amersham Pharmacia Biotech). Chromatographic solvents were as follows: (A) 100 mM ammonium acetate and (B) 100% acetonitrile. HPLC conditions were: 4°C, 1 ml/min flow rate and absorbance measuring at λ = 277 nm. The gradient method consisted of 2 CV (Column Volume) solvent A, a gradient >10 CV to 6% solvent B, held 2 CV solvent B and column re-equilibration. To determine retention times of standards, 10 μl mix of six ribonucleosides (pseudouridine, cytidine, uridine, 5-methylcytidine, guanosine and adenosine), each at the concentration of 1 mM, was separated.

RESULTS

m⁵C₃₄ in ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ is formed only in intron-containing precursors

Green et al. (20) isolated five ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ genes that slightly differ in intron sequence. Inspection of the Genomic tRNA database revealed that there are five human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ genes containing introns with nucleotide sequences identical or nearly identical to those published by Green et al (20) and Karwowska and Szweykowska-Kulinska (6).

The only human ${\text{tRNA}}_{(\text{C}*\text{AA})}^{\text{Leu}}$ sequence with the identified modified nucleosides is known from HeLa cells (21). Its nucleotide sequence corresponds exactly to the nucleotide sequence of the mature domain of the isolated human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ genes from HeLa cells with the exception of one additional G nucleotide in the tRNA^Leu gene extra arm coding sequence (6). However, Harada et al. (21) was not able to identify the nature of modified cytidine in position 34. Storbel and Abelson (5) have shown that this modification in the yeast tRNA corresponds to m⁵C and that its formation is intron-dependent. Using PCR, we have previously isolated two intron-containing ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ genes from human blood (6). One of the isolated human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ genes was studied further. It contains 22 bp intron. The putative structure of pre-tRNA^Leu derived from this gene is shown in Figure 1a. The structure of the prolongated anticodon stem containing intron is proposed according to the model of yeast intron-containing ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ (5). The putative model of the extended anticodon stem shows that (i) the second (A₃₅) and third (A₃₆) anticodon nucleotides and G₃₇ can be base-paired with the complementary nucleotides of the intron; (ii) the majority of the intron can be arranged in two hairpin-stem and loop structures (arm I and arm II, see Figure 1); (iii) the 5′ splice site junction is located at the boundary of the stem–loop of the first arm in the prolongated anticodon stem; (iv) the 3′ splice site junction is located in the unpaired region. To test whether C₃₄ in human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ is modified to m⁵C and whether its introduction depends on the presence of 22 nt long intron, the tRNA^Leu gene with the intron and a mutant of the tRNA^Leu gene without (Figure 1b) were transcribed in vitro using T7 RNA polymerase and [5′-α³²P]ATP. Radiolabelled transcripts were then incubated in the presence of nuclear, cell-free HeLa extract which contained the activity necessary for m⁵C synthesis. The modified transcripts were gel-purified and analysed for their m⁵C content after a complete hydrolysis of RNA with RNase T₂. Since RNase T₂ produces nucleoside-3′-monophosphates, the labelled phosphate is exclusively derived from the nearest adenosine. This ensures that only five cytidine nucleotides will be labelled in the tRNA^Leu without intron and six in the intron-containing precursor (Figure 1), since they have 3′-flanking adenosines. The only m⁵C that was found in the sequenced human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ was C₄₈, but this nucleoside is accompanied at the 3′ side by guanosine and thus C₄₈ is not labelled (21). Nucleotides were resolved in a thin-layer two-dimensional chromatographic system that separates modified nucleotides from their unmodified counterparts. Figure 2 shows the autoradiograms of T₂ hydrolysates derived from intron-less and intron-containing pre-tRNA^Leu. Only pre-tRNA containing intron gives rise to 3′-m⁵CMP (Figure 2a and b). It shows that the only cytosine that can be methylated in the intron-containing pre-tRNA^Leu is C₃₄. Thus, the first cytosine of the anticodon of human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ is methylated in intron-dependent manner.

Figure 1.

Cloverleaf structure of ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ precursor containing intron (a), mature tRNA (b) and two mutants with partially deleted introns (derived from plasmids pHLIVS2 ΔI1, pHLIVS2 ΔI2) (c and d). Anticodon sequence is underlined, arrows point to splice sites.

Figure 2.

Analysis of m⁵C₃₄ content in intron-less (a) and intron-containing (b) human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ by two-dimensional thin layer chromatography. Pre-tRNAs were incubated in the cell-free nuclear HeLa extract. tRNA^Leu precursors were synthesized and treated as described in Materials and Methods.

The structure of a prolongated anticodon stem and nucleotide sequence that surrounds the position to be modified are crucial for m⁵C₃₄ formation in ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$

To learn more about the sequence and structure specificity of m⁵C₃₄ methylation in tRNA^Leu precursor, a number of mutants were generated where wild-type nucleotides occupying positions 32, 33, 35, 36 and 37 were exchanged for other nucleotides. Moreover, two mutants were constructed with partially deleted intron sequences (Figure 1c and d) and four mutants in which either the base-pairing in the prolongated anticodon stem was disrupted or base pairs were reversed. All mutant pre-tRNAs listed in Table 1 were examined for their capacity to be substrates for m⁵C₃₄ formation in HeLa cell-free nuclear extract.

Table 1.

Efficiency of C₃₄ to m⁵C₃₄ conversion in HeLa extract

Mutant name	Mutation site	Relative amount of m5C34 (%)
pHLIVS2	wt	100
Mini-helix	pre-tRNALeu anticodon stem-containing intron	0 (undetectable)
pHLIVS2 Δ I	Intron deleted	0
pHLIVS2 Δ I 1	Intron arm I deleted	0
pHLIVS2 ΔI 2	Intron arm II deleted	0
pHLIVS2 A32	C32 → A32	85
pHLIVS2 G32	C32 → G32	0
pHLIVS2 U32	C32 → U32	85
pHLIVS2 A33	U33 → A33	85
pHLIVS2 C33	U33 → C33	10
pHLIVS2 G33	U33 → G33	10
pHLIVS2 C35	A35 → C35	0
pHLIVS2 G35	A35 → G35	0
pHLIVS2 U35	A35 → U35	0
pHLIVS2 C36	A36 → C36	0
pHLIVS2 G36	A36 → G36	0
pHLIVS2 U36	A36 → U36	0
pHLIVS2 A37	G37 → A37	0
pHLIVS2 C37	G37 → C37	0
pHLIVS2 U35A11i	A35 → U35, U11i → A11i	0
pHLIVS2 U37	G37 → U37	0
pHLIVS2 G36Ci10	A36 → G36, Ui10 → Ci10	85
pHLIVS2 C37Gi9	G37 → C37, Ci9 → Gi9	0
pHLIVS2 Gi9Ai10Ai11	Ci9 → Gi9, Ui10 → Ai10, Ui11 → Ai11	0

³²P-labelled pre-tRNAs^Leu were incubated in cell-free nuclear extract for 20 min at 30°C. The intron-containing pre-tRNA^Leu and its mutants were analysed for the presence of m⁵C₃₄ after T₂ digestion by thin layer chromatography as described in Materials and Methods. The efficiency of m⁵C₃₄ formation was calculated as described in Materials and Methods.

A mini-helix consisting of prolongated anticodon stem was not methylated at the C₃₄ position. The removal of the intron arm I or II resulted in the complete abolishment of C₃₄ methylation. Thus, m⁵C₃₄ formation depends on the structure of the intron-containing prolongated anticodon stem. The exchange of C₃₂ to A or U had no major effect on the efficiency of m⁵C₃₄ formation (85% of the wild-type efficiency). However, the exchange of C₃₂ to G resulted in the inhibition of m⁵C₃₄ methylation. The transversion of U₃₃ to A did not significantly influence the formation of m⁵C₃₄. However, the transversion of U₃₃ to G and transition to C strongly reduced the level of m⁵C₃₄ formation. All mutations in positions 35, 36 and 37 resulted in m⁵C₃₄-deficient pre-tRNAs^Leu. From these experiments, a consensus sequence for m⁵C₃₄ methylation in pre-tRNA^Leu was established: C/A/U₃₂-U/A₃₃-C₃₄-A₃₅-A₃₆-G₃₇. Careful inspection of the prolongated anticodon stem secondary structure reveals that A₃₅, A₃₆ and G₃₇ can base pair with intron nucleotides U_i11U_i10 and C_i9, respectively. To learn more about the sequence and structure dependence of m⁵C₃₄ methylation within the prolongated anticodon stem, additional mutations were introduced in which potential base pairs were reversed or base-pairing was disrupted. The reversion of base pair A₃₅–U_i11 to U₃₅–A_i11 did not promote m⁵C₃₄ formation. The reversion of base pair A₃₆–U_i10 to G₃₆–C_i10 resulted in 85% efficiency of m⁵C₃₄ formation, showing that no A₃₆ per se is indispensable for modification but N₃₆–N_i10 base-pairing within the moiety of the prolongated anticodon stem is crucial. However, like in the case of the base pair A₃₅–U_i11, the reversion of G₃₇–C_i9 to C₃₇–G_i9 resulted in a complete abolishment of m⁵C₃₄ formation. In this case, not the base-pairing but the identity of guanosine in position 37 (and adenosine in the position 35), and possibly their pairing with C_i9 and U_i10, respectively, is crucial for the modification to occur. Disruption of all three base pairs A₃₆–U_i11, A₃₇–U_i10, and G₃₇–C_i9 by the introduction of noncomplementary nucleotides within the intron sequence, i.e. A_i11, A_i10 and G_i9, leads to the inhibition of m⁵C₃₄ formation. Thus, both the sequence and structure of the prolongated anticodon stem are crucial for the introduction of methyl group on C₃₄ of pre-tRNA^Leu.

NSUN2 encodes yeast TRM4 orthologue

The BLAST similarity search of human UniGene database at NCBI (22) using the yeast Trm4p (enzyme that introduces m⁵C into yeast ${\text{pre-tRNA}}_{\text{CAA}}^{\text{Leu}}$ in an intron-dependent manner) as a query enabled us to identify EST sequences of a putative human orthologue. The putative protein sequence encoded by the contig sequence, constructed from identified ESTs using the CAP3 program (23), shows 35% identity to the yeast sequence. A comparison of the contig sequence with the human genome revealed only one gene with an identical putative coding sequence (NCBI accession no. NM_017755), located on the human chromosome 5, with precise position 5p15.31. This gene is called NSUN according to HUGO Gene Nomenclature Committee and contains 19 exons, and encompasses ∼34 000 nt. Based on the results of the analyses reported in this article, we propose to rename it as hTRM4. Bioinformatic analysis of amino acid sequence of putative m⁵C₃₄ MTases and the alignment of regions forming the active site in yeast and human Trm4 enzymes were already published by Bujnicki et al. (24). Sequence analysis using the programs predictNLS (25), SignalP (26) and PSORT II (http://psort.hgc.jp) revealed 60–70% probability of nuclear localization of hTrm4. However, we did not find any classical nuclear localization signals (NLS) within the protein sequence.

Using the cDNA library from HeLa cells, we isolated a 2304 bp cDNA encoding hTrm4 protein. Its nucleotide sequence was confirmed to be identical to the virtually constructed cDNA contig from EST libraries and to exon sequences of the identified human gene.

Human Trm4 catalyses the formation of m⁵C₃₄ in human ${\text{pre-tRNA}}_{\text{CAA}}^{\text{Leu}}$

Since our efforts to overexpress the hTrm4 protein in different prokaryotic expression systems were unsuccessful (data not shown), we decided to investigate hTrm4 activity in yeast strain BY4742 with the deleted endogenous TRM4 gene. The deletion of this gene has no phenotypic effect on cell growth (27). Δtrm4 yeast strain BY4742 was transformed with hTrm4 cDNA that was cloned into yeast pYES2 plasmid. The same yeast strain was also transformed with pYES2 plasmid containing yeast Trm4p cDNA. Low-molecular weight RNA was isolated from both yeast lines, digested with P1 nuclease and digestion products were dephosphorylated. Nucleosides were separated on HPLC chromatography using RPC18 column. Figure 3 presents the chromatographic elution profiles of nucleosides from yeast lines: wild type, Δtrm4, Δtrm4 transformed with plasmid carrying hTrm4 cDNA and Δtrm4 transformed with plasmid carrying yeast Trm4p cDNA. In the elution profile of nucleosides obtained from low-molecular weight RNA from Δtrm4 + yeast Trm4p, the level of m⁵C was restored to that achieved for m⁵C for wild-type yeast strain. In the Δtrm4 yeast strain, m⁵C nucleosides are undetectable. In the case of Δtrm4 + hTrm4 cDNA, we observed only a low level of m⁵C. However, this small peak appeared repeatedly in all consecutive experiments. The low level of m⁵C methylation may be due to several (mutually not exclusive) reasons: (i) hTrm4 protein recognizes a small group of substrates that are normally methylated by the yeast Trm4p enzyme, (ii) the expressed hTrm4 enzyme has only residual activity, e.g. because of non-native folding or post-translational modification or (iii) hTrm4 localization in yeast is not fully correct. To test whether hTrm4 exhibits activity toward human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ containing intron, we prepared total protein extracts from Δtrm4, Δtrm4 + hTrm4 cDNA, Δtrm4 + Trm4p cDNA and wild-type yeast lines, and conducted activity tests similar to those in the HeLa cell-free nuclear extract. Figure 4 shows the m⁵C₃₄ analyses of human intron-less and intron-containing ${\text{pre-tRNAs}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ in the wild-type yeast strain (Figure 4a), Δtrm4 (Figure 4b), Δtrm4 + hTrm4 cDNA (Figure 4c) and Δtrm4 + Trm4p cDNA yeast lines (Figure 4d). In the case of the wild-type yeast strain, human pre-tRNA^Leu containing intron is a substrate for endogenous MTase while pre-tRNA^Leu without intron is not. When the protein extract from Δtrm4 yeast line was used, there was no m⁵C in intron-containing and intron-less pre-tRNAs. However, when we used the protein extract prepared from Δtrm4 + Trm4p cDNA, human intron-containing pre-tRNA^Leu was methylated at C₃₄, demonstrating that in the wild-type yeast strain TRM4 MTase was responsible for m⁵C₃₄ formation. In the case of Δtrm4 + hTrm4 cDNA, human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ was methylated when its precursor contained an intron. Finally, yeast ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ containing intron is methylated at C₃₄ in the S-10 yeast extract prepared from the Δtrm4 + hTrm4 strain (Figure 4e). Figure 5 shows the kinetics of m⁵C₃₄ formation in human intron-containing ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ in the extract prepared from the Δtrm4 + hTrm4 cDNA yeast line. The efficiency of m⁵C₃₄ formation in yeast pre-tRNA^Leu in the same yeast line is about the same (∼45%, data not shown). Several mutants of the human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ were tested in yeast extracts containing the human Trm4 enzyme. In all cases, we obtained similar results as in the case of HeLa cell-free nuclear extract (data not shown). These biochemical experiments definitely prove that the human gene identified by us encodes a functional counterpart of the yeast Trm4p enzyme.

Figure 3.

Detection of m⁵C in bulk low-molecular weight RNA extracted from yeast wild-type strain BY4742 (b), BY4742 ΔTRM4 (c), BY4742 transformed with pYES2 containing yeast Trm4p cDNA (d) and BY4742 ΔTRM4 transformed with pYES2 containing hTrm4 cDNA (e). Separation of a test nucleoside mixture (Ψ, U, C, m⁵C, G and A) is presented in (a). The total RNA fraction was completely hydrolysed by nuclease P1 and the resulting nucleotides dephosphorylated by CIAP. Analysis of nucleosides was performed as described in Materials and Methods.

Figure 4.

Analysis of m⁵C₃₄ content in human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ without intron and containing intron by 2-D thin layer chromatography. Pre-tRNAs were incubated in yeast S-10 extract derived from wild-type strain (a), BY4742 ΔTRM4 (b), BY4742 ΔTRM4 transformed with pYES2-containing hTrm4 cDNA (c) and BY4742 ΔTRM4 transformed with pYES2-containing yeast Trm4p cDNA (d). (e) Incubation of yeast ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ containing intron in S-10 extract prepared from BY4742 ΔTRM4 transformed with pYES2-containing hTrm4 cDNA. tRNA^Leu precursors were synthesized and treated as described in Materials and Methods.

Figure 5.

Kinetics of m⁵C₃₄ formation in human intron-containing ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$. ³²P-labelled pre-tRNA^Leu transcripts were incubated with S-10 yeast extract prepared from BY4742 Δtrm4 + hTrm4 cDNA at 30°C for 10, 20, 40 and 90 min and analyzed for the relative amount of m⁵C residues at position 34. The measurements were taken as described in Materials and Methods.

Human Trm4 has narrower substrate specificity than its yeast orthologue

Apart from methylating C₃₄ in yeast pre-tRNA^Leu containing intron, yeast Trm4p is required for C₄₈ and C₄₉ methylation in different yeast tRNAs. Using total protein extracts from yeast lines described in the preceding chapter, we examined hTrm4 ability to methylate cytosines at positions 48/49. hTrm4 is not able to introduce a methyl group at position C₄₈ of human tRNA^Leu, both at the level of intron-containing and intron-less precursors (Figure 6a), while the yeast Trm4p methylates human intron-less precursor and does not recognize intron-containing substrate (Figure 6b). Additionally, we analysed m⁵C₄₈ and m⁵C₄₉ formation in yeast ${\text{tRNA}}_{\left(\text{AGA}\right)}^{\text{Ser}}$ and ${\text{tRNA}}_{\left(\text{GUC}\right)}^{\text{Asp}}$, respectively. We show that neither of them is methylated at C₄₈ or C₄₉ by hTrm4 and both are substrates for yeast Trm4p (Figure 6c and d). Control experiments carried out in the Δtrm4 strain show no detection of m⁵C_48/49 formation in any tRNA tested (data not shown). Thus, human Trm4 substrate specificity is narrower than in the case of its yeast orthologue.

Figure 6.

Analysis of m⁵C₄₈ (a–c) and m⁵C₄₉ (d) content in human and yeast tRNAs using hTrm4 and yeast Trm4p enzymes by 2D TLC. Pre-tRNAs were labelled using [α-³²P] GTP during T7 RNA polymerase transcription, gel purified and incubated in S-10 extract. (a) BY4742 Δtrm4 + hTrm4 yeast strain and human pre-tRNA^Leu, (b) BY4742 Δtrm4 + yeast Trm4p and human pre-tRNA^Leu, (c) BY4742 Δtrm4 + hTrm4 (left) and BY4742 Δtrm4 + Trm4p (right) and yeast ${\text{tRNA}}_{\left(\text{AGA}\right)}^{\text{Ser}}$, and (d) BY4742 Δtrm4 + hTrm4 (left), BY4742 Δtrm4 + Trm4p (right) and yeast ${\text{tRNA}}_{\left(\text{GUC}\right)}^{\text{Asp}}$. Yeast ${\text{tRNA}}_{\left(\text{GUC}\right)}^{\text{Asp}}$ and ${\text{tRNA}}_{\left(\text{AGA}\right)}^{\text{Ser}}$ genes do not contain introns.

DISCUSSION

In this article, we describe the identification, cloning and enzyme activity of the corresponding gene product encoding human Trm4 enzyme. This is the first report showing intron-dependent methylation of human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ and identification of human gene encoding tRNA methylase responsible for this reaction. The hTRM4 gene was identified on the basis of its amino acid sequence similarity to a known Trm4p MTase from yeast. Our studies provide experimental validation of the phylogenetic analysis of RNA:m⁵C MTases by Bujnicki et al. who predicted the product of the NSUN2 gene as the closest human homologue of yeast Trm4p (24). hTRM4 is a single-copy gene, located on the human chromosome 5, with the precise position 5p15.31. The hTrm4 protein is a member of an evolutionarily conserved family, which includes genuine and putative RNA:m⁵C MTases from all three domains of life, such as the human nucleolar proliferation-associated antigen p120 (NOL1) and its homologues. Wu et al. (27) reported that the yeast Trm4p was found to be localized in the nucleus, including the nucleolus, and was concentrated at the nuclear periphery. Our experiments have shown that its human homologue hTrm4 is always localized in the nucleoplasm and in nucleoli of HeLa cells, and in several cases we also observed a concentration of hTrm4 at the nuclear periphery (data not shown). These data have been recently confirmed by Frye and Watt (12).

The presence of an intron is required for the m⁵C₃₄ formation in human ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ as demonstrated in this paper by the maturation of intron-less and intron-containing tRNA precursor in HeLa cell-free nuclear extract. In earlier studies, Harada et al. (21) showed that this position in the mature human ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ is occupied by a hypermodified derivative. The authors suggest that this is a 2′-O-methylated and further hypermodified cytidine derivative (21). Very similar results were obtained by Randerath et al. (28) when leucine tRNAs were isolated and sequenced from rat tumour, Morris hepatoma 5123D cells. Our studies show that the first stage of C₃₄ modification process in HeLa cells is the m⁵C₃₄ formation. However, the final chemical structure of this modification remains to be characterized. In addition to intron-dependent m⁵C₃₄ formation, we have shown that the activity of hTrm4 strictly depends on a specific sequence surrounding the cytosine to be modified and on the intron structure. Similar results were obtained studying the yeast Trm4p enzyme by Storbel and Abelson (5). However, our experiments have shown that minisubstrate composed of the anticodon stem–loop extended by the intron was not methylated at C₃₄. In contrast, Motorin and Grosjean (10) presented data that analogous minihelix of yeast ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ was an efficient substrate for tRNA:m⁵C-methyltransferase. These results imply that there are differences in the ability of enzyme–substrate recognition between the human and fungi Trm4 enzymes. The nucleotide sequence that surrounds C₃₄ in human tRNA precursor is more restrictive for m⁵C₃₄ formation than in the case of yeast. The substitution of adenosine from the middle position of the anticodon to uridine did not promote m⁵C₃₄ formation in human tRNA precursor while in the case of yeast amber leucine-inserting pre-tRNA containing intron (anticodon CUA), m⁵C₃₄ was formed. Additionally, our experiments have shown that the reversion of base pair A₃₅–U_i11 to U₃₅–A_i11 also resulted in the m⁵C₃₄ deficient ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$. Moreover, human MTase responsible for the introduction of methyl group at the anticodon wooble cytosine (hTrm4) requires a consensus nucleotide sequence C/A/U₃₂-U/A₃₃-C₃₄-A₃₅-A₃₆-G₃₇ as a prerequisite for m⁵C₃₄ formation. Thus, all nucleotides from the anticodon loop that are present upstream from the intron sequence are essential for pre-tRNA^Leu modification. These results again strengthen the conclusion about the differences in the ability of enzyme–substrate recognition between human and yeast Trm4 enzymes.

Motorin and Grosjean showed (10) that the yeast Trm4p enzyme catalysed m⁵C formation at four distinct sites in different tRNAs: 34, 40, 48 and 49. m⁵C was found at positions 34 and 40 only in two yeast tRNAs [${\text{tRNA}}_{\left(\text{CUA}\right)}^{\text{Leu}}$ (Sup53) and ${\text{tRNA}}_{\left(\text{GAA}\right)}^{\text{Phe}}$, respectively], whereas most elongator tRNAs bore either m⁵C₄₈ or m⁵C₄₉, but never both in the same tRNA molecule. Motorin and Grosjean concluded that yeast Trm4p MTase is responsible for cytosine conversion in all yeast tRNAs and pre-tRNAs, while two other identified MTases encoded in S.cerevisiae genome are required for proper rRNA m⁵C modification. We transformed yeast strain BY4742 deprived of the TRM4 gene with yeast Trm4p cDNA and human Trm4 cDNA. In the Δtrm4 + Trm4p cDNA strain, the level of m⁵C in low-molecular RNA molecules was restored to that of the wild-type yeast strain (see Figure 3). However, in the Δtrm4 + hTrm4 cDNA strain, the level of m⁵C was hardly detectable. It clearly shows that human enzyme is not able to recognize the majority of yeast tRNA substrates. Since our experiments show that yeast and human Trm4 MTases both recognize mammalian ${\text{pre-tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ containing intron, the conclusion can be drawn that the yeast enzyme has much broader substrate specificity than the human Trm4. We confirm this statement showing that hTrm4 is not able to introduce m⁵C₄₈ at the level of intron-containing and intron-less human pre-tRNA^Leu precursor. Moreover, we show that it does not methylate different yeast tRNA precursors at positions 48 and 49. In the light of these observations, it is interesting to note that Bujnicki et al. (24) identified two additional human putative orthologues of yeast Trm4p, which have probably originated from duplications and subsequent functional specializations of the original TRM4 gene in the animal lineage. One or both of these so-far uncharacterized genes may be responsible for some of the methylations carried out by the yeast Trm4p, but not exhibited by its human counterpart. This speculation can explain a very low level of m⁵C in the Δtrm4 + hTrm4 strain. Summarizing our results concerning the hTrm4 enzyme, it can be concluded that the enzyme acts on intron-containing ${\text{tRNA}}_{\left(\text{CAA}\right)}^{\text{Leu}}$ precursors, it is more sensitive to the nucleotide sequence that surrounds the position to be modified than its yeast homologue Trm4p and it is equally sensitive to the anticodon prolongated, intron-containing structure. Moreover, its sensitivity to substrate specificity is more constrained than in the case of yeast homologue—Trm4p.