|Home | About | Journals | Submit | Contact Us | Français|
The essential Saccharomyces cerevisiae tRNAHis guanylyltransferase (Thg1p) is responsible for the unusual G−1 addition to the 5′ end of cytoplasmic tRNAHis. We report here that tRNAHis from Thg1p-depleted cells is uncharged, although histidyl tRNA synthetase is active and the 3′ end of the tRNA is intact, suggesting that G−1 is a critical determinant for aminoacylation of tRNAHis in vivo. Thg1p depletion leads to activation of the GCN4 pathway, most, but not all, of which is Gcn2p dependent, and to the accumulation of tRNAHis in the nucleus. Surprisingly, tRNAHis in Thg1p-depleted cells accumulates additional m5C modifications, which are delayed relative to the loss of G−1 and aminoacylation. The additional modification is likely due to tRNA m5C methyltransferase Trm4p. We developed a new method to map m5C residues in RNA and localized the additional m5C to positions 48 and 50. This is the first documented example of the accumulation of additional modifications in a eukaryotic tRNA species.
tRNAs from all organisms are extensively modified, with numerous alterations of base and ribose moieties at different positions (47). In the yeast Saccharomyces cerevisiae, 25 different chemical modifications of tRNA have been described. These occur at 34 positions, and each tRNA bears ~11 modifications (22). Although many of these modifications and the corresponding modification enzymes are highly conserved in different organisms, only three modification enzymes are known to be essential in yeast. Gcd10p/Gcd14p catalyzes formation of m1A58, which occurs in numerous tRNA species and is essential for tRNAiMet (3). Tad2p/Tad3p catalyzes adenosine deaminase activity to form I34, which affects seven yeast tRNAs (13). Thg1p catalyzes tRNAHis guanylyltransferase activity to add an extra guanine nucleotide to the 5′ end of tRNAHis at position −1, called G−1 (17).
A unique feature of all tRNAHis species is the presence of an extra G−1 residue opposite the discriminator base at position 73. Among more than 500 characterized tRNA species, only one other tRNA species, tRNAPhe from Tetrahymena pyriformis mitochondria, has an additional residue at its 5′ end (43). The extra G−1 residue of tRNAHis is generated in two very different ways. In prokaryotes, chloroplasts, and plant mitochondria, G−1 is genome encoded and retained in the mature tRNAHis because RNase P cleaves at an unusual position to form an 8-bp aminoacyl stem containing G−1-C73 (7, 34). By contrast, for eukaryotic cytoplasmic tRNAHis and at least one animal mitochondrial tRNAHis, G−1 is not genome encoded (27) but instead is added posttranscriptionally.
We demonstrated previously that the essential protein Thg1p has tRNAHis guanylyltransferase activity in vitro and is required for G−1 addition to tRNAHis in vivo (17). We used a biochemical genomics screen (30) to show that glutathione S-transferase-Thg1p copurifies with yeast tRNAHis guanylyltransferase activity and then showed that yeast Thg1p expressed and purified from Escherichia coli has the correct activity in vitro. This reaction is highly specific for tRNAHis, and the 3′→5′ extension of a polynucleotide chain catalyzed by Thg1p is highly unusual. The only other similar 3′→5′ polynucleotide extension activity is that found in the mitochondria of the amoeboid protist Acanthamoeba castellani (36) and the chytridiomycete fungus Spizellomyces punctatus (6), which are likely responsible for editing the 5′ ends of tRNAs (28). Thg1p adds G−1 to monophosphorylated tRNAHis (p-tRNAHis) by a ligase-like mechanism, involving formation of an Ap-p-tRNAHis-adenylylated intermediate, and removal of the AMP moiety during subsequent G−1 addition (17); however, Thg1p lacks any of the six characteristic motifs found in the superfamily of ligases/mRNA capping enzymes (45). We also demonstrated that Thg1p is required for tRNAHis guanylyltransferase in vivo, by showing that as cells are depleted of Thg1p they accumulate tRNAHis lacking G−1 (17).
Here we investigate the cellular events that occur as cells stop growing when depleted of Thg1p, with the objective of explaining the essential function of Thg1p and of probing the physiological responses of loss of Thg1p and of G−1 from tRNAHis. Four results are described. First, we show that the accumulation of tRNAHis lacking G−1 is accompanied by coincident appearance of uncharged tRNAHis, which may explain why Thg1p is essential. Second, we show that Thg1p-depleted cells appear to activate the GCN4 pathway, which normally responds to amino acid or purine starvation by increased expression of a number of genes involved in amino acid biosynthesis (21). Third, we show by fluorescence in situ hybridization (FISH) that tRNAHis accumulates in the nucleus in Thg1p-depleted cells. Fourth, we found unexpectedly that the tRNAHis of Thg1p-depleted cells accumulates additional 5-methylcytidine (m5C) modifications, well after the loss of G−1 and aminoacylation. The m5C modification requires the tRNA m5C-methyltransferase Trm4p, which catalyzes all other known m5C modifications of yeast tRNAs (31), and we have localized the additional m5C by development of a new RNA mapping technique. This accumulation of m5C in tRNAHis is, to our knowledge, the first example of additional tRNA modifications observed during a cellular response in eukaryotes, and its timing suggests that the accumulation of m5C is an indirect consequence of lack of Thg1p. The additional m5C might be caused by a prolonged presence of uncharged tRNA in the nucleus, or it might be part of a regulatory response.
Strains 13050 (trm4-Δ) and 13642 (gcn2-Δ) are derivatives of BY4742 and are from Research Genetics. Strains WG18 (thg1-Δ0:kanMX [CEN URA3 PGAL10-THG1]) and WG12 (THG1+ [CEN URA3 PGAL10-THG1]) were described previously (17). Strain BY4741 and the los1-Δ derivative (YDO5055) are from Open Biosystems. Strain EE1b-6, harboring the thermosensitive Ran-GAP allele rna1-1, was previously described (52). To construct a trm4-Δ strain conditionally lacking Thg1p, we first switched the kanMX cassette in strain WG18 to natMX, essentially as described elsewhere (51), to generate strain WG29 (relevant genotype, thg1-Δ0:natMX PGAL-THG1). Then, the trm4-Δ0:kanMX cassette and flanking sequence (−279 to +297 relative to TRM4 ends) was PCR amplified from strain 13050 with primers (TRM4UP279, 5′-TAGCGAGAAGAAGTAGATGCATTTC; TRM4DN297, 5′-GATAAAAAATTCTCTGTCTCAATCCATG) and transformed into strain WG29 to generate WG30 (relevant genotype, thg1-Δ0::natMX PGAL-THG1 trm4-Δ0::kanMX). Strain WG32 (relevant genotype, thg1-Δ0::natMX PGAL-THG1 gcn2-Δ0::kanMX) was constructed in a similar way after PCR amplification of the gcn2-Δ0:kanMX cassette and flanking sequences (−241 to +184 relative to the GCN2 ends) from strain 13642 with primers (GCN2UP241, 5′-GAGATAAGAACCTGGTGTAATCAATC; GCN2DN184, 5′-GAAGTAGACCATCTCATTAAACCTAC). Control strains WG28 (trm4-Δ0::kanMX THG1+ PGAL-THG1) and WG31 (gcn2-Δ0:kanMX THG1+ PGAL-THG1) were constructed similarly.
Strains were maintained on synthetic medium lacking uracil and then grown for analysis in YP medium (1% yeast extract, 2% peptone) containing 2% glucose (YPD) or 2% galactose (YPGal).
Low-molecular-weight bulk RNA used for purification of individual tRNAs and for bisulfite sequencing was prepared as described Gu et al. (17). Bulk RNA used for analysis of tRNA aminoacylation by Northern blotting was isolated under acidic conditions at 4°C as described previously (41). Bulk RNA used for mRNA analysis by reverse transcriptase PCR (RT-PCR) or Northern blotting was isolated as described elsewhere (35). Then, RNA was incubated with 40 U DNase I (Roche) in 150 μl of buffer containing 10 mM Tris-Cl (pH 7.5) and 5 mM MgSO4 at 25°C for ~30 min, followed by phenol extraction and ethanol precipitation. tRNA was deacylated by incubation in 100 μl 0.1 M Tris-Cl (pH 9.0) for 30 min at 37°C, followed by precipitation and resuspension of RNA in 10 mM sodium acetate (pH 4.5) and 1 mM Na2-EDTA. tRNAHis and tRNAGlyGCC were purified from low-molecular-weight RNA as described elsewhere (24), with 5′-biotinylated oligomers (BioHis, 5′-/biotin/GCCATCTCCTAGAATCGAACC; BioGly, 5′-/biotin/TGGTGCGCAAGCCCGGAATCGAACC).
To analyze the 3′ end of purified tRNAHis, RNA was *pCp labeled and then sized by 4 M urea-10% polyacrylamide gel electrophoresis (PAGE), with a hydrolysis ladder as described elsewhere (26) or treated with 0.5 U/μl RNase T1 (Industrial Research, New Zealand) in buffer containing 50 mM Tris-Cl (pH 7.5), 2 mM EDTA, and 0.4 μg/μl bulk RNA at 37°C for 1 h, followed by 4 M urea-20% PAGE. Nucleosides were analyzed and quantified as described previously (12, 24), using a C18 high-performance liquid chromatography (HPLC) column. Northern blot assays of tRNA to detect aminoacylation were performed essentially as described elsewhere (41). RNA was resolved on a 6.5% acid gel, electroblotted onto a Hybond N+ membrane (Amersham Biosciences), cross-linked by UV, and probed with the following specific primers: P1exP1 (5′-ACTAACCACTATACTAAGA) for tRNAHisGUG, Gly3′ (5′-AAGCCCGGAATCGAACCGG) for tRNAGlyGCC, ArgP1 (5′-TAGCCAGACGCCGTGAC) for tRNAArgACG, Lys31P1 (5′-TAAAAGCCGAACGCTCTACC) for tRNALysUUU, TyrP1 (5′-CGAGTCGAACGCCCGAT) for tRNATyrGUA, Mito tRNAHis (5′-CGAACTCAGATTTAACGCA) for mitochondrial tRNAHisGUG, and 5s RNA (5′-TGGTAGATATGGCCGCAACC) for 5S RNA.
A 40% bisulfite solution was freshly prepared by dissolving 4.05 g of sodium bisulfite (Sigma) in 5.5 ml of water, adjusting the pH to 5.1 with 6.3 M sodium hydroxide, adding 33 μl of 200 mM hydroquinone (Sigma), and adjusting the volume to 10 ml with water (15). Sixty micrograms of bulk RNA was incubated in 75 μl bisulfite solution for 30 min at 95°C and desalted by passage through a Micro Bio-spin 6 chromatography column (Bio-Rad). Then, RNA was desulfonated by incubation with 0.5 M Tris-Cl (pH 9.0) at 37°C for 1 h, followed by addition of a 1/10 volume of 10 M ammonium acetate and 20 μg glycogen and precipitation with 2.5 volumes of ethanol. Remaining m5C residues in tRNAHis were mapped by primer extension, using an appropriate primer modified for expected C-to-U changes from nucleotides (nt) 76 to 59 (tRNAHisform5C18nt, 5′-TAATACCATCTCCTAAAA), followed by 4 M urea-12.5% PAGE and PhosphorImager (Molecular Dynamics) visualization. Poison primer extensions contained ddGTP but no dGTP.
For RT-PCR, first-strand cDNA synthesis was performed with Moloney murine leukemia virus RT according to the manufacturer's instructions (Invitrogen). Then, RT was inactivated, and cDNA was diluted 16-fold in 10 mM Tris-Cl (pH 7.5) and amplified in 40-μl mixtures containing 1 to 2 μl of template cDNA, 3 mM MgCl2, 1× PCR buffer (10 mM Tris-Cl, pH 8.3, 50 mM KCl), 0.25 mM deoxynucleoside triphosphate (dNTP), 1 μl Taq polymerase, 0.5 μM of each specific primer pair (in which either the forward or reverse primer was 32P labeled), and a similarly labeled ACT1 primer pair as a control. PCR products were resolved on a 10% nondenaturing polyacrylamide gel. To confirm linearity of RT-PCR, a standard curve was generated using different amounts of RNA for first-strand synthesis. To quantify results, signals were first normalized to ACT1 and then normalized to the corresponding ratio in wild-type cells 6 h after glucose shift. Primers were the following: ACT1m (5′-TGGGTTTGGAATCTGCCGGTATTG) and ACT13 (5′-TTAGAAACACTTGTGGTGAACGATAG) for ACT1; LYS15m (5′-AGACACTACCAACCCTCACAACCC) and LYS13 (5′-CTATCGAACAATTTCTTGGCTCTAACC) for LYS1; ARG4m5 (5′-AAGATGGAAGCTGCTCTCACGATGG) and ARG4m3 (5′-AGCGGTTCCACCAGTAGCATCTCG) for ARG4; HIS5m (5′-GATCAACGGGGGTGACAATGTCTTG) and HIS53 (5′-TTATTCATTGGCCAGCTTATATAACG) for HIS5; EFB15m (5′-TGGTTGCTAACGTCAAGGCCATCG) and EFB13 (5′-ATATCGGTAGATTGGACGTGGTCTTC) for EFB1. For Northern blot assays, ~20 μg bulk RNA was resolved on a denaturing agarose gel, followed by transfer of RNA to a Hybond N+ membrane and hybridization to oligonucleotide probes (ACT13 for ACT1, LYS13 for LYS1, and ARG4m3 for ARG4).
FISH was performed as described previously (42), with modifications as described in reference 48, using previously described probes to tRNAMet, tRNATyr, poly(A)-containing RNA, and a new probe to tRNAHis (5′-TCCTAGAATCGAACCAGGGTTTCATCGGCCACAACGATGTGTACTAACCACTATACTAAG).
A Nikon Microphot-FX microscope was used to observe fluorescence on the slides, and a Sensys charge-coupled device camera (Photometrics, Tucson, AZ) with QED software (QED Imaging, Pittsburgh) was used to capture the images. Adobe Photoshop 5.0 was used for image assembly.
Thg1p depletion might cause cell death because the accumulation of tRNAHis lacking G−1 impairs aminoacylation, as has been shown in vitro (32, 40), or because some other aspect of tRNA processing or function is affected. To address this question, we used a conditional mutant strain bearing a single THG1 gene under PGAL promoter control (relevant genotype, thg1-Δ [CEN PGAL-THG1], referred to as thg1-Δ), and examined tRNA when cells were switched from permissive medium containing galactose (YPGal) to nonpermissive medium containing glucose (YPD), in which THG1 transcription is repressed. Cells begin to grow more slowly after 17 h in YPD medium and accumulate tRNAHis lacking G−1 between 12 and 20 h after glucose shift, consistent with the loss of Thg1p activity in extracts (17).
Thg1p-depleted cells accumulate uncharged tRNAHis coordinately with the loss of G−1. To analyze tRNA aminoacylation, we isolated bulk RNA under acidic conditions, which stabilize the aminoacyl bond of tRNA, and compared RNA mobility on acidic polyacrylamide gels, before and after base treatment to remove the amino acid. Presence of the amino acid results in slower mobility of the tRNA. As shown in the Northern blot assay in Fig. Fig.1,1, aminoacylation of tRNAHis in Thg1p-depleted cells is similar to that in control strains at early times (0 and 4.5 h) after glucose shift (lanes a to d and i to l) but is substantially reduced 16 h after the shift (lanes m and n) and almost completely absent 28 h after the shift (lanes o and p). Furthermore, tRNAHis levels are not substantially reduced during Thg1p depletion, relative to levels of the 5S RNA control. We note that the faster migration of tRNAHis from Thg1p-depleted cells is not due to loss of G−1, because transcripts of 76-mer tRNAHis containing G−1 and 75-mer tRNAHis lacking G−1 have similar mobilities in acidic gels (lanes q and r).
Loss of aminoacylation in Thg1p-depleted cells is specific for cytoplasmic tRNAHis. The aminoacylation status of cytoplasmic tRNALys, tRNATyr, tRNAGly, and tRNAArg does not change measurably during Thg1p depletion, although the observed basal aminoacylation levels differ for each species (Fig. (Fig.1).1). Thus, loss of aminoacylation of tRNAHis is not due to some general effect on aminoacylation caused by Thg1p depletion.
Loss of aminoacylation of cytoplasmic tRNAHis is likely not due to loss of histidyl-tRNA synthetase (HisRS). We were able to probe HisRS activity in Thg1p-depleted cells, because yeast has only one gene encoding HisRS proteins for aminoacylation of both cytoplasmic tRNAHis and mitochondrial tRNAHis and because mitochondrial tRNAHis has a genome-encoded G−1 that renders it insensitive to loss of Thg1p. Mitochondrial tRNAHis is charged to the same extent both in control and in Thg1p-depleted cells (Fig. (Fig.1,1, lanes a to h and i to p). Therefore, HisRS is fully active in mitochondria during Thg1p depletion, and we infer that HisRS is present and active in the rest of the cell.
One other possible explanation for the lack of aminoacylated tRNAHis in Thg1p-depleted cells would be the lack of a mature CCA 3′ end, since CCA is necessary for aminoacylation of tRNA. This could occur if G−1 were a determinant in vivo for CCA addition to tRNAHis. To address this possibility, we purified cytoplasmic tRNAHis from Thg1p-depleted cells, after base treatment to deacylate any remaining aminoacylated tRNA, and examined its size after 3′-end labeling.
tRNAHis from Thg1p-depleted cells has a 3′ end of the same size as CCA-containing tRNAHis from control cells (Fig. (Fig.2).2). As expected, tRNAHis from control cells appears to have G−1 at its 5′ end (Fig. (Fig.2A,2A, lanes b to e), based on the hydrolysis ladder of this tRNA (lane a), which allows positioning of the characteristic gap representing known 2′-O-methylated residues at positions 4 and 18. By contrast, tRNAHis from Thg1p-depleted cells appears to be 1 nt smaller. Early after glucose shift (0 and 4.5 h), almost all the tRNAHis is the same length as in control cells (Fig. (Fig.2A,2A, lanes f and g). However, at later times a substantial fraction of the tRNAHis is 1 nucleotide shorter, about 60% after 16 h, and 85% after 28 h (Fig. (Fig.2A,2A, lanes h and i). Since we had previously shown by primer extension that this thg1-Δ strain accumulates tRNAHis lacking G−1 over the same time course (17), we conclude that tRNAHis has lost G−1 and that the 3′ end contains CCA.
To prove explicitly that tRNAHis has a mature-size 3′ end in Thg1p-depleted cells, we treated the tRNA with RNase T1 to release the labeled 3′-terminal oligonucleotide after G71 and compared its size to that from tRNAHis of control cells. As shown in Fig. Fig.2B,2B, the 3′ oligonucleotides of tRNAHis from Thg1p-depleted and control cells are the same size after RNase T1 treatment (lanes b to e and f to i) and the same size as the oligonucleotide from synthetic 75-mer tRNAHis after similar treatment (lane j). This result provides compelling evidence that tRNAHis in Thg1p-depleted cells contains a mature CCA end and that lack of aminoacylation is not due to a defect in its 3′-end maturation.
One other change that might result from loss of G−1 in tRNAHis is alteration of its modification status. Cytoplasmic tRNAHis contains 2′-O-methyladenosine (Am); 2′-O-methylguanosine (Gm); 1-methylguanosine (m1G); three pseudouridines (Ψ); three dihydrouridines (D); 5-methylcytidine (m5C); and ribothymidine (rT). To determine if the nucleoside modification status of tRNAHis is altered in Thg1p-depleted cells, we purified tRNAHis and quantified its nucleosides using HPLC.
We found unexpectedly that tRNAHis from Thg1p-depleted cells has elevated levels of m5C. As shown in Fig. Fig.3A,3A, the modified nucleosides are resolved well enough to quantify them, except for rT, which is hidden by the G peak, and m1G/Gm, which migrate together and can only be quantified as a group. Dihydrouridine is only observed at a different wavelength (55), but it is also well resolved. Quantification shows that levels of Ψ, Am, D, and m1G/Gm remain almost constant in Thg1p-depleted cells, although there is some mild fluctuation in the m1G/Gm levels (Table (Table1).1). By contrast, the peak corresponding to m5C is distinctly larger in Thg1p-depleted cells at late times after glucose shift (Fig. (Fig.3B).3B). This peak has the retention time and spectrum characteristic of m5C, suggesting that it represents additional m5C modification. Quantification reveals 1 mole of m5C in tRNAHis from control cells (as expected) and from thg1-Δ strains 4.5 h after glucose shift (Table (Table1);1); however, the amount of m5C increases to 1.3 mol 17 h after glucose shift and to 2.1 mol after 30 h (Table (Table1).1). The increased m5C content is specific for tRNAHis, because another purified tRNA (tRNAGly) shows no increased m5C content after Thg1p depletion (Fig. (Fig.3D;3D; Table Table1)1) and because m5C levels of bulk RNA do not change (data not shown).
Detailed examination of the timing of changes in tRNAHis in Thg1p-depleted cells shows that loss of aminoacylation and G−1 appear coincidentally but the accumulation of additional m5C is delayed. Based on Fig. Fig.11 and and22 (derived from one experiment) and Fig. Fig.33 (derived from another experiment), the m5C accumulation appears to be delayed, since after 16 to 17 h in YPD medium ~60% of tRNAHis from Thg1p-depleted cells lacks G−1 (Fig. (Fig.2,2, lane h) and is not aminoacylated (Fig. (Fig.1,1, lane m), but m5C levels have only increased by ~30% (Table (Table1).1). To examine the timing explicitly, we measured all three parameters from the same experiment (Fig. (Fig.4).4). Our results show that loss of G−1 and aminoacylation occur with very similar time courses (as measured by the midpoint of the curves), whereas the accumulation of m5C in tRNAHis is distinctly delayed, by as much as 8 h. Thus, m5C accumulation cannot be the cause of loss of aminoacylation of tRNAHis and is instead a consequence of the loss of Thg1p.
Since Trm4p is responsible for other known m5C modifications in yeast tRNA (31), it seemed likely that Trm4p is also responsible for the additional m5C modification occurring during Thg1p depletion. Thus, we introduced a trm4 deletion into a thg1-Δ strain and examined m5C levels in Thg1p-depleted cells. As expected, tRNAHis has 1 mole of m5C in wild-type cells, no detectable m5C in trm4-Δ strains, and one extra mole of m5C in Thg1p-depleted TRM4+ cells 31 h after glucose shift (Fig. (Fig.5A5A and data not shown). Since tRNAHis from a trm4-Δ thg1-Δ strain also has no detectable m5C after 31 h in glucose, we conclude that Trm4p is responsible for the extra m5C modification in Thg1p-depleted cells.
We developed a new method to map m5C modifications in tRNA, based on an m5C DNA mapping method (15). When bisulfite-treated RNA is processed at a pH compatible with RNA, cytidine and derivatives (3-methylcytidine, 2′-O-methylcytidine, and N4-acetylcytidine) are quantitatively converted to uridine and the corresponding derivatives (data not shown). However, m5C is resistant to bisulfite treatment, suggesting that sites of m5C modification can be mapped by primer extension. One caveat in this analysis is the unexpected sensitivity of Ψ to bisulfite treatment, which also causes a primer extension block (data not shown); however, as shown below, this does not affect mapping.
Primer extension shows that the new m5C modifications occur at C48 and C50 of tRNAHis, flanking the known m5C at position 49. Figure Figure5C5C shows a primer extension of bisulfite-treated bulk RNA. The sequence from control cells (lanes g to j) is that expected from nt 39 to 54 of tRNAHis after conversion of cytidine to uridine, and there is a single major band in the G lane, indicating m5C49, as well as a possible minor band at position 48 (lane h). Also as expected, primer extension shows tRNAHis from a trm4-Δ strain (lanes b to e) has no m5C (lane c); however, tRNAHis from Thg1p-depleted cells has the expected m5C49, as well as two other m5C residues at positions 48 and 50 (lane m). Furthermore, we observe only these three m5C residues between G−1 and A58 (data not known). Thus, the extra m5C residue in Thg1p-depleted cells maps to C48 and C50 and requires Trm4p.
We quantified the m5C at position 50 by a poison primer extension assay employing ddGTP but no dGTP. In this way, primer extension will quantitatively stop at the first m5C or other cytidine derivatives resistant to bisulfite treatment. Poison primer extension of tRNAHis from control cells indicated a single stop at position 49 (Fig. (Fig.5D,5D, lane e), whereas extension of tRNAHis from Thg1p-depleted cells indicated stops at both positions 50 and 49 (lane a). Quantification indicated that 0.6 mol of m5C is located at C50. The remaining 0.4 mol of m5C in tRNAHis of Thg1p-depleted cells is located at C48, rather than in the region covered by the primer (nt 59 to 76), since analysis of an oligonucleotide released from nt 59 to 76 of tRNAHis by directed RNase H-mediated cleavage reveals no m5C residue in this region (data not shown).
Examination of individual species of tRNAHis confirmed the location of the additional m5C residues and suggested that additional m5C modification occurs first at C50. To do this, we cloned cDNA from tRNAHis by RT-PCR after bisulfite treatment to convert cytidine, but not m5C, to uridine and sequenced individual clones. Of 13 isolates examined, 6 had cytidine at position 49, 5 had cytidine at positions 49 and 50, and two had cytidine at positions 48, 49, and 50. This distribution demonstrates that all three cytidine residues can be modified in a given tRNA and suggests that modification at C48 generally follows modification at position C50.
Since Thg1p-depleted cells accumulate tRNAHis lacking aminoacylation, it seemed likely that this uncharged tRNAHis might activate the GCN4 pathway. In yeast, amino acid starvation or lack of functional tRNA synthetases derepresses translation of the transcription factor Gcn4p, which in turn activates transcription of a battery of genes encoding amino acid biosynthetic enzymes (19, 20). Activation of Gcn4p is likely mediated by uncharged tRNAs, which activate the Gcn2p kinase to phosphorylate eIF2α, leading ultimately to derepression of GCN4 translation (21). Thus, although the uncharged tRNAHis in Thg1p-depleted cells also lacks G−1 and contains an extra m5C at later times, it might also activate the GCN4 pathway.
Two lines of evidence suggest that Thg1p-depleted cells activate the GCN4 transcription control pathway. First, the mRNA levels of genes known to be induced by the GCN4 pathway are elevated in Thg1p-depleted cells, as shown by RT-PCR analysis (Fig. (Fig.6A)6A) and Northern analysis (Fig. (Fig.7).7). RT-PCR analysis indicated that LYS1, HIS5, and ARG4 mRNA levels are each elevated between 19-fold and 35-fold relative to ACT1 mRNA levels in Thg1p-depleted cells (Fig. (Fig.6A).6A). For each of these genes, the increase in mRNA levels is not evident in Thg1p-depleted cells after 6 h in YPD medium (lane q), is pronounced at 18 h (lane r), and is elevated further after 28 h (lane s). By contrast, mRNA levels of each of these three genes change little, if at all, in control cells (lanes h to k). Furthermore, each of two genes known not to be affected by the GCN4 pathway, EFB1 (encoding elongation factor B1) and ACT1, are not induced in Thg1p-depleted cells (Fig. (Fig.6B).6B). Northern analysis (Fig. (Fig.7)7) confirmed the induction of ARG4 and LYS1 in Thg1p-depleted cells.
Second, activation of the GCN4 pathway in Thg1p-depleted cells is, as expected, mostly dependent on Gcn2p. To test the involvement of Gcn2p, we introduced a GCN2 deletion into the thg1-Δ strain and measured mRNA levels of the same GCN4-regulated genes and negative controls. As shown in Fig. Fig.6A,6A, the dramatic increase in mRNA levels of HIS5, ARG4, or LYS1 observed in Thg1p-depleted cells was substantially reduced when Gcn2p was absent (lanes t to w); however, there was still a residual increase in mRNA levels (about two- to threefold, relative to ACT1 mRNA). As shown below, this residual increase in mRNA levels may be due to the presence of tRNAHis in the nucleus, which can activate the GCN4 pathway independent of Gcn2p (37).
In vertebrates and in yeast, inhibition of tRNA aminoacylation results in tRNA nuclear accumulation (14, 29, 41). Since Thg1p depletion prevents aminoacylation of tRNAHis by HisRS, we anticipated that such depletion would result in nuclear tRNAHis pools. To test this we employed FISH using an oligonucleotide complementary to tRNAHis labeled at the 3′ end with digoxigenin-11-dUTP and detected its location with fluoresceinated antidigoxigenin (42, 48).
As anticipated, tRNAHis is distributed throughout the cytoplasm when a wild-type strain (BY4741) is grown in YPD medium (Fig. (Fig.8A,8A, top row, left) and has an increased nuclear tRNAHis signal in los1-Δ cells, which lack the tRNA nuclear exportin (Fig. (Fig.8A,8A, top row, right). As shown in Fig. Fig.8A8A (right, rows 3 and 4), thg1-Δ cells (WG18) show prominent nuclear pools of tRNAHis 14 h and 20 h after glucose shift, but not after 2 h (right, row 2), whereas control parent cells (WG12) have the same uniform distribution of tRNAHis as the wild-type strain (Fig. (Fig.8A,8A, left, rows 2 to 4). This time course of appearance of nuclear tRNAHis pools in thg1-Δ cells mirrors the effect of Thg1p depletion upon loss of G−1 and loss of aminoacylation of tRNAHis.
Since depletion of Thg1p specifically affects aminoacylation of tRNAHis (Fig. (Fig.1),1), we evaluated the subcellular distributions of other tRNAs and mRNA to determine if the cellular distribution of tRNAs mirrors this cognate tRNA-specific effect. As anticipated, both tRNATyr and poly(A)-containing RNA are distributed throughout the nucleus and cytosol of wild-type cells (Fig. (Fig.8B,8B, top row), but both accumulate in nuclei of rna1-1 cells (Fig. (Fig.8B,8B, row 3), which have a defective RanGTP:RanGDP nucleus/cytoplasm gradient (11). In contrast, and also as expected, los1-Δ cells accumulate tRNATyr but not poly(A)-containing RNAs in the nucleus (Fig. (Fig.8B,8B, row 2). Neither control cells nor cells depleted of Thg1p for 14 h accumulate tRNATyr (Fig. (Fig.8B,8B, rows 4 and 5), tRNAMet (data not shown), or poly(A)-containing RNAs in the nucleus (Fig. (Fig.8B,8B, rows 4 and 5). Thus, depletion of Thg1p appears to have a cognate-specific effect, causing specifically tRNAHis to accumulate in the nucleus.
We have shown that a major consequence of depletion of Thg1p from yeast is the accumulation of uncharged tRNAHis that accompanies accumulation of tRNAHis lacking G−1. The accumulation of uncharged tRNAHis is specific, since each of four other control cytoplasmic tRNAs is aminoacylated normally in this strain. The accumulation of uncharged tRNAHis occurs over nearly the same time course as the accumulation of tRNAHis lacking G−1; in each case the midpoint of the curve occurs in 16 to 17 h. Moreover, no other obvious physical alterations of tRNAHis are observed in this timeframe. As determined by 3′-end labeling of purified tRNA, tRNAHis from Thg1p-depleted cells contains an intact CCA end. Thus, lack of G−1 does not preclude the 3′-end maturation of tRNAHis. Furthermore, the modification status of tRNAHis is virtually normal in Thg1p-depleted cells, although additional m5C appears later. Since loss of G−1 from tRNAHis occurs at the same time as loss of growth (17) and loss of aminoacylation, we conclude that the likely cause of death of Thg1p-depleted cells is the lack of aminoacylated tRNAHis. A recent study reports that Thg1p interacts with the origin recognition complex and suggests that Thg1p has a role in nuclear division and budding (39); it is not clear how this phenomenon is related to Thg1p G−1 addition activity.
The simplest interpretation of our observations is that lack of G−1 in tRNAHis in Thg1p-depleted cells is the direct cause of lack of aminoacylation in vivo. This is consistent with previous studies showing that G−1 is a critical determinant for aminoacylation of tRNAHis in vitro by yeast tRNAHis synthetase (32, 40), as is also true of E. coli tRNAHis synthetase (18). Since the loss of G−1 and aminoacylation in tRNAHis also closely follows the loss of growth of Thg1p-depleted cells, it seems likely that lack of charged tRNAHis is the limiting factor for cell growth. Uncharged tRNAHis is unlikely to be caused by lack of active HisRS in the cell. HisRS is present and active in Thg1p-depleted cells, since mitochondrial tRNAHis remains fully aminoacylated as Thg1p is depleted. Mitochondrial tRNAHis has a genome-encoded G−1 residue that is, presumably, retained during tRNA maturation, as occurs in bacteria (8, 34), obviating the need for Thg1p.
The accumulation of tRNAHis in nuclei of thg1-Δ cells is inferred to be caused by failure to aminoacylate this tRNA, because in yeast (42) and Xenopus laevis oocytes (29) disruption of tRNA synthetase function leads to nuclear accumulation of substrate tRNA. However, the mechanism for nuclear accumulation of tRNAHis is not entirely clear. On one hand, lack of G−1 addition to nuclear tRNAHis could prevent its aminoacylation in the nucleus and thereby inhibit its export to the cytosol, since Thg1p resides in the nucleus as well as in the cytoplasm (23) and it has been reported that there is histidyl-tRNAHis in the nucleus (49). This is consistent with our observation that some of the GCN4 pathway activation caused by lack of Thg1p is Gcn2p independent at a comparable time (18 h) (Fig. (Fig.6),6), since it is known that there is a Gcn2-independent pathway for GCN4 activation that acts in the nucleus (37). On the other hand, G−1 depletion could occur in the cytoplasm, leading to loss of aminoacylation and movement of tRNA from the cytosol to the nucleus (retrograde tRNA nuclear import), as has recently been reported (44, 50).
It is interesting that the uncharged tRNAHis that accumulates in the nucleus of Thg1p-depleted cells is apparently not subject to massive degradation. Recently, Anderson and coworkers (25) described a tRNA degradation pathway that acts in the nucleus to degrade pre-tRNAiMet lacking m1A58. We observe little obvious degradation of tRNAHis in Thg1p-depleted cells (Fig. (Fig.1),1), suggesting that, despite the absence of histidine charging and its presence in the nucleus, tRNAHis is not a substrate for large-scale degradation by this pathway. One possible reason might be that the tRNAHis is in its native conformation and is therefore immune from this degradation system.
Since Gcn2p is located in the cytoplasm (23) and thought to function there (37), the large component of GCN4 pathway activation that is Gcn2p dependent at 18 and 28 h (Fig. (Fig.6)6) suggests that a portion of uncharged tRNAHis is located in the cytoplasm. This could occur because tRNA export is not absolutely dependent on tRNA charging (4, 29), or because some population of cytoplasmic tRNAHis loses its G−1 during growth in Thg1p-depleted cells, leading to a requirement for repair of the 5′ end by Thg1p. The CCA-adding enzyme has such a repair role in yeast (53). Regardless of the mechanism by which tRNAHis ends up in the cytoplasm, we note that the amount of Gcn2p-dependent activation of the GCN4 pathway caused by tRNAHis from Thg1p-depeleted cells is very similar to activation caused by other stimuli, such as 3-aminotriazole, that activate this pathway (33). Since the GCN4 pathway is activated in response to 3-aminotriazole by binding of uncharged but otherwise mature tRNAHis to Gcn2p to activate it to phosphorylate the alpha subunit of eIF2 (21), we conclude that uncharged tRNAHis lacking G−1 can also serve the same role.
Surprisingly, we found that tRNAHis accumulates additional m5C in Thg1-depleted cells, well after the loss of G−1 and aminoacylation. The extra m5C was mapped to C50 (0.6 mol) and C48 (0.4 mol) by development and use of a new RNA m5C mapping technique using bisulfite, adapted from a corresponding method for DNA (15). These additional m5C sites in tRNAHis flank the known modification site at C49, and C50 is a new site for m5C modification of yeast tRNAs, although this site is modified in tRNA of some other organisms (47). Since Trm4p is responsible for all other known m5C modifications of yeast tRNA at positions 34, 40, 48, and 49 (31), and since trm4 mutants do not accumulate additional m5C in Thg1p-depleted cells, we presume that Trm4p catalyzes formation of the additional m5C modification at both C48 and C50.
Although there have been reports that queuine and Y-base modifications are reduced in transformed cells (16, 38), and other reports have suggested the possibility of lower modification levels in yeast tRNA under certain conditions (5, 46), the accumulation of extra m5C in tRNAHis of Thg1-depleted cells is, to our knowledge, the first documented case of additional tRNA modification appearing during a cellular response in eukaryotes. Recently, Björk and colleagues reported a gain-of-function mutation in Salmonella enterica that results in the synthesis of new tRNA modifications that affect tRNA function (10). The additional m5C modification described here might arise simply as a result of the prolonged presence of uncharged tRNAHis in the nucleus, where Trm4p is known to be localized (54). Alternatively, the additional m5C might arise as a response of some regulatory pathway that is initiated as a result of the accumulation of uncharged tRNAHis. Consistent with the hypothesis of a regulatory role for tRNA modification, it was recently observed that human protein kinase B (Akt) and RSK can phosphorylate and inactivate Trm8p (9), one of two subunits required for m7G modification of tRNA (1, 2), suggesting that altered modification levels may occur during regulatory responses in the cell.
It is somewhat of a puzzle why the m5C modification at C48 and C50 is incomplete. It is possible that Trm4p has difficulty modifying these sites. It is also possible that the incomplete modification of tRNAHis reflects modification of the portion of cellular tRNAHis that is located in the nucleus, either the newly synthesized tRNAHis or possibly cytoplasmic tRNAHis that returns to the nucleus during Thg1p depletion (44, 50). It is also somewhat of a mystery why the m5C modification is delayed relative to other consequences of loss of Thg1p. The delay could be directly related to the cause of inefficient modification, or it could be part of a regulatory response.
We are grateful to A. Alexandrov and J. Jackman for valuable advice.
This research was supported by grants GM52347 to E.M.P. and GM27930 to A.K.H. from the National Institutes of Health.