Cytoplasmic and mitochondrial tRNA nucleotidyltransferase activities are derived from the same gene in the yeast Saccharomyces cerevisiae.

ATP (CTP):tRNA-specific tRNA nucleotidyltransferase is an enzyme required for the synthesis of functional tRNAs in eukaryotic cells. Neither the tRNA genes in the nucleus nor in organelles encode the CCA end, so it must be added post-transcriptionally. The gene that codes for the enzyme that adds the CCA end to nuclear coded tRNAs in Saccharomyces cerevisiae has been isolated (Aebi, M., Kirchner, G., Chen, J.-Y., Vijayraghavan, U., Jacobson, A., Martin, N. C., and Abelson, J. (1990) J. Biol. Chem. 265, 16216-16220). We now demonstrate that there is a mitochondrial tRNA nucleotidyltransferase activity in yeast and that it is a matrix enzyme. A comparison of purified mitochondrial enzyme with its cytoplasmic counterpart revealed no differences. These results suggest that proteins responsible for this step in the maturation of tRNAs in the nucleus and mitochondria might be identical and coded by the same nuclear gene. Accumulation of shortened mitochondrial as well as cytoplasmic tRNAs in a strain with a temperature-sensitive tRNA nucleotidyltransferase is consistent with this hypothesis. Alteration of the wild type gene such that amino-terminal truncated proteins are produced leads to a defect in mitochondrial function and a decrease in mitochondrial nucleotidyltransferase activity. This provides a direct demonstration that one gene provides this enzyme activity for the biosynthesis of tRNAs in both the nuclear/cytoplasmic and mitochondrial compartments in yeast.


Saccharomyces cerevisiae"
(Received for publication, February 18, 1992) Jeou Yuan ChenS, Paul B. M. Joyce  ATP (CTP):tRNA-specific tRNA nucleotidyltransferase is an enzyme required for the synthesis of functional tRNAs in eukaryotic cells. Neither the tRNA genes in the nucleus nor in organelles encode the CCA end, so it must be added post-transcriptionally. The gene that codes for the enzyme that adds the CCA end to nuclear coded tRNAs in Saccharomyces cerevisiae has been isolated (Aebi, M., Kirchner, G., Chen We now demonstrate that there is a mitochondrial tRNA nucleotidyltransferase activity in yeast and that it is a matrix enzyme. A comparison of purified mitochondrial enzyme with its cytoplasmic counterpart revealed no differences. These results suggest that proteins responsible for this step in the maturation of tRNAs in the nucleus and mitochondria might be identical and coded by the same nuclear gene. Accumulation of shortened mitochondrial as well as cytoplasmic tRNAs in a strain with a temperature-sensitive tRNA nucleotidyltransferase is consistent with this hypothesis. Alteration of the wild type gene such that aminoterminal truncated proteins are produced leads to a defect in mitochondrial function and a decrease in mitochondrial nucleotidyltransferase activity. This provides a direct demonstration that one gene provides this enzyme activity for the biosynthesis of tRNAs in both the nuclear/cytoplasmic and mitochondrial compartments in yeast. Precursor RNAs made from transfer RNA genes must be processed in a series of steps to yield mature products. Nuclear and organellar precursor RNAs must be trimmed a t their 5' and 3' ends. Base modifications as well as the post-transcriptional addition of the CCA end must also occur (reviewed in Ref. 1). In addition, some nuclear tRNA gene transcripts contain intervening sequences that must be removed. The removal of intervening sequences and the end maturation steps occur in the nucleus in Xenopus (2) as do some of the base modifications (3, 4). Precursor tRNAs in yeast that * This work was supported by National Institutes of Health Grant GM27597 (to N. C. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ contain intervening sequences and are nuclear restricted have base modifications and CCA ends (5,6) suggesting that these processing steps and the enzymes that carry them out are nuclear localized in yeast as well. Other base modifications occur in the cytoplasm (3, 7) as does repair of the 3"terminal A of the CCA end (8). Organellar tRNA precursor processing occurs inside organelles, and both mitochondrial (9-13) and chloroplast (14)(15)(16) extracts containing tRNA processing activities have been described. In cases where tRNAs are coded by nuclear genes but imported into mitochondria (17)(18)(19)(20)(21), it is not clear whether processing occurs before, during, or after import into the organelle. Regardless of the location, it is clear that tRNA biosynthetic pathways have many similarities and require enzymes with analogous functions in multiple cellular locations. There are two mechanisms whereby enzymes that carry out analogous functions in eukaryotic cells are provided to multiple cellular locations. Many isoenzymes carry out identical biochemical reactions but are structurally distinct and arise from different genes. A smaller class of enzymes, recently named sorting isozymes (22), is made up of proteins which carry out analogous reactions in more than one cellular location but arise from the same gene. In yeast, enzymes that are shared between the mitochondria and cytoplasm and are coded by the same gene include fumarase (23), isopropylmalate synthetase (24), and valyl (25) and histidyl tRNA synthetases (26). W,W-Dimethylguanosinespecific tRNA methyltransferase is also a member of this class, but is unique in that it is shared by the mitochondria and the nucleus and does not appear to be present in the cytoplasm (27), whereas isopentenyl pyrophosphate:tRNA isopentenyl transferase (28) appears in the nucleus, mitochondria, and cytoplasm.' As described above, ATP (CTP):tRNA-specific nucleotidyltransferase activity is required for tRNA biosynthesis in multiple cellular compartments and for repair of tRNAs in the cytoplasm. A comparison of the enzymes isolated from the mitochondria and the rest of the cell in yeast suggested to us that the proteins responsible for nucleotidyltransferase activity in these compartments might be identical. Analysis of tRNA biosynthesis in a mutant with temperature-sensitive nucleotidyltransferase activity indicated that both mitochondrial and cytoplasmic tRNAs were affected at the restrictive temperature. Site-directed mutagenesis of the previously identified and characterized CCAl gene from yeast (29) was used to make alterations in the gene that result in amino-terminal truncations in the protein produced from this gene. Strains carrying the altered genes are respiratory-deficient as judged by their growth characteristics on non-fermentable carbon sources, and they have decreased mitochondrial but not cytoplasmic nucleotidyltransferase activity. These results dem-L. Hunter, N. C. Martin, and A. K. Hopper, unpublished observation.

14879
onstrate that the same gene codes for all of the nucleotidyltransferase activity in yeast and thus ATP (CTP):tRNAspecific nucleotidyltransferase joins the sorting isozyme class in yeast.
Enzyme Assays and Isolations-Mitochondria were prepared as described previously (11) except that, in the experiment reported in Table 111, the mitochondria were further purified by sucrose gradient centrifugation (30). For the experiment involving digitonin treatment, the mitochondria were suspended in 10 mM KPO,, 0.6 M sorbitol, pH 7.2, and mixed with increasing concentrations of digitonin in the same buffer. After 5 min the sample was centrifuged in a microcentrifuge to separate a supernatant and pellet fraction, and the pellets were suspended in the same buffer containing 1% Tween 20. Assays and isolation procedures for nucleotidyltransferase were as described by Chen et al. (31). Cytochrome oxidase was assayed as described in Ref. 32, except that the difference between the absorbance at 550 and 540 nm was used to determine activity. Assays for malate dehydrogenase (33) and glucose-6-phosphate dehydrogenase activity (34) were as described.
Northern Analyses-Small RNAs were isolated, transferred, and detected by Northern analysis as described by Aebi et al. (29). The oligonucleotides 5' ATCTTCTGGTTCGCAGCCAG and 5' GCGCCTGACCTTTTGGCTTC, complementary to cytoplasmic tRNA& and mitochondrial tRNAPr", respectively, were used for Northern analyses. The oligonucleotides were labeled at their 5' ends with T, polynucleotide kinase. Hybridization was at 54 "C overnight, followed by three 3-min washes in 0.8 M NaC1,0.08 M sodium citrate at room temperature.
Site-directed Mutagenesis and Transformations-Site-directed mutagenesis was carried out according to the procedure of Geisselsoder et al. (35). The oligonucleotide used to change the ATG at position CCGTAGGATATCCAAAAAAAG. The oligonucleotide used to +1 of the CCAl gene (accession number 505612) to an ATC was change the ATG a t position 28 was AGCAGCACTATTCTGCAG-TAG. These genes were transferred from pJDB207 (29) into YCp50 and introduced into 352-1A by the LiOAC procedures of Ito et al. (36). Transformants were maintained on uracil minus media but were grown for four generations on rich media containing 1% yeast extract, 2% Bacto-peptone, 2% galactose, and 0.2% dextrose, which resulted in increased yield and quality of mitochondria for comparing activity produced by the mutant genes.

ATP (CTP)
:tRNA-specific nucleotidyltransferase is clearly required for mitochondrial tRNA biosynthesis because the CCA end is not encoded in tRNA genes. Although the activity is quite easy to assay in mitochondrial extracts, care must be taken to ascertain the activity is inside the mitochondria and not a result of cytoplasmic nucleotidyltransferase contamination of mitochondrial preparations. The results of three experiments collectively demonstrate the enzyme activity we measure in mitochondrial preparations is sequestered and that it behaves as expected for a matrix enzyme. We assayed intact and broken mitochondria for nucleotidyltransferase by adding radiolabeled CTP and substrate tRNA to identical samples prior to and after freeze thawing. The substrate tRNA does not cross an intact mitochondrial membrane so any labeling of this added tRNA by intact mitochondria must be due to enzymes external to the organelle. Freeze-thawing of the identical sample disrupts the integrity of the organelle and the activity we measure increases dramatically. This result demonstrates that the activity is sequestered. Centrifugation of the extract separates soluble enzymes from membrane-bound enzymes. Cytochrome c oxidase is found in the pellet, while the nucleotidyltransferase is not (Table I), suggesting that the enzyme is soluble. Finally, the release of the enzyme by increasing concentrations of digitonin parallels the release of the known matrix enzyme malate dehydrogenase (Fig. 1). Therefore, we conclude that the enzyme activity we measure is indeed inside the mitochondria and appears to be a matrix enzyme. We have highly purified the mitochondrial enzyme (Table  11) using a procedure we previously developed for the purification of the nuclear/cytoplasmic enzyme (30). We characterized the organelle enzyme with regard to its K,,, for ATP and CTP, its pH optimum, and its isoelectric point, and none of these characteristics differed significantly from those of the cytoplasmic enzyme (data not shown). In addition, when the two enzymes were compared by sodium dodecyl sulfate-polyacrylamide gel electrophoresis they appeared identical (Fig.  2). This apparent identity, and the precedent for the same gene providing enzymes to mitochondria and the rest of the cell, led us to hypothesize that the CCAl gene known to code for the enzyme in the nuclear/cytoplasmic compartment (29) also coded for the mitochondrial activity.  One prediction of this hypothesis is that shorter mitochondrial tRNAs should accumulate when strain 352-1A, containing a temperature-sensitive allele of the CCAl gene, ccal-1 (29), is grown at the restrictive temperature. To determine if ccal-1 affects mitochondrial tRNA biosynthesis we carried out the following experiment. A culture of 352-1A cells was grown at the permissive temperature, then divided, and one half was continued at the permissive temperature, while the other was shifted to the restrictive temperature. At various time intervals we removed samples, prepared tRNA, and probed for either cytoplasmic or mitochondrial tRNA using oligonucleotide probes specific for a cytoplasmic tRNAArC and a mitochondrial tRNA"'". The results presented in Fig. 3 demonstrate that shorter tRNAs do accumulate in both cellular compartments, although the amount of shorter tRNA relative to longer tRNA is greater in the cytoplasm than in the mitochondria. The result of this experiment is consistent with our hypothesis, but it could also be explained if the decrease in cytoplasmic protein synthesis caused by the lack of a CCA end on tRNA led to a decrease in the synthesis of a mitochondrial specific isoform of the enzyme.
Homology to the amino terminus of the E. coli enzyme begins at amino acid 90 of the longest open reading frame predicted by the yeast gene sequence (29). This suggested to us that some of the first 90 amino acids might not be necessary for enzyme activity. The importance of amino-terminal sequences to the import of nuclear coded mitochondrial proteins (37,38) is well known, and we reasoned that we might remove enough of the amino-terminal amino acids to prevent localization of the protein to mitochondria without abolishing the  Table I1 and cytoplasmic enzyme isolated as described in Ref. 30 and analyzed by polyacrylamide gel electrophoresis. Lane 1, cytoplasmic enzyme; lane 2, a mixture of mitochondrial and cytoplasmic enzyme; lane 3, mitochondrial enzyme. Numbers a t the right refer to molecular weight of the size standards ( l a n e 4 ) . essential nuclear/cytoplasmic function of this enzyme. The lack of nucleotidyltransferase activity in mitochondria would result in an inability of the cells to carry out mitochondrial protein synthesis so that they would be unable to grow on non-fermentable carbon sources.
We altered the CCAI gene such that we changed the first or the first two ATGs to ATC or ATC and CAG (Fig. 4A). Plasmids carrying no CCAl gene (YCp50), a wild type copy of the CCAl gene (CCAI ATG-ATG), or CCAl genes with either the first ATG (CCA 1 ATC-ATG) or the first and second ATG missing (CCAI ATC-CAG) were transformed into the ccal-l mutant. T o determine whether these altered genes could complement the temperature-sensitive lethal phenotype, we tested their ability to grow on glucose medium at 23 and 37 "C. All of the plasmids containing the CCAl gene were able to grow at both temperatures demonstrating that sufficient nucleotidyltransferase activity to support life is made from the mutant genes (Fig. 4B). We then tested the ability of these cells to grow under conditions that require mitochondrial function by use of glycerol plates. As can be seen in Fig.  4C, cells transformed with the wild type gene grow fine on glycerol a t 23 "C. Growth on glycerol at 37 "C is poor for all of the transformants. Nonetheless, it is clear that there is no growth a t 37 "C when the vector alone is present and cells transformed with the gene predicted to produce a protein with a 9-amino acid amino-terminal truncation (CCAl ATC-ATG) or with a 17-amino acid amino-terminal truncation (CCAI ATC-CAG) show impaired growth on glycerol compared with those containing the wild type gene (CCAI ATG-ATG). The cells with the 9-amino acid truncation, however, seemed to show better growth on glycerol than the cells producing the 17-amino acid truncation protein. Thus the altered genes cannot compensate for the temperature-sensitive phenotype on glycerol medium, whereas they can do so on glucose medium. This demonstrates that the CCAl gene product is required for mitochondrial function.
If the glycerol-negative growth phenotype of the cells transformed with mutant genes is due to a lack of import of the CCA enzyme, then the enzyme activity in mitochondria from transformants with mutant genes should be decreased. The ccal-I strain transformed with vector alone, with the vector carrying the wild type CCAl gene, or with the vector carrying the CCAl gene without the first ATG or without the first two ATGs were grown for four generations a t 23 "C in rich media to assure a good yield of mitochondria from respiratory competent cells. Mitochondria were isolated and the CCA enzyme measured. We have already shown that the enzyme from the temperature-sensitive mutant is much less active than wild type enzyme regardless of temperature of assay (29). This means that, in vitro, the activity measured in the transform-

FIG. 3. Accumulation of shorter
tRNAs in a strain with the ccal-1 allele. Cultures of a strain carrying the ccal-1 allele were grown a t either 23 or 37 "C. tRNA samples taken at hourly intervals were separated by polyacrylamide gel electrophoresis, transferred to paper, and hybridized with nuclear and mitochondrial tRNA-specific probes. ants is derived from the plasmid and not from the chromosomal temperature-sensitive ccal-1 allele. Cells containing vector alone had the least mitochondrial CCA activity (Table  111). Mitochondria from the strain carrying the wild type gene had 500 units of CCA enzyme activitylmg of mitochondrial protein, whereas those carrying the mutant genes had about 10-fold less activity. In contrast, cytoplasmic CCA activity is detected in extracts from all of the transformants except when only the vector is present. All of the mitochondria had comparable levels of cytochrome oxidase activity which confirms that at 23 "C mitochondrial biogenesis proceeds. There was little cytoplasmic contamination in these mitochondria. No glucose-6-phosphate dehydrogenase activity was detected in the mitochondrial extracts (data not shown).

DISCUSSION
We have demonstrated that yeast mitochondria contain a nucleotidyltransferase activity. Although this was the expected result since the yeast mitochondrial tRNA genes do not encode the CCA end, detection of the activity had not been reported previously. The fact that it behaves as a matrix enzyme is also expected, since mitochondrial tRNA biosynthesis occurs in this compartment and the tRNAs function there. Our comparison of the activities from mitochondria and the rest of the cell, although not extensive, did suggest that the enzyme in different compartments could be the same. The results, demonstrating that mitochondrial tRNAs are affected when strains carrying the ccal-l allele are shifted to the nonpermissive temperature, are consistent with this idea. The results of the site-directed mutagenesis experiments are unambiguous; a mitochondrial phenotype can be caused by altering the open reading frame such that amino-terminal truncated proteins would be formed. Mitochondrial nucleotidyltransferase activity, but not cytoplasmic activity, is markedly affected by these changes.
Nucleotidyltransferase activity has also been detected in wheat mitochondrial extracts (12), but the relationship of this activity to that found in the rest of the cell is not known. One interesting possibility is that mitochondria, chloroplasts, and the nuclear/cytoplasmic compartments all share the same gene product. Rat liver mitochondria are reported to contain 30% of the total nucleotidyltransferase activity, but the activities in the two compartments differ in their properties (39,40). It is possible that these differences would not persist in purified preparations and the differences observed are more apparent than real. Alternatively, maybe the unusual structures of some mammalian mitochondrial tRNAs relative to cytoplasmic tRNAs would preclude sharing of this particular enzyme between the mitochondria and the rest of the cell. At least one enzyme, fumarase, appears to be coded by a single gene but shared between the mitochondria and cytoplasm in mammals (41,42).
The MOD5 (22), FUMl (23), LEU4 (24), VAL1 (25), H T S l (26), and T R M l (43) genes contain two in-frame ATG codons that are both used to produce proteins differing in the presence or absence of an amino-terminal extension. The longer proteins are either exclusively (22)(23)(24)(25)(26) or more efficiently (43) imported into mitochondria, while the shorter form of the protein fulfills non-mitochondrial requirements for the activity. There are three in-frame ATGs in the CCAl gene prior to sequences that are homologous to the E. coli enzyme so different sized proteins could, in theory, be produced from the CCAl gene. Yet, all of the enzyme appears to be the same size in yeast. The fact that a phenotype is observed when the first ATG is removed demonstrates that it is used as a translational start site in the wild type gene. There are several explanations for our observation that mitochondrial and cytoplasmic activities comigrate on 10% polyacrylamide gels. The first is that translation does start at multiple ATGs, but our gels did not separate the different sized proteins that would result. The second is that all of the enzyme is produced from the first ATG, but only a portion of it is imported into mitochondria. Only a portion of the longer form of the MOD5 gene product is targeted to mitochondria and what remains in the cytoplasm is active (22). The third possibility is that there are two forms of the protein produced by the CCAl gene, and the longer form, like most proteins imported into the mitochondrial matrix, is processed to remove amino-terminal sequences upon import (37). Such a processing event could result in a mitochondrial protein about the same size as a cytoplasmic protein even though they were originally initiated a t different ATGs. Analysis of the mRNAs coded by this gene, amino-terminal sequences of the proteins from different cellular compartments, and additional site-directed mutagenesis experiments will be necessary to understand how the CCAl gene provides nucleotidyltransferase to multiple compartments.
The evidence that the enzyme is in mitochondria in yeast is direct (Table I, Fig. 1). While there have been no studies that have measured the activity in cytoplasm versus nuclei in yeast, in Xenopus oocytes about 30% of oocyte tRNA nucleotidyltransferase is in the nucleus (44). Despite the lack of direct activity measurements, it is clear that precursors that contain intervening sequences already have their CCA ends added (5,6). Thus CCA addition precedes the removal of intervening sequences which is known to occur in the nucleus in yeast (45). Nucleotidyltransferase participates in the repair of the CCA end in yeast (8), and we presume that this occurs in the cytoplasm. Certainly the activity is present in the cytoplasm of rat liver cells (40) and Xenopus oocytes (44). Thus the CCAl gene product must function in three cellular compartments. Our results demonstrate that amino-terminal sequences do play a role in providing the activity to mitochondria. The enzyme at 62.5 kDa is slightly larger than the 60-kDa size commonly given as the size which would preclude diffusion into nuclei (46), but even much smaller proteins contain nuclear import signals that increase their rate of import into nuclei (47). A scan of the protein sequence does not immediately identify a stretch of basic amino acids similar to other known nuclear targeting sequences (48). Future experiments will be necessary to address the nuclear import of this enzyme and to determine how a portion of it comes to remain in the cytoplasm for tRNA repair.