A 2′-Phosphotransferase Implicated in tRNA Splicing Is Essential in Saccharomyces cerevisiae *

The last step of tRNA splicing in the yeastSaccharomyces cerevisiae is catalyzed by an NAD-dependent 2′-phosphotransferase, which transfers the splice junction 2′-phosphate from ligated tRNA to NAD to produce ADP-ribose 1"-2" cyclic phosphate. We have purified the phosphotransferase about 28,000-fold from yeast extracts and cloned its structural gene by reverse genetics. Expression of this gene (TPT1) in yeast or in Escherichia coli results in overproduction of 2′-phosphotransferase activity in extracts. Tpt1 protein is essential for vegetative growth in yeast, as demonstrated by gene disruption experiments. No obvious binding motifs are found within the protein. Several candidate homologs in other organisms are identified by searches of the data base, the strongest of which is inSchizosaccharomyces pombe.

The last step of tRNA splicing in the yeast Saccharomyces cerevisiae is catalyzed by an NAD-dependent 2phosphotransferase, which transfers the splice junction 2-phosphate from ligated tRNA to NAD to produce ADPribose 1؆-2؆ cyclic phosphate. We have purified the phosphotransferase about 28,000-fold from yeast extracts and cloned its structural gene by reverse genetics. Expression of this gene (TPT1) in yeast or in Escherichia coli results in overproduction of 2-phosphotransferase activity in extracts. Tpt1 protein is essential for vegetative growth in yeast, as demonstrated by gene disruption experiments. No obvious binding motifs are found within the protein. Several candidate homologs in other organisms are identified by searches of the data base, the strongest of which is in Schizosaccharomyces pombe.
tRNA splicing is essential in both the yeast Saccharomyces cerevisiae and humans, since both organisms contain tRNA gene families whose members all contain intervening sequences. Yeast has 10 such intron-containing tRNA gene families (of the approximately 45 total tRNA gene families) (see Ref. 1 for review), and humans have at least one intron-containing tRNA gene family (2). Since all known eukaryotic nuclear-encoded tRNA Tyr genes contain introns, it is likely that tRNA splicing is essential in all eukaryotes for processing of tRNA genes. tRNA introns are invariably located 1 base 3Ј of the anticodon, and this location is critical for the the first step of splicing. In both Xenopus and yeast the endonuclease binds the mature domain of the precursor tRNA, measures the length of the anticodon stem to locate the intron (3,4), and excises it if the structure at the 3Ј splice site is correct (5,6). The products of the reaction are exons bearing 2Ј-3Ј cyclic phosphates and 5Ј-hydroxyl groups at their ends, as shown in Fig. 1 (7,8).
Joining of the exons involves a ligase that generates a mature sized tRNA bearing a splice junction 2Ј-phosphate (9). The ligase from yeast catalyzes four distinct chemical steps to effect ligation: the 2Ј-3Ј cyclic phosphate at the end of the 5Ј exon is opened to a 2Ј-phosphate by a cyclic phosphodiesterase activity; the 5Ј-OH at the beginning of the 3Ј exon is phosphorylated by a polynucleotide kinase activity in the presence of GTP; the 5Ј-phosphate is activated by adenylylation from ATP; and then ligation occurs with loss of the adenylate moiety (9 -11). The result of ligation is a mature sized tRNA bearing a splice junction 2Ј-phosphate (see Fig. 1). A ligase present in wheat germ (12,13), Chlamydomonas (14), and humans (15) also generates splice junctions with a 2Ј-phosphate, and the wheat germ protein is very similar in catalytic activities to the yeast enzyme (16,17). Since removal of the 2Ј-terminal phosphate prevents the yeast ligase from working in vitro (10,18), the 2Ј-phosphate is likely formed at the splice junction when this ligase acts in vivo. The yeast ligase is known to be responsible for tRNA splicing in yeast, since conditional ligase mutants accumulate unligated tRNA exons under nonpermissive conditions (19). However, a second ligase, which uses a completely different chemical reaction and does not generate a splice junction 2Ј-phosphate, has been implicated in tRNA splicing in humans in vitro (20,21) and in Xenopus oocytes in vivo (22).
Removal of the splice junction 2Ј-phosphate occurs by a highly unusual reaction: a 2Ј-phosphotransferase transfers the splice junction phosphate to NAD, forming the novel NAD derivative, ADP-ribose 1Љ-2Љ cyclic phosphate (ApprϾp) 1 (23). Two lines of evidence support the claim that the yeast enzyme catalyzes this step in the cell (24,25). First, the phosphotransferase is highly specific for substrates bearing an internal 2Ј-phosphate; an oligonucleotide bearing an internal 2Ј-phosphate is efficiently dephosphorylated, whereas oligonucleotides terminating with 5Ј-, 3Ј-, 2Ј-, or 2Ј-3Ј cyclic phosphates are not detectably dephosphorylated. Second, this is the only activity detected in crude extracts that can efficiently remove the 2Јphosphate from ligated tRNA. A similar 2Ј-phosphotransferase has been described in HeLa cell extracts; like the yeast enzyme the HeLa enzyme is highly specific for substrates with internal 2Ј-phosphates and is the only activity that can efficiently dephosphorylate 2Ј-phosphorylated ligated tRNA (25). Moreover, it is likely that the phosphotransferase can act in vivo on tRNA substrates: Xenopus oocytes injected with 2Ј-phosphorylated ligated tRNA catalyze formation of ApprϾp concomitant with dephosphorylation (23).
To begin to study the role of the phosphotransferase in yeast, we have purified the protein and cloned its structural gene (TPT1; tRNA 2Ј-phosphotransferase). Phosphotransferase was purified ϳ28,000-fold, the N-terminal amino acid sequence was determined, and the appropriate DNA was isolated by colony hybridization of a yeast genomic library. The identified ORF was expressed in Escherichia coli and shown to catalyze 2Јphosphotransferase activity, implying that phosphotransferase is a single catalytic polypeptide. 2Ј-Phosphotransferase is essential for vegetative growth in yeast, as demonstrated by analysis of strains with chromosomal deletions in the TPT1 gene. The sequence of the phosphotransferase does not reveal any obvious binding or catalytic motifs. Several significantly similar ORFs are identified by searches of the data base, including a particularly strong one in Schizosaccharomyces pombe.

EXPERIMENTAL PROCEDURES
Preparation of Ligated tRNA-Ligated tRNA Phe with a 32 P-labeled splice junction 2Ј-phosphate was prepared by in vitro endonucleolytic cleavage and ligation (with partially purified enzymes) of an [␣-32 P]ATP-labeled pre-tRNA Phe transcript (26). The 32 P-labeled pre-tRNA Phe transcript (340 Ci/mmol) was derived from T7 RNA polymerase transcription of a plasmid-borne copy of the end-matured pre-tRNA Phe gene (27).
2Ј-Phosphotransferase Assay-Transfer of the 2Ј-phosphate from ligated tRNA to NAD to form ApprϾp was performed as described by McCraith and Phizicky (26) in 10 l reaction mixtures in phosphotransferase buffer (20 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 2.5 mM spermidine, 0.1 mM DTT, and 0.4% Triton X-100) containing 1-4 fmol of ligated tRNA, 1 mM NAD, and 1 l of 2Ј-phosphotransferase diluted in Buffer B containing 100 g/ml BSA. Reaction mixtures were incubated at 30°C for 20 min and applied to polyethyleneimine-cellulose plates that were developed in 2 M sodium formate, pH 3.6, to separate ApprϾp from the tRNA (26). One unit of activity corresponds to the amount of phosphotransferase required to transfer 50% of the 2Ј-phosphate from 1 fmol of ligated tRNA to NAD in 20 min, determined by serial dilution.
Growth of Yeast-YP medium contains 1% yeast extract and 2% peptone, minimal medium is described by Sherman (29), and 5-fluoroorotic acid medium is described by Boeke et al. (30). Strain JHRY-20-2Ca was grown in a 100-liter fermentor (New Brunswick model IF130) at 30°C to an A 600 of 4 in YP medium containing 2% glucose, 0.083 g/ml streptomycin, 0.033 g/ml penicillin, 0.5% ethanol, and 0.17% polyethylene glycol, and cells were chilled with ice and harvested in a Westfalia Clarifier (Centrico Inc., Northvale, NJ). To measure phosphotransferase expression, strain JHRY-20-2Ca transformed with either the 2 LEU2 vector yEPlac181 or with pGMC5 (yEPlac181 TPT1 ϩ ) was grown at 30°C in minimal medium lacking leucine to A 600 ϭ 0.45. SC466 transformed with either the GAL10 promoter vector pBM150 (CEN URA3 P GAL10 ) or with pGMC21 (pBM150 P GAL10 -TPT1) was grown overnight in minimal medium containing 2% raffinose and lacking uracil, and cells were inoculated directly into 75-ml cultures of YP ϩ 2% galactose or YP ϩ 2% glucose and grown for 2.5 generations. Cells were harvested, washed, resuspended, and made into extracts as described below for the purification of phosphotransferase.
Disruptions of the TPT1 gene were generated by insertion of a LEU2 HindIII fragment (obtained from a polylinker-inserted LEU2 SspI chromosomal fragment, modified to contain HindIII ends) into one of two positions within the TPT1 gene, as illustrated in Fig. 4B: to replace the HindIII fragment of 226 nucleotides around the ATG translation start of pGMC7, generating pGMC22 (tpt1-⌬1::LEU2), or to replace the 744 base pair HindIII-AflII fragment of pGMC7, after filling in the AflII end and adding a HindIII linker, generating pGMC17 (tpt1-⌬2::LEU2).
pGMC10 contains the TPT1 ORF bounded by engineered EcoRI sites, ligated into pSP72. It was generated by polymerase chain reaction amplification of the TPT1 gene from the ATG of the ORF to a site 240 nucleotides downstream of the TAA termination codon (see Fig. 4B), using Taq polymerase and primers dP-1 (AGAGGAATTCACAGGTG-GCCGAAGGTGCTGCC) and dP-2 (GACGAATTCATGCGCCAGGTAC-TACAAAAAG), followed by EcoRI digestion, ligation into the vector, and sequencing of transformants to identify an otherwise unaltered gene. The EcoRI fragment of pGMC10 was inserted into pBM150 (34) to place the TPT1 gene under control of the yeast GAL10 promoter (pGMC21) and into pKK223-3 (34) to place the TPT1 gene under control of the E. coli tac promoter (pGMC9).
Expression of Tpt1 Protein in E. coli-E. coli strain RZ510 (relevant genotype lac i SQ ) was transformed with pKK223-3 vector or with pGMC9 to express the TPT1 gene. 50-ml cultures were grown at 37°C in L broth to an A 600 of 0.8, induced with 1 mM isopropyl-␤-D-thiogalactopyranoside for 1 h, and cells were harvested, resuspended in 2 ml of buffer containing 50 mM Tris 7.5, 1 mM EDTA, 450 mM NaCl, 5 mM DTT, and 10% glycerol, and sonicated on ice in 10 s bursts for 1 min. Extracts obtained after centrifugation (12.5 mg/ml) were aliquoted, frozen, and thawed to measure phosphotransferase activity.
Protein Concentration and Visualization-Protein concentration was determined using Bradford reagent (Bio-Rad). Polypeptides were visualized after SDS-polyacrylamide gel electrophoresis by silver staining, as described (35).
Protein was precipitated by the addition of solid ammonium sulfate to 80% saturation, followed by centrifugation at 14,700 ϫ g for 45 min. 2Ј-Phosphotransferase Gene Implicated in tRNA Splicing glycerol) containing 59% saturated ammonium sulfate, and 2Ј-phosphotransferase activity was recovered by two successive extractions of the 59% pellet with 2 liters of Buffer A containing 46% saturated ammonium sulfate, followed by centrifugation to remove the pellet. Protein in the combined supernatants was reprecipitated by the addition of solid ammonium sulfate (to 80% saturation) and centrifugation, and the pellet was resuspended in 500 ml of Buffer B (20 mM Tris, pH 7.5, 2 mM EDTA, 4 mM MgCl 2 , 1 mM DTT, and 10% (v/v) glycerol) containing 350 mM NaCl, and dialyzed overnight against 20 liters of this buffer. Dialyzed protein was applied to a 250 ml blue Sepharose CL-6B column (Pharmacia Biotech Inc.) equilibrated in the same buffer, washed, and active fractions (700 ml) flowed through the column. These fractions were pooled, diluted 2-fold with Buffer B lacking NaCl, reapplied to the blue Sepharose column equilibrated with Buffer B containing 0.175 M NaCl, and retained protein was eluted with a 2-liter linear gradient of Buffer B from 0.175 to 2 M NaCl. 2Ј-Phosphotransferase eluted at 450 -500 mM NaCl.
Active fractions (350 ml) from the blue Sepharose column were dialyzed against Buffer B containing 40 mM NaCl, applied to a 250-ml heparin-Sepharose column, and activity was eluted with a gradient of Buffer B containing 40 -900 mM NaCl. Active fractions (25 ml) were loaded directly onto a 15-ml hydroxylapaptite column (DNA grade Bio-Gel HTP, Bio-Rad) equilibrated in Buffer C (20 mM Tris, pH 7.5, 0.5 mM EDTA, 1 mM MgCl 2 , 10% (v/v) glycerol, 1 mM ␤-mercaptoethanol, and 50 mM NaCl), and phosphotransferase activity was eluted with a linear gradient of Buffer C containing 0 -0.25 M K 2 HPO 4 (pH 7.5 with phosphoric acid). Peak fractions from the hydroxylapatite column (8 ml) were dialyzed against Buffer B containing 40 mM NaCl and 20% glycerol, applied to a 0.5-ml orange A Sepharose column (Amicon Division, W. R. Grace & Co., Beverly, MA), and 2Ј-phosphotransferase was eluted with a gradient of Buffer B from 40 to 700 mM NaCl. Active fractions, (250 l each), were dialyzed individually against Buffer B containing 50% (v/v) glycerol and 55 mM NaCl and subsequently stored at Ϫ20°C. Little loss of activity was observed over 3 months.
A streamlined purification yielded material that was about 4-fold less pure. Crude extracts were dialyzed against Buffer B containing 0.15 M NaCl, loaded directly on a 500-ml blue Sepharose column, and retained protein was eluted with a 2.5-liter gradient of Buffer B containing 150 mM to 1.5 M NaCl. Active fractions (which elute at higher salt concentration (0.6 M) than in the preparation above) were then subjected to chromatography on heparin-agarose, hydroxylapatite, and orange A Sepharose as described above.
Protein Sequencing-100 -200 pmol of purified phosphotransferase was resolved on a 12% SDS-polyacrylamide gel, transferred electrophoretically to polyvinylidene difluoride membranes, stained with Coomassie Blue R-250, destained, and used for sequencing essentially as described by Matsudaira (36). Transfer was done in a Bio-Rad transblot apparatus at room temperature for 4 h at 0.65 mA/square cm or 300 mA in transfer buffer containing 39 mM glycine, 48 mM Tris base, and 20% methanol.
UV Cross-linking of tRNA to Phosphotransferase Fractions-RNAprotein cross-linking reactions were assembled in 15 l of phosphotransferase assay buffer containing 200 g/ml BSA, 180,000 cpm of spliced tRNA, 750 units (approximately 5 ng) of 2Ј-phosphotransferase from the orange A column peak, 5 mM AMP, and no NAD. Reaction mixtures were preincubated for 15 min at 30°C, and proteins were cross-linked to RNA for 10 min at room temperature, using a UV source at 254 nm (UV-C Bleit 155 lamp, Spectronics Corp., Westbury, NY) with an intensity of 2 milliwatts/cm 2 . Samples were supplemented with 55 l of 10 mM Tris, pH 7.5, and digested with 7.5 units of ribonuclease T1 at 50°C for 45 min. Proteins were precipitated at Ϫ20°C by the addition of trichloroacetic acid to 10%, followed by centrifugation to pellet the protein, washing with ice cold acetone, resuspension in SDS-polyacrylamide gel electrophoresis loading buffer, boiling for 15 min, and electrophoresis on a 12% SDS-polyacrylamide gel.

Purification of 2Ј-Phosphotransferase Implicates a 30-kDa
Polypeptide-The purification of 2Ј-phosphotransferase yields a prominent polypeptide of 30 kDa, which comigrates with activity. This is illustrated in Fig. 2, in which fractions from the final purification step were analyzed for both polypeptides and phosphotransferase activity. The amount of 30-kDa polypeptide closely parallels phosphotransferase activity in different fractions, both across the final orange A column (compare fractions 20 -36 in Figs. 2, A and B) and in the peak fractions from the heparin-agarose and hydroxylapatite columns of the purification ( Fig. 2A). Three other minor polypeptides are visible in this preparation of 2Ј-phosphotransferase activity, with apparent molecular masses of 52, 45, and 20 kDa. Neither of the chromatographic profiles of the 52-kDa or the 45-kDa polypeptides corresponds to the observed activity peak from the orange A column (Fig. 2); however, the profile of the 20-kDa polypeptide does appear to correspond to the observed activity peak from this column. The same 30-kDa polypeptide, but not the 20-kDa polypeptide, also copurifies with activity using a streamlined purification procedure (see "Experimental Procedures"), and if the material from the penultimate step of this streamlined procedure is chromatographed on DEAE or on another heparin-agarose column (with isocratic elution) instead of the orange A column. Thus it seemed likely that the 30-kDa polypeptide is the limiting component responsible for 2Ј-phosphotransferase activity.
The 30-kDa Protein Cross-links to Spliced tRNA-The suggestion that the 30-kDa polypeptide is a component of the phosphotransferase is supported by the observation that it can be cross-linked to its substrate. As shown in Fig. 3, ligated tRNA bearing a 2Ј-phosphate, prepared from [␣-32 P]ATP-labeled pre-tRNA transcript, is cross-linked to a polypeptide of 30 kDa (band B1) from the peak orange A fraction, as visualized after RNase T1 treatment of the sample and subsequent electrophoresis through an SDS-polyacrylamide gel. This crosslinking requires both phosphotransferase and UV illumination. No UV-dependent cross-linking is observed to BSA, which was deliberately present at a 600-fold higher concentration than 2Ј-phosphotransferase and to the contaminating 52-and 45-kDa polypeptides. A band migrating at around 20 kDa (band B2) was present in all lanes, and thus represents an RNase T1-resistant background. Based on the apparent molecular mass of the cross-linked polypeptide (band B1), we conclude that the 30-kDa polypeptide comigrating with activity in the purification is the cross-linked protein. Separate experiments with less purified fractions demonstrate that the interaction of the 30-kDa polypeptide with tRNA is specific: cross-linking occurs with ligated tRNA bearing a 2Ј-phosphate, but not with dephosphorylated ligated tRNA (data not shown). Furthermore, cross-linking is inhibited by an excess of unlabeled synthetic substrate (U p pU, which contains a 2Ј-phosphate) (23,37) at a concentration of 50 M, but is not inhibited by the corresponding nonsubstrate (UpU) at a concentration of 10 mM (data not shown). Since the same size polypeptide copurifies with activity and cross-links to tRNA substrate, it is likely that this is one of the components of the phosphotransferase. Since, in addition there is reasonably good yield at each step of the purification (see Table I), and there is only one major copurifying polypeptide, it is likely that this polypeptide is the only catalytic subunit of the phosphotransferase. This conclusion is substantiated further below.
Isolation of the Phosphotransferase Gene-To clone the structural gene for the phosphotransferase, we sequenced the Nterminal end of the 30-kDa protein in the final orange A column, after transfer of the polypeptides to polyvinylidene difluoride paper and excision of the appropriate region. We obtained 14 amino acids of N-terminal sequence. The same sequence was also obtained from two other experiments with equivalent, but less pure, material obtained from the stream-lined purification procedure, which contained slightly different contaminants. Furthermore, the yield of amino acids obtained from the initial steps of the sequencing run corresponded roughly to the number of moles of 30-kDa polypeptide subjected to sequencing. Thus, we were likely sequencing the major polypeptide present at 30 kDa and not a minor contaminant.
The 2Ј-phosphotransferase gene (TPT1) was isolated by reverse genetics. The sequence of 14 amino acids was compared with the information in the yeast data base, and a single perfect match was found at the N-terminal end of an ORF of 26.2 kDa, located on chromosome XV. The sequence of this ORF is shown in Fig. 4A. No other near matches were found in the yeast data base. An oligonucleotide probe was designed from the data base DNA sequence, and this probe was used to isolate the gene from a yeast genomic library by colony hybridization, as described under "Experimental Procedures." A schematic of a portion of chromosome XV that was isolated is illustrated in Fig. 4B, showing the position of the TPT1 ORF and neighboring ORFs. Plasmids containing the phosphotransferase gene, constructed as described under "Experimental Procedures," were then used to confirm that the gene encodes 2Ј-phosphotransferase.
TPT1 Encodes the 2Ј-Phosphotransferase-Overproduction of Tpt1 protein in yeast results in overproduction of 2Ј-phosphotransferase activity. This was established in two experiments, which are summarized in Table II. First, high gene dosage results in overproduction of activity. Extracts made from a strain bearing a high copy plasmid containing the gene and its regulatory regions (2-m TPT1) have about 55-fold more phosphotransferase activity than extracts from the same strain bearing the plasmid vector alone (Table II). Second, placing the open reading frame under control of a regulatable promoter in yeast results in regulated overproduction of the phosphotransferase activity. To this end, a plasmid was constructed with the TPT1 open reading frame immediately downstream of the GAL10 promoter and transcription start site, as described under "Experimental Procedures." A strain bearing this plasmid (P GAL10 -TPT1) overproduces phosphotransferase about 20-fold when grown in galactose, which induces transcription, compared with the phosphotransferase produced when cells are grown in glucose, which represses transcription (Table II). These two experiments demonstrate that expression of the TPT1 gene is the limiting factor determining the observed 2Ј-phosphotransferase activity in yeast extracts.
To establish that TPT1 encodes the 2Ј-phosphotransferase, the gene was expressed in E. coli. As shown in Table III, extracts from an E. coli strain bearing the expression vector pKK223-3 have little, if any, detectable 2Ј-phosphotransferase activity. However, extracts from an E. coli strain containing the TPT1 ORF fused immediately downstream of the hybrid trp-lac promoter of pKK223-3 have 10 6 -fold more 2Ј-phosphotransferase activity than the control extracts when expression is induced in the presence of isopropyl-␤-D-thiogalactopyranoside. Since the TPT1 gene encodes a protein of 26.2 kDa, which has 2Ј-phosphotransferase activity, and its size is nearly the same as that identified in the purification (Fig. 2), and in crosslinking experiments (Fig. 3), it is highly likely that Tpt1 protein is the phosphotransferase protein.
2Ј-Phosphotransferase produced by expression of the TPT1 gene in E. coli is very similar to that purified from yeast. Both proteins require NAD to catalyze removal of the splice junction 2Ј-phosphate from ligated tRNA, and the NAD dependence is similar (half-maximal activity at 1 unit of Tpt1 protein requires about 10 -20 M NAD). Furthermore, Tpt1 protein expressed in E. coli also transfers the splice junction 2Ј-phosphate from ligated tRNA to NAD to produce ApprϾp, as measured by  (lanes a and b) or appropriately diluted buffer (lanes c and d) was incubated with 180,000 cpm of ␣-32 P-labeled-spliced tRNA in phosphotransferase buffer containing 200 g/ml BSA and no NAD, illuminated with 254-nm light, and cross-linked proteins were detected after RNase T1 digestion by electrophoresis on an SDS-polyacrylamide gel, as described under "Experimental Procedures." B1, the major cross-linked species observed with phosphotransferase, migrating as a doublet; B2, a background RNase T1-resistant UV-independent band observed in all lanes.
comigration of the product (in several different thin layer chromatography systems) with ApprϾp made by the purified yeast protein, as well as by phosphatase resistance of the product (data not shown). Similarly, Tpt1 protein produced in E. coli, like the yeast enzyme, prefers substrates with tRNA structure: for each protein 2Ј-phosphorylated tRNA is about 50-fold more efficient a substrate than a synthetic oligonucleotide (pUp-U p pU), which contains a 2Ј-phosphate. 2 Assuming that E. coli extracts do not fortuitously supply missing factors that can act in concert with Tpt1 protein, it is likely that 2Ј-phosphotransferase is a single polypeptide that can recognize its substrates and catalyze the complete phosphotransfer reaction.
Tpt1 Protein Is Essential for Vegetative Growth-Removal of the 2Ј-phosphate from ligated tRNA is likely to be critical for tRNA function. Since the splice junction 2Ј-phosphate is 1 base 3Ј of the anticodon, its bulk and charge would be expected to interfere with anticodon recognition and thus impair growth. If Tpt1 protein is the enzyme that catalyzes removal of the splice junction 2Ј-phosphate from ligated tRNA in vivo, then cells lacking this protein would likely be dead (or very sick).
Gene disruption experiments demonstrate that the TPT1 gene is essential. To show this, we did a standard one-step gene disruption experiment, in which we replaced one allele of the TPT1 gene in a diploid with a copy of the LEU2 gene, as described under "Experimental Procedures," and sporulated the resulting diploid. Two separate gene disruptions were made by replacement of a fragment spanning the ATG of the ORF with the LEU2 gene: in tpt1-⌬1::LEU2, a 226-nucleotide HindIII fragment extending from Ϫ174 in the promoter to ϩ53 in the coding region was replaced; and in tpt1-⌬2::LEU2 a 744 nucleotide HindIII-AflII fragment from Ϫ174 in the promoter to ϩ570 in the coding region was replaced (see Fig. 4B). Southern analysis confirmed in both cases that the chromosomal DNA from the transformant diploids contained one normal sized copy of the TPT1 gene and one copy of the TPT1 gene that was altered by the presence of the LEU2 gene (data not shown). In SC804 (relevant genotype: TPT1 ϩ /tpt1-⌬1::LEU2) 6 tetrads were examined; all segregated two live:two dead spores, and all the live spores lacked the LEU2 marker. Similarly in SC805 (relevant genotype: TPT1 ϩ /tpt1-⌬2) seven tetrads were examined and all segregated two live Leu Ϫ spores and two dead spores. These are the expected results if disruption of the TPT1 gene is lethal. Since, in addition, microscopic examination of the nonsurviving spores demonstrated that they germinated and grew into microcolonies, these results suggest that the TPT1 gene is essential.
To confirm that the lethality of the disruptions was caused by lack of the TPT1 gene itself, we demonstrated that the TPT1 gene on a plasmid could complement the deletion and suffice for viability. As shown in Table IV, SC804 (tpt1-⌬1::LEU2/TPT1 ϩ ) carrying a plasmid bearing the TPT1 gene on either a single copy plasmid (CEN URA3) or a multi-copy plasmid (2-m URA3) could readily segregate Leu ϩ (tpt1 Ϫ ) spores as long as the spores also contained a URA3 ϩ TPT1 ϩ plasmid; by contrast, SC804 carrying just a URA3 plasmid segregated only Leu Ϫ spores. Furthermore, only the TPT1 gene itself was required to complement the tpt1-⌬1::LEU2 disruption, since Leu ϩ Ura ϩ spores could be readily recovered if the sporulated diploid carried a URA3 CEN plasmid containing only the TPT1 ORF (fused to the GAL10 promoter) and 236 nucleotides of downstream DNA (to allow for termination of transcription). Since the TPT1 ORF is the only ORF on this plasmid, and it still complements the lethality of the tpt1 disruption, it is highly likely that the TPT1 gene is essential. Furthermore, as expected if the TPT1 gene is essential for vegetative growth, cells bearing the TPT1 deletion require the plasmid-borne TPT1 gene for continued viability. Whereas wild type cells bearing a URA3 plasmid can easily lose such plasmids when the URA3 gene is selected against on media containing 5-fluoroorotic acid, tpt1-⌬1::LEU2 haploid strains bearing the TPT1 gene on a URA3 plasmid cannot lose the plasmid and therefore die on media containing 5-fluoroorotic acid (see Table IV). This is the expected result if expression of the TPT1 gene is necessary for vegetative growth.
Analysis of the TPT1 Gene-The phosphotransferase is a 230-amino acid protein, with a calculated molecular weight of 26,196 and an isoelectric point of 9.30. It is a highly charged protein; fully 30% of the residues are potentially ionized at neutral pH, and there is a substantial excess of basic residues (12 Arg, 20 Lys, 12 His) over acidic residues (11 Asp,14 Glu).
Examination of the amino acid sequence with either GCG or Prosite, or by visual inspection, reveals little about possible domains of the protein. No well characterized RNA binding motifs are found in the Tpt1 protein, including the RNP1, arginine-rich (ARM), RGG, K homology (KH), and doublestranded RNA binding motifs (see Ref. 38 for review). Similarly, there are no sequences identical to the RNA binding motifs of the class I aminoacyl tRNA synthetases, which are involved in binding the amino acid acceptor stem of the tRNA (39,40), and no marked similarities with the less well characterized class II aminoacyl tRNA synthetase sequences (41). Among the tRNA synthetases there is a lack of a consensus site for binding the tRNA anticodon (42,43). The best match to an NAD binding domain is to that of diphtheria toxin (44). The residues of diphtheria toxin (and by analogy the related exotoxin A from Pseudomonas aeruginosa) that contact NAD are characterized by the sequence HGTXXXYXXSIXX(X)GXQ/ RXP/R. A reasonable match is found in the S. cerevisiae sequence beginning at amino acid 117 (HGTNLQSVIKIIES-GAISP). The other well characterized NAD binding site contains the sequence GXGXXG/A, which is found in many dehydrogenases at the end of the first ␤-sheet of a ␤␣␤ ADPbinding fold (45)(46)(47), and which is also found in NAD-requiring poly(ADP-ribose) polymerases (48); no perfect matches are located in the TPT1 sequence. There are also no obvious similarities with the sequence of NAD-dependent DNA ligases, includ-2 S. Spinelli and E. M. Phizicky, unpublished results. ing the region around the adenylylation site (49). The Tpt1 amino acid sequence is conserved in ORFs from several different eukaryotes as well as in an ORF from E. coli. The closest similarity is to an ORF in S. pombe, found on clone c2C4 of the S. pombe sequencing project being done at the Sanger Center in the United Kingdom (http://www.sanger-.ac.uk/ϳyeastpub/svw/pombe.html.). The predicted amino acid sequence of this gene is shown in Fig. 5. The S. pombe ORF is 34% identical and 57% similar to that of S. cerevisiae, with three distinct blocks of 10 or more amino acids where the identity approaches or exceeds 80%. The conservation of sequence extends over the entire length of the S. cerevisiae protein, suggesting that the S. pombe ORF might be a functional homolog of the Tpt1 protein. Tpt1 protein also shares significant conserved amino acid sequence with an ORF in E. coli and several ESTs from higher eukaryotes, as illustrated in Fig. 5. The amount of conserved sequence, and the fact that the conservation is largely in the same regions between all the potential proteins, suggest that these ORFs form a family. DISCUSSION We have cloned the S. cerevisiae TPT1 gene, which encodes the 2Ј-phosphotransferase activity implicated in the last step of tRNA splicing: removal of the splice junction 2Ј-phosphate from ligated tRNA. This gene was cloned from the purified protein by reverse genetics and demonstrated to be authentic by overproduction of the activity in both yeast and E. coli under regulated promoter control. 2Ј-Phosphotransferase appears to be a single catalytic polypeptide (Fig. 1, Table III). The purified protein is the predominant silver-staining band in highly purified preparations, and the bacterially expressed protein catalyzes the same NAD-dependent 2Ј-phosphotransferase reaction in extracts from E. coli. Since formation of ApprϾp is at least a two-step chemical reaction, this single polypeptide likely carries out both steps, if both steps are enzyme-catalyzed.
2Ј-Phosphotransferase is essential for vegetative growth in S. cerevisiae, since a diploid of genotype tpt1-⌬1::LEU2/TPT1 ϩ could not segregate a LEU2 spore unless an exogenous source of phosphotransferase was present on a plasmid, and since the TPT1-containing plasmid could not then be lost from the strain (Table IV). Based on its biochemical activity (23)(24)(25)(26), the lethality caused by a lack of phosphotransferase in yeast is due either to a failure to complete the dephosphorylation step of tRNA splicing or to the lack of ApprϾp in the cell. We favor the former hypothesis. Since all members of each of 10 tRNA gene families have introns in yeast (1), lack of phosphotransferase ought to lead to the accumulation of 2Ј-phosphorylated tRNAs for all members of these gene families. Presence of the 2Јphosphate 1 base 3Ј of the anticodon would seem likely to impair tRNA function at some stage of translation. However, it is conceivable that lack of ApprϾp is also deleterious. To our knowledge this splicing reaction is the only biochemical pathway leading to ApprϾp formation; if it or its downstream  ϪLeucine 500 ϫ 10 3 SC466 ϩ pBM150 (P GAL10 vector) YP ϩ glucose 6.1 ϫ 10 3 SC466 ϩ pBM150 (P GAL10 vector) YP ϩ galactose 6.8 ϫ 10 3 SC466 ϩ pGMC21 (P GAL10 -TPT1) Y P ϩ glucose 8.6 ϫ 10 3 SC466 ϩ pGMC21 (P GAL10 -TPT1) Y P ϩ galactose 170 ϫ 10 3 2Ј-Phosphotransferase Gene Implicated in tRNA Splicing metabolic products has a cellular role, then the failure to make this product might also impair growth. We have recently isolated conditional tpt1 Ϫ mutants to begin to ascertain the consequences of a lack of the protein. 3 The cellular role of ApprϾp would most easily be addressed by altering its levels in vivo.
We have previously identified a highly specific cyclic phosphodiesterase that can convert ApprϾp or rϾp to the correspond-ing ribose-1-P derivative in yeast extracts (50), an activity that appears to be related to a similar activity in wheat germ (50,51). A cyclic phosphodiesterase from Arabidopsis with very similar properties has recently been cloned (52); analysis of its function in Arabidopsis or of the function of the corresponding gene in yeast may directly address the question of the role of ApprϾp in cells.
A crude calculation based on the purification (Table I) indicates that there is on the order of 10 times more phosphotransferase protein in a yeast cell than there is of the other splicing enzymes: tRNA ligase protein and endonuclease (34,53). Endonuclease and ligase may form a complex in vivo, based on their similar populations within the cell, their localization in similar subdomains of the nucleus (7,54), and the concerted splicing reaction observed in vitro with tRNA precursors (55). If such a complex includes the phosphotransferase, there is likely an excess of uncomplexed phosphotransferase. This is consistent with the observation that there is at least 12 times as much 2Ј-phosphotransferase activity in cells as is necessary for normal growth. Since the GAL10 promoter is tightly repressed by glucose-containing medium, it might be expected that a tpt1-⌬1::LEU2 strain with a P GAL10 -TPT1 plasmid would die on glucose. Unfortunately this is not the case; cells with only a galactose-regulated TPT1 gene have wild type growth rates after prolonged growth in glucose. Under these conditions, phosphotransferase activity is down 12-fold from that observed in wild type cells 2 ; thus there is an excess of phosphotransferase in the cell.
The high degree of conservation of the S. cerevisiae Tpt1 sequence with sequences from S. pombe, mouse, rice, and E. coli suggests strongly that these proteins constitute a family. Given the similarities of sequence, it seems reasonably likely that the S. pombe ORF, and perhaps the mouse and rice ORFs, encode functional Tpt1 proteins; these ORFs all share the same regions of conserved sequence with the S. cerevisiae protein (Fig. 5) and similar or more extensive homologies with one another (data not shown). If so, then the regions of conserved sequence presumably contain the as yet uncharacterized binding and catalytic domains of the protein. We note that the putative NAD binding site identified by comparison of the S. cerevisiae sequence to the diphtheria toxin NAD binding site (44) is retained in all the sequences. Further experiments will be required to define this and the other domains.
In the context of a gene family, the presence of a highly conserved E. coli ORF is striking. Since E. coli is not known to splice tRNAs, or to have a ligase like that in yeast that gener- 3  coli (sp͉P39380 YJII_ECOLI), mouse (GenBank™ number W65960); and rice (dbj͉D15111͉RICC0076A). Not shown aligned is a human sequence ((GenBank™ accession numbers H39778, H43264, W23913), which is similar to the others, but has a large insert and a region of duplicated alignment. Black shadings with white lettering, amino acids which are conserved in the S. cerevisiae sequence and any of the other ORFs; *, translation termination signal; -, gaps in the alignment. 2Ј-Phosphotransferase Gene Implicated in tRNA Splicing ates splice junctions with a 2Ј-phosphate, it seems unlikely that this protein encodes a tRNA 2Ј-phosphotransferase. However, the E. coli ORF might encode a related catalytic or binding activity. Understanding the function and/or the biochemical activity of the E. coli protein might therefore lead to an understanding of the origin of the unusual activity catalyzed by the yeast 2Ј-phosphotransferase.