Separation and characterization of 5'- and 3'-tRNA processing nucleases from rat liver mitochondria.

The 5'- and 3'-tRNA processing nucleases have been isolated from rat liver mitochondria. The two activities co-purified through heparin-agarose and phenyl-Sepharose columns and then efficiently separated on a DEAE-cellulose column. The 5' processing nuclease was found in the flow-through fraction, and the 3' processing activity eluted with 0.5 M KCl. Both enzymes were greater than 500-fold purified over the high speed supernatant of a mitoplast extract. The 159-base pre-tRNATyr used as a substrate in this study was synthesized in vitro and contained the Escherichia coli suppressor III tRNATyr plus a 49-base leader sequence and a 25-base trailing sequence. The 5' processing nuclease converted the pre-tRNATyr into two discrete RNA species, identified as the 5'-processed intermediate and the 5' flanking fragment, by endonucleolytic cleavage at the 5' end of the mature tRNATyr sequence. The 3' processing nuclease was inactive with the intact pre-tRNATyr as substrate but efficiently converted the 5'-processed intermediate to the mature tRNATyr, indicating an obligatory order of processing in which 5' maturation was necessary before cleavage by the 3' processing nuclease could occur. The mitochondrial enzymes exhibited optimal activity in the presence of about 2 mM Mg2+, but both enzymes were nearly fully active without addition of exogenous Mg2+ to the reaction mixtures. In contrast, a partially purified 5' processing endonuclease present in the postmitochondrial cytosolic fraction required higher [Mg2+] for activity, thus providing a means for differentiating between these similar enzyme activities obtained from the cytosolic and mitochondrial fractions.

The 5'-and 3'-tRNA processing nucleases have been isolated from rat liver mitochondria. The two activities co-purified through heparin-agarose and phenyl-Sepharose columns and then efficiently separated on a DEAE-cellulose column. The 5' processing nuclease was found in the flow-through fraction, and the 3' processing activity eluted with 0.5 M KCl. Both enzymes were greater than 500-fold purified over the high speed supernatant of a mitoplast extract. The 159base pre-tRNATYr used as a substrate in this study was synthesized in vitro and contained the Escherichia coli suppressor I11 tRNATy' plus a 49-base leader sequence and a 25-base trailing sequence. The 5' processing nuclease converted the pre-tRNATYr into two discrete RNA species, identified as the 5"processed intermediate and the 5' flanking fragment, by endonucleolytic cleavage at the 5' end of the mature tRNATYrsequence. The 3' processing nuclease was inactive with the intact pre-tRNATYr as substrate but efficiently converted the B'-processed intermediate to the mature tRNATY', indicating an obligatory order of processing in which 5' maturation was necessary before cleavage by the 3' processing nuclease could occur. The mitochondrial enzymes exhibited optimal activity in the presence of about 2 mM Mg"', but both enzymes were nearly fully active without addition of exogenous Mg2+ to the reaction mixtures. In contrast, a partially purified 5' processing endonuclease present in the postmitochondrial cytosolic fraction required higher [Mg2+] for activity, thus providing a means for differentiating between these similar enzyme activities obtained from the cytosolic and mitochondrial fractions.
The current concept of mammalian mitochondrial RNA synthesis is that transcription of both the heavy and light strands of mtDNA is initiated in the noncoding region near the displacement loop, resulting in the production of polycistronic RNA species (for a review see Clayton, 1984). This model was strongly supported by the identification of the major heavy and light strand promoters that have been localized proximal to each other in the region near the displacement loop (Chang and Clayton, 1984;Bogenhagen et d . , 1984; * These studies were supported in part by Grant GM 22597 from the National Institutes of Health. 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. $Present address: Merck, Sharp, and Dohme Research Laboratories, West Point, PA 19486. §Recipient of United States Public Health Service Research Career Development Award Grant CA00676 from the National Institutes of Health, Department of Health and Human Services, during portions of this study. To whom correspondence should be addressed. Chang et al., 1985Chang et al., , 1986. The spatial arrangement of the genes in mammalian mtDNA is extremely economical in that there are few, if any, noncoding nucleotides separating the gene sequences (Anderson et al., 1981(Anderson et al., ,1982Bibb et al., 1981).
A common feature among these genomes is that the genes for rRNAs and most of the protein-encoding genes are flanked by contiguous or nearly contiguous tRNA genes. This observation led to the prediction that most of the required nucleolytic processing of primary mitochondrial transcripts is likely to be carried out by one or several endonucleases that recognize tRNA structures and cleave precisely at the 5' and 3' ends of the tRNA sequences. Cytoplasmic 5'-and 3'-tRNA processing nucleases with these properties have been identified in eukaryotic cells (Frendewey et al., 1985;Castaiio et al., 1985). Characterization of these enzymes in i n vitro assays suggests that nuclear pre-tRNA transcripts are trimmed by precise 5'-and 3'-tRNA endonucleases that function in temporal sequence with 5' processing preceding cleavage at the 3' processing site.
In prokaryotes, 5' processing of tRNA precursors is carried out by the well studied RNase P (Kole and Altman, 1982), a precise endonuclease containing a protein moiety and a catalytically active RNA subunit (Guerrier-Takada et al., 1983).
Maturation at the 3' ends of prokaryotic pre-tRNAs is less well characterized, but both endo-and exonucleases appear to be involved (Kole and Altman, 1982).
Yeast mitochondria contain an RNase P-like enzyme composed of a nuclear-encoded protein moiety and an essential mitochondrial 9 S RNA known to be the product of the mitochondrial tRNA synthesis locus. This enzyme accurately removed the 5' leader sequence from a homologous precursor tRNA by endonucleolytic scission (Hollingsworth and Martin, 1986). The first described mammalian mitochondrial tRNA processing nuclease was an RNase P-like enzyme isolated from HeLa cells (Doersen et al., 1985). This enzyme exhibited the same specificity as Escherichia coli RNase P using the precursor to E. coli suppressor tRNATy' as substrate. The HeLa cell mitochondrial enzyme was sensitive to pretreatment with micrococcal nuclease and Pronase and was differentiated from a similar activity obtained from a cytosolic fraction primarily on the basis of chromatic differences. Heretofore, mitochondrial enzymes involved in the nucleolytic maturation at the 3' end of tRNA precursors have not been described.
In this report, we describe the purification and separation of two rat liver mitochondrial enzymes that sequentially process the 5' leader and 3' trailer sequences from an in vitro synthesized precursor transcript containing the E. coli suppressor tRNATY'. The optimal Mg2+ and KC1 concentrations for the mitochondrial 5'-and 3'-tRNA processing nucleases are similar, but differ substantially from the optimal conditions required by a partially purified 5'-tRNA processing activity obtained from a postmitochondrial cytosolic fraction.
FIG. 1. Nucleotide sequence and probable secondary structure of the precursor transcript used as a substrate in this study. The transcript was produced in oitro by SP6 polymerase from pGEM 1 transcription vector containing the gene for the E. coli suppressor I11 tRNAsr as described under "Experimental Procedures." The complete sequence has been numbered sequentially from the 5' to the 3' end, and every 10th ribonucleotide is identified by a bold dot. The arrows delineate the 5' and 3' ends of the mature tRNAS'.
Portions of these results were presented previously in abstract form (Manam and Van Tuyle, 1986).

EXPERIMENTAL PROCEDURES AND RESULTS'
Partial Purification and Separation of the 5'and 3'-tRNA Processing Nucleases from Rat Liver Mitochondria-Mitochondria were isolated from rat liver tissue by differential centrifugation and treated with digitonin to remove the outer mitochondrial membrane to reduce nuclear and cytosolic contamination of the mitochondrial enzyme preparations. The resulting inner membrane-matrix preparations (mitoplasts) were lysed with 0.5% Triton X-100 in the presence of 0.35 M KC1 and clarified by centrifugation. An in vitro synthesized 159-base transcript (pre-tRNATyr) ( Fig. 1) containing the E. coli suppressor I11 tRNATyr was used to screen this supernatant fraction for tRNA processing nucleases. As shown in Fig.  2, the supernatant fraction converted the pre-tRNATy' into a product of about 85 bases, consistent with the length of the mature suppressor tRNA5'. The supernatant fraction was then applied to a column of heparin-agarose, washed with 0.1 M KCl, and eluted stepwise with 0.25 and 0.5 M KCl. Assay of the 0.25 M KC1 fraction produced two products of 110 and 85 bases (Fig. 2). Subsequent analyses of the 110-and 85base products confirmed their identity as the 5'-processed intermediate and the mature tRNATyr, respectively (see the Miniprint). To further purify the processing nucleases, the 0.25 M KC1 fraction from heparin-agarose was applied directly to a phenyl-Sepharose column. The 5'-and 3'-tRNA processing activities co-eluted at the 50% (v/v) ethylene glycol step (Fig. 3). The final step in the purification made use of a DEAE-cellulose column to which the active fractions from the phenyl-Sepharose column (50% ethylene glycol step) were applied. The material that was not retained by the DEAEcellulose (flow-through fraction) contained an activity that converted the pre-tRNATY' primarily to the 110-base inter- FIG. 2. Assay of the mitochondrial tRNA processing nucleases in the initial supernatant fraction and after heparinagarose column fractionation using E. coli pre-tRNATSr as the substrate. The methods used in the heparin-agarose column fractionation and the details of the assay are presented under "Experimental Procedures."

c P R E -t R N A T y '
mediate ( Fig. 4A). Fingerprint analysis and 5' end analysis (see the Miniprint) showed that the 110-base product contained the mature 5' end of the tRNATyr and the unprocessed 3' trailer sequence. The DEAE-cellulose column fractions were reassayed using the 5"processed 110-base intermediate as the substrate (Fig. 4B). An activity that efficiently converted the 5'-processed precursor to the mature tRNATyr was 10274 Rat Mitochondrial 5'-and 3'-tRNA Processing Nucleases A . The methods used in the DEAE-cellulose column fractionation and the details of the assay are described under "Experimental Procedures." A , column fractions were assayed with intact pre-tRNATYr as substrate; B, column fractions were assayed with the 5"processed pre-tRNATYr produced by preincubation of the intact pre-tRNATY' with the separated 5' processing nuclease present in the flow-through fraction.

ELUTION BUFFER F R A C T I O N
identified in the fraction eluting at 0.5 M KC1 employed in the stepwise elution. Thus, the 5' and 3' processing nucleases were virtually completely separated in this final chromatography step. Furthermore, the apparent requirement of the 3' processing nuclease for a 5"processed intermediate as substrate strongly suggests that there exists an obligatory ordered sequence in the action of these enzymes initiated by the 5' processing nuclease followed by the 3' nuclease reaction.

DISCUSSION
The 5'-and 3'-tRNA processing nucleases have been isolated and separated from extracts of rat liver mitoplasts. The 5' processing enzyme was shown to be an endonuclease in that it cleaved the pre-tRNATyr at the 5' end of the mature tRNATy' sequence yielding the leader sequence and a 110base 5"processed intermediate. Endonucleolytic cleavage at the 5' end of pre-tRNA sequences appears to be a common feature in tRNA processing in both prokaryotic (Kole and Altman, 1982) as well as eukaryotic nuclear (Koski et al., 1976;Garber and Altman, 1979;Kline et al., 1981;Akaboshi et al., 1980) and mitochondrial (Hollingsworth and Martin, 1986;Doersen et al., 1985) systems. The mitochondrial 3' processing nuclease exhibited no detectable activity with the intact pre-tRNATy' transcript, but converted the 5'-processed intermediate to the mature tRNATyr. It appears that the 3' processing nuclease is restricted from cleaving at the 3' processing site of substrates containing a 5' leader sequence due to either steric hindrance by the leader sequence or due to a requirement for the presence of the mature 5' end that is likely to be adjacent to the 3' cleavage site in the folded tRNA structure. This apparent requisite order in the sequence of the processing reactions in vitro is consistent with the 5'-and 3'-tRNA processing enzymes from Xenopus laevis (Castaiio et dl., 1985), Bombyx mori (Garber and Altman, 1979), KB cells (Zasloff et al., 1982), and Drosophila (Frendewey et al., 1985).
The major product produced by the 3' processing nuclease was the mature tRNATy' possessing the E. coli gene-encoded terminal trinucleotide CCA. However, secondary products that were several nucleotides shorter on the 3' end were also identified. This apparent staggered cleavage by the 3' nuclease may have been exacerbated by the encoded CCA sequence present in the precursor transcript. None of the known mammalian mitochondrial tRNA genes encode for the terminal CCA of the acceptor stem (Clayton, 1984), thus requiring addition of these three nucleotides subsequent to 3' end processing of the encoded mitochondrial tRNA sequence. Thus, in vivo the probable role of the 3' processing nuclease would be to cleave the 5"processed intermediate precisely at the 3' end of the encoded tRNA sequence, i.e. one nucleotide beyond the mature 5' end on the duplex acceptor stem, in preparation for addition of the CCA extension by tRNA nucleotidyltransferase (Mukerji and Deutscher, 1972). It is also logical that the 3' processing nuclease must recognize terminally added CCAs as requisite extensions of mature tRNAs in vivo and thus leave them intact. We, therefore, suggest that the staggered cleavage observed with the 3' processing nuclease probably reflects complications introduced by the E. coli pre-tRNATy' substrate and that the products obtained represent (a) a dominant cleavage at the 3' end of the mature tRNATyr due to recognition of the terminal CCA, and ( b ) cleavage within the CCA sequence a t a distance one or two nucleotides from the processed 5' terminus closely juxtaposed on the duplex acceptor stem, in keeping with the predicted cleavage site in mitochrondrial systems.
An alternative explanation for the observed staggered cleavage by the 3' processing nuclease is that this enzyme might be an exonuclease and that the staggered ends may be the result of incomplete exonucleolytic processing.
We believe this is unlikely since products intermediate in size between the mature tRNATyr and the 110-base precursor were not observed during assay of the more highly purified preparations of the enzyme, even under conditions of high substrate-toenzyme ratios employed during measurement of activity units. However, confirmation of an endonucleolytic mode of action for this enzyme will require demonstration of removal of an intact trailer sequence.
For direct comparison, a cytosolic 5' processing nuclease was partially purified from the postmitochondrial supernatant of rat liver extracts. The total units of activity of this enzyme present in the cytosolic fraction were only about four times that of the comparable enzyme obtained from the mitochondrial extract. This enzyme also produced an endonucleolytic scission at the 5' end of the mature tRNA sequence of the pre-tRNATy' substrate. The optimal concentrations of Mg2+ and K' required by the cytosolic enzyme were higher than those required by either the mitochondrial 5' or 3' processing nucleases. In fact, the mitochondrial enzymes remained highly active in the absence of added Mg2+ to the reaction mixtures. Similar differences in the requirement for Mg2' were also observed between the cytosolic and mitochondrial RNase P enzymes isolated from HeLa cells (Doersen et al., 1985).
In studies of mitochondrial enzymes that are known to have cytosolic or nuclear counterparts, care must be taken to ensure that the mitochondrial fraction is free of the corresponding Rat Mitochondrial 5'-and 3'-tRNA Processing Nucleases 10275 extramitochondrial components. The importance of carefully separating the subcellular fractions was emphasized by the study of the RNase P-like enzymes from HeLa cells (Doersen et al., 1985) in which there was a tremendous excess of total RNase P activity in the cytosolic fraction, compared to that found in mitochondria. Although this concern appears to be of lesser magnitude with our rat liver system, in that the ratio of total 5' processing nuclease activity found in the cytosol uersus the mitochondria was only about 4:1, we have nonetheless taken the special precaution of removing the outer mitochondrial membrane and washing the resultant mitoplasts prior to preparation of the mitochondrial enzyme extracts. With these identical procedures, mitoplast extracts were found previously (Ledwith et al., 1986) to contain high levels of the mitochondrial DNA polymerase y and were shown to be virtually free of the nuclear DNA polymerases a and p. We believe that these results, in conjunction with the observed differences in optimal reaction conditions required by the 5' processing nucleases isolated from the cytosolic and mitoplast fractions in this report, provide strong evidence that the tRNA processing nucleases obtained from the mitoplast preparations are mitochondrial enzymes.
To facilitate the enzyme assays during our purification of the mitochondrial 5' and 3' processing nucleases, we chose to use the E. coli pre-tRNATYr as the substrate because of its high potential for forming normal tRNA secondary structure. In contrast, mammalian mitochondrial tRNA sequences exhibit many unusual features (Anderson et al., 1981(Anderson et al., ,1982Bibb et al., 1981;Gortz and Feldmann, 1982;Roe et al., 1982;Brown and Simpson, 1982;Sekiya et at., 1980;Wolstenholme et al., 1982;Pepe et al., 1982;Grosskopf andFeldmann, 1981a, 1981b;Kobayashi et al., 1980Kobayashi et al., ,1981Koike et al., 1982;Saccone et al., 1981). For example, they are less G+C-rich than their cytosolic or prokaryotic counterparts, and many lack the "constant" bases known to be involved in the stabilization of tertiary folding. It would seem, therefore, that mitochondrial tRNAs have considerably less potential for stabilizing higherorder structures. Furthermore, the length of the TW-loop varies among mitochondrial tRNAs; the universal sequence GTqCRA is variable; and in the case of tRNA Ser the dihydrouridine arm is completely missing. Thus, these potential substrates for the mitochondrial 5' and 3' processing nucleases are highly variable and present a challenging array of recognition features. It will now be of considerable interest to analyze the breadth of specificity of our purified processing nucleases with precursor transcripts containing various rat mitochondrial tRNA sequences.

AGY'
Rat L ' -f l a n k i n g sequence: t h e r e s p e c t i v e RNA sampler, each containing 70bg o f carnet' tRNA were The products were separated on PEI-cellulose plates ~n 0.7% potaSslum phosphate.   The products were electrOphoreSed on an a n a l y t i c a l 20% polyacrylamlde 7M urea gel and Visualized by autoradiography  a Assayed a f t e r DEAE-cellulose chromatography Assayed after heparin-agarose chroaatograDhy