Mitochondrial tRNA 5′-Editing in Dictyostelium discoideum and Polysphondylium pallidum*

Background: Secondary structure modeling predicts 5′-editing of mitochondrial tRNAs in amoebozoans Dictyostelium discoideum and Polysphondylium pallidum. Results: Experimental verification reveals unexpected instances of editing and species-specific variations. Conclusion: Mitochondrial tRNA 5′-editing is a complex process in which 3′–5′ RNA polymerases play a key role. Significance: Further biochemical characterization of mitochondrial tRNA 5′-editing reactions and enzymes is indicated. Mitochondrial tRNA (mt-tRNA) 5′-editing was first described more than 20 years ago; however, the first candidates for 5′-editing enzymes were only recently identified in a eukaryotic microbe (protist), the slime mold Dictyostelium discoideum. In this organism, eight of 18 mt-tRNAs are predicted to be edited based on the presence of genomically encoded mismatched nucleotides in their aminoacyl-acceptor stem sequences. Here, we demonstrate that mt-tRNA 5′-editing occurs at all predicted sites in D. discoideum as evidenced by changes in the sequences of isolated mt-tRNAs compared with the expected sequences encoded by the mitochondrial genome. We also identify two previously unpredicted editing events in which G-U base pairs are edited in the absence of any other genomically encoded mismatches. A comparison of 5′-editing in D. discoideum with 5′-editing in another slime mold, Polysphondylium pallidum, suggests organism-specific idiosyncrasies in the treatment of U-G/G-U pairs. In vitro activities of putative D. discoideum editing enzymes are consistent with the observed editing reactions and suggest an overall lack of tRNA substrate specificity exhibited by the repair component of the editing enzyme. Although the presence of terminal mismatches in mt-tRNA sequences is highly predictive of the occurrence of mt-tRNA 5′-editing, the variability in treatment of U-G/G-U base pairs observed here indicates that direct experimental evidence of 5′-editing must be obtained to understand the complete spectrum of mt-tRNA editing events in any species.

Transfer RNA species undergo a variety of post-transcriptional processing events to produce the mature tRNA molecules that function in translation (1). These processing steps may include tRNA editing reactions, which result in changes in the tRNA sequence from the sequence encoded by the tRNA gene. Multiple types of editing reactions have been observed to occur at locations throughout a tRNA, including at the 5Ј and 3Ј ends and within the tRNA body (2)(3)(4)(5)(6)(7)(8)(9)(10)(11). These reactions include base deamination wherein C or A nucleotides are edited to become U or I, respectively, and insertion/deletion editing wherein unencoded nucleotides are added to specific locations or encoded nucleotides are removed and replaced. 5Ј-tRNA editing, first described in the mitochondria of Acanthamoeba castellanii, is an example of the latter type (12). In 5Ј-editing, pre-tRNA transcripts contain one to three mismatched nucleotides at the 5Ј-end of the aminoacyl-acceptor stem, and these mismatches are predicted to result in suboptimal tRNA function (see Fig. 1). The mismatched nucleotides are absent in the sequenced mature mt-tRNA, 2 which is suggestive of post-transcriptional editing (12)(13)(14). The editing reaction minimally comprises a two-step process in which first the mismatched nucleotides are removed from the pre-tRNA transcript and second the correct Watson-Crick base-pairing nucleotides are added back to the 5Ј-truncated tRNA (Fig. 1). This 5Ј-end repair step apparently requires a 3Ј-5Ј RNA polymerase capable of using the 3Ј-aminoacyl-acceptor stem nucleotides as a template for 5Ј-nucleotide addition (15)(16)(17). The requirement for this apparently non-standard enzyme chemistry in the form of a polymerase that acts in the 3Ј-5Ј direction, opposite to all known DNA/RNA polymerases, impeded identification of the 5Ј-editing enzyme(s) for nearly two decades.
Recently, however, the first two candidate enzymes for participation in mitochondrial 5Ј-editing, members of the tRNA His guanylyltransferase (Thg1) superfamily, were identified in the eukaryotic slime mold Dictyostelium discoideum (18,19). Thg1-like proteins (TLPs) from Bacteria and Archaea exhibit a proficient ability to use the template-dependent 3Ј-5Ј polymerase function to repair 5Ј-truncated tRNA species; the latter species mirror the truncated tRNA repair intermediates inferred to occur during mt-tRNA 5Ј-editing (20,21). Using in vitro enzyme assays, two different D. discoideum TLPs * This work was supported, in whole or in part, by National Institutes of Health (DdiTLP3 and DdiTLP4) showed equivalent abilities to repair each of two different tRNAs that were predicted to undergo tRNA 5Ј-editing in D. discoideum mitochondria (15). Whether these two enzymes participate collaboratively or separately in tRNA 5Ј-editing in vivo remains unknown.
Experimental evidence confirming the existence of tRNA 5Ј-editing (by comparison of actual mt-tRNA sequences with the encoded mt-tDNA genes and identification of changes in the 5Ј-nucleotides) has so far only been obtained for a small number of eukaryotes, including A. castellanii, Spizellomyces punctatus, Physarum polycephalum, Monoblepharella sp., and Harpochytrium sp., but not in D. discoideum (6, 12-14, 22, 23). In D. discoideum, the existence of 5Ј-editing has been inferred based on the presence of mismatched nucleotides in the appropriate positions of the aminoacyl-acceptor stem in the encoded mt-tDNA sequences, but 5Ј-editing per se has not yet been demonstrated experimentally in this organism. Thus, to fully elucidate the composition and function of the enzymes that carry out the 5Ј-editing reaction, it is necessary to know the editing status of the mitochondrial tRNA species in this amoebozoan.
In this work, we carried out a comprehensive characterization of the mitochondrial 5Ј-editing system in D. discoideum by sequencing all eight mt-tRNAs initially predicted to undergo editing in vivo. Additionally, we sequenced several mt-tRNAs that are not predicted to be edited in D. discoideum, revealing the presence of two previously unexpected editing events whereby G-U base pairs are efficiently edited to Watson-Crick base-paired nucleotides. This first example of editing of G-U base pairs in the absence of bona fide non-Watson-Crick mismatches in the aminoacyl-acceptor stem prompted us to undertake a closer investigation of editing activity with respect to G-U base pairs. For this, we compared 5Ј-editing in D. discoideum with editing in another slime mold, Polysphondylium pallidum, where this process has also been predicted but not experimentally verified. In doing so, we identified different patterns of G-U base pair editing that suggest mechanistic differences between 5Ј-editing reactions in divergent organisms. In vitro characterization of repair activities catalyzed by DdiTLP3 and DdiTLP4 confirms that both enzymes are able to repair mt-tRNAs that have been shown by direct sequencing to undergo 5Ј-editing (both the previously proposed as well as newly revealed edited tRNAs) as well as a control tRNA that is not known to be edited in vivo. The general lack of tRNA substrate specificity exhibited by either enzyme during repair of 5Ј-truncated tRNAs suggests that tRNA substrate selection during 5Ј-editing is not likely to be exerted at the 3Ј-5Ј polymerase step of the reaction.
The circularized RNA was resuspended in 10 l of double distilled H 2 O, and 2 l was used as template for reverse transcription by Superscript III reverse transcriptase (Invitrogen). RNA (ϳ2 g) and an appropriate cDNA-RT oligonucleotide (Table 1) for each tRNA (0.2 M) were annealed in a 10-l reaction by heating to 95°C for 5 min followed by slow cooling to room temperature for 30 min; extension reactions were assembled using the entire volume from the annealing reaction and contained 1ϫ first strand buffer (Invitrogen), 5 mM DTT, a 0.5 mM concentration of each dNTP, and 200 units of Superscript III reverse transcriptase in a final volume of 20 l. Reactions were incubated at 50°C for 1 h, and then enzyme was heat-inactivated by incubation at 75°C for 10 min. A portion of FIGURE 1. Schematic of mt-tRNA editing in D. discoideum. 5Ј-Editing of mt-tRNA occurs in at least two steps in which 5Ј-mismatched nucleotides are removed from the 5Ј-end of the mt-tRNA, producing a presumed 5Ј-truncated mt-tRNA intermediate. Biochemical data suggest that the 5Ј-truncated mt-tRNA is then acted on by one of the TLP enzymes encoded by the D. discoideum nuclear genome, thus repairing the 5Ј-truncated end via 3Ј-5Ј polymerase activity using the 3Ј-side of the aminoacyl-acceptor stem as the template for polymerase activity. The presence of an intact 3Ј-CCA end on the tRNA prior to editing is shown for clarity (and is supported by the sequences obtained in this work), but the precise order of 3Ј-processing and 5Ј-editing events has not been determined.
each reaction mixture (3 l) was used as the template for PCR using reverse transcriptase and 3Ј-oligonucleotides (Table 1) to amplify the ligated 5Ј-and 3Ј-end junction with iProof polymerase (Bio-Rad) in 50-l reactions. PCRs were carried out with 2-min incubation at 98°C followed by five cycles as follows: 30 s at 98°C, 30 s at 55°C, and 20 s at 72°C and then 25 cycles of 10 s at 98°C, 30 s at 65°C, and 20 s at 72°C. Cycling was followed by 10 min at 72°C. RT-PCR products of the correct size were extracted from 10% acrylamide, 4 M urea gels by soaking the crushed gel slice in elution buffer containing 0.5 M ammonium acetate, 0.1% SDS, and 5 mM EDTA followed by phenol:chloroform extraction and ethanol precipitation of the extracted DNA.
The entire volume of resulting purified DNA was 5Ј-phosphorylated in reactions with T4 polynucleotide kinase (New England Biolabs) in 1ϫ T4 polynucleotide kinase buffer and 1 mM ATP, incubated at 37°C for 30 min, and heat-inactivated at 75°C for 10 min. 5Ј-Phosphorylated PCR products were purified by phenol:chloroform extraction and ethanol precipitation and ligated into a pUC19 vector that was blunt end-linearized by treatment with SmaI (New England Biolabs) and dephosphorylated with calf intestinal phosphatase (Invitrogen). After transformation, individual clones were used to isolate plasmid DNA, which was then sequenced. The majority of obtained sequences contained mature 3Ј-CCA end sequences; only sequences of these 3Ј-end matured mt-tRNAs are reported here.
Sequencing of P. pallidum mt-tRNAs-P. pallidum strain PPHU8 (ATCC 44421) was cultured in a modified ATCC 712 PYG formulation, and a fraction enriched in mitochondria was isolated by cell disruption followed by differential centrifugation. Mitochondria were sedimented from the postnuclear supernatant by a 20-min centrifugation at 13,000 ϫ g. RNA was isolated from mitochondrial pellets by a phenol:cresol extraction protocol (14), precipitated from the aqueous phase with ethanol, and treated with RNase-free DNase I before being reisolated and stored at Ϫ20°C.
Circularization of tRNAs followed published protocols (14, 24): 1 l of the resulting circularized tRNA was diluted with 1 l of 5Ј RT primer (1 pmol/l; Table 1) in a 21.5-l final volume and incubated at 90°C for 2 min. The tRNA-primer mixture was allowed to cool to room temperature over a 20-min period and transferred to ice for 15 min. A final reaction volume of 30 l contained 21.5 l of cooled tRNA-primer mixture, 0.1 volume of avian myeloblastosis virus buffer (Amersham Biosciences), 1.8 l of 1 mM dNTPs (60 M final concentration of each dNTP), and 10 units of avian myeloblastosis virus reverse transcriptase (Amersham Biosciences) and was incubated at 45°C for 45 min. For PCR of RT products, 20 pmol of the applicable 5Ј and 3Ј primers ( Table 1) and 1 l of cDNA were used for each reaction. Reaction mixtures (50 l) consisted of Thermo Buffer (New England Biolabs; 5 l), 0.2 mM dNTPs (250 M final concentration of each dNTP), and 2.5 units of Taq DNA polymerase (5 units/l; New England Biolabs). PCRs (carried out in a PerkinElmer Life Sciences GeneAmp PCR System 2400 thermocycler) were initiated with a 3-min incubation at 94°C followed by 35 cycles as follows: 30 s at 94°C, 30 s at 50 or 55°C (depending on primer melting temperature), and 40 s at 72°C. Cycling was followed by 12 min at 72°C. PCR products were electrophoresed in 2% agarose gels. For TOPO TA cloning with pCR 2.1 (Invitrogen), PCR products were ligated and subsequently cloned according to the manufacturer's protocol.
Tobacco Acid Pyrophosphatase (TAP) Treatment prior to Circularization-Where indicated, total RNA was treated with TAP to convert 5Ј-triphosphorylated ends to 5Ј-monophosphorylated ends that are appropriate for ligation. Reactions (100 l) contained 10 g of total RNA (isolated as above) and 4 units of TAP (Epicenter) in 1ϫ TAP buffer as provided by the manufacturer and were incubated at 37°C for 1 h. Reaction products were purified by phenol:chloroform extraction and ethanol precipitation, and the RNA was resuspended in 10 l of double distilled H 2 O for use in the circularization reactions as described above.
Cloning and Transcription of 5Ј-Truncated D. discoideum mt-tRNAs for in Vitro Assays-In vitro transcripts for mt-tRNA substrates that do not initiate with a G nucleotide (Glu UUC and Tyr GUA ) were produced as 5Ј-hammerhead ribozyme constructs; a unique targeting sequence was engineered into the  (1) and - (2) CAAACGAATACTCTAACCAAC MAY 30, 2014 • VOLUME 289 • NUMBER 22

Analysis of tRNA 5-Editing in Mitochondria
ribozyme so that the indicated 5Ј-end sequence of each mt-tRNA is obtained after co-transcriptional cleavage (15,25). The Cys GCA and Trp CCA transcripts both initiate with G ϩ2 and, therefore, were constructed for direct transcription without a 5Ј-hammerhead sequence. Sequence-verified clones were used as templates for in vitro transcription by T7 RNA polymerase (15) and then 5Ј-phosphorylated with T4 polynucleotide kinase and [␥-32 P]ATP. Phosphatase Protection Assays for 5Ј-End Repair-5Ј-32 P-Labeled mt-tRNAs Glu UUC , Tyr GUA , Trp CCA , and Cys GCA were used as substrates for phosphatase protection assays to test for 5Ј-nucleotide addition. Reactions were performed as described (15); NTPs or dNTPs were included at 0.1-1 mM concentration in the assays as indicated. Reactions were initiated by addition of enzyme (at the concentrations indicated in each figure) to labeled tRNA and NTP and were carried out in 25 mM HEPES, pH 7.5, 10 mM MgCl 2 , 3 mM DTT, 125 mM NaCl, 0.2 mg/ml BSA buffer. Digestion with the indicated nuclease (RNase A or RNase T1) followed by calf intestinal phosphatase treatment was carried out as described (20), and reaction products were visualized by TLC and imaged using a Typhoon imager.

Eight tRNA Substrates Containing Mismatched Base Pairs in
the Aminoacyl-acceptor Stem are 5Ј-Edited in D. discoideum-Eighteen mt-tRNA species are encoded in the mitochondrial genome of D. discoideum, and gene sequences for eight of these tRNAs clearly indicate the presence of up to three mismatched nucleotides in the predicted aminoacyl-acceptor stems (Fig. 2). To determine whether these tRNAs undergo 5Ј-editing to repair the mismatches, we isolated tRNA from axenically grown D. discoideum and sequenced the 5Ј-end of each of the eight predicted 5Ј-editing substrates. To obtain tRNA sequences, we used a previously described method in which the bulk tRNA preparations are treated with T4 RNA ligase to join the tRNA 5Ј-and 3Ј-ends, and then the sequence of the ligated region is amplified by RT-PCR using specific primers that target each tRNA (22,26). The resulting PCR products were cloned into a SmaI-digested plasmid for sequencing of individual clones.
In the first set of sequenced clones (Table 2), we clearly observed evidence for 5Ј-editing of six of the eight mt-tRNA species that encode mismatched nucleotides in the acceptor stem (Leu UAG , Leu UAA , Gln UUG , Ile GAU(1) , Ile GAU (2) , and Tyr GUA ). In each case, sequences of the isolated mt-tRNAs contained Watson-Crick base-paired nucleotides in place of the originally encoded mismatched nucleotides, suggesting a mechanism of 5Ј-editing in D. discoideum similar to that observed in other protozoa (Fig. 1). In one case (tRNA Leu UAA ) two of four clones appeared to reflect partial editing in that the 3Ј-5Ј repair component of the editing enzyme had apparently added the correct C ϩ3 and U ϩ2 nucleotides but not the final C ϩ1 nucleotide to complete the aminoacyl-acceptor stem ( Table 2). Because the number of clones obtained for each mt-tRNA varied, likely due to differences in efficiency of reverse transcription of each mt-tRNA, it is possible that similar partially edited intermediates exist for other tRNAs but were not detected. Nonetheless, sequences that demonstrate the existence of 5Ј-editing in D. discoideum were readily obtained for each of these six tRNAs, and the observation of these fully edited sequences suggests that the 5Ј-edited tRNAs likely comprise a significant fraction of the mt-tRNA pool in vivo.
For the remaining two mt-tRNAs in which 5Ј-editing was expected to occur (Ile CAU and Glu UUC ), the initially sequenced clones did not reveal sequences of the fully edited mt-tRNAs.
For Ile CAU , we obtained two clones, both of which were unedited (i.e. retained the ϩ1-72 mismatch), and for Glu UUC , we obtained three clones, one of which retained the ϩ1-72 mismatch, whereas the other two showed partial editing in which the N ϩ1 nucleotide had apparently been removed but Identification of mt-tRNA 5-editing sites in D. discoideum a Fourteen of 18 annotated mt-tRNAs were sequenced for this work; tRNAs Lys UUU , Asn GUU , Arg UCU , and His GUG do not contain annotated mismatches and were not sequenced. b Aminoacyl-acceptor stem sequences of annotated mt-tRNA genes (shown in italics and labeled as tDNA with 5Ј-and 3Ј-ends as indicated) are compared with sequences of clones obtained through this work. Predicted mismatched nucleotides are shown in lowercase bold letters; U-G/G-U base pairs in the annotated or cloned sequences are indicated by underlined nucleotides. Replacement nucleotides are indicated by red letters. c Editing status is indicated as follows: E, edited, observed changes were all to Watson-Crick base-paired nucleotides; PE, partially edited, evidence obtained for removal of mismatches and/or partial addition of Watson-Crick base-pairing nucleotides; NE, not edited, no evidence for editing was observed in the obtained sequences. All sequences were derived from fully mature 3Ј-CCA-containing mt-tRNA as indicated.

Analysis of tRNA 5-Editing in Mitochondria
the correct Watson-Crick-pairing G ϩ1 had not yet been added (Table 2).
To further probe the editing of these mt-tRNAs, we considered a possibility based on the biochemical reaction catalyzed by the presumptive 3Ј-5Ј polymerase component of the 5Ј-editing enzyme. The 3Ј-5Ј polymerases are members of the Thg1 superfamily of enzymes, which use the 3Ј-hydroxyl of an incoming NTP to attack an activated 5Ј-phosphate on the tRNA substrate (27,28). Thus, the immediate product of the 3Ј-5Ј addition reaction would be a tRNA with a nucleoside triphosphate at its 5Ј-end (Fig. 1). In Saccharomyces cerevisiae, the Thg1 enzyme removes the 5Ј-pyrophosphate from the added nucleotide to generate a 5Ј-monophosphate end, but the ability of TLPs to similarly catalyze the removal of a terminal 5Ј-triphosphate has not been investigated. We reasoned that if TLPs catalyze 5Ј-repair but do not efficiently remove the 5Ј-pyrophosphate from some of the edited mt-tRNAs then the tRNAs would not be circularized efficiently by T4 RNA ligase (because ligase requires a 5Ј-monophosphate) and thus might not be amplified from the tRNA pool. To test this possibility, we treated the in vivo isolated tRNA with TAP to remove any 5Ј-triphosphorylated ends prior to ligation and then proceeded with RT-PCR amplification for the two missing tRNAs.
Consistent with our hypothesis, after TAP treatment, we readily obtained fully edited clones of both tRNAs (Ile CAU and Glu UUC ) for which we were previously unable to observe editing ( Table 2). The complete shift in the population after TAP treatment (from wholly unedited to substantially edited) suggests that the presence of a 5Ј-triphosphate was a significant barrier to circularization of at least these two tRNAs. The isolation of the other six circularized tRNAs in edited form even without TAP treatment may mean that the removal of the 5Ј-triphosphate after repair is more efficient for some substrates than for others or may be due to differences in stability of the 5Ј-triphosphate in certain mt-tRNAs.
Species-specific Differences in Editing of U-G/G-U Base Pairs during 5Ј-Editing-For one of the mt-tRNAs sequenced above (Ile GAU(1) ), we observed that the editing activity went beyond the correction of the two mismatched base pairs at A ϩ1 xG 72 and A ϩ2 xA 71 to also "repair" the U ϩ3 -G 70 base pair in the aminoacyl-acceptor stem ( Table 2 and Fig. 2). The ability of the 5Ј-editing activity to act on U-G/G-U base pairs despite their prevalence in normal tRNA structures has been observed previously as in A. castellanii where two examples of U-G editing were demonstrated, both of which apparently occur during the course of the editing reaction to repair other mismatches in the tRNAs (14).
To further assess the generality of editing of U-G/G-U base pairs, sequences of mt-tRNAs were determined from another eukaryotic slime mold, P. pallidum. In this organism, 5Ј-editing is also predicted on the basis of mismatches observed in five of 11 mtDNA-encoded tRNAs, and two of these five predicted substrates contain U-G/G-U base pairs in varying contexts within the first three base pairs of the aminoacyl-acceptor stem (Fig. 3). In P. pallidum, as in D. discoideum, isolated mt-tRNA sequences showed evidence of widespread 5Ј-editing ( Table 3). The three mt-tRNAs that contain mismatches but lack U-G/ G-U base pairs (Ile GAU , Leu UAA , and Asn GUU ) were observed to be mostly edited as expected (Table 3). However, in contrast to the observation of substantial editing of U-G/G-U pairs at multiple positions in A. castellanii and D. discoideum, editing of these base pairs in P. pallidum appears to be much rarer. For the FIGURE 3. Summary of verified mt-tRNA editing events in P. pallidum. Aminoacyl-acceptor stem sequences of 11 mtDNA-encoded tRNA species in P. pallidum are shown. Clones containing the corresponding mt-tRNA genes were isolated from P. pallidum strain PPHU8 mtDNA and sequenced (36). Sequence comparison indicates that the mitochondrial genome of strain PPHU8 is essentially identical to that of P. pallidum strain PN500 (GenBank accession number EU275726.1). The top row includes mt-tRNAs predicted to be substrates for 5Ј-editing (based on the presence of identifiable mismatches) prior to this work; the bottom row includes mt-tRNAs that do not contain mismatched nucleotides in the first three positions of their aminoacyl-acceptor stem. Mismatched nucleotides in the originally encoded mt-tRNA genes are indicated in red; alterations that result in Watson-Crick base-paired nucleotides verified here are indicated by red nucleotides shown to the left of each arrow. U-G/G-U base pairs in any of the first three positions of the aminoacyl-acceptor stem are indicated in blue; changes at any of these nucleotides, if observed here, are also indicated by red nucleotides shown to the left of each arrow. Partial editing is indicated by small arrows with the replacement nucleotide that is observed in some but not all clones indicated in parentheses next to the arrowhead.
two tRNAs that contain mismatches as well as U-G/G-U pairs (Ala UGC and Phe GAA ), little or no editing of the U-G/G-U base pairs was observed (Table 3). Thus, the ability to repair U-G/ G-U base pairs in the course of 5Ј-editing seems to follow tRNA-or species-specific rules, perhaps depending on organism-specific features of the editing machinery.
Identification of Two Additional mt-tRNA Editing Substrates in D. discoideum-We chose several of the D. discoideum mt-tRNAs that did not contain mismatches in the encoded sequences, including tRNAs Pro UGG , Trp CCA , and Cys GCA , to serve as controls for the above sequencing reactions because we expected to obtain the gene-encoded mt-tRNA sequences for these species (Fig. 2). However, Pro UGG and Trp CCA , which contain U-G and G-U base pairs, respectively, at the ϩ1-72 position but no other non-Watson-Crick base pairs, unexpectedly also showed evidence of 5Ј-editing (Table 2). In both cases, the encoded nucleotide at the ϩ1 position was not present; instead the sequenced mt-tRNA contained the correct Watson-Crick base-pairing ϩ1 nucleotide as indicated in Fig. 2. This is the first reported example of 5Ј-editing occurring in the absence of a bona fide mismatched base pair and implies that the nuclease component of the 5Ј-editing enzyme (as yet unidentified) must rely on additional features beyond simply the presence of a non-Watson-Crick mismatch.
In light of this observation, sequences of several other mt-tRNAs that similarly contain U-G/G-U base pairs but no identifiable mismatches were examined. Sequences were obtained in the same way for P. pallidum Glu UUC , which contains a G ϩ2 -U 71 base pair, and D. discoideum Ala UGC , Phe GAA , and Met CAU , all with G-U/U-G base pairs at the ϩ3-70 position. None of the G-U base pairs were observed to be edited in any of the sequences obtained from either organism (Tables 2 and 3). Thus, the total number of mt-tRNA editing substrates now verified in D. discoideum stands at 10 instead of the eight species that were initially predicted to undergo 5Ј-editing based on the presence of mismatches in their encoded sequences (Fig. 2).
The sequences obtained for tRNA Cys GCA suggest some additional complexity in the editing reaction. This tRNA contains no mismatches or U-G/G-U base pairs, so we again expected to isolate sequence identical to that of the mtDNA-encoded tRNA. However, when the standard ligation and amplification procedures were followed, we quantitatively obtained mt-tRNA clones (five of five) that lacked the C ϩ1 nucleotide but instead began with a 5Ј-terminal G ϩ2 (Table 2). We then treated the same tRNA with TAP prior to the ligation reactions as described above for Ile CAU and Glu UUC . Interestingly, we readily obtained clones from the TAP-treated RNA, and 15 of 15 of these were full-length, including C ϩ1 . This intriguing observation suggests that at least some of the full-length tRNA Cys may also contain the 5Ј-triphosphorylated end that is inhibitory to circularization. Further characterization of the processing reactions that generate mt-tRNA Cys is required to understand whether the 5Ј-editing machinery plays a role in metabolism of this, or any other, non-mismatch-containing substrates in vivo.
In Vitro Activities of DdiTLP3 and DdiTLP4 Correlate with the Observed in Vivo Editing Pattern-Previous in vitro assays demonstrated that both DdiTLP3 and DdiTLP4 catalyze effi- Identification of mt-tRNA 5-editing sites in P. pallidum a Six of 11 annotated mt-tRNAs were sequenced for this work; tRNAs Arg UCU , Gln UUG , Ile CAU , Met CAU , and His GUG do not contain annotated mismatches and were not sequenced. b Aminoacyl-acceptor stem sequences of annotated mt-tRNA genes (shown in italics and labeled as tDNA with 5Ј-and 3Ј-ends as indicated) are compared with sequences of clones obtained through this work. Predicted mismatched nucleotides are shown in lowercase bold letters; U-G/G-U base pairs in the annotated or cloned sequences are indicated by underlined nucleotides. Edited nucleotides are indicated by red letters. c Editing status is indicated as follows: E, edited, observed changes were all to Watson-Crick base-paired nucleotides; PE, partially edited, evidence obtained for removal of mismatches and/or partial addition of Watson-Crick base-pairing nucleotides; NE, not edited, no evidence for editing was observed in the obtained sequences. All sequences were derived from fully mature 3Ј-CCA-containing mt-tRNA as indicated.
cient repair of two D. discoideum mt-tRNAs (Ile CAU and Leu UAG ) (15), but from this limited number of substrates, it is difficult to discern whether the two candidate enzymes exhibit non-overlapping substrate specificities that could perhaps explain the presence of two putative editing enzymes in D. discoideum. Thus, the in vitro activities of purified DdiTLP3 and DdiTLP4 were tested with four additional substrates, including two mt-tRNAs previously predicted to undergo editing (Glu UUC and Tyr GUA ), one of the newly identified mt-tRNAs (Trp CCA ), and a control mt-tRNA (Cys GCA ) that is not known to undergo editing. Repair activities were tested using substrates that mimic 5Ј-truncated intermediates predicted to occur after the nuclease step of the 5Ј-editing reaction (i.e. for Glu UUC , the mismatched U ϩ1 is removed, and the substrate transcript initiates with C ϩ2 ) (Fig. 4A). The tested substrates all contained an intact 3Ј-CCA end because the observation of partially edited and unedited intermediates among in vivo tRNA sequences, which all contain mature 3Ј-CCA ends, indicates that there is no obligatory order between 5Ј-editing and 3Ј-CCA addition reactions. A phosphatase protection assay used extensively to characterize 3Ј-5Ј addition with other tRNAs was performed with each 5Ј-end-labeled substrate incubated in the presence of purified enzyme, ATP (required to activate the 5Ј-monophosphate for addition), and other NTP(s) as necessary to complete the aminoacyl-acceptor stem of each mt-tRNA. In these assays, nucleotide addition products are observed as phosphatase-resistant labeled oligonucleotides that can be separated by TLC from the inorganic phosphate generated by phosphatase treatment of the unreacted substrate. The ability of each of the four Thg1-related enzymes encoded in D. discoideum (DdiThg1, DdiTLP2, DdiTLP3, and DdiTLP4) and S. cerevisiae Thg1 (SceThg1) to add the indicated N ϩ1 nucleotide and thus repair the 5Ј-truncated mt-tRNAs was determined (Fig. 4). As observed previously, only DdiTLP3 and DdiTLP4 catalyze efficient repair of any of these substrates, consistent with their ability to catalyze the observed editing reactions documented above by direct mt-tRNA sequencing. Moreover, the similar dependence of the observed activities on the concentration of DdiTLP3 and DdiTLP4, now demonstrated for five of the 10 known edited mt-tRNAs in D. discoideum, reveals no substantial differences between the catalytic activities of either enzyme that might indicate tRNA-specific roles for DdiTLP3 or DditTLP4 in editing. Finally, both enzymes catalyzed proficient repair of the 5Ј-truncated tRNA Cys GCA , which is not predicted to undergo editing in vivo, suggesting that the tRNA specificity of the editing reactions is not likely to be dictated by the polymerase component of the editing holoenzyme but perhaps is associated with the unidentified nuclease.

DISCUSSION
Here, we demonstrate that extensive 5Ј-editing of mitochondrial tRNA occurs in D. discoideum and P. pallidum (Tables 2  and 3 and Figs. 2 and 3). In D. discoideum, mt-tRNA sequences reveal that 10 of the 18 mtDNA-encoded tRNAs undergo 5Ј-editing, including one instance of editing of a G-U base pair and two cases of U-G editing ( Fig. 2 and Table 2). However, in the slime mold P. pallidum, although 5Ј-editing of mismatched nucleotides is also observed (for five of 11 annotated mt-tRNAs), U-G/G-U base pairs are observed to be edited much more rarely if at all (Fig. 3 and Table 3). Therefore, there are significant differences in the treatment of U-G/G-U base pairs by the 5Ј-editing machinery in different species or in the context of different mt-tRNAs, differences that remain to be investigated. We also demonstrate that in vitro 5Ј-end repair reactions catalyzed by the candidate D. discoideum editing enzymes (DdiTLP3 and DdiTLP4) are consistent with the observed editing events, including the newly identified repair of single U-G/G-U base pairs (Fig. 4).
To date, 5Ј-editing of mismatches encoded in mt-tRNA has been experimentally verified in seven eukaryotic species, including multiple members of Amoebozoa (D. discoideum, P. pallidum, P. polycephalum, and A. castellanii) and Fungi (S. punctatus, Monoblepharella sp., and Harpochytrium sp.) (6, 12-14, 22, 23). In each case, the presence of mismatched nucleotides encoded in any of the first three positions of the aminoacyl-acceptor stem is highly predictive of the occurrence of 5Ј-editing to repair these mt-tRNAs, and the majority of mature mt-tRNA sequences reveal that removal and repair of the 5Ј-mismatched nucleotides has indeed occurred. However, the results presented here demonstrate that the outcome of 5Ј-editing in any organism cannot be completely predicted on the basis of genome sequencing alone. Among these seven species are examples in which U-G/G-U base pairs are either unedited, partially edited, or fully edited in various contexts, and this variable editing (Fig. 5) does not appear to be a predictable feature for individual mt-tRNAs. Repair of U-G/G-U base pairs in at least two D. discoideum mt-tRNAs in the absence of any other non-Watson-Crick mismatched nucleotides, as documented here, suggests the possibility of an even larger potential pool of mt-tRNA editing substrates beyond just mismatch-containing tRNA. Thus, experimental data, such as direct sequencing of mt-tRNA, are required to rigorously assess the consequences of 5Ј-editing in any organism.
The rules that govern how U-G/G-U base pairs are treated in the context of the 5Ј-editing reaction are apparently not universal. In D. discoideum, we observed robust editing of the U ϩ3 -G 70 base pair of tRNA Ile GAU(1) to a C ϩ3 -G 70 pair, which occurred concurrently with repair of the mismatched base pairs located at the first two positions of this mt-tRNA. Of the six other species in which 5Ј-editing has been demonstrated, five of these (A. castellanii, P. pallidum, S. punctatus, Monoblepharella sp., and Harpochytrium sp.) contain at least one tRNA where a G-U pair would be similarly encountered during repair of other mismatched nucleotides; however, there are substantial differences between the observed editing patterns of these tRNAs (Fig. 5). For example, Monoblepharella tRNA Met , like D. discoideum tRNA Ile GAU (1) , contains mismatches at ϩ1-72 and ϩ2-71 along with a U ϩ3 -G 70 pair; however, only ϳ40% of the isolated Monoblepharella mt-tRNA contained the fully repaired sequences (including C ϩ3 -G 70 ), whereas the remainder retained the U-G base pair (23). In addition, examples of mt-tRNA in which G-U base pairs are completely unedited are observed in P. pallidum and Harpochytrium despite the presence of other mismatched nucleotides. The function of 5Ј-editing of G-U/U-G base pairs is not readily apparent because it is known that these types of base pairs are well tolerated in normal tRNAs. For G-U/U-G base pairs that are edited during the If evidence for 5Ј-editing has been observed in any of the sequences obtained, this is indicated by arrows with the replacement nucleotides shown in red to the left of each arrow. Partial editing is indicated by small arrows with the corrected nucleotide that is observed in some but not all clones indicated in parentheses next to the arrowhead. The observed extent of 5Ј-editing of the indicated U-G/G-U base pairs (based on the number of edited sequences of the total number of sequences obtained for each mt-tRNA) is shown as a percentage below each mt-tRNA. The sequence of A. castellanii Met CAU was obtained from a PCR product instead of individual clones and thus is presumed to reflect 100% editing (14).
course of editing other bona fide mismatches, it is possible that there is no inherent "benefit" to the tRNA to this editing process. However, for instances of G-U/U-G editing that occur in the absence of other mismatches (see below), there may be some important effect on tRNA structure or function that is not yet apparent due to the limited number of available sequences. This issue remains to be further investigated.
Even when the position of the G-U/U-G base pair is considered (i.e. whether it is located immediately next to the mismatch versus removed by one canonical base pair), a generalizable "rule" for editing of G-U base pairs is not evident. Although the G ϩ3 -U 70 in P. pallidum tRNA Ala is separated by one standard Watson-Crick base pair from the mismatch at ϩ1-72 and is not edited, the U ϩ3 -G 70 pair in A. castellanii tRNA Leu UAA is located in a similar sequence context and is fully edited (Fig. 5). Furthermore, some G-U or U-G base pairs are located directly adjacent to a mismatched base pair and are efficiently edited to Watson-Crick pairs (i.e. those from D. discoideum Ile GAU(1) , A. castellanii Met CAU , and Monoblepharella Pro UGG or Gly UCC ), whereas other base pairs in a similar context are either partially edited (P. pallidum Phe GAA and Monoblepharella Met CAU and Glu UUC ) or wholly unedited (Harpochytrium Met CAU ). The different outcomes in terms of editing may reflect differences in catalytic activities of TLPs or the editing nucleases from different organisms. Thus, biochemical characterization of different editing enzymes from multiple species that carry out 5Ј-editing will be required to directly address this question.
This work revealed editing of the ϩ1-72 G-U/U-G base pairs of D. discoideum tRNA Pro and tRNA Trp , which is to our knowledge the only demonstrated example of mt-tRNA 5Ј-editing that occurs in the absence of any other obvious mismatched nucleotides. This observation is consistent with previous biochemical data obtained with the partially purified mitochondrial 5Ј-editing enzyme in A. castellani in which the active fraction exhibited nucleotide incorporation activity with commercially obtained Escherichia coli tRNA Tyr (17). Because the E. coli tRNA lacks any mismatched nucleotides in the aminoacyl-acceptor stem, one possible explanation for this result would be a constitutive removal/repair activity that is not strictly dependent on the presence of mismatched nucleotides in the stem but instead acts more broadly on tRNA in vivo. The existence of editing activities that do not require mismatched nucleotides would not be apparent from simple sequence comparisons because such editing would involve no changes in the identity of the encoded bases in the aminoacyl-acceptor stem.
Relevant to this possibility is the observation reported here that for D. discoideum tRNA Cys full-length mt-tRNA sequences could not be obtained without first treating the tRNA with TAP (presumably to remove additional ligation-inhibiting 5Ј-phosphates) ( Table 2). A possible explanation for this striking difference is that the mature tRNA Cys GCA present in D. discoideum contains a 5Ј-triphosphorylated C ϩ1 nucleotide that is inhibitory to ligation, although the presence of other unknown 5Ј-phosphate modifications that inhibit ligation cannot be excluded. Because the standard cleavage reaction catalyzed by the 5Ј-maturation enzyme RNase P would generate a 5Ј-monophosphorylated C ϩ1 nucleotide, an alternative mechanism may have generated the 5Ј-end of mt-tRNA Cys . One intriguing pos-sibility is that C ϩ1 is added by the same enzymes that catalyze 5Ј-editing in vivo. We note that tRNA Cys is the only one of the "unedited" mt-tRNAs in D. discoideum that contains a C at the ϩ1 position and that other tRNAs containing C ϩ1 are known to be miscleaved with some frequency by RNase P in bacteria (29). In this scenario, RNase P miscleavage could generate a 5Ј-truncated substrate (tRNA Cys initiating at G ϩ2 ) that would be repaired by DdiTLP3 and/or DdiTLP4 as was observed with the in vitro activity assays (Fig. 4).
The question of the true identity of the 5Ј-editing enzyme in vivo in D. discoideum remains unresolved by these assays, and based on these in vitro results, it remains possible that either DdiTLP3, DdiTLP4, or both enzymes together are required to participate in editing in the cell. Intriguingly, both DdiTLP3 and DdiTLP4 were observed to catalyze a single major addition product corresponding to the correct full-length mt-tRNA in the in vitro assays, suggesting that the enzymes possess some mechanism to control the 3Ј-5Ј addition reaction and avoid addition of excess nucleotides to the 5Ј-end of the tRNA during repair. The molecular basis for this activity and whether these enzymes utilize a strategy similar to that of the yeast Thg1 enzyme, which uses removal of a 5Ј-pyrophosphate residue from the added nucleotides to limit addition (30), remain to be addressed through further experiments. The structures of eukaryotic Thg1 and a bacterial TLP have been solved and indicate that Thg1/TLP family enzymes are tetramers, which has also been confirmed by solution measurements (31,32). Therefore, it is possible that a heterotetramer containing both DdiTLP3 and DdiTLP4 comprises the active form of the editing enzyme. In vitro mixing experiments in which equal concentrations of both DdiTLP3 and DdiTLP4 were mixed prior to initiating reactions did not reveal any significant stimulation of 5Ј-repair activity beyond that observed for either enzyme alone (data not shown), but these observations do not rule out the possibility that such a heterotetramer forms co-translationally or after mitochondrial import. DdiTLP3, but not DdiTLP4, is predicted to contain an N-terminal tag for mitochondrial localization; however, localization of either of these enzymes has not been experimentally determined and would be required to confirm unequivocally or rule out a role in mitochondrial tRNA 5Ј-editing for either enzyme in living cells (33).
In D. discoideum, five of the 10 edited mt-tRNAs and one unedited mt-tRNA have now been shown to undergo efficient repair by both DdiTLP3 and DdiTLP4 (Ref. 15 and Fig. 4). Although substrate-specific activities of either enzyme on one or more of the remaining edited tRNAs could yet be identified that might explain the presence of two possible editing enzymes in D. discoideum, the weight of the evidence obtained to date argues against this possibility. Instead, the 5Ј-repair component of the editing enzyme seems to act broadly on any 5Ј-truncated mt-tRNAs that are generated by the nuclease activity, suggesting that any substrate recognition component of the 5Ј-editing enzyme would appear to be exerted at the 5Ј-nuclease step. This behavior contrasts sharply with the strong tRNA substrate selectivity exhibited by the eukaryotic Thg1 members of the 3Ј-5Ј polymerase enzyme family that exhibit a Ͼ10 4 -fold preference for nucleotide addition to tRNA His over other tRNA species (34). The apparent lack of inherent tRNA substrate selectivity also opens the door to participation of TLPs in editing-like activities that do not depend on the presence of mismatched nucleotides but instead may involve alternative tRNA processing or quality control pathways. This may be particularly important with respect to the as-yet-unidentified functions of TLP enzymes in Bacteria and Archaea, which do not contain any 5Ј-mismatches in their encoded tRNAs (20,21,35). Studies to characterize the biological function of these enzymes are underway and may reveal pathways for general maintenance of tRNA 5Ј-end integrity.