A Disease-causing Point Mutation in Human Mitochondrial tRNAMet Results in tRNA Misfolding Leading to Defects in Translational Initiation and Elongation*

The mitochondrial tRNA genes are hot spots for mutations that lead to human disease. A single point mutation (T4409C) in the gene for human mitochondrial tRNAMet (hmtRNAMet) has been found to cause mitochondrial myopathy. This mutation results in the replacement of U8 in hmtRNAMet with a C8. The hmtRNAMet serves both in translational initiation and elongation in human mitochondria making this tRNA of particular interest in mitochondrial protein synthesis. Here we show that the single 8U→C mutation leads to a failure of the tRNA to respond conformationally to Mg2+. This mutation results in a drastic disruption of the structure of the hmtRNAMet, which significantly reduces its aminoacylation. The small fraction of hmtRNAMet that can be aminoacylated is not formylated by the mitochondrial Met-tRNA transformylase preventing its function in initiation, and it is unable to form a stable ternary complex with elongation factor EF-Tu preventing any participation in chain elongation. We have used structural probing and molecular reconstitution experiments to examine the structures formed by the normal and mutated tRNAs. In the presence of Mg2+, the normal tRNA displays the structural features expected of a tRNA. However, even in the presence of Mg2+, the mutated tRNA does not form the cloverleaf structure typical of tRNAs. Thus, we believe that this mutation has disrupted a critical Mg2+-binding site on the tRNA required for formation of the biologically active structure. This work establishes a foundation for understanding the physiological consequences of the numerous mitochondrial tRNA mutations that result in disease in humans.

The mitochondrial tRNA genes are hot spots for mutations that lead to human disease. A single point mutation (T4409C) in the gene for human mitochondrial tRNA Met (hmtRNA Met ) has been found to cause mitochondrial myopathy. This mutation results in the replacement of U8 in hmtRNA Met with a C8. The hmtRNA Met serves both in translational initiation and elongation in human mitochondria making this tRNA of particular interest in mitochondrial protein synthesis. Here we show that the single 8U3 C mutation leads to a failure of the tRNA to respond conformationally to Mg 2؉ . This mutation results in a drastic disruption of the structure of the hmtRNA Met , which significantly reduces its aminoacylation. The small fraction of hmtRNA Met that can be aminoacylated is not formylated by the mitochondrial Met-tRNA transformylase preventing its function in initiation, and it is unable to form a stable ternary complex with elongation factor EF-Tu preventing any participation in chain elongation. We have used structural probing and molecular reconstitution experiments to examine the structures formed by the normal and mutated tRNAs. In the presence of Mg 2؉ , the normal tRNA displays the structural features expected of a tRNA. However, even in the presence of Mg 2؉ , the mutated tRNA does not form the cloverleaf structure typical of tRNAs. Thus, we believe that this mutation has disrupted a critical Mg 2؉ -binding site on the tRNA required for formation of the biologically active structure. This work establishes a foundation for understanding the physiological consequences of the numerous mitochondrial tRNA mutations that result in disease in humans.
Human mitochondria are subcellular organelles that produce more than 90% of the energy required by the cell. The mitochondrial genome encodes 13 proteins necessary for energy production, two rRNAs and all of the 22 tRNAs required for the synthesis of these proteins (1,2). Mammalian mitochondrial tRNAs have several unusual features that distinguish them from canonical tRNAs. In many cases, they lack a number of the conserved or semi-conserved nucleotides that play important roles in creating the L-shaped tertiary structure of prokaryotic and eukaryotic cytoplasmic tRNAs (3). There is little detailed structural information on these tRNAs. No data are currently available that examine the structure of mammalian mitochondrial tRNAs with single nucleotide resolution. However, chemical and enzymatic probing has lead to the idea that these tRNAs have retained the basic cloverleaf structure of canonical tRNAs but that they lack several conserved tertiary interactions leading to a weaker three-dimensional structure (4 -8). In particular, a number of the long range interactions between the Dand T-arms of the tRNAs appear to be missing.
All 22 tRNAs that function in mammalian mitochondria are encoded in the mitochondrial DNA. Considerable interest in mitochondrial tRNAs centers on the occurrence of diseases arising from mutations in their genes that lead to maternally inherited genetic disorders (9 -12). The diseases associated with mitochondrial tRNA mutations may arise from failure in the processing of the tRNA (13), from reduced stability of the tRNA (14,15), from a reduction in aminoacylation (12,16,17), from a reduced ability of the mutated aminoacyl-tRNA to interact with mitochondrial elongation factor Tu (EF-Tu mt ) 3 (the corresponding prokaryotic factor is also designated EF1A) (16), and from the failure of the tRNA to be correctly modified leading to translational defects (18).
Normally, protein biosynthetic systems have two tRNA Met species. One is used solely for initiation, and the other functions in polypeptide chain elongation. Animal mitochondria are quite unusual in that they contain a single gene for tRNA Met , which functions in both polypeptide chain initiation and chain elongation. As a result of this dual role, mitochondrial Met-tRNA Met must be recognized by the mitochondrial Met-tRNA transformylase (MTF mt ) and be brought as fMet-tRNA Met to the ribosome for translational initiation (19). In addition, Met-tRNA Met must interact with elongation factor EF-Tu mt and bind to the A-site of the ribosome during translational elongation. Thus, this tRNA Met is of central importance in mitochondrial translation.
Human tRNA Met has a number of interesting features (Fig.  1A). The D-loop is somewhat small and lacks the G residues at positions 18 and 19 that facilitate interactions with the T-loop in the tertiary structure. The first position of the anticodon contains the rare modified base 5-formylcytidine. This modification may play a role in the unusual codon recognition requirements of this tRNA, which must recognize both AUG and AUA codons. The minor loop is short lacking the usual G47, and the T-stem has two adjacent pyrimidine:pyrimidine pairs (U-U and U-⌿). Furthermore, the T-loop lacks the T⌿C sequence and contains only six nucleotides instead of the normal seven. These unusual structural features suggest that human mitochondrial tRNA Met may have an intrinsically weak tertiary structure.
Three interesting point mutations (T4409C, A4435G, and G4450A) occur in the gene for human tRNA Met (hmtRNA Met ). The T4409C mutation (Fig. 1A) results in a U8 to C change at the corner of the acceptor stem and D-stem of hmtRNA Met . This mutation leads to mitochondrial myopathy resulting in dystrophic muscles and exercise intolerance (20). The A4435G mutation leads to the change of A37 to G37 in the anticodon loop of the tRNA (21). This mutation acts as a modulator of Leber's Hereditary Optic Neuropathy increasing the severity of this condition when it arises because of other mutations in the mitochondrial DNA. The G4450A mutation leads to loss of the final base pair in the T-stem (Fig. 1A). This mutation presents as splenic lymphoma, is largely confined to lymphocyte cells, and results in severely abnormal mitochondria leading to serious defects in energy production (22). A systematic examination of the structural and biochemical consequences of these mutations is lacking. Here we examine the structure of human mitochondrial tRNA Met and probe the effects of the 8U3 C mutation on the structure and function of this tRNA.

EXPERIMENTAL PROCEDURES
RNA Synthesis-Human mitochondrial tRNA Met transcripts for aminoacylation experiments were prepared by in vitro transcription using the hammerhead ribozyme construct described previously (23). The hmtRNA Met was purified by denaturing (10%) PAGE (29:1 acrylamide:bisacrylamide prepared with 7 M urea, 90 mM Tris borate, 2 mM EDTA), visualized by UV shadowing, excised from the gel, and recovered by passive elution in water followed by ethanol precipitation. hmtRNA Met transcripts for selective 2Ј-hydroxyl acylation analyzed by primer extension (SHAPE) experiments were prepared in the context of the structure cassette as described (24). D-and T-half-molecules were chemically synthesized (Dharmacon), purified, and analyzed as described previously (25).
Purification of E. coli Methionyl-tRNA Synthetase (MetRS) and Human Mitochondrial MetRS (hmMetRS)-A saturated overnight culture of JM109 cells carrying the pQE60-Escherichia coli MetRS plasmid construct (kindly provided by Uttam RajBhandary, Massachusetts Institute of Technology) was grown at 37°C in 2ϫ YT media (20 ml) supplemented with 50 g/ml ampicillin and used to inoculate 2 liters of 2ϫ YT media (50 g/ml ampicillin). The cells were grown at 37°C for 4 h (A 600 ϭ 0.6), induced with 50 M isopropyl ␤-D-thiogalactopyranoside and then grown at 37°C for 4 h post-induction. The cells were harvested by centrifugation at 4,000 rpm for 30 min. The cell pellet was resuspended in 500 ml of 10 mM Tris-HCl, pH 7.6, and then re-collected by low speed centrifugation. The cell pellet was fast frozen and stored at Ϫ80°C until use.
Cells carrying a plasmid encoding the His 6 -tagged human mitochondrial MetRS were grown as described (23). The cells were lysed as described above, and the hmMetRS was further purified as described (23).
Purification of Bovine Mitochondrial Methionyl-tRNA Transformylase (MTF mt )-E. coli BL21 cells, carrying the pET15-bovine MTF mt plasmid construct, were grown as described (19). Cells were harvested and lysed as described for E. coli MetRS above. The protein was purified as described for the E. coli MetRS except that the buffers contained 50 mM Tris-HCl, pH 7.6. The purified protein sample was dialyzed against 2 volumes of 500 ml of MTF dialysis buffer (20 mM Tris-HCl, pH 7.6, 100 mM KCl, 10% glycerol, and 3 mM ␤ME) for 1 h, fastfrozen, and stored at Ϫ80°C.
Assay for the Aminoacylation of Human Mitochondrial tRNA Met -The aminoacylation reactions for both the normal and 8U3 C mutated tRNA Met transcripts were performed essentially as described (23). Reaction mixtures (100 l) contained 50 mM Tris-HCl, pH 7.6, 2.5 mM MgCl 2 , 2.5 mM ATP, 0.  were separated on Partisil LK5D TLC plates (Whatman) with a butanol:acetic acid:water (4:1:1) mixture. TLC plates were visualized by phosphorimaging (GE Healthcare) and the spots were analyzed using the ImageQuant program.
Binding of Human Mitochondrial Met-tRNA Met to Bovine Mitochondrial EF-Tu (EF-Tu mt )-EF-Tu mt was prepared as described (26), except that the cells were lysed as described above for E. coli MetRS, and the high speed centrifugation step was omitted. Where indicated the normal U8 or 8U3 C mutated hmtRNAs were phosphorylated with cold ATP using polynucleotide kinase (New England Biolabs) prior to large scale aminoacylation.
To measure ternary complex formation, reaction mixtures (50 l) were prepared as reported (27) except that 20 mM Hepes-KOH, pH 7, and the indicated amounts of EF-Tu mt were used. The reactions were incubated for 15 min at 0°C or 6 min at 37°C as indicated. Free [ 35 S]Met-hmtRNA Met was digested by a 30-s incubation with 10 g of RNase A, and the reaction was terminated by the addition of cold 5% trichloroacetic acid. Following a 10-min incubation on ice, the [ 35 S]Met-hmtRNA Met precipitate was collected on nitrocellulose filters and quantified by liquid scintillation counting. Determination of the K d for ternary complex formation was carried out as described previously (28).
Degradation of hmtRNA Met in a Mitochondrial Extract-Bovine mitoplasts (0.2 g) were prepared as described (29). Mitoplasts were lysed in buffer (2 ml) containing 15 mM Tris-HCl, pH 7.6, 40 mM KCl, 6 mM MgCl 2 , 6 mM ␤ME, 0.8 mM EDTA, and 1.6% Triton X-100 by hand homogenization. The extract was clarified by centrifugation at 10,000 rpm for 15 min at 4°C. Normal U8 and 8U3 C mutated hmtRNA Met were 32 P-endlabeled using polynucleotide kinase (New England Biolabs) according to the manufacturer's instructions. Reaction mixtures (10 l) contained 100 nM normal U8 or 8U3 C mutated [ 32 P]hmtRNA Met and the indicated amount of the extract in the lysis buffer above. Incubation was for 10 min at 37°C, and cold trichloroacetic acid precipitation was used to quantitate the amount of [ 32 P]hmtRNA Met remaining.
Selective 2Ј-Hydroxyl Acylation Analyzed by Primer Extension (SHAPE) Analysis of Normal U8 and 8U3 C Mutated hmtRNA Met Transcripts-Normal U8 or 8U3 C mutated hmtRNA Met (12 pmol, 0.33 M) in 36 l of nuclease-free water (Ambion) was incubated at 50°C for 2 min and then cooled on ice for 2 min. The RNA was divided into 2 aliquots of 4 pmol (12 l) and 8 pmol (24 l). Folding buffer (6 l; 333 mM Hepes-KOH, pH 8, 333 mM NaCl) was added to the 4 pmol of RNA, and folding buffer with 20 mM Mg 2ϩ (12 l) was added to the 8 pmol of RNA, and the two samples were incubated at 37°C for 20 min. To 1 l of 100 mM 1-methyl-7-nitroisotoic anhydride (1M7) in anhydrous DMSO or 1 l of anhydrous DMSO (control), 9 l (2 pmol) of folded RNA was added and allowed to react at 37°C for 70 s (5 half-lives). The balance of the RNA folded in the presence of Mg 2ϩ (18 l) was divided into 2 aliquots of 2 pmol (9 l) each and stored at 37°C for sequencing. MgCl 2 (1 l; 64 mM) was added to the RNA treated with the folding buffer in the absence of Mg 2ϩ . Radiolabeled oligonucleotide (0.3 M; 3 l; 5Ј-32 P-GAACCGGACCGAAGCCCG, obtained from the Nucleic Acids Core Facility at University of North Carolina) was added to the 1M7-treated, DMSO-treated, or untreated RNA (2 pmol), and the samples were incubated at 65°C for 5 min and then at 35°C for 20 min for primer annealing. To each reaction, reverse transcription buffer (6 l; 250 mM KCl, 167 mM Tris-HCl, pH 8.3, 17 mM dithiothreitol, and 0.42 mM each dNTP) was added. Then either ddCTP or ddTTP (1 l; 5 mM; Amersham Biosciences) was added to the untreated RNA. After heating to 52°C, reverse transcriptase (1 l; 200 units; Superscript III, Invitrogen) was added, and the primer extension reactions were performed at 52°C for 5 min. Reactions were quenched with 4 M NaOH (1 l) and heated at 95°C for 5 min. For gel analysis, a gel loading solution (29 l; 138 mM unbuffered Tris-HCl, 73% (v/v) formamide, 2 mM Tris borate, 86 mM EDTA, pH 8, with xylene cyanol and bromphenol blue) was added, and the samples were heated at 95°C for an additional 5 min. The cDNA products from the ϩ and Ϫ 1M7 and sequencing reactions were separated by denaturing gel electrophoresis (10% polyacrylamide). Samples on gels (21 cm ϫ 40 cm ϫ 0.4 mm) were subjected to electrophoresis at 1400 V for ϳ2.5 h. Gels were visualized by phosphorimaging (GE Healthcare). The ϩ and Ϫ 1M7 band intensities were quantified using SAFA (30) and corrected for signal drop-off (31). SHAPE reactivities were normalized by subtracting intensities for the Ϫ1M7 control from the ϩ1M7 reaction and dividing each by the average reactivity of the most reactive 7% of the nucleotides. To facilitate comparison of the normal U8 and 8U3 C mutant tRNAs, the two data sets were normalized to the reactivity of the -CCA end nucleotides. The reactivity of each nucleotide was assigned a value between 0 and 1. Nucleotides fall into one of four categories (32, 33) as follows: unreactive (0.000 -0.055), low reactivity (0.055-0.110), moderately reactive (0.110 -0.220), or highly reactive (0.220 -1.000).
Structural Studies of hmtRNA Met Half-molecules-Reconstitution of hmtRNA Met from U8 and 8U3 C D-half-molecules with T-half-molecules required Mg 2ϩ and was assessed by gel mobility shift assays using native 15% PAGE in Tris borate buffer (89 mM Tris base, 89 mM boric acid, pH 8.3) with and without 3 mM Mg 2ϩ at 4°C (25). The concentration of the D-half-molecule was held constant at 31.2 M, whereas the concentration of the T-half-molecule was varied from 4.2 to 112 M.
UV-Monitored Thermodynamic Experiments-The halfmolecule RNA samples were dissolved in the above Tris borate buffer used for the PAGE experiments to obtain an RNA concentration of 1.2 M. MgCl 2 was added to a concentration of 3 mM. UV-monitored thermal denaturations and re-naturations were replicated 10 times and monitored by measuring UV absorbance (260 nm) using a Cary 3 spectrophotometer as published (34,35). The data points were averaged over 20 s and recorded with a temperature change of 1°C per min from 4 to 90°C. The one most inconsistent of the 10 melting transitions (either a denaturation or renaturation) was discarded from each set, and the resulting data were averaged on a point-bypoint basis.
For UV melts of the hmtRNA Met transcripts, the normal U8 and the 8U3 C hmtRNA Met were dialyzed against water using 10-kDa cutoff dialysis cups (Stratagene). The U8 and 8U3 C hmtRNA Met transcripts were diluted to 0.5 M in a buffer containing 10 mM NaCl and 10 mM Hepes-KOH, pH 8.0. The thermal denaturation of the tRNAs was monitored by UV absorbance at 260 nm using a Cary 3 spectrophotometer. Data points were recorded once per min from 4 to 95°C with a temperature change of 1°C per min. Following thermal renaturation, 6 mM Mg 2ϩ was added to the U8 and 8U3 C transcripts, and the UV-monitored thermal denaturation experiments were repeated. (23) have shown that the transcript of mitochondrial tRNA Met has aminoacylation properties similar to those observed with the native tRNA. Thus, it was possible to use the normal transcript and a transcript containing the 8U3 C mutation for studies on the effect of the mutation on the properties of the tRNA. The 8U3 C mutation leads to a myopathy presumably arising from a reduction in translational activity in mitochondria. To determine the biochemical consequence of the 8U3 C mutation, the abilities of the U8 and 8U3 C hmtRNA Met transcripts to be aminoacylated by the human mitochondrial methionyl-tRNA synthetase (hmMetRS) were tested. Aminoacylation is an early step required for the tRNA to be used in either the elongation or initiation phase of protein synthesis and is thus of central importance for protein synthesis in mitochondria. The normal U8 transcript was aminoacylated as expected (23); however, the 8U3 C mutation caused a significant reduction in the rate of aminoacylation of the tRNA by hmMetRS (Fig. 1B). This observation provides one clear rationale for the failure of this tRNA to function in mitochondrial protein biosynthesis.

Aminoacylation of the Normal and 8U3 C Mutated tRNA Met -Previous studies
Not unexpectedly, the normal hmtRNA Met was aminoacylated by the E. coli MetRS (Fig. 1C). Interestingly, whereas the 8U3 C hmtRNA Met was poorly aminoacylated by the hmMetRS, it was not aminoacylated at all by the E. coli MetRS (Fig. 1C) suggesting that the mutated tRNA had a significantly altered structure. The hmMetRS is believed to be both structurally and functionally homologous to its prokaryotic counterpart (23). However, this work demonstrates that the hmMetRS is less discriminatory than E. coli MetRS for the structure of the tRNA. Because a major determinant in the recognition of tRNA Met by the MetRS is thought to lie in the anticodon sequence that is unchanged (36), the weak aminoacylation most likely reflects significant structural alterations in the tRNA as a result of the mutation.
Formylation of Normal and 8U3 C Mutated Met-tRNA Met -The defective aminoacylation of the 8U3 C hmtRNA Met made it difficult to assess the effects of the mutation on additional steps in protein biosynthesis. However, small amounts of the aminoacylated 8U3 C mutated hmtRNA Met could be isolated, permitting a limited investigation of additional steps in translation.
In the mammalian mitochondrial system, the Met-tRNA Met must be formylated by the mitochondrial transformylase (MTF mt ) to be used in initiation (19,37). The abilities of the U8 and 8U3 C Met-tRNA Met to be formylated were tested by incubation of the [ 35 S]Met-tRNA with the bovine MTF mt in the presence of folinic acid to provide the formyl group. The conversion of [ 35 S]Met to [ 35 S]fMet was monitored by TLC (Fig. 2). The normal hmtRNA Met formylates well, but interestingly, formylation of the 8U3 C mutated tRNA is barely detectable (Fig. 2). The inability of the 8U3 C mutated tRNA to be efficiently aminoacylated and formylated suggests there would be little fMet-tRNA Met available for initiation of protein synthesis.
Binding of Normal and 8U3 C Mutated Met-tRNA Met to EF-Tu mt -The hmtRNA Met must be partitioned between the initiation and elongation stages of protein synthesis. To function in elongation, the formylation step must be by-passed and the Met-tRNA Met must form a ternary complex with EF-Tu mt (also known as mitochondrial EF1A). The binding of aminoacyl-tRNA to EF-Tu mt is normally studied at 0 -4°C to protect the aminoacyl bond from hydrolysis (38,39). Because the strength of the interaction of Met-hmtRNA Met with EF-Tu mt had never been determined, these experiments were begun with an investigation of the binding of EF-Tu mt to the normal Met-tRNA Met . Because the transcript begins with a 5Ј-terminal A rather than the G residue preferred by the T7 RNA polymerase, the construct used for transcription contained a hammerhead ribozyme and T7 promoter gene sequence upstream of the tRNA sequence. Following transcription, the hammerhead ribozyme was cleaved leaving behind a tRNA beginning with a 5Ј-OH rather than the 5Ј-phosphate normally present in tRNAs. The presence of a 5Ј-phosphate is not required for aminoacylation or formylation of this tRNA (data not shown). The formation of the ternary complex with E. coli EF-Tu is not affected by the 5Ј-phosphate (40), but no studies were available on the importance of this group for EF-Tu mt . The first issue addressed was whether the ability of EF-Tu mt to bind the aminoacylated tRNA in the ternary complex was affected by the presence of the 5Ј-phosphate. For this experiment, the aminoacylated transcript carrying the 5Ј-OH was tested in ternary complex formation and compared with an aminoacylated transcript that carried a 5Ј-phosphate following phosphorylation by polynucleotide kinase. As indicated in Fig. 3A, the Met-tRNA transcript carrying the 5Ј-phosphate formed a ternary complex with EF-Tu mt efficiently. However, the transcript with the 5Ј-OH was noticeably less active in interacting with EF-Tu mt . This observation indicates that, unlike E. coli EF-Tu, the interaction of EF-Tu mt with Met-tRNA benefits from contact with the phosphate at the 5Ј-end of the aminoacyl-tRNA. Examination of the residues in the binding pocket for the aminoacyl-tRNA in the bacterial and mitochondrial factors reveals that Glu-287 in E. coli EF-Tu is replaced by the oppositely charged residue Arg-335 in EF-Tu mt (27). This change alters the charge balance in the region of EF-Tu in close proximity to the 5Ј-end of the tRNA in the ternary complex and may provide an additional stabilizing interaction enhancing the binding of EF-Tu mt to the conformationally fragile mitochondrial tRNAs.
Analysis of the binding curve between EF-Tu mt and the phosphorylated Met-hmtRNA Met indicates a binding constant of 27 Ϯ 6 nM at 0°C. This value is in good agreement with the K d value observed for the binding of several other mitochondrial aminoacyl-tRNAs to EF-Tu mt (41).
The interaction of the 8U3 C Met-hmtRNA Met was then investigated at 0°C using the phosphorylated transcript (Fig.  3B). Rather surprisingly, the mutated tRNA was as effective as the normal transcript in interacting with EF-Tu mt under these conditions suggesting that the 8U3 C Met-tRNA had adopted a tRNA-like conformation recognized by this factor. This observation appeared to be in conflict with the poor activity of the 8U3 C hmtRNA Met in aminoacylation and formylation. However, the previous assays were carried out at 37°C, whereas the ternary complex assay was carried out at 0°C. Hence, ternary complex formation was tested at 37°C with both the phosphorylated U8 and 8U3 C Met-hmtRNA Met species (Fig. 3C). At this higher temperature, the 8U3 C Met-hmtRNA Met was almost inactive in ternary complex formation, whereas the U8 Met-hmtRNA Met had significant activity in ternary complex formation. These data suggest a temperature-dependent destabilization of the structure of the 8U3 C mutated tRNA. Structural probing provided insight into this question as described below.  Chemical Probing of the Structure of the Normal U8 tRNA Met in the Presence and Absence of Mg 2ϩ -The structures of the normal U8 and 8U3 C mutant hmtRNA Met transcripts were probed using SHAPE chemistry at 37°C (24,42). SHAPE chemistry relies on the differential reactivity of the nucleotide 2Ј-OH to the small molecule 1M7 (24,42). Nucleotides in flexible, single-stranded regions are preferentially modified by the reagent, revealing different conformational states of residues within the RNA. The chemically treated tRNA transcripts were reverse-transcribed using a radiolabeled primer. Sites of modification were identified based on their ability to obstruct reverse transcription. The cDNA products were analyzed on a denaturing polyacrylamide gel. When residues in unstructured or flexible regions of the tRNA are modified, the reverse transcriptase stops, leading to a labeled band observed on the gel one residue shorter than the position of the modification. Comparison of the position of these bands to a sequencing ladder allows the identification of residues susceptible to modification (32). The reactivity of each nucleotide toward 1M7 was quantified, and the relative reactivity of each nucleotide was assigned a value between 0 and 1. The higher the value, the greater the propensity the nucleotide has to be in an unstructured region. This information allowed us to develop a model of the tRNA structure (43).
The reactivity of 1M7 is not Mg 2ϩ -sensitive; hence, SHAPE chemistry was used to examine the influence of Mg 2ϩ on the structure formed by the normal and 8U3 C mutant hmtRNA Met transcripts. In the presence of Mg 2ϩ , a number of distinct structural features were observed for the normal hmtRNA Met (Fig. 4, A and B). The 2Ј-OH of residues predicted to form the acceptor stem and the D-stem were not accessible to the reagent. This indicated that they were largely paired as expected. Only a single residue in the D-loop was highly reactive suggesting that most of the residues in this region of the tRNA were involved in tertiary interactions that reduced their reactivity. The anticodon stem was clearly protected, but as expected residues in the anticodon loop were available for modification. The variable loop again showed limited modification indicating tertiary interactions. These interactions were expected to take place primarily with residues in the D-loop as would be expected from tertiary contacts observed in canonical tRNAs. The T-stem apparently formed despite the presence of the U-U pairs. However, in contrast to other tRNAs (33), residues in the T-loop were quite reactive indicating that they were accessible in the tertiary structure of this tRNA.
The folding of the hmtRNA Met structure required the presence of Mg 2ϩ (Fig. 4A, lanes 1 and 3 compared). Mg 2ϩ not only stabilizes the negatively charged backbone but is known to bind to specific sites of canonical cytoplasmic tRNAs to aid in folding their tertiary structures. In the absence of Mg 2ϩ , the hmtRNA Met structure was open and highly reactive to 1M7, indicating that both secondary and tertiary interactions had been lost (Fig. 4A, lane 1).
A more thorough analysis of the SHAPE data through quantitation of the individual nucleotide reactivities provided additional insight into the three-dimensional structure formed by the unusual hmtRNA Met . The SHAPE reactivity pattern was suggestive of a folded cloverleaf-shaped tRNA and was similar to the reaction pattern observed with other tRNAs (33) (Fig.  5A). Superimposition of the nucleotide reactivities on the L-shaped tRNA structure (43) (Fig. 5B) provided additional information about possible tertiary interactions within the tRNA. For clarity, the nucleotides have been colored according to their level of reactivity. As expected, the most reactive portion of the tRNA was the anticodon loop. The reactivity data support a structure in which many of the conserved canonical tertiary interactions are preserved. For example, nucleotides U8, A14, and A21 which, in canonical tRNAs form a triple base pair, were all modestly reactive or nonreactive in hmtRNA Met . Other possible preserved tertiary interactions are shown in Fig.  5B and include the expected interaction between nucleotides in the D-loop and the variable loop such as between A15 and U48.
Despite the likely presence of many conserved tertiary interactions in hmtRNA Met , other interactions are probably not occurring because of the shortened sequence of this tRNA. For example, interactions between the D-and T-loops may be different from canonical tRNAs because the D-loop is short and lacks the common GG sequence, whereas the T-loop is only six nucleotides instead of the universal seven nucleotides found in classical tRNAs. These deviations from classical tRNAs indicate that hmtRNA Met cannot have the G18-U55 and G19-C56 interactions that normally stabilize the corner of the L-shaped structure (3). Also, A58 in the T-loop was not reactive, whereas its expected partner U54 showed significant reactivity. This suggests that the canonical tertiary interaction of U54 with A58 may not occur. The tertiary interactions occurring in the T-loop of the hmtRNA Met may be significantly different from that of the canonical cytoplasmic tRNAs.
Chemical Probing of the Structure of the 8U3 C Mutated tRNA in the Presence and Absence of Mg 2ϩ -SHAPE analysis of the 8U3 C mutated hmtRNA Met carried out at 37°C in the presence of Mg 2ϩ indicated that all of the nucleotides in the tRNA were susceptible to modification except the three G-C pairs in the anticodon stem. The single nucleotide mutation (8U3 C) at the corner of the acceptor stem and D-stem of the hmtRNA Met thus resulted in a drastic loss of structure that is seen even in the presence of Mg 2ϩ (Fig. 4C,  lane 3). The regions of the D-loop, the variable loop, acceptor stem, and T-stem all exhibited an increased reactivity to the reagent 1M7. For example, all of the residues in the D-loop and most of the residues in the variable loop were now reactive. Furthermore, the residues in the acceptor stem, which were base-paired in the normal U8 tRNA, were reactive. These changes indicated that the tertiary structure and a large portion of the secondary structure had been lost because of the mutation. As expected, the 8U3 C mutant hmtRNA Met had essentially no structure in the absence of Mg 2ϩ , as was observed for the normal hmtRNA Met (Fig.  4C, lane 1). Even the anticodon stem had lost its base-paired structure. Clearly, the folding of the mutated tRNA was not responding to the presence of Mg 2ϩ and thus may have lost one or more Mg 2ϩ -binding sites critical to the structure. It is likely that the loss in structure observed upon mutation of 8U3 C results from a loss of stabilizing tertiary interactions that were present in the normal hmtRNA Met and that are dependent on the presence of Mg 2ϩ . It should be noted that, alternatively, the tRNA could be folding into a dynamic mixture of transient conformations that do not reflect the structure of the normal U8 hmtRNA Met .
Analysis of the 8U3 C mutated tRNA clearly showed a global increase in reactivity (Fig. 5, C and D). This increase was seen in all of the stems except the anticodon stem suggesting a loss of these secondary structural elements. Both the D-and T-loops also significantly increased in reactivity, suggesting a loss in stabilizing tertiary interactions. The variable loop, which makes extensive tertiary interactions in the U8 hmtRNA Met , became highly reactive in the 8U3 C tRNA. Although the mutated 8U3 C residue itself showed only a slight increase in reactivity, its tertiary interaction partners, A14 and A21, became highly reactive suggesting that they were unable to form the necessary stabilizing tertiary interactions in this region of the tRNA. When these reactivities were superimposed on the L-shaped hmtRNA Met structure, it was clear that changing the single U8 nucleotide to a C lead to a loss of both secondary and tertiary interactions resulting in a tRNA that was largely unstructured (Fig. 5D).
Stability of Normal and 8U3 C Mutated hmtRNA Met in a Mitochondrial Extract-A number of mutated mitochondrial tRNAs are more rapidly degraded in vivo than are the normal FIGURE 5. Tertiary interactions in the hmtRNA Met and the effect of mutating U8 to C on these interactions. A, histogram showing the reactivity of nucleotides in the U8 hmtRNA Met transcript. Nucleotides with high, medium, low, and no reactivity are shown in red, orange, green, and black, respectively (see "Experimental Procedures"). Structural regions in the tRNA are shown above the histogram. The numbering shown is from the Sprinzl system (53). B, L-shaped structure of hmtRNA Met . Possible tertiary interactions occurring in the tRNA are shown and are based on conserved interactions that occur in the canonical cytoplasmic tRNAs. The nucleotides are colored based on reactivity as in A above. C, reactivity of nucleotides in 8U3 C hmtRNA Met . The histogram is labeled and colored as in A above. D, unstructured form of the 8U3 C mutated hmtRNA Met . The colors of the nucleotides are based on their reactivity to 1M7 as in the histogram in C. In contrast to the reactivity of the nucleotides of the normal U8 hmtRNA Met to 1M7, the reactivity data for the 8U3 C mutant hmtRNA Met clearly did not fit the L-shaped structure.
tRNAs (43,44). To determine how the 8U3 C mutation might affect the stability of the tRNA within mitochondria, the degradation of the normal and 8U3 C mutated hmtRNA Met was monitored in an extract prepared from the mitochondrial matrix. For this analysis, the U8 and 8U3 C transcripts were labeled at the 5Ј-end with 32 P-phosphate and incubated with varying concentrations of the mitochondrial extract. The amount of the tRNA remaining intact was determined by precipitation with trichloroacetic acid. The 8U3 C mutated tRNA was more readily degraded by the mitochondrial nucleases than the normal U8 tRNA (Fig. 6). When visualized on a sequencing gel, it was apparent that the normal tRNA was degraded in discrete locations corresponding to loops, whereas the mutated tRNA was degraded throughout the body of the tRNA (data not shown). The differential degradation of the two tRNAs was in agreement with the SHAPE data showing an overall lack of structure in the mutated tRNA. Furthermore, these data suggested that, in addition to being incorrectly folded, the mutated tRNA will also be more rapidly degraded within the milieu of the mitochondrial matrix.
Association of T-half-molecules with U8 and 8U3 C D-half-molecules-To further probe the effect of the 8U3 C mutation on the structure of the tRNA in the presence and absence of Mg 2ϩ , D-and T-domain hmtRNA Met half-molecules were chemically synthesized with the two pseudouridines, ⌿, and either U8 or 8U3 C in the D-half-molecule (Fig. 7A). Previously, we demonstrated that incubation of the normal D-and T-half-molecules of bovine mtRNA Met in the presence of Mg 2ϩ resulted in reconstitution of the mtRNA Met (25). The reconstituted tRNA could be observed as a slow moving band under native conditions. The effect of the 8U3 C mutation on the reconstitution of the human mtRNA Met was assessed using PAGE migration in the presence and absence of Mg 2ϩ .
In PAGE, when the slower migrating mtRNA Met D-halfmolecule containing the normal U8 was titrated with the faster migrating T-half-molecule at 3 mM Mg 2ϩ , a gel shift was observed that corresponded to the formation of a complex between the two half-molecules (Fig. 7C) (25). In fact, two new bands attributed to the reconstituted tRNA appeared to increase with increasing amounts of the T-halfmolecule. This increase was accompanied by the expected decrease in the D-half-molecule (Fig. 7C). The two gel bands attributed to the reconstituted mtRNA Met contained both half-molecules, as determined by denaturing PAGE (data not shown). The slower moving mtRNA Met (ϳ70% of the reconstituted tRNA) migrated in a fashion similar to that of E. coli cytoplasmic tRNA (Fig. 7C), and to the migration found for the complex of bovine mtRNA Met half-molecules (25). The faster moving band representing some 30% of the reconstituted human mtRNA Met migrated faster than the D-halfand slower than the T-half-molecule. It is believed to be folded into a second, but non-native, conformation. In the absence of Mg 2ϩ , the migration of the D-half-molecule was more rapid than in the presence of Mg 2ϩ suggesting a Mg 2ϩdependent conformational change in the half-molecule. In the absence of Mg 2ϩ , the two half-molecules could not form a complex (Fig. 7, C and D).
The mutated 8U3 C D-half-and normal T-half-molecules were incapable of forming a complex either in the presence or absence of Mg 2ϩ (Fig. 7, E, F, and B). The migration of the mutant 8U3 C D-half-molecule was much faster than that of the normal D-half-molecule in the presence of Mg 2ϩ , suggesting a conformational difference in the half-molecule resulting from the 8U3 C change. In addition, the rapid migration of the 8U3 C D-half-molecule was not affected by Mg 2ϩ . These results suggested that the 8U3 C D-half-molecule had lost the ability to bind one or more structurally important Mg 2ϩ ions. The hmtRNA Met transcripts were subjected to gel electrophoresis under native conditions and in the presence and absence of Mg 2ϩ (data not shown). In the presence of Mg 2ϩ , the 8U3 C mtRNA Met transcript migrated more slowly than the normal transcript. In the absence of Mg 2ϩ , only a slight difference in migration was observed. These data again indicated that the 8U3 C and normal mtRNA Met responded differently to Mg 2ϩ .
Thermal Denaturation of the Mutated tRNA-The thermal stabilities of the U8 and 8U3 C D-half-molecules and the normal T-half-molecule were also determined in the presence and absence of Mg 2ϩ . The addition of counterions, particularly Mg 2ϩ , to RNA stabilizes its structure. The PAGE results suggested that the 8U3 C D-half-molecule had lost the ability to bind one or more critical Mg 2ϩ . The U8 half-molecule gave different melting profiles in the presence and absence of Mg 2ϩ (Fig. 8A). In contrast, the 8U3 C D-half-molecule had the same thermal melting profile in both the presence and absence of Mg 2ϩ (Fig. 8B) indicating that it was unable to respond to the addition of this counterion. This observation again argues that the mutation led to the loss of a Mg 2ϩbinding site.
The normal T-half-molecule exhibited a major thermal transition that was stabilized by the presence of Mg 2ϩ (Fig. 8C). This was surprising considering that the T-stem contained two adjacent U-U mismatches (Fig. 1A). However, tandem U-U mismatches form one of the most stable internal loops in RNA and, despite the stem distortion that results, their presence actually stabilizes duplex RNA (45,46). Thermal denaturation of the intact U8 hmtRNA Met transcript demonstrated the importance of Mg 2ϩ for the correct folding of hmtRNA Met . In the absence of Mg 2ϩ , the U8 tRNA FIGURE 6. Stability of the normal and 8U3C mutated hmtRNA Met in a mitochondrial extract. The percentage of trichloroacetic acid-precipitable counts for the normal U8 transcript (diamonds) and the 8U3 C mutated (squares) hmtRNA Met remaining after incubation of 32 P-labeled transcripts with increasing amounts of a mitochondrial matrix extract. began melting at low temperatures but demonstrated a major transition at around 30°C (Fig. 9A, blue). In the presence of Mg 2ϩ , the U8 tRNA was stably folded until about 50°C where a major transition occurred (Fig. 9A, pink). The shift in melting temperature in the presence of Mg 2ϩ demonstrated that the structure was stabilized by the presence of Mg 2ϩ . Conversely, melting the 8U3 C hmtRNA Met transcript did not show a strong dependence on Mg 2ϩ . In both the presence and absence of Mg 2ϩ , the 8U3 C transcript failed to show a major thermal transition, although significant hyperchromicity was observed (Fig. 9B). Clearly the mutant 8U3 C tRNA did not show the same structural response to Mg 2ϩ as did the normal U8 tRNA. In addition, the lack of a major thermal transition for the 8U3 C mutated tRNA agrees with the results of the SHAPE experiments, which indicated that the 8U3 C mutated tRNA lacks significant structure.

DISCUSSION
Considerable information is now available on the sequences of mammalian mitochondrial tRNAs. With some exceptions, these tRNAs can be drawn as cloverleaf structures and are thought to be able to fold into a three-dimensional structure that resembles the classical L-shape of canonical tRNAs (43). In mammals, the mitochondrial tRNAs are A/U-rich reflecting the base composition of the genomes. In general, these tRNAs are shorter (71-72 nucleotides in length) than bacterial or eukaryotic cytoplasmic tRNAs (about 76 nucleotides for class I tRNAs). The D-loop tends to be smaller and most of them lack the GG sequence that is involved in tertiary interactions in normal tRNAs (8). The T-loop varies from five to nine residues deviating from the highly conserved seven residues found in cytoplasmic tRNAs. The cryo-EM study on the structure of the bovine mitochondrial ribosome (47) shows a tRNA tightly bound at the P-site. The structure of this tRNA can be fit into the crystallographic structure of a cytoplasmic tRNA except in the region of the elbow perhaps reflecting the smaller sizes of the T-and D-loops of many mitochondrial tRNAs. The D-and T-stems also vary significantly in size and often contain mismatches. Searches for tertiary interaction networks corresponding to those found in other tRNAs have not been successful, and no compensating set of interactions can be deduced from sequence alignments (43). As discussed in more detail under "Results," our data indicate that hmtRNA Met forms a cloverleaf structure but that D-loop and T-loop interactions cannot follow the pattern observed in canonical tRNAs.
More than 130 pathogenic mutations (see MITOMAP available online) have been observed in mitochondrial tRNA genes (48). Numerous studies have been carried out in efforts to delin- eate the underlying causes for the defects observed in genetic diseases arising from mutations in mitochondrial tRNA genes (43). In many cases, numerous deleterious biochemical effects are observed with the mutated tRNAs making it difficult to assess the precise cause of the pathogenicity. There are several examples in which structural perturbations are believed to arise as a result of the mutations (14,49). One of the more complete studies recently published (50) examined mutations in tRNA Tyr . These studies showed that mutation of G15 and A22 in the D-loop as well as U54 in the T-loop resulted in structural alterations in the mutated tRNAs.
These mutations are thought to disrupt tertiary interactions and lead to the loss of aminoacylation of these tRNAs.
The data presented here indicate that the single point mutation 8U3 C in hmtRNA Met causes the loss of stable structure at physiological temperatures. The inability of the 8U3 C mutated tRNA to fold properly severely impedes aminoacylation, formylation, and interaction with EF-Tu mt . Consequently, the mutated tRNA cannot function effectively in either translational initiation or elongation. With a significant proportion of the heterogeneous population of mitochondria having the 8U3 C mutation, the result is the pathogenic manifestation of disease.
Residue U8 plays an important role in forming and stabilizing the tertiary structure of tRNA. Early work (51) indicated that Mg 2ϩ also plays a critical role in stabilizing the structure of the D-arm and its tertiary interactions. The crystal structure of yeast tRNA Phe clearly shows that Mg 2ϩ is tightly and site-specifically coordinated in binding pockets formed by tertiary interactions within the tRNA (52). It is believed that formation of these binding pockets precedes Mg 2ϩ binding, and that the subsequent binding of Mg 2ϩ then stabilizes the three-dimensional architecture of the tRNA (52). A single tightly coordinated Mg 2ϩ is bound at the elbow of tRNA Phe . The binding site for this ion is formed by nucleotides in the D-arm and at the corner of the D-stem and acceptor stem (Fig. 10A). The phosphate of U8 contacts this Mg 2ϩ through a bridging water molecule. U8 is also involved in a triple non-Watson-Crick pairing interaction with nucleotides A14 and A21 forming the binding pocket (Fig. 10B). Interestingly, in canonical tRNAs, U8 is highly conserved and is considered a universal nucleotide (Fig.   FIGURE 8. Effect of Mg 2؉ on the thermal stabilities of the normal and 8U3C mutated hmtRNA Met D-half-molecules and the normal T-halfmolecule. Normal U8 (A), 8U3 C (U8C) mutated D-half-molecules (B), and normal T-half-molecule (C) were subjected repeatedly to thermal denaturation and renaturation monitored at 260 nm. The average of nine melt profiles conducted on different days are plotted for each RNA in the absence (blue) and presence (pink) of 3 mM Mg 2ϩ . FIGURE 9. Thermal denaturation of the U8 and 8U3C transcripts in the presence and absence of Mg 2؉ . A, thermal denaturation of the U8 transcript was monitored at 260 nm in the absence (blue) and presence (pink) of 6 mM Mg 2ϩ . B, 8U3 C (U8C) transcript in the absence (blue) and presence (pink) of 6 mM Mg 2ϩ was subjected to thermal denaturation monitored at 260 nm.
10C) underlying the critical role of this nucleotide in forming the site-specific Mg 2ϩ binding pocket that leads to a correctly folded, functionally L-shaped tRNA. Our work suggests that mutation of this single nucleotide to cytosine (8U3 C) in hmtRNA Met prevents formation of this Mg 2ϩ binding pocket, resulting in a tRNA that fails to fold in the presence of Mg 2ϩ . The loss of the structure of the tRNA could occur because of a disruption in the base pairing between nucleotides U8 and either A14 or A21, which in turn would destabilize the sharp turn and the Mg 2ϩ binding pocket. Alternatively, loss of structure could result from an inability to properly coordinate the Mg 2ϩ ion via a water molecule in the pocket. Without the stabilizing effect of Mg 2ϩ bound at this site, a drastic loss of structure results leading to a tRNA that does not effectively participate in protein biosynthesis. Although it is not possible to directly measure the specific binding of an individual Mg 2ϩ ion, the data presented here are in agreement with this basic hypothesis.
Of the 22 tRNAs present in human mitochondria, 16 of them have U at position 8 and 14 of them have the U8-A14-A21 combination seen in hmtRNA Met and canonical tRNAs. It is of interest to note that the most common mutation in human mitochondrial tRNAs (the A3243G mutation in the tRNA Leu gene associated with MELAS) is the mutation of A14 to G. The extremely deleterious effect of this mutation is also likely to arise from an inability to form the tertiary interaction involving the U8-A14-A21 base triple and Mg 2ϩ coordination. The absence of tertiary interactions between the D-and T-loop of the hmtRNA Met suggests a substantial role for U8 and its associated Mg 2ϩ -binding site in stabilizing the three-dimensional structure of the mitochondrial tRNA.