A natural non-Watson-Crick base pair in human mitochondrial tRNAThr causes structural and functional susceptibility to local mutations.

Six pathogenic mutations have been reported in human mitochondrial tRNAThr (hmtRNAThr); however, the pathogenic molecular mechanism remains unclear. Previously, we established an activity assay system for human mitochondrial threonyl-tRNA synthetase (hmThrRS). In the present study, we surveyed the structural and enzymatic effects of pathogenic mutations in hmtRNAThr and then focused on m.15915 G > A (G30A) and m.15923A > G (A38G). The harmful evolutionary gain of non-Watson-Crick base pair A29/C41 caused hmtRNAThr to be highly susceptible to mutations disrupting the G30-C40 base pair in various ways; for example, structural integrity maintenance, modification and aminoacylation of tRNAThr, and editing mischarged tRNAThr. A similar phenomenon was observed for hmtRNATrp with an A29/C41 non-Watson-Crick base pair, but not in bovine mtRNAThr with a natural G29-C41 base pair. The A38G mutation caused a severe reduction in Thr-acceptance and editing of hmThrRS. Importantly, A38 is a nucleotide determinant for the t6A modification at A37, which is essential for the coding properties of hmtRNAThr. In summary, our results revealed the crucial role of the G30-C40 base pair in maintaining the proper structure and function of hmtRNAThr because of A29/C41 non-Watson-Crick base pair and explained the molecular outcome of pathogenic G30A and A38G mutations.

INTRODUCTION

Mitochondria are present in most eukaryotic cells and generate the majority of cellular energy via oxidative phosphorylation (1). Mitochondria have their own genome and protein synthesis system. The human mitochondrial genome encodes 13 essential subunits of respiratory chain complexes, two rRNAs, and 22 mitochondrial tRNAs (hmtRNAs) (2). In the mitochondrial translation system, the RNA components are supplied by the mitochondria, whereas the protein components, such as ribosomal proteins, aminoacyl-tRNA synthetases (aaRSs), and various factors, are encoded in nucleus and transported into the mitochondria from the cytoplasm (3).

In human mitochondria, all amino acids occupy only one corresponding hmtRNA, except for two copies for hmtRNASer and hmtRNALeu. To date, more than 250 pathogenic mutations associated with mitochondrial diseases have been reported in hmtRNAs genes (http://www.mitomap.org/MITOMAP) (4), indicating the essential role of hmtRNAs in mitochondrial functions. These mutations affect the stability and functions of tRNAs (5,6). For example, the mitochondrial myopathy, encephalopathy, lactic acidosis, and stroke-like episodes syndrome (MELAS)-associated mutations in hmtRNALeu (UUR) result in marked reduction in aminoacylation and a fragile structure (7). The U48C mutation causes T-stem slip in hmtRNALeu (CUN) (8). The A8344G mutation in hmtRNALys causes the abolition of the m5s2U modification, resulting in severe translation failure in myoclonic epilepsy with ragged red fibers (MERRF) (9,10). The G52A and A57G mutations in hmtRNALeu (CUN) and hmtRNALeu (UUR) lead to lower stability and decreased tRNA charging capacity (11). These observations imply that functional defects of tRNAs play a key role in mitochondrial translation and molecular pathogenesis.

The pivotal role of tRNA is to generate aminoacyl-tRNA, catalyzed by aaRS, as the first step of protein biosynthesis. The aminoacylation reaction requires a high level of accuracy and provides a critical checkpoint for translational quality control (12). Some aaRSs have evolved editing functions to ensure removal of incorrect aa-AMPs (pre-transfer editing) or mis-charged tRNAs (post-transfer editing) (13,14). A slight decrease in cytoplasmic aminoacylation accuracy could cause an intracellular accumulation of mis-folded proteins and upregulation of protein chaperones in neurons, causing severe mammalian neurodegeneration (15). Similarly, impaired mitochondrial aminoacylation accuracy leads to embryonic lethality (16).

We have cloned TARS2, encoding the human mitochondrial threonyl-tRNA synthetase (hmThrRS) precursor, and purified the mature form of hmThrRS. hmThrRS is active in hmtRNAThr aminoacylation and utilizes tRNA-dependent pre-transfer editing and post-transfer editing to maintain translational quality control. Establishing such an aminoacylation and editing determination system enables us to further study the molecular basis of the potential pathogenic mechanism for hmtRNAThr in amino acid charging. So far, six mutations in the hmtRNAThr gene have been reported as pathogenic. A patient carrying the m.15915 G > A (G30A) mutation in hmtRNAThr gene displayed hearing loss, muscle weakness, mental retardation, and seizures (17). The m.15923A > G (A38G) mutation caused lethal infantile mitochondrial myopathy (LIMM) and the patients showed hypoglycemia, lactic acidosis, and sudden multisystem failure (18). The m.15924A > G (A39G) mutation led to lethal respiratory chain defects and was fatal in infancy (19). A patient carrying the Δ15940U (ΔU60) mutation showed Leber's hereditary optic neuropathy (LHON) (20). The m.15950 G > A (G70A) mutation was associated with Parkinson's disease, which could contribute to dopaminergic nerve cell death (21). A patient carrying the m.15951A > G (A71G) mutation exhibited loss of vision (22). However, their molecular mechanisms need to be explored.

tRNAs acquire a large variety of modifications during the maturation process. Among these, the N6-threonylcarbamoyladenosine (t6A) modification, which is often located at position 37, aroused our interest. The t6A modification is found in almost all tRNAs reading ANN codons (N being one of the four nucleotides A, C, G and U) of all three kingdoms: bacteria, eukaryotes, and archaea, and also in mitochondria and chloroplasts (23). The t6A modification facilitates codon-anticodon interaction and promotes translational fidelity (24,25). Deficiency of the t6A modification leads to an obvious increase in frameshifting events in cells (26). Although the enzymes responsible for the t6A modification of mammalian mitochondrial tRNAs have not been identified, the in vitro t6A biosynthesis reaction using Saccharomyces cerevisiae Sua5 and Qri7 have been established (27). Sua5 catalyzes threonine, bicarbonate, and ATP to generate the intermediate threonylcarbamoyl-AMP (TCA) (28). TCA is then released from Sua5 and is bound by Qri7 to participate in the t6A modification (27). According to previous reports, one of the pathogenic mutations of hmtRNAThr, A38G, is next to t6A37 of tRNAs. The t6A modification occurs at A37 in bovine mitochondrial tRNAThr (bmtRNAThr), suggesting that hmtRNAThr likely has a t6A modification. We are interested in whether hmtRNAThr could be t6A-modified in vitro and, if so, what are the effects of these pathogenic mutations on the modification of t6A37?

In the present study, based on initial structural and functional screen of all pathogenic mutations, we focused on the G30A and A38G mutations, which are located in the anticodon stem and loop of hmtRNAThr (Figure 1A). We revealed the intrinsic susceptibility of the anticodon stem in terms of structural integrity, aminoacylation, editing, and tRNA modification, which was caused by an evolutionary non-Watson–Crick A29/C41 base pair in hmtRNAThr. Furthermore, A38 contributes significantly to aminoacylation and editing, and is a determinant for the t6A modification. Our results provide clues to the potential molecular mechanisms of pathological hmtRNA mutations.

Figure 1.

Cloverleaf structure and stability of hmtRNAThr and its mutants. (A) Cloverleaf structure of hmtRNAThr with pathogenic mutations. (B) Analysis of the conformations of the pathogenic mutants under native and denaturing conditions. (C) Tm values of pathogenic mutants. (D) Aminoacylation of pathogenic related mutants. hmtRNAThr (), -G30A (), -A38G (), -A39G (), -U60 (), -G70A (), -A71G () and without enzyme addition as control (). Error bars indicate the standard deviations.

MATERIALS AND METHODS

Materials

l-Thr, l-Ser, dithiothreitol, nucleoside triphosphates (NTPs), guanosine monophosphate (GMP), tetrasodium pyrophosphate, Tris-base, MgCl2, NaCl and inorganic pyrophosphate were purchased from Sigma (St Louis, MO, USA). [14C]Thr was obtained from Biotrend Chemicals (Destin, FL, USA). [α-32P]ATP and [γ-32P]ATP were obtained from Perkin Elmer Inc. (Waltham, MA, USA). T4 DNA ligase, T4 PNK (polynucleotide kinase), RNase T1, RNase S1, and restriction endonucleases were obtained from Thermo Scientific (Pittsburgh, PA, USA). Phusion high-fidelity DNA polymerase was purchased from New England Biolabs (Ipswich, MA, USA). Ni2+-NTA (nitrilotriacetic acid) Superflow was purchased from Qiagen Inc. (Germany). Pyrophosphatase (PPiase) was obtained from Roche Applied Science (China). The dNTP mixture was purchased from Takara (Japan). Oligonucleotide primers were synthesized by Biosune (China). Escherichia coli Rosetta (DE3) cells were purchased from Stratagene (Santa Clara, CA, USA).

tRNA gene cloning and transcription

The mitochondrial genomes of humans and cows contain only one gene encoding mtRNAThr with a UGU anticodon. Construct pTrc99b-T7-hmtRNAThr was obtained as described previously (29). The DNA sequence of the T7 promoter and the bovine mitochondrial tRNAThr (bmtRNAThr) gene were obtained by ligating six chemically synthesized DNA fragments together, and then ligating the product into plasmid pTrc99b (pre-cleaved with EcoRI/BamHI) to construct pTrc99b-T7-bmtRNAThr. The DNA fragment (TranszymeTrp) containing the DNA sequence of the T7 promoter, hammerhead ribozyme, and the hmtRNATrp gene was cloned from eight chemically synthesized DNA fragments using the same protocol as that for bmtRNAThr (30). Gene mutagenesis was performed according to the protocol provided with the KOD-plus mutagenesis kit. Transcripts were obtained using the T7 RNA polymerase run-off procedure, as described previously (31). The tRNA concentration was determined by UV absorbance at 260 nm and the extinction coefficient was calculated from the sequence of each tRNA. The hmtRNAThr transcript has an excellent amino acid accepting capacity (∼1400 pmol/A260), suggesting that it is correctly folded without modification.

Gene cloning, mutagenesis, expression, and protein purification

The yeast Sua5 and Qri7 coding sequences were amplified from genomic DNA obtained from yeast cells and cloned into vector pET28a(+) with an N-terminal His6-tag and named pET28a(+)-Sua5 and pET28a(+)-Qri7. pET28a(+)-Qri7 was used as the template to construct N-terminal truncation mutant pET28a(+)-Qri7-ΔN29, which encoded the mature form of ScQri7. pET28a(+)-TARS2-ΔN19 was obtained as detailed previously (29). The WARS2 coding sequence was amplified from cDNA, obtained by reverse-transcription polymerase chain reaction (RT-PCR) from total RNA obtained from HEK293T cells, and cloned into pET28a(+), with an N-terminal His6-tag, between EcoRI and XhoI sites. The recombinant plasmid was named pET28a(+)-WARS2. pET28a(+)-WARS2 was used as the template to construct pET28a(+)-WARS2-ΔN18 with the sequence encoding the mitochondrial targeting signal deleted (32). The ScSua5, ScQri7-ΔN29, hmThrRS-ΔN19, and hmTrpRS-ΔN18 genes were expressed in E. coli Rosetta (DE3) cells. Protein overexpression was induced at an A600 ≈ 0.6 with 200 μM isopropyl β-d-1-thiogalactopyranoside (IPTG) at 20 °C overnight. Purification of the ScSua5, ScQri7-ΔN29, hmThrRS-ΔN19 and hmTrpRS-ΔN18 proteins was performed as described previously (29). The protein concentrations were determined by the A280 of the enzyme solution.

Aminoacylation assay

Aminoacylation kinetics were performed at 37°C in a reaction mixture containing 60 mM Tris–HCl (pH 7.5), 10 mM MgCl2, 5 mM dithiothreitol (DTT), 0.1 mg/ml bovine serum albumin (BSA), 2.5 mM ATP, 100 μM [14C]Thr, (0.5–10) μM wild-type hmtRNAThr or its variants, and 200 nM hmThrRS, as described previously (33). Aminoacylation of hmtRNATrp with [14C]Trp was carried out at 37°C in a reaction mixture containing 50 mM Tris–HCl (pH 8.0), 10 mM MgCl2, 1 mM KF, 4 mM ATP, 100 μM [14C]Trp, 5 μM hmtRNATrp or its mutants and 50 nM hmTrpRS.

[32P]-labeling of mtRNAThr and preparation of Ser-[32P]mtRNAThr

An editing-defective ThrRS was used to mischarge mtRNAThrs with Ser to obtain Ser-mtRNAThr (22). We labeled the mtRNAThrs and their variants with α-32P-ATP to obtain 32P labeled tRNAs and aminoacyl-tRNAs, as described previously (29).

tRNA secondary structure probing

tRNA transcripts were labeled with 10 pmol of γ-32P-ATP and T4 PNK enzyme (NEB). Labeled tRNAs were purified on a 12% PAGE gel containing 8 M urea and eluted overnight at 4°C in 600 μl of elution buffer (0.2 M NaAc, pH 5.2). The labeled tRNAs were then precipitated using ethanol, and the pellets were dissolved in 40 μl of 5 mM MgCl2 buffer. Before use, each tRNA transcript was folded by incubation at 80°C for 2 min and then chilled on ice for 5 min.

Structure probing assays were performed as described previously (11,34). In native conditions, labeled tRNAs were digested with RNase S1 at 37°C for 5 min in 5 μl of 10 mM Tris–HCl (pH 7.0), 10 mM MgCl2, 50 mM KCl, and RNase S1 (2.5 or 5 units) (35,36). Under denaturing conditions, the labeled tRNAs were digested with RNase T1 (0.05 units) in 5 μl of 10 mM sodium citrate (pH 5.0), 1 mM EDTA, and 3.5 M urea at 37°C for 5 min. Alkaline ladders were obtained by digestion of the labeled transcript by 5 μl of 80 mM Na2CO3/NaHCO3 (pH 9.0) for 5 min at 80°C. All the reactions were stopped by adding 5 μl of stop mixture containing 0.6 M sodium acetate (pH 5.0) and 3 mM EDTA. Cleaved fragments were analyzed on 15% denaturing PAGE containing 8 M urea. Electrophoresis was carried out at 4°C and constant 30 W (∼1500 V) for 4.5 h. Gels were dried for 2 h at 80°C using a Gel Dryer (Bio-Rad), and detected using a phosphorimager (Fujifilm).

AMP formation assay

The AMP formation assay was performed using thin-layer chromatography (TLC) at 37°C in a reaction mixture containing 60 mM Tris–HCl (pH 7.5), 10 mM MgCl2, 5 mM DTT, 0.1 mg/ml BSA, 10 U/ml PPiase (Roche), 40 mM Ser, 3 mM [α-32P]ATP and 2 μM hmThrRS in the presence of hmtRNAThr or its variants, as detailed previously (29).

Post-transfer editing assay

Post-transfer editing of wild-type pre-formed Ser-[32P]hmtRNAThr or Ser-[32P]bmtRNAThr and its variants by hmThrRS was performed as described previously (29).

Biolayer interferometry

The dissociation constant (Kd) was obtained through biolayer interferometry using an Octet RED system (ForteBio). Samples or buffer were dispensed into 96-well plates at 200 μl per well. The operating temperature was maintained at 25°C. Proteins were diluted into kinetic buffer [50 mM potassium phosphate (pH 5.5), 50 mM MgCl2, and 0.002% Tween-20] and immobilized on anti-Ni–NTA sensor tips. The hmtRNAThr or its variants were diluted with the same buffer into a range of different concentrations. The Kd values were obtained by fitting the processed data with the 1:1 model in the Octet analysis software with R2 > 0.99 and are shown as mean ± error of fit.

In vitro assay for t6A modification of hmtRNAThr

The reaction was performed at 30°C in a reaction mixture containing 50 mM Tris–HCl (pH 8.0), 200 mM NaCl, 5 mM MnCl2, 1 mM NaHCO3, 2 mM DTT, 2.5 mM ATP, 100 μM [14C]Thr, 5 μM hmtRNAThr or its variants and 2 μM ScSua5 and ScQri7. Samples of the reaction mixture were removed at specific time points, quenched on Whatman filter pads, and equilibrated with 5% trichloroacetic acid. The pads were washed three times for 10 min each with 5% trichloroacetic acid and then with 100% ethanol. The pads were then dried and quantified using a scintillation counter (Beckman Coulter).

RESULTS

Screening of pivotal sites among pathogenic mutations of hmtRNAThr based on potential structural alterations and aminoacylation defects

To date, six mutations have been reported to be pathogenic in the hmtRNAThr gene [see Mamit-tRNA database (http://mamit-trna.u-strasbg.fr/Summary.asp)]: G30A, A38G, A39G, G70A, A71G, and U60 deletion (ΔU60) (Figure 1A). The G30A, A38G, and A39G mutations are located in the anticodon stem or loop. The G70A and A71G mutations are in the amino acid-accepting stem. The ΔU60 mutation is in the T loop. However, their effects on the structure and functions of hmtRNAThr are poorly understood.

To investigate the effect of the above mutations on the structure of hmtRNAThr, we transcribed hmtRNAThr and its six single-point mutants, and analyzed them by PAGE under native and denaturing conditions. Wild-type hmtRNAThr and all mutants including one-nucleotide difference (ΔU60) migrated with the same rates on denaturing urea-PAGE (Figure 1B). On native PAGE, hmtRNAThr-G30A and -G70A migrated slightly but obviously slower compared with the wild-type hmtRNAThr (Figure 1B). These results suggested that the G30A and G70A mutations changed the conformation of hmtRNAThr. The structural alteration could be also detected using the melting temperature (Tm) value of tRNA. We compared the Tm value of wild-type hmtRNAThr and its various mutants, as shown in Figure 1C. Compared with the Tm value of wild-type hmtRNAThr, those of hmtRNAThr-G30A, -G70A and -A71G decreased by 23.5°C, 11°C and 5°C, respectively; however that of hmtRNAThr-A38G increased by 4.5°C; and those of hmtRNAThr-ΔU60 and -A39G showed no obvious change. The data showed that the G30A and G70A mutations decreased the stability of hmtRNAThr markedly, and correlated with the migration assay on native PAGE. The Tm value of hmtRNAThr-A38G mutant indicated a more rigid conformation.

The central role of tRNA is the generation of aminoacyl-tRNA, catalyzed by its cognate aminoacyl-tRNA synthetase, to supply the materials of protein biosynthesis. To understand the effect of these mutations on the charging capacity of hmtRNAThr, we performed aminoacylation of these hmtRNAThrs by hmThrRS. Four mutations (A39G, ΔU60, G70A and A71G) had little effect, whereas the G30A and A38G mutations caused an obvious decrease in the charging capacity of hmtRNAThr (Figure 1D). Further aminoacylation kinetics of hmThrRS for the wild-type hmtRNAThr, hmtRNAThr-G30A and -A38G showed that the kcat values for both hmtRNAThr-G30A and -A38G were decreased, while their Km values were increased by ∼3-fold compared with that of hmtRNAThr. In addition, the catalytic efficiencies (kcat/Km) of hmThrRS for hmtRNAThr-G30A and -A38G were only 14% and 21% of that for hmtRNAThr, respectively (Table 1).

Table 1.

Aminoacylation kinetics of hmThrRS for various hmtRNAThr mutants

tRNA	k cat (s−1)	K m (mM)	k cat/Km (×10−3)	Relative kcat/Km (%)
hmtRNAThr	0.09 ± 0.01	1.05 ± 0.01	86	100
-G30A	0.04 ± 0.003	3.37 ± 0.09	12	14
-C40A	0.06 ± 0.004	4.85 ± 0.3	12	14
-C40U	0.08 ± 0.005	3.65 ± 0.2	22	26
-C40G	0.03 ± 0.001	2.95 ± 0.4	10	12
-G30A/A29G	0.11 ± 0.03	1.40 ± 0.03	95	110
-G30A/C40U	0.13 ± 0.05	1.36 ± 0.03	96	111
-A38G	0.05 ± 0.007	2.81 ± 0.2	18	21
-A38C	0.07 ± 0.005	2.24 ± 0.03	31	36
-A38U	0.08 ± 0.005	3.92 ± 0.1	20	23

The results represent the average of at least two independent trials with the standard deviations indicated.

hmtRNAThr-G30A displayed alterations in shifting in the native gel and amino acid charging, and had a decreased Tm value. The A38G mutation, located in a critical loop region, reduced amino acid charging and increased the Tm value (in contrast to other mutations). The other four point mutations slightly influenced shifting in the native gel and the Tm value and had no or negligible effects on amino acid charging, suggesting they likely affected mitochondrial translation via different mechanisms in downstream steps (such as elongation factor binding, ribosome binding, or decoding). Therefore, we selected the G30A and A38G mutations to study the potential effects of mutations on the structure and functions of hmtRNAThr.

The G30–C40 base pair is essential for maintaining the anticodon stem structure of hmtRNAThr because of non-Watson–Crick base pair A29/C41

Interestingly, the middle of the anticodon stem of hmtRNAThr transcript has a non-Watson–Crick base pair, A29/C41 (Figure 2A). The severe effect of the G30A mutation on the local structure and function of hmtRNAThr suggested that, with a natural non-Watson–Crick A29/C41 base pair in the middle, the G30–C40 base pair in the stem should be crucial to maintain the structure of the anticodon stem and loop. Therefore, we constructed C40A, C40G, and C40U mutations to disrupt the G30–C40 Watson–Crick base pair; G30A/C40U to form A30–U40, a new Watson–Crick base pair; G30A/A29G with the G29-C41 Watson–Crick base pair, and A30–C40 non-Watson–Crick base pair in the context of the wild-type hmtRNAThr transcript (Figure 2A). Electrophoretic mobility results showed that the hmtRNAThr-C40A, -C40G, and -C40U mutants with a non-Watson–Crick base pair migrated more slowly than the wild-type hmtRNAThr and as slowly as hmtRNAThr-G30A (Supplementary Figure S1). However, the mutants hmtRNAThr-G30A/C40U with Watson–Crick base pair A30-U40 and hmtRNAThr-G30A/A29G with one Watson–Crick base pair G29-C41 and one non-Watson–Crick base pair A30-C40 migrated as fast as the wild-type hmtRNAThr (Supplementary Figure S1). Further, analysis of the Tm values showed that those of the hmtRNAThr-G30A, -C40A, -C40G and -C40U mutants were decreased markedly compared with that of the wild-type hmtRNAThr (Figure 2B). However, the Tm values of the hmtRNAThr-G30A/C40U and -G30A/A29G mutants were similar to that of the wild-type hmtRNAThr (Figure 2B). The data from the hmtRNAThr-G30A/A29G mutant suggested that the structurally detrimental effect of the G30A mutation was caused by the A29/C41 non-Watson–Crick base pair in hmtRNAThr. The results showed that at least one Watson–Crick base pair in the third and fourth base pairs of the anticodon stem is necessary to maintain the conformation of hmtRNAThr.

Figure 2.

G30-C40 base pair is essential for hmtRNAThr stability. (A) Anticodon loop and stem structure of hmtRNAThr and its mutations. (B) Tm values of hmtRNAThr mutants.

A nuclease probing assay was then carried out with unpaired G residues-specific RNase T1 and single-stranded specific RNase S1 to reveal the effect of the G30A mutation on the structure of hmtRNAThr. The results showed that the main RNase S1 cleavage sites were at positions 34 and 35 in the anticodon loop of the wild-type hmtRNAThr (Figure 3A and B). However, the RNase S1 cleavage sites for hmtRNAThr-G30A obviously shifted to positions 30 and 31 in the anticodon stem, suggesting that the G30A mutation disrupted the base pair between U31 and A39, and thus disturbed the stem structure (Figure 3A and B). The main RNase S1 cleavage site in hmtRNAThr-G30A/C40U shifted again to positions 34 and 35 (Figure 3A and B), indicating that the stem structure was rescued by an additional C40U mutation. A minor portion of hmtRNAThr-G30A/C40U was also digested at positions 30 and 31 (Figure 3A and B), suggesting that the weaker Watson–Crick base pair A30–U40 is less efficient in maintaining the stem structure compared with the stronger Watson–Crick base pair G30–C40 in hmtRNAThr. Our data showed that, at least from viewpoint of structure, the fourth base pair of the anticodon stem should be a Watson–Crick base pair (either G–C or A–U), even G–U was not tolerated (as shown by the decreased Tm of the hmtRNAThr-C40U mutant). Once the fourth base pair was disrupted, the local structure was altered; however, this could be compensated for by restoring the third Watson–Crick base pair of the stem (as illustrated by the hmtRNAThr-G30A/A29G mutant).

Figure 3.

Enzymatic probing analysis of hmtRNAThr, -G30A and -G30A/C40U. (A) Probing was performed using various RNase S1 concentrations. Lane C, control; lane G, ladder digested by RNase T1 under denaturing conditions; lane OH−, alkaline digestion. (B) Structure analysis by enzymatic probing. The red stars indicate the main RNase S1 cleavage sites.

The G30A mutation influences tRNA binding and aminoacylation by hmThrRS

The above data clearly showed the critical role of G30–C40 in maintaining the local structure of the anticodon stem of hmtRNAThr. Whether the structural alteration in the anticodon stem influences its interaction with hmThrRS was not clear; therefore, we used biolayer interferometry (BLI) and immobilized hmThrRS on a sensor chip to detect the affinity between hmThrRS and hmtRNAThr. The results showed that the Kd value of hmThrRS for hmtRNAThr-G30A was 1.5-fold weaker than that for wild-type hmtRNAThr (Table 2). Similarly, the Kd values for both hmtRNAThr-C40A and -C40G were increased, showing that the affinity between hmThrRS and hmtRNAThr was weakened upon disrupting the G30–C40 base pair as hmtRNAThr-G30A mutant (Table 2). The Kd value for hmtRNAThr-C40U was slightly higher, indicating a weaker affinity to hmThrRS (Table 2). The affinities for hmtRNAThr-G30A/C40U and hmtRNAThr-G30A/A29G were comparable to that of the wild-type hmtRNAThr (Table 2). The above data suggested that the G30A mutation weakens the interaction between hmThrRS with hmtRNAThr by altering the conformation of the anticodon stem.

Table 2.

Binding affinities of hmThrRS or ScQri7 with various hmtRNAThr mutants

hmThrRS	tRNA	ScQri7
K d (μM)		K d (nM)
1.8 ± 0.3	hmtRNAThr	26.9 ± 1.2
3.1 ± 0.4	-G30A	4.1 ± 0.2
4.3 ± 0.3	-C40A	11.8 ± 0.5
2.4 ± 0.1	-C40U	11.0 ± 0.3
3.6 ± 0.2	-C40G	3.3 ± 0.1
1.6 ± 0.2	-G30A/C40U	25.5 ± 1.0
1.4 ± 0.4	-G30A/A29G	29.1 ± 2.1
2.3 ± 0.1	-A38G	31.2 ± 1.9
3.7 ± 0.2	-A38C	51.7 ± 1.6
4.3 ± 0.4	-A38U	47.4 ± 1.1

The results represent the average of at least two independent trials with the standard deviations indicated.

We next assayed and compared the aminoacylation activity of hmThrRS for hmtRNAThr and its mutants. The results showed that, among the mutants of C40, the tRNA charging level of hmtRNAThr-C40G was the most significantly decreased (Figure 4A), while those of the hmtRNAThr-C40A and -C40U mutants were comparable to that of hmtRNAThr-G30A, indicating that G40 was most unfavorable at this site. However, the charging levels of hmtRNAThr-G30A/C40U and -G30A/A29G were rescued to that of the wild-type hmtRNAThr (Figure 4A). These results suggested that disruption of the G30–C40 base pair caused the reduction in the aminoacylation level (Figure 4A), which otherwise could be restored by re-forming the G29–C41 base pair. Further analysis of the aminoacylation kinetics showed that the Km values of hmThrRS for all hmtRNAThr-C40A, -C40U, and -C40G mutants were increased by 3–5-fold, with a decreased kcat. The catalytic efficiencies of hmThrRS for the three mutants were only 14%, 26%, and 12% of that for hmtRNAThr, respectively (Table 1). However, both the Km and kcat values for the double-site mutants, hmtRNAThr-G30A/C40U and -G30A/A29G, were similar to those of the wild-type hmtRNAThr (Table 1); thus, the catalytic efficiency of hmThrRS was not affected by the double-site mutations. Altogether, the above data revealed that the G30A mutation of hmtRNAThr decreases the affinity between hmThrRS and hmtRNAThr by disrupting the G30–C40 Watson–Crick base pair of hmtRNAThr, which could be compensated for by a Watson–Crick base pair G29–C41. In the third and fourth base pairs of the anticodon stem of hmtRNAThr, at least one Watson–Crick base pair is necessary to maintain the structure of the anticodon stem and loop. It also suggested that the conformation, but not the nucleotide sequence, in the anticodon stem determines the proper recognition by hmThrRS, because hmtRNAThr-G30A/C40U and -G30A/A29G are well qualified as hmtRNAThr.

Figure 4.

Aminoacylation and editing properties of hmThrRS for hmtRNAThr-G30A. (A) Aminoacylation of hmtRNAThr (), -G30A (), -C40A (), -C40U (), -C40G (), -G30A/C40U () and -G30A/A29G () catalyzed by hmThrRS, without enzyme addition as a control (). (B) AMP formation by hmThrRS in the presence of hmtRNAThr () or its mutants hmtRNAThr-G30A (), -C40A (), -C40U (), -C40G (), without tRNA addition as a control (). (C) AMP formation by hmThrRS in the presence of hmtRNAThr-G30A/C40U () or -G30A/A29G () mutant, without tRNA addition as a control (). (D) Post-transfer editing of hmThrRS for mischarged hmtRNAThr () or its mutants hmtRNAThr-G30A (), -G30A/C40U () and -G30A/A29G (), without enzyme addition as a control (). Error bars indicate the standard deviations.

Disrupting the G30–C40 Watson–Crick pair induces a decrease in editing by hmThrRS

hmThrRS contains an editing domain to deacylate mis-charged tRNA and ensure the fidelity of protein synthesis. As stated previously, hmThrRS clears mis-acylated Ser-tRNAThr via its post-transfer editing function, whose abolition resulted in generation of Ser-tRNAThr (29). In addition, tRNA functions as a stimulator to promote tRNA-dependent pre-transfer editing. To investigate whether the G30A mutation affected the editing function of hmThrRS, we performed Ser-included AMP formation assays (measuring the total editing function, including tRNA-independent, tRNA-dependent pre-transfer editing, and post-transfer editing) in the absence or presence of hmtRNAThr or the hmtRNAThr-G30A mutant. The results showed that the AMP formation in the presence of hmtRNAThr-G30A decreased significantly compared with that of hmtRNAThr (Figure 4B). The observed AMP formation rate constant (kobs) in the presence of hmtRNAThr-G30A was only 43% of that in the presence of hmtRNAThr (Table 3). To explore whether the editing-impairment was due to structural alteration of the anticodon stem, we performed AMP formation assays in the presence of hmtRNAThr-C40A, -C40G, and -C40U. The results showed that the kobs values decreased to different extents (Figure 4B and Table 3). However, the double-site mutants, hmtRNAThr-G30A/C40U and -G30A/A29G, did not affect the total editing activity of hmThrRS (Figure 4C and Table 3), again suggesting the importance of at least one Watson–Crick base pair at 30–40 and 29–41 for the correct structure of hmtRNAThr and the editing function of hmThrRS. In addition, the results of a deacylation assay showed that the post-transfer editing activity of hmThrRS for hmtRNAThr-G30A was slightly decreased compared with that of the wild-type hmtRNAThr, while the double-site mutations G30A/C40U and G30A/A29G in hmtRNAThr did not affected the post-transfer editing activity of hmThrRS (Figure 4D).

Table 3.

Observed rate constants of AMP formation by hmThrRS with Ser in the presence of wild-type hmtRNAThr or its mutants

tRNA	k obs × 10−3(s−1)	Relative kobs (%)
hmtRNAThr	33.3 ± 0.6	100
-G30A	14.5 ± 0.4	43
-C40A	20.1 ± 0.1	60
-C40U	26.3 ± 0.6	79
-C40G	15.1 ± 0.3	45
-G30A/A29G	34.2 ± 0.6	103
-G30A/C40U	36.3 ± 1.7	109
-A38G	16.8 ± 0.6	50
-A38C	20.6 ± 0.3	61
-A38U	19.9 ± 0.3	60

The results represent the average of two independent trials with the standard deviations indicated.

G30A mutation in bmtRNAThr has little effect on its functions because of its natural G29–C41 base pair

Interestingly, bmtRNAThr is similar to hmtRNAThr in sequence and structure in the anticodon stem, except for a Watson–Crick base pair, G29–C41 (Figure 5A). We constructed the G30A mutation in bmtRNAThr and measured the accepting activity of bmtRNAThr-G30A. The data showed that the bmtRNAThr-G30A mutant had a comparable aminoacylation level to bmtRNAThr (Figure 5B), which was different from the obvious decreased activity of hmtRNAThr-G30A (Figure 4A). To reveal the potential importance of the G29–C41 base pair of mtRNAThr, we further constructed bmtRNAThr-G29A and -G30A/G29A mutants to mimic hmtRNAThr and the hmtRNAThr-G30A mutant (Figure 5A). The results showed that bmtRNAThr-G30A/G29A lost its accepting activity (Figure 5B), which was consistent with that of the G30A mutation in hmtRNAThr. Meanwhile, bmtRNAThr-G29A had little effect on the charging activity (Figure 5B). Furthermore, we performed an AMP formation assay to detect whether these bmtRNAThr mutations affected the editing properties of hmThrRS. The data showed that among the mutants, only the double-site mutant bmtRNAThr-G30A/G29A decreased the editing activity of hmThrRS (Figure 5C and D). Our results further indicated that at least one Watson–Crick base pair in the third and fourth base pairs of the anticodon stem is crucial for aminoacylation and the editing activities for mitochondrial tRNAThr; if these two base pairs were non-Watson–Crick base pairs, the functions of mtRNAThr would be disrupted.

Figure 5.

Disrupting the G30-C40 base pair of bmtRNAThr does not affect the aminoacylation and editing properties of hmThrRS. (A) Anticodon stem and loop structure of hmtRNAThr, bmtRNAThr, and its mutants. (B) Aminoacylation of bmtRNAThr () and its mutants bmtRNAThr-G30A (), -G29A () and -G30A/G29A () catalyzed by hmThrRS, without enzyme addition as a control (). (C) AMP formation of hmThrRS in the presence of bmtRNAThr or its mutants. (D) Graphic representations of AMP formation in (C), bmtRNAThr (), -G30A (), -G29A () and -G30A/G29A (). Error bars indicate the standard deviations.

The G30A mutation in hmtRNATrp causes structure fragility, which arises from a natural A29/C41 non-Watson–Crick base pair

By analyzing the secondary structure of all human mitochondrial tRNAs, we found that besides hmtRNAThr, only hmtRNATrp has a natural A29/C41 non-Watson–Crick base pair (Figure 6A). The electrophoretic mobility results showed that hmtRNATrp-G30A also migrated slower than hmtRNATrp in a native gel (Supplementary Figure S2), which could be rescued by restoring the G29–C41 base pair, because hmtRNATrp-G30A/A29G migrated as fast as hmtRNATrp (Supplementary Figure S2). Furthermore, the Tm value of the hmtRNATrp-G30A mutant was markedly decreased and similarly rescued by a further A29G mutation, as shown for the double-site mutant hmtRNATrp-G30A/A29G (Figure 6B). The above results clearly showed that the natural A29/C41 non-Watson–Crick base pair made the structure of hmtRNATrp sensitive to the G30A mutation, similar to the case in hmtRNAThr. We also investigated the effect of the G30A mutation on the Trp-accepting capacity mediated by the purified recombinant human mitochondrial tryptophanyl-tRNA synthetase (hmTrpRS) (Figure 6C). The G30A mutation did not affect the aminoacylation of hmtRNATrp (Figure 6D), implying that hmtRNATrp recognition by hmTrpRS was not reliant on the anticodon stem. The above data confirmed that the structural integrity of hmtRNATrp was also dependent on the G30–C40 base pair because of a natural A29/C41 non-Watson–Crick base pair. Indeed, the m. 5540 G>A (hmtRNATrp G30A) mutation is associated with a progressive encephalopathy and cytochrome c oxidase deficiency (37).

Figure 6.

The G30A mutation causes structural flexibility of hmtRNATrp. (A) Cloverleaf structure of hmtRNATrp and its mutants. (B) Temperature curves of hmtRNATrp and its mutants measured at 260 nm. (C) Analysis of purified mature hmTrpRS. (D) Aminoacylation levels of hmtRNATrp () and -G30A () catalyzed by hmTrpRS, without enzyme addition as a control (). Error bars indicate the standard deviations.

A38 is an important element for hmtRNAThr charging, the interaction of hmThrRS and hmtRNAThr, and the editing function of hmThrRS

Besides the G30A mutation, the A38G mutation also impaired the tRNA charging capacity in the initial screening assays (Figure 1C), implying that A38, in the anticodon loop, is a potentially key element in aminoacylation. We changed A38 to C or U, obtaining hmtRNAThr-A38C or -A38U. A time-course curve of the aminoacylation reaction showed that the Thr-accepting ability of all the substitution mutants of hmtRNAThr-A38C, -A38U, and -A38G obviously decreased compared with that of hmtRNAThr (Figure 7A). Furthermore, the aminoacylation kinetic parameters showed that the Km values of hmThrRS for the hmtRNAThr-A38C, -A38U and -A38G mutants increased by about 3-fold compared with that of hmtRNAThr, while the kcat values for the three mutants decreased slightly (Table 1). The catalytic efficiencies for hmtRNAThr-A38C, -A38U and -A38G were 36%, 23%, and 21% of that of hmtRNAThr, respectively (Table 1). Furthermore, the affinities between hmThrRS and hmtRNAThr or hmtRNAThr-A38C, -A38U and -A38G mutants were also measured and compared using BLI. The data showed that the Kd value of hmThrRS for hmtRNAThr-A38G was comparable to that for hmtRNAThr, while the Kd values for hmtRNAThr-A38C and -A38U increased by ∼2-fold (Table 2). In addition, we performed the AMP formation assays to investigate the role of A38 in the editing function of hmThrRS. The results showed that the AMP formation with hmtRNAThr-A38G decreased markedly compared with that of hmtRNAThr (Figure 7B). The kobs of hmThrRS for hmtRNAThr-A38G were only 50% of that for hmtRNAThr (Table 3). Similarly, both hmtRNAThr-A38C and -A38U resulted in a reduction in AMP formation (Figure 7B, Table 3). Moreover, the results from the deacylation assay of Ser-tRNAThr showed that among the three mutants, hmtRNAThr-A38G led to a marked decrease in the post-transfer editing activity of hmThrRS (Figure 7C). The above data indicated that A38 plays an important role in hmtRNAThr/hmThrRS interactions, tRNA recognition, and editing properties of hmThrRS.

Figure 7.

A38 is a key element for tRNA charging and editing of hmThrRS. (A) Aminoacylation levels of hmtRNAThr () and its mutants hmtRNAThr-A38G (), -A38C () and -A38U () catalyzed by hmThrRS, without enzyme addition as a control (). (B) Quantification of AMP formation in the presence of hmtRNAThr (), -A38G (), -A38C () or -A38U (), without tRNA addition as a control (). (C) Post-transfer editing of hmThrRS for mischarged hmtRNAThr (), -A38G (), -A38C () and -A38U (), without enzyme addition as a control(). Error bars indicate the standard deviations.

G30A and A38G mutations induce a marked decrease in t6A modification in hmtRNAThr

tRNA modification is a crucial step in the tRNA life cycle, and contributes to its structure and function. Defects in tRNA modification frequently result in mitochondrial diseases (10,38). Whether the pathogenic mutations G30A and A38G of hmtRNAThr affected modifications of hmtRNAThr was unknown. Although the complete post-transcriptional modification pattern of hmtRNAThr has not been reported, post-transcriptional modifications in bmtRNAThr have been determined, including m1A9, m3C32, t6A37, Ψ67 and m5C72 (39). We proposed that hmtRNAThr probably had a similar modification pattern to bmtRNAThr, except for Ψ67, because A67 is present in hmtRNAThr. The t6A modification at position 37 aroused our interest for the following reasons. First, G30A and A38G are close to A37. Second, the t6A37 modification is essential to maintain the correct conformation of the anticodon loop (24,40). Third, in Saccharomyces cerevisiae, the enzymes ScSua5 and ScQri7 catalyze the t6A modification at position 37 of tRNAs (27). Therefore, whether the t6A modification in human tRNAs can be catalyzed by these two enzymes and whether this modification exists in hmtRNAThr need to be investigated. We purified ScSua5 and ScQri7 and showed that ScSua5/ScQri7 could catalyze the t6A modification of hmtRNAThr efficiently (Figure 8A). To confirm that the t6A modification occurs at position A37, we mutated A37 to C, G and U, respectively. Under the same conditions, t6A could not be detected in the three substitutions of A37 (Figure 8A), indicating the t6A37 also exists in hmtRNAThr, as in bmtRNAThr. However, the t6A modification level of hmtRNAThr-G30A was significantly decreased compared with that of hmtRNAThr (Figure 8B). The observed rate constant (kobs) of the t6A modification in the hmtRNAThr-G30A was only 41% of that of hmtRNAThr (Table 4). As described above, three substitutions of another base for C40 in the anticodon stem decreased the thermal stability of these tRNAs (Figure 2C); and the t6A modifications of hmtRNAThr-C40A, -C40U and -C40G were also decreased (Figure 8B and Table 4). However, the two double-site mutants, hmtRNAThr-G30A/C40U and -G30A/A29G, had similar t6A modification levels compared with hmtRNAThr, because of the presence of one Watson–Crick base pair in the third and fourth base pairs in the anticodon stem (Figure 8B and Table 4). ScSua5/ScQri7 decreased the modification of t6A in hmtRNAThr-G30A with lowest Tm. The above data showed that the t6A modification occurs at position 37 of hmtRNAThr and is associated with the stability of its anticodon stem. In addition, we found that the t6A modification has a negligible (or a slightly inhibitory) effect on its aminoacylation (Supplementary Figure S3).

Figure 8.

Both the G30A and A38G mutations cause decreased t6A37 modification. (A) t6A modification of hmtRNAThr (), -A37G (), -A37C () and -A37U () mutants, without enzyme addition as a control (). (B) t6A biosynthesis of hmtRNAThr () and its mutants hmtRNAThr-G30A (), -C40A (), -C40U (), -C40G (), -G30A/C40U () and -G30A/A29G () catalyzed by ScSua5 and ScQri7, without enzyme addition as a control (). (C) t6A modification levels of hmtRNAThr (), -A38G (), -A38C () and -A38U () mutants, without enzyme addition as a control (). Error bars indicate the standard deviations.

Table 4.

Observed rate constants of t6A modification in wild-type hmtRNAThr or its mutants by ScSua5/ScQri7

tRNA	k obs × 10−3(min−1)	Relative kobs (%)
hmtRNAThr	34.6 ± 0.1	100
-G30A	14.2 ± 1.9	41
-C40A	16.7 ± 0.7	48
-C40U	20.6 ± 1.3	60
-C40G	5.7 ± 0.3	16
-G30A/A29G	38.6 ± 1.9	111
-G30A/C40U	29.5 ± 1.7	85

The results represent the average of two independent trials with the standard deviations indicated.

We also measured the t6A modification level of hmtRNAThr-A38G. The data showed that the A38G mutation abolished the t6A modification (Figure 8C). Further, we found that both the A38C and A38U mutations abolished the t6A modification (Figure 8C). These results revealed that, on the one hand, A38 is a critical determinant for the t6A modification in human mitochondrial tRNAs, at least for hmtRNAThr; and on the other hand, abolition of the t6A37 modification of hmtRNAThr is also likely to contribute to the pathogenesis of the A38G mutation.

G30A and A38G mutations alter the interaction between ScQri7 and hmtRNAThr

In the t6A modification reaction, ScSua5 catalyzes the first step of TCA intermediate formation; ScQri7 catalyzes the second step, involving binding TCA and the tRNA substrate to form t6A (27). To explore whether the mutations affected the affinity between ScQri7 and hmtRNAThr, we measured the Kd values of ScQri7 to hmtRNAThr and its mutants using BLI. The Kd value of ScQri7 with hmtRNAThr was 26.9 ± 1.2 nM, and with hmtRNAThr-G30A was 4.1 ± 0.2 nM, indicating that the affinity of the enzyme for the hmtRNAThr-G30A mutant was 6-fold stronger than that to hmtRNAThr (Table 2). In addition, the Kd values of ScQri7 to all hmtRNAThr-C40 mutants, hmtRNAThr-C40A, -C40U and -C40G, were markedly decreased compared with that to hmtRNAThr (Table 2), indicating a stronger affinity between the enzyme and hmtRNAThr after alteration in structure of the anticodon stem. However, the Kd values of hmtRNAThr-G30A/C40U in which the fourth base pair in the stem is a Watson–Crick base pair and hmtRNAThr-G30A/A29G, in which the third is a Watson–Crick base pair, were similar to that of hmtRNAThr (Table 2), again indicating the critical role of an intact stem structure in the proper interaction between ScQri7 and hmtRNAThr. Moreover, only a moderately weakened affinity of ScQri7 to hmtRNAThr-A38G was determined (Table 2); while those to hmtRNAThr-A38C and -A38U were ∼2-fold weaker than that to hmtRNAThr (Table 2). Our data indicated that the rigid structure controlled by the third or fourth Watson–Crick base pair in the anticodon stem, and A38 in anticodon loop, are important elements for the interaction between ScQri7 and hmtRNAThr, and for the t6A37 modification.

DISCUSSION

We further expanded the sequences with a broader species spectrum from thirteen eukaryotes including primates, mammals, non-mammalian vertebrates, and invertebrate. Their evolutionary relationship was constructed based on divergence times calculated by Timetree (Supplementary Figure S4A) (41). Detailed analysis of the sequences of the anticodon stem showed that the third base pair is always a Watson–Crick pair in the mtRNAThrs from invertebrate, non-mammalian vertebrates, mammals, and primates, except for gorilla and human mtRNAThrs (Supplementary Figure S4B), suggesting that the gain of A29/C41 non-Watson–Crick base pair in gorilla and human mtRNAThrs is a later evolutionary event. The fourth base pair (G30–C40) is highly conserved in all mtRNAThrs (Supplementary Figure S4B). In addition, the second base pair is U28-A42 in some lower eukaryotes, which is different to the C28–G42 in hmtRNAThr. We found that the C28U/G42A mutation had no effect on hmtRNAThr migration, thermal stability, and aminoacylation (data not shown). Thus, we hypothesized that the A29/C41 mismatch is a newly evolved non-Watson–Crick pair in mitochondrial tRNAThrs from Gorilla gorilla and Homo sapiens, while a Watson–Crick base pair is present in those from other representative eukaryotes (this work) or in lower eukaryotes, such as Saccharomyces cerevisiae and Schizosaccharomyces pombe (42). All nucleotides in the anticodon stem, particularly the last three base pairs, have been suggested to be critical for a proper helical conformation during genetic code decoding (43). Therefore, it is reasonable that the introduction of the A29/C41 non-Watson–Crick base pair in hmtRNAThr highlights the essential role of the downstream strong G30–C40 base pair in regulating tRNA structure and functions. The Tm value of hmtRNAThr with A30–U40 was much larger than that of tRNA with the G30–U40 wobble base pair, suggesting that G30–U40 is less stable than A30–U40, at least in these positions. Consistently, both aminoacylation and editing activities at the presence of hmtRNAThr with G30-U40 were reduced, possibly because of lower stability and/or the altered pit of the anticodon stem helix with G30–U40. Mutations causing two mismatches at 29/41 and 30/40 positions would impair the structural integrity and have a direct influence on various functional steps during the tRNA life cycle. Consistently, the detrimental effect of the mismatch between bases 29 and 41 in hmtRNATrp makes it very sensitive to local mutations leading to a mismatch between 30 and 40 bases. In fact, the hmtRNATrp G30A mutation leads to syndromes with a progressive encephalopathy and cytochrome c oxidase deficiency (37), probably because of an altered tRNA structure in the anticodon stem. The presence of a Watson–Crick pair between bases 29 and 41 in mitochondrial tRNAThrs from other species suggested that these tRNAs are also tolerant to the single-point mutation disrupting the G29–C41 or G30–C40 base pairs, as revealed in bmtRNAThr. The evolutionary advantage and possible role of the newly acquired A29/C41 non-Watson–Crick base pair is not clear. One possibility is that the introduced A29/C41 non-Watson–Crick base pair might fine-tune the flexibility of the anticodon stem of mtRNAThrs from Gorillas and humans, and contribute to its mRNA decoding capacity on mitochondrial ribosomes.

Recently, the general sequence features near the anticodon stem and loop based on 400 elongator tRNAs from bacteria, eukaryotes and archaea have been summarized (43). hmtRNAThr exhibits both conservation and divergence from canonical tRNAs in the anticodon stem. Interesting, despite only Watson–Crick base pairs being tolerated between bases 29–41, hmtRNAThr has evolved the A29/C41 non-Watson–Crick pair, which subsequently makes the G30–C40 base pair extremely sensitive to non-Watson–Crick mutations (43). Indeed, once a Watson–Crick base pair is re-formed between bases 29 and 41, this sensitivity of the G30–C40 base pair is relieved. Lastly, the nucleotide at position 31 of tRNAs prefer A>C>G>>U (43); however, hmtRNAThr has U31, leading to A39, but not G39 or ψ39, starting the helical stem (43).

The anticodon stem of Saccharomyces cerevisiae mitochondrial tRNAThr (ScmtRNAThr) is an important element for the binding and catalysis of ScmThrRS to ScmtRNAThr (44). The identity nucleotides in tRNAThr from E. coli, S. cerevisiae, and T. thermophilus have been reported, including G1-C72, C2-G71, U3-A70, G34, U35, G36 and (or) U73 (45–49). Our previous study showed that G35 and U36 were the identity determinants in hmtRNAThr (29). Based on the data in the present work, we suggested that G30 is important in maintaining a proper conformation of anticodon stem and that A38 is a key element for tRNA aminoacylation and editing.

The t6A modification strengthens the codon-anticodon pairing and facilitates translational fidelity (24,25). The presence of a t6A modification in bmtRNAThr strongly suggested that it also exists in hmtRNAThr. Our results showed that hmtRNAThr could be t6A37-modified. The G30A mutation resulted in marked reduction in the t6A modification, which could be rescued by G30A/C40U or G30A/A29G mutations, implying the importance of the intact structure of the anticodon stem in t6A modification. hmtRNAThr (anticodon UGU) belongs to the intermediate group of decoding tRNAs because the codon box contains a weak A1–U36 pair in the first position (43). The t6A modification is essential to stabilize the weak codon-anticodon pair A1–U36 and avoid miscoding in the codon boxes (43). Defects of the t6A modification often lead to an increased frequency of frameshift events (26,50). Therefore, the G30A mutation in the anticodon stem of hmtRNAThr was likely to cause a decrease in the efficiency and/or fidelity of mitochondrial protein synthesis, which could explain the clinical symptoms of patients carrying the G30A mutation in hmtRNAThr, namely hearing loss, muscle weakness, and focal COX deficiency (17). Interestingly, A38 also functions as a critical nucleotide determinant in the t6A modification, at least for hmtRNAThr. The t6A modification contributes to the formation of canonical U-turn structure of the anticodon loop by preventing U33–A37 base pairing, which suggested that the A38G mutation in hmtRNAThr probably affected the structure of the anticodon loop (40). In addition, C32 and A38 tend to form a non-Watson–Crick pair during decoding (43). The A38G mutation would make the C32 and G38 form a strong Watson–Crick pair, which might also affect the structure of the anticodon loop, consistent with the increased Tm value of the A38G mutant. U36 and A37 bases have been suggested to be the identity determinants in t6A formation in Xenopus laevis oocyte cytoplasm (51). Despite A38 being a critical element in t6A formation in hmtRNAThr and tRNAs in Xenopus laevis oocyte cytoplasm (51), G38 is found in human and bovine mitochondrial t6A37-contained tRNAIles (4,39). Therefore, the systematic study of identity determinants of t6A37 formation in mitochondria should be further explored. The deficiency in aminoacylation and t6A modification of the A38G mutant implied low translational fidelity and efficiency in mitochondria, which might contribute to the symptoms of complex III and IV deficiency in tissues and sudden multisystem failure (18).

In summary, we propose a model of the potential molecular pathogenesis of the G30A and A38G mutations, which alter the anticodon stem and/or loop structure of hmtRNAThr. The two mutations result in decreased aminoacylation, editing, and t6A modification, which likely lead to low accuracy and efficiency of mitochondrial protein synthesis (Figure 9).

Figure 9.

Model for the potential pathogenic mechanism of G30A and A38G mutations in hmtRNAThr. In human mitochondria, the hmtRNAThr transcript was modified, including the t6A modification; the fully modified hmtRNAThr was charged with Thr by hmThrRS with the resultant Thr-hmtRNAThr transferred to the mitochondrial ribosome for protein synthesis. The G30A mutation decreased the aminoacylation and editing activities of hmThrRS and the t6A modification level by disrupting the crucial G30–C40 base pair and thus altering the helical conformation of hmtRNAThr. The A38G mutation decreased the aminoacylation and editing of hmThrRS and, importantly, abolished the synthesis of t6A37, possibly forming a strong C32–G38 base pair. Decreased aminoacylation, editing, and downregulated or abolished t6A37 formation, which is critical for stabilizing the codon-anticodon interaction, separately or collectively had detrimental effects on the speed, accuracy, and/or fidelity of the genetic code flow in human mitochondria. Other effects of the mutations on the tRNA functions (such as other modifications, interaction with the ribosome, etc.) may also contribute to the observed pathogenesis.

Click here for additional data file.