Human mitochondrial tRNA quality control in health and disease

Abstract

Mutations in human mitochondrial tRNA genes are associated with a number of multisystemic disorders. These single nucleotide substitutions in various domains of tRNA molecules may affect different steps of tRNA biogenesis. Often, the prominent decrease of aminoacylation and/or steady-state levels of affected mitochondrial tRNA have been demonstrated in patients' tissues and in cultured cells.

 

Similar effect has been observed for pathogenic mutations in nuclear genes encoding mitochondrial aminoacyl-tRNA-synthetases, while over-expression of mitochondrial aminoacyl-tRNA synthetases or elongation factor EF-Tu rescued mutated tRNAs from degradation.

In this review we summarize experimental data concerning the possible regulatory mechanisms governing mitochondrial tRNA steady-state levels, and propose a hypothesis based on the tRNA channelling principle. According to this hypothesis, interaction of mitochondrial tRNA with proteins ensures not only tRNA synthesis, maturation and function, but also protection from degradation. Mutations perturbing this interaction lead to decreased tRNA stability.

Keywords: :

• aminoacyl-tRNA-synthetases
• elongation factor Tu
• human diseases
• mitochondria
• tRNA
• tRNA channelling

Mitochondria are essential organelles that are present in virtually all eukaryotic cells and contain their own DNA and translation system. The human mitochondrial genome, a circular DNA molecule of 16 659 bp,¹ encodes 2 rRNAs, 22 tRNAs and 13 subunits of oxidative phosphorylation complexes, localized within the inner mitochondrial membrane and providing the lion share of the cellular ATP pool.² In order to synthesize these 13 polypeptides, the mitochondria should import from the cytoplasm more than 150 proteins needed for mitochondrial (mt) DNA replication and transcription, mt-tRNAs modification and aminoacylation, as well as ribosomal proteins and translation factors. The mitochondrial translation system uses only 22 mt-encoded tRNAs (mt-tRNAs), the minimal known set of tRNAs sufficient for decoding all the triplets, due to the non-universal mitochondrial genetic code.³

Exclusive importance of the proper functioning of mt-tRNAs is illustrated by an observation that mutations in these small molecules are hotspots for mitochondrial pathogenesis, including myopathies, neurodegenerative pathologies and multisystemic disorders caused by defects in mt oxidative phosphorylation. Mutations in mt-tRNA genes contribute to the etiology of more than half of disorders caused by mutations in mt-DNA, while tRNA genes comprise only 10% of the mt genome.^1,4

Mutations in mt-tRNA genes usually result in interference with proper synthesis of mitochondrial proteins and in respiratory chain deficiencies. These mutations may result in improper maturation of the mt-tRNA molecules, in reduced ability to be charged by the cognate amino acid or in inability to decode corresponding codons (reviewed in refs. 5 and 6). Frequently these mutations reduce mt-tRNA steady-state levels, implying that there is a strict quality control of tRNA in mitochondria. In other words, mitochondria should provide some means of specifically recognizing and scavenging abnormal tRNA molecules, at the same time protecting the normal tRNAs from degradation. The mechanism underlying this quality control of mt-tRNAs has not been thoroughly investigated. We assume that this process can be achieved by a channeling process similar to that taking place within the cytosol, suggesting that at every step from transcription and maturation to aminoacylation and association with the ribosome, mt-tRNA exists in a complex with proteins; otherwise it is subjected to degradation. Thus, we hypothesize that increased turnover of non-aminoacylated/non-channeled mt-tRNA molecules provides a universal mechanism for the quality control of the tRNA pool within the human mitochondria.

Channeling Theory

The channeling theory was proposed by Smith in 1975 and further refined by Deutscher and coauthors.⁷ According to this theory, tRNAs aminoacylated in the cytosol are directly transferred from the aminoacyl-tRNA synthetases (ARS) to the elongation factor, and then to the ribosomes, without dissociation into the cellular matrix.⁸ The theory was later expanded to the nuclear-cytosolic channeling. tRNAs encoded and synthesized in the nucleus, where they undergo aminoacylation, are allowed to leave the nucleus if they are properly charged.^9,10 The nuclear tRNA export receptor Los1p binds the aminoacylated tRNA cargo together with RanGTP,¹¹ and this complex is then transported through the nuclear pore complex (NPC). Scyl1 collects aminoacyl-tRNAs from Los1p on the cytoplasmic side of the NPC and channels them to elongation factor which further delivers it to the ribosome.¹²

In summary, the nuclear channeling is required for tRNA maturation and aminoacylation, shipping aminoacyl-tRNA from the nucleus to the cytosol and its final delivery to the ribosome. In the cytosol, channeling mediates the transfer of de-acylated tRNA from the ribosome directly to the aminoacyl-tRNA-synthetases (ARS), then the association of aminoacylated tRNAs with elongation factor and subsequent transfer of aminoacyl-tRNAs to the ribosomes. During all these stages tRNA stays protein-bound and thus is protected from degradation.

Transcription and Maturation of mt tRNA

Human mitochondrial genome is transcribed polycistronically from two heavy (H)-strand promoters and one light (L)-strand promoter.¹³ Following transcription, the 5′ and 3′ ends of the tRNA molecules are formed by endonucleolytic cleavage of the precursor polycistronic RNA by RNase P and tRNase Z, respectively. The CCA triplet needed at the 3′ end of all mature tRNAs is added by CCA-adding enzyme following 3′ end processing.¹⁴ Thus, the genes of mt tRNA serve as “punctuation signals” for processing of mtDNA polycistronic transcripts.¹⁵

Mt-tRNA genes are located in three transcription units that are transcribed at different rates. The short H-strand unit, including rRNA region and two tRNA genes, is transcribed twice as fast as the L-strand, containing eight tRNA genes, and 25 times more frequently than the entire H-strand unit destined to produce 12 tRNAs.¹⁶ However, King and Attardi demonstrated that the steady-state levels of the different mt tRNAs are remarkably uniform, indicating the importance of post-transcriptional downregulation of “over-expressed” molecules.¹⁷ Authors suggested that the majority of tRNA sequences transcribed from the rRNA region and L-strand may decay before any processing occurs. A plausible mechanism would involve “the presence of factors capable of stabilizing the limited amounts of newly formed tRNAs.” This is the very first formulation of the idea of mt tRNA quality control by channeling in healthy cells. Furthermore, they proposed that the “ideal candidates for these stabilizing factors would be the tRNA synthetases.” We can also hypothesize that the interaction with the CCA-adding enzyme can be a crucial step in mt tRNA post-transcriptional control, and only those tRNAs that undergo CCA addition can escape the degradation pathway. Supporting this hypothesis is the finding that addition of CCA prevents the tRNAs from degradation by tRNaseZ.¹⁴

Final tRNA maturation is provided by modifications of several nucleosides. In yeast cells, tRNA modifications actively shape the unstructured tRNA precursor into the functional molecule. Noteworthy, during this processes the maturing tRNA is transferred from one modification enzyme to another.¹⁸ This kind of channeling has not been demonstrated for mt-tRNA, but can be suggested as well. In cytosol, hypomodified tRNAs are degraded via the rapid tRNA decay pathway.¹⁹ This degradation proceeds as a consequence of destabilized three-dimensional tRNA structure followed by increased accessibility to nucleases.²⁰

Mt-tRNAs are less well structured and less extensively modified than the cytosolic ones, however, post-transcriptional modifications play especially important role for these tRNAs. Increased plasticity is a common characteristic of AU-rich animal mt-tRNAs, thus, the classical cloverleaf structure needed for correct aminoacylation may be reached only following a number of modification steps (reviewed in ref. 21). Accordingly it was shown that unmodified human mt tRNA^Lys, tRNA^Leu and tRNA^Asp cannot fold properly and adopt non-canonical hairpin structures.²² The lack or decrease of some post-transcriptional modifications is considered as one of the effects of pathogenic mutations causing MERRF and MELAS in human mt tRNA^Lys and tRNA^Leu23 (described also below). Finally, mutations in nuclear Pus1 and TRMU genes, encoding mt-tRNA modifying enzymes, can result in mitochondrial disease.²⁴ The levels of several tRNAs were decreased in fibroblasts of patients bearing mutations in TRMU.²⁵ The yeast mto2 (homolog of TRMU) null mutants are also characterized by reduced amounts of several mt-tRNAs as well as significant reduction in their aminoacylation.²⁶ To explain this effect, we can propose two possibilities: (1) instability of hypomodified mt-tRNAs due to improper folding, as discussed above; (2) inability of mutant enzymes TRMU and Pus1 to interact with mt-tRNAs, and thus to protect them from degradation nor channel them to the next protein partner, ARS.

Aminoacylation and Steady-State Levels of Mutated mt-tRNAs

Analysis of the recent data regarding mitochondrial functional disorders caused by mutations in mt-tRNA genes and in nuclear genes of mt-tRNAs binding proteins, led us to hypothesize that rapid degradation of free mt-tRNAs not bound to any protein partner provides their quality control. Here we present data demonstrating that all known mutations in mt-tRNAs which decrease their aminoacylation are associated with the corresponding decrease in their steady-state levels presumably due to reduced ability of mutated tRNA to be bound to- and protected by aminoacyl-tRNA-synthetases (ARS).

The A3243G mutation in the D-loop of tRNA^Leu(UUR) associated with MELAS syndrome, and A8344G substitution in the T-loop of tRNA^Lys, associated with MERRF syndrome, are the most frequent and the most extensively studied point mutations in mt tRNA genes. Chomyn et al.²⁷ investigated the pathogenic mechanism of these transitions by construction of trans-mitochondrial cybrid cell lines (widely used cellular models of human mt disorders, obtained by fusing mtDNA-depleted immortalized cells with enucleated cytoplasts from patients bearing mutant mtDNA). In MELAS cybrid cells, they observed more than 50% decrease in mt-tRNA^Leu(UUR) levels and 50% reduction in their aminoacylation level compared with the control cells. By combining the decrease in the amount of mt-tRNA^Leu(UUR) and the reduction in the efficiency of aminoacylation of this tRNA in different mutant cell lines, the absolute levels of aminoacylated tRNA^Leu(UUR) were calculated to be 25–30% of the average level in the controls. These data can indicate that the decrease in steady-state level of mt-tRNA^Leu(UUR) was caused by degradation of the deacylated form of the mutated tRNA .

Similar results were obtained in cells containing the T3271C mutation localized within the anticodon stem of tRNA^Leu(UUR).²⁸ The steady-state level was decreased by 70% due to shorter half-life of the mutant tRNA^Leu(UUR). The level of aminoacylated tRNA^Leu(UUR) was decreased by 50%.

The amount of mutant tRNA^Lys in MERRF cybrid cells and the aminoacylation efficiency were also decreased relative to wild type cells,²⁹ the effect being dependent on the nuclear context of the cells.³⁰ Noteworthy, the respiration deficiency in A8344G MERRF cybrids cells has been partially rescued by targeting into mitochondria tRNA^Lys molecules of yeast origin.³¹ Recently, the same approach has been applied to MELAS mutation, using the recombinant tRNA designed to be importable into mitochondria and aminoacylated by mt-LeuRS³² and thus, probably, able to enter into mitochondrial channeling.

The G8313A mutation in D-stem of mitochondrial tRNA^Lys was also associated with decrease in tRNA^Lys aminoacylation by 55%, while the level of mitochondrial tRNA^Lys was decreased to 10%.³³

Yet another study by Toompuu et al.³⁴ reports the results regarding 7472insC mutation in the V-loop of tRNA^Ser_UCN. In cybrid cells, this mutation resulted in a 70% decrease in the steady-state level of tRNA^Ser_UCN and a 25% decrease in aminoacylation compared with the wild-type cells. Processing of the precursor molecules was not affected. These results, as well as all the data presented here on different mt-tRNA point mutations, can indicate that deacylated mutant tRNA molecules were mostly degraded, as opposed to the aminoacylated ones.

This explanation matches also the data derived from the investigation of the C1624T mutation in the D-step region of mt-tRNA^Val that causes Leigh syndrome.³⁵ The authors concluded that the decrease in steady-state mt-tRNA^Val levels in the C1624T cell lines was caused by degradation of the deacylated form of the mutated tRNA.

To study the stability of mt-tRNA^Ile with the A4269G substitution in anticodon loop,³⁶ cybrid cells carrying this pathogenic mutation were cultured in ethidium bromide containing medium to inhibit mitochondrial transcription. The authors monitored the mt-tRNAs degradation in vivo and found it to be dramatically higher in mutant compared with wild-type cells: half-life (T_1/2) of 3.6 and 33 h, respectively. There was a 30% decrease in the aminoacylation level of the mutated tRNA. Interestingly, when the degradation rate was measured in vitro, the T_1/2 of the wild-type tRNAs^Ile has changed from 33 h to 10 min and was similar to that of mutant tRNA.

Similarly, the myopathy-associated G5703A mutation in the mt-tRNA^Asn gene resulted in marked decrease in tRNA^Asn steady-state levels.³⁷ This reduction was not due to the accumulation of intermediate transcripts or decreased transcription. Nevertheless, the in vitro measurement of degradation kinetics showed that the wild-type tRNA^Asn was degraded slightly faster than the mutated one. As in the previous case, this apparent paradox could be easily explained in frame of our hypothesis that suggests that both the mutated and the wild-type molecules, being deacylated before the assay, were withdrawn from the channeling and thus equally unprotected from RNases.

Taken together, all currently available data on stability and aminoacylation levels of mutated human mt-tRNA demonstrate that pathogenic mutations resulted in 50–90% decrease in the steady-state level of mutant tRNA and 25–50% decrease of its aminoacylation. The absolute degree of aminoacylation in mutant forms presents 10–25% of the average level in the controls, indicating that the decrease in steady-state level of mutant mt-tRNA is mostly caused by degradation of the deacylated form of the mutated molecules. This point is in perfect correlation with channeling hypothesis.

Pathogenic Mutations in Mitochondrial Aminoacyl-tRNA-Synthetases Lead to Decreased Steady-State Levels of the Cognate tRNAs

ARS is a group of enzymes responsible for specific attachment of an amino acid to its cognate tRNA, thus performing the key step of translation (reviewed in ref. 38). Human mitochondrial ARS are encoded by nuclear genes (named LARS2 for mt leucyl-RS, RARS2 for mt arginyl-RS, SARS2 for mt seryl-RS, VARS2 for mt valyl-RS and so on) and imported into mitochondria from the cytoplasm.

A pathogenic mutation in RARS2, the gene encoding mitochondrial arginyl-tRNA synthetase, resulted in a striking reduction in the amount of the mt-tRNA^Arg in patient’s fibroblasts.³⁹ The residual tRNA^Arg was almost fully aminoacylated. The authors concluded that uncharged mt-tRNA^Arg transcripts are unstable.

Similar results were obtained in our laboratory when we detected a pathogenic mutation c.1169A > G in SARS2, encoding mitochondrial seryl-tRNA synthetase.⁴⁰ An almost unique property of SARS2 is that it has two isoacceptor substrates: tRNA^Ser_UCN and tRNA^Ser_AGY. The total amount of tRNA^Ser_AGY in immortalized lymphocytes derived from patients bearing the pathogenic mutation in SARS2 was reduced to 10–20% of the control lymphocytes. The residual pool of this tRNA was non-acylated. We concluded that mutation in SARS2 significantly impairs the ability of the enzyme to aminoacylate tRNA^Ser_AGY leading to degradation of the uncharged tRNA molecules. In contrast to tRNA^Ser_AGY, the amount of tRNA^Ser_UCN isoacceptor remained nearly normal. Although the separation of the charged and uncharged forms of tRNA^Ser_UCN in an acid gel could not be achieved, we assumed that the level of aminoacylation of the tRNA^Ser_UCN isoacceptor was not significantly decreased. Thus, a particular mutation in SARS2 selectively interferes with aminoacylation of tRNA^Ser_AGY while conserving the activity toward tRNA^Ser_UCN close to normal.

To our knowledge, there are no available data concerning the absolute amounts of ARS in mitochondria of normal or mutant cells. For pathogenic mutations in mt ARS, the catalytic activity was measured only for DARS2, aspartyl-RS mutants.⁴¹ For mutant enzyme, the 400-fold decreased of enzymatic activity was mainly due to reduced catalytic rate, whereas tRNA binding properties were not affected. We can propose, that mutant enzyme (normally presenting in mitochondria in catalytic amounts) can interact with cognate tRNA, but stays blocked in tRNA-DARS2 complex due to very low velocity of aminoacylation reaction. This may explain the degradation of the remaining part of corresponding mt-tRNA, unable to interact with blocked ARS and thus subjected to degradation.

The data on mutated mt ARSs provide the most convincing arguments for our hypothesis. Indeed, the reduced stability of mt tRNA molecules is a result of their inability to interact with the mutant ARS. This conclusion is supported by data on the overexpression of mt ARS in human and yeast cells.

Overexpression of Mitochondrial Aminoacyl-tRNA Synthetases Rescued the Mutated tRNAs from Degradation

Base substitutions in mt-tRNA genes equivalent to human pathogenic mutations, introduced into yeast mitochondria by biolistic transformation (an approach allowing the foreign DNA delivery into yeast mitochondria, adapted for the introduction of the desired mutations into mt tRNA genes), result in severe respiratory chain defects. This makes S. cerevisiae a convenient model to study the molecular aspects of human mitochondrial diseases.⁴² Genetic screens for multicopy suppressors of corresponding respiration phenotypes led to identification of two tRNA interactors—cognate mt ARS and mt translation elongation factor EF-Tu,⁴³ both of which alleviate the defects in respiration, associated with mutations in mt tRNA genes.

Substitutions in yeast tRNA^Val, equivalent to the human C1624T mutation, and in tRNA^Leu, equivalent to the MELAS C3256T mutation, resulted in a slow growth on respiratory substrates.⁴⁴ This phenotype could be suppressed by overexpression in yeast cells of either human genes VARS2 and LARS2 (encoding mt valyl-RS and mt leucyl-RS) or their yeast orthologs (VAS1 and NAM2 respectively). The steady-state level of the mutated tRNA^Val was restored by overexpression of yeast VAS1 and partially recovered by overexpression of the human VARS2 as well as yeast non-cognate NAM2. The authors suggested that mt LeuRS and ValRS act by increasing the steady-state levels of the both mutated tRNAs, the cross-suppressing action of ARSs is likely to be mediated by stabilizing the altered structure of the mutated tRNA in a chaperone-like manner. We cannot exclude also the possibility of mis-acylation of tRNA^Val by non-cognate LeuRS and vice versa. Both “chaperone” and “mis-acylation” interpretations support the hypothesis that the interaction of tRNA with ARS increases the steady-state level of the tRNA by providing protection from nucleases.

In line with yeast data it was recently shown that in human cells respiratory deficiency associated with mt-tRNA mutations can be suppressed by overexpression of cognate ARS genes. In fact, overexpression of human LARS2 in cybrid cells carrying the A3243G MELAS mutation resulted in increased amounts of mt-tRNA^Leu(UUR) without increase in the proportion of aminoacylated tRNA.⁴⁵ However, the increase in steady-state levels of mt-tRNA^Leu(UUR) resuled in 44% increase in levels of aminoacylated mt-tRNA^Leu(UUR). Overexpression of human LARS2 in other MELAS cybrid cell lines led to 2-fold increase in the percentage of aminoacylated tRNA^Leu(UUR).⁴⁶ The authors suggested that the discrepancy between the two studies may be attributed to different nuclear backgrounds of the cybrid cells.

In accordance with data showing decreased stability of deacylated form of mt-tRNA^Val carrying C1624T mutation as compare with its aminoacylated form, Rorbach et al.³⁵ demonstrated that overexpression of cognate VARS2 led to partial restoration of the mt-tRNA^Val levels in cybrid cells.

In all above cases, ARSs overexpression led to partial recovery of the steady-state levels of mutated tRNAs. This effect can be explained only by stabilization of tRNA structure upon interaction with ARS and, therefore, tRNA protection from degradation. Interestingly, this interaction is not necessarily lead to aminoacylation, since in yeast model system it was shown that recombinant forms of ARS lacking enzymatic activity have been able to suppress mutant mt-tRNA deficiencies.⁴⁴

The Effect of Mitochondrial Elongation Factor Tu on Mutant mt-tRNA Stability

Mitochondrial elongation factor Tu (mt EF-Tu) was another protein identified in yeast screening as suppressor of mutations in mt tRNA genes. This protein forms a ternary complex with aminoacyl-tRNA and GTP following by delivery of the charged tRNA to the A-site of the ribosome during the elongation phase of protein synthesis. The successful codon:anticodon interaction leads to the hydrolysis of GTP releasing EF-Tu:GDP complex from the ribosome.

Using Saccharomyces cerevisiae mutant strain defective in the 3′-end processing of the mitochondrial tRNA^Asp transcript, it was shown that overexpression of mitochondrial EF-Tu or AspRS restored the level of the mutated tRNA^Asp and rescued the defective respiratory phenotype.⁴⁷

Similarly, yeast mt EF-Tu overexpression rescued the pathological effect of mutation in yeast mt-tRNA^Leu_UUR equivalent to the human MELAS mutations. RNA hybridization analysis confirmed that mt-tRNA^Leu_UUR levels has been restored to normal in cells overexpressing mt EF-Tu. Interestingly, mt EF-Tu overexpression also restored normal levels of another mt-tRNA (tRNA^Val) that was un-detectable in the mutant strain.⁴²

Furthermore, suppression of MELAS mutations by overexpression of elongation factors was investigated in human cells.⁴⁸ In this system overexpression of human mt EF-Tu led to partial restoration of protein synthesis deficiency in myoblasts of MELAS patients. There were no data on the steady-state levels of the mutant tRNAs, however, this effect could be anticipated given the results obtained from the yeast model.

The question how the level of mutated mt-tRNA can be rescued by overexpression of elongation factor is really intriguing. In fact, as discussed above, aminoacylation of mutant mt-tRNA is never completely ablated. We suggest that the level of mt-tRNA can be increased by overexpression of EF-Tu, since elongation factor mediates the release of aa-tRNAs from the complex with their cognate ARS, as it was shown for cytosolic protein (reviewed in ref. 49). This may have two consequences: (1) increase of amount of ARS molecules available for interaction with free mt-tRNA; (2) EF-Tu, as a part of channelling mechanism, protects the bound tRNA molecules from degradations.

It should be noted that mt EF-Tu fails to rescue defective phenotype in several cybrid cells, namely those containing the 7472insC mutation in the mt-tRNA^Ser_UCN. Modest overexpression of mt EF-Tu in these cells had no effect on the decreased abundance of tRNA^Ser_UCN.³⁴ The level of EF-Tu overexpression in various experimental systems may be critical for the manifestation of the rescue effect. Alternatively, various efficiencies of rescue can be accounted for by different levels of residual aminoacylation of mt-tRNAs in cells carrying different mutations. Accordingly, mutations that suppress aminoacylation to higher degree may result in mutant tRNA that cannot form a complex with mt EF-Tu, irrespective of the extent of EF-Tu overexpression.

All together, the observations summarized here delineate the possible role of the mitochondrial translation factor EF-Tu in protection of mt-tRNA from degradation presumably due to the channeling effect.

Possible Mechanisms of mt tRNA Degradation

Only limited progress has been made in the understanding of mechanisms of RNA degradation in mammalian mitochondria (reviewed in refs. 50 and 51). For mt-RNA, including tRNA, an attractive model of internal polyadenylation-dependent degradation has been proposed.⁵⁰ In bacteria, polyadenylation plays a central role in RNA degradation, which is thought to be initiated by endoribonucleolytic cleavage of RNA molecule, followed by the addition of poly(A) tails to the end of the cleavage product, referred to as internal polyadenylation. Internal polyadenylation targeting RNA molecules for rapid 3′–5′ exonucleolytic degradation has been shown not only for mammalian mt mRNA molecules, but also for mt-tRNAs. Of note, contrary to mRNA molecules, mt-tRNAs undergo only internal, but not 3′- end polyadenylation. Next, it was shown that CCA added to the 3′ end of mature tRNA serve as an anti-determinant for tRNase Z recognition.¹⁴ Therefore, it can be hypothesized that tRNase Z is responsible for endoribonucleolytic cleavage of non-mature mt-tRNA molecules (those without CCA at 3′-end), promoting their subsequent internal polyadenylation and degradation. Which enzymes could potentially accomplish the subsequent steps of this process, internal polyadenylation and exonucleolytic degradation? The homogeneity of the internal poly(A) tails in mt-tRNA molecules suggests that they can be formed by poly(A) polymerase (PAP), identified in human mitochondria.⁵² The most likely exoribonuclease that degrades internally polyadenylated mt-tRNAs appears to be polynucleotide phosphorylase (PNPase). However, PNPase role in mt RNA degradation was recently questioned by demonstration of its localization on the inter-membrane surface of the inner mitochondrial membrane and possible role in RNA import into human mitochondria.⁵³

So far the best characterized mt RNA decay system has been described for the yeast S. cerevisiae. Yeast mitochondrial degradosome complex consists of RNA helicase SUV3 and of exoribonuclease DSS1.⁵⁴ The human ortholog, hSuv3p, is an ATP-dependent multisubstrate helicase localized within the mitochondrial matrix.⁵⁵ Yet, the mitochondrial exoribonuclease that would constitute the partner of hSuv3p, has not been identified. Thus, there are many open questions regarding the molecular mechanisms of tRNA decay in human mitochondria.

In any case, tRNA degradation occurs via an enzymatically driven process. We suggest that nucleases form a part of channeling, since tRNA molecules that cannot proceed to the next step of channeling (due to mutation in tRNA or in corresponding protein partner) should be directly transferred to the degradation machinery. We are not aware of documented examples of tRNA degradation in vivo without implication of nucleases, since tRNA molecules are rather stable and can be degraded only at high pH values or in the presence of heavy metals, normally not detected in cells, especially inside the mitochondria. Recently discovered cleavage of cytosolic tRNAs in halves upon stress conditions is catalyzed by released from the vacuole (in yeast) or endocytosed (in human cells) endonucleases (reviewed in ref. 56), however, the function of this endonucleolytic process is not to decrease the levels of cellular tRNAs, but rather to create new regulatory molecules. Since this process can be activated by oxidative stress, directly connected with mitochondrial dysfunctions, the stability of mt-tRNA would be very interesting to determine.

Possible degradation or fragmentation of mt-tRNA in the conditions of apoptosis or oxidative stress is not easy to detect using the common methodology. New technologies can accelerate the progress of our knowledge on the involvement of mt-tRNA and its fragments in signaling and stress response pathways. For instance, RNA deep-sequencing⁵⁷ permitting to identify all the RNA molecules present in the cell even in very low amounts as well as their fragments and degradation products, can be used to trace the various stages of tRNA degradation in normal conditions and under pathological circumstances. Specific tRNA chips approach,⁵⁸ applied for studies of tissue-specific individual tRNA expression,⁵⁹ has never been used for mt-tRNA. This micro-array method seems to be very perspective to evaluate the steady-state and aminoacylation levels of various mt-tRNAs in human cells containing pathogenic mutations in mtDNA or in nuclear genes involved in mitochondrial channeling process.

Conclusions

In this review we tried to summarize the available data concerning the possible regulatory mechanisms of mitochondrial tRNA steady-state levels at different stages of mt-tRNA pathway: synthesis, maturation and function in protein synthesis. It seems probable that tRNA channeling takes place in human mitochondria. Indeed, at each step of its “life cycle,” mt-tRNA should form a complex with its corresponding protein, protecting it from undergoing nucleolytic degradation. Mutations in mt-tRNA or in its protein interactors disturb the tRNA channeling, resulting in tRNA hydrolysis and consequent decrease in its steady-state level.

Understanding the fundamental mechanisms regulating mt-tRNA in health and in pathological conditions will significantly impact on the prevention and cure of the human mitochondrial disorders, often devastating, for which no treatment is currently available.

Abbreviations:
mt	=	mitochondrial
ARS	=	aminoacyl-tRNA-synthetase
NRC	=	nuclear pore complex
MELAS	=	mt encephalomyopathy, lactic acidosis and stroke-like episodes
MERRF	=	myoclonic epilepsy with ragged red fibers

Acknowledgments

We are grateful to Julia Yaglom for the critical reading of the manuscript. We thank our colleagues and collaborators for their contributions. N.E. was supported by grants AFM, FRM and ANR.