5 S rRNA and tRNA import into human mitochondria. Comparison of in vitro requirements.

In vivo, human mitochondria import 5 S rRNA and do not import tRNAs from the cytoplasm. We demonstrated previously that isolated human mitochondria are able to internalize a yeast tRNA(Lys) in the presence of yeast soluble factors. Here, we describe an assay for specific uptake of 5 S rRNA by isolated human mitochondria and compare its requirements with the artificial tRNA import. The efficiency of 5 S rRNA uptake by isolated mitochondria was comparable with that found in vivo. The import was shown to depend on ATP and the transmembrane electrochemical potential and was directed by soluble proteins. Blocking the pre-protein import channel inhibited internalization of both 5 S rRNA and tRNA, which suggests this apparatus be involved in RNA uptake by the mitochondria. We show that human mitochondria can also selectively internalize several in vitro synthesized versions of yeast tRNA(Lys) as well as a transcript of the human mitochondrial tRNA(Lys). Either yeast or human soluble proteins can direct this import, suggesting that human cells possess all factors needed for such an artificial translocation. On the other hand, the efficiency of import directed by yeast or human protein factors varies significantly, depending on the tRNA version. Similarly to the yeast system, tRNA(Lys) import into human mitochondria depended on aminoacylation and on the precursor of the mitochondrial lysyl-tRNA synthetase. 5 S rRNA import was also dependent upon soluble protein(s), which were distinct from the factors providing tRNA internalization.

Mitochondria, although containing their own genome, import the vast majority of their macromolecular components from the cytoplasm. If the mechanisms of pre-protein import are well understood, the import of nuclear-coded RNAs into mitochondria was investigated to a much lesser extent. Targeting of RNA into mitochondria though not universal is widely spread among organisms (1)(2)(3)(4)(5). Mitochondrial import of transfer RNAs was found in plants, protists, some lower animals, and fungi. The number of imported tRNA species varies from one (in yeast) to the totality (in trypanosomatids), and tRNA import mechanisms seem to differ from one organism to another. We have shown previously that, in the yeast Saccharomyces cerevisiae, import of a single tRNA CUU Lys (further referred to as tRK1) 1 occurs via formation of a complex with the precursor form of the mitochondrial lysyl-tRNA synthetase (pre-MSK) and requires the intactness of pre-protein import apparatus (6 -8).
In mammalians, no tRNA import has been reported, but several other RNAs are thought to be targeted into mitochondria. One of these is the RNA component of RNase MRP, a site-specific endoribonuclease supposed to be involved in primer RNA cleavage during replication of mitochondrial DNA and to be present in the organelle in a very low amount (9). It was hypothesized that the process of mitochondrial DNA replication requires a very low number of MRP RNA molecules per mitochondrial genome (10 -12). The presence of MRP RNA in the mitochondria was recently strongly supported by quantitative analysis of mitochondrial RNAs in HeLa cells (13).
The second candidate for import into mammalian mitochondria is the RNA component of RNase P, an endoribonuclease involved in processing of 5Ј ends of tRNAs (14). The gene coding for the RNA component of RNase P, found in a number of mitochondrial genomes, is absent in human mitochondrial DNA. On the other hand, the activity of partially purified RNase P of human mitochondria was found to be nucleasesensitive. The situation was controversial, because an RNAindependent RNase P activity associated with human mitochondria was also reported (15). However, the most recent data confirm the RNA-dependent character of RNase P activity in human mitochondria and mitochondrial localization of the nuclear-encoded 340-base RNA component of this enzyme (13).
A third nuclear-encoded RNA associated with mammalian mitochondria is the 5 S ribosomal RNA (16). This small (119 -122 bases) RNA is not encoded by the mitochondrial genome in mammals; however, it was detected by Northern hybridization in highly purified bovine mitochondria and mitoplasts. Furthermore, it was demonstrated that introduction of a tag sequence into the nuclear 5 S rRNA gene did not inhibit the import of the 5 S RNA, and the mutant RNA was detected in highly purified mitoplasts of transfected human cells (17).
The function of imported 5 S rRNA in mammalian mitochondria is unknown. In plants, algae, and some protozoans, a 5 S rRNA is encoded by the mtDNA and was found to be a component of the mitochondrial ribosome. By contrast, in fungi and animals (where it is not encoded by the organellar genome), no 5 S rRNA has been detected in mitoribosomes. There are several reports describing isolation and characterization of mammalian mitoribosomes (18 -20). These preparations (which lacked 5 S rRNA) were active in poly(U)-directed phenylalanine polymerization but failed to direct translation of natural mRNAs in vitro. The finding that a fraction of nuclear-coded 5 S rRNA is associated with mammalian mitochondria raises the question whether it is a component of the fully functional mitoribosome lost during ribosome isolation or it has another function in the organelle.
Another important question is how the negatively charged RNA molecule can cross the hydrophobic environment of the mitochondrial double membrane in countercurrent of proton gradient. Mitochondrial import of RNA was studied on models of trypanosomatids and yeast and concerned mostly transfer RNAs (3,21). It is commonly agreed that human mitochondria do not import tRNAs in vivo, although this conclusion is mainly based on the presence of all tRNA genes needed for mitochondrial translation in the mitochondrial genome (22). However, in yeast S. cerevisiae, also possessing tRNA genes sufficient for translation, a single tRNA, tRK1, is imported into mitochondria in vivo (6). Its import depends on the cytoplasmic precursor of the pre-MSK and other(s) soluble cytosolic factors not yet identified (5,7,23). We have shown recently that, in the presence of pre-MSK, protein extracts of human cells direct the import of tRK1 into isolated human mitochondria (24). This fact may signify that the human cell possesses the most of the machinery needed for tRNA import, even if this pathway does not exist in vivo. Here we describe a specific in vitro assay for 5 S rRNA import into human mitochondria and show that its internalization proceeds by a mechanism similar to that found for tRNA in yeast.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-Escherichia coli strain XL1blue (Stratagene) or Stabl2 (Life Technologies, Inc.) were used for cloning purposes, as described by the manufacturers. Epicurian coli strain BL21-CodonPlus (DE3)-RIL (Stratagene) was used to overproduce KRS, pre-MSK, and MSK. S. cerevisiae strain YPH499 (25) was used to isolate mitochondria. ⌬Tom20 IRV yeast strain bearing the pMRS-A plasmid (MSK1 gene cloned in pRS416 vector) was used to isolate yeast import directing proteins (ScIDPs) (26). Yeast strain W⌬M was used to prepare extracts lacking pre-MSK (ScIDPs°). This strain was constructed from W303 strain by deleting the MSK1 gene using the KanMX4 cassette replacement technique (27). Cultivation of yeast cells was done as described (28). HepG2 cell line (provided by R. Lightowlers) was used to isolate human mitochondria. HepG2 or HeLa cells were used to isolate human import directing proteins (HmIDPs). Human cells were cultured in Dulbecco's modified Eagle's medium with nonessential amino acids, Earle's salts, and L-glutamate, buffered with sodium bicarbonate, and supplemented with 10% fetal calf serum at 37°C and 5% CO 2 .
All plasmids for tRNA expression in vitro were constructed by insertion of a corresponding polymerase chain reaction-amplified gene in pUC118 under control of a T7 promoter and with addition of a BstNI site (which creates a 3Ј-CCA-OH terminus after in vitro transcription). Two versions containing U in 5Ј-terminal positions (r8 and r2), which is unfavorable for transcription catalyzed by T7 RNA polymerase, were cloned so that the tRNA sequence was preceded by a hammerhead ribozyme sequence. Transcription of these constructs gives rise to a "tranzyme" molecule (29) the autocatalytic activity of which liberates the corresponding tRNAs bearing a correct terminal nucleotide. The cloverleaf structures and mutations are shown in Fig. 8A. Mutations were introduced as described (30).
Isolation and Characterization of Mitochondria-Yeast mitochondria were isolated as described previously (8,31). Human mitochondria were isolated as described elsewhere (32) with minor modifications (schematically presented in Fig. 1). After the second round of high speed centrifugation of mitochondria, they were suspended in Breakage Buffer (0.44 M mannitol, 20 mM Tris-HCl (pH 7.0), 20 mM NaCl, 1 mM EDTA) and centrifuged through two layers of sucrose (0.6 and 1.85 M). Mitochondria were collected on the top of the 1.85 M layer and harvested by high speed centrifugation. The integrity of mitochondria was checked by a citrate synthase assay (33). Mitochondria used for import were at least 85% of integrity. Contamination with cytosolic membranes was checked by measuring the activity of Ca 2ϩ -dependent ATPase (34). Purified mitochondria contained less than 1% of total ATPase activity. Measurement of the respiratory control ratio of isolated mitochondria was done with succinate as a substrate (33). Mitochondria used for import assay had respiratory control ratio between 3.5 and 5.0. The integrity of mitochondrial proteins localized in different organellar compartments was controlled by Western analysis, using antibodies against yeast ATP-ADP carrier or yeast mtHSP70 (provided by N. Pfanner) or against subunits I and II of human cytochrome c oxidase (Molecular Probes) for human mitochondria.
Enzymes and Proteins Directing the Import (IDPs)-Pre-MSK was produced by overexpression of the MSK1 gene in E. coli using the pET3a vector. The recombinant protein was purified from inclusion bodies and refolded as suggested by the manufacturer (Stratagene). The protein used was at 85% pure. MSK (the mature form) was expressed and purified in a same way, but the first 32 codons of the MSK1 gene, corresponding to a targeting signal predicted by the PSORT program (35), were deleted. MSK was at 90% pure. KRS was expressed from pET3a and His 6 tag-purified onto nickel-nitrilotriacetic acid spin columns (Qiagen). Mammalian cytoplasmic aminoacyl-tRNA synthetases (aaRS) were a partially purified multi-aaRS complex (36) from bovine heart. Mammalian mitochondrial lysyl-tRNA synthetase was a partially purified from bovine heart (LysRSmt). aaRS and LysRSmt were kindly provided by L. Sydorik. MSK versions were identified by Western analysis using polyclonal antibodies against the recombinant Msk1p (kindly provided by A. Tzagoloff).
Isolation, Labeling, and Analysis of RNAs-tRK1 and tRK2 were purified as described previously (7,38). In vitro transcription was done using T7 RNA polymerase (Promega). r2 and r8 tranzymes were generated as described in Ref. 29. Synthetic RNAs were excised from a 40-cm-long 12% denaturing gel permitting single base resolution, and were refolded by one cycle of heat denaturing-refolding in the presence of 0.5 mM Mg 2ϩ . We identified 5Ј-terminal nucleotides of the tranzymes r2 and r8 by P1 nuclease hydrolysis (39) with subsequent thin layer chromatography (40). To verify 3Ј-termini of synthetic tRNAs, we used aminoacylation tests. Aminoacylation by the yeast cytosolic lysyl-tRNA synthetase (KRS) was done as in Ref. 7 and by bovine heart aaRS as in Ref. 37. Aminoacylation of tr3 was done by pure recombinant MSK as in Ref. 6, and trKHm was aminoacylated by partially purified mitochondrial lysyl-tRNA synthetase from bovine heart. Taking into account the results of thin layer chromatography and high levels of aminoacylation (50 -100%, as indicated in Table I), we can affirm that tRNA transcripts used in import assays had predicted termini and were correctly folded. To purify human and yeast 5 S rRNA, human 5.8 S rRNA, and tRNAs, total RNA was prepared with Trizol reagent (Life Technologies, Inc.) and separated on 12% denaturing polyacrylamide gels. Bands corresponding to the needed RNA were excised and extracted (41). RNAs were checked for purity by electrophoresis and Northern analysis. Before import assays, all RNAs were fully 5Ј end-32 P-labeled by T4 polynucleotide kinase, gel-purified, and refolded.
RNA Import Assays-The standard assay (100 l) contained mitochondria (50 g of mitochondrial protein), 3 pmol of 5Ј end-32 Plabeled RNA and 10 g of IDPs in Import Buffer: 0.44 M mannitol, 20 mM HEPES-KOH (pH 6.8), 20 mM KCl, 2.5 mM MgCl 2 , 1 mM ATP, 5 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 0.1 mM diisopropyl fluorophosphate, 0.1 mM L-lysine, 0.5 mM phosphoenolpyruvate, and 4 units of pyruvate kinase. The import assay was carried out at 30°C for 20 min, which corresponds to the logarithmic phase of RNA uptake (42). After treatment with a mixture of nucleases (20 units/ml micrococcal nuclease, 50 g/ml RNase A, and 25 units/ml phosphodiesterase), mitoplasts were generated by treatment with digitonin, at 100 g/1 mg of mitoprotein. After incubation for 15 min in ice, mitoplasts were harvested by centrifugation and washed twice as described (43). Mitoplast quality was controlled by Western analysis with antibodies against an outer membrane protein, porin (Calbiochem). In average, mitoplasts lose 50 -60% of porin with respect to intact mitochondria. Mitoplasts were then lysed in 1% SDS, 0.1 M sodium acetate (pH 4.8), and 0.05% diethyl pyrocarbonate at 60°C; mtRNA was phenol-extracted and separated by denaturing gelelectrophoresis; and import was quantified by scanning in a phosphorimager (Fuji, Bas2000).
MSK-binding Assay-The assay was as described in Ref. 23 with several modifications: 10 5 cpm of 5Ј 32 P-labeled RNA was mixed with 5 g of IDPs in the final volume of 50 l of Import Buffer without mitochondria. The mixture was incubated for 10 min at 20°C, and insoluble aggregates were removed by a brief centrifugation. 50 l of Lysis Buffer (0.1% Nonidet P-40, 0.1 M Tris-HCl (pH 7.5)) and 1 l of anti-MSK antibodies was added to the supernatant. After incubation for 2 h at 4°C, 10% v/v of protein A-Sepharose beads were added and the mixture was shacked for 1 h at 4°C. Sepharose beads were harvested by a brief centrifugation, washed with Lysis Buffer, and radioactivity was determined by Cerenkov counting. As a negative control, the same assay was done with bovine serum albumin instead of IDPs, as a positive one labeled aminoacylated tRK1 and ScIDPs were used.
Pre-protein Import Assay-Pre-MSK gene was polymerase chain reaction-amplified by using two oligonucleotides; the 5Ј-terminal included the T7 promoter and the N-terminal sequence of pre-MSK, CCCTGATCATAAT-ACGACTCACTATAGGGAGCCACCATGAATGTGCTGTTAAAAAGA-CGCAG, and the 3Ј-terminal included the termination codon GGGAATTC-CATATGAATGTGCTGTTAAAAAGACGC. 35 S-Labeled pre-MSK was synthesized in vitro in a coupled transcription-translation T7 polymerase-dependent system (Promega). The import assay was done in the same conditions as RNA import with exception of GIP-blocking experiments, where DTT was omitted to avoid reducing disulfide bridges in bovine pancreas trypsin inhibitor (BPTI). After 20 min of incubation, 50 g/mg mitochondrial protein of proteinase K were added and the mixture was incubated for additional 5 min at 20°C. Phenylmethylsulfonyl fluoride was then added to 0.5 mM, mitochondria were harvested by centrifugation and suspended in the sample buffer and the proteins were separated by electrophoresis in 10% Laemmli's polyacrylamide gel (44).
Blocking of Pre-protein Import-To dissipate the membrane charge, 10 -20 mol of uncouplers (valinomycin or oligomycin) were added to the standard import mixture, which was incubated in ice for 5 min before addition of labeled RNA and IDPs. The mixtures contained 70 mM KCl in the case of valinomycin and 0.5 mM KCN in the case of oligomycin treatment (45). To remove outer membrane-associated receptors, mitochondria were treated with 10 g of proteinases (trypsin or proteinase K)/mg of mitoprotein (45), harvested by centrifugation, and washed two times with Breakage Buffer containing soybean inhibitor of trypsin.
To block the pre-protein import channel GIP (general insertion pore), we designed a covalent conjugate that represents 32 N-terminal amino acids (MLSNLRILLNKALRKAHTSMVRNFRYGKPVQC-NH 2 ) corresponding to the mitochondrial targeting signal of rat ornithine transcarbamoylase (OTC) coupled by a heterobifunctional reagent MBS to free amino groups of BPTI. The coupling was proceeded by using the thiol group of a unique C-terminal cysteine of OTC (see the scheme in Fig. 4A). Coupling was performed in Neosystems. Mitochondria were incubated with OTC-BPTI conjugate (40 -200 nmol/import assay) at 20°C for 10 min before addition of IDPs and labeled RNA.

S rRNA Is Imported into Isolated Human Mitochondria-
We previously described an assay permitting specific import of yeast cytoplasmic tRK1 into isolated yeast mitochondria (31) and, more recently, demonstrated that isolated human mitochondria can internalize tRK1 in the same conditions (24). Here we tested whether small human RNAs can be internalized by isolated human mitochondria. The sequence of manipulations is presented in Fig. 1. The import reaction contained energized mitochondria of HepG2 cells, soluble proteins termed IDPs, and 5Ј end-32 P-labeled RNA. Mitochondria were devoid of endoplasmic reticulum membranes (judged by the absence of Ca 2ϩ -dependent ATPase activity) and were at Ͼ85% integrate (estimated by a citrate synthase assay). After a 15-min incubation, the external RNA was removed by treatment with nucleases, generation and purification of mitoplasts. The mitoplasts obtained by this procedure were only partially devoid of the outer membrane (estimated by quantitative Western analysis, which has shown a 50 -60% loss of an outer membrane marker, porin). Protected RNA was analyzed by gel electrophoresis and autoradiography. At different steps, we checked the presence of contamination with nuclear (U3 snRNA) or cytoplasmic (cytoplasmic tRNA Met ) RNAs by Northern hybridization (Fig. 1, right panel). Initial preparation of mitochondria contained a small amount of U3 snRNA (1% of the total pool) and of cytoplasmic tRNA Met (0.2%); however, treatment with nucleases completely removed nuclear and cytoplasmic contaminating RNAs, whereas the internal marker, the mitochondrial tRNA Cys , and the partially imported 5 S rRNA were protected. Generation of mitoplasts and treatment with nucleases caused a decrease (by 40%) of the signal for mitochondrial tRNA Cys in comparison with the initial mitochondrial preparation, which might be the result of a partial disruption of mitochondria during the procedure. Taking into account that this tRNA (encoded in mitochondrial DNA) is present exclusively in the mitochondria in vivo, we used the corresponding coefficient (k ϭ 1.67) for all the further quantitative estimations. The obtained values for purity of mitochondria are in agreement with previous reports (13).
After incubation of 5Ј end-labeled human 5 S rRNA with mitochondria in the presence of HmIDPs and ATP, we observed that a portion of the externally added RNA was protected from nucleases ( Fig. 2). Protected 5 S rRNA becomes detectable within 10 min, and the amount of protected full-length 5 S rRNA reaches the plateau at 20 min (not shown). Protected 5 S rRNA was observed only upon addition of partially fractionated protein extract of human cells (HmIDPs) and was not detected in the absence of soluble proteins. Detection of labeled 5 S rRNA after treatment with nucleases cannot be explained by protection with proteins present in HmIDPs, because the same treatment in the absence of mitochondria leads to a complete hydrolysis of the RNA (Fig. 2). Labeled 5 S rRNA remains protected after disruption of the outer membrane and purification of mitoplasts, in a same way as internal mitochondrial RNA marker (tRNA Cys , Fig. 1). After disruption of the mitochondrial membranes by a detergent, labeled 5 S rRNA becomes accessible to the nucleases (Fig. 2). These results suggest that the externally added 5 S rRNA was internalized by the mitochondria. Generation of mitoplasts had no significant effect onto the amount of protected 5 S rRNA in comparison with nuclease-treated mitochondria. Therefore, protected RNA is associated either with the matrix component or the inner membranes, or both. 5 S rRNA internalization by human mitochondria requires the presence of ATP (Fig. 2), which can not be replaced by other NTP or nonhydrolyzed ATP analog (not shown). This result means that internalization requires the energy of ATP hydrolysis.
Surprisingly, ScIDPs were also found to direct the import of 5 S rRNA, although the efficiency of the import was significantly lower (6 -8 times), whereas simultaneous addition of HmIDPs and ScIDPs had no synergetic effect (Fig. 2). This cannot be explained by a differential stability of the 5 S rRNA in the presence of human or yeast proteins. In fact, incubation of labeled RNA with either HmIDPs or ScIDPs in the absence of mitochondria resulted in a comparable level of degradation (5-15% for the band corresponding to the nondegraded RNA, see Fig. 2, right panel). Therefore, the needed cytoplasmic factor(s) seem to be conserved in both organisms, but either the human factor(s) are more active to deliver 5 S rRNA into human mitochondria, or they were less concentrated in ScIDPs preparations.
To understand if RNA import in the conditions used was specific for 5 S rRNA, we tested other human small cytoplasmic RNAs, which are not imported in vivo. We found that human cytoplasmic tRNAs and human 5.8 S rRNA were not imported, neither in the absence of proteins, nor in the presence of ScIDPs or HmIDPs (Fig. 2). Absence of detectable protection was because of the absence of internalization of RNAs by the mitochondria, and not the differential degradation of RNAs by IDPs, because the RNA stability test demonstrated that they all were as stable as the 5 S rRNA in the presence of proteins (Fig. 2, right panel). We tested other individual tRNAs of different origin (yeast and bacterial) and found that they were not internalized in the conditions used (not shown). We can conclude that the assay developed is specific for a naturally imported RNA, 5 S rRNA, and can therefore be used to study the mechanisms of its internalization.
Quantification of in Vivo and in Vitro Import of 5 S rRNA-Combination of Northern hybridization (Fig. 1, right panel) and in vitro import assays (Fig. 2) permits to estimate the efficiency of 5 S rRNA import in vivo and in vitro. By scanning hybridization signals and taking into account the degradation coefficient (k ϭ 1.67, see above), we can deduce that 0.9% of the total cellular 5 S rRNA was associated with the mitochondrial compartment (inner membrane and matrix). This value can be considered as the in vivo import efficiency. The amount of total cellular 5 S rRNA was deduced from comparison of the densities of the bands corresponding to 5 S rRNA and to the control RNA (100 ng of pure 5.8 S rRNA; Fig.  1, left panel). For in vivo analysis, mitochondria were isolated from 3.5 ϫ 10 7 cells. Taking into account that the efficiency of import in vivo is 0.9%, we could estimate that each cell contain 3.2 ϫ 10 4 molecules of 5 S rRNA associated with mitochondria. As a matter of fact, evaluating the average number of mitochondrial ribosomes per cell as 3.4 ϫ 10 4 (46 -48), it appears that the number of molecules of imported 5 S rRNA in vivo may be sufficient for fitting all the organellar ribosomes.
3 pmol of fully 5Ј end-labeled 5 S rRNA was used in each in vitro import assay with 50 g of mitochondrial protein. This amount of mitochondrial protein corresponds to 7.0 ϫ 10 5 disrupted cells. Each cell contains, on average, 400 mitochondria (the exact number of mitochondria in cultured human cells varies depending on the phase of cell cycle (Refs. 13, 49, and 50)). Assuming that the cells were fully disrupted, we can estimate that 2.8 ϫ 10 8 mitochondria were present in each import assay. We compared the density of the band corresponding to the full-length form of protected 5 S rRNA with an aliquot of the input labeled RNA (Fig. 2). Taking into account the 5% level of RNA degradation in the presence of HmIDPs (see Fig. 2, left panel) and the degradation coefficient, we estimated the efficiency of 5 S rRNA import in vitro directed by HmIDPs as 1.0 Ϯ 0.1%. This value is underestimated, because only 60 -80% of the protected RNA corresponded to the nondegraded form of 5 S rRNA. This can be because of degradation of RNA inside the mitochondria after the import, but not outside, because no such degraded band was observed with nonimported RNAs (5.8 S rRNA or tRNAs, Fig. 2). Furthermore, the amount of protected RNAs of lower size was always proportional to that of the full-length form. Therefore, for all further evaluations of import efficiency, we measured densities of the bands corresponding to full-size RNAs.
Considering the estimations described above, one can conclude that the efficiencies of 5 S rRNA import in vivo (0.9%) and in vitro (1.0 Ϯ 0.1%) were similar. 5 S rRNA Import Depends on the Pre-protein Import Apparatus-The need of soluble proteins to direct 5 S rRNA into human mitochondria could be explained by a mechanism described previously for mitochondrial import of tRNA in yeast, where it is co-imported with a mitochondrially targeted preprotein (7,8). To verify this hypothesis, we checked if the pre-protein import apparatus is involved in 5 S rRNA internalization.
Pre-protein import machinery requires the intactness of the outer membrane-exposed proteins of the GIP and the electrochemical potential across the mitochondrial inner membrane (51)(52)(53). Human mitochondria pretreated with proteases failed to internalize 5 S rRNA (Fig. 3A), suggesting that proteins exposed on mitochondrial outer membrane are required for importing this RNA. Reversible dissipation of the membrane charge by treatment of mitochondria with valinomycin (a potassium ionophore) (45) leads to an inhibition of 5 S rRNA internalization (Fig. 3B). An irreversible block of the respiratory chain with oligomycin in the presence of potassium cyanide also inhibited the import (Fig. 3B). Taken together, it appears that, for internalization of 5 S rRNA, human mitochondria need ATP hydrolysis energy, membrane charge, and protease-sensitive outer membrane receptors. Such requirements are similar to these described for protein import into mitochondria. On the other hand, this does not prove directly the involvement of pre-protein import machinery, because similar requirements were found for tRNA mitochondrial import in trypanosomatids, but the RNA and protein import receptors were still proven to be distinct (54 -57).
To address directly the implication of the pre-protein import apparatus in RNA import, we designed a system of blocking of GIP by a nonunfoldable protein. To construct it, the BPTI was conjugated with the N-terminal signal peptide of mitochondrially targeted OTC (Fig. 4A). The N-terminal peptide directs the conjugate OTC-BPTI to the mitochondrial outer membrane receptors, whereas BPTI portion blocks the GIP because of its FIG. 2. Isolated human mitochondria specifically import 5 S rRNA. Autoradiographs of 5Ј end-labeled RNAs protected from nucleases after incubation with isolated HepG2 mitochondria are presented. Input, aliquots (1%) of labeled RNAs; arrows indicate full-length RNAs. cyto tRNA, human cytoplasmic tRNAs; 5S rRNA and 5.8S rRNA, natural human 5 and 5.8 S ribosomal RNAs. Particular conditions of import experiments are explained at the top of the autoradiographs: w/o mitochondria, the assay including nucleases treatment was done in the absence of mitochondria; w/o protein, the assay was done without addition of proteins; w/o ATP, the assay was done without addition of ATP; Triton X-100, after a standard import assay, the detergent was added to 0.1% before addition of nucleases. Equal amounts of HmIDPs or ScIDPs were added in each assay. The assays without mitochondria and without ATP and Triton X-100 were done with HmIDPs. 3 pmol of fully labeled RNAs were added in each assay. The slots of the middle and lower panel represent the results of import assay with 5.8 S rRNA and tRNAs in conditions corresponding to these of the upper panel (input, HmIDPs, ScIDPs, and HmIDPsϩScIDPs). Right bottom, quantification of RNA stability. Labeled RNAs were incubated in the presence of IDPs in the same conditions as in the import assay in the absence of mitochondria. The value indicates the percentage of the full-length RNA after 20 min with respect to the assay without proteins. disulfide bridges preventing unfolding (58,59). To demonstrate that pre-protein import was inhibited by OTC-BPTI, we used a labeled pre-protein, pre-MSK. Isolated yeast and human mitochondria in the conditions used for 5 S rRNA import were able to internalize the in vitro synthesized pre-protein (Fig. 4B). As expected, pre-incubation of the mitochondria with the conjugate totally inhibited the in vitro import of pre-MSK both into yeast and into human mitochondria (Fig. 4B). At high concentrations, 200 nmol/assay, BPTI by itself has a slight inhibitory effect onto pre-MSK import. A possible explanation would be that commercial BPTI contained contaminants that partially inhibited the import. However, at 200 nmol/assay, BPTI alone inhibited the import of pre-MSK only by 10 Ϯ 5%, whereas OTC-BPTI completely abolished it (Fig. 4B), which proves that the conjugate specifically affected pre-protein apparatus.
Pre-incubation of human mitochondria with OTC-BPTI resulted in a strong inhibition of 5 S rRNA internalization (Fig.  4C). Likely in the test of pre-MSK import (Fig. 4B), BPTI itself has a slight inhibitory effect. However, at 200 nmol/assay, inhibition by BPTI was only 15 Ϯ 5%, whereas inhibition by OTC-BPTI was complete. From these results, it seems obvious that the GIP channel is involved in 5 S rRNA import.
Human Proteins Can Direct Import of a Yeast tRNA Lys into Human Mitochondria-We have shown previously that isolated human mitochondria can internalize a yeast tRNA Lys , tRK1 (24). This process, although specific, was completely artificial because, first, human mitochondria do not import tRNAs in vivo, and, second, tRK1 is a yeast tRNA. However, comparison of requirements for tRK1 internalization and these for 5 S rRNA import reveals an evident similarity. HmIDPs supplemented with pre-MSK were found to direct internalization of tRK1 (24). In these previous experiments, HmIDPs were isolated by the same procedure as ScIDPs, directing tRK1 import into isolated yeast mitochondria (42,60). We tested here if we could direct tRK1 into isolated human mitochondria without addition of pre-MSK by changing the conditions of HmIDPs isolation (Fig. 5A). We varied two parameters of protein solubilization, pH and salt concentration, and added the step of DEAE cellulose chromatography. Only HmIDPs extracted at low pH (6.5) were active in tRK1 import assay (Fig. 5A). HmIDPs isolated at low ionic strength, found previously to be optimal for isolation of ScIDPs, were active only upon addition of pre-MSK. Increasing the concentration of NaCl results in HmIDPs that can direct tRK1 import without addition of pre-MSK (Fig. 5A). The effect of ionic strength can be explained by dissociation of macromolecular aggregates leading to solubilization of import factor(s). In all further experiments of importing tRNA derivatives into human mitochondria, we used the HmIDP preparations obtained at 150 mM NaCl and at pH 6.5.
To test whether tRK1 import involves the pre-protein channel, we blocked the GIP by OTC-BPTI conjugate. We observed for tRK1 the same effect as for 5 S rRNA; addition of 200 nmol of conjugate prior to the import assay completely blocked the import, whereas BPTI alone at this concentration inhibited the import of tRK1 only to 10 Ϯ 5% (Fig. 5B). We conclude that human cells contain all protein components needed for artificial tRK1 import into mitochondria and that the mechanism of its translocation is similar to that used by 5 S rRNA.
tRNA Lys Import into Human Mitochondria Depends on Lysyl-tRNA Synthetases-In yeast, tRK1 is imported in its aminoacylated form and its aminoacylation is catalyzed by the corresponding cytosolic lysyl-tRNA synthetase, KRS (7). We verified if this observation remains true for tRK1 import into human mitochondria directed by HmIDPs. Import of tRK1 did not depend on addition of mammalian complex of aaRS or yeast KRS when the tRNA was aminoacylated prior the import. In contrast, the deacylated version of tRK1 was very poorly imported without addition of aminoacyl-tRNA synthetases, whereas addition of mammalian aaRS or KRS strongly increased its import (Fig. 6A). Low import efficiency of deacylated tRK1 (10% with respect to a fully aminoacylated tRNA) can be explained either by a noncomplete deacylation or by a partial aminoacylation of the tRNA during the import reaction. Enhancing effects of aaRS and KRS were similar. Furthermore, tRK1 can be aminoacylated with the lysine by mammalian cytosolic aminoacyl-tRNA synthetases with an efficiency similar to that for KRS (Fig. 6A). These results mean that tRK1 must be aminoacylated prior to import into human mitochondria and that human lysyl-tRNA synthetase can provide this aminoacylation.
Pre-MSK is essential for tRK1 import in yeast mitochondria (7, 23). We used this established fact to estimate the stoichiometry of tRK1/pre-MSK interactions. ScIDPs°(yeast protein extracts lacking pre-MSK) cannot direct tRK1 into human mi-tochondria without addition of pre-MSK. Addition of increasing amounts of pre-MSK lead to an increase of the amount of imported tRK1 (Fig. 6B, upper panel). By using the in vitro synthesized 35 S-labeled protein, the efficiency of pre-MSK import in vitro was estimated as 20 Ϯ 5% (Fig. 6B, middle panel). Dependence of the amount of imported tRK1 on the amount of imported pre-MSK had a linear part in the range of 0 -60 fmol of importable pre-MSK and reached the plateau at 90 fmol (Fig.  6B, bottom panel). The plateau can be explained by limiting concentrations of other factor(s) needed for import. Considering the linear part of the curve, one can suggest that imported tRK1 and imported pre-MSK were in equimolar ratio, which suggests that one molecule of pre-MSK directs one molecule of tRK1 into the mitochondria.
We next tested if in HmIDPs there was present the functional analogue of pre-MSK (by this term, we understand a protein which would retain the function of targeting tRK1 into human mitochondria). Anti-MSK antibodies reveal a polypeptide of 65 kDa in human mitochondria (Fig. 6C). This corresponds to the predicted molecular mass of the human mitochondrial LysRS (61). The same polypeptide is recognized in HmIDPs (Fig. 6C). One can suppose that human and yeast The "input" corresponds to the same amount of labeled pre-MSK that was used in the import assay. C, LysRSmt is required for tRK1 import into human mitochondria. Two left panels, Western analyses of human mitochondrial proteins (HM), yeast mitochondrial proteins (SM), and HmIDPs with anti-MSK antibodies. Pure pre-MSK was used as a positive control. Right panel, effect of anti-MSK antibodies onto tRK1 import. Autoradiographs of 5Ј end-labeled tRK1 protected from nucleases after incubation with isolated HepG2 mitochondria in the presence of HmIDPs are presented. Nonspecific antibodies (at the same amount as anti-MSK) were used as a negative control (at the bottom). Dilutions of the antibodies are indicated above the autoradiographs. mitochondrial LysRS share common antigenic determinants. Treatment of HmIDPs with increasing amounts of anti-MSK antibodies resulted in a progressive loss of their capacity to direct tRK1 import into human mitochondria (Fig. 6C, left  panel). One can suggest that the precursor of human mitochondrial lysyl-tRNA synthetase (pre-LysRSmt) is able to fulfill the role of pre-MSK as tRK1 import factor. 5 S rRNA and tRNA Lys Import Is Directed by Distinct Factors-The same HmIDP and ScIDP preparations that direct tRK1 import can also direct import of 5 S rRNA. Although conditions permitting internalization of 5 S rRNA and tRK1 are similar, import of 5 S rRNA was pre-MSK-independent, because, in contrast to the case of tRK1 import, treatment of HmIDPs with increasing amounts of anti-MSK antibodies had no significant effect on 5 S rRNA import (compare Figs. 6C and 7A).
To see if 5 S rRNA and tRK1 import could involve common factors, we fractionated ScIDPs and HmIDPs by differential (NH) 2 SO 4 precipitation and tested fractions for their capacity to direct import of either tRK1 or 5 S rRNA into human mitochondria. The fraction FSc4 of ScIDPs contains pre-MSK (as determined by Western analysis; data not shown), whereas FSc6 was needed to direct import of aminoacylated tRK1 into yeast mitochondria when supplemented either by pre-MSK or by FSc4 fraction (Fig. 7B). This result means that, in FSc4, only pre-MSK is essential for tRK1 import, whereas FSc6 contains all other essential factor(s). As expected, FSc4 alone was unable to direct tRK1 import. On the other hand, FSc4 fraction directs import of 5 S rRNA (Fig. 7B). A similar result was obtained for HmIDPs; FHm4 was able to direct 5 S rRNA import but not tRK1 import, whereas the fraction FHm5 supplemented with pre-MSK directed tRK1 import but not that of 5 S rRNA (Fig. 7B). These results give the evidence that protein factors directing the import of tRK1 and 5 S rRNA into isolated human mitochondria are distinct.
RNA Import Efficiency Depends on the Origin of Protein Factors-As demonstrated above, human cells possess all factors to direct import of either human 5 S rRNA or yeast tRK1 into mitochondria. We compared a panel of RNAs for their capacity of import into human mitochondria in the presence of either HmIDPs or ScIDPs. The first group comprised two natural 5 S rRNAs, one isolated from the yeast S. cerevisiae, and the other from human cells HepG2. The second group represented two natural cytoplasmic tRNA Lys from yeast, tRK1 (imported in vivo) and tRK2 (nonimported in vivo). The third group included two tranzyme versions of yeast tRNA Lys (29). r8 had a sequence of tRK1 with one mutation in the anticodon (C34U) and two mutations in the aminoacceptor stem (G1U, C72A). This version should not be imported in yeast mitochondria (60). r2 had a sequence of tRK2 with one mutation in the anticodon (U34C), which is predicted to confer import capacity to this normally nonimported tRNA (23). The last group included two T7 transcripts with sequences corresponding to the yeast and human mitochondrial tRNA Lys , tr3 and trKhm, correspondingly (Fig. 8A). No protected RNA was detected in the absence of IDPs in any case (not shown). On the other hand, several RNA species were protected with different efficiencies in the presence of either ScIDPs or HmIDPs (Fig. 8, B and C). RNAs and transcripts were verified for their stability in the presence of HmIDPs and ScIDPs, which were found comparable (less than 10% of variation in each case; see Fig. 8D). We also checked if RNAs were protected by proteins present in IDP preparation and found that this was not the case. Taking into account the importance of aminoacylation for tRK1 import, all the tested tRNA versions were aminoacylated with their cognate aminoacyl-tRNA synthetases before the import assay. The level of aminoacylation of the analyzed tRNAs ranged between 50 and 100% (Table I). These control experiments suggest that the difference in import efficiencies was not because of differential protection, degradation, or level of aminoacylation. Therefore, by using equal amounts of IDPs in all assays, we can compare yeast and human factors for their import directing efficiency with respect to different RNA species.
Human 5 S rRNA can be directed into mitochondria either by human or yeast IDPs; however, the efficiency of import directed by HmIDPs was 6 times higher than that of ScIDPs ( Fig. 8B and Table I). In yeast, 5 S rRNA was hypothesized to be nonimported in vivo. Yeast 5 S rRNA differs from the human one in 35-40 (of 121) positions (Fig. 8A); however, it still shares the secondary structure common for all eukaryotic 5 S rRNA (62). We found that yeast 5 S rRNA was weakly imported by HmIDPs, but not by ScIDPs ( Fig. 8B and Table I). On can observe, therefore, that the best efficiency on import in vitro corresponds to the in vivo imported human RNA and protein factors of human origin. tRK1 was internalized both by ScIDPs and HmIDPs (Fig. 8C, Table I). Surprisingly, HmIDPs were even more efficient in directing the yeast tRNA into human mitochondria. This effect can be explained by a less efficient import of the carrier protein of yeast origin (pre-MSK) with respect to the human analogue, pre-LysRSmt. This suggestion is supported by the fact that ScIDPs are more efficient than HmIDPs in directing tRK1 into yeast mitochondria (data not shown). Furthermore, when analyzing pre-MSK import into mitochondria (Fig. 4B), we observed that the amount of imported pre-MSK was 2 times higher in yeast mitochondria than in human ones. One can suggest that pre-MSK directs tRK1 into the human mitochondria less efficiently than the human pre-LysRSmt.
tRK2 was not imported in the conditions used (Fig. 8C, Table  I). In the yeast tRNA mitochondrial import system, the first position of the anticodon (C34) and the first base pair of the aminoacceptor stem (G1:C72) determine mitochondrial import of tRNA Lys (23,60). As expected, introduction of an U1:A72 pair plus U34 in tRK1 (r8) leads to a loss of import directed by either ScIDPs or HmIDPs.
Introduction of U34C mutation in tRK2 confers to this normally nonimported tRNA a capacity of import into yeast mitochondria (23,60). We observed that r2 was also internalized by FIG. 7. tRK1 and 5 S rRNA import into human mitochondria requires distinct factors. A, the effect of anti-MSK antibodies onto 5 S rRNA import, as in Fig. 6C. B, analysis of ScIDPs and HmIDPs fractions for their import directing capacities. Autoradiographs of 5Ј end-labeled RNAs protected from nucleases after incubation with isolated HepG2 mitochondria are presented. When indicated, 20 ng of recombinant pre-MSK was added.
human mitochondria (Fig. 8C). As for tRK1, the efficiency of import was higher with HmIDPs than with ScIDPs, but the difference was more important (Table I). Therefore, differences of pre-MSK and pre-LysRSmt import efficiencies cannot fully explain this result. One can suppose that other nonidentified factor(s) might also influence the import efficiency.
All imported tRNA versions studied previously were based on tRK1 and tRK2 sequences. We tested if mitochondrial tRNA species, normally restricted to the mitochondrial compartment, could be internalized in vitro. We observed that in vitro syn-  c Aminoacylation value corresponds to the maximal level (%) of tRNA that can be aminoacylated by a cognate aminoacyl-tRNA synthetase. KRS was used to aminoacylate tRK1, tRK2, r2, and r8; tr3 was aminoacylated by purified recombinant MSK (mature form) and trKHm with partially purified mitochondrial LysRS. thesized transcripts with the sequences corresponding to the yeast or human mitochondrial tRNA Lys can also be directed into isolated human mitochondria (Fig. 8C, Table I). tr3, corresponding to the yeast mitochondrial tRNA Lys , was imported in the presence of either HmIDPs or (slightly better) ScIDPs. trKhm, corresponding to the human mitochondrial lysine-tRNA, was directed into human mitochondria by HmIDPs but not by ScIDPs. This result cannot be explained by the difference in pre-MSK and pre-LysRSmt import efficiencies, and may reflect different affinities of these RNAs to other import factors.
We exploited the fact that human pre-LysRSmt was recognized by anti-MSK antibodies to compare affinities of the tRNAs to these proteins by using a co-immunoprecipitation assay described previously (23) (MSK binding, Table I). The most important observation is that all tRNA versions with low or absent MSK binding are not imported. This might mean that all the tested versions use pre-LysRS as a targeting factor. Second, all versions of tRNA Lys with MSK binding above a certain threshold level (Ͼ0.2, Table I) were imported. Finally, we did not find direct correlation between the efficiency of MSK binding and efficiency of import. These results suggest that the import efficiency also depends on additional import factors. DISCUSSION RNA import into mitochondria is a quasi-universal process; however, the type and the number of imported RNAs vary among species to a surprisingly significant extent. Even when RNAs of the same type (tRNAs) are concerned, the mechanism of import is also hypothesized to differ (3,21) and, taking into account different approaches used to study different cases, very difficult to compare. The function of mitochondrially imported RNAs is also sometimes difficult to assign. For example, the imported yeast tRNA Lys was, for a long time, thought not to participate in organellar translation. However, our recent studies demonstrated that tRK1 may be a part of mitochondrial protein synthesis apparatus (24). Mitochondrial function of RNA components of RNase P and MRP endonuclease was also a subject of discussion, mostly because of their low concentration in mitochondria (10,15). However, recent experiments proved that the number of molecules of these RNAs associated with the mitochondrial compartment might be sufficient to satisfy RNA processing (13). Import of nuclear encoded 5 S rRNA into mammalian mitochondria was discovered several years ago (16). Mitochondrial content of 5 S rRNA was never carefully quantified; however, it appeared that this RNA was more abundant than RNA components of RNase P of MRP. Mammalian mitochondrial ribosomes were thought to be devoid of 5 S rRNA (19). On the other hand, in several other organisms, mitochondrial ribosomes possess 5 S rRNA. Furthermore, although active on poly(U) templates, isolated mammalian ribosomes were never proved to be functional onto natural mRNAs. All these considerations, done a priori, make difficult the assignment of the function for 5 S rRNA in mammalian mitochondria and the possibility of its association with mitoribosomes remains open. This study aimed (a) to quantify 5 S rRNA import in vivo, as a first approach to its mitochondrial function; (b) to develop a specific in vitro assay, as a first approach of studying the import mechanism; (c) to compare biochemical requirements of 5 S rRNA import with a better understood system of tRNA mitochondrial targeting in yeast.
Our estimations show that the number of 5 S rRNA molecules associated with the mitochondrial compartment of the cell is comparable with the predicted average number of mitoribosomes. This result is not proof that mitochondrial 5 S rRNA participates in constitution of the mitoribosome and only indicates on such a possibility. On the other hand, the number of 5 S rRNA molecules associated with each mitochondrion is high enough to suppose a significant biological role for this RNA in the organelle.
We demonstrated previously that a yeast tRNA, tRK1, can be specifically internalized by isolated human mitochondria (24). Here we describe an assay for 5 S rRNA import in vitro. This internalization was found as specific as in vivo, because no other natural human RNA tested were imported, and its efficiency was comparable with that found in living cells. Therefore, experimental conditions used permit to properly model the import pathway in vitro. 5 S rRNA import required ATP energy, the electrochemical potential across the mitochondrial membrane and soluble protein factors present in a partially fractionated extract of human cells (HmIDPs). Such requirements are similar to those found for tRNA import into yeast mitochondria and to those permitting to deliver a yeast tRNA into human mitochondria (24). This finding is remarkable taking into account a considerable diversity of RNA mitochondrial import mechanisms among species (3,21).
In yeast, tRK1 import depends on pre-protein translocation complexes, including the outer membrane-localized Tom20p and the inner membrane-localized Tim44p proteins (8). Here we show that blocking the GIP complex by a nonimportable analogue of pre-protein (OTC-BPTI) blocks both pre-protein import and RNA import into human mitochondria. This result, taken together with the need of soluble factors, can be interpreted as an involvement of imported pre-proteins in RNA mitochondrial targeting. The commonly agreed paradigm stipulates that proteins are unfolded during translocation across the mitochondrial membranes. We therefore are forced to suppose that either RNA import factors serve only to target the RNA toward the mitochondria, or the carrier protein has strong local contacts with the imported tRNA which are not disrupted by the translocation apparatus. In this context, one can mention the existence of alternative mechanisms for translocation of macromolecules across other cellular membranes that do not require unfolding (63). It may occur that one of these could be implicated in RNA import.
Contrary to an obvious similarity of the requirements for translocation of tRNA and 5 S rRNA, the soluble proteins directing their import are distinct. It is not possible yet to say which protein(s) are targeting 5 S rRNA into mitochondria. There is more evidence in what concerns artificial tRNA import. We found previously that the precursor of the mitochondrial lysyl-tRNA synthetase (pre-MSK) is essential for tRK1 targeting into yeast mitochondria (7). This pre-protein can also function as a targeting/translocation factor for tRK1 import into isolated human mitochondria (24). We show here that interaction of tRK1 and pre-MSK during the import proceeds in an equimolar ratio, which strongly support the hypothesis that the pre-protein serves as a specific carrier for this tRNA. Our results suggest that the precursor of the human mitochondrial LysRS (pre-LysRSmt) could replace pre-MSK. At some extent this result is surprising, because it was shown recently that, in human cells, one gene codes for both mitochondrial and cytosolic LysRS (61). This is an opposite situation with respect to S. cerevisiae, where mitochondrial and cytosolic LysRS are coded by distinct genes, MSK1 and KRS1. All the three proteins, Msk1p, Krs1p, and LysRSmt, belong to the class IIb of aminoacyl-tRNA synthetases (64). As deduced from the crystal structure of the E. coli enzyme, the molecule of a LysRS can be divided into four regions: the N-terminal domain responsible for anticodon recognition, and the C-terminal catalytic domain, which, in turn, may be divided into the first half containing conserved motifs 1 and 2, the insertion domain, and the second half with the conserved motif 3 (65). Multiple sequence alignments show that the catalytic core is highly conserved among known LysRS, whereas N-terminal domains and insertion parts of C-terminal domains show very low similarity among species (66). Comparison of potential secondary structures and sequence alignments of Msk1p, Krs1p, and LysRSmt with the LysRS of E. coli by 3D-PSSM program (67) shows that the percentage of identical amino acids and the SAWTED E-values (68) were closer in the pair Krs1p/LysRSmt for fulllength proteins and C-terminal domains, and in the pair Msk1p/LysRSmt for N-terminal domains (Table II). This fact indicates that the N-terminal domain of pre-MSK and pre-LysRSmt, which is dispensable for aminoacylation activity (66), may be involved in tRK1 import. It is noteworthy that, in vivo, none of the cytoplasmic human tRNA Lys are imported into mitochondria. Therefore, the particular way of RNA-protein interaction leading to tRK1 import and involving yeast or human mitochondrial LysRS does not concern each tRNA Lys species. Among several tested natural RNAs, 5 S rRNA, tRK1, tRK2, human tRNAs (in this study), yeast tRNA i Met , tRNA Phe of E. coli (24), only species imported in vivo were well internalized by isolated mitochondria, tRK1 (imported in yeast), and human 5 S rRNA (imported in human cells). On the other hand, testing synthetic derivatives of other tRNA Lys revealed that some additional versions could also be imported into human mitochondria. We demonstrated here that not only derivatives of yeast cytoplasmic tRNA Lys can be imported (as expected from previous experiments with yeast mitochondria), but also derivatives of mitochondrial tRNAs (yeast and human tRNA Lys ) or yeast 5 S rRNA. The efficiency of import depended on interaction with targeting proteins and on import efficiency of these proteins. The situation was similar to that found in other organisms where various mutant versions of imported or even nonimported RNAs could also be targeted into mitochondria (3). It becomes clear that the pathway of RNA targeting into mitochondria exhibits significant flexibility, even in organisms importing a very restricted number of RNA species. This flexibility and the availability of in vitro models open a possibility to set up a system of in vivo targeting of engineered RNAs into human mitochondria. Such a system may have potential biomedical use, because a large number of neuromuscular diseases are associated with mutations in mitochondrial DNA (69,70). It would be attractive to use the RNA import to correct the effect of these mutations. The feasibility of such system is supported by our recent finding (24) that, in yeast, a mutation in mitochondrial DNA can be suppressed by the tRNA imported from the cytoplasm.