Stepwise Transfer of tRNA through the Double Membrane ofLeishmania Mitochondria*

Import of tRNA into Leishmaniamitochondria involves transfer through a double membrane barrier. To examine whether specific sorting mechanisms for individual tRNAs direct them to different mitochondrial compartments, the distribution of tRNA transcripts, internalized in vitro, was examined by suborganellar fractionation. Significant amounts of tRNATyrwere localized in the matrix and on the outer face of the inner mitochondrial membrane. With time, the matrix:membrane ratio increased. Translocation through the inner membrane apparently required the presence of a specific signal in the D arm of tRNATyr, and tRNAGln(CUG), lacking this sequence, was excluded. Hydrolysis of ATP was necessary at both the outer and inner membranes. However, the protonophores carbonylcyanidem-chlorophenylhydrazone and nigericin, the K+ionophore valinomycin, and the F1F0 ATPase inhibitor oligomycin had only marginal effects on uptake through the outer membrane but severely inhibited inner membrane translocation, indicating the unusual requirement of both the electrical and chemical components of the electromotive force generated across the inner membrane. The results are consistent with a mechanism involving stepwise transfer of tRNA through distinct outer and inner membrane channels.

Import of tRNA into Leishmania mitochondria involves transfer through a double membrane barrier. To examine whether specific sorting mechanisms for individual tRNAs direct them to different mitochondrial compartments, the distribution of tRNA transcripts, internalized in vitro, was examined by suborganellar fractionation. Significant amounts of tRNA Tyr were localized in the matrix and on the outer face of the inner mitochondrial membrane. With time, the matrix:membrane ratio increased. Translocation through the inner membrane apparently required the presence of a specific signal in the D arm of tRNA Tyr , and tRNA Gln (CUG), lacking this sequence, was excluded. Hydrolysis of ATP was necessary at both the outer and inner membranes. However, the protonophores carbonylcyanide m-chlorophenylhydrazone and nigericin, the K ؉ ionophore valinomycin, and the F 1 F 0 ATPase inhibitor oligomycin had only marginal effects on uptake through the outer membrane but severely inhibited inner membrane translocation, indicating the unusual requirement of both the electrical and chemical components of the electromotive force generated across the inner membrane. The results are consistent with a mechanism involving stepwise transfer of tRNA through distinct outer and inner membrane channels.
The mitochondrial genomes of many protozoal species, including leishmania, trypanosomes, and plasmodium, are unusual for their apparently total lack of tRNA genes (1)(2)(3). To sustain the translation of organellar mRNAs, a large number of nuclear-encoded tRNAs are imported from the cytoplasm (4 -10). Mitochondrial import of one or more tRNAs has also been documented in tetrahymena (11), yeast (12), and several species of higher plants (13,14).
To understand the mechanism of import, we have developed an in organello system from leishmania (15). It was shown that import is ATP-dependent and specific for RNA sequence (15,16). A purine-rich oligonucleotide motif in the D arm of tRNA was identified as an import signal in vivo (17) as well as in vitro (18). A 15-kDa protein associated with the outer mitochondrial membrane acts as carrier or receptor for direct import of tRNA, without the mediation of soluble factors (19).
Analysis of intact mitochondria or mitoplasts for tRNA internalized in vivo or in vitro does not address the question of the distribution of the RNA in the various intramitochondrial compartments, viz. the outer membrane, intermembrane space, inner membrane, or matrix. We have previously reported UTP labeling of in vitro imported small RNAs (15), presumably by the matrix-localized terminal uridylyl transferase (20). A fraction of tRNA Ile associated with the mitochondrial fraction in vivo was found to be resistant to micrococcal nuclease in the presence of digitonin, which selectively solubilizes the outer mitochondrial membrane, suggesting the matrix location of this tRNA (17). It is not known whether different tRNAs are sorted to different locations once inside the mitochondrion, in a manner analogous to the sorting of imported proteins, and if so, whether separate matrix localization and sorting signals exist within the tRNA structure. The study of intramitochondrial distribution is important as it may enable a distinction to be made between 1) a concerted mechanism of import through a single transport channel spanning both the outer and inner membranes, and 2) a stepwise mechanism involving separate outer and inner membrane translocation channels.
In this study, mitochondria were fractionated into various compartments for the study of the distribution of imported RNA, specifically, the sequence specificity and energy requirements of translocation through the inner membrane. The results support the notion of stepwise insertion of RNA through the outer and inner mitochondrial membranes.
Enzymes and Inhibitors-Carbonylcyanide m-chlorophenylhydrazone (CCCP), oligomycin, valinomycin, and nigericin (all from Sigma) were dissolved in ethanol and diluted 100-fold into import reactions. The sodium salt of carboxyatractyloside (Sigma) was dissolved in water and similarly diluted. Mitochondria (80 -100 g of protein) were preincubated with inhibitor (plus 0.1 M KCl in the case of valinomycin) for 15 min on ice before dilution with an equal volume of import buffer, as indicated. Rabbit muscle myokinase (Roche Molecular Biochemicals), an ammonium sulfate suspension, was recovered by microcentrifugation and suspended in 10 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 1 mM EDTA, 50% glycerol before use. Import Incubation-Unless otherwise stated, washed mitochondria (80 -100 g of protein) were incubated at 37°C for 15 min with 100 fmol of 32 P-labeled import substrate in 10 mM Tris-HCl, pH 7.5, 10 mM MgAc 2 , 2 mM dithiothreitol, and 4 mM ATP in a total volume of 20 l. Then RNase A (2.5 g/ml) and RNase T1 (50 units/ml) were added and incubation continued for an additional 15 min at the same temperature. Mitochondria were washed in cold STE and recovered by centrifugation.
Submitochondrial Fractionation-RNase-treated mitochondria containing internalized radiolabeled RNA were incubated with 320 M digitonin (Roche Molecular Biochemicals) in 10 l of STE for 15 min on ice (23). Alternatively, mitochondria were subjected to hypotonic shock in 20 l of 1 mM Tris-HCl, pH 8.0, 1 mM EDTA (15) for 15 min on ice, followed by addition of 2.5 l of 2 M sucrose. Mitoplasts were separated from the intermembrane space fraction by microcentrifugation at 3450 ϫ g for 3 min at 4°C. The mitoplasts were suspended in 10 l of freeze-thaw buffer (0.6 M sucrose, 10 mM Tris-HCl, pH 7.5, 1 mM EDTA; Ref. 24) and subjected to three freeze-thaw cycles, each consisting of freezing at Ϫ70°C followed by rapid thawing at 37°C. The inner membrane and matrix fractions were separated by centrifugation at 7000 ϫ g for 5 min at 4°C. RNA was recovered from each fraction by guanidinium isothiocyanate extraction and isopropanol precipitation (18). To quantify the amount of RNA, an aliquot was spotted on DEAE anion exchange paper (DE 81, Whatman) and counted. Alternatively, the RNA was resolved by urea-PAGE (15), followed by counting of the dried gel band.
Enzyme Assays-Kynurenine hydroxylase and succinate dehydrogenase were spectrophotometrically assayed as described previously (16,19). To assay malate dehydrogenase, reactions (1 ml) containing 96 mM potassium phosphate, pH 7.4, 150 M NAD ϩ , and mitochondrial extract were initiated by the addition of 260 M sodium malate and the reduction of NAD ϩ followed by the increase in absorbance at 340 nm (⑀ ϭ 6.22 mM Ϫ1 cm Ϫ1 ).
Measurement of Membrane Potential-The uptake of rhodamine 123 was used as a measure of the mitochondrial membrane potential (25,26). Reactions (1 ml) containing 10 mM Tris-HCl, pH 8.0, 10 mM MgAc 2 , 1 mM dithiothreitol, 220 mM sucrose, 1 mM EDTA, and 2.63 M rhodamine 123 (Sigma) were placed in a cuvette, and the absorbances at 516 and 495 nm were measured. Mitoplasts (derived from 200 g of mitochondria), 4 mM ATP, and 50 M CCCP were sequentially added, and the absorbances at the above wavelengths continuously recorded over time.

RESULTS
Optimization of the Ribonuclease Protection Assay-Previous import assays (15,16,18,19) had employed relatively high concentrations of RNase and a suboptimal incubation temperature (25°C) to digest excess (unimported) RNA. Under these conditions we sometimes observed partial degradation of imported RNA, possibly by residual RNase during the postimport washing and mitochondrial lysis steps. 2 In order to make the assay more reliable in terms of RNA intactness and yield, the RNase concentration was lowered, whereas the incubation temperature was raised to 37°C to increase the rate of degradation. A RNase titration experiment (Fig. 1A) showed that at 37°C, protection of intact RNA was observed using only 2.5% of the original enzyme concentration. At lower concentrations, there was evidence of incomplete endonucleolytic cleavage leading to smearing in the lane, whereas at higher concentrations, the amount of protected RNA was reduced and finally eliminated. These results emphasize the importance of controlling the RNase step for successful observation of import.
The amount of protected RNA increased in a time-dependent manner up to 20 min of incubation of the mitochondria with radiolabeled tRNA Tyr and was completely sensitive to RNase if the mitochondria were lysed with Triton X-100 ( Fig. 1B), as expected for uptake into the membrane-bound organelle.
Separation of Mitochondrial Compartments-In order to assess the intramitochodrial distribution of tRNAs imported in vitro, a simple scheme for separating the various intramitochondrial compartments was developed ( Fig. 2A). Mitochondria were treated with digitonin, a detergent that selectively solubilizes the outer membrane (23). Under the conditions employed, the outer membrane marker kynurenine hydroxylase (19), but not the inner membrane marker succinate dehydrogenase (19), was solubilized (Table I), attesting to the selectivity of the digitonin treatment.
After separation of the digitonin-soluble fraction (outer membrane plus intermembrane space) by centrifugation, the particulate fraction (mitoplasts), containing membrane vesicles of more or less uniform size, 2 was subjected to freeze-thaw cycles (24) to liberate the matrix contents, leaving behind an insoluble fraction enriched with the inner membrane. Light microscopy of this fraction revealed the presence of large, presumably fused, membrane lamellae, with no evidence of intact mitoplasts. 2 As expected, malate dehydrogenase was predominantly located in the soluble fraction and succinate dehydrogenase in the particulate fraction (Table I). (Some malate dehydrogenase activity was also detected in the digitonin-soluble fraction; a similar result has been obtained with mammalian mitochondrial preparations (23) and probably represents contamination with the cytosolic form of the enzyme.) The mitochondrial enzyme terminal uridylyl transferase (TUTase), implicated in RNA editing mechanisms (20), was also released by freeze-thaw lysis, confirming its presence in the matrix. 2 Treatment of mitochondria with digitonin resulted in enhanced sensitivity of inner membrane proteins, such as succinate dehydrogenase, to protease treatment, whereas the ma-trix marker malate dehydrogenase remained unaffected (Fig.  2B), indicating the intactness of the inner membrane.
To determine whether the mitoplasts obtained by this procedure can develop a membrane potential (⌬), digitonintreated preparations were incubated with rhodamine 123. The uptake of this lipophilic cation into energized mitochondria is strictly dependent upon the presence of ⌬ across the inner membrane and is accompanied by a red shift in the absorption spectrum (25,26). When mitoplasts were added to rhodamine 123, a rapid and transient red shift was observed (Fig. 2C), due to incomplete energization by endogenous substrates (26). Addition of ATP resulted in a more sustained red shift (Fig. 2C), indicating dye uptake into the mitoplasts. Further addition of CCCP, a protonophore that dissipates ⌬, reversed the red shift (Fig. 2C), confirming the development of an ATP-induced proton gradient across the mitoplast membrane.
Intramitochondrial Distribution of tRNA Tyr -Mitochondria were incubated with radiolabeled tRNA Tyr transcript, then digested with RNase. After washing, submitochondrial fractions were separated as above and analyzed for the presence of RNA.
In the presence of ATP, mitochondria accumulated internalized tRNA Tyr in different submitochondrial compartments (Fig.  3A). The majority of the RNA (68% in this experiment) was in the matrix, 31% was bound to the inner membrane, and less than 1% was in the intermembrane space. RNase treatment of the digitonin-treated mitoplasts resulted in nearly complete digestion of the membrane-bound RNA, whereas the RNA in the matrix was resistant (Fig. 3A), as expected. Thus, the RNA bound to the inner membrane is exposed to the intermembrane space. Essentially identical results were obtained when mitoplasts were prepared by hypotonic shock instead of digitonin. 2 The intramitochondrial distribution of RNA changed with time ( Fig. 3B). At early times of incubation, the majority of the RNA was in the inner membrane fraction, whereas after 10 min at 37°C, increasing amounts appeared in the matrix, with a concomitant reduction in the membrane-bound fraction, whereas the total amount internalized increased only marginally. These results indicate that translocation through the inner membrane is slower than that through the outer membrane.
The effect of temperature on the intramitochondrial distribution of tRNA Tyr was examined (Fig. 3C). Total uptake increased sharply between 5 and 15°C, but rose only marginally thereafter. In contrast, transfer into the matrix was favored at 35 and 45°C. Near the optimum growth temperature of the promastigote form of Leishmania (25°C), the RNA was equally distributed between the inner membrane and matrix compartments (Fig. 3C), reflecting a reduced rate of matrix transfer at the lower temperature. 2 Sequence Discrimination at the Inner Membrane-It was pre-  viously shown that uptake of RNA through the outer membrane (as measured by RNase protection assays of intact mitochondria) requires recognition of a conserved motif with the consensus sequence UGGYAGRRY in the D arm of tRNA (see Fig. 4) by the outer membrane receptor, TAB (18,19). To examine whether such a sequence is sufficient for transfer into the matrix, a deletion derivative of tRNA Tyr , containing the 5Ј-terminal 39 nucleotides, including the D arm, was tested for its intramitochondrial distribution. Entry of tRNA Tyr (1-39) into the matrix was better than that of the parental molecule (Fig. 4, A ad B). The time course of intramitochondrial distribution was also similar (Fig.  5). At early times of incubation, almost all of the RNA was bound to the inner membrane, whereas after 10 min at 37°C, the matrix content increased and the amount of membrane-associated RNA decreased. This kinetic pattern is consistent the presence of an intermediate, inner membrane-bound state. Recently, we have observed 2 efficient matrix localization of an oligoribonucleotide containing only nucleotides 5-27 of tRNA Tyr , i.e. the D arm hairpin loop (18). The outer membrane system discriminates between tRNA Tyr , which is imported in vivo (7) and contains the conserved D arm sequence (18), and tRNA Gln (CUG), which is not detectable in mitochondrial RNA (7) and lacks the conserved D arm signal (16) (see Fig. 4). However, low levels (about 25% of that of tRNA Tyr ) of transfer of tRNA Gln (CUG) through the outer membrane could still be observed in vitro (19), indicating that this discrimination is not absolute. When the intramitochondrial distribution of the internalized tRNA Gln was examined, 90 -95% was found to be confined to the inner membrane; little or no RNA was located in the matrix (Fig. 4C). Similarly, very low levels of ␦ RNA, a synthetic transcript that lacks the import signal and is transferred to an insignificant extent through the outer membrane (18), were found to be exclusively on the inner membrane (Fig. 4D).
These results indicate that 1) the D arm signal is sufficient to penetrate the inner membrane, and 2) the inner membrane has an even greater selectivity for RNA sequence than the outer membrane.
Requirement of ATP Hydrolysis at the Inner and Outer Membranes-Total uptake of tRNA Tyr increased with the concentration of added ATP until saturation was reached at 3-4 mM (Fig.  6A). The nonhydrolyzable analogs AMPPCP (Fig. 6A) and AMPPNP 2 were unable to replace ATP, illustrating the requirement of cleavage of the ␤-␥ pyrophosphate bond.
To determine the sites of ATP action, mitochondria were incubated with RNA and ATP in the presence of AMP and the nonpenetrant enzyme myokinase, to scavenge the external ATP via the reaction ATP ϩ AMP 3 2ADP. This treatment resulted in inhibition of outer membrane transfer (Fig. 6B, left  panel). When mitochondria were preloaded with ATP and then incubated with RNA in the absence of further ATP addition, normal levels of outer membrane transfer were observed, but myokinase inhibited this process, indicating that some or all of the ATP must be exported from the matrix or intermembrane space in order to sustain transfer (Fig. 6B, left).
When mitochondria were incubated with carboxyatractyloside, a specific inhibitor of the adenine nucleotide translocator on the inner membrane (27), and then ATP and RNA were added, import was inhibited (Fig. 6B, right). This suggests that accumulation of ATP in the matrix via the adenine nucleotide translocator is important. Mitochondria preloaded with ATP, washed, and subsequently challenged with RNA in the presence of carboxyatractyloside were also inhibited (Fig. 6B, right). Because external ATP is absent and matrix ATP is prevented from translocating to the intermembrane space by carboxyatractyloside, this result is consistent with the requirement of ATP in the washed and incubated with RNA and import buffer (no ATP) for 45 min at 25°C in the absence (bar 4) or presence (bar 5) of AMP and myokinase. In the right panel, mitochondria were preincubated without (bar 1) or with (bar 2) 100 M carboxyatractyloside and then diluted to twice the volume with RNA, import buffer, and 4 mM ATP for import incubation. The other reactions were performed with ATP-preloaded mitochondria in the absence (bar 3) or presence (bar 4) of 50 M carboxyatractyloside. In each case, total RNA uptake was measured by the DEAE paper method. C, the role of ATP hydrolysis at the inner membrane. Mitochondria were incubated with 100 fmol of 32 P-labeled tRNA-Tyr for 15 min at 37°C in the absence or presence of 4 mM ATP or 4 mM AMPPCP. After RNase treatment, mitochondria were fractionated into intermembrane space (hatched bars), inner membrane (open bars), and matrix (filled bars) fractions. RNA in each fraction was determined by urea-PAGE followed by gel band counting and expressed as a percentage of the total amount internalized. The total uptake values obtained in the absence of ATP, in the presence of ATP, and in the presence of AMPPCP, were 0.39, 1.62, and 0.34 fmol, respectively. intermembrane space or at the outer membrane.
The effect of ATP hydrolysis on the intramitochondrial distribution of tRNA Tyr was examined (Fig. 6C). Addition of ATP caused a 7-fold increase in the level of RNA in the matrix but only a 1.6-fold increase in the level of inner membrane-bound RNA. In the presence of AMPPCP, translocation into the matrix was reduced to negligible levels, whereas 84% remained associated with the inner membrane. Taken together, these results demonstrate the requirement of ATP hydrolysis for transfer of RNA through both the outer and inner membranes.
Role of the Electromotive Force across the Inner Membrane-Many mitochondrial transport processes are energetically driven by the electromotive force across the inner membrane generated by vectorial proton movement coupled to respiratory electron transport. These include translocation of proteins (28,29) and of various low molecular weight metabolites (27). A conceivable role of ATP in RNA import is to generate a proton gradient by the hydrolytic activity of the oligomycin-sensitive F 1 F 0 ATPase, which subsequently drives transfer through the membrane.
This possibility was tested by observing the effect of various respiratory inhibitors on ATP-dependent total uptake, as well as the intramitochondrial distribution of tRNA Tyr . Initial titrations with the protonophore CCCP, which dissipates the proton gradient (30) and hence the electromotive force, revealed only insignificant inhibition (less than 2-fold) of total uptake even at high inhibitor concentrations (up to 200 M), 2 suggesting that outer membrane transfer does not require an electromotive force. However, more than 90% of the internalized RNA was restricted to the inner membrane, very little being found in the matrix (Fig.  7A); the membrane-bound form was located on the outer surface, exposed to the intermembrane space. 2 Clearly, insertion through the inner membrane requires a proton gradient.
Oligomycin, at a concentration of up to 50 M, which is 50 -100 times the dose for 50% inhibition of leishmania mitochondrial ATPase (31,32), could inhibit total uptake about 2-fold, but it caused a 10-fold reduction in inner membrane translocation (Fig. 7B), indicating a role of the F 1 F 0 ATPase in transfer to the matrix.
The electromotive force consists of two components: the membrane potential (⌬), caused by charge separation across the membrane, and the pH gradient (⌬pH), due to the difference in proton concentration on the two sides (30). In the presence of valinomycin, a K ϩ ionophore that reduces ⌬ but not ⌬pH (30), transfer of tRNA Tyr into the matrix was inhibited (Fig. 7B). Nigericin, which carries out an electroneutral K ϩ -H ϩ exchange, thereby reducing ⌬pH without any effect on ⌬ (30), had a similar inhibitory effect (Fig. 7B). The results suggest that both the electrical and chemical components of the electromotive force are necessary for translocation of tRNA through the inner membrane. DISCUSSION We describe here a modified import assay that monitors the distribution of internalized RNA in different submitochondrial compartments. Several modifications of previous assay methods (15,16,18,19) were introduced. The import incubation was carried out at 37 instead of 25°C. At the higher temperature, the rate but not the yield of import in vitro was higher, and entry into the matrix was facilitated (Fig. 3). The broad temperature range over which import is active may be a reflection of the adaptability of the parasite to grow at different temperatures as it passes through invertebrate and mammalian hosts. The postimport RNase treatment was also carried out at 37°C using only 2.5% of the concentration of RNase previously employed (Fig. 1); these conditions were sufficient to digest excess RNA and at the same time minimized the risk of RNA degradation during subsequent fractionation steps. Third, the combination of digitonin and freeze-thaw lysis of mitoplasts allows the intramitochondrial compartments to be rapidly and easily separated (Fig. 2) without the need for prolonged ultracentrifugation steps.
Selective permeabilization of the outer membrane with digitonin or by hypotonic shock revealed the presence of a significant fraction of internalized tRNA associated with the outer side of the inner membrane, i.e. facing the intermembrane space (Fig. 3). ible with concerted models of import that predict translocation intermediates spanning both membranes and enclosed in a transport pore consisting of inner and outer membrane proteins. This type of general insertion pore located at inner membraneouter membrane contact sites conducts cytoplasmic proteins into the mitochondrial matrix (33).
In concerted models of import, the responsibility for selection of the correct import substrate lies with receptors located at the outer membrane, which recognize appropriate import signals. Thus, misrecognition at this step should lead to entry of the inappropriate substrate into the matrix. Our previous studies showed that the outer membrane system can discriminate between tRNA Tyr and tRNA Gln (CUG), but not absolutely (19). However, the residual level of tRNA Gln passing through the outer membrane is almost exclusively restricted to the inner membrane (Fig. 4). In contrast, derivatives of tRNA Tyr retaining the conserved D arm import signal (18) pass into the matrix at least as effectively as the entire molecule (Fig. 4). Because this matrix targeting sequence interacts with TAB, a protein located on the outer membrane (18,19), one possibility is that the TAB-D arm complex passes through the outer membrane and subsequently interacts with the inner membrane channel for matrix entry. According to this scenario, RNAs, such as tRNA Gln , that do not interact with TAB on the mitochondrial surface (18) would be effectively excluded from the matrix.
Although ATP hydrolysis is required for translocation of RNA through both the outer and inner membranes (Fig. 6), the roles of ATP at these two steps are different. Notably, inhibition of the proton-pumping F 1 F 0 ATPase with oligomycin, or dissipation of the proton gradient with CCCP or nigericin, results in severe inhibition of inner membrane translocation but only a marginal effect on transfer through the outer membrane (Fig. 7). Because nigericin, through electroneutral K ϩ -H ϩ exchange, affects only the chemical component (⌬pH) of the electromotive force (30), it is likely that transfer of RNA through the inner membrane is driven by the excess proton concentration in the intermembrane space. Symport of polyanionic RNA with protons would be an effective charge-balancing mechanism necessary to pump the RNA against the membrane potential (negative inside) of respiring mitochondria. An analogous example is that of the inner membrane phosphate transporter, which is coupled to the proton gradient by P i -H ϩ symport (27).
The inhibition of inner membrane translocation by valinomycin (Fig. 7), which neutralizes the K ϩ gradient and thus reduces membrane potential (⌬), exposes the paradoxical situation that a negatively charged membrane is required for transfer of negatively charged RNA. Several possibilities may be considered to explain this. 1) Import of RNA is coupled to export of some other anion such that the process is electrogenic, with the matrix losing net negative charge. An example of this type of transport is the ⌬-requiring exchange of matrix ATP for cytoplasmic ADP catalyzed by the adenine nucleotide translocator (27). 2) RNA is complexed with a positively charged protein factor that moves electrophoretically into the negatively charged matrix. This mechanism has been invoked to explain the transfer of the amphipathic helix constituting the import signal of matrix-targeted proteins (33).
Whereas most inner membrane transport processes, including that of proteins, require either the ⌬ or ⌬pH component of the electromotive force, transport of tRNA may be unusual in requiring both. This is probably a consequence of the high negative charge of the molecule as well as of its sequencespecific interaction with a positively charged species. According to this hypothesis, ⌬ would be required for initiation, and ⌬pH for completion of transfer of the phosphodiester backbone through the inner membrane.
The co-import model of mitochondrial tRNA uptake (34) is a concerted mechanism involving the passive transfer of RNA bound to a matrix-targeted carrier protein through the general insertion pore. However, the results presented here support a stepwise mechanism of import of tRNA through the two mitochondrial membranes. The first step consists of TAB-D arm recognition at the outer membrane followed by ATP-dependent transfer to the intermembrane space. The RNA then rapidly associates with the outer surface of the inner membrane. We assume that inner membrane binding is fast, because significant accumulation of RNA in the intermembrane space fraction was not observed, even at early times of incubation (Figs. 3 and  5). Subsequently, the RNA (possibly as an RNP complex) is transferred to the matrix by a process requiring the D arm signal as well as both the chemical and electrical components of the electromotive force. Attempts to define the biochemical components of the membrane-bound machinery are currently in progress.