Import of proteins into mitochondria. Energy-dependent uptake of precursors by isolated mitochondria.

The post-translational import of cytoplasmically synthesized precursors of proteins of the mitochondrial matrix and inner membrane has been studied in vitro with isolated yeast mitochondria and reticulocyte lysates programmed with yeast mRNA. Import of precursors into isolated mitochondria is not a nonspecific side reaction: up to 70% of the in vitro-synthesized precursor of the  &subunit of FIATPase can be processed to the mature form and rendered protease-resistant during incubation with mitochondria. Furthermore, precursors which were processed to their mature form prior to incubation with mitochondria do not become protease-resistant under similar conditions. Finally, a brief trypsin treatment of mitochondria prior to incubation with labeled F1 psubunit precursor abolished uptake of this precursor into the mitochondria. In vitro import occurs into the correct intramitochondrial location: this was determined by the use of chelators such as GTP and bathophenanthroline disulfonate which inhibit the cleavage of precursors to their mature form by a hypotonic extract from yeast mitochondria or by lysed mitochondria. The processing of these precursors by intact mitochondria, however, is not affected by such inhibitors. Since these inhibitors can cross the outer, but not the inner, mitochondrial membrane, at least part of the precursor molecule is translocated across an intact inner membrane before cleavage to the mature form occurs. Subfractionation of mitochondria after in vitro import provides further evidence for the proper localization of proteins imported into the mitochondrial matrix, intermembrane space, and membrane fraction. Import of precursors into isolated yeast mitochondria thus exhibits the topological specificity of the process as it occurs in living cells. The in vitro import of precursor polypeptides requires energy: in respiring mitochondria, import is blocked either by protonophores or by valinomycin plus potassium, but not by oligomycin; in cyanide-inhibited mitochondria supplemented with external ATP, import is blocked by carboxyatractyloside or by oligomycin. This shows that an electrochemical gradient across the inner mitochondrial membrane, and not ATP itself, is required for protein import.

The post-translational import of cytoplasmically synthesized precursors of proteins of the mitochondrial matrix and inner membrane has been studied in vitro with isolated yeast mitochondria and reticulocyte lysates programmed with yeast mRNA. Import of precursors into isolated mitochondria is not a nonspecific side reaction: up to 70% of the in vitro-synthesized precursor of the &subunit of FIATPase can be processed to the mature form and rendered protease-resistant during incubation with mitochondria. Furthermore, precursors which were processed to their mature form prior to incubation with mitochondria do not become protease-resistant under similar conditions. Finally, a brief trypsin treatment of mitochondria prior to incubation with labeled F1 psubunit precursor abolished uptake of this precursor into the mitochondria.
In vitro import occurs into the correct intramitochondrial location: this was determined by the use of chelators such as GTP and bathophenanthroline disulfonate which inhibit the cleavage of precursors to their mature form by a hypotonic extract from yeast mitochondria or by lysed mitochondria. The processing of these precursors by intact mitochondria, however, is not affected by such inhibitors. Since these inhibitors can cross the outer, but not the inner, mitochondrial membrane, at least part of the precursor molecule is translocated across an intact inner membrane before cleavage to the mature form occurs. Subfractionation of mitochondria after in vitro import provides further evidence for the proper localization of proteins imported into the mitochondrial matrix, intermembrane space, and membrane fraction. Import of precursors into isolated yeast mitochondria thus exhibits the topological specificity of the process as it occurs in living cells.
The in vitro import of precursor polypeptides requires energy: in respiring mitochondria, import is blocked either by protonophores or by valinomycin plus potassium, but not by oligomycin; in cyanide-inhibited mitochondria supplemented with external ATP, import is blocked by carboxyatractyloside or by oligomycin. This shows that an electrochemical gradient across the inner mitochondrial membrane, and not ATP itself, is required for protein import.
8 To whom requests for reprints should be sent.
Most mitochondrial proteins are coded by nuclear genes, translated on cytoplasmic ribosomes, and imported into mitochondria (1,2). Many of these polypeptides are initially synthesized as larger precursors which can be detected by pulse-labeling whole cells or by translating isolated mRNA in a reticulocyte lysate (1-4). These in vitro-synthesized precursors can be cleaved to their corresponding mature form and rendered insensitive to externally added proteases during incubation with intact mitochondria (1-3). Translocation across mitochondrial membranes and processing to the mature size can thus occur post-translationally. This uptake mechanism has been termed "vectorial processing" (1) to set it apart from the obligately co-translational transport of secretory proteins across the endoplasmic reticulum ("vectorial translation").
Despite the initial evidence for in uitro import by yeast mitochondria, it remained uncertain whether or not this process resembles protein import into mitochondria in uiuo. In order to resolve this point, the following questions had to be answered. Is in vitro uptake efficient or merely an insignificant side reaction? Does it exhibit the characteristic energy dependence which is typical of the import process in intact yeast cells (5)? Are the precursors transported to their correct intramitochondrial location? Can mitochondria take up precursors that have been processed before being incubated with the mitochondria?
This study has attempted to answer these questions. First, a convenient import assay was developed which eliminated the need for subsequent protease treatment of the isolated mitochondria. Second, the specificity of import was assessed by a newly developed procedure for fractionating yeast mitochondria into their major compartments (6). Third, import of precursors into the matrix space was probed by a method which does not require subfractionation of mitochondria. Fourth, the energy dependence of import was studied under conditions in which the only energy donor was added ATP.
The results of this work indicate that the import of precursor polypeptides into isolated mitochondria reflects the import process as it occurs in living yeast cells. A study of this in vitro system has yielded new information on the import mechanism.

MATERIALS AND METHODS
Yeast Strain and Fractionation-The wild type Saccharomyces cereuisiae strain D273-10B (a, ATCC 25657) was used throughout this study. For the preparation of mHNA and mitochondria, the cells were grown to early logarithmic phase in a semisynthetic medium containing 2% lactate and 0.1% glucose (6). The isolation of crude mRNA (3), spheroplasts (S), mitochondria (6). and mitoplasts (6) has been described.
A crude preparation of matrix protease was prepared from yeast mitochondria essentially as reported (7) and partially purified as will he described elsewhere.' Subfractionation of yeast mitochondria was done as outlined in the accompanying paper ( 6 ) .
Cell-free Protein Synthesis-Cell-free protein synthesis was programmed by crude yeast mRNA in a rabbit reticulocyte lysate (8) untreated with nuclease. Protein synthesis was stopped by chilling to 0 "C and centrifugation (143,000 X g, 40 min) to remove polysomes. Aliquots of translated lysate were frozen in liquid N, and stored at -80 "C. Prior to incubation with mitochondria, aliquots were filtered through a Sephadex G-25 column which had been equilibrated with 0.15 M KCI, 20 mM Hepes2/KOH, pH 7.4. For processing precursors in the lysate prior to incubation with mitochondria, a partially purified hypotonic extract of yeast mitochondria' was incubated with the translated lysate for 20 min at room temperature. Ortho-phenanthroline was then added to 2 mM in order to inhibit further processing (7) and the lysate was filtered.
I n Vitro Import Conditions-The in vitro import assay contained The mixture was incubated for 30 min at 27 "C with shaking and then chilled on ice. For protease treatment, the sample was divided in half: one-half received trypsin (final concentration, 120 pg/ml) and the other half 1 mM TLCK and 1 mM phenylmethylsulfonyl fluoride. After 30 min at 0 "C, trypsin activity was arrested by the addition of 1.2 mg/ml of soybean trypsin inhibitor and 1 mM TLCK. Mitochondria were then reisolated by centrifugation (l0,OOO X g for 10 min) and the pellet as well as the supernatant were dissociated in 3% SDS, 1 mM phenylmethylsulfonyl fluoride for 3 min at 95 "C. The dissociated samples were diluted 14-fold with Triton buffer (1% Triton X-100,150 mM NaCI, 5 n I M EDTA, 50 mM Tris-CI, pH 8.0) and subjected to quantitative immunoprecipitation. Incubations with a hypotonic extract or with lysed mitochondria were as described above except that the total mixture was dissociated at once after the 30-min incubation a t 27 "C. In experiments in which the effect of inhibitors on import was tested, the mitochondria (10 mg/ml) were pretreated with the inhibitors for 10 min on ice. The final inhibitor concentration stated under "Results" was then obtained once the other components of the incubation were added. Trypsin treatment of mitochondria prior to import was done with either 4 or 10 pg of trypsin/ml and 5 mg of mitochondria/ml for 10 min a t 0 "C and stopped by the addition of 100 pg/ml of soybean trypsin inhibitor and 1 mM TLCK. The mitochondria were then reisolated and washed by centrifugation (l0,ooO x g for 10 min). As a control, mitochondria were incubated with trypsin that had been pretreated with trypsin inhibitor.
Quantitative Immunoprecipitation-After dissociation in SDS, each supernatant from the in vitro import assay and an aliquot of unincubated lysate was mixed with 200 pg of SDS-dissociated unlabeled yeast mitochondrial protein. This ensured that all samples subjected to immunoprecipitation contained the same amount of any mitochondrial antigen which might compete with the labeled polypeptides for binding the immunoglobulins. Each sample (10-15 m l ) was mixed with an equal volume (50-70 pl) of antiserum, shaken for 16 h a t 4 "C, and then mixed with an equal volume (250-350 pl) of a 10% w/v suspension of glutaraldehyde-fixed Staphylococcus aureus cells (9). Washing and elution of the immunoprecipitates (3), preparation of subunit-specific antisera ( 6 ) , electrophoresis in SDS-polyacrylamide slab gels (lo), and fluorography (11) were done by published procedures. Whenever possible, results were quantified both by densitometric scanning of the fluorographs and by directly counting ["Slmethionine in gel slices. The gel slices were fvst soaked in distilled water for 30 min, incubated overnight a t 60 "C in 0.8 mi of NCS tissue solubilizer, cooled, and counted with 10 ml of a toluenebased scintillation fluid. Fluorograms were scanned with a Bausch-Lomb scanner and traces were quantified with a Hewlett-Packard integrator.
concentration was kept at 0.6 M. Spheroplasts were labeled uniformly with ["Slmethionine as in Ref. 5, except that the labeling time was 30-60 min. Cytochrome b2 was assayed as L-lactate-ferricyanide reductase (15). Bathophenanthroline disulfonate and carboxyatractyloside were purchased from Sigma and Boehringer Mannheim, respectively.

A Significant Proportion of a Given in Vitro-synthesized
Precursor Can Be Imported and Protected Post-translationally by Isolated Mitochondria-Carefully isolated yeast mitochondria were incubated with ["S]methionine-labeled in vitro translation products. Half of the mixture was subsequently incubated with trypsin to digest "5S-labeled polypeptides accessible to the added protease. The other half served as the control. The mitochondria were then reisolated from each aliquot and supernatants and mitochondria were analyzed for radiolabeled F1 P-subunit (a matrix polypeptide).
The results of one such experiment are shown in Fig. 1. Mitochondria from the control aliquot contained both the precursor and the mature form of the F, P-subunit. In contrast, mitochondria from the protease-treated aliquot contained only the mature form, suggesting that the labeled forms associated with the control mitochondria have different locations: the precursor is apparently outside and the mature subunit inside the mitochondrial membranes. In this experiment, the supernatant contained only the precursor (lane 2 ) , as did the lysate prior to incubation (not shown). (The relative amounts of mature and precursor forms recovered from the supernatant depend upon the intactness of the yeast mitochondria, since the soluble matrix processing protease (7) leaks out of partially damaged mitochondria and converts psubunit precursor to mature ,&subunit outside the mitochondria.) The efficiency of the trypsin treatment in degrading any nonprotected ,&subunit polypeptides is confirmed by the fact that no F I ,&subunit or its precursor could be recovered from the supernatant following proteolysis (lane 4, Fig. 1). Similar results were obtained for the F1 CY-and y-subunits, for cytochrome c oxidase subunit V (an inner membrane protein), for cytochrome b2 (an intermembrane space enzyme), and for the M, = 29,000 "porin" protein of the outer membrane (not shown).
Four fluorographs like the one pictured in Fig. 1 were quantified (see "Materials and Methods"). The results show that 60-70s of the in vitro-synthesized FI P-subunit can be protected from proteolysis upon incubation with mitochondria ( Table I). The ability of the mitochondria to withstand the trypsin treatment varies, however. Since incubation in reticulocyte lysate and protease treatment destroys about 30% of the intermembrane space enzyme cytochrome b2 (not shown), it clearly damages the mitochondrial outer membrane and perhaps the inner membrane as well. As a consequence, the percentages for protected radiolabeled polypeptides given in Table I may be an underestimation. Loss of ["Slmethionine resulting from cleavage of the (presumably) NHZ-terminal extension (17) would cause an additional underestimation.
GTP Inhibits the Processing of Precursors by a Hypotonic Extract, but Not by Intact Mitochondria-A matrix-located, chelator-sensitive mitochondrial protease is responsible for the maturation of those precursors that are transported partly or completely across the inner membrane. The protease is quantitatively released by hypotonic extraction of isolated yeast mitochondria (7). When extract obtained from 200 pg of mitochondria (the amount of mitochondria in a standard import assay) is incubated with a translated, gel-filtered reticulocyte lysate, nearly 90% of the FI P-subunit precursor is

TABLE I Quantitation of precursor import into isolated yeast mitochondria
Experiments similar to the one in Fig. 1 were quantified as described under "Materials and Methods." The amount of radiolabeled precursor immunoprecipitable from lysate before incubation with mitochondria is taken as 100%.
All 2 7 8 was recovered as precursor.
converted to the mature form (see Figs. 2 and 3). G T P inhibits the processing up to 90%. Earlier work (7) has shown that ATP and other chelators of divalent cations (o-phenanthroline, EDTA) are also effective inhibitors of the processing activity. Besides ATP and GTP, CTP and guanyl-5"yl imidodiphosphate also inhibited processing, although with less efficiency. Since the effect of ATP and GTP on the protease  3 (right). GTP inhibits processing of the F, 8-subunit precursor by a matrix fraction and by lysed mitochondria, but not by intact mitochondria. Two hundred pg of either intact or hypotonically disrupted mitochondria were incubated for 30 min at 27 "C with [""Slmethionine-labeled precursors as described earlier, with increasing amounts of GTP adjusted to pH 7.0. A hypotonic extract prepared from 200 pg of yeast mitochondria was incubated with radiolabeled precursors under identical conditions, except that 2.5 pg of efrapeptin was added per ml to prevent excessive hydrolysis of the added GTP by FI-ATPase. After incubation, the intact mitochondria were reisolated and dissociated in 3% SDS. The incubations containing lysed mitochondria or extract were dissociated directly in SDS. All samples were immunoprecipitated for FI P-subunit and analyzed by SDS-polyacrylamide gel electrophoresis, fluorography, and quantitation of the fluorograms. 100% is defined in each case as the radioactivity recovered as the mature form of FI P-subunit in the samples containing no GTP.
can be overcome by adding excess Mg", the inhibition is likely to result from chelation (28).
In contrast to the effect in a hypotonic extract from mitochondria, GTP has no significant inhibitory effect on the processing of FI /?-subunit by intact mitochondria (Fig. 3).
This GTP resistance is not caused by a component present in the mitochondrial membranes, since processing of the /?-subunit precursor by disrupted mitochondria is GTP-sensitive (Fig. 3). The observed processing of the /?-subunit precursor by intact mitochondria in the presence of GTP must then occur in the matrix compartment into which G T P cannot readily diffuse (18). G T P can, however, readily penetrate into the intermembrane space, since the outer membrane permits passive diffusion of molecules with molecular weights up to several thousands (19). Only the processing enzyme which is protected from the inhibitory concentration of G T P by an intact inner membrane can cleave mitochondrial precursors of matrix and inner membrane polypeptides. Processing by mitochondria in the presence of external G T P can thus be used as a convenient assay for the import of precursors into the matrix. The validity of this assay is based upon the fact that the outer, and not the inner, membrane of yeast mitochondria is permeable to GTP. Although our evidence suggests that GTP acts as a chelator here, proof of this is not essential for interpretation of the results.
Bathophenanthroline disulfonate is another potent inhibi- IMS, intermembrane space fraction; MA. matrix fraction; ME, the insoluble membrane fraction (both the inner and outer membranes) after release of the matrix and intermembrane tor of the soluble processing protease and its charge should prevent its diffusion into the matrix space of intact mitochondria. Indeed, 0.5 mM bathophenanthroline disulfonate blocks processing by a hypotonic mitochondrial extract completely, but has little effect on processing by intact mitochondria (not shown).
Proteins Are Imported to Their Correct Zntramitochondriaf Location-When isolated yeast mitochondria are first allowed to take up in vitro-synthesized precursors and then fractionated into matrix space, intermembrane space, and a membrane fraction, labeled matrix proteins are found in the matrix fraction but not the intermembrane space fraction, whereas the labeled intermembrane space enzyme cytochrome b:! is found in the intermembrane space, but not the matrix. Labeled membrane proteins are found only in the membrane fraction ( Fig. 4 and Table 11). The low levels of labeled matrix enzymes in the intermembrane space fraction and of intermembrane space polypeptides in the matrix fraction can be fully accounted for by cross-contamination of these fractions (Table  HA). However, cross-contamination can only partly account for the result that roughly half of the imported matrix and intermembrane space enzymes are recovered in the membrane fraction. It is unknown whether these membrane-associated proteins represent assembly intermediates or a side reaction. Some of the precursors may be trapped "in transit" across the membranes. Nonspecific adsorption may be an additional contributing factor. Import and processing of in uitro-synthesized precursors by isolated mitoplasts sheds further light on the localization of matrix and intermembrane space polypeptides. The mitoplasts retain a large portion of their outer membrane even though >90% of an intermembrane space marker enzyme is released (6) and they are able to process the precursors of both FI P-subunit and cytochrome bz to their corresponding mature forms (Table 111). When mitoplasts are reisolated through a 0.625 M sucrose cushion from the standard incubation mixture for in vitro import, mature radiolabeled FI psubunit is recovered with the mitoplast pellet, whereas processed radiolabeled cytochrome b2 is released to the suspending medium. This suggests that the processed FI p-subunit is sequestered by the mitoplasts in the matrix space, whereas the processed cytochrome b:! remains outside the mitochondrial inner membrane. With whole mitochondria, radiolabeled processed cytochrome b2 is recovered completely with the mitochondrial pellet (Table III), presumably because the processed polypeptide is sequestered in the intermembrane space.
The recovery of a large part of processed radiolabeled cytochrome bs with membranes after import with whole mitochondria (see Table 11) might, therefore, reflect nonspecific adsorption of the processed cytochrome b:! to outer or inner membranes. When the mitoplasts are reisolated through the sucrose cushion, approximately 50 mM KC1 and 30 mM Hepes/ KOH, pH 7.4, are present in the incubation mixture. The release of intermembrane space from whole mitochondria, on the other hand, is done with 0.1 M mannitol and 10 mM Tris-C1, pH 7.4. The higher ionic strength present during the mitoplast reisolation may be sufficient to eliminate the adsorption of processed cytochrome b2 to the membranes, as is seen in the fractionation results.
In order to check the submitochondrial distribution of non-

Distribution of unlabeled marker enzymes and of [:'"S/methionine-labeled importedproteins after the fractionation of yeast mitochondria
In A, 100% is defined as the total enzyme activity in the whole translated lysate was determined by immunoprecipitating the desired mitochondria recovered after incubation in the translated reticulocyte protein from an aliquot of lysate which had been mixed with the lysate. Cytochrome b2 was measured as L-lactate-ferricyanide reduc-equivalent amount of unlabeled mitochondrial protein as was used in tase. The distribution of cytochrome c oxidase subunit 111 was deter-the in vitro import assay. For computing the distribution of a given mined by immunoreplication and scanning the radioautograms. In B, labeled protein among the mitochondrial subfractions, the radioactivone fractionation experiment, in which all six of the proteins listed ities recovered in the matrix, the intermembrane space, and the were immunoprecipitated from the various fractions, was quantitated membranes were summed and considered 100%. This figure ranged by slicing the dried polyacrylamide gels and counting the '% radio-from 86 to 114% of the corresponding value in column 3, which activity. The total amount of any one polypeptide present in the represents the recovery in whole mitochondria. ' This polypeptide is not made as a larger precursor (not shown).

TABLE I11
In uitro-synthesized cytochrome bz is processed, but not sequestered, by mitoplasts In uitro-synthesized polypeptides were incubated with 200 pg of either mitoplasts or mitochondria under the standard conditions for in vitro import, including 5 mM GTP. After 30 min a t 27 "C, each incubation mixture was layered over a 0.625 M sucrose cushion and centrifuged for 15 min at 30,000 X g. Pellets and supernatants were dissociated with 3% SDS a t 95 "C. Samples were quantitatively immunoprecipitated for cytochrome b? and F1 P-subunit. The immunoprecipitates were analyzed by SDS-gel electrophoresis and fluorography, and the fluorograms were quantified by densitometric scanning. 100% refers to the total amount of immunoprecipitable ["%] methionine-labeled polypeptide recovered from each incubation condition. F, P-subunit specifically adsorbed in vitro translation products, we followed the distribution of the glycolytic enzyme glyceraldehyde-3-Pdehydrogenase. As shown in Table 11, in uitro-synthesized glyceraldehyde-3-P dehydrogenase (a cytosolic enzyme) binds only slightly to mitochondria, but whatever is bound is recovered in the membrane fraction. This distribution shows that an in vitro translation product nonspecifically adsorbed to mitochondria appears neither in the matrix nor the intermembrane space fractions. Attempts to separate the inner and outer membranes after in vitro import were only partly successful, since incubation of the mitochondria with a reticulocyte lysate drastically lowered the recovery of outer membrane. Still, it was possible to show that in uitro-synthesized cytochrome c oxidase subunit V becomes associated only with the inner, and not with the outer, membrane (not shown). The submitochondrial distribution of imported polypeptides argues strongly that polypeptide import by isolated mitochondria mirrors the import process in vivo.
Processed Precursor Is Not Imported by Isolated Mitochondria-The results described so far indicate that processing of the precursors to matrix and inner membrane enzymes occurs after, not before, import. This implies that import might be blocked if the precursors are artificially processed before they are incubated with isolated mitochondria. This is indeed the case (Fig. 5 ) . A translated reticulocyte lysate was incubated with purified processing protease isolated from yeast mitochondria. About 75% of the in uitro-synthesized precursor of subunit V of cytochrome c oxidase was thereby converted to the mature form as determined by SDS-gel electrophoresis and counting of the precursor and mature polypeptide band. As a control, another aliquot of the labeled lysate was incubated with processing protease which had been inhibited by o-phenanthroline. Both aliquots were filtered through a Sephadex G-25 column and incubated with isolated mitochondria under the usual conditions of in vitro import. In this way, one lysate contained mainly processed polypeptides and the other only the precursors. After a subsequent trypsin treatment of the mitochondria, mature subunit V was protected from proteolysis in the control sample, but not in the sample in which the precursors had been processed to their mature forms prior to incubation. Since the processing activity is purified from a soluble hypotonic extract and subunit V is a hydrophobic membrane protein (21), our preparation of processing enzyme is free of unlabeled mature subunit V, which could dilute the radiolabeled subunit V and thereby complicate interpretation of the experiment. I n Vitro Import Requires a Protease-sensitive Component Prior processing of the cytochrome c oxidase subunit V precursor blocks its import into isolated mitochondria. [:'"SI Methionine-labeled precursors were synthesized in vitro and precursor polypeptides in half of the lysate were processed to their mature size by the addition of a partially purified hypotonic extract from yeast mitochondria.' The processing protease was inhibited after 20 min by 2 mM o-phenanthroline. For the control lysate, the inhibitor and extract were mixed prior to addition to the lysate. Both samples were then filtered through Sephadex G-25 and used for the import assay as described under "Materials and Methods." Translocation across the mitochondrial membranes was checked by the resistance of the radiolabeled polypeptides to externally added trypsin (120 pg/ ml). Aliquots of processed and unprocessed lysate, and the supernatants and pellets from the in oitro import, were dissociated in 3 6 SDS and subjected to quantitative immunoprecipitation for cytochrome c oxidase subunit V (su V). Samples were analyzed by gel electrophoresis and fluorography. A photograph of the fluorogram is shown. 1, cytochrome c oxidase subunit V precursor from lysate; 2, subunit V from the mitochondrial pellet which was not protease-treated; 3, subunit V from the supernatant corresponding to lane 2; 4, subunit V from the protease-treated mitochondria; 5, supernatant corresponding to 4.

on the Outer Face of the Mitochondrial Outer Membrane-
Mitochondria that had been exposed to as little as 10 pg of trypsin per ml a t 0 "C for 10 min prior to the incubation with in vitro-synthesized precursors did not import precursors nor cleave them to their mature forms (Table IV). After trypsin treatment, at least 95% of the intermembrane space enzyme cytochrome br was still inaccessible to externally added cytochrome c: ' showing that the outer membrane had remained largely intact. Since import by the treated mitochondria was strongly inhibited, a protein exposed on the cytoplasmic face of the outer membrane may be necessary for the uptake of precursors into mitochondria.
In Vitro Import, but Not Cleavage, of the Precursor Is Energy-dependent-We have previously reported that, in intact yeast cells, maturation of mitochondrial precursor polypeptides is blocked by conditions which lower the ATP level in the matrix (5). Protein import by isolated mitochondria also exhibits this characteristic energy dependence. By removing small molecules from a labeled reticulocyte lysate (see "Materials and Methods"), we could reconstitute an in vitro import system that was completely dependent on added ATP (Fig. 6). In this experiment, import of the FI P-subunit was assessed by the presence of processed, labeled F, P-subunit in the mitochondria (see preceding sections). The mitochondrial pellet and supernatant incubated without ATP show only the precursor, most of which remains in the supernatant. After incubation in the presence of ATP and an ATP-regenerating system, however, 52% of the radiolabeled FI /?-subunit precur-

Trypsin-treated isolated mitochondria are unable to import
precursor polypeptides Isolated mitochondria were incubated for 10 min at 0 'C with the concentrations of trypsin indicated below. A IO-fold excess of soybean trypsin inhibitor and 1 mM TLCK were added. The mitochondria were subsequently used for the in vitro import assay. Maturation and uptake of FI /?-subunit or cytochrome br were determined by a second protease treatment, immunoprecipitation, SDS-polyacrylamide gel electrophoresis, and fluorography as described earlier (see "Materials and Methods"). In each case, the amount of mature, protected polypeptide was quantified by densitometric scanning of the fluorogram. The amount of mature polypeptide recovered with untreated mitochondria is defined as 1006. This value ranged from 32-43%, of the total available precursor of the respective polypeptide. ' In this experiment, 5 mM GTP was included in the in vitro import assay and import was assayed as in Fig. 3. sor is recovered as mature form with the mitochondrial pellet. In order to show that ATP is required for translocation, and not cleavage, of the precursor, the labeled lysate was incubated with hypotonically disrupted mitochondria: nearly complete conversion of the precursor to the mature form was observed even in the absence of added ATP. Thus, energy is required for translocation of the precursors, not for their cleavage.

-ATP
Import of Precursors into the Mitochondrial Matrix Requires a n Electrochemical Gradient across the Inner Membrane-ATP-driven import of the FI /3-subunit precursor is blocked by carboxyatractyloside, a specific inhibitor of adenine nucleotide translocation across the mitochondrial inner membrane (Fig. 7). This shows that the added ATP acts from within the matrix. (In these experiments, KCN was added to block oxidative phosphorylation resulting from respiration of endogenous substrates.) This confirms the results obtained earlier with intact yeast cells (5).
ATP-dependent import of the FI /3-subunit precursor is also inhibited by oligomycin which blocks the mitochondrial ATPase complex (Fig. 7). Thus, the added ATP must be hydrolyzed by this complex for import to occur. The most likely explanation for this would be that ATP is used to generate an electrochemical gradient across the mitochondrial inner membrane. Indeed, ATP-dependent import is abolished if any such gradient is collapsed by valinomycin plus K' or by carbonyl cyanide m-chlorophenyl hydrazone (Fig. 7). In these experiments, oligomycin was added as well to block rapid hydrolysis of ATP. The essential role of an electrochemical gradient across the inner membrane is further documented by the observation that oligomycin does not block import of the FI /3-subunit if energy is supplied by respiration instead of by added ATP (Fig. 8). This shows that the inhibitory effect of oligomycin with added ATP as energy donor does not merely reflect a general disruptive effect of oligomycin on mitochon- drial integrity. Valinomycin, on the other hand, blocks import even in respiring mitochondria. Polarographic measurements clearly confirmed that the inhibitors employed here act in the expected manner: oligomycin blocks stimulation of respiration by ADP, but not that by the uncoupler carbonyl cyanide mchlorophenyl hydrazone or by valinomycin plus K+ (not shown).

DISCUSSION
In order to understand how proteins are imported into mitochondria, one must first be able to reproduce the import process in vitro. It had already been shown that isolated mitochondria can process in vitro-made mitochondrial precursor polypeptides and that the resulting mature polypeptides are inaccessible to externally added proteases (1, 2). However, it remained to be shown that this in vitro process is an acceptable representation of mitochondrial protein import in vivo. The present work suggests that this is indeed the case.
The relatively simple in vitro system described here exhibits several features which characterize protein import by mitochondria in living cells. First, a large fraction of a given in vitro-made precursor is taken up by the mitochondria. Second, polypeptides are transported to their correct mitochondrial location. Third, import of precursors into or across the mitochondrial inner membrane requires energy.
Study of this in vitro system has yielded new information on the import process. For example, it was possible to show that precursors processed before incubation with mitochondria are not imported. In other words, uptake occurs before, not after, processing. Independent evidence for this comes from the demonstration that the processing protease for precursors imported across the inner membrane is located in the matrix and that the extracted protease is not energy-dependent (7). It was also found that import could be abolished by treating isolated mitochondria with low amounts of trypsin, Protein Import by Isolated Yeast Mitochondria 13041 which apparently alters but does not disrupt the outer membrane. This treatment also abolished binding of precursors to outer membrane vesicles which had been subsequently purified from these mitochondria. These results suggest that import requires protein "receptors" on the outer face of the mitochondrial outer membrane (1). The in vitro system described here was particularly useful for identifying the energy requirement of protein import into mitochondria. In our earlier work we had found that processing of precursors in intact yeast cells required the presence of ATP in the mitochondrial matrix. Since in vivo processing still occurs in rho-mutants which lack respiration as well as a functional mitochondrial ATPase complex (22), we suggested that it was ATP itself, rather than an energized membrane, that functioned as an energy source (5). Results from an in vitro import system using Neurospora mitochondria showed that carbonyl cyanide m-chlorophenyl hydrazone blocks the conversion of the in vitro-synthesized ADP/ATP translocator to a protease-resistant form (23). However, since carbonyl cyanide m-chlorophenyl hydrazone, in the absence of oligomycin, depletes the matrix of ATP as well as collapsing the membrane potential, these results were inconclusive as to the immediate energy donor.
The results obtained with the in vitro system here show that the immediate energy donor is not ATP itself, but an electrochemical gradient across the mitochondrial inner membrane. In this respect, protein import into mitochondria resembles the transport of proteins into or across the cytoplasmic membrane of Escherichia coli (24, 25). Since valinomycin-treated mitochondria appear to catalyze a H+/K' exchange (26), it is not possible to tell from the available data whether protein import requires a pH gradient, an electrical potential, or both.
These results explain our earlier observation that a buildup of precursors to mitochondrial proteins can be seen more easily in rho-yeast cells than in wild type cells (5). The fact that protein import is not completely blocked in these mutants suggests that they can generate an electrochemical gradient across the mitochondrial inner membrane by some process(es) other than respiration or ATP hydrolysis. For example, the ATP/ADP exchange via the adenine nucleotide translocator appears to be coupled to a membrane potential (27). Since any electrochemical gradient generated by the translocator would be small, import of polypeptides should require a significantly smaller electrochemical gradient than ATP synthesis requires.