Early Events in the Transport of Proteins into Mitochondria IMPORT COMPETITION BY A MITOCHONDRIAL PRESEQUENCE*

Studies with a synthetic presequence peptide, FIB1-20, corresponding to the NHz-terminal 20 amino acids of the F1.ATPase B-subunit precursor (pFIB) show that although this peptide binds avidly to phospholipid bi- layers it does not efficiently compete for import of full-length precursor into mitochondria, Ki = 100 p~ (Hoyt, D. W., Cyr, D. M., Gierasch, L. M., and Douglas, M. G. (1991) J. BioC. Chem. 266, 21693-21699). Herein we report that longer FIB presequence peptides and F1821-51+3 compete for mitochondrial import at 1000-, 250-, and 25-fold lower concentrations, respectively, than A longer peptide, F1B1-51+3, no more effective as an import competitor than F1B1-32+2. Both minimal length and amphiphilic character required in order for FIB peptides to block mitochondrial import. longer FIB peptides at a step common to all precursors since they import precursors a- 8- subunits ADP/ATP protein.

subunits and the ADP/ATP carrier protein. Dissipation of membrane potential (A$) across the inner mitochondrial membrane is observed in the presence of FIBpeptides, but this mechanism alone does not account for the observed import inhibition. F1B1-32+2 and 21-61+3 block import of pFl@ 100% at peptide concentrations which dissipate A$ less than 25%. In contrast, experiments with valinomycin demonstrate that when mitochondrial A$ is reduced 25% import of pFl@ is inhibited only 25%. Therefore, at least 75% of maximal import inhibition observed in the presence of F1B1-32+2 and F1B21-51+3 does not result from dissipation of A$. Import inhibition by Fl@-peptides is reversible and can be overcome by increasing the amount of fulllength precursor in import reactions. FIB presequence peptides and full-length precursor are therefore likely to compete for a common import step. Presequence dependent binding of pFIB to trypsin-sensitive elements on the outer mitochondrial membrane is insensitive to inhibitory concentrations of FIB presquence peptide. We conclude that import inhibition by FIB presequence peptides is competitive and occurs at a site beyond initial interaction of precursor proteins with mitochondria.
Most mitochondrial proteins are synthesized on cyto-* This work was supported by National Institutes of Health Grant GM36537. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Supported by Postdoctoral Fellowship NC-90-F4 from the American Heart Association, North Carolina Affiliate.
To whom correspondence should be addressed.
plasmic polyribosomes and must be imported either co-or post-translationally into the appropriate suborganellar compartment. Targeting information is typically specified by a stretch of amino acids present at the amino terminus of precursor prqteins, the presequence, which is usually cleaved by a metalloprotease upon entry into the mitochondrial matrix (for reviews see Hart1 et ul., 1989;Pfanner and Neupert, 1990;Verner and Schatz, 1988). Gene fusion studies demonstrate that the presequence alone is necessary and sufficient to direct mitochondrial import of both authentic precursors and non-mitochondrial passenger proteins (Hurt et al., 1984;Honvich et al., 1985;Emr et al., 1986). However, presequences contain no consensus sequence for recognition by the import apparatus (von Heijne, 1986). Instead, sequence analysis of mitochondrial targeting signals reveals that presequences contain an abundance of polar, hydrophobic, and basic amino acids with a marked absence of acidic residues (von Heijne, 1986). Based on the regular spacing of basic residues within presequences, it is predicted that presequences do, however, share the ability to organize into amphiphilic structures in apolar environments such as mitochondrial membranes (von Heijne, 1986). This prediction was proven correct when interactions between synthetic presequence peptides and model membranes were studied (Roise et al., 1986;Hoyt et al., 1991). Further, importance of amphiphilicity in the import signal has been demonstrated by experiments in which artificial mitochondrial presequences, made from serine, leucine, and arginine, were found to direct import of passenger proteins fused to them if the presequences could form amphiphilic structures (Allison and Schatz, 1986;Roise et al., 1988). Thus, structural requirements for a functional mitochondrial import signal appear rather modest. This, however, makes it difficult to understand how specificity of intracellular delivery is achieved. An approach to this problem is to study synthetic presequences in both model and biological systems. Studies with the synthetic presequence of the F1-ATPase' @-subunit precursor (pF1@), the NH2-terminal 20 amino acids, indicate this presequence can adopt an amphiphilic a-helical conformation in hydrophobic environments and can bind tightly to phospholipid monolayers. This presequence peptide, however, does not penetrate the hydrophobic core of the bilayers to disrupt the integrity of acidic phospholipid vesicles (Hoyt et al., 1991). Interestingly, the pFl@ presequence alone does not efficiently compete for in vitro import of full-length precursor (Ki 100 The abbreviations used are: F1-ATPase, soluble portion of the mitochondrial inner membrane bound ATPase complex; DiS-C3-5, 3',3'-dipropylthiadicarbocyanine iodine. A$, electrical potential across the inner mitochondrial membrane; HPLC, high performance DTT, dithiothreitol; BSA, bovine serum albumin; HEPES, 442-liquid chromatography; MOPS, 4-morpholinepropanesulfonic acid; hydroxyethy1)-1-piperazineethanesulfonic acid.

21700
This is an Open Access article under the CC BY license. Hoyt et al., 1991). However, studies with the preornithine carbamyltransferase presequence, amino acids 1-27, show that this peptide can completely block mitochondrial import at 5-10 PM, and collapse A# across the inner mitochondrial membrane unless excess reticulocyte lysate is present in the import reactions (Gillespie et al., 1985). The presequence of cytochrome oxidase subunit IV (COX IV), amino acids 1-25, was found to disrupt artificial membranes as well as uncouple mitochondria (Roise et al., 1986). On the other hand, shorter COX IV presequence peptides, retaining targeting information, were found to efficiently compete for import without apparently altering membrane integrity (Glaser and Cumsky, 1990).

PM,
Since fusion of the first 14 amino acids of pFIP to dihydrofolate reductase (DHFR) is sufficient to direct in vitro mitochondrial import of this normally cytosolic protein (Walker et al., 1990), it is difficult to understand why the FIB presequence peptide does not efficiently compete for import. Several features of pFIP import, however, suggest why this result is observed. Like other precursors, pFIP can be translocated into mitochondria post-translationally in a sequential series of events which include binding to the outer mitochondrial membrane, interaction with outer membrane proteins (MOM 19, MAS 70, and ISP 42;Sollner et al., 1989;Hines et al., 1990;Vestweber et al., 1989) and membrane potential (A+)dependent insertion through the inner membrane membrane (Gasser et al., 1982;Schleyer et al., 1982). Translocation is coupled to an ATP-dependent interaction with mitochondrial HSP 70 (Kang et al., 1990) and is followed by cleavage of the presequence and Hsp6O assisted assembly into the F,.ATPase complex in the mitochondrial matrix (Cheng et al., 1989). However, pFIP import differs in that insertion of this precursor through the inner membrane requires cytosolic ATP hydrolysis in addition to A# (Chen and Douglas, 1987a;Pfanner et al. 1987;Eilers et al., 1987). Import of other precursors, with the exception of the ADP/ATP carrier protein , requires ATP but only in the matrix (Hwang and Schatz 1990;Miller and Cumsky, 1991). Cytosolic ATP consumed in import could be hydrolyzed by molecular chaperones during interactions with pFIP and possibly by other soluble or membrane-associated translocation factors (Ohta and Schatz 1984;Chen and Douglas, 1987b;Pfanner and Neupert, 1987;Deshaies et al., 1988;Murakami et al., 1988;Sheffield et al., 1990;Murakami and Mori, 1990). Since factor-dependent ATP hydrolysis is partially responsible for insertion of the pFIP presequence into mitochondrial membranes, one explanation for the lack of import inhibition by the pFIP presequence peptide, F1P1-20, may be that more than the presequence is required for high affinity interaction with soluble or membrane-bound import components required for insertion into mitochondrial membranes.
To test the relationship between length of presequence peptides and their ability to block mitochondrial import, we synthesized several FIB presequence peptides of varying length and tested their ability to compete for in vitro mitochondrial protein import. Herein we report that presequence peptides, of varying amphiphilicity, containing 14 amino acids in addition to the presequence, block import half-maximally at 25-1000-fold lower concentrations than a peptide containing the presequence alone. Import competition was deemed specific in that it was reversible and could be competed by full-length precursor protein. Additionally, presequence peptides, at concentrations which completely block import, had no apparent effect on binding of full-length precursors to mitochondria. Thus, import competition appears to occur at a point beyond the initial interaction of precursor proteins with mitochondria.
DNA Techniques-Transformation of E. coli strain MC1066 and preparation of small scale plasmid DNA were as described (Maniatis et al., 1982). Restriction digestions were as described by the commercial suppliers. Linearized plasmid DNA for in vitro transcription was extracted two times with chloroform/phenol (l:l), one time with chloroform/isoamyl alcohol (24:l). DNA was then precipitated with 70% ethanol and isolated by centrifugation. The resultant pellet was rinsed with 70% ethanol, dried, and then resuspended in 10 mM Tris, pH 7.4, made 1 mM in EDTA at 1 gg/ml.
In Vitro Transcription and Translation-The genes coding for precursors to FIB, the F1-ATPase a-subunit (pFla) and the ADP/ ATP carrier protein (pAAC) were placed under control of the phage T7 polymerase promoter by ligation into plasmid pT7-2 (Promega) as described previously: pT,ATP, (Chen and Douglas, 1987a), pT7ATP1 (Takeda et al., 1986), and pT7AAC1 (Smagula and Douglas, 1988). The respective plasmids, pT7ATP,, pT7ATP1, and pT7AAC1 were linearized with HindIII, BamHI, and EcoRI. Two to 5 pg of linearized plasmid DNA was added to transcription reactions containing T? polymerase (Promega Corp., Madison, WI) as described earlier (Chen and Douglas, 1987a). Products of transcription reactions were extracted two times with chloroform/phenol (l:l), one time with chloroformlisoamyl alcohol (24:l). mRNA was then precipitated by addition of sodium acetate (300 mM) and ethanol (70%) and isolated by centrifugation. The resultant pellet was rinsed with 70% ethanol, dried, and then resupended in 50 pl of RNase-free H, O (Maniatias et al., 1982). Purified mRNA transcripts (5 pl) were used immediately in translation reactions or were stored frozen at -80 "C. Biosynthesis of 35S-labeled precursor protein from mRNA transcripts was carried out in nuclease-treated reticulocyte lysate exactly as instructed by the supplier (Promega Corp., Madison, WI) except 35S-translabel (1059 Ci/mmol, ICN Biomedicals, Irvine CA) was substituted for [35S]methionine. Translation reactions were used immediately in import reactions or were stored frozen at -80 "C.
I n Vitro Import Experiments-Cells grown in lactate medium were harvested, and mitochondria were isolated as described (Gasser et al., 1982) except after isolation they were resuspended to 5 mg/ml in 1.2 M Sorbitol made 1 mM in EDTA and 10 mM MOPS buffer, pH 7.4 (SEM), aliquoted into 250-pl fractions, and flash frozen in liquid N,. In vitro import of "S-labeled precursor protein was carried out in 100-pl reaction mixtures containing 200 mM sucrose, 10 mM HEPES, pH 7.4, 100 mM potassium acetate, 2 mM magnesium acetate, 2 mM DTT, 1 mM ATP, 10 mM succinate, 25 mM creatine phosphate, 2.5 mg/ml of creatine phosphokinase, 3% BSA, and mitochondria (40-60 pg), thawed on ice just before use. Except where noted, mitochondria were preincubated on ice with or without peptide in import buffer for 5 min prior to start of the import reactions by addition of %labeled precursor and shift to 25 "C. Following incubation at 25 'C for 20 min, mitochondria were reisolated, resuspended in SEM, reisolated again and then analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Laemmli, 1970) and fluorography (Chamberlin, 1979). Quantitation of fluorograghs was achieved using model 300A Laser densitometer (Molecular Dynamics, Sunnyvale, CA). Quantitation of x-ray film exposed to the same gel for various time periods confirmed linearity of the observed signal.
Synthetic FIB Peptides-Presequence peptides were synthesized by Dr. David Klapper of the Department of Microbiology at University of North Carolina and Immuno-Dynamics (La Jolla, CA). Following synthesis and hydrogen fluoride cleavage all peptides were HPLC purified on a C4-Vydac column. The mobile phase consisted of 0.1% trifluoroacetic acid and acetonitrile mixed 1:l. Column buffer was removed by freeze drying peak fractions, and peptides were stored in a vacuum desiccator until use. The sequence and amino acid composition of peak fractions was confirmed by standard techniques. Pep-tide stock solutions (1 mM) were made by weighing out peptide and then dissolving it in 10 mM HEPES, ph 7.4. Typically, 2 ml of peptide stock solutions were made, aliquoted, and then freeze dried. Concentrations of peptide stock solutions were verified by quantitative amino acid analysis. Just hefore use, peptide solutions were reconstituted to their original volume with water. All peptides were freely soluhle in water.
Measurement of Mitochondrial Membrane Potrntial-Qualitative changes in A$ hrought on hy F,/j presequence peptides were assayed by monitoring increases in fluorescence of the cationic fluorescent dye :1,:~-dipropylthiadicarhocyanine (DiS-C3-5, Molecular Prohes, Eugene, OR) after peptide addition reaction mixtures containing dyeloaded mitochondria. D I S C -5 binds mitochondria exhihiting A$ with a resultant decrease in DiS-C.7-5 fluorescence. Agents which reduce A$ cause release of DiS-C3-5 from mitochondria which is exhibited by an increase in DiS-C3-5 fluorescence (Sims et al., 1974). Mitochondria (0.60 mg/ml) were incuhated in 3-ml reaction mixtures containing 200 mM sucrose. 10 mM HRI'ES. pH 7.4, 100 mM potassium acetate, 2 mM magnesium acetate, 2 mM DTT, 1 mM ATP, 10 mM succinate, 3% RSA, and 2.0 P M DiS-C3-5, added from an ethanolic stock (150 pM). Reaction mixtures were maintained at 20 "C and stirred continuously through out the incuhation period. Changes in fluorescence were monitored with a Perkin-Elmer model MPF-3 Fluorometer equipped with a Hitachi-OPD 33 strip chart recorder. Excitation was set a t 620 nm and emission a t 680 nm. Under these conditions mitochondria maintained a memhrane potential for a t least 25 min.

RESULTS
Peptide Length Is Important for Import Competition-Previous results from in uitro competition studies with the synthetic peptide F1B1-20 demonstrate that the FIB presequence alone does not efficiently compete for mitochondrial import ( K , s 100 p~) .
This could result from inadequate information or length within Fl/31-20 to allow for high affinity binding to import factors or failure of the peptide to insert deep enough into the outer membrane to gain access to import components. T o test the length dependence of peptide inhibition of mitochondrial import, peptides consisting of the 19 amino acid presequence of FIB plus adjacent amino acids in the mature protein were synthesized and HPLC purified (see "Experimental Procedures"). Depicted in Fig. 1 is the wild type sequence for the first 51 amino acids of pFIP and the synthetic Fl@ peptides used in this study. Where noted, additional amino acids have been included to allow for iodination and chemical coupling of the peptides in other studies.

FlII(l411.3 ~P R L~A T S R A I F I A A I O S A~~S I S~~~~S~~~-~~~~~~~
FIG. 1. Amino acid sequences of synthetic FIB peptides used in this study. The arrow hetween lysine I 9 and glutamine 20 in the wild t.ype pFIB sequence marks the presequence cleavage site. Regions of undrrlinrd sequence denote amino acids engineered into the peptides to allow for iodination of the peptides in other studies. All peptides were HPLC purified prior to use (see "Experimental Procedures"). Import competition by FIB peptides was tested in an in uitro import assay utilizing ,""S-labeled Fi(j precursor protein translated in reticulocyte lysate and mitochondria isolated from S. cereuisiae (see "Experimental Procedures"). This assay is based on quantitat,ion of &Ldependent processing of the precursor protein to the mature form which occurs only if the precursor is translocated into the mitochondrial matrix. Assays with F1/31-20 (Fig. 2) and the control peptide F1/j33-49+6 (not shown) confirmed earlier results which demonstrated that neither peptide is an effective import inhibitor (Hoyt et al., 1991). However Fl[j1-32+2, which contains 14 amino acids in addition to the presequence, competed for import in a dose-dependent manner, and half-maximal inhibition occurred a t around 0.12 P M (Fig. 2). Similar results were obtained with F1[j1-.51+3 (Fig. 2). This K, represents about a 1000-fold increase in sensitivity of import to presequence containing peptide when compared to F1P1-20 and is in the range of those observed for import competition by purified full-length precursor proteins (Pfaller et al., 1988;Sheffield et al., 1990). Others have shown that import information in addition to the presequence is present within the first 34 amino acids of preF,P (Redwell et al., 1987) and thus, FIG. 2. Import of full-length pFIO is inhibited by FIB presequence peptides. A , isolated yeast mitochondria (GOO pg/ml) were incuhated in reartions mixtures containing 200 mM surrose, 10 mM HEI'ES, pH 5.4, 3 Y HSA. 1 0 0 mM potassium acetate. 2 mM magnesium acetate, 2 mM DTT, 1 mM ATP, 10 mM succinate, 2.5 mM creatine phosphate, 2.5 mg/ml creatine phosphokinase. "S-laheled precursor translated in reticuloc.yte lysate, and Fld presequence peptide as indicated in a final volume of 100 pl. Mitochondria were preincuhated on ice in import huffer. with and without peptide. 5 min prior to start of import reactions hv addition of '"S-laheled precursor and shift to 25 "C. Following incuhation for 20 min. mitochondria were reisolated, washed, and analyzed for import as descrihed under "Experimental Procedures." H . fluorographed gels were quantitated for mature Fld formed during import reactions hy laser densitometry.
Values are expressed as of mature protein formed in the nhsencr of peptide, set arhitrarily at 100";'. and are plotted Lvrsus the concentration of respective peptide. It should he noted that in the presence of Fl/j peptides none of the precursor protein found associated with mitochondria was protected from digestion by proteinase K (not shown), indicating that FIB presequence peptides inhihit import and not processing of the precursor to the mature form under thcsc experimental ronditions. this may be the cause for increased sensitivity of import to the longer FIB peptides. To address this possibility two mutant peptides of identical length to F1P1-32+2 were synthesized, F1@l-32SQ+2 and 21-51+3 (Fig. 1). FIP1-32SQ+2 contained conservative replacements of the lysine and arginine at positions 30 and 31 with serine and glutamine, respectively. These basic residues are predicted to extend the amphiphilic structure of the presequence and thus strengthen the import signal (Bedwell et al., 1987). F1P21-51+3 lacks the presequence portion of pFIP but still retains some predicted ability to form an amphiphilic helix (see under "Discussion"). When tested in competition assays, a 4-fold higher concentration of FIP1-32SQ+2 was required to half-maximally block import of fulllength precursor (Ki = 0.5 p~) when compared to F1P1-32+2 ( Fig. 2). However, 35 times more F1P21-51+3 was required to compete for import than F1/31-32+2 (Fig. 2). It is noteworthy that F1P21-51+3 (Ki = 4 p~) remained 25 times more potent as an import competitor than F1P1-20 (K' % 100 pM, Fig. 2).
Hence, additional length appears important, perhaps more important, than the ability to form an extended amphiphilic helix in conferring ability of synthetic FIP peptides to compete for mitochondrial protein import.
Peptides Compete for Mitochondrial Import in a Reversible Manner-Some, but not all, amphiphilic peptides interact with and disrupt lipid bilayers (Segrest et al., 1990). Disruption of the inner mitochondrial membrane and resultant dissipation of the electrochemical gradient required to drive protein translocation into mitochondria has been associated with inhibition of mitochondrial import by a few synthetic presequence peptides (Gillespie et Roise et al., 1986;Glaser and Cumsky, 1990). To examine the possibility that FIB peptides were acting in a similar manner, dose-dependent inhibition of mitochondrial import and changes in mitochondrial A$l assayed with the cationic fluorescent dye DiS-C3-5, were compared under conditions used to assay in vitro mitochondrial import (see "Experimental Procedures"). Presented in Fig. 3, A-E, are curves representing increases in fluorescence, corresponding to decreases in A$, observed upon addition of respective FIP peptides to reaction mixtures containing DiS-C3-5-loaded mitochondria. These values are expressed in percent of the maximal increase in fluorescence observed upon addition of the K' ionophore valinomycin (2.5 pg/ml) to reaction mixtures. Also shown is the corresponding effect of respective F1P peptides on import of pFIP assayed under similar conditions. In most cases, F1@ peptides dissipated mitochondrial A$ to about 80% of maximal values observed with valinomycin at concentrations below 5 p~ (Fig.  3, C-E). F1P1-20 and F1P21-51+3 are the exceptions (Fig. 3, B and F ) .
Dissipation of A$ by F,P peptides made it difficult to resolve the extent of import inhibition due to competition from that resulting from reduced membrane potential. However, in all cases import inhibition by FIP peptides exceeded the effect on A$ (Fig. 3, B-F). This is in contrast to the effect observed with valinomycin plus K' which simply dissipates A$ across the mitochondrial membrane (compare Fig. 3, A to B-8'). The peptides FIB 1-32+2 and 21-51+3 blocked import 50 and 100% at peptide concentrations which dissipate A$ less than 5 and 25%, respectively (Fig. 3, C and F ) . Titration experiments with valinomycin ( Fig. 3A) demonstrate, however, that when mitochondrial A$ is reduced 25% import is inhibited only 25%. Thus, at least 75% of maximal import inhibition observed in the presence of F1P1-32+2 and F1P21-51+3 is likely to result from competition for a specific import step.
The data in Fig. 3 indicate that some but not all import competition noted with the longer FIB peptides might be due Values (% inhibition) are expressed as percent inhibition of mature FIB formation as compared mature FIP formed in import reactions in the absence of peptide which was arbitrarily set at 0% in each case.
to dissipation of the threshold A+ required to maintain import competency of mitochondria. It was reasoned that if membrane potential is necessary for import, the movement of competitor peptides through the import apparatus should compete for a portion of A$ required for efficient import. If this were so two conditions should be demonstrable to document the specificity of import inhibition by FIB peptides. The inhibition by peptides should be reversible and the inhibition should be competed by full-length precursor. To test the first condition we asked if inhibition of import by FlPl-51+3 was reversible. Mitochondria were first incubated with 0 or 1.5 FM peptide, a concentration which dissipates mitochondrial A$ about 65% (Fig. 30), and 35S-labeled F,P precursor in reticulocyte. After initial incubation, mitochondria from respective reaction mixtures were reisolated, washed, resuspended in import buffer, and divided into three equal parts. One-third was assayed directly for import in the first incubation, another was reincubated with untranslated reticulocyte lysate in the absence of peptide, and the final third was reincubated with additional 35S-labeled precursor in reticulocyte lysate in the absence of peptide. In the primary incubation, FI@1-51 reduced import to about 10% of control levels (Fig. 4). However, import competence was restored to control levels after wash and reincuhation of peptide treated mitochondria (Fig. 4). Control experiments with valinomycin indicate that maturation of precursor in the second incubation is dependent upon A$ and thus represents precursor imported into the mitochondrial matrix (not shown). Hence, these data demonstrate that import inhibition by Fl[3 peptides is transient and does not result in permanent damage to mitochondria.
In a second experiment we asked if import inhibition by FIB peptide could he overcome by increasing the amount of full-length precursor protein present in import reactions. This was accomplished by incubating mitochondria in separate reaction mixtures with 1 and 30 p1 of reticulocyte lysate containing '"'S-labeled FIBor FIR-subunit precursors.
Twenty-nine pl of untranslated reticulocyte was added to reaction mixtures containing 1 p1 of programed lysate to make all reactions chemically equivalent. The presence of Fl[jl-51+3 reduced import of pFltv and pFI& about 90 and 75%, respectively, when 1 p1 of labeled precursor was present in import reactions (Fig. 5 , R and C). However, when 30 pl of pFln was present in a parallel import reaction, import was inhibited only 50% hy F1@1-51+3 (Fig. 5, E and F ) . pF1/3 was less effective a t competing with FIB peptide for import inhibition (Fig. 5 , E and F ) , this may indicate that the import apparatus has a higher affinity for pFln than pF1& Both pFln and pFl@ were imported a t higher efficiencies when 30 times precursor was present in import reactions (Fig. 5 , R and E ) . It is not clear why this result was observed. I t could he that increased import eff'iciency results from a greater percentage of precursor being availahle for import a t high precursor concentration because a lower percentage of total precursor is tied up in nonspecific interactions during import reactions.

FIG. 4. Inhibition of import by Fl,31-R1+3 is reversible. A ,
mitochondria ((io0 pg/ml) were incubated with reticulocyte lysate containing""S-lnt)eled pF,/j in 300 pI of reaction mixtures as described in the legend t o Fig. 2. After initial incubatinn for 20 min at 25 "C, reactions were divided into three equal parts, mitochondria were reisolated from each, and then resuspended in import buffer and either analyzed directly for import in the lirst incubation or reincubated in 100 pl of import reactions a t 25 "C for 25 min. One aliquot of mitochondria from each initial import condition (no addition and + peptide) was reincuhated. in the ahsence of F,P peptide, with untranslated reticulocyte lysate or lysate containing I" S-labeled precursor. *denotes the peptide concentration mitochondria were treated with in the first inrnt)ation. After the serond incubation mitochondria were reisolated and analyzed for import as descrihed under "Experimental J'rocedures." N , quantitation of hands corresponding to the mature protein in the Iluorographed gel. To calculate import of pF,b into mitochondria in the second incutmtion, mature protein present in mitochondria resultant from the initial import reaction was subtracted from mature protein present in mitochondria after reincubation with ."S-laheled pF,/j. (/,,'-(' and F -I ) = import in the second reaction). Irrespective of this, from these data it is clear that mitochondria remain import competent in t.he presence of F1p peptides and that import inhihition is likely to result from competition between full-length precursor and peptide for specific import step(s).
Since Fl[31-51 inhihited import of two presequence containing peptides we next asked what effect peptide had on import of a precursor which lacked a presequence. The adenine nucleotide transporter (pAAC), which contains several internal targeting domains, was used for this purpose. Import of pAAC unlike pFIB does not require a memhrane potential across the mitochondrial inner memhrane to deliver the precursor to a protease protected location. However, memhrane insertion of pAAC into the inner membrane does require A$ across the inner membrane (Pfanner ~t al., 1987). In earlier studies, the delivery of pAAC to its transmemhrane location in the inner membrane was readily determined by defining the extent to which the protease protected form becomes resistant to extraction hy sodium carhonate at pH 11.5 Smagula and Douglas, 1988). After the simultaneous import of the pFln and the pAAC protein, the proteins associated with mitochondria and protected from external protease were respectively characterized for their extractahility by sodium carhonate. Shown in the center section of Fig.  6 (no additions), are three lanes displaying total imported protein and the protein of the total remaining in the supernatant or within the pellet following extraction with sodium carbonate, respectively. In the ahsence of peptide, 50"; of input pFln was imported to a protease protected space. Addition of F1b1-51+3 hlocked entry of pFIn to a protected space by ahout 95% (Fig. 6). In cont,rast, import of pAAC is reduced only about 207;. Quantitation of the total imported AAC protein which became resistant to carhonate extraction revealed that the presence of Fl[31-51+3 had no effect on the energetics required for membrane insertion. Fig. 6 shows that slightly over 50% of the total protected pAAC is transmembrane, independent of the presence of the inhihitory peptide. In contrast, in the ahsence of membrane potential (+ valinomycin), the amount of pAAC resistant to carhonate extraction was reduced to 16% of the total protected protein. Import of pFln was not detected in the ahsence of memhrane potential. It should be noted that at 10 p~ Fl/~1-5l+3 import of pAAC to a non-carbonate extractahle space was inhihited by ahout 75% (not shown). These results indicate that mitochondria remain import competent for pAAC in the presence of FIB]- incubation. mitochondria were reisolated. and resuspended in 100 pI of SEM buffer containing 100 pg/ml proteinase K and incubated for 30 min on ice. Phenylmethylsulfonyl fluoride was then added to 1 mM and reaction mixtures were split into 2 aliquots. One aliquot was made 0.1 m in sodium carhonate, pH 11.5, and incubated on ice for 30 min while the other was stored on ice until analysis for import. Following incuhation with sodium carbonate, soluhle proteins were separated for integral memhrane proteins hy centrifugation in a Reckman airfuge for 30 min a t 30 psi. and hoth the pellet and supernatants were analyzed for import as descrihed under "Experimental Procedures." When added Fl[fl-51+3 and valinomycin were 1.5 and 2.0 p~, respectively. Precursor protein associated with washed mitochondrial pellets was quantitated hy scanning the fluorographed gel with a densitometer.
51+3 at concentrations which completely block import of presequence containing precursors. Further, these data infer that pAAC and presequence-containing precursors have different affinities for the import apparatus and may require A$ to different extents to drive insertion into the inner membrane.

FIB Peptides Block Import at a
Step beyond Binding to Mitochondria-Fl@ presequence peptides completely block import without diminishing binding of pFl@ to mitochondria (Fig. 2). Additionally, inhibition of pFln import by FIB presequence peptides is accompanied by a large accumulation of precursor protein on the mitochondrial surface which is not observed in the absence of peptide (Fig. 5, R, C and E , F ) .
These results are of interest since binding of precursor proteins to the outer mitochondrial memhrane is proposed to involve specific interaction of presequences with trypsinsensitive import receptors (Pfanner et al., 1991). To examine Fl@ precursor binding to mitochondria in greater detail, we tested presequence dependence and trypsin sensitivity of binding as criteria for specificity. We compared the binding the pFl@ mutant F1@A4-37, which is missing amino acids 4-37, to that of wild type pFl@ (Fig. 7). In deenergized mitochondria (+ valinomycin) 14 and 20% of input pFl/3 hound to mitochondria in the absence and presence Fl/31-51+3, respectively (Fig. 7, R and C). When import by energized mitochon-  Fig. 2. Mock treated mitochondria were incuhated on ice for 30 min with trypsin (0.1 mg/mgof mitorhondria) which was previously inactivated with a 2-fold weight exress of hovine trypsin inhihitor. Trypsin-treated mitochondria were incuhnted as ahove except hovine trypsin inhihitor was added after incuhation. Mitochondria were then reisolated and resuspended t o the original concentration of 5 mg/ml in SEM and used for import nssnys. \\'hen present, valinomycin and Fl/fl-51+3 were 2.0 and 1.5 p~, respectively. fmw A represents hinding ohserved in the ahsenre o f mitochondria in import reactions. Ihnds on the fluorographed gels rorresponding to precursor and mature protein were qunnt itnted hy laser densitometry and are expressed in arhitrary units. dria was completely blocked by Fl/jl-51+3, the amounts of precursor hound to mitochondria were the same as those hound in the absence of peptide (Fig. 7 , I1 andR). Binding F1[jA4-37 to mitochondria was 14-and 50-fold lower than that observed with pFI& under energized and deenergized conditions, respectively ( R versus G and D versus I ) . Thus, results with F1/3.14-37 show that although binding of precursor protein to mitochondria is not competed by presequence peptide, binding is presequence-dependent.
The role of import receptors in mediating the ohserved mitochondrial binding of pFl/3 was addressed hy comparison of pFID binding to control mitochondria and to mitochondria digested with trypsin a t concentrations which reduce import hut do not disrupt mitochondrial integrity (not shown). Trypsin treatment of mitochondria reduced import 78% (Fig. 8, I) and H ) , hut residual import remained sensitive to peptide (Fig. 8, H a n d I ) . Consistent with pFIP binding to high affinity sites on the outer membrane, binding of pFIS to mitochondria was reduced about 50-70% by trypsin treatment as compared to mock treated controls (Fig. 8, H versus F, E ucrsus I , and D versus H). Binding of pFl/3 to trypsin-treated mitochondria was insensitive to FIB peptide (Fig. 8, H versus I ) . Although trypsin digestion reduced pFl@ binding to mitochondria, it appears hound precursor is imported with the same efficiency observed with mock treated mitochondria. This is indicated by the observation that the ratio of precursor to mature protein found associated with hoth control and digested mitochondria is constant (Fig. 8, D and H ) . At present it is not clear whether residual binding and import is mediated by outer membrane import proteins not digested by trypsin or by a bypass route (Pfaller et al., 1988). Nonetheless, pFIP binding to mitochondria is 50-70% dependent on trypsinsensitive elements, is independent of A$, and presequencedependent. However, we observe that binding to the mitochondrial surface is insensitive to Fl[3 presequence peptide. Thus, we conclude that import competition hv Fl/3 peptides occurs a t a site beyond initial binding of precursor proteins to mitochondria.

DISCUSSION
The role performed by mitochondrial presequences in orchestrating specific protein delivery and translocation events within the cell remains obscure. One of several key questions which remains is how these sequences which share only the potential to form an amphiphilic structure mediate specific binding to mitochondria. In the present studv we have examined presequence-dependent binding of the Fl/j-suhunit precursor to isolated mitochondria and the ability of synthetic peptides composed of the NH2-terminal residues of pFl@ to compete for this binding. We find, quite unexpectedly, that presequence dependent binding to mitochondria is not competed for in any detectable manner by Fl@ presequence peptides. However, efficient competition for import of the bound precursor can occur at a point beyond binding to receptor components on the organelle surface. Earlier studies demonstrated that although the Fl@ presequence peptide, F1P1-20, binds with high affinity to phospholipid vesicles, K, = 50 X lo-' M, this peptide is a weak import competitor, Ki 2 100 X 10"j M (Fig. 2, Hoyt et al., 1991). In present study we find that three different F,@ peptides, all 34 amino acids in length, F1@1-32+2, F1@1-32 SQ+2, and F1@21-51+3 compete for import half-maximally at 1000-, 250-, and 25-fold lower concentrations, respectively, than F1P1-20, (Fig.  9B). Qualitative and quantitative predictors of peptide amphiphilic character, Shiffer-Edmundson helical wheel plots and the algorithm of Segrest (Shiffer and Edmundson, 1967;Segrest et al., 1990), respectively, indicate F1@1-20 has the highest predicted hydrophobic moment/residue and charge density on its polar face of all peptides used in this study (Fig.   9, A and B ) . Thus, ability to form an amphiphilic a-helical structure with high charge density appears less important than overall length in conferring ability of FIP presequence peptides to inhibit mitochondrial import. This is supported by the observation that F1@21-51+3 has a calculated hydrophobic moment/residue which is 50% that of F1@1-20, but is 25 times more potent as an import competitor (Fig. 9B). On the other hand, F1@1-32+2 has about 30% more amphiphilic character, exhibits a higher charge density, and is 33 times more potent as an import competitor than F1@21-51+3 (  (Shiffer and Edmundson, 1967). Shaded circles represent basic amino acids.

B, tabulation of the inhibition constants and calculated values for
the mean hydrophobic moment/residue and mean hydrophobicity/ residue of non-polar face. Inhibition constants were determined from data presented in Fig. 2. The other values were calculated using an algorithm designed to predict the amphiphillic character of proteins (Segrest et al., 1990). 9, A and B ) . Based on the above observations we propose that both minimal length and minimal amphiphilic character are required for functional import signals. In the case of the F1@subunit precursor, the amphiphilicity of the presequence appears sufficient to direct import but the minimal length necessary for function requires additional residues. This would explain the observation that 14 amino acids of the FIB presequence can direct in vitro import of DHFR (Walker et al., 1990) but the synthetic 20-residue F1@ presequence peptide fails to efficiently compete for import ( Fig. 2; see also Hoyt et al., 1991). Import competition by Fl@l-51+3 and other synthetic presequence peptides (Gillespie et Glaser and Cumsky, 1990) appears to take place within the contact site very near or at the membrane potential-dependent step. We propose that the peptide itself competes for energy-dependent transport of the full-length precursor in the contact site. This model would propose that when inhibitory peptide concentration is sufficient to saturate the A#-dependent translocation device, import of peptide would consume a large portion of A+ necessary to drive transport of other proteins. Upon completion of transport or removal of excess peptide, membrane potentials would reestablish for further energy-dependent activities. Several observations in the present study support this view. First, F1@1-20 binds avidly to membranes but is not able to penetrate deeply into phospholipid bilayers and disrupt phospholipid vesicles (Hoyt et al., 1991). Second, although dissipation of mitochondrial A# was observed in the presence of longer Fl@ peptides less than 25% of maximal import inhibition can be attributed to this mechanism when import inhibition by F1@1-32+2 and F1@21-51+3 is compared to that by the uncoupler valinomycin (Fig. 3, A and C). Third, unlike valinomycin-treated mitochondria, if peptide-treated mitochondria are washed and reincubated in the presence of additional precursor, efficient uptake of newly added precursor is observed (Fig. 4). Fourth, in the presence of Fl@ peptide at concentrations sufficient to completely block import of pFla, A#-dependent import of pAAC into the mitochondrial inner membrane is not affected. Fifth, peptide-dependent inhibition of import is reduced by addition of a 30-fold more full-length precursor to import reactions (Fig. 5). The inclusion of additional substrate for energy-dependent transport would not be expected to overcome inhibition by peptide if the full-length substrate and inhibitory peptide were acting at different places in the membrane. Thus, this result confirms that the two forms of the precursor directly compete in some manner at a specific import step We also observe that peptide consisting of the import signal failed to inhibit binding of lysate translated full-length precursors to mitochondria (Figs. 2, 5, 7, and 8). These observations suggest that recognition events on the mitochondrial surface which may require outer membrane proteins (Sollner et al., 1989(Sollner et al., , 1990Hines et al., 1990) may involve more than the presequence. Support for this conclusion is provided by the following observations. First, when import of F1a was blocked by peptide Fl@l-51+3 a significantly larger amount of precursor was found associated with mitochondria than in the absence of peptide (Fig. 5, B, C, E, and F ) . Similarly, at concentrations of Fl@ peptide which completely block import of pF,@ no decrease in the binding of Fl@ precursor to mitochondria is observed (Fig. 7). This association of pFl@ with mitochondria is dependent on the presequence (Fig. 7), independent of a membrane potential across the inner membrane (Fig. 7) and greatly enhanced by trypsin-sensitive elements on the mitochondrial surface (Fig. 8). This mitochondrial association of precursor on the surface appears (Figs. 7 and 8) as it has in earlier studies to be on the import pathway (Reizman et al., 1983). Following a first incubation in the presence of an inhibitory concentration of Fl@l-51+3, 30-40% of the bound but non-imported Fl/3 precursor is imported following removal of peptide and reincubation of mitochondria (not shown).
In comparing results presented here with those of previous studies with synthetic presequence peptides, we find that our results are in close agreement with observations made in import competition studies with synthetic truncated forms of the Cox IV presequence which still retain targeting information. (Glaser and Cumsky, 1990). As in our experiments, dissipation of mitochondrial A+ by Cox IV presequence peptides was not responsible for import inhibition, and a competitive mechanism for import inhibition was indicated (Glaser and Cumsky, 1990). Additionally, import inhibition by Cox IV presequence peptides was not associated with a reduction in binding of precursor proteins to mitochondrial membranes (Glaser and Cumsky, 1990). Further experiments showed that Cox IV presequence peptide blocks import of pAAC at a site beyond translocation into the outer mitochondrial membrane but before insertion into the inner mitochondrial membrane (Glaser and Cumsky, 1990). Fl@ presequence peptides were found to block pAAC import at a similar site.' Thus, results from import competition studies with pAAC support interpretations from binding experiments that specific inhibition of mitochondrial import by synthetic presequence peptides occurs at a site beyond initial interaction of precursor proteins with mitochondria.
The present data are consistent with a model in which the presquence would serve to organize soluble factor(s) onto the precursor either co-or post-translationally. The factor(s) in this complex would maintain the precursor in a transport competent state and assist in targeting the precursor to the mitochondrial surface. It is quite possible that import receptors recognize both structural features of the precursor and soluble factor(s) to mediate specific binding. Such a model was recently proposed for the role of secB in facilitating binding of precursor proteins to the export apparatus in E. coli. (Hartl et al., 1990).
Failure of synthetic Fl/3 peptides to block binding of precursor proteins to mitochondria indicates they do not present a conformation or organization which promotes the binding of factors necessary for high affinity binding to import receptors. Instead they probably gain access to membrane-bound import components directly. This is indicated by ability of FIP peptides to compete for residual import in trypsin treated mitochondria (Fig. 8, H and I). The ability of mitochondrial precursors to be imported independent of the import receptor system is not without precedent. Presequence dependent import of cytochrome oxidase subunit Va (COX Va) precursor occurs at 100% of control rates in mitochondria treated with trypsin at concentrations which abolish import of other precursors (Miller and Cumsky, 1991). F1@ peptides also compete for import of COX Va demonstrating they share a common import step (not shown). Interestingly COX Va import lacks or requires very low levels of ATP hydrolysis in the cytosol (Miller and Cumsky, 1991) demonstrating a relationship between bypass of trypsin-sensitive import factors and lack of a cytosolic ATP requirement for import of an authentic mitochondrial precursor. The above data suggest that proteinaceous import factors on the mitochondrial surface serve to recognize and discharge factors associated with precursor proteins which require factors for import and that some precursor proteins which are naturally import competent may D. M. Cyr and M. G. Douglas, unpublished observations. not require the assistance of ATP-dependent translocation factors or proteinaceous import receptors.