Mft52, an acid-bristle protein in the cytosol that delivers precursor proteins to yeast mitochondria.

We have identified a novel protein, Mft52, in the cytosol of yeast cells. Mft52 has a two-domain structure that includes a receptor-like carboxyl-terminal “acid-bristle” domain, which binds basic, amphipathic mitochondrial targeting sequences. Native Mft52, purified from the cytosol of yeast cells, is found as a large particle eluting in the void volume of a Superose 6 gel filtration column. Fusion proteins, consisting of mitochondrial targeting sequences fused to nonmitochondrial passenger proteins, are targeted to mitochondria in wild-type yeast cells, but defects in the gene encoding Mft52 drastically reduce the delivery of these proteins to the mitochondria. We propose that Mft52 is a subunit of a particle that is part of a system of targeting factors and molecular chaperones mediating the earliest stages of protein targeting to the mitochondria.

We have identified a novel protein, Mft52, in the cytosol of yeast cells. Mft52 has a two-domain structure that includes a receptor-like carboxyl-terminal "acidbristle" domain, which binds basic, amphipathic mitochondrial targeting sequences. Native Mft52, purified from the cytosol of yeast cells, is found as a large particle eluting in the void volume of a Superose 6 gel filtration column. Fusion proteins, consisting of mitochondrial targeting sequences fused to nonmitochondrial passenger proteins, are targeted to mitochondria in wild-type yeast cells, but defects in the gene encoding Mft52 drastically reduce the delivery of these proteins to the mitochondria. We propose that Mft52 is a subunit of a particle that is part of a system of targeting factors and molecular chaperones mediating the earliest stages of protein targeting to the mitochondria.
Eukaryotic cells are compartmentalized. Ribosomes in the cytosol translate proteins bound for all intracellular compartments, and newly synthesized proteins have to be sorted and delivered to their correct subcellular destination. The fate of secretory proteins is determined from the time that the aminoterminal signal sequence emerges from the ribosome; a signalrecognition particle binds the hydrophobic signal sequence of the nascent secretory protein, designating the ribosome-nascent chain complex for delivery to the endoplasmic reticulum (1)(2)(3)(4). Because almost all precursor proteins destined for the mitochondria also carry a targeting sequence at the extreme amino terminus, we have begun a search for cytosolic factors that might bind mitochondrial precursors at a similarly early stage of translation, promoting the delivery of nascent precursor proteins to the mitochondria.
At least two molecular chaperones are known to stimulate the import of precursor proteins into mitochondria: HSP70 1 (5-7) and MSF (8,9). However, neither of these chaperones specifically recognizes the targeting sequence of mitochondrial precursor proteins; even MSF, which is selective in its binding of protein substrates, binds to mitochondrial precursors that do not carry amino-terminal targeting sequences and binds poorly if at all to isolated targeting sequence peptides (9). By definition, chaperones such as HSP70 and MSF would be recruited to the nascent precursor protein after most of the polypeptide had been synthesized.
It is clear that precursor proteins could begin translocation across the mitochondrial membranes before translation is complete. Indeed, ribosomes can be visualized on the mitochondrial surface in situ (10) and have been shown to be actively translating mitochondrial precursor proteins (10,11). It is also clear that the cotranslational import of precursor proteins in vitro is an efficient process, ensuring that every molecule of precursor made will be imported into the mitochondria (12,13). What is not yet clear is how the ribosome, programmed to translate a mitochondrial precursor protein, could be recognized and then delivered to the surface of the mitochondria in vivo.
Here we identify a novel cytosolic protein, Mft52, that functions in the delivery of proteins to the mitochondria. In particular, Mft52 is required for the delivery of fusion proteins such as COXIV-DHFR, where the only targeting information is encoded at the extreme amino terminus of the precursor protein, the first part of the precursor to emerge from the ribosome. Mft52 has an "acid-bristle" domain, homologous to that found in the presequence-binding subunits of the import receptor on the mitochondrial surface (14,15). We propose that Mft52 functions to usher nascent precursor proteins to the mitochondria.

Expression of Mft52 in Escherichia coli-
The open reading frame encoding Mft52 was amplified by polymerase chain reaction using the primers 5Ј-G GCG GGA TCC ATG GCT CTG TCA CAA AAA CAA ATA G-3Ј and 5Ј-CAG CGG ATC CAA GCT TGC ATT ATA CGT GGT CAT TT-3Ј and ligated into the plasmid pQE9 (Qiagen) to produce a hexahistidine-tagged form of Mft52. The hexahistidine-tagged protein was expressed in a lon Ϫ strain of E. coli (a kind gift of Sabine Gratzer) and purified using Ni-NTA resin (Qiagen) according to the manufacturer's instructions. Truncated Mft52, was engineered by polymerase chain reaction with the mutagenic primer 5Ј-AAA AGG CAG AGA TCA AGC TTG GAA GTC AAT ACT AT-3Ј. The construct was subcloned into the vector pQE9 and expressed as described above. To incorporate the acid-bristle domain of Mft52 into a GST-fusion protein, the SpeI-XhoI fragment of MFT1 was cloned into the plasmid pGEX-KG (a kind gift of Donald Smith) cut with XbaI and XhoI. This results in fusion of the Mft52 fragment from Leu 175 -Lys 392 to GST.
Antibody Production-Monoclonal antibodies specific for Mft52 were produced from mice immunized with hexahistidine-tagged Mft52. In addition, sheep were injected with the peptide Mft52(T 347 -E 365 ) conjugated to diphtheria toxoid (Chiron Mimotopes) to generate epitopespecific antibodies. The anti-peptide antibodies specifically recognize both Mft52 and Tom20. Antibodies raised in rabbits to cytochrome b 2 , dihydrofolate reductase, and 80 S ribosomes were a kind gift from Jeff Schatz.
Tryptic Digest and Mass Spectrometry-Polyacrylamide gel slices containing approximately 5 g of protein were reduced in the presence * This project was supported by a grant from the Australian Research Council (to T. L.) and an Australian Postgraduate Research Award (to T. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. Tel.: 61 3 9479 2156; Fax: 61 3 9479 2467; E-mail: T.Lithgow@latrobe.edu.au. 1 The abbreviations used are: HSP70, 70-kDa heat shock protein; MSF, mitochondrial import stimulating factor; COXIV-DHFR, fusion protein composed of subunit IV of the cytochrome oxidase complex fused to dihydrofolate reductase; PAGE, polyacrylamide gel electrophoresis; MFT, mitochondrial fusion protein targeting factor; GST, glutathione S-transferase; F 1 ␤-lacZ, fusion protein of ␤-subunit of the F 1 ATPase and ␤-galactosidase; Tom20 and Tom22, 20-kDa and 22-kDa, respectively, translocation component in the mitochondrial outer membrane. of dithiothreitol and alkylated with 4-vinylpyridine (16). The proteins were digested in situ with 0.2 g of trypsin at 37°C for 16 h. The tryptic fragments were extracted by sonication in 1% trifluoroacetic acid and then in 0.1% trifluoroacetic acid/60% acetonitrile. The extracts were pooled, lyophilized, dissolved in 4 mM ammonium acetate/0.1% formic acid, and chromatographed by reversed phase-high performance liquid chromatography. Analysis of the tryptic digest was by on-line electrospray mass spectrometry.
Synthetic Peptides and in Vitro Binding Assay-The synthetic peptide representing the targeting sequence from subunit IV of cytochrome oxidase (COXIV; MLSLRQSIRFFKPATRTLCSSR) was synthesized commercially (Chiron Mimotopes). Synthetic peptides representing the mitochondrial targeting sequence of chaperonin 60 (CPN60; MLR-LPTVLRQMRPVSRALAPHLTRAYC), a non-amphipathic analog of the COXIV sequence (SYNB2; MLSRQQSQRQSRQQSQRQSRYLL), a highly-basic segment from ribosomal protein S6 (S6 (229 -249) ; AKRRRLSSLRASTSKSEESQK) and the uncharged, amphipathic peptide SCC1-19M were kind gifts from Peter Høj, Jeff Schatz, Paul Jenö, and Katsuyoshi Mihara, respectively. Purified hexahistidine-tagged Mft52 was diluted in binding buffer (50 mM sodium phosphate, pH 7.0) and bound to Ni-NTA resin by incubation on ice for 30 min. The immobilized Mft52 was incubated with the indicated amount of peptide in binding buffer (100 l) for 30 min on ice. The resin was collected by centrifugation, washed twice in binding buffer, and analyzed by SDS-PAGE (17) and silver staining. To competitively inhibit precursor binding to Mft52, synthetic peptides (25 M) were preincubated with GST-Mft52 in 50 mM potassium phosphate (pH 7.4) for 30 min on ice, and in vitro translated 35 S-labeled Su9-DHFR was added for a further 30 min. Unbound material was washed away with 50 mM potassium phosphate, pH 7.4, and the bound precursor was analyzed by SDS-PAGE and fluorography. The plasmid encoding Su9-DHFR was a kind gift from Klaus Pfanner.
Deletion of the MFT1 Gene-To delete the MFT1 gene (18) from haploid yeast cells, the polymerase chain reaction-amplified DNA fragment in pQEMFT1 was digested with PstI and EcoRV, and a PstI-XbaI (blunt) fragment of the LEU2 gene was ligated into the plasmid, deleting the sequence encoding Mft52 after the codon corresponding to Leu 75 . The LEU2⌬mft1 fragment was transformed into the yeast strain JK9 -3d to generate the ⌬mft1 strain (YTHB2: Mat␣, ura3, leu2, his4, trp1, LEU2⌬mft1). To test that the hexahistidine-tagged form of Mft52 is functional, the EcoRI-HindIII fragment containing the modified open reading frame was subcloned from pQE9 into pBluescript (Stratagene), and then the EcoRI-ClaI fragment was subcloned into the yeast expression vector YPGE (19). Yeast cells (YTHB2) were transformed with the plasmid and tested for growth on rich media at 37°C. Yeast cells carrying the YPGE::MFT1 plasmid overexpress Mft52 (3-5-fold as judged by immunoblot analysis) and grow as fast as wild-type cells at 37°C.
Subcellular Fractionation-Yeast cells were grown in semisynthetic lactate media to mid-log phase and are converted to spheroplasts (20). To prepare the cytosolic fraction, spheroplasts were homogenized in RNasin buffer (20 mM Hepes, pH 7.4, 0.6 M sorbitol, 5 M vanadylcomplex, 1.25 g/ml leupeptin, 0.75 g/ml antipain, 0.25 g/ml chymostatin, 5 g/ml pepstatin, and 0.5 mM phenylmethylsulfonyl fluoride) and centrifuged for 10 min at 10,000 rpm (Sorvall SS-34 rotor) and then for 30 min at 55,000 rpm (Beckman TL100.3 rotor). The protein concentration of the cytosolic fraction was around 20 mg/ml. Mitochondria prepared from the 10,000 rpm pellet were purified on Nycodenz gradients and stored frozen as described (21). Protein concentration was determined by Coomassie Blue staining (22), and 100 g of protein were typically used per lane for immunoblot analysis.
The cytosol fraction (0.5 ml) was loaded onto either a Superdex-75 column or a Superose 6 column equilibrated in 50 mM sodium phosphate, 50 mM sodium chloride, pH 7.0, and the eluant was monitored by absorbance at 280 nm. Fractions (1 ml) were collected and assayed by SDS-PAGE and immunoblot analysis. In control runs, to rule out the presence of phospholipid micelles in the extract, 0.1% Brij58 was added to the column buffer.
Miscellaneous-DNAsis software (Hitachi) was used for sequence alignments and secondary structure predictions. The plasmid YCpF 1 ␤-lacZ (C␤Z1) was a gift from Scott Emr, and the plasmid YEpCOXIV-DHFR (pKSE) was a gift from Jeff Schatz.

RESULTS
The MFT1 Gene Encodes Mft52, a Protein That Enhances Precursor Delivery to the Mitochondria-To isolate yeast mutants compromised in the targeting of proteins to the mitochon-dria, Emr et al. (23) devised an elegant selection system based on a toxic lacZ fusion protein. In wild-type yeast cells, efficient delivery of the F 1 ␤-lacZ fusion protein (composed of the aminoterminal targeting sequences of the ␤-subunit of the F 1 -ATPase fused to ␤-galactosidase) from the cytosol to the mitochondria leads to a respiratory growth defect. Mutations in two genes, MFT1 and MFT2, lead to a reduction in the delivery of the F 1 ␤-lacZ fusion protein to mitochondria (24).
Fusion proteins consisting of mitochondrial targeting sequences fused to the passenger protein dihydrofolate reductase (DHFR) have been important tools in dissecting the protein import machinery. In particular, proteins such as COXIV-DHFR that carry a minimal mitochondrial targeting sequence have enabled us to distinguish the subunits of the import receptor that specifically interact with mitochondrial targeting sequences (Tom20 and Tom22; Refs. 15 and 25-27) from those that recognize more global aspects of a precursor protein (Tom37 and Tom70; Ref. 28).
To determine the extent of the protein-targeting defect in mft1-1 cells, we transformed wild-type and mutant mft1-1 cells with pKSE, a plasmid encoding COXIV-DHFR (29). Subcellular fractionation of wild-type and mutant mft1-1 cells revealed that delivery of fusion proteins to the mitochondria is generally compromised in mft1-1 mutant yeast cells (Fig. 1A). Thus, in the case of these proteins whose only targeting information is the basic, amphipathic amino terminus, the protein encoded by the MFT1 gene is required for precursor delivery to the mitochondria.
Antibodies were raised against the protein encoded by the MFT1 gene, to determine its subcellular location in yeast cells. A protein with an apparent molecular mass of M r 52,000 was identified in the cytosol of yeast cells (Fig. 1B, lane 5). According to the standard nomenclature for components in the mitochondrial protein import pathway (30), we refer to this protein as Mft52. Mft52 Is a Component of a Cytosolic Particle-To determine the native molecular size of Mft52, we subjected a yeast cytosolic extract to gel-filtration chromatography. Native Mft52 is not a monomer, because it elutes in the void fraction of a Superdex-75 column ( Fig. 2A). To better resolve the size of Mft52, we used a Superose 6 column. Fig. 2B shows only the earliest fractions from this column. As a marker for resolution of the column, the clathrin heavy chain, Chc1, is shown (identified by amino-terminal sequencing). Immunoblot analysis of all of the column fractions revealed that native Mft52 was recovered quantitatively in the void fraction of the Superose 6 column (i.e. 7.5-8.0 ml, corresponding to an apparent native molecular size greater than 5000 kDa), ahead of the clathrin triskelions. Note that the leading edge of the clathrin peak overlaps the void fraction, so that some Chc1 is seen in these fractions.
The huge apparent size of the native protein was surprising, given that the Mft52 subunit purified from E. coli chromatographs at 50 -100 kDa as would be expected for a monomer or dimer (data not shown, see below). We conclude that native Mft52 is a subunit of a large particle in the cytosol of yeast cells.
Many of the other proteins present in the void fraction of the Superose 6 column appear to be subunits of one or more ribonucleoprotein particles. After pretreatment of the cytosolic extract with RNase, these proteins (in the range 10 -45 kDa by SDS-PAGE) are removed from the void fraction (Fig. 2C). Mft52 is unaffected by RNase treatment, coeluting in the void fraction of the RNase-treated cytosol with proteins of subunit size 65, 90, and 170 kDa.
Sequence Homology of Mft52 to Acid-bristle Proteins-Sequence analysis revealed that Mft52 has two short regions of sequence similarity to the yeast proteins Tom20 and Tom22 (Fig. 3A), and antibodies raised to the peptide Mft52(T 347 -E 365 ) cross-react with Tom20 on Western blots (data not shown). Tom20 and Tom22 are partner subunits of the import receptor, and both have been shown to be in direct contact with the basic, amphipathic targeting sequence of mitochondrial precursor proteins bound at the mitochondrial surface (25)(26)(27). The acidbristle sequences in Tom20 and Tom22 define the sites responsible for recognition and binding of targeting sequences (15). The sequence segments shared between Tom20/22 and Mft52 correspond precisely to the acid-bristle sequence segments. In Mft52, the two acid-bristle sequences are predicted to sit in tandem in a carboxyl-terminal, four-helix domain (data not shown).
Domain Structure of Mft52-To produce sufficient Mft52 to investigate the predicted domain structure, we added a hexahistidine tag and overexpressed the protein in E. coli. The hexahistidine-tagged form of Mft52 was functional, because it complemented the temperature-sensitive defects of ⌬mft1 yeast cells (data not shown, see "Materials and Methods"). Recombinant Mft52 was purified from the soluble fraction of E. coli cells, but a substantial proportion was always recovered as a 34-kDa fragment (Fig. 3B). The 34-kDa form was identified as an amino-terminal fragment of Mft52 by amino-terminal sequence analysis. Thus, in addition to the acid-bristle domain, Mft52 has an amino-terminal, protease-resistant domain, which can be purified from E. coli.
To identify the protease-sensitive region that defines this domain, we prepared a complete trypsin cleavage map of both the intact (52 kDa) and fragmented (34 kDa) forms of Mft52. Mass spectrometry of the complete trypsin digests revealed that the most carboxyl-terminal peptide that could be recovered from the 34-kDa fragment corresponded to the sequence ending at Arg 287 . This suggests that in Mft52, Met 1 -Arg 287

FIG. 2. Mft52 is a component of a large particle in the cytosol of yeast cells.
A, the cytosolic fraction (5 mg of protein) was chromatographed on a Superdex-75 column (100-ml bed volume), and elution of Mft52 was monitored by immunoblotting (upper panel). Replica samples were loaded onto SDS-polyacrylamide gel for analysis by Coomassie Blue staining (lower panel). As calibration controls, blue dextran (equivalent to a 5000-kDa globular protein) elutes at 45 ml, immunoglobulin (150 kDa) elutes at 55 ml, serum albumin (66 kDa) elutes at 64 ml. B, the cytosolic fraction (12 mg of protein) was chromatographed on a Superose 6 column (bed volume, 25 ml), and the early fractions were assayed by SDS-PAGE and silver staining. The identity of Mft52 was confirmed by immunoblotting (data not shown). As calibration controls, blue dextran (equivalent to a 5000 kDa globular protein) elutes at 7.5 ml), thyroglobulin (660 kDa) elutes at 11 ml, and apoferritin (440 kDa) elutes at 13 ml. The heavy chain of clathrin (Chc1) was identified by amino acid sequencing. C, yeast cytosol extract (12 mg of protein) was treated with 40 g of RNase A for 30 min at 30°C and then chromatographed on the Superose 6 column. The eluant fraction after 7.5 ml is compared from extracts treated (ϩ) or not treated (Ϫ) with RNase. The identity of Mft52 was confirmed by immunoblotting, and Chc1 was identified by amino-terminal sequencing. The size of the other three proteins in the fraction is indicated.
includes the protease-resistant domain.
To be certain that the putative amino-terminal fragment represents a protease-resistant domain, we engineered a Hin-dIII restriction site in the open reading frame at the codons corresponding to Met 282 and Val 283 and expressed the truncated Mft52 in E. coli. The truncated Mft52, purified from E. coli, comigrates with the 34-kDa fragment on polyacrylamide gels (Fig. 3B), and had a mass of 34,324 Da as determined by mass spectrometry. Allowing for the amino acids included in the cloning procedure, this corresponds exactly to the predicted mass of the protease-resistant domain Met 1 -Val 283 . Gel filtration chromatography allowed us to estimate the native size of the purified Mft52 subunit as 50 -100 kDa, suggesting that it is a globular monomeric, or perhaps dimeric, protein (data not shown).
Mft52 Binds Mitochondrial Targeting Sequences-The sequence similarity to Tom20/Tom22 suggested that Mft52 might bind mitochondrial targeting sequences. To test this directly, purified Mft52 was incubated with synthetic peptides representing a functional mitochondrial targeting sequence (COXIV; Ref. 29) and a nonfunctional mutant targeting sequence (SCC) from mitochondrial cytochrome P-450 (SCC1-19M; Refs. 31 and 32). Fig. 4 shows the binding of the COXIV targeting sequence by Mft52. Binding of the COXIV peptide was saturable, with half-maximal binding at a peptide concentration of around 10 Ϫ6 M (data not shown), but Mft52 fails to bind the control SCC peptide (Fig. 4A).
To further define the requirements for peptide binding, we grafted the acid-bristle domain of Mft52 onto GST. A binding assay was set up with an in vitro translated precursor protein Su9-DHFR (consisting of the 69-amino acid targeting sequence from subunit 9 of the mitochondrial ATPase fused to DHFR; Ref. 33). The acid-bristle domain of Mft52 can mediate the binding of Su9-DHFR, but the basic amphipathic sequences from COXIV and CPN60 compete for binding (Fig. 4B). As controls for specificity of binding to Mft52, neither the uncharged, amphipathic peptide SCC nor the basic, hydrophillic peptides S6 (229 -249) (34) or SYNB2 (15) could compete effectively for precursor binding.
We, therefore, conclude that Mft52 is a subunit of an oligomeric particle, and that the Mft52 subunit can mediate binding of basic, amphipathic mitochondrial targeting sequences.

Targeting of Proteins to the Mitochondria: the MFT Genes-
The process of mitochondrial protein import has been the subject of intense investigation. Details on the mechanisms by which precursor proteins can be recognized by the import receptor on the mitochondrial surface and translocated across the mitochondrial membranes are becoming more and more clear (35)(36)(37)(38). But how does a nascent mitochondrial precursor find its way from a ribosome to the surface of its target organelle? It seems unlikely that this is achieved exclusively by diffusion through the cytosol after translation is complete. We are attempting to define components of the protein targeting machinery of the cell that can direct nascent precursor proteins toward the mitochondria.
Many precursor proteins are assisted in their delivery to the mitochondria by molecular chaperones such as MSF and HSP70. However, a second class of precursors, typified by COXIV-DHFR and similar fusion proteins, do not use these chaperones, suggesting that an ATP-independent targeting factor recognizes the targeting sequence of these fusion proteins (28,39). Two subunits of the import receptor, Tom20 and Tom22, provide acidic domains responsible for recognition and binding of the basic, amphipathic targeting sequence of both "classes" of mitochondrial precursor protein. Thus, fusion proteins like COXIV-DHFR provide a specific means to identify those cytosolic targeting factors that interact directly with the mitochondrial targeting sequence.
We have shown that in mft1 mutant yeast cells, COXIV-DHFR fails to be delivered to the mitochondria, and we have identified and characterized the cytosolic Mft52 protein encoded by the MFT1 gene. Mft52 has two segments of homology to the acid bristles of the import receptor subunits Tom20 and Tom22. The Tom20 and Tom22 acid-bristle segments are important in the binding of targeting sequences at the surface of the mitochondria (15). In Mft52, these amino acid segments sit within a domain, defined by secondary structure prediction and by protease mapping.
Additional Components of the Targeting Particle-Mutations in one other gene, MFT2, also reduced delivery of the F 1 ␤-lacZ fusion protein to mitochondria (24). We are currently testing the hypothesis that the MFT2 gene encodes another subunit of the Mft52 particle. The huge apparent size of the Mft52-containing particle suggests that it is not globular; 40 S ribosomes (a globular particle of approximately 4000 kDa) were recovered in the high-speed pellet of subcellular preparations, whereas the cytosolic fraction containing Mft52 was collected in the supernatant. The anomalous behavior of the Mft52 particle under gel filtration strongly suggests that the particle is asymmetric, like the clathrin triskelions (40), and our preliminary analysis of the particle containing Mft52 by sucrose density gradient centrifugation supports this notion. We are currently attempting further purification of the particle to precisely define its subunit structure.
Recently, a genetic screen for factors that could promote the targeting of the yeast protein Mod5 to the mitochondria revealed four new genes that can influence the process (41). Mutations in any one of these genes, named MDP1-MDP4 for Mitochondria Down Protein import mutants, caused an accumulation of the Mod5 reporter protein in the cytosol of the mutant cells. MDP2 has previously been cloned as VRP1, a protein involved in maintaining the actin cytoskeleton, and MDP3 is identical to PAN1, another actin-binding protein (42). These intriguing results suggest that a correctly assembled cytoskeleton is required for the efficient delivery of proteins to the mitochondria in vivo. It will be of interest to learn whether the other MDP genes, MDP1 and MDP4, might be allelic to either MFT1 or MFT2, and whether the particle containing Mft52 might use the cytoskeleton to assist protein delivery to the mitochondria.
Mft52 as a Key Component in a Hierarchy of Targeting Factors-Although yeast mutants deficient for Mft52 fail to deliver fusion proteins containing amino-terminal mitochondrial targeting sequences to the mitochondria, "natural" precursor proteins, such as the intact F 1 ␤ protein, are delivered; mft1 cells remain viable at 25°C, and steady-state levels of F 1 ␤ remain normal in mft1 mutant cells (24). In addition to the aminoterminal presequence, "natural" mitochondrial precursor proteins must have additional elements of targeting information, perhaps in the mature region of the precursor, which allow them to be delivered by chaperones such as HSP70 (5,6,7,43) and MSF (8,9,44), even in the absence of Mft52. Fusion proteins like COXIV-DHFR, which rely solely on the aminoterminal targeting sequence, are dependent on Mft52 to ensure their delivery to the mitochondria. Nascent secretory proteins emerging from the ribosome are bound by components of the targeting machinery. Perhaps the first targeting component to access the emerging nascent chain is nascent chain-associated complex (45), which binds indiscriminately to nascent chains destined for the endoplasmic reticulum, mitochondria, or cytosol (46). We propose that the Mft52 particle acts subsequently, specifically binding mitochondrial targeting sequences and committing these precursor proteins for delivery to the mitochondria. Indeed, a nascent mitochondrial precursor translated in a wheat germ lysate can be specifically cross-linked to both nascent chain-associated complex and a protein of similar size to Mft52 (46).
The involvement of nascent chain-associated complex and the Mft52 particle in protein targeting during translation does not imply that two completely distinct, cotranslational and posttranslational, mechanisms exist for protein import into mitochondria. Rather, we suggest that components such as nascent chain-associated complex and Mft52 initiate a targeting process that could be completed cotranslationally. However, in cases where a substantial length of polypeptide has been synthesized, other chaperones such as HSP70 and MSF would be able to bind the nascent precursor to ensure its delivery to the mitochondria as an import-competent polypeptide. With the identification of Mft52, it is now possible to test this model and dissect the sequence of events from the initiation of translation of a precursor protein until its delivery to the import receptor on the mitochondrial surface.