(Received for publication, August 20, 1996, and in revised form, December 5, 1996)
From the School of Biochemistry, La Trobe University,
Bundoora 3083, Australia and § Department of Biology,
Hamilton College, Clinton, New York 13323
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.
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 amino-terminal signal sequence emerges from the ribosome; a signal-recognition particle binds the hydrophobic signal sequence of the nascent secretory protein, designating the ribosome-nascent chain complex for delivery to the endoplasmic reticulum (1-4). 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: HSP701 (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.
The open reading
frame encoding Mft52 was amplified by polymerase chain reaction using
the primers 5-G GCG GGA TCC ATG GCT CTG TCA CAA AAA CAA ATA G-3
and
5
-CAG CGG ATC CAA GCT TGC ATT ATA CGT GGT CAT TT-3
and ligated into
the plasmid pQE9 (Qiagen) to produce a hexahistidine-tagged form of
Mft52. The hexahistidine-tagged protein was expressed in a
lon
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 5
-AAA AGG CAG AGA
TCA AGC TTG GAA GTC AAT ACT AT-3
. 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 Leu175-Lys392 to GST.
Monoclonal antibodies specific for Mft52 were produced from mice immunized with hexahistidine-tagged Mft52. In addition, sheep were injected with the peptide Mft52(T347-E365) conjugated to diphtheria toxoid (Chiron Mimotopes) to generate epitope-specific antibodies. The anti-peptide antibodies specifically recognize both Mft52 and Tom20. Antibodies raised in rabbits to cytochrome b2, dihydrofolate reductase, and 80 S ribosomes were a kind gift from Jeff Schatz.
Tryptic Digest and Mass SpectrometryPolyacrylamide gel slices containing approximately 5 µg of protein were reduced in the presence of dithiothreitol and alkylated with 4-vinylpyridine (16). The proteins were digested in situ with 0.2 µg 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 AssayThe 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; MLRLPTVLRQMRPVSRALAPHLTRAYC), 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 µl) 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 µM) were preincubated with GST-Mft52 in 50 mM potassium phosphate (pH 7.4) for 30 min on ice, and in vitro translated 35S-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 GeneTo 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 Leu75. The
LEU2mft1 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.
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 µM vanadyl-complex, 1.25 µg/ml leupeptin, 0.75 µg/ml antipain, 0.25 µg/ml chymostatin, 5 µg/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 µg 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.
MiscellaneousDNAsis software (Hitachi) was used for
sequence alignments and secondary structure predictions. The plasmid
YCpF1-lacZ (C
Z1) was a gift from Scott Emr, and the
plasmid YEpCOXIV-DHFR (pKSE) was a gift from Jeff Schatz.
To isolate yeast mutants compromised
in the targeting of proteins to the mitochondria, 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
F1-lacZ fusion protein (composed of the amino-terminal
targeting sequences of the
-subunit of the F1-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
F1
-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 Mr 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 ParticleTo 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 ProteinsSequence
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(T347-E365) 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 (25-27). The
acid-bristle 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 Arg287. This suggests that in Mft52, Met1-Arg287 includes the protease-resistant domain.
To be certain that the putative amino-terminal fragment represents a protease-resistant domain, we engineered a HindIII restriction site in the open reading frame at the codons corresponding to Met282 and Val283 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 Met1-Val283. 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 SequencesThe 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 106 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.
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 (35-38). 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 ParticleMutations in
one other gene, MFT2, also reduced delivery of the
F1-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 itochondria
own
rotein 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.
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 F1 protein, are delivered;
mft1 cells remain viable at 25 °C, and steady-state
levels of F1
remain normal in mft1 mutant
cells (24). In addition to the amino-terminal 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 amino-terminal
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.
We thank Scott Emr, in whose lab the mft1 mutants were first characterized, for strains, plasmids, and encouragement to initiate this project. Thanks also to Sabine Rospert and Jeff Schatz for ideas, critical discussions, plasmids, and antisera. We thank Alfons Lawen for critical discussion; Rosemary Condron for excellent assistance with protein sequencing; Joan Hoogenraad for monoclonal antibody production; Kaye Truscott for advice with the fast protein liquid chromatography; and Marilyn Anderson, Nick Hoogenraad, and Stewart Nuttall for critical reading of the manuscript.