Binding to Chaperones Allows Import of a Purified Mitochondrial Precursor into Mitochondria*

Refolding of the acid-unfolded precursor to mitochondrial aspartate aminotransferase (pmAAT) is inhibited when cytosolic Hsc70 is included in the refolding reaction (Artigues, A., Iriarte, A., and Martinez-Carrion, M. (1997) J. Biol. Chem. 272, 16852–16861). At low molar excess of Hsc70 pmAAT is recovered in insoluble aggregates containing equal amounts of Hsc70. However, in the presence of a large excess of Hsc70, refolding of pmAAT is still arrested, but the enzyme remains in solution. Similar behavior was observed with two other cytosolic chaperones, bovine Hsp90 and yeast Ydj1. Coimmunoprecipitation of pmAAT using Hsc70 antibodies confirmed the formation of soluble Hsc70-pmAAT complexes at high concentrations of the chaperone. Data from analytical centrifugation, sedimentation in glycerol gradients, and partial purification of the soluble complexes indicate that multiple Hsc70 molecules bind per pmAAT polypeptide chain. The absence of catalytic activity together with the protease susceptibility of pmAAT bound to Hsc70, Hsp90, or Ydj1 suggest that these chaperones bind and maintain pmAAT in a partially unfolded state, analogous to the import-competent conformation of the protein synthesized in cell-free extracts. Remarkably, the purified pmAAT bound to Hsc70 or Ydj1, but not to Hsp90, is imported by isolated import quots of mitochondrial g/ml L -tosylamido-2-phenylethylchloromethylketone (TPCK)-trypsin min the (sample or (sample of 0.1% Triton X-100. trypsin digestion of 0.5 (cid:2) of with an equal volume of SDS-PAGE sample SDS-PAGE and to a

The majority of mitochondrial proteins are encoded by the nuclear genome and synthesized on cytoplasmic ribosomes, most of them as precursor forms with N-terminal targeting sequences that are usually removed by a specific peptidase in the matrix. Import through the outer and inner membrane translocation complexes requires energy in the form of ATP hydrolysis in the matrix and an electrochemical potential across the inner mitochondrial membrane, as well as the participation of several mitochondrial chaperones including the import motor, mitochondrial Hsp70 (reviewed in Refs. 1 and 2). Although some proteins might be imported cotranslationally (3,4), for the majority of proteins import can take place posttranslationally, both in vitro and in vivo (1,5). In addition, it is well known that mitochondrial precursors are not imported into mitochondria in their native state (6,7). The degree of folding of the import substrate compatible with translocation varies among precursors. Some proteins can be presented to mitochondria in their native conformation, and mitochondria are able to unfold them during translocation (8). These proteins usually do not require ATP hydrolysis outside the mitochondria for translocation. The import of many other mitochondrial precursors, including the precursor for rat liver aspartate aminotransferase (pmAAT) 1 used in this work, requires extramitochondrial ATP (9,10), probably for their release from complexes with cytosolic chaperones. It is likely that these mitochondrial precursors never fold in the cytosol and engage the translocation machinery when still in a partially unfolded state. In fact, the folding of pmAAT synthesized in a cell-free extract is inhibited by cytosolic factors, and once the protein is folded mitochondria can no longer import it (9). The interaction with cytosolic chaperones might prevent not only premature folding but also aggregation and proteolysis of the incompletely folded polypeptides. The requirement for cytosolic proteins may not be universal because some precursor proteins can be imported in vitro into mitochondria in the absence of cytosolic extracts (6,11). It is clear that the requirements for the delivery of newly synthesized precursors to the outer mitochondrial membrane differ among precursor proteins and may depend on factors such as overall hydrophobicity or oligomeric state of the native mature protein.
Cytosolic components reportedly performing this chaperone function for mitochondrial precursors include members of the Hsp70 and Hsp40 (DnaJ-related) family of molecular chaperones (1). The constitutively expressed cytosolic member of the Hsp70 family of molecular chaperones (Hsc70) is involved in a broad spectrum of cellular processes (12)(13)(14). Its chaperone activity is regulated by a number of co-chaperones, including members of the Hsp40 family that stimulate its ATPase activity and promote substrate association with Hsp70 (14 -16). Both Hsc70 and Hsp40 bind to nascent chains as they elongate on cytoplasmic ribosomes (17)(18)(19), and evidence has been accumulating in support of a general role for these chaperones in the translocation of mitochondrial proteins. Depletion of Hsp70 proteins in yeast mutants interfered with the biogenesis of mitochondrial proteins (20), and the yeast DnaJ homologue Ydj1 was initially isolated as MAS5, a mutation that causes defects in mitochondrial import in yeast (21). Furthermore, both Hsc70 and the DnaJ homologues dj2 or dj3 appear to be required for import of preornithine decarboxylase translated in vitro into mammalian mitochondria (22)(23)(24). Additional cytosolic components such as the presequence binding factor (PBF) (25) and the mitochondrial import stimulation factor (MSF) (26) may have a more specific role as targeting factors.
On the other hand, eukaryotic Hsp90s are unique among molecular chaperones in that they display considerable specificity for their substrate proteins. Although purified Hsp90 has been shown to prevent the in vitro aggregation of some unfolded proteins, the majority of their known substrates in vivo are signal transduction proteins, including a variety of hormone and growth factor receptors (27)(28)(29). In addition, Hsp90 has been shown to bind proteins thought to be involved in targeted protein traffic in yeast via their tetratricopeptide repeat domains (30). An early report (31) described the isolation of a fraction from rabbit reticulocyte lysate (RRL) containing Hsp90 and Hsp70, which both stimulated import into mitochondria of a hybrid precursor protein and mediated formation of Hsp90-steroid receptor heterocomplexes. Yet the involvement of Hsp90 on protein translocation remains obscure.
We have shown previously that cytosolic Hsc70 binds to pmAAT both during translation in a cell-free extract (32) and in vitro refolding of the acid-unfolded protein (33). Subsequently we identified seven putative Hsc70-binding sites in the pmAAT polypeptide (34) which included the cleavable presequence peptide in addition to several peptide regions in the mature portion of the protein. The complexes obtained at low molar excess of Hsc70 over purified unfolded pmAAT were insoluble and formed aggregates containing equal amounts of the two proteins (33). Thus, a question remained as to whether Hsc70 was able to bind simultaneously to several of the binding sites in pmAAT. Here we report that in the presence of high concentrations of Hsc70, purified pmAAT forms soluble complexes containing multiple Hsc70 molecules bound to a single pmAAT polypeptide. Soluble complexes can also be assembled with purified yeast YDJ1 and mammalian Hsp90 and unfolded pmAAT. The import competence of these complexes is chaperone-specific. Mitochondria were unable to import Hsp90-bound pmAAT, whereas import from the complexes with Hsc70 or Ydj1 requires the presence of reticulocyte lysate. This contrasts previous reports on the import of another matrix protein, preadrenodoxin, from preassembled complexes with Hsp70 or MSF which could take place in the absence of a cytosolic extract or even ATP (35). To our knowledge this is the first report of in vitro import of a mitochondrial protein from a preassembled Hsp40-precursor complex.

EXPERIMENTAL PROCEDURES
Protein Purification-The expression in Escherichia coli and purification of the histidine-tagged protein pmAAT(His 6 )Tyr, a protein construct containing a tag of six histidines and a tyrosine residue fused to the C-terminal end of pmAAT, was carried out as described previously (36) using metal chelation chromatography in a Ni 2ϩ -nitrilotriacetic acid (Ni-NTA)-agarose column (Qiagen). The mature form of the protein, mAAT(His 6 )Tyr, was obtained by incubating the precursor with trypsin (1:100 trypsin/pmAAT(His 6 )Tyr molar ratio) to remove the Nterminal presequence peptide, followed by chromatography on a CM-Sephadex column. Protein concentration was estimated from the absorbance at 280 nm using the calculated molar absorption coefficient of 55,610 M Ϫ1 cm Ϫ1 (36) and M r ϭ 48,540 and 45,824 for the precursor and mature protein, respectively. Because all the experiments reported in this work were performed using this tagged protein construct, for brevity and clarity of exposure we will omit the suffix (His 6 )Tyr and we will refer to these proteins as pmAAT for the precursor or mAAT for the mature form. Hsc70 was purified from bovine brain following published procedures (37). Protein concentration was estimated using the published ⑀ (280 nm) ϭ 47,800 M Ϫ1 cm Ϫ1 and M r ϭ 70,000 (38). Hsp90 was purified from bovine brain as described previously (39). Ydj1 was expressed in E. coli and purified according to Cyr et al. (40). Hsp90 and Ydj1 protein concentrations were determined from the intensity of Coomassie Blue-stained SDS-PAGE gels, using mAAT as standard. The purified proteins were stored at 4°C.
Protein Labeling-Radiolabeled pmAAT was prepared by incubating the protein (200 g) with 2 M Na 125 I (1 Ci) in 100 l of refolding buffer (100 mM Hepes, 0.1 mM EDTA, 1 M dithiothreitol) in an IODO-GEN-coated tube (Sigma) for 30 min at room temperature. The reaction was quenched by addition of 5 l of 100 mM NaI and 5 l of 4 mM hydroxymethyl acetate, followed by extensive dialysis in refolding buffer. The final specific radioactivity of the labeled protein was 300,000 dpm/g. Syncatalytic modification of pmAAT with 5-iodoacetamidofluorescein (5-IAF) was achieved by incubation of pmAAT (1 mg/ml) in refolding buffer with 2 mM 5-IAF in the presence of 50 mM of the substrate analogue ␣-methyl aspartate. After incubation for 4 h at room temperature, the excess of 5-IAF and ␣-methyl aspartate was removed by ultrafiltration through a Centricon 30 (Amicon). The resulting AF-labeled pmAAT (AF-pmAAT) contains a strong absorbance reporting group (⑀ 492 nm ϭ 75,000 M Ϫ1 cm Ϫ1 ) suitable to follow the sedimentation profiles of pmAAT and its complex with Hsc70 during analytical ultracentrifugation. The concentration of the modified proteins was estimated from the Coomassie Blue-stained SDS-PAGE gels by comparing the intensity of the protein bands with standards of known concentrations of unlabeled mAAT.
Formation of Soluble Complexes of pmAAT with Different Molecular Chaperones-Acid unfolding of 125 I-or 5-AF-labeled pmAAT was performed by incubation in 2 mM Tris acidified to pH 2.0 with HCl for 90 min at room temperature. The final protein concentration was calculated from the specific radioactivity of the 125 I-labeled protein. Binding of pmAAT to Hsc70, Hsp90, or Ydj1 was accomplished by diluting the acid-denatured protein with 10 volumes of ice-cold refolding buffer containing a given chaperone followed by incubation for 90 min at 10°C. The final concentration of pmAAT monomer in the reaction was 2.5 ϫ 10 Ϫ3 M monomer, and the concentration of each molecular chaperone was 1.8 M, unless indicated otherwise.
Trypsin Susceptibility of pmAAT Bound to Hsc70 -Trypsin digestion was performed at 4°C by addition of a concentrated stock solution of TPCK-trypsin in 1 mM HCl to give a final Hsc70/trypsin molar ratio of 200:1. In control experiments using Hsc70 alone, we found that with this trypsin/Hsc70 ratio about 90% of Hsc70 remains intact even after a 1-h incubation at 4°C. To evaluate the extent of proteolysis of both chaperone and pmAAT, samples were analyzed by SDS-PAGE and visualized either in a Molecular Dynamics PhosphorImager TM ( 125 I-pmAAT) or by staining with Coomassie Blue (Hsc70).
Immunoprecipitation-Immunoprecipitation reactions of Hsc70-125 I-pmAAT complexes (60 l) were performed by addition of an equal volume of rabbit anti-Hsc70 polyclonal antibodies diluted 1:1 in PBS buffer, 5 mM EDTA, and 0.1% Triton X-100. The same amount of preimmune serum was used as a control. Following incubation for 4 h at 10°C, 60 l of protein A-agarose (Repligen) was added to each reaction, and the samples were incubated for an additional hour at 4°C with gentle shaking. The suspensions were centrifuged at maximum speed on a microcentrifuge. The supernatants were diluted in SDS sample buffer, and the pellets, following an additional wash in phosphatebuffered saline, 0.1% Triton X-100, 5 mM EDTA, were resuspended in SDS sample buffer and heated at 60°C for 20 min. The agarose beads were pelleted by centrifugation, and the supernatants were analyzed by SDS-PAGE and PhosphorImaging.
Analytical Ultracentrifugation-Determination of the sedimentation coefficient of the AF-labeled pmAAT alone or in complex with Hsc70 was done on a Beckman XL-A analytical ultracentrifuge, using an AN-60 Ti analytical rotor. Samples were centrifuged at 30,000 or 15,000 rpm for AF-pmAAT alone or Hsc70-AF-pmAAT complexes, respectively. Sedimentation coefficients were estimated from the absorbance boundary profiles at 492 nm (the max of fluorescein) using the software provided with the instrument.
Sedimentation in Glycerol Gradients-Native Hsc70 (1.8 M) and pmAAT (2.5 ϫ 10 Ϫ3 M) as well as samples of Hsc70-pmAAT complex formation reactions were layered on top of a 5-ml 10 -30% glycerol gradient in refolding buffer. The gradients were preformed on a Biocomp model 106 gradient maker following the manufacturer's instructions. Samples were centrifuged at 55,000 rpm for 4 h on a TLX Beckman ultracentrifuge using a TLX-L rotor. Following fractionation of the gradients on a Retriever II (Isco), the fractions were analyzed for the presence of 125 I-pmAAT by counting aliquots on a Beckman Gamma 5500 counter and for the presence of Hsc70 by SDS-PAGE followed by staining the gels with Coomassie Blue.

Import of Purified pmAAT into Isolated Mitochondria-Mitochondria
were isolated from male Wistar rat liver as described previously (41) and resuspended in MESH buffer (220 mM mannitol, 0.1 mM EDTA, 70 mM sucrose, 20 mM HEPPS, pH 7.4). The import reaction was performed by mixing 40 l of Hsc70-125 I-pmAAT complexes with 40 l of freshly isolated mitochondria (4 mg/ml final concentration) and 40 l of RRL or MESH buffer containing ATP and an ATP-regenerating system (200 M MgATP, 2 mg/ml creatine kinase, 500 M creatine phosphate, 400 mM M malate, 400 M succinate, 200 M NADH, in MESH buffer, pH 7.4). The reaction mixture was incubated for 30 min at 30°C. After stopping import by chilling on ice, an aliquot was taken and mixed with an equal volume of SDS-PAGE sample buffer. The remaining of the reaction was centrifuged at 16,000 ϫ g for 4 min, and the pellet containing the reisolated mitochondria was washed twice with 200 l of MESH buffer and finally resuspended in an adequate volume of MESH buffer. Aliquots of this mitochondrial fraction were either left untreated and mixed directly with SDS-PAGE sample buffer (sample 1) or treated with 20 g/ml L-tosylamido-2-phenylethylchloromethylketone (TPCK)trypsin for 30 min on ice either in the absence (sample 2) or presence (sample 3) of 0.1% Triton X-100. The trypsin digestion was terminated by addition of 0.5 l of phenylmethylsulfonyl fluoride, and the samples were mixed with an equal volume of SDS-PAGE sample buffer. All samples were subsequently analyzed by SDS-PAGE and exposure to a PhosphorImager screen. The amount of protein bound to mitochondria (sample 1), imported into mitochondria (sample 2), or imported and properly folded (sample 3) was estimated from the intensity of the corresponding radiolabeled protein band and is expressed as percentage relative to the intensity of the pmAAT band in the unfractionated import reaction.

Formation of Soluble Complexes Between pmAAT and
Hsc70 -The protein construct used in this work contains an LEHHHHHHY tag fused to the C-terminal end of pmAAT. The addition of this C-terminal tail does not alter the overall stability, catalytic properties, or refolding ability of the enzyme (data not shown), but it provides a convenient tool for the purification and detection of the protein. The histidine tag allows for rapid purification of the protein using metal affinity resins, whereas the C-terminal tyrosine can be easily iodinated for radiolabeling purposes.
Hsc70 binds refolding intermediates of acid-denatured pmAAT and prevents further folding and reactivation of the enzyme (33). The nature of the complexes formed depends on the concentration of Hsc70 in the refolding reaction. At a low molar excess of Hsc70, the resulting complexes contain stoichiometric amounts of pmAAT and Hsc70 and form insoluble aggregates that can be removed by centrifugation (Fig. 1). However, the fraction of pmAAT recovered in the supernatant increases with increasing concentrations of Hsc70 until all of the pmAAT remains in solution at molar ratios of Hsc70 over pmAAT of 200 or higher (Fig. 1). Yet no transaminase activity is detected in the supernatant, which indicates that the soluble pmAAT is not completely folded. The inactive soluble pmAAT could be immunoprecipitated using Hsc70 polyclonal antibodies ( Fig. 2), which confirms that the pmAAT in the supernatant is indeed bound to Hsc70. The solubility of these complexes is marginal, and they readily precipitate when the sample is concentrated slightly after preparation (data not shown). Native pmAAT, either alone or in the presence of an excess of Hsc70, did not precipitate with anti-Hsc70 antibodies (data not shown). Similar concentration-dependent formation of insoluble or soluble complexes was observed with two other chaperones, Hsp90 and Ydj1 (Fig. 1).
Stoichiometry of the Soluble Hsc70-pmAAT Complexes-A possible explanation for the different behavior of the complexes formed at low and high concentrations of Hsc70 is that the soluble complexes contain multiple Hsc70 molecules associated with a single pmAAT polypeptide. To determine the stoichiometry of the complexes, we first isolated the complex from the large excess of Hsc70 present in the complex formation reaction by metal chelation chromatography in a Ni-NTA column taking advantage of the presence of the histidine tag at the C-terminal end of pmAAT. In control experiments we established that Hsc70 alone does not bind to the column and elutes with the loading buffer. On the other hand, histidine-tagged native pmAAT binds very tightly but can be eluted with 110 mM imidazole. When Hsc70-pmAAT complexes were loaded on the column, the majority of Hsc70 eluted with the loading buffer, as expected considering the large excess of Hsc70 present in the mixture. SDS-PAGE analysis of the material bound to the resin , Hsp90 (f), or Ydj1 (E) as indicated under "Experimental Procedures." After incubation for 90 min at 4°C, the radioactivity remaining in the supernatants was measured after centrifugation of the samples at 20,000 rpm on a microcentrifuge. The amount of 125 I-pmAAT recovered in the supernatants was expressed as percentage relative to the total amount of radiolabeled protein added. The arrow indicates the fraction of radioactive pmAAT recovered in the supernatant when acid-unfolded protein is diluted in refolding buffer alone and allowed to refold for 90 min.

FIG. 2. Coimmunoprecipitation of pmAAT with Hsc70 antibodies.
Immunoprecipitation reactions of Hsc70-125 pmAAT soluble complexes were performed as indicated under "Experimental Procedures" using rabbit preimmune or anti-Hsc70 whole antiserum. Both supernatants (Sn) and pellets (P) were resuspended in SDS sample buffer to the same final volume and analyzed by SDS-PAGE. The dried gels were exposed overnight to a PhosphorImager screen, and the radiolabeled protein bands were detected on a Molecular Dynamics PhosphorImager TM .
showed the presence of Hsc70 together with pmAAT (data not shown). Following visualization either by Coomassie Blue staining (Hsc70) or autoradiography ( 125 I-pmAAT), the amount of each protein bound to the Ni-NTA resin was estimated from the intensity of the corresponding electrophoretic bands, using known concentrations of each protein as standards (see "Experimental Procedures" for details). According to these measurements, we found that on average between 5 and 7 molecules of Hsc70 are bound to each pmAAT polypeptide chain. This figure represents only a rough approximation because several factors such as dissociation of Hsc70 from low affinity binding sites during purification of the complex could influence the estimate. In any case, it is clear that several Hsc70 molecules must bind to a single pmAAT polypeptide in order to prevent aggregation and subsequent precipitation of the complex. Moreover, this stoichiometry is in good agreement with the number of Hsc70-binding sites (seven) identified in the pmAAT sequence by screening a synthetic peptide library (34). Interestingly, attempts at eluting the proteins from the Ni-NTA resin with high concentrations of imidazole were unsuccessful. Only solubilization with SDS-PAGE sample buffer released the bound proteins. A possible explanation for these findings is that the tight binding of the Hsc70-pmAAT complexes to the resin involves not only specific interactions of the histidine tag with the Ni 2ϩ immobilized in the Ni-NTA resin but also nonspecific interactions of nonnative pmAAT with the solid matrix. This conclusion is supported by the observation that in vitro translated wild type pmAAT, which is also partially unfolded, binds not only to Ni-NTA resins but also to several other agarose-based solid supports. 2 Sedimentation Analysis of the Soluble Hsc70-pmAAT Complexes-The hydrodynamic properties of these soluble complexes were further characterized using ultracentrifugation methods. To follow the distribution of pmAAT during boundary sedimentation velocity experiments in the presence of high concentrations of Hsc70, pmAAT was selectively labeled with 5-IAF at Cys-166 (42). Upon labeling, native pmAAT loses ϳ50 -60% of the original activity. This chromophore was selected because of its high molar extinction coefficient (75,000 M Ϫ1 cm Ϫ1 ) at 492 nm which allowed the detection of small amounts of pmAAT during ultracentrifugation by absorption optics. When native AF-pmAAT is analyzed, it sediments with an s value of 7.5 S, essentially identical to the sedimentation coefficient of the unlabeled wild type enzyme. This s value corresponds to a globular protein with a molecular mass of ϳ90 kDa, which is close to the theoretical mass of the dimeric protein (97,080) calculated from its amino acid composition (36). Following acid unfolding of AF-pmAAT and incubation with an excess of Hsc70, the Hsc70-bound AF-pmAAT sediments with an s value of 12.5 S, substantially higher than that obtained for the native protein. This increase in the sedimentation velocity coefficient relative to the native dimer is consistent with the formation of large complexes containing several Hsc70 molecules associated with each pmAAT polypeptide. Aliquots were analyzed by SDS-PAGE and Coomassie Blue staining or visualization in a PhosphorImager to detect Hsc70 or pmAAT, respectively. As a control, native pmAAT was treated under similar conditions. The slight increase in electrophoretic mobility of native pmAAT upon treatment with trypsin results from hydrolysis after Arg-28 at the C-terminal end of the presequence peptide (9).
Because of the nonnative structure of pmAAT in these complexes, it is likely that they have a larger partial specific volume than the compact, denser native dimer.
Close examination of the sedimentation profiles reveals a complex behavior of the Hsc70-pmAAT complexes, which deviates from the behavior expected for the sedimentation of ideal species and suggests the presence of either a mixture of complexes with different stoichiometries or the existence of multiple binding equilibria as a consequence of mass transport changes during sedimentation. Sedimentation analysis in glycerol gradients shows that although native pmAAT remains at the top of a 10 -30% glycerol gradient, Hsc70-pmAAT complexes are distributed throughout the gradient (Fig. 3) thus confirming the existence of a heterogeneous mixture of Hsc70-pmAAT complexes.
Protease Susceptibility of Hsc70-bound pmAAT-Native pmAAT is very resistant to proteolysis, and for this reason trypsin susceptibility has been used previously as the criterion to follow the folding of the protein synthesized in cell-free extracts (9) or to establish the partially unfolded conformation of pmAAT bound to the chaperonin GroEL (36). Thus, analysis of the trypsin susceptibility of radiolabeled pmAAT provides a sensitive tool to test the conformational state of pmAAT bound to Hsc70, even in the presence of an excess of the chaperone. Unfortunately, native Hsc70 is also hydrolyzed by trypsin. In a series of control experiments using different concentrations of trypsin and incubation conditions, we found that at a molar ratio of Hsc70/trypsin of 200:1 over 95% of Hsc70 remains intact after incubation for 30 min at 4°C. Under these conditions, trypsin removes only the presequence peptide from native pmAAT producing the mature form of the enzyme with slightly higher electrophoretic mobility (Fig. 4, lane 8) (9). By contrast, pmAAT bound to Hsc70 is rapidly hydrolyzed by trypsin. After incubation with trypsin for 30 min at 4°C, most of the protein has disappeared from the gel (Fig. 4, lanes 4 -6), whereas a large fraction of the Hsc70 present in the reaction survives this incubation (Fig. 4, lanes 1-3). pmAAT remains trypsin-susceptible even after an overnight incubation of the Hsc70-pmAAT complex at 10°C. During the digestion of pmAAT, a series of intermediate fragments appear that are similar in size to those formed following trypsin treatment of GroEL-pmAAT complexes or of the protein freshly translated in RRL (36). This tryptic pattern suggests that pmAAT bound to Hsc70 remains in a partially unfolded conformation still FIG. 5. Import of pmAAT bound to different molecular chaperones into isolated mitochondria. Complexes of 125 I-pmAAT with either Hsc70 (A), Ydj1 (B), or Hsp90 (C) were prepared as indicated under "Experimental Procedures" using a 700-fold molar excess of chaperones over pmAAT subunit. Import reactions were started by mixing 40 l of preassembled complexes with 80 l of a 1:1 mixture of RRL and mitochondria (4 mg/ml final protein concentration). The reactions were incubated for 30 min at 30°C in the presence (ϩ⌬⌿) or absence (Ϫ⌬⌿, mitochondria treated with 1 M valinomycin) of a membrane potential as described under "Experimental Procedures." After reisolating mitochondria by centrifugation, aliquots were either left untreated (lanes 2 and 6) or treated with 20 g/ml TPCK-trypsin for 30 min on ice in the absence (lanes 3 and 7) or presence (lane 4) of 0.1% Triton X-100. All samples were analyzed by SDS-PAGE, and the radioactive protein bands were visualized in a PhosphorImager. Aliquots of the unfractionated import reaction (Total) were diluted 1:4 with MESH buffer prior to analysis by SDS-PAGE (lanes 1 and 5).
susceptible to proteolysis but already containing structural domains with restricted accessibility to trypsin. The nonnative conformation of pmAAT bound to Hsc70 is probably the main factor responsible for the aforementioned propensity of the complexes to precipitate upon concentration.
Import of Chaperone-bound pmAAT into Isolated Mitochondria-Earlier studies (9) have shown that wild type pmAAT synthesized in vitro in RRL can be efficiently imported into isolated rat liver mitochondria and processed to the mature form by removal of the presequence peptide. Identical results were obtained for the in vitro translation and mitochondrial import of the histidine-tagged construct used in this study (data not shown). For both wild type (9) and histidine-tagged pmAAT, the yield of import gradually decreases as the translation product synthesized in RRL is allowed to fold into a native-like conformation by incubating the translation reaction at 20°C. Accordingly, purified native pmAAT is not imported into mitochondria either (data not shown). Thus, both purified and in vitro translated pmAAT must be presented to mitochondria in a nonnative conformation in order to be taken up efficiently by the organelle. Given the crude nature of the cell-free extract used for translation of pmAAT, a number of different cytosolic components might be involved in preserving the import-competent conformation of the newly synthesized protein.
Our attempts at using acid-unfolded purified pmAAT as import substrate failed because all of the protein precipitated when diluted directly into the import reaction containing mitochondria (data not shown). On the other hand, because pmAAT bound to Hsc70, YDJ1 or Hsp90 is maintained in solution indefinitely in a partially unfolded conformation, it meets one of the requirements for translocation into mitochondria, and therefore it might represent a suitable import substrate.
To test this possibility, aliquots of the pre-assembled complexes prepared using radiolabeled 125 I-pmAAT were incubated with isolated energized mitochondria under conditions similar to those used with in vitro translated pmAAT (i.e. in the presence of RRL, 30 min at 30°C; see "Experimental Procedures" for details). With all three chaperone-pmAAT complexes, a fraction of radioactive pmAAT coprecipitates with mitochondria after fractionating the import reaction by centrifugation, which suggests that the chaperone-bound import substrate binds to the organelle (Fig. 5, A-C, lane 2). Complete internalization of the imported protein, however, requires that the bound protein becomes inaccessible and therefore resistant to trypsin and that it is processed to the shorter mature form with higher electrophoretic mobility. These two premises are only fulfilled by pmAAT bound to Hsc70 or YDJ1. As shown in Fig. 5, mitochondria isolated from import reactions incubated with Hsc70-pmAAT or YDJ1-pmAAT contain a mature-like radioactive protein that survives treatment with trypsin (Fig.  5, A and B, lane 3). From the intensity of this protein band, we estimated that between 15 and 20% of the total pmAAT added to the reaction is imported by mitochondria. The majority of the imported protein (over 90%) remains intact after disrupting the mitochondrial membranes with detergent so that trypsin can have access to the internalized protein (Fig. 5, A and B, lane 4). This indicates that pmAAT is not only imported and processed but also folds correctly into the protease-resistant conformation of the native protein. Deenergizing mitochondria by dissipating the membrane potential across the inner membrane by addition of 1 M valinomycin completely abolished import (Fig. 5, A  and B, lane 7), in agreement with previous results (9) showing that both a membrane potential and intramitochondrial ATP are necessary for translocation of pmAAT into the mitochondrial matrix. By contrast, all of the radioactive protein found in the mitochondrial pellet from reactions containing Hsp90-pmAAT was digested by trypsin (Fig. 5C, lane 3), which is consistent with its location on the outer surface of mitochondria and therefore accessible to the externally added protease. Thus, pmAAT in complex with Hsp90 is not a suitable substrate for import into mitochondria. The order of addition of the different reaction components (RRL, mitochondria, and complexes) had no effect on the yield of protein import (data not shown).
The import of purified pmAAT bound to Hsc70 or YDJ1 into mitochondria resembles in many respects the translocation of Hsc70-pmAAT complex in refolding buffer was prepared, and aliquots of each dilution were incubated with mitochondria (4 mg/ml protein concentration) and RRL for 30 min at 30°C. After reisolating mitochondria, aliquots were analyzed by SDS-PAGE, and the intensity of the associated radiolabeled protein bands was determined using a Phos-phorImager. The amount of protein imported was estimated by comparing the intensity of the trypsin-resistant 125 I-mAAT band with that of a standard of 125 I-pmAAT of known specific radioactivity. The amount of total pmAAT added was calculated from the concentration of pmAAT used in the preparation of the complex and the corresponding dilution.
FIG. 7. Time course of import of the Hsc70-bound pmAAT into mitochondria. Import reactions were performed at 30°C by mixing aliquots of the Hsc70-125 I-pmAAT complex prepared as described in Fig.  6 with mitochondria (4 mg/ml protein concentration) in the presence of RRL. At the indicated reaction times, a 20-l aliquot was withdrawn, and further import was quenched by chilling on ice. Mitochondria were reisolated by centrifugation and analyzed by SDS-PAGE and visualization in a PhosphorImager. The fraction of protein imported was estimated from the intensity of the radiolabeled mature protein band associated with trypsin-treated mitochondria. Data are expressed as percentage relative to the intensity of the band corresponding to the total pmAAT added to the reaction. the protein freshly translated in RRL (9). For instance, import of pmAAT is concentration-dependent, and the amount of protein imported reaches a plateau when 300 -400 ng of pmAAT are incubated with 1.5 mg of mitochondria (Fig. 6). At this optimal concentration, ϳ20% of the added protein is imported which corresponds to a maximum uptake of about 40 ng (or ϳ1 pmol) of pmAAT per mg of mitochondrial protein. The fraction of protein imported remains essentially the same (ϳ20%) at all the concentrations tested below the plateau region, which suggests that only a portion of the population of chaperone-bound pmAAT molecules is competent for import. This yield of import is considerably lower than that of pmAAT freshly synthesized in RRL (60 -70% of the total translation product added to the reaction). On the other hand, analysis of the time course of import carried out at the optimal conditions described above (300 ng of Hsc70-bound pmAAT, 1.5 mg of mitochondrial protein in the presence of RRL at 30°C) shows that the amount of protein imported increases with time and reaches a maximum at about 30 min (Fig. 7). The import rate can be approximated by a first-order reaction with a t1 ⁄2 of about 10 min, essentially identical to that reported earlier for the in vitro translated protein (9). Finally, addition of an excess of unlabeled preassembled Hsc70-pmAAT complexes inhibits dramatically the import of in vitro translated, radiolabeled pmAAT (data not shown). This competition suggests that the import of the two types of substrates uses the same set of components in the translocation machinery of mitochondria and follows a similar pathway.
Import of Hsc70-bound pmAAT Requires Additional Components from RRL-The import experiments described above were designed to reproduce as closely as possible the conditions used for the import of pmAAT synthesized in vitro in RRL, i.e. the reactions containing mitochondria and the chaperone-pmAAT substrate were supplemented with a concentration of RRL comparable with that present in the import reaction set up with the protein translated in vitro. The question still remained as to whether RRL was necessary for the import of pmAAT from the preformed complex with Hsc70. To address this issue, RRL was omitted from the reaction, and the Hsc70-pmAAT complex was incubated with mitochondria in the presence of import buffer containing ATP and respiratory substrates to maintain the mitochondria energized. As shown in Fig. 8A, some radioactive pmAAT coprecipitated with mitochondria under these conditions, but it retained its precursor size (lane 2) and susceptibility to trypsin (lane 3). Thus the protein associated with mitochondria was not properly imported. This result indicates that RRL is necessary for import of pmAAT into mitochondria. Because some pmAAT is found in the mitochondrial pellet, we reasoned that perhaps components from the lysate were required for import but not for the initial binding of the protein to the protein translocation machinery on the mitochondrial membrane. However, when the mitochondria containing bound pmAAT were isolated and incubated again in the presence of RRL, the bound pmAAT remained accessible to trypsin (Fig. 8A, lane 6). Thus import of the protein precipitating with mitochondria cannot be restored by subsequent addition of RRL. This negative result is not due to mitochondria being damaged during the first incubation at 30°C. Mitochondria that had been preincubated for 30 min at 30°C in import buffer containing ATP were able to import pmAAT from its complex with Hsc70 in the presence of RRL with the same efficiency as untreated mitochondria (Fig. 8B). Similar results were obtained with the Ydj1-pmAAT complex (data not shown). Thus, binding to Hsc70 or Ydj1 in a partially unfolded conformation is not sufficient for translocation of pmAAT into isolated mitochondria. Additional cytosolic components present in RRL are necessary for both productive binding and subsequent translocation across the mitochondrial membranes.

DISCUSSION
The solubility of the complexes obtained using purified pmAAT and Hsc70 depends on their stoichiometry. Whereas 1:1 complexes form insoluble aggregates, those containing 4 -7 chaperone molecules associated with each pmAAT chain remain in solution. Preliminary analysis of the insoluble complexes shows that Hsc70 is bound to a unique site in the pmAAT polypeptide. 3 This binding site spans residues Val-283 to Lys-288, and it corresponds to the central portion of the synthetic peptide pm-31 (residues Ala-280 to Pro-293), one of the Hsc70-binding regions previously identified in pmAAT (34).
Thus, it appears that because of kinetic competition between folding and binding to Hsc70, only one of the binding sites is occupied by the chaperone. The preferential binding of Hsc70 to the Val-276 to Lys-288 peptide region may reflect either differ-3 A. Artigues, A. Iriarte, and M. Martinez-Carrion, manuscript in preparation.
FIG. 8. RRL is required for import of Hsc70-bound pmAAT into isolated mitochondria. A, complexes of Hsc70 with 125 I-pmAAT were assembled as described in Fig. 6 and mixed with freshly isolated mitochondria (4 mg/ml final protein concentration) in import buffer containing 67 mM MgATP and an ATP-regenerating system but not RRL. After incubation for 30 min at 30°C mitochondria were reisolated and resuspended in import buffer. Half of the sample was supplemented with an equal volume of RRL (ϩRRL) and the other half was left untreated (ϪRRL). Both samples were incubated for 30 min at 30°C. Aliquots of the corresponding unfractionated reactions (Total) and mitochondrial fractions (Mitochondria) were analyzed by SDS-PAGE either before (Ϫ) or after treatment with trypsin (ϩ). The radioactive protein bands were visualized using a PhosphorImager. B, mitochondria were preincubated for 30 min at 30°C in the absence of import substrate, mixed with Hsc70-125 I-pmAAT in the presence of RRL, and further incubated for 30 min at 30°C. Aliquots of the complete import reaction (Total) and mitochondrial fraction (Mitochondria) were analyzed as described above for A. ences in affinity of Hsc70 for the various binding sites in pmAAT or differences in the level of structure and/or accessibility of the binding peptides in the early folding intermediates. In the native dimer, several residues from this peptide are located at the subunit interface and in contact with the Nterminal arm from the other subunit (43). Binding of Hsc70 exclusively to this region of the chain might interfere with formation of the dimer and perhaps even with further folding of the monomer. Partially folded conformations of proteins are prone to aggregation, and as a result the arrested pmAAT folding intermediate trapped by Hsc70 precipitates together with the attached chaperone.
On the other hand, binding of Hsc70 molecules to additional binding sites could increase the solubility of the complex by protecting exposed hydrophobic surfaces in the partially folded intermediate and thereby stabilizing the protein against aggregation. Analysis of the substrate specificity of Hsc70 indicated that indeed this chaperone recognizes peptide sequences enriched in hydrophobic residues (44 -46) which would be usually present in the interior of a correctly folded protein. These results favor a model in which several Hsc70 molecules would have to bind to the nascent pmAAT chain in vivo in order to maintain the protein partially unfolded but in solution in the highly concentrated medium of the cell cytosol. Furthermore, it has been predicted that unfolded polypeptides contain on average high affinity binding sites for Hsc70 every 40 residues (46). Thus, during protein translation in the cell, the vectorial elongation of the polypeptide chain on the ribosome may allow the binding of multiple Hsc70 molecules. Under in vitro conditions in which the full-length polypeptide collapses to a compact folding intermediate, all the possible binding sites are presented to the chaperone simultaneously. Increasing the concentration of chaperone accelerates the rate of the second-order binding reaction so that binding to additional sites can take place before they are buried in the folding chain.
The translocation of proteins into organelles such as mitochondria and endoplasmic reticulum is also vectorial with the N-terminal signal sequence emerging first at the trans side of the translocation channel. Therefore, the putative binding sites in the passenger protein will also be exposed sequentially to the chaperones inside the organelles. The two models proposed for the mechanism of action of organellar Hsp70 in mediating protein translocation into mitochondria and endoplasmic reticulum ("trapping" and "actively pulling" models) (47) also predict that several Hsp70 molecules will need to bind to the translocating polypeptide.
Interestingly, yeast Ydj1 and mammalian Hsp90 show a similar concentration-dependent formation of insoluble or soluble complexes with pmAAT. It is likely that they also bind to several regions in the mAAT sequence although the exact stoichiometry of the soluble complexes was not determined. The biological relevance of the recognition and binding of pmAAT by Hsp90 in vitro is not clear. Hsp90 is a highly abundant cytosolic protein whose cellular function and mechanism are not well understood. As discussed in the Introduction, this chaperone is required for the activation of steroid hormone receptors and other signal transduction molecules (27) and may also have a role in the conformational regulation of a variety of proteins. Although it has been shown that Hsp90 shows chaperone activity in vitro with some proteins (48), genetic analyses in yeast strongly suggest that Hsp90 does not play a general role in the folding of newly synthesized proteins in the cell (14). Thus, the binding of pmAAT to Hsp90 is probably not relevant in vivo where chaperones are likely to be much more specific in their selection of substrates. This conclusion is supported by the fact that Hsp90-bound pmAAT is not imported by mitochondria.
In addition to regulating the ATPase cycle of Hsp70, Hsp40 family members (including yeast Ydj1 and mammalian dj2) are able to recognize and bind unfolded proteins to facilitate their subsequent transfer to Hsp70 for folding (49,50). Although there is still controversy as to whether the mammalian DnaJ homologues dj1 or dj2 bind nascent chains in vivo (18,51), recent studies (52) in yeast show that the ability of cytosolic Hsp40s to bind unfolded substrates appears to be an essential function in vivo. The results presented in this work confirm that indeed Ydj1 is able to bind a chemically unfolded protein, pmAAT, in the absence of other chaperones. More importantly, pmAAT in complex with either Ydj1 or Hsc70 not only retains a partially unfolded conformation but can also be imported by mitochondria. As mentioned in the Introduction, protein unfolding is one of the requisites for translocation into mitochondria (6,9). Although mitochondria can simultaneously unfold and import some folded substrates (8), many precursors may never fold in the cytosol prior to import; the question then arises as to how the unfolded conformation is preserved and protected. Studies with yeast mutants have shown that Hsp70 proteins and Ydj1 are involved in the import of several polypeptides into mitochondria (20,53). More recently, using a system of chaperone depletion from and readdition to RRL, it was reported that both Hsc70 and the DnaJ homologues dj2 or dj3 are required for import of preornithine decarboxylase translated in vitro into mammalian mitochondria (22)(23)(24). Interestingly, the readdition of Hsc70 or dj2 after translation was complete did not restore import of precursor synthesized in chaperone-depleted lysate, suggesting that the chaperones must be present during synthesis of the precursor protein and therefore interact cotranslationally with the nascent chain in order to be effective. Thus the Hsc70-dj2/dj3 chaperone system is probably not involved directly in the translocation step but rather in stabilizing the unfolded precursor from aggregation until it engages the translocation machinery in the mitochondrial membrane.
Our results with the preassembled complexes of pmAAT support this chaperoning role for Hsc70 and Ydj1. In both cases, Hsc70 or Ydj1 alone does not support import of nonnative pmAAT. Import is only achieved with the collaboration of additional components from the lysate. The identity and mechanism of action of these factors is unknown. The inability of Hsp90 to support import indicates that some of the interactions with the putative cytosolic factors might be chaperone-specific. Interestingly, these import reactions are somewhat heterogeneous, containing chaperones from either bovine brain (Hsc70) or yeast (Ydj1) and the cytosolic extract from rabbit reticulocytes. We should point out that yeast Ydj1 is highly similar to the mammalian Hsp40s dj2 and dj3 (47% identity and about 70% sequence similarity between Ydj1 and human dj2), both of which have been detected in RRL (22,24). Moreover, Ydj1 has been shown to interact not only with yeast Hsp70s but also with bovine brain Hsp70 (54). Finally, dj2, dj3, and Hsc70, the only or major form of Hsp70 in RRL (22), are highly conserved among mammals (the human and rat proteins are 99% identical) (54). For these reasons, the heterologous Hsc70 and Hsp40 proteins used in this work can substitute for the endogenous chaperones in the lysate that might play a similar role during in vitro synthesis of pmAAT in RRL (32).
Nascent pmAAT exists in RRL as a heterogeneous population of partially folded species that appear to interact dynamically with a number of RRL factors (55) including Hsc70. However, the interaction of Hsc70 with nascent pmAAT is transient, and after 30 min of translation, only 10% of the newly synthesized chains coimmunoprecipitate with Hsc70 (32). Yet, at this point the precursor is soluble and in an import-competent state which suggests that other factors make take over from Hsc70 and interact with the newly synthesized protein in an ATP-dependent manner. The preassembled complexes prepared with purified pmAAT and Hsc70 (or Ydj1) resemble to some extent the initial complexes formed with the nascent chain on the ribosome. Because the full-length pmAAT in the preassembled complexes can be imported upon addition of lysate, the putative cytosolic factors must act downstream from the Hsc70/Hsp40 chaperones, and therefore in vivo they probably interact with the nascent chain posttranslationally. Possible candidates for ATP-dependent cytosolic factors include the PBF (25) and MSF (26), both of which have affinity for presequence peptides and might cooperate with Hsp70. Yet, because Hsp70 and MSF can independently support import of this precursor through two different pathways with the precursor binding to the outer membrane translocon component TOM70 in the presence of MSF and to TOM20 in the presence of Hsc70 (56), the extent of their cooperation in vivo is unclear. Interestingly, and in contrast to our results with pmAAT, import of preadrenodoxin bound to either Hsp70 or MSF did not require the addition of a cytosolic extract (35). Unfortunately, very little is known about the structure or mechanism of action of PBF, and we have been unable to detect interaction between MSF and pmAAT either with the purified proteins or in in vitro translation reactions. 4 Thus, the identity of the cytosolic factors acting posttranslationally as chaperones/targeting factors for the import of pmAAT remains elusive. The preassembled complexes of pmAAT with Hsc70 and Ydj1 described in this work could be used as targets for the isolation of some of these factors from reticulocyte lysate.
In summary, complexes of nonnative pmAAT with yeast Ydj1 and mammalian Hsc70 and Hsp90 could be assembled in vitro using purified proteins. pmAAT in complex with Hsc70 and Ydj1 was suitable for import into mitochondria although only in the presence of other unknown cytosolic factors. Interestingly, the partially unfolded pmAAT bound to Hsp90 was not imported by mitochondria, which suggests that either the folding intermediates trapped by this chaperone lack the inherent properties needed for binding to and import by mitochondria or, more likely, Hsp90 is not involved in protein translocation in vivo and consequently is not recognized by the hypothetical factors required for import.