Mitochondrial import receptors Tom20 and Tom22 have chaperone-like activity.

Mitochondrial preproteins are synthesized in the cytosol with N-terminal signal sequences (presequences) or internal targeting signals. Generally, preproteins with presequences are initially recognized by Tom20 (translocase of the outer membrane) and, subsequently, by Tom22, whereas hydrophobic preproteins with internal targeting signals are first recognized by Tom70. Recent studies suggest that Tom70 associates with molecular chaperones, thereby maintaining their substrate preproteins in an import-competent state. However, such a function has not been reported for other Tom component(s). Here, we investigated a role for Tom20 in preventing substrate preproteins from aggregating. In vitro binding assays showed that Tom20 binds to guanidinium chloride unfolded substrate proteins regardless of the presence or absence of presequences. This suggests that Tom20 functions as a receptor not only for presequences but also for mature portions exposed in unfolded preproteins. Aggregation suppression assays on citrate synthase showed that the cytosolic domain of Tom20 has a chaperone-like activity to prevent this protein from aggregating. This activity was inhibited by a presequence peptide, suggesting that the binding site of Tom20 for presequence is identical or close to the active site for the chaperone-like activity. The cytosolic domain of Tom22 also showed a similar activity for citrate synthase, whereas Tom70 did not. These results suggest that the cytosolic domains of Tom20 and Tom22 function to maintain their substrate preproteins unfolded and prevent them from aggregating on the mitochondrial surface.

Mitochondrial preproteins are synthesized in the cytosol with N-terminal signal sequences (presequences) or internal targeting signals. Generally, preproteins with presequences are initially recognized by Tom20 (translocase of the outer membrane) and, subsequently, by Tom22, whereas hydrophobic preproteins with internal targeting signals are first recognized by Tom70. Recent studies suggest that Tom70 associates with molecular chaperones, thereby maintaining their substrate preproteins in an import-competent state. However, such a function has not been reported for other Tom component(s). Here, we investigated a role for Tom20 in preventing substrate preproteins from aggregating. In vitro binding assays showed that Tom20 binds to guanidinium chloride unfolded substrate proteins regardless of the presence or absence of presequences. This suggests that Tom20 functions as a receptor not only for presequences but also for mature portions exposed in unfolded preproteins. Aggregation suppression assays on citrate synthase showed that the cytosolic domain of Tom20 has a chaperone-like activity to prevent this protein from aggregating. This activity was inhibited by a presequence peptide, suggesting that the binding site of Tom20 for presequence is identical or close to the active site for the chaperone-like activity. The cytosolic domain of Tom22 also showed a similar activity for citrate synthase, whereas Tom70 did not. These results suggest that the cytosolic domains of Tom20 and Tom22 function to maintain their substrate preproteins unfolded and prevent them from aggregating on the mitochondrial surface.
Most mitochondrial proteins are synthesized on cytosolic ribosomes as preproteins with N-terminal signal sequences (presequences) or internal targeting signals. During or following synthesis, many preproteins associate with molecular chaperones that maintain preproteins in a loosely folded translocation-competent conformation (1)(2)(3). Preproteins are then targeted to the mitochondria and imported into the organelle. A dynamic protein complex, termed the translocase of the outer membrane of mitochondria (TOM), 1 is responsible for recogniz-ing and translocating preproteins into the organelle. In yeast, this complex includes the import receptors Tom20 and Tom70 and a general import pore that contains an import receptor (Tom22), a channel protein (Tom40), and small Toms (4 -8). In mammal, Tom20 (9 -11), Tom70 (12,13), Tom22 (14,15), Tom40 (16), and small Toms (16,17) have been identified and characterized. Preproteins with N-terminal presequences are initially recognized by the receptor Tom20, whereas preproteins with internal targeting signals, such as inner membrane carrier proteins, are recognized preferentially by Tom70. Preproteins are then transferred to the general import pore and translocated into mitochondria.
Tom70 functions as a chaperone-docking receptor for the import of membrane preproteins (18). Cytosolic chaperones Hsc70 or Hsp90 carrying a substrate protein dock onto a tetratricopeptide repeat (TPR) domain of Tom70 and transfer preproteins to another set of TPR domains that serve as a binding site for the substrate preproteins (18,19). The association of chaperones with Tom70 prevents substrate preproteins from aggregating until the preproteins are transferred to the general import pore for subsequent membrane translocation (18).
For Tom20-dependent preproteins, it is unclear how these are prevented from aggregating on the mitochondrial surface. Recently, we found that the cytosolic arylhydrocarbon receptor interacting protein (AIP), a FKBP52 homolog, interacts with human Tom20 and that AIP stabilizes pre-ornithine transcarbamylase (pOTC), a presequence-containing preprotein, and facilitates its import into mitochondria (20). AIP, which has a chaperone-like activity, appears to transfer a preprotein-Hsc70 complex to Tom20. However, the Hsc70 is not incorporated into the Tom20-preprotein complex when the preprotein is transferred to Tom20 (20,21). Therefore, we hypothesize that Tom20 itself has a function to prevent preproteins from aggregating on the mitochondrial surface.
Human Tom20 has a presequence binding site that mainly consists of a TPR domain and a glutamine-rich segment (22). The binding groove of Tom20 contains hydrophobic residues and interacts with the hydrophobic face of amphiphilic presequence peptides. However, Tom20 shows little specificity for binding to presequence peptides (23). Indeed, the cytosolic domain of Tom20 is able to bind to hundreds of mitochondrial presequences, although no obvious sequence homology exists among them (24). Furthermore, peptide scan analysis suggests that Tom20 binds not only to the presequences but also to the mature portion of preproteins (25). Furthermore, Tom20 binding is not restricted to presequence-containing preproteins but can also associate with preproteins with internal targeting signals such as the phosphate carrier protein and mitochondrial porin (25,26).
In the present study, we report that the cytosolic domain of Tom20 serves as a preprotein receptor that binds not only to presequences but also to the unfolded mature portion of preproteins. An aggregation suppression assay indicated that Tom20 has a chaperone-like activity to prevent a substrate protein, citrate synthase (CS), from thermal aggregation. Furthermore, the cytosolic domain of Tom22 showed a similar activity, whereas that of Tom70 did not. These results suggest that the cytosolic domains of Tom20 and Tom22, but not Tom70, maintain their substrate preproteins unfolded on the mitochondrial surface and prevent them from aggregation.
Expression and Purification of Proteins-Glutathione S-transferase (GST)-fused human Tom20 and Tom22 proteins were expressed and purified as described (14,27). To obtain the cytosolic domain of human Tom70, the pQE30-His 6 -cTom70 plasmid was transformed into SG13009 cells (Qiagen), and the expressed protein was purified, as the native protein, by metal chelation chromatography. To obtain pOTC-GFP-His 6 and GFP-His 6 proteins, the pET-30a(ϩ)-pOTC-GFP-His 6 or pET-30a(ϩ)-GFP-His 6 plasmid was transformed into SG13009 cells, respectively. The expressed proteins were purified by metal chelation chromatography under denaturing conditions in the presence of 6 M guanidinium chloride.
In Vitro Binding Assay-Purified GST-fused proteins were absorbed onto glutathione-Sepharose beads in binding buffer (20 mM Hepes-KOH, pH 7.4, 50 mM KCl, 1 mM MgCl 2 , and 0.1 mg/ml bovine serum albumin). A reticulocyte lysate containing 35 S-labeled proteins was mixed with the beads in the binding buffer (total 300 l). Binding reaction was performed for 30 min at 25°C with gentle shaking. Unbound proteins were removed by centrifugation in an Ultrafree-MC centrifugal filter unit (Millipore Corp., Bedford, MA), and the retained beads were washed once with binding buffer. Bound proteins were eluted by adding elution buffer A (50 mM Tris-HCl, pH 8, and 15 mM glutathione), and the eluate was subjected to SDS-PAGE. Radioactivity in the gels was visualized and quantified using a FUJIX BAS2000 image plate analyzer (Fuji Film Co., Tokyo, Japan). Elution of proteins was checked by staining with Coomassie Brilliant Blue R-250.
The purified proteins pOTC-GFP-His 6 , GFP-His 6 , and CS were also subjected to the binding assay. They were denatured in 6 M guanidinium chloride and diluted into binding buffer containing GST-fused proteins prebound to glutathione-Sepharose beads. The native form of CS solved in 20 mM Hepes-KOH, pH 7.4, was also subjected to the binding assay. Binding reaction and elution were performed as described above. The eluate was subjected to SDS-PAGE, and proteins were detected by immunoblot analysis. Chemiluminescence signals were visualized using ECL kits (Amersham Biosciences) and detected using a Las-1000 Plus Bioimage analyzer (Fuji Photo Film, Tokyo, Japan).
Chaperone Assay-Thermal denaturation of CS (0.15 M) was carried out by incubation at 43°C in 40 mM Hepes-KOH, pH 7.5 (29). Aggregation was measured by light scattering in a F4500 fluorescence spectrophotometer (Hitachi, Tokyo, Japan) equipped with a thermostated cell holder at an excitation and emission wavelength of 360 nm. For SDS-PAGE analysis, aggregation was performed at 43°C for 10 min. Samples were then cooled on ice for 5 min. Aggregated proteins were recovered by centrifugation and subjected to SDS-PAGE followed by staining with Coomassie Brilliant Blue R-250.

RESULTS
The Cytosolic Domain of Tom20 Binds to Both the Presequence and the Unfolded Mature Portion of Preproteins-To examine whether Tom20 serve as a receptor only for the presequence or for the whole preprotein, a GST-capture binding assay was performed (Fig. 1). We first confirmed our previous observation (27) that Tom20 binds to a presequence (Fig. 1A). Reticulocyte lysates containing in vitro translated proteins were incubated with GST fusions prebound with glutathione-Sepharose beads, and GST fusions and the bound proteins were then eluted with reduced glutathione. pOTC-GFP is a protein in which the presequence of pOTC is fused with GFP (28). Tom20-(25-145) is a GST-fused protein in which GST was N-terminally fused with residues 25-145 of Tom20 (27) (see also Fig. 2A). When in vitro translated pOTC-GFP and GFP were subjected to the binding assay, pOTC-GFP bound to Tom20, whereas GFP did not. This result, together with our previous observations (27), indicate that Tom20 binds to the presequence portion of pOTC-GFP but not to the GFP domain. We showed previously that the GFP domain of pOTC-GFP is stably folded when it is translated in rabbit reticulocyte lysate (27). Therefore, Tom20 apparently does not bind to the folded form of GFP.
We then asked whether Tom20 can bind to the unfolded form of GFP. Purified GFP and pOTC-GFP denatured in 6 M guanidinium chloride were subjected to the binding assay (Fig. 1B). Surprisingly, unfolded GFP as well as unfolded pOTC-GFP FIG. 1. Tom20 (T20) binds to unfolded proteins. A, translation mixture containing 35 S-labeled pOTC-GFP or GFP (10 l) was incubated with GST or Tom20-(25-145) (0.6 nmol) prebound with glutathione-Sepharose beads in binding buffer (total, 300 l). After washing, GST derivatives and bound proteins were eluted and subjected to SDS-PAGE and fluorography. % Input represents the percentage of input 35 S-labeled proteins. B, purified pOTC-GFP-His 6 (pOTC-GFP-6His) or GFP-His 6 (GFP-6His) proteins (50 M), denatured in 6 M guanidinium chloride (2 l), were subjected to the binding assay as in A. The eluates were subjected to SDS-PAGE and immunoblot analysis using GFP antiserum. % Input represents the percentage of input pOTC-GFP-His 6 or GFP-His 6 . C, native or denatured CS (5 M, 2 l) was subjected to the binding assay as in A. The eluates were subjected to SDS-PAGE and immunoblot analysis using an anti-CS antibody. % Input represents the percentage of input CS.
Tom20 and Tom22 Have Chaperone-like Activity bound to Tom20. This suggests that Tom20 binds to the unfolded GFP polypeptide regardless of the presence or absence of the presequence portion. Thus, Tom20 appears to have a specificity for unfolded proteins.
To confirm this, we used another protein, citrate synthase (Fig. 1C). CS is a mitochondrial protein that is initially synthesized as a precursor form with an N-terminal presequence. The precursor form of CS is imported into the mitochondria and processed into the mature form. Therefore, purified CS from pig heart has no presequence. When native and unfolded forms of purified CS were subjected to the binding assay, the denatured form of CS bound to Tom20, whereas the native form of the enzyme did not. This suggests that Tom20 can recognize the unfolded CS but not the folded enzyme. Taken together with the observation regarding GFP, Tom20 has a general specificity for unfolded proteins. These results suggest that Tom20 serves as a receptor for unfolded preproteins on the mitochondrial surface.
Tom20 Has a Chaperone-like Activity-The observation that Tom20 binds to unfolded proteins suggests that Tom20 has a chaperone activity. Therefore, we examined whether Tom20 can prevent a substrate protein from aggregation by using an aggregation suppression assay (Fig. 2). Native CS was incubated with Tom20-(25-145) ( Fig. 2A) at 43°C, and the aggregation of CS was monitored by measuring the light scattering (Fig. 2B). When CS alone was incubated, it began to aggregate after 4 min and reached a plateau at ϳ15 min. Aggregation of CS was not affected by the addition of GST. In contrast, when CS was incubated with equimolar Tom20-(25-145), aggregation was markedly suppressed. The addition of Tom20-(25-145) in 3.3-fold excess led to a stronger suppression. Thermal aggregation of CS was also analyzed by SDS-PAGE (Fig. 2C). A large amount of aggregated CS was observed when CS was incubated with GST. In contrast, when CS was incubated with Tom20-(25-145), aggregation of CS was much reduced. These results indicate that the cytosolic domain of Tom20 has a chaperone-like activity to suppress thermal aggregation of CS. This activity may be important for Tom20 to maintain substrate preproteins unfolded on a mitochondrial surface and prevent them from aggregation.
The deletion mutants of the cytosolic domains of Tom20, Tom20-(25-125), and Tom20-(25-105) ( Fig. 2A) were also ex- Tom20 and Tom22 Have Chaperone-like Activity amined for chaperone-like activity (Fig. 2B). Tom20-(25-125) suppressed aggregation of CS to the same degree as Tom20-(25-145). However, Tom20-(25-105) suppressed the aggregation less effectively. In SDS-PAGE analysis, the aggregation of CS was efficiently reduced by Tom20-(25-125) as well as by Tom20-  and was partially reduced by Tom20-(25-105) (Fig. 2C). These results indicate that Tom20 has chaperone-like activity and that the region 106 -125 of Tom20 is important for effective chaperone-like activity. The region 106 -125 is partially involved in the binding site for presequence peptides (22). Indeed, deletion of this region in Tom20 significantly reduced the binding of presequence peptides to Tom20 (27). It is likely that the binding site for presequence peptides is identical or close to the region responsible for chaperone-like activity.
The Presequence Binding Site of Tom20 Is Responsible for a Chaperone-like Activity-Presequence peptides of preproteins directly bind to Tom20 (22). Therefore, if the binding site of Tom20 for presequence peptides is also responsible for the chaperone-like activity, the addition of a presequence peptide will block the active site and inhibit the chaperone-like activity.
To check this possibility, we investigated the effect of a presequence peptide on the chaperone-like activity of Tom20 (Fig. 3). When CS was incubated with Tom20-(25-145), aggregation of CS was suppressed as in Fig. 2B, indicating again that Tom20 has a chaperone-like activity (Fig. 3A). The addition of a control peptide (PTH-(69 -84) peptide) did not influence the chaperone-like activity of Tom20- . In contrast, the addition of a presequence peptide increased CS aggregation, suggesting that the presequence peptide inhibits the chaperone-like activity of Tom20. The presequence peptide itself did not increase CS aggregation (data not shown). In SDS-PAGE analysis, the addition of the presequence peptide increased the amount of aggregated CS (Fig. 3B), confirming that the chaperone-like activity of Tom20 is inhibited by the presequence peptide. These results suggest that the binding site of Tom20 for presequences is also responsible for the chaperone-like activity.
The Cytosolic Domain of Tom22 Also Has a Chaperone-like Activity-Because the preproteins initially recognized by Tom20 are subsequently transferred to Tom22, Tom22 as well as Tom20 may have a chaperone-like activity. Therefore, we examined a chaperone-like activity of Tom22 (Fig. 4). A series of GST-fused Tom22 deletion mutants (Fig. 4A) were examined for aggregation suppression of CS (Fig. 4B). When CS was incubated with Tom22-(1-71), the aggregation of CS was strongly suppressed, suggesting that the cytosolic domain of Tom22, as well as Tom20, has a chaperone-like activity. Tom22-(1-62) and Tom22-(1-48) also suppressed the aggregation. These results indicate that the region 1-48 of Tom22 is important for the chaperone-like activity. In contrast, Tom22-(102-142) did not suppress the aggregation, suggesting that the intermembrane space (IMS)-faced domain of Tom22 has no chaperone-like activity. SDS-PAGE analysis also showed that the aggregation of CS is largely suppressed by a series of the cytosolic domain mutants of Tom22 but not by the IMS domain (Fig. 4C). In the binding assay, denatured CS associated with the cytosolic domain of Tom22 but not with the IMS-faced domain (Fig. 4D). Native CS did not bind to either domain. These results indicate that the cytosolic domain of Tom22 binds to unfolded protein. Thus, Tom22 also has a function to maintain substrate preproteins in an unfolded state and prevent them from aggregation.
Tom70 Has Only a Weak Chaperone-like Activity for CS-Finally, we examined whether the cytosolic domain of Tom70 can suppress the thermal aggregation of CS (Fig. 5). When CS was incubated with Tom70-(95-608), a histidine-tagged cytosolic domain of Tom70, the turbidity started to increase after 4 min (Fig. 5A). During the incubation from 4 to 10 min, the turbidity observed in the presence of Tom70-(95-608) was a little lower than control (CS alone). However, after ϳ10 min the turbidity in the presence of Tom70-(95-608) increased much higher than control, probably due to thermal aggregation of Tom70-(95-608) itself. In SDS-PAGE analysis, the aggregated CS was reduced a little by the addition of Tom70-(95-608) (Fig. 5B). These results suggest that the cytosolic domain of Tom70 has only a weak activity to suppress the aggregation of CS.

DISCUSSION
Although the importance of cytosolic chaperones in mitochondrial protein import has been established (1-3), there are few studies on the chaperone-like function of the receptor proteins on the mitochondrial surface. In the present study, we found that the cytosolic domain of Tom20 binds to unfolded proteins regardless of the presence or absence of presequence (Fig. 1). This suggests that Tom20 functions as a receptor not only for presequences but also for the unfolded portions of mature preproteins. Although a number of mitochondrial preproteins are known to be maintained in an unfolded state during their cytosolic transport and during translocation into mitochondria, the extent to which preproteins are in a folded or unfolded state on the mitochondrial surface was not known. Our finding suggests that preproteins are maintained unfolded throughout the import process from the cytosol to the inside of mitochondria. On the other hand, some preproteins are transferred to the mitochondrial surface in a folded state. They are then imported into the mitochondria by sequential unfolding of the preprotein driven by the pulling force applied by mitochondrial Hsp70 (30). Also in this case, Tom20 and Tom22 may serve as receptors for partially unfolded portions of the preproteins.
The present results also suggest that Tom20 has a chaperone-like activity capable of suppressing the thermal aggregation of CS (Fig. 2). This activity may be important for maintaining mitochondria-targeted preproteins unfolded on the mitochondrial surface and suppressing their aggregation. Because preproteins that failed to be translocated into mitochondria are rapidly degraded (20,31), the activity of Tom20 may increase the efficiency of preprotein import by preventing preproteins from degrading. The importance of this activity for the mitochondrial protein import was suggested by our recent find- Tom20 and Tom22 Have Chaperone-like Activity ing on AIP (20). The chaperone-like activity of AIP appears to stabilize matrix precursor proteins such as pOTC and facilitate its import into mitochondria. The importance of chaperones in protein transport is also illustrated by the role of trigger factor as a ribosome-associated protein that stabilizes the translocationcompetent form of a secretory precursor protein in Escherichia coli (32,33). Trigger factor exhibits chaperone-like activity to suppress the thermal aggregation of substrate proteins (34).
In the current studies, the active site of Tom20 responsible for chaperone-like activity was suggested to be identical or close to the binding site for a presequence peptide (Fig. 3). In general, chaperone proteins have a hydrophobic surface for binding to their unfolded substrate proteins. Because the binding groove of Tom20 for presequences is composed of hydrophobic residues (22), it is possible that the groove also serves as a binding site for unfolded proteins. More than one Tom20 molecule may sequentially bind to a single preprotein molecule to prevent it from aggregation. We suppose that a Tom20 molecule first binds to the presequence of a preprotein, and then another Tom20 molecule(s) binds to the mature portion. Then the preprotein, dissociated from the Tom20 molecules, is transferred to Tom22 and is translocated across the outer membrane through the general import pore. Mitochondrial preproteins may be strictly distinguished from other proteins by cooperative recognition by Tom20 and other receptors such as Tom22 and/or by sequential recognition of the multiple sites in presequences by these receptors.
Furthermore, we also found that the cytosolic domain of Tom22 has a chaperone-like activity (Fig. 4). Because preproteins initially recognized by Tom20 are subsequently trans- Tom20 and Tom22 Have Chaperone-like Activity ferred to Tom22, it is not unexpected that Tom22 has a chaperone-like activity. Tom22 is contained in a large complex forming a general import pore. Therefore, Tom22 may maintain unfolded preproteins on a general import pore when preproteins are transferred from Tom20. In contrast, we found only a weak chaperone-like activity of Tom70 when CS was used as a substrate protein (Fig. 5). Because Tom70 preferentially recognizes preproteins with internal targeting signals, such as the ADP/ATP carrier protein, it is not unexpected that Tom70 cannot recognize CS as its substrate. Our finding of chaperone-like activities of Tom20 and Tom22 will contribute to further understanding of the properties and functions of these receptors. Tom20 and Tom22 Have Chaperone-like Activity