Functional Staging of ADP/ATP Carrier Translocation across the Outer Mitochondrial Membrane*

The ADP/ATP carrier (AAC) is the major representative of the inner membrane carrier proteins of mitochondria that are synthesized without cleavable presequences. The characterization of the import pathway of AAC into mitochondria has mainly depended on an operational staging system. Here, we introduce two approaches for analyzing the import of AAC, blue native electrophoresis and folding-induced translocation arrest, that allow a functional staging of AAC transport across the outer membrane. (i) Blue native electrophoresis permits a direct monitoring of the receptor stage of AAC and its chase into mitochondria. Binding to this stage requires the receptor protein Tom70 but not Tom37 or Tom20. (ii) A fusion protein between AAC and dihydrofolate reductase can be selectively arrested in the general import pore complex of the outer membrane by ligand induced folding of the passenger protein. Cross-linking demonstrates that the arrested preprotein is in close contact not only with several receptors and Tim10 but also with the channel protein Tom40, providing the first direct evidence that cleavable preproteins and carrier preproteins interact with the same outer membrane channel. The staging system presented here permits a molecular dissection of AAC transport across the outer mitochondrial membrane, relates it to functional units of the translocases, and indicates a coordinated and successive cooperation of distinct translocase subcomplexes during transfer of the preprotein.

Although many mitochondrial preproteins are synthesized with cleavable amino-terminal targeting sequences (presequences), several groups of mitochondrial proteins, in particular membrane proteins, are synthesized without presequences (1)(2)(3)(4). Instead, these noncleavable preproteins possess internal targeting information. An example is the ADP/ATP carrier (AAC), 1 which is a major representative of the large class of carrier proteins of the inner membrane. The current information shows that the import route of carrier proteins into mitochondria is largely distinct from that of cleavable preproteins.
For the import of the hydrophobic inner membrane carriers, the following mitochondrial transport components have been assigned (17). The carrier preproteins are targeted via the receptor Tom70 (18,19) in cooperation with Tom37 (20); Tom20 can also be involved in the recognition of carriers (21,22). The transport pathways of carrier preproteins and cleavable preproteins merge at Tom5 of the GIP complex (13). After crossing the outer membrane, the carrier import route separates completely from that of cleavable preproteins (23). Carrier preproteins are bound by the Tim9-Tim10 complex in the intermembrane space (24,25) and are transferred to the carrier translocase of the inner membrane, consisting of the peripheral membrane protein Tim12 and the integral proteins Tim22 and Tim54 (26 -29). Insertion into the inner membrane requires a ⌬ and is followed by assembly to the mature dimeric form of the carrier protein (30,31).
Most studies on the import pathway of carrier preproteins made use of a system of import stages that were defined by the differential protease sensitivity of AAC translocation intermediates in isolated mitochondria (23,32,33). Accumulation of preproteins was achieved by dissipation of the ⌬ and lowering of the ATP levels in the in vitro import reactions. The following stages were defined (see also Fig. 1A): stage I, AAC bound to cytosolic cofactors; stage II, AAC bound to the mitochondrial surface; stage III, AAC was largely protected against externally added protease despite a dissipation of the inner membrane ⌬; stage IV, AAC was inserted into the inner membrane; stage V, AAC was dimeric and mature. This operational staging system was developed before the import components of the Tom and Tim machineries were identified. The assignment of Tom and Tim components to individual import stages is far from being complete. It is likely that the stages only represent snapshots of a carrier import pathway and that experimental gaps exist along the pathway.
For this report, we focused on the transport route of AAC across the outer membrane that poses several open questions. It is unknown which of the receptors Tom70, Tom37, and Tom20 are actually involved in the initial recognition of carrier preproteins at the mitochondrial surface because the initial binding step could not be directly monitored. Moreover, it is unclear whether the receptor Tom22 comes into direct contact with carrier preproteins or plays more of an indirect role in their import. On the one hand, prebinding of antibodies against Tom22 to isolated mitochondria showed an inhibitory effect on carrier import (9), and a reduction of the mitochondrial levels of Tom22 impaired carrier import (34). On the other hand, in vitro binding studies with purified receptors demonstrated an interaction of carrier preproteins with Tom70 and Tom20, but not with Tom22 (22). Although it is thought that the carrier preproteins are transported through the same outer membrane channel (Tom40) as cleavable preproteins, direct evidence for such a contact between Tom40 and a carrier preprotein has not been obtained so far due to the inability to stably arrest a carrier preprotein in the outer membrane import pore. In this study, we employed two approaches for a functional characterization of the carrier pathway. Blue native electrophoresis permitted a direct monitoring of the receptor stage, demonstrating that only Tom70, but not Tom37 or Tom20, was required for the initial binding. A carboxyl-terminal fusion of AAC with dihydrofolate reductase (DHFR) yielded a protein that could be selectively and stably arrested in the GIP complex by methotrexate-induced folding of the DHFR domain. This arrested preprotein yielded the first evidence for a close proximity of AAC with Tom22 and Tom40. We compare the previous operational staging with the new functional staging of AAC import, which permits a molecular dissection of carrier transport across the outer mitochondrial membrane.

EXPERIMENTAL PROCEDURES
Import of Preproteins into Isolated Mitochondria-The Saccharomyces cerevisiae strains used in this study are shown in Table I. Mitochondria were isolated from yeast cells grown in YPG media (1% yeast extract, 2% bactopeptone and 3% glycerol) (35,36), resuspended in SEM buffer (250 mM sucrose, 1 mM EDTA, 10 mM MOPS, pH 7.2) and stored at Ϫ80°C in 10 mg/ml aliquots. Mitochondrial preproteins were obtained by in vitro transcription using SP6 polymerase and in vitro translation using rabbit reticulocyte lysate (Amersham) in the presence of [ 35 S]methionine/cysteine (37,38).
For the in vitro import of preproteins, mitochondria were diluted in bovine serum albumin-containing buffer (3% (w/v) fatty acid-free bovine serum albumin, 80 mM KCl, 5 mM MgCl 2 , 10 mM MOPS/KOH, pH 7.2) in the presence of 2 mM ATP and 2 mM NADH. Where indicated, methotrexate was also added to a final concentration of 20 M. The mitochondrial membrane potential was dissipated by the addition of 8 M antimycin A, 20 M oligomycin, and 1 M valinomycin (Sigma) prior to import reactions. Rabbit reticulocyte lysate containing 35 S-labeled preproteins (2.5-5% (v/v) of import reaction) was incubated with yeast mitochondria (25-50 g of protein) at 25°C. Valinomycin (1 M) was added to stop import. Where indicated, samples were treated with proteinase K (50 g/ml) on ice for 15 min. The protease was inactivated by the addition of 1 mM phenylmethylsulfonyl fluoride, and samples were incubated for an additional 10 min at 4°C. Following centrifugation and washing in SEM buffer, mitochondrial pellets were lysed in the appropriate detergent-containing buffer and separated by SDS-or blue native-polyacrylamide gels.
Analysis of AAC Stages-For the formation of a stage II AAC intermediate, 25 units/ml apyrase (Fluka) was added separately to 35 Slabeled AAC in reticulocyte lysate and mitochondria in import buffer lacking externally added ATP and containing 8 M antimycin A, 20 M oligomycin, and 1 M valinomycin (Sigma). Samples were incubated at FIG. 1. Dissection of import stages of AAC by blue native electrophoresis. A, operational staging of the import of AAC into mitochondria according to Pfanner and Neupert (32) and Pfanner et al. (33). Cytosolic AAC most likely bound to molecular chaperone(s) (C) is termed stage I. Arrest of AAC with mitochondrial receptors (R) by depletion of ATP represents stage II, whereas translocation through the outer membrane into the intermembrane space (IMS) in the presence of ATP but absence of a membrane potential (⌬) represents stage III. The presence of the ⌬ results in the Tim machinery-dependent insertion of AAC into the inner membrane (stage IV), where it assembles into its native dimeric form (stage V). B, separation of AAC translocation intermediates by blue native electrophoresis. [ 35 S]AAC in reticulocyte lysate was incubated with isolated mitochondria in the absence (lane 1) or presence (lanes 2 and 3) of ATP and the absence (lanes 1 and 2) or presence (lane 3) of ⌬ in order to arrest AAC at the appropriate stages. Samples were halved and treated with (lanes 4 -6) or without (lanes 1-3) proteinase K (Prot. K) and solubilized in digitonin-containing buffer and complexes separated by blue native PAGE prior to phosphorimage analysis. The AAC stages represented by blue native PAGE are indicated. As a control to verify that the final stage of AAC import (stage V) represents native AAC, mitochondria were solubilized in digitonin-containing buffer, subjected to blue native PAGE, and subsequently immunodecorated with antibodies directed specifically against AAC (lane 7). ade2-101 his3-⌬200 leu2-⌬1 ura3-52 trp1-⌬63 lys2-801 tom20::URA3 7 MR100 ade2-101 his3-⌬200 leu2-⌬1 ura3-52 trp1-⌬63 lys2-801 tom37::LEU2 This study 25°C for 10 min prior to the addition of preprotein to the mitochondria. For the formation of a stage III AAC intermediate, 35 S-labeled AAC in reticulocyte lysate containing endogenous ATP levels, was added to mitochondria previously dissipated of a membrane potential by the addition of 1 M valinomycin. For the formation of chasable AAC intermediates, mitochondria and reticulocyte lysates were depleted of ATP according to Glick (39) prior to import at 25°C for 10 min. Dissipation of the membrane potential was achieved by addition of 5 M carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone. Samples were divided and centrifuged before resuspension of mitochondrial pellets in an equal volume of bovine serum albumin-containing buffer with or without an ATP regenerating system (2 mM ATP, 100 g/ml creatine kinase, 5 mM creatine phosphate) and in the presence of 2 mM NADH or 5 M carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone as indicated. Samples were incubated for a further 10 min at 25°C prior to centrifugation.
Construction of AAC-DHFR Plasmid for in Vitro Transcription/ Translation-The open reading frame of yeast AAC2 (40) lacking a stop codon was amplified by polymerase chain reaction using Vent polymerase (New England Biolabs Inc.) with yeast genomic DNA as template and the primers AAC2-1 (5Ј-CTTCAGAATTCATACATTAACATAC-3Ј) and AAC2-2 (5Ј-AGATCTGGATCCTTTGAACTTCTTACCAAACAA-3Ј), containing EcoRI and BamHI restriction enzyme sites, respectively. The open reading frame of mouse DHFR was amplified by polymerase chain reaction using the primers DHFR-A (5Ј-AGATCTGGATCCATG-GTTCGACCATTGAACTG-3Ј) and DHFR-B (5Ј-ATAGCATGCTTAGT-CTTTCTTCTCGTAGAC-3Ј), containing BamHI and SphI restriction enzyme sites, respectively. The polymerase chain reaction products were extracted with phenol/chloroform and precipitated with ethanol prior to digestion with EcoRI and BamHI for the AAC product or with BamHI and SphI for the DHFR product. The samples were electrophoresed on 1% NuSieve GTG agarose (FMC Bioproducts), excised, and used directly in a three-way ligation reaction with the plasmid pGEM-4Z (Promega) previously cut with EcoRI and SphI enzymes. The BamHI site linking the AAC and DHFR open reading frames encoded the amino acid residues Gly and Ser.
Disruption of TOM37-To delete the coding region of the TOM37 gene, the LEU2 marker was ligated between a 800-base pair DNA fragment corresponding to the 5Ј-noncoding region of TOM37 and a 400-base pair DNA fragment of the 3Ј-noncoding region and cloned into pGEM-4Z. The cloned DNA was amplified by polymerase chain reaction and used to transform the diploid S. cerevisiae strain YPH501. Leucineprototrophic cells were sporulated and subjected to spore analysis. Following sporulation, the genotypes of the spores were analyzed by growth on selective media. Standard procedures were used for manipulations of DNA and yeast strains (47,48).
Cross-linking and Immunoprecipitation-Mitochondria containing accumulated 35 S-labeled AAC or AAC-DHFR were reisolated through a sucrose cushion, washed and incubated with 0.4 mM ethylene glycol bis(succinimidylsuccinate) (EGS) as described previously (38). After reisolation, mitochondria were boiled in SDS-containing buffer prior to dilution in Triton X-100 buffer and then subjected to immunoprecipitation (38). Samples were electrophoresed on a gradient Tris-Tricine gel (49) prior to phosphorimage analysis.
Miscellaneous-Antisera against an oligopeptide derived from Tom37 were prepared by coupling the peptide to keyhole limpet hemocyanin (Calbiochem, Frankfurt, Germany) (50) and injection into rabbits.
For salt and alkaline extraction of mitochondrial proteins, mitochondrial pellets (50 g) were resuspended in 100 l of SEM, 0.5-2 M NaCl, or 100 mM Na 2 CO 3 and incubated on ice for 15 min prior to centrifugation and separation into pellet and supernatant fractions (51). Samples were subjected to SDS-PAGE and immunodecoration. SDS-PAGE was performed according to Laemmli (52). Immunodecoration was per-formed according to standard procedures (46), and detection was achieved using the enhanced chemiluminescence method (Amersham Pharmacia Biotech). For detection of radiolabeled proteins, the dried gel or polyvinylidene difluoride membrane was exposed to phosphorimage storage cassettes prior to phosphorimage analysis (Molecular Dynamics).

Import Stages of ADP/ATP Carrier Dissected by Blue Native
Electrophoresis-The preprotein of AAC was synthesized in vitro in rabbit reticulocyte lysates in the presence of [ 35 S]methionine/cysteine. Mitochondria were isolated from yeast wildtype cells. For accumulation of AAC at stage II (Fig. 1A), reticulocyte lysate and mitochondria were depleted of ATP, and the inner membrane potential was dissipated. The reticulocyte lysate was incubated with the mitochondria at 25°C. The mitochondria were reisolated, lysed with the detergent digitonin, and subjected to blue native electrophoresis and digital autoradiography (43,44,53). AAC accumulated at stage II and migrated at a broad range at the vicinity of 400 -500 kDa (Fig.  1B, lane 1). When the mitochondria were treated with proteinase K before lysis, the 400 -500-kDa band disappeared (Fig. 1B,  lane 4), demonstrating that the AAC was accumulated at the mitochondrial surface.
For accumulation of AAC at stage III (Fig. 1A), the membrane potential was dissipated, yet ATP was kept at a high level. After incubation, lysis and separation of native protein complexes by blue native electrophoresis, AAC migrated in a band in the low molecular mass range (below 60 kDa) (Fig. 1B,  lane 2). This band remained intact also after treatment of mitochondria with proteinase K (Fig. 1B, lane 5), confirming the typical property of a stage III intermediate (32,33).
Furthermore, in the presence of ATP and a ⌬, AAC was completely imported into the inner membrane and assembled  1 and 3) an ATP-regenerating system in the presence (lanes 1, 2, 5, and 7) or the absence (lanes 3, 4, 6, and 8) of the protonophore carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone. Samples were incubated a second time at 25°C for 10 min (2nd inc. (Chase)) to allow the chase of the AAC intermediates to proceed. Reactions were centrifuged, and mitochondrial pellets were solubilized in digitonin-containing buffer before analysis by blue native PAGE and subsequent phosphorimaging.
to the dimeric stage V form (Fig. 1A). In blue native electrophoresis, the in vitro imported AAC detected by digital autoradiography (Fig. 1B, lanes 3 and 6) assembled into a form indistinguishable from the in vivo imported AAC form that was detected by Western blotting (Fig. 1B, lane 7) (31,53).
We asked whether the import stages II and III dissected by blue native electrophoresis represented true translocation intermediates, i.e. if they could be chased to subsequent import stages III and V (stage IV is only transiently observed as kinetic intermediate at low temperature and thus cannot be separated by blue native electrophoresis (32)). AAC was accumulated at stage II, and mitochondria were reisolated. The sample was split into halves. In one portion, the ATP level was kept low and thus the preprotein remained at stage II (Fig. 2,  lane 1). In the other portion, ATP was added to the reisolated mitochondria, whereas the ⌬ was still dissipated. Thereby, the preprotein was efficiently transported to the low molecular mass form, indicative of the stage III intermediate (Fig. 2, lane  2). AAC was also accumulated at stage II but in the presence of a ⌬. Samples were halved, and ATP was added to one sample resulting in the chase of AAC from stage II (Fig. 2, lane 3) to stage V (Fig. 2, lane 4).
In addition, AAC was first accumulated at stage III, the mitochondria were reisolated and split into two portions. In one sample, the ⌬ remained dissipated by inclusion of the protonophore carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone in the second incubation (Fig. 2, lane 5). In the other sample, a ⌬ was restored by omitting the protonophore in the second incubation and inclusion of an energy-regenerating system. Thereby the stage III intermediate was chased to the dimeric stage V form (Fig. 2, lane 6). A small amount of AAC remained at the stage III position after the second incubation (Fig. 2, lane 6). We noticed a similar situation when the first incubation was performed in the presence of ATP and a ⌬, whereas the second incubation was in the absence of a ⌬ (Fig.  2, lane 7). When an energy-regenerating system restored ⌬ (and ATP levels) in the latter situation, also the remaining stage III intermediate was chased to the mature stage V form (Fig. 2, lane 8), indicating that also this stage III form represented a true intermediate. In some experiments, bands at the range of 400 kDa were observed in variable and low amounts. These bands were not related to AAC import and apparently represent nonspecifically 35 S-labeled mitochondrial complexes.
We conclude that blue native electrophoresis permits a direct monitoring of the transport of AAC from stage II to stage III and to the mature form, stage V. Stages II and III represent true translocation intermediates because they are efficiently chased into mitochondria.
Accumulation of AAC at the Receptor Stage II Requires Tom70 but Not Tom37 or Tom20 -To test which receptor subunits were required for the accumulation of AAC at stage II, we isolated mitochondria from yeast strains lacking either TOM70 (21) or TOM20 (7). tom70⌬ mitochondria completely lacked the stage II intermediate of AAC (Fig. 3A, lane 2), whereas tom20⌬ mitochondria generated this intermediate as wild-type mitochondria did (Fig. 3A, lanes 3 and 1, respectively). It must be emphasized that detailed previous work showed that in tom70⌬ mitochondria, the remainder of the Tom machinery, including Tom20 and the GIP complex with Tom22 and Tom40, was fully intact and competent for the import of cleavable  3 and 4), or 100 mM Na 2 CO 3 (lanes 5 and 6) and separated into pellet (P) and supernatant (S) fractions prior to SDS-PAGE, Western blotting, and immunodecoration with antibodies specific for Tom70, Tom40, Tom37, and Tim10 as indicated. Tom37 remained similarly membrane-associated to treatment of mitochondria with 2 M NaCl. C, Tom37 migrates in an outer membrane complex distinct from Tom70, Tom20, and the GIP complex. Isolated S. cerevisiae wild-type mitochondria were lysed in digitonin buffer and subjected to blue native PAGE in the first dimension and SDS-PAGE in the second dimension according to Dekker et al. (44). Following electrophoresis, proteins were blotted and subsequently immunodecorated with antibodies specific for Tom proteins as indicated. The positions of the various complexes are indicated. D, mitochondria from wild-type (WT) S. cerevisiae or from mutant strains lacking Tom37 (tom37⌬) or Tom70 (tom70⌬) were subjected to SDS-PAGE and Western blotting and immunodecorated with antibodies specific for Tom37 or Tom70 as indicated. E, formation of the stage II intermediate of AAC in tom37⌬ mitochondria. The stage II AAC intermediate was accumulated in wild-type (WT) S. cerevisiae mitochondria or mitochondria from the strain lacking Tom37 (tom37⌬) as described above and subjected to blue native PAGE and phosphorimaging analysis. preproteins (7,21,44). tom20⌬ mitochondria, however, have a strongly decreased level of Tom22 because Tom20 is required for the biogenesis of the preprotein of Tom22 (11,54,55). The efficient formation of the stage II intermediate in tom20⌬ mitochondria thus indicates that Tom22 is not limiting for generation of the receptor stage of AAC.
Gratzer et al. (20) identified Tom37 as a further subunit of the Tom machinery. They showed a genetic interaction between TOM70 and TOM37 and an involvement of Tom37 in the import of AAC. Tom37 is accessible to protease added to intact mitochondria, i.e. it is exposed on the mitochondrial surface (20). Further information on the association of Tom37 with the outer membrane and its requirement for import stages of AAC is scarce, probably because of the difficulty of generating a monospecific antibody against Tom37. We found that a peptide corresponding to the residues 299 -311 of Tom37 coupled to hemocyanin led to production of a monospecific rabbit antiserum against Tom37 and used this antiserum to characterize basic properties of Tom37. (i) Tom37 remained membraneassociated upon treatment of mitochondria with up to 2 M sodium chloride (Fig. 3B, lane 3). However, when the mitochondria were treated at alkaline pH to extract soluble proteins and peripheral membrane proteins (56), Tom37 was extracted from the membranes (Fig. 3B, lane 6). As a control, Tim10, an intermembrane space protein that is only loosely associated with the inner membrane, was similarly extracted at pH 11.5 (Fig. 3B, lane 6), whereas Tom70 and Tom40, as well as all other Tom proteins studied so far, remained in the membrane sheets (Fig. 3B, lane 5) (4). Tom37 is thus the first Tom protein that behaves as a peripheral membrane protein. (ii) Blue native electrophoresis of digitonin-lysed mitochondria followed by SDS-PAGE in the second dimension separates the Tom machinery into distinct subcomplexes as shown in Fig. 3C: Tom70 at 150 -220 kDa; Tom20 at 60 -120 kDa; and the GIP complex containing Tom40, Tom22, and the small Tom proteins (13,44) at ϳ400 kDa. Tom37 migrated in a ϳ250-kDa region in blue native electrophoresis and thus was distinct from the other Tom subcomplexes (Fig. 3C). (iii) To test the proposal that Tom37 cooperates with Tom70 in binding of AAC (20), we constructed a yeast strain with a deletion of the TOM37 gene. The resulting tom37⌬ cells are viable (20). The lack of Tom37 was confirmed by immunodecoration with anti-Tom37 antibodies (Fig. 3D). Surprisingly, the tom37⌬ mitochondria were fully competent for generation of the stage II intermediate of AAC as determined by blue native electrophoresis (Fig. 3E). Thus, these results show that Tom37 is not an integral membrane protein but is stably associated with the outer membrane via an interaction with as yet unidentified components. Tom37 is not required for the initial binding of AAC to Tom70 but apparently operates at a later stage of AAC import.
We conclude that the formation of the stage II intermediate of AAC strictly requires Tom70. Tom37 and Tom20, and probably also Tom22, are not needed for the initial binding of the AAC preprotein to mitochondria. electrophoresis at a very low molecular mass, below 60 kDa (Figs. 1 and 2). Because the known Tom components migrate above 60 kDa and the GIP complex at 400 kDa (Fig. 3C) (44), this demonstrates that the stage III-AAC is not associated stably enough with any of the Tom components to withstand separation by native electrophoresis. The arrest of a stage III intermediate is achieved only at the level of the inner membrane, that is by dissipation of the ⌬. We therefore reasoned that under these conditions the major portion of an AAC molecule is transferred so far into the intermembrane space that the interaction with the GIP complex is minimized.

An AAC-DHFR Fusion Protein Can Be Arrested in the Tom
In order to obtain an AAC that could be stably arrested in the GIP complex, we constructed a fusion protein between AAC and the enzyme DHFR (Fig. 4A). AAC-DHFR synthesized in reticulocyte lysate efficiently associated with mitochondria and was transported to a protease-protected location in the presence of a ⌬ (Fig. 4A, lanes 5-7). Also in the absence of a ⌬, part of AAC-DHFR became resistant to added protease, resembling the property of a stage III intermediate (Fig. 4A, lane 8).
To directly test if AAC-DHFR was transported to the inner membrane in a ⌬-dependent manner, we used blue native electrophoresis. Only in the presence of a ⌬, a product was formed (Fig. 4B, lane 2) that migrated slightly slower than dimeric AAC (Fig. 4B, lane 4), indicating a ⌬-dependent assembly of AAC-DHFR. The specific ligand methotrexate stabilizes the folding state of DHFR and thereby prevents complete import of cleavable preproteins that carry DHFR as passenger protein (59,60). Methotrexate indeed strongly inhibited the ⌬-dependent assembly of AAC-DHFR (Fig. 4C, lane 2). The effect was specific because the import of authentic AAC was not affected by addition of methotrexate (Fig. 4C, lane 4).
In the absence of ATP and a ⌬, AAC-DHFR was accumulated as a stage II intermediate comparable to, but at a higher molecular mass range, as the authentic AAC stage II intermediate (Fig. 4D, lane 1). In the presence of ATP and methotrexate, AAC-DHFR was efficiently accumulated in a distinct high molecular mass region of 450 -500 kDa (Fig. 4D, lane 2). The accumulation of AAC-DHFR at this position was independent of a ⌬ (Fig. 4D, lane 3), indicating that the accumulation did not involve the inner membrane translocase. Indeed, accumulated AAC-DHFR was fully sensitive to externally added proteinase K independently of ⌬ (Fig. 4E, lanes 5 and 6). A comparison of AAC-DHFR arrest in wild-type and tom70 mitochondria revealed a strong reduction in the mutant mitochondria (Fig. 4E, lanes 3 and 4), indicating that AAC-DHFR preferentially utilizes Tom70 as receptor, as does authentic AAC (19,21). In this case, the addition of ATP to the import reaction enabled a fraction of AAC-DHFR to bypass the Tom70-bound stage II formation (21). The sizes of AAC-DHFR of ϳ55 kDa and of the GIP complex of ϳ400 kDa largely account for the observed size of the arrested intermediate of 450 -500 kDa. These results suggested that in the presence of methotrexate, AAC-DHFR was accumulated as an outer membrane-spanning intermediate in the GIP complex. For a direct demonstration, the following cross-linking approach was performed.
Cross-linking of AAC and AAC-DHFR to Tom Proteins and Tim10 -We first probed the molecular environment of authentic AAC accumulated at stage II or stage III. Mitochondria with stage II-AAC were treated with the amino-reactive homobifunctional cross-linking reagent EGS (Fig. 5A, lane 1). The mitochondria were then lysed with SDS and subjected to immunoprecipitation under stringent conditions with antibodies directed against Tom70, Tom40, or Tim10. As expected, Tom70 represented the major cross-linked products of AAC at stage II (Fig. 5A, lane 2) (61). Only weak cross-linked products that precipitated with anti-Tom40 or anti-Tim10 were observed (Fig. 5A, lanes 3 and 4). Mitochondria with stage III-AAC yielded a different cross-linking pattern. Tim10 represented a major cross-linked product (Fig. 5A, lanes 5 and 8) in agreement with the observations of Koehler et al. (29) and Sirrenberg et al. (26), whereas anti-Tom70 and anti-Tom40 precipitated only cross-linked products of low abundance (Fig. 5A,  lanes 6 and 7).
The fusion protein AAC-DHFR was then accumulated across the outer membrane in the presence of ATP and methotrexate. Addition of EGS generated several cross-linked products (Fig.  5B, lane 2) that were specifically recognized by anti-Tom antibodies. Besides Tom70 (Fig. 5B, lane 3), Tom40 represented a major cross-linked product (Fig. 5B, lane 4). Moreover, Tom22 and Tom20 were identified among the cross-linked products (Fig. 5B, lanes 5 and 6). Finally, AAC-DHFR was also crosslinked to Tim10 (Fig. 5B, lane 7), demonstrating that AAC- DHFR was reaching into the intermembrane space. This crosslinking approach demonstrates that the arrested preprotein spans the GIP complex and provides the first direct evidence that AAC in transit is in close proximity to Tom22 and Tom40. DISCUSSION This study presents two new approaches to analyze the translocation of the ADP/ATP carrier across the outer mitochondrial membrane. It thereby extends the operational staging of AAC import used so far ( Fig. 6; see also Fig. 1A) and links the import stages to the function of the translocase components (functional staging).
By the use of blue native electrophoresis, the initial contact of AAC with the mitochondrial surface as well as the subsequent import can be directly monitored. The formation of this receptor stage (stage II) selectively requires the receptor Tom70 (Fig. 6). Three other Tom proteins, Tom37, Tom20, and Tom22, expose large domains to the cytosol and have been proposed to function as import receptors. Tom37 has been assumed to function as a receptor subunit together with Tom70 (20). We show that Tom37 is not required at this early import stage and so must function at an later import step. In addition, we find that Tom37 is firmly associated with the outer membrane, yet in contrast to all other known Tom proteins, is a peripheral membrane protein. Tom20 is similarly not required for the initial binding of AAC to mitochondria. Additionally, because tom20⌬ mitochondria are strongly reduced in the mitochondrial content of Tom22 because of the dependence of Tom22 biogenesis on Tom20 (10,11,44,53), this suggests that Tom22 is also not limiting for formation of the stage II intermediate. This is in agreement with the observation that Neurospora crassa mitochondria with a reduced level of Tom22 were able to bind AAC to the outer membrane surface (34). Together these results suggest that of the four mitochondrial receptor subunits (3,4,55), only Tom70 is strictly required for the first binding step of AAC. However, under some circumstances, such as in mitochondria lacking Tom70, AAC can bypass this initial binding step and utilizes Tom20 and Tom22 as receptors although in this case, the efficiency of AAC binding and import is strongly reduced (21) (Fig. 6).
The receptor Tom70 migrates at ϳ150 -220 kDa in a native gel, whereas Tom70-bound AAC migrates at ϳ400 -500 kDa. How can this difference be explained? The cytosolic form of AAC, a hydrophobic protein, is not a soluble monomer, but is present in larger oligomeric complexes, together with cytosolic cofactors/chaperones (52,(62)(63)(64)(65). Release of a preprotein from some chaperones, such as the mitochondrial import stimulating factor that directs preproteins to Tom70, has been shown to require ATP (63)(64)(65). The binding of AAC to Tom70 occurs at low ATP levels, suggesting that cytosolic chaperones are still bound to the preprotein. In the subsequent ATP-dependent transport step across the outer membrane, the cytosolic factors are removed from the preprotein. This can be monitored by the use of blue native electrophoresis because the stage III intermediate migrates at a small molecular mass, below 60 kDa.
Because each Tom subcomplex migrates at a molecular mass larger than 60 kDa (44), AAC at stage III is not stably associated with any Tom protein and evidently not with the 400-kDa GIP complex. AAC at stage III is translocated across the GIP and is mainly exposed to the intermembrane space (23,66). As AAC at stage II is largely confined to Tom70, these findings explain why the operational staging system failed in providing direct evidence for an interaction of AAC with the import channel Tom40. We compared cross-linking of AAC to import components at stage II and stage III. Only very weak crosslinking to Tom40 was observed at both stages, whereas the major cross-linked products contained Tom70 at stage II and FIG. 6. Functional staging of import of AAC into mitochondria. Comparison of the operational staging and functional staging of AAC. The operational staging is based on the methods used to accumulate preproteins (32,33). Stage IV represents a kinetic intermediate. The functional staging relates transport stages to the function of import components. Based on the work presented here, the various studies on the characterization of import components are incorporated into a hypothetical model of the complete import pathway of AAC as major representative of the carrier import route. During transfer of the preprotein, the functional units of import components, indicated by the functional stages, cooperate in a coordinated and successive manner (see last paragraph under "Discussion" for further details and references). Tom5 is closely associated with Tom40 (13) and is thus not further labeled in the figure. Cpn, cytosolic chaperone; IM, inner membrane; IMS, intermembrane space; MSF, mitochondrial import stimulating factor; OM, outer membrane. small Tims at stage III. Although this weak cross-linking of AAC to Tom40 may be seen as the first experimental indication for a direct contact of the two proteins, the low efficiency raises questions about the significance of the cross-linking result.
In order to obtain a stable transport intermediate of AAC in the GIP, we generated a fusion protein between AAC and DHFR at the carboxyl terminus that was selectively arrested across the outer membrane by stabilizing the DHFR domain with the ligand methotrexate. The fusion protein followed the authentic carrier import pathway, including targeting via Tom70, generation of a stage II intermediate, and membrane potential-dependent import and assembly in the absence of methotrexate. The methotrexate-induced accumulation of AAC-DHFR in the GIP complex is independent of the presence or absence of the inner membrane potential ⌬. AAC-DHFR thus represents the first intermediate that selectively spans the Tom channel in a stable manner, because a stable arrest of presequence-containing preproteins in the Tom channel requires the concomitant ⌬-dependent accumulation in the Tim17-Tim23 channel of the inner membrane at translocation contact sites (43,67,68). Cross-linking demonstrates that the accumulated AAC-DHFR is in close proximity to several Tom proteins (Tom70, Tom20, Tom22, and Tom40), as well as to Tim10, providing the first direct evidence that Tom22 and Tom40 are in the immediate vicinity of AAC in transit. These results indicate that the channel Tom40 is important for import of both cleavable preproteins and carrier preproteins and support the view of a direct role of Tom22 in the import of carrier preproteins. Although carrier preproteins synthesized in vitro do not interact with the purified cytosolic domain of Tom22 (22), short peptides derived from a carrier preprotein were found to specifically bind to this cytosolic domain, 2 indicating that unfolded segments of carrier preproteins can interact with Tom22. This fits to the view that Tom22 is not involved in the first recognition step of a carrier preprotein. After the initial binding by Tom70, the carrier preproteins are unfolded during the translocation process and thus can expose unfolded segments for binding to Tom22.
We propose that the cross-linking of arrested AAC-DHFR to a remarkably large number of Tom proteins and to the intermembrane space protein Tim10 points to an important characteristic of the import process. Although the initial recognition of the chaperone-bound preprotein at the mitochondrial surface is probably mediated by the single component Tom70, the translocation across the outer membrane involves a coordinated cooperation of several subcomplexes in the membrane and on the cis (cytosolic) and trans (intermembrane space) sides, including the receptor subcomplexes of Tom70 and Tom20, the GIP complex containing Tom22 and the import channel Tom40, and the Tim9-Tim10 complex in the intermembrane space. By incorporation of the studies on the identification and characterization of individual import components (13, 18, 19, 21, 23-29, 31, 61, 64), we propose a model in which the import of carrier preproteins is mediated by a successive interaction of functional units of the import machinery (depicted in Fig. 6). AAC bound to cytosolic chaperones binds to the receptor Tom70. In an ATP-dependent reaction, the chaperones are released, and Tom70 contacts the GIP complex and donates the AAC to the import channel Tom40 in a process involving Tom20, Tom22, and Tom5 on the cis side and small Tims on the trans side. The Tim9-Tim10 complex with bound AAC contacts the carrier translocase of the inner membrane, including Tim12 on the intermembrane space side and the membraneintegrated Tim22-Tim54. Little is known about the mechanism of ⌬-dependent insertion of AAC into the inner membrane by the carrier translocase and the assembly to the mature form. A detailed analysis of this inner membrane insertion will depend on the generation of a selective and stable translocation intermediate in the inner membrane carrier translocase as it has been possible now with the outer membrane GIP complex.