The carboxyl-terminal third of the dicarboxylate carrier is crucial for productive association with the inner membrane twin-pore translocase.

The carrier proteins of the mitochondrial inner membrane consist of three structurally related tandem repeats (modules). Several different, and in some cases contradictory, views exist on the role individual modules play in carrier transport across the mitochondrial membranes and how they promote protein insertion into the inner membrane. Thus, by use of specific translocation intermediates, we performed a detailed analysis of carrier biogenesis and assessed the physical association of carrier modules with the inner membrane translocation machinery. Here we have reported that each module of the dicarboxylate carrier contains sufficient targeting information for its transport across the outer mitochondrial membrane. The carboxyl-terminal module possesses major targeting information to facilitate the direct binding of the carrier protein to the inner membrane twin-pore translocase and subsequent insertion into the inner membrane in a membrane potential-dependent manner. We concluded that, in this case, a single structural repeat can drive inner membrane insertion, whereas all three related units contribute targeting information for outer membrane translocation.

The carrier proteins of the mitochondrial inner membrane consist of three structurally related tandem repeats (modules). Several different, and in some cases contradictory, views exist on the role individual modules play in carrier transport across the mitochondrial membranes and how they promote protein insertion into the inner membrane. Thus, by use of specific translocation intermediates, we performed a detailed analysis of carrier biogenesis and assessed the physical association of carrier modules with the inner membrane translocation machinery. Here we have reported that each module of the dicarboxylate carrier contains sufficient targeting information for its transport across the outer mitochondrial membrane. The carboxyl-terminal module possesses major targeting information to facilitate the direct binding of the carrier protein to the inner membrane twin-pore translocase and subsequent insertion into the inner membrane in a membrane potentialdependent manner. We concluded that, in this case, a single structural repeat can drive inner membrane insertion, whereas all three related units contribute targeting information for outer membrane translocation.
All nuclear-encoded mitochondrial proteins contain intrinsic targeting signals that direct them from their site of synthesis in the cytosol to their intended functional location. The nature of these signals, however, is not uniform from one protein to the next but falls into two main classes, either cleavable aminoterminal signal sequences (presequences) or multiple noncleavable internal signals (1)(2)(3)(4)(5)(6)(7). The translocase of the outer mitochondrial membrane (TOM) 1 complex recognizes and imports precursor proteins containing both types of targeting signal, whereas further import requires one of two translocases of the inner membrane (the TIM23 and TIM22 complexes) depending on the nature of the precursor protein (1,3,(5)(6)(7). Precursors with a presequence are imported via the TIM23 complex, whereas precursors with internal targeting signals that are destined for the mitochondrial inner membrane are imported via the essential Tim9 -10 complex of the intermembrane space and the TIM22 complex. The large hetero-oligomeric TIM22 complex is a twin-pore translocase (8) consisting of the integral membrane proteins Tim22, Tim54, and Tim18 and the peripheral membrane proteins of the small Tim family, Tim9, -10, and -12 (9 -17).
A major representative of precursor proteins with internal targeting signals is the metabolite carrier protein family of the mitochondrial inner membrane. Carrier proteins fulfill vital functions for eukaryotic cells, as they are critical for metabolite exchange between mitochondria and the cytosol by forming a transport route across the tightly sealed inner mitochondrial membrane. Members of the carrier protein superfamily share common structural features; they are ϳ300 amino acids long and divided into three tandem repeats (modules) of similar length (18,19). The structure of the mitochondrial ADP/ATP carrier (AAC) solved recently by x-ray crystallography revealed six transmembrane ␣-helices that form a barrel in the membrane (20). The structural fold of each repeat is similar and joined by two short loops exposed to the intermembrane space.
The question as to how the carriers themselves are transported to mitochondria and inserted into the inner membrane has drawn much attention. In vitro, the import of carrier proteins can be trapped at various stages along the way and has permitted a substantial dissection of the pathway (21)(22)(23). The newly synthesized carrier protein first binds to molecular chaperones in the cytosol (Stage I). The carrier protein then transfers to a major import receptor, Tom70, on the surface of mitochondria to which it remains bound in the absence of ATP (Stage II) (23)(24)(25)(26)(27). If the membrane potential (⌬) across the inner membrane is fully dissipated but ATP is supplied, the carrier protein moves further along the import pathway crossing the outer membrane through the TOM complex and associates with the Tim9 -10 complex of the intermembrane space (Stage III) (11-14, 23, 28). An association of the carrier protein with components of the inner membrane TIM22 complex is also detected under these conditions (8,13,29). If the ⌬ is, however, only partially dissipated, the carrier precursor accumulates at the stage of inner membrane insertion associating with the TIM22 complex (Stage IV) (8). A full ⌬ provides sufficient energy for the carrier precursor to insert into the inner membrane where subsequent assembly into a dimeric form occurs (Stage V) (23,30).
Beyond this analysis, several studies, using different carrier proteins, fusions, and truncated constructs, have addressed the question of where the targeting information for the different transport steps resides within the precursor. It appears that the targeting determinants, which drive outer membrane receptor binding, translocation through the general import pore, and even association with members of the small Tim family, are contained within each structural repeat (29,31,32). At least for receptor recruitment and outer membrane translocation, there is cooperation between each of these domains (31). Peptide binding scans of carriers revealed Tom70 and Tim9 -10 binding sites within all three repeats (33)(34)(35). There is, however, controversy regarding the nature of carrier targeting signals that direct the later stages of carrier translocation from the intermembrane space to insertion into the inner membrane. Using the uncoupling protein 1, it was initially described that the first repeat alone can facilitate targeting and membrane insertion of a chimeric protein (36), whereas an independent study (37) using the same carrier reports that the middle repeat contains the complete targeting signal to direct association with the inner membrane translocation machinery and, hence, membrane insertion. Although the carboxyl-terminal two-thirds of AAC has been implicated in facilitating the later stages of import into mitochondria (38,39), a more detailed study reports that the import signals that promote inner membrane insertion were specifically contained within the third module (29). In contrast to these data, a recent report suggests that the third module alone is insufficient to promote membrane insertion but rather that all three modules of the carrier precursor are required to allow membrane insertion via the TIM22 complex (32). Surprisingly, truncated forms of AAC fused to mouse dihydrofolate reductase were unable to follow the carrier transport pathway via the Tim9 -10 and TIM22 complexes but were instead mistargeted to the presequence translocase (TIM23 complex), which directs them into the matrix (32). However, the major drawback of these studies, collectively, has been the lack of an experimentally accessible translocation intermediate during insertion into the inner membrane, i.e. a transport intermediate accumulated at the TIM22 complex. Thus, despite all efforts the role of the different segments of the carrier in protein insertion into the inner membrane has remained enigmatic.
Recently, by manipulation of the ⌬ across the inner membrane, the precursor of the dicarboxylate carrier 1 (DIC) could be accumulated at the inner membrane TIM22 translocase (Stage IV) (8). This observation provided, for the first time, the possibility of directly analyzing the molecular determinants responsible for the late stages of carrier import leading to membrane insertion. We generated DIC constructs containing each module either singularly or in pairs. We showed that the third module of DIC contains the dominant targeting signals that allow binding to the twin-pore translocase and promotion of import from Stage III to V.

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-Previously described yeast strains YPH499 (WT) (Mata ade2-101 his3-⌬200 leu2-⌬1 ura3-52 trp1-⌬63 lys2-801) (40) and PRY19 (Tim18 ProtA ) (Mata ade2-101 his3-⌬200 leu2-⌬1 ura3-52 trp1-⌬63 lys2-801 tim18::TIM18-TEV-ProtA) (  , and DIC II,III (95-298) were amplified by PCR using Vent polymerase (New England BioLabs Inc.) and yeast genomic DNA as the template. For DIC constructs with amino-terminal deletions (DIC II , DIC III , DIC II,III ), oligonucleotide primers containing the SP6 RNA polymerase binding sequence directly followed by the start codon ATG and the nucleotide sequence encoding the first few amino acids of the desired DIC construct were used. Carboxyl-terminal deletion constructs (DIC I , DIC II , DIC I,II ) were amplified using primers complementary to the desired region with the addition of an engineered in-frame stop codon. For the non-truncated ends of DIC constructs, oligonucleotide primers complementary to the 5Ј-or 3Ј-untranslated regions of the DIC1 gene were used as appropriate. To generate DIC I,III (1-94 and 198 -298), the single modules DIC I and DIC III were first amplified by PCR from genomic DNA with flanking engineered endo-nuclease restriction sites EcoRI and BamHI and BamHI and XbaI, respectively. Following restriction digestion of DIC I , DIC III , and pGEM4Z, a three-way ligation was used to generate pGEM-DIC I,III . A short linker of G and S separates DIC I from DIC III in this construct. Radioactive precursors were synthesized by in vitro transcription using SP6 RNA polymerase (Stratagene) and PCR-generated DNA templates amplified from genomic DNA or pGEM-DIC I,III , followed by in vitro translation in the presence of [ 35 S]methionine/cysteine using rabbit reticulocyte lysate.
In Vitro Import into Isolated Mitochondria-Mitochondria were isolated as described previously (41) from PRY19 and the corresponding wild-type cell, YPH499. Standard import of radiolabeled carrier precursor into isolated yeast mitochondria was performed as described previously (42). For non-standard import reactions (⌬-modified), mitochondria (50 g of protein) were diluted to a final concentration of 0.5 g/l in import buffer (1% (w/v) fatty acid-free bovine serum albumin, 250 mM sucrose, 80 mM KCl, 20 mM KH 2 PO 4 , 5 mM MgCl 2 , 5 mM methionine, 10 mM MOPS-KOH, pH 7.2). Prior to import, samples were incubated for 5 min at 25°C and supplemented with 2 mM ATP, 2 mM NADH, 5 mM creatine phosphate, and 0.1 mg/ml creatine kinase. For partial uncoupling of mitochondria, 20 M oligomycin was added to the import buffer to prevent the reverse action of F o F 1 -ATPase and either 10 or 30 M carbonyl cyanide m-chlorophenylhydrazone (CCCP). Import (25°C) was initiated by addition of rabbit reticulocyte lysate containing radiolabeled precursor protein. Reactions were terminated after 15 min by addition of 1 M valinomycin. Samples were treated with 50 g/ml proteinase K for 15 min on ice, and then the protease was inactivated with 2 mM phenylmethylsulfonyl fluoride. Mitochondria were reisolated and either subjected to analysis by SDS-PAGE or solubilized in ice-cold 1% digitonin buffer (1% (w/v) digitonin, 0.1 mM EDTA, 50 mM NaCl, 10% (v/v) glycerol, 1 mM phenylmethylsulfonyl fluoride, 20 mM Tris-HCl, pH 7.4) and subjected to blue native-PAGE (BN-PAGE). For control imports, radiolabeled precursors were imported into either fully energized or ⌬-depleted mitochondria, solubilized in 0.5% (v/v) Triton X-100, and treated with proteinase K as above. Insoluble material was removed by centrifugation, and mitochondrial proteins were trichloroacetic acid-precipitated prior to analysis by SDS-PAGE and digital autoradiography.
Localization of Imported DIC Precursors-Following import of 35 Slabeled full-length or mutant DIC precursors, mitochondria were reisolated and resuspended in 200 l of SEM (250 mM sucrose, 1 mM EDTA, and 10 mM MOPS-KOH, pH 7.2) and then divided in half. To induce swelling and hence rupture of the outer membranes, one sample of mitochondrial suspension was diluted with four volumes of 10 mM MOPS-KOH buffer, pH 7.2. As a control, the other half of the mitochondrial suspension was diluted with 4 volumes of SEM buffer. After incubation on ice for 30 min, mitochondria (mitoplasts and control) were reisolated by centrifugation, resuspended in 100 l of SEM buffer, and proteinase K-treated as described above. Mitochondrial proteins were separated by SDS-PAGE and subsequently subjected to digital autoradiography or immunodecoration following Western blotting.
Miscellaneous-Standard protocols were used for BN-PAGE, SDS-PAGE, and Western blotting. Immune complexes were detected by enhanced chemiluminescence (Amersham Biosciences). Digital autoradiography using PhosphorImager technology (Amersham Biosciences) was used to analyze radiolabeled proteins. Non-relevant gel lanes were removed digitally.

All Modules of DIC Translocate across the Mitochondrial
Outer Membrane-To investigate the location of targeting signals in carrier proteins that direct their translocation into the mitochondrial inner membrane, we made mutants of DIC containing one or two structural repeats, referred to as modules (Fig. 1A). Resistance to proteinase K added to the outside of mitochondria following import was used as a measure of precursor translocation across the outer membrane. Full-length and DIC mutants were efficiently synthesized in rabbit reticulocyte lysate in the presence of [ 35 S]methionine/cysteine (Fig.  1B, lane 1). Next, lysates were mixed with wild-type yeast mitochondria in the presence or absence of a ⌬ across the inner membrane. After incubation at 25°C, mitochondria were left untreated or treated with proteinase K, reisolated, and then analyzed by SDS-PAGE. Radioactive full-length and mutant DIC were found associated with mitochondria in the presence and absence of a ⌬ (Fig. 1B, lanes 2 and 3). Moreover, for both full-length and mutant DIC significant amounts were also recovered with protease-treated mitochondria both in the presence or absence of a ⌬ (Fig. 1B, lanes 4 and 5), suggesting that they were efficiently imported across the outer mitochondrial membrane. To ensure that reisolated radioactive signals represented translocated precursor and not just an intrinsic resistance of each mutant to proteinase K, control imports were performed. Following import of each DIC construct independently, mitochondria were solubilized with Triton X-100 and proteinase K-treated. Proteinase K-resistant precursor was recovered for constructs DIC I , DIC II , and DIC III (Fig. 1B, lanes 8  and 9), but not to the same level as obtained in intact mitochondria. Thus, all DIC constructs were able to translocate across the mitochondrial outer membrane both in the presence and absence of a ⌬.
Depletion of the ⌬, by addition of the potassium ionophore valinomycin, abolishes translocation of the precursor into the inner membrane. The precursor is arrested in transport at Stage III. Under these conditions the amount of precursor that is protected against proteinase K reflects the portion that has passed the outer membrane but is not yet inserted into the inner membrane. When a ⌬ is present, the precursor is also transported across the outer membrane to a protease-protected environment but is additionally able to insert into the inner membrane. To determine the efficiency of translocation across the outer membrane under both ⌬ conditions, a quantitation was performed. Surprisingly, all DIC mutants were transported across the outer membrane more efficiently than fulllength DIC, particularly under conditions of a full ⌬ (Fig. 1C, compare lane 1 with lanes 2-7). In agreement with published data (30), the import of full-length DIC into mitochondria in the absence of a ⌬ was only marginally reduced compared with fully energized mitochondria (Fig. 1C, compare lanes 1 and 8). However, the ⌬-dependent import of the DIC mutants was variable, with the most prominent effect observed for module III in which there was a reduction of ϳ50% when the ⌬ was fully dissipated as compared with full ⌬ (Fig. 1C, compare  lanes 4 and 11).
We concluded that all modules of DIC contain sufficient targeting information for import across the outer membrane into the intermembrane space in both fully energized or ⌬dissipated mitochondria. Whereas translocation across the outer membrane was only partially ⌬-dependent, module III contains a strong ⌬-responsive element. The ability of each individual DIC module or module combination to direct insertion of carrier proteins into the inner membrane via the TIM22 complex under conditions of variable ⌬ could thus be examined in intact mitochondria.
DIC Modules Form Stage IV Intermediates-Recent work has raised doubt whether individual carrier modules follow the carrier pathway by engaging with the TIM22 translocase once they have passed through the outer membrane (32). To deter-mine whether any single or combined modules of DIC contain sufficient targeting information to drive the late stages of carrier import (from III to V), their ability to form a Stage IV intermediate was investigated. We recently showed that dur-  2 and 3); the second set was treated with proteinase K (lanes 4 and 5). As a control, radiolabeled precursors were imported into fully energized mitochondria, solubilized with 0.5% Triton X-100, and either left untreated (lanes 6 and 7) or proteinase K-treated (lanes 8 and 9). Reisolated mitochondria and radiolabeled precursors (10% of added lysate, lane 1) were subjected to SDS-PAGE and analyzed by digital autoradiography. Bands marked with an asterisk most likely represent truncated products of DIC arising from internal initiations of translation. C, quantitation of mitochondrial import. Import reactions were performed as described for panel B. Quantitation of the digital autoradiograms from proteinase K-treated mitochondria was performed with Image-Quant 1.2 (Amersham Biosciences). The intrinsic proteinase K resistance of constructs DIC I , DIC II , and DIC III was subtracted from import signals. Bars indicate the S.E. of the means.
ing import full-length DIC binds to the TIM22 translocation machinery, forming a high molecular mass complex that is discernable on BN-PAGE. This Stage IV complex ( Fig. 2A) represents a productive translocation intermediate and accumulates with greatest efficiency under conditions of low ⌬ that can be generated in organello by the addition of the protonophore CCCP (8). We predicted that because all DIC con-structs cross the outer membrane (Fig. 1, B and C) and are thus accessible for binding to the TIM22 complex the DIC module or modules containing the TIM22 complex binding elements should be readily identified. The six DIC constructs plus fulllength DIC (control) were imported into wild-type mitochondria in the presence of 30 M CCCP. Mitochondria were proteinase K-treated to remove unimported surface-exposed precursors, reisolated, solubilized in digitonin, and then analyzed by BN-PAGE (Fig. 2B). As expected, full-length DIC formed the high molecular mass Stage IV intermediate, as well as the characteristic low molecular mass Stage III intermediate (which dissociates from the TOM complex during the BN-PAGE run) (43) and fully imported dimeric DIC (Stage V) (Fig.  2B, lane 1). The most striking observation was that all DIC mutants containing module III formed a Stage IV intermediate (Fig. 2B, lanes 4, 6, and 7), whereas modules I and II alone failed to form a visible Stage IV complex (Fig. 2B, lanes 2 and  3). Interestingly, DIC modules I and II combined formed a stable Stage IV intermediate (Fig. 2B, lane 5).
Targeting Determinants for Stage IV Formation Predominantly Reside in Module III of DIC-We have shown previously that the full-length DIC precursor binds to the TIM22 complex tightly enough to survive isolation of the complex from mitochondria and subsequent BN-PAGE analysis. Thus, to investigate whether or not the high molecular mass complexes observed on BN-PAGE (Fig. 2B) represented an association of carrier mutants with the TIM22 complex as opposed to nonspecific associations, we isolated the complex following import of the DIC mutants into mitochondria under conditions of variable membrane potential. Radiolabeled precursors of DIC mutants were accumulated in mitochondria carrying Protein Atagged Tim18 in the presence (full and intermediate levels) and absence of a ⌬. Following proteinase K treatment to remove unimported precursor, the TIM22 complex was purified by affinity chromatography on IgG-Sepharose from reisolated mitochondria solubilized in digitonin buffer. Following cleavage of Tim18 from the Protein A tag, the isolated complex was analyzed by BN-PAGE. Indeed, the high molecular mass complexes observed on BN-PAGE (Fig. 2B) represented Stage IV intermediates as the isolated TIM22 complex contained the radiolabeled DIC mutant precursors (Fig. 3A, lanes 13-28). These data clearly show that DIC mutants containing module III (DIC III , DIC II,III , and DIC I,III ) as well as DIC I,II associate with the TIM22 complex. Interestingly, even with a low background noise characteristic of an isolated protein complex on BN-PAGE, there was no evidence of an interaction between DIC I or DIC II with the TIM22 complex (Fig. 3A, lanes 5-12).
A quantitation was performed to investigate the relative efficiency of Stage IV formation of each DIC mutant under different ⌬ conditions. Bearing in mind that the mutant DIC precursors had variable competency for crossing the outer membrane (Fig. 1C), we developed an experimental strategy (Fig. 3B, upper panel) to carefully determine the relative efficiency of Stage IV formation compared with the amount of carrier precursor that translocated across the outer membrane (Import control). In addition, because some of the DIC mutants may have dissociated from the TIM22 complex during the BN-PAGE run (Fig. 3A, low molecular mass signals, lanes [13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28], all quantitations were taken from SDS-PAGE analysis of isolated Stage IV DIC intermediates (Fig. 3B, upper panel). Consistent with our previous findings, full-length DIC formed Stage IV intermediates in a ⌬-responsive manner (Fig. 3B,  lanes 1-4), forming the previously defined "tethered" (no ⌬) and "docked" states (intermediate ⌬) (8 A full ⌬ promotes efficient membrane insertion and assembly of the carrier precursor into a dimeric form (Stage V). This general scheme is based on data drawn from import studies on both AAC and DIC (8, 9, 11, 13, 21-23, 28, 30). B, some DIC constructs form a high molecular complex on BN-PAGE equivalent to a Stage IV intermediate. 35 Slabeled carrier constructs were imported into wild-type mitochondria (50 g of protein) in the presence of 30 M CCCP, treated with proteinase K, and solubilized in 1% digitonin buffer prior to analysis by BN-PAGE. Asterisk denotes nonspecific complexes.
all the single module DIC mutants, DIC III bound to the TIM22 complex with a capacity close to the level of full-length (Fig. 3B,  lanes 13-16), whereas the structural equivalents DIC I and DIC II did not bind at all (Fig. 3B, lanes 5-12), consistent with the results described above (Figs. 2B and 3A). The binding pattern in response to ⌬ was, however, different for DIC III compared with full-length as maximum association with the TIM22 complex occurred in the absence of a ⌬ (Fig. 3B, lane  16) rather than at an intermediate level. Interestingly, when modules I and II were combined a propensity to bind the TIM22 complex was reestablished (Fig. 3B, lanes 17-20). Although DIC II,III could also bind the TIM22 complex, its ability to bind in the absence of ⌬ was lower than for DIC III . In addition, the pattern of binding in relation to the ⌬ indicated that further translocation from Stage IV to V was seemingly retarded by the presence of module II because under conditions of full ⌬ a significant portion of the carrier mutant remained bound to the TIM22 complex (Fig. 3B, lane 25). The pattern of binding in response to ⌬ of the artificial carrier construct DIC I,III resembled that of full-length DIC but with much reduced efficiency at intermediate levels of ⌬ (Fig. 3B, lanes 21-24).
These data collectively provide strong evidence that the major targeting elements responsible for driving import from Stage III to IV are contained within module III of DIC. This targeting element responds to the ⌬ in a manner similar to full-length DIC. The inability of DIC modules I and II to interact with the TIM22 complex can, to some degree, be overcome when these modules are combined.

FIG. 3. A single module of DIC contains sufficient targeting information for TIM22 complex association.
A, specific association of DIC modules with the TIM22 complex. 35 S-labeled DIC carrier constructs were imported into Tim18 ProtA mitochondria in the presence of either valinomycin (1 M) or CCCP (concentrations as indicated). Mitochondria were then treated with proteinase K, solubilized in 1% digitonin buffer, and subjected to affinity purification using IgG-Sepharose. Purified complexes were separated by BN-PAGE and associated radiolabeled DIC constructs visualized by digital autoradiography. B, quantitation of TIM22 complex binding by DIC constructs. Flow diagram (upper panel) indicating the experimental procedure used to assess the relative efficiency of TIM22 complex binding by DIC constructs (Stage IV isolation) compared with translocation across the outer membrane (Import control) and the corresponding quantitations (lower panel). The import reactions and isolation procedure were essentially as described for panel A, except that samples were separated by SDS-PAGE. Quantitations were performed with ImageQuant 1.2 (Amersham Biosciences). Bars indicate the S.E. of the means.
The ⌬-sensitive Module III of DIC Is Not Sufficient for Full Inner Membrane Carrier Insertion-Following binding to the TIM22 complex, a carrier protein inserts into the membrane in a ⌬-dependent manner where it assembles into its functional dimeric form. Because the amount of Stage IV intermediate that isolated with the TIM22 complex generally decreased with increasing ⌬, we speculated that the carrier precursors inserted into the inner membrane in a productive manner. Thus, to assess the ability of the DIC constructs to insert into the membrane, we investigated their protease sensitivity in swollen mitochondria (mitoplasts). When a full-length carrier precursor is membraneinserted it becomes largely resistant to proteinase K treatment in mitoplasts with just some clipping of the protein occurring. We used attainment of protease resistance in mitoplasts as a basis for analyzing ⌬-dependent membrane insertion.
Full-length and DIC mutants were synthesized in rabbit reticulocyte lysate in the presence of [ 35 S]methionine/cysteine and then mixed with wild-type yeast mitochondria in the presence (full and intermediate) and absence of a ⌬ across the inner membrane. Following incubation at 25°C, reisolated mitochondria were split. One half of the sample was subjected to osmotic swelling to form mitoplasts, and the other half was left untreated. The mitoplasts and control mitochondria were proteinase K-treated, reisolated, and analyzed by SDS-PAGE. In mitoplasts with a full ⌬ a clipped form of full-length DIC was evident, indicative of membrane-inserted carrier (Fig. 4A, lanes   5-7). At first glance it seemed apparent that all DIC mutants were at least partially inserted in the inner membrane (Fig. 4A, lanes [5][6][7][8]; this was unexpected, at least for DIC I and DIC II , which did not even reach Stage IV. A possible explanation for the observed protease resistance would be that mitoplasting was incomplete. However, this was not the case, as intermembrane space-exposed proteins Tim50 and Tim10 were fully degraded upon proteinase K treatment of swollen mitochondria, whereas control matrix-located Mge1 remained proteaseprotected (Fig. 4A, lanes 11 and 12). It was evident that some DIC constructs have resistance to proteinase K in Triton X-100treated mitochondria (DIC I , DIC II , and DIC III ) (Fig. 4A, lane  10) as do all DIC constructs in ⌬-depleted mitoplasts (Fig. 4A, lane 8) that represent nonspecific background signals, because a membrane potential is required for insertion of carrier proteins.
To obtain a true assessment of ⌬-dependent membrane insertion for each of the DIC constructs, a quantitation was performed taking into account the nonspecific radioactive signals detected in the absence of ⌬. In addition, for each ⌬ variable the level of protein insertion was assessed relative to outer membrane-translocated precursor. As expected for fulllength DIC, a large fraction of the precursor inserted into the membrane under conditions of full and moderate ⌬ (Fig. 4B,  lanes 1 and 2). Surprisingly, although DIC III formed the Stage IV intermediate reasonably well compared with full-length DIC FIG. 4. Module III of DIC promotes inner membrane insertion. A, inner membrane insertion of DIC constructs. Isolated wild-type mitochondria were incubated with 35 S-labeled DIC constructs for 15 min at 25°C in the presence of either valinomycin (1 M) or CCCP (concentrations as indicated). After import, mitochondria were divided in half, reisolated, and resuspended in either SEM buffer (mitochondria) or MOPS buffer (mitoplasts). Both mitochondria (lanes 1-4) and mitoplasts (lanes 5-8) were treated with proteinase K and analyzed by SDS-PAGE and digital autoradiography. As a control, radiolabeled precursors were imported into fully energized mitochondria, solubilized with 0.5% Triton X-100, and either left untreated (lane 9) or treated with proteinase K (lane 10). Control immunodecoration with specific antiserum of proteinase K-treated mitochondria and mitoplasts indicates swelling was complete (lanes 11 and 12). B, quantitation of ⌬-dependent inner membrane insertion of DIC constructs. Import reactions were as described in the legend to panel A and under "Experimental Procedures." The yield of membrane-inserted precursor was assessed relative to outer membrane-translocated precursor. For quantitations, radioactive signals obtained for DIC constructs in the absence of ⌬ were subtracted from the signals obtained from energized mitochondria to obtain a true assessment of ⌬-dependent membrane insertion. Quantitations were performed with ImageQuant 1.2 (Amersham Biosciences). Bars indicate the S.E. of the means. (Fig. 3B), this did not translate to efficient membrane insertion (Fig. 4B, lanes 10 and 11). Inefficient membrane insertion was also observed for DIC I,II (Fig. 4B, lanes 13 and 14). If, however, an additional module were combined with module III, membrane insertion was significantly improved (Fig. 4B, lanes 16 -21). In the case of DIC II,III it reached a level greater than that obtained for full-length (Fig. 4B, compare lanes 1-3 with 19 -21). For all mutant precursors that contained module III, membrane insertion increased with the strength of ⌬.
In summary, the translocation of DIC from Stage IV to V as a minimum requires ⌬-sensitive import signals contained within module III of the protein. The addition of a second structural repeat to module III, however, improves membrane insertion ϳ2-3-fold under conditions of full membrane potential. DISCUSSION To address the controversy regarding the location of targeting signals that direct inner membrane insertion of carriers, we used the knowledge that full-length DIC binds tightly to the TIM22 complex as a means to investigate which region or regions of DIC contain the targeting information necessary for recognition and stable binding to this inner membrane translocase. Our results clearly showed that the major targeting information that mediates translocation via the carrier pathway is contained within module III of DIC, as all mutants containing this module (DIC III , DIC I,III , and DIC II,III ) associated with the TIM22 complex. Module III bound with greatest efficiency in the absence of ⌬. This may reflect a strong ability of this module to bind a component of the TIM22 complex. Individual modules I or II did not associate with the TIM22 complex under any condition investigated, clearly indicating that alone they do not contain sufficient targeting information for inner membrane translocation even though all DIC modules were shown to transport across the outer membrane. On the other hand, the double module mutant DIC I,II retained an ability to form a Stage IV complex, indicating that an additional targeting element may be contained within the joining loop of modules I and II or, alternatively, that signals that are separated in the linear sequence of DIC come together to form a TIM22 complex-specific targeting signal.
Although module III of DIC contained sufficient targeting information for binding the TIM22 complex under the full range of ⌬ conditions examined, the single module alone did not lead to a very efficient or perhaps stable membrane insertion. If, however, a second module were attached to module III, i.e. the natural module II or the artificially attached module I, then insertion was 2-3-fold better with a full membrane potential. It is possible that modules I or II of DIC alone may not bind strongly enough to the TIM22 complex to withstand the isolation procedure employed and therefore could actually contain targeting information that is not detected using our method. However, determination of membrane insertion of DIC I and DIC II revealed that these mutant constructs failed to incorporate into the membrane, supporting our belief that alone they do not contain sufficient targeting determinants to drive the late stages of import. In contrast DIC I,II retained an ability to insert into the membrane, albeit inefficiently. Thus, although module III of DIC may not exclusively direct the late stages of import, the targeting information contained within this module promotes the most productive import.
In summary, we have used a new approach to examine carrier translocation from Stage IV to V. Our data suggest that the dominant targeting signals for DIC are contained within module III. These results support the findings of Endres et al. (29), who report that module III of AAC directs insertion into the inner membrane in a strictly ⌬-dependent manner. Taken together, it appears that a single structural repeat or module predominately drives inner membrane translocation of carriers from Stage III to V via the TIM22 translocase. This is in contrast to outer membrane translocation to which each independent carrier module can contribute. It remains to be seen whether the major findings obtained in this study apply to all members of the carrier family.