Intramolecular Disulfide Bond of Tim22 Protein Maintains Integrity of the TIM22 Complex in the Mitochondrial Inner Membrane*

Background: Tim22 is a central component of the mitochondrial inner membrane protein insertion machinery TIM22 complex. Results: Lack of the disulfide bond of Tim22 destabilizes Tim22 and impairs substrate protein assembly. Conclusion: The disulfide bond of Tim22 has a role in stabilization of the TIM22 complex, which is important for the TIM22 protein assembly pathway. Significance: Tim40(Mia40)/Erv1-independent disulfide bond formation contributes to protein stability in mitochondria. Mitochondrial proteins require protein machineries called translocators in the outer and inner membranes for import into and sorting to their destination submitochondrial compartments. Among them, the TIM22 complex mediates insertion of polytopic membrane proteins into the inner membrane, and Tim22 constitutes its central insertion channel. Here we report that the conserved Cys residues of Tim22 form an intramolecular disulfide bond. By comparison of Tim22 Cys → Ser mutants with wild-type Tim22, we show that the disulfide bond of Tim22 stabilizes Tim22 especially at elevated temperature through interactions with Tim18, which are also important for the stability of the TIM22 complex. We also show that lack of the disulfide bond in Tim22 impairs the assembly of TIM22 pathway substrate proteins into the inner membrane especially when the TIM22 complex handles excess amounts of substrate proteins. Our findings provide a new insight into the mechanism of the maintenance of the structural and functional integrity of the TIM22 complex.

Mitochondrial proteins require protein machineries called translocators in the outer and inner membranes for import into and sorting to their destination submitochondrial compartments. Among them, the TIM22 complex mediates insertion of polytopic membrane proteins into the inner membrane, and Tim22 constitutes its central insertion channel. Here we report that the conserved Cys residues of Tim22 form an intramolecular disulfide bond. By comparison of Tim22 Cys 3 Ser mutants with wild-type Tim22, we show that the disulfide bond of Tim22 stabilizes Tim22 especially at elevated temperature through interactions with Tim18, which are also important for the stability of the TIM22 complex. We also show that lack of the disulfide bond in Tim22 impairs the assembly of TIM22 pathway substrate proteins into the inner membrane especially when the TIM22 complex handles excess amounts of substrate proteins. Our findings provide a new insight into the mechanism of the maintenance of the structural and functional integrity of the TIM22 complex.
Eukaryotic cells contain highly developed membrane structures called organelles. Each organelle consists of a characteristic set of proteins and lipids to exert its specialized functions. Mitochondria are essential organelles consisting of four compartments, the outer membrane (OM), 3 inner membrane (IM), intermembrane space (IMS), and matrix. Because nearly all the mitochondrial proteins are synthesized in the cytosol, they have to be imported into mitochondria, sorted to one of the four submitochondrial compartments, where they function, and attain their functional native conformation, which is often facilitated by assembly into the membrane or multiprotein complex. Previous studies have revealed that these processes are mediated by elaborate protein machineries called translocators in the OM and IM (1)(2)(3)(4). For instance in the OM, the TOM40 complex functions as a general entry gate for most mitochondrial proteins, and the TOB/SAM complex in the OM mediates assembly of a more specialized set of proteins, mainly ␤-barrel proteins, in the OM (5)(6)(7)(8).
In the IM, the TIM23 complex mediates translocation across or insertion into the IM for proteins with an N-terminal cleavable presequence, which contains a matrix-targeting signal, with the assistance of mitochondrial Hsp70-associated motor and chaperone (MMC) proteins in the IM and matrix (9,10). Another translocator complex in the IM, the TIM22 complex, mediates insertion of polytopic or multispanning membrane proteins into the IM with the assistance of soluble small Tim chaperones in the IMS (11). The TIM22 complex consists of membrane-embedded core components, Tim22, Tim54, Tim18, and Sdh3, and peripheral chaperones, Tim9, Tim10, and Tim12 (12)(13)(14)(15)(16)(17)(18)(19). Tim9, Tim10, and Tim12 form a complex when associated with the TIM22 core complex, whereas Tim9 and Tim10 form a soluble heterohexamer in the IMS (20). Tim22 is a central component of the TIM22 complex and forms an insertion channel, which probably accommodates loop structures of transmembrane segments of substrate polytopic membrane proteins (21). Tim18 and Sdh3, which are homologous with each other, are involved in the assembly of the TIM22 complex on their own (19). Tim54 appears to be an adaptor protein that bridges the TIM22 core complex and small Tim proteins through direct interactions with Tim10 (22,23).
Tim40/Mia40 and Erv1 constitute a disulfide relay system in the IMS, thereby facilitating oxidative folding and import of IMS proteins, which generally contain twin Cys motifs such as CX 3 C or CX 9 C (24 -28). Tim40 transiently binds to the incoming substrate proteins in the IMS to form a mixed disulfide intermediate. Subsequently, Erv1, an FAD-containing sulfhydryl oxidase, oxidizes Tim40 to release substrate proteins to the IMS with transfer of a disulfide bond into substrate proteins (24 -28).
Although previous studies have revealed a number of IMS proteins or protein domains that contain disulfide bond(s), which are likely transferred from Tim40, there are still many proteins or protein domains in the IMS that contain Cys residues whose formation of disulfide bonds remains to be experimentally confirmed (38,39). Here we find that the conserved Cys residues in Tim22 form an intramolecular disulfide bond. Genetic and biochemical analyses reveal that the disulfide bond of Tim22 stabilizes Tim22 through its interactions with Tim18, and this stabilization is important for the stability and functions of the TIM22 complex. Our findings provide a new insight into the mechanism of the maintenance of the structural and functional integrity of the TIM22 complex.
4-Acetamido-4Ј-maleimidylstilbene-2,2Ј-Disulfonic Acid (AMS) Modification-Yeast cells were cultivated in SCD (ϪTrp) to mid-log phase and harvested by centrifugation. The cells were incubated in 1 ml of 5% TCA (trichloroacetic acid) on ice for 10 min and centrifuged at 5,800 ϫ g for 5 min. The pellets were vortexed in 100 l of 5% TCA containing 100 l of glass beads for 30 s and diluted with 900 l of 5% TCA. After removal of glass beads, proteins were precipitated by centrifugation at 13,000 ϫ g for 5 min at 4°C and washed once with ice-cold acetone. The precipitated proteins were suspended in buffer (1% SDS, 50 mM Tris-HCl, pH 7.5) with or without 30 mM dithiothreitol (DTT) and incubated for 30 min at 30°C. Proteins were precipitated again with TCA and incubated in AMS-containing buffer (1% SDS, 50 mM Tris-HCl, pH 7.5, 15 mM AMS) for 15 min at 37°C.
In Vitro Mitochondrial Protein Import-Mitochondria were suspended in import buffer (250 mM sucrose, 10 mM MOPS-KOH, pH 7.2, 80 mM KCl, 2 mM ATP, 20 mM NADH, 12 mM creatine phosphate, 120 g/ml creatine kinase, 2 mM methionine, 5 mM MgCl 2 , 5 mM DTT, 2.5 mM potassium P i , pH 7.4, 1% BSA) and preincubated for 2 min at 25°C. 35 S-labeled substrate proteins (wild-type and mutant Tim22 and Tim23), which were synthesized with homemade reticulocyte lysate or unlabeled Tim23-FLAG, which was synthesized with wheat germ extracts (BioSieg, Inc.), were then added to the mitochondrial suspension and incubated at 25°C unless otherwise stated. After stopping the import reaction by the addition of 10 g/ml valinomycin, mitochondria were treated with 50 g/ml proteinase K for 10 min on ice to digest substrate proteins outside mitochondria.
Cycloheximide Chase-Yeast cells were cultivated in SCD (ϪTrp) at 30°C to mid-log phase, and protein synthesis was stopped by the addition of 200 g/ml cycloheximide. The cells were further cultivated, and whole cell extracts were prepared (32) after the indicated periods of time. The whole cell extracts were analyzed by SDS-PAGE and immunoblotting.

Tim22 Forms an Intramolecular Disulfide Bond-Evidence
has accumulated that a number of proteins form disulfide bond(s) in the mitochondrial IMS, usually with the aid of the Erv1-Tim40 disulfide bond relay system (23)(24)(25)(26)(27). Tim22 is a central channel-forming component of the TIM22 complex, the insertion machinery for polytopic IM proteins. Because Tim22 has two Cys residues that are well conserved among different organisms (Fig. 1A), we wondered whether Tim22 forms an intramolecular disulfide bond in the IMS. We thus compared the migration rates of Tim22 on SDS-PAGE gels between reducing (in the presence of ␤-mercaptoethanol (␤-ME)) and nonreducing (in the absence of ␤-ME) conditions followed by immunoblotting with anti-Tim22 antibodies (Fig.  1B). Tim22 migrated faster under the nonreducing condition than the reducing condition, suggesting that Tim22 forms a disulfide bond. To confirm this further, we tested the presence of a free thiol group by incubating cell extracts with a thioltrapping reagent, AMS and detected Tim22 by immunoblotting. AMS can be covalently attached to a free thiol group in a protein, which accompanies a 500-Da increase in the size of the modified protein. The size of Tim22 increased upon AMS treatments only when proteins were pretreated with DTT (Fig. 1C, DTTϩ, AMSϩ), but it did not increase without pretreatment with DTT (Fig. 1C, DTTϪ, AMSϩ). As a control, we analyzed AMS modification of another IM protein, Tim23, which is a Tim22 homolog and contains three Cys residues. Upon AMS treatment, Tim23 was modified even without DTT pretreatment, and the migration rates of AMS-modified Tim23 with or without DTT pretreatment were the same. These results show that Tim22, but not Tim23, forms a disulfide bond and does not contain free thiol group in mitochondria.
We next constructed yeast strains containing Tim22 mutants with one or two Cys 3 Ser replacement(s) (C42S, C141S, or C42/141S). The single Cys 3 Ser mutants, Tim22-C42S and Tim22-C141S, and the double Cys 3 Ser mutant, Tim22-C42/141S, all exhibited similar migration rates, under both nonreducing and reducing conditions, eliminating the possibility that two Cys residues independently contribute to a disulfide bond formation or that Tim22 forms a disulfidelinked dimer (Fig. 1D). To rule out the trivial possibility that Tim22 forms a disulfide bond only after solubilization of mitochondria, we preincubated mitochondria with different concentrations of DTT on ice or at 30°C, washed off DTT, and then solubilized the mitochondria under the nonreducing condition. The proteins were analyzed by SDS-PAGE in the absence of ␤-ME followed by immunoblotting with antibodies against Tim22. Tim22 was detected as a reduced form after pretreatment with Ն30 mM (30°C) or Ն90 mM (0°C) DTT (Fig. 1E). Taken together, we conclude that Tim22 forms an intramolecular disulfide bond in mitochondria.
Effects of Cys 3 Ser Mutations on Assembly of Tim22 into the TIM22 Complex-We next asked what the roles of the disulfide bond of Tim22 are. We first examined whether lack of the disulfide bond affects the assembly of Tim22 into the TIM22 complex. We synthesized radiolabeled wild-type Tim22 and its Cys 3 Ser mutants (C42S, C141S, and C42/141S) in vitro and imported them into mitochondria isolated from wild-type cells. As reported previously, imported wild-type Tim22 formed a Tim22 dimer and was subsequently assembled into the 300-kDa TIM22 complex in a manner dependent on the membrane potential across the IM (⌬⌿) (23) (Fig. 2A). In contrast, the Tim22 mutants, Tim22-C42S, Tim22-C141S, and Tim22-C42/ 141S, hardly formed a dimer and were not assembled into the 300-kDa TIM22 complex efficiently ( Fig. 2A). The amounts of Tim22 assembled into the TIM22 complex were normalized by those of imported Tim22 (SDS-PAGE) and plotted against time (Fig. 2C). These results indicate that both Cys-42 and Cys-141 are important for proper Tim22 assembly into the TIM22 complex through formation of the dimer.
We then compared the in vitro assembly of radiolabeled wild-type Tim22 into the TIM22 complex in mitochondria isolated from the wild-type, tim22-C42S, tim22-C141S, and tim22-C42/141S strains. Interestingly, assembly of wild-type Tim22 into the TIM22 complex was accelerated in tim22-C42S, tim22-C141S, and tim22-C42/141S mitochondria as compared with that in wild-type mitochondria (Fig. 2, B and D). We also noticed that the Tim22 dimer was hardly detected during the Tim22 assembly in tim22-C42S, tim22-C141S, and tim22-C42/ Identical residues and similar residues are marked with asterisks and semicolons/periods, respectively. The conserved cysteine residues are highlighted in black. B, whole cell extracts were prepared from wild-type yeast cells under reducing (ϩ␤-ME) or nonreducing (Ϫ␤-ME) condition and analyzed by SDS-PAGE followed by immunoblotting. C, whole cell extracts from wild-type yeast cells were treated with AMS as described under "Experimental Procedures," and proteins were analyzed by SDS-PAGE and immunoblotting with anti-Tim22 and anti-Tim23 antibodies. D, proteins in mitochondria isolated from wild-type and tim22 Cys 3 Ser mutant cells were analyzed by SDS-PAGE under reducing (ϩ␤-ME) or nonreducing (Ϫ␤-ME) condition followed by immunoblotting with anti-Tim22 antibodies. E, mitochondria isolated from wild-type yeast cells were incubated with the indicated concentrations of DTT on ice or at 30°C for 30 min. After washing with SEM buffer, proteins were analyzed by SDS-PAGE under nonreducing conditions and immunoblotting with anti-Tim22 antibodies.   141S mitochondria (Fig. 2B), probably due to rapid consumption of the Tim22 dimer intermediate for uptake by the TIM22 complex (23). These results suggest that the lack of the disulfide bond in Tim22 alters the quaternary structures of the TIM22 complex in such a way that newly imported wild-type Tim22 can exchange with pre-existing mutant Tim22 much faster than in wild-type mitochondria.

Conserved Cys Residues of Tim22
Are Important for Stabilization of the TIM22 Complex-It was reported previously that destabilization of the TIM22 complex by mutations in each of the subunits facilitates integration of newly imported, corresponding subunits into the TIM22 complex (23). The above results of the in vitro import of Tim22 into wild-type and tim22 mutant mitochondria suggest that the quaternary structure of the TIM22 complex is relaxed in the absence of the disulfide bond in Tim22, so that the exchange between the newly imported Tim22 and pre-existing Tim22 in the TIM22 complex is facilitated. If this is the case, wild-type Tim22 with a disulfide bond may have a higher affinity for other subunits of the TIM22 complex, Tim54 and Tim18, rather than the Tim22 Cys 3 Ser mutants lacking the disulfide bond. We thus com-pared the steady-state levels of the subunits of the TIM22 complex between wild-type and tim22 mutant mitochondria at 30 and 37°C. Protein levels of the subunits of the TIM22 complex (Tim22, Tim54, and Tim18), the TIM23 complex (Tim50, Tim23, and Tim17), the TOM40 complex (Tom70 and Tom22), matrix chaperones (Hsp60 and Mdj1), and carrier proteins (AAC and PIC) are all similar between wild-type and tim22 Cys 3 Ser mutant mitochondria at 30°C (Fig. 3A). However, the steady-state levels of the Tim22 mutants (Tim22-C42S, Tim22-C141S, and Tim22-C42/141S) decreased when yeast cells were cultured at elevated temperature (37°C) for 4 h before the isolation of mitochondria (Fig. 3B).
We next analyzed the levels of the 300-kDa TIM22 complex by blue-native PAGE (BN-PAGE) after solubilization of mitochondria with 1% digitonin. The apparent sizes of the TIM22 complex detected with anti-Tim22 antibodies were slightly smaller in tim22 Cys 3 Ser mutant mitochondria than in wildtype mitochondria (Fig. 3C). When anti-Tim54 antibodies were used for detection of the TIM22 complex, additional smaller complexes were observed for tim22 Cys 3 Ser mutant mitochondria (Fig. 3C, asterisks). As negative controls, the amounts of the TOM40 complex in the OM and the TIM23 complex in the IM were comparable between wild-type and tim22 Cys 3 Ser mutant mitochondria. The amounts of the TIM22 complex detected with antibodies against Tim22, Tim18, and Tim54 also decreased after treatment at 37°C (Fig. 3D). These results indicate that C42S, C141S, and C42/141S mutations in Tim22 destabilize the TIM22 complex at 30°C, and lead to decreased levels of the 300-kDa TIM22 complex after treatment at 37°C. This is consistent with the idea that Cys 3 Ser mutations or most likely the lack of the conserved disulfide bond in Tim22 cause(s) destabilization of the TIM22 complex, which instead facilitates exchange of pre-existing Tim22 with newly imported Tim22. Disulfide Bond of Tim22 Is Important for Tim22-Tim18 Interactions-We further asked which interactions among the subunits of the TIM22 complex were destabilized by the Tim22 Cys 3 Ser mutations. We isolated mitochondria from wildtype and tim22 Cys 3 Ser mutant cells expressing C-terminally FLAG-tagged Tim18 instead of authentic Tim18. The mitochondria were solubilized with digitonin and subjected to coimmunoprecipitation using anti-FLAG agarose. The levels of Tim22 and Tim54 co-immunoprecipitated with Tim18-FLAG were comparable between wild-type and tim22 Cys 3 Ser mutant mitochondria (Fig. 4A). Tom70 was used as a negative control (Fig. 4A). However, when mitochondria were preincubated for 10 min at 37°C before solubilization, the amounts of Tim22 and Tim54 co-immunoprecipitated with Tim18-FLAG were significantly reduced (Fig. 4B). The reduction of the coimmunoprecipitated proteins may simply reflect the decreased amounts of total Tim22 Cys 3 Ser mutants after heat treatment (Fig. 4B, Total). To distinguish the effects of the absence of the disulfide bond on possible decreased intersubunit inter-  actions from the decreased amounts of Tim22, we overexpressed wild-type or mutant Tim22 proteins from multicopy plasmids and performed similar co-immunoprecipitation experiments with or without preincubation at 37°C. Now due to overexpression, total amounts of wild-type and mutant Tim22 are comparable between wild-type and tim22 Cys 3 Ser mitochondria even after heat treatment at 37°C (Fig. 4, C and  D, Total). Nevertheless, the amount of Tim22 co-immunoprecipitated with Tim18-FLAG was still significantly reduced after heat treatment for tim22 Cys 3 Ser mutant mitochondria, although reduction of co-immunoprecipitated Tim54 is marginal (Fig. 4D). These results indicate that the disulfide bond of Tim22 is important for Tim22-Tim18 interactions in the TIM22 complex at elevated temperature.
Tim22 Is Susceptible to Degradation due to the Loss of the Disulfide Bond-When wild-type Tim22 was overexpressed, the reduced form of Tim22 was observed, yet it was gone after heat treatment at 37°C (Fig. 4, C and D). Therefore, the reduced form of Tim22 is likely unstable and degraded efficiently at elevated temperature. We next performed cycloheximide chase experiments to compare the stability or degradation of wildtype and mutant Tim22 as well as Tim18 and Tim54 in the TIM22 complex. Briefly, we stopped protein synthesis in wildtype and tim22 Cys 3 Ser mutant cells by the addition of a ribosomal inhibitor, cycloheximide. Then, whole cell extracts were prepared from the cells cultivated for different time periods (1-3 h) at 30°C or 37°C after translation inhibition to follow the fate of each protein. Immunoblotting of Tim22 showed that Tim22 Cys 3 Ser mutants were rapidly degraded as compared with wild-type Tim22, and rapid degradation was more prominent at 37°C than at 30°C (Fig. 5). We also observed accelerated degradation of Tim18 at 37°C in tim22 Cys 3 Ser mutant cells. However, degradation rates of another subunit of the TIM22 complex, Tim54, and a control protein, Tom70, did not significantly differ between wild-type and tim22 Cys 3 Ser mutant cells (Fig. 5). These results clearly indicate that the disulfide bond plays an important role in stabilizing Tim22 especially at elevated temperature.
Protein Assembly via the TIM22 Complex Is Impaired in tim22 Cys 3 Ser Mutant Mitochondria-We asked whether the disulfide bond of Tim22 plays important roles in protein import and assembly via the TIM22 complex. We thus performed in vitro import assays using an IM protein, Tim23, as a substrate for insertion into the IM via the TIM22 complex. Radiolabeled wild-type Tim23 was incubated with mitochondria isolated from wild-type and tim22 Cys 3 Ser mutant cells at 37°C. The samples were separated into halves. One half was treated with proteinase K to digest Tim23 outside the mito-FIGURE 6. Import and assembly of Tim23 are impaired in tim22 Cys 3 Ser mutant mitochondria. A, 35 S-labeled Tim23 was imported into wild-type and tim22 Cys 3 Ser mutant mitochondria at 37°C for the indicated times. Imported and assembled Tim23 were analyzed by SDS-PAGE (uppermost panels) and BN-PAGE (central panel) followed by radioimaging. Imported proteins (lowermost left panel) and assembled proteins (lowermost right panel) were quantified. The amounts of the longest time point for wild-type mitochondria are set to 100%. B, C-terminally FLAG-tagged Tim23 was synthesized with wheat germ extract and imported into wild-type and tim22 Cys 3 Ser mutant mitochondria at 25°C for the indicated times. To assess the amounts of imported Tim23, the mitochondria were treated with proteinase K. Imported Tim23-FLAG was analyzed by SDS-PAGE followed by immunoblotting with the anti-FLAG antibody. 10% cont. indicates 10% of protein used for the in vitro import assay. Lower panel, imported Tim23-FLAG was quantified. The amount of the longest time point for wild-type mitochondria is set to 100%. chondria and analyzed by SDS-PAGE to assess the amounts of Tim23 imported into mitochondria; the other half was solubilized with digitonin and analyzed by BN-PAGE to monitor its assembly into the TIM23 complex. When analyzed by BN-PAGE, imported Tim23 was assembled into the 100-kDa TIM23 core complex consisting of Tim23 and Tim17 efficiently as well as the TIM23 holo-complex containing Tim21 and/or Tim50 with higher molecular weights (Fig. 6A). Assembly of imported Tim23 into the TIM23 complex was less efficient in tim22 Cys 3 Ser mutant mitochondria. Although import of radiolabeled Tim23 into mitochondria was only slightly affected by tim22 Cys 3 Ser mutation (Fig. 6A), this defect became enhanced when we used excess amounts of C-terminally FLAG-tagged Tim23, which was synthesized in vitro with wheat germ extracts, for in vitro import assays (Fig. 6B). Therefore, the disulfide bond of Tim22 is required for efficient import and assembly of multispanning IM proteins such as Tim23 via the TIM22 complex.
Overexpression of Carrier Proteins Compromises Growths of tim22 Cys 3 Ser Mutant Cells-Our in vitro experiments suggest that the disulfide bond of Tim22 is important for protein import and assembly through the TIM22 pathway. Because the TIM22 protein import pathway is essential for mitochondrial biogenesis, hence yeast cell growth, we wondered whether Cys 3 Ser mutations of Tim22 affect cell growth. Because defects of the TIM22 pathway caused by the Cys 3 Ser mutations of Tim22 are more prominent when excess amounts of substrate proteins are imported (Fig. 6B), we tested the effects of overexpression of metabolite carrier proteins such as AAC, PIC, and DIC, which are inserted into the IM via the TIM22 complex, on the growths of tim22 Cys 3 Ser mutant cells. Although tim22 Cys 3 Ser mutant cells grew normally on fermentable and nonfermentable media at all temperatures we tested as compared with wild-type cells (Fig. 7A), overexpression of AAC, PIC, or DIC exacerbated the growths of tim22 Cys 3 Ser mutant cells more significantly than that of wild-type cells (Fig. 7B). In contrast, overexpression of an OM protein, Tom22, or a matrix-targeted protein, pb 2 (167)⌬19-DHFR, a fusion protein consisting of the matrix-targeting signal derived from cytochrome b 2 followed by mouse dihydrofolate reductase (DHFR), did not affect the growths of tim22 Cys 3 Ser mutant cells at all (Fig. 7B). These results support our idea that the disulfide bond of Tim22 is functionally important especially when the TIM22 complex handles excess amounts of its substrate proteins.

DISCUSSION
In this study, we found that two conserved Cys residues of Tim22 form an intramolecular disulfide bond, which is important for stability and functions of the TIM22 complex. To assess the roles of the disulfide bond in Tim22, we constructed three tim22 Cys 3 Ser mutant cells, in which one or both conserved Cys residues are replaced with Ser. Lack of the disulfide bond in Tim22 renders interactions between Tim22 and Tim18 weak in the TIM22 complex especially at elevated temperature. These weakened Tim22-Tim18 interactions apparently lead to rapid exchange of Tim22 assembled in the TIM22 complex with a pool of free Tim22 (e.g. newly imported Tim22), to rapid deg-radation of the Tim22 mutants in the IM, and to destabilization of the TIM22 complex. Besides, the TIM22 complex with destabilized Tim22 mutants is defective in its function in assembly of polytopic IM proteins in the IM, especially when it has to handle excess amounts of substrate proteins, which is reflected in the impaired cell growth upon overexpression of carrier proteins.
During preparation of this manuscript, Chacinska and colleagues (34) reported that Tim22 forms a disulfide bond and that Tim22 Cys 3 Ser mutants fail in assembly into the TIM22 complex in vitro, which is essentially the same as we observed here. They also showed that Tim40/Mia40 drives import of Tim22 into the IMS through formation of the mixed disulfide intermediates between Tim22 and Tim40, although Tim22 does not contain a typical Cys motif for the Tim40 pathway such as CX 3 C or CX 9 C. Interestingly, they reported that the redox state of Tim22 is not affected in tim40-F311E and erv1-2 cells, suggesting that the Tim40-Erv1 system is not required for oxidation of Tim22 (34). Consistently, we observed normal disulfide bond formation of Tim22 even when Tim40 or Erv1 was depleted by GAL7 promoter shut-off (35,36), although the steady-state level of Tim22 somehow decreased after depletion of Tim40 (data not shown). We also confirmed by in vitro import assays that newly imported Tim22 is mainly present as an oxidized form even in Tim40-depleted or Erv1-depleted mitochondria (data not shown).
Then what facilitates disulfide formation of newly imported Tim22 in mitochondria? Chacinska and colleagues (34) showed that ⌬⌿ is essential for the disulfide bond formation of Tim22 and that the oxidized form of newly imported Tim22 preferentially interacts with Tim18. These results suggest that the disulfide bond formation occurs after assembly of Tim22 into the TIM22 complex. Interestingly, when we overexpressed wildtype Tim22, not only oxidized Tim22 but also reduced Tim22 was detected in isolated mitochondria (Fig. 4C). This may suggest that excess Tim22 molecules that overflowed from the TIM22 complex may not form a disulfide bond. Supporting this idea, our co-immunoprecipitation experiments revealed stronger interactions of oxidized Tim22 with Tim18 than reduced Tim22 (Fig. 4C). In addition, the reduced form of Tim22 was unstable because it is rapidly degraded after heat treatment of mitochondria overexpressing Tim22 (Fig. 4D). On the basis of these observations, we propose that Tim22 spontaneously forms a disulfide bond between the conserved Cys residues after proper folding, which is achieved by correct assembly into the TIM22 complex and brings the Cys residues in close proximity (Fig. 8). In other words, even wild-type Tim22 cannot fold properly into its native conformation until it interacts with its proper binding partner, Tim18, in the TIM22 complex, and such incompletely folded Tim22, which cannot form a disulfide bond, becomes prominent when Tim22 is overexpressed. Although Tim22 has been often assumed to have four hydrophobic transmembrane (TM) segments, precise membrane topology of Tim22 has not been experimentally determined so far. Possible TM segments are predicted by the TMpred program to be residues 50 -69, 81-99, 129 -146, and 174 -191, although the second predicted TM segment (residues 81-99) is less hydrophobic (37). We can assume, on the basis of the sequence similarity between Tim22 and Tim23, that the N terminus of Tim22 faces the IMS (12). If Tim22 contains four TM segments, Cys-41, which is likely present in the IMS, needs to be inserted into the IM to form a disulfide bond with Cys-141, which is localized near the matrix side in the third TM helix. Alternatively, if Tim22 contains only three TM segments, leaving the second predicted TM segment uninserted into the IM, the disulfide bond may be formed more easily because Cys-141 should be close to the IMS side in the third predicted TM helix, which could be more consistent with our finding of the disulfide bond between Cys-42 and Cys-141. Clearly, biochemical and structural analyses to determine the membrane topology of Tim22 are essential for further understanding of the functional details of the TIM22 complex in mitochondrial protein import.