The mitochondrial outer membrane contains two protein translocators: the TOM40 and TOB/SAM complexes. Mdm10 is distributed in the TOB complex for β‐barrel protein assembly and in the MMM1 complex for tethering of the endoplasmic reticulum and mitochondria. Here, we establish a system in which the Mdm10 level in the TOB complex—but not in the MMM1 complex—is altered to analyse its part in β‐barrel protein assembly. A decrease in the Mdm10 level results in accumulation of in vitro imported Tom40, which is a β‐barrel protein, at the level of the TOB complex. An increase in the Mdm10 level inhibits association not only of Tom40 but also of other β‐barrel proteins with the TOB complex. These results show that Mdm10 regulates the timing of release of unassembled Tom40 from the TOB complex, to facilitate its coordinated assembly into the TOM40 complex.
A class of mitochondrial outer membrane proteins, β‐barrel proteins take a specific route involving the TOM40 and TOB/SAM complexes for their membrane insertion in β‐barrel structures (Pfanner et al, 2004; Paschen et al, 2005; Endo & Yamano, 2009). Four β‐barrel membrane proteins—Tom40, porin, Tob55/Sam50 and Mdm10—have been identified so far in yeast mitochondria (Imai et al, 2008). Tom, Tob, Sam and Mas have been used to name the proteins that are involved in β‐barrel protein assembly, but the divergent nomenclature is not yet unified. Hereafter, we thus describe different names in parallel for the reader's convenience. Tom40 is a central subunit of the TOM40 complex, which is an outer membrane translocator, and forms a β‐barrel structure to function as an import channel. Assembly of the in vitro synthesized Tom40 precursor into the TOM40 complex in isolated mitochondria occurs through several steps (Model et al, 2001). Tom40 crosses the outer membrane through the TOM40 complex, associates with the TOB complex (Tob55/Sam50, Mas37/Sam37/Tom37 and Tom38/Sam35/Tob38) and is then assembled into the final TOM40 complex. The other three β‐barrel proteins—porin, Tob55/Sam50 and Mdm10—do not accumulate an intermediate with the TOB complex during their assembly, although they crucially require the TOB complex for their membrane insertion. Recent studies have shown that Mdm10 is also a subunit of the TOB complex, but, at the same time, constitutes the MMM1 complex tethering the endoplasmic reticulum (ER) and mitochondria to function in lipid biosynthesis (Meisinger et al, 2004, 2007; Kornmann et al, 2009). Interestingly, although Mdm10 is a subunit of the TOB complex, Mdm10 was shown to facilitate assembly of Tom40 after its dissociation from the TOB complex (Meisinger et al, 2004). How can a subunit of the assembly complex mediate the Tom40 assembly after its evacuation from the complex?
In this study, we reinvestigated the role of Mdm10 in the β‐barrel protein assembly by altering the protein level of Mdm10 in the TOB complex. We found that an increase in the level of Mdm10 blocks the association of all β‐barrel proteins with the TOB complex, whereas a decrease in the level of Mdm10 available for association with the TOB complex impairs the dissociation of Tom40, but not the other β‐barrel proteins, from the TOB complex. These results suggest that dynamic association of Mdm10 with the TOB complex promotes the early step, rather than the later step, of the assembly of Tom40 by regulating the timing of release of unassembled Tom40 from the TOB complex, facilitating the coordinated assembly of Tom40 into the final TOM40 complex.
Mdm10 assists exit of Tom40 from the TOB complex
Recent findings that the MMM1 tethering complex, which comprises Mdm10, Mdm12 and Mmm1, mediates lipid transfer between the ER and mitochondria (Kornmann et al, 2009) raises the possibility that deletion of the MDM10 gene would cause pleiotropic effects on various lipid‐dependent processes, which will hamper assessment of the precise role of Mdm10 in the β‐barrel protein assembly. In connection to this, we noticed that the growth phenotypes and protein levels of components of the TOM40 and TOB complexes change differently in varying parental backgrounds after disruption of the MDM10 gene (data not shown). This is probably because complete depletion of Mdm10 leads to significant reduction in the levels of cardiolipin and phosphatidylethanolamine (Fig 1A; Osman et al, 2009), which are important for the maintenance and assembly of the membrane protein complexes (Kutik et al, 2008; Tamura et al, 2009). We thus attempted to lower the Mdm10 protein level—which is about 3% of that of Tom40 in wild‐type cells (data not shown)—without changing the lipid levels of cardiolipin or phosphatidylethanolamine, or the protein levels of other TOM40 and TOB components, by mutagenizing the β‐signal of Mdm10. The β‐signal is a short carboxy‐terminal sequence that probably functions as a sorting signal of mitochondrial β‐barrel proteins (Kutik et al, 2008). We chose the mdm10‐KY mutant (K484A,Y491A), which did not show obvious growth defects (supplementary Fig S1A online) and contained normal levels of cardiolipin and phosphatidylethanolamine (Fig 1A). This is probably because the amount of Mdm10 associated with Mmm1 or Mdm12 was only affected mildly in KY mitochondria (described later), despite the total amount of Mdm10 being reduced significantly (Fig 1B). KY mitochondria contained normal levels of the TOM40 and TOB components (Fig 1B), which form the normal TOM40 and TOB complexes (supplementary Fig S1B,C online), although a partly dissociated TOM40 complex increased slightly in KY mitochondria (supplementary Fig S1B online, 100 kDa bands). KY mitochondria retained full import ability (supplementary Fig S1D online) and nearly normal mitochondrial morphology (supplementary Fig S2A,B online). We thus decided to use the KY mitochondria for further analyses of the effects of Mdm10 reduction on β‐barrel protein assembly.
Radiolabelled Tom40 and porin were synthesized in vitro and incubated with mitochondria isolated from wild‐type or KY cells. Mitochondria were then solubilized with digitonin and subjected to blue‐native polyacrylamide gel electrophoresis (BN‐PAGE) analyses to follow the assembly of Tom40 and porin in the outer membrane. Assembly of Tom40 in wild‐type mitochondria proceeds through two assembly intermediates before the final 450 kDa complex is formed: the 250 kDa assembly I intermediate involving the TOB complex and the less‐characterized 100 kDa assembly II intermediate (Fig 1C, left panel; Model et al, 2001; Ishikawa et al, 2004). In KY mitochondria, although Tom40 formed assembly I with the TOB complex normally, it stayed as assembly I for longer during incubation, resulting in reduced formation of assembly II and the final 450 kDa complex (Fig 1C). To confirm the inefficient dissociation of Tom40 from the TOB complex in KY mitochondria, we performed a two‐step assembly experiment for Tom40 (Fig 1D). Radiolabelled Tom40 was accumulated kinetically at the stage of assembly I by incubation for 5 min at 30°C. The mitochondria were reisolated and subjected to chase incubation at 25°C for various times, and the fate of assembly I was analysed by BN‐PAGE. Indeed, the rate of dissociation of Tom40 from the TOB complex (that is, the amount of assembly I) decreased for KY mitochondria as compared with wild‐type mitochondria at 25°C. The inefficient chase of Tom40 from the TOB complex led to retarded formation of assembly II and of the subsequent final TOM40 complex. However, efficiency of the assembly of porin (Fig 1C, right panel), Tob55/Sam50 or Mdm10 (supplementary Fig S1E online) in the outer membrane was not impaired and even increased slightly in KY mitochondria. These results suggest that, in contrast to the previous observation (Meisinger et al, 2004), Mdm10 is primarily involved in the early step of the Tom40 assembly pathway by promoting dissociation of Tom40 from the TOB complex, although we cannot rule out the possibility that the KY mutation also slightly affected the later step of the Tom40 assembly.
Increased Mdm10 blocks the substrate–TOB association
If Mdm10 mediates dissociation of Tom40 from the TOB complex, an increase in the level of Mdm10 will accelerate dissociation of Tom40 from the TOB complex. We thus constructed a yeast strain that allows the controlled expression of Mdm10 from the galactose‐inducible GAL1 promoter. Unexpectedly, those cells showed significant growth defects on galactose‐containing medium, which induced overexpression of Mdm10 (supplementary Fig S3A online). This is in contrast to the overexpression of components of the MMM1, TOB or TOM40 complexes, which caused no or only moderate growth defects (supplementary Fig S3A online).
We then isolated mitochondria from those cells 2 h (Mdm10↑) or 6 h (Mdm10↑↑) after the change from galactose‐free to galactose‐containing medium. The level of Mdm10 increased fourfold and more than tenfold in Mdm10↑ and Mdm10↑↑ mitochondria, respectively, whereas the levels of other components of the translocator complexes and β‐barrel proteins were not affected (Fig 2A; supplementary Fig S3B online), and normal lipid contents and nearly normal mitochondrial morphology were attained (supplementary Fig S4 online). We observed only slight accumulation of the precursor form of pHsp60(50)‐EGFP3 and pMdj1 in vivo, which probably reflected the decrease in level of Tom40 (supplementary Fig S3B online).
We analysed the effects of Mdm10 overexpression on the import and assembly of β‐barrel proteins in vitro (Fig 2B–D). We confirmed that overexpression of Mdm10 did not affect the import of precursors to the matrix and inner membrane, or the assembly of subunits of the TOM40 complex—except for Tom40—in the outer membrane (supplementary Fig S3C,D online); it had been proposed that assembly of Tom5, Tom6, Tom7 and Tom22 depended on the TOB complex (Stojanovski et al, 2007). When assembly of Tom40 was followed by BN‐PAGE, formation of assembly I was significantly impaired, resulting in a reduction of the final TOM40 complex formation in Mdm10↑ mitochondria (Fig 2B). More surprisingly, assembly of other β‐barrel proteins, porin and Tob55/Sam50, into the outer membrane was also impaired in Mdm10↑ mitochondria (Fig 2B). Proteinase K treatment of wild‐type and Mdm10↑ mitochondria after import of Tom40, Tob55/Sam50 and porin showed that translocation across the outer membrane of those proteins was retarded by overexpression of Mdm10 (Fig 2C), which is consistent with the previous observation that association of β‐barrel proteins with the TOB complex was coupled to translocation across the outer membrane (Wiedemann et al, 2003). We obtained similar results for the import of Mdm10 into Mdm10↑ mitochondria (supplementary Fig S5 online). Therefore, whereas depletion of Mdm10 impaired only dissociation of Tom40, but not other β‐barrel proteins, from the TOB complex, overexpression of Mdm10 inhibited association of Tom40 and other β‐barrel proteins with the TOB complex.
Competition between Mdm10 and β‐barrel proteins
Why does the overexpression of Mdm10 inhibit association of β‐barrel proteins, including Tom40, with the TOB complex instead of accelerating Tom40's dissociation from the TOB complex? It was reported that the TOB complex exists as a major 200 kDa TOBcore complex without Mdm10 and as a minor 350 kDa TOBholo complex with Mdm10 (Meisinger et al, 2004). Indeed, Fig 3A shows that the 350 kDa TOBholo complex and the 200 kDa TOBcore complex significantly increased and decreased, respectively, in Mdm10↑↑ mitochondria as compared with wild‐type mitochondria. A decrease in the amount of TOBholo complex to even 58% (Fig 3A) resulted in significantly impaired association of β‐barrel proteins with the TOB complex (Fig 2B). Therefore, the simplest interpretation of these results would be the competition model, in which Mdm10 and premature β‐barrel proteins are mutually exclusive on the TOB complex, so that overexpressed Mdm10 might prevent entry of premature β‐barrel proteins into the TOB complex. Indeed, the assembly I band for Tom40 on a BN‐PAGE gel was shifted by the addition of FLAG antibody when the FLAG‐epitope tag was attached to Mas37/Sam37, Tob55/Sam50 and Tom38/Sam35, but not to Mdm10 or Mdm12 (Fig 3B). Accessibility of the FLAG tag on Mdm10 to the FLAG antibody was confirmed (supplementary Fig S6A online). We also found that, although Mdm10 was in direct contact with Tob55/Sam50, newly imported Mdm10 was not in association with pre‐existing endogenous Mdm10, indicating that each TOBholo complex contains a single Mdm10 molecule (supplementary Fig S6B,C online). In contrast to the carbonate‐extractable assembly I intermediate of premature Tom40 (Model et al, 2001), overexpressed Mdm10 resisted carbonate extraction (supplementary Fig S6D online) and behaved differently to overexpressed Tob55/Sam50 (supplementary Fig S6E–G online). These results indicate that Mdm10 in the TOBholo complex that competes with other premature β‐barrel proteins is not a mere assembly intermediate but a dynamic constituent of the TOB complex.
Therefore, assembly I of Tom40 does not contain Mdm10; that is, Tom40 forms the assembly I intermediate with the TOBcore complex, but not with the TOBholo complex. The mutual exclusiveness of Mdm10 with Tom40 (and probably other β‐barrel proteins) in association with the TOB complex explains the result that conversion of the TOBcore complex to TOBholo complex by overexpression of Mdm10 blocks the entry of newly imported Tom40 and other β‐barrel proteins into the TOB complex. It also explains the reason why association of Mdm10 with assembly I causes release of Tom40 from the TOB complex, which is one of the rate‐limiting steps in the Tom40 assembly process. By contrast, as assembly of β‐barrel proteins other than Tom40 was not affected by Mdm10 reduction or complete depletion (Fig 1C; Meisinger et al, 2004), they do not seem to require Mdm10 for efficient exit from the TOB complex. This is consistent with the well‐known fact that β‐barrel proteins other than Tom40 do not accumulate at the level of the TOB complex during their assembly processes.
What happens to the association of Mdm10 with the MMM1 tethering complex when the level of Mdm10 is altered? We confirmed by using immunoprecipitation that, whereas the amount of Mdm10 co‐immunoprecipitated with Tob55/Sam50 and Tom38/Sam35 was increased by overexpression of Mdm10, the amount of Mdm10 and Mmm1 associated with Mdm12 remained unaffected (Fig 4A). We also confirmed that, even under the conditions of a reduced level (about 30% of wild type) of Mdm10, the amount of Mdm10 associated with Mdm12 was not reduced significantly as compared with the amount associated with Tom38/Sam35 (Fig 4B). Overexpression of Mdm10 thus shifts the equilibrium of the TOB complex, but not the MMM1 complex, from its Mdm10‐free form to Mdm10‐containing form.
Mdm10 was originally identified as a protein involved in mitochondrial morphology and distribution (Sogo & Yaffe, 1994). However, its functions in various cellular processes, including β‐barrel membrane protein assembly, lipid biosynthesis, and tethering of the ER and mitochondria, were also revealed (Meisinger et al, 2004; Osman et al, 2009; Kornmann et al, 2009). These diverse functions of Mdm10 probably arise from its dynamic distribution among distinct subpopulations corresponding to the subunits of TOBholo and MMM1 complexes, and the form in association with Tom7 but not with the TOB or MMM1 complex (data not shown; Meisinger et al, 2006). Therefore, the pleiotropic defects of mdm10Δ mitochondria might well reflect the alteration of the Mdm10 levels in the different subpopulations. Here, we thus established a method to alter the level of Mdm10 in the TOB complex, with minimized effects on the level of Mdm10 in the MMM1 complex, the protein levels of the TOM40 or TOB complex, and the lipid levels of cardiolipin and phosphatidylethanolamine. By using this method, we analysed the function of Mdm10 as a dynamic constituent of the TOB complex in the assembly of an import channel protein Tom40 and other β‐barrel proteins.
Although Mdm10 was reported to mediate a late step of the Tom40 assembly (Meisinger et al, 2004), we found that Mdm10 is primarily involved in the earlier step—which is quantitatively summarized in Fig 2D. Mdm10 appears to primarily assemble with the MMM1 complex, which is less abundant than the TOB complex (data not shown), and the remaining pool of Mdm10 is available for association with the TOB complex. Tom7, which is probably overflown from the TOM40 complex owing to variation in unsynchronized expression of the components of the TOM40 complex, regulates the association of Mdm10 with the TOB complex (Meisinger et al, 2006).
Here, we propose a model of the Tom40 assembly regulated by Mdm10 (Fig 5). After crossing the outer membrane through the TOM40 channel, Tom40 and other β‐barrel proteins associate with the TOBcore complex with the aid of small Tim proteins as chaperones in the intermembrane space. Then, association of Mdm10 with assembly I promotes dissociation of Tom40 from the TOB complex, whereas other β‐barrel proteins can leave the TOBcore complex without the assistance of Mdm10 (Fig 2D). Mdm10 seems to occupy the site in the TOBcore complex that can be used for binding of Tom40 and other β‐barrel proteins because Mdm10 and other β‐barrel proteins are mutually exclusive on the TOB complex (Fig 3B; Meisinger et al, 2004). Recently, a similar role in chase of Tom40 from the TOB complex was proposed for Mas37/Sam37 (Chan & Lithgow, 2008). Mdm10 might cooperate with Mas37/Sam37 to promote dissociation of Tom40 from the TOB complex because overexpression of Mdm10 apparently stabilized Mas37/Sam37 in the TOB complex (supplementary Fig S7 online).
Why does Tom40, but not other β‐barrel proteins, require Mdm10 to leave the TOB complex? After being released from the TOB complex, the premature Tom40 molecule should require a highly coordinated assembly with the other subunits of the TOM40 complex to form the final TOM40 complex (Model et al, 2001). Therefore, an attractive hypothesis is that Mdm10 in cooperation with Tom7 might contribute to adjusting the timing of Tom40 release to availability of the other subunits for assembly with Tom40. This model should be further examined in future studies.
Yeast growth conditions. Cells were grown in YP (1% yeast extract, 2% polypeptone), S (0.67% yeast nitrogen base without amino acids) or SC (0.67% yeast nitrogen base without amino acids, 0.5% casamino acid) containing 2% glucose, 2% galactose or 2% lactate with appropriate supplements.
Yeast strains and plasmids. The yeast strains, plasmids and PCR primers used in this study are described in supplementary Tables S1–S3 online.
Lipid analyses. Cells grown in SC containing 2% glucose medium in the presence of 32Pi for more than 20 h were shifted to lactate (plus 0.2% glucose) medium with 32Pi and cultivated for 6 h at 23°C. Phospholipids were extracted from whole‐cell lysate and separated by thin‐layer chromatography as described previously (Vaden et al, 2005).
In vitro import, BN‐PAGE and immunoprecipitation. In vitro import, BN‐PAGE and immunoprecipitation were performed as described in the supplementary information online.
Supplementary information is available at EMBO reports online (http://www.emboreports.org).
Conflict of Interest
The authors declare that they have no conflict of interest.
Supplementary Figs S1–S7
We thank the members of the Endo laboratory for discussion. We acknowledge the support of this work by Grants‐in‐Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and a Core Research for Evolutional Science and Technology grant from the Japan Science and Technology Corporation. K.Y. was a Research Fellow of the Japan Society of the Promotion of Science.
- Copyright © 2010 European Molecular Biology Organization