|Home | About | Journals | Submit | Contact Us | Français|
99% of all mitochondrial proteins are synthesized in the cytosol, from where they are imported into mitochondria. In contrast to matrix proteins, many proteins of the intermembrane space (IMS) lack presequences and are imported in an oxidation-driven reaction by the mitochondrial disulfide relay. Incoming polypeptides are recognized and oxidized by the IMS-located receptor Mia40. Reoxidation of Mia40 is facilitated by the sulfhydryl oxidase Erv1 and the respiratory chain. Although structurally unrelated, the mitochondrial disulfide relay functionally resembles the Dsb (disufide bond) system of the bacterial periplasm, the compartment from which the IMS was derived 2 billion years ago.
Mitochondria consist of two aqueous compartments, the matrix and the intermembrane space (IMS).2 The mitochondrial genome codes only for roughly a dozen different proteins, and the vast majority of mitochondrial proteins are nuclear encoded. Even simple organisms such as bakers' yeast contain several hundred matrix proteins. Matrix proteins are synthesized in the cytosol as precursors with N-terminal presequences (also referred to as matrix-targeting signals), which are processed following translocation. The import of matrix-destined preproteins is mediated by translocases in the outer membrane (TOM complex) and inner membrane (TIM23 complex) in an ATP- and membrane potential-dependent process (for review, see Refs. 1–3). Proteomic studies suggest that positively charged presequences are consistently found on all matrix proteins, although in some cases they are not proteolytically removed (4, 5).
So far, ~50 different proteins were identified in the IMS of yeast mitochondria, and the list of IMS proteins is rapidly growing (6). The functions of these proteins are diverse. In addition to components involved in mitochondrial respiration, the IMS contains many proteins that transport proteins, metabolites, lipids, or metal ions between both mitochondrial membranes. In addition, several pro-apoptotic components are stored in the IMS and released when the cell death program is triggered (7).
Some IMS proteins are synthesized in the cytosol as preproteins carrying bipartite presequences that consist of a matrix-targeting signal followed by a hydrophobic sorting region. The latter serves as a stop-transfer sequence that is inserted into the inner membrane during protein import and removed by IMS-located proteases (8). Proteins with bipartite presequences embark on the general matrix-directed protein-targeting pathway, from which they are redirected into the IMS. However, many IMS proteins lack N-terminal targeting signals and are sorted into the IMS on a unique import route that differs in many respects from the matrix-targeting pathway. Import of many of these proteins relies on the mitochondrial disulfide relay, which is introduced below.
The IMS proteins Mia40 (mitochondrial IMS import and assembly pathway 40 kDa) and Erv1 (essential for respiratory growth and viability 1) represent the central components of the disulfide relay system. Both proteins are ubiquitously present in eukaryotes and are highly conserved. They are essential for viability in yeast, and conditional mutants show severe defects in the biogenesis of mitochondria and, presumably as secondary effects, in other cellular activities (9–14).
Mia40 binds directly to imported IMS proteins and therefore might serve as an intramitochondrial import receptor. Mia40 comprises a highly conserved domain of ~8 kDa. The structures of this domain of the human and yeast Mia40 proteins were recently solved by NMR and crystallography (15–17). This domain contains six invariant cysteine residues: a redox-active CPC motif (−290 mV) is followed by a twin CX9C motif (see below) that forms two structural disulfides (Fig. 1A). The CPC motif is part of a short helix that is connected via a flexible region to a rigid helix-loop-helix region that is stabilized by the twin CX9C motif. The two helices form a hydrophobic binding groove that is positioned in close proximity to the redox-active disulfide bond. This groove serves as binding region for substrates that presumably are recognized by so-called MISS (for mitochondrial IMS-sorting signal; also referred to as ITS for IMS-targeting signal) sequences (see below) (18, 19).
In fungi, but not in animals and plants, Mia40 proteins contain N-terminal linker regions that anchor them in the inner membrane (13, 14). In yeast, the N-terminal anchor region is not essential (13), and its function is less clear. It is conceivable that the addition of a bipartite import signal might be advantageous under anaerobic conditions when the oxidation-driven import via the mitochondrial disulfide relay pathway might be less efficient.
The oligomeric state of Mia40 is not known. On blue native gels, yeast Mia40 migrates at 150–180 kDa (12), suggesting that Mia40 is part of a larger complex. Alternatively, Mia40 might migrate exceptionally slow on these gels; on denaturing SDS-polyacrylamide gel, the 40-kDa protein migrates at ~60–70 kDa (12, 13). The recombinant expression of the conserved domain of Mia40 did not reveal any evidence for dimer or oligomer formation.
The second component of the disulfide relay, Erv1, was initially identified as a component that can stimulate liver regeneration (20). In these experiments, parts of the livers of rats were removed, and factors of liver extracts that improved regeneration of the organ were identified. Rat Erv1 was found to serve as a growth factor for liver regeneration, and in mammals, Erv1 is therefore also referred to as augmenter of liver regeneration (ALR), growth factor erv1-like (GFER), or hepatopoietin. Whether Erv1 plays a physiological role in liver development is not known.
Erv1 consists of two domains, an N-terminal less structured region (shuttle domain) and a C-terminal FAD-binding domain (Fig. 1B) (10, 21, 22). Several structures of the FAD-binding domain of the human and yeast Erv1 proteins were solved by NMR and crystallization, revealing a conserved four-bundle structure holding the FAD factor noncovalently bound in a hydrophobic, deeply embedded pocket (15, 16, 23–26). A surface-exposed CXXC motif is located in proximity to the isoalloxazine ring of the FAD, which is efficiently oxidized by electron transfer to the FAD. A second redox-active CXXC motif is part of the flexible N-terminal domain of Erv1. Erv1 forms a dimer in which the cysteine pairs of one N-terminal domain and one FAD domain from two opposing subunits are in close proximity, allowing an efficient intermolecular oxidation of the N-terminal CXXC motif (27–30). The N-terminal region serves as shuttle domain that interacts with Mia40 (see below).
The FAD domains in some Erv1 proteins (for example, the Arabidopsis thaliana homolog) contain a short but well defined hydrophobic tunnel through which oxygen has access to N5 of the FAD isoalloxazine ring system (21, 25). Oxygen can thereby directly oxidize the FAD moiety, which leads to the production of hydrogen peroxide. Mammalian Erv1 proteins lack this hydrophobic tunnel, presumably to prevent hydrogen peroxide formation. In fungi and most likely also in metazoa, FAD oxidation is mediated by cytochrome c, a highly abundant component in the IMS. In vitro, collision of reduced Erv1 with oxidized cytochrome c allows rapid electron exchange (31–33). At least in yeast, cytochrome c is presumably the predominant oxidant of Erv1 because mutants lacking cytochrome c or cytochrome c oxidase accumulate Mia40 in its reduced form. Thus, electrons from the disulfide relay pathway are channeled into the respiratory chain and presumably contribute to the formation of the mitochondrial membrane potential and ATP production. However, in comparison with the electrons originating from NADH and FADH2, their contribution is very minor.
Yeast cells can grow anaerobically. It is not clear how Erv1 oxidation is achieved in the absence of oxygen. Interestingly, it was reported that cytochrome c becomes essential upon oxygen depletion (34). This suggests that cytochrome c is required for Erv1 oxidation under these conditions, but it remains unclear how cytochrome c is oxidized anaerobically.
Hot13 is a small cysteine-rich IMS protein that improves the performance of the disulfide relay to some degree (35). In contrast to Mia40 and Erv1, Hot13 is dispensable in yeast, and Hot13-deficient cells show no obvious defects. In vitro experiments suggest that Hot13 can remove zinc ions from newly imported proteins and from Mia40, thereby improving protein oxidation in the IMS (36, 37).
Whether they are directed to the matrix or the IMS, mitochondrial preproteins can be imported into isolated mitochondria in vitro. Hence, an obligatory coupling of protein synthesis and protein translocation (such as found for most proteins of the mammalian endoplasmic reticulum) does not apply for mitochondrial protein import. Nevertheless, protein synthesis and protein import still might be kinetically coupled, i.e. proteins might start to be imported while they are still synthesized. This might be achieved by the attraction of nascent chains of polysomes to the surface of mitochondria so that proteins that are produced by downstream ribosomes are synthesized in direct proximity of TOM complexes. In agreement with this concept, it was shown that mRNAs for mitochondrial proteins are enriched in mitochondrial fractions isolated from yeast cells and that, at least for certain transcripts, this enrichment depends on the presence of the preprotein receptor Tom20 (38, 39). In addition, yeast mitochondria carry the mRNA-binding protein Puf3 on their surface (40–42). Puf3-binding sites were identified in several 5′-UTR regions of mRNA encoding mitochondrial proteins, including some IMS proteins such as Cox17, Cox23, and Pet191 (Fig. 2). Whether Puf3 binding to mRNA leads to co-translational protein import or has other functions in the expression of mitochondrial proteins is not known.
Substrates need to be in a reduced and unfolded state to be imported into the IMS (43–45). For cytosolic precursors of the small Tim proteins Tim9 and Tim10, it was reported that bound zinc ions stabilize the reduced state of cysteine residues prior to import. However, it is not known whether the zinc ions are removed before import or if proteins enter the IMS in a zinc-containing form (36).
The protein-conducting channel of the TOM complex is believed to serve as a general entry site for preproteins into mitochondria. Also IMS proteins apparently employ the TOM pore, as blocking the TOM complex with large amounts of matrix-targeted proteins prevented protein import into the IMS (44). Direct binding of substrates of the mitochondrial disulfide relay to TOM receptors was, however, not shown. Mia40 plays an essential role in the import of these proteins and serves as a docking factor that binds substrates during or directly after their translocation across the outer membrane (Fig. 2).
Preproteins bind to Mia40 by use of specific binding sequences (19). These so-called MISS motifs (consensus aromatic-XX-hydrophobic-hydrophobic-XXC) serve as recognition sites that determine which positions of the incoming polypeptides are bound by Mia40 (18, 19). The MISS motifs thereby presumably position the sequences in the binding cleft of Mia40 in an orientation that allows the interaction of the redox-active CPC motif of Mia40 with the cysteine residue in the substrate (15, 17). In this reaction, a mixed disulfide of both proteins is formed as a reaction intermediate (12, 13, 46). It is conceivable that this strong contact is used to improve preprotein translocation across the outer membrane. In the imported substrate proteins, the sequence around the MISS motif forms a helical region, and NMR studies suggested that helix formation is induced or stabilized by Mia40 binding (16). In a final step, formation of an intramolecular disulfide bond leads to the release of the substrates from Mia40 (46). Because only unfolded proteins are able to traverse the TOM pore, oxidative folding traps the proteins stably in the IMS.
Most substrate proteins contain two disulfide bonds. At least in vitro, Mia40 can introduce both disulfide bonds in sequential reactions (30). In the presence of oxygen, semi-oxidized proteins are also further oxidized without additional factors in vitro, but whether this reflects the in vivo situation is unclear (15, 16). Mutants in which individual cysteines are mutated can still be oxidized as long as the MISS signal is not destroyed; in this case, only one disulfide bond is formed (15, 30).
In vitro, glutathione plays a critical role in substrate oxidation by Mia40 (30, 46). The precise role of glutathione is not clear, but the presence of 5–10 mm glutathione strongly increases both Mia40-mediated substrate oxidation in the reconstituted system and Mia40-dependent protein import into the IMS. How can a reductant increase the rate of protein oxidation? It was observed that, in the absence of glutathione, Mia40 accumulated in mixed disulfides, which represented kinetically trapped oxidation intermediates. It is conceivable that, in these intermediates, non-native disulfides are formed, and Mia40 can be released only by the help of an isomerase/reductase activity. Glutathione could fulfill this function and thereby serve as a lubricant that ensures that the oxidation pathway will finally lead to a native folded protein. This would also explain why most Mia40 substrates contain two disulfide bonds because these will strongly stabilize the oxidized conformations and make them resistant to reduction by glutathione. Indeed, IMS proteins such as small Tim proteins and twin CX9C proteins (see below) were shown to have very negative redox potentials (−340 to −310 mV) and are reduced only by high concentrations of dithiothreitol or glutathione if boiled in urea or SDS (47). Because the strongly stimulating effect of glutathione was observed only in vitro, it cannot be excluded that in vivo additional reductants contribute to the folding of IMS proteins (48).
The N-terminal shuttle domain of Erv1 is specifically designed to interact with Mia40, whereas the redox-active cysteine pair in the FAD domain is inaccessible to Mia40 (30, 49, 50). The sequence around the N-terminal cysteine pair is conserved and forms a helix that resembles the region around the MISS signal in imported proteins. Presumably, imported proteins and Erv1 alternately interact with Mia40, thereby cycling it between oxidized and reduced states (Fig. 2) (50). The redox potentials of the substrate proteins (−340 to −310 mV for Tim9, Tim10, Tim13, and Cox17) were found to be very similar to that of the N-terminal domain of Erv1 (−320 mV) (29, 33, 45, 51–53). However, the redox pair in the FAD domain of Erv1 is significantly more positive (−150 mV), hence oxidizing, and presumably drives the oxidation of substrate proteins. It was suggested that Mia40 and Erv1 are associated with each other, forming a ternary complex with the substrate to mediate substrate oxidation with high efficiency (54).
The majority of the substrates of the mitochondrial disulfide relay are small proteins with simple helix-loop-helix folds in which the helices are connected by two disulfide bonds (55–58). The cysteines forming these disulfides are spaced by either three or nine amino acid residues. Consequently, the proteins are called twin CX3C and twin CX9C proteins, respectively. In yeast, 15 mitochondrial twin CX9C motif proteins and five mitochondrial twin CX3C motif proteins were identified, most of which are conserved from plants to humans (59–61). The oxidoreductase Mia40 also contains a twin CX9C motif, but it is much larger compared with other members of the twin CX9C family (59). Notably, bioinformatics analyses suggest the number of twin CX9C proteins in human cells to be at least twice as high (62), whereas the number of twin CX3C proteins in animals and fungi is five.
The members of the twin CX3C family are also referred to as small Tim proteins, a group of import components that are ubiquitously expressed in eukaryotes. Small Tim proteins function as chaperones that facilitate the transport of hydrophobic membrane proteins from the TOM channel through the IMS to their insertion sites at the inner and outer membranes (Fig. 3A) (63–65). The small Tim proteins were shown to form three distinct heterohexameric complexes of ring-like structure: two soluble complexes formed by Tim9 and Tim10 and by Tim8 and Tim13, respectively, and one Tim9-Tim10-Tim12 complex that is associated with the membrane-embedded TIM22 translocase of the inner membrane (47, 58, 66, 67). Formation of the disulfide bonds within the small Tim proteins was shown to be a prerequisite for their assembly into these complexes (43).
Notably, a mutation of one of the cysteine residues in the human homolog of Tim8, DDP1/TIMM8a, has been shown to be the cause of the Mohr-Tranebjaerg (MTS/DFN-1) or deafness/dystonia syndrome, a progressive neurodegenerative disorder (68, 69). In this disease, TIMM8a is unstable and cannot be detected in patient fibroblasts. Due to a critical function of the Tim8-Tim13 complex in the import of Tim23, the central component of the preprotein translocase of the inner membrane, the absence of TIMM8a results in impaired protein import into mitochondria and, as a consequence, in severe pleiotropic mitochondrial dysfunction.
Although the small Tim proteins play a consistent role in the import of proteins into mitochondria, the functions of twin CX9C proteins appear to be rather heterogeneous (Fig. 3). Many of the twin CX9C proteins are linked to the assembly of the cytochrome c oxidase of the respiratory chain (Fig. 3D) (59, 60). For example, the copper chaperone Cox17 delivers copper to subunits 1 and 2 (Cox1 and Cox2) of cytochrome c oxidase and has been shown in vitro to bind Cu(I) in a redox-regulated fashion (70, 71). Cox19 and Cox23, which are structurally related to Cox17, are also involved in copper delivery, although they may not bind copper directly (72, 73). Likewise, the proteins Cmc1 and Cmc2 have recently been shown to be involved in the maturation of cytochrome c oxidase (74, 75).
The two twin CX9C proteins Mdm35 and Som1 are presumably not involved in cytochrome c oxidase assembly. Mdm35 is important for the import of the Ups1, Ups2, and Ups3 proteins into the IMS; these three factors that are critical for mitochondrial lipid homeostasis (Fig. 3B) (76, 77). In the absence of Mdm35, the three Ups proteins are unstable and are substrates for the i-AAA protease Yme1 and Atp23, two IMS-localized peptidases. The formation of a stable complex between Mdm35 and the Ups proteins ensures their stable accumulation in the IMS.
The protein Som1 is the third subunit of the Imp (inner membrane peptidase) complex (Fig. 3E) (78–81). The Imp peptidase processes nuclear encoded substrates (such as cytochrome b2, the cytochrome b5 reductase Mcr1, the glycerol-3-phosphate dehydrogenase Gut2, and cytochrome c1) as well as the mitochondrially encoded protein Cox2. Deletion mutants of SOM1 show defects in the processing of several Imp substrates (78, 82).
In addition to the twin CX3C and twin CX9C proteins, the IMS harbors other proteins containing disulfide bonds. These proteins include the superoxide dismutase Sod1 and its copper chaperone Ccs1, the sulfhydryl oxidase Erv1, the complex III subunits Qcr6 and Rieske protein Rip1, the thioredoxin-like proteins Sco1 and Sco2, and the proteins Cox11 and Cox12.
Sod1 and Ccs1 form part of the antioxidative system that dismutates superoxide anions to hydrogen peroxide (Fig. 3F). Sod1 is a dimeric copper- and zinc-containing protein that contains one disulfide bond per subunit. The insertion of this disulfide bond and of the copper ion is facilitated by Ccs1 (83–85). The majorities of both proteins are present in the cytosol; however, small amounts are also found in the IMS of mitochondria (84, 86, 87). It was proposed that Ccs1 mediates the import of Sod1 into the IMS because up-regulation of Ccs1 in the IMS results in an increase of Sod1 in this compartment (84). Like Sod1, Ccs1 contains one structural disulfide bond. This disulfide bond is introduced during import by the mitochondrial disulfide relay (88, 89, 100).
In addition to its two redox-active CXXC motifs, Erv1 contains a structural disulfide bond that is critical for its stability (Fig. 3C). This disulfide bond stabilizes the four-helix bundle of Erv1 that is responsible for the binding of the redox cofactor FAD. Notably, Erv1 lacks a mitochondrial presequence and is a substrate of the mitochondrial disulfide relay itself (60, 90). However, it remains unclear whether Mia40 introduces the structural disulfide into Erv1.
Complex III of the respiratory chain also harbors two disulfide-containing proteins that face the IMS: Qcr6 (subunit 8 in mammals) and the Rieske iron-sulfur protein Rip1 (Fig. 3D). Qcr6 and subunit 8 are composed of two antiparallel helices that are connected by one or two disulfide bonds, respectively (91), thereby resembling the structure of twin CX9C proteins. In the catalytic subunit Rip1, the [2Fe-2S] iron-sulfur cluster is held between two loops of the protein that are connected by a disulfide bond (92). This disulfide bond is found in Rieske proteins of all eukaryotes and is critical for the enzymatic activity of complex III (93).
In addition, the proteins Cox11 and Cox12 are disulfide-containing proteins of the IMS (Fig. 3D). Cox12 is part of cytochrome c oxidase, and its CX9C-CX10C cysteine pattern closely resembles the twin CX9C motifs. As in these proteins, the cysteines form two parallel disulfide bonds. Cox11 is an assembly factor for cytochrome c oxidase that is required for copper insertion (94). It has been proposed that the protein can also exist as a dimer that is stabilized by a disulfide bond (95). The function of Cox11 in copper transfer from Cox17 to Cox1 might be catalytically similar to that of Sco1 and Sco2 in the copper transfer from Cox17 to Cox2 (96). Sco1 and Sco2 are membrane proteins that expose a domain with a thioredoxin-like fold in the IMS (Fig. 3D) (52, 97, 98). This fold also contains a CX3C motif that has recently been shown to be redox-active in vivo. It was proposed that Sco2 acts as a thiol-disulfide oxidoreductase that can oxidize the copper-coordinating cysteine residues in Sco1 during Cox2 maturation (99). It remains unclear whether the mitochondrial disulfide relay is involved in the reoxidation of Sco2.
The function of the mitochondrial disulfide relay for the import or folding of these proteins is not well understood. Proteins such as Sco1/Sco2, Cox11, Erv1, Rip1, and Sod1/Ccs1 are of very diverse structure, suggesting that the protein-folding capacity of Mia40 is not limited to simply structured helix-loop-helix proteins. It will be a major task in the future to study the relevance of the mitochondrial disulfide relay for the biogenesis and function of the wide range of IMS proteins.
*This work was supported by grants from the Deutsche Forschungsgemeinschaft and the Landesschwerpunkt Rheinland-Pfalz on Membrane Transport (to J. M. H. and J. R.). This is the fifth article in the Thematic Minireview Series on Redox Sensing and Regulation.
2The abbreviations used are: