The Cox2 subunit of Saccharomyces cerevisiae cytochrome c oxidase is synthesized in the mitochondrial matrix as a precursor whose leader peptide is rapidly processed by the inner membrane protease following translocation to the intermembrane space. Processing is chaperoned by Cox20, an integral inner membrane protein whose hydrophilic domains are located in the intermembrane space, and Cox20 remains associated with mature, unassembled Cox2. The Cox2 C-tail domain is exported post-translationally by the highly conserved translocase Cox18 and associated proteins. We have found that Cox20 is required for efficient export of the Cox2 C-tail. Furthermore, Cox20 interacts by co-immune precipitation with Cox18, and this interaction requires the presence of Cox2. We therefore propose that Cox20 binding to Cox2 on the trans side of the inner membrane accelerates dissociation of newly exported Cox2 from the Cox18 translocase, promoting efficient cycling of the translocase. The requirement for Cox20 in cytochrome c oxidase assembly and respiratory growth is partially bypassed by yme1, mgr1 or mgr3 mutations, each of which reduce i-AAA protease activity in the intermembrane space. Thus, Cox20 also appears to stabilize unassembled Cox2 against degradation by the i-AAA protease. Pre-Cox2 leader peptide processing by Imp1 occurs in the absence of Cox20 and i-AAA protease activity, but is greatly reduced in efficiency. Under these conditions some mature Cox2 is assembled into cytochrome c oxidase allowing weak respiratory growth. Thus, the Cox20 chaperone has important roles in leader peptide processing, C-tail export, and stabilization of Cox2.
CYTOCHROME c oxidase is composed of three subunits encoded in the mitochondrial genome (mtDNA) and eight subunits encoded in the nuclear genome in the budding yeast Saccharomyces cerevisiae. In addition to the genes for these subunits, at least 20 other nuclear yeast genes are specifically required for synthesis of the mitochondrially coded subunits and post-translational steps in assembly of the active enzyme (Barrientos et al. 2002; Herrmann and Funes 2005; Fontanesi et al. 2006).
The second largest subunit of cytochrome c oxidase, Cox2, is a mitochondrial gene product whose acidic N-terminal and C-terminal domains are translocated through the inner membrane from the matrix to the intermembrane space (IMS) and flank two transmembrane helices (Tsukihara et al. 1996). Cox2 topogenesis is of particular interest since its hydrophilic domains are the largest known to be exported through the inner membrane. In budding yeast, localized membrane-bound translation of the COX2 mRNA, specifically activated by Pet111 (Green-Willms et al. 2001; Naithani et al. 2003), produces a precursor, pre-Cox2, with a short N-terminal leader peptide (Pratje et al. 1983). The pre-Cox2 N-tail is co-translationally exported by Oxa1 (He and Fox 1997; Hell et al. 1998), a highly conserved inner membrane translocase that is also required for C-tail export (reviewed in Bonnefoy et al. 2009). Once in the IMS, the pre-Cox2 leader peptide is rapidly processed by the inner membrane protease (IMP) (Nunnari et al. 1993; Jan et al. 2000), in a reaction chaperoned by Cox20 (Hell et al. 2000).
The acidic Cox2 C-tail is exported to the IMS by a mechanism that is distinct from N-tail export and appears to be post-translational (He and Fox 1997; Fiumera et al. 2007). C-tail export depends specifically upon another highly conserved inner membrane translocase, Cox18 (Saracco and Fox 2002), which is paralogously related to Oxa1, as well as to bacterial YidC proteins (Funes et al. 2004). In addition, Cox2 C-tail export requires Mss2 and is promoted by Pnt1, two proteins that interact with Cox18 (He and Fox 1999; Broadley et al. 2001; Saracco and Fox 2002). Interestingly, overproduction of Oxa1 in a mutant lacking Cox18 results in some export of the Cox2 C-tail, although Cox2 remains unassembled and the cells fail to respire (Fiumera et al. 2009). This result suggested that Cox18 has an assembly function in the IMS that overproduced Oxa1 cannot carry out, in addition to its translocation function. One possibility here is that Cox18 could promote interaction of the exported Cox2 C-tail with an assembly factor in the IMS. One candidate for such a factor is Cox20.
Cox20 has previously been shown to be a 205-amino-acid integral mitochondrial inner membrane protein. It has two centrally located transmembrane helices flanked by hydrophilic domains in the intermembrane space (Hell et al. 2000). Cox20 interacts directly with pre-Cox2 and promotes its processing. Since this interaction depends upon export of pre-Cox2 by Oxa1, it appears to involve domains in the IMS (Hell et al. 2000). In addition, Cox20 remains associated with unassembled mature Cox2, suggesting that it has roles in cytochrome c oxidase assembly downstream of pre-Cox2 processing (Hell et al. 2000; Preuss et al. 2001; Herrmann and Funes 2005), possibly including Cox2 metallation (Rigby et al. 2008). Cox20 has therefore been described as a chaperone, although it has no detectable similarity to other well-characterized chaperones or domains of known function.
In this study we have investigated the role of Cox20 in the export and assembly of Cox2. We find that Cox20 is required for efficient export of the Cox2 C-tail and that it interacts with the translocase Cox18 but only when Cox2 is present. In addition, Cox20 stabilizes mature but unassembled Cox2.
Materials and Methods
Yeast strains and genetic analysis of pseudorevertants
S. cerevisiae strains used in this study are listed in Table 1. All strains are congenic to D273-10B (ATCC 25657). Nuclear genes were manipulated using standard methods (Guthrie and Fink 1991). Transformation of plasmids and PCR products into yeast was accomplished with the EZ transformation kit (Zymo Research). Complete media (YPA) containing adenine, dextrose (D), ethanol plus glycerol (EG), or raffinose (R) were prepared as previously described (Guthrie and Fink 1991). Complete synthetic media (CSM) and CSM lacking specific growth factors were purchased from Bio101 Systems. COX20 was modified to encode a protein tagged with three MYC epitope at its C terminus, but with no other changes, by the pop-in pop-out strategy as described (Schneider et al. 1995). The plasmid pOXA1-W56R-ADH1 (Supekova et al. 2010) was obtained from F. Supek and P. G. Schultz
Independent spontaneous pseudorevertants of the cox20Δ strain LEE83 were isolated by plating cells from distinct clonal cultures on YPAEG plates and incubating at 30° for 7 to 10 days. A single pseudorevertant was picked from each clonal culture, purified, and mated to cox20Δ (SCS194), cox20Δmgr1Δ (LEE145), cox20Δmgr3Δ (LEE100), and cox20Δyme1Δ (LEE106). The resulting diploids were isolated and their ability to grow on YPAEG was assessed to determine that the mutations were recessive and to score their ability to complement the known genes.
Analysis of mitochondrial proteins
To examine export of the C terminus of Cox2, we employed strains whose mtDNA encodes a version of Cox2 bearing C-terminal HA-epitopes (Saracco and Fox 2002). Mitoplasts were prepared by osmotic shock from purified mitochondria as described (Glick 1995; Glick and Pon 1995). For each sample of mitoplasts, the equivalent of 75 μg of mitochondrial protein was treated with 20 μg/ml proteinase K, or mock treated, as described (Saracco and Fox 2002), except that after protease treatment samples were directly resuspended in 20 mM HEPES pH 7.4, 0.6 M sorbitol, 10% trichloroacetic acid, and 2 mM PMSF and incubated at 60° for 10 min.
For co-immune precipitation experiments, crude mitochondria were prepared as described (Diekert et al. 2001). One milligram of mitochondrial protein was solubilized in 1 ml of 1% digitonin, 100 mM NaCl, 20 mM Tris pH 7.4, and 1 mM PMSF for 10 min on ice and then centrifuged at 16,000 × g for 10 min at 4°. Five percent of the clarified supernatant was precipitated using Strataclean resin (StrataGene). The remaining supernatant was incubated with anti-hemagglutinin (anti-HA) antibody conjugated beads (Roche). After incubation for 1 hr at 4°, the beads were washed three times in the digitonin buffer before proteins were eluted in 5× Laemmli SDS sample buffer.
Protein samples to be analyzed by Western blotting were separated on 12% or 16% SDS-PAGE gels as indicated and transferred to Immobilon-P polyvinylidene difluoride membrane (Millipore). Antibodies were anti-HA monoclonal antibody clone 3F10 (Roche) (except the experiment of Figure 1, where anti-HA-HRP conjugate (Roche) was used), anti-c-myc (anti-Myc) monoclonal antibody clone 9E10 (Roche), polyclonal anti- Cit1 (citrate synthase), polyclonal anti-Yme1, and monoclonal anti-Cox2 CCO6. Anti-mouse IgG-HRP (BioRad) or anti-rabbit IgG-HRP (BioRad) secondary antibodies were applied after washing and detected by the ECL and ECL+ detection methods (GE Healthcare).
In vivo pulse-labeling of mitochondrial translation products was carried out as described (Bonnefoy et al. 2001) in cells grown on YPAR medium, except that there was no chase incubation, and mitochondria were prepared as described (Diekert et al. 2001). Radiolabeled mitochondrial proteins were separated on 16% SDS PAGE gels, which were dried and autoradiographed.
Cox20 is required for Cox2 C-tail export independently of cytochrome c oxidase assembly or Cox2 N-tail processing
Cox20 is known to be required for the N-terminal processing of pre-Cox2 and to remain associated with mature Cox2 prior to its assembly into cytochrome c oxidase (Hell et al. 2000). Before Cox2 can be assembled, the C-tail must be translocated from the matrix to the intermembrane space. We asked whether Cox20 plays a role in the export of the Cox2 C-tail. Mitochondria were purified from strains whose mtDNA encodes a variant of Cox2 with three HA-epitope tags at its C terminus and then converted to mitoplasts lacking the outer membrane. This epitope is accessible to protease added to wild-type mitoplasts, whose Cox2 C-tail has been translocated through the inner membrane (Saracco and Fox 2002; Fiumera et al. 2007) (Figure 1). Protease digestion of mitoplasts from a cox20Δ strain did not destroy the C-terminal epitope, but did shorten the tagged protein by digestion of the exported Cox2 N-tail domain (Figure 1). The C-tail epitope was digested by protease when the mitoplasts were solubilized by the addition of the mild detergent octyl glucoside. Thus, Cox20 is required for export of the Cox2 C-tail domain through the inner membrane, but not for the mechanistically distinct (He and Fox 1997) N-tail export process.
Hell et al. (2000) had previously observed a defect in Cox2 C-tail export in the cox20Δ mutant. However, they attributed this effect to decreased inner membrane potential caused by the lack of cytochrome c oxidase in the mutant. We investigated this possibility further by examining the topology of the HA epitope in strains lacking either Cox1 or Cox3, the other two core subunits of cytochrome c oxidase. In both cases the C-terminal epitope on Cox2 was accessible to protease in mitoplasts from the mutants (Figure 1), confirming that Cox2 C-tail export is not prevented indirectly by the absence of cytochrome c oxidase activity (Saracco and Fox 2002).
Since Cox20 is required for N-terminal processing of the pre-Cox2 leader peptide, we next asked whether this processing is required for C-tail export. Pre-Cox2 is processed by cleavage after residue 15 by the Imp1 subunit of the inner membrane protease complex (Pratje et al. 1983; Behrens et al. 1991; Nunnari et al. 1993). Neither the cox2-N15I processing site mutation (Saracco 2003; Perez-Martinez et al. 2009) nor the imp1Δ mutation, both of which prevent pre-Cox2 leader peptide processing, prevented Cox2 C-tail export as measured by protease sensitivity of the Cox2 C-terminal HA epitopes in mitoplasts (Figure 1). (While some HA-reactive Cox2 was detected in the imp1Δ mitoplasts treated with protease, this is largely attributable to inefficient mitoplasting, as evidenced by the partial protection of the IMS marker Yme1.)
Cox20 interacts dynamically with Cox18 in a Cox2-dependent manner
Cox18 is the inner membrane translocase responsible for moving the Cox2 C-tail into the intermembrane space (Saracco and Fox 2002; Bonnefoy et al. 2009). To further investigate the role of Cox20 in translocation, we tested for physical interaction between Cox18 and Cox20 by co-immune precipitation. Mitochondria were isolated from cells containing epitope-tagged Cox20-Myc and either tagged Cox18-HA or wild-type Cox18. Digitonin solubilized extracts were immunoprecipitated with anti-HA antibody bound to agarose beads, and the precipitates were analyzed by Western blotting (Materials and Methods). Cox20–Myc coprecipitated with Cox18–HA but was not precipitated from extracts containing untagged Cox18 (Figure 2).
Cox20 and Cox18 could either interact in a stable complex or could interact dynamically during the translocation of the Cox2 C-tail. To distinguish between these possibilities we asked whether Cox20–Myc and Cox18–HA co-immune precipitation depends on the presence of Cox2. Cox2 translation was prevented by introducing the cox2-20 mutation (Torello et al. 1997; Bonnefoy et al. 2001) into the mtDNA of a strain containing Cox20–Myc and Cox18–HA. A mitochondrial extract from this strain was analyzed as described above, revealing that the interaction between Cox20 and Cox18 was disrupted when Cox2 synthesis was blocked (Figure 2). Thus, mitochondrially coded Cox2 appears to bridge the interaction between the translocase Cox18 and the chaperone-like protein Cox20.
The absence of Cox20 can be partially bypassed by disruption of the i-AAA protease complex
Cox20 is required for respiratory growth (Figure 3, Hell et al. 2000). However, cox20Δ mutant cells plated on nonfermentable medium yielded spontaneous weakly respiring pseudorevertants. We suspected that these mutations might affect the activity of the i-AAA protease, on the basis of a previous study of mutations that allow overproduced Oxa1 to partially suppress a cox18Δ mutation (Fiumera et al. 2009). Since Cox20 and Cox18 appear to function together, we tested the phenotype of double mutants containing the cox20Δ mutation and mgr1Δ, mgr3Δ, or yme1Δ (Figure 3). In a cox20Δ background, deletion of YME1, MGR1, or MGR3 caused weak respiratory growth (Figure 3). However, this weak respiratory growth was dependent upon IMP1.
We found that 36 independent spontaneous cox20Δ pseudorevertants (Materials and Methods) contained recessive nuclear mutations, since they yielded nonrespiring diploids when mated with an otherwise wild-type cox20Δ strain. Twenty-three of these pseudorevertants appear to be due to mgr3 mutations since they produced respiring diploids when mated to cox20Δ mgr3Δ strain, indicating failure to complement, but nonrespiring diploids when mated to either cox20Δ mgr1Δ or cox20Δ yme1Δ strains. Tetrad analysis confirmed that five of these pseudorevertants were indeed caused by mutations tightly linked to mgr3Δ. Ten of the remaining spontaneous pseudorevertants failed to complement after being mated with a cox20Δ mgr1Δ. The remaining three pseudorevertants produced nonrespiring diploids when mated to cox20Δ double mutant strains containing mgr1Δ, mgr3Δ, or yme1Δ mutations, indicating that they complemented all three. The spontaneous mutations in these three pseudorevertants fall into a single complementation group and are linked to each other, indicating that they affect a single, as-yet unidentified, additional gene. This screen is apparently not saturated, since none of the pseudorevertants contained mutations that failed to complement yme1Δ.
Imp1-dependent processing of pre-Cox2 in cox20Δ mutants occurs in the absence of i-AAA protease activity
Processing of the pre-Cox2 N-terminal leader peptide is essential for assembly of cytochrome c oxidase, since mitochondrial mutations that alter the processing site cause a nonrespiratory growth phenotype in strains with wild-type nuclear genomes (Bonnefoy et al. 2001; Saracco 2003; Perez-Martinez et al. 2009). The fact that one function of Cox20 is to assist in this Imp1-dependent processing event (Hell et al. 2000) raised the question of how the pre-Cox2 N terminus is processed in cox20Δ stains lacking i-AAA protease activity. We first confirmed that mature Cox2 does accumulate at steady state in the weakly respiring cox20Δ mgr1Δ, cox20Δ mgr3Δ, and cox20Δ yme1Δ double mutants by Western blotting of mitochondrial proteins probed with an anti-Cox2 antibody. While processing was much less efficient in the double mutants than in wild type, each of the weakly respiring double mutants contained significant amounts of mature Cox2 (Figure 4A). The generation of mature Cox2 under these conditions is still dependent upon Imp1 activity since a cox20Δ yme1Δ imp1Δ triple mutant, which has a nonrespiratory growth phenotype (Figure 3), accumulated the same pre-Cox2 species as an imp1Δ single mutant (Figure 4A).
All of the strains deficient in i-AAA protease activity also accumulated detectable amounts of a smaller Cox2 species (indicated by * in Figure 4A). Since the i-AAA protease is known to be responsible for degradation of unassembled Cox2 (Nakai et al. 1995; Pearce and Sherman 1995; Weber et al. 1996), the smaller Cox2 species appears to be a breakdown product of pre-Cox2 that accumulates in the absence of i-AAA activity. Interestingly, the cox20Δ mutant accumulated substantially less pre-Cox2 than either the imp1Δ mutant or any of the cox20Δ strains deficient in i-AAA protease (Figure 4). These data suggest that Cox20 directly stabilizes pre-Cox2 against degradation by the i-AAA protease. Alternatively, pre-Cox2 whose C-tail has not been exported may be more labile due to its aberrant topology.
In contrast to the accumulation of immunologically detectable mature Cox2 in weakly respiring cox20Δ mutants deficient in i-AAA protease, processing of pre-Cox2 in the absence of Cox20 was not observed during 35S-pulse labeling of mitochondrial translation products in the same strains (Figure 4B). Thus, while Imp1-dependent processing of pre-Cox2 occurs in the double mutants (Figure 4A), it is too slow to be detectable during pulse labeling. These data are consistent with the proposal that Cox20 serves as a chaperone to present pre-Cox2 to the inner membrane protease complex, but is not a component of that complex (Hell et al. 2000).
Import of a variant form of pre-Cox2 from the cytoplasm does not bypass the requirements for Cox20 and Imp1 in cytochrome c oxidase biogenesis
We attempted to dissect further the activities of Cox20 using a plasmid-borne mutant version of the COX2 gene, that had been recoded for expression from the nucleus. This recoded COX2 gene (denoted N-COX2 in Figure 5), specifying the amino acid substitution W56R and the mitochondrial targeting signal of the Oxa1 protein, was obtained on a high-copy plasmid termed pOXA1–W56R–ADH1 (Supekova et al. 2010). We transformed this plasmid into the nuclei of mutant strains (Figure 5). As expected, it strongly complemented the respiratory growth defect of a cox2Δ mutation in mtDNA. Furthermore, import of the nuclearly encoded, cytoplasmically synthesized variant of Cox2 also complemented the respiratory defect of a nuclear pet111Δ mutant that is specifically unable to translate the mitochondrial COX2 mRNA (Figure 5). In addition, import of this protein bypassed the requirement for the Cox2 C-tail translocase Cox18, suggesting that the assembly pathway for the imported variant of Cox2 does not involve import into the matrix and subsequent reexport of the C-tail. However, import of this protein in the absence of Cox20, Imp1, or Oxa1 did not promote respiratory growth (Figure 5) or the assembly of active cytochrome c oxidase as assayed by reduction of tetramethyl-p-phenylenediamine (Mcewen et al. 1985; unpublished data).
The experiments presented here argue strongly that Cox20 has an important role in Cox2 C-tail topogenesis, in addition to its previously established function in pre-Cox2 leader peptide processing. First, in the absence of Cox20, mitochondrially synthesized unassembled pre-Cox2 accumulates with its N-tail domain in the IMS and its C-tail domain in the matrix, showing that Cox20 is required for efficient C-tail translocation through the inner membrane. Second, this defect in Cox2 C-tail export cannot be due to the defect in pre-Cox2 processing caused by the absence of Cox20, since C-tail export was not affected by either cox2 or imp1 mutations, which also prevent pre-Cox2 processing. Finally, Cox20 co-immune precipitates with the Cox2 C-tail-specific translocator Cox18 from mitochondria synthesizing pre-Cox2, but not from mitochondria lacking a functional COX2 gene. Thus, the mitochondrial gene product Cox2 appears to bridge a dynamic interaction between Cox20 and Cox18.
These data are consistent with the findings of Hell et al. (2000), who reported that a significant fraction of the Cox2 C-tail, newly synthesized in isolated cox20Δ mitochondria, remained on the matrix side of the inner membrane after pulse labeling, while the N-tail was efficiently exported. Our findings that the vast majority of Cox2 molecules accumulated in a cox20Δ mutant have an N-out, C-in topology, while both N- and C-tails of Cox2 are efficiently exported in other mutants that lack cytochrome c oxidase activity, argue that the export defect in cox20Δ mitochondria is not a secondary effect due to reduced membrane potential, as previously suggested (Hell et al. 2000).
On the basis of these results, we propose that Cox20 binding to Cox2 accelerates dissociation of newly exported Cox2 from the Cox18 translocase on the IMS side of the inner membrane. Thus, while not directly involved in membrane translocation per se, Cox20 is required for efficient cycling of the Cox18 translocase.
Unassembled Cox2 is largely degraded by the i-AAA protease Yme1 (Nakai et al. 1995; Pearce and Sherman 1995; Weber et al. 1996; Graef et al. 2007), which is bound to the inner membrane facing the IMS (Leonhard et al. 1996). Yme1 is associated with two other proteins, Mgr1 and Mgr3, that recognize and deliver at least some substrates for degradation by Yme1 (Dunn et al. 2006; Dunn et al. 2008). A study of interspecies Yme1 chimeras comprising combinations of homologous domains of the S. cerevisiae and Neurospora crassa proteins identified Cox20 as a factor influencing the pathways by which unassembled Cox2 substrate could reach the i-AAA proteolytic domain (Graef et al. 2007).
Our results strongly suggest that a third critical chaperone function of Cox20 is to protect as-yet-unassembled Cox2 from degradation by the i-AAA protease complex during the assembly process downstream of export. Furthermore, they indicate a strong dependence of Cox2 degradation upon the putative substrate recognition factors Mgr1 and Mgr3, at least in the absence of Cox20. In cells lacking Cox20, the steady-state level of accumulated pre-Cox2 was extremely low, and those cells exhibited a tight nonrespiratory phenotype. However, in cells lacking Cox20 as well Yme1, Mgr1, or Mgr3, the steady-state levels of pre-Cox2 and mature Cox2 were increased, and those cells exhibited weak respiratory growth reflecting the assembly of low levels of cytochrome c oxidase. Consistent with these findings, Hell et al. (2000) observed that virtually all residual unassembled pre-Cox2 or mature Cox2 was associated with Cox20 when cytochrome c oxidase assembly was blocked by either an imp1Δ or a cox4Δ mutation, respectively. The C-terminal domain of Cox20 is basic, which may facilitate its interaction with the acidic N- and C-tails of Cox2 following their export to the IMS.
Cox20 is required for normal rates of pre-Cox2 processing by the Imp1 subunit of the inner membrane protease. However, slow Imp1-dependent processing does occur in the absence of Cox20 if pre-Cox2 is stabilized by loss of the i-AAA protease. In contrast, the requirement for Imp1 is not bypassed by elimination of the i-AAA protease.
Although Cox20 is also required for normal export of the Cox2 C-tail domain, low levels of export and assembly into cytochrome c oxidase are detectable in the absence of Cox20 if i-AAA activity is also absent. On the basis of the model that Cox20 promotes dissociation of exported Cox2 C-tail from the Cox18 translocase, we suggest that in the absence of Cox20 the i-AAA protease rapidly degrades unprotected Cox2 C-tail that is associated with Cox18. In the absence of both Cox20 and the i-AAA protease, slow dissociation of exported Cox2 from Cox18 can occur and leads to limited cytochrome c oxidase assembly. Consistent with these ideas, we have found that respiratory growth of a cox20Δ yme1Δ double mutant is improved by overproduction of Cox18 (unpublished results).
Supekova et al. (2010) recently described a recoded and modified version of the COX2 gene that supports respiratory growth when allotopically expressed from the yeast nucleus in a cox2Δ mutant. As expected, this gene, termed here N-COX2, bypassed the requirement for Pet111, which normally activates translation of the COX2 mRNA inside mitochondria. Interestingly, it also bypassed the requirement for the Cox2 C-tail-specific export translocase Cox18. This finding suggests that the cytoplasmically synthesized C-tail domain of allotopically expressed pre-Cox2 remains in the intermembrane space during import into mitochondria, rendering Cox18 superfluous. However, N-COX2 did not bypass the requirement for Cox20. Since expression of N-COX2 required Imp1, Cox20 is presumably required to chaperone the cleavage of imported pre-Cox2 by the inner membrane protease. Cox20 may also be required for stabilization of unassembled imported Cox2, and possibly chaperoning its further assembly.
Previous work in this laboratory has shown that overproduced Oxa1 can partially bypass the requirement for Cox18 in export of the Cox2 C-tail domain, but not the requirement for Cox18 in cytochrome c oxidase assembly (Fiumera et al. 2009). This observation suggested that in addition to translocating the Cox2 C-tail through the inner membrane, Cox18 normally delivers the C-tail in a fashion promoting assembly, possibly by interacting with downstream assembly factors. This study supports this idea, and indicates that Cox20 may be an assembly factor that can acquire the exported Cox2 C-tail from Cox18, but not from overproduced Oxa1. (On the basis of the experiments with N-COX2, we postulate that Cox20 can interact with the Cox2 C-tail imported from the cytoplasm in the absence of Cox18.) Interestingly, mgr1Δ or mgr3Δ mutations, but not yme1Δ, allowed overproduced Oxa1 to promote weak cytochrome c oxidase assembly and respiratory growth in the absence of Cox18. These findings suggested that in the absence of the substrate recognition factors Mgr1 and/or Mgr3, the AAA+ protein Yme1 could function as a chaperone for Oxa1-exported Cox2, instead of degrading it (Fiumera et al. 2009). In light of these results, it is tempting to speculate that the role of Cox20 vis-à-vis the Cox18 translocase is carried out by latent Yme1 chaperone activity when the Cox2 C-tail is aberrantly exported by overproduced Oxa1.
Cox18 is highly conserved. Indeed the homologous proteins from humans (Gaisne and Bonnefoy 2006), fission yeast (Gaisne and Bonnefoy 2006), N. crassa (Funes et al. 2004), and E. coli (Preuss et al. 2005) can partially complement a cox18Δ in S. cerevisiae at the level of respiratory growth. A putative Cox20 homolog (NCBI NP_932342.1) has also been identified in humans (Herrmann and Funes 2005). A mammalian ortholog of Cox20 could not have a role in Cox2 processing since mammalian Cox2 proteins are not synthesized as precursors (Steffens and Buse 1979; Anderson et al. 1982). However, a mammalian Cox20 protein could participate with Cox18 in Cox2 C-tail export as well as Cox2 stabilization and assembly into cytochrome c oxidase.
We thank F. Supek and P. G. Schultz for the plasmid pOXA1-W56R-ADH1. L.E.E. and S.A.S were supported in part by National Institutes of Health (NIH) Training Grant T32 GM007617. This study was supported by NIH grant R01 GM29362 to T.D.F.
Communicating editor: S. Dutcher
- Received October 10, 2011.
- Accepted November 2, 2011.
- Copyright © 2012 by the Genetics Society of America