Oxa1-Ribosome Complexes Coordinate the Assembly of Cytochrome c Oxidase in Mitochondria^*

Yeast Strains and Growth Media

All yeast strains used in this study are derivatives of W303-1A (MAT a, ade2 ura3 leu2 his3 trp1). The OXA1, MDM38, and MBA1 genes were deleted by LEU2-, KanMX-, and HIS3-containing cassettes, respectively. For construction of the Oxa1 linker mutants, the sequences corresponding to the OXA1 promoter and amino acids 1–317 and 318–402 were separately amplified by PCR and cloned into the SacI and XhoI sites of a pRS424 vector. In the resulting Oxa1 sequence, the insertase domain (amino acids 1–317) and the ribosome-binding domain (amino acids 318–402) were separated by restriction sites for BamHI and PstI. These sites were opened to insert the Nsp1 linker fragment sequences that were amplified from the pSF362-pQE80N plasmid; these sequences corresponded to residues 2–101 and 2–201 of the Nsp1^FS mutant protein (23). The Oxa1 variants expressed from the resulting constructs were named Oxa1¹⁰⁰ and Oxa1²⁰⁰, respectively. These plasmids or a control plasmid with the wild-type OXA1 gene was transformed into Δoxa1 single mutants or Δoxa1 Δmba1 double mutants.

Yeast cultures were grown at 30 °C in 1% yeast extract and 2% peptone or in minimal synthetic medium supplemented with 2% glycerol, 2% galactose, or 2% glucose (24). Mitochondria were isolated as described previously (24).

Analysis of Mitochondrial Translation Products

Mitochondrial translation products were radiolabeled in whole cells (in vivo) or in isolated mitochondria (in organello) as described previously (25). To assess the accessibility of cysteine residues, mitochondria were incubated in 0.6 m sorbitol, 0.5 m NaCl, and 20 mm Tris (pH 7.4) in the presence of 1 mm mPEG-12 or mPEG-24 for 15 min at 4 °C. Prolonged incubation at temperatures above 20 °C should be avoided to prevent modification of cysteine residues in the matrix. Modification of cysteine residues was stopped by the addition of 2 mm DTT.

Immunoprecipitation and Co-immunoprecipitation

To isolate the Cox20-containing protein complex, translation products were radiolabeled in 120 μg of isolated mitochondria. Mitochondria were washed and lysed in ice-cold lysis buffer (1% digitonin, 20 mm Tris (pH 7.4), 150 mm NaCl, and 1 mm EDTA). The extract (200 μl) was cleared by centrifugation (10 min, 16,000 × g, 4 °C) before protein A-Sepharose beads and 5 μl of anti-Cox20 serum were added. After incubation for 1 h at 4 °C, the beads were extensively washed and analyzed by SDS-PAGE. For immunoprecipitation of Cox2, mitochondria were lysed under denaturing conditions in 20 μl of 1% SDS that was diluted 25-fold with 0.1% Triton X-100, 20 mm Tris (pH 7.4), 150 mm NaCl, 2 mm PMSF, and 5 mm EDTA.

Blue Native (BN) PAGE and Cytochrome c Oxidase Assembly Assay

To follow assembly of cytochrome c oxidase, radiolabeled precursor proteins were imported into mitochondria. Cox5a and Cox13 were synthesized in the presence of [³⁵S]methionine in reticulocyte lysate (26). Import into isolated yeast mitochondria was performed in import buffer (250 mm sucrose, 10 mm MOPS/KOH (pH 7.2), 80 mm KCl, 2 mm KH₂PO₄, 5 mm MgCl₂, 5 mm methionine, 3% BSA, 2 mm NADH, 2 mm ATP, 5 mm creatine phosphate, and 0.1 mg/ml creatine kinase) at 25 °C. Import was stopped by the addition of a 1% mixture of 2.5 mm antimycin A, 0.25 mm valinomycin, and 5 mm oligomycin in ethanol. After in vitro import, mitochondria were washed with 250 mm sucrose, 20 mm MOPS/KOH (pH 7.2), and 1 mm EDTA and processed for SDS- or BN-PAGE analysis, followed by detection of radiolabeled proteins by digital autoradiography.

For BN-PAGE, mitochondria were solubilized in 1% digitonin, 20 mm Tris-HCl (pH 7.4), 5 mm EDTA, 100 mm NaCl, 10% (w/v) glycerol, and 2 mm PMSF for 30 min at 4 °C. The lysate was cleared at 20,000 × g for 15 min at 4 °C. After the addition of 10× loading dye (5% Coomassie G-250, 500 mm 6-aminohexanoic acid, and 100 mm BisTris (pH 7.0)), the supernatant was separated on a 4–13% polyacrylamide gradient gel.

Insertion of Linkers Causes Growth Defects on Non-fermentable Carbon Sources

Oxa1 shares a hydrophobic core region with bacterial YidC proteins that comprises five transmembrane regions. In contrast to its bacterial homologs, Oxa1 contains a C-terminal stretch region of ∼100 amino acid residues that protrudes into the matrix. This region is of α-helical amphipathic structure and is rich in positively charged residues (11–15). It was speculated that the C-terminal region of Oxa1 tethers the ribosome to the insertion complex so that nascent chains are threaded efficiently across the membrane. To characterize the relevance of the ribosome-binding domain of Oxa1 for membrane biogenesis in mitochondria, we constructed Oxa1 expression plasmids in which we introduced a pair of restriction sites into the OXA1 sequence after codon 317 (Fig. 1A). These sites were used to construct Oxa1 variants into which spacers of 100 and 200 residues were inserted directly C-terminal of the fifth transmembrane domain (Oxa1¹⁰⁰ and Oxa1²⁰⁰). For these linker regions, we chose sequences corresponding to positions 2–101 and 2–201, respectively, of the Phe-to-Ser variant of the nucleofilament Nsp1 (23). In the corresponding Nsp1 sequence, phenylalanine residues were exchanged for serine residues resulting in an intrinsically unstructured nature of this sequence. The reasoning for using the Phe-to-Ser variant of the Nsp1 sequence was to physically separate the membrane domain from the ribosome-binding region of Oxa1 without adding a novel structured domain to the Oxa1 protein.

These Oxa1 variants were introduced into yeast strains lacking either the endogenous OXA1 gene or lacking both OXA1 and MBA1. When mitochondria were isolated from the resulting mutants, the Oxa1 variants could be detected by Western blotting at the expected sizes and at levels that were comparable with the Oxa1 levels in wild-type mitochondria (Fig. 1B).

To test the functionality of these variants, we spotted cultures of these strains on plates containing fermentable (glucose) or non-fermentable (glycerol) carbon sources (Fig. 1C). Whereas all strains grew well on glucose, they differed considerably in their ability to grow under respiratory conditions: All Oxa1 variants complemented the Δoxa1 mutant, indicating that, in the presence of Mba1, the spacers did not compromise mitochondrial protein biogenesis. In contrast, in the mutants lacking Mba1, the spacers in Oxa1 caused a strong growth defect that correlated with the lengths of the insertion: growth of the Oxa1¹⁰⁰ mutant was ∼10-fold reduced, and the Oxa1²⁰⁰ mutant was even more severely affected (Fig. 1C).

Mdm38 is an inner membrane protein that, in addition to Oxa1 and Mba1, binds to mitochondrial ribosomes (51, 52). Expression of Oxa1 in Δoxa1 Δmdm38 double mutants partially repressed the phenotype of the double mutant, confirming earlier studies demonstrating that Δmdm38 cells show reduced respiratory activity. In this case, as in the Δoxa1 single mutant, the spacer variants did not further reduce the ability to grow on glycerol. This points to a rather specific cooperative role for Mba1 and the C-terminal region of Oxa1 in mitochondrial membrane protein biogenesis.

Insertion of Linkers into Oxa1 Leads to Reduced Levels of Respiratory Chain Complexes

The reduced growth of the linker-containing strains pointed to a defect in respiratory activity. To test this more directly, the different strains were grown on galactose medium. Mitochondrial fractions were then isolated, and the oxygen consumption was measured using a Clark electrode (Fig. 2A). The results correlated well with the growth phenotype, showing decreased oxygen consumption with increasing lengths of the linker segments. Oxa1 plays a crucial role in the biogenesis of cytochrome c oxidase, and direct measurements of the cytochrome c oxidase activity indeed revealed strong defects in the mutants. In the Δmba1 mutants expressing Oxa1, the cytochrome c oxidase activity levels were reduced to ∼50%, which correlated well with previous results published for Δmba1 strains (21). In the presence of the linker stretches, the cytochrome c oxidase was further reduced and dropped to ∼28% in the Oxa1²⁰⁰ strain.

Next, we analyzed the steady-state levels of cytochrome c oxidase subunits by Western blotting (Fig. 2, C and D). The insertion mutants showed considerably reduced levels of Cox1 and, to a slightly lesser degree, Cox2, whereas nuclear encoded proteins of the inner membrane (Cox4, Coa1, Coa3, and Atp5) or the matrix (Aco1) were not affected. It should be noted that the Cox2 protein accumulated in the mature form, whereas in Δoxa1 mutants, the maturation of Cox2 is blocked (16, 17). Due to the proteolytic instability of the precursor form, no Cox2 was detectable in Δoxa1 Δmba1 mitochondria (Fig. 2C). From this, we conclude that the insertion of a flexible linker between the membrane domain and the ribosome-binding domain of Oxa1 strongly interferes with the biogenesis or stability of cytochrome c oxidase.

Insertion of Linkers into Oxa1 Does Not Impair Processing of Cox2

The Oxa1 insertase was shown to be required for membrane insertion of Cox2 (16). In yeast, Cox2 is synthesized with an N-terminal leader peptide that is removed by the Imp1 protease after translocation of the N terminus into the IMS (27, 28). Mutants lacking Oxa1 fail to insert Cox2 into the inner membrane, which therefore accumulates in the matrix in its precursor form and is rapidly degraded (16, 29). To follow the maturation of newly synthesized Cox2, we labeled mitochondrial translation products with [³⁵S]methionine in whole yeast cells (in vivo) after blocking cytosolic translation with cycloheximide (Fig. 3A). In cells lacking Oxa1, the levels of Cox1, Cox2, and Cox3 and, to some degree, also those of other translation products, were strongly diminished (Fig. 3A). The small amounts of Cox2 that were synthesized in Δoxa1 cells were present in the precursor form. In cells carrying the linker-containing variants of Oxa1, no changes in the expression patterns were found, showing that even the longer linker does not prevent maturation and hence insertion of Cox2.

Next, we analyzed mitochondrial protein synthesis in the strains lacking Mba1. Δmba1 mutants have been shown to exhibit insertion defects for Cox2, although here still considerable amounts of Cox2 are mature (21). As shown in Fig. 3B, mutants lacking Oxa1 and Mba1 did not process newly synthesized Cox2. However, the expression of linker-containing Oxa1 variants strongly relieved this block, and about half of the produced Cox2 protein was found in the mature form. Interestingly, the block in Cox2 processing was similar for all three Oxa1 variants. From this, we conclude that the defect in the biogenesis of cytochrome c oxidase in the linker-containing Oxa1 mutants is not due to defects in processing of the Cox2 precursor.

Next, we asked whether insertion of the linkers into Oxa1 renders the insertion of Cox2 more dependent on the membrane potential. To test this, we incubated mitochondria isolated from different yeast mutants in the presence of increasing concentrations of valinomycin, which dissipates the proton gradient across the inner membrane. We added [³⁵S]methionine to radiolabel translation products (“in organello translation”). This again showed that about half of the produced Cox2 proteins accumulated in the precursor form in the Δmba1 mutants expressing the Oxa1 variants (Fig. 3C). The degree of Cox2 processing was not further diminished upon the addition of valinomycin. In summary, these data indicate that, in the presence of Mba1, even the 200-residue spacer did not impair the insertion of the N terminus of Cox2 by Oxa1. In the absence of Mba1, some of the Cox2 protein accumulated in the precursor form; however, this species was not considerably increased in the presence of the linker mutants of Oxa1. This suggests that the defect in the biogenesis of cytochrome c oxidase is not caused by impaired processing of newly synthesized Cox2.

Insertion of Linkers into Oxa1 Does Not Prevent Protein Insertion into the Inner Membrane

The observed maturation of the Cox2 precursor is a good indication that the insertion of the N terminus of Cox2 is not impaired in the Oxa1 linker mutants. However, this does not rule out defects in the translocation of the large C-terminal IMS domain of Cox2. In earlier studies, protein translocation was assessed by protease accessibility assays, which can be problematic due to the inherent inefficiency to rupture the outer membrane of respiratory-deficient mutants by hypotonic swelling. We therefore set out to develop a swelling-independent assay to monitor inner membrane protein topology. In bacteria, a topology assay has been established that uses the modification of cysteine residues by membrane-impermeable alkylating reagents (30, 31). To assess the accessibility of protein thiols in mitochondria, we employed the alkylating maleimide derivatives mPEG-12 and mPEG-24 (Fig. 4A), which bind irreversibly and specifically to cysteine residues, thereby adding molecular masses of 0.6 and 1.2 kDa per moiety, respectively. Both molecules are not membrane-permeable. Therefore, we tested whether they are able to enter the IMS, e.g. through the pores of β-barrel proteins in the outer membrane that facilitate the diffusion of molecules of up to 3–5 kDa. Mitochondria were incubated with mPEG-12 and mPEG-24, and the migration of cysteine-containing proteins was analyzed by Western blotting (Fig. 4B). Although we did not observe size shifts for matrix proteins (Yah1 and Mdj1), proteins of the IMS (Cmc1) or inner membrane proteins containing IMS-exposed domains (Cox2) shifted after exposure to these alkylating reagents. With this, we could show that mPEG-12 and mPEG-24 are able to pass the outer mitochondria membrane but cannot cross the inner membrane. Hence, these alkylating reactions specifically modify regions of mitochondrial proteins that are exposed to the IMS.

We next radiolabeled translation products in isolated mitochondria before we added mPEG-12 and mPEG-24 (Fig. 4C). Incubation with the alkylating maleimide compounds reduced the mobility of translation products except for the matrix protein Var1 (which contains one cysteine residue), consistent with the inaccessibility of matrix proteins. The observed size shifts of most hydrophobic translation products indicated that they expose cysteine residues to the IMS. We tested the accessibility of newly synthesized translation products in the different Oxa1 mutants (Fig. 4D). To identify the nature of the different shifted bands, we performed immunoprecipitation experiments with antibodies against Cox2. Cox2 is the translation product that comprises the largest IMS domains; it contains four cysteine residues, two of which are directly adjacent to the transmembrane domains (Fig. 4D, inset). In wild-type mitochondria, mPEG-24 treatment caused a significant shift of Cox2 from 34 to 39 kDa (Fig. 4, C and D, lanes 3 and 4), consistent with a modification of all four cysteine residues. A size shift was also observed with endogenous Cox2 as detected by Western blotting (Fig. 4B). The same fully modified Cox2 species was observed in the Δmba1 strains, regardless which Oxa1 variant was present (Fig. 4D, lanes 8, 12, and 16). In addition, lower migrating bands appeared that presumably were derived from the precursor form of Cox2, in which only one or two cysteine residues were alkylated (indicated by asterisks). From this, we conclude that the Oxa1¹⁰⁰- and Oxa1²⁰⁰-expressing strains contain fractions of Cox2 in which both IMS domains are translocated across the inner membrane, confirming that a defect in membrane insertion is not the cause of the cytochrome c oxidase deficiency of these strains. Moreover, also Cox1 is alkylated in mitochondria of all four strains, indicating that also this protein is inserted into the inner membrane.

Insertion of Linkers into Oxa1 Does Not Prevent Binding of Cox2 to Cox20

Because the reduced levels of cytochrome c oxidase were apparently not caused by defects in the synthesis or membrane insertion of their subunits, we asked whether steps farther downstream in their biogenesis were affected in these mutants. In the biogenesis pathway of Cox2, membrane insertion of the protein is followed by binding to the Cox2-specific chaperone Cox20 (32, 33). To assess the binding of newly synthesized Cox2 to Cox20, we radiolabeled translation products in mitochondria of the different strains, lysed the mitochondria, and performed co-immunoprecipitation experiments with Cox20-specific antibodies or preimmune serum as a control (Fig. 5). In extracts of wild-type mitochondria, the newly synthesized Cox2 and, to a lesser degree, Cox3 were co-isolated with anti-Cox20 antibodies. Similarly, the mature forms of Cox2 and Cox3 were pulled down with Cox20 from extracts of the linker mutants (Fig. 5, arrows). From this, we conclude that insertion of the linkers into the Oxa1 sequence does not block the interaction of Cox2 and Cox3 with Cox20, pointing to a role for Oxa1 in a later step of cytochrome c oxidase assembly.

Insertion of Linkers into Oxa1 Impairs Assembly of the Cytochrome c Oxidase Complex

The strong defect in the biogenesis of cytochrome c oxidase in the Oxa1 spacer mutants is apparently not caused by a block of protein insertion. We therefore asked whether these mutants show defect in later stages of cytochrome c oxidase biogenesis, in particular in the complex assembly. The enzymes of the respiratory chain are associated with large supercomplexes presumably to increase the efficiency of channeling electrons from one complex to the next (34). These supercomplexes can be separated by BN-PAGE. To follow the assembly of cytochrome c oxidase experimentally, we radiolabeled the nuclear encoded subunits Cox5a and Cox13 in vitro and followed the assembly after import into isolated mitochondria by BN-PAGE. Cox5a (Fig. 6A) and Cox13 (Fig. 6B) were efficiently imported into the Oxa1 mutant mitochondria, but their assembly into the supercomplexes of the respiratory chain occurred only with considerable reduced kinetics and efficiency. This indicates that the insertion of linker regions into the Oxa1 sequence caused a defect in complex assembly. The strongly reduced levels of these supercomplexes were confirmed by Western blotting of isolated mitochondria. As shown in Fig. 7A, mitochondria lacking Oxa1 and Mba1 contained almost no detectable supercomplexes. Upon expression of wild-type Oxa1, considerable levels of the different supercomplexes III₂, III₂/IV, III₂/IV₂, and V₂ were present. Whereas the III₂ and V₂ supercomplexes, which contain two cytochrome bc₁ complexes or two ATPase complexes, respectively, were also present in the Oxa1¹⁰⁰ and Oxa1²⁰⁰ strains, the levels of the supercomplexes containing cytochrome c oxidase were severely diminished. Thus, spatial separation of the insertion domain and the ribosome-binding region of Oxa1 did not interfere with the insertion of mitochondrial translation products into the inner membrane but rather with a much later stage in the biogenesis of cytochrome c oxidase, indicating that the ribosome binding to Oxa1 has a critical function in coordinating cytochrome c oxidase assembly in the inner membrane of mitochondria.