Evidence for the association of yeast mitochondrial ribosomes with Cox11p, a protein required for the CuB site formation of cytochrome c oxidase
- First Online:
- Cite this article as:
- Khalimonchuk, O., Ostermann, K. & Rödel, G. Curr Genet (2005) 47: 223. doi:10.1007/s00294-005-0569-1
- 120 Views
Cytochrome c oxidase is the terminal enzyme of the mitochondrial (mt) respiratory chain. It contains copper ions, which are organized in two centres, CuA and CuB. The CuA site of subunit Cox2p is exposed to the mt intermembrane space, while the CuB site of subunit Cox1p is buried in the inner mt membrane. Incorporation of copper into the two centres is crucial for the assembly and activity of the enzyme. Formation of the CuB site is dependent on Cox11p, a copper-binding protein of the mt inner membrane. Here, we experimentally prove that Cox11p possesses a Nin–Cout topology, with the C-terminal copper-binding domain exposed in the mt intermembrane space. Furthermore, we provide evidence for the association of Cox11p with the mt translation machinery. We propose a model in which the CuB site is co-translationally formed by a transient interaction between Cox11p and the nascent Cox1p in the intermembrane space.
KeywordsCytochrome c oxidaseCopper metabolismMitochondriaCox11pMitoribosomesSaccharomyces cerevisiae
Copper is an essential nutrient required for the activity of a number of enzymes with diverse biological functions (for a review, see Puig and Thiele 2002). The delivery of copper to cytochrome c oxidase (COX), the terminal enzyme of the mitochondrial (mt) respiratory chain, is one of the crucial steps for the assembly and function of the enzyme (Poyton and McEwen 1996). Copper ions of COX are organized in two centres. The binuclear, mixed-valency CuA site is formed by two copper ions and is located in the intermembrane space (IMS)-exposed domain of subunit Cox2p. The heme a3-CuB site, which consists of one Cu ion situated next to heme a3 in the subunit Cox1p, is buried in the inner mt membrane (IMM) 13 Å below the surface (Tsukihara et al. 1995).
In Saccharomyces (Sac.) cerevisiae, several proteins involved in copper delivery to COX have been identified, among them Cox17p, Sco1p, Sco2p and Cox11p. Sco1p, which is anchored in the IMM with a copper-binding domain exposed in the IMS (Krummeck 1992; Beers et al. 1997), is believed to be engaged in the insertion of copper into the CuA site of COX (Glerum et al. 1996; Lode et al. 2000). Sco2p, which is highly similar to Sco1p but not essential for respiratory growth, may also be involved in this step (Glerum et al. 1996; Lode et al. 2002). However, its exact role has not yet been defined.
Cox17p was postulated to act as a copper shuttle between the cytosol and the IMS (Beers et al. 1997). However, Maxfield et al. (2004) demonstrated that a modified form of Cox17p, which is anchored in the IMM and exposed to the IMS, is also functional, arguing against the proposed role as copper shuttle. Recently, Horng et al. (2004) were able to show, by in vitro studies and by expressing a soluble derivative of Sco1p in the yeast cytosol, that Cox17p donates copper to Sco1p. The same study revealed that Cox17p is also the copper donator of Cox11p. This evolutionarily conserved protein is a copper-binding protein which plays an essential role in delivering copper to the CuB site of COX (Tzagoloff et al. 1990; Hiser et al. 2000; Carr et al. 2002; Carr and Winge 2003). Cox11p has a molecular mass of 34 kDa (28 kDa for the mature form, according to Tzagaloff et al. 1990) and is tightly associated with the IMM. As Cox17p transfers copper to Cox11p, its copper-binding motif in the C-terminal part is likely to be exposed to the IMS, as is the case for Sco1p. Mutations in COX11 elicit COX deficiency and the finding that COX11 is allelic to the PSO7 gene suggests an additional role in the oxidative stress response (Pungartnik et al. 1999).
A first indication for an essential role of COX11p in the formation of the CuB centre of COX came from studies on the Rhodobacter sphaeroides homologue of Cox11p (Hiser et al. 2000). These studies also revealed that RsCox11p is necessary for proper alignment of hemes a and a3 and the stability of the COX magnesium/manganese centres. Recent investigations on Sac. cerevisiae Cox11p showed that it forms a homodimer via its C-terminal domain. Each monomer co-ordinates one Cu(I) via three thiolate ligands (Carr et al. 2002). The two copper ions in the dimer exist in a binuclear cluster and appear to be ligated by three highly conserved cysteines. Mutations in any of these residues cause a reduction in Cu(I)-binding and subsequent respiratory deficiency. In addition, a conserved methionine (M224) is necessary for copper binding. Presumably, it plays a role in metal ion transfer or stabilization of the copper site without binding in the inner co-ordination sphere. The mutational data are in agreement with a high-resolution structure of the Cox11p homologue of Sinorhizobium meliloti (Banci et al. 2004).
The genome sequence of the fission yeast Schizosaccharomyces (Sch.) pombe revealed the presence of two homologues of COX11 (cox11+, cox11b+). Similarly, two COX11 homologues have been detected in the human genome (HsCOX11), one of which, however, is predicted to be a pseudogene (Petruzella et al. 1998). Interestingly, both Sch. pombe proteins (SpCox11p, SpCox11b p) contain long N-terminal extensions of more than 500 amino acids, which exhibit a significant degree of identity to Sac. cerevisiae Rsm22p. This protein is a component of the mt ribosome and necessary for respiratory growth (Saveanu et al. 2001; Carr et al. 2002; Gan et al. 2002). Fusion of SpCox11p with the Sch. pombe homologue of Rsm22p may hint at a link between the formation of the CuB site of COX and the mt translation machinery.
In this study, we present experimental data demonstrating the Nin–Cout topology of Cox11p in the IMM, thus confirming the prediction of previous studies (Carr and Winge 2003). We further address the putative link between mt translation and formation of the CuB centre. The results of sucrose gradient centrifugation and co-immunoprecipitation experiments suggest that Cox11p may be directly or indirectly associated with mt ribosomes. A chimeric protein, with the N-terminal part derived from Sco1p and the C-terminal part from Cox11p, is non-functional. Therefore the N-terminal part of Cox11p seems to be important for its function.
We discuss that insertion of copper into the CuB site of COX may be directly coupled with Cox1p synthesis via Cox11p.
Materials and methods
Strains and media
Escherichia coli strain XL1-Blue (Stratagene) was used for cloning procedures and plasmid propagation. Media were as described by Sambrook et al. (1989).
The Sac. cerevisiae strains used were BY4741 (MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0), Y06479 (MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0, cox11:KanMX4; EUROSCARF) and YSC1070-662427 (MATα, his3Δ1, leu2Δ0, met15Δ0, COX11-3HA; Open Biosystems). Yeast cells were grown in YPD, YPLac or synthetic minimal medium lacking uracil and supplemented with 1% glucose or ethanol and 3% glycerol as a carbon source (Kaiser et al. 1994).
Constructs and plasmids
Primers used in this study. Introduced sequences of the respective restriction enzymes are in lowercase. Overlapping sequences are shown in italics
Sequence (5′ to 3′)
Introduced restriction site
TAT TTA gga tcc ATG ATA AGA ATA TGT CCC ATT G
TAT TTA ctc gag TTA ATT TGA GTT GTC TTT CCT TG
ACA AGG AAA GAC AAC TCA AAT CTG GTT CCG CGT GGA
TCC ACG CGG AAC CAG ATT TGA GTT GTC TTT CCT TGT
TAT TTA tct aga ATG ATA AGA ATA TGT CCC ATT GTT AGA TCT AAG GTT
TAT TTA ctc gag CTA TTA GCG GCC GCA CTG AGC AGC
TCT TAT TTC TTC AAC GCC ATT TGT GCT CGT
TAT TTA tct aga ATG CTG AAG TTG TCA AGA AGT
ACG AGC ACA AAT GGC GTT GAA GAA ATA AGA
Preparation of mitochondria
Yeast cells were grown at 30°C to the early stationary phase and mitochondria were prepared as described by Daum et al. (1982).
Carbonate extraction of mt proteins
Alkaline extraction of Cox11p-3HA and Sco1N-Cox11Cp-3HA was performed according to Fujiki et al. (1982), with slight modifications. First, 300–500 μg of mt protein were resuspended in 500 μl of 0.1 M sodium carbonate solution, pH 11.5, incubated on ice for 30 min and centrifuged at 165,000 g for 1 h at 2°C. The supernatant and the pellet, which was resuspended in 1 ml of 10 mM Tris-HCl, pH 7.5, were precipitated with 10% trichloroacetic acid (TCA), washed twice with ice-cold 80% acetone, dissolved in SDS-sample buffer, subjected to SDS-PAGE and analysed by immunoblotting.
Proteinase K protection assay
Mitochondria were resuspended in 0.65 M manitol, 20 mM Tris-HCl, pH 7.5, 1 mM EDTA, diluted with 20 mM Tris-HCl pH 7.5, 1 mM EDTA to a final manitol concentration of 0.1 M and incubated on ice for 25 min. The resulting mitoplasts were spun down at 12,000 g for 10 min at 2°C, resuspended in 0.65 M manitol, 20 mM Tris-HCl, pH 7.5, 1 mM EDTA and aliquots of 100 μg were used for Proteinase K treatment.
Proteinase K (1–200 μg ) was added to mitoplasts in the presence or absence of 20 mM EDTA, pH 8.0. The addition of EDTA was immediately before Proteinase K treatment.
Samples were incubated on ice for 20 min and the reaction was stopped by the addition of 4 mM PMSF (Sigma, Mo., USA) or AEBSF (AppliChem). The pellets obtained after centrifugation at 20,000 g for 10 min at 2°C were dissolved in SDS-sample buffer, subjected to SDS-PAGE and analysed by immunoblotting.
This analysis was performed according to Szyrach et al. (2003), with slight modifications. mt proteins (500–1,500 μg) were lysed in 1% or 2% digitonin, 20 mM Tris-HCl, pH 7.2, for 15 min at 4°C in the presence of 10 mM MgSO4. The extracts were cleared by centrifugation for 10 min at 18,000 g at 2°C, loaded onto continuous 20–40% sucrose gradients (5 ml) containing 20 mM Tris-HCl, pH 7.2, 0.1% digitonin, 20 mM dithiothreitol (DTT), 4 mM AEBSF and 10 mM MgSO4 and centrifuged at 148,000 g for either 1 h or 4 h at 2°C. Nineteen fractions of 270 μl were collected and aliquots of 4 μl from each fraction were taken to determine the absorption at 260 nm, to establish the mitoribosomal profile. The proteins were precipitated by 10% TCA and subjected to SDS-PAGE and Western blot analysis.
mt proteins (300 μg) were isolated from the cox11Δ strain Y06479 transformed with p416Sc COX11HA and lysed with 1% digitonin, 20 mM Tris-HCl, pH 7.2, for 15 min at 4°C in the presence of 10 mM MgSO4. The extract was cleared by centrifugation for 10 min at 18,000 g at 2°C and incubated with an anti-HA affinity matrix (Roche, USA) for either 1 h or overnight on a rotator at 4°C. The beads were washed according to the protocol of the manufacturer, resuspended in SDS-sample buffer, boiled, spun down and the supernatant was subjected to SDS-PAGE and analysed by Western blotting.
Western blot analysis
Separation of proteins by SDS-PAGE was performed as described by Laemmli (1970). The proteins were transferred onto a polyvinylidene difluoride membrane (Millipore or Amersham Pharmacia Biotech, Chalfont, UK) and probed with antibodies directed against the HA epitope (Roche), Cox2 and Cox3p (Molecular Probes), aconitase (Aco1p; a kind gift from R. Lill, Marburg), Sco1p (Buchwald et al. 1991), alcohol dehydrogenase (Adh1p; kindly gifted by C. Walch-Solimena, Dresden), MrpL36p (kindly provided by J.M. Herrmann, Munich) and Pet123p (a kind gift from T.D. Fox, Ithaca, N.Y., USA). Bound antibodies were detected with horseradish peroxidase-conjugated secondary antibodies and visualised with the ECL-Plus kit (Amersham Pharmacia Biotech).
Cloning procedures and standard DNA techniques were performed as described by Sambrook et al. (1989). Yeast cells were transformed by the lithium acetate procedure (Schiestl and Gietz 1989). Sequencing was done using the dideoxy-chain termination method (Sanger et al. 1977) with 5′ IRD 800-labeled primers (MWG) and the Thermo Sequenase fluorescent labeled primer cycle sequencing kit with 7-deaza-dGTP (Amersham Pharmacia Biotech).
Cox11p is an integral membrane protein with Nin–Cout topology
As outlined in the Introduction, Cox11p is reported to be firmly associated with the IMM (Tzagoloff et al. 1990). In line with this observation, the structure of the Sin. meliloti Cox11p homologue revealed the presence of a transmembrane domain (Banci et al. 2004). Bioinformatic analysis of Sac. cerevisiae Cox11p using the SMART program (Letunic et al. 2002) predicts that the amino acid residues 83–105 form a single transmembrane helix.
We applied the method of alkaline extraction to define whether Cox11p is an integral membrane protein or loosely associated with the membrane (Fujiki et al. 1982).
Mitochondria of strain YSC1070-662427 expressing Cox11p-3HA from its authentic promoter were treated with 0.1 M sodium carbonate as described in the Materials and methods.
A Proteinase K protection assay was performed to experimentally prove the predicted Nin–Cout topology (see Introduction). Mitochondria were purified from strain YSC1070-662427, converted to mitoplasts and treated with increasing amounts of Proteinase K, as described in the Materials and methods.
We were interested to test whether the accessibility for Proteinase K of Cox11p-3HA is influenced by bound copper. To this end, the Proteinase K protection assay was performed in the presence of 100 mM DTT or in the presence of 20 mM EDTA (data not shown). The results were identical to that obtained in the absence of DTT or EDTA (Fig. 2b). We conclude that the Proteinase K resistance of Cox11p is not affected by bound copper.
Cox11p co-fractionates with mt ribosomes in sucrose gradients
Aconitase (Aco1p) was mainly detected in the low-density top fractions, whereas the membrane protein Cox2p was mainly present in fractions of higher density. The highest concentration of Cox2p was observed in fraction 5, which may reflect the assembled COX.
MrpL36p and Pet123p were detected in fractions of low and middle density (fractions 2–4, 7–9) and those of the highest density (fractions 16–18). The distribution probably reflects the free proteins (top fractions) and both disassembled (intermediate fractions) and assembled ribosomes (bottom fractions), indicating that the conditions used do not allow a clear distinction between fully assembled ribosomes and their subunits. The distribution of Cox11p-3HA was very similar to that of MrpL36p, with the highest concentration present in the top fractions and a significant portion in the bottom fractions. As this profile differs from that observed for Cox2p, it can be shown that the presence of Cox11p in the high-density fractions is not due to contamination with membrane fragments.
To exclude the possibility of protein aggregation, we modified the gradient conditions by shortening the time of centrifugation to 1 h (Fig. 3b). Again, the distribution of Cox11p-3HA was similar to that of MrpL36p and Pet123p (Fig. 3b). The accumulation of the various proteins in fraction 6 might result from membrane contamination in this preparation.
The finding of a fraction of Cox11p co-migrating with mitoribosomes suggests a direct or indirect interaction of the copper chaperone with the translation machinery.
Co-immunoprecipitation of Cox11p and MrpL36p
Cox11p-3HA can be detected in the lysate, in unbound material after incubation with the matrix and in the precipitated fraction (Fig. 4, upper panel, lanes 2, 3, 7), but not in the negative control (non-incubated beads; lane 1). Antibodies directed against aconitase (Aco1p), Cox2p and Cox3p were used as controls. None of these proteins seems to interact with Cox11p, because in no case was a signal detected in lane 7. Incubation with antibodies directed against MrpL36p, however, yielded a faint signal of the respective molecular mass (20 kDa) in lane 7. A stronger signal was obtained after overnight incubation (Fig. 4, right panel). The detection of Pet123p yielded an extremely weak signal, detectable only upon long exposure (data not shown). The result is in favour of the proposal that Cox11p interacts directly or indirectly with the mt protein synthesis machinery.
Replacement of the N-terminal part of Cox11p by its Sco1p counterpart results in respiratory deficiency and lack of association with mitoribosomes
The N-terminal part of Cox11p possesses a highly charged amino acid stretch adjacent to the transmembrane helix. Such stretches have been reported to participate in protein–protein interactions, particularly with mitoribosomes (Szyrach et al. 2003). To test whether this is also true for Cox11p, we replaced the N-terminal moiety of Cox11p (including the TM domain) by the respective counterpart of Sco1p, which has a similar topology. Figure 5a shows a scheme of the chimeric protein (Sco1N-Cox11Cp-3HA).
The Δcox11 transformants expressing the chimera are respiratory-deficient at all temperatures tested (25, 30, 37°C), demonstrating that the N-terminal part of Cox11p is important for its function.
Interestingly, the distribution profile of the Sco1N-Cox11Cp-3HA differs from that of the native Cox11p in that it does not co-migrate with mitoribosomes (data not shown).
Our data show that the N-terminal part of Cox11p is important for its function. The results are in line with the idea that the N −terminal part of Cox11p defines a site which is crucial for the observed association with mitoribosomes.
In this paper, we show by proteinase K digestion of mitoplasts that Cox11p has a Nin–Cout topology in the IMM. This result confirms the prediction of Carr et al. (2002) that the copper-binding site in the C-terminal part of Cox11p protrudes into the IMS. Recent data by Horng et al. (2004), which were obtained by in vitro experiments and a yeast cytosolic expression system, strongly suggest that the primary copper donator of Cox11p is Cox17p. This copper-binding protein is partly present in the IMS and has been shown to be functional even when anchored in the IMM (Maxfield et al. 2004). Therefore, it is very likely that a direct interaction between Cox17p and Cox11p in the IMS mediates the copper transfer. However, Horng et al. (2004) were not able to detect a stable interaction of these proteins. Possibly, these interactions exist only transiently and require cross-linking in order to be detected.
Cox11p exhibits a number of similarities with Sco1p, which is engaged in the formation of the CuA site. Like Sco1p, Cox11p exhibits a Nin–Cout topology. Both proteins are anchored by a single transmembrane domain in the IMM and form homodimers (Nittis et al. 2001; Carr et al. 2002). Both Cox11p and Sco1p possess a copper-binding site in the IMS-exposed C-terminal part and receive copper ions from Cox17p. However, the inability of crosswise complementation and the observation that both proteins seem not to interact with each other leads to the conclusion that the processes of formation of the CuA and CuB centres appear to be mechanistically independent (Leary et al. 2004).
How does Cox11p mediate insertion of Cu(I) into the CuB site of COX, which is deeply buried in the IMM? One possibility is that the incorporation of copper into the CuB site may occur during the synthesis of Cox1p. The observation that the Sch. pombe homologues contain N-terminal extensions with significant homology to the Sac. cerevisiae mitoribosomal protein Rsm22p may hint at a link between mt translation and formation of the CuB site. Fusion of two genes to yield a fusion protein with the activities of both single proteins is not without precedent in Sch. pombe and can hint at a co-operative action of the two proteins. For example, Cox15p, which is involved in the biosynthesis of heme a, is fused to Yah1p, an enzyme engaged in electron transfer in mitochondria (Barros et al. 2001; Bureik et al. 2002). The functional link between both proteins became evident from the finding that Yah1p acts as the electron acceptor for Cox15p in the course of heme O oxidation (Barros et al. 2002; Carr and Winge 2003).
In the case of Rsm22p, immunological methods have so far provided no experimental proof of interaction with Cox11p. However, our data indicate an association of Cox11p with mitoribosomes. We observe that part of Cox11p co-fractionates with the fraction of assembled mitoribosomes in sucrose gradients. It seems unlikely that Cox11p is part of a distinct high-molecular-weight complex of similar size as mitoribosomes. The detection of Cox11p but not Cox2p in the bottom fractions of the gradient clearly demonstrates that the presence of Cox11p is not due to membrane vesicles. The different fractionation profiles of Cox2p and Cox11p indicate that Cox11p is not associated with the COX complex.
A second line of evidence for an association of Cox11p with mitoribosomes comes from our finding of the co-immunoprecipitation of Cox11p and MrpL36p, a constituent of the large mitoribosomal subunit. It remains to be clarified whether a large subunit alone or the entire ribosome is required for this interaction. A recent proteomic approach to identify components of the mitoribosomes by mass spectrometry failed to identify Cox11p (Gan et al. 2002). Interestingly, other ribosome-associated proteins like Oxa1p (Jia et al. 2003; Szyrach et al. 2003) or translational activator proteins (Krause-Buchholz et al. 2004) were also not identified in this study. Obviously, the conditions used for the isolation and purification of mitoribosomes used in this approach were too stringent to maintain the association with peripherally associated proteins. It seems likely that the interaction of Cox11p with mitoribosomes is weak and possibly indirect, mediated by additional components. Cox11p possesses only a short N-terminal part protruding into the mt matrix. The site of potential interactions with the mt ribosome or “linker” proteins is therefore limited to this region, or—in case of a membrane protein as a “linker”—to the transmembrane segment. Currently, we are studying the effect of the N-terminal deletions of Cox11p on its ability to interact with mitoribosomes.
The authors would like to thank R. Lill (Marburg) T.D. Fox (Ithaca, N.Y.), C. Walch-Solimena (Dresden) and J.M. Herrmann (München) for antibodies, M. Pielenz for plasmid pJR1-31XL ScSCO1 and U. Krause-Buchholz for helpful comments and advice.