Journal list menu
Submitochondrial distributions and stabilities of subunits 4, 5, and 6 of yeast cytochrome oxidase in assembly defective mutants
Abstract
The concentration and submitochondrial distribution of the subunit polypeptides of cytochrome oxidase have been studied in wild type yeast and in different mutants impaired in assembly of this respiratory complex. All the subunit polypeptides of the enzyme are associated with mitochondrial membranes of wild type cells, except for a small fraction of subunits 4 and 6 that is recovered in the soluble protein fraction of mitochondria. Cytochrome oxidase mutants consistently display a severe reduction in the steady-state concentration of subunit 1 due to its increased turnover. As a consequence, most of subunit 4, which normally is associated with subunit 1, is found in the soluble fraction. A similar shift from membrane-bound to soluble subunit 6 is seen in mutants blocked in expression of subunit 5a. In contrast, null mutations in COX6 coding for subunit 6 promote loss of subunit 5a. The absence of subunit 5a in the cox6 mutant is the result of proteolytic degradation rather than regulation of its expression by subunit 6. The possible role of the ATP-dependent proteases Rca1p and Afg3p in proteolysis of subunits 1 and 5a has been assessed in strains with combined mutations in COX6, RCA1, and/or AFG3. Immunochemical assays indicate that another protease(s) must be responsible for most of the proteolytic loss of these proteins.
1 Introduction
Cytochrome oxidase mutants of Saccharomyces cerevisiae share a common phenotype, characterized by pronounced reductions in the steady-state concentrations of some constituent polypeptides, particularly hydrophobic products of the mitochondrial genetic system. The loss of these constituents is the result of proteolysis and occurs even in strains unable to complete assembly of the holoenzyme for lack of a prosthetic group [1-3]. These observations suggest that only the fully formed complex is protected from the action of endogenous proteases. The exceptions are mutations in subunits 6a, 6b, and 8 that do not affect the stability of the enzyme [4-6].
The X-ray crystallographic structure of bovine cytochrome oxidase has provided a wealth of information about the subunit interactions in the mitochondrial enzyme [7]. The yeast enzyme probably has a very similar structure since its subunit composition is almost identical to that of bovine cytochrome oxidase [8]. In the present study we attempt to explain the biochemical properties of different cytochrome oxidase mutants in the context of the recently reported structure.
2 Materials and methods
2.1 Yeast strains and growth media
The strains of S. cerevisiae used in this study are listed in Table 1 . The compositions of the media used to grow yeast have been described [1]. Mitochondria were isolated by the method of Faye et al. [10]from yeast grown in 2% galactose, 1% yeast extract, and 1% peptone.
| Strain | Genotype | Source |
|---|---|---|
| W303-1A | a ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 | a |
| W303-1B | α ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 | a |
| W303ΔCOX4 | α ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 cox4∷URA3 | This study |
| W303ΔCOX5 | α ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 cox5∷HIS3 | This study |
| W303ΔCOX6 | α ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 cox6∷URA3 | This study |
| W303ΔCOX5ΔCOX6 | α ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 cox5∷HIS3 cox6∷URA3 | aW303ΔCOX5×W303ΔCOX6 |
| aW303ΔRCA1 | a ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 rca1∷URA3 | [9] |
| aW303ΔAFG3 | a ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 afg3∷HIS3 | [9] |
| aW303ΔRCA1ΔAFG3 | a ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 rca1∷URA3 afg3∷HIS3 | [9] |
| W303ΔRCA1ΔCOX6 | α ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 rca1∷URA3 cox6∷URA3 | W303ΔCOX6×aW303ΔRCA1 |
| W303ΔAFG3ΔCOX6 | a ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 afg3∷HIS3 cox6∷URA3 | W303ΔCOX6×aW303ΔAFG3 |
| W303ΔRCA1ΔAFG3ΔCOX6 | a ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 rca1∷URA3 afg3∷HIS3 cox6∷URA3 | W303ΔCOX6×aW303ΔRCA1ΔAFG3 |
| aW303ΔPET309 | a ade2-1 his3-1,15 leu2-3,112 trp1-1 ura3-1 pet309∷HIS3 | This study |
- a aDr. Rodney Rothstein, Department of Human Genetics, Columbia University.
2.2 Strain constructions
Null alleles of PET309, COX4, COX5, and COX6 were introduced in the respiratory competent strains S. cerevisiae W303-1A or W303-1B. PET309 was disrupted by removing a 1.2 kb fragment between the two BamHI sites in the gene and replacing it with a 1.8 kb BamHI fragment containing HIS3. The COX4 disrupted allele was made by deleting the internal 114 bp KpnI fragment and inserting a 1 kb KpnI fragment containing URA3 in the gap. COX5 was disrupted by removal of a 576 bp BglII–PstI, of which 298 bp consisted of coding sequence and inserting the HIS3 gene on a 1 kb BamHI fragment. The COX6 disruption was a simple insertion of a 1 kb HindIII fragment with URA3 at the HindIII site in the COX6 coding sequence.
To make the lacZ fusion, a 571 bp PstI fragment (−273 to +298), containing the upstream and part of the coding region of COX5, was ligated to the PstI site of the integrative plasmid YIp366 [11]. The resultant plasmid (pG46/ST20) with an in-frame fusion to lacZ was linearized at the unique BstEII site in the LEU2 gene of the plasmid. The linear plasmid was integrated at the leu2 loci of W303-1A, W303ΔCOX5 and W303ΔCOX5ΔCOX6. Three independent transformants, selected for leucine prototrophy, were assayed for β-galactosidase activity [12].
2.3 Miscellaneous procedures
Standard procedures were used for the preparation and ligation of DNA fragments, and for transformation and recovery of plasmid DNA from E. coli [13]. Protein concentrations were determined by the method of Lowry et al. [14]. For Western blot analysis, proteins were separated either on 12% [15]or 16.5% acrylamide gels [16]. After transfer to nitrocellulose, the blots were reacted with antibodies specific for different cytochrome oxidase subunits, followed by exposure to 125I-protein A, according to the protocol of Schmidt et al. [17].
3 Results and discussion
3.1 Submitochondrial distribution of 4, 5a, and 6 in cytochrome oxidase mutants
Subunits 4, 5a, and 6 of yeast cytochrome oxidase are encoded by the nuclear genes COX4, COX5a, and COX6, respectively [18-20]. These proteins are part of cytochrome oxidase and are present in a stoichiometry of 1 : 1 : 1 [21]. Subunit 4 is located on the matrix side of the inner membrane. It does not penetrate into the lipid bilayer and is associated with subunits 1 and 3 of the complex [7]. This is also true of subunit 6, except that it interacts with subunits 5a and 7a, both of which span the membrane once and have domains extending into the matrix compartment that provide the anchoring sites for subunit 6. Subunit 1 and 5a are complexed through their hydrophobic transmembrane domains [7].
The distribution of subunits 4, 5a, and 6 between the soluble and membrane phases of mitochondria from respiratory competent yeast was examined by Western blot analysis using an antibody against the holoenzyme. Mitochondria of the respiratory competent haploid strain W303-1B were disrupted by a brief sonic treatment to release the soluble matrix proteins. Immunoblots of the starting mitochondria and of the membrane vesicles and soluble protein fraction show these constituents to be present largely in the membrane vesicles. A small percentage of subunits 4 and 6 (but not subunit 5a), however, is also detected in the soluble fraction (Fig. 1 ). These may represent a pool of unassembled subunits and/or subunits loosely bound to partially assembled complexes. Subunits 7a, 7, and 8 are not separated in the gel and are detected as a single composite band in the membrane fraction.
A similar analysis was carried out on several different mutants blocked in cytochrome oxidase assembly. Mutations in PET309, shown to be necessary for maturation/stability of the mitochondrial subunit 1 mRNA, lead to a deficiency of subunit 1 [22]. Western blot assays of cytochrome oxidase subunits in a pet309 null mutant show that while the mitochondrial concentration of subunit 4 is comparable to that of wild type, it is recovered almost entirely in the soluble protein fraction (Fig. 1). The absence of membrane bound subunit 4 is consistent with its association with subunits 1 and 3. The inability of the pet309 mutant to express subunit 1 also elicits a secondary deficiency in subunits 2 and 5a. The absence of subunit 2 must be the result of proteolysis since pulse-labeling experiments indicate that this constituent is synthesized in pet309 [22]. The additional loss of subunit 5a in pet309 probably stems from a higher turnover rate as well, but the effect is not as pronounced. Subunit 5a has a single transmembrane domain and therefore remains membrane bound in the mutants as well as wild type. Although the concentration of subunit 6 in the pet309 mitochondria is not significantly different from the wild type, like subunit 4 most of it is present in the soluble fraction. A partial explanation for this may lie in the reduced level of subunit 5a with which it normally interacts. A similar phenotype is seen in the cox4 null mutant, which also displays a somewhat lower concentration of subunit 5a and an absence of subunit 2 (Fig. 1).
Yeast have two isoforms of subunit 5 encoded by COX5a and COX5b [23]. Since subunit 5a, expressed from COX5a, is by far the more abundant of the two, mutations in this gene elicit a strong respiratory defect. Nonetheless, expression of the minor subunit 5b homolog from the second gene allows cox5a mutants to make a low but measurable level of cytochrome oxidase. This is also evident from the immunoblots of the mitochondrial fractions isolated from a cox5a null mutant (Fig. 1). In this mutant, the faint signal detected at the position of subunit 5 corresponds to the protein derived from COX5b. Aside from the absence of subunit 2 (for the same reason discussed above), the most striking feature of the cox5a mutant is the almost complete recovery of subunit 6 as a soluble protein. This observation suggests that either the interaction with subunit 7a is too weak to anchor subunit 6 to the membrane or that subunit 7a itself is also proteolytically degraded in this mutant. The fact that a low level of subunit 4 is present in the membrane fraction also indicates some cytochrome oxidase assembly in this mutant.
3.2 Protection of subunit 5a by subunit 6
The steady-state concentration of cytochrome oxidase subunits was also examined in the cox6 null mutant W303ΔCOX6. Antibodies against either the holoenzyme or purified subunit 5 revealed that the cox6 mutant had virtually undetectable levels of subunit 5a in mitochondria and in the post-mitochondrial supernatant fraction. The faint signal detected in mitochondria is probably subunit 5b, since a signal of equal intensity is seen in the cox5a null mutant W303ΔCOX5 (Fig. 2 ).
The severe depletion of subunit 5a in the cox6 mutant background could be due to (1) failure of the protein to be imported, (2) decreased transcription or translation, (3) high rates of proteolytic turnover. The absence of subunit 5a in the post-mitochondrial supernatant fraction makes a transport defect improbable. Northern analysis of polyA-enriched RNA showed COX5a mRNA abundance to be approximately the same in the mutant and wild type, excluding a transcriptional defect (data not shown). The possibility that translation of the COX5a mRNA might be regulated by subunit 6 was tested with lacZ fusions. A lacZ construct containing 273 bp of the COX5 5′-untranslated region was integrated at the leu2 locus of the respiratory competent haploid yeast W303-1A, and of the respiratory deficient mutants W303ΔCOX5 and W303ΔCOX5ΔCOX6. The β-galactosidase activities measured in the three different strains were not significantly different (data not shown), confirming that the cox6 mutation does not affect transcription of COX5a and is also unlikely to exert a regulatory effect on translation of the mRNA.
The results of the Northern analysis and the lacZ fusion experiments argue strongly against a direct role of subunit 6 in expression of subunit 5a, but rather suggest that proteolysis of this protein is the most likely explanation for its absence in cox6 mutants. Protection of subunit 5a by subunit 6 against endogenous proteases would imply that the two proteins are physically associated with one another, even in mutants unable to form the holoenzyme. In pet309 and cox4 mutants subunit 6 is found in the soluble protein fraction of disrupted mitochondria (Fig. 1). This indicates that the interaction of subunits 5a and 6 is too labile to be maintained even under the relatively mild sonic treatment used to fragment mitochondrial membranes. Attempts to detect such a complex by using detergents instead of physical means to disrupt mitochondria were also unsuccessful.
The results of the immunoblot assays are diagrammatically summarized in Fig. 3 . It is of interest to note that the more hydrophobic constituents such as subunits 1, 2, 5a (and probably subunit 3) appear to be more prone to proteolytic degradation in assembly-arrested mutants. Subunits 4 and 6, whose associations with the inner membrane are entirely through contacts with integral membrane proteins, are released into the matrix and are relatively stable. This suggests that partially assembled hydrophobic proteins, perhaps because of their deleterious effect on membrane based processes, are rapidly removed by the action of local proteases.
3.3 Is subunit 5a degradation prevented in rca1 and afg3 mutants?
RCA1/YTA12 and AFG3/YTA10 code for mitochondrial inner membrane proteins [24, 25]that belong to the AAA-family [26]. Recent evidence indicates that Rca1p and Afg3p are subunits of an ATP-dependent protease [27]responsible for proteolytic clearing of improperly assembled proteins of mitochondria. Mutations in these genes have been shown to slow the rate of proteolytic degradation of some inner membrane proteins [28, 29].
To further explore the biochemical basis for the subunit 5a deficiency, and in particular to assess the possible roles of Rca1p and/or Afg3p in degradation of this subunit, the cox6 null mutation was introduced into strains with either single or double mutations in RCA1 and AFG3. These strains were used to measure the steady-state concentration of subunit 5a in mitochondria. The results of the Western blot analyses indicate that mutations in RCA1 or AFG3 alone or in combination do not appear to spare subunit 5a (Fig. 4 ). In agreement with earlier results [9], mutations in the two ATP-dependent proteases by themselves do not appreciably affect the mitochondrial concentration of subunit 5a (Fig. 3).
Subunit 1 is not detected in the rca1 and afg3 single or double mutants. The absence of subunit 1 is very probably due to proteolysis, since pulse labeling experiments indicate that the rca1 mutants synthesize this protein [24]. This is also true of afg3 mutants, although in this case synthesis of subunit 1 is not as efficient as in wild type [28, 30]. The proteolytic loss of subunit 1 in the rca1 and afg3 mutants independent of the cox6 mutation is consistent with earlier evidence indicating that such mutants are defective in assembly of cytochrome oxidase [9, 24]. The lack of a clear effect of the rca1 or afg3 mutations on the steady-state concentrations of subunits 1 or 5a in the cox6 background suggests that a protease(s) other than Rca1p and Afg3p must be responsible for the increased turnover of these membrane constituents when assembly of cytochrome oxidase is blocked.
Acknowledgements
This research was supported by grant GM50187 from the National Institutes of Health, United States Public Health Service. D.M.G. is the recipient of a Medical Research Council of Canada post-doctoral fellowship.
