ATP22, a nuclear gene required for expression of the F0 sector of mitochondrial ATPase in Saccharomyces cerevisiae.

Expression of the mitochondrial proton-translocating ATPase of Saccharomyces cerevisiae has been shown to depend on chaperones that target the F1 and F0 sectors of this inner membrane complex. Here we report a new gene, designated ATP22 (reading frame YDR350C on chromosome IV), that provides an essential function in the assembly of F0. ATP22 was cloned by transformation of C208/L2, a strain previously assigned to complementation group G99 of a collection of respiration-defective nuclear pet mutants. C208/L2 and the other atp22 mutants have oligomycin-insensitive F1-ATPase, suggesting that the lesion is confined to F0. This is supported by the sedimentation properties of the mutant ATPase and results of immunochemical analysis of F0 subunit polypeptides. Northern analysis of ATPase transcripts and in vivo pulse labeling of the mitochondrial translation products in the mutant indicate normal expression of subunits 6, 8, and 9, the three mitochondrial gene products of F0. Atp22p therefore functions at a post-translational stage in assembly of F0. Localization studies indicate Atp22p to be a component of the mitochondrial inner membrane. Protease protection experiments further indicate that Atp22p faces the matrix side of the membrane where most of the ATPase proteins are located and assembled.

The proton-translocating ATPase of mitochondria consists of two functionally distinct parts, F 1 and F 0 . The F 1 -ATPase catalyzes the reversible synthesis of ATP from ADP and inorganic phosphate. It consists of 5 distinct polypeptides with a stoichiometry of 3␣:3␤:␥:␦:⑀ (1,2). This extrinsic protein of the mitochondrial inner membrane is attached by means of a double stalk to F 0 , the proton-transferring hydrophobic component of the inner membrane (3,4). In most bacteria, F 0 consists of 3 subunits. The exceptions are photosynthetic bacteria that have an F 0 with 4 subunits. The F 0 of Saccharomyces cerevisiae is composed of 9 different polypeptides (5,6). Three other AT-Pase-associated subunits have been shown recently (7) to be required for dimerization of the yeast enzyme. The most abundant constituent of F 0 is a low molecular weight proteolipid (subunit 9 or subunit C) that forms the proton channel of the complex (8,9).
Studies of respiration-deficient yeast mutants have disclosed the existence of at least two nuclear gene products necessary for the expression of the F 1 -ATPase. ATP11 and ATP12 code for chaperones that interact with the ␤ and ␣ subunits of F 1 , respectively, thereby minimizing their aggregation through nonspecific hydrophobic interactions (10,11). A number of nuclear genes have also been implicated in assembly of yeast F 0 . At least 4 genes have been shown to be required for expression of the mitochondrially encoded subunits 6, 8, and 9 (12)(13)(14)(15). Another nuclear gene, ATP10, codes for a subunit 6-specific chaperone of the inner membrane protein (16).
Most ATPase-deficient mutants have an unstable mitochondrial genome causing them to accumulate Ϫ and o mutants. 1 ATPase mutants also incur secondary losses of the bc 1 and cytochrome oxidase complexes. Even though this phenotype is fairly widespread among pet mutants (e.g. leaky mitochondrial protein synthesis mutants), it nonetheless is a good starting point for identifying strains with lesions in the ATPase. In continuing efforts to catalogue nuclear gene products that contribute to the maintenance of respiration-competent mitochondria, we have screened the class of pleiotropic pet mutants (17) for defects in ATPase. In this communication we report a new gene ATP22, which is required for the biogenesis of the ATPase. The encoded product, Atp22p, is a mitochondrial protein that provides an essential function in assembly of the F 0 sector.

MATERIALS AND METHODS
Yeast Strains and Growth Media-The genotypes and sources of the S. cerevisiae strains used in this study are listed in Table I. The compositions of the media for growth of yeast have been described elsewhere (20).
Preparation of Yeast Mitochondria and ATPase Assays-Mitochondria were prepared by the method of Faye et al. (21) except that Zymolyase 20,000 instead of Glusulase was used to convert cells to spheroplasts. ATPase activity was assayed by measuring release of inorganic phosphate from ATP at 37°C in the presence and absence of oligomycin (22). For localization of Atp22p, mitochondria were prepared by the method of Glick (23).
Cloning and Sequencing of ATP22-ATP22 was cloned by transformation of the respiration-deficient mutant C208/L2 (MAT␣ leu2-3,112 atp22-1) with a yeast genomic library consisting of partial Sau3A fragments of nuclear DNA from strain D273-10B/A1 cloned in YEp13 (24). Transformation of 10 8 cells with 10 g of library DNA yielded a single leucine-independent and respiration-competent clone (C208/L2/T1). The plasmid (pG99/T1) conferring respiratory competence to the mutant was amplified in Escherichia coli RR1 and used to subclone the gene. The insert of subclone pG99/ST4 was sequenced by the method of Maxam and Gilbert (25) following single strand separation of 5Ј endlabeled restriction fragments. All the restriction sites used for labeling were crossed from neighboring sites. The sequence of ATP22 is identical to reading frame YDR350C on chromosome IV.
Disruption of ATP22-The following strategies were used to delete most of the ATP22 coding sequence. pG99/ST4 was digested with NdeI, removing a 1.7-kb fragment internal to the gene. The remaining sequence was blunt-ended with Klenow polymerase and ligated to a 12-nucleotide-long BglII linker. The religated plasmid was digested with BglII removing an additional 190 nucleotides of 5Ј sequence and 70 nucleotides of coding sequence. The BglII-gapped plasmid was ligated to a 1-kb BamHI fragment containing the yeast HIS3 gene. The resultant plasmid pG99/ST11 was used to isolate a 1.4-kb linear fragment with the atp22::HIS3 null allele. The null allele was substituted for wild type ATP22 in the respiration-competent diploid strain W303 by the one-step gene replacement method (26). Histidine-independent transformants were sporulated and tetrads dissected to obtain the haploid atp22 null mutants W303⌬ATP22 and aW303⌬ATP22. A second deletion allele was made by first amplifying ATP22 with 153 nucleotides of upstream and 108 nucleotides of downstream sequences. Following cloning in pGEM-T Easy (Promega, Madison, WI), the plasmid was digested with MfeI and AvrII to remove 940 nucleotides of the gene. The gapped plasmid was ligated to an EcoRI-XbaI fragment containing HIS3. This construct was used to transform the respiration-competent haploid strain LL20. Construction of the ATP22-BIO Fusion Gene-The following PCR primers were used to amplify ATP22: 5Ј-TGAAGATCTATTCTGCGCG and 5Ј-GGCGGGATCCATTTAGACTTTTCAACGTC. The 2.1-kb product, containing 90 nucleotides of 5Ј sequence followed by the entire coding sequence except for the termination codon, was digested with BglII and BamHI and ligated in-frame to the 270-nucleotide sequence coding for the biotinylation signal of bacterial transcarboxylase (27) in the shuttle plasmid YEp352Bio6 (28). The fusion gene was transferred as an XmaI-HindIII fragment to the multicopy shuttle plasmid YEp351 (29) and the integrative plasmid YIp351 (29). The latter construct (pG99/ST14) was linearized at the ClaI site of LEU2 in the plasmid and integrated at the homologous locus of the atp22 mutant C208/L2 (26).
Construction of trpE-ATP22 Fusion Gene-The sequence of ATP22 coding for the 243 carboxyl-terminal amino acids was isolated as a 1.2-kb Sau3A-HindIII fragment and cloned in pATH20 linearized with BamHI and HindIII (30). This plasmid expressed a 64-kDa fusion protein consisting of the amino-terminal half of the E. coli anthranilate synthase and the carboxyl-terminal 243 residues of Atp22p. The insoluble fusion protein was solubilized in SDS and partially purified by chromatography on a BioGel A 0.5 column (Bio-Rad) prior to immunization of rabbits.
Miscellaneous Procedures-Standard methods were used for the preparation and ligation of DNA fragments and for transformation and recovery of plasmid DNA from E. coli (31). Proteins were separated on SDS-PAGE in the buffer system of Laemmli (32). Antibody-antigen complexes were visualized Western blots with 125 I-protein A (33) or by a secondary antibody with the SuperSignal chemiluminescent substrate kit (Pierce). Protein concentrations were determined by the method of Lowry et al. (34).

RESULTS
Phenotype of atp22 Mutants-The atp22 mutants reported here are part of a collection of respiration-deficient pet mutants of S. cerevisiae (17). Complementation group G99 of this collec-tion consists of 9 independent isolates all of which were complemented by a o tester strain indicating that the respiratory deficiency stems from recessive mutations in a nuclear gene. The mutants tend to degrade to secondary o or Ϫ mutants as a result of deletion in their mtDNA. The frequency of o / Ϫ mutants with deleted mitochondrial genomes ranged from 5 to 75% in the five strains tested (Table II). Increased instability of mitochondrial DNA is one of the characteristics of ATPase mutants (35,36). The most stable mutant, N417/L2, grew slowly on glycerol/ethanol, probably because the mutation in this strain causes only a partial loss of function.
Assays of ATPase in isolated mitochondria indicated that the mutant enzyme is oligomycin-insensitive (Table II). This phenotype is associated with mutations that prevent the F 1 -ATPase from interacting with F 0 and can be elicited by lesions in subunits of F 0 , proteins required for expression of one of the three mitochondrially encoded subunits of F 0 (12,13), and accessory factors that function in F 0 assembly (36). Additionally, because of the dependence of F 0 assembly on the endogenously expressed subunits, loss of oligomycin sensitivity can also be a consequence of mutations that impair mitochondrial translation (17). Cytoplasmic petite mutants deficient in mitochondrial protein synthesis assemble catalytically active and oligomycin-insensitive F 1 -ATPase, which exists as a soluble matrix protein (38).
The ATPase defect of atp22 mutants is not due to the accumulation of large numbers of o / Ϫ clones. For example, the ATPase activity measured in C290 and N417/L2 was insensitive to oligomycin, even though the percentage of o / Ϫ cells in the cultures used to prepare mitochondria was only 30 and 5%, respectively (Table II). Like other ATPase mutants, atp22 mutants are partially deficient in cytochromes a, a 3 , and b and have reduced NADH and succinate oxidase activities (Fig. 1A, Table II) (13,14).
A requirement of Atp22p for expression of the mitochondrial ATPase genes ATP6 (OLI2), ATP8 (AAP1) (44), and ATP9 (OLI1) was also excluded. Northern analysis of mitochondrial transcripts confirmed the presence of the fully processed ATP9 and the bistronic ATP6/ATP8 mRNAs (Fig. 1B). The two ATP6/ ATP8 transcripts seen in yeast mitochondria result from processing at two different 5Ј sites (43). The mitochondrial ATPasespecific mRNAs were also studied in an atp22 null mutant constructed in strain LL20 which produces ϳ50% o / Ϫ mutants. In this background the ATP9 transcript normalized to the 21 S rRNA was 96% of wild type. The normalized longer ATP6/ATP8 transcript was 90% and the shorter transcript was

ATP22 Is Required for Expression of Mitochondrial ATPase
51% of wild type. The lack of requirement of Atp22p for expression of the mitochondrial ATPase genes is also supported by in vivo translation assays indicating that atp22 mutants synthesize subunits 6, 8, and 9 of F 0 (Fig. 1C). Finally, the sequence of ATP22 does not correspond to any known subunit of F 1 or F 0 (see below). Based on these results, Atp22p is likely to play a role at a post-translational stage of F 0 assembly. Cloning and Disruption of ATP22-The ATP22 gene was cloned by transformation of C208/L2 (MAT␣ leu2-3,112 atp22-1), a derivative of C208, with a yeast genomic plasmid library. One of 4,000 leucine-independent transformants obtained was rescued for the growth defect on non-fermentable carbon sources. The plasmid pG99/T1, responsible for restoration of FIG. 1. Phenotype of atp22 mutants. A, spectra of mitochondrial cytochromes in wild type and an atp22 mutant. Mitochondria of the wild type strain D273-10B/A1 (D273) and the atp22 mutant N417/L2 were extracted with potassium deoxycholate at a protein concentration of 5 mg/ml (39), and difference spectra of the extracts oxidized with potassium ferricyanide and reduced with sodium dithionite were recorded at room temperature. The positions of the ␣-absorption bands of cytochromes a, a 3 , b, c, and c 1 are indicated. B, Northern analysis of mitochondrial ATP6 and ATP9 transcripts. Mitochondria from the wild type strains D273-10B (ATP6 panel) and W303-1A (ATP9 panel), and from the atp22 mutants N417/L2 and C290 were extracted (20), and approximately equivalent amounts of RNA were separated by electrophoresis on a 1% non-denaturing agarose gel. After transfer to DBM paper (41), the blot was hybridized to 32 P-labeled (42) ATP6 and ATP9 probes consisting of the entire genes. The ATP6 probe hybridizes to two transcripts of 4.4 and 4.8 kb, both of which contain ATP8 and ATP6 but differ at their 5Ј end (43). C, the wild type strain D273-10B/A1 and the two atp22 mutants were labeled with [ 35 S]methionine in the presence of cycloheximide to inhibit cytoplasmic protein synthesis (37). The cells were denatured with trichloroacetic acid, extracted with SDS, and the labeled proteins separated by SDS-PAGE on a 17.5% polyacrylamide gel. The proteins were transferred to a nitrocellulose membrane and exposed to x-ray film. The mitochondrial translation products are identified in the margin as follows: ribosomal protein Var1, subunits 1 (Cox1p), subunit 2 (Cox2p), subunit 3 (Cox3p) of cytochrome oxidase, cytochrome b (Cyt. b), and subunit 6 (Atp6p), subunit 8 (Atp8p), and subunit 9 (Atp9p) of the oligomycin-sensitive ATPase. The percentage of o/Ϫ mutants were 63% for C290 and 16% for N417/L2.   (29), and the resultant constructs were tested for their ability to confer respiratory competence to C208/L2. Based on the ability of the subclones to rescue the mutant phenotype, the gene was localized between a HindIII and a BglII site of pG99/ST1 (Fig. 2A). The sequence of this region revealed a single reading frame corresponding to open reading frame YDR350C/TCM10 on chromosome IV. Because TCM10, the previous three letter designation for this gene, is not based on any known function of the protein, we propose ATP22 as a more appropriate alternative designation based on earlier names of mitochondrial ATPase-related genes. A mutant allele of ATP22 was constructed by replacing most of the coding sequence with the yeast HIS3 gene (see "Materials and Methods"). A linear fragment of DNA containing the null allele was used to transform the respiration-competent diploid strain W303. Two independent His ϩ transformants were sporulated, tetrads dissected, and the meiotic products scored for their growth phenotype. Seven complete tetrads obtained from the two transformants (a/␣W303⌬ATP22) showed a 2:2 segregation of growth on non-fermentable substrates (glycerol/ethanol). In each instance the loss of respiratory competence co-segregated with the histidine prototrophy. The respiratory defect of the haploid segregants was complemented by a o mutant but not by atp22 testers, indicating close linkage of the point mutation to the atp22::HIS3 null allele. Two haploid atp22 segregants, W303⌬ATP22 and aW303⌬ATP22, were also checked for stability of their mtDNA. In both cases less than 1% of freshly grown cultures of either strain had wild type mtDNA ( ϩ ). A similar deletion in strain LL20, however, had a less severe effect on mtDNA. In this genetic background only 50% of vegetatively grown cells were o/Ϫ . Nonetheless, because of their high genome instability, atp22 null mutants were of limited usefulness for biochemical studies.
Analysis of ATPase-The absence of oligomycin-sensitive ATPase in atp22 mutants was most consistent with a lesion in the F 0 sector of the complex. This is supported by the sedimentation properties of the ATPase in the atp22 mutant. In agreement with previous studies (45), the ATPase extracted with Triton X-100 from wild type mitochondria sediments as the F 1 -F 0 complex with an estimated mass of 500 kDa (Fig. 3A). The distribution of ATPase activity (48) following centrifugation of a similar extract from the mutant mitochondria indicated a mass of ϳ310 kDa, a value within experimental error of the mass of the yeast F 1 -ATPase determined by this method as reported previously (45).
A lesion in the F 0 sector was also supported by the steadystate concentrations of ATPase subunits. Some components of F 0 , particularly subunit 6, are unstable in mutants that fail to assemble the F 1 -F 0 complex (16). Western analysis indicated that mitochondria of atp22 mutants had normal amounts of the ␣ subunit of F 1 but were almost totally depleted of subunit 6 of F 0 (Fig. 3B). The atp13 (aep2) mutant N9-168, included as a control, also displays the absence of subunit 6, even though this FIG. 3. Characterization of the ATPase in atp22 mutants. A, sedimentation analysis of mitochondrial ATPase. Mitochondria were prepared from the wild type strain D273-10B/A1 (D273) and from the atp22 mutant N417/L2. The preparation of mitochondria and all subsequent manipulations including centrifugation of the sucrose gradients were done at room temperature to prevent cold depolymerization of F 1 (46). The mitochondria were suspended in 20 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 2 mM ATP, and 1 mM phenylmethylsulfonyl fluoride at a protein concentration of 10 mg/ml. To the mitochondrial suspension was added a 10% solution of Triton X-100 to a final concentration of 0.4%. The mixture was centrifuged at 150,000 ϫ g av , and 0.5 ml of each supernatant was mixed with 0.2 mg of ␤-galactosidase and applied to 4.8 ml of a 7-20% sucrose gradient prepared in a buffer containing 10 mM Tris-HCl, pH 7.5, 2 mM ATP, 0.1 mM EDTA, and 0.1% Triton X-100. The gradients were centrifuged for 4 h at 65,000 rpm in a Beckman SW65 rotor and were fractionated into 16 equal fractions, which were assayed for ATPase (0.1 ml) and ␤-galactosidase (0.02 ml) (47). B, Western analysis of F 1 and F 0 subunits. Mitochondria proteins (20 g) from the wild type strain D273-10B/A1 (D273), from the atp22 mutants N417/L2, C290, and C326, and from the atp13 mutant N9-168 were separated by SDS-PAGE on a 12% polyacrylamide gel. Following transfer to nitrocellulose, the blot was probed with antibodies to the ␣ subunit of F 1 (F 1 -␣) and subunit 6 of F 0 (Atp6p). The antibody-antigen complexes were visualized with the SuperSignal chemiluminescent substrate kit (Pierce). The percentages of o/Ϫ mutants were 5% for N417/L2, 30% for C290, 73% for C326, and 17% for N9-168.

FIG. 4. Mitochondrial localization of Atp22p.
A, total mitochondrial proteins (50 g) of the wild type strain W303-1A, the atp22 null mutant aW303⌬ATP22 (⌬ATP22), and the atp22 mutant C208/L1 transformed with pG99/T1 (C208/L1/T1) were separated by SDS-PAGE on a 12% polyacrylamide gel and probed with an antibody against Atp22p. The blot was treated with 125 I-protein A and exposed to x-ray film. The migration of molecular mass standards is indicated in the left-hand margin. B, mitochondrial proteins (40 g) of C208/L2 harboring the ATP22-BIO fusion gene on a high copy plasmid (ST13) and integrated into chromosomal DNA in single copy (ST14) were separated on a 12% gel as in A. Following transfer to nitrocellulose, biotinylated proteins were stained with avian conjugated to peroxidase (50). gene targets subunit 9 (12,13). The low level of subunit 6 detected in N417/L2 is consistent with the slow growth phenotype of this mutant. Because synthesis of subunit 6 is not affected in the mutants (Fig. 1C), its low steady-state level is most likely the result of increased protein turnover as a result of impaired F 0 assembly.
Localization of Atp22p-Atp22p is a 71-kDa basic protein with an overall hydrophilic character. There are, however, several hydrophobic stretches in the protein of sufficient length to qualify for transmembrane domains (not shown). No homologues of Atp22p, except for other species of Saccharomyces, were detected in the most recent protein data banks.
Atp22p was localized in mitochondria immunochemically and by expressing it as a fusion protein containing a carboxylterminal 7-kDa polypeptide with a biotinylation signal (49). The antibody recognized a protein of ϳ70 kDa in the mitochon-drial fraction of a wild type strain but not the atp22 null mutant (Fig. 4A). The greatly increased signal seen in a mutant transformed with ATP22 on a multicopy plasmid further confirmed this band to be Atp22p. Atp22p was not detected in the post-mitochondrial supernatant fraction of wild type yeast or of the transformant (data not shown). A biotinylated protein of the expected size (ϳ7 kDa larger than the native Atp22p) was also detected in mitochondria of the atp22 mutant C208/L2 transformed with the ATP22-BIO fusion gene on a high copy plasmid or integrated in single copy at the LEU2 locus of chromosomal DNA (Fig. 4B). Both transformants were complemented for the respiratory defect indicating the presence of the carboxyl-terminal biotinylated peptide did not interfere with the activity of the protein.
The presence of several hydrophobic domains in Atp22p suggested it might be a membrane protein. This was supported by , and 2 ml of each suspension was disrupted by sonic (Sonic) irradiation for 7 s with a Branson Sonifier microtip. The suspension was centrifuged at 100,000 ϫ g av for 20 min. The supernatant (S) was collected, and the membrane pellet (P) was suspended in a final volume of 2 ml of STE. The submitochondrial membrane fraction was mixed with an equal volume of 0.2 M sodium carbonate (Carb.), incubated on ice for 20 min, and centrifuged at 100,000 ϫ g av for 20 min. The supernatant (S) was collected, and the membrane pellet (P) was resuspended in the starting volume of STE. Mitochondria (40 g of protein) and equivalent volumes of supernatant and submitochondrial membranes were separated on a 12% polyacrylamide gel and processed as in Fig. 4A. B, mitochondria were prepared from the wild type strain D273-10B/A1 by the method of Glick (23). The mitochondria were suspended in 0.6 M sorbitol, 20 mM Hepes, pH 7.4, at a protein 8 mg/ml in 0.6 M sorbitol, 20 mM Hepes, pH 7.5 (SH). Mitoplasts were prepared by diluting the mitochondrial suspension with 8 volumes of 20 mM Hepes, pH 7.5. The mitochondria were also diluted with 8 volumes of 0.6 M sorbitol, 20 mM Hepes, pH 7.5, as a control. Proteinase K (Prot. K) was added to one-half of each sample after dilution to a final concentration of 100 g/ml. After incubation for 60 min on ice, phenylmethylsulfonyl fluoride was added to a final concentration of 2 mM, and the mitochondria and mitoplasts were recovered by centrifugation at 100,000 ϫ g av for 10 min. The pellets were suspended in 0.6 M sorbitol, 20 mM Hepes, pH 7.5, and precipitated by addition of 0.1 volume of 50% trichloroacetic acid and heated for 10 min at 65°C. Mitochondrial (Mt) and mitoplast (Mp) proteins (40 g) were separated by SDS-PAGE on a 12% polyacrylamide gel, transferred to nitrocellulose, and probed with antibody against Atp22p and cytochrome b 2 (Cyt b 2 ). Antibody-antigen complexes were visualized by a second reaction with 125 I-protein A and exposure to x-ray film.
FIG. 6. Solubility properties of Atp22p. A, extraction of submitochondrial particles with potassium deoxycholate (DOC). Mitochondria from the wild type strain W303-1A were converted to submitochondrial particles as in Fig. 5A. The membranes were suspended in 0.6 M sorbitol, 20 mM Tris-HCl, pH 7.5, 0.5 mM EDTA (STE) at a final concentration of 12 mg of protein/ml. The suspension was adjusted to 1 M KCl and the indicated concentrations of potassium deoxycholate. After centrifugation at 100,000 ϫ g av for 20 min, the supernatants were collected, and the pellets were resuspended in the starting volume of STE. Equivalent volumes of the pellets and supernatants were separated by SDS-PAGE on a 12% polyacrylamide gel. Proteins were transferred to nitrocellulose and probed with antibodies against Atp22p and Sco1p (40) as in Fig. 4A. B, separation of Atp22p on isopycnic gradients. Mitochondria (Mito.) of the wild type strain W303-1A were converted to submitochondrial (SMP) particles by sonic irradiation as in Fig. 5A. The mitochondrial suspensions were also adjusted to 1% potassium deoxycholate (DOC) in the presence of 1 M KCl. Intact mitochondria (6 mg of protein) and equivalent amounts of mitochondria disrupted by sonic irradiation or detergent were applied on top of three separate discontinuous gradients containing a total of 5 ml of the indicated concentrations of sucrose. The gradients were centrifuged for 3 h at 265,000 ϫ g av in a Beckman SW65Ti rotor. Fourteen equal fractions were collected, and fractions 3-14 as well as the samples applied to the gradients (lane C) were analyzed by SDS-PAGE on a 12% polyacrylamide gel. Following transfer to nitrocellulose, the blot was probed with antibody against Atp22p and subunit 5 of cytochrome oxidase (Cox5p) as in Fig. 4A. its resistance to extraction with sodium carbonate and cofractionation with submitochondrial particles following sonic disruption of mitochondria (Fig. 5A). In contrast to the intermembrane marker cytochrome b 2 , most of which was released after conversion of mitochondria to mitoplasts, all the Atp22p remained associated with the mitoplasts (Fig. 5B). Treatment of mitochondria and mitoplasts with proteinase K under conditions that cause loss of proteins that face the cytoplasmic side of the inner membrane had no effect on Atp22p. The presence of Atp22p in a protease-protected compartment of mitoplasts together with its failure to be released from sonically disrupted mitochondria suggests that it is a membrane protein facing the matrix side of the inner membrane.
Unlike most mitochondrial membrane proteins, Atp22p is not solubilized with deoxycholate. A titration indicated that even at a final concentration of 1% deoxycholate, none of the Atp22p was extracted (Fig. 6A). Under the same conditions, increasing concentrations of the detergent caused a progressive removal of the inner membrane protein Sco1p (40) from the membranes. The ineffectiveness of deoxycholate to solubilize Atp22p was also evident in its banding behavior on isopycnic gradients. Centrifugation of mitochondria or submitochondrial particles through a step sucrose gradient showed that Atp22p banded at densities similar to that of Cox5p, a subunit of the inner membrane marker cytochrome oxidase. When mitochondria were treated with 1% deoxycholate, Atp22p was concentrated at a denser region of the gradient, whereas most of the cytochrome oxidase was shifted to a region of lesser density. The increased density is probably due to extraction of the bulk phospholipids but not of Atp22p from the membrane by deoxycholate.

DISCUSSION
Nuclear genes of S. cerevisiae reported previously (12)(13)(14)(15) to affect F 0 assembly are involved in processing/translation of the mitochondrial ATPase-specific transcripts for subunits 6, 8, and 9. The exception is Atp10p, a mitochondrial inner membrane protein that functions post-translationally as a chaperone of subunit 6 (16,36). The ATP22 product reported here is the second example of a mitochondrial protein with a posttranslational role essential for F 0 assembly. This is supported by the presence in atp22 mutants of the fully processed ATP6, -8, and -9 mRNAs and their normal translation when cycloheximide-inhibited cells are pulse-labeled with a radioactive precursor. Furthermore, the presence in the mutants of catalytically active F 1 -ATPase excludes a role of Atp22p in assembly of this oligomeric protein. The nearly complete absence under steady-state conditions of subunit 6 in the mutants further argues for the importance of Atp22p for biogenesis of the F 1 -F 0 complex. This evidence, however, does not necessarily mean that Atp22p is involved in biogenesis of subunit 6 because mutations that prevent formation of F 0 , independent of their specific functions, produce a similar phenotype. For example, the atp13 (aep2) mutant N9-168 is also grossly deficient in subunit 6 despite the fact that the product of this gene is involved in expression of subunit 9 (12,13).
Although the precise function of Atp22p is not clear at present, its localization and topology in the inner membrane of mitochondria is consistent with a role in assembly of the F 0 sector. Unlike ATP10 for which homologues exist in plants, some fungi, but not animals, ATP22 appears to be present only in the genus Saccharomyces. Searches of current protein data banks have failed to reveal even distant relationships to proteins of other organisms. This suggests that the function of Atp22p is related to some unique feature of F 0 in this yeast.