Mutational studies with Atp12p, a protein required for assembly of the mitochondrial F1-ATPase in yeast. Identification of domains important for Atp12p function and oligomerization.

The Atp12p protein of Saccharomyces cerevisiae is required for assembly of the F1 moiety of the mitochondrial ATP synthase. The current work has used mutant forms of Atp12p in an effort to learn about amino acids and/or domains that are important for the action of the protein. In one set of studies, the mutant atp12 genes were cloned and sequenced from 13 independent isolates of chemically mutagenized yeast. Of the 10 different mutant alleles that were identified, 9 (8 nonsense and 1 frameshift) lead to the early termination of the protein. A single missense mutation that substitutes lysine for Glu-289 was identified in two of the atp12 strains. Analysis of several Atp12p variants, each with different substitutions at Glu-289, showed that the functional activity of Atp12p is compromised when non-acidic residues are introduced at position 289 in the sequence. In other work, deletion analysis led to the assignment of two domains in Atp12p; the functional domain of the protein was mapped to the sequence between Gln-181 and Val-306, and a structural domain (Asp-307 through Gln-325) was identified that confers Atp12p the ability to oligomerize with other proteins in mitochondria.

The Atp12p protein of Saccharomyces cerevisiae is required for assembly of the F 1 moiety of the mitochondrial ATP synthase. The current work has used mutant forms of Atp12p in an effort to learn about amino acids and/or domains that are important for the action of the protein. In one set of studies, the mutant atp12 genes were cloned and sequenced from 13 independent isolates of chemically mutagenized yeast. Of the 10 different mutant alleles that were identified, 9 (8 nonsense and 1 frameshift) lead to the early termination of the protein. A single missense mutation that substitutes lysine for Glu-289 was identified in two of the atp12 strains. Analysis of several Atp12p variants, each with different substitutions at Glu-289, showed that the functional activity of Atp12p is compromised when nonacidic residues are introduced at position 289 in the sequence. In other work, deletion analysis led to the assignment of two domains in Atp12p; the functional domain of the protein was mapped to the sequence between Gln-181 and Val-306, and a structural domain (Asp-307 through Gln-325) was identified that confers Atp12p the ability to oligomerize with other proteins in mitochondria.
The ATP synthase of mitochondrial, chloroplast, and bacterial membranes is a key enzyme involved in energy production (1)(2)(3). The enzyme is composed of a catalytic moiety F 1 , attached peripherally to an integral membrane component F 0 . The subunit composition of F 0 varies among species, while F 1 has been highly conserved in evolution. In most organisms studied F 1 contains five different subunits in the stoichiometric ratio: ␣ 3 ␤ 3 ␥␦⑀ (1-3). X-ray diffraction data for the F 1 have been obtained with the mitochondrial enzyme from both bovine (4) and rat liver (5,6). The ␣ and ␤ subunits alternate in a hexagonal array (4,5) surrounding the amino and carboxyl termini of the ␥ subunit (4). The ␦ and ⑀ subunits are not visible in the current crystal structures. These subunits likely reside at the base of the enzyme since they are required for binding F 1 to F 0 (7,8).
In Saccharomyces cerevisiae the F 1 subunits are encoded by nuclear genes (9 -13) and, with the exception of ⑀ (11), are synthesized as precursors containing an amino-terminal mitochondrial targeting sequence that is cleaved during import (14). The reactions of sorting the F 1 subunits into mitochondria and the subsequent folding reactions within the organelle are mediated by heat-shock proteins (Hsps) 1 that serve as "molecular chaperones" in the cell (15). With respect to the biogenesis of mitochondrial proteins (such as the F 1 ), the combined activities of the cytoplasmic and mitochondrial Hsp70-Hsp40 chaperone pairs are proposed to mediate the translocation of unfolded proteins into mitochondria; folding the polypeptide chain in the matrix is then facilitated by the Hsp60 and Hsp10 proteins (for review, see Refs. 15 and 16). In the specific case of F 1 biogenesis, the final steps in the enzyme formation require two proteins, Atp11p and Atp12p, neither of which has significant sequence homology with other proteins in the data banks (17)(18)(19). Yeast mutants that are deficient for either Atp11p or Atp12p accumulate the F 1 ␣ and ␤ subunits in large protein aggregates instead of forming the enzyme oligomer (17). In contrast to Hsp60-deficient strains, which show aggregation of both the mature and precursor forms of the F 1 ␤ subunit (20), only the mature form of the F 1 subunits is observed in atp11 and atp12 mutants (17). For this reason, Atp11p and Atp12p are suggested to act at a point in the F 1 assembly pathway that is downstream from the Hsp60 step.
In considering the type of action elicited by Atp11p and Atp12p during F 1 assembly, it is informative to compare the phenotypes of atp11 and atp12 strains with those of yeast that have null mutations in the individual F 1 structural subunit genes. For instance, yeast that are deficient for the ␣ subunit harbor the ␤ subunits as aggregated proteins; likewise, the ␣ subunit aggregates in the ␤ subunit null strains (17). As mentioned above, aggregation of the F 1 ␣ and ␤ subunits is the signature phenotype of atp11 and atp12 strains. In contrast, the ␣ and ␤ subunits remain soluble in mitochondria of ␥ (13), ␦ (12), or ⑀ (11) null mutants, despite the fact that the absence of any of these subunits blocks F 1 assembly. Moreover, in ␥ subunit-deficient yeast, the ␣ and ␤ subunits sediment in linear sucrose gradients to the position where ␣␤ dimers would be expected (13). Thus, it appears that aggregation of the F 1 ␣ and ␤ subunits prevails only under conditions in which ␣␤ dimerization is not possible, such as in the ␣ or ␤ subunit null strains, or in yeast that lack a protein (i.e. Atp11p or Atp12p) whose function may be to mediate ␣/␤ oligomerization. The proposal that Atp11p and Atp12p serve as chaperones during F 1 biogenesis is supported by the fact that Atp11p and Atp12p are present in mitochondria at levels that are several orders of magnitude lower than the amount of F 1 ␣ and ␤ subunit protein (21) 2 which is in accord with the fact that there is only a small pool of unassembled F 1 proteins in the steady state (22).
The present paper reports on features of the Atp12p protein that are important for its action as an F 1 assembly factor. The primary translation product of the ATP12 gene is a 37-kDa precursor protein, which is cleaved to generate the mature polypeptide (33 kDa) following import into mitochondria (18). In previous work Atp12p was shown to be a soluble protein of the mitochondrial matrix that sediments in linear sucrose gradients as an oligomer of 70 -80 kDa (18). The current work presents evidence that the Atp12p oligomer observed in mitochondria is likely heterogeneous in content and indicates a region in the Atp12p primary structure that is important for intermolecular associations of the protein. In addition, the functional domain of Atp12p was identified and shown to include an acidic amino acid that is required for optimal activity of the protein.
DNA Sequencing-The oligonucleotide primers used for sequencing and for constructing atp12 deletion/truncation mutants are listed in Table II. Genomic DNA was purified from yeast and served as the template for PCR amplification of the atp12 gene using the primers 1 and 9. The 1048-bp PCR product was digested with BamHI and SstI and ligated at these restriction sites in the yeast/E. coli shuttle vector, YEp352 (25). The atp12 gene carried in the plasmids was sequenced by the dideoxy method (26) with Sequenase (U. S. Biochemical Corp.). Both strands of two separate PCR products were sequenced for each atp12 mutant.
Yeast Plasmid Constructions-The plasmids used in this study are described in Table III. Atp12p is numbered from 1 to 325, where residue 1 is the initiator methionine in the primary translation product (18). In cases where sequences were removed from the 3Ј end of the ATP12 gene, the plasmids and the encoded products are named according to where the mutant protein is truncated. This is indicated by the singleletter code and number (18) of the last retained amino acid. For mutant proteins that have deletions from the amino terminus, the plasmids and the encoded products are designated by the ⌬ symbol, followed by the span of ATP12 codons removed. The number of amino acids that are deleted from the carboxyl or the mature amino terminus of the mutant Atp12p proteins is indicated in Table III. In naming the plasmids coding for mutant Atp12p proteins harboring missense mutations, the single-letter code is used to indicate the amino acid substitution at the designated position in the Atp12p primary structure (18). The details of the plasmid constructions are provided in Footnote 3. All of the plas- 3 Wild type Atp12p is encoded in the multi-copy (2 ) plasmid, pG57/ ST4 (constructed as described in Ref. 18). Plasmids coding for Atp12p deleted for sequences from the carboxyl terminus are described as follows. Atp12(V306)p is encoded by pG57/V306. This plasmid carries a 1.0-kb SstI-HincII fragment of DNA that was prepared from pG57/ST4 and ligated to the 2vector, YEp352 (25), after the vector was opened with EcoRI, made blunt-ended with Klenow, and then digested with SstI. Atp12(V283)p is coded for in pG57/V283. For this plasmid construction, a 1.0-kb SstI-RsaI fragment was purified from pG57/ST4 and ligated to YEp352, which was prepared by first digesting with XbaI, filling in the site with Klenow, and then cutting with SstI. Atp12(P239)p is encoded by pG57/P239. For this construction, a plasmid carrying the atp12-8 nonsense allele cloned from yeast strain P366 was opened with SstI, and the 3Ј end was made blunt using T 4 DNA polymerase and Klenow. The resultant DNA was digested with BglII (this restriction site is upstream from the nonsense mutation in the mutant atp12-8 allele). Klenow was then used to blunt the BglII restriction site, after which the double blunt-ended DNA fragment was self-ligated to produce plasmid pG57/P239. Plasmids coding for Atp12p proteins deleted for sequences from the amino terminus are described as follows. In all of these plasmids, the 5Ј-truncated ATP12 coding sequences are ligated in-frame with DNA coding for the mitochondrial targeting sequence of Atp11p (amino acids 1-39, see Ref. 21). As the first step, a plasmid (pG13L) was constructed with the coding sequence for the Atp11p leader peptide. To make pG13L, a 340-bp SstI-SmaI DNA fragment was prepared from the plasmid pG13⌬40 -75s (27) and ligated to YEp352 that was cut with the same restriction enzymes. Atp12(⌬1-81)p is encoded by pG57⌬1-81. For this construction, a 1.0-kb DNA fragment was purified from pG57/ST4 after cutting the plasmid with BfaI, filling in the site with Klenow, and digesting with HindIII. The linear DNA fragment was then ligated to YEp352, which was prepared by opening at the BamHI site, blunt-ending the vector with Klenow, and cutting with HindIII. The resultant plasmid was used to prepare a 1.0-kb SmaI-HindIII fragment, which was then ligated to pG13L that was digested with the same enzymes to produce pG57⌬1-81. There are two amino acid residues (Gly and Asp) encoded at the junction between the ATP11 and ATP12 sequences in pG57⌬1-81 that results from the addition of the sequence, GGGGAT, during construction of this plasmid. Atp12(⌬1-124)p is coded for in pG57⌬1-124. To make this plasmid, an 896-bp NsiI-HindIII fragment prepared from pG57/ST4 was ligated to PstI,HindIII-cut pG13L. There are seven codons (Gly-Asp-Pro-Leu-Glu-Ser-Thr) inserted in between the ATP11 and ATP12 sequences in pG57⌬1-124 due to the addition of the sequence, GGGGATCCTCTAGAGTCGACC, during the construction of this plasmid. Atp12(⌬1-180)p is encoded by pG57⌬1-180. For this plasmid construction, a 730-bp FspI-HindIII DNA fragment was purified from pG57⌬1-124 and ligated to HincII,HindIII-cut pUC18. The resultant plasmid was used to prepare a 740-bp SmaI-HindIII fragment, which was then ligated to SmaI,HindIII-cut pG13L to produce pG57⌬1-180. There are six amino acids (Gly-Asp-Pro-Leu-Glu-Ser) encoded at the junction between the ATP11 and ATP12 sequences in pG57⌬1-180 that result from adding the sequence, GGGGATCCTCTA-GAGTCG, during this plasmid construction. Atp12(⌬1-224)p is coded for in pG57⌬1-224. To make this plasmid, a 600-bp EcoRV-HindIII DNA fragment was purified from pG57/ST4 and ligated to SmaI,Hin- mids whose construction utilized PCR (p657/E289D, p657/E289A, pG57/E289Q, pG57/ST21) were sequenced to verify that there were no artificial mutations introduced in the ATP12 coding region due to errors made by the Taq polymerase.
Preparation of Yeast Mitochondria-Yeast were grown aerobically in liquid YPGal at the temperature indicated in the experiment to early stationary phase. The method of Faye et al. (28) was used to prepare mitochondria with the exception that Zymolyase, instead of Glusulase, was used to digest the cell wall. Phenylmethylsulfonyl fluoride (PMSF) was added to 10 g/ml final concentration during the cell-breaking step to minimize proteolysis.
Solubilization of Atp12p from Mitochondria-Two different methods, both of which give comparable results, were used to prepare mitochondrial extracts containing solubilized Atp12p. For some experiments, mitochondria were suspended at 10 mg/ml in 10 mM Tris-HCl, pH 8.0, and sodium deoxycholate was added to 1 mg/ml to permeabilize the membranes. Following a 15-min incubation at 0°C, the suspension was centrifuged for 30 min at 4°C at 50,000 rpm in a Beckman 70Ti rotor. Alternatively, mitochondria were suspended at 7-8 mg/ml in 0.4 ml of the buffer specified in the experiment and exposed to four 10-s bursts, with cooling in between, of sonic irradiation at 40% output (Branson dIII-cut pG13L. The construction of plasmids that code for Atp12p proteins harboring missense mutations is described as follows. Plasmid pG57/E289K encodes Atp12(E289K)p. This plasmid carries the atp12-5 missense allele that was cloned (using PCR) from the mutant strain E822 and ligated in the BamHI,SstI sites of YEp352. The multi-copy plasmids encoding the Atp12(E289D)p, Atp12(E289A)p, and Atp12(E289Q)p proteins were obtained by site-directed mutagenesis. As part of this work, an 880-bp SstI-Sau3AI fragment was purified from pG57/ST4 and ligated to SstI,BamHI-cut YEp352 to produce plasmid pG57/Sau, which codes for the amino three-fourths of Atp12p. The site-directed changes were made using PCR with plasmid pG57/ST4 as the DNA template. For the E289A mutation, primers 20 and 17 (Table  II) were used to make a 141-bp product, which was then cut with BglII and NheI, and primers 12 and 14 (Table II) were used to produce a 169-bp PCR that was subsequently digested with HindIII and NheI. The two PCR products were ligated to BamHI,HindIII-cut pG57/Sau to give plasmid pG57/pE289A. Similar strategies were used to create the pG57/E289D and pG57/E289Q plasmids. To make pG57/E289D, primers 20 and 18 (Table II) were used for synthesis of the 141-bp PCR product, which was digested with BglII and XbaI, and primers 12 and 15 (Table II) were used to make the 169-bp PCR product, which was cut using HindIII and XbaI. For pG57/E289Q, the 141-bp PCR product was obtained using primers 20 and 19 (Table II) and then cut using BglII and PstI, and the 164-bp PCR product was synthesized using primers 12 and 16 (Table II) and subsequently digested with HindIII and PstI. Three-piece ligations were then done to put the appropriate pairs of 141-and 169-bp PCR products into the BamHI and HindIII sites of pG57/Sau yielding pG57/E289D and pG57/E289Q. YHistag-Atp12p is the name for the Histag-Atp12p protein that is produced in yeast and is encoded in plasmid pG57/ST21. This protein utilizes the Atp11p mitochondrial leader sequence as the targeting signal for the mature Atp12p protein that is tagged with a (6x)histidine sequence at the amino terminus. To make this plasmid, the primers 11 and 13 (Table II) were used for PCR with the bacterial plasmid pHISATP12 (this plasmid is described under "Materials and Methods") as the DNA template. The 1.0-kb PCR product was digested with SmaI and BamHI and added to a three-piece ligation with a 340-bp SstI-SmaI fragment purified from pG13⌬40 -75s (27) and SstI,BamHI-cut YEp352 to produce pG57/ST21. The construction of ATP12 plasmids with vectors of the yeast twohybrid system is described as follows. Plasmids pG57/ST22 and pG57/ ST23 code for Histag-Atp12p in, respectively, the pAS2-1 and pACT-2 vectors purchased from CLONTECH. For these constructions, a 1-kb SmaI-BamHI fragment was purified from pG57/ST21 and ligated either to pAS2-1 or pACT-2, each of which was prepared by first digesting with NcoI, blunt-ending with Klenow, and digesting with BamHI.
Gal4p-AD-Histag-Atp12p e a With the exception of pG57/ST4 (Ref. 18), the present study is the source for all of the plasmids. The multi-copy (2) vector, YEp352 (25), was used for all plasmid constructions.
b These values refer to the number of amino acids that are deleted from the amino terminus of the mature protein, which is the form of Atp12p that is generated when the mitochondrial leader sequence is removed.
c The numbers refer to the ATP12 codons that were removed; in each case the remaining portion of the ATP12 gene is ligated to the Atp11p mitochondrial leader sequence (ATP11 codons 1-39, Ref. 21). d Fusion protein formed between the DNA binding domain of Gal4p and Histag-Atp12p. e Fusion protein formed between the activation domain of Gal4p and Histag-Atp12p. sonifier, model 450). The sonicated samples were then centrifuged as described above. The disruption of mitochondria (by means of detergent or sonic irradiation) was done in the presence of PMSF (0.5 mM), leupeptin (1 g/ml), and pepstatin (1 g/ml) to minimize proteolysis of the solubilized proteins.
Purification of Histag-Atp12p from E. coli Expression Systems-Two different plasmids were employed for the overproduction of Histag-Atp12p in bacteria. In one case, PCR was used to create the bacterial plasmid, pHISATP12, which encodes the mature form of Atp12p (without the mitochondrial leader sequence (18)) that carries a (6x)histidine sequence at the amino terminus. For this construction, a yeast plasmid coding for wild type Atp12p (pG57/ST4, Table III) served as the template for PCR with the primers 9 and 10 (Table II) to produce a 910-bp DNA fragment. The PCR product was digested with SmaI and SstI and ligated to EheI,SstI-cut pPROEX TM HTa to produce pHISATP12. E. coli cells carrying pHISATP12 were grown in a 0.5-liter LB/ampicillin culture at 30°C to mid-log phase. Following a 4-h induction with 0.5 mM isopropylthiogalactoside, the cells were harvested, sonically irradiated in buffer L (50 mM Tris-HCl, pH 8.5, 10 mM 2-mercaptoethanol, 1 mM PMSF), and centrifuged at 6,000 rpm for 10 min in a Sorvall SA600 rotor at 4°C. Recombinant Histag-Atp12p partitions in the insoluble fraction at this step. The inclusion bodies were dissolved in 8 M urea, 20 mM Tris-HCl, pH 8.0, 1 mM PMSF, 10 mM 2-mercaptoethanol, after which the urea-solubilized solution was clarified by centrifugation (6,000 rpm, 10 min), rapidly diluted 20-fold in buffer (20 mM Tris-HCl, pH 8.0, 10 mM 2-mercaptoethanol), and loaded on a DEAE column (bed volume ϭ 6 ml). The column was washed with 20 mM Tris-HCl, pH 8.0, and eluted with a linear 0 -2 M NaCl gradient in 20 mM Tris-HCl, pH 8.0. The fractions containing Histag-Atp12p were pooled and brought to 80% saturation with (NH 4 ) 2 SO 4 . The protein precipitate was dissolved in 10 ml of buffer L and loaded on a Ni-NTA column (bed volume ϭ 2 ml). The affinity column was washed sequentially with 20 ml of buffer A (20 mM Tris-HCl, pH 8.5, 100 mM KCl, 10 mM 2-mercaptoethanol, 10% glycerol, 20 mM imidazole), 15 ml buffer of B (20 mM Tris-HCl, pH 8.5, 500 mM KCl, 10 mM 2-mercaptoethanol, 10% glycerol), and 15 ml of buffer A. Buffer C (20 mM Tris-HCl, pH 8.5, 100 mM KCl, 10 mM 2-mercaptoethanol, 10% glycerol, 100 mM imidazole) was then used to selectively elute Histag-Atp12p, which yielded the protein 95% pure (see lane 1, Fig. 1). The second expression system employed the bacterial vector, pMAL-C2 (New England Biolabs), to direct the synthesis of a chimeric protein in which maltose binding protein (MBP) is fused immediately proximal to the (6x)histidine sequence in Histag-Atp12p. The plasmid pMBPHISATP12 was created by subcloning a 1-kb SmaI-HindIII fragment encoding the sequence for Histag-Atp12p from the plasmid pG57/ST21 (Table III) into the pMAL-C2 vector that was prepared as an XmnI,HindIII-cut DNA. E. coli cells carrying the plasmid pMBPHISATP12 were grown in 2 liters of LB/ampicillin at 37°C to the exponential phase of growth, after which expression from the plasmid was induced for 3 h using 1 mM isopropylthiogalactoside. Next, the cells were harvested and suspended in 50 ml of lysis buffer (10 mM Na 2 HPO 4 , pH 7.0, 30 mM NaCl, 0.25% Tween 20, 10 mM 2-mercaptoethanol, 10 mM EDTA, 10 mM EGTA). Following three cycles of freezing and thawing, the cell suspension was sonically irradiated, brought to 0.5 M NaCl, and centrifuged at 9,000 rpm in a Sorvall SA600 rotor. The crude extract (containing the recombinant fusion protein) was diluted five times with column buffer (10 mM Na 2 HPO 4 , pH 7.0, 500 mM NaCl, 0.25% Tween 20, 10 mM 2-mercaptoethanol, 1 mM EGTA, 1 mM azide) and loaded at a flow rate of 1 ml/min on an Amylose column (bed volume ϭ 3 ml) that was pre-equilibrated with the same column buffer. The column was washed with 9 ml of fully supplemented column buffer and then with 15 ml of column buffer omitted for Tween 20, and the MBP-Histag-Atp12p fusion protein was finally eluted with 15 ml of column buffer (detergentfree) that contained 10 mM maltose in 95% pure form. The Coomassiestained gel in Fig. 2 shows a sample of the purified MBP-Histag-Atp12p protein (lane 2) from the pooled column fractions.
Affinity Precipitation of Histag-Atp12p from Mitochondrial Extracts-Mitochondria were suspended in 50 mM Tris-HCl, pH 8.5, 10 mM 2-mercaptoethanol, 1 mM PMSF, and soluble extracts were prepared by sonic irradiation (see above). A 50-l slurry of Ni-NTA resin was pre-equilibrated with buffer M (10 mM Tris-HCl, pH 8.0, 5 mM imidazole, 140 mM NaCl, 1% Triton X-100, 1 mM PMSF) and added to 0.14 ml of sonic supernatant containing solubilized mitochondrial proteins. The suspension was mixed end-over-end for 30 min at room temperature and then centrifuged for 3 min in a microcentrifuge at room temperature. The supernatant was collected, and the Ni-NTA beads were washed five times for 5 min with 0.14 ml of buffer M, and finally suspended in SDS-gel loading buffer in preparation for Western analysis.
Sedimentation Analysis of Atp12p-Mitochondria were suspended in 20 mM Tris-HCl, pH 8.0, and soluble extracts were prepared by sonic irradiation (see above) at 4°C. Molecular weight markers (hemoglobin and myokinase, or hemoglobin and lipoamide dehydrogenase) were added to the solubilized protein samples, and the mixtures were centrifuged through 7-20% linear sucrose gradients under the conditions of sedimentation described (18). Previously described methods (18,21) were used to define the positions of myokinase, hemoglobin, and lipoamide dehydrogenase peaks in the gradients. The gradient fractions were assayed by Western analysis for Atp12p.
Chemical Cross-linking-For cross-linking with amine-reactive bifunctional reagents, purified recombinant Histag-Atp12p was incubated at 0.4 mg/ml in 20 mM Na 2 HPO 4 , pH 7.5, 150 mM NaCl with either 5 mM disulfosuccinimidyl tartarate, 5 mM ethylene glycolbis(sulfosuccinimidylsuccinate), or 5 mM dithiobis(sulfosuccinimidylpropionate) at 22°C for 30 min. The modification reactions were quenched by the addition of Tris-HCl, pH 8, to 50 mM concentration and denatured in SDS-gel loading buffer in preparation for electrophoresis. For experiments that employed glutaraldehyde, the purified Histag-Atp12p protein was incubated in 10 mM Tris-HCl, pH 7.5, at 0.4 mg/ml with 0.05, 0.1, or 0.5% glutaraldehyde at 22°C for 10 min, at which time SDS-gel loading buffer was added, and the samples were loaded on SDS-polyacrylamide gels. In control experiments, 0.5 mg/ml hemoglobin was incubated under the same conditions with 0.5% glutaraldehyde.
Assays-Protein concentrations were estimated by the method of Lowry et al. (30). ATPase activity was measured by the colorimetric determination of inorganic phosphate as described previously (27).
Miscellaneous Procedures-Standard techniques were used for restriction endonuclease analysis of DNA, purification and ligation of DNA fragments, and transformations of and recovery of plasmid DNA from E. coli (31). Yeast transformations employed the LiAc procedure (32). The method of Laemmli (33) was used for SDS-polyacrylamide gel electrophoresis. Western blotting followed the procedure of Schmidt et al. (34). Antibodies against the full-length, mature Atp12p protein were prepared as described (18) and used at a dilution of 1:100.

Characterization of the atp12 Mutants Obtained by Chemical
Mutagenesis-Thirteen independent yeast isolates, with mutations in the ATP12 gene, were obtained by chemical mutagenesis with nitrosoguanidine and ethyl methanesulfonate (24). These strains are respiratory-deficient due to a defect in the F 1 -ATPase assembly pathway and fail to utilize non-fermentable carbon sources for growth (17). The biochemical properties of some of the mutant atp12 strains were reported previously (18). In the current work, the mutant genes were cloned and sequenced from each of the atp12 strains to determine the type of genetic lesions that alter Atp12p function (see under "Materials and Methods"). This analysis disclosed a single missense allele (carried in strains E822 and E823) and nine mutant alleles (eight nonsense and one frameshift) that cause early termination of the Atp12p protein (Table IV). The genetic lesions in the latter strains should lead to the synthesis of Atp12p with deletions of 29, 48, 76, 86, 172, 187, 223, 271, or 274 amino acid residues from the carboxyl terminus (Table IV). The positions of the chemically induced mutations are indicated in the Atp12p sequence shown in Fig. 3, using arrows that are flagged with the name of the mutant strain (highlighted in the black boxes). Western analysis (data not shown, see also Ref. 18) shows the presence of the mutant form of Atp12p in mitochondrial samples prepared from the atp12 strains, C264, E822, E695, N242, and P366. Such proteins are deleted for up to 86 amino acids from the carboxyl terminus of Atp12p. Mutant Atp12p proteins that are predicted to have 172 or more amino acids removed from the carboxyl terminus were not detected in Western blots of total mitochondrial protein.
Respiratory Properties of Yeast That Produce Genetically Engineered Forms of Atp12p with Deletions of Sequences from the Carboxyl or the Amino Terminus-The atp12 nonsense alleles (see above) encode truncated forms of Atp12p, which are inactive when produced in single copy from the chromosome. To investigate the possibility that carboxyl sequences could be dispensed with if the level of truncated Atp12p is raised in the cell, a series of multi-copy plasmids were constructed with atp12 genes that have deletions of sequences from the 3Ј end of the gene (Table III). These plasmids direct the synthesis of the mutant proteins, Atp12(P239)p, Atp12(V283)p, and Atp12-(V306)p, which are missing 86, 42, and 19 amino acids, respectively, from the carboxyl terminus (Table III). The position of the last retained amino acid in these mutant proteins is indicated with an arrow in the Atp12p sequence shown in Fig. 3.
The properties of the plasmid-borne Atp12p mutant proteins were evaluated in the genetic background of a respiratory-deficient yeast strain that harbors a disruption at the ATP12 locus (aW303⌬ATP12 (18)) ( Table V). None of the mutant proteins that have sequences deleted from the carboxyl terminus conferred to the host strain the ability to grow within 48 h using ethanol-glycerol (EG), a non-fermentable carbon source (Table V). Low levels of respiratory activity were observed only for the strain that produces Atp12-(V306)p, which eventually grows on EG plates after 3-4 days at 30°C. Western analysis established that the plasmidproduced Atp12(V306)p, Atp12(V283)p, and Atp12(P239)p proteins were present in mitochondria isolated from the respective yeast transformants at 30 -80% of the wild type level ( Fig. 4 and Table V). Similar results were obtained when mitochondrial extracts (solubilization with 0.1% sodium deoxycholate), rather than total mitochondria, were analyzed by Western blots (data not shown). With respect to the Western analysis it is important to note that while the antigen used to raise the polyclonal Atp12p antiserum in previous studies (18) included the full sequence for mature Atp12p, there is no proof that epitopes for the antibody are distributed along the entire length of the protein. Thus, truncated forms of Atp12p might, in fact, be missing epitopes or could have an altered structure that compromises epitope presentation. It is for these reasons that the levels of the truncated forms Atp12p detected by Western analysis of mitochondria (Table V) are taken to represent the minimal amount of the mutant proteins in the organelle. In other experiments, the mitochondrial samples prepared from the mutant yeast were assayed for ATPase activity in the absence and presence of oligomycin (Table V). Since the amount of oligomycin-sensitive ATPase activity indicates the amount of F 1 that is properly assembled and bound to F 0 , this analysis provides an indication of whether or not the resident Atp12p protein is competent for mediating F 1 assembly. Virtually no ATPase activity (Ͻ5% wild type level) was detected in the yeast transformants that overproduce the Atp12(V306)p, Atp12-(V283)p, and Atp12(P239)p proteins (Table V), indicating that the removal of sequences from the carboxyl terminus of Atp12p severely compromises its action in the F 1 assembly pathway.
Deletion mutagenesis was also employed to analyze the functional significance of amino-terminal sequences of Atp12p us- ing aW303⌬ATP12 as the yeast host for recombinant plasmids. For this work, multi-copy plasmids were constructed in which 5Ј-end-deleted portions of the ATP12 coding region were fused in-frame with the sequence for the mitochondrial targeting signal of the Atp11p protein (21) to encode the mutant proteins, Atp12(⌬1-81)p, Atp12(⌬1-124)p, Atp12(⌬1-180)p, and Atp12-(⌬1-224)p (Table III). Following cleavage of the Atp11p targeting signal, these mutant Atp12p proteins are missing 51, 94, 150, or 194 residues from the amino terminus of mature Atp12p. The first amino acid of Atp12p in these proteins is, respectively, Leu-82, Cys-125, Gln-181, and Ile-225 (see Fig. 3).
Of the four mutant proteins that are deleted for aminoterminal sequences, two were shown to confer respiratory competence (e.g. growth on EG plates) to aW303⌬ATP12 within 24 -48 h of incubation at 30°C (see results for Atp12(⌬1-81)p and Atp12(⌬1-124)p, Table V). Western analysis performed with mitochondria from these yeast transformants disclosed that only Atp12(⌬1-124)p was observed at an appreciable level (18% relative to the control) in samples of total mitochondrial protein (Fig. 4, see data for cells grown at 30°C in Table V) and in soluble extracts of mitochondria (data not shown). The yeast transformant that produces this protein was also shown to display 14% the control level of oligomycin-sensitive ATPase activity in mitochondria isolated from cells grown in galactose at 30°C (Table V). Despite the fact that Atp12(⌬1-81)p was not detected in Western blots of isolated mitochondria, yeast carrying the plasmid for this protein showed 78% the wild type level of oligomycin-sensitive ATPase activity (Table V). The evidence for high levels of assembled F 1 in yeast that produce Atp12(⌬1-81)p suggests that the mutant protein is active in vivo. To explain the absence of Atp12(⌬1-81)p from Western blots, we suggest that either epitopes for the antibody are masked in the mutant protein or that the mutant protein might be unusually unstable in vitro.
Two of the yeast transformants that produce truncated forms of Atp12p were found to be temperature-sensitive for respiratory function. At 23°C, yeast that overproduce either Atp12(⌬1-124)p or Atp12(⌬1-180)p showed wild type growth  Atp12p was produced from a multi-copy plasmid. b Yeast were scored for growth after 48 h following replica plating from glucose to ethanol/glycerol media. ϩϩ, good growth; ϩ, moderate growth; Ϫ, no growth.
c The amount of Atp12p protein was determined using an AMBIS 4000 imaging system to scan Western blots of total mitochondrial protein. The values obtained for the protein band intensities were corrected to account for the size of the mutant proteins relative to full-length Atp12p. The mean values from the analysis of two separate Western blots are reported; a representative blot that shows data for cells grown at 30°C is given in Fig. 4 on EG plates. Moreover, there was 60% of the control level of oligomycin-sensitive ATPase activity measured in mitochondria isolated from these transformants, following their growth at 23°C in galactose media (Table V). The effect of temperature is most notable for yeast that synthesize Atp12(⌬1-180)p, as this strain is completely respiratory-deficient at 30°C (see Table V). The enhancement or acquisition of respiratory activity at 23°C was not accompanied by an increased amount of Atp12(⌬1-124)p or Atp12(⌬1-180)p detected by Western analysis in samples of isolated mitochondria (see Table V). Again, altered epitope display or marked instability in vitro offer potential explanations for the difficulties in detecting forms of Atp12p deleted for amino-terminal sequences in Western blots. Respiratory Properties of Yeast That Produce Atp12p with Mutations at Glu-289 -The identification of a Glu-289 3 Lys missense mutation in the Atp12p sequence of two independent atp12 isolates (E822, E823; Table IV) suggests that Glu-289 may be important for the activity of the protein. For this reason, the effects of changes at Glu-289 were further analyzed by generating three additional substitutions; E289D, E289A, and E289Q (see Footnote 3 for method details). The respiratory phenotypes of yeast that produce Atp12p with substitutions at Glu-289 are given in Table VI. Under conditions in which Atp12p is produced from multi-copy plasmids in the genetic background of the disruption strain aW303⌬ATP12, all four mutant proteins were detected by Western analysis at near wild type levels in isolated mitochondria (data not shown) and in soluble extracts of deoxycholate-treated mitochondria (Fig.  5). The mutant protein Atp12(E289K)p fails to confer respiratory function (growth on EG plates) to the host strain aW303⌬ATP12, even when produced from a multi-copy plasmid (Table VI). In this case only 11% of the wild type level of oligomycin-sensitive ATPase activity can be detected in isolated mitochondria. This observation is in accord with the report that the ability of yeast to utilize non-fermentable substrates for growth correlates with the presence of at least 15% mitochondrial ATPase activity (35). Yeast transformants harboring the atp12 mutations E289A or E289Q show normal growth on EG plates but exhibit 60 -70% reduction in the level of mitochondrial oligomycin-sensitive ATPase activity (Table  VI). In the case of the mutation E289D, wild type levels of mitochondrial ATPase activity are observed whether the mutant protein is produced from a multi-copy plasmid (Table VI) or from a single copy plasmid (data not shown). Since all four mutant proteins are detected in mitochondria at near wild type levels, the data suggest that the reduced activity of Atp12p harboring either the E289K, E289A, or E289Q mutations is not due to a reduction in the amount of the mutant proteins but is likely to be a consequence of having a non-acidic residue at position 289 in the sequence.
Sedimentation Analysis of Plasmid-borne Atp12p Variants-The molecular mass of mitochondrial Atp12p was estimated from its sedimentation properties in linear sucrose gradients to be in the range of 70 -80 kDa (18). The molecular mass of the mature protein, as estimated from its migration in SDS gels relative to the precursor (18), is 33 kDa. These results suggest that the protein observed in sucrose gradients is oligomeric. Sedimentation analysis was used to determine if the ability of Atp12p to form higher ordered oligomers is lost when sequences are deleted from the amino or carboxyl terminus or as a result of mutations at Glu-289. Soluble mitochondrial extracts were prepared from the yeast transformants that overproduce mutant forms of Atp12p and centrifuged through linear sucrose gradients in the presence of molecular weight standards. Only the deletion mutants of Atp12p, whose presence was detected in Western blots of mitochondrial samples (see Table V), were analyzed in these studies, i.e. Atp12-(V306)p, Atp12(V283)p, Atp12(P239)p, and Atp12(⌬1-124)p. The sizes predicted for monomers of mutant forms of the protein are given in Table VII. The three proteins that are missing carboxyl sequences (Atp12(V306)p, Atp12(V283)p, and Atp12-(P239)p) co-sediment with myokinase (M r ϭ 21,000) in sucrose gradients (Fig. 6). This result suggests that these three mutant proteins are present as monomers in the mitochondrial matrix. On the basis of sequence, the size of the Atp12(P239)p monomer is comparable to that of Atp12(⌬1-124)p (Table VII), yet the sedimentation profiles for these two proteins are significantly different (Fig. 6). The apparent larger size noted for the amino-terminal deletion mutant suggests that Atp12(⌬1-124)p forms an oligomer in vivo (Table VII). With respect to the mutant proteins that are substituted at Glu-289, the sedimentation behavior of all four proteins (Fig. 6) was shown to be comparable with that of wild type mitochondrial Atp12p (18). These results indicate that the E289K, E289D, E289A, and E289Q mutations have no effect on Atp12p oligomerization.
Analysis of the Oligomeric State of Atp12p-Efforts were made to determine if mitochondrial Atp12p is a homo-or hetero-oligomer. As part of this work, Histag-forms of Atp12p synthesized in E. coli were used to evaluate the size of Atp12p in the absence of other mitochondrial proteins. Control experiments had shown that the presence of the (6x)histidine tag sequence does not alter the functional properties of Atp12p. Such experiments employed an atp12 null mutant transformed with a yeast plasmid (pG57/ST21, Table III) that encodes a protein chimera in which the Atp11p leader sequence is used to sort the tagged Atp12p protein (called YHistag-Atp12p for Yeast Histag-Atp12p) to mitochondria. The (6x)histidine sequence is retained on YHistag-Atp12p following cleavage of the heterologous leader peptide as indicated by the fact that YHistag-Atp12p (but not the non-tagged form of the protein) is selectively precipitated from mitochondrial extracts with Ni-NTA beads (procedure described under "Materials and Methods," data not shown). Notably, the atp12 null mutation is complemented by the plasmid-borne YHistag-Atp12p as indicated by the ability of the transformed strain to grow on ethanol-glycerol plates at the wild type rate and to exhibit wild type levels of oligomycin-sensitive mitochondrial ATPase activity FIG. 4. Western blot of mitochondria prepared from yeast transformants cultured at 30°C, which produce truncated forms of Atp12p from plasmids. Mitochondria were prepared from aW303⌬ATP12 transformants that produce the Atp12p proteins indicated in the figure. The following amounts of total mitochondrial protein were loaded in each lane of a 12% SDS-polyacrylamide gel: wild type Atp12p, Atp12(V306)p, Atp12(V283)p, 20 g; Atp12(P239)p, 40 g; Atp12(⌬1-81, ⌬1-124, ⌬1-180, ⌬1-224)p, 60 g. Gel electrophoresis and Western blotting were done as described under "Materials and Methods." After probing with antibody against Atp12p, the blot was reacted with 125 I-protein A and exposed to x-ray film. The migration of molecular mass markers (given in kilodaltons) is indicated in the lefthand margin of the figure.
Histag-Atp12p was overproduced in E. coli from the plasmid pHISATP12 and purified to homogeneity (see Fig. 1). The sedimentation behavior of Histag-Atp12p (mass ϭ 35.7 kDa) in 7-20% sucrose gradients (Fig. 7) is significantly different from that of Atp12p that is lacking the last 19 amino acids, which migrates as a monomer (see sedimentation profile for Atp12(V306)p (mass ϭ 30.9 kDa) in Fig. 6). The size difference (ϳ5 kDa) between Histag-Atp12p and Atp12(V306)p is too small to account for the different migration of the two proteins in this experiment; similar work performed with the native and biotin-tagged forms of Atp11p indicated comparable sedimentation profiles for the proteins despite a difference of ϳ9 kDa in their size (21). The sedimentation profile of bacterially produced Histag-Atp12p is much more similar to that of the wild type mitochondrial protein (18) and to the mitochondrial Atp12p variants substituted at Glu-289 (Fig. 6), which sediment in sucrose gradients as oligomers. This analysis suggested that the native form of Atp12p is a homo-oligomer. However, the results from additional experiments conflict with this proposal. Studies with purified bacterial Histag-Atp12p that were designed to visualize a covalently linked Atp12p dimer gave negative results. SDS gels run with purified Histag-Atp12p in the absence of 2-mercaptoethanol did not show evidence of disulfide-linked dimers; such linkages were predicted as possible since there are three cysteines in the Atp12p polypeptide chain. In other work, three amine-reactive bifunctional chemical reagents (disulfosuccinimidyl tartarate, ethylene glycolbis(sulfosuccinimidylsuccinate), dithiobis(sulfosuccinimidylpropionate)), and glutaraldehyde, were used to crosslink purified recombinant Histag-Atp12p as described under "Materials and Methods"; there was no evidence in SDS gels for cross-linked Histag-Atp12p dimers generated by any of the chemical modification reactions (data not shown). Notably, under the same conditions cross-linking of hemoglobin by glutaraldehyde was observed suggesting that the experiment was properly designed to identify the presence of oligomers. Additional evidence against the formation of Atp12p homo-oligomers was obtained using the yeast two-hybrid screen (29), in which the two plasmids used in the assay both carried the gene for Histag-Atp12p. No activation of the lacZ reporter gene was observed suggesting that Atp12p monomers do not interact under the conditions of this assay. Although the negative result with the two-hybrid screen is not final proof against Atp12p homo-oligomerization, it is in agreement with the results from the cross-linking experiments that were performed with a battery of chemical modifiers.
The possibility was considered that the apparent migration of recombinant Histag-Atp12p as a homodimer in sucrose gradients (Fig. 7) was due to an artifact imposed by the purification procedure, which included extraction from inclusion bodies under denaturing conditions followed by renaturation, and efforts were made to overproduce the protein in bacteria in soluble form. To this end, a recombinant fusion protein formed between maltose binding protein (MBP) and Histag-Atp12p was found to remain soluble following bacterial cell lysis (see under "Materials and Methods"). Purified MBP-Histag-Atp12p a Atp12p was produced from a multi-copy plasmid. b Yeast were scored for growth after 48 h following replica-plating from glucose to ethanol/glycerol media. ϩϩ, good growth; ϩ, moderate growth; Ϫ, no growth.
c The amount of Atp12p protein was determined using an AMBIS 4000 imaging system to scan Western blots of total protein in mitochondrial extracts. The mean values from the analysis of two separate Western blots are reported; one of the blots is shown in Fig. 5.  Atp12p was produced from a multi-copy plasmid. b These values are for "mature" Atp12p, which is produced following cleavage of the mitochondrial sorting signal.
c Evidence for the monomeric or oligomeric state of Atp12p is based on sedimentation analysis that was performed with two separate mitochondrial samples. The data from one of the experiments are shown in Fig. 6. ( Fig. 2) (mass ϭ 77 kDa) sediments in 7-20% sucrose gradients to a position between the hemoglobin (mass ϭ 66 kDa) and lipoamide dehydrogenase (mass ϭ 100 kDa) marker proteins (Fig. 7). This result indicates that the fusion protein migrates as a monomer. Thus, cumulatively, the results from crosslinking studies, the yeast two-hybrid screen, and sedimentation analysis of recombinant MBP-Histag-Atp12p suggest that pure, homogeneous Atp12p does not form homo-oligomers and that the oligomeric form of Atp12p observed in mitochondrial samples most likely originates from the interaction of Atp12p with one or more different gene products. DISCUSSION Sequence analysis of 13 independent isolates of yeast harboring chemically induced mutations in the ATP12 gene identified eight nonsense alleles, one frameshift mutation that leads to premature termination of the protein, and a Glu-289 3 Lys substitution in the Atp12p primary structure (Table IV). The low frequency of missense mutations (1 in 10 alleles) is not a common feature of the pet mutant collection (24) from which the atp12 strains were obtained. A plausible reason why more missense mutations are not present among the atp12 strains is because the genetic screen that was used originally to select the strains (24) was based on the ability of cells to grow on nonfermentable carbons (ethanol-glycerol, EG). Thus, only the most severe mutations, which prevent Atp12p from assembling F 1 in amounts sufficient to support growth on EG plates, were selected by the screen. Since most mutations found in the atp12 genes produce deletions, it would appear that Atp12p is relatively resilient to substitutions of individual residues. However, it was found that at least one position (residue 289) in the amino acid sequence may be of particular importance. Our study indicates that an acidic residue is required at this position for optimal Atp12p activity.
Identification of Domains in Atp12p-A schematic map of wild type Atp12p is presented in the upper part of Fig. 8. The mitochondrial targeting sequence (black-shaded region in the wild type Atp12p map) is estimated to be 30 amino acids long (Met-1 through Leu-30) on the basis of the relative migrations of the precursor and mature forms of Atp12p in SDS gels (18). Two domains were mapped in the mature protein using deletion mutants of Atp12p (for reference, Fig. 8 also includes schematic protein maps for the Atp12p deletion mutants and provides a summary of the characteristics of these proteins). The sequence between Gln-181 and Val-306 constitutes the functional domain of the protein (see hatched region in the wild type Atp12p protein map). The amino-terminal boundary of this domain was assigned on the basis of the results obtained with the mutant protein, Atp12(⌬1-180)p (see Atp12(⌬1-180)p map in Fig. 8). The overproduction of Atp12(⌬1-180)p confers conditional growth properties to the atp12 disruption strain. At the permissive temperature (23°C) this mutant protein fosters assembly of a functional mitochondrial ATPase and thereby enables the host strain to use non-fermentable carbon sources for growth. No evidence of F 1 assembly was found in yeast that produce Atp12p with a longer deletion from the amino terminus (see Atp12 (⌬1-224)p, Fig. 8). The carboxyl-terminal boundary of the functional domain in Atp12p is assigned to Val-306, The numbers above each blot indicate fractions of the gradient; sucrose density increases going from right (high numbers) to left (low numbers). Spectrophotometric and enzymatic assays, respectively, were used to define the peak positions in the gradients for hemoglobin (Hb, filled arrowhead) and myokinase (MK, open arrowhead) as described (21). Western analysis followed the procedures described in the legend to Fig. 4.  2). The experiment with purified Histag Atp12p (loaded as 2 g in 200 l of Tris-HCl, pH 7.5) was performed as described in the legend to Fig. 6. The conditions for the sedimentation experiment with the MBP-Histag-Atp12p fusion protein (loaded as 2 g in 200 l of Tris-HCl, pH 7.5) were the same with the following exceptions. First, 10 mM maltose was included in the buffer to prevent aggregation of the maltose binding protein, which oligomerizes in the absence of maltose (36). Second, lipoamide dehydrogenase (100 kDa) was included with hemoglobin as a size standard in the gradient; lipoamide dehydrogenase activity was assayed as described (18). Finally, 15 fractions of identical volume were collected from the bottom of the tube. Western analysis to determine the position of the recombinant Atp12p proteins in the gradients was done as described in the legend to Fig. 4, and the results are illustrated as in Fig. 6. which is the last retained amino acid in the mutant protein, Atp12(V306)p (see Atp12(V306)p map in Fig. 8). Notably, this mutant protein is significantly impaired for activity, as indicated by the phenotype of the yeast transformant that produces it from a multi-copy plasmid (Table V). When cultured in media containing the fermentable sugar, galactose, there was no evidence for functional mitochondrial ATPase (Table V). However, this strain displays partial respiratory function (which means there must be some level of assembled ATPase) as indicated by the slow growth phenotype on EG plates (discussed under "Results"). Therefore, it is proposed that the minimal sequence necessary for the action of Atp12p in F 1 assembly is contained within the Atp12(V306)p protein.
The results from sedimentation analysis with mitochondrial samples that harbor variant forms of Atp12p identified a second domain in the protein, which is important for intermolecular associations. Native Atp12p is observed to form a higherordered oligomer in mitochondria (18). Deletions made from the amino terminus, and substitutions at Glu-289 in the sequence, did not show evidence for interfering with the ability of Atp12p to oligomerize (see data for Atp12(⌬1-124)p and the E289K, E289D, E289A, and E289Q proteins in Fig. 6). However, removing as few as 19 amino acids from the carboxyl terminus yields a mutant protein (Atp12(V306)p) that sediments like a monomer (Fig. 6). On this basis the carboxylterminal sequence between Asp-307 and Gln-325 is designated as a discrete domain in Atp12p that is involved in its oligomerization. The loss of this domain, which reduces the ability of Atp12(V306)p to form stable oligomers, likely explains the defective behavior of the protein (see above).
Studies performed to determine if mitochondrial Atp12p is a homo-or hetero-oligomer provided conflicting results initially. A preparation of (6x)histidine-tagged Atp12p produced in E. coli was found to show sedimentation properties similar to the native mitochondrial protein (Fig. 7), which was interpreted to be representative of homo-oligomerization. However, such putative homo-oligomers were not visualized in cross-linking studies performed with the purified recombinant protein.
There was also no evidence for intermolecular associations of Atp12p monomers using the yeast two-hybrid system. The ambiguity in the data was resolved by examining the sedimentation properties of a chimera (Fig. 2) between maltose binding protein and Histag-Atp12p. The MBP-Histag-Atp12p fusion protein purified from bacteria was shown to migrate as a monomer in linear sucrose gradients (Fig. 7). The results of this analysis, along with the chemical modification and genetic studies described above, argue against the formation of homo-oligomers of Atp12p. On this basis we suggest that the native form of Atp12p observed in mitochondrial samples is a hetero-oligomer.