Identification of a Nuclear Gene ( FMC1 ) Required for the Assembly/Stability of Yeast Mitochondrial F 1 -ATPase in Heat Stress Conditions*

We have identified a yeast nuclear gene ( FMC1 ) that is required at elevated temperatures (37 °C) for the forma-tion/stability of the F 1 sector of the mitochondrial ATP synthase. Western blot analysis showed that Fmc1p is a soluble protein located in the mitochondrial matrix. At elevated temperatures in yeast cells lacking Fmc1p, the a -F 1 and b -F 1 proteins are synthesized, transported, and processed to their mature size. However, instead of being incorporated into a functional F 1 oligomer, they form large aggregates in the mitochondrial matrix. Identical perturbations were reported previously for yeast cells lacking either Atp12p or Atp11p, two specific assembly factors of the F 1 sector (Ackerman, S. H., and Tzagoloff, A. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 4986–4990), and we show that the absence of Fmc1p can be efficiently compensated for by increasing the expression of Atp12p. However, unlike Atp12p and Atp11p, Fmc1p is not required in normal growth conditions (28– 30 °C). We propose that Fmc1p is required for the proper folding/stability or functioning of Atp12p

F 1 F o -ATP synthases play a major role in cellular energy production. They are found in the plasma membranes of bacteria, thylakoid membranes of chloroplasts, and in the inner membrane of mitochondria. They use a proton gradient across their host membrane to produce ATP from ADP and inorganic phosphate (1,2). This enzyme contains two distinct parts, called F o and F 1 . The F o mediates the transmembrane transport of protons, and the synthesis of ATP takes place on the F 1 .
The F 1 contains five different types of subunits in the stoichiometric ratio ␣ 3 ␤ 3 ␥␦⑀ (3,4). The three-dimensional structures of F 1 from bovine heart (5), rat liver (6) and yeast (7) show that the ␣and ␤-subunits alternate in a hexagonal array with a central cavity occupied by the amino and carboxyl termini of the ␥-subunit. The interfaces between the ␣and ␤-subunits form three catalytic and three noncatalytic nucleotide binding sites.
In the yeast Saccharomyces cerevisiae, the F 1 subunits are encoded in the nucleus (8 -12), synthesized in the cytoplasm, imported into mitochondria as unfolded polypeptide chains (13), and then folded in the mitochondrial matrix with the help of Hsp60p and Hsp10p (14). The oligomerization of the F 1 monomers is assisted by two proteins called Atp12p and Atp11p. These interact directly with the ␣-F 1 and ␤-F 1 proteins, respectively (15,16). In yeast strains lacking either Atp11p or Atp12p, the ␣-F 1 and ␤-F 1 proteins aggregate in the mitochondrial matrix (17). Thus it is believed that Atp12p and Atp11p facilitate the formation of ␣␤ heterodimers by protecting these two F 1 subunits from non-productive interactions (16).
We report in this study the identification of Fmc1p, a novel protein required for the formation or stability of the F 1 oligomer. Like Atp11p and Atp12p, its absence also results in the aggregation of the ␣-F 1 and ␤-F 1 proteins. However, this is seen only at elevated temperatures (37°C), whereas Atp11p and Atp12p are required both in normal (28 -30 C°) and heat stress conditions. Interestingly, the formation/stability of the F 1 oligomer was restored in cells lacking Fmc1p by increasing the expression of Atp12p. We propose that Fmc1p assists the folding/stability or functioning of Atp12p and that this role becomes essential at elevated temperatures.

Construction of Yeast Strains Carrying a Null Mutation in FMC1
FMC1 was deleted according to the procedure described by Wach (18

Isolation of ATP12 as a Multicopy Suppressor of the Null Allele ⌬fmc1
The ⌬fmc1 strain MC6 was transformed with a partial HindIII digest of yeast chromosomal DNA cloned into the URA3-containing multicopy vector pEMBLYe23 (a gift of Dominique Thomas, Gif-sur-Yvette). About 100,000 Ura ϩ transformants were selected on SC medium lacking uracil and were then replica-plated onto glycerol medium (N3) prewarmed for at least 2 h at 37°C. After an incubation of 5 days at 37°C, clones that showed good growth were retested for their growth on glycerol at 37°C after curing the plasmids they contained on SC medium supplemented with 0.1% of 5-fluoroorotic acid. When growth on glycerol at 37°C was plasmid-dependent, the suppressor plasmids were recovered in E. coli and tested again by retransformation of ⌬fmc1 cells. Several rescuing plasmids contained the same insert, a HindIII-HindIII 1599-base pair fragment corresponding to a segment of chromosome X between coordinates 87249 and 88848 (this plasmid has been called pLE10). Since ATP12 was the only gene present on this fragment, it can be concluded that it was responsible for the suppressor activity of pLE10.

Genetic and Molecular Biology Methods
Genetic experiments were carried out as described by Rose et al. (21). Standard recombinant DNA techniques were used as described by Sambrook et al. (22). Yeast transformations were performed using described procedures (23). Yeast genomic DNA was isolated as described previously (24). Yeast mitochondrial RNAs were isolated as described by di Rago et al. (25)

Biochemical Techniques
Mitochondria were prepared by the enzymatic method described by Guérin et al. (29). Alkaline extraction of mitochondrial proteins with sodium carbonate at pH 11.5 was performed as described by Rouillard et al. (30). Protein concentrations were determined by the procedure of Lowry et al. (31) in the presence of 5% SDS using bovine serum albumin as standard. The specific ATPase activity was measured at pH 8.4 as described by Somlo (32). Oxygen consumption rates were measured in the respiratory medium (0.65 M mannitol, 0.3 mM EGTA, 3 mM Tris phosphate, 10 mM Tris maleate, pH 6.75) as described by Rigoulet and Guérin (33). Variations in transmembrane potential (⌬⌿) were evaluated in the same medium by measurement of the fluorescence quenching of rhodamine 123 with an SFM Kontron fluorescence spectrophotometer (34). Immunoprecipitation experiments were made from 2 mg of mitochondrial proteins as described by Todd et al. (35). Polyclonal antibodies raised against the ␣-F 1 subunit were added to the 100,000 ϫ g supernatant of the 0.375% Triton X-100 mitochondrial extract. The immunoprecipitated proteins were washed with a buffer containing 0.1% Triton X-100 (w/v), 150 mM NaCl, 10 mM sodium phosphate, pH 7.0. The final pellet was dried under vacuum and dissolved in 20 l of dissociation buffer devoid of reducing agent. SDS-PAGE was performed according to Laemmli (36). The slab gel was silver-stained as described by Ansorge (37). Western blot analyses were performed as described previously (38). Sedimentation analysis in sucrose gradients was performed as described by Ackerman and Tzagoloff (17), except that the gradient was centrifuged at 48,000 rpm for 3 h in a Beckman SW55 Ti rotor. Polyclonal antibodies raised against yeast ATP synthase subunits, Aac2p, Atp12p, and Fmc1p were used after dilution 1:10,000; 1:10,000, 1:2,500, and 1:5,000, respectively. ProBlott membranes were incubated with peroxidase-labeled antibodies at a dilution of 1:10,000 and revealed with the ECL ϩ reagent of Amersham Pharmacia Biotech. The chloroform/methanol extraction of subunits 6, 8, and 9 of ATP synthase was as described by Michon et al. (39). Mitochondrial proteins were labeled in cycloheximide-blocked cells according to Claisse et al. (40). Mitoplasts were prepared according to Daum et al. (41).

Production of anti-Fmc1p Antibodies
Anti-Fmc1p antibodies were prepared by Eurogentec (Seraing, Belgium) with the synthetic peptide NH 2 -YNPGNKLTQDEK as an immunogen.

RESULTS
During a systematic program of functional analysis of newly discovered yeast genes, we found that a null allele of the ORF YIL098c (which we have called FMC1 for formation of mitochondrial complexes 1) impairs the ability of yeast to grow at 37°C on media containing a nonfermentable carbon source such glycerol or ethanol (in W303-1B and with HIS3 as the inactivation marker) (Fig. 1). Three lines of evidence show that this phenotype is due to the inactivation of FMC1 as follows. (i) The correct integration of the deletion cassette at the FMC1 locus was confirmed by Southern blot analysis (not shown); (ii) the mutant phenotype and histidine prototrophy cosegregated 2:2 in tetrads originating from heterozygous ⌬fmc1/ϩ cells (not shown); and (iii) after transformation with a plasmid containing FMC1 (pLL1), ⌬fmc1 cells recovered a normal growth on ethanol or glycerol at 37°C. As oxidative phosphorylation is essential for growing yeast on glycerol or ethanol, these data indicated that FMC1 may be needed for the formation or functioning of the yeast energy transducing system at elevated temperatures.
We first aimed to determine whether FMC1 is needed for pre-mRNA splicing or the propagation of mtDNA at elevated temperatures. To this end, we constructed a ⌬fmc1 strain containing an intronless mtDNA (strain MC6, see Table I). This strain was still unable to grow on glycerol or ethanol at 37°C (Fig. 1), showing that the absence of respiratory growth is not due to a defect in mitochondrial pre-mRNA splicing.
The propagation of the mtDNA by the mutant was analyzed by determining the production of cytoplasmic petites 2 (rho Ϫ / rho°cells) during fermentation at elevated temperatures. Fresh cells were grown at 28°C in glycerol (selecting for a rho ϩ mitochondrial genome), then grown for 5-6 generations in glucose at 28 or 37°C. The cultures were then plated at 28°C on a glycerol medium containing 0.1% glucose (on this medium 1 The abbreviations used are: PCR, polymerase chain reaction; F o and F 1 , integral membrane and peripheral portions of ATP synthase; ⌬⌿, transmembrane electrical potential; DCCD, dicyclohexylcarbodiimide; PAGE, polyacrylamide gel electrophoresis; mtDNA, mitochondrial DNA. 2 A petite designates a mutant yeast cell that fails to grow on a non-fermentable carbon source (glycerol or ethanol). On a 2% glycerolmedium containing low amounts of glucose (0.1%), such cells can makeonly small size colonies and are therefore called petites. cells that fail to produce ATP by oxidative phosphorylation give very small colonies). This analysis was performed for wild type and ⌬fmc1 strains, with either a wild type (intron-containing) or intronless mitochondrial genome. The results presented in Table II show that deletion of FMC1 gene leads to a 5-7-fold increase in petite production in the presence of both an introncontaining and intronless mitochondrial genome at 37°C. However, it is clear that the level of petite production seen at 37°C, especially in the presence of an intronless mitochondrial genome, is not sufficient to explain the respiratory-deficient phenotype. We conclude that the increased production of petites is probably a secondary effect of the ⌬fmc1 mutation.
When the ⌬fmc1 strain was grown for a longer time on glucose at 37°C (up to 15 generations), higher levels of petites were not observed. This suggests that the ⌬fmc1 petites grow poorly at 37°C, preventing them from completely taking over the cultures. To test this, ⌬fmc1 cells were made rho°by growing them in the presence of ethidium bromide at 28°C and then transformed with a plasmid bearing FMC1 or with the corresponding empty vector. The resulting transformants were then tested for their growth on glucose at 37°C. At 37°C growth of the ⌬fmc1 rho°, cells transformed with the empty vector was severely affected compared with that of the ⌬fmc1 rho°cells transformed with the cloned FMC1 gene (Fig. 2). Consistent with this, we found that rho ϩ ⌬fmc1 cells grew poorly at 37°C in the presence of ethidium bromide (not shown). Thus when S. cerevisiae lacks FMC1, it needs a wild type mtDNA to grow normally by fermentation at elevated temperatures.
Next we analyzed the expression of the mtDNA in the ⌬fmc1 mutant with an intronless mtDNA (to reduce problems associated with petite production). In a Northern blot analysis, probes specific for the genes CoxI, CoxII, ATP9, ATP6, and 21 S RNA produced similar radioactive signals in the mutant and wild type, indicating that FMC1 is not involved in transcription of the mitochondrial genes at elevated temperatures (Fig. 3A). We then determined the ability of ⌬fmc1 cells to synthesize the mitochondrially encoded proteins at 37°C. This was done with whole cells grown by fermentation for five to six generations at 37°C. These were incubated at 37°C for varying times (5, 15, or 30 min) with radiolabeled Met ϩ Cys in the presence of cycloheximide (an inhibitor of cytoplasmic protein synthesis that has no effect on the mitochondrial ribosomes). Mitochondrial membranes were then prepared and analyzed by SDS-PAGE and autoradiography. This revealed that the mutant grown at 37°C was able to normally synthesize six of the eight proteins encoded by the mtDNA (cytochrome b, Cox1, Cox2, Cox3, Var1, and ATP6), whereas the radioactive signals corresponding to ATP9 and ATP8 were significantly weaker in comparison to the wild type (even when the time of incorporation of the radiolabeled amino acids was reduced to 5 min) (Fig. 3B). In the experimental conditions used, it is difficult to determine whether the rate of synthesis of ATP9 and ATP8 was reduced in the mutant and/or whether they were less efficiently incorporated into mitochondrial membranes or more prone to proteolytic degradation (see below).
Respiratory, ATPase, and Proton-Pumping Activities of ⌬fmc1 Mitochondria-To further understand the role of FMC1 in oxidative phosphorylation, we analyzed the energy conversion capacity of mitochondria isolated from cells lacking this gene. For this, we used a ⌬fmc1 strain containing an intronless mtDNA. Mitochondria isolated from the mutant and wild type strains grown at 28°C exhibited the same oxygen consumption rates (see Table III). By contrast, when the mutant was grown at 37°C, its respiratory activity was substantially reduced (by 70%) in comparison to the wild type strain cultivated at the same temperature.
Mitochondria from the mutant and wild type strains cultivated at 28°C had approximately the same DCCD-sensitive ATP hydrolytic activity (Table III). By contrast, ⌬fmc1 cells grown at 37°C had a very low mitochondrial ATPase activity (less than 10% of the control). This residual activity was essentially insensitive to DCCD and also about two times lower than that of mitochondria isolated from wild type rho°cells cultivated at 37°C. Thus, the F 1 sector of the ATP synthase is severely affected in yeast cells lacking FMC1 grown at 37°C, whereas it is fully active when these cells are grown at 28°C.
It should be noted that the ATPase activity measured for the wild type grown at 37°C was much less sensitive to DCCD than that of wild type cells grown at 28°C (see Table III). This suggests that a substantial part of the F 1 sector is not physically or functionally coupled with the F o sector when wild type yeast is cultivated at elevated temperatures. This observation has been reported previously (42).
The proton-pumping activities of ⌬fmc1 mitochondria were probed by fluorescence quenching of rhodamine 123. The Fig. 4 shows the results obtained with mitochondria isolated from mutant and wild type cells grown at 37°C. For the wild type, the addition of ethanol produced a fluorescence quenching of the dye, which was transiently decreased by adding 50 M ADP, thus reflecting an electrogenic exchange of internal ATP against external ADP and a proton influx through the F o during phosphorylation of the added ADP. By contrast, with mutant mitochondria, although ethanol was still able to energize the membrane, a subsequent addition of ADP could not substantially decrease the membrane potential (⌬⌿). Changes in ⌬⌿ mediated by the ATPase proton-pumping activity were analyzed after energizing mitochondria by ethanol, an activation step necessary to remove the natural inhibitory peptide (IF1) of the mitochondrial ATPase, which would otherwise in- hibit ATPase activity (43). As expected, with wild type mitochondria, inhibition by KCN of the proton pumping by the respiratory chain resulted in a collapse of ⌬⌿, and subsequent addition of ATP promoted a fluorescence quenching of the dye that was DCCD-sensitive. With mutant mitochondria, an addition of ATP after collapsing ⌬⌿ by KCN promoted a lower increase in ⌬⌿ that was almost insensitive to DCCD.
Taken together, the results of these different analyses show that oxidative phosphorylation is severely affected at elevated temperatures when the FMC1 gene is absent, with the most dramatic consequences seen at the level of the ATP synthase.
FMC1 Is Required for the Assembly or Stability of the ATP Synthase at Elevated Temperatures-As described above, the respiratory growth defect of the ⌬fmc1 strain is probably due to a failure in mitochondrial ATP synthesis. To determine whether this was caused by a block in the assembly of the ATP synthase, we examined the steady state levels of several subunits of this enzyme in the mutant grown at elevated temperatures. This was done either by Western blot analysis of whole mitochondria or by silver staining of mitochondrial proteins extracted with a mixture of chloroform and methanol (Fig. 5, A  and B). The levels of the F1-␣ and F1-␤ proteins were apparently not affected by the absence of Fmc1p. By contrast, the amounts of all the other subunits tested (␥, ␦, su.4, su.d, oligomycin sensitivity-conferring protein (OSCP), ATP6, ATP8, ATP9, su.i, su.g, and su.f) were strongly reduced in the mutant. Since synthesis of the mitochondrially encoded ATP6 is normal in the mutant (see above), the near complete absence of this protein at the steady state is probably due to proteolytic degradation. A higher susceptibility to proteolysis is probably also responsible for the virtual absence of the two other subunits of mitochondrial origin (ATP9 and ATP8), although these two proteins may be synthesized at a slower rate (see above). We do not know whether the missing subunits of nuclear origin are synthesized in the ⌬fmc1 mutant, but it is reasonable to assume that their absence also results from proteolytic degradation after to a block in the assembly of the enzyme.
It has been shown previously that in the absence of either Atp12p or Atp11p, two chaperones specifically involved in the assembly of the F 1 sector of the ATP synthase, the F 1 -␣ and F 1 -␤ proteins accumulate normally, but instead of being incorporated into a functional F 1 oligomer, they form large aggregates in the mitochondrial matrix (17). Interestingly, a significant portion of the F 1 -␣ and F 1 -␤ proteins also behaved as large protein aggregates in the ⌬fmc1 mutant, as shown by sucrose gradient analysis (Fig. 5C).
Immunoprecipitation with antibodies against the F 1 -␣ protein were made to see whether this protein was associated with the remaining ATP synthase subunits in the ⌬fmc1 mutant. The immunoprecipitates obtained from mitochondria isolated from the mutant grown at 37°C contained only the F1-␣ and F1-␤ subunits (not shown). This indicates that these two proteins are part of an entity that is not, or is poorly, associated with the remaining subunits of the enzyme.
Fmc1p Is a Soluble Mitochondrial Protein-An analysis of the 155-amino acid (18,352 Da) sequence deduced from FMC1 using the P-sort program of Nakai and Kanehisa (44) indicates that Fmc1p is a mitochondrial protein. This prediction was confirmed with the use of antibodies raised against a 14-amino acid peptide corresponding to the nucleotide sequence of FMC1 between positions 390 and 432. A Western blot analysis of mitochondrial proteins isolated from wild type cells using these antibodies produced a specific 14-kDa signal that could not be detected in the ⌬fmc1 mutant (Fig. 6A). The apparent size of the protein indicates that Fmc1p is synthesized as a precursor containing an amino-terminal presequence of ϳ4 kDa. After an osmotic disruption of both the outer and inner mitochondrial membranes in the presence of carbonate, the protein was recovered in a water-soluble form (Fig. 6B). After disruption of just the outer membrane by a mild osmotic treatment of mitochondria, a portion of the immunological signal was recovered in the mitoplast fraction and preserved after treatment of the mitoplasts with proteinase K (the recovery of a part of the signal in the supernatant fraction is probably due to partial damage of the mitoplasts) (Fig. 6C). These data indicate that Fmc1p is a protein of the mitochondrial matrix, either free or loosely bound to the inner face of the inner mitochondrial membrane.
The ⌬fmc1 Mutant Can Be Rescued by Increasing the Copy Number of ATP12-The results described above show that Fmc1p is a mitochondrial protein needed at elevated temperatures for the assembly or stability of the F 1 sector of the ATP synthase. To gain more insight into its function, we decided to determine whether the loss of Fmc1p can be compensated for by overexpressing another yeast protein(s), a current approach to identifying proteins with related cellular functions. To this end, the ⌬fmc1 mutant was transformed with a yeast wild type genomic library in a high copy number vector, and the resulting transformants were tested for their growth on glycerol at 37°C. In that way, we found that overexpression of Atp12p efficiently rescued the ⌬fmc1 phenotype (Fig. 7).
Somewhat surprisingly, ⌬fmc1 cells transformed with ATP12 on a low copy number vector also show a nearly wild type growth on glycerol at 37°C. Thus, a small increase in the expression of Atp12p may be apparently sufficient to overcome the absence of Fmc1p at elevated temperatures. However, in vitro, the F 1 -ATPase activity was found to be restored only partially (50% in comparison to the wild type control; data not shown). By contrast, this activity was fully restored with ATP12 on a high copy number vector. It is well known that a substantial decrease of the ATP synthase activity (up to 80%) has only marginal effects on the growth of yeast on non-fermentable carbon sources at temperatures above 20°C (45). This could explain that despite their reduced F 1 -ATPase activity, ⌬fmc1 cells transformed with ATP12 on a low copy number vector have a wild type growth on glycerol at 37°C.
These results are particularly interesting. First, given the very specific action of Atp12p in the assembly of the F 1 oligomer (16,17), they strongly suggest that Fmc1p has a function also confined to this process. Second, because the ⌬fmc1 mutant exhibits at elevated temperatures all the characteristics of the ⌬atp12 mutant and because the absence of Fmc1p can be overcome by increasing the production of Atp12p, a logical view is that the function of Atp12p is impaired at elevated temperatures when Fmc1p is missing. Consistent with this, we found that the abundance of Atp12p was significantly reduced in FIG. 2. In the absence of mtDNA, the ⌬fmc1 mutant grows poorly by fermentation at elevated temperatures. The ⌬fmc1 mutant was made rho°by growing at 28°C in the presence of ethidium bromide and was then transformed with a plasmid containing the FMC1 gene (pLL1) or the corresponding empty vector (pRS316). The transformants were grown for 2 days at 28°C in a glucose medium lacking uracil (SC-ura). The cultures were diluted, and 5 l of each dilution were spotted on SC-ura. The plates were incubated for 3 days at 37°C and then photographed.

FIG. 3. Analysis of the expression of mitochondrially encoded genes in the ⌬fmc1 mutant at elevated temperatures. Panel A,
Northern blot analysis of mitochondrial transcripts. Total mitochondrial RNAs were isolated from wild type FMC1 ϩ (strain MC1) and mutant ⌬fmc1 (strain MC6) grown for 6 -8 generations in galactose (YPGALA) at 37°C. The RNAs were separated in a formaldehydecontaining agarose gel and then transferred to a nitrocellulose membrane. The same blot was hybridized successively with 32 P-radiolabeled DNA probes specific for cytochrome b (Cyt b), CoxI, CoxIII, ATP6, and 21 S rRNA genes. Panel B, mitochondrial protein synthesis. Wild type FMC1 ϩ (strain MC1) and mutant ⌬fmc1 (strain MC6) cells were grown in galactose (YPGALA) for 6 -8 generations at 37°C and then incubated as indicated for 5Ј or 15Ј at 37°C with [ 35 S]methionine and [ 35 S]cysteine in the presence of cycloheximide. Mitochondrial membranes were then isolated and analyzed by SDS-PAGE and an autoradiograph of the gel (500,000 cpm were loaded on each lane). The photograph on the right is from a long run gel to increase the resolution of Cox3 and ATP6. mitochondria isolated from ⌬fmc1 cells cultivated at 37°C (Fig. 8). We have also shown that the multi-copy suppressor relationship between FMC1 and ATP12 is not reciprocal, as the ⌬atp12 mutant remained unable to grow on glycerol after transformation with FMC1 cloned in a high copy number vector. Altogether these data suggest that Fmc1p may be required for the formation/stability or functioning of Atp12p at elevated temperatures.
Since Atp11p cooperates with Atp12p in the assembly of the F 1 oligomer (15-17), we decided to determine directly whether Atp11p could, like ATP12, rescue the ⌬fmc1 mutant. ATP11 may have been poorly represented in the library we used and, hence, not isolated in the search of multicopy suppressors. We therefore constructed a high copy number vector containing ATP11. After transformation with this plasmid, the ⌬fmc1 mutant remained unable to grow on glycerol at 37°C (not shown). Also, we found that the ⌬atp11 mutant was not rescued by overexpression of Fmc1p. Thus, contrary to Atp12p, it seems that the proper folding/stability or functioning of Atp11p at elevated temperatures does not require the presence of Fmc1p. DISCUSSION We have identified a novel nuclear-encoded yeast mitochondrial protein and showed that its presence is required for the assembly/stability of the F 1 sector of the ATP synthase in heat stress conditions. In its absence and at elevated temperatures, the ␣-F 1 and ␤-F 1 proteins are synthesized, transported, and processed to their mature size, but instead of being incorporated into a functional F 1 oligomer, they form large aggregates in the mitochondrial matrix. Identical defects in the assembly of the F 1 oligomer were observed previously for yeast cells lacking either Atp12p or Atp11p (17). However, unlike Fmc1p, these two latter proteins are required for the formation of the F 1 oligomer not only at elevated temperatures but also in normal growth conditions.
The oxygen consumption rate of mitochondria isolated from the ⌬fmc1 mutant grown at elevated temperatures was strongly reduced (by 70%) in comparison to the wild type. This is probably a secondary consequence of the defect in the assembly of the F 1 oligomer. Indeed a decreased respiratory activity was also seen for strains carrying null mutations in the genes of ␣-F 1 , ␤-F 1 (8,9,46), Atp11p, and Atp12p (17).
Interestingly, the ⌬fmc1 mutant recovered the ability to assemble the F 1 oligomer at elevated temperatures by increasing the expression of Atp12p, suggesting that the function of Atp12p may be compromised at elevated temperatures when Fmc1p is lacking. Consistent with this, the steady state level of Atp12p was found to be substantially reduced in mitochondria isolated from ⌬fmc1 cells grown at elevated temperatures.  Based on these observations, a reasonable hypothesis is that Fmc1p helps the folding/stability or functioning of Atp12p at elevated temperatures.
The molecular mass of native Atp12p, estimated from its sedimentation properties in sucrose gradients, is at least twice as great as that of the monomer (47). Chemical modifications and two-hybrid genetic studies argue against the formation of oligomers of Atp12p (48). Furthermore, in a strain unable to express Atp11p, the native size of Atp12p was not found to be modified, indicating that Atp11p and Atp12p are not part of the same complex (49). Based on the results reported here, it will be particularly interesting to see whether Fmc1p belongs to or is required for the formation of the Atp12 oligomer. Experiments to resolve this question are in progress.
Atp12p and Atp11p interact directly with the F 1 -␣ and F 1 -␤ proteins, respectively, and this is presumed to facilitate the formation of ␣␤ heterodimers by protecting these two subunits of the ATP synthase from non-productive interactions (15,16). If Fmc1p actually mediates the ␣␤ dimerization by assisting Atp12p at elevated temperatures, how can we explain that  (39). The proteins of the organic phase were separated on a 15% SDS-PAGE and silver-stained. Panel C, the mitochondria were sonicated and then centrifuged through a discontinuous sucrose gradient as described by Ackerman and Tzagoloff (17). 15 l of each fraction were analyzed by SDS-PAGE and Western blot with antibodies against the ␣-F 1 , ␤-F 1 , and Aac2p proteins. OSCP, oligomycin sensitivity-conferring protein. Panel C, mitoplasts were prepared from fresh wild type mitochondria (4 mg) as described by Daum et al. (41). The pellet and supernatant fractions were each divided in two parts. One part of each fraction was treated with 20 g/ml proteinase K at 4°C for 15 min. Equivalent amounts of proteins from the four samples were analyzed by SDS-PAGE and Western blot with the indicated antibodies. antibodies against the F 1 -␣ protein coimmunoprecipitated the F 1 -␤ protein in mitochondrial extracts of the ⌬fmc1 mutant. As will be presented elsewhere, the F 1 -␣ and F 1 -␤ proteins were found by immunocytochemistry to be part of the same inclusion bodies in mitochondria of the ⌬fmc1 mutant. 3 Thus, even when they fail to interact properly, the F 1 -␣ and F 1 -␤ proteins still remained associated physically, which could explain their coimmunoprecipitation from the ⌬fmc1 mutant.
Severe perturbations at the level of the F o sector of the ATP synthase were seen in the ⌬fmc1 mutant. It is not known whether or not the F o sector is assembled in the ⌬atp11 and ⌬atp12 mutants. We believe that Fmc1p is not directly required for the assembly of this part of the enzyme. Rather the perturbations at the level of the F o sector in the ⌬fmc1 mutant were probably a secondary consequence of the defect in the assembly of the F 1 sector. Indeed as in the ⌬fmc1 mutant, the three mitochondrially encoded subunits of the ATP synthase, which are all essential components of the F o sector, failed to accumulate in a mutant carrying a null mutation in the F 1 -␤ protein gene (42). Thus, whereas the F 1 sector can assemble independently of the F o sector, for example in yeast cells lacking mtDNA, it appears that the formation or stability of the F o sector may depend on the presence of an assembled F 1 sector. A sequential assembly of the F 1 and F o sectors could be of critical importance in preventing the depolarization of the inner mitochondrial membrane during the assembly of the ATP synthase complex.
At 37°C, the growth of the ⌬fmc1 mutant on glucose was severely affected when it lacked wild type mitochondrial DNA. This phenotype is most likely related to the defect in the assembly of the F 1 oligomer. Indeed, null mutations in the genes of ␣-F 1 , ␤-F 1 , Atp12p, and Atp11p also impair the growth of yeast lacking mtDNA (for review, see Ref. 50). The basis of the so-called "petite negativity" conferred by mutations affecting the F 1 sector is not totally understood, but evidence suggest that in the absence of a functional respiratory chain, the ATP hydrolytic activity of the F 1 sector would be needed for a correct polarization of the inner mitochondrial membrane and, hence, for the biogenesis of mitochondria (50,51).
Taken together, the data reported in this study suggest that Fmc1p functions only at elevated temperatures. If this is true, one might expect that the protein would only be expressed in such conditions. However, this was not found to be case (accumulation of the mRNA and the protein are about the same at 28 -30°C and 37°C, data not shown). There are cases known where a protein whose accumulation is not influenced by temperature plays a critical role in heat stress conditions. For example, subunit 6 of yeast complex III is dispensable at 28°C, whereas at 37°C, its absence has dramatic consequences on the assembly/stability of this complex (52). Thus it is possible that Fmc1p is involved in the assembly/stability of the F 1 -ATPase at all temperatures but that this role becomes critical only in heat stress conditions.