|Home | About | Journals | Submit | Contact Us | Français|
F1-ATPase is a rotary molecular machine with a subunit stoichiometry of α3β3γ1δ1ε1. It has a robust ATP-hydrolyzing activity due to effective cooperativity between the three catalytic sites. It is believed that the central γ rotor dictates the sequential conformational changes to the catalytic sites in the α3β3 core to achieve cooperativity. However, recent studies of the thermophilic Bacillus PS3 F1-ATPase have suggested that the α3β3 core can intrinsically undergo unidirectional cooperative catalysis (T. Uchihashi et al., Science 333:755-758, 2011). The mechanism of this γ-independent ATP-hydrolyzing mode is unclear. Here, a unique genetic screen allowed us to identify specific mutations in the α and β subunits that stimulate ATP hydrolysis by the mitochondrial F1-ATPase in the absence of γ. We found that the F446I mutation in the α subunit and G419D mutation in the β subunit suppress cell death by the loss of mitochondrial DNA (ρo) in a Kluyveromyces lactis mutant lacking γ. In organello ATPase assays showed that the mutant but not the wild-type γ-less F1 complexes retained 21.7 to 44.6% of the native F1-ATPase activity. The γ-less F1 subcomplex was assembled but was structurally and functionally labile in vitro. Phe446 in the α subunit and Gly419 in the β subunit are located on the N-terminal edge of the DELSEED loops in both subunits. Mutations in these two sites likely enhance the transmission of catalytically required conformational changes to an adjacent α or β subunit, thereby allowing robust ATP hydrolysis and cell survival under ρo conditions. This work may help our understanding of the structural elements required for ATP hydrolysis by the α3β3 subcomplex.
Mitochondria are the powerhouses of the cell that synthesize the majority of ATP required to support cellular activities under aerobic conditions. Although mitochondria are known to be essential for cell viability and mitochondrial stress induces cell death, the key mitochondrial factors that determine cell viability under specific stress conditions are not well established. One of the classic inducers of mitochondrial stress is the loss of mitochondrial DNA (mtDNA). In the budding yeast Saccharomyces cerevisiae and some other closely related species, petite mutants with extensively deleted mtDNA (ρ−) or a complete loss of mtDNA (ρo) are viable (1–5). However, most yeast species, known as petite negative, do not survive the loss of mtDNA on exposure to mutagens. The loss of cell viability is not caused by a bioenergetic crisis. For instance, a petite-negative yeast such as Kluyveromyces lactis can survive on glucose medium after disruption of the cytochrome c gene but cannot survive the elimination of mtDNA (6). K. lactis is therefore capable of synthesizing ATP to support cell growth via fermentative glycolysis, so lethality from ρo (ρo lethality) is caused by factors other than ATP depletion.
Our previous studies have shown that specific nuclear mutations can suppress lethality from ρo in K. lactis. These mutations, referred to as mgi (for mitochondrial genome integrity), occur in the α, β, and γ subunits of the mitochondrial FoF1-ATP synthase (7–9). The ATP synthase is a rotary molecular machine that synthesizes ATP from ADP and inorganic phosphate by using the proton gradient generated by the electron transport chain (10, 11). It consists of the transmembrane Fo domain and the F1-ATPase extending into the mitochondrial matrix. The catalytic F1 contains nine polypeptides with the stoichiometry of α3β3γδε. The readmission of protons from the intermembrane space to the mitochondrial matrix through Fo drives the rotation of γ within the α3β3 hexameric ring, which induces the conformational changes necessary for the synthesis of ATP (12, 13). The F1-ATPase can be dissociated from the membrane-embedded Fo. In free F1, γ rotates in the reverse direction, allowing hydrolysis but not synthesis of ATP. It has been believed that the rotating γ plays a key role in coordinating sequential conformational changes in the three catalytic sites, which confers a highly cooperative and robust mechanism for ATP hydrolysis.
In yeast cells with ρo (ρo cells), Fo is eliminated because of the loss of the mtDNA-encoded Atp6, -8, and -9 subunits. The gain-of-function mgi mutations confer on the remaining F1 domain a novel property that permits cell survival. A widely accepted model is that the mutant F1 remains active in hydrolyzing ATP−4 that is glycolytically generated in the cytosol before being imported into the matrix. The mutant F1, but not the wild type F1, hydrolyzes ATP−4 to ADP−3, which is subsequently exported to the cytosol. This reversed exchange of cytosolic ATP−4 against the matrix ADP−3 via the adenine nucleotide translocase is electrogenic, which allows the mitochondria to maintain the minimal membrane potential (Δψm) required for mitochondrial biogenesis (14–16). A large genetic screen has identified 35 mgi alleles in K. lactis (9). A similar mutation in the γ subunit of F1-ATPase that suppresses the slow-growth phenotype of ρo cells in the yme1 background has also been reported in Saccharomyces cerevisiae (17). Interestingly, all the mutable residues are mapped to two specific regions, with one suggested to be a molecular bearing for the rotation of γ within the α3β3 hexamer and the other being proximal to the membrane close to the conserved DELSEED loop in β and its equivalent in α. At these two particular locations, all the mutated amino acids are situated on the interface between the α, β, and γ subunits. This suggests that the suppressor mutations affect subunit-subunit interactions. The mutant F1 appears to have a lower Km for ATP than the wild-type enzyme (18). The altered kinetic property may allow the mutant complexes to continue to hydrolyze ATP in ρo cells in which the matrix ATP concentration might be low. By introducing the mgi mutations into the S. cerevisiae F1 complex, the Mueller group found that the mutant ATP synthase is uncoupled (19). Four mutant F1 complexes were crystallized, and their structures were analyzed (20). Interestingly, the mgi mutations were found either to affect substrate (phosphate) binding or to reduce the steric hindrance imposed by the β subunit when γ rotates in the α3β3 core.
In an attempt to identify nuclear mutations that potentially suppress ρo lethality in the absence of ATP hydrolysis by F1-ATPase, we screened suppressor mutations in a strain lacking the γ subunit. Surprisingly, this unique genetic screen yielded two mutations that are again mapped to the α and β subunits. These mutations apparently confer a robust ATP-hydrolyzing activity in the absence of γ. This finding provides strong in vivo evidence supporting the newly emerging paradigm that robust catalysis can take place in the α3β3 core.
Yeast cells were grown in complete glucose medium containing 1% yeast extract, 2% Bacto peptone, and 2% glucose (YPD) or in minimum medium containing 6.7 g/liter Bacto yeast nitrogen base without amino acids (Difco) and 2% glucose. Nutrients essential for auxotrophic strains were added at 25 μg/ml for bases and 50 μg/ml for amino acids. Complete glycerol-ethanol medium and glycerol medium contained 1% yeast extract, 2% Bacto peptone, and 2% glycerol with or without 2% ethanol (YPGE and YPG, respectively). Where indicated, ethidium bromide (EB) was added at a concentration of 16 μg/ml. G418 medium was YPD plus G418 at 200 μg/ml. For sporulation of K. lactis, ME medium contained 5% malt extract and 2% Bacto agar.
The yeast strains used in this study are listed in Table 1. Standard procedures were used for the construction of K. lactis strains. Transformation of K. lactis was performed by the lithium acetate-dimethyl sulfoxide method as previously described in detail (21). For selection of G418-resistant transformants, transformed cells were grown at room temperature overnight before being plated on G418 medium. Disruption of the chromosomal genes was achieved by a one-step gene replacement procedure. To construct the atp3Δ::URA3 allele, pBS-MGI5, a pBluescript-based plasmid containing the entire ATP3 gene, was digested with ClaI. The 0.75-kb ClaI fragment comprising the N-terminal two-thirds of the ATP3 sequences and part of the promoter region was deleted and replaced by the 1.1-kb URA3 gene (see Fig. 1B). The resulting plasmid, pMGI5Δ::URA3, was digested with EcoRI to release the 1.5-kb atp3Δ::URA3 cassette, which was subsequently used for disruption of ATP3 in various strains. The atp3::kan cassette was isolated from plasmid pMGI5::kan/3 in which ATP3 was disrupted by insertion of kan at codon 161.
The plasmids pCXJ4-MGI2 and pCXJ4-MGI1/7 are based on the K. lactis integrative vector pCXJ4 (X. J. Chen, unpublished data) and contain K. lactis ATP1 (KlATP1) and K. lactis ATP2 (KlATP2), respectively. The two plasmids were linearized with ClaI, which cuts within the K. lactis LEU2 gene, before being integrated onto the chromosome by selecting for Leu+ transformants. pUK-KlMGI2/HP and pCXJ22-KlATP2 are multicopy plasmids for K. lactis containing KlATP1 and KlATP2, respectively.
CK392/7 (atp1-7) and CK391/18 (atp2-12) are random spores derived from CW15-2A and CW14-1A, respectively. Replacement of ATP3 by atp3::kan in the two strains yielded CK401 and CK400, respectively. Successive disruption of ATPδ and ATPε in CK401/1 and CK400 by the atpδ::URA3 and atpεΔ::LEU2 cassettes produced the triple-disrupted strains CK412/1 and CK410/6, respectively.
For His6 tagging of ATP1 in the atp3Δ background, we constructed the plasmid pCXJ4-sym2.1(atp1-7)His, in which the allele with the F-to-I change at position 446 encoded by atp1 (atp1F446I) was tagged with His6 on the C terminus. The plasmid was linearized with ClaI within the LEU2 gene and integrated into the leu2 locus of CK333 (αΔ γΔ) by selecting for Leu+ transformants. This generated strain TL2/2. Correct integration of the plasmid was confirmed by PCR-based genotyping.
Mitochondria were isolated as described by Boldogh and Pon (22). Yeast cells were grown in high-glucose minimum medium, washed with Tris-SO4-dithiothreitol (DTT) buffer, treated with 5 mg zymolyase per gram of yeast to create spheroplasts, and then broken with a 40-ml Dounce homogenizer. Mitochondria were collected by differential centrifugation.
His6-tagged F1-ATPases were purified by Ni column chromatography as described by Mueller et al., with modifications (23). Briefly, crude mitochondria were resuspended in sonication buffer (0.25 M sucrose, 20 mM HEPES-KOH, pH 7.4, 5 mM ε-amino caproic acid, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride [PMSF]) at a ratio of 1:2, by volume. Sonication was performed on ice 4 times for 1 min each time, with 1-min intervals used between each sonication. The inverted mitochondrial solution was centrifuged at 6,400 rpm at 4°C for 5 to 10 min. The supernatant was then collected and added to a Ni-nitrilotriacetic acid (NTA) slurry containing 50% beads and 50% buffer A (0.05 M NaCl, 0.25 M sucrose, 20 mM HEPES-KOH, pH 7.4, 5 mM ε-amino caproic acid, and 1 mM PMSF plus 3% buffer B [0.1 M imidazole, 0.05 M NaCl, 0.25 M sucrose, 20 mM HEPES-KOH, pH 7.4, 5 mM ε-amino caproic acid, 1 mM PMSF]). The ratio of sample solution to Ni-NTA slurry was 60:1, by volume. The solution was gently rocked at 4°C for 4 to 5 h before it was centrifuged at 2,000 × g for 1 min at 4°C. The beads were washed twice with buffer A containing 3% buffer B, and F1-ATPases were eluted from the Ni beads 4 times using 0.3 ml buffer B per elution. The wild-type F1-ATPase was also purified as described by Mueller and coworkers before being analyzed by size exclusion chromatography (23).
Blue native (BN)-PAGE was performed using a Life Technologies (Carlsbad, California) native PAGE Novex bis-Tris gel system as recommended by the manufacturer. The nickel column pull-down products were separated on a 3 to 12% polyacrylamide gradient gel at 4°C. The final concentration of the native PAGE G-250 sample additive used was 0.5%. No other detergent was needed. Western blotting was performed using antibodies specific for the α or β subunit of the yeast F1-ATPases. Signals were detected using an enhanced chemiluminescence detection method (Roche and Thermo Scientific). For BN-PAGE of the mitochondrial lysate, the final concentration of digitonin used to solubilize mitochondria was 2%, and the concentration of native PAGE G-250 sample additive used was 1/3 of the digitonin concentration.
The His6-tagged γ-less F1-ATPases from strains TL1/1 and TL2/2 were prepared as described above. The sonication buffer used to resuspend crude mitochondria contained no EDTA to prevent the stripping of Ni from the column. Sonication was performed on ice 4 times at half cycle for 30 s each time, with 30-s intervals used between each sonication. The solution containing inverted mitochondria was centrifuged at 13,000 rpm at 4°C for 30 min. Clear supernatant was collected and incubated with 1 ml of Ni-NTA slurry containing 50% beads and 50% buffer A for 30 to 60 min at 4°C. The mixture was then passed through a vertical column, washed 3 times with buffer A, and eluted 6 times with 500 μl buffer B containing 200 mM instead of 100 mM imidazole. The proteins were concentrated and analyzed by size exclusion chromatography on a calibrated Superdex 200 10/30 column equilibrated with 50 mM NaCl, 20 mM HEPES, pH 7.4, 5 mM ε-amino caproic acid, 1 mM DTT, 1 mM ATP, 250 mM sucrose, and 1 mM EDTA. Peak fractions were analyzed by SDS-PAGE and Western blotting.
ATPase activity was measured as described by Taussky and Schorr (24), with modifications. Five to 10 μl (approximately 0.1 to 0.2 mg) of mitochondria was added to 3 ml of 10 mM Tris-HCl (pH 8.2), 5 mM ATP, 2 mM MgCl2, and 200 mM KCl with or without oligomycin at 10 μg/ml. The reaction mixtures were incubated at 30°C for 10 min before quenching with 10% SDS. All samples were gently vortexed to avoid inactivation of enzymes. ATPase activity is expressed as μmol ATP hydrolyzed/min/mg protein.
In S. cerevisiae, the γ subunit is required for the assembly of F1-ATPase (25). In an attempt to isolate mutations that suppress ρ° lethality from K. lactis in the absence a functionally active F1, we screened for EB-resistant (EBr) colonies from CK204 disrupted in the ATP3 gene encoding the γ protein. We have previously reported that CK204 is petite negative and unable to form colonies on EB, which specifically eliminates mtDNA (7) (see below). CK204 was thus plated on EB medium and incubated for 7 days at 28°C. Fast-growing EBr papillae were scored. The EBr clones all lacked mtDNA, as confirmed by Southern blot analysis of total DNA prepared from the mutant cells by using total K. lactis mtDNA as a probe (data not illustrated).
Genetic analysis was performed to identify putative suppressor mutations that permitted the survival of the γΔ and ρo cells. The four EB-resistant isolates CK204/EB2, CK204/EB3, CK204/EB4, and CK204/EB8 were crossed to the wild-type strain CK56-16C. The resulting diploids, CW13, CW14, CW15, and CW16, respectively, were sporulated, and 12 asci were dissected from each strain. Phenotypic examination of the segregants indicated that respiratory deficiency, as judged from failure to grow on nonfermentable YPGE medium, segregated 2+:2− in all tetrads. The respiration-deficient phenotype cosegregated with Ura+, marking the disruption of ATP3 in the parental strain CK204. When segregants were spotted onto EB plates that were incubated for 5 days at 28°C, two interesting observations were made. First, approximately 50% of respiration-deficient segregants were resistant to EB, suggesting the presence of a mutation that is capable of suppressing ρo lethality in the absence of the γ subunit, as exemplified by CK307 and CK308 (Fig. 1A). Second, about 50% of the respiration-competent segregants formed papillae on EB plates, as exemplified by CW14-4D and CW15-6C. However, these strains did not have the same level of EB resistance as CK307 and CK308. When these apparently weak mutations were combined with disruption of ATP3, strains (e.g., CK307 and CK308; Fig. 1A) that had high levels of resistance to EB were formed. As described below, we subsequently found that the suppressor mutations in CK204/EB4 occur in ATP1, encoding the α subunit of F1-ATPase. Consequently, the mutation was designated atp1-7. As the mutations in strains CK204/EB2, CK204/EB3, and CK204/EB8 have the same change in the ATP2 gene, encoding the β subunit of F1-ATPase, this allele was referred to as atp2-12.
Both the atp1-7 and atp2-12 mutants were respiration competent in the presence of γ, and no significant difference in the growth rate on glycerol plates compared with that of wild-type strain PM6-7A was noticeable (Fig. 2). These cells did not form spontaneous respiration-deficient mutants on YPD medium (data not shown). As shown in Fig. 1A, EBr colonies arose at a high frequency for strains CW15-6C (atp1-7) and CW14-4D (atp2-12). Quantitative estimation revealed that CW15-6C and CW14-4D formed EBr colonies at frequencies of 5.6 ×10−2 and 2.6 × 10−2, respectively, which are significantly higher than the frequency of 1.1 × 10−4 found in K. lactis wild-type strain PM6-7A. EBr papillae were also observed in the atp1-7 atp2-12 double mutant CK314-2B at a frequency of 8.0 × 10−2, which was higher than the frequencies in the atp1-7 and atp2-12 single mutants (Fig. 1A). These observations support the idea that the atp1-7 and atp2-12 mutants have poor survival after the elimination of mtDNA by EB. The disruption of ATP3 from these cells improved cell viability (see below).
As the atp3::URA3 allele in CK204 used for the isolation of the atp1-7 and atp2-12 alleles was constructed by a simple insertion of the URA3 marker at codon 161 (7), it might be possible that a truncated version of the γ protein with the N-terminal sequence is produced and incorporated into the F1 complex. To address this issue, we constructed the atp3Δ::URA3 allele in which the promoter sequence and majority of the coding region of ATP3 were replaced by URA3 (Fig. 1B). This cassette was introduced into CW15-6C (atp1-7) and CW14-4D (atp2-12) to replace the wild-type ATP3. The correct replacement was confirmed by Southern blot analysis of genomic DNA. As shown in Fig. 1C, the strains with the ATP3 deletion, CK308 and CK307, had a band of 1.5 kb that hybridized to the 1.5-kb EcoRI-EcoRI ATP3 probe instead of a band of 1.27 kb in a wild-type control strain, which is expected for the correct replacement of the chromosomal ATP3 by atp3Δ::URA3. Because both CK308 and CK307 grew on EB medium (Fig. 1A), the data further confirm that the activity of the atp1-7 or atp2-12 allele in suppressing ρo lethality is enhanced by the complete lack of the γ protein. When ATP3 was disrupted in CK314-2B (atp1-7 atp2-12), the resulting strain, CK326, also displayed an EBr (or petite-positive) phenotype (data not illustrated).
To identify the mutations that suppress ρo lethality in the γ-less strains, we first considered whether the ATP1 and ATP2 genes, encoding the α and β subunits of F1-ATPase, respectively, might have been mutated. Indeed, we found that the mutations occurred in the two genes. When plasmid pCXJ4-MGI2, carrying the ATP1 gene, was integrated into the LEU2 locus of CW15-6C (atp1-7), the formation of papillae on EB medium was suppressed in the resulting strain, CK295 (Fig. 1A), whereas transformation by pCXJ4-MGI1/7, bearing ATP2, did not suppress the EBr papilla phenotype (data not illustrated). The results suggest that the mutation in CW15-6C might occur in ATP1. This allele, initially derived from CK204/EB4, was thus tentatively designated atp1-7. Likewise, when pCXJ4-MGI1/7, bearing the ATP2 gene, was integrated into CW13-4C, CW14-4D, and CW16-7C, derived from CK204/EB2, CK204/EB3, and CK204/EB8, respectively, the resulting strains, CK292 (Fig. 1A), CK291, and CK293, respectively (data not illustrated), did not show the EBr papilla phenotype. Introduction of pCXJ4-MGI2 (ATP1) did not suppress the formation of EBr papillae in the three strains. As the isolates carry the same lesion in ATP2 (see below), the mutation was designated atp2-12.
Further experiments were performed to confirm that the γ-independent ρo lethality suppressor mutations occur in ATP1 and ATP2. First, we found that, in contrast to the atp1-7 and atp2-12 single mutants, which are respiration competent, the double mutant CK314-2B failed to grow on glycerol medium (Fig. 2). However, transformation of CK314-2B with the plasmids pUK-KlMGI2/HP and pCXJ22-KlATP2, carrying the KlATP1 and KlATP2 genes, respectively, could partially restore growth on YPG (Fig. 2). The complementation of respiratory deficiency by the two genes suggests that the two mutations in CK314-2B may have occurred in ATP1 and ATP2.
Second, we crossed CW15-2A (atp1-7) to CK141-4A (atp1-2) and CW14-4D (atp2-12) to CK98-8A (atp2-3). K. lactis cells carrying the atp1-2 or atp2-3 allele are viable with ρo and resistant to EB (7, 8). The resulting diploid strains, CK325 and CK312, respectively, were sporulated, and 12 tetrads were dissected from each strain. When the segregants were examined for their response to EB, it was found that in all tetrads, two spores were EBr, while the other two spores formed EBr papillae on EB plates (data not illustrated). These data indicate that the mutation in CW15-2A is allelic to ATP1, and the one in CW14-4D is allelic to ATP2.
The ATP1 gene from CK204/EB4 (atp1-7) was amplified by PCR, and the complete coding region was sequenced. Compared with the sequences in the wild-type allele in the parental strain PM6-7A used for the isolation of atp1-7, two mutations were found. The codons 435 and 446 were changed from GCC and ACC to ACC and ATC, respectively. These changes convert Ala435 into Thr and Phe446 into Ile. To delineate which mutation is responsible for ρo lethality suppression in the γΔ background, alleles with the single mutations A435T and F446I in the α subunit (αA435T and αF446I, respectively) were generated by site-directed in vitro mutagenesis. The mutant alleles were cloned into the K. lactis vector pCXJ4 and integrated into the LEU2 locus of CK333 (αΔ γΔ) by selecting for Leu+ transformants. Following the test on EB plates, it was found that only the αF446I allele and not the αA435T allele conferred the EBr phenotype. We thus concluded that in the initial atp1-7 mutant, the αF446I mutation is responsible for the suppression of ρo lethality in γΔ cells. The amino acid αPhe446 is highly conserved in the α subunit of F1-ATPase in organisms from bacteria to humans (Fig. 3A).
Likewise, we amplified the ATP2 gene from the EBr isolates CK204/EB2, CK204/EB3, and CK204/EB8. The entire coding region of the gene was sequenced and compared with that of the wild type. A single mutation that changed codon 419 from GGT to GAT, which converted Gly419 to Asp, was found in the three isolates. βGly419 is also highly conserved during evolution (Fig. 3A).
To assess whether the two other small subunits of F1, δ and ε, play a role in the phenotypic manifestation of the atp1-7 and atp2-12 alleles in suppressing ρo lethality, we disrupted the ATPδ (26) and ATPε (27) genes. Inactivation of either δ or ε in strain CK401/1 (atp1-7 γΔ) yielded CK408 (atp1-7 γΔ δΔ) and CK409 (atp1-7 γΔ εΔ), respectively, which were both respiration deficient (Fig. 4). More importantly, CK408 and CK409 remained EBr with a growth rate comparable to that of CK401/1 (atp1-7 γΔ) on the EB plates. Likewise, disruption of either δ or ε in CK400 (atp2-12 γΔ) yielded CK405 (atp2-12 γΔ δΔ) and CK406 (atp2-12 γΔ εΔ), respectively. Both CK405 and CK406 remained resistant to EB (Fig. 4).
Strains with triple disruptions were constructed in the atp1-7 and atp2-12 backgrounds. The ATP3, ATPδ, and ATPε genes were successively inactivated in CK392/7 and CK391/18 to generate CK412/1 (atp1-7 γΔ δΔ εΔ) and CK410/6 (atp2-12 γΔ δΔ εΔ), respectively. When examined on EB medium, we found that both CK412/1 and CK410/6 are resistant to EB (Fig. 4). The results indicate that γ, δ, and ε are all dispensable for the suppressor phenotype of the atp1-7 and atp2-12 alleles.
We speculated that the atp1-7 and atp2-12 mutations in the α and β subunits may increase the ATP hydrolysis activity of the F1 subcomplex lacking γ. The γ-independent ATPase activity may be too low with the wild-type α and β to suppress ρo lethality. To test this idea, we measured the ATPase activity of isolated mitochondria in the absence and presence of oligomycin. As shown in Table 2, the γ-less strain CK306 expressing the wild-type α and β subunits retained only 5.1% of the native F1-dependent ATPase activity. The activity was increased to 21.7% and 44.6% in strains CK308 (atp1-7 γΔ) and CK307 (atp2-12 γΔ), respectively. These activities are oligomycin insensitive, as expected, because the γ-less complex would not be physically coupled to Fo, where oligomycin binds.
We found that the atp1-7 atp2-12 double mutant CK314-2B had only 10.8% of the wild-type ATPase activity. It was also respiration deficient on YPG medium in the presence of γ (Fig. 2). The two mutations were apparently incompatible for both hydrolysis and synthesis of ATP, although the single mutants retained partial respiratory growth. The disruption of ATP3 in the double mutant (CK326) slightly increased ATPase activity. This rather subtle change is apparently sufficient for keeping ρo cells viable. Interestingly, the atp1-7 and atp2-12 single mutants had oligomycin-sensitive ATPase activity higher than that of the wild type in the presence of the γ subunit, despite the fact that they were unable to maintain cell viability under ρo conditions. It is apparent that the γ subunit stimulates ATP hydrolysis under ρ+ conditions but inhibits the suppressor activity of the atp1-7 and atp2-12 mutants under ρo conditions.
We attempted to purify the mutant γ-less subcomplexes so that their biochemical properties could be further characterized. A His6 tag was introduced at the C terminus of the wild-type ATP1 and the mutant atp1-12 alleles. These alleles were integrated into the leu2 locus of CK333 (αΔ γΔ). The untagged ATP1 was also chromosomally integrated as a control. As expected, all the strains remained respiration deficient because of the loss of γ (Fig. 5A). The reintroduction of the His6-tagged atp1-7, but not the His6-tagged wild-type ATP1, conferred a robust growth phenotype on EB medium. The strong suppressor phenotype was therefore caused by the atp1-7 allele and not the His6 tag.
Isolated mitochondria were sonicated to release the His6-tagged γ-less F1 subcomplexes, which were purified by nickel column chromatography. We found that Atp1-His6 and Atp1F446I-His6 were successfully pulled down together with the β subunit (Fig. 5B). This suggests that the α subunit is associated with β and likely to form a subcomplex in vivo to hydrolyze ATP in the absence of γ. We found that in the pull-down products, the stoichiometry between α and β varied substantially in different experiments. It is likely that the subcomplexes are unstable in vitro (see below) and that the unassembled α and β subunits may have precipitated out at different rates during the course of the pull-down procedure. We also introduced the His6-tagged atp2-12 allele in a βΔ γΔ double mutant and attempted to pull down the α subunit. Atp2G419D-His6 was poorly recovered by nickel column chromatography, suggesting that the C terminus of the β subunit is not as accessible as that of the α protein.
Surprisingly, we could not detect significant ATP-hydrolyzing activity with the pull-down products from either ATP1 or atp1-7 cells. This suggests that the γ-less subcomplex may be structurally and functionally labile in vitro. By applying the pull-down products to BN-PAGE, we found that the α and β subunits were present either in a monomeric form or in a wide range of molecular species of >720 kDa, which is indicative of aggregate formation (Fig. 5C). No band of ~300 kDa suggestive of an α3β3 complex was visible.
Western blot analysis of mitochondrial lysates showed that the α and β subunits were not degraded in the absence of the γ subunit (Fig. 6A). To exclude the possibility that the in vitro instability of the γ-less F1-ATPase described above was not an artifact of the harsh conditions applied during the pull-down procedure and was not caused by His6 tagging, we directly analyzed digitonin-solubilized mitochondria from untagged strains by BN-PAGE. As shown in Fig. 6B, the wild-type ATP synthase was resolved in both the monomeric and dimeric forms. In the γ-less strains, regardless of the presence or absence of a suppressor mutation, the untagged α and β subunits were detected in a wide zone of high molecular mass, indicative of protein aggregation. Thus, the γ-less F1 immediately released from mitochondria is as unstable as that in the pull-down samples. These observations confirm that the γ-less F1 complexes, which putatively hydrolyze ATP in mitochondria, are labile in vitro under the gel electrophoresis conditions.
We finally examined the assembly state of the γ-less F1 complexes by directly applying the pull-down products on a size exclusion column. The resulting fractions were analyzed by Western blot analysis (Fig. 7A and andB).B). Indeed, we found that the F1 subunits did not form the distinct peak of ~300 kDa expected for an α3/β3 complex. Instead, they were distributed in fractions with molecular masses corresponding to species ranging from monomers to large aggregates. No significant difference in the assembly state was found with or without the presence of the atp1-7 allele in the γ-less complexes. We also observed that the size variation of the γ-less F1 deduced from size exclusion chromatography was apparently greater than that from the BN-PAGE experiment (Fig. 6B). In the latter case, molecular species of <720 kDa were barely visible. These proteins may be sensitive to electrophoresis and form molecular species that do not enter the gel. Overall, the data showed that the αβ subcomplexes are very labile and can be readily dissociated and aggregated. This property may account for the loss of ATPase activity in vitro.
The F1-ATPase is an extremely robust molecular machine. ATP hydrolysis by the three catalytic β subunits drives the rotation of the central γ rotor at a speed of as high as 10,000 rounds per minute (28). This unusual capacity is achieved by a high degree of cooperativity between the three catalytic sites that undergo sequential conformational changes. It has long been believed that the central γ rotor plays a key role in mediating cooperative ATP hydrolysis (29). The γ subunit is situated within the α3β3 ring and is proposed to determine the conformational and catalytic states of β subunits during the catalytic cycle. ATP binding and hydrolysis by one catalytic β subunit drive the unidirectional rotation of the rotor γ subunit through β-γ interactions. The rotating γ in turn induces conformational changes in the next β subunit to enhance its catalytic activity. However, accumulating evidence has emerged to support the idea that the α3β3 subcomplex can intrinsically catalyze cooperative ATP hydrolysis independently of γ. Early studies have shown an assembled and catalytically active γ-less complex in the thermophilic bacterium PS3 F1-ATPase (30, 31). Genetic studies also supported the assembly of an active F1 lacking γ in S. cerevisiae mitochondria (32). The mitochondrial γ-less F1 has an undetectable ATPase activity (25), whereas the isolated PS3 α3β3 subcomplex showed 20 to 25% of the ATPase activity of the α3β3γ complex. Although the PS3 α3β3 subcomplex is easily dissociated when the enzyme is diluted, incubated at high and low temperatures, and subjected to native polyacrylamide gel electrophoresis, it appears to exhibit similar cooperative kinetics as α3β3γ. The data support the idea that kinetic cooperativity is an intrinsic property of the α3β3 core (31) and that γ is critical for structural stabilization rather than an essential component for cooperativity. In contrast, a subsequent study reported very limited cooperativity by the α3β3 subcomplex (33).
More recently, the Noji group of researchers used high-speed atomic force microscopy and observed ATP-induced dynamic conformational changes in the β subunits of an α3β3 complex from the thermophilic bacterium PS3 (34, 35). At a given time, only one β subunit assumes the open state for ATP binding. When changed to the closed state, the neighboring β subunit is converted to the open state. A unidirectional propagation of the open state was elegantly documented. These observations provide direct evidence for a cooperative catalytic mechanism within the α3β3 core. The γ central stalk is, no doubt, very important for the assembly of ATP synthase and for coupling proton-driven rotation in Fo to conformational changes in β during ATP synthesis, but it is not absolutely required for catalytic cooperativity during ATP hydrolysis by α3β3. This is consistent with an early observation that the association of α with a nucleotide-bound β subunit induces asymmetry in F1 lacking the central γ rotor (36).
In the present report, we show that the γ-less F1-ATPase from K. lactis has a barely detectable ATPase activity in isolated mitochondria, like the γ-less mutant of S. cerevisiae (25). This activity is enhanced by the αF446I and βG419D mutations, where the mutant γ-less F1 complexes retain 21.7 and 44.6% of the wild-type γ-containing F1 activity, respectively. The mutation-induced recovery of the ATPase activity can be phenotypically scored by the maintenance of cell viability in K. lactis under ρo conditions. The dramatically increased ATP hydrolysis activity supports the presence of a robust cooperativity between the catalytic sites independently of the central γ rotor, as a unisite catalytic mode would be expected to have an activity 105- to 106-fold lower than that of a cooperative mechanism (37, 38). We have not been able to determine the catalytic kinetics of the γ-less enzymes with or without the presence of suppressor mutations. These subcomplexes seem to be extremely labile. They form complexes of various sizes and are readily dissociated and aggregated in vitro, as revealed by BN-PAGE and size exclusion chromatography. The γ-less subcomplexes would be expected to be at least partially assembled in vivo to hydrolyze ATP, which allows the survival of ρo cells. It is possible that the protein-dense environment and/or the presence of F1-specific chaperones (e.g., Atp11, Atp12, and Fmc1) (39–44) in the mitochondrial matrix may stabilize the γ-less subcomplexes in vivo.
Several explanations for the stimulation of ATP hydrolysis by the αF446I and βG419D mutations in the γ-less mitochondrial F1-ATPases may be offered. First, as αPhe446 and βGly419 are located on the α-β interface, it is possible that the mutations structurally stabilize the γ-less complexes. However, on the basis of currently available data, this is unlikely to be the case. We found that Atp1-His6 can pull down the β subunit as efficiently as Atp1F446I-His6. In other words, wild-type α and β subunits are likely to be as efficient as their mutant counterparts at interaction with their assembly partners in vivo, but it might be expected that in both cases the resulting complexes would be more unstable than native F1. Furthermore, size exclusion chromatography confirmed that the γ-less F1 forms molecular species of various sizes, regardless of the presence or absence of the αF446I mutation.
Alternatively, it can be speculated that the αF446I and βG419D mutations in the K. lactis F1-ATPase stimulate ATP hydrolysis activity by altering the α-β interactions, with this alteration increasing the cooperativity within the α3β3 subcomplex. The mechanisms of γ-independent ATP hydrolysis and catalytic cooperativity are currently unclear (45). Gly419 in the yeast F1 is located on the N-terminal edge of the DELSEED loop in the β subunit and Phe446 is located on the N-terminal edge of an equivalent loop in the α subunit (46) (Fig. 3A). These loops connect two long α helices and have previously been proposed to mediate torque transmission through interactions with γ. They are rich in negatively charged residues that may interact with the positive surface, known as the “ionic track,” on the γ subunit (47). However, several studies have shown that the DELSEED loop may not be critical for torque transmission and rotational catalysis. Replacement of the acidic residues in the DELSEED loop by alanine has little effect on the rotary torque (48). The Bacillus PS3 F1-ATPase with deletions in the DELSEED loop retains ATP hydrolysis activity through a rotational mechanism (49). The bacterial enzyme with the disruption of all predicted interactions between the DELSEED loop and γ by mutagenesis can still catalyze unidirectional rotation (50). Thus, specific interactions between the DELSEED loop and γ are not essential for cooperativity and the kinetic power of F1-ATPase.
We speculate that in K. lactis, the αF446I and βG419D mutations may alter the α-β interactions, thereby improving catalytic cooperativity in the γ-less conformation. Although a structure of the yeast α3β3 ring is currently unavailable, it would seem reasonable to predict that the N-terminal edge of the DELSEED loops in the α and β subunits, where Phe446 and Gly419, respectively, are located, may alter interactions with the adjacent β and α subunits. The Yoshida group has previously solved the α3β3 subcomplex of the thermophilic Bacillus PS3 F1-ATPase (51). In the absence of γ, the nucleotide-free α3β3 subcomplex exists in a perfect 3-fold symmetry. The αPhe398 residue, equivalent to αPhe446 in K. lactis at the N-terminal edge of the DELSEED loop, is on the interface with the N-terminal α helix of the DELSEED loop in the adjacent β subunit, with an estimated distance of less than 6 Å (Fig. 3B and andC).C). It is possible that in the K. lactis γ-less enzyme, the replacement of αPhe446 with isoleucine alters the interactions with the N-terminal α helix on the β subunit, which changes catalytic kinetics or cooperativity. Likewise, replacement of βGly419 (or βGly388 in Bacillus F1) with aspartic acid at the N-terminal edge of the DELSSED loop may alter the length and properties of the N-terminal α helix of the DELSEED loop in the β subunit. This may change the interactions with the C-terminal edge of the DELSEED loop on the adjacent α subunit. Because the DELSEED loops are expected to be structurally flexible, it is currently difficult to predict how the mutations exactly affect the α-β interactions during ATP hydrolysis in the γ-less subcomplexes.
The suppression of ρo lethality by the αF446I and βG419D mutations is much more effective in the γ-less configurations. We showed that the presence of γ inhibits the in vivo suppressor activity of the mutant alleles (Fig. 1A). The occupancy of the central cavity by γ appears to mask the prosurvival effects of the suppressor mutations. Consistent with its low ATP hydrolysis activity, the wild-type K. lactis α3β3 subcomplex is unable to suppress ρo lethality. It is possible that the γ subunit acts as a mediator for stimulating α-β interactions during cooperative catalysis by the native F1-ATPase. The αF446I and βG419D mutations may bypass the requirement for γ by directly stimulating α-β interactions.
It remains mysterious with regard to the wild-type α3β3γ complex, which is active in ATP hydrolysis but is unable to suppress ρo lethality in vivo. As previously suggested, the wild-type enzyme may have a high Km for ATP, which would not allow hydrolysis of sufficient ATP in ρo cells where ATP is scarce (18). In this context, we cannot completely exclude the possibility that in the γ-less configuration the αF446I and βG419D mutations promote cell survival by lowering the Km for ATP, which increases ATP turnover and therefore the membrane potential in ρo cells by a mechanism independent of the ATP synthase complex. The γ-less K. lactis F1-ATPase is highly unstable in vitro. For future studies, detailed kinetic properties of the γ-less enzymes may be tested by introducing similar mutations in a more stable enzyme, such as the thermophilic Bacillus PS3 α3β3 subcomplex. Alternatively, the ATP-hydrolyzing activity of the α3β3γ and α3β3 complexes may be inhibited in vivo under ρo conditions and the inhibition may be relieved by the suppressor mutations.
Smith and Thorsness have previously reported that the βG227S mutation in S. cerevisiae F1-ATPase suppresses the slow-growth phenotype of cells in which the ATP3 gene encoding the γ subunit is disrupted (52). Because the atp3 mutant is unable to maintain mtDNA, cell survival would require a robust ATP hydrolysis by F1 to maintain the mitochondrial inner membrane potential through the electrogenic ATP−4 (cytosol)/ADP−3 (matrix) exchange. The γ-less F1 is likely ineffective in ATP hydrolysis, but it may be increased by the βG227S mutation. Indeed, it was found that ATP hydrolysis was increased from 5.6% of that for the native F1 in cells expressing the wild-type α and β to 19% of that in cells expressing the wild-type α and the βG227S allele. βGly227 is located close to the conserved arginine finger domain of the active site. The βG227S mutation may potentially increase ATP hydrolysis in the absence of the γ subunit. Interestingly, the arginine finger has previously been proposed to sense the presence (or absence) of the γ-phosphate of ATP and to contribute to the cooperativity of the stator ring in F1-ATPase (53, 54).
In summary, although the exact mechanism remains to be determined, our data provide in vivo evidence for robust ATP-hydrolyzing activity by γ-less mitochondrial F1-ATPases. This is possible only when specific mutations are introduced at the N-terminal edge of the DELSEED loop in either the α or the βsubunit. The data provide unambiguous evidence for the importance of these two sites in promoting ATP hydrolysis in a γ-independent manner. These findings are consistent with the notion that an effective α-β interaction is critical for ATP hydrolysis by the α3β3 ring. This mode of ATP hydrolysis is reminiscent of that of some AAA-ATPases whose catalytic activity is independent of a central rotor (45). Thus, the genetic system that we developed in K. lactis could be further explored in the future for effectively dissecting the mechanism of α-β interactions without the interference from the γ subunit.
We thank Tom Duncan's laboratory for help with the ATPase assay, Richard Cross and Ed Berry for insightful discussion, and Jean Velours for providing antibodies against the F1-ATPase subunits.
This work was supported by National Institutes of Health grant R01AG023731.
Published ahead of print 6 September 2013