Manipulations in the Peripheral Stalk of the Saccharomyces cerevisiae F₁F₀-ATP Synthase

Abstract

The Saccharomyces cerevisiae F₁F₀-ATP synthase peripheral stalk is composed of the OSCP, h, d, and b subunits. The b subunit has two membrane-spanning domains and a large hydrophilic domain that extends along one side of the enzyme to the top of F₁. In contrast, the Escherichia coli peripheral stalk has two identical b subunits, and subunits with substantially altered lengths can be incorporated into a functional F₁F₀-ATP synthase. The differences in subunit structure between the eukaryotic and prokaryotic peripheral stalks raised a question about whether the two stalks have similar physical and functional properties. In the present work, the length of the S. cerevisiae b subunit has been manipulated to determine whether the F₁F₀-ATP synthase exhibited the same tolerances as in the bacterial enzyme. Plasmid shuffling was used for ectopic expression of altered b subunits in a strain carrying a chromosomal disruption of the ATP4 gene. Wild type growth phenotypes were observed for insertions of up to 11 and a deletion of four amino acids on a nonfermentable carbon source. In mitochondria-enriched fractions, abundant ATP hydrolysis activity was seen for the insertion mutants. ATPase activity was largely oligomycin-insensitive in these mitochondrial fractions. In addition, very poor complementation was seen in a mutant with an insertion of 14 amino acids. Lengthier deletions yielded a defective enzyme. The results suggest that although the eukaryotic peripheral stalk is near its minimum length, the b subunit can be extended a considerable distance.

Introduction

F₁F₀-ATP synthase is the principal enzyme for ATP generation in most organisms. The overall structure and mechanism of the enzyme is highly conserved. The catalytic sites are located in the F₁ sector, and ATP synthesis is driven by proton translocation through the F₀ sector. Proton translocation results in movement of the rotor stalk that extends through the center of F₁, and it is this rotation that results in the conformational changes in F₁ that account for the binding change mechanism (1–3). The function of the peripheral stalk is to hold F₁ against the movement of the rotor stalk (4).

Although the prokaryotic and eukaryotic peripheral stalks apparently serve the same core function, the subunit structures are very different. In bacteria, the peripheral stalk of the F₁F₀-ATP synthase is primarily composed of two identical b subunits that are thought to be in an extended α helix. The b subunits each transit the membrane once as part of F₀, extend alongside one noncatalytic α/β subunit interface to the top of F₁ where the carboxyl terminus of at least one b subunit interacts with the δ subunit (5, 6). The hydrophilic portion of two subunits, b subunits are thought to be in a parallel coiled-coil conformation from the surface of the membrane to the top of F₁ (7). Support for this can be found in the right-handed coiled-coil structure of the peripheral stalks of the related enzyme, the A₁A₀-ATP synthase of Thermus thermophilus (8). In contrast, the eukaryotic peripheral stalk is composed of four different subunits named OSCP, h (or F₆ in Bos taurus), d, and b in Saccharomyces cerevisiae. The S. cerevisiae ATP4 gene encodes a b subunit that differs substantially in both structure and primary sequence from the bacterial b subunit. The b subunit traverses the membrane twice and extends in a largely α helical conformation up one side of F₁ (9, 10). In the B. taurus peripheral structure (11), the membrane-extrinsic portion of the b subunit participates in protein-protein interactions with the d and F₆ subunits. Finally, the carboxyl-terminal region of the b subunit associates with the OSCP located at the top of the F₁ sector (12). The major differences in subunit composition between the eukaryotic and prokaryotic peripheral stalks raised a question about whether the two stalks have fundamentally different physical and functional characteristics.

Previous work done in our laboratory showed that the E. coli enzyme displayed a high degree of tolerance for substantial insertions and deletions within the b subunit in the tether domain where the protein emerges from the surface of the membrane. At the extremes, F₁F₀-ATP synthase activity was detectable in Escherichia coli enzymes with b subunits containing deletions of 11 and insertions of up to 14 amino acids (13, 14). An active F₁F₀-ATP synthase was also observed containing two b subunits with a size difference of up to 14 amino acids (15). These results were interpreted as evidence that for a considerable degree of flexibility within the E. coli peripheral stalk, not only were substantial changes in length tolerated, but chimeric proteins were constructed using sequences from b subunits of distantly related organisms within a functional F₁F₀-ATP synthase (16, 17).

In the present work, the length of the S. cerevisiae peripheral stalk was manipulated to determine whether the F₁F₀-ATP synthase had the same tolerance to insertions and deletions as the prokaryotic enzyme. A synthetic ATP4 gene (ATP4_syn) was designed to facilitate site-directed mutagenesis and expressed in S. cerevisiae with an ATP4 gene disruption. Characterization of the mutants suggested that the S. cerevisiae peripheral stalk possesses a limited tolerance for changes in its length.

EXPERIMENTAL PROCEDURES

Materials and Media

Molecular biology enzymes and oligonucleotides were obtained from New England Biolabs and Invitrogen, respectively. Reagents, chemicals, and media were acquired from Sigma-Aldrich, Fisher Scientific, and MP Biomedicals.

YPD medium was used as a rich medium for growth of S. cerevisiae. The medium consisted of 1% Bacto-yeast extract, 2% Bacto-peptone, and 2% glucose. The carbon source of this rich medium can be substituted with either glycerol, in the case of YPG,² or galactose, in the case of YPGal. Complete synthetic drop-out medium, denoted CSM nutrient, was used to select for auxotrophic markers with S. cerevisiae. The defined medium omitted the appropriate nutrient for the auxotrophic marker. The medium contained 0.67% yeast nitrogen base, 2% glucose, and 0.077% nutrient drop-out mix (MP Biomedicals). Sporulation medium contained a limited amount of glucose and nutrients in order to induce sporulation of a diploid S. cerevisiae strain. Sporulation medium had 1% potassium acetate, 0.1% Bacto-yeast extract, 0.05% glucose. Solid S. cerevisiae medium was made by addition of 1.6% agar to any of the previously described media.

Construction of ATP4_syn Expression Vectors

The ATP4_syn shuttle vectors were constructed by a multiple-step protocol. The ATP4_syn gene was synthesized by Genscript (Piscataway, NJ) and was originally received as an insert in the plasmid pUC57 (Fig. 1). Flanking restriction sites EcoRI and NotI were used to facilitate moving the A4TP4_syn gene into shuttle vector pRS313. PCR primer pairs, AW114/AW115 and AW116/AW117, were used to amplify the 150 bp upstream and downstream regions of the ATP4 gene from S. cerevisiae genomic DNA (Table 1). Restriction sites included in both primer pairs facilitated movement of the DNA into positions flanking the ATP4_syn gene, resulting in expression shuttle vectors pAW33 (Ap^r, ATP4_syn, and HIS3) and pAW36 (Ap^r, ATP4_syn, and URA3). The plasmids were confirmed by sequencing.

FIGURE 1.

The ATP4_syn gene. A, restriction map of the ATP4_syn gene. The distance of 100 bp is marked by a bar. B, complete sequence of ATP4_syn gene. Nucleotides changed from the wild type ATP4 to insert enzyme restriction sites through silent mutations and to alter codon preference are highlighted. Restriction sites are underlined, and the name for the enzyme for each restriction site is indicated below each sequence.

TABLE 1.

Primers used in S. cerevisiae mutagenesis experiments

Primer designation	Sequencea
AW114	CCCCCCCGTCGACGTGTTGTGACGCAACTGC
AW115	CCCCCCCCCCATGGAAGGACAACGAACACCTTGGC
AW116	CCCGCGGCCGCTCACAACAGTAACTGCG
AW117	GGGGGAGCTCGCCAATGGTTCTATCCAAAAGGG
AW83	CAAGAAGAGATATATAACCTGAGCATCC
AW84	CAATCACGACGCTTTTTCTCTTCAC

^a The restriction sites for SalI, NcoI, NotI, and SacI are underlined, boldface, underlined and italicized, and boldface and italicized, respectively.

Plasmid Complementation of a ΔATP4 Strain

Primers AW83 and AW84 were used to amplify the KanMx6 cassette and flanking DNA from a haploid ATP4 disruption strain obtained from ATCC (catalogue no. 4012750) (Table 1). The linear PCR product was transformed directly into BY4743, a diploid strain homozygous for the wild type ATP4 gene. S. cerevisiae transformations were performed essentially as described by Geitz et al. (18). Incorporation of the KanMx6 cassette into one ATP4 allele of the cells was detected by screening for growth on solid YPD media supplemented with 250 μg/ml (active) G418 (YPD+G418) and YPG medium. Colony PCR was performed was used as an initial screening tool to determine genotypes of candidate strains (19). Isolated genomic DNA was used for sequencing to confirm genotypes (20).

Plasmid pAW36 (ATP4_syn, Ap^r, and URA3) was transformed into strain AW2 (MATa/MATα his⁻ ura⁻ atp4::KanMx6/+) (Table 2). Sporulation was induced by a shift to solid sporulation medium. The resulting tetrads were dissected using a micromanipulator and haploid colonies were screened for growth on three types of media: (i) solid CSM without uracil (CSM-URA) medium, (ii) solid YPD+G418 medium, and (iii) solid YPG to test for functional ATP synthase activity. Colonies were genotyped for a KanMx6 cassette disruption of the chromosomal ATP4 gene and the presence of the plasmid-borne ATP4_syn gene. The strain was tested for mating type using Cold Springs Harbor Method, found to be Mata and was named AW3 (Mata his⁻ ura⁻ atp4::KanMx6, pAW36 (ATP4_syn, Ap^r, and URA3)).

TABLE 2.

Growth phenotype and genotype of ATP4_syn strains generated

Designation	ATP4 gene	Genotype	Growth characteristicsa
			YPGD	YPG
AW3	ATP4syn	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pAW36 (URA3+, ATP4)	+++	+++
AW4	ATP4syn	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pAW33 (HIS3+, ATP4)	+++	+++
AW8	ATP4synΔ4	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pAW41 (HIS3+, atp4Δ4)	+++	+++
AW6	ATP4synΔ7	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pAW35 (HIS3+, atp4Δ7)	±	−
AW9	ATP4synΔ11	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pCJB4 (HIS3+, atp4Δ11)	±	−
AW10	ATP4synΔ14	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pCJB2 (HIS3+, atp4Δ14)	±	−
AW11	ATP4syn+4	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pAW42 (HIS3+, atp4+4)	+++	+++
AW5	ATP4syn+7	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pAW34 (HIS3+, atp4+7)	+++	+++
AW12	ATP4syn+11	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pAW44 (HIS3+, atp4+11)	+++	+++
AW14	ATP4syn+14	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pAW45 (HIS3+, atp4+14)	++	++
AW7	ΔATP4	MATa his3Δ1 leu2Δ0 ura3Δ0, atp4::KanMx6, pRS313 (HIS3+)	±	−

^a +++, large, healthy colonies; ++, medium size colonies; ±, very small colonies; −, no growth.

Construction of Mutant Plasmids

Mutagenic oligonucleotides were designed to either insert or delete 4, 7, 11, or 14 codons (Fig. 2). Each mutant ATP4_syn gene was constructed in essentially the same manner by ligation-mediated site-directed mutagenesis (21). Mutant ATP4_syn expression plasmids were transformed into the S. cerevisiae strain AW3 (his⁻ ura⁻ atp4::KanMx6, pAW36 (Ap^r, URA3, and ATP4_syn)) to yield cells that contained both the pAW36 plasmid (Ap^r, URA3, and ATP4_syn) and a mutant plasmid (Ap^r, HIS3, and mutant ATP4_syn). The pAW36 plasmid was selected against by replica plating the S. cerevisiae cells on solid YPD+ 0.1% 5-fluoroorotic acid medium. After selection on solid YPD+5-fluoroorotic acid media, surviving cells are patched on to solid CSM-His (CSM without histidine) medium. Genomic DNA from selected patches grown on solid CSM-His medium were isolated and sequenced for the presence of the mutant ATP4_syn gene.

FIGURE 2.

Construction of insertion and deletion mutations. A (top), wild type synthetic sequence and deduced amino acid sequence corresponding to amino acids 73–97. Deduced nucleotide and amino acid sequences for each of the deletion mutations are shown. B (top), wild type synthetic sequence and deduced amino acid sequence corresponding to amino acids 82–97. Mutations are denoted on the left. Restriction sites used for mutagenesis and screening purposes are underlined and labeled.

Preparation of Mitochondria

For mitochondrial preparation, the strains were grown in YPGal as a carbon source to allow for growth of the ρ⁻ strains. Biological F₁F₀-ATP synthase activity was determined by plating serial dilutions on plates containing rich media with 2% glycerol and 0.1% glucose (YPGD) and incubating at 30 °C for 3 days.

Preparation of mitochondria from each mutant strain was done by scaling down the method of Meisinger et al. (22). A 5-ml overnight culture grown at 30 °C was used to innoculate 50 ml of YPGal and allowed to grow to stationary phase. 500 ml of YPGal was inoculated with between 5–20 ml of the stationary culture and grown for 20 h.

To make spheroplasts, cells were harvested, washed once with sterile double-distilled H₂O, and weighed. The cells were suspended at 2 ml buffer/g wet cells in prewarmed DTT buffer (100 mm Tris-H₂SO₄, pH 9.4, 10 mm DTT). The suspension was incubated with gentle agitation (80 rpm) for 20 min at 30 °C. The cells were recovered by centrifuged and washed with zymolyase buffer (ZB) (1.2 m sorbitol, 20 mm potassium phosphate, pH 7.4) at 7 ml/g wet cells. Cells were suspended in ZB containing zymolyase 20-T (Fisher Scientific) at 7 ml/g wet cells and incubated at 30 °C with slow shaking for 30–45 min. After zymolyase treatment, the cells were harvested and washed in ZB at 7 ml/g wet cells.

To lyse the cells, the pellet was carefully suspended at 6.5 ml/g wet cells precooled homogenization buffer (0.6 m sorbitol, 10 mm Tris-HCl, pH 7.4, 1 mm EDTA, and 1 mm PMSF). All steps were performed at 4 °C. The spheroplasts were lysed with 15 strokes of a glass Teflon homogenizer. Afterward the homogenate was diluted 2-fold with homogenization buffer and centrifuged at 1500 × g for 5 min to pellet cell debris and nuclei. The supernatant was transferred to a clean centrifuge tube and centrifuged at 4000 × g for 5 min. To recover the mitochondrial fraction, the supernatant was transferred to a second clean centrifuge tube and centrifuged for 15 min at 12,000 × g. The pellet was gently suspended in 250 μl of 4 °C SEM buffer (250 mm sucrose, 1 mm EDTA, 10 mm MOPS-KOH, pH 7.4). The mitochondria were stored at 4 °C until ATPase assays or a Western blot could be performed. Remaining mitochondria were flash frozen and stored at −80 °C. Protein concentration was determined using the bicinchoninic acid method (23).

Assays of F₁F₀-ATP Synthase

Assays of ATP hydrolysis were performed essentially as described by Tzagoloff and Ackerman (24). Mitochondrial protein samples (30 μg protein/well) were loaded onto a 12% Tris-glycine SDS Bio-Rad TGX Gel. Following electrophoresis, the mitochondrial proteins were transferred to a PVDF membrane (Bio-Rad) by electroblotting (110 V for 20 min at 4 °C). Detection was performed using the Rodeo ECL Western kit (Affymetrix), including the washing of the membrane with Rodeo Blocker. The rabbit antiserum against the S. cerevisiae b subunit was a generous gift from Dr. Rosemary Stuart (Marquette University). The rabbit antiserum against the b subunit was used at a concentration of 1:1000. The secondary antibody, anti-rabbit IgG (Amersham Biosciences) was used at a 1:30,000 concentration. The antibody was detected using chemiluminescence (Affymetrix). The signals were visualized using high performance chemiluminescence film (GE Healthcare) and a Kodak X-Omat. Signal strength was assessed using Un-Scan-It gel digitizing software (Silk Scientific, Inc.).

RESULTS

Plasmid Complementation with ATP4_syn

To construct a strain lacking a functional ATP4 gene, it was disrupted by a linear piece of DNA containing the KanMx6 cassette by homologous recombination. However, the deletion of the ATP4 gene from the genome was known to cause S. cerevisiae cells to go either ρ⁻ or ρ⁰ (25). To overcome this problem, a scheme was devised to maintain a copy of the native ATP4 gene or the ATP4_syn gene in the cells during all steps involved in chromosomal ATP4 gene deletion strain construction.

Wild type diploid strain, BY4743 (ATCC), was transformed with a linear PCR product containing the KanMx6 cassette flanked by an ATP4 sequence. The ATP4 sequence provided a means for homologous recombination with a single ATP4 allele. Transformants were selected for growth on YPD+G418 medium, genomic DNA was prepared and sequenced. The resulting heterozygous strain was named AW2 (MATa/MATα his⁻ ura⁻ atp4::KanMx6/+). Plasmid pAW36 (ATP4_syn, Ap^r, and URA3) was transformed into strain AW2. Sporulation was induced resulting in tetrads. Haploid colonies were genotyped for the KanMx6 cassette disruption of the chromosomal ATP4 gene and the presence of the plasmid-borne ATP4_syn gene. The strain was subsequently tested for mating type, found to be Mata, and named AW3 (Mata his⁻ ura⁻ atp4::KanMx6, pAW36 (ATP4_syn, Ap^r, URA3)). Plasmid complementation of the chromosomal disruption was tested by patching AW3 (ATP4_syn) on YPG medium. The strain grew robustly at a level comparable with the wild type parental strain BY4743. This showed that the synthetic ATP4_syn gene allowed phenotypically normal F₁F₀-ATP synthase activity in vivo. A second strain AW4 (Mata his⁻ ura⁻ atp4::KanMx6, pAW36 (ATP4_syn, Ap^r, and HIS3)) was constructed by plasmid shuffling to serve as a positive control for subsequent experiments.

Mutant Strain Construction

Insertion and deletion mutagenesis of the E. coli peripheral stalk suggested a considerable degree of plasticity, but it was unknown whether this property was shared with the eukaryotic enzyme (13, 14). To test the eukaryotic stalk, a comparable set of mutations were designed to lengthen and shorten the b subunit of S. cerevisiae. The ATP4_syn genes were constructed with 4–14 codons either deleted or inserted. In the case of the insertions, the coding sequence was duplicated with alternative codons for the amino acids wherever possible. The number of codons inserted or deleted was based on the periodicity of an α helix. The site for the manipulations was selected by inspection of the high-resolution structure of the B. taurus peripheral stalk (11) and the cryoelectron microscopy image of the S. cerevisiae F₁F₀ complex (26). In both structures, the peripheral stalk narrowed to the width of a single α helix where the b subunit is likely to emerge from the membrane. This narrowing of the peripheral stalk roughly coincides with the region in the E. coli enzyme where much of the insertion and deletion work was done. The S. cerevisiae and B. taurus b subunits share 44% sequence similarity. The target within the peripheral stalk was chosen by aligning the deduced amino acid sequence of the B. taurus b subunit with that of the S. cerevisiae b subunit (Fig. 3). This alignment revealed that the narrow region in the B. taurus peripheral stalk structure was homologous to amino acids b_K78–A94 in S. cerevisiae. However, two short segments of highly of conserved amino acids appear at b_D79–A83 and b_V100–D108, so mutations were designed to minimize modification of these sites. Therefore, the segment of the ATP4_syn gene encoding amino acids b_A81–A94 was selected as the target for mutagenesis.

FIGURE 3.

Amino acid sequence alignment of B. taurus and S. cerevisiae b subunits. The protein sequences were obtained from the Pubmed protein data base and were aligned using protein BLAST. The method used was the compositional adjustment method. The two sequences had 24% sequence identity and 44% sequence similarity.

Growth Properties of Mutant Strains

Recombinant ATP4_syn genes were tested for the ability to complement the disruption of the endogenous ATP4 gene in vivo by growth on YPG and YPGD. Use of glycerol as a carbon source requires a functional, intact F₁F₀-ATP synthase, and inclusion of a limited amount of dextrose in YPGD media allowed for slow growth of petite mutants.

The insertion and deletion strains showed dramatically different growth phenotypes. Deletions of seven or more amino acids showed no growth on glycerol and little growth on YPGD medium, suggesting the petite phenotype (Fig. 4A). In contrast, the ATP4_synΔ4 strain grew well on YPG and YPGD and was comparable with that of the control AW4 (ATP4_syn) strain. Insertions of up to eleven amino acids also showed robust growth on both types of media (Fig. 4B). Significantly, less growth was observed for the ATP4_syn+14 strain, indicating the enzyme was just barely capable of sustaining biologically significant function.

FIGURE 4.

Growth of ATP4_syn mutant strains on YPGD. Cultures were serially diluted from 10⁶ cells/ml to 100 cells/ml or 10⁷ cells/ml to 10³ cells/ml. Plates were placed in an incubator at 30 °C and grown for 3 days. A, deletions in b subunit. B, insertions in b subunit.

Expression of Recombinant b Subunits

It was likely that levels of expression and/or import of the recombinant b subunits might be affected due to the introduction of mutations. Western blots were performed on mitochondria isolated from the mutant strains using anti-b subunit antiserum to determine steady-state levels of the recombinant b subunits (Fig. 5). The positive control from strain AW4 (ATP4_syn) showed a strong signal indicating that the synthetic gene product was expressed and imported into the mitochondria (Fig. 5, A and B, lane 2). Levels of the ATP4_synΔ4 subunit were comparable with ATP4_syn. The ATP4_synΔ7 and ATP4_synΔ11 subunits were detected at much lower levels (Fig. 5B). A band from ATP4_synΔ14 subunit was only observable by prolonged exposure of the film.

FIGURE 5.

Immunoblot analysis of ATP4_syn gene mutant mitochondria. Mitochondrial proteins were separated on a SDS gel, and proteins were transferred to a PVDF membrane. The presence of a b subunit was detected using anti-serum against the b subunit of S. cerevisiae. To ensure that the anti-b antiserum was specific a negative control from a strain carrying an empty vector was included, and no band was detected on the Western blot. A, lane 1, null control; lane x, empty; lane 2, ATP4_syn; lane 3, ATP4_synΔ4; lane 4, ATP4_synΔ7; lane 5, ATP4_synΔ11; lane 6, ATP4_synΔ14. B, lane 1, null control; lane x, empty; lane 2, ATP4_syn; lane 3, ATP4_syn+4; lane 4, ATP4_syn+7; lane 5, ATP4_syn+11; lane 6, ATP4_syn+14.

In contrast, the ATP4_syn+4 subunit was present at levels comparable with the positive control. As the b subunit was further lengthened, expression levels decreased. Although both of the strains expressing ATP4_syn+7 and ATP4_syn+11 subunits had a normal biological phenotype, the level of expression was only 60% of that of the positive control. Extension to 14 amino acids showed a further decrease of expression levels to about 50% of the positive control. This 10% decrease in expression resulted in a disproportionate change in growth phenotype. Although successfully imported into the mitochondria, it seems that the ATP4_syn+14 subunit may not be efficiently incorporated into an F₁F₀ complex. Therefore, expression level of the b subunit did not necessarily directly correlate to growth phenotypes of the mutant strains.

F₁F₀-ATPase Activity

In mitochondria-enriched samples, the greatest single source of ATP hydrolysis activity is attributed the F₁F₀-ATP synthase. Traditionally, the amount of P_i release that is specifically assigned to the enzyme is the amount of activity lost by addition of the F₁F₀-ATP synthase inhibitor, oligomycin. Oligomycin inhibits the enzyme by interacting with the F₀ proton channel at a site thought to be located at the interface between the a subunit and the ring of c subunits (27).

In the absence of oligomycin, strain AW4 (ATP4_syn) mitochondrial-fractions had significantly greater ATP hydrolysis activity than the negative control strain (Table 3). Mitochondria prepared from the ATP4_syn disruption strains displayed ATPase activity comparable with the negative control. Although the ATP4_synΔ4 strain grew well on YPG media, mitochondrial fractions prepared from the strain exhibited little or no F₁F₀-ATPase activity. The evidence suggested that the ATP4_synΔ4 enzyme was clearly intact and active in vivo, but the complex was apparently unstable during preparation of the mitochondrial sample. In the presence of oligomycin, the negative control had negligible oligomycin sensitivity (2%). Most of the deletion strains showed no appreciable oligomycin sensitivity (Table 3).

TABLE 3.

S. cerevisiae F₁F₀-ATP synthase activity for deletion mutants

b Subunit	ATPase activity (μmol Pi min−1 mg protein−1)		% Oligomycin-sensitive
	− Oligomycin	+ Oligomycin
ATP4syn	5.0 ± 1.1	1.7 ± 0.4	66%
ATP4syn Δb	2.3 ± 0.9	2.1 ± 0.7	9%
ATP4synΔ4	3.9 ± 0.6	3.5 ± 0.9	10%
ATP4synΔ7	3.3 ± 0.8	2.1 ± 0.4	35%
ATP4synΔ11	1.5 ± 0.5	1.2 ± 0.4	18%
ATP4synΔ14	2.3 ± 0.8	2.1 ± 0.8	8%

In stark contrast to the deletion strains, the insertion strain mitochondrial samples possessed abundant ATPase activity (Table 4). Strains that contained b subunits with insertions up to eleven amino acids possessed ATPase activity that was indistinguishable from the positive control. Even the ATP4_syn+14 sample displayed activity substantially greater than the negative control value. In the presence of oligomycin, the strain AW4 (ATP4_syn) had a 57% decrease in ATP hydrolysis activity, a value comparable with that reported in the literature (28, 29). Surprisingly, the ATP4_syn+4, ATP4_syn+11, and ATP4_syn+14 strains showed little oligomycin sensitivity, suggesting that oligomycin inhibition was lost due to the insertion mutations.

TABLE 4.

S. cerevisiae F₁F₀-ATP synthase activity for insertion mutants

b Subunit	ATPase activity (μmol Pi min−1 mg protein−1)		% Oligomycin-sensitive
	− Oligomycin	+ Oligomycin
ATP4syn	5.6 ± 0.5	2.3 ± 0.9	59%
ATP4syn Δb	3.8 ± 0.6	3.7 ± 0.6	2%
ATP4syn+4	5.8 ± 0.6	5.1 ± 0.7	12%
ATP4syn+7	5.7 ± 0.5	3.0 ± 0.3	47%
ATP4syn+11	5.8 ± 0.6	5.1 ± 0.7	10%
ATP4syn+14	4.6 ± 0.6	4.3 ± 0.7	7%

Interestingly, both the ATP4_synΔ7 and ATP4_syn+7 mitochondria reproducibly showed considerable oligomycin sensitivity. As noted before the ATP4_synΔ7 has the petite phenotype, whereas the ATP4_syn+7 possesses a wild type growth phenotype. We suspect that these particular changes have a minimal effect on the orientation of the b subunit relative to the a subunit within the F₀ sector. This may account for the oligomycin sensitivity unique to those two subunits.

DISCUSSION

Here, we report the construction, expression, and characterization of a series of mutations in the b subunit of S. cerevisiae F₁F₀-ATP synthase. Plasmid shuffling was used to achieve the ectopic expression of altered b subunits in a S. cerevisiae strain carrying a chromosomal disruption of the ATP4 gene. Insertions of up to 11 amino acids yielded a wild type phenotype for growth on YPG medium, and mitochondrial-enriched fractions had readily detectable levels of the recombinant subunits with abundant ATP hydrolysis activity. This activity showed surprisingly little oligomycin sensitivity when considered in the context of growth phenotype. Additionally, expression of the ATP4_syn+14 b subunit provided limited complementation to a strain carrying a disrupted ATP4 gene. In contrast, most of the deletion mutations resulted in a loss of F₁F₀-ATP synthase function. Only the smallest deletion subunit, ATP_synΔ4, yielded a positive growth phenotype on YPG medium. The F₁F₀ complex with this shortened subunit had no apparent activity in vitro, suggesting enzyme instability during preparative procedures.

Previous work done in our laboratory showed that the E. coli F₁F₀-ATP synthase could accept large changes in length of the b subunits (13, 14). Those experiments were interpreted as evidence that considerable plasticity was an inherent property of the bacterial peripheral stalk. The experiments presented in this paper suggest that the eukaryotic peripheral stalk has similar plasticity, at least with respect to a short deletion and lengthy insertions. Changes in length ranging from the deletion of four to the insertion of 11 amino acids in the b subunit yielded an active enzyme. Assuming the predicted α helical conformation remains intact, this represents a change in length of ∼18 Å (Fig. 6A). Although this may not seem like a large change in length, the measured distance from the top of the membrane to the bottom of F₁ is ∼40 Å. Therefore, these mutations result in a change of 45% in the length of the peripheral stalk segment between the F₁ and F₀ sectors in an intact F₁F₀ complex. The additional three amino acids in the ATP4_syn+14 subunit resulted in substantially less enzyme.

FIGURE 6.

Structural models of recombinant peripheral stalks. A, the model is based on Protein Data Bank code 2CLY. The b, d, and F₆ subunits are shown with the b subunit central to each model. The structure on the far left shows the wild type, B. taurus complex (11). The vertical bars indicate the areas of the insertions and deletion, with the numbers to the left of the bars indicating the distance. Models labeled Δ4, +7, and +11 represent deletion and insertions in functional S. cerevisiae peripheral stalks. B, changes in the angles required in order to accommodate the insertion and deletion b subunits. The figure was rendered in MacPyMOL from Protein Data Bank codes 2CLY and 2WSS (11, 12). The composite image was made in Adobe Photoshop.

The evidence from experiments involving deletions is less clear. In E. coli, deletion of seven amino acids was essentially wild type, and longer deletions resulted in the assembly of fewer F₁F₀-ATP synthases (13). In S. cerevisiae, deletions of seven or more amino acids resulted in an apparent total loss of enzyme function. Although there was abundant ATP synthesis activity apparent in vivo, even the ATP4_synΔ4 b subunit F₁F₀ complex proved unstable during preparation of the mitochondria. For the longer deletions, it is unclear whether the enzyme failed to assemble with the deletion subunits or whether the F₁F₀ complex was unstable. In any case, once the b subunit is too short its interactions within the F₁F₀ complex are irrevocably compromised. The lack of F₁F₀-ATP synthase is associated with deformation or loss of the mitochondria leading to an overall deficiency in energy metabolism (30). The evidence would seem to suggest that the wild type S. cerevisiae b subunit is very near its minimum functional length. Considering the sequence similarity between S. cerevisiae and B. taurus b subunits, especially within the segment crystallized by Dickson et al. (11), it appears likely that eukaryotic F₁F₀-ATP synthases evolved to favor the shortest practicable peripheral stalks.

One of the most surprising findings was the lack of oligomycin sensitivity associated with the majority of the mutant ATP4_syn F₁F₀-ATP synthases. With the exception of the ATP4_syn+7 enzyme, all of the mutant strains that contained functional F₁F₀-ATP synthases were essentially insensitive to oligomycin. The result suggests that a subtle conformational change occurred within the F₁F₀ complex due to incorporation of the ATP4_syn+4, ATP4_syn+11, and ATP4_syn+14 subunits. The shift appeared to be propagated within the F₀ sector to the oligomycin binding site at the proton channel. Importantly, the shift did not cause any apparent loss in channel function, and only inhibitor binding seemed to be affected. There are two important implications of this observation. First, this suggests that oligomycin sensitivity is not a valid measure of F₁F₀-ATPase activity in enzymes with b subunit mutations and by extension alterations in other peripheral stalk subunits. Further support for the conclusion is found in a study involving the successful functional replacement of the h subunit by the B. taurus F₆ subunit. This replacement resulted in an oligomycin insensitive enzyme (31). Second, propagation of a conformation shift in the b subunit implies that changes in peripheral stalk length may not be fully accounted for by structural changes within the stalk itself.

Recent physical biochemistry experiments have indicated that within the context of the entire enzyme, the E. coli peripheral stalk is much less compliant than the central stalk (32, 33). Insertion of 11 amino acids within the b subunit yielded in a more flexible peripheral stalk, but not to a level that it reaches the pliability of the central stalk (35). A recent high resolution structure of the B. taurus F₁-peripheral stalk complex showed the possibility of two hinges B. taurus peripheral stalk (Fig. 6B) (12). One is located in the OSCP subunit on top of the F₁ complex. The second appears to be located in the region of b₁₄₆ in the B. taurus enzyme. Additionally, the cryoelectron microscopy structure shows the peripheral stalk narrowing as it emerges from the membrane to a diameter consistent with a single α helix provided by the b subunit (34). This is the region where the insertions and deletions within the b subunit were constructed. The narrowing of the peripheral stalk, along with the current experiments that show considerable plasticity is consistent with a third hinge in the area where the b subunit emerges from the membrane. This suggests that localized regions of higher plasticity are necessary for the peripheral stalk to be incorporated into the complex. Once the enzyme is assembled, the peripheral stalk then assumes a stiff conformation relative to the central stalk.

Two hinges within the b subunit, one located near the surface in the membrane and the other alongside F₁ at b₁₄₆, might act as pivot points for the b subunit during F₁F₀ assembly. This suggests a hinged stiff model that accommodates the present understanding of the peripheral stalk. Modest changes in the angle of these hinges would alter the pitch of the peripheral stalk, allowing incorporation of a shortened or a lengthened b subunit into the F₁F₀ complex (Fig. 6B). Alterations in pitch may contribute to the loss oligomycin sensitivity observed with peripheral stalk recombinant subunits.