The pathogenic m.8993 T > G mutation in mitochondrial ATP6 gene prevents proton release from the subunit c-ring rotor of ATP synthase

Abstract

The human ATP synthase is an assembly of 29 subunits of 18 different types, of which only two (a and 8) are encoded in the mitochondrial genome. Subunit a, together with an oligomeric ring of c-subunit (c-ring), forms the proton pathway responsible for the transport of protons through the mitochondrial inner membrane, coupled to rotation of the c-ring and ATP synthesis. Neuromuscular diseases have been associated to a number of mutations in the gene encoding subunit a, ATP6. The most common, m.8993 T > G, leads to replacement of a strictly conserved leucine residue with arginine (aL₁₅₆R). We previously showed that the equivalent mutation (aL₁₇₃R) dramatically compromises respiratory growth of Saccharomyces cerevisiae and causes a 90% drop in the rate of mitochondrial ATP synthesis. Here, we isolated revertants from the aL₁₇₃R strain that show improved respiratory growth. Four first-site reversions at codon 173 (aL₁₇₃M, aL₁₇₃S, aL₁₇₃K and aL₁₇₃W) and five second-site reversions at another codon (aR₁₆₉M, aR₁₆₉S, aA₁₇₀P, aA₁₇₀G and aI₂₁₆S) were identified. Based on the atomic structures of yeast ATP synthase and the biochemical properties of the revertant strains, we propose that the aL₁₇₃R mutation is responsible for unfavorable electrostatic interactions that prevent the release of protons from the c-ring into a channel from which protons move from the c-ring to the mitochondrial matrix. The results provide further evidence that yeast aL₁₇₃ (and thus human aL₁₅₆) optimizes the exit of protons from ATP synthase, but is not essential despite its strict evolutionary conservation.

Introduction

Mitochondrial disorders result from defects in oxidative phosphorylation (OXPHOS) (1,2). Oxidative phosphorylation uses the energy obtained from the catabolism of the food we ingest for the synthesis of adenosine triphosphate (ATP), which is the main energy source for the cellular processes that sustain life (3). OXPHOS typically involves five multiprotein complexes (I–V) all of which are embedded or anchored in the inner membrane of mitochondria. Complexes I–IV, which constitute the electron transport chain (ETC), obtain electrons from the oxidation of carbohydrates and fatty acids, which are ultimately used to reduce oxygen to water. In the process of oxidation/reduction reactions in the ETC, protons are pumped out of the matrix to the intermembrane space (IMS) establishing a proton motive force across the mitochondrial membrane. The proton motive force is used by the ATP synthase, Complex V, which couples the return of protons into the matrix with phosphorylation of ADP forming ATP (4). Any defect in the activity or biogenesis of the OXPHOS system will affect high-energy demanding tissues and organs and especially the brain, which consumes two-thirds of the dozens of kilograms of ATP that our body uses each day.

The human mitochondrial genome encodes 13 of the 90 subunits of the OXPHOS system, and a number of RNAs required for their synthesis inside the mitochondria. The remaining subunits are encoded in the nuclear genome and post-translationally imported into the mitochondrion. Despite the preponderance of nuclear genes required for formation and functioning of the OXPHOS system, mutations in the mitochondrial genome have been most frequently identified in mitochondrial disorders, presumably because of the exposure to this genome to damaging reactive oxygen species (ROS) (5) and the poor effectiveness of the mitochondrial DNA (mtDNA) repair systems (6). An added complexity for diseases due to mutations in the mtDNA is that the mitochondrial genome is present in up to 10 000 copies in a single cell (7,8). Thus, pathogenic mutations in the mtDNA usually co-exist in patient’s cells and tissues along with wild-type copies of mtDNA (heteroplasmy). This genetic heterogeneity can make it difficult to understand precisely how and to what extent these mutations compromise mitochondrial function and thus human health.

Yeast Saccharomyces cerevisiae is an unicellular fungus amenable to manipulation of the mitochondrial genome (9) and is unable to stably maintain mitochondrial heteroplasmy (10). Owing to its efficient capacity to produce ATP by the fermentation of sugars, this yeast can survive mutations that inactivate oxidative phosphorylation. We have exploited these attributes to investigate the consequences of one of the most common mtDNA mutation (m.8993 T > G) responsible for human mitochondrial diseases, including neuropathy ataxia retinitis pigmentosa (NARP) and maternally inherited leigh syndrome (MILS) (11–23). The m.8993 T > G mutation is in the gene encoding subunit a of the ATP synthase (ATP6) and changes a strictly conserved leucine residue to arginine, aL₁₅₆R. Subunit a and an oligomeric ring of identical c-subunits (c-ring) are responsible for the transport of protons from cytoplasm to the matrix coupled to the synthesis of ATP (24–27). Previously, we showed that an equivalent of the m.8993 T > G mutation in yeast (aL₁₇₃R) severely compromised (by 90%) mitochondrial ATP synthesis (28). Here, we report nine intragenic pseudo-reversions of this mutation that restore to various degrees ATP synthase function. Considering the recently described high-resolution structures of yeast ATP synthase (25,29), these findings suggest the molecular bases of the primary pathogenic effects of the m.8993 T > G mutation and provide novel information on the proton pathway within the membrane domain of ATP synthase.

Results

Selection of genetic suppressors of the aL₁₇₃R mutation

Genetic reversion of the mutant phenotype induced by aL₁₇₃R can result from a second mutation within the ATP6 gene (intragenic suppression) or in a different gene (extragenic suppression) located in mtDNA (intragenomic suppression) or nuclear (intergenomic suppression) DNA. Owing to the inability of S. cerevisiae to maintain mitochondrial heteroplasmy, genetic suppressors of aL₁₇₃R in mtDNA will be present in all mtDNA molecules (homoplasmy). With a haploid strain (in which nuclear genes are in single copy), nuclear suppressors can be dominant or recessive, whereas only dominant suppressors can be selected from a diploid strain where each nuclear gene is in two copies. We, therefore, used in this study diploid (MR12) and haploid (MR14) yeast strains for the isolation of genetic suppressors of the aL₁₇₃R mutation.

From diploid strain MR12

The m.8993 T > G mutation converts leucine at position 156 of human subunit a with arginine (aL₁₅₆R). The corresponding residue in the mature form of yeast subunit a is aL₁₇₃ (aL₁₈₃ in the subunit a polypeptide precursor from which the first 10 N-terminal residues are removed during its maturation (Fig. 4A)) (30). In the mutant diploid strain, MR12, the corresponding leucine codon TTA was replaced with the arginine codon AGA (aL₁₇₃R, TTA₁₈₃AGA) (see Materials and Methods). As expected from the severe consequences of the aL₁₇₃R change on ATP synthase (28), MR12 grew very poorly on non-fermentable substrates as compared with the parent strain, MR11 (Fig. 1A).

Figure 4.

Molecular basis of the detriment effect of aL₁₇₃R. (A) Sequence alignment of parts of aH5 (157–191) (right) and aH6 (a.a. 215–245) of subunits a (left) from various origins The shown sequences are from Saccharomyces cerevisiae (S.c.), Yarrowia lipolytica (Y.l.), Schizosaccharomyces pombe (S.p.), Polytomella sp (P.sp.), Arabidopsis thaliana (A.t.), Bos taurus (B.t.), Sus scrofa (S.s.) and Homo sapiens (H.s.). Positions of first-site and second-site reversions in haploid and diploid strains are indicated. (B) Overall top view from the matrix of the a/c₁₀-ring with the pathway along which protons move from the intermembrane space (IMS) to the mitochondrial matrix. The side chains of residues (depicted with balls and sticks) and the n-side (purple) and p-side (green) pockets involved in this transfer are shown. Positions of first-site and second-site reversions and crucial residues for proton translocation are indicated. (C) View from the entry-to-exit channels of amino acid residues surrounding aL₁₇₃. (D) View of the subunit a portion at the interface with the c-ring are drawn as balls and sticks side chains of highly conserved and important residues. Residue side chains coloured in green on a pale green background form an hydrophobic barrier between the two hydrophilic clefts; those strictly conserved are underlined. The entry and exit gates for protons to and from the c-ring are encircled, and their routes within the two half-channels of subunit a are depicted by the green and purple arrowhead lines. (E) View in the same orientation as (C) of the indole nitrogen of aW₁₇₃ presumably responsible for a partial proton shortage between to the hydrophilic clefts in the aL₁₇₃W strain. (F) While the aliphatic part of aR₁₇₃ contributes to the hydrophobic sealing of the n- and p-side clefts, its guanidinium group prevents the exit of protons from the c-ring in proximity of aY₂₄₁. (G) The position of the guanidium group of aK₁₇₃ and its hydrophobic alkyl chain enable proton release from the c-ring in the p-side cleft. (H) Replacing aR₁₆₉ by S or M at the proton exit gate possibly shift and partially restore proton flow despite the presence of aR₁₇₃. The residue color code for all the panels is green for hydrophobicity, cyan and magenta for entry and exit channels, respectively, and yellow for second-site reversions of aL₁₇₃R.

Figure 1.

Influence of the subunit a mutations on yeast respiratory growth. (A) Fresh glucose cultures of the diploid (MR12) and haploid (MR14) aL₁₇₃R mutants, their revertants (designated by their mutation(s) in subunit a) and parental strains (MR11 and MR6, respectively) were serially diluted and spotted onto rich glucose and glycerol plates. The shown plates were scanned after 4 days incubation at 28°C. (B) Growth of the same strains in liquid glycerol medium monitored with the bioscreen system over a period of 80 h during which cell densities (OD_600nm) were taken every 20 min.

Mutant derivatives of MR12 that suppressed or reverted the disease mutation were selected from 22 colonies of MR12 individually grown in glucose media. Cells (10⁸) of each culture were spread on an agar plate containing glycerol as the sole carbon source. On average, 10 colonies grew on each selection plate, from which only 1 or 2 were taken for further analysis. In this way, most of the retained isolates were genetically independent (i.e. not daughter cells). These strains all grew significantly better on medium containing glycerol as the carbon source as compared with the original aL₁₇₃R mutant strain (Fig. 1A). Some of the revertants grew comparable to the wild-type yeast whereas others grew much less efficiently. Because of the nature of the initial mutation, none of the isolates reverted the AGA codon to initial leucine TTA codon as this would have required an extremely rare (about 10⁻¹⁴) double nucleotide change (AGA₁₈₃TTA), and there was no possibility to revert to another leucine codon by a single nucleotide substitution. Sixteen revertants had a mutation that altered the TTA codon into a different amino acid codon (first-site reversion) with AGA₁₈₃ATA (aR₁₇₃M) occurring in 12 clones, AGA₁₈₃AAA (aR₁₇₃K) in 3 clones and AGA₁₈₃AGT (aR₁₇₃S) in 1 clone (Table 1). In the six remaining revertants, a single nucleotide change was identified in another ATP6 codon (second-site reversion): AGA₁₇₉AGT (aL₁₇₃R + aR₁₆₉S) in 3 clones, GCT₁₈₀CCT (aL₁₇₃R + aA₁₇₀P) in 1 clone, GCT₁₈₀GGT (aL₁₇₃R + aA₁₇₀G) in 1 clone and ATC₂₂₆AGC (aL₁₇₃R + aI₂₁₆S) in 1 clone. Upon sporulation, all the revertants produced asci in which all four spores had a better respiratory growth than spores issued from MR12 (aL₁₇₃R) indicating that a nuclear mutation was not contributing to the improved growth phenotype. For the sake of clarity, the first-site suppressors (aR₁₇₃M, aR₁₇₃K and aR₁₇₃S) will be identified relative to the wild-type subunit a sequence (aL₁₇₃M, aL₁₇₃K and aL₁₇₃S).

Table 1.

Intragenic suppressors of aL₁₇₃R

Codon change	a.a change	From MR12	From MR14
First-site reversions
AGA183ATA	𝑎R173M	12	32
AGA183AAA	𝑎R173K	3	5
AGA183AGT	𝑎R173S	1	0
AGA183TGA	𝑎R173W	0	1
Second-site reversions
AGA179AGT	𝑎R169S	3	0
AGA179ATA	𝑎R169M	0	1
GCT180CCT	𝑎A170P	1	0
GCT180GGT	𝑎A170G	1	0
ATC226AGC	𝑎I216S	1	0

From haploid strain MR14

Using the same procedure used for the isolation of revertants from MR12, we isolated revertants from the previously described aL₁₇₃R haploid strain MR14 (MR6 is the parent), which had the same mutant codon (TTA₁₈₃AGA) (28). We identified four different intragenic pseudo-reversions: AGA₁₈₃ATA (aR₁₇₃M) in 32 clones; AGA₁₈₃AAA (aR₁₇₃K) in 5 clones; AGA₁₈₃TGA (aR₁₇₃W) in one clone and AGA₁₇₉ATA (aL₁₇₃R + aR₁₆₉M) in one clone. While the three first-site reversions conferred a good growth on glycerol medium, the strain with the second-site reversion did not grow as well on non-fermentable carbon sources (Fig. 1B).

As expected, the revertant mutations isolated from MR12 and MR14 overlapped. The effect of the mutation aL₁₇₃R in both strains was indeed suppressed with replacement of L₁₇₃ with Lys or Met, and the same second-site suppressor site was identified in codon for aR₁₆₉ and the resultant growth phenotypes were comparable.

Influence on the OXPHOS system of the genetic suppressors of aL₁₇₃R (isolated from strain MR14)

During the course of this study, we determined that MR11, and thus, all its derivatives (MR12 and revertants), contained an additional mutation (cL₅₃F) in the mitochondrial gene encoding subunit c (ATP9), which was determined to substantially compromise ATP synthase function (c.f. Supplementary Material, Info, Figs S1–5). We, therefore, investigated the functional influence of the genetic suppressors only isolated from strain MR14, which did not have the subunit c mutation. However, as cL₅₃ is far (12.4 Å) from aL₁₇₃ in yeast ATP synthase structure (Supplementary Material, Fig. S3C), and because of the overlap between the suppressors identified from MR12 and MR14 (Table 1), we are confident that those isolated from MR12 correct the detrimental consequences of aL₁₇₃R even in the presence of cL₅₃F (see below).

Assembly of OXPHOS complexes

The influence on the assembly/stability of the four genetic suppressors of aL₁₇₃R (aR₁₇₃W, aR₁₇₃K, aR₁₇₃M, and aL₁₇₃R + aR₁₆₉M) selected from strain MR14 was investigated by BN-PAGE of mitochondrial samples solubilized with digitonin (Fig. 2A–E), and by evaluation of the steady state levels of subunit a by western blot analysis (Fig. 2G). ATP synthase complexes were visualized in BN gels by ATPase activity (Fig. 2A), Coomassie blue staining (Fig. 2B), and by using antibodies against subunit c (Atp9) (Fig. 2C), subunit a (Atp6) (Fig. 2D) and the α subunit of F₁ (Atp1) (Fig. 2E). In the conditions used, ATP synthase monomers were much more abundant than oligomers of ATP synthase. There is evidence that most, if not all, of the ATP synthase subunits can assemble in the absence of subunit a, as was observed in yeast strains lacking the ATP6 gene or that of a protein, Atp10, specifically required to assemble subunit a (31,32). However, the resulting assemblies dissociate easily as evidenced by the presence of free F₁ and c-ring in BN-gels. As was already observed (28), the ATP synthase accumulated less efficiently in the aL₁₇₃R strain compared with WT yeast, but free F₁ and c-ring were not detected indicating that the mutated subunit a assembles normally. Since unassembled yeast subunit a is rapidly degraded (33,34) the steady state levels of subunit a in denaturing gels provide better estimates of the levels of fully assembled ATP synthase than BN-PAGE. Based on the level of subunit a, the amount of holo-ATP synthase in aL₁₇₃R samples was at 58% versus WT (Fig. 2G). Since mitochondrial biogenesis is an energy-consuming process, the difference was likely the consequence of the extremely low ATP producing capacity of the aL₁₇₃R mitochondria (10% versus WT (31), see below), rather than specific defects in the assembly of the ATP synthase. Also, partial ATP synthase intermediates were not detected in the samples of the revertants. Good accumulation of subunit a was observed in mitochondria from the aR₁₇₃W, aR₁₇₃K, and aR₁₇₃M strains (77–91% versus WT), whereas the aL₁₇₃R + aR₁₆₉M strain (54%) had a deficit in subunit a similar to that observed in the aL₁₇₃R mutant (58%). The amounts of ATP synthase monomers in BN-gels shown in Figure 2C–E were rather consistent with the levels of subunit a in denaturing gels (Fig. 2F).

Figure 2.

Influence of the subunit a mutations on ATP synthase assembly and accumulation levels. Mitochondria (250 μg) isolated from wild-type yeast MR6, the aL₁₇₃R strain MR14 and its revertants (designated by their mutation(s) in subunit a) were solubilized with digitonin (2 g/g protein) and separated by BN-PAGE in a 3–12% polyacrylamide gel. ATP synthase complexes (monomer, V₁ and oligomers, V_n) were in-gel detected with their ATPase activity (A), Coomassie blue (B), and with antibodies against subunits Atp9/c (C), Atp6/a (D) and Atp1/αF₁ (E) after their transfer to a PVDF membrane. (F) The amounts of ATP synthase monomers, expressed in %WT and normalized to the content of porin in the samples, were estimated from Fig. 2C-E (with ImageJ). (G) Mitochondrial proteins were separated by SDS-PAGE (10 μg per lane), transferred to a nitrocellulose membrane, and probed with antibodies against the indicated proteins. The strains corresponding to lanes 1–6 are indicated. The shown protein detection assays are representative of at least three independent experiments.

We further evaluated the abundance of cytochrome oxidase (Complex IV) by measuring the steady levels of the Cox2 subunit of this complex this protein is degraded when unassembled (35). Secondarily, yeast ATP synthase defective mutants also have a reduced content in cytochrome oxidase except for those that have a large impact on proton leaks through F_O (28,31,36–39), indicating that the biogenesis of this complex is modulated by the activity of F_O (40). Mitochondria from the aL₁₇₃R and the aL₁₇₃R + aR₁₆₉M strains showed a large drop in the content of cytochrome oxidase (80% versus wild type), whereas it was much more abundant in the aR₁₇₃W, aR₁₇₃K and aR₁₇₃M strains (Fig. 2F).

Mitochondrial respiration and ATP synthesis

We next investigated the influence on respiration (oxygen consumption) and ATP synthesis of the genetic suppressors of aL₁₇₃R, in intact (osmotically protected) mitochondria. Oxygen consumption was measured with NADH and ascorbate/TMPD as sources of electrons. The transfer of electrons to oxygen from NADH involves the NADH dehydrogenase located at the external surface of mitochondria (Nde1) and cytochrome b-c₁ reductase and cytochrome oxidase, whereas ascorbate/TMPD delivers electrons directly to cytochrome oxidase. NADH was used alone (State 4, where respiration is induced only by the passive permeability to protons of the inner membrane), in the presence of ADP (State 3 or phosphorylating conditions) or the uncoupler agent CCCP (thus without any membrane potential, conditions where respiration is maximal). The rate of ATP synthesis was measured in the presence of NADH and a large excess of external ADP.

As already reported (28), electrons from NADH and ascorbate/TMPD were transferred slowly to oxygen in aL₁₇₃R mitochondria (20% versus WT, see Table 2), because of their low content in cytochrome oxidase (see above), and the rate of ATP synthesis was reduced in similar proportions (Table 2). These activities were minutely increased by the second-site suppressor of aL₁₇₃R (aR₁₆₉M), which is in line with its very modest respiratory growth improvement versus aL₁₇₃R (Fig. 1B). On the other hand, consistent with their good growth on glycerol medium (Fig. 1), the levels of respiration and ATP synthesis were considerably improved in the first-site revertants (aR₁₇₃W, aR₁₇₃K, aR₁₇₃M) (Table 2).

Table 2.

Mitochondrial respiration and ATP synthesis

Strain	Respiration nmol O. min−1.mg−1 (% WT in parenthesis)				ATP synthesis nmol Pi. min−1.mg−1
	NADH	NADH+ADP	NADH+CCCP	Asc/TMPD +CCCP	-Oligo	+Oligo	F1FO % WTa	F1FO % WTb
WT	329 ± 26 (100)	650 ± 73 (100)	1550 ± 96 (100)	2809 ± 299 (100)	640 ± 100	15 ± 3	100	100
𝑎L173R	137 ± 17 (42)	132 ± 32 (20)	279 ± 38 (18)	489 ± 66 (17)	48 ± 13	10 ± 3	<10%	10
𝑎L173W	341 ± 14 (104)	675 ± 43 (104)	1558 ± 78 (101)	2548 ± 80 (91)	488 ± 27	16 ± 1	76	83
𝑎L173K	265 ± 24 (81)	422 ± 27 (65)	986 ± 69 (64)	1400 ± 34 (50)	326 ± 36	16 ± 2	50	64
𝑎L173M	247 ± 14 (75)	436 ± 30 (67)	1052 ± 80 (68)	1327 ± 208 (47)	405 ± 26	17 ± 4	62	75
𝑎R169M	206 ± 12 (63)	188 ± 28 (29)	342 ± 39 (22)	526 ± 77 (19)	75 ± 15	22 ± 5	<10%	16

Mitochondria were isolated from cells grown for 5–6 generations in rich galactose medium (YPGalA) at 28 °C. Reaction mixes for assays contained 0.15 mg/mL protein, 4 mM NADH, 150 μM ADP (for respiration assays), 1 mM ADP (for ATP synthesis assays), 12.5 mM ascorbate, 1.4 mM N,N,N,N,-tetramethyl-p-phenylenediamine (Asc/TMPD), 4 μM CCCP, 4 μg/mL oligomycin (oligo). F₁F_O represents the specific ATP synthesis by F₁F₀-ATP synthase relative to WT, per mg of protein (^a) and after normalization to the relative contents of subunit a (^b) determined in Figure 2F.

Mitochondrial ATP hydrolysis

We tested ATP hydrolysis in non-osmotically-protected mitochondria (this to relax the enzyme from any membrane potential that would limit its ATP hydrolytic activity). ATPase was measured at pH 8.4 to prevent binding of the natural peptide inhibitor IF1 to F₁ (41). Oligomycin inhibits the ATPase activity of fully assembled and coupled mitochondrial ATP synthase, but not, for instance, free F₁-ATPase. In samples with the wild-type enzyme, oligomycin inhibits ATPase activity by about 80% (Table 3). Any oligomycin-insensitive activity is normally due to other ATPases in the preparation. The mitochondrial samples from the aL₁₇₃R mutant and three of its revertants (aL₁₇₃W, aL₁₇₃M, aL₁₇₃K) showed good oligomycin-sensitive ATPase activity. Surprisingly, while ATP synthesis activity from aL₁₇₃R was severely diminished, the ATPase activity was not decreased and ATPase was sensitive to oligomycin. This result seems to separate ATPase activity from ATP synthesis activity. In contrast, the ATPase activity of mitochondria isolated from yeast second-site suppressor strain, aL₁₇₃R + aR₁₆₉M, was just half the rate relative to the wild type, and it was only inhibited 25% by oligomycin. Thus, while the second-site suppressor mutation aR₁₆₉M decreased the ATPase activity and oligomycin sensitivity, it increased the rate of ATP synthesis relative to the mitochondria isolated from the strain with aL₁₇₃R in an otherwise wild-type background.

Table 3.

Mitochondrial ATPase activity

Strain	ATPase activity μmol Pi. min−1.mg−1
	-Oligo	+Oligo	% Inhib.
WT	6.851 ± 0.292	1.623 ± 0.179	76
𝑎L173R	4.767 ± 0.321	1.517 ± 0.496	68
𝑎L173W	6.881 ± 0.058	1.092 ± 0.079	84
𝑎L173K	6.509 ± 0.042	1.405 ± 0.321	78
𝑎L173M	7.064 ± 0.042	1.329 ± 0.213	81
𝑎R169M	3.832 ± 0.267	2.826 ± 0.346	26

The ATPase assays were performed from mitochondrial samples that had been frozen at −80°C in the absence of osmotic protection at pH 8.4, in the ATPase buffer (see section Materials and Methods).

Mitochondrial membrane potential

We further investigated (using intact mitochondria buffered at pH 6.8) the effect of the suppressor amino acid changes had on the ability of the mitochondria to generate a transmembrane potential gradient (ΔΨ). We assessed this using Rhodamine 123, a cationic dye whose fluorescence is quenched relative to the ΔΨ. The ΔΨ was made by the electron transport chain by oxidation of NADH formed by the oxidation of ethanol. The resulting potential gradient can be diminished by the ATP synthase with the phosphorylation of ADP (Fig. 3A). The addition of ADP normally results in a transient fluorescence increase that progressively disappears until all of the ADP is phosphorylated (Fig. 3A). Thus, this assay evaluates how tightly proton transport is coupled with ATP synthesis and the ability of the ATP synthase to phosphorylate ADP. It is to be noted that the activity of the electron transport chain needs to be strongly diminished (by at least 80%) to observe a reduced ability to create a potential gradient with the oxidation of ethanol. Accordingly, consistent with their respective rates of oxygen consumption (Table 2), mitochondria from the aL₁₇₃R mutant responded poorly to the addition of ethanol whereas those from the other tested strains generated a higher ΔΨ (Table 2). Consistent with the respective rates of ATP synthesis (Table 2), the potential induced by ethanol in mitochondria from the aL₁₇₃R and aL₁₇₃R + aR₁₆₉M strains did not change when ADP was added, whereas it did so in those from the first-site revertants (aL₁₇₃W, aL₁₇₃M, aL₁₇₃K) (Fig. 3A). This indicates that while the ATP synthase from aL₁₇₃R can hydrolyze ATP and ATPase is sensitive to oligomycin, it cannot establish a proton gradient with ATP hydrolysis. KCN inhibits the electron transport chain by inhibiting cytochrome oxidase. Normally, KCN addition after ATP synthesis results in a partial loss of membrane potential because the newly formed ATP is hydrolyzed by the ATP synthase, which pumps protons outside the mitochondrion (as evidenced by the loss of the remaining ΔΨ after a further addition of oligomycin (Fig. 3A)). Mitochondria isolated from the strains with the first-site revertants clearly showed this biphasic loss of ΔΨ, whereas the membrane potential was largely and rapidly lost after KCN addition in mitochondria from the aL₁₇₃R and aL₁₇₃R + aR₁₆₉M strains, which further reflects the differences in the activity of the ATP synthase.

Figure 3.

Influence of the subunit a mutations on ATP synthase-dependent variations in the electrical potential across the mitochondrial inner membrane. Variations in mitochondrial transmembrane potential (ΔΨ) were monitored by fluorescence quenching of Rhodamine 123 in intact mitochondria from wild-type yeast MR6, the aL₁₇₃R strain MR14 and its revertants (designated by their mutations(s) in subunit a as indicated). (A) Tests of ΔΨ consumption by the ATP synthase upon addition of ADP to mitochondria energized with electrons from ethanol. (B) ATP-driven proton pumping assays. The additions were 25 μg/mL rhodamine 123, 150 μg/mL mitochondrial proteins (Mito), 10 μL ethanol (EtOH), 75 μM ADP, 2 mM potassium cyanide (KCN), 0.2 mM ATP, 4 μg/mL oligomycin (oligo) and 4 μM carbonyl cyanide-m-chlorophenyl hydrazone (CCCP). The shown fluorescence traces are representative of at least three independent experiments.

We also examined ATP-driven proton-pumping activity of ATP synthase in the presence of large excess of externally added ATP (Fig. 3B). To this end, mitochondria were energized with ethanol and the resulting ΔΨ was collapsed by adding KCN, releasing inhibitor IF1 from the ATP synthase. ATP was then added less than 1 min later, before IF1 rebinding (42) to determine if ATP hydrolysis will generate a ΔΨ. As expected, a large oligomycin-sensitive ΔΨ variation was observed in mitochondria isolated from the wild-type strain. This was also observed in mitochondria from the first-site revertant strains (aL₁₇₃W, aL₁₇₃M, aL₁₇₃K), but a much reduced proton-pumping activity was observed in mitochondria from aL₁₇₃R and aL₁₇₃R + aR₁₆₉M strains (Fig. 3B). Thus, again, while the ATP synthase of the aL₁₇₃R mutant showed a good oligomycin-sensitive ATP hydrolytic activity in the absence of any membrane potential (Table 2), it fails to work in the reverse mode in intact mitochondria. Possibly, the aL₁₇₃R mutation prevents the any significant movement of protons through the F_O against a proton gradient.

Structural modeling of the subunit a mutations

Subunits a and c are major components of Fo, which are directly responsible for the movement of protons coupled to the synthesis of ATP by the ATP synthase. Starting from the N-terminus, yeast subunit a has a membrane spanning α-helix (aH1) followed by a 4-helix bundle (aH3-aH6) horizontally wrapped around the c-ring rotor (Fig. 4B) (25). The two domains are linked on the matrix side of the membrane by an amphipathic α-helix (aH2). Two charged residues (aR₁₇₆ in aH5 and cE₅₉ in cTM2) facing each other near the middle of the membrane at the a/c-ring interface are essential for activity (25,26). Two hydrophilic pockets enable protons from the p-side of the membrane (intermembrane space) to reach cE₅₉ and be released after an almost complete rotation of the c-ring on the n-side (matrix) of the membrane (Fig. 4B). The n-side cleft has a funnel shape 15 Å long (from aR₁₇₆ to aD₂₄₄ or aE₁₆₂), 8 Å wide (from aD₂₄₄ to aE₁₆₂) and 16 Å deep (from aS₁₆₅ to the C-terminus of subunit a) (Fig. 4B). It is bordered by polar and electrically charged residues of aH6 (aS₂₄₀, aY₂₄₁, aK₂₄₃, aD₂₄₄, aH₂₄₉) and aH5 (aE₁₆₂, aS₁₆₅, aR₁₆₉, aS₁₇₂) (Fig. 4D), two of which (aE₁₆₂ and aD₂₄₄) are essential for moving protons out of the n-side cleft (25). The side chain of aR₁₇₆, which is oriented towards the n-side cleft, helps through electrostatic interactions discharged subunit c monomer to rotate and be reloaded with a proton in the p-side cleft (25,43–45) (Fig. 4D). In addition to aL173, nine other residues are strictly conserved in subunits a: (i) aP₁₀₆ allows the turn between aH3 and aH4; (ii) A₁₈₃ ensures the shortest contact with the c-ring; (iii–vi) four hydrophilic residues line the bottom of the two half channels (aN₁₀₀, aN₁₈₀ and aQ₂₃₀ in the entry one, and aY₂₄₁ in the exit one); and (vii–ix) three leucine residues (aL₁₇₃, aL₁₇₇ and aL₂₃₇) that together with three other highly but not strictly conserved hydrophobic residues (aI₁₂₅, aV₂₃₃ and aW₂₃₄) form an hydrophobic plug that prevents proton shortage between the two channels (Fig. 4A and C).

Residue aL₁₇₃ altered by the m.8993 T > G mutation is in aH5 just one α-helix turn from aR₁₇₆ and aR_169, and faces the essential glutamate in the c-ring involved in proton translocation, cE₅₉ (Fig. 4D). Because of the proximity of aL₁₇₃ to aR₁₇₆ and cE₅₉, it is not very surprising that its replacement with arginine compromises the function of F_O (Fig. 4F). Without large conformational changes in subunit a, the aliphatic moiety of aR₁₇₃ side chain preserves the hydrophobic barrier between the two proton half channels, whereas its guanidinium group increases electropositivity in the exit channel (Fig. 4D and F). As a result, proton discharge in the n-cleft and/or rotation of the c-ring might be expected to be affected. Consistent with this, the activity of the ATP synthase was substantially improved when aR₁₇₃ was replaced with uncharged residues (aS₁₇₃, aM₁₇₃, aW₁₇₃) (Fig. 4E). In apparent contradiction, the activity was also substantially recovered when aR₁₇₃ was replaced with lysine (aK₁₇₃). However, being further from the exit channel, the electric influence of the amine group of aK_173, in particular on aD₂₄₄ and aE₁₆₂, is less important (Fig. 4G). Furthermore, the pKa of lysine is significantly lower than that of arginine, and within the local environment, the lysine may be uncharged. Partial recovery of ATP synthase activity by the aR₁₆₉M/S changes further supports that the positive charge of aR₁₇₃ compromises F_O function (Fig. 4H). Notably, the aR₁₆₉M mutation was also shown to compensate for another human pathogenic subunit a change in yeast (aW₁₂₆R) presumably by preserving proper interactions between aR₁₇₆ and cE₅₉ (46). The proximal second-site reversions of aL₁₇₃R (aA₁₇₀P, aA₁₇₀G) might similarly attenuate the detrimental effects of aR₁₇₃ by structurally shifting its side chain.

Although substantially restored with aM₁₇₃, aW₁₇₃ or aK₁₇₃, the measured ATP synthesis rate was diminished by 25–50% versus wild-type yeast (Table 2). For aM₁₇₃ and aK₁₇₃, the rates of ATP synthesis and electron transfer to oxygen were similarly diminished, indicating a reduction in the rate of proton transport through the F_O possibly because of a reduced hydrophobicity around aR₁₇₆ that makes the exit of protons towards the matrix and/or rotation of the c-ring less efficient. With aW₁₇₃, there is a substantial (20–25%) drop in ATP production rate despite the preservation of a normal rate of electron transfer to oxygen. This indicates that the proton flow through the F_O is only minimally affected. Indeed, as already mentioned above, yeast ATP synthase defective mutants show a slower rate of oxygen consumption only when F_O activity is compromised (40). A possible explanation for the effects of aL₁₇₃W is a partial proton shortage due to a distortion by the indole group of tryptophane of the hydrophobic barrier that seals the two half proton channels (Fig. 4E).

Conclusion

With the recently described cryo-EM structures of F₁F_O ATP synthases from various mitochondrial origins (25–27,44,45), it has become feasible to map at a molecular level discrete structural changes induced by mutations of this enzyme involved in human diseases. However, even with this information and a detailed knowledge of their biochemical consequences alone, it can be difficult to understand how ATP synthase function is compromised by the mutations. Genetic suppressors provide another input to help determine the mechanisms of pathogenic mutations. This approach is easily done in the yeast S. cerevisiae and this has proved successful in a number of previous studies (46–49). We herein applied it to the most common pathogenic mutation (m.8993 T > G, aL₁₇₃R) of the mitochondrial ATP6 gene encoding the subunit a of ATP synthase. As we have shown (28), the leucine-to-arginine change (aL₁₇₃R) leads to a 90% drop in yeast mitochondrial ATP synthesis, indicating a severe functional impairment of F_O, but without defects in the assembly/stability of subunit a. Nine different amino acid changes of varying suppressor activity, at the original mutation site (aL₁₇₃M, aL₁₇₃S, aL₁₇₃K and aL₁₇₃W) or in another position of subunit a (aR₁₆₉M, aR₁₆₉S, aA₁₇₀P, aA₁₇₀G and aI₂₁₆S) were here identified. The results support the hypothesis that the positively charged guanidine group of aR₁₇₃ prevents the release of protons from the c-ring into the n-side hydrophilic cleft of F_O likely due to unfavorable electrostatic interactions with two residues (aR₁₇₆ and cE₅₉), which are essential for the coupled movement of protons across the mitochondrial inner membrane with rotation of the c-ring.

Residue aL₁₇₃ is not directly involved in proton conduction within the F_O as it is located outside the n-side cleft and because its aliphatic side chain cannot exchange protons. Our results suggest, however, that aL₁₇₃ is indirectly involved in the proton-conduction pathway. Indeed, none of the suppressors fully restored ATP synthase function. For aK₁₇₃ and aM₁₇₃, F_O functions poorer compared to the wild-type enzyme, aW₁₇₃ is responsible for a partial proton shortage between its two hydrophilic clefts, and respiratory growth is compromised with aS₁₇₃. Using a similar suppressor genetics approach, we concluded that two other strictly conserved leucine residue of subunit a (aL₂₃₇ and aL₂₄₂) targeted by pathogenic mutations (m.9176 T > G, m.9176 T > C, and m.9191 T > C) also indirectly improve proton conduction within the wild-type F_O (47,49).

In a previous study (48), we also provided evidence that another pathogenic mutation of subunit a (m.8969G > A; aS₁₆₅N) leads to the establishment of hydrogen bonds that interferes with the proton conduction involving aE₁₆₂. Remarkably, the m.8969G > A mutation was efficiently suppressed by long-range interactions upon mutations on the p-side of the membrane (at position 190 on aH5) (48). Similarly, the aL₁₇₃R mutation was partially suppressed by a mutation (aI₂₁₆S), which is close to residue 190 (on aH6) in the structure (Fig. 4A). This is an interesting finding that holds promise for the development of therapeutic molecules acting from the p-side of the membrane (thus without the need to cross the inner membrane).

Studies of human cells and tissues containing the m.8993 T > G mutation led to conflicting conclusions: a block in F_O-mediated proton conduction (21), defects in ATP synthase assembly/stability (13,15), or impaired coupling between F₁ and F_O (18,50). The present yeast-based study provides evidence that the leucine-to-arginine change induced by this mutation prevents the release of protons from the c-ring due to unfavorable electrostatic interactions with two catalytic residues of F_O (aR₁₇₆ and cE₅₉) without any defect in ATP synthase assembly/stability and proton shortage within the F_O.

Materials and Methods

Growth media and genotypes

The media used for growing yeast were: YPGA (1% bacto yeast extract, 1% bacto peptone, 2% glucose, 60 mg/l adenine), YPGA10 (1% bacto yeast extract, 1% bacto peptone, 10% glucose, 60 mg/l adenine), YPGalA (1% bacto yeast extract, 1% bacto peptone, 2% galactose, 60 mg/l adenine), YPGlyA (1% bacto yeast extract, 1% bacto peptone, 2% glycerol, 60 mg/l adenine), WO (2% glucose, 0.67% yeast nitrogen base with ammonium sulfate from Difco), SP1: 0.1% glucose, 0.25% yeast extract, 50 mM potassium acetate. Solid media were made with 2% bacto agar (Difco, Becton Dickinson). The genotypes of the strains used in this study are listed in Table 4. Growth curves were determined with the Bioscreen C™ system (Piscataway, NJ).

Table 4.

Genotypes and origin of yeast strains

Strain	Nuclear genotype	mtDNA	Source
MR6	MAT𝑎 ade2–1 his3–11,15 trp1–1 leu2–3112 ura3–1 CAN1 arg8::HIS3	ρ+ WT	(31)
MR14	MAT𝑎 ade2–1 his3–11,15 trp1–1 leu2–3112 ura3–1 CAN1 arg8::HIS3	ρ+atp6-L173R	(28)
DFS160	MATα ade2–101 ura3–52 leu2∆ arg8::URA3 kar1–1	ρ0	(64)
KL14-4A	MAT𝑎 his1 trp2	ρ+oli1r-1-1 (atp9-L53F) capr-1-321 parr-1-454	(65) This study
SDC31	MATα ade2–101 ura3–52 leu2∆ arg8::URA3 kar1–1	ρ−atp6-L173R	(28)
MR11	MATα/𝑎 ade2–101/+ ura3–52/+ leu2∆/+ arg8::URA3/+ kar1–1/+ +/his1 +/trp2	ρ+oli1r-1-1 (atp9-L53F) capr-1-321 parr-1-454	This study
MR12	MATα/𝑎 ade2–101/+ ura3–52/+ leu2∆/+ arg8::URA3/+ kar1–1/+ +/his1 +/trp2	ρ+atp6-L173R oli1r-1-1 (atp9-L53F) capr-1-321 parr-1-454	This study
RMR12/1	Revertant isolated from MR12	ρ+atp6-L173M	This study
RMR12/6	Revertant isolated from MR12	ρ+atp6-L173K	This study
RMR12/11	Revertant isolated from MR12	ρ+atp6-L173S	This study
RMR12/8	Revertant isolated from MR12	ρ+atp6-L173R + R169S	This study
RMR12/13	Revertant isolated from MR12	ρ+atp6-L173R + A170P	This study
RMR12/20	Revertant isolated from MR12	ρ+atp6-L173R + A170G	This study
RMR12/5	Revertant isolated from MR12	ρ+atp6-L173R + I216S	This study
RMR14/4	Revertant isolated from MR14	ρ+atp6-L173W	This study
RMR14/6	Revertant isolated from MR14	ρ+atp6-L173K	This study
RMR14/12	Revertant isolated from MR14	ρ+atp6-L173M	This study
RMR14/48	Revertant isolated from MR14	ρ+atp6-L173R + R169M	This study

Construction of the aL₁₇₃R diploid strain (MR12)

As described in (28), the plasmid (pSDC31) harboring the ATP6 gene with the aL₁₇₃R mutation (TTA₁₈₃AGA) was introduced into the mitochondria of the strain DFS160 that is entirely devoid of mtDNA (ρ⁰). The resulting ρ⁻ synthetic strain (SDC31) was crossed to the ρ⁺ strain KL14-4A. Diploids cells were selected in a minimal glucose medium (WO) and allowed to divide for at least 15 generations for homoplasmy of ρ ⁺ mtDNA recombinants carrying the atp6-L₁₇₃R mutation. Single diploid clones were selected on solid minimal medium (WO) and tested for growth on rich glycerol medium (YPGlyA). Out of 1000 clones, 5 (called MR12) carried the aL₁₇₃R mutation and all 5 grew very slowly on glycerol in comparison with the control diploid strain (MR11) obtained by crossing DFS160 to KL14-4A.

Selection of genetic suppressors of aL₁₇₃R

Genetic suppressors of aL₁₇₃R were selected from diploid strain MR12 and a haploid strain (MR14) carrying the same mutation that was previously described (28). To ensure genetic independence of the isolates, the strains were streaked for single colonies on YPGA plates and dozens of subclones were picked up and individually grown for two days in 10% glucose (YPGA). Glucose was removed from the cultures by washings twice with water and 10⁸ cells from each culture were spread on rich glycerol (YPGlyA) plates. The plates were incubated at 28°C for at least fifteen days. A maximum two revertants per plate were selected for further analysis. The revertants were purified by subcloning on YPGA and their ATP6 gene was PCR-amplified and sequenced with two pairs of primers: (i) 5’-TAATATACGGGGGTGGGTCCCTCAC (Forward, from nucleotide position −100 upstream of the ATP6 start codon) and 5’-CTGCATCTTTTAAATATGATGCTG (Reverse, from nucleotide position +743 downstream of the ATP6 start codon); (ii) 5’-GTATGATTCCATACTCATTTG (Forward, from nucleotide position +337 downstream of the ATP6 start codon) and 5’-GGGCCGAACTCCGAAGGAGTAAG (reverse, from nucleotide position +218 downstream of the ATP6 stop codon).

Miscellaneous procedures

Mitochondria were isolated by the previously described enzymatic method (51) from yeast cells grown until middle exponential phase (2–3 × 10⁷ cells/mL) in rich galactose medium (YPGalA) at 28°C. Protein concentration was determined by the Lowry method (52) in the presence of 5% SDS. Oxygen consumption rates were measured with a Clark electrode using 150 μg/mL of mitochondrial proteins in a respiration buffer (0.65 M mannitol, 0.36 mM ethylene glycol tetra-acetic acid, 5 mM Tris/phosphate, 10 mM Tris/maleate pH 6.8) as in (53). The additions were 4 mM NADH, 12.5 mM ascorbate (Asc), 1.4 mM N, N, N′, N′-tetramethyl-p-phenylenediamine (TMPD), 150 μM ADP and 4 μM carbonyl cyanide m-chlorophenylhydrazone (CCCP). The transmembrane potential (ΔΨ) was evaluated in respiration buffer and 150 μg/mL of mitochondrial protein, by monitoring fluorescence quenching of Rhodamine 123 (25 μg/mL) using an SAFAS Monaco spectrophotometer as described in (54). The additions were 10 μL ethanol (EtOH), 75 μM ADP, 2 mM potassium cyanide (KCN), 0.2 mM ATP, 4 μg/mL oligomycin (oligo) and 4 μM CCCP. ATP synthesis rate measurements were performed with 150 μg/mL of mitochondrial proteins in respiration buffer supplemented with 4 mM NADH and 1 mM ADP in a thermostatically controlled chamber at 28°C, as described in (31). Aliquots of 50 μL were withdrawn from the oxygraphy cuvette every 15 s and the reaction was stopped with 7% (w/v) perchloric acid, 25 mM EDTA. The samples were then neutralized to pH 6.8 by adding 2 M KOH/0.3 M MOPS, and ATP was quantified using a luciferin/luciferase assay (ATPlite kit from Perkin Elmer) and the CLARIO star microplate reader (from BMG LABTECH). The participation of the F₁F_O-ATP synthase in ATP production was assessed by adding oligomycin (20 μg/mg of protein). ATPase activity measurements were performed with 150 μg/mL of mitochondrial proteins in Somlo ATPase buffer (0.2 M KCl, 3 mM MgCl₂, 10 mM Tris–HCl, adjusted to pH 8.4) as described (55). The reaction was started by the addition of 50 μL 0.1 M ATP. After 2 min, the reaction was stopped and proteins were precipitated by adding TCA 100% and ATP hydrolysis was measured by measuring inorganic phosphate using ammonium molybdate solution and measuring the absorbance at 610 nm (56). Oligomycin sensitive ATP hydrolysis of F₁F_O-ATPase was assessed by adding oligomycin (2 μg/mL). Blue native polyacrylamide gel electrophoresis (BN-PAGE) analyses was done as described (57). Briefly, mitochondrial extracts solubilized with 2 g digitonin/g protein were separated in a 3–12% acrylamide continuous gradient gel. ATP synthase complexes were detected in-gel by their ATPase activity as described (58), or by western blot after transferred to polyvinylidene difluoride (PVDF) using polyclonal antibodies against Atp6/subunit a (a gift from J. Velours), Atp1/subunit α (a gift from J. Velours) and Atp9/subunit c after 1:5000 dilution (59). SDS-PAGE analyses of mitochondrial proteins was performed according to (60). 1:500 dilutions of monoclonal antibodies against Cox2 and Porin (Molecular Probes) were used. The tested proteins were revealed using peroxidase-labeled antibodies (Promega) at a 1:5000 dilution and the ECL reagent of Pierce™ ECL western blotting substrate (ThermoScientific). Immunological signal quantification was performed using ImageJ (61).

Amino-acid alignments and structural modeling

Multiple sequence alignment of ATP synthase a-subunits of various origins was performed using Clustal Omega (62). Molecular views of subunit a and c₁₀-ring were obtained from the dimeric Fo domain of S. cerevisiae ATP synthase (pdb_id: 6b2z; 3.6 Å resolution (25)). All structure figures were drawn using PyMOL molecular graphic system (63).

Funding

A Ph.D. fellowship from China Scholarship Council (CSC) to X.S.; PhD fellowships from CNRS and Retina foundation, respectively to M.Bi. and M.R.; the Association Française contre les Myopathies (#22382 to D.T.T.) and J.P.dR.; the National Science Center of Poland (NSCP 2016/23/B/NZ3/02098 to R.K.); from National Institutes of Health (R35GM131731 to D.M.M).

Author contributions

X.S., M.R., M.Bi., R.K. and N.E.G. isolated, sequenced and biochemically characterized yeast strains under the supervision of D.T.T. and J.P.dR. and with the technical support of M.Bo. and F.G. a.d. performed the structural modeling analyses. X.S., a.d., R.K., M.Bi., D.T.T., J.P.dR. and D.M.M. analyzed the data. X.S., a.d., D.T.T., J.P.dR. and D.M.M. wrote the paper. D.T.T. and J.P.dR. designed the research. Conflict of interest statement: The authors declare no competing or financial interests.