Arginine 107 of yeast ATP synthase subunit g mediates sensitivity of the mitochondrial permeability transition to phenylglyoxal

Modification with arginine-specific glyoxals modulates the permeability transition (PT) of rat liver mitochondria, with inhibitory or inducing effects that depend on the net charge of the adduct(s). Here, we show that phenylglyoxal (PGO) affects the PT in a species-specific manner (inhibition in mouse and yeast, induction in human and Drosophila mitochondria). Following the hypotheses (i) that the effects are mediated by conserved arginine(s) and (ii) that the PT is mediated by the F-ATP synthase, we have narrowed the search to 60 arginines. Most of these residues are located in subunits α, β, γ, ϵ, a, and c and were excluded because PGO modification did not significantly affect enzyme catalysis. On the other hand, yeast mitochondria lacking subunit g or bearing a subunit g R107A mutation were totally resistant to PT inhibition by PGO. Thus, the effect of PGO on the PT is specifically mediated by Arg-107, the only subunit g arginine that has been conserved across species. These findings are evidence that the PT is mediated by F-ATP synthase.

and mutagenesis to identify the unique arginine of F-ATP synthase subunit g (of a total of 60 conserved residues) that mediates the PTP-modulating effect of PGO in yeast. This finding is a step forward in the molecular understanding of PTP regulation and further supports the hypothesis that the PTP forms from F-ATP synthase.

The effects of PGO on the PTP have been conserved but differ between species
Consistent with the effect of PGO on the PTP of rat liver mitochondria, PGO desensitized the PTP to Ca 2ϩ in mouse liver mitochondria (MLM) (Fig. 1A) as well as in mitochondria from Saccharomyces cerevisiae in the presence of ETH129, a selective ionophore that allows Ca 2ϩ equilibration across the inner membrane in yeast mitochondria (Fig. 1B). In either case, the desensitizing effect can be appreciated from the increased Ca 2ϩ retention capacity (CRC) of PGO-treated mitochondria (Fig. 1, A and B). Unexpectedly, PGO caused instead sensitization of the pore to Ca 2ϩ in permeabilized human HEK293 (Fig.  1C) and Drosophila S 2 R ϩ cells (Fig. 1D). Sensitization of PTP opening by PGO in human and Drosophila cells was not due to digitonin permeabilization, as it was also observed in mitochondria isolated from these cells (results not shown). These findings indicate that the ability to modulate the PTP by PGO has been conserved across species. Under the assumptions (i) that the observed differences depend on specific features of the PTP rather than on the existence of speciesspecific reactive sites and (ii) that these sites are located on the F-ATP synthase, we narrowed the search of the reactive arginine(s) from 135 (yeast) to 60 arginines that are conserved between F-ATP synthases of S. cerevisiae, Homo sapiens, Mus musculus, and Drosophila melanogaster (the positions of these residues in the yeast, human, and mouse enzymes are reported in Table 1).

Chemical modification by PGO does not affect catalytic activity, oligomycin sensitivity, assembly, and subunit composition of F-ATP synthase
Due to the large number of arginine residues and their specific relevance to enzymatic activity of F-ATP synthase, we assessed whether treatment with PGO affects the catalytic properties of the enzyme in MLM and yeast mitochondria (YM). In MLM, a trend toward a slight inhibition of ATP hydrolysis was observed (but this did not reach statistical significance), and the catalytic activity maintained its sensitivity to oligomycin ( Fig. 2A), and respiratory stimulation by ADP was not affected (Fig. 2B). Similarly, in YM, both ATP hydrolytic activity and respiratory stimulation by ADP were unaffected by PGO (Fig. 2, C and D). Blue native gel separation and in-gel activity staining after digitonin extraction did not reveal any major effect of PGO on F-ATP synthase oligomeric state in MLM (Fig. 2E). In these experiments, we used MLM because they are easy to purify and because the mouse enzyme is virtu- Representative traces are shown. Bars, CRC of DMSO-(black) or PGO-pretreated (red) isolated mitochondria or permeabilized cells. CRC of DMSO-treated mitochondria was set as 1 unit. Results are mean Ϯ S.E. (error bars) of at least three independent experiments. *, p Ͻ 0.05 versus vehicle; **, p Ͻ 0.01 versus vehicle. One-way ANOVA was used for statistical analysis.

Subunit g Arg-107 of F-ATPase regulates PTP
ally identical to the human and yeast species for composition of conserved arginines ( Table 1). Analysis of dimers and monomers resolved by SDS-PAGE and silver staining demonstrated that treatment with PGO did not alter the pattern and relative abundance of F-ATP synthase subunits (Fig. 2F). Based on these results, the arginine residues of the F 1 sector, subunit a, and c-ring were not investigated further. Taking advantage of the yeast model and of the species-specific response to PGO, we performed a set of mutagenesis experiments in yeast cells to identify the arginine residue(s) conferring the modulatory effect to PGO.

OSCP does not confer sensitivity to the effects of PGO
We first investigated whether OSCP, which contains two conserved arginine residues (Table 1), could be involved in conferring the species specificity to the effects of PGO, which are opposite in human and yeast mitochondria (Fig. 1). Deletion of the ATP5 gene encoding for OSCP (⌬ATP5) prevented F-ATP synthase assembly and growth on glycerol, both features being rescued by reexpression of the WT yeast gene (⌬ATP5 ϩ yOSCP) (Fig. 3, A-C). Replacement with the human OSCP gene (⌬ATP5 ϩ hOSCP) was instead essentially unable to rescue F-ATP synthase assembly unless the yeast mitochondrial targeting sequence replaced the human sequence in the construct (⌬ATP5 ϩ yhOSCP). Indeed, only the chimeric protein allowed assembly of a functional F-ATP synthase complex, which grew normally on glycerol (Fig. 3, A-C). The rescued mutant displayed an increased CRC in response to PGO (i.e. the same response of WT yeast mitochondria) (Fig. 3D). Thus, OSCP does not confer species specificity to the effects of PGO on the PTP.

Arginine 107 of yeast F-ATP synthase subunit g mediates the effect of PGO on the mitochondrial permeability transition
Both subunit e and g possess a conserved arginine residue (Arg-8 and Arg-107 in the yeast sequence, respectively, Fig. 4). These residues correspond to Arg-15 and Arg-96, respectively, in the human sequence (Table 1). To test whether these resi-dues are involved in PTP modulation by PGO, we first analyzed deletion mutants of subunit g (⌬ATP20) and of subunit e (⌬TIM11). The effect of PGO was completely abolished in the ⌬ATP20 deletion mutant lacking subunit g (Fig. 5A), and a similar, albeit somewhat smaller, effect was observed with the ⌬TIM11 mutant lacking subunit e (Fig. 5B). To explore the role of individual arginines, we generated specific point mutants and tested their CRC. Remarkably, the R107A mutation in subunit g dramatically blunted the effect of PGO (Fig. 5C), whereas the R8A mutation in subunit e was ineffective, with an identical sensitivity to PGO as the WT species (Fig. 5D). This result was puzzling because subunit e only has one conserved arginine in position 8, and therefore we would have expected the same result following either deletion of the subunit e gene or the R8A mutation. This apparent contradiction was resolved by analysis of the expression of the e and g subunits in the ⌬ATP20 and ⌬TIM11 mutants. Indeed, deletion of the ATP20 gene led to the expected disappearance of subunit g only, whereas deletion of the TIM11 gene caused the disappearance of both subunit e and g (Fig. 5E). Both Arg-8 and Arg-107 are located next to the GXXXG dimerization domains (Figs. 4 and 5F), and we suspect that this is the region where the PTP forms.

The ATP20 Arg-107 mutation does not affect growth properties, dimerization, and subunit composition of F-ATP synthase
Given the importance of subunits e and g in both F-ATP synthase dimerization and PTP formation in yeast (19), we characterized the key features of the R107A mutants. Yeast cells can grow by fermentation in glucose-rich medium, whereas they need ATP generated from oxidative phosphorylation in glycerol medium. The ATP20 R107A mutant grew both on the glucose and glycerol (Fig. 6A), indicating that the mutation did not influence the function of F-ATP synthase. The ⌬TIM11 and ⌬ATP20 mutants expectedly did not display dimers after digitonin extraction and blue native PAGE (Fig. 6B). Mitochondria  6B) with identical subunit composition, as judged from SDS-PAGE and silver staining (Fig. 6C). Thus, the R107A replacement in subunit g did not influence the assembly, activity, and stability of F-ATP synthase.

PGO reacts with Arg-107 of subunit g and affects the pore only in mitochondria with the WT subunit
We next tested whether the PGO adduct affects the properties of the PTP. One emerging issue is pore size, which is affected by the deletion of lateral stalk subunits OSCP and b even when the CRC is apparently unaffected (24). We therefore measured Ca 2ϩ -induced swelling in yeast mitochondria in the presence of ETH129 to allow electrophoretic Ca 2ϩ uptake. The swelling rate was significantly inhibited by PGO in mitochondria reexpressing WT subunit g but not the R107A mutant, which displayed a lower rate of swelling (Fig. 7, A-C). Because yeast subunit g also contains an arginine residue at position 106, we employed synthetic test peptides corresponding to the 88 -114 amino acid segment of WT or containing the R106A replacement. Following reaction with PGO, the peptides were analyzed by MALDI-TOF MS. The results revealed that one adduct with a molecular mass of 116 Da was formed on both the WT and the R106A modified peptide (Fig. 7D). This molecular mass corresponds to one molecule of PGO covalently attached to the guanidine moiety of arginine. Formation of one single adduct in the WT peptide carrying two adjacent arginines suggests that reaction of Arg-107 with PGO may reduce the reactivity of Arg-106. . Values refer to state 3/state 4 ratio and are expressed as mean Ϯ S.E. of three independent experiments. C, ATP hydrolysis rate of YM isolated from WT strain BY4743 pretreated with DMSO (black bar) or with the indicated concentrations of PGO (red bars). Values have been subtracted for the oligomycin-insensitive fraction and expressed as nmol of ATP/mg/min. Results are mean Ϯ S.E. of at least three independent experiments. D, oxygen consumption rate of YM isolated from WT strain BY4743 pretreated with DMSO (black bar) or with the indicated concentrations of PGO (red bars) in the presence of NADH and ADP. Values refer to the ratio of OCR (ADP) and OCR (NADH) and are expressed as mean Ϯ S.E. of three independent experiments. E, isolated MLM pretreated with the concentrations of PGO labeling each lane were separated by BN-PAGE as described under "Experimental procedures" and revealed with Coomassie Blue and in-gel activity staining for ATP hydrolysis to identify dimers (D), monomers (M), and F 1 sector of F-ATP synthase. F, dimers and monomers were cut out from BN-PAGE gels and subjected to SDS-PAGE and silver-stained. All sets of lanes in E and F are from the same gels.

Subunit g Arg-107 of F-ATPase regulates PTP Expression of human subunit g in yeast prevents PGO-dependent desensitization of PTP
PGO desensitizes the yeast PTP to Ca 2ϩ (thus increasing the CRC), whereas it sensitizes it in human HEK293 cells (thus decreasing the CRC; see Fig. 1). We exploited this phenotypic difference to test whether expression of human subunit g is able to switch the inhibitory effect of PGO in yeast mitochondria to the inducing effect seen in human mitochondria. Deletion of the ATP20 gene encoding for yeast subunit g prevented dimerization of F-ATP synthase, which could be rescued by both expression of ATP20 and of ATP5L (which encodes human subunit g) (Fig. 8A). As for WT yeast (compare with Fig.  1), PGO increased the CRC of ⌬ATP20 mitochondria reexpressing ATP20, whereas replacement with ATP5L totally prevented the desensitizing effects of PGO, which rather decreased the CRC (Fig. 8B), as also seen in permeabilized HEK293 cells. These results suggest that subunit g confers species specificity to the effects of PGO on the PTP.

Arg-107 of subunit g is a target for modification by high glucose, phenylglyoxal, and methylglyoxal
Advanced glycation end-products (AGE) originate from nonenzymatic modification of lysine and arginine residues that takes place after long-term exposure to high concentrations of glucose or to reactive compounds like methylglyoxal (MGO), a byproduct of glycolysis that alters protein structure, stability, and function (26 -28). Yeast cells grown at high glucose concentration undergo protein glycation largely due to generation of high levels of MGO (29), with a relevant increase in argpyrimidine-modified proteins (30,31). Arginine residues are frequently found in enzyme sites involved in substrate binding and catalysis, and their modification by MGO results in the formation of AGE (28). Consistent with these toxic effects, the growth rate of yeast cells was slowed down in high-glucose medium (Fig. 9A). At normal concentrations of glucose, toxicity was mimicked by PGO and MGO (Fig. 9A). Remarkably, yeast strains lacking subunit g or harboring the subunit g R107A mutation were significantly more resistant to PGO toxicity (Fig.  9B).

Discussion
The molecular nature of the PTP is a long-standing mystery of mitochondrial biology. In the past, we have used relatively selective sulfhydryl, histidine, and arginine reagents to define the properties of discrete sites conferring PTP regulation by the proton electrochemical gradient, divalent cations, quinones, and oxidative stress (8). The recent hypothesis that the PTP may originate from a specific, Ca 2ϩ -induced conformation of F-ATP synthase (16 -19) provides a unique opportunity to identify these sites in molecular terms. Indeed, mutagenesis of specific residues or deletion of specific subunits of F-ATP synthase should modify predictably the properties of the PTP and/or the effect of reagents acting at these sites. This is a formidable challenge because F-ATP synthase is a complex multisubunit enzyme (32-37) whose assembly is affected by deletion of specific subunits (23,24). The present study is part of our , and ATP20 subunits. B, serial dilution spotting assay of the indicated genotypes. Cell suspensions were grown to A 600 nm ϭ 1; serially diluted; plated on medium containing 2% glucose (Glu), 2% Glu drop-out uracil (Ϫuracil), and 2% glycerol; and incubated at 30°C. Pictures were taken after 1-2 days. C, yeast mitochondrial proteins of the indicated genotypes were separated by BN-PAGE as described under "Experimental procedures," and revealed with Coomassie Blue and in-gel activity staining for ATP hydrolysis to identify dimers (D), monomers (M), and F 1 sector of F-ATP synthase. D, CRC of yeast mitochondria from ⌬ATP5 ϩ yOSCP (black bars) and ⌬ATP5 ϩ yhOSCP (red bars) pretreated with the indicated concentrations of PGO. Data are expressed as CRC/CRC 0 ratio, in which CRC 0 refers to the CRC of DMSO-treated mitochondria. Results are mean Ϯ S.E. (error bars) of at least three independent experiments. All sets of lanes in A and C are from the same gels.

Subunit g Arg-107 of F-ATPase regulates PTP
efforts at identifying such regulatory sites and, as a corollary, at (dis)proving the general hypothesis that the PTP originates from F-ATP synthase. So far, we have provided evidence (i) that Ca 2ϩ triggers the PTP by substituting Mg 2ϩ at the catalytic site, causing a conformational change that is transmitted through OSCP to the lateral stalk and eventually to the inner membrane (21), and (ii) that PTP blockade by matrix H ϩ is mediated by the unique, highly conserved histidine residue of OSCP subunit (22). Here, we have been able to identify a unique arginine residue that mediates the effects of PGO on the Ca 2ϩ sensitivity of the PTP in yeast mitochondria.
Our previous studies had defined the PTP-modulatory effects of arginine-selective reagents like glyoxals and 2,3-butanedione (12)(13)(14)(15). Although glyoxals can also react with other residues (such as lysines and cysteines) (38), it has been possible to trace their effects on the PT to modification of arginines with reasonable certainty (12)(13)(14)(15). In rat liver mitochondria, the outcome depended on the specific glyoxal used, with PTP inhibition by MGO or PGO and PTP activation by OH-PGO (12)(13)(14)(15). Studies with a variety of compounds with similar reactivity but different overall physical-chemical properties led to the conclusion that glyoxals react with the same arginine residue(s) and

Subunit g Arg-107 of F-ATPase regulates PTP
that the main determinants for the effects on the PTP are charge and hydrogen-bonding capacity of the adducts (12)(13)(14)(15). Here, we have added the very relevant piece of information that PGO has species-specific effects on the PTP. Having identified Arg-107 of subunit g as the unique target of PGO in yeast, we suspect that the species-specific effects may depend on differences in the primary structure of F-ATP synthase, which appear to be particularly prominent in subunits e and g (Fig. 4).
A large fraction of the 60 conserved arginines of F-ATP synthase is located in the catalytic sector (26 in subunit ␣ and 17 in subunit ␤) and in the F 0 subunits ␥, ⑀, c, and a (11 residues). The membrane extrinsic region (F 1 sector) of the F-ATP synthase complex is composed of 3␣ and 3␤ and of one each of the ␥, ␦, and ⑀ subunits. The 3␣3␤ hexamer contains the catalytic sites responsible for ATP synthesis or hydrolysis (32) and associates through the ␥, ␦, and ⑀ subunits with a ring of hydrophobic c subunits (the c-ring) in the membrane domain. The rotating c-ring is in contact with the static a subunit, which maintains a specific pathway for translocation of protons through the membrane (34). A conserved arginine residue of subunit a plays a critical role in blocking a futile proton shortcut and in facilitating proton transfer at the a/c-ring interface via a carboxylic group in the c-ring (39). Our data do not reveal whether the arginine residues of the catalytic sector react with glyoxals, but they do indicate that, if present, these adducts do not play a direct role in PTP modulation.
OSCP plays a critical role in the assembly and function of F-ATP synthase (40,41) and has a well-characterized regulatory role in PTP modulation in mammals by providing a site for CyPD binding, which results in sensitization of the PTP to Ca 2ϩ (17,42,43). Replacement of yeast with human OSCP did not reverse PTP sensitization to Ca 2ϩ by PGO, however, suggesting that also OSCP arginines are not critical for PTP modulation by glyoxals.
F-ATP synthase complexes associate into dimers, which are essential for the formation of oligomers and generation of the cristae (44 -46). Subunits e and g are essential for this process, and deletion of either subunit leads to dimer and cristae disruption (46 -52). The e and g subunits are also important for regulation of the Ca 2ϩ sensitivity of yeast PTP, which undergoes desensitization after their genetic ablation (19). The fact that their contribution has been so far addressed in null mutants inevitably limits our understanding of their putative function in PTP formation. Conversely, site-directed mutagenesis of specific residues that does not alter assembly, subunit composition, and enzymatic activity of the enzyme allows interrogation of their role in PTP formation and regulation.
The yeast R107A mutant of subunit g described here shows F-ATP synthase dimers with normal hydrolytic activity, yet it displays a PTP that is desensitized to Ca 2ϩ (data not shown) and insensitive to PGO and a decreased swelling rate in sucrose. This last finding suggests that Arg-107 of subunit g may be

Subunit g Arg-107 of F-ATPase regulates PTP
important for solute permeation through the PTP, providing a further clue as to a possible site where the pore forms. Of note, replacement of yeast subunit g with its human counterpart confers the "human" phenotype to yeast, suggesting that the species-specific effect of PGO is more likely to depend on structural differences between the yeast and human F-ATP synthase than on the Arg-107-PGO adduct as such.
In mammals, dysregulated PTP opening is a primary cause of cell death initiated via matrix swelling and rupture of the outer membrane followed by release of proapoptotic factors (cytochrome c), whereas its inhibition appears often to be protective (8). Whether the pore plays a similar role in yeast is still a matter of debate. Yeast does not possess a mitochondrial Ca 2ϩ uniporter to mediate rapid Ca 2ϩ overload; nor does it form the apoptosome complex required for the activation of effector caspases. However, yeast mitochondria can undergo membrane permeabilization in response to Ca 2ϩ and oxidative stress (53,54), leading to the release of cytochrome c, which certainly causes decreased respiration and ATP production. These events, together with the activation of the yeast metacaspase Ycalp, are proximal events in yeast programmed cell death (54). Thus, as for mammals, PTP opening might represent an important cell death pathway also for yeast, which can undergo apoptosis and aging (55).
AGE arising from carbonyl stress contribute to many pathological states, such as diabetes and age-related diseases (26 -28). It is apparent that adducts form at many sites other than subunit g, yet our findings suggest that altered PTP features may play a role in glucose-dependent toxicity also in human cells, as was already suggested (56). Whether additional arginine residues mediate the glyoxal-specific effects described in rat mitochondria and human cells (12)(13)(14)(15) remains to be established. However, the demonstration that a single conserved arginine residue of subunit g mediates the effects of PGO on the PTP in yeast is another step toward a molecular understanding of PTP regulation and evidence that the latter forms from a specific conformation of the F-ATP synthase.

Reagents and cells
PGO hydrate, MGO, oligomycin, rotenone, succinic acid, pyruvate, alamethicin, ATP, ADP, EGTA, CaCl 2 , digitonin, NADH, ETH129, galactose, and sucrose were from Sigma (Milan, Italy). Cyclosporin A was purchased from Calbiochem. Ca 2ϩ Green-5N was from Invitrogen (Milan, Italy). HEK293T were obtained from the American Tissue Culture Collection (ATCC) and cultured in Dulbecco's modified Eagle's medium (Lonza, Basel, Switzerland) supplemented with fetal bovine serum (10%) and penicillin and streptomycin (1%) (Thermo Fisher Scientific). Drosophila Schneider S2 cells were obtained from Dr. Michael Forte (Oregon Health and Science University) and cultured as described (57). The Saccharomyces cerevisiae strains BY4743, ⌬TIM11, and ⌬ATP20 were purchased from Thermo Scientific, and mitochondria were isolated as described (19). Antibodies for TIM11, ATP20, and ␥ subunits Figure 6. The ATP20 R107A mutation does not affect dimerization, assembly, and function of F-ATP synthase. A, serial dilution spotting assay of BY4743, ⌬TIM11, ⌬ATP20, ATP20 WT, ATP20 R107A strains. Cell suspensions were grown to A 600 nm ϭ 1; serially diluted; plated on medium containing 2% Glu, 2% Glu drop-out uracil (Ϫuracil), and 2% glycerol; and incubated at 30°C. Pictures were taken after 1-3 days. B, yeast mitochondrial proteins of the indicated genotypes were separated by BN-PAGE as described under "Experimental procedures" and revealed with Coomassie Blue and in-gel activity staining for ATP hydrolysis to identify dimers (D), monomers (M), and the F 1 sector of F-ATP synthase. C, dimers and monomers were cut out from BN-PAGE, subjected to two-dimensional SDS-PAGE, and revealed by silver staining. All sets of lanes in B and C are from the same gels.

Subunit g Arg-107 of F-ATPase regulates PTP
were a kind gift of Dr. Marie-France Giraud (University of Bordeaux, France). Antibody for the OSCP subunit was purchased from Santa Cruz Biotechnology, Inc. (sc-365162).

Yeast mutant generation
To generate ⌬ATP5 yeast expressing the human or yeast OSCP (ATP5) sequence, the ATP5 gene was first deleted in W303 diploids by substituting the genomic sequence with the Kan cassette of pFA6a-KanMX4 vector. Then diploids were allowed to sporulate in 2% potassium acetate medium, resulting tetrads were dissected, and correct clones were checked by semiquantitative PCR using appropriate primers. Yeast ATP5 gene (yOSCP) was cloned in pFL38-URA vector after amplification using the following primers that include upstream and downstream regulatory sites (endogenous promoter and termination sequence): 5Ј-CACGACGTTGTAAAACGACGGCC-AGTGAATTCGAGCTCGGTACCCTGCCGTCGTCATAA-AGTGGAC (forward) and 5Ј-TTACGCCAAGCTTGCATGC-CTGCAGGTCGACTCTAGAGGATCCCCGTTTGCCTGG-ATACACGAAC (reverse). Cloned products were confirmed by sequence analysis. Human OSCP (hOSCP) cDNA (NM_ 001697) was purchased from GenScript and amplified using the following primers: 5Ј-CAGTGTGCTGGAATTCAACACA-ATGGCTGCCCCAGCAGTGTCC (forward) and 5Ј-GATA-TCTGCAGAATTTTAGACAATCTCCCGCATAGCCCTG (reverse). To substitute with the yeast putative mitochondrial import sequence at the N terminus of the OSCP gene (yhOSCP), we used the forward primer 5Ј-CAGTGTGCTGG-AATTCAACACAATGTTTAATAGAGTCTTTACCAGGT-CATTTGCATCAAGCTTAAGAGCTGCTGCTTCCAAAG-CTGCTGCTCCCCCTCCTGTTCAGGTATACGGTATTG and the same reverse primer as for hOSCP. PCR products were cloned in pYES2-URA previously digested with EcoRI-HF using the In-Fusion HD cloning kit (Clontech), and resulting  Fig. 1B except that Ca 2ϩ green-5N was omitted. A and B, PTP-dependent mitochondrial swelling was evaluated as the decrease of absorbance at 540 nm, and the fraction of mitochondria that underwent the PT was assessed as described by Petronilli et al. (63). Additions of 50 M Ca 2ϩ and 5 M alamethicin are indicated by arrows. Ca 2ϩ was omitted in the light blue traces. Mitochondria isolated from ATP20 WT (A) and ATP20 R107A yeast (B) were treated with DMSO (black traces) or 4 mM PGO in DMSO (red traces). Representative traces of at least three independent experiments are shown. C, rate of mitochondrial swelling triggered by 50 M Ca 2ϩ . ATP20 WT or ATP20 R107A mitochondria were pretreated with DMSO (black bars) or 4 mM PGO (red bars). The rate of mitochondrial swelling is expressed as the fraction of swollen mitochondria/min. Data are mean Ϯ S.E. (error bars) of at least three independent experiments. *, p Ͻ 0.05 versus WT (ATP20 WT); ***, p Ͻ 0.001 versus vehicle. One-way ANOVA was used for statistical analysis. D, MALDI-TOF mass spectra of synthetic peptides of the amino acid segment comprising Arg-107. The mass spectra of the parent peptides are shown in the top windows, and that of the PGO-modified peptides is shown in the bottom window. Masses in the spectra are M ϩ H ϩ . The 116-Da adducts are marked with arrows.
The QuikChange Lightning site-directed mutagenesis kit (Agilent) was used to harvest the mutants, and the dsDNA templates we used were TIM11 and ATP20 WT genes created in pFL38 plasmid. After the mutations were confirmed by sequencing, a one-step transformation protocol (58) was used to transform the plasmid into ⌬TIM11 or ⌬ATP20 S. cerevisiae strain and plated on the selective medium 2% Glu (drop-out uracil) and incubated in 30°C for 2 days.

Mouse liver mitochondria preparation
All of the isolation procedures were performed at 4°C or on ice. Mouse livers were cut into pieces in the isolation buffer containing 250 mM sucrose, 10 mM Tris-HCl, 0.1 mM EGTA, pH 7.4. The homogenate suspension was centrifuged at 690 ϫ g for 10 min. The supernatant was collected and centrifuged at 7000 ϫ g for 10 min. The pellet was suspended in the isolation buffer and centrifuged at 9400 ϫ g for 5 min. The mitochondria pellet was resuspended in about 400 l of isolation buffer. The biuret test was used for determining protein concentration.

Yeast mitochondria isolation
Yeast mitochondria were isolated as described previously (19). Briefly, yeast cells were cultured in 400 ml of medium containing 1% yeast extract, 1% bacto-polypeptone, and 2% galactose overnight at 30°C. Yeast cells were harvested by centrifugation at 2000 ϫ g for 5 min at room temperature. The cell pellet was incubated in 0.1 M Tris-SO 4 , pH 9.4, buffer, supplemented with 10 mM DTT for 15 min at 37°C. The incubation was terminated by centrifugation at 2000 ϫ g for 5 min and washed once with 1.2 M sorbitol, 20 mM P i , pH 7.4, buffer. The cell pellet was resuspended in the above sorbitol buffer supplemented with zymolyase (0.4 mg/g pellet) and then incubated for 45 min at 30°C. The incubation was terminated by centrifugation at 2000 ϫ g for 5 min at 4°C and washed once with sorbitol buffer. The cell pellet was resuspended in cold buffer containing 0.6 M mannitol, 10 mM Tris-HCl, 0.1 mM EDTA, pH 7.4, and homogenized with a Potter homogenizer. The supernatant was collected and centrifuged at 2000 ϫ g for 5 min at 4°C, and then the supernatant was centrifuged at 12,000 ϫ g for 10 min at 4°C. The mitochondrial pellet was harvested in mannitol medium. Protein concentration was quantified by the A 280 of 0.6% SDS-solubilized mitochondria.

Cell permeabilization
The culture medium of human HEK293 cells was discarded, and cells were washed once with Dulbecco's PBS. Cells were

Subunit g Arg-107 of F-ATPase regulates PTP
treated with 0.25% trypsin, and the reaction was terminated by the addition of culture medium. Drosophila S 2 R ϩ cells were detached with a scraper. Cell suspensions were centrifuged at 500 ϫ g for 5 min and washed once with medium containing 130 mM KCl, 10 mM MOPS-Tris, 1 mM P i , 0.1 mM EGTA, pH 7.4. The cell pellet was resuspended in a medium containing 130 mM KCl, 10 mM MOPS-Tris, 1 mM P i , 1 mM EGTA, pH 7.4, and supplemented with 100 M digitonin and incubated for 10 min on ice at a density of 2 ϫ 10 7 cells/ml (HEK293 cells) or supplemented with 150 M digitonin and incubated on ice for 20 min at a density of 6 ϫ 10 7 cells/ml (Drosophila S 2 R ϩ cells). The incubations were terminated by dilution with 15-20 ml of medium containing 130 mM KCl, 10 mM MOPS-Tris, 1 mM P i , 0.1 mM EGTA, pH 7.4, followed by centrifugation at 500 ϫ g for 5 min at 4°C. The cell pellet was washed once with 5 ml of the above KCl medium at 4°C. The final pellet was resuspended in 250 mM sucrose, 20 M EGTA, 10 mM HEPES-KOH, pH 8.0, for reaction with PGO.

Chemical modification with PGO
Freshly isolated mouse liver or yeast mitochondria (1 mg/ml) or 2 ϫ 10 7 /ml permeabilized cells were incubated with PGO in 250 mM sucrose, 20 M EGTA, 10 mM HEPES-KOH, pH 8.0, for 15 min at 25°C. The reaction was terminated by decreasing the pH to 6.8 with HEPES and cooling to 4°C. Mitochondria were precipitated by centrifugation at 8000 ϫ g for 6 min and washed once with 250 mM sucrose, 0.1 mM EGTA, 0.5 mg/ml BSA, 10 mM Tris-HCl, pH 7.4. Mitochondria or cells were finally resuspended in the assay medium.

Mitochondrial swelling assay
Isolated yeast mitochondria were suspended in 250 mM sucrose, 10 mM Tris-MOPS, 20 M EGTA, 1 mM NADH, 5 M ETH129, 0.5 mg/ml BSA, 2 mM P i , pH 7.4, at a concentration of 0.5 mg of protein/ml in a final volume of 2 ml. Mitochondrial swelling was measured as the decrease of the turbidity (apparent absorbance at 540 nm) using a Cary 100 UV-visible spectrophotometer from Agilent Technologies.

Western blotting
Isolated mitochondria were lysed in 150 mM NaCl, 50 mM Tris-HCl, pH 7.4, 2 mM EDTA, 2 mM EGTA, 1% Triton X-100, 10 mM ␤-mercaptoethanol, 10% glycerol, 0.3% Nonidet P-40, 0.04% bromphenol blue, supplemented with protease inhibitor mixture (Sigma) at 2.5 g of protein/l and boiled for 5 min. The above lysates were loaded onto NuPAGE TM 12% BisTris protein gels (Invitrogen), and proteins were separated by electrophoresis in NuPAGE TM MOPS SDS running buffer (Invitrogen) for 3 h at 20 mA at 4°C. Resolved proteins were transferred to nitrocellulose membranes in NuPAGE TM transfer buffer for 1.25 h at 30 V at 4°C, followed by membrane blocking with 5% (w/v) dry milk. The membrane was incubated overnight with the antibodies for TIM11, ATP20, and ␥ subunits overnight at 4°C. Immunoreactive bands were detected by chemiluminescence using an Uvitec Cambridge instrument.

Blue native gel electrophoresis
Isolated yeast mitochondria were suspended in 150 mM potassium acetate, 30 mM HEPES, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, supplemented with 1.5% (w/v) digitonin, pH 7.4, at 10 mg/ml, followed by centrifugation at 100,000 ϫ g for 25 min at 4°C with a Beckman TL-100 rotor. Supernatants were collected and supplemented with native PAGE 5% G-250 sample additive (Invitrogen) and then rapidly applied onto NativePAGE TM Novex TM 3-12% BisTris protein gels (Invitrogen). Electrophoresis was carried out in the Dark Blue cathode buffer at 150 V for 20 min and in the Light Blue cathode buffer at 250 V for 2 h. After electrophoresis, gels were stained with Coomassie Blue or used for in-gel activity staining of F-ATP synthase as described previously (19).

ATP hydrolysis assay
MLM or YM pretreated by PGO were suspended in 30 mM sucrose, 50 mM KCl, 50 mM Tris-HCl, 2 mM EGTA, 4 mM MgCl 2 and 2 mM phosphoenolpyruvate, supplemented with 4 units/ml pyruvate kinase, 3 units/ml lactate dehydrogenase, 2 mM ATP, and 0.3 mM NADH. Treatment with 4 M oligomycin was performed at 37°C for 15 min. The assay was carried out at 37°C at a protein concentration of 40 g ϫ ml Ϫ1 in a final volume of 0.2 ml. ATP synthesis was measured as the decrease of absorbance of NADH at 340 nm using an Infinite 200 PRO multimode microplate reader.

Oxygen consumption rate
A Clark oxygen electrode was used to detect the oxygen consumption rate of isolated mitochondria. MLM were incubated in 250 mM sucrose, 10 mM Tris-MOPS, 10 M EGTA, pH 7.4, at a final concentration of 0.5 mg of protein/ml in a final volume of 2 ml. Further additions were 5 mM succinate as substrate, 1 mM P i , and 0.2 mM ADP. YM were incubated in 250 mM sucrose, 10 mM Tris-MOPS, 20 M EGTA, 5 mM P i , 0.5 mg/ml BSA, pH 7.4, at a final concentration of 0.25 mg of protein/ml in a final volume of 2 ml. Further additions were 1 mM NADH, 0.2 mM ADP, and 4 M oligomycin.

Mass spectrometry of g subunit peptides
Peptides (amino acids 88 -114 of subunit g) modeled on the native sequence or containing alanine at position 106 were purchased from Caslo ApS (Kongens Lyngby, Denmark). Peptides were dissolved at a concentration of 100 M in 10 mM Hepes-K ϩ , pH 8.0, and allowed to react with 2 mM PGO at room temperature for 60 min. The reaction was stopped by the addition of 0.5% TFA. The peptides were separated from the reaction mixture by reverse phase chromatography using C18 ZipTips (Millipore) pre-equilibrated with 0.5% TFA. Bound peptides were eluted with 10 l of 60% acetonitrile and 0.5% TFA. The eluate was concentrated to 1 l under a stream of nitrogen, mixed with 1 l of a saturated solution of ␣-cyano-4-hydroxycinnamic acid, and applied to the MALDI target plate. MALDI-TOF was performed on a Bruker Ultraflextreme (Bremen, Germany) mass spectrometer operating in positive reflectron mode. Calibration of the instrument was performed using peptides of known mass.

Secondary structure assignment
This was performed pairing predictions from PSIPRED (59) and FELLS (60), whereas transmembrane regions were predicted with TMHMM (61). Default parameters were selected. The three-dimensional structure was visualized with Chimera (62).