Assessing Actual Contribution of IF1, Inhibitor of Mitochondrial FoF1, to ATP Homeostasis, Cell Growth, Mitochondrial Morphology, and Cell Viability*

Background: IF1 inhibits ATPase activity of mitochondrial FoF1-ATP synthase. Results: Although IF1 alleviates ischemic injury, the cell can grow normally, manage to maintain ATP levels, and keep mitochondria morphology intact without IF1. Conclusion: IF1 helps ATP homeostasis, but activated glycolysis can cover deficiency of IF1. Significance: Integrated regulation of mitochondrial ATP synthesis is crucial for metabolic dynamism. FoF1-ATP synthase (FoF1) synthesizes ATP in mitochondria coupled with proton flow driven by the proton motive force (pmf) across membranes. It has been known that isolated IF1, an evolutionarily well conserved mitochondrial protein, can inhibit the ATP hydrolysis activity of FoF1. Here, we generated HeLa cells with permanent IF1 knockdown (IF1-KD cells) and compared their energy metabolism with control cells. Under optimum growth conditions, IF1-KD cells have lower cellular ATP levels and generate a higher pmf and more reactive oxygen species. Nonetheless, IF1-KD cells and control cells show the same rates of cell growth, glucose consumption, and mitochondrial ATP synthesis. Furthermore, contrary to previous reports, the morphology of mitochondria in IF1-KD cells appears to be normal. When cells encounter sudden dissipation of pmf, the cytoplasmic ATP level in IF1-KD cells drops immediately (∼1 min), whereas it remains unchanged in the control cells, indicating occurrence of futile ATP hydrolysis by FoF1 in the absence of IF1. The lowered ATP level in IF1-KD cells then recovers gradually (∼10 min) to the original level by consuming more glucose than control cells. The viability of IF1-KD cells and control cells is the same in the absence of pmf. Thus, IF1 contributes to ATP homeostasis, but its deficiency does not affect the growth and survival of HeLa cells. Only when cells are exposed to chemical ischemia (no glycolysis and no respiration) or high concentrations of reactive oxygen species does IF1 exhibit its ability to alleviate cell injury.

F o F 1 -ATP synthase (F o F 1 ) synthesizes ATP in mitochondria coupled with proton flow driven by the proton motive force (pmf) across membranes. It has been known that isolated IF1, an evolutionarily well conserved mitochondrial protein, can inhibit the ATP hydrolysis activity of F o F 1 . Here, we generated HeLa cells with permanent IF1 knockdown (IF1-KD cells) and compared their energy metabolism with control cells. Under optimum growth conditions, IF1-KD cells have lower cellular ATP levels and generate a higher pmf and more reactive oxygen species. Nonetheless, IF1-KD cells and control cells show the same rates of cell growth, glucose consumption, and mitochondrial ATP synthesis. Furthermore, contrary to previous reports, the morphology of mitochondria in IF1-KD cells appears to be normal. When cells encounter sudden dissipation of pmf, the cytoplasmic ATP level in IF1-KD cells drops immediately (ϳ1 min), whereas it remains unchanged in the control cells, indicating occurrence of futile ATP hydrolysis by F o F 1 in the absence of IF1. The lowered ATP level in IF1-KD cells then recovers gradually (ϳ10 min) to the original level by consuming more glucose than control cells. The viability of IF1-KD cells and control cells is the same in the absence of pmf. Thus, IF1 contributes to ATP homeostasis, but its deficiency does not affect the growth and survival of HeLa cells. Only when cells are exposed to chemical ischemia (no glycolysis and no respiration) or high concentrations of reactive oxygen species does IF1 exhibit its ability to alleviate cell injury.
F o F 1 -ATP synthase (F o F 1 ) 2 in mitochondria synthesizes ATP coupled with proton flow driven by proton motive force (pmf) across membranes, which is generated by the respiratory chain. Demand for ATP and supply of oxygen and oxidation substrates for the respiratory chain can vary in living cells, and activity of F o F 1 must be regulated to meet these situations. When pmf becomes insufficient for ATP synthesis due to ischemia or other reasons, ATP production becomes more dependent on glycolysis. At the same time, however, F o F 1 would start to catalyze the back reaction, ATP hydrolysis. Therefore, cells must somehow prevent this futile ATP hydrolysis of F o F 1 , and IF1 has been proposed to take this task. IF1 was isolated as a protein factor that showed inhibitory effects on the ATP hydrolysis activity of F o F 1 (1). It is a nuclear-coded mitochondrial protein and is conserved evolutionarily from yeast to mammals (2)(3)(4). IF1 forms a dimer at acidic pH (ϳ6.7) and exhibits inhibitory effects, but a tetramer form, which is formed at basic pH (ϳ8.0), cannot interact with F o F 1 (5,6). The crystal structure of the F 1 portion of F o F 1 with bound IF1 shows that the C-terminal ␣-helix of IF1 is inserted into the interface between the ␣and ␤-subunits of F 1 (7).
Recently, in vivo studies have revealed an important role of IF1 in energy metabolism in living cells. It was found that IF1 is overexpressed in various human cancer cells, and it has been proposed that one role of IF1 is as an additional molecular switch for the metabolic shift from oxidative phosphorylation to glycolysis in tumor cells (8). When cells are deprived of oxygen and glucose, IF1-suppressed cells lose cellular ATP and die more rapidly than control cells, suggesting a role of IF1 in maintaining ATP concentration ([ATP]) in energy crises (9). Several groups reported that IF1 facilitates dimer formation of F o F 1 in the inner mitochondrial membranes (10 -12) and contributes to cristae formation (9,(13)(14)(15)(16)(17)(18). This profound effect of IF1 on mitochondrial morphology remains controversial (19,20). Reports have indicated that overexpression of IF1 decreased the magnitude of pmf and increased the rate of respiration (9), whereas the opposite results have also been reported (8). Most of previous observations described above were made for the cells in which IF1 knockdown and overexpression are transient, and there remains ambiguity over whether the observed phenomena are only transient, during adaptation to IF1-deficient situations, and not inherent to cells already adapted. Transient knockdown also is inconvenient for observing cells over long periods of time, such as observing cell growth and cell viability. In this report, we generated HeLa cells in which IF1 expression is suppressed permanently (IF1-KD cells) and compared their energy metabolism and cell physiology with control cells. Under optimum growth conditions, even though cellular [ATP] in IF1-KD cells is low and reactive oxygen species (ROS) production is elevated, IF1-KD cells and control cells apparently show the same cell growth, glucose consumption, mitochondrial ATP synthesis, and mitochondria morphology. When pmf is lost, IF1-KD cells maintain cellular [ATP] after recovery from a transient drop by activated levels of glycolysis and show no defect in survival in the absence of pmf. Thus, IF1 contributes to [ATP] homeostasis, but cells can grow and survive without IF1 as long as glucose is available. When both glycolysis and respiration are blocked or high levels of ROS are produced, IF1 helps cells to endure longer.

MATERIALS AND METHODS
Cells and Mitochondria-HeLa cells (from Japanese Collection of Research Bioresources) were grown in DMEM (Nissui Pharmaceutical) supplemented with 10% fetal bovine serum (Invitrogen). For the glucose deprivation experiment, the dialyzed fetal bovine serum (Invitrogen) was supplemented into glucose-free DMEM (Cell Science & Technology Institute). Mitochondrial fractions were prepared as described (21).
Retroviral Gene Transduction-An shRNA-expressing retroviral vector of pSuper.retro.puro (Oligoengine) was used for gene knockdown. The target sequences for shRNA were predicted by an application of iRNAi (mekentosj freeware). The sequence for IF1-KD was 5Ј-GAGCACAGAGTAGAGA ACAACTGGCAGCT-3Ј. Preparation and transduction of retroviruses were carried out as described (22).
Electrophoresis-After lysis of the fractions and SDS-PAGE, the gel was blotted with Bjerrum buffer supplemented with 20% ethanol and 0.037% SDS by semi-dry blotter (Bio-Rad) at 12 V for 30 min at room temperature. Antibodies were as follows: anti-␣, anti-␤, anti-⑀, and anti-IF1 antibodies were purchased from Molecular Probes (A21350), Molecular Probes (A21351), Abnova (H00000514-M01), and MitoScience (MS506), respectively. For analysis of the native F o F 1 complex, clear native polyacrylamide gel electrophoresis was used. Mitochondrial fractions (1 g/l proteins) were suspended with a solubilization buffer (50 mM imidazole-HCl, pH 7.0, 50 mM NaCl, 2 mM 6-aminohexanoic acid, 1 mM EDTA, and 3 g/l of digitonin) at 4°C. The suspension was incubated on ice for 10 min and centrifuged at 100,000 ϫ g for 10 min at 4°C. The supernatant was mixed with 0.1 volume of a loading buffer (50% glycerol and 0.1% Ponceau S). After electrophoresis, the gel was immersed in a denaturation buffer (48 mM Tris base, 39 mM glycine, 20% ethanol, and 1% SDS) for 10 min at room temperature and was subjected to Western blotting.
Assays-Cellular [ATP] was measured as follows. Cells were passaged onto a six-well culture plate and cultivated. After incubation for a day, they were washed by ice-cold phosphatebuffered saline twice and by 500 l of 0.4 M perchloric acid. The plate was floated on liquid nitrogen, and cells were frozen. After thawing on ice, cells were scraped and centrifuged at 18,000 ϫ g for 10 min at 4°C. The supernatant (400 l) was mixed with 40 l of 4 M K 2 CO 3 , and cells were incubated on ice for 10 min. Finally, cells were centrifuged at 18,000 ϫ g for 10 min at 4°C, and the amount of ATP in the supernatant was measured by luciferase assay. The protein concentration of the lysate was also measured by BCA protein assay (Thermo Scientific). ATP synthesis was measured by MASC (Mitochondrial Activity of Storeptolysin O-permeabilized Cell) assay (22,23). Briefly, cells were passaged to a 96-well white plate and incubated overnight. The cells were permeabilized by streptolysin O, and their cytosolic components were washed out. Luciferase, luciferin, 0.1 mM ADP, and 10 mM P i were supplemented, and the reaction was started by addition of 10 mM succinic acid. Luciferin emission was monitored by a luminometer (Berthold Technologies). Glucose consumption by cells was measured as follows. Cells passaged at 7.5 ϫ 10 4 cells/well of a six-well plate were incubated for 2 days in normal DMEM or DMEM containing 15 M carbonyl cyanide 3-chlorophenylhydrazone (CCCP). Then, the culture medium was collected to measure the remaining glucose, and the cells were collected to count living cells by trypan blue exclusion assay. Concentration of medium glucose was measured by phenol sulfate method (24).
[ATP] Monitoring-Real-time monitoring of cytoplasmic [ATP] levels was carried out with ATeam (25). Cells were passaged at 7.7 ϫ 10 4 cells per 3.5-cm glass-bottomed dish within 2 ml of phenol red-free DMEM and were incubated overnight. The next day, 0.5 ml of the culture medium was collected, and 50 M CCCP, 10 g/ml oligomycin, and/or 10 mM 2-deoxyglucose were added as indicated. The dishes were set in a microchamber (Olympus MI-IBC-IF) on a fluorescent microscope (IX-81 Olympus) and were incubated for 10 min at 37°C in air containing 5% CO 2 . FRET signals were measured for 10 min, and then 0.5 ml of the collected medium was added into the dishes. The FRET signals were measured for another 60 min (one frame per min). The measurement was performed with a FRET filter set (XF88-2 Optoscinece), 20ϫ object immersion lens (UPLANSAPO 20XO Olympus; numerical aperture, 0.85), ND12 filter, and 250-ms exposure time, and the images were scanned by a cooled CCD camera (ORCA-R2 Hamamatsu Photonics). All of the data analysis was done by MetaMorph software (Molecular Devices).
Cell Images-To visualize mitochondria, cells were passaged at 7.7 ϫ 10 4 cells per 3.5-cm glass-bottomed dish and were incubated overnight. The dishes were set in a micro-chamber (Olympus MI-IBC-IF) on a fluorescent microscope (Olympus IX-81) at 37°C under 5% CO 2 in air and tetramethylrhodamine ethyl ester (TMRE) was added at 15 nM (final concentration). After a 20-min incubation, images of cells were recorded, and the fluorescent intensities were obtained. Three independent experiments were carried out, and TMRE intensities of Ͼ400 cells were quantitated. The p value between the IF1-KD and the control-treated cells is Ͻ0.01. To observe electron microscopic images, cells were grown on plastic dishes and fixed by 2% paraformaldehyde and 2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) at room temperature for 30 min. Next, cells were exposed into 2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) at 4°C overnight. The samples were post-fixed with 2% osmium tetroxide in 0.1 M phosphate buffer (pH 7.4) at 4°C for 1 h. They were anhydrated by ethanol and embedded in Quetol-812 resin at 60°C for 48 h. They were stained by 2% uranyl acetate at room temperature for 15 min and by lead staining solution at room temperature for 3 min. Electron microscopic images were shot by JEOL JM1200EX at 80 kV.
Cell Viability-In the viability test in CCCP, cells passaged at 1.5 ϫ 10 5 cells/well of a six-well plate were incubated for 2 days in normal DMEM or DMEM containing 15 M CCCP. Then, cells were collected, and living cells were counted by trypan blue exclusion assay. In the Paraquat tolerance assay, 5000 cells per well of a 96-well plate were passaged and were incubated overnight. The next day, the media were changed to those with/ without paraquat. After incubation for another day, the viable cells were measured by 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium assay (Cell Counting Kit-8; Dojindo).

RESULTS
Stable Knockdown of IF1-Stable IF1 knockdown HeLa cells (IF1-KD cells) were established by retroviral gene transduction. IF1-KD cells express only a small amount of IF1 in mitochondria, less than ϳ10% of control cells, without affecting expression of ␣-, ␤-, and ⑀-subunits of F o F 1 (Fig. 1A). The content of the F o F 1 complex in mitochondria in IF1-KD cells is also similar to that of those in control cells. Under the conditions we tested, the dimer of F o F 1 was hardly seen in the blue native gel electrophoresis in either IF1-KD cells or control cells (Fig. 1B). To avoid the possibility that the concentration of detergent (digitonin in our case) was too high to preserve the dimer form, we tested several concentrations of digitonin for solubilization of mitochondria, but the results were the same.  Fig. 2A). This implies that IF1 is working in control cells to block ATP hydrolysis activity of F o F 1 even under optimum growth conditions and that IF1-KD cells are hydrolyzing more ATP. More ATP hydrolysis by F o F 1 that accompanies the pumping out of protons from mitochondria should build up a higher pmf. Indeed, when the mitochondrial membrane potential (⌬), which constitutes pmf together with transmembrane ⌬pH, is estimated from TMRE images (see Fig.  3D), it is clear that IF1-KD cells generate a higher ⌬ than control cells (Fig. 2B). ROS levels in IF1-KD cells, assessed by dichlorofluorescein, a fluorescent probe for ROS, is two times higher than in control cells (Fig. 2C), probably as a result of higher ⌬.
Normal Growth, ATP Synthesis, and Mitochondria Morphology of IF1-KD Cells-Despite lower [ATP] and higher ROS production, IF1-KD cells can grow at the same rate as control cells (Fig. 3A). Then, the ability of mitochondria to synthesize ATP was compared by MASC assay (see "Materials and Methods") (22). The results show that mitochondria in IF1-KD cells can synthesize ATP at nearly the same rate as control cells (Fig. 3B), indicating that the ATP synthesis reaction catalyzed by F o F 1 is not affected by IF1. The observed ATP synthesis is catalyzed by F o F 1 but not by other metabolism because the synthesis is completely inhibited by oligomycin, a specific inhibitor of mitochondrial F o F 1 . Note that this result cannot exclude the possibility that IF1 plays a role in preventing ATP hydrolysis because the reaction mixtures of the above experiments contain ADP and P i but very small amounts of ATP. We extensively examined population, shape, and cristae structures of mitochondria in IF1-KD cells in electron micrograph and TMRE images, but we could not find significant differences from those in control cells and found them indistinguishable (Fig. 3, C and D). Also,  Decrease in [ATP] upon Loss of pmf in IF1-KD Cells-IF1 has been assumed to play a role in preventing ATP hydrolysis by F o F 1 under conditions where pmf is small or null. Next, to know whether it really happens in living cells, cytoplasmic [ATP] is monitored in real-time with a recently developed ATP-sensing probe, ATeam. ATeam is a protein consisting of three domains: yellow fluorescence protein, a bacterial protein capable of ATP binding, and cyan fluorescence protein. ATP binding to the middle domain induces changes of FRET efficiency between the two fluorescent domains (25). IF1-KD cells and control cells were infected with ATeam-expressing retrovirus, and their stable cell lines were established. Cells were incubated in a microchamber on a fluorescent microscope, and after 10 min, CCCP, an uncoupler that dissipates pmf, was added.
[ATP] in control cells was apparently unaffected by CCCP and remained constant. On the other hand, [ATP] in IF1-KD cells dropped upon addition of CCCP and recovered to its original level in 10 min (Fig. 4A). This drop is attributed to the ATP hydrolysis promoted by F o F 1 in the absence of pmf because it was not observed when oligomycin, a specific inhibitor of F o F 1 , was included in the solution (Fig. 4B). When the solution contained CCCP and 2-deoxyglucose, which blocks glycolysis, [ATP] in both IF1-KD cells and control cells dropped and never recovered (Fig. 4C). The FRET decrease in IF1-KD cells is roughly 1.5-fold faster than in control cells, giving an approximation that uninhibited F o F 1 would be responsible for one third of ATP consumption in IF1-KD cells. All of these results suggest that activated glycolysis is responsible for both the prevention of CCCP-induced drop of [ATP] in control cells and the recovery from CCCP-induced drop of [ATP] in IF1-KD cells.
Activated Glycolysis in IF1-KD Cells in Absence of pmf-Next, we measured glucose consumption in the absence and presence of CCCP. As shown (Fig. 5A), both control and IF1-KD cells consume similar amounts of glucose in the absence of CCCP. The consumption is increased for both cells when CCCP is present, but the extent of increase is much greater for IF1-KD cells than for control cells, reflecting that more glucose would be required to compensate wasteful ATP hydrolysis by uninhibited F o F 1 in IF1-KD cells. Given that [ATP] is maintained in IF1-KD cells, IF1-KD cells would be expected to survive well in the presence of CCCP. This is indeed the case; both control and IF1-KD cells hardly die after a 24-h incubation in CCCP (Fig. 5B). Thus, even if dissipation of pmf of mitochondria in IF1-KD cells induces a transient drop of [ATP] and an increase in glucose consumption, IF1-KD cells can still endure pmf-null situations just as well as control cells can.
Death of IF1-KD Cells under Fatal Conditions-The effect of IF1 on cell viability is evident only under fatal conditions where even control cells are dying. When all the energy sources for cells are deprived, that is, glycolysis is blocked by 2-deoxyglucose and respiration by potassium cyanide (KCN), cells cannot make any ATP and start to die. The time course of progression of cell death reveals that IF1-KD cells die at a faster rate than the control cells (Fig. 6A). The difference is not drastic but is still significant. Thus, IF1, in this situation mimicking ischemia, helps cells to survive longer. This observation is consistent with the results of transient IF1 knockdown as reported (9). Another fatal condition is inclusion of paraquat, a poison that catalyzes efficient ROS production in cells. Again, IF1-KD cells are more vulnerable to paraquat (Fig. 6B). Higher background ROS levels in IF1-KD cells (Fig. 2C) may contribute to the higher susceptibility to paraquat.

DISCUSSION
Cells Grow Normally without IF1-Taking advantage of HeLa cells with permanent IF1 knockdown, we show that IF1deficient cells can grow at the same rate under optimum growth conditions as control cells. Low cellular [ATP] and high ROS production in IF1-KD cells do not hamper cell growth. It was reported that a mutant Saccharomyces cerevisiae that lacks IF1 can grow at the same rate as wild-type cells (26), suggesting that normal cell growth in the absence of IF1 is common to yeast and mammals. In other words, IF1 is not essential for normal cell growth.
IF1 Does Not Affect Mitochondria Morphology-Previous studies reported that knockdown of IF1 caused stimulation of autophagy, decrease in mitochondrial volume fraction in the cell, and decrease in cristae density in mitochondria (10,11,13,14). However, none of these were observed for IF1-KD cells with certainty in our study. We cultured stable IF1-KD cells for more than one year, but no obvious change of morphology in cellular organelles, including mitochondria, was noticed throughout this period. One possible reason for the different results is that previous works used transient knockdown via RNAi, whereas we used stable, permanent knockdown cells established by retrovirus infection. If the morphological changes in mitochondria are transient ones to respond to new situations where IF1 is no longer available, they would escape our observation. Our observations of mitochondria in IF1-KD HeLa cells are consistent with our recent findings that IF1 knock-out mice have mitochondria with apparently normal morphology. 3 In addition, we observed that the amount of F o F 1 dimer is the same between control and IF1-KD cells. As reported (19,20,27), our observation does not support the indication that IF1 facilitates formation of F o F 1 dimer, which then helps cristae formation (9).
IF1 Is Rapidly Recruited upon pmf Loss-The real-time monitoring of cellular [ATP] by a newly developed ATeam unveiled how HeLa cells maintain [ATP] when they are exposed to hypoxic conditions that cause loss of pmf in mitochondria. Upon loss of pmf, [ATP] in IF1-KD cells drops sharply, whereas [ATP] in control cells is apparently unchanged (Fig. 4A). This [ATP] drop in IF1-KD cells is not observed when oligomycin is present (Fig. 4B), ensuring that ATP hydrolysis activity of uninhibited F o F 1 without available IF1 is responsible for the drop. In the case of control cells, this means that IF1 in mitochondria immediately prevents F o F 1 from performing ATP hydrolysis, which otherwise could occur under conditions where ATP is ample and pmf is poor, and maintains [ATP] without drop. Thus, the rapid action of IF1 during cellular energy crises (e.g. ischemia) is demonstrated for the first time.

IF1-deficient Cells Can Maintain [ATP] after pmf Loss by Glycolysis-[ATP]
in IF1-KD cells recovers from the transient drop to original levels 10 min after the pmf loss ( Fig. 4A) even though hydrolysis of ATP by uninhibited F o F 1 continues. This should be due to elevated glycolysis because glucose consumption by IF1-KD cells increases nearly three times when pmf is lost (Fig. 4D). Similarly, glycolysis in the control cells would be elevated immediately after the pmf loss to produce more ATP to compensate for the loss of the ATP supply from mitochondria. It is not known what signal induces activation of glycolysis in IF1-KD and control cells. It has been well established that key enzymes in the glycolytic pathway can be allosterically activated when [ATP] drops (28). However, control HeLa cells maintain [ATP] without a transient [ATP] drop. We point out an intriguing possibility that the loss of pmf itself triggers activation of glycolysis through a mechanism that is yet unknown.