Altered Lipid Synthesis by Lack of Yeast Pah1 Phosphatidate Phosphatase Reduces Chronological Life Span*

Background: Pah1 phosphatidate phosphatase regulates the synthesis of triacylglycerol and membrane phospholipids. Results: The lack of Pah1 causes increases in phospholipid synthesis, ATP consumption, and oxidative stress and a decrease in chronological life span. Conclusion: Pah1 plays a role in chronological life span by regulating lipid synthesis. Significance: Pah1-mediated regulation of triacylglycerol and phospholipid syntheses is important for chronological life span. In Saccharomyces cerevisiae, Pah1 phosphatidate phosphatase, which catalyzes the dephosphorylation of phosphatidate to yield diacylglycerol, plays a crucial role in the synthesis of the storage lipid triacylglycerol. This evolutionarily conserved enzyme also plays a negative regulatory role in controlling de novo membrane phospholipid synthesis through its consumption of phosphatidate. We found that the pah1Δ mutant was defective in the utilization of non-fermentable carbon sources but not in oxidative phosphorylation; the mutant did not exhibit major changes in oxygen consumption rate, mitochondrial membrane potential, F1F0-ATP synthase activity, or gross mitochondrial morphology. The pah1Δ mutant contained an almost normal complement of major mitochondrial phospholipids with some alterations in molecular species. Although oxidative phosphorylation was not compromised in the pah1Δ mutant, the cellular levels of ATP in quiescent cells were reduced by 2-fold, inversely correlating with a 4-fold increase in membrane phospholipids. In addition, the quiescent pah1Δ mutant cells had 3-fold higher levels of mitochondrial superoxide and cellular lipid hydroperoxides, had reduced activities of superoxide dismutase 2 and catalase, and were hypersensitive to hydrogen peroxide. Consequently, the pah1Δ mutant had a shortened chronological life span. In addition, the loss of Tsa1 thioredoxin peroxidase caused a synthetic growth defect with the pah1Δ mutation. The shortened chronological life span of the pah1Δ mutant along with its growth defect on non-fermentable carbon sources and hypersensitivity to hydrogen peroxide was suppressed by the loss of Dgk1 diacylglycerol kinase, indicating that the underpinning of pah1Δ mutant defects was the excess synthesis of membrane phospholipids.

In Saccharomyces cerevisiae, Pah1 phosphatidate phosphatase, which catalyzes the dephosphorylation of phosphatidate to yield diacylglycerol, plays a crucial role in the synthesis of the storage lipid triacylglycerol. This evolutionarily conserved enzyme also plays a negative regulatory role in controlling de novo membrane phospholipid synthesis through its consumption of phosphatidate. We found that the pah1⌬ mutant was defective in the utilization of non-fermentable carbon sources but not in oxidative phosphorylation; the mutant did not exhibit major changes in oxygen consumption rate, mitochondrial membrane potential, F 1 F 0 -ATP synthase activity, or gross mitochondrial morphology. The pah1⌬ mutant contained an almost normal complement of major mitochondrial phospholipids with some alterations in molecular species. Although oxidative phosphorylation was not compromised in the pah1⌬ mutant, the cellular levels of ATP in quiescent cells were reduced by 2-fold, inversely correlating with a 4-fold increase in membrane phospholipids. In addition, the quiescent pah1⌬ mutant cells had 3-fold higher levels of mitochondrial superoxide and cellular lipid hydroperoxides, had reduced activities of superoxide dismutase 2 and catalase, and were hypersensitive to hydrogen peroxide. Consequently, the pah1⌬ mutant had a shortened chronological life span. In addition, the loss of Tsa1 thioredoxin peroxidase caused a synthetic growth defect with the pah1⌬ mutation. The shortened chronological life span of the pah1⌬ mutant along with its growth defect on non-fermentable carbon sources and hypersensitivity to hydrogen peroxide was suppressed by the loss of Dgk1 diacylglycerol kinase, indicating that the underpinning of pah1⌬ mutant defects was the excess synthesis of membrane phospholipids.
The synthesis of lipids is an energy-consuming process that cells engage in throughout their growth (1)(2)(3). In logarithmically growing yeast, 2 a diverse set of membrane lipids (e.g. PS 3 and its derivatives PE and PC, PI and its derivatives phosphoinositides and sphingolipids, and phosphatidylglycerophosphate and its derivatives PG and CL) are synthesized from the precursor PA via the liponucleotide intermediate CDP-DAG ( Fig. 1) (2,3). As cells exhaust nutrients and progress into the stationary phase (e.g. quiescence), PA is channeled to DAG for the synthesis of TAG ( Fig. 1) (4 -6). Under certain growth conditions (e.g. supplementation with choline or ethanolamine), the DAG derived from PA is also used for the synthesis of PC or PE by way of the CDP-choline or CDP-ethanolamine branches, respectively, of the Kennedy pathway ( Fig. 1) (3). Quiescent cells may resume logarithmic growth if supplemented with fresh medium, and the TAG synthesized during the transition to stationary phase is mobilized to DAG and free fatty acid that are converted back to PA for conversion to CDP-DAG and the synthesis of membrane phospholipids ( Fig. 1) (2,3,(7)(8)(9)(10).
The partitioning of PA to CDP-DAG and DAG in yeast is largely controlled by Pah1 PAP (11), a peripheral membraneassociated enzyme that catalyzes the Mg 2ϩ -dependent dephosphorylation of PA to produce DAG (2,3,(12)(13)(14). Pah1 PAP activity is relatively low in logarithmically growing cells, whereas the activity is induced as cells progress into the stationary phase (6). Pah1 PAP activity is regulated on multiple levels to control lipid metabolism. The expression of PAH1 is controlled by growth phase and nutrient status (6,15), and the catalytic efficiency of Pah1 PAP is controlled by membrane lipids (16,17), nucleotides (18), and by its phosphorylation and dephosphorylation (19 -22). The phosphorylation status of the enzyme also controls its association with the membrane (19,20,(23)(24)(25) as well as its abundance through proteasomal degradation (26 -28). * This work was supported, in whole or in part, by National Institutes of Health PAP activity is governed by the DXDXT catalytic motif within a haloacid dehalogenase-like domain in Pah1 that is conserved in mammalian lipin PAP enzymes (11,29,30). Genetic and biochemical studies of Pah1 and its mammalian orthologs (i.e. lipins) have revealed that the PAP enzyme is a major regulator of lipid homeostasis and cell physiology (12)(13)(14)(31)(32)(33). The importance of Pah1 in lipid metabolism is indicated by diverse phenotypes of the pah1⌬ mutant, many of which are related to the increased level of PA and the decreased levels of DAG and TAG (11, 29, 34 -36). The increased level of PA induces the expression of UAS INO -containing phospholipid synthesis genes and an increase in membrane phospholipid synthesis (6,11,34,35,37). Considering that PAP activity is elevated as yeast cells progress to the stationary phase (6), the effect of the enzyme loss on phospholipid synthesis is greater in the stationary phase than in the exponential phase (6,11,35). The increased levels of phospholipids in the pah1⌬ mutant are presumably responsible for its aberrant expansion of the nuclear/ER membrane (29,34). The reduced levels of DAG and TAG in the pah1⌬ mutant correlate with a reduced number of lipid droplets (35,36), and the inability to synthesize TAG results in the accumulation of fatty acids giving rise to growth sensitivity to palmitoleic and oleic acids (35).
The pah1⌬ mutant also exhibits phenotypes whose molecular basis in connection with its altered lipid metabolism is not yet clear. It is defective in vacuole fusion and exhibits small fragmented vacuoles as opposed to a large vacuole in the stationary phase (38). Cells lacking Pah1 are defective in cell wall integrity and easily ruptured by sonication (39,40). In addition, the pah1⌬ mutant exhibits a high mannose-to-glucose ratio, a high level of N-acetylglucosamine, and is hypersensitive to K1 killer toxin (39). Moreover, pah1⌬ mutant cells are temperature-sensitive (11,34,41), are unable to grow on glycerol, a non-fermentable carbon source (11,41), and exhibit apoptotic and necrotic phenotypes (35). The inability to grow on glycerol has suggested that pah1⌬ mutant cells are respiratory-deficient (41). In this work, we confirmed that the mutant is defective in growth on glycerol and other non-fermentable carbon sources, but this phenotype was ascribed to reduced levels of ATP through its overconsumption for lipid synthesis rather than due to a defect in oxidative phosphorylation. Moreover, the pah1⌬ mutant was susceptible to oxidative stress, which together with reduced ATP contributed to a shortened chronological life span. These phenotypes of the pah1⌬ mutant were partially complemented by loss of the Dgk1 DAG kinase, indicating that an underlying mechanism is the excess synthesis of membrane phospholipids.

Experimental Procedures
Materials-All chemicals were reagent grade or better. Growth medium components were obtained from Difco Laboratories. BacTiter-Glo TM microbial cell viability assay kit was purchased from Promega. Phusion high fidelity DNA polymerase and the DNA gel extraction kit were purchased from New England Biolabs and Qiagen, respectively. Carrier DNA for yeast transformation was from Clontech. DNA size ladders, molecular mass protein standards, electrophoresis reagents, Triton X-100, Bio-Safe Coomassie G-250, and protein assay reagents were from Bio-Rad. PVDF membrane and the enhanced chemifluorescence Western blotting substrate were from GE Healthcare. Life Technologies, Inc., was the source of mouse anti-porin, anti-carboxypeptidase Y, anti-OxPhos antibody against subunit III of complex IV and anti-phosphoglycerate kinase antibodies and the source of MitoSOX TM Red mitochondrial superoxide indicator, tetramethylrhodamine methyl ester, and 3-12% polyacrylamide gradient gels. Alkaline phosphatase-conjugated goat anti-mouse IgG antibodies and goat anti-rabbit IgG antibodies were from Pierce and Thermo Scientific, respectively. Aprotinin, ATP, benzamidine, bovine serum albumin, DTT, hydrogen peroxide, leupeptin, lyticase, pepstatin, PMSF, sodium azide, and xylenol orange were purchased from Sigma. Radiochemicals and primulin were from PerkinElmer Life Sciences and MP Biomedicals, respectively. Acrylamide solutions and scintillation counting supplies were from National Diagnostics. Silica gel 60 and LK5D TLC plates were obtained from EM Science and Whatman, respectively. Lipids were obtained from Avanti Polar Lipids with the exception of heptadecanoic acid, which was from Alfa Aesar.
Strains and Growth Conditions-The yeast strains used in this study are listed in Table 1. Standard methods were used for culturing yeast (42,43). Cells were grown at 30°C in synthetic complete (SC) medium, which contained 2% glucose as the carbon source, YPD (1% yeast extract, 2% peptone, 2% glucose) medium or YPEG (1% yeast extract, 2% peptone, 0.95% ethanol, 3% glycerol) medium. The ethanol in YPEG medium allows for growth of pah1⌬, which otherwise does not occur in medium with glycerol as the sole carbon source. Where indicated, glucose was replaced in SC or YPD media with 2% ethanol, 3% glycerol, 2% acetate, or 2% lactate. Growth of cells in liquid medium was measured spectrophotometrically at A 600 nm . For measurement of growth on solid medium, liquid cultures were adjusted to A 600 nm ϭ 0.67, followed by 10-fold serial dilutions. The serially diluted cell suspensions were spotted onto solid medium, and growth was scored after incubation for 2-3 days. For the analysis of chronological life span (44), cultures in SC medium were diluted in fresh medium at A 600 nm ϭ 0.01 and grown to the exponential phase (A 600 nm ϭ 1). The exponential phase cells were diluted in fresh medium at A 600 nm ϭ 0.1 and grown for 2 days to the stationary phase. The stationary phase culture (day 0 in chronological life span experiments) was then incubated for 10 days during which aliquots were taken on a daily basis and plated onto YPD agar plates. Colonies formed after incubation for 2 days were counted as being produced from viable cells. The viability at day 0 was set at 100%.
DNA Manipulations and Construction of Mutants-The plasmids used in this study are listed in Table 1. Standard methods were used for isolation of chromosomal and plasmid DNA, for digestion and ligation of DNA, and for PCR amplification of DNA (42). Transformations of Escherichia coli (42) and yeast (45) were performed as described previously. Yeast deletion mutations were generated by the method of one-step gene replacement (46). For construction of the dgk1⌬ mutant (YPY3), the parental strain W303-1A was transformed with the dgk1⌬::HIS3 disruption cassette that was amplified by PCR from the genomic DNA of the dgk1⌬::HIS3 mutant in the RS453 strain background (47). The yeast transformant exhibiting histidine prototrophy was confirmed for the deletion of DGK1 by PCR analysis. For construction of the dgk1⌬ pah1⌬ mutant (YPY4), strain YPY3 was transformed with the pah1⌬::URA3 disruption cassette that was released from pGH317 by digestion with XbaI and SphI (11). The dgk1⌬ transformant exhibiting uracil prototrophy was confirmed for the deletion of PAH1 by PCR analysis. For construction of the tsa1⌬ mutant (YPY5), the parental strain W303-1A was transformed with the tsa1⌬::kanMX4 disruption cassette that was amplified by PCR from the genomic DNA of tsa1⌬::kanMX4 mutant in the BY4741 strain background (yeast deletion consortium). The pah1⌬ tsa1⌬ mutant strain YPY6 was constructed by transformation of the pah1⌬ mutant strain GHY57 with the tsa1⌬::kanMX4 disruption cassette. These tsa1⌬ and pah1⌬ tsa1⌬ mutations were confirmed by PCR analysis and DNA sequencing.
Preparation of Cell Extracts-All steps to prepare cell extracts were performed at 4°C. Yeast cultures were harvested at 1,500 ϫ g for 5 min, washed with water, and resuspended in lysis buffer (50 mM Tris-HCl, pH 7.5, 0.3 M sucrose, 10 mM 2-mercaptoethanol, 0.5 mM PMSF, 1 mM benzamidine, 5 g/ml aprotinin, 5 g/ml leupeptin, and 5 g/ml pepstatin). The cell suspension was added with glass beads (0.5-mm diameter) and then subjected to five repeats of a 1-min burst and a 2-min cooling using a BioSpec Products Mini-Beadbeater-16 (48). The disrupted cells were centrifuged at 1,500 ϫ g for 10 min to separate unbroken cells and cell debris (pellet) from cell extracts (supernatant). The protein concentration was determined by the method of Bradford (49) using bovine serum albumin as a standard.
Isolation of Mitochondria and Analysis of Morphology-Mitochondria were isolated from disrupted spheroplasts by differential centrifugation followed by sucrose gradient centrifugation according to the procedures described by Meisinger et al. (50). Western blot analysis with anti-porin (mitochondrial marker), anti-carboxypeptidase Y (vacuole marker), anti-phosphatidylserine synthase (ER marker) (51), and anti-phosphoglycerate kinase (cytosol marker) antibodies showed that the mitochondrial preparations used in the study were highly purified and essentially devoid of contaminating fractions. Purified mitochondria were prepared for electron microscopy (52) and visualized using a JEOL 1200EX transmission electron microscope at 60 kV and captured with Gatan Orius 830 digital imaging system. Mitochondrial tubulation in cells expressing plasmid pYX142-mtGFP (53) was examined for fluorescence using a Zeiss LSM 710 confocal microscope.
PAGE and Western Blot Analysis-SDS-PAGE (54) using 10% slab gels and Western blotting (55, 56) using a PVDF membrane were performed as described previously. Primary mouse and rabbit antibodies were routinely used at a dilution of 1:1,000. Alkaline phosphatase-conjugated goat anti-mouse and anti-rabbit IgG antibodies were used at a dilution of 1:5,000. Immunecomplexesweredetectedusingtheenhancedchemifluorescence Western blotting substrate. Fluorimaging was used to acquire fluorescence signals from immunoblots, and the intensities of the images were analyzed using ImageQuant software. Blue native-PAGE (57) of digitonin-lysed mitochondria (58) was performed with a 3-12% linear gradient slab gel at 4°C for 20 h. Following electrophoresis, the polyacrylamide gel was stained with Coomassie Blue or electroblotted onto a PVDF membrane. The protein complexes on the PVDF membrane were detected by immunoblotting with anti-OxPhos antibody against subunit III of complex IV (58). Data were analyzed with Quantity One software from Bio-Rad.

Pah1 Phosphatidate Phosphatase in Chronological Life Span
Lipid Extraction and Analysis-Lipids were extracted from purified mitochondria by the method of Bligh and Dyer (59). Mitochondrial phospholipids were separated on silica gel plates by two-dimensional TLC (60). The separated lipids were stained with primulin (61), subjected to fluoroimaging, and quantified with ImageQuant software. The identity of phospholipids was confirmed by comparison with standards. The fatty acid composition of the mitochondrial lipids after deacylation was determined by gas-liquid chromatography (35) using heptadecanoic acid as an internal standard. The mitochondrial phospholipid molecular species were analyzed by liquid chromatography-electrospray ionization quadrupole time-of-flight mass spectrometry (62,63). For radiolabeling of cellular neutral lipids and total phospholipids, cells were grown in SC medium to the late exponential phase (A 600 nm ϳ1). The cells were harvested, washed, and resuspended in SC medium with glycerol as the carbon source and [2-14 C]acetate (1 Ci/ml). After incubation for 2 or 4 h, the radiolabeled cells were harvested; the lipids were extracted (59) and separated by one-dimensional TLC on Silica Gel 60 plates (64). [2-14 C]Acetate was used as a standard to calculate the radioactivity of the radiolabeled lipids. For radiolabeling of phosphoinositol-containing sphingolipids, cells were grown to the stationary phase in SC medium containing myo-[2-3 H]inositol (10 Ci/ml). The lipids were extracted and deacylated, and the radiolabeled sphingolipids were separated by one-dimensional TLC on silica gel LK5D plates (65)(66)(67). The radiolabeled lipids on TLC plates were visualized by phosphorimaging and quantified with ImageQuant software.
Measurements of Oxygen Consumption, Mitochondrial Membrane Potential, ATP, Superoxides, and Lipid Hydroperoxides-Cells were diluted 10-fold in fresh medium and assayed for oxygen consumption using an oxygen electrode in a 500-l chamber (68). Mitochondrial oxygen consumption was confirmed with controls that contained 0.05% sodium azide. Mitochondrial membrane potential was measured with tetramethylrhodamine methyl ester, a cationic red-orange fluorescent dye that is readily sequestered by active mitochondria. Cells were washed twice with phosphate-buffered saline, pH 7.0, and then incubated for 30 min at 30°C in buffer containing 40 M tetramethylrhodamine methyl ester. After washing with phosphate-buffered saline, 10 4 -labeled cells were measured for fluorescence. ATP was measured with the BacTiter-Glo TM microbial cell viability assay kit. Cells were resuspended in 80 l of sterile water and mixed with an equal volume of BacTiter-Glo TM reagent in 96-well opaque plates. After incubation for 3 min, the luminescence produced was measured. ATP (0.1-100 M) was used as a standard in the assay. Superoxides were measured with MitoSOX Red, a mitochondrial superoxide indicator. Cells were washed twice with phosphate-buffered saline, pH 7.0, and incubated for 30 min at 30°C in the buffer containing 5 M MitoSOX Red. After washing twice with phosphatebuffered saline, 10 4 -labeled cells were measured for fluorescence. Lipid hydroperoxides were measured by the ferrous oxidation-xylenol orange complex assay using hydrogen peroxide as a standard (69).
Enzyme Assays-Oligomycin-sensitive F 1 F 0 -ATP synthase activity was measured by following the reverse ATPase reaction using a coupled spectrophotometric assay (70,71). The reaction mixture contained 50 mM HEPES-KOH, pH 8.0, 5 mM MgSO 4, 2.5 mM ATP, 2.5 mM phosphoenolpyruvate, 0.3 mM NADH, 2 g/ml antimycin, 50 g/ml pyruvate kinase, 50 g/ml lactate dehydrogenase, and 20 g of mitochondria with and without 2 g/ml oligomycin. Sod1 and Sod2 superoxide dismutase activities were determined in polyacrylamide gels by nitro blue tetrazolium-negative staining (72,73). Catalase activity was measured by following the decomposition of hydrogen peroxide at A 240 nm (74). The reaction mixture contained 50 mM potassium phosphate buffer, pH 7.0, 20 mM hydrogen peroxide, and 20 g of cell extract. All enzyme assays were performed in triplicate and were linear with time and protein concentration. A unit of ATPase was defined as nanomoles/min and catalase was defined as micromoles/min.
Analyses of Data-Statistical analyses were performed with SigmaPlot software. The p values of Ͻ0.05 were taken as a significant difference.

pah1⌬ Mutant Exhibits a Growth Defect on Non-fermentable
Carbon Sources-One of the distinct phenotypes shown by the pah1⌬ mutant is the lack of growth on glycerol (11,41). Because non-fermentable carbon sources are metabolized through mitochondrial respiration, the growth defect of the pah1⌬ mutant has been attributed to respiratory deficiency (41). To further confirm this phenotype, we examined the growth of the pah1⌬ mutant on acetate, ethanol, and lactate (Fig. 2). As on glycerol, the pah1⌬ mutant did not show growth on acetate or lactate. Although the pah1⌬ mutant grew on ethanol, its growth was much less than that of the wild type control.
Morphology, Phospholipid Composition, and Supercomplex Formation in Mitochondria of the pah1⌬ Mutant-The growth defect of the pah1⌬ mutant on non-fermentable carbon sources raised a question about its mitochondrial morphology and phospholipid composition. Accordingly, these attributes were examined with purified mitochondria. Electron microscopic analysis of the mitochondria purified from wild type and pah1⌬ mutant cells grown in YPEG medium showed no major difference in morphology. The mitochondrial morphology of cells grown on YPEG or in SC medium was also examined by FIGURE 2. pah1⌬ mutant exhibits a growth defect on non-fermentable carbon sources. Wild type and pah1⌬ mutant cells were grown at 30°C to saturation in YPD medium. The cultures were washed and resuspended in water at A 600 nm ϭ 0.67. After 10-fold serial dilutions, 5 l of each cell suspension was spotted onto YP agar medium containing the indicated carbon source, followed by incubation for 3 days (2% glucose) or 5 days (2% ethanol, 2% acetate, 2% lactate, or 3% glycerol). The data are representative of three independent experiments. fluorescence microscopy using a mitochondrion-targeted GFP (53). In this analysis, the mitochondria of the pah1⌬ mutant exhibited the tubulated morphology that is characteristic of stationary phase cells with mitochondrial function (53).
Mitochondrial phospholipids (e.g. CL and PE) play important roles in mitochondrial function that include electron transport, oxidative phosphorylation, and stabilization of the respiratory supercomplexes (58,(75)(76)(77). Accordingly, we examined the phospholipid composition of the mitochondria purified from pah1⌬ cells grown to the stationary phase in SC medium. This analysis showed that the mass of mitochondrial lipids of the pah1⌬ mutant (e.g. 221 Ϯ 27 g of lipid/mg of protein) was not significantly different from that of the wild type control (e.g. 233 Ϯ 13 g of lipid/mg of protein). The composition of the major phospholipids PC, PE, PI, and PS, as well as the mitochondrion-specific phospholipid CL, was not significantly affected by the pah1⌬ mutation (Fig. 3A). However, the pah1⌬ mutation caused the amount of PA to decrease by 42% (Fig. 3A). The amount of the minor mitochondrial phospholipid PG was too low for detection in this analysis. The fatty acid composition of total mitochondrial lipids was analyzed by gas-liquid chromatography. As described previously for whole cell extracts (35), the major fatty acyl species of the mitochondria of wild type and pah1⌬ mutant cells were palmitic, palmitoleic, stearic, and oleic acids (Fig. 3B). The effects of the pah1⌬ mutation were a relatively small (20%) increase in palmitoleic acid and a corresponding small decrease (15%) in oleic acid. The relative amounts of palmitic acid and steric acid were not significantly affected by the pah1⌬ mutation.
The molecular species of the mitochondrial phospholipids were determined by liquid chromatography-electrospray ionization quadrupole time-of-flight mass spectrometry (Fig. 4). This analysis distinguishes the molecular species of a given phospholipid class that have different m/z values and thus identifies each by the total number of carbons and the degree of unsaturation in the acyl chains. By normalizing the mass of mitochondria prior to lipid extraction, we were able to examine differences in the individual phospholipid molecular species between the pah1⌬ mutant and the wild type control. The major changes (e.g. ϳ2-fold) caused by the pah1⌬ mutation in phospholipids with two fatty acyl chains were increases in species with 32 carbons (e.g. PE, PS, and PG) and decreases in species with 34 carbons (e.g. PS and PA) and 36 carbons (e.g. PE and PA). For CL with four fatty acyl chains, the pah1⌬ mutation caused increases (e.g. ϳ2-fold) in many of the species with 64 and 66 carbons.
Respiratory supercomplexes are major mitochondrial components for oxidative phosphorylation that transfer electrons and create a proton gradient for ATP synthesis (78). In Saccharomyces cerevisiae, the electron transport chain complexes III and IV associate to form the supercomplexes III 2 IV (trimer) and III 2 IV 2 (tetramer) (79). The phospholipid CL is essential for organization of complexes III and IV into a supercomplex in intact yeast mitochondria (58). Although the pah1⌬ mutation did not affect the amount of CL in the mitochondria, it did result in elevated levels of unsaturated fatty acyl chains (see above). Because of this result, we examined the respiratory supercomplexes in the mitochondria of pah1⌬ mutant cells grown in YPEG medium. The purified mitochondria were solubilized with digitonin followed by blue native-PAGE. Immunoblot analysis showed that the wild type and pah1⌬ mutant contained the trimeric and tetrameric forms of respiratory supercomplexes, which migrated at the expected size (Fig. 5A). Compared with the wild type, however, the pah1⌬ mutant exhibited a 30% reduction in the levels of the respiratory supercomplexes (Fig. 5B).
Mitochondrial Oxygen Consumption, Membrane Potential, and F 1 F 0 -ATP Synthase Activity of the pah1⌬ Mutant-The pah1⌬ mutant (3.05 Ϯ 0.4 nmol O 2 /min/A 600 nm ) showed no major difference from the wild type control (2.7 Ϯ 0.7 nmol O 2 /min/A 600 nm ) in the rate of oxygen consumption. Likewise, the mitochondrial membrane potential of the pah1⌬ mutant (1.4 Ϯ 0.3 arbitrary units of fluorescence) was not significantly different from that of the wild type (1.2 Ϯ 0.2 arbitrary units of fluorescence). In oxidative phosphorylation, the production of ATP occurs by mitochondrial F 1 F 0 -ATP synthase coupled with the H ϩ flux (80 -82). The enzyme also catalyzes the reverse reaction (i.e. ATP hydrolysis in the absence of membrane potential or a pH gradient (83)). We measured the activity of F 1 F 0 -ATP synthase as its ATPase activity from the mitochondria of wild type and pah1⌬ mutant cells. This analysis showed that the mitochondrial F 1 F 0 -ATPase activity of the pah1⌬ mutant (0.55 Ϯ 0.08 nmol/min/mg) was not significantly different from that of the wild type control (0.58 Ϯ 0.07 nmol/min/ mg). For this assay, ATP must be transported into the mitochondrial matrix for exchange with ADP (84). Thus, the level of ATPase activity exhibited by the pah1⌬ mutant indicates that the mutation affects neither F 1 F 0 -ATPase nor the ATP/ADP transport function of the mitochondria. pah1⌬

Mutant Exhibits a Decrease in Cellular ATP That Inversely Correlates with an Increase in Cellular Membrane
Phospholipids-The cellular levels of ATP were measured in wild type and pah1⌬ mutant cells during growth on SC medium (Fig. 6). During the exponential phase of growth, the ATP levels of the pah1⌬ mutant were similar to those of the wild type control. As cells progressed to the post-diauxic and stationary phases of growth, the amounts of ATP in both the wild type and pah1⌬ mutant cells declined. However, the levels of ATP in the mutant cells rapidly lowered (e.g. 45% at 24 h) when compared with those in the wild type control (Fig. 6B). At the 48-h time point, the reduced ATP levels in the pah1⌬ mutant and wild type were not significantly different. The decreased level of ATP in the pah1⌬ mutant correlates with 2-fold increases in the total cellular phospholipids and fatty acids when compared with wild type cells (6,35). In a complementary experiment, we examined the levels of ATP and lipids in wild type and the pah1⌬ mutant. Cells were grown to exponential phase in SC medium and then transferred to fresh SC medium containing glycerol as the carbon source. The shift to the glycerol-containing medium halted the growth of the pah1⌬ mutant, consistent with the lack of growth on solid growth medium containing glycerol as the carbon source. Two and 4 h following the transfer to the glycerol-containing medium, the cellular ATP level of  A, mitochondria from wild type and pah1⌬ mutant cells grown in YPEG were solubilized in digitonin, and the proteins were separated by blue native-PAGE. Following the electrophoresis, the proteins were transferred to a PVDF membrane, which was then probed with antibodies raised against subunit III of complex IV. B, amounts of the tetramer (III 2 IV 2 ) and trimer (III 2 IV) supercomplexes were analyzed with Quantity One software. A representative Coomassie Blue-stained gel is shown in A, and the quantitation data shown in B is the average of three experiments Ϯ S.D. (error bars). *, p Ͻ 0.05 versus WT. AU, arbitrary units. pah1⌬ mutant cells was ϳ2-fold lower than that of the wild type control (Fig. 7A). The reduction of ATP levels in the pah1⌬ mutant correlated with the time-dependent increases in cellular levels of radiolabeled phospholipids (e.g. 3.3-fold at 4 h) and fatty acids (e.g. 2.3-fold at 4 h) (Fig. 7B). In addition, the pah1⌬ mutant exhibited decreases in the levels of DAG (e.g. 41% at 4 h) and TAG (e.g. 98% at 4 h) (Fig. 7B).
Complex Sphingolipid Composition of the pah1⌬ Mutant-The pah1⌬ mutation elevates all of the major membrane phospholipids in the stationary phase of growth (35). Among these phospholipids, PI is used as a substrate in two steps of complex sphingolipid synthesis (85,86). Accordingly, we questioned whether sphingolipid synthesis was affected by the increased level of PI in the pah1⌬ mutant. Wild type and pah1⌬ mutant cells were grown to the stationary phase in the presence of myo-[2-3 H]inositol followed by lipid extraction, the deacylation of phospholipids, and the separation of the sphingolipids by TLC. The levels of complex sphingolipids in the pah1⌬ mutant were not significantly different from those of the wild type control (Fig. 8A). The compositional analysis indicated that the pah1⌬ mutation caused small decreases in the amount of inositolphosphoceramide (20%) and mannosylinositolphosphoceramide (10%) but an increase (30%) in the amount of mannose (inositol-P) 2 -ceramide (Fig. 8B).
pah1⌬ Mutant Exhibits Increased Levels of Superoxides and Lipid Hydroperoxides, Has Reduced Levels of Superoxide Dismutase 2 and Catalase Activity, and Is Hypersensitive to Hydrogen Peroxide-Superoxides are generated by electron leaks as by-products during oxidative phosphorylation and react readily with macromolecules (e.g. lipids, protein, and nucleic acids), causing detrimental effects on cellular structure and function (87)(88)(89). The reduced levels of respiratory supercomplexes in the pah1⌬ mutant raised a question about the levels of reactive oxygen species. We examined the levels of mitochondrial superoxide in SC medium-grown stationary phase cells by staining with MitoSOX Red. This analysis showed that the superoxides in the pah1⌬ mutant were 3-fold greater than that of the wild type control (Fig. 9A). Because the reactive oxygen species are known to cause oxidative damage to cellular macromolecules, we questioned whether the levels of cellular lipid hydroperoxides were affected by the pah1⌬ mutation. The total lipid fraction was isolated from stationary phase cells grown in SC medium and were measured for hydroperoxides by a colorimetric assay using the ferrous oxidation-xylenol orange complex. This analysis showed that the cellular content of lipid hydroperoxides in the pah1⌬ mutant was 3-fold higher than that of wild type cells (Fig. 9B).
Superoxide dismutase, which catalyzes the conversion of superoxide to hydrogen peroxide, was measured in SC medium-grown stationary phase cells. The mitochondrion-associated Sod2 superoxide dismutase activity was ϳ50% lower in the pah1⌬ mutant when compared with the wild type control, whereas the cytosolic Sod1 activity was not significantly af-   fected in the pah1⌬ mutant (Fig. 10A). The cell extract was also assayed for the activity of catalase, the enzyme that decomposes hydrogen peroxide to water and oxygen. This analysis showed that the pah1⌬ mutant had a 32% reduction in catalase activity (Fig. 10B). Given these observations, we questioned whether the growth of the pah1⌬ mutant would be affected by hydrogen peroxide. Hydrogen peroxide was added to SC or YPD media plates, and the growth of wild type and pah1⌬ mutant cells on the growth medium was scored after 3-5 days of incubation. This analysis showed that the pah1⌬ mutant was hypersensitive to hydrogen peroxide (Fig. 11). Higher concentrations of hydrogen peroxide were required to inhibit the growth of the mutant on rich YPD medium.
Loss of Tsa1 Thioredoxin Peroxidase Causes a Synthetic Growth Defect with the pah1⌬ Mutation-Tsa1 thioredoxin peroxidase is a major antioxidant enzyme in yeast that protects cells against oxidation systems capable of generating reactive oxygen species (90,91). The lack of TSA1 has been shown to cause synthetic lethality with the lack of PAH1 (92). In the genetic background examined here, the pah1⌬ tsa1⌬ mutant was viable when grown in SC medium (Fig. 12A). However, the double mutant grew to a lower cell density when compared with the tsa1⌬ mutant, which itself grew to a lower density when compared with the pah1⌬ mutant or the wild type control (Fig. 12A). We examined the effect of the tsa1⌬ mutation on the hydrogen peroxide sensitivity of pah1⌬ mutant cells. In this experiment, we utilized the concentration of hydrogen peroxide (e.g. 0.1 mM) that was not inhibitory to the pah1⌬ mutant but was inhibitory to the tsa1⌬ mutant (Fig. 12B). The pah1⌬ mutation exacerbated the inhibitory effect of hydrogen peroxide on the growth of the tsa1⌬ mutant (Fig. 12B), indicating a synthetic growth defect caused by the pah1⌬ and tsa1⌬ mutations.
dgk1⌬ Mutation Suppresses the Shortened Chronological Life Span, the Hypersensitivity to Hydrogen Peroxide, and the Growth Defect on Non-fermentable Carbon Sources of the pah1⌬ Mutant-We examined the effect of the pah1⌬ mutation on the chronological life span, which is defined as the length of time that non-dividing cells survive (44). Wild type and pah1⌬ mutant cells were grown to the stationary phase in SC medium, and then over time they were examined for their ability to form colonies on rich YPD growth medium. The wild type control lost about 50% viability 8 days after reaching quiescence (Fig. 13). In contrast, it took only 3 days for the pah1⌬ mutant to lose the same extent of viability; no pah1⌬ mutant cell was viable 8 days after reaching quiescence (Fig. 13).
The Dgk1 DAG kinase counterbalances the activity of Pah1 PAP by catalyzing the CTP-dependent conversion of DAG to PA (47). Several pah1⌬ mutant phenotypes, which include an irregular nuclear/ER membrane expansion, reduced lipid droplet formation, and increased cellular phospholipid content, are    OCTOBER 16, 2015 • VOLUME 290 • NUMBER 42 complemented by the dgk1⌬ mutation (35,36,47). We questioned whether the dgk1⌬ mutation also suppresses some of the phenotypes described here. The dgk1⌬ mutation extended the chronological life span of the pah1⌬ mutant by 2 days (Fig. 13). In fact, the dgk1⌬ mutation by itself caused an increase in chronological life span; ϳ50% of the culture was still viable 10 days after quiescence. The deletion of DGK1 also complemented the growth defects of pah1⌬ mutant cells exposed to hydrogen peroxide (Fig. 14A) and permitted growth of the mutant on medium with non-fermentable carbon sources (Fig.  14B). The dgk1⌬ mutant itself grew on non-fermentable carbon sources and was insensitive to growth inhibition by hydrogen peroxide.

Discussion
In S. cerevisiae, Pah1 PAP plays a crucial role in the synthesis of the storage lipid TAG, and in the regulation of membrane phospholipid synthesis (2,3,12,13). The loss of Pah1 PAP activity, which affects the levels of both its substrate PA and product DAG, results in striking changes in the composition of cellular lipids, such as a great reduction in the levels TAG, an accumulation of fatty acids, and a significant increase in membrane phospholipids (6,11,35). These phenotypic changes in lipid metabolism are directly or indirectly coupled to other phenotypes, such as an irregular expansion of the nuclear/ER membrane, marked vacuole fragmentation, and increased fatty acid-induced lipotoxicity (11,29,34,35,38). Similarly, in mammalian cells, loss of lipin PAP activity results in metabolic disorders that include lipodystrophy, insulin resistance, peripheral neuropathy, rhabdomyolysis, and inflammation (30,(93)(94)(95)(96)(97)(98)(99)(100)(101).
The initial characterization of the pah1⌬ 4 mutant indicated a respiratory-deficient phenotype because the cells are unable to grow on glycerol as the carbon source (41). In this work, we confirmed that the pah1⌬ mutant has a growth defect on non-fermentable carbon sources, but we could not attribute this growth defect to respiratory deficiency; the mutant did not exhibit major changes in oxygen consumption rate, membrane potential, F 1 F 0 -ATP synthase activity, or gross mitochondrial morphology. That the pah1⌬ mutant is not respiratory-deficient is also indicated by the lack of a petite phenotype (41), which is typical of mutants in lipid metabolism that are respiratory-deficient (76,102,103).
Although mutations in PAP-encoding PAH1 (11) and Lpin1 (104) genes in yeast and mice, respectively, cause similar alterations in lipid metabolism, their effects on some physiological functions may differ. In contrast to the pah1⌬ mutant, which does not exhibit a clear defect in mitochondrial morphology and function, the lipin 1-deficient mouse accumulates large, misshapen mitochondria in muscle whose disorganized cristae become evident after statin treatment. In addition, the lipin 1-deficient muscle is significantly defective in increasing oxygen consumption in response to a mitochondrial uncoupler (104). These phenotypes of lipin 1 deficiency are ascribed to a defect in mitochondrial autophagy that is activated through the DAG-dependent protein kinase D-Vps34 phosphoinositide 3-kinase signaling pathway (104).
The pah1⌬ mutation did not have a major effect on the content of the major mitochondrial phospholipids CL and PE, which have overlapping involvements in mitochondrial functions that include oxidative phosphorylation (105). However, the mitochondrial phospholipid that was affected by the pah1⌬ mutation was PA. This was an unexpected result because the major effect of the pah1⌬ mutation on total cellular membrane phospholipids is an increase in PA content (11,35). The reduction in PA does not appear to be due to its increased utilization; the mass of the mitochondrial phospholipids was not affected by the pah1⌬ mutation. That the mitochondrial lipid mass was not affected by the pah1⌬ mutation also differs from the effect that the mutation has on total cellular lipids whose amounts increase in mutant cells (6,11,35). Because PA is synthesized in the ER (3), it is unclear whether or not the reduction in PA 4 pah1⌬ was previously known as smp2⌬.  content was a consequence of a defect in ER to mitochondrial trafficking (106) or due to some other metabolic processes. In mammalian cells, the level of PA is thought to be involved in mitochondrial fusion and fission events (107,108). An increase in PA content, as mediated by phospholipase D activity, induces mitochondrial fusion/tubulation (107), whereas the depletion of PA, as mediated by phospholipase A 1 activity, induces mitochondrial fission/fragmentation (108). Based on these observations, we speculated that the pah1⌬ mutant with a reduced PA content in mitochondria might exhibit a fragmented mitochondrial morphology in stationary phase cells, but this is not the case as the mutant exhibits typical mitochondrial tubulation morphology.
Among the mitochondrial functions of CL is the requirement for the formation/stability of respiratory supercomplexes (58,77). The levels of CL and its molecular species indicate that the pah1⌬ mutant is not defective in the synthesis and remodeling of the mitochondrial phospholipid. Despite its increases in the levels of CL molecules with unsaturated fatty acyl chains, the pah1⌬ mutant was shown to contain a reduced (30%) level of respiratory supercomplexes. In light of the recent finding that remodeling of the CL fatty acyl groups does not affect mitochondrial morphology or oxidative phosphorylation (109), it is unclear whether the changes in CL molecular species have an effect on the respiratory supercomplexes. Respiratory supercomplexes are thought to play a role in the enhancement of electron flow and thereby to prevent the formation of excess oxygen radicals (78,110). Thus, it is possible that the reduced level of the supercomplexes in the pah1⌬ mutant might contribute to the elevated level of superoxide.
Despite the fact that the pah1⌬ mutant is not respiratorydeficient, it has a reduced amount of ATP. ATP is required for the synthesis of fatty acyl-CoA molecules as well as for the synthesis of CTP, the essential nucleotide required for the synthesis of phospholipids (1,3,111). The reduction in ATP exhibited an inverse correlation with the increase in cellular membrane phospholipid synthesis that occurs when pah1⌬ mutant cells progress into the stationary phase (6,11,35). This relationship was also observed when late exponential phase glucose-grown cells were transferred to glycerol-containing growth medium. We posit that the decreased ATP content in pah1⌬ mutant cells is the result of its overconsumption for membrane phospholipid synthesis, and as a consequence, the reduced level of ATP contributes to the growth defect on non-fermentable carbon sources and a reduced chronological life span.
To generate ATP in mitochondria, electrons are transferred to oxygen through the electron transport chain, and during this process, reactive oxygen species are produced that can cause damage to macromolecules, ultimately resulting in the loss of cell viability (68,(112)(113)(114). In this regard, we discovered that stationary phase pah1⌬ mutant cells exhibit elevated levels of superoxide and cellular lipid hydroperoxides. Contributing factors include the reduced level of respiratory supercomplexes (see above) and the reduced activities of Sod2 superoxide dismutase and catalase that are responsible for removing reactive oxygen species. In addition, the elevated level of membrane phospholipids, which are prone to oxidation (115), provides an explanation why the pah1⌬ mutant might be more susceptible to oxidative stress. Consequently, pah1⌬ mutant cells were hypersensitive to hydrogen peroxide. That the pah1⌬ mutant is more prone to oxidative stress than the wild type is further supported by the synthetic growth defect imparted by the tsa1⌬ mutation.
Oxidative stress and accelerated aging are intimately related (110,116,117). Indeed, the pah1⌬ mutant, which is prone to oxidative stress, had a significantly reduced chronological life span. The loss of Dgk1 DAG kinase in the pah1⌬ mutant partially complemented its defect in chronological life span. That the dgk1⌬ mutation suppresses the increased synthesis of membrane phospholipids in the pah1⌬ mutant, but does not alleviate its defect in TAG synthesis (47), indicates that the shortened chronological life span of the pah1⌬ mutant is due, at least in part, to the increased synthesis of membrane phospholipids. This hypothesis is further supported by the fact that the dgk1⌬ mutation also alleviates the hypersensitivity of the pah1⌬ mutant to hydrogen peroxide and allows for its growth on non-fermentable carbon sources.
Although the work presented here supports the notion that the overconsumption of ATP for lipid synthesis and oxidative stress contributes to reduced chronological life span, we cannot rule out other mechanisms that may come into play as a result of the altered regulation of lipid synthesis that occurs in the pah1⌬ mutant. It is known that a defect in sphingolipid synthesis, as mediated by pharmacological inhibition or down-regulating expression of serine palmitoyltransferase (the first enzyme in the sphingolipid synthesis pathway (86)), extends chronological life span in yeast (118). This process involves the down-regulation of the Pkh1/2-Sch9 signaling pathway (118), which in turn activates the Snf1/AMP kinase signaling pathway and down-regulates the protein kinase A and target of rapamycin complex 1 signaling pathways (119). Because PI, one of the major phospholipids that is elevated in the pah1⌬ mutant (35), is the direct precursor for the synthesis of inositolphosphoceramide and mannose (inositol-P) 2 -ceramide (85,86), we considered the hypothesis that the reduced chronological life span in the pah1⌬ mutant might also be ascribed to an increase in membrane sphingolipids. However, despite its elevated PI content, the pah1⌬ mutant did not exhibit major changes in sphingolipids. Additional studies will be required to further explore other mechanisms (e.g. the target of rapamycin complex 1 signaling pathway) that might be involved in controlling the chronological life span of the pah1⌬ mutant.
In conclusion, the work reported here emphasizes the importance of Pah1 PAP in regulating the coordinate synthesis of TAG and membrane phospholipids. In particular, this work reveals a novel role of Pah1 PAP in the chronological life span that is contributed by energy expenditure and oxidative stress.
Author Contributions-Y. P. performed the experiments and prepared the manuscript. E. M. performed the analyses of mitochondrial respiratory supercomplexes and morphology by electron microscopy. T. A. G. analyzed the mitochondrial phospholipid molecular species by mass spectrometry. G. S. H. and G. M. C. directed the research and contributed to the preparation of the manuscript. All authors analyzed the results and approved the final version of the manuscript.