Inorganic Pyrophosphatase Defects Lead to Cell Cycle Arrest and Autophagic Cell Death through NAD+ Depletion in Fermenting Yeast*

Background: The cellular consequences of inorganic pyrophosphate excess in a eukaryotic cell are unknown. Results: Saccharomyces cerevisiae cells depleted of inorganic pyrophosphatase Ipp1p on respiratory carbon sources undergo cell cycle arrest, but fermenting cells undergo NAD+ depletion-induced autophagy and die. Conclusion: Inorganic pyrophosphatase depletion can cause cell death through autophagy. Significance: This is the first work detailing the cellular consequences of intracellular pyrophosphate accumulation in eukaryotes. Inorganic pyrophosphatases are required for anabolism to take place in all living organisms. Defects in genes encoding these hydrolytic enzymes are considered inviable, although their exact nature has not been studied at the cellular and molecular physiology levels. Using a conditional mutant in IPP1, the Saccharomyces cerevisiae gene encoding the cytosolic soluble pyrophosphatase, we show that respiring cells arrest in S phase upon Ipp1p deficiency, but they remain viable and resume growth if accumulated pyrophosphate is removed. However, fermenting cells arrest in G1/G0 phase and suffer massive vacuolization and eventual cell death by autophagy. Impaired NAD+ metabolism is a major determinant of cell death in this scenario because demise can be avoided under conditions favoring accumulation of the oxidized pyridine coenzyme. These results posit that the mechanisms related to excess pyrophosphate toxicity in eukaryotes are dependent on the energy metabolism of the cell.

Inorganic pyrophosphatases are required for anabolism to take place in all living organisms. Defects in genes encoding these hydrolytic enzymes are considered inviable, although their exact nature has not been studied at the cellular and molecular physiology levels. Using a conditional mutant in IPP1, the Saccharomyces cerevisiae gene encoding the cytosolic soluble pyrophosphatase, we show that respiring cells arrest in S phase upon Ipp1p deficiency, but they remain viable and resume growth if accumulated pyrophosphate is removed. However, fermenting cells arrest in G 1 /G 0 phase and suffer massive vacuolization and eventual cell death by autophagy. Impaired NAD ؉ metabolism is a major determinant of cell death in this scenario because demise can be avoided under conditions favoring accumulation of the oxidized pyridine coenzyme. These results posit that the mechanisms related to excess pyrophosphate toxicity in eukaryotes are dependent on the energy metabolism of the cell.
Many biosynthetic reactions produce inorganic pyrophosphate (PP i ) as a by-product. The advantage of this seems to lie in a greater ⌬GЈ 0 stored in the ␣,␤-phosphoanhydride bond of nucleoside triphosphates compared with that in the ␤,␥-phosphoanhydride bond and the fact that PP i is rapidly hydrolyzed to orthophosphate, driving both the kinetics and energetics of anabolic reactions toward biosynthesis (1). For this reason, PP i homeostasis is exceedingly important for the cell. Removal of PP i from the cytosol is carried out by two main non-homologous enzyme types: soluble inorganic pyrophosphatases (sPPases) 4 and proton-translocating membranebound pyrophosphatases (H ϩ -PPases) (2,3). The first of these is found as the sole enzyme that keeps cytosolic PP i at very low levels in animals and fungi. Soluble PPases may share this function with their membrane-bound counterparts in plants, protists, and some bacteria and archaea; albeit until now only the involvement and proper location of the membrane-bound H ϩ -PPases have been proved in these organisms (2,3). Not surprisingly then, a genetic defect leading to the absence of sPPase activity has been found to cause Escherichia coli to stop cell proliferation (4) and to be inviable in the budding yeast Saccharomyces cerevisiae (5)(6)(7)(8). In the case of E. coli, this growth defect was shown to be non-lethal (4). However, no study to date has dealt with the physiological basis accounting for the lethality of sPPase deficiency in eukaryotes or how it affects the cell cycle in either prokaryotes or eukaryotes.
Up to now, the cellular importance of PP i homeostasis has only been patchily and scantily studied. In bacteria, it has been observed that the level of intracellular PP i does not correlate directly with that of sPPase polypeptide; on the contrary, it remains constant until the amount of this enzyme drops below a minimum threshold (9), suggesting post-translational regulation of its catalytic activity or a constitutive excess of this enzyme under physiological conditions. In mammals, alteration of sPPase levels is associated to several illnesses, many related to calcium phosphate homeostasis (10,11). Interestingly, human sPPase PPA1 is now being found associated with tumor cells. Thus, cytosolic levels of this protein have been shown to be increased in several proteomics studies dealing with cancer tissues from different origins, such as lung and colon (12,13). In addition, plants have a very intricate PP i metabolism: they possess a number of structurally and catalytically diverse PP i -utilizing proteins like the membrane-bound H ϩ -PPases, several PP i -dependent glycolytic enzymes, and multiple sPPase isoforms (2,3,14). It has been observed that photosynthetic carbon assimilation and plant metabolism in general are greatly affected by changes in the levels of soluble PPases (15)(16)(17). However, we cannot start to understand plant PP i metabolism and its influence on economically important issues, such as seed oil or carbohydrate yield, if we are unsure how simpler unicellular organisms behave in this respect.
In this report, we show for the first time that a conditional genetic defect in the yeast IPP1 gene induces massive cell death in fermenting cultures but only induces cell cycle arrest in respiring cells. Furthermore, we describe the physiological mechanism for cell death and the metabolic step that causes it. In addition we show that under respiratory conditions cell cycle arrest is reversible. These results may open the way to new approaches in metabolism-related cell bioengineering and chemotherapy and underline the importance of so-called housekeeping cell processes like PP i homeostasis.
Cell Death and Viability Assays-Cell viability was assessed by colony forming ability. Briefly, approximately 500 cells as estimated by A 600 were plated onto the appropriate solid medium, and survival was scored as the percentage of colonies formed from the total inoculated cells. Because different conditions could lead to alterations in the number of yeast cells estimated by absorbance due to size differences, estimations were corrected by counting cells on a hemocytometer. Death was estimated on propidium iodide-stained cells by flow cytometry as cells bearing DNA contents less than 1C (sub-G 1 ) (21).
Flow Cytometry and Microscopy-Cells were stained with propidium iodide essentially as described by Sazer and Sherwood (22) and analyzed on a Coulter Epics XL apparatus as described (23).
For microscopical analysis, cells were visualized using a Leica DM 6000B type fluorescence microscope. Cell size was evaluated from differential interference contrast microphotographs as the apparent area covered by the cell evaluated using the NIH ImageJ program (24). Percentages of budded cells were estimated from microphotographs; a minimum of 150 cells per treatment were counted. TUNEL assays were done essentially as described (25) using a commercial kit (Roche Applied Science, catalog number 11684817910). Cells were stained by incubating with 5 g/ml dihydrorhodamine 123 for 2 h under appropriate growth conditions and then viewed under a rhodamine filter. FM4-64 staining was done as described (26); estimation of the percentage of cells showing solid vacuoles on FM4-64 staining was done as for budded cells above.
Cell Extracts and Enzymatic Assays-Cytosol-enriched fractions were obtained from exponentially growing cells by breaking them using glass beads (0.5-mm inner diameter) in a buffer containing 0.33 M sorbitol, 25 mM Tris-HCl (pH 7.5), 2 mM EDTA, 2 mM DTT, 1 mM PMSF, 1 mM benzamidine, and 1 mM ⑀-aminocaproic acid. Cells were subjected to five 1-min bursts on a Vortex mixer at full speed with 1-min intervals on ice between bursts. Total homogenates were cleared of unbroken cells and cell debris by low speed centrifugation (5 min at 500 ϫ g) followed by high speed centrifugation for 15 min (20,000 ϫ g). The pyrophosphatase activity assay was performed as described previously (8), and released phosphate was determined as described (27). Activity in the presence of 2 mM potassium fluoride, corresponding to unrelated diphosphatase activities, was subtracted from the data. These activities represented less than 5% of total PP i hydrolytic activity measured in control cells.
Protein Determination and Western Blotting-Protein determination was done using a dye binding-based assay from Bio-Rad (catalog number 500-0006) according to the manufacturer's instructions and using ovalbumin as a standard. Proteins were separated by SDS-PAGE using standard procedures. Proteins were then transferred to nitrocellulose filters and probed with antibodies raised against the sPPase PPA I from the microalga Chlamydomonas reinhardtii (14) or a commercial anti-GFP antibody (Clontech, catalog number 632460). Proteins were visualized on x-ray films using horseradish peroxi-dase-coupled secondary antibodies and a chemiluminescence kit (Millipore, catalog number WBKLS0100).
Quantitation of Metabolites-For PP i quantitation, YPC3 cells were grown on the appropriate medium for 24 h under semicontinuous culture conditions as stated above. Approximately 10 9 cells were collected by centrifugation, washed with ice-cold water, and extracted with 1 ml of 4% perchloric acid. To aid cell breakage, cells were subjected to three 1-min Vortex bursts in the presence of glass beads (0.5-mm inner diameter). After decanting glass beads and debris by centrifugation, extracts were neutralized by adding sequentially 140 l of 5 M KOH and 60 l of 1 M Tris-HCl (pH 7.5) and centrifuged again at top speed in a tabletop centrifuge for 30 min to decant potassium perchlorate salts. PP i was measured in the supernatants using a commercial kit (Molecular Probes, catalog number E-6645) according to the manufacturer's instructions. Pyridine nucleotide coenzyme extractions were performed as follows. Briefly, YPC3 and W303-1a cells were grown under semicontinuous conditions for 18 h (SG and CR conditions) or 36 h (MR and galactose controls). Then approximately 2.5 ϫ 10 7 cells were washed with ice-cold water, resuspended in 200 l of either 50 mM NaOH (NADH extraction) or 50 mM HCl (NAD ϩ extraction), and heated at 60°C for 30 min on a heat block. After this, extracts were neutralized and clarified by centrifugation. Coenzymes were determined using a coupled assay consisting of 10 mM Bicine, pH 8.0, 15 units/ml yeast alcohol dehydrogenase (Sigma-Aldrich, catalog number A-7011), 0.4 mg/ml phenazine methosulfate, 0.25 mg/ml thiazolyl blue formazan, and 3% ethanol. Pyridine nucleotide coenzyme levels were spectrophotometrically estimated at 570 nm as the kinetic production of reduced formazan.
RNA Methods-Yeast RNA was extracted using the hot phenol method (28). One microgram of RNA was subjected to cDNA synthesis using a Quantitect reverse transcription kit (Qiagen, catalog number 205311) according to the manufacturer's instructions. PCR was done using 1 l of newly synthesized cDNA in a 20-l tube. Sequences of the oligonucleotides used as primers are available upon request. The total number of cycles was set to 25 to avoid saturation. Amplified DNA was separated on agarose gels under standard conditions, photographed using a Gel Doc XRϩ System (Bio-Rad, catalog number 170-8195) and quantified using Quantity One v4.6.2 software (Bio-Rad, catalog number 170-9600).
Miscellaneous Methods-Experiments were typically done in triplicate. The actual number of independent experiments is included in the figure legends (n). When data from representative experiments are shown, this is stated in each case. Statistical analysis was done using unpaired t tests with a cutoff of p Յ 0.05.

RESULTS
Effect of Ipp1p Deficiency on Cell Viability-We have previously shown that a S. cerevisiae engineered strain in which the gene IPP1 encoding its cytosolic sPPase is governed by a GAL1 inducible/repressible promoter (8) is viable as long as a plasmid-borne heterologous H ϩ -PPase is expressed when the GAL1 promoter is repressed. An improved version of this mutant (YPC3) (18) has been used to study the kinetics of events related to Ipp1p deficiency to delineate the cellular importance of PP i homeostasis. Phosphate and carbon metabolisms are long known to interact in their regulation (for example, see Ref. 29). Also, viability is increasingly associated to carbon source quality and availability in many organisms, in particular to the dichotomy between fermentation and respiration and more recently to calorie availability (30 -32). Having this in mind, we compared YPC3 control cells (grown on GAL1inducing conditions, 2% galactose) with others placed under three different repressing carbon source conditions: SG, CR, and MR. As expected, after 24 h, YPC3 cells accumulated 0.31 Ϯ 0.05, 0.31 Ϯ 0.05, 0.21 Ϯ 0.02, and 0.02 Ϯ 0.01 nmol of PP i /10 6 cells in SG, CR, MR, and galactose control conditions, respectively; i.e. SG and CR cells displayed approximately 16-fold and MR cells displayed approximately 10-fold greater levels of PP i than those observed under galactose conditions. On the other hand, Ipp1p polypeptide levels decreased sharply upon switch of YPC3 cells from galactose (control) to glucose irrespective of its concentration (Fig. 1A, SG and CR). In addition, the switch to MR conditions induced a slower rate of Ipp1p disappearance. This is in agreement with the GAL1 promoter being inactive due to the lack of its transcription factor, Gal4p, when cells are grown on this carbon source but not actively repressed by Gal80p and Mig1p as it is under glucose conditions (33). In any case, Ipp1p polypeptide levels were undetectable on Western blots 6 h after the switch to glucose and 24 h after the switch to MR. These results were confirmed by PP i hydrolytic activity assays (Fig. 1B). Six hours after the switch to SG or CR or 24 h after the switch to MR, no sPPase activity was observed. Concomitant with this, growth was slowed to a halt after 18 h under SG and CR conditions and 36 h under MR condition (Fig. 1C), whereas galactose controls showed a constant growth rate even after 48 h of semicontinuous culture.
To ascertain whether viability was compromised and the stage of sPPase activity at which this reduction occurs, we also followed the ability of these cells to form colonies (Fig. 1D). Albeit enzymatic activity was undetectable much earlier, cell viability was not observed to drop until 18 h after switching to SG or CR condition, and after 21 h, it was observed to plateau and reach a minimum at approximately 27% of plated cells. Similarly, the viability of MR cells dropped severely at 36 h and reached the same lowest value as SG and CR cells after 48 h (Fig.  1D). In subsequent experiments where only a single time point was assayed, unless otherwise indicated, all determinations were done on YPC3 cells incubated for 18 h under SG or CR condition or for 36 h under MR condition.
Effect of Ipp1p Deficiency on Cell Cycle-We were interested in determining the consequences that a defect in PP i homeostasis had on the cell cycle. To this end, we compared the cell cycle profiles of cells incubated under control, SG, CR, and MR conditions ( Fig. 2A, left panel). In this case, a differential behavior was evident between fermenting and respiring cells. Under both fermenting conditions (SG and CR) cells arrested their cycle abruptly; additionally, a great proportion of cells showed a DNA content smaller than that corresponding to a haploid intact genome (sub-G 1 ), indicative of cell death (21). Conversely, MR cells showed no accumulation of sub-G 1 cells  despite displaying a nearly equally severe cell cycle arrest. In the case of SG and CR, a discernible sub-G 1 population increase over control cells could be observed as early as 18 h after the carbon source switch ( Fig. 2A, right panel) and reached up to 25% of the cells in a culture. However, under MR conditions, no meaningful increase in the proportion of sub-G 1 cells with respect to control cells was observed even 56 h after the glycerol switch (6.5 Ϯ 1.3 and 2.2 Ϯ 0.2% for MR and control cells, respectively; n ϭ 10,000 cells). Because propidium iodide staining does not provide enough accuracy to discern between a G 1 and an S phase cell cycle arrest by flow cytometry in budding yeast, we measured the amounts of CLN3, CLB5, and CLB3 cyclin mRNAs by semiquantitative RT-PCR as markers of G 1 , S, and G 2 phases of the yeast cell cycle, respectively (Fig. 2B). Both SG and CR conditions showed a marked depletion of all cyclins studied, a situation usually associated to G 0 (32). MR cells displayed an accumulation of CLB5 cyclin mRNA and an overall profile similar to that observed in hydroxyurea-treated cells, indicative of S phase arrest, and in sharp contrast to the profile observed in cells where G 1 arrest was induced by addition of ␣-factor (Fig. 2B, right panel).
Morphology and Characteristics of Ipp1p-deficient SG and CR Cells-In agreement with cyclin mRNA results, under the microscope (Fig. 3A), most MR cells showed large buds compared with control cells (percentages of budded cells ϮS.D. from a representative experiment were 46.1 Ϯ 5.9 and 72.2 Ϯ 5.8% for galactose control and MR, respectively; n ϭ 2), whereas SG and CR cells were largely unbudded (5.0 Ϯ 3.1 and 5.0 Ϯ 5.2% for SG and CR, respectively). Strikingly, they also showed a characteristic morphology: they were much bigger than control or MR cells and appeared to contain an enlarged vacuole that filled much of the cell interior. When the apparent cellular area observable under the microscope was quantified, MR and control cells showed no statistically significant differences between themselves, whereas SG and CR cells were approximately 2-fold larger than controls (Fig. 3A, bottom panel). Nuclear DNA degradation and hence large sub-G 1 populations are characteristic features of apoptosis. In this case, nuclear DNA degradation frequently gives rise to the appearance of a characteristic laddering on genomic DNA agarose gels due to internucleosomal fragmentation. However, when genomic DNA from SG and CR cells was electrophoresed, a continuous smear was observed instead (Fig. 3B). TUNEL assays were also negative on SG, CR, and MR cells (data not shown). Reactive oxygen species are also typical inducers and markers of programmed cell death (34). However, both SG and CR cells did not show any discernible increases in reactive oxygen species as measured by dihydrorhodamine 123 fluorescence (Fig. 3C). In marked contrast, MR cells showed a clear fluorescent staining indicative of reactive oxygen species accumulation; this feature was not observed on the parental W303-1a strain grown under the same conditions.
The prevalence of vacuolization in SG and CR cells led us to test whether autophagy was induced. To test this hypothesis, a GFP-ATG8 fusion construct was transformed into YPC3, and the proteolytic processing of ATG8 was followed by Western blot as a marker of autophagy. Control cells kept in galactose and MR cells did not show any processing of the GFP-Atg8p fusion, whereas SG and CR cells showed a clear band corresponding to liberated GFP in addition to the full fusion. Also, bands of intermediate molecular weight were visible, probably reflecting partial degradation of GFP-Atg8p (Fig. 4A). A similar band pattern was displayed by a nitrogen starvation positive control of autophagy.
The fluorescent vital dye FM4-64 has been shown to stain differentially the acidic vacuoles of autophagic yeast cells due to accumulation of membrane material (26). When FM4-64 was added to cultures of SG and CR cells, a pattern of solid vacuoles was observed, similar in appearance to what is observed in nitrogen-starved yeast cells (Fig. 4B). The proportion of cells showing solid vacuoles under these conditions was similar to that of cells bearing sub-G 1 amounts of DNA. On the contrary, galactose-grown control and MR cells did not show any apparent accumulation of FM4-64 inside vacuoles and retained a normal morphology.
Reversibility of Cell Cycle Arrest in MR Cells-In contrast to SG and CR cells, yeast under MR did not show any signs of cell death. We therefore tested whether these cells were in fact arrested in a reversible way and whether removal of excess PP i could promote recovery. We first tested whether cells grown under MR conditions for 48 h and then placed in fresh YPGalactose liquid medium could resume growth. Indeed, although cells maintained under MR conditions (fresh YPGlycerol medium) could not proliferate, A 600 clearly increased in YPGalactose cultures (Fig. 5A). Nevertheless, these cultures showed an approximately 6-h lag phase. Probably, during this period MR cells lost their cytosolic PP i due to diffusion and had their GAL1 promotors released from repression (Fig. 5A). We also tested whether the capacity for producing colonies was restored. Cells kept under MR conditions for 48 h that were placed in a releasing liquid medium (YPGalactose) prior to plating on galactose-containing agar plates increased their capacity  Survival of cells grown on 2% galactose and later refreshed again in galactose medium for the duration of the experiment (mock) was considered 100% (357 colonies). Asterisks denote significant statistical differences (p Յ 0.05). C, irreversibility of cell cycle arrest under glucose conditions. YPC3 cells were kept under control (2% galactose), SG, CR, and MR conditions for 24 (SG and CR) or 48 h (MR) prior to being inoculated into fresh complete medium identical to the initial medium (Ϫrelease) or supplemented instead with 2% galactose (ϩrelease). After the indicated times, the ability to form individual colonies was tested. Survival of cells grown on 2% galactose and later refreshed again in galactose medium for the duration of the experiment (mock) was considered 100% (380 colonies) Closed bars,0hinrelease medium; open bars,8hinrelease medium. For all panels, error bars denote S. E. (n ϭ 3). A.U., absorbance units.
to form colonies in a time-dependent manner as opposed to those kept in YPGlycerol (Fig. 5B). This recovery was exclusive of cells that had been subjected to MR conditions. When cells under the SG or CR condition were subjected to an 8-h release in fresh YPGalactose medium, no recovery was observed in their colony forming capacity (Fig. 5C).

Molecular Mechanisms Causing Cell Death under SG and CR
Conditions-Yeast fermentative metabolism is stoichiometrically taut in regard to NAD ϩ /NADH homeostasis. Because NAD ϩ biosynthetic enzymes have long been known to be sensitive to excess PP i (35), we hypothesized that the cell levels of pyridine nucleotide coenzymes could be affected in YPC3 cells under SG, CR, and MR conditions. Indeed, levels of NAD ϩ dropped significantly in YPC3 cells grown for 18 h under SG and CR conditions and for 36 h under MR condition compared with a parental strain grown in the same conditions (Table 1). Concomitantly, YPC3 cells showed decreased NAD ϩ /NADH ratios with respect to wild-type cells when grown under SG and CR conditions. Strikingly, YPC3 MR cells displayed an NAD ϩ / NADH ratio close to that observed in wild-type cells under the same conditions; this is due to a reduction in the levels of NADH that is not observed under SG or CR condition. Depletion of cellular NAD ϩ levels has been described to induce massive autophagy and cell death in mammalian cells (36). To test whether such a mechanism was responsible for the induction of cell death in YPC3 yeast cells under SG and CR conditions, we supplemented the growth culture medium with 10 mM acetaldehyde so that yeast cells could oxidize additional amounts of NADH independently from fermentation. Under these conditions, CR cells showed a 3-fold increase in colony forming capacity, reaching levels close to those found for control cells (Fig. 6A, top panel). Cells kept under SG condition showed a similar behavior (16.3 Ϯ 0.3 and 63.0 Ϯ 3.5% for untreated and acetaldehyde-treated cells 18 h after medium switch, respectively). This effect was accompanied by a drastic reduction in the percentage of cells displaying sub-G 1 characteristics as well as a cell size that was indistinguishable from cells kept in galactose (Fig. 6A, middle and bottom panels, respectively). Moreover, this pattern was repeated in cells transformed with an expression plasmid carrying the ORF of the gene NQR1 encoding a plasma membrane-bound NADH oxidoreductase (19). Cells overexpressing this enzyme oxidize greater amounts of NADH under fermenting conditions in a coenzyme Q-dependent but alcohol dehydrogenase-independent manner. Yeast YPC3 cells transformed with this construct and subjected to caloric restriction conditions again showed clear differences in survival rate, proportion of cells in sub-G 1 , and cell size (Fig. 6B,  top, middle, and bottom panels, respectively). However, under SG, cells displayed smaller differences despite the overexpression of NQR1 (survival figures were 54.4 Ϯ 2.3 and 68.4 Ϯ 3.0% for empty plasmid-bearing and NQR1-overexpressing cells 18 h after medium switch, respectively).
Cell Death Process under SG and CR Conditions-Autophagy has been shown to be a cellular process that under certain circumstances can lead to cell death (37). We investigated the involvement of autophagy in the cell death induction and execution taking place under SG and CR conditions. To ascertain whether autophagy was a cause of death and not simply a con-comitant process, we made use of two YPC3-derived yeast strains affected in their capability to complete autophagy. Strains defective for ATG8 are unable to promote the formation of the autophagosome (38). We also deleted VPH1, an essential subunit of the vacuolar V-ATPase; this H ϩ pump is required for vacuolar acidification and was early shown to be paramount for completion of the autophagic process (39). Cells that had their IPP1 gene under the control of GAL1 promoter and devoid of Atg8p or vacuolar V-ATPase activity showed no loss of viability when placed under SG conditions as measured by their capacity to form colonies (Fig. 7, A and B, top panels). Furthermore, they showed dramatic decreases in sub-G 1 cells under flow cytometry (Fig. 7, A and B, middle panels), and the cell size was reduced to almost control levels when compared with galactose-grown control cells (Fig. 7, A and B, bottom panels). Similar results were obtained under CR conditions (survival figures were 42.5 Ϯ 3.2 and 88.9 Ϯ 7.0% for wild-type and atg8⌬ cells 18 h after medium switch, respectively, on the one hand and 17.2 Ϯ 3.9 and 111.8 Ϯ 6.3% for wild-type and vph1⌬ cells 18 h after medium switch, respectively, on the other hand).

DISCUSSION
Despite intensive research, no thorough information yet exists regarding the cellular implications of some basic metabolic processes. One such field is PP i metabolism. Despite PP i being involved in pathophysiological bone resorption, suspected to play a role in cancer, and on the whole assumed to affect nearly all aspects of cell anabolism, the effects an alteration of PP i homeostasis has at the cellular level are still unknown. Although eukaryotes and bacteria present obvious differences, the scant information available from studies in E. coli is assumed to hold true for eukaryotes. As a result, it is generally considered that an excess of PP i would stop cell growth due to an overall inhibition of anabolism and that this situation would be reversible upon PP i removal from the cell (4). The present work depicts a very different scenario in eukaryotic cells. In yeast, depletion of cytosolic sPPase activity brings cell death but only in the case of cells that obtain their energy from fermentation. This cell death is accompanied by cell cycle arrest in the G 1 /G 0 phase, and both cell death induction and cell cycle arrest are irreversible.
The vast majority of the PP i -producing reactions have been proved to be inhibited by an excess of this metabolite in vitro (1). Moreover, even unrelated reactions are affected due to the side effects of attaining high concentrations of PP i , such as sequestration of Mg 2ϩ ions (1), making elimination of excess PP i a critical issue in cell metabolism. In this regard, the nicotinic acid-mononucleotide adenylyltransferase step in the biosynthesis of NAD ϩ was the first major biochemical reaction described both to produce this metabolite and to be inhibited by it (35). Other PP i -producing biological reactions are also affected by an excess of PP i , for example, the polymerization of macromolecules, such as proteins, RNA, and DNA. However, despite an important and still open discussion on the influence of PP i levels on the fidelity of nucleotide or amino acid incorporation into macromolecules (40,41), it is unknown whether any of them are especially sensitive to excess PP i and therefore candidates for the growth defects observed in bacteria and  yeast. As we show in the present work, under some conditions, one of the cellular processes most sensitive to excess PP i seems to be the homeostasis of nicotinamide. In yeast, fermentation places cells in a delicate redox equilibrium. On the one hand, nicotinamide coenzyme concentrations are low ([NAD ϩ ] ϩ [NADH] is approximately 1 mM), and as a result, most of the cytosolic coenzyme is probably committed to cycle between its oxidized and reduced forms in glycolysis. On the other, this process is stoichiometrically rigid; i.e. the capacity to oxidize NADH by acetaldehyde reduction using alcohol dehydrogenase is equivalent to the amount that was previously reduced by the glyceraldehyde-3-phosphate dehydrogenase, and thus, the cell lacks any flexibility to alter its NAD ϩ /NADH ratio through metabolism. This means that any further needs for NAD ϩ must be covered by biosynthesis of new coenzyme molecules. It is noteworthy that in the cell there are several reactions that consume NAD ϩ and eventually destroy it, e.g. ADP-ribose transferases, cADP-ribose synthases, sirtuins, and in the case of mammals poly(ADP-ribose) polymerases. At least in mammals, under some conditions, such as DNA damage by genotoxic drugs, these processes can even deplete cellular NAD ϩ contents (42). Both known pathways for NAD ϩ biosynthesis, salvage and de novo biosynthesis, include steps where PP i is formed and hence where an excess of this compound can act as an effective inhibitor. Noticeably, the rate-limiting step in NAD ϩ biosynthesis, shared by both de novo and salvage pathways, is the adenylation of nicotinic acid mononucleotide, a reversible reaction among the first ones described to be sensitive to excess PP i (1,35). In addition, the transfer of a phosphoribosyl moiety to yield nicotinamide/nicotinic acid mononucleotide and PP i in the salvage pathway is also reversible (43). The salvage pathway is considered the primary pathway for NAD ϩ synthesis in yeast (44). Which of these steps is primarily responsible for the observed cell death induction in fermenting cells is still unknown. Research on this subject may offer a potential way to control cell proliferation.
Caloric restriction has been shown to lengthen lifespan and affect many cellular processes in nearly all living model systems with the exception of plants (44). The best accepted mechanism relies on the increase of the concentration of NAD ϩ at the expense of NADH. Sirtuins, NAD ϩ -dependent histone deacetylases, are derepressed by a relative excess of NAD ϩ , and gene expression is altered through transcriptional silencing, resulting in changes in lifespan. In yeast, the presence of glucose at a low concentration (typically 0.5%) is enough to induce a mixed metabolism comprising fermentation and respiration that helps maintaining a high NAD ϩ /NADH ratio and increase lifespan (45,46). In the case of cytosolic sPPase depletion, we have shown that increasing NAD ϩ availability in the cell inhibited the induction of autophagy and cell death. Despite this, caloric restriction could not increase survival, extent of cell death induction, or its time course. Noticeably, overexpression of an alternative NADH oxidase, such as Nqr1p, was only able to alleviate cell death under caloric restriction conditions. This suggests that nicotinamide coenzyme homeostasis must be sternly pushed toward NAD ϩ production or NADH reoxidation (e.g. addition of high concentrations of acetaldehyde) to alleviate cell death effectively and that milder strategies, such as caloric restriction and Nqr1p overexpression, are only effective when in combination.
Although cell death is usually accompanied by prior cell cycle arrest, a general inhibitory effect of anabolism by excess PP i could be perfectly understood with an asynchronous arrest of cell functions. Our data show that under both fermentative (SG and CR) and respiratory (MR) conditions cell cycle arrest is part of the cellular responses to cytosolic sPPase depletion. In the case of fermentative conditions, G 0 /G 1 arrest is observed. This could be a response to a deficiency in nicotinamide coenzymes. Subsequent induction of autophagy seems to reinforce this idea. In the case of cells maintained in glycerol as a carbon source, an S phase arrest may suggest that DNA duplication together with nicotinamide coenzyme biosynthesis is one of the biochemical processes most sensitive to an excess of PP i in the cell. Further work is being carried out in our laboratory to shed light on this topic.
All in all, the present report constitutes the first cell physiology study on the alteration of PP i homeostasis in eukaryotes in terms of its effects on cell cycle and cell death induction. These data have clear importance to understand which processes are most likely affected by an excess of cellular PP i and may also be useful to design new alternative therapeutic approaches against diseases like cancer or bone disorders.