Yeast Pah1p Phosphatidate Phosphatase Is Regulated by Proteasome-mediated Degradation*

Background: Yeast Pah1p phosphatidate phosphatase is most active in stationary phase to produce diacylglycerol for triacylglycerol synthesis. Results: Pah1p is degraded in stationary phase and stabilized by proteasome inhibitor MG132, mutations in proteasome function and ubiquitination, and elevated levels of phosphatidate. Conclusion: Pah1p is regulated by proteasome-mediated degradation. Significance: Regulation of Pah1p stability/degradation is important to lipid homeostatic mechanisms. Yeast PAH1-encoded phosphatidate phosphatase is the enzyme responsible for the production of the diacylglycerol used for the synthesis of triacylglycerol that accumulates in the stationary phase of growth. Paradoxically, the growth phase-mediated inductions of PAH1 and phosphatidate phosphatase activity do not correlate with the amount of Pah1p; enzyme abundance declined in a growth phase-dependent manner. Pah1p from exponential phase cells was a relatively stable protein, and its abundance was not affected by incubation with an extract from stationary phase cells. Recombinant Pah1p was degraded upon incubation with the 100,000 × g pellet fraction of stationary phase cells, although the enzyme was stable when incubated with the same fraction of exponential phase cells. MG132, an inhibitor of proteasome function, prevented degradation of the recombinant enzyme. Endogenously expressed and plasmid-mediated overexpressed levels of Pah1p were more abundant in the stationary phase of cells treated with MG132. Pah1p was stabilized in mutants with impaired proteasome (rpn4Δ, blm10Δ, ump1Δ, and pre1 pre2) and ubiquitination (hrd1Δ, ubc4Δ, ubc7Δ, ubc8Δ, and doa4Δ) functions. The pre1 pre2 mutations that eliminate nearly all chymotrypsin-like activity of the 20 S proteasome had the greatest stabilizing effect on enzyme levels. Taken together, these results supported the conclusion that Pah1p is subject to proteasome-mediated degradation in the stationary phase. That Pah1p abundance was stabilized in pah1Δ mutant cells expressing catalytically inactive forms of Pah1p and dgk1Δ mutant cells with induced expression of DGK1-encoded diacylglycerol kinase indicated that alteration in phosphatidate and/or diacylglycerol levels might be the signal that triggers Pah1p degradation.

Insights into the synthesis, turnover, and regulation of membrane phospholipids and lipid droplet storage lipids have come from studies using the model eukaryote yeast Saccharomyces cerevisiae (1)(2)(3)(4)(5). PAH1-encoded PAP 2 (Pah1p), which catalyzes the Mg 2ϩ -dependent dephosphorylation of PA to form DAG and P i (6), has emerged as a key lipid homeostatic enzyme in yeast (1,2,7). Pah1p is responsible for production of the DAG used for de novo synthesis of TAG and membrane phospholipids, as well as the DAG that facilitates the formation of lipid droplets (8 -13). In addition, Pah1p is responsible for controlling the levels of PA that affect growth of the nuclear/ER membrane, assist membrane-associated Scs2p in excluding the Opi1p repressor from the nucleus where it attenuates expression of phospholipid synthesis genes, and that regulate trafficking of proteins and enzymes required for vacuole homeostasis and fusion (1, 2, 14 -18). APP1 (11), DPP1 (19), and LPP1 (20), which also encode PAP activity in S. cerevisiae, are not involved in de novo lipid synthesis (7). Instead, these activities presumably serve other cellular functions that include lipid signaling and endocytosis (7,11,21).
Pah1p is conserved throughout evolution with counterpart enzymes in humans (8,22), mice (23,24), flies (25,26), worms (27), and plants (28,29). In fact, the discovery that PAH1 encodes a PAP enzyme in yeast (8,30) made studies on the orthologous PAP enzymes in higher eukaryotic organisms possible (7,21). The importance of understanding the genetic and biochemical regulations of PAP expression and function, respectively, is highlighted by the observations that defects in mammalian lipin PAP enzymes underlie metabolic disorders that include obesity, lipodystrophy, peripheral neuropathy, myoglobinuria, and inflammation (31).
Much is known about the biochemical regulation of Pah1p from S. cerevisiae (7,21). Its PAP activity is dependent on the DIDGT catalytic motif within a haloacid dehalogenase-like domain in the enzyme (8,15). Anionic phospholipids (e.g. CDP-DAG and phosphatidylinositol) stimulate PAP activity through a mechanism that increases the enzyme's affinity for PA (32), whereas sphingoid bases (e.g. sphinganine and phytosphingosine) inhibit activity by a mechanism that excludes PA from the enzyme (33). The nucleotides ATP and CTP inhibit PAP activity by a complex mechanism that affects both the V max and K m values for PA and by chelation of the cofactor Mg 2ϩ (34). Given the activation/inhibition constants and the cellular concentrations of these effector molecules, their regulatory roles in controlling PAP activity and lipid metabolism are thought to be physiologically relevant (21).
Perhaps the most important manner by which Pah1p function is modulated involves phosphorylation and dephosphorylation and the translocation of the enzyme from the cytosol to the membrane (7,21). Pah1p is a relatively abundant lipid metabolic enzyme that is primarily found in the cytosol as a phosphoprotein (16,(35)(36)(37)(38), and it is a substrate for multiple protein kinase phosphorylations (e.g. Pho85p-Pho80p, Cdc28p-cyclin B, protein kinase A, protein kinase C, and casein kinase II) that cause a loss of enzyme function (14,16,(35)(36)(37)39). For Pah1p to interact with its membrane-bound substrate PA, it must translocate to the nuclear/ER membrane where it is dephosphorylated by the Nem1p-Spo7p protein phosphatase complex (14, 16, 35-37, 40, 41), a process facilitated by an acidic domain at the C-terminal region of Pah1p (42). The dephosphorylation of Pah1p leads to its binding to the membrane via an N-terminal amphipathic helix that allows dephosphorylation of PA to generate DAG (41). In addition to controlling Pah1p location, phosphorylation by some protein kinases (e.g. Pho85p-Pho80p and protein kinase A) directly inhibits PAP activity, whereas dephosphorylation stimulates activity (16,35,37).
PAH1 expression is regulated at the transcriptional level in response to growth nutrients. PAH1 is induced throughout growth in cells grown with glucose as the carbon source, and in the stationary phase of growth the induction in the expression of PAH1 is enhanced by inositol supplementation (10). These regulations are mediated by the Ino2p, Ino4p, and Opi1p regulatory circuit and by the transcription factors Gis1p and Rph1p (10). The expression of PAH1 is also induced in the exponential phase by zinc deficiency through a mechanism that involves the Zap1p zinc-responsive transcription factor (43). Additionally, microarray data indicate that PAH1 is induced upon transition from glucose-based fermentative growth to glycerol-and ethanol-based respiratory growth (44). The growth phase-and zincmediated inductions of PAH1 expression correlate with elevated levels of PAP activity (10,43). On the one hand, the induced expression of PAP activity in zinc-depleted exponential phase cells is responsible for increased synthesis of phosphatidylcholine via the DAG-based CDP-choline (e.g. Kennedy) pathway (43). On the other hand, the induced expression of PAP activity in zinc-replete stationary phase cells is responsible for increased synthesis and accumulation of TAG that occurs at the expense of membrane phospholipid synthesis (10).
Paradoxically, the zinc-and growth phase-mediated inductions of PAH1 and PAP activity do not correlate with Pah1p abundance (10,43). In fact, Pah1p is hardly detectable when its PAP activity is greatest (10,35,36,43). In this work, we explored the basis for this enigma with respect to growth phase regulation. We found that Pah1p is a reasonably stable protein in exponential phase cells, but it was degraded by a proteolytic activity present in stationary phase cells that was prevented by the proteasome inhibitor MG132. An analysis of Pah1p abundance in mutants defective in proteasome function indicated that the enzyme is subject to degradation by the proteasome. Pah1p abundance was also stabilized in cells with elevated PA content.

EXPERIMENTAL PROCEDURES
Materials-All chemicals were reagent grade. Difco was the source of growth medium supplies. The yeast deletion consortium strains were purchased from Invitrogen. DNA gel extraction and plasmid DNA purification kits were obtained from Qiagen. The Yeastmaker TM transformation kit was from Clontech. Radiochemicals were from PerkinElmer Life Sciences. National Diagnostics supplied the scintillation counting supplies and acrylamide solutions. Electrophoresis reagents and molecular mass standards were from Bio-Rad. Polyvinylidene difluoride membrane and the enhanced chemifluorescence Western blotting detection kit were purchased from GE Healthcare. Thermo Scientific, Invitrogen, Pierce, and Roche Applied Science were the suppliers of alkaline phosphataseconjugated goat anti-rabbit antibodies, mouse anti-phosphoglycerate kinase (Pgk1p) antibodies, alkaline phosphatase-conjugated goat anti-mouse antibodies, and mouse anti-HA antibodies, respectively. Avanti Polar Lipids and EM Science were the sources of lipids and silica gel 60 TLC plates. Sigma was the source of ampicillin, aprotinin, benzamidine, carbobenzoxy-L-leucyl-L-leucyl-L-leucinal (MG132), cycloheximide, leupeptin, 2-mercaptoethanol, pepstatin, phenylmethylsulfonyl fluoride, SDS, and Triton X-100.
Strains and Growth Conditions- Table 1 contains a list of the strains and plasmids used in this work. Plasmid amplification and maintenance were performed in Escherichia coli strain DH5␣. E. coli cells were grown at 37°C in LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl, pH 7). Ampicillin (100 g/ml) was added to select for cells carrying plasmids. S. cerevisiae cells were routinely grown at 30°C in YEPD medium (1% yeast extract, 2% peptone, 2% glucose) or in standard synthetic complete (SC) medium supplemented with 2% glucose (45,46). The appropriate amino acids were omitted from SC medium for the selection of plasmids (45,46). The SC medium used for the growth of strain BY4741 and its mutant derivatives was modified for optimum growth (47). For Pah1p stability studies, cycloheximide (100 g/ml) was added to SC medium to arrest translation. The proteasome inhibitor MG132 (50 M) was added to cultures in 0.1% dimethyl sulfoxide. Cells expressing DGK1 under control of the GAL1/10 promoter were grown to exponential phase in SC medium with 2% raffinose as the carbon source; DGK1 expression was induced by supplementation with 2% galactose. Cell numbers in liquid cultures were determined spectrophotometrically at an absorbance of 600 nm. The growth medium was supplemented with agar (2% for yeast, 1.5% for E. coli) for growth on plates.
Preparation of Cell Extracts and Subcellular Fractionation-All steps were performed at 4°C. Cell-free extracts were prepared by disruption of cells with glass beads (0.5-mm diameter) in a Biospec Products Mini BeadBeater-16 (48). The disruption buffer contained 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 0.3 M sucrose, 10 mM 2-mercaptoethanol, 1 mM NaF, and a mixture of protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 5 g/ml aprotinin, 5 g/ml leupeptin, and 5 g/ml pepstatin). Glass beads and cell debris were removed by centrifugation at 1,500 ϫ g for 5 min to obtain the cell extract. The cell extract was centrifuged at 100,000 ϫ g for 1 h. The pellet was resuspended in the cell disruption buffer to the same volume as the supernatant fraction. Protein concentration was determined by the Coomassie Brilliant Blue dye-binding method of Bradford (49) using bovine serum albumin as the standard.
PAP Activity Assay-PAP activity was measured by following the generation of water-soluble 32 P i from chloroform-soluble [ 32 P]PA for 20 min at 30°C (48). [ 32 P]PA was enzymatically synthesized from DAG and [␥-32 P]ATP using E. coli DAG kinase, and the radioactive product of the reaction was purified by TLC (48). The PAP reaction mixture contained 50 mM Tris-HCl buffer, pH 7.5, 1 mM MgCl 2 , 0.2 mM [ 32 P]PA (10,000 -15,000 cpm/nmol), 2 mM Triton X-100, and enzyme protein in a total volume of 0.1 ml. PAP activity assays were conducted in triplicate, and the average standard deviation was Ϯ5%. The reactions were linear with time and protein concentration. A unit of PAP activity was defined as the amount of enzyme that catalyzed the dephosphorylation of 1 nmol of PA/min. Specific activity was defined as units/mg of protein.
Western Blot Analysis-SDS-PAGE (50) and Western blotting (51,52) with polyvinylidene difluoride membrane were performed by standard protocols. Rabbit anti-Pah1p (36) and mouse anti-Pgk1p antibodies were used at a concentration of 2 g/ml. The dilutions used for rabbit anti-Dgk1p (53) and mouse anti-HA antibodies were 1:1,500 and 1:1,000, respectively. Secondary alkaline phosphatase-conjugated goat antirabbit and anti-mouse IgG antibodies were used at a dilution of 1:5,000. Immune complexes were detected using the enhanced chemifluorescence Western blotting detection kit. Fluorimaging was used to acquire images from immunoblots, and the densities of the images were quantified with ImageQuant software. Signals were in the linear range of detectability.
Labeling and Analysis of Lipids-Steady-state labeling of lipids with [2-14 C]acetate was performed as described previously (54). Lipids were extracted from labeled exponential and stationary phase cells by the method of Bligh and Dyer (55). Lipids were analyzed by one-dimensional TLC on silica gel plates (56). The identity of radiolabeled lipids on TLC plates was confirmed by comparison of their migration with that of standards after exposure to iodine vapor. Radiolabeled lipids were visualized by phosphorimaging analysis and were quantified using ImageQuant software.
Data Analyses-The Student's t test (SigmaPlot software) was used to determine statistical significance, and p values Ͻ 0.05 were taken as a significant difference.

Pah1p Abundance Declines in a Growth Phase-dependent
Manner-Recent studies have shown that PAH1 expression and Pah1p PAP activity are induced throughout growth, and this regulation is responsible for the synthesis and accumulation of TAG in the stationary phase (10). Here, we examined the effect of growth on Pah1p abundance. As described previously (10), wild type cells exhibited a growth phase-dependent induction in PAP activity (Fig. 1A). Unexpectedly, the induced expression of PAP activity did not correlate with an increase in enzyme abundance. Instead, the levels of Pah1p decreased when wild type cells progressed from the exponential to stationary phases of growth (Fig. 1B). The levels of PAP activity and Pah1p abundance were also examined in pah1⌬ dpp1⌬ lpp1⌬ mutant cells expressing PAH1 HA from multicopy plas- mid pGH312. In this way, the contributions of DPP1 and LPP1 that also add to the increase in PAP activity during growth (10) were eliminated, and at the same time, the levels of Pah1p were overexpressed to enhance detection with anti-Pah1p antibodies. The pah1⌬ dpp1⌬ lpp1⌬ mutant possesses the APP1 gene that encodes a fourth PAP activity (11), but as described previously (10), this gene did not contribute significantly to the growth phase-mediated induction in PAP activity (Fig. 1C). The overexpression of PAH1 correlated with an increased level of PAP activity in the exponential phase when compared with the endogenous activity of wild type cells (Fig. 1A), and the overexpressed level of activity was induced in a growth phase-dependent manner (Fig. 1C). Similarly to that observed in wild type cells, the abundance of overexpressed Pah1p declined in a growth phase-dependent manner (Fig. 1D). The anti-Pah1p antibodies used in this study are directed against the C-terminal portion of the protein (36). Thus, we considered the possibility that the inability to observe Pah1p during the latter stages of growth was due to proteolytic cleavage of the C-terminal residues for which the antibodies were raised. Previous work has shown that a C-terminal truncation of Pah1p retains PAP activity (8); therefore, it is possible that a truncated form of Pah1p may still be present and active in stationary phase cells while escaping detection with the anti-Pah1p antibodies. To address this possibility, the blots with samples derived from the cells expressing N-terminal HAtagged Pah1p from plasmid pGH312 were also probed with anti-HA antibodies. This analysis, however, revealed the same growth phase-dependent loss of Pah1p as that observed on the Western blot using the anti-Pah1p antibodies (data not shown).
Despite the protease inhibitors present in the cell disruption buffer, the high concentration of proteases in stationary phase cells (57,58), coupled with the high temperature used to denature proteins for SDS-PAGE, could lead to increased degradation of Pah1p. To address this possibility, equal amounts of extracts from exponential and stationary phase cells were mixed and subjected to Western blot analysis. The Pah1p signal detected in the mixed sample was equal in intensity to that found for the exponential phase extract (Fig. 1E). Thus, the extract from stationary phase cells did not possess a proteolytic activity that degraded Pah1p from exponential phase cells.
We examined the stability of Pah1p in the exponential phase of growth where the protein could still be detected. Wild type cells were treated with cycloheximide to arrest translation; extracts were prepared, and the levels of Pah1p were analyzed by Western blotting. Despite the decrease in Pah1p abundance in the later phases of growth, there was only a slight increase in the decay rate of Pah1p with respect to growth phase (Fig. 2). The half-life of Pah1p in cells grown for 10, 12, and 14 h was 10, 14, and 16% shorter, respectively, than that of cells grown for 8 h.
A Protease Activity from the 100,000 ϫ g Pellet Fraction of Stationary Phase Cells Degrades Purified Recombinant Pah1p and the Degradation Is Prevented by the Proteasome Inhibitor MG132-Extracts from exponential and stationary phase cells were prepared in the absence of protease inhibitors and fractionated into the 100,000 ϫ g supernatant and pellet fractions. Incubation of recombinant Pah1p with the fractions from exponential phase cells did not have a major effect on the stability of the purified enzyme (Fig. 3A). However, incubation with the 100,000 ϫ g pellet from stationary phase cells resulted in degradation of Pah1p that was dependent on the time of incubation (Fig. 3A) and the amount of the pellet fraction used in the incubation (Fig. 3B). Because of the fact that 20 S protea-  somes sediment with the 100,000 ϫ g pellet fraction of eukaryotic cells (59), we questioned whether the degradation of the recombinant Pah1p was due to the proteolytic activity of proteasomes. MG132 is a known inhibitor of proteasome activity (60), and accordingly, we tested if this reagent would affect the stability of Pah1p. Indeed, MG132 prevented the time-dependent degradation of Pah1p in a concentration-dependent manner (Fig. 3C).
Pah1p Abundance in Stationary Phase Cells Is Stabilized by the Proteasome Inhibitor MG132-We sought evidence that MG132 could prevent the degradation of Pah1p in vivo. For this experiment, we utilized the pdr5⌬ mutant that lacks the multidrug efflux pump (61). The pdr5⌬ mutation prevents efflux of MG132 and thus facilitates the inhibition of the proteasome degradation pathway in vivo (60,62). The effect of MG132 on Pah1p stability was examined in pdr5⌬ mutant cells expressing PAH1 HA on multicopy plasmid pGH312. As discussed above, the overexpression of Pah1p facilitated its detection by Western blot analysis. MG132 treatment prevented the growth phase-dependent degradation of Pah1p (Fig. 4B). Because MG132 caused a reduction in the growth of pdr5⌬ mutant cells (Fig. 4A), we questioned whether the increase in Pah1p stability might be influenced by the reduction in cell numbers. Therefore, in another approach, pdr5⌬ mutant cells (without plasmid pGH312) were grown to late exponential phase prior to treatment with MG132, and the abundance of the endogenously expressed Pah1p was examined after the treatment. This regime also eliminated a possible concern that the effectiveness of MG132 is reduced upon long incubation periods (60). The cell density was not significantly affected after MG132 was introduced into the culture medium (Fig.  4C), and the treatment stabilized the levels of Pah1p in the cells (Fig. 4D).

Pah1p Abundance Is Stabilized in Mutants with Impaired Proteasome and Ubiquitination Functions-Because
Pah1p was stabilized in the presence of the proteasome inhibitor MG132, we examined the abundance of the enzyme in mutants (rpn4⌬, blm10⌬, ump1⌬, and pre1 pre2) that are known to exhibit defects in proteasome function (63)(64)(65)(66). Rpn4p is a transcriptional activator that is required for the induction of proteasome genes (67,68). Blm10p is an activator of the enzymatic activity of the core particle of the proteasome and is required for its maturation (64, 69 -71). Ump1p is a maturation factor required for coordination of proteasome assembly and activation (65). Pre1p and Pre2p contribute to the chymotrypsin-like activity of the proteasome (66,72). With the exception of the pre1 pre2 mutant, whose isogenic parent was strain WCG4a, the genetic background of the mutants was strain BY4741. We noted that the amount of Pah1p in the exponential phase varied among the mutant strains (Fig. 5A). We also noted that the relative amount of Pah1p in the stationary phase of strain BY4741 (25 Ϯ 7%) was slightly greater than that observed for wild type strain WCG4a (17 Ϯ 2%). Because of these differences, we assessed the effect of each mutant on Pah1p stability by comparing the amount of enzyme in the stationary phase relative to the amount of enzyme in the exponential phase. The rpn4⌬, blm10⌬, ump1⌬, and pre1 pre2 mutations caused increases in the amount of Pah1p in the stationary phase relative to the amount in the exponential phase (Fig. 5A). Quantification of the data in Fig. 5A indicated that the increase in Pah1p abundance in the stationary phase of the mutants relative to the amount of enzyme in the stationary phase of the wild type parent strains was 200, 225, 304, and 840%, respectively (Fig. 5B).
Effect of the pre1 pre2 Mutations on PAP Activity and Lipid Composition-Given that Pah1p was most stabilized in the stationary phase of pre1 pre2 mutant cells, we questioned what effect this would have on PAP activity and lipid composition. The PAP activity in the stationary phase was not any greater than the induced level of activity that is normally shown in wild type cells (Figs. 1A and 6B) (10,85). As described previously for wild type cells (8,10,86), the amount of TAG increased (600%) in the stationary phase when compared with the amount in the exponential phase (Fig. 6C). Although the amount of Pah1p in the stationary phase of pre1 pre2 cells was much greater (840%) than that found in the stationary phase of wild type cells (Figs. 5 and 6A) and despite the fact that Pah1p is the only PAP enzyme responsible for the synthesis and accumulation of TAG that occurs in the stationary phase (10), the growth phase-mediated increase in TAG content observed for the mutant was only 200% (Fig. 6C). Also, the growth phase-mediated decrease (57%) in phospholipid content of the pre1 pre2 mutant was greater than that observed (25%) for the wild type control (Fig.  6C). Consequently, the amounts of TAG and phospholipids in the stationary phase of pre1 pre2 mutant cells were both 50% less than that found in the wild type control. The most striking effect of the pre1 pre2 mutation on lipid composition was the fatty acid content (Fig. 6C). The amount of fatty acids in the stationary phase was 12-fold greater when compared with that in the exponential phase; however, for the wild type control, the increase in fatty acid content between the stationary and exponential phase was only 40% (Fig. 6C).
Effects of the D398E and D400E Mutations in Pah1p and the Induced Expression of Dgk1p on Pah1p Abundance-Because PA content, as mediated by Pah1p PAP activity, regulates cel-lular processes that include nuclear/ER membrane expansion and the transcriptional regulation of phospholipid synthesis genes (8,14,15,87), we examined the hypothesis that PA might also influence the process of Pah1p degradation. The abundance of Pah1p was examined in pah1⌬ mutant cells that express the catalytic site D398E and D400E mutations that abrogate the PAP activity of the enzyme (15). In these cells, the PA level is elevated due to the pah1⌬ mutation and resulting loss of PAP activity, and at the same time, Pah1p is still present for its examination by Western blot analysis (15). The amounts of the D398E and D400E mutant enzymes were stabilized in the stationary phase when compared with the wild type enzyme (Fig. 7A). As discussed above, the stability of Pah1p was quantified by comparing the amount of enzyme in the stationary phase relative to the amount of enzyme in the exponential phase. The increases in Pah1p abundance in the stationary phase of the D398E and D400E mutants relative to the amount of enzyme in the stationary phase of the control were 227 and 262%, respectively (Fig. 7B).  were quantified with ImageQuant software. For the wild type control and each mutant, the relative amount of Pah1p/Pgk1p in exponential phase was arbitrarily set at 100%, and the amount of enzyme in stationary phase was relative to that in the exponential phase. The data shown in the figure is the average of three independent experiments Ϯ S.D. (error bars). C, dgk1⌬ mutant cells bearing plasmid YCplac111-GAL1/10-DGK1 or vector control were grown in SC medium with 2% raffinose to the exponential phase. 2% galactose was then added to the culture medium to induce the expression of DGK1. Extracts were prepared from the cells at the point when galactose was added to the growth medium (exponential phase, E) and 24 h after supplementation (stationary phase, S). Anti-Pah1p, anti-Pgk1p (loading control), or anti-Dgk1p antibodies were used for Western blotting.  (115). D, levels of Pah1p and Pgk1p were quantified with ImageQuant software. The relative amount of Pah1p/Pgk1p in exponential phase was arbitrarily set at 100%, and the amount of enzyme in stationary phase was relative to that in the exponential phase. The data shown in the figure is the average of three independent experiments Ϯ S.D. (error bars).
The effect of elevated PA content on Pah1p was also examined in dgk1⌬ mutant cells that expressed DGK1 from the inducible GAL1/10 promoter. DGK1 encodes a CTP-dependent DAG kinase that catalyzes the conversion of DAG to PA (53,87). Cells that overexpress Dgk1p (as directed by induction of GAL1/10-DGK1) exhibit an increase in PA content at the nuclear/ER membrane (41,87). The abundance of Pah1p was examined before and after the induction of DGK1. The induced expression of Dgk1p correlated with a greater amount of Pah1p in stationary phase cells when compared with stationary phase cells without DGK1 induction (Fig. 7C). The amount of Pah1p in the stationary phase of the induced cells was 700% more that the amount of enzyme in stationary phase cells with the vector control (Fig. 7D).

DISCUSSION
The expression of PAH1 and its encoded PAP activity, along with the concomitant synthesis and accumulation of TAG, are greatest when cells progress from the exponential to stationary phases of growth (8 -10, 86). Yet, when the gene is most highly expressed and the enzyme is most active, we were unable to observe a corresponding increase in Pah1p abundance. We reasoned that the basis for this enigma was the degradation of Pah1p by a proteolytic activity present in stationary phase cells. The Pah1p expressed and purified from E. coli was degraded upon incubation with the 100,000 ϫ g pellet fraction of stationary phase cells, whereas the enzyme was not degraded when incubated with the same fraction of exponential phase cells. MG132, an inhibitor widely used to examine protein degradation via the proteasome (60), prevented the degradation of the recombinant enzyme. Moreover, the effect of MG132 on stabilizing Pah1p abundance was recapitulated in vivo; both endogenously expressed and plasmid-mediated overexpressed levels of Pah1p were more abundant in the stationary phase of cells treated with MG132. Because of the stabilizing effect of MG132, we examined Pah1p abundance in mutants defective in proteasome function and ubiquitination. Although most of the mutations resulted in the stabilization of Pah1p in the stationary phase, the pre1 pre2 mutations that eliminate nearly all (Ͼ95%) of the chymotrypsin-like activity of the 20 S proteasome (66,72) had the greatest effect on enzyme levels. This result supported the notion that Pah1p is subject to degradation by way of the proteasome. The ubiquitin ligase mutations did not stabilize Pah1p to the same extent as that of the mutants defective in proteasome function. However, this might be explained if the ubiquitin ligase enzymes have overlapping functions with respect to the ubiquitination of Pah1p, whereas mutants lacking the ubiquitin isopeptidase Doa4p are unable to maintain normal levels of ubiquitin (63,81,82). Additional studies are needed to confirm a ubiquitinated form of Pah1p and identify the sites of modification.
Why is Pah1p degraded if its PAP activity is so important to the synthesis of TAG in the stationary phase? Lack of PAP activity (e.g. due to pah1⌬ or nem1⌬ mutations) results in a dramatic loss (ϳ90%) in TAG content and several deleterious phenotypes (8, 9, 12-15, 17, 88). Yet, an excess of PAP activity (e.g. due to the overexpression of phosphorylation-deficient Pah1p or overexpression of the Nem1p-Spo7p protein phosphatase complex) causes a lethal phenotype (14,16,36). This lethality is presumably due to the loss of PA needed for phospholipid synthesis and/or the toxic effects of DAG (8,36,89). These observations point to the fact that PAP activity must be fine-tuned to control the PA/DAG balance at the nuclear/ER membrane (7). In fact, the lethal phenotype caused by excess PAP activity is suppressed by the overexpression of DGK1-encoded DAG kinase activity (87), whereas some deleterious phenotypes caused by the lack of PAP activity are suppressed by the loss of DAG kinase activity (e.g. pah1⌬ dgk1⌬ double mutation) (7,87). We posit that proteasome-mediated degradation of Pah1p alleviates the toxic effects of excess PAP activity and plays a role in the complex mechanisms that maintain the PA/DAG balance.
The pre1 pre2 mutant has been a useful tool for studying proteasome-mediated degradation of both ubiquitinated and nonubiquitinated proteins (66, 90 -94). Because Pah1p was stabilized in this mutant, we examined the effects of the pre1 pre2 mutations on PAP activity and lipid composition. We expected that the TAG content might be elevated in the mutant; instead, the mutations had the opposite effect, as well as causing a decrease in phospholipids and a massive increase in fatty acids. Moreover, the elevated amount of Pah1p in pre1 pre2 mutant cells did not yield a commensurate increase in PAP activity. However, we were not surprised as it is the location of the enzyme and not its abundance that determines its in vivo function. This indicated that a compensatory mechanism (e.g. increase in phosphorylation or decrease in dephosphorylation) might have attenuated the PAP activity of the stabilized enzyme to preserve normal cell physiology. This attenuation in PAP activity in the stationary phase of the mutant might be a contributing factor for why the TAG content was not elevated when compared with the control. In addition, the dramatic increase in fatty acids indicated that the pre1 pre2 mutations most likely affected additional enzymes in lipid metabolism. For example, the inhibition of acyltransferases used for the synthesis of PA and TAG and/or the stimulation of lipases used for the hydrolysis of phospholipids and TAG could account for the alterations in lipid content of the mutant. That Hmg2p hydroxymethylglutaryl-CoA reductase (95) and Ole1p ⌬-9 fatty acid desaturase (96) are also subject to proteasome degradation indicates that this form of regulation is more broadly applicable to lipid metabolism in yeast. Further substantiating this idea, studies documenting the proteomics of ubiquitination in exponential phase yeast have identified multiple proteins involved in lipid metabolism (e.g. Opi1p, Scs2p, Gpt2p, Slc1p, and Cho1p) (97). Pah1p was not identified as a ubiquitinated protein in the study; as indicated above, its degradation occurred in the late exponential to stationary phase of growth.
Previous work has shown that phosphorylation/dephosphorylation regulates Pah1p stability/degradation (35,37). Pah1p is subject to multiple phosphorylations (16,(35)(36)(37)39) that presumably occur in the exponential phase of growth, and at least some of these phosphorylations (e.g. by Pho85p-Pho80p and protein kinase A) stabilize enzyme abundance (35,37). It is also known that in late exponential phase cells, a phosphorylationdeficient form (7 alanine mutations of the Pho85p-Pho80p phosphorylation sites) of Pah1p is less stable than wild type Pah1p, although a hyperphosphorylated form of the enzyme in nem1⌬ mutant cells lacking Nem1p-Spo7p protein phosphatase activity is more stable than the wild type enzyme (14,35,37). Thus, the phosphorylation of Pah1p in the exponential phase provides an explanation as to why the mixing of extracts from exponential and stationary phase cells did not result in the degradation of Pah1p from exponential phase cells. Additionally, the Pah1p was expressed in E. coli and therefore not subject to endogenous phosphorylations and was not protected from degradation by the 100,000 ϫ g pellet fraction of stationary phase cells. The fact that Pah1p in exponential phase cells was a relatively stable protein with a half-life of Ͼ70 min was consistent with the notion that phosphorylation prevents the degradation of the enzyme. Overall, these observations support the hypothesis that the dephosphorylated and more active form of Pah1p is degraded by the proteasome. It is known that the processes of phosphorylation/dephosphorylation, ubiquitination, and proteasome degradation may be closely related (98). Future studies will examine these possible relationships.
In addition to serving as lipid metabolic intermediates, the substrate and product of the PAP reaction are signaling molecules that regulate a variety of cellular processes (99 -108). In particular, PA exerts its signaling functions via diverse mechanisms, such as direct binding and modulation of enzymatic activity, membrane tethering, and through changes in membrane structure (103,104). A variety of PA targets have been identified in mammalian cells that include protein kinases (e.g. protein kinase C and mechanistic target of rapamycin) and lipid kinases (e.g. phosphatidylinositol-4-phosphate 5-kinase and sphingosine kinase) (103,107). In S. cerevisiae, PA targets the transcriptional repressor Opi1p and the SNARE protein Spo20p and stimulates Cho1p phosphatidylserine synthase activity (1,18,103,109). Because of the importance of PA as a signaling molecule in S. cerevisiae (1,2,18), we considered the possibility that it might govern the abundance of Pah1p. That Pah1p was at least partially stabilized in pah1⌬ mutant cells expressing catalytically inactive forms of Pah1p or dgk1⌬ mutant cells overexpressing Dgk1p DAG kinase activity supports the hypothesis that PA content signals enzyme stability/ degradation. How this signaling occurs is unknown, but we envisage that PA levels might regulate components of the proteasome degradation pathway, the protein kinase and phosphatase activities that modulate Pah1p abundance through phosphorylation/dephosphorylation, and/or the process by which phosphorylated Pah1p is recruited to the membrane for its dephosphorylation (Fig. 8). In this working model, the PA signal that controls Pah1p stability/degradation is ultimately governed by its own PAP activity.