Yeast PAH1-encoded phosphatidate phosphatase controls the expression of CHO1-encoded phosphatidylserine synthase for membrane phospholipid synthesis

The PAH1-encoded phosphatidate phosphatase (PAP), which catalyzes the committed step for the synthesis of triacylglycerol in Saccharomyces cerevisiae, exerts a negative regulatory effect on the level of phosphatidate used for the de novo synthesis of membrane phospholipids. This raises the question whether PAP thereby affects the expression and activity of enzymes involved in phospholipid synthesis. Here, we examined the PAP-mediated regulation of CHO1-encoded phosphatidylserine synthase (PSS), which catalyzes the committed step for the synthesis of major phospholipids via the CDP–diacylglycerol pathway. The lack of PAP in the pah1Δ mutant highly elevated PSS activity, exhibiting a growth-dependent up-regulation from the exponential to the stationary phase of growth. Immunoblot analysis showed that the elevation of PSS activity results from an increase in the level of the enzyme encoded by CHO1. Truncation analysis and site-directed mutagenesis of the CHO1 promoter indicated that Cho1 expression in the pah1Δ mutant is induced through the inositol-sensitive upstream activation sequence (UASINO), a cis-acting element for the phosphatidate-controlled Henry (Ino2–Ino4/Opi1) regulatory circuit. The abrogation of Cho1 induction and PSS activity by a CHO1 UASINO mutation suppressed pah1Δ effects on lipid synthesis, nuclear/endoplasmic reticulum membrane morphology, and lipid droplet formation, but not on growth at elevated temperature. Loss of the DGK1-encoded diacylglycerol kinase, which converts diacylglycerol to phosphatidate, partially suppressed the pah1Δ-mediated induction of Cho1 and PSS activity. Collectively, these data showed that PAP activity controls the expression of PSS for membrane phospholipid synthesis.

The Saccharomyces cerevisiae 2 PAH1-encoded PAP 3,4 (EC 3.1.3.4) and CHO1-encoded PSS 5 (EC 2.7.8.8) are two of the most highly regulated enzymes in lipid metabolism (1,2). The PAP reaction is the committed step for the synthesis of the neutral lipid TAG, whereas the PSS reaction is the committed step in the CDP-DAG pathway for the de novo synthesis of the major membrane phospholipids PC and PE (1, 2) (Fig. 1). The DAG derived from the PAP reaction is also used in the CDPcholine and CDP-ethanolamine branches of the Kennedy pathway for the synthesis of PC and PE, respectively, when cells defective in PSS or other enzymes in the CDP-DAG pathway are supplemented with choline or ethanolamine (1, 2) (Fig. 1). For catalytic function, both Cho1 and Pah1 are required to associate with the membrane where their phospholipid substrates reside; Cho1 is an integral membrane enzyme in the ER (3)(4)(5)(6), whereas Pah1 is a peripheral membrane enzyme that translocates from the cytosol to the nuclear/ER membrane (7)(8)(9).
Analyses of the pah1⌬ and cho1⌬ mutants lacking Pah1 and Cho1, respectively, have shed light on the importance of PAP and PSS activities in lipid metabolism and cell physiology. The pah1⌬ mutation increases the level of the PAP substrate PA but decreases the levels of the PAP product DAG and its derivative TAG (10 -12). The lack of Pah1 causes a variety of phenotypes that include the derepression of phospholipid synthesis genes (e.g. INO1, OPI3, and INO2), the increase of phospholipid synthesis, the expansion of the nuclear/ER membrane, the susceptibility to fatty acid-induced toxicity, and the reduction of lipid droplet formation (10 -14). The impact of the Pah1 deficiency on overall cell physiology is further exemplified by the fact that pah1⌬ mutant cells cannot grow on non-fermentable carbon sources (i.e. respiratory deficiency) (10,15) as well as at elevated temperatures (10,13,15). The mutant cells are hypersensitive to oxidative stress, have a shortened chronological life span (16), and exhibit defects in cell wall integrity (17,18) and vacu-ole fusion (19). Some of the pah1⌬ phenotypes require the function of Dgk1 (12,14,20), the CTP-dependent DGK that phosphorylates DAG to form PA (Fig. 1). The cho1⌬ mutant lacks the ability to synthesize PS (21,22) and thus requires the supplementation of choline or ethanolamine to synthesize PC or PE by the Kennedy pathway (1,2). Studies with cells lacking Cho1 have revealed that PS is required for protein kinase C function (23,24), tryptophan transport (25), vacuole function and morphogenesis (26), and direction of endocytic proteins to the plasma membrane (27).
Cho1 and Pah1 are regulated for their functions in lipid metabolism by genetic mechanisms. The expression of CHO1 is elevated in the exponential phase when cells are grown in the absence of the phospholipid precursor molecules inositol, choline, ethanolamine, and serine (28 -32) as well as in the presence of the essential nutrient zinc (33). In contrast, the CHO1 expression is reduced by inositol supplementation in the exponential phase, and this regulation is enhanced by the addition of choline, ethanolamine, or serine in the growth medium (28 -32). The gene expression is also reduced by zinc depletion from the growth medium (33) or when cells progress from the exponential to the stationary phase of growth (34,35). These regulations of the CHO1 expression are mediated by the Henry regulatory circuit, which involves the Ino2-Ino4 complex that drives transcriptional activation through its binding to the UAS INO element and the PA-regulated repressor Opi1, which inhibits the function of the activator complex through its interaction with Ino2 (1,2,36). The CHO1 transcriptional regulation plays a role in the partitioning of CDP-DAG for the synthesis of PI and PS and the PS-derived synthesis of PE and PC in the CDP-DAG pathway (1,2,36).
The expression of PAH1 is also regulated by some of the same growth conditions that regulate the CHO1 expression, but with an opposite effect (37)(38)(39). The transcriptional regulation of PAH1 involves the transcription factors Ino2, Ino4, Opi1, Gis1, and Rph1 (for inositol and growth phase regulation) as well as the transcription factor Zap1 (for zinc-mediated regulation) (37)(38)(39). The induction of the PAH1 transcription in the stationary phase or in response to zinc depletion correlates with the elevation of PAP activity (37,38). On the one hand, the induced expression of PAP activity in zinc-replete stationary phase cells is responsible for increased synthesis and accumu-lation of TAG that occurs at the expense of phospholipid synthesis (37). On the other hand, the induced expression of PAP activity in zinc-depleted exponential phase cells is responsible for increased synthesis of PC via the CDP-choline branch of the Kennedy pathway (38).
The loss of Pah1 and its PAP activity (e.g. pah1⌬ mutation) increases the level of PA and its utilization for the synthesis of membrane phospholipids (10,37,40), raising a question whether the expression of Cho1 and its PSS activity is regulated through cell growth. Here, we showed that the pah1⌬ mutation results in the induction of Cho1 and PSS activity in a growthdependent manner from the exponential to the stationary phase and that the elevation of the CHO1 expression is controlled through the UAS INO element in the promoter. The mutation in the core consensus of UAS INO abolished the induction of Cho1 and PSS activity, resulting in the suppression of the pah1⌬ phenotypes in lipid synthesis, nuclear/ER membrane morphology, and lipid droplet formation. Collectively, this work advances the understanding of how PAP activity regulates phospholipid synthesis through the transcriptional regulation of the PSS enzyme.

The pah1⌬ mutation induces the expression of Cho1 and PSS activity
We examined the effect of growth phase on the expression of CHO1 and its encoded PSS activity in wild type and pah1⌬ mutant cells (Fig. 2). In this work, the level of Cho1 was examined by immunoblotting to analyze the CHO1 expression  because of the caveat that mRNA abundance does not necessarily correlate with protein abundance. The anti-Cho1 antibody recognizes two forms of Cho1 that differ in their electrophoretic mobility (41). The 30-kDa form represents Cho1 phosphorylated by protein kinase A at Ser-46 and Ser-47, whereas the 27-kDa form represents the unphosphorylated form of the enzyme (41). In wild-type cells, the levels of Cho1 ( Fig. 2A) and its PSS activity (Fig. 2B) were reduced in the stationary phase when compared with the exponential phase. As described previously (34), PSS activity was 3.6-fold lower in the stationary phase than in the exponential phase. Compared with wild type, the pah1⌬ mutant had 2.2-fold higher PSS activity in the exponential phase. In addition, the pah1⌬ mutant showed a 2.4-fold induction of the enzyme activity from the exponential to the stationary phase. Consequently, in the stationary phase, the PSS activity of the pah1⌬ mutant was 19-fold higher than the enzyme activity of wild type. The elevation of PSS activity in the pah1⌬ mutant correlated with the increase of the Cho1 level. These results indicate that the induced expression of Cho1 and PSS activity in the pah1⌬ mutant is up-regulated during growth from the exponential to the stationary phase.

Induced expression of Cho1 and PSS activity in the pah1⌬ mutant is mediated by the CHO1 UAS INO element
The up-regulation of the Cho1 level in the pah1⌬ mutant suggested that the enzyme is controlled at the level of its gene expression. The CHO1 promoter contains a UAS INO element as well as other putative regulatory elements. To identify the relevant element by an unbiased approach, we constructed CHO1 alleles that contain different lengths of the promoter by 5Ј-nested deletion and analyzed their expression on a low-copy plasmid in the pah1⌬ mutant. We reasoned that if the plasmidborne CHO1 allele contains a UAS, its expression would be up-regulated like the endogenous expression of Cho1 in the pah1⌬ mutant. We did not analyze the plasmid-borne CHO1 alleles in the cho1⌬ pah1⌬ mutant background. Supplementation of choline or ethanolamine, which is required for viability of cells containing the cho1⌬ mutation, alters the regulation of phospholipid synthesis (21,22). Accordingly, pah1⌬ cells trans-formed with the CHO1 allele were grown to the exponential and stationary phases and were examined for the Cho1 level by immunoblot analysis (Fig. 3). Compared with the pah1⌬ mutant that expresses the endogenous CHO1 (i.e. vector control), its transformants containing the CHO1 allele with the 200-bp or longer promoter exhibited higher levels of Cho1 in both the exponential and the stationary phases of growth (Fig.  3A). However, the increase of the Cho1 level was not shown by the expression of the CHO1 allele with the 100-bp or shorter promoter, indicating that the UAS element is located between Ϫ200 and Ϫ100 of the CHO1 promoter. Further truncation analysis of the 100-bp region showed that the Cho1 level of the pah1⌬ mutant was increased by the CHO1 allele with the 170-bp or longer promoter, but not with the 150-bp promoter (Fig. 3B), indicating that the UAS element affecting expression is located between Ϫ170 and Ϫ150. Analysis of the 20-bp region of the CHO1 promoter revealed that it contains the UAS INO element (Ϫ163 to Ϫ154).
To determine whether the UAS INO element is required for the induction of CHO1 expression in the pah1⌬ mutant, we examined the effect of its mutation (CATGTGAAAG 3 TTTTTTAAAG, mutations underlined) (42) in the plasmidborne CHO1. Compared with the wild type allele, the mutant allele of CHO1 lacking the consensus sequence of UAS INO did not significantly increase the Cho1 level in the pah1⌬ mutant (Fig. 3C). To further confirm this result, the UAS INO mutation was introduced to the chromosomal CHO1 of wild type and the pah1⌬ mutant. Immunoblot analysis (Fig. 4A) showed that the cho1 mutant had a significant reduction of the Cho1 level in the exponential phase. Compared with the pah1⌬ mutant, the cho1 pah1⌬ mutant showed a great reduction of the Cho1 level in the exponential and stationary phases (Fig. 4A). In fact, its Cho1 level was almost identical to that of the cho1 mutant. The effect of the UAS INO mutation on the Cho1 level correlated with the decrease of PSS activity (Fig. 4B). In the exponential phase, PSS activity of the cho1 pah1⌬ mutant was 5.2-and 8.4fold, respectively, lower than that of wild type and the pah1⌬ mutant; in the stationary phase, the enzyme activity was 1.2-and 12-fold, respectively, lower than that of wild type and the pah1⌬ mutant. Taken together, these results indicate that the UAS INO element is responsible for the induction of Cho1 expression in the pah1⌬ mutant and its growth-dependent upregulation from the exponential to the stationary phase.

Effects of the CHO1 UAS INO mutation on lipid composition of the pah1⌬ mutant
We explored the effects of the CHO1 UAS INO mutation on the changes of lipid composition imparted by the pah1⌬ mutation. Yeast cells were grown in SC medium containing [2-14 C]acetate for steady-state labeling of lipids (43,44). The radiolabeled lipids were extracted from exponential and stationary phase cells and analyzed by TLC for total lipids and phospholipids. As described previously (12), the pah1⌬ mutant had a reduced level of TAG (1.8-and 2.4-fold, respectively) but an increased level of phospholipids (2.1-and 7.8-fold, respectively) in the exponential and stationary phases (Fig. 5). The alterations in the levels of TAG and phospholipids were suppressed by the CHO1 UAS INO mutation in the exponential phase ( Fig. 5A) and partially suppressed in the stationary phase (Fig. 5B). In addition, the pah1⌬ mutation caused a high increase in the levels of sterol esters (6.7-fold) and fatty acids (20-fold) in the stationary phase (10,37). These effects of the pah1⌬ mutation were also suppressed by the CHO1 UAS INO mutation (Fig. 5B). As for individual phospholipid species, the pah1⌬ mutant showed an increase in the levels of PC (1.9-and 13.4-fold), PE (3-and 6-fold), PI (2.5-and 7.8-fold), PS (1.5-and 6.4-fold), and PA (2.4-and 6-fold) in the exponential (Fig. 6A) and stationary ( Fig. 6B) phases, respectively. The CHO1 UAS INO mutation suppressed the pah1⌬-mediated increase in the levels of PC, PE, PS, and PA in the exponential phase but had little effect on the level of PI.

Effects of the CHO1 UAS INO mutation on nuclear/ER membrane morphology, lipid droplet formation, and temperature sensitivity of the pah1⌬ mutant
The pah1⌬ mutant has an irregularly shaped nucleus with the expansion of the membrane that is attributed to the increase of phospholipid synthesis (11,13). Because the CHO1 UAS INO mutation exerts a suppressive effect on the pah1⌬mediated increase of phospholipid synthesis, we examined the nuclear/ER membrane morphology of the cho1 pah1⌬ mutant (Fig. 7A). The expression of the Sec63-GFP fusion protein was used to mark the nuclear/ER membrane in the cell (13). Compared with wild type, the cho1 mutant showed little difference in the nuclear/ER membrane morphology and contained a round nucleus. As described previously (13), most of the pah1⌬ cells showed the irregular, expanded nuclear/ER morphology. In contrast, the cho1 pah1⌬ cells (93%; Fig. 7B) had the roundshaped nucleus in the exponential phase, indicating that the CHO1 UAS INO mutation suppresses the nuclear/ER membrane phenotype of the pah1⌬ mutant. The suppressive effect was less strong in the stationary phase, and the number of cho1 pah1⌬ cells containing a round nucleus was reduced by 53%.  Lipids were extracted and separated by the one-dimensional thin-layer chromatography system for neutral lipids, and the images were subjected to ImageQuant analysis. The data are reported as counts/min found in a particular lipid per 10 6 cells. The amount of label found in the lipids from the stationary phase cultures was reduced relative to that found from the exponential phase because of the turnover of lipids during growth and incorporation of the label into other metabolites. The data are means Ϯ S.D. (error bars) from triplicate determinations. Erg, ergosterol; ErgE, ergosterol ester; FA, fatty acid; PL, phospholipids.
We also examined the mutants for the formation of lipid droplets by staining with the fluorescent dye BODIPY 493/503 (Fig. 8A). The wild-type and cho1 cells had a similar number of lipid droplets and showed a ϳ1.5-fold increase in the organelle number from the exponential to the stationary phase (Fig. 8B). Compared with wild type, the pah1⌬ mutant showed a significant reduction (2-and 4.5-fold, respectively) of lipid droplet formation in the exponential and stationary phases. In addition, its lipid droplet formation was not increased in the stationary phase. This phenotype of the pah1⌬ mutant was suppressed by the CHO1 UAS INO mutation, and the cho1 pah1⌬ mutant had a number of lipid droplets similar to that of the wild type in the exponential phase. In the stationary phase, the cho1 pah1⌬ mutant showed a reduction in lipid droplet formation, but its lipid droplet number was still higher than that of wild type. This result indicates that, in the stationary phase of growth, the CHO1 UAS INO mutation has a partial suppressive effect on the defect of the pah1⌬ mutant in lipid droplet formation.
The lack of growth at the elevated temperature (e.g. 37°C) is characteristic of the pah1⌬ mutant (10,13,15). Unlike the phenotypes described above, temperature sensitivity of the pah1⌬ mutant in the exponential and stationary phases was not suppressed by the CHO1 UAS INO mutation (Fig. 9).

Effect of the dgk1⌬ mutation on the induced expression of Cho1 and PSS activity of the pah1⌬ mutant
The lack of Dgk1 (i.e. the dgk1⌬ mutation) suppresses the pah1⌬ phenotypes (e.g. increased phospholipid synthesis, nuclear/ER membrane expansion, and induced INO1 expression) that correlate with an increase in the PA level (20,20). Accordingly, we examined the effect of the dgk1⌬ mutation on the Cho1 expression and PSS activity of wild type and the pah1⌬ mutant (Fig. 10). The dgk1⌬ mutation introduced into wild type did not have a major effect on the expression of Cho1 (Fig. 10A) and PSS activity (Fig. 10B). In contrast, it showed a partial suppressive effect on the Cho1 expression and PSS activity of the pah1⌬ mutant in the stationary phase, but not in the exponential phase. For example, the PSS activity of the dgk1⌬ pah1⌬ mutant in the stationary phase was 32% lower than that of the pah1⌬ mutant. Yet, the PSS activity of the double mutant was 7-fold higher than the enzyme activity of wild type in the stationary phase.
Because both PAP and DGK control the PA level (10,20), we examined the effect of the dgk1⌬ mutation on the PA level of the pah1⌬ mutant. As described previously (12,20), the pah1⌬ mutation caused an increase in the PA content in both the exponential (2.2-fold) and stationary (2.7-fold) phases of growth. The dgk1⌬ mutation suppressed the pah1⌬ effect on PA content (Fig. 10C). We also examined the effect of the dgk1⌬ mutation on temperature sensitivity exhibited by the pah1⌬ mutant. Like the pah1⌬ mutant, the dgk1⌬ pah1⌬ mutant showed no growth at 37°C, indicating that temperature sensitivity of the pah1⌬ mutant is not suppressed by the loss of DGK activity (Fig. 10D).

Discussion
PAP and PSS catalyze reactions in the overlapping pathways of lipid biosynthesis (Fig. 1). The enzyme reactions are linked to the lipid intermediate PA, which is the substrate of PAP and the precursor of CDP-DAG used as the substrate of PSS (Fig. 1). The conversion of PA to DAG and CDP-DAG appears to be a major regulatory step that bifurcates the lipid biosynthetic pathway into two branches (i.e. TAG synthesis and phospholipid synthesis) (10,12,37). The changes in PAP activity are directly correlated with TAG synthesis but are inversely correlated with phospholipid synthesis during cell growth (37). Accordingly, yeast cells have a lower PAP activity in the exponential phase when the rate of phospholipid synthesis is high, whereas they have a higher PAP activity in the stationary phase when the rate of phospholipid synthesis is low (37). The elevation of phospholipid synthesis in the pah1⌬ mutant, which causes the expansion of the nuclear/ER membrane, is related to the increased availability of the precursor PA by the defect of its conversion to DAG as well as by the increased expression of phospholipid synthesis genes (10,12,13). We showed in this study that the expression of CHO1-encoded PSS, which catalyzes the committed step in the CDP-DAG pathway of de novo phospholipid synthesis, is highly induced in the pah1⌬ mutant and that the enzyme induction is increased in the stationary phase when the CHO1 expression in wild type is reduced. That the induction of CHO1 expression in the pah1⌬ mutant was mediated through the UAS INO element supports the notion that the elevation of the PA level reduces the Opi1-mediated transcriptional repression of the UAS INO -containing genes as per the Henry regulatory circuit (Fig. 11). PA is known to sequester Opi1 at the nuclear/ER membrane (2,45), and the phospholipid synthesis genes (e.g. INO1, INO2, and OPI3) previously known to be induced by the pah1⌬ mutation (13) contain the UAS INO element (2,46,47).
The CHO1 UAS INO mutation in the pah1⌬ mutant abolished the up-regulation of the Cho1 level and PSS activity, suppressing the increase of phospholipid synthesis and nuclear/ER membrane as well as the decrease of TAG synthesis and lipid droplet formation. These observations highlight the role of the CHO1-encoded PSS in the synthesis of membrane phospholipids and the role of the PAH1-encoded PAP in the negative regulation of the CHO1 expression. The effects of the CHO1 UAS INO mutation on the pah1⌬ phenotypes were pronounced in the exponential phase of growth when PSS activity is expressed at its highest level (34). The partial suppression of the pah1⌬ phenotypes in the stationary phase suggests that the higher availability of PA may compensate for the reduction of the PSS level.
Although the PA level of the pah1⌬ mutant was reduced by the CHO1 UAS INO mutation in the stationary phase, it was still higher than that of wild type cells (Fig. 6B). This difference in the PA content would be expected to attenuate the Opi1 function. As indicated above, it has been known that transcription of other UAS INO -containing phospholipid synthesis genes (e.g. INO1 and OPI3) is induced in the pah1⌬ mutant (13,48). Although it is unknown whether transcriptional induction of these genes leads to the increased expression and function of their encoded proteins, the lack of total suppression of PC content, especially in the stationary phase, is consistent with an increase in phospholipid methyltransferase activity encoded by OPI3. Phospholipid synthesis genes (e.g. CKI1 (49), CPT1 (50), EKI1 (51), and EPT1 (52)) in the Kennedy pathway are also subject to transcriptional regulation through the UAS INO element, but their encoded activities are not expected to play a  major role in this regulation because growth medium was not supplemented with choline and/or ethanolamine (1,2).
The increase of the PI content in the pah1⌬ mutant was not suppressed by the CHO1 UAS INO mutation in the exponential or stationary phase of growth. Whereas the expression of PIS1encoded PI synthase is not regulated through Opi1 (28,53), the INO1 gene encoding the inositol-3-phosphate synthase, which is crucial for the synthesis of inositol and thus for PI synthesis (54), is regulated by Opi1 (28, 55). Thus, the induced expression of INO1 in response to the pah1⌬ mutation (13) is consistent with the increased availability of inositol for PI synthesis. However, if the induced expression of INO1 is translated to an increase in the level and activity of inositol-3-phosphate synthase, it would not be massive because pah1⌬ mutant cells do not excrete inositol (10).
The pah1⌬ mutation also causes a dramatic increase in the amounts of fatty acids and ergosterol ester, primarily in the stationary phase of growth (10) (Fig. 5). Although the mecha-   (2), and the bifurcation of PA for the synthesis of TAG and phospholipids via the CDP-DAG pathway. Under growth conditions (e.g. exponential phase; left) whereby Pah1 expression (green highlight) and PAP activity are low (37), the level of PA is elevated, and the Opi1 repressor is tethered to the nuclear/ER membrane via its interactions with PA and Scs2. The inhibition of Opi1 for its nuclear entry allows for the transcriptional activation (boldface arrow) of CHO1 by the Ino2-Ino4 complex and the inductions of Cho1 (pink highlight) and PSS activity for increased phospholipid synthesis (large letters) via the CDP-DAG pathway (this work). The reduced rate of TAG synthesis in the exponential phase of growth (37) is indicated with small gray letters. Under growth conditions (e.g. stationary phase; right) whereby Pah1 expression (pink highlight) and PAP activity are high (37), the synthesis of TAG is elevated (large letters) and the PA level is reduced (small gray letters). The reduction of the PA level allows for the dissociation of Opi1 from the nuclear/ER membrane and its entry into the nucleus, where it represses the transcriptional activation of CHO1 (thin gray arrow and letters) by inhibiting the function of the Ino2-Ino4 activator complex through its binding to Ino2. The repression of Cho1 (green highlight) and PSS activity results in the reduction of phospholipid synthesis (small gray letters). nism for these changes is unclear, the elevated levels of the lipids were almost reduced to the levels of wild type by alleviating the induced expression of PSS activity. We hypothesize that the PA content, as regulated by the pah1⌬ and cho1 pah1⌬ mutations, is the basis for the changes in the levels of fatty acids and ergosterol ester. This assertion is supported by the fact that fatty acid synthesis, as mediated by the ACC1, FAS1, and FAS2 genes, is under the control of the Henry regulatory circuit (56,57) and that the ARE1 and ARE2 genes, which encode the acyl-CoA sterol acyltransferase enzymes, contain a putative UAS INO element in their promoters. Accordingly, additional studies are warranted to addresses the expression regulation of those genes.
The lack of the DGK1-encoded DGK suppresses the pah1⌬ phenotypes that are ascribed to the increase of the PA level (20,58), which include the induced expression of UAS INO -containing genes and increased phospholipid content, nuclear/ER membrane expansion, decreased lipid droplet formation, and reduced chronological life span (10, 12-14, 16, 20, 59). As discussed above, we consider that the increase of the PA level is responsible for induced expression of Cho1 and PSS activity in the pah1⌬ mutant. In further support of this, we showed that the dgk1⌬ mutation suppressed the pah1⌬ mutant for its induced Cho1 expression and PSS activity, but only in the stationary phase of growth with a partial effect. Thus, the PA-mediated regulation of the CHO1 expression is more complex than was expected.
Of the pah1⌬ phenotypes examined in this study, temperature sensitivity was not suppressed by the CHO1 UAS INO mutation or by the dgk1⌬ mutation. These observations support the notion that the inability of the pah1⌬ mutant to grow at the elevated temperature is not related to a change in the PA level but rather to other changes (e.g. decrease in DAG production) caused by the loss of PAP activity. Additional pah1⌬ phenotypes that are not affected by the dgk1⌬ mutation include the decrease of TAG synthesis, sensitivity to fatty acid toxicity (12), and vacuole fragmentation (60).
In addition to the genetic regulation discussed above, the PAP and PSS activities of Pah1 and Cho1, respectively, are biochemically regulated by the substrates and products of their reactions. The PSS substrate CDP-DAG stimulates PAP activity by a mechanism that increases its affinity for PA (61), whereas the PAP substrate PA stimulates PSS activity by a mechanism that increases its affinity for CDP-DAG (62). Moreover, the PAP product DAG inhibits PSS activity by a noncompetitive mechanism (62). The two enzymes are also regulated by phosphorylation. Protein kinase A phosphorylates Pah1 (63) and Cho1 (41,64) and inhibits their PAP and PSS, activities, respectively. For Cho1, the posttranslational modification stabilizes its abundance in the cell for the net effect of stimulating the synthesis of PS (41). The phosphorylationmediated regulations of the two enzymes are more complex and are discussed elsewhere (1). In addition, the regulations of Pah1 and Cho1 are interrelated with the metabolism of nucleotides and sphingolipids (1,2,65).
In summary, the work reported here advances our understanding of how the PAP enzyme reaction regulates phospholipid synthesis through the transcriptional regulation of the PSS enzyme. Clearly, the long-term genetic regulations, coupled with the short-term biochemical regulations of the PAP and PSS enzymes are critical to controlling the balance between the synthesis of TAG for stasis and the synthesis of membrane phospholipids for cell growth.

Materials
All chemicals were reagent grade or better. Culture medium components were from BD Difco. DNA purification kits were from Qiagen. Restriction endonucleases, modifying enzymes, and DNA polymerases were from New England Biolabs. DNA size standards (1 Kb Plus DNA ladder) were from Invitrogen. Oligonucleotides, nucleotides, protease inhibitors, Triton X-100, Ponceau S stain, Aerococcus viridans glycerol-3-phosphate oxidase, horseradish peroxidase, and bovine serum albumin were from Sigma-Aldrich. Carrier DNA for yeast transformation was from Clontech. Bradford protein assay reagent, protein size standards, and electrophoretic reagents were from Bio-Rad. PVDF membrane and the enhanced chemifluorescence substrate for Western blotting were from GE Healthcare. Alkaline phosphatase-conjugated goat anti-rabbit IgG antibodies (product 31340, lot NJ178812), alkaline phosphataseconjugated goat anti-mouse IgG antibodies (product 31322, lot PB1815636), Triton X-100 (Surfact-Amps, product 28314), Amplex Red, and BODIPY 493/503 were from Thermo Fisher Scientific. Pseudomonas sp. lipoprotein lipase was from Wako. Radiochemicals were from PerkinElmer Life Sciences. Phospholipids were from Avanti Polar Lipids. Liquid scintillation mixtures were from National Diagnostics. Silica gel 60 TLC plates were from EMD Millipore, and LK5D partisil silica gel TLC plates were from Whatman.

Strains and growth conditions
The strains used in this work are listed in Table 1. Yeast cells were grown at 30°C in YEPD medium (1% yeast extract, 2% peptone, 2% glucose) or in SC medium containing 2% glucose (66). For selection of yeast cells containing plasmids, appropriate amino acids were omitted from SC medium. Escherichia coli DH5␣ was used for plasmid maintenance and amplification and was grown at 37°C in LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl, pH 7.4). E. coli transformants containing plasmids were selected by ampicillin (100 g/ml) resistance. Solid media for the growth of E. coli and yeast contained agar at a concentration of 1.5 and 2%, respectively. Cell numbers in liquid cultures were estimated spectrophotometrically by measuring absorbance at 600 nm.

DNA manipulations
Standard methods were used for the isolation of plasmid and yeast genomic DNA and for the manipulation of DNA with restriction enzymes, DNA ligase, and modifying enzymes (66,67). PCRs were optimized as described by Innis and Gelfand (68). Yeast and E. coli transformations were performed by standard protocols. DNA sequencing reactions were performed according to the dideoxy method with Taq DNA polymerase and analyzed by automated DNA sequencer (service provided by GENEWIZ, South Plainfield, NJ).

Construction of yeast mutants
The cho1⌬::URA3 mutant was derived from the yeast strain W303-1A by one-step gene replacement (69). The W303-1A strain was transformed with a 2.9-kb cho1⌬::URA3 disruption cassette that was released from pGH442 by digestion with SacI and HindIII, and the resulting transformants were selected on the SC-Ura medium supplemented with 1 mM choline. The cho1⌬ mutant was identified from the Ura ϩ transformants by choline auxotrophy and by PCR analysis of gene disruption. The cho1 mutant, which contains a defective CHO1 UAS INO element, was derived from the cho1⌬::URA3 mutant by onestep gene replacement. The cho1⌬ mutant was transformed with a 2.3-kb CHO1 DNA that was released from pGH440m by digestion with SacI and HindIII, and the resulting transformants were selected on 5-fluoroorotic acid-containing medium. The gene replacement in the 5-fluoroorotic acidresistant transformant was confirmed by DNA sequencing of the PCR-amplified CHO1. The pah1⌬ cho1 mutant was derived from the cho1⌬ mutant by one-step gene replacement. The  cho1 mutant was transformed with a 3-kb pah1⌬::URA3 disruption cassette that was released from pGH317 by digestion with the XbaI and HindIII. The pah1⌬ mutation was confirmed from the Ura ϩ transformants by temperature sensitivity and PCR analysis of gene disruption.

Preparation of cell extracts
All steps were performed at 4°C. Yeast cell pellets were resuspended in breaking buffer (50 mM Tris-HCl, pH 7.5, 0.3 M sucrose, 10 mM 2-mercaptoethanol, 0.5 mM phenylmethanesulfonyl fluoride, 1 mM benzamidine, 5 g/ml aprotinin, 5 g/ml leupeptin, and 5 g/ml pepstatin). Glass beads (0.5-mm diameter) were added to the cell suspension, and the mixture was vigorously agitated using the Mini-BeadBeater-16 (Biospec Products). Unbroken cells and glass beads were precipitated by centrifugation at 1,500 ϫ g for 10 min, and the supernatant was transferred to a new tube for use as cell extracts. Protein concentration of cell extracts was estimated by the method of Bradford (70) using bovine serum albumin as the standard.

SDS-PAGE and immunoblot analysis
SDS-PAGE (71) and immunoblotting (72)(73)(74) using PVDF membrane were performed as described previously. The samples for immunoblotting were normalized to total protein loading. Ponceau S staining was used to monitor the protein transfer from the polyacrylamide gels to the PVDF membrane. Immunoblot analysis of the protein Cho1 was performed with anti-Cho1 antibody raised against the N-terminal portion of the protein (41) at a concentration of 0.25 g/ml. Goat antirabbit IgG antibody conjugated with alkaline phosphatase was used as a secondary antibody at a dilution of 1:5,000. Immune complexes were detected on immunoblots using the enhanced chemifluorescence Western blotting reagents as described by the manufacturer. Images were acquired by fluorimaging analysis. Immunoblotting signals were in the linear range of detectability.

Enzyme assays
All assays were conducted at 30°C in a total volume of 0.1 ml. PSS activity was measured by following the incorporation of water-soluble [3-3 H]serine (10,000 cpm/nmol) into chloroform-soluble [3-3 H]PS (75)(76)(77). The enzyme reaction contained 50 mM Tris-HCl (pH 8.0), 0.6 mM MnCl 2 , 4 mM Triton X-100, 0.2 mM CDP-DAG, and 0.5 mM serine (75)(76)(77). The enzyme assays were conducted in triplicate, and the average S.D. of the assays was Ϯ5%. The reactions were linear with time and protein concentration. A unit of PSS activity was defined as the amount of enzyme that catalyzed the formation of 1 nmol of product/min.

Radiolabeling and analysis of lipids
The steady-state labeling of lipids with [2-14 C]acetate was performed as described previously (43). Lipids were extracted (78) from the radiolabeled cells and then separated by onedimensional TLC for neutral lipids (79) or phospholipids (80). The resolved lipids were visualized by phosphorimaging and quantified by ImageQuant software using a standard curve of [2-14 C]acetate. The identity of radiolabeled lipids was con-firmed by comparison with the migration of authentic standards visualized by staining with iodine vapor.

Analysis of PA
PA was analyzed by the fluorometric coupled enzyme assay of Morita et al. (81) with minor modifications (82). The cellular lipids were extracted (78), solubilized with purified Triton X-100 (Surfact-Amps), and digested with lipoprotein lipase. The PA derived from glycerol 3-phosphate was coupled to the formation of resorufin with glycerol-3-phosphate oxidase to produce hydrogen peroxide, which is reduced by peroxidase using Amplex Red.

Microscopy
For nuclear/ER membrane morphology analysis, cells were grown at 30°C and collected at the exponential and stationary phases of growth. The cells were resuspended in a reduced volume of the same medium. For the analysis of lipid droplets, cells were grown in the same medium and collected at the exponential and stationary phases, stained for 30 min with 2 M BODIPY 493/503, and washed with phosphate-buffered saline (pH 7.4). The average number of cells with normal nuclear/ER membrane structure (i.e. round-to oval-shaped circle) or the number of lipid droplets per cell was scored from Ն3 fields of view (Ն150 cells). The fluorescence images were observed under a microscope (Nikon Eclipse Ni-U) with a long pass green fluorescent protein filter, captured by the DS-Qi2 camera with the imaging software NIS-Elements BR.

Analyses of data
SigmaPlot software was used for the statistical analysis of data. The p values Ͻ 0.05 were taken as a significant difference.