The Saccharomyces cerevisiae Lipin Homolog Is a Mg2+-dependent Phosphatidate Phosphatase Enzyme*♦

Mg2+-dependent phosphatidate (PA) phosphatase (3-sn-phosphatidate phosphohydrolase, EC 3.1.3.4) catalyzes the dephosphorylation of PA to yield diacylglycerol and Pi. In this work, we identified the Saccharomyces cerevisiae PAH1 (previously known as SMP2) gene that encodes Mg2+-dependent PA phosphatase using amino acid sequence information derived from a purified preparation of the enzyme (Lin, Y.-P., and Carman, G. M. (1989) J. Biol. Chem. 264, 8641–8645). Overexpression of PAH1 in S. cerevisiae directed elevated levels of Mg2+-dependent PA phosphatase activity, whereas the pah1Δ mutation caused reduced levels of enzyme activity. Heterologous expression of PAH1 in Escherichia coli confirmed that Pah1p is a Mg2+-dependent PA phosphatase enzyme and showed that its enzymological properties were very similar to those of the enzyme purified from S. cerevisiae. The PAH1-encoded enzyme activity was associated with both the membrane and cytosolic fractions of the cell, and the membrane-bound form of the enzyme was salt-extractable. Lipid analysis showed that mutants lacking PAH1 accumulated PA and had reduced amounts of diacylglycerolanditsderivativetriacylglycerol.ThePAH1-encoded Mg2+-dependent PA phosphatase shows homology to mammalian lipin, a fat-regulating protein whose molecular function is unknown. Heterologous expression of human LPIN1 in E. coli showed that lipin 1 is also a Mg2+-dependent PA phosphatase enzyme.

ylsulfonyl fluoride, benzamidine, aprotinin, leupeptin, pepstatin, isopropyl ␤-D-thiogalactopyranoside, and Triton X-100 were purchased from Sigma. Lipids were obtained from Avanti Polar Lipids. Silica gel thin-layer chromatography plates were from EM Science. Protein assay reagents, electrophoretic reagents, and protein standards were purchased from Bio-Rad. Mouse monoclonal anti-HA antibodies were from Roche Applied Science. Alkaline phosphatase-conjugated goat anti-mouse antibodies were purchased from Pierce. Hybond-P polyvinylidene difluoride membrane and the enhanced chemifluorescence Western blotting detection kit were purchased from Amersham Biosciences. Scintillation counting supplies were purchased from National Diagnostics.
Strains and Growth Conditions-The strains used in this work are listed in Table 1. Yeast cells were grown in YEPD medium (1% yeast extract, 2% peptone, 2% glucose) or in synthetic complete (SC) medium containing 2% glucose at 30°C as described previously (31,32). For selection of yeast cells bearing plasmids, appropriate amino acids were omitted from SC medium. The growth medium used for the analysis of lipids and for testing the inositol excretion phenotype lacked inositol (33). Growth of the ino1 mutant was used as an indicator of inositol excretion, and the inositol excreting opi1 mutant (34) was used as a positive control. Plasmid maintenance and amplifications were performed in Escherichia coli strain DH5␣, and protein expression was performed in the strain BL21(DE3)pLysS. E. coli cells were grown in LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl, pH 7.4) at 37°C, and ampicillin (100 g/ml) was added to select for the bacterial cells carrying plasmids. For growth on solid media, agar plates were prepared with supplementation of either 2% (yeast) or 1.5% (E. coli) agar. Cell numbers in liquid media were determined spectrophotometrically at an absorbance of 600 nm. For heterologous expression of yeast PAH1 and human LPIN1, E. coli BL21(DE3)pLysS cells bearing pGH313 and pGH318, respectively, were grown to A 600 ϭ 0.5 at room temperature in 500 ml of LB medium containing ampicillin (100 g/ml) and chloramphenicol (34 g/ml). The culture was incubated for 1 h with 1 mM isopropyl ␤-Dthiogalactopyranoside to induce the expression of His 6 -tagged Pah1p and lipin 1.
DNA Manipulations, Amplification of DNA by PCR, and DNA Sequencing-Standard methods were used to isolate plasmid DNA and yeast genomic DNA and for manipulation of DNA using restriction enzymes, DNA ligase, and modifying enzymes (32). Yeast (35) and E. coli (32) transformations were performed by standard protocols. PCR reactions were optimized as described by Innis and Gelfand (36). DNA sequencing reactions were performed by the dideoxy method using Taq DNA polymerase (32) and analyzed by automated DNA sequencer.
Construction of Plasmids-The plasmids used in this work are listed in Table 1. The S. cerevisiae PAH1 gene (SMP2/YMR165C) was cloned by PCR. A 3.8-kb DNA fragment that contains the entire coding sequence (2.6 kb) of PAH1, the 5Ј-untranslated region (0.7 kb), and the 3Ј-untranslated region (0.5 kb) was amplified from the genomic DNA template of S. cerevisiae strain BY4742. The PAH1 DNA fragments were digested with XbaI/SphI and inserted into plasmid YEp351 at the same restriction enzyme sites. The multicopy plasmid containing PAH1 was named pGH311. The PAH1 gene was used to construct PAH1 HA , in which the sequence for an HA epitope tag (YPYDVPDYA) was located after the start codon. The 0.7-and 3.1-kb PAH1 DNA fragments that contain the HA tag at the 3Ј and 5Ј ends, respectively, were amplified by PCR. These DNA fragments were digested with XbaI/AatII and AatII/ SphI, respectively, and inserted into YEp351 at the XbaI/SphI sites. The multicopy plasmid containing PAH1 HA was named pGH312. For expression of PAH1 in E. coli, the entire coding sequence of PAH1 was amplified by PCR using plasmid pGH311 as a template. The PCR product (ϳ2.6 kb) was digested with HindIII to produce 0.6-and 2.0-kb DNA fragments, which were then digested with NdeI and XhoI, respectively. The NdeI-HindIII and HindIII-XhoI DNA fragments were inserted into pET-15b at the NdeI/XhoI sites. The E. coli expression plasmid containing the His 6 -tagged PAH1 was named pGH313. For expression of human LPIN1 (accession number NM_145693) in E. coli, the entire coding sequence of the gene was amplified from a full-length LPIN1 cDNA clone (OriGENE Technologies, Inc.) by PCR using primers with add-on restriction enzyme sites (MluI before start codon/XhoI after stop codon). The PCR product (2.7 kb) was digested with MluI, filled with Klenow, and digested with XhoI. The LPIN1 DNA fragment was ligated with pET-15b that was digested with NdeI, filled with Klenow, and digested with XhoI. The E. coli expression plasmid containing the His 6 -tagged LPIN1 was named pGH318.
Construction of the pah1⌬ Mutant and the pah1⌬ dpp1⌬ lpp1⌬ Triple Mutant-URA3 DNA (1.4 kb) was amplified from plasmid pRS406 by PCR using primers with add-on restriction enzyme sites. The PCR products were digested with Tth111I and SpeI and inserted into the plasmid pGH311 that was digested with the same restriction enzymes to remove 80% of the PAH1 coding sequence. The resulting plasmid, which contains a 3-kb PAH1 deletion cassette (pah1⌬::URA3), was named pGH317. Deletion of the PAH1 gene in the yeast chromosome was performed by the method of one-step gene replacement (37). The PAH1 deletion cassette was released from plasmid pGH317 by digestion with XbaI and SphI and used to transform strains W303-1A and TBY1. The resulting transformants were selected on SC-uracil medium. Disruption of the PAH1 gene in uracil prototrophs was examined by PCR analysis of genomic DNA using primers that flank the inserted URA3 gene. Preparation of the Cytosolic and Membrane Fractions from S. cerevisiae-All steps were performed at 4°C. Yeast cells were suspended in 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. Cells were disrupted with glass beads (0.5 mm diameter) using a Biospec Products Mini-BeadBeater-8 as described previously (38). Unbroken cells and glass beads were removed by centrifugation at 1,500 ϫ g for 10 min. The cell lysate was centrifuged at 100,000 ϫ g for 1 h to separate cytosolic (supernatant) from membrane fractions (pellet). The membranes were suspended in the same buffer used to disrupt cells. Protein concentration was estimated by the method of Bradford (39) using bovine serum albumin as the standard.
Purification of His 6 -tagged Pah1p and Human Lipin 1-All steps for protein purification were performed at 4°C. E. coli cells expressing His 6 -tagged Pah1p and lipin 1 were washed once with 20 mM Tris-HCl, pH 8.0, buffer and suspended in 20 ml of 20 mM Tris-HCl, pH 8.0, buffer containing 0.5 M NaCl, 5 mM imidazole, and 1 mM phenylmethylsulfonyl fluoride. Cells were disrupted by a freeze-thawing cycle and by two passes through a French press at 20,000 pounds/square inch. Unbroken cells and cell debris were removed by centrifugation at 12,000 ϫ g for 30 min at 4°C. The supernatant (cell lysate) was gently mixed with 2 ml of 50% slurry of Ni 2ϩ -NTA-agarose for 2 h. The Ni 2ϩ -NTA-agarose/enzyme mixture was packed in a 10-ml Poly-Prep column and washed with 20 ml of 20 mM Tris-HCl, pH 8.0, buffer containing 0.5 M NaCl, 45 mM imidazole, 10% glycerol, and 7 mM 2-mercaptoethanol. His 6 -tagged proteins were then eluted from the column in 1-ml fractions with a total of 5 ml of 20 mM Tris-HCl, pH 8.0, buffer containing 0.5 M NaCl, 250 mM imidazole, 10% glycerol, and 7 mM 2-mercaptoethanol. Enzyme preparations were dialyzed against 20 mM Tris-HCl, pH 8.0, buffer containing 10% glycerol and 7 mM 2-mercaptoethanol and stored at Ϫ80°C. (40) and immunoblotting (41) using PVDF membrane were performed as described previously. For detection of the HA-tagged Pah1p, mouse monoclonal anti-HA antibodies were used at a dilution of 1:1,000. Goat anti-mouse IgG-alkaline phosphatase conjugates were used as secondary antibodies at a dilution of 1:5,000. The HA-tagged proteins were detected on immunoblots using the enhanced chemifluorescence Western blotting detection kit as described by the manufacturer. Images were acquired by FluorImaging analysis.

SDS-PAGE and Immunoblot Analysis-SDS-PAGE
Preparation of Labeled Substrates-[ 32 P]PA was synthesized enzymatically from DAG and [␥-32 P]ATP with E. coli DAG kinase as described by Carman and Lin (42). [␤-32 P]Diacylglycerol pyrophosphate was synthesized enzymatically from PA and [␥-32 P]ATP with Catharanthus roseus PA kinase as described by Wu et al. (19).
Enzyme Assays-Mg 2ϩ -dependent PA phosphatase activity was measured by following the release of water-soluble 32 P i from chloroform-soluble [ 32 P]PA (10,000 cpm/nmol) as described by Carman and Lin (42). The reaction mixture contained 50 mM Tris-HCl buffer, pH 7.5, 1 mM MgCl 2 , 0.2 mM PA, 2 mM Triton X-100, and enzyme protein in a total volume of 0.1 ml. Diacylglycerol pyrophosphate phosphatase activity was measured by following the release of water-soluble 32 P i from chloroform-soluble [␤-32 P]diacylglycerol pyrophosphate (10,000 cpm/ nmol) as described by Wu et al. (19). The reaction mixture contained 50 mM citrate buffer, pH 5.0, 0.1 mM diacylglycerol pyrophosphate, 2 mM Triton X-100, and enzyme protein in a total volume of 0.1 ml. All enzyme assays were conducted in triplicate at 30°C. The average standard deviation of the assays was Ϯ 5%. The reactions were linear with time and protein concentration. A unit of enzymatic activity was defined as the amount of enzyme that catalyzed the formation of 1 nmol of product per minute.
Labeling and Analysis of Lipids-Steady-state labeling of phospholipids and neutral lipids with 32 P i and [2-14 C]acetate, respectively, were performed as described previously (43)(44)(45)(46). Lipids were extracted from labeled cells by the method of Bligh and Dyer (47). Phospholipids were analyzed by two-dimensional thin-layer chromatography on silica gel plates using chloroform/methanol/ammonium hydroxide/water (45:25:2:3, v/v) as the solvent system for dimension one and chloroform/ methanol/glacial acetic acid/water (32:4:5:1, v/v) as the solvent system for dimension two (48). Neutral lipids were analyzed by one-dimensional thin-layer chromatography on silica gel plates using the solvent system hexane/diethyl ether/glacial acetic acid (40:10:1, v/v) (49). The identity of the labeled lipids on thin-layer chromatography plates was confirmed by comparison with standards after exposure to iodine vapor. Radiolabeled lipids were visualized by phosphorimaging analysis. The relative quantities of labeled lipids were analyzed using Image-Quant software.
Data Analysis-Kinetic data were analyzed with the EZ-FIT enzyme kinetic model-fitting program (50), and statistical analyses were performed with SigmaPlot software. p values Ͻ 0.05 were taken as a significant difference.

RESULTS
Identification of the S. cerevisiae PAH1 Gene Encoding a Mg 2ϩ -dependent PA Phosphatase Enzyme-Mg 2ϩ -dependent PA phosphatase was partially purified (through the Mono Q chromatography step) from sodium cholate-treated membranes as described by Lin and Carman (22). An SDS-polyacrylamide gel slice containing a protein with a molecular mass of 91 kDa was subjected to trypsin digestion followed by amino acid sequence analysis of peptide fragments by matrix-assisted laser desorption ionization tandem time-of-flight mass spectrometry. 3 Unambiguous amino acid sequence information obtained from 23 peptides ( Fig. 1) matched perfectly with the deduced amino acid sequence of SMP2 (YMR165C) in the Saccharomyces Genome Data Base. We renamed this gene PAH1 (phosphatidic acid phosphohydrolase) because its protein product has the molecular function of a PA phos-

PAH1-encoded Mg 2؉ -dependent PA Phosphatase
phatase (phosphohydrolase) enzyme. The PAH1 sequence is found on chromosome XIII and does not have any sequence motifs suggesting introns in the gene, and its deduced protein product consists of 862 amino acids. Pah1p contains a conserved haloacid dehalogenase (HAD)-like domain (51,52) in the middle of the protein sequence ( Fig.  1), which contains a DXDXT (residues 398 -402) motif that is found in a superfamily of Mg 2ϩ -dependent phosphatase enzymes with diverse substrate specificity (51,52).
To examine the hypothesis that PAH1 encodes a Mg 2ϩ -dependent PA phosphatase, we examined the levels of the enzyme activity from cell extracts derived from exponential phase cells. In the pah1⌬ mutant, the Mg 2ϩ -dependent PA phosphatase activity was reduced by 34% when compared with the activity found in the wild type parent ( Fig. 2A). We also examined the effect of the pah1⌬ mutation on Mg 2ϩ -dependent PA phosphatase activity in the dpp1⌬ lpp1⌬ mutant to eliminate the contributions of the DPP1-encoded and LPP1-encoded Mg 2ϩ -independent PA phosphatase activities that are still active under the assay conditions (e.g. 1 mM MgCl 2 ) used for the Mg 2ϩ -dependent activity (22,24,29). The Mg 2ϩ -dependent PA phosphatase activity in the cell extract of pah1⌬ dpp1⌬ lpp1⌬ triple mutant cells was 30% lower that the activity in the dpp1⌬ lpp1⌬ double mutant ( Fig. 2A). The remaining Mg 2ϩ -dependent PA phosphatase activity in the triple mutant must be attributed to yet another gene that codes for a PA phosphatase enzyme.
The multicopy plasmid containing the PAH1 gene directed a 4-fold overexpression of PA phosphatase activity when compared with dpp1⌬ lpp1⌬ cells not bearing the plasmid ( Fig. 2A). As would be expected, the PA phosphatase activity directed by the PAH1 gene was dependent on the presence of Mg 2ϩ ions in the assay buffer. A PAH1 HA allele was constructed and cloned into a multicopy plasmid for identification of Pah1p by immunoblotting. The HA-tagged version of the enzyme was functional and exhibited the same levels of Mg 2ϩ -dependent PA phosphatase activity in the dpp1⌬ lpp1⌬ mutant as the untagged enzyme. Immunoblot analysis showed that the HA-tagged PA phosphatase (Pah1p HA ) migrated as a 124-kDa protein upon SDS-PAGE (Fig. 2B).
The 91-kDa protein that was used to identify the PAH1 gene was isolated from the membrane fraction of yeast (22). Yet, Pah1p does not contain any transmembrane-spanning regions. Localization studies with a Pah1p fused with green fluorescent protein indicated that the Pah1p is present throughout the cytoplasm (18). Given this information, the association of Mg 2ϩ -dependent PA phosphatase activity with the cytosolic and membrane fractions of the cell was examined. As described previously (14,15), most (66%) of the membrane-associated PA phosphatase activity in wild type cells was attributed to the DPP1 and LPP1 gene products ( Fig. 2A). Analysis of Mg 2ϩ -dependent PA phosphatase activity in the pah1⌬ and pah1⌬ dpp1⌬ lpp1⌬ mutants and in the dpp1⌬ lpp1⌬ mutant overexpressing the PAH1 gene showed that the PAH1-encoded enzyme was found in both the cytosolic and membrane fractions ( Fig. 2A). About 70% of the PAH1-encoded Mg 2ϩdependent PA phosphatase present in the membrane fraction was extracted with 0.5 M NaCl (Fig. 3). This result indicated that the PAH1encoded enzyme could associate with membranes as a peripheral membrane protein (53). The membrane-associated Mg 2ϩ -dependent PA phosphatase activity present in the pah1⌬ dpp1⌬ lpp1⌬ triple mutant was also salt-extractable (Fig. 3).
Mg 2ϩ -dependent and -independent forms of PA phosphatase have been characterized as being sensitive or insensitive to the thioreactive agent NEM (12,20). We examined the effect of NEM on the Mg 2ϩ -dependent PA phosphatase activity in the cytosolic fraction of wild type, dpp1⌬ lpp1⌬, and pah1⌬ dpp1⌬ lpp1⌬ cells. NEM inhibited the PA phosphatase activity in wild type cells and in dpp1⌬ lpp1⌬ mutant cells by 26 and 27%, respectively (Fig. 4). In the pah1⌬ dpp1⌬ lpp1⌬ triplet mutant, 90% of the PA phosphatase was inhibited by NEM (Fig. 4). These results indicated that the PAH1-encoded PA phosphatase was an NEM-insensitive enzyme whereas the remaining PA phosphatase activity in the triplet mutant was an NEM-sensitive enzyme.
Heterologous Expression of the PAH1-encoded Mg 2ϩ -dependent PA Phosphatase in E. coli-The overexpression of Mg 2ϩ -dependent PA phosphatase activity in dpp1⌬ lpp1⌬ cells bearing the PAH1 gene on a multicopy plasmid indicated that PAH1 encodes a Mg 2ϩ -dependent PA phosphatase enzyme. However, this result did not rule out the possibility that the PAH1 gene was a regulatory gene whose product controlled the expression or activities of PA phosphatase enzymes. We used heterologous expression of the yeast PAH1 gene in E. coli to test the hypothesis that the PAH1 gene was the structural gene encoding a Mg 2ϩdependent PA phosphatase enzyme. The purified His 6 -tagged Pah1p migrated as a 114-kDa protein upon SDS-PAGE (Fig. 5A). This protein catalyzed the dephosphorylation of PA in a protein concentration-dependent manner (Fig. 5B). The specific activity of the recombinant PA phosphatase enzyme was 3,000 nmol/min/mg. If we assume that the specific activity of the PAH1-encoded Mg 2ϩ -dependent PA phosphatase in the cell extract of yeast is 0.5 nmol/min/mg (based on the data in Figs. 2 and 4), then the specific activity of the purified recombinant enzyme represents a 6,000-fold enrichment of the enzyme.
We examined the basic enzymological properties of the purified recombinant Mg 2ϩ -dependent PA phosphatase enzyme. Optimum enzyme activity was found at pH 7.5 (Fig. 6A). No enzyme activity was observed when MgCl 2 was omitted from the standard reaction mixture (Fig. 6B). PA phosphatase activity exhibited a dose-dependent requirement for MgCl 2 with maximum activity at a final concentration of 1 mM (Fig. 6B). NEM (1-20 mM) did not affect the activity of the enzyme. The effect of Triton X-100 on Mg 2ϩ -dependent PA phosphatase activity is shown in Fig. 6C. The addition of Triton X-100 to the assay mixture resulted in the apparent inhibition of activity characteristic of surface dilution kinetics (54). The function of Triton X-100 in the assay for Mg 2ϩ -dependent PA phosphatase (22,23) as well as many other lipiddependent enzymes is to form a mixed micelle with the lipid substrate providing a surface for catalysis (54). Since the PAH1-encoded Mg 2ϩdependent PA phosphatase exhibited surface dilution kinetics, the kinetic analysis of the enzyme was performed using Triton X-100/PAmixed micelles. Accordingly, the concentration of PA in the mixed micelles was expressed as a surface concentration in mol % as opposed to a molar concentration (54). In this experiment, the enzyme was measured such that Mg 2ϩ -dependent PA phosphatase activity was only dependent on the surface concentration of PA (i.e. at a molar PA concentration of 0.2 mM) and independent of the molar concentration of PA (54). As described for the Mg 2ϩ -dependent PA phosphatase purified from yeast (25), the enzyme exhibited positive cooperative kinetics with respect to the surface concentration of PA (Fig. 6D). Analysis of the kinetic data according to the Hill equation yielded a Hill number of 3 and a K m value for PA of 3 mol %. A major distinction between the Mg 2ϩ -dependent and -independent PA phosphatase enzymes is the ability of the Mg 2ϩ -independent enzymes to utilize a variety of lipid phosphate substrates such as diacylglycerol pyrophosphate (12,14,15,20). As described previously for the Mg 2ϩ -dependent PA phosphatase purified from yeast (19), the PAH1-encoded enzyme did not utilize diacylglycerol pyrophosphate as a substrate.
Phenotypic Properties of the pah1⌬ Mutant-A mutation in PAH1 (SMP2) was first identified in a screen for mutants that exhibit increased stability of heterologous plasmids (55). More recently, Santos-Rosa et al. (56) have shown that pah1⌬ (smp2⌬) mutants have enlarged, irregularly shaped nuclei with projections that associate with the peripheral ER. This phenotype has been attributed to increased membrane phospholipid synthesis because the INO1, INO2, and OPI3 phospholipid synthesis genes are derepressed in the pah1⌬ (smp2⌬) mutant background (56). The derepression of INO1 in the pah1⌬ (smp2⌬) mutant prompted us to examine the mutant for the inositol excretion phenotype (34). Inositol excretion is due to the overexpression of INO1-encoded inositol-3-phosphate synthase activity and massive production of inositol (57)(58)(59). However, the pah1⌬ mutant did not exhibit the inositol excretion phenotype. As described previously (56), the pah1⌬ mutant grew more slowly than wild type cells at 30°C and exhibited a temperature-sensitive phenotype at 37°C (Fig. 7). Interestingly, the dpp1⌬ lpp1⌬ double mutant, which is defective in nearly all of the Mg 2ϩ -independent PA phosphatase activity in yeast (15), grew equally   as well as wild type cells at both 30 and 37°C (Fig. 7). The dpp1⌬ lpp1⌬ mutations, however, slightly exacerbated the temperature-sensitive phenotype of the pah1⌬ mutation (Fig. 7).
We questioned whether the pah1⌬ mutation would affect the cellular levels of PA. In these experiments, cells were grown to the exponential and stationary phases of growth in medium lacking inositol. Both growth phases were examined because Mg 2ϩ -dependent PA phosphatase activity is elevated in stationary phase cells (46,60). Inositol was omitted from the growth medium to preclude the regulatory effects that this phospholipid precursor has on phospholipid synthesis (5,61). Phospholipids were extracted from wild type and pah1⌬ mutant cells that were labeled to steady state with 32 P i and analyzed by two-dimensional thin-layer chromatography. In exponential and stationary phase cells, the pah1⌬ mutation caused increases in the cellular levels of PA of 122 and 160%, respectively (Fig. 8). Introduction of the dpp1⌬ lpp1⌬ mutations into the pah1⌬ mutant background resulted in yet a further accumulation of PA in exponential and stationary phase cells (Fig. 8). The PA levels in the pah1⌬ dpp1⌬ lpp1⌬ triple mutant increased by 233 and 480% in the exponential and stationary phases, respectively, when compared with the wild type control (Fig. 8). The pah1⌬ mutation also affected the composition of major phospholipids in the exponential phase; the mutation caused a decrease in phosphatidylcholine (43%) and increases in phosphatidylethanolamine (50%) and phosphatidylinositol (80%) (Fig. 8A). Similar effects on phospholipid composition were observed in the pah1⌬ dpp1⌬ lpp1⌬ triple mutant (Fig. 8A). The decreased content of phosphatidylcholine might be attributed to the decrease in DAG needed for phosphatidylcholine synthesis via the CDP-choline pathway whereas the increased phosphatidylinositol content might be attributed to the INO1-mediated increase in the phosphatidylinositol precursor inositol (5). The effect of the pah1⌬ mutation on phospholipid composition in stationary phase cells was not as great as that in exponential phase (Fig. 8B).
DAG is used by DAG acyltransferase enzymes to produce TAG (6,62,63). If the PAH1-encoded Mg 2ϩ -dependent PA phosphatase contributes to the production of this pool of DAG, then the pah1⌬ mutation might be expected to affect TAG composition. This question was addressed by labeling cells with [2-14 C]acetate followed by the extraction and analysis of neutral lipids by thin-layer chromatography. We analyzed TAG in both the exponential and stationary phases of growth because the level of TAG is elevated in stationary phase cells (60,64). In exponential phase cells, the TAG content of the pah1⌬ mutant was 62% lower than that of the wild type parent (Fig. 9A). The effect of the pah1⌬ mutation on TAG content was even more dramatic in stationary phase cells; TAG levels dropped by 92% when compared with the control (Fig.  9B). That the defect in PA phosphatase activity in the pah1⌬ mutant was responsible for the reduction in TAG content was supported by the   S. cerevisiae wild type, pah1⌬,  pah1⌬ dpp1⌬ lpp1⌬, and dpp1⌬ lpp1⌬ cells. The indicated cells were grown to saturation in YEPD medium and diluted to a density of 2 ϫ 10 7 cells per ml at 30°C. Serial dilutions (1:10) of the cells were spotted (5 l) onto YEPD plates, and growth was scored after 2 days of incubation at 30 and 37°C. decreased levels of DAG in both the exponential (40%) and stationary (52%) phases of growth (Fig. 9).
Sterols and free fatty acids were also affected by the pah1⌬ mutation in exponential and stationary phase cells. The ergosterol ester content in the pah1⌬ mutant increased by 88 and 117% in exponential and stationary phase cells, respectively, when compared with the wild type control (Fig. 9). The level of ergosterol decreased in the exponential phase of the pah1⌬ mutant by 47%, whereas in the stationary phase, the level of ergosterol was not significantly affected by the mutation (Fig. 9). The fatty acid content of the pah1⌬ mutant in exponential and stationary phase cells increased by 43 and 125%, respectively, when compared with the control (Fig. 9). The increased levels of ergosterol ester and fatty acids were presumably due to the decreased ability of pah1⌬ mutant cells to utilize fatty acids for TAG synthesis. The introduction of the dpp1⌬ lpp1⌬ mutations in the pah1⌬ mutant background did not have a major effect on the changes in TAG and DAG that were brought about by the pah1⌬ mutation itself. However, some of the effects of the pah1⌬ mutation on sterols and fatty acids were enhanced by the dpp1⌬ lpp1⌬ mutations (Fig. 9). The dpp1⌬ lpp1⌬ mutations by themselves did not have a major effect on neutral lipid composition in either the exponential or stationary phases of growth (Fig. 9).
Heterologous Expression of the Human LPIN1 cDNA in E. coli and Identification of Lipin 1 as a Mg 2ϩ -dependent PA Phosphatase Enzyme-The protein product of the human LPIN1 gene (i.e. lipin 1) shares sequence homology with yeast Pah1p (30,56). Although the molecular function of lipin 1 is unknown, it is known that this protein in mice plays a major role in fat homeostasis (30,(65)(66)(67)(68)(69). Accordingly, we questioned the possibility that lipin 1 might be a Mg 2ϩ -dependent PA phosphatase enzyme. To test this hypothesis, we expressed human LPIN1 cDNA in E. coli. Purified His 6 -tagged human lipin 1 was assayed for Mg 2ϩ -dependent PA phosphatase activity. The results of this assay showed that human lipin 1 was in fact a Mg 2ϩ -dependent PA phosphatase enzyme. The specific PA phosphatase activity of the purified protein was 1,600 nmol/min/mg. This level of activity was comparable with that of the purified recombinant PAH1-encoded Mg 2ϩ -dependent PA phosphatase enzyme. The enzymological characterization of the LPIN1-encoded Mg 2ϩ -dependent PA phosphatase will be the subject of separate report.

DISCUSSION
The Mg 2ϩ -dependent form of PA phosphatase is postulated to be the enzyme involved in the de novo synthesis of TAG and phospholipids (via the Kennedy pathway) in S. cerevisiae (6,12). However, this notion has not been established because a gene encoding this enzyme is unknown. In this work using a reverse genetic approach, we identified PAH1 as the gene encoding a Mg 2ϩ -dependent PA phosphatase enzyme. Overexpression of the PAH1 gene in S. cerevisiae resulted in elevated levels of Mg 2ϩ -dependent PA phosphatase activity, whereas the deletion of the gene resulted in the reduction of this enzyme activity. Moreover, cells containing the pah1⌬ mutation accumulated PA and had reduced amounts of DAG and its acylated derivative TAG. The effects of the pah1⌬ mutation on TAG content were most evident in the stationary phase of growth where the synthesis of TAG predominates over the synthesis of phospholipids (60,64). Likewise, the pah1⌬ mutation showed the most striking effects on phospholipid composition in the exponential phase of growth where the synthesis of phospholipids dominates over TAG synthesis (60,64). The heterologous expression of the S. cerevisiae PAH1 gene in E. coli confirmed that Pah1p possessed Mg 2ϩdependent PA phosphatase activity. Moreover, the enzymological properties of the recombinant Mg 2ϩ -dependent PA phosphatase were very similar to those of the 91-kDa enzyme previously purified from S. cerevisiae (22,23,25). Collectively, these data provided conclusive evidence that the S. cerevisiae PAH1 gene is a bona fide structural gene encoding a Mg 2ϩ -dependent PA phosphatase in S. cerevisiae and that this enzyme does in fact generate DAG for lipid synthesis.
The 91-kDa form of Mg 2ϩ -dependent PA phosphatase used to identify the PAH1 gene was a proteolytic product of a larger sized enzyme (24). In fact, the sequence information derived from the 91-kDa enzyme lacked sequences at the C-terminal end of the protein (Fig. 1). The  . Neutral lipid composition of the S. cerevisiae wild type, pah1⌬, pah1⌬ dpp1⌬ lpp1⌬, and dpp1⌬ lpp1⌬ cells in the exponential and stationary phases of growth. The indicated cells were grown to the exponential (A) and stationary (B) phases of growth in the presence of [2-14 C]acetate (1 Ci/ml). Lipids were extracted, separated by one-dimensional thin-layer chromatography, and the images were subjected to ImageQuant analysis. The percentages shown for the individual lipids were normalized to the total 14 C-labeled chloroform-soluble fraction, which also contained phospholipids and unidentified neutral lipids. Each data point represents the average of three experiments Ϯ S.D. Abbreviations: Erg, ergosterol; ErgE, ergosterol ester; FA, fatty acid.
The PAH1-encoded Mg 2ϩ -dependent PA phosphatase contains an HAD-like domain with a catalytic DXDXT motif found in a superfamily of Mg 2ϩ -dependent phosphatase enzymes (51,52). In contrast, the DPP1-and LPP1-encoded Mg 2ϩ -independent PA phosphatases contain a catalytic motif consisting of the consensus sequences KXXXXXXRP (domain 1), PSGH (domain 2), SRXXXXXHXXXD (domain 3), which is shared by a superfamily of lipid phosphatases that do not require Mg 2ϩ ions for activity (70 -72). Distinctive phosphatase motifs found in the different types of PA phosphatase provide an explanation as to why attempts to identify a Mg 2ϩ -dependent PA phosphatase gene by sequence homology to a Mg 2ϩ -independent PA phosphatase have been unsuccessful. Thus, while both forms of PA phosphatase catalyze the same overall reaction, it is expected that their catalytic mechanisms would be different. Another major difference between the two types of PA phosphatase enzymes is the nature in which they associate with membranes. The DPP1-and LPP1-encoded Mg 2ϩ -independent PA phosphatases are integral membranes proteins with six transmembrane-spanning regions (14,15). On the other hand, the PAH1encoded Mg 2ϩ -dependent PA phosphatase does not have any transmembrane-spanning regions. This enzyme was associated with both the membrane and cytosolic fractions of the cell, and the membrane-bound enzyme was a peripheral membrane protein.
Mg 2ϩ -dependent and -independent PA phosphatase enzymes have been classified as being NEM-sensitive and NEM-insensitive, respectively (10,12,73,74). However, sensitivity to NEM is not an appropriate criterion to classify the two types of PA phosphatase enzymes because each type contains both NEM-sensitive and NEM-insensitive enzymes. For example, the DPP1-encoded phosphatase is NEM-insensitive (19), whereas the LPP1-encoded phosphatase is NEM-sensitive (20). Likewise, the PAH1-encoded Mg 2ϩ -dependent PA phosphatase activity was insensitive to NEM, whereas most of the Mg 2ϩ -dependent PA phosphatase activity remaining in the pah1⌬ dpp1⌬ lpp1⌬ triple mutant was sensitive to NEM.
Pah1p (Smp2p) has been recently identified as a factor that coordinates phospholipid synthesis with nuclear/ER membrane growth (56). This conclusion is based on the correlation between massive nuclear/ER membrane expansion and the derepression of the phospholipid synthesis genes INO1 (involved in phosphatidylinositol synthesis), OPI3 (involved in phosphatidylcholine synthesis), and INO2 (a positive phospholipid synthesis transcription factor) in pah1⌬ (smp2⌬) mutant cells (56). Chromatin immunoprecipitation analysis indicates that Pah1p (Smp2p) interacts with the promoters of the INO1, OPI3, and INO2 genes, suggesting its role as a transcription factor in the regulation of phospholipid synthesis (56). Previous studies have shown that the expression of these genes, which contain a UAS INO element, is controlled by the positive transcription factors Ino2p and Ino4p and by the negative transcription factor Opi1p (4, 5, 61). Maximum expression of these genes is driven by the interaction of an Ino2p-Ino4p complex with the UAS INO element in their promoters, whereas expression of these genes is attenuated by interaction of Opi1p with DNA-bound Ino2p (75). The repressive effect of Opi1p is most dramatic when cells are supplemented with inositol (4, 5, 61). The molecular function (i.e. Mg 2ϩdependent PA phosphatase activity) of Pah1p identified in this work suggests that it might control the expression of phospholipid synthesis genes by controlling the levels of PA.
Loewen et al. (76) have shown that reduced PA concentration, brought about by inositol supplementation, promotes the translocation of Opi1p from the nuclear/ER into the nucleus where it interacts with Ino2p to repress INO1 expression (76). By analogy, the same mechanism should apply to the regulation of other UAS INO -containing genes including OPI3 and INO2. The reduction in PA concentration in response to inositol supplementation has been attributed to increased PI synthesis (76), which draws upon the PA pool in the biosynthetic pathway (5). The PA-mediated regulation of Opi1p function might also be explained by involvement of the PAH1-encoded Mg 2ϩ -dependent PA phosphatase. Previous studies have shown that Mg 2ϩ -dependent PA phosphatase activity is elevated in inositol-supplemented cells (46), and as shown here, the PAH1 gene product played a role in controlling the cellular levels of PA. In addition, Pah1p (Smp2p) exists in S. cerevisiae as a phosphorylated protein, and it is dephosphorylated by an Nem1p-Spo7p protein phosphatase complex (56). Analysis of mutants defective in the protein phosphatase complex indicates that dephosphorylation of Pah1p (Smp2p) is required for normal expression of INO1, OPI3, and INO2 and nuclear/ER membrane growth (56). Based on these observations, the phosphorylation state of Pah1p might control its Mg 2ϩ -dependent PA phosphatase activity and/or the cellular location of the enzyme (e.g. membrane versus cytosolic) and thus the PA-mediated regulation of Opi1p and UAS INO -containing genes. Additional studies are required to address these hypotheses.
Lipin 1 has been identified as a factor that controls fat metabolism in mammalian cells (30,(65)(66)(67)(68)(69). In a mouse model, lipin 1 deficiency prevents normal adipose tissue development that results in lipodystrophy and insulin resistance, whereas excess lipin 1 promotes obesity and insulin sensitivity (30,65). Despite the importance of lipin 1, the mechanism by which it affects lipodystrophy and obesity has been an enigma due to the lack of information on the molecular function of the protein.
In this work, we found that human lipin 1 is a Mg 2ϩ -dependent PA phosphatase, the penultimate enzyme in the pathway to synthesize TAG. This finding provides a mechanistic basis for how lipin 1 regulates lipid metabolism in mammalian cells. Moreover, this work indicated that Mg 2ϩ -dependent PA phosphatase activity might be an important pharmacological target to control lipid metabolism in humans.