Phosphatidylethanolamine Has an Essential Role inSaccharomyces cerevisiae That Is Independent of Its Ability to Form Hexagonal Phase Structures*

Two yeast enzymes, Psd1p and Psd2p, catalyze the decarboxylation of phosphatidylserine to produce phosphatidylethanolamine (PtdEtn). Mitochondrial Psd1p provides ∼90% of total cellular phosphatidylserine decarboxylase activity. When thePSD1 gene is deleted, the resultant strain(psd1Δ) grows normally at 30 °C in glucose and in the absence of exogenous choline or ethanolamine. However, at elevated temperature (37 °C) or on the nonfermentable carbon source lactate, the growth of psd1Δ strains is minimal without ethanolamine supplementation. The reduced growth and viability correlate with a PtdEtn content below 4% of total phospholipid. These results suggest that there is a critical level of PtdEtn required to support growth. This theory is supported by growth data revealing that a psd1Δ psd2Δ dpl1Δ strain can only grow in the presence of ethanolamine. In contrast, a psd1Δ psd2Δstrain, which makes low levels of PtdEtn from sphingolipid breakdown, can be rescued by ethanolamine, choline, or the ethanolamine analogue propanolamine. psd1Δ psd2Δ cells grown in 2 mm propanolamine accumulate a novel lipid, which was determined by mass spectrometry to be phosphatidylpropanolamine (PtdPrn). PtdPrn can comprise up to 40% of the total phospholipid content in supplemented cells at the expense of phosphatidylcholine and PtdEtn. The absolute level of PtdEtn required for growth when PtdPrn is present appears to be 1% of the total phospholipid content. The essential function of the PtdEtn in the presence of propanolamine does not appear to be the formation of hexagonal phase lipid, insofar as PtdPrn readily forms hexagonal phase structures detectable by31P NMR.

In Saccharomyces cerevisiae the aminoglycerophospholipids, phosphatidylethanolamine (PtdEtn) 1 and phosphatidylcholine (PtdCho), are synthesized by two main pathways. When yeast are grown in the absence of exogenous ethanolamine or choline, the aminoglycerophospholipids are synthesized via the de novo pathway shown in Fig. 1. Phosphatidylserine (PtdSer) is first synthesized in the endoplasmic reticulum (ER) and its associated membranes by PtdSer synthase (Pss1p) (1) and is subsequently decarboxylated to form PtdEtn by PtdSer decarboxylase 1 (Psd1p) in the mitochondria (2) and by PtdSer decarboxylase 2 (Psd2p) in the Golgi/vacuole (3). The PtdEtn is exported from the sites of decarboxylation to the ER and subsequently methylated to form PtdCho (2,4,5). When ethanolamine and choline are present, PtdEtn and PtdCho are synthesized via the Kennedy pathway, which uses CDP-ethanolamine and CDP-choline intermediates (5). PtdEtn derived from the Kennedy pathway can also be methylated to form PtdCho (4,5). Phosphoethanolamine that serves as an intermediate in the Kennedy pathway can also be synthesized as a consequence of sphingolipid degradation via dihydrosphinogsine-1-phosphate metabolism using Dpl1p (6,7).
Early genetic examination of yeast aminoglycerophospholipid synthesis by Atkinson et al. (8) isolated mutants requiring ethanolamine or choline for growth. All of the mutants isolated belonged to a single complementation group designated cho1 that had defects in Pss1p. The auxotrophic requirements of the cho1 mutants were not only fulfilled by ethanolamine or choline, but also by monomethylethanolamine or dimethylethanolamine (8). Yeast strains with defects in both PtdSer decarboxylases (psd1⌬ psd2⌬) are also auxotrophic for either ethanolamine or choline (3), but mutants having defects in both methylation enzymes (pem1/cho2 pem2/opi3) are stringently auxotrophic for choline (5). These nutrient requirements suggested that PtdCho, rather than PtdSer or PtdEtn, was an essential lipid. All of the described mutants, however, contained a wild-type dihydrosphingosine-1-phosphate lyase gene (DPL1) permitting low levels of PtdEtn synthesis without ethanolamine supplementation.
In designing an experimental approach to study lipid transport in yeast, the psd1⌬, psd2⌬ and double psd1⌬ psd2⌬ null alleles were constructed. Two assumptions were made based on experimental data: that the psd1⌬ and psd2⌬ single null strains grow without auxotrophic requirements (2,3), and that ethanolamine auxotrophs isolated in screens of mutagenized cells die because of the inability to produce enough PtdCho. However, our recent work with the temperature-sensitive mutants derived from the psd1⌬ strains demonstrated that the psd1⌬ parental strain is an ethanolamine auxotroph at 36°C (9). Also recently, Birner et al. (10) demonstrated that psd1⌬ strains are ethanolamine auxotrophs when grown on the nonfermentable carbon source lactate. In this report, we examined the growth phenotype of a psd1⌬ strain and show that it is a conditional ethanolamine auxotroph that does not die because of a lack of PtdCho, but rather because of a depletion of PtdEtn. We also provide evidence that PtdEtn, or one of its metabolites, is essential for growth of S. cerevisiae and this essential function is independent of the ability of PtdEtn to form hexagonal phase structures.

EXPERIMENTAL PROCEDURES
Materials-All simple salts, buffers, amino acids, nutritional supplements, and solvents were purchased from either Sigma or Fisher. Yeast media components (yeast extract, peptone, and nitrogen base without amino acids) were purchased from Difco or American Biorganics Inc. Propanolamine (Prn) and monomethylethanolamine (MME) were purchased from Aldrich.
Phospholipid standards for thin layer chromatography (TLC), dioleoyl-PtdCho, and dioleoyl-PtdEtn were obtained from Avanti Polar Lipids or Sigma. Thin layer Silica Gel H plates were purchased from Analtech Corp, and thin layer Silica Gel 60 plates were purchased from Merck. The radiochemical [3-3 H]serine was purchased from Amersham Biosciences, Inc.
Yeast Strains and Growth Methods-The strains utilized in these studies and their genotypes are shown in Table I. The growth media for yeast, YPD and minimal medium, contained 2% glucose (SC) or 2% lactate (SL) as indicated (11). Adenine (20 mg/liter), uracil (20 mg/liter), and ethanolamine (Etn, 2 mM) were routinely added to the YPD to give YPDAUE. Etn, choline (Cho), MME, and Prn were prepared as 0.5 M stocks (pH 6 -7) that were filter-sterilized.
[ 3 H]Serine Labeling of Yeast Strains-All yeast strains were initially grown at 30°C in SC or SL containing 2 mM Etn, to mid-log phase. Subsequently, cells were diluted to an A 600 nm of 0.3 in SC or SL without serine, containing 10 Ci of [ 3 H]serine/ml. For specific incubation conditions, refer to table legends. After incubation, the cells were precipitated with trichloroacetic acid (10% final concentration) and combined with ϳ10 mg of carrier cells. Pellets were washed twice with cold water, and the lipids were extracted and analyzed by TLC as previously described (2,3). The TLC plates were sprayed with 0.2% (w/v) 8-anilino-1-napthalenesulfonic acid, and the lipids were visualized under UV light. Lipids were scraped from the TLC plates, and radioactivity was quantified by liquid scintillation counting.
Lipid Phosphorus Measurement-Strains were grown under the con-ditions described in the table legends. The cells were harvested, and the lipids were extracted as described above except carrier cells and trichloroacetic acid were not added. For quantification of mitochondrial lipids, crude and purified mitochondria fractions were isolated using the method of Glick and Pon (12). PtdCho, PtdIns, PtdSer, PtdEtn, cardiolipin, and other lipids were separated by two-dimensional TLC on Silica 60 plates in chloroform/methanol/acetic acid (13/5/2 v/v/v), followed by chloroform/methanol/formic acid (13/5/2 v/v/v). For analysis of the psd1⌬ psd2⌬ lipid content, the two-dimensional TLC was performed using chloroform/methanol/ammonium hydroxide (65/35/5 v/v/v) followed by chloroform/acetic acid/methanol/water (75/25/5/2.2 v/v/v). Lipids were visualized by iodine vapor and scraped into glass tubes. Phosphorus was quantified by the Rouser method (13). Liquid Chromatography/Mass Spectrometry (LC/MS)-LC/MS and tandem mass spectrometry (LC/MS/MS) was carried out in a Sciex API-IIIϩ triple quadrupole mass spectrometer (PE-Sciex, Thornhill, Toronto, Canada). This machine was equipped with an electrospray ionization source interfaced to a gradient HPLC system. The electrospray ionization spray voltage was 4500 V, the orifice potential maintained at 75 V, and the collisional offset potential was 20 eV for tandem experiments. For collisional induced decomposition experiments, the argon gas thickness was 235 ϫ 10 13 molecules/cm 3 . Samples from two-dimensional TLC were injected into a liquid stream of methanol with 1 mM ammonium acetate flowing at 0.02 ml/min. To further separate some samples, normal phase HPLC was used on a silicic acid column (150 ϫ 2.0 mm, Phenomenex, Torrance, CA) with a gradient from hexane/isopropanol (30:40) to hexane/isopropanol/water (30:40:7, with 1 mM ammonium acetate) over 20 min at a flow rate of 0.2 ml/min.
Synthesis of Dioleoyl-phosphatidylpropanolamine (PtdPrn)-Dioleoyl-PtdPrn was synthesized by a transphosphatidylation reaction using dioleoyl-PtdCho as a substrate (14). 100 mg of dioleoyl-PtdCho was incubated in a two-phase diethyl ether, 165 mM sodium acetate, acetic acid (pH 5.5), with 100 mM calcium chloride system with 25% propanolamine in the presence of phospholipase D, isolated from Savoy cabbage (15), for 1 h at room temperature. The reaction was stopped by the addition of 0.5 M EDTA, and the ether was evaporated under a stream of N 2 gas. The lipid was extracted by a Bligh and Dyer extraction (16). The dioleoyl-PtdPrn was isolated by preparative TLC on Silica 60 plates developed in chloroform/acetic acid/methanol/water (75/25/5/2.2 v/v/v/ v). The lipid was eluted from the silica gel with a Bligh and Dyer monophase and then extracted by phase separation. The chloroform phase was recovered and the solvent evaporated under a stream of N 2 gas. Residual acetic acid was removed from the isolated lipid by resuspension and evaporation of chloroform. The concentration of dioleoyl-PtdPrn was determined by lipid phosphorus measurement, and the purity was assessed by TLC and mass spectrometry. 31 P NMR Spectroscopy-Dioleoyl-PtdEtn, dioleoyl-PtdCho and dioleoyl-PtdPrn (40 -50 mg) were hydrated in 500 l of 20 mM HEPES (pH 7) containing 100 mM sodium chloride and 10% deuterium oxide and vortexed at room temperature. The pH of the lipid mixture was checked FIG. 1. Schematic diagram of aminoglycerophospholipid biosynthesis in yeast. In the absence of exogenous ethanolamine and choline, aminoglycerophospholipids are synthesized via the de novo pathway. Serine is incorporated into PtdSer by PtdSer synthase (Pss1p) in the ER. The resultant PtdSer is then transported to the location of PtdSer decarboxylase 1 (Psd1p) at the inner mitochondrial membrane or PtdSer decarboxylase 2 (Psd2p) in the Golgi/vacuole, where it is converted to PtdEtn. PtdEtn produced by Psd1p or Psd2p is then transported from the mitochondria or Golgi/vacuole back to the ER. Within the ER, PtdEtn is methylated to PtdCho by the sequential action of PtdEtn methyltransferase 1 and 2 (Pemp1 and Pemp2). PtdEtn can also be synthesized from sphingolipid-derived phosphoethanolamine originating from the activity of dihydrosphingosine-1phosphate lyase (Dpl1p). Alternatively, when Etn and Cho are present, PtdEtn and PtdCho are synthesized via the Kennedy pathway (not shown in detail). and adjusted as necessary. The 31 P NMR spectra were recorded on a Varian Inova 600-MHz spectrometer operating at 243 MHz for 31 P. The free induction decays were accumulated by employing 21-s 90°pulse, and 320 -1600 scans with a 30-kHz sweep width and a 1.0-s interpulse delay. The spectra were collected in the presence of proton decoupling. The temperatures of the samples were maintained at Ϫ15°C to 10°C by a variable temperature control unit and equilibrated 30 min prior to the data acquisition.

RESULTS
Growth Characteristics of the psd1⌬ Strain-Yeast express two PtdSer decarboxylase enzymes: Psd1p, which resides in the inner mitochondrial membrane (2); and Psd2p, which co-localizes to the vacuole and Golgi compartments (3). Strains containing a psd1⌬ mutation are dependent upon Psd2p and Dpl1p activity for growth in the absence of exogenous ethanolamine or choline (3). When the psd1⌬ strain was originally characterized, it displayed wild-type growth in the absence of ethanolamine (2). This characterization, however, was only done in liquid glucose medium at 30°C or on solid medium at 30°C with glucose. When we were analyzing temperaturesensitive mutants in this genetic background, we discovered that the psd1⌬ parental strain is an ethanolamine auxotroph at elevated temperature in liquid medium (9). To further characterize the growth of the psd1⌬ strain, we analyzed the auxotrophic requirements in various carbon sources at 30°C and 37°C in both liquid and solid media. The growth characteristics of the psd1⌬ strain in liquid glucose medium are shown in Fig. 2. At 30°C ( Fig. 2A), as described previously (2), the psd1⌬ strain displays wild-type growth regardless of ethanolamine or choline supplementation. However, at 37°C (Fig. 2B), the psd1⌬ strain shows minimal growth without ethanolamine or choline supplementation, but either ethanolamine or choline supplementation restores wild-type growth.
The addition of the dpl1⌬ allele to the psd1⌬ strain makes the strain strictly dependent upon Psd2p for PtdEtn synthesis in the absence of ethanolamine. The psd1⌬ dpl1⌬ strain, when pregrown in the presence of ethanolamine and then shifted to ethanolamine/choline-free medium, displays almost wild-type growth at 30°C (Fig. 3A). However, if the psd1⌬ dpl1⌬ strain is grown in ethanolamine/choline-free medium for five to seven generations and then diluted to a lower absorbance, it can no longer grow in the same medium (Fig. 3B). These observations indicate that cells accumulate sufficient intracellular ethanolamine pools to support growth for several generations, but the depletion of these pools leads to arrest of growth. At 37°C the growth characteristics of the psd1⌬ dpl1⌬ strain are identical to the psd1⌬ stain (data not shown).
The observed ethanolamine auxotrophy of the psd1⌬ and psd1⌬ dp1l⌬ is dependent upon how the growth is assessed. psd1⌬ and psd1⌬ dpl1⌬ cells grow on solid synthetic glucose medium in the absence of ethanolamine at both 30°C and 37°C (data not shown). This apparent lack of ethanolamine auxotrophy at 37°C for psd1⌬ cells, and at 30°C and 37°C for psd1⌬ dpl1⌬ cells on solid media may be caused by recycling of excreted choline or ethanolamine. Recent evidence (17) demonstrates that choline recycling following PtdCho formation from the de novo pathway can be highly efficient.
Previous work from our laboratory showed that the psd1⌬ strain exhibits a greater tendency to produce petite, rho Ϫ cells, compared with wild-type cells (2). The production of petite cells is normally the result of loss of mitochondrial function. Because Psd1p is a mitochondrial enzyme and PtdEtn is enriched in this organelle, we examined the growth of the psd1⌬ strain on the nonfermentable carbon source lactate, to determine whether a requirement for functional mitochondria alters the nutrient requirements of this strain. We observed almost identical growth characteristics as described by Birner et al. (10) for both the psd1⌬ and psd1⌬ dpl1⌬ strains (data not shown). The psd1⌬ strain at both 25°C and 30°C does not grow in the absence of ethanolamine or choline, but supplementation with either nutrient restores wild-type growth. At 37°C the psd1⌬ strain does not grow regardless of nutrient supplementation. This ethanolamine auxotrophy was also observed for the psd1⌬ strain on solid lactate medium and in liquid glycerol medium at 30°C. Thus, the psd1⌬ strain is an ethanolamine auxotroph under all conditions that require functional mitochondria. Analysis of the de Novo Aminoglycerophospholipid Synthesis in the psd1⌬ Strain-The effects of the psd1⌬ mutation upon FIG. 2. The psd1⌬ strain requires ethanolamine or choline for optimal growth at 37°C in glucose medium. Log phase wild-type (SEY6210) and psd1⌬ (RYY52) cells, grown at 30°C in SC without ethanolamine were harvested by centrifugation and washed twice with SC. Cells were re-inoculated to give an absorbance of 0.02 in SC (Ⅺ, E), SC plus 2 mM ethanolamine (f, q), or SC plus 2 mM choline (OE, ࡗ). Cell growth was monitored at 30°C (A) or 37°C (B) by culture absorbance at 600 nm. Representative data are shown; similar results were seen in four other experiments. Analysis of Aminoglycerophospholipid Composition of the psd1⌬ Strain-Because the alterations in aminoglycerophospholipid synthesis displayed by the psd1⌬ strain were similar in both permissive and nonpermissive conditions, we examined the steady-state levels of the aminoglycerophospholipids to determine the cause of the ethanolamine auxotrophy. The size of the aminoglycerophospholipid pools was determined by phosphorus analysis. Results shown in Table III are the percentage of total lipid phosphorus observed in PtdSer, PtdCho, PtdIns, and PtdEtn from cells grown in glucose medium in the absence or presence of ethanolamine at 30°C and 37°C. Under all conditions analyzed, the content of PtdCho in the psd1⌬ and psd1⌬ dpl1⌬ strains were similar or elevated compared with the wild-type strain. The PtdEtn content, however, was decreased by 50 -70% in both psd1⌬ and psd1⌬ dpl1⌬ cells compared with wild-type cells at 30°C, in the absence or presence of ethanolamine. At 37°C the PtdEtn content was decreased in both psd1⌬ and psd1⌬ dpl1⌬ strains by 80% without nutrient addition, and by 4 -11% in the presence of ethanolamine compared with wild-type cells. The decreased PtdEtn content in both strains was compensated for by an increase in PtdIns. Under these experimental conditions, the lipid composition of the psd1⌬ dpl1⌬ strain did not vary significantly from that of the psd1⌬ strain.
The data in Table IV show the sizes of the aminoglycerophospholipid pools from whole cell homogenates and the crude mitochondria fraction from cells grown in lactate medium at 25°C. The results we observed displayed similar trends to that described by Birner et al. (10), but the exact percentages of each lipid pool were different. This difference is most likely caused by variation of growth conditions and experimental procedures. We grew cells in minimal lactate medium with controlled amounts of ethanolamine, whereas Birner et al. (10) used YPL medium. YP medium contains enough choline and ethanolamine to rescue the ethanolamine auxotrophy of the psd1⌬ strain (data not shown). The phospholipid composition of the crude mitochondrial fraction (Table IV) for both the psd1⌬ and wild-type strains is similar to that observed for the whole cell homogenate. The PtdCho pool in psd1⌬ cells grown in lactate medium was elevated compared with wild-type cells in the absence or presence of ethanolamine, and the PtdEtn content was decreased by 70 -85%. The decrease in PtdEtn levels in psd1⌬ cells is compensated for by increases in PtdIns and PtdSer content. The addition of ethanolamine to the growth media does not dramatically alter the aminoglycerophospholipid composition of either strain.
Because alterations in the aminoglycerophospholipid composition of the psd1⌬ cells do not explain their inviability at elevated temperature or in lactate medium, we compared the absolute PtdEtn pool sizes from all conditions (Tables III and  IV). Under conditions where the psd1⌬ strain is an ethanolamine auxotroph (glucose at 37°C, and lactate at 25°C), the PtdEtn content of this strain is under 4%; however, under permissive conditions, psd1⌬ cells have a PtdEtn content between 8 and 30%. Thus, growth of the psd1⌬ strain appears to correlate with a PtdEtn content greater than 4%, when PtdEtn is required for PtdCho synthesis. Only in glucose medium at 37°C does the presence of ethanolamine restore wild-type PtdEtn content in the psd1⌬ strain.
Characterization of the Growth Phenotype of psd1⌬ psd2⌬ and psd1⌬ psd2⌬ dpl1⌬ Strains-Because the psd1⌬ strain appears to have limiting PtdEtn when grown at elevated temperature in glucose medium or in nonfermentable carbon sources, we tested the hypothesis that yeast require a critical level of PtdEtn for growth. Fig. 4 shows the growth characteristics in glucose medium at 30°C of a psd1⌬ psd2⌬ strain that makes low levels of PtdEtn from sphingolipid breakdown in the absence of ethanolamine using Dpl1p, and a psd1⌬ psd2⌬ dpl1⌬ strain that cannot make PtdEtn in the absence of ethanolamine. If PtdEtn is essential for growth, then the psd1⌬ psd2⌬ dpl1⌬ strain should not grow in the presence of choline. The psd1⌬ psd2⌬ strain is an ethanolamine auxotroph (Fig.  4A), but can grow in the presence of either ethanolamine or choline. The psd1⌬ psd2⌬ dpl1⌬ strain also cannot grow in the absence of ethanolamine and, unlike the psd1⌬ psd2⌬ strains, cannot grow in the presence of choline. The requirement for ethanolamine in psd1⌬ psd2⌬ dpl1⌬ strain cannot be compensated for by either ethanolamine analogue MME or Prn (Fig.   FIG. 3. The psd1⌬ dpl1⌬ strain displays some dependence on ethanolamine for growth in glucose at 30°C. Log phase wild-type (SEY6210) and psd1⌬ dpl1⌬ (JCY272) cells, grown at 30°C in SC medium with (A) or without (B) 2 mM ethanolamine were harvested by centrifugation and washed twice with SC. Cells were re-inoculated to give an absorbance of 0.02 in SC (Ⅺ, E), SC plus 2 mM ethanolamine (q), or SC plus 2 mM choline (ࡗ). Cell growth was monitored at 30°C by culture absorbance at 600 nm. Representative data are shown; similar results were seen in four other experiments.

Analysis of lipid content from strains grown in glucose at 30°C and 37°C
Wild-type (SEY6210), psd1⌬ (RYY52), and psd1⌬ dpl1⌬ (JCY272) cells were grown to mid-log phase at 30°C in SC medium plus ethanolamine (2 mM). Cells were harvested by centrifugation, washed twice with SC, and resuspended at an absorbance of 0.05 to 0.2 in SC or SC plus 2 mM ethanolamine and incubated at 30°C or 37°C for 8 h. Cells were harvested and lipids quantified. Results are the percentage of total lipid phosphorus in each phospholipid pool. Data are expressed as the mean Ϯ S.E. from three to seven experiments. Significance was determined using a two-tailed t test compared with wild-type under similar conditions (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). Strain

Analysis of lipid content from cell homogenates and crude mitochondria purified from cells grown in lactate at 25°C
Wild-type (SEY6210) and psd1⌬ (RYY52) cells were grown to mid-log phase at 30°C in SL medium plus ethanolamine (2 mM). Cells were harvested by centrifugation, washed twice with SL and resuspended at an absorbance of 0.05 to 0.2 in SL or SL plus 2 mM ethanolamine and incubated at 25°C for 24 h. Cells were harvested, mitochondria were purified, and lipids were quantified. Results are the percentage of total lipid phosphorus in each phospholipid pool. Data are expressed as the mean Ϯ S.E. from three to five experiments. Significance was determined using a two-tailed t test compared with wild-type under similar conditions (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). Strain

H]Serine incorporation into the aminoglycerophospholipids for strains grown in either glucose or lactate at 30°C and 37°C
Wild-type (SEY6210) and psd1⌬(RYY52) cells were grown to mid-log phase at 30°C in SC or SL medium plus ethanolamine (2 mM). Cells were harvested by centrifugation, washed once with SC or SL, and resuspended at an absorbance of 0.2 in SC lacking serine or SL lacking serine medium containing 10 Ci of [ 3 H]serine/ml. Cells were incubated at 30°C and 37°C for 4 h (for glucose-grown cells) or 12 h (for lactate-grown cells). Cells were harvested, and amount of labeled PtdSer, PtdEtn, and PtdCho was quantified. Results shown are percentage of total radiolabel incorporated into each phospholipid and the total cpm [ 3 H]serine incorporated into all phospholipids. Data are expressed as the mean Ϯ S.E. from three to five experiments. Significance was determined using a two-tailed t-test compared with wild-type at 30°C (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001).  4B), but the psd1⌬ psd2⌬ strain grows normally in the presence of either compound. It was surprising that propanolamine could not rescue the psd1⌬ psd2⌬ dpl1⌬ strain, as this analogue should theoretically closely mimic ethanolamine because it has a linear structure and contains a primary amine.
To further characterize the nutrient requirements of the psd1⌬ psd2⌬ dpl1⌬ strain, we examined the ethanolamine concentration dependence in the absence (Fig. 5A) or presence (Fig. 5B) of 2 mM propanolamine. With ethanolamine supplementation alone, concentrations of this nutrient of 0.5 mM or higher gave optimal growth. If the psd1⌬ psd2⌬ dpl1⌬ strain is grown with 2 mM propanolamine supplementation, the ethanolamine requirement can be reduced to 50 M. Therefore, propanolamine can fulfill some, but not all, of the ethanolamine requirements within the cell.
Analysis of Propanolamine Usage in the psd1⌬ psd2⌬ Strain-Because the ethanolamine requirement of the psd1⌬ psd2⌬ strain can be replaced by propanolamine, we next examined the metabolism of this precursor to phospholipid by comparing the aminoglycerophospholipid composition of the psd1⌬ psd2⌬ cells grown in the presence of propanolamine with that of cells grown in the presence of ethanolamine. After separating the lipids on two-dimensional TLC (Fig. 6A), two differences were apparent with the propanolamine-grown cells. The cells grown in propanolamine accumulate a new lipid that is predicted to be PtdPrn, and these cells lack phosphatidyldimethylethanolamine (PtdEtn(Me) 2 ). The lipid composition of psd1⌬ psd2⌬ cells grown in propanolamine or ethanolamine was quantified by phosphorus analysis (Fig. 6B). For the cells grown in ethanolamine, PtdEtn comprises 14%, PtdCho comprises 38%, and PtdEtn(Me) 2 comprises 4% of the total lipid phosphorus. When grown in propanolamine, PtdPrn makes up 40% of the total lipid, with PtdCho reduced to 9.5% and PtdEtn to 1%. A similar lipid composition was observed in psd1⌬ psd2⌬ dpl1⌬ cells grown in the presence of 2 mM propanolamine and 50 M ethanolamine (data not shown). These data show that the psd1⌬ psd2⌬ and psd1⌬ psd2⌬ dpl1⌬ strains readily incorporate propanolamine into PtdPrn, and suggest that this lipid may replace some but not all of the cellular requirements for PtdEtn. The data also demonstrate that, under the appropriate conditions, steady-state PtdEtn levels as low as 1% of total lipid can support log phase growth.
To confirm the identity of the PtdPrn, we purified the lipid by two-dimensional TLC and further analyzed it using electrospray ionization mass spectrometry. The major positive ion in the purified lipid sample was m/z 702, which is consistent with [M ϩ H] ϩ for a 32:2 acyl phospholipid, if the head group was phosphopropanolamine. Following collision-induced decomposition (MS/MS) (Fig. 7), a loss of 155 mass units from the [M ϩ H] ϩ m/z 702 was observed. This would be the expected loss for a phosphopropanolamine moiety, based on a loss of 141 mass units observed for PtdEtn (18). These data did not exclude the possibility that the sample was phosphatidylmonomethylethanolamine (PtdEtn(Me)) because it would be expected to give identical data. To prove that the lipid synthesized by the psd1⌬ FIG. 4. The psd1⌬ psd2⌬ dpl1⌬ strain is an ethanolamine auxotroph that cannot be rescued by choline. A, log phase psd1⌬ psd2⌬ (PTY44) and psd1⌬ psd2⌬ dpl1⌬ (HKY44) cells grown at 30°C in SC medium with 2 mM ethanolamine were harvested by centrifugation and washed twice with SC. Cells were re-inoculated to give an absorbance of 0.02 in SC (Ⅺ, E), SC plus 2 mM ethanolamine (f, q), or SC plus 2 mM choline (OE, ࡗ). Cell growth was monitored by culture absorbance at 600 nm. B, psd1⌬ psd2⌬ and psd1⌬ psd2⌬ dpl1⌬ cells were plated onto SC, SC plus 2 mM Etn, SC plus 2 mM Cho, SC plus 2 mM MME, or SC plus 2 mM Prn and grown for 3 days at 30°C. Representative data are shown; similar results were seen in three other experiments. psd2⌬ strain in the presence of propanolamine was indeed PtdPrn, normal phase LC/MS was performed with the isolated yeast lipid and a PtdEtn(Me) standard. There was a 7-min difference between the elution time of the PtdPrn compared with the PtdEtn(Me) standard (data not shown), demonstrating the yeast lipid is not PtdEtn(Me).
We also analyzed the purified PtdCho spot from the propanolamine-treated sample by MS/MS to determine whether it contained any methylated-PtdPrn species. The data indicated that greater than 90% of the lipid in the PtdCho spot was PtdCho, and it contained only traces of trimethyl-PtdPrn (data not shown). These results indicate that the PtdEtn methyltrans-ferases (Pem1p and Pem2p) can use some PtdPrn as a substrate, but PtdEtn is used more efficiently.
Analysis of the Physical Properties of PtdPrn-Under physiological conditions unsaturated PtdCho, PtdSer, and PtdIns all form bilayers when hydrated, but only PtdEtn forms a hexagonal phase in vitro (19 -24). These hexagonal phase structures have been proposed to be important for membrane fluidity and membrane fusion (25)(26)(27). Thus, one apparent requirement of cells for PtdEtn may be a consequence of the role of this lipid in forming hexagonal phases. We next sought to examine whether PtdPrn could also form hexagonal phases and replace PtdEtn in this regard. To determine the physical structure PtdPrn prefers, we performed 31 P NMR analysis on synthetic dioleoyl-PtdPrn, using the method of Cullis et al. (19). The dioleoyl-PtdPrn was hydrated at neutral pH, and its 31 P NMR spectra were compared with that of dioleoyl-PtdCho and dioleoyl-PtdEtn (Fig. 8). As described previously (19,28), dioleoyl-PtdEtn undergoes a phase shift from the bilayer conformation to the hexagonal phase with increasing temperature. The midpoint for the phase shift is near Ϫ5°C. We also found that PtdPrn readily forms hexagonal phase structures and does so at a lower temperature than PtdEtn. We were not able to define conditions in which the analog formed only the bilayer phase because the NMR probe temperature cannot go below Ϫ15°C. The dioleoyl-PtdCho gave the expected spectra at 10°C that indicates the presence of the lamellar-bilayer phase. From these data we conclude that PtdPrn forms hexagonal phase structures similar to those of PtdEtn. Thus, the inability of the analog to replace PtdEtn in the psd1⌬ psd2⌬ dpl1⌬ strains is likely caused by a function of PtdEtn that is independent of hexagonal phase formation or by a lower propensity to form bilayer phases. The data also demonstrate that a hexagonal phase lipid, PtdPrn, can comprise 40% of the total membrane phospholipids, and cells still display wild-type growth. DISCUSSION We have previously constructed strains containing a null allele for either or both the PSD1 and PSD2 genes (2,3,29). The original characterization of these strains showed that either the psd1⌬ or the psd2⌬ strain remained prototrophic for ethanolamine and retained some PtdSer decarboxylase activity, whereas the double mutant was an ethanolamine auxotroph with no measurable PtdSer decarboxylase activity. The psd1⌬ psd2⌬ strain, like pss1/cho1 mutants, could be rescued with either ethanolamine or choline (3,8). Because either nutrient could rescue cells that lacked the de novo aminoglycerophospholipid synthetic pathway, it was concluded that most probably PtdCho and not PtdSer or PtdEtn was essential for cell growth (4). However, the current studies and recent work by Birner et al. (10) demonstrate that the psd1⌬ strain is an ethanolamine prototroph in glucose medium, and that this strain becomes auxotrophic for ethanolamine at 37°C or when cultured on nonfermentable carbon sources such as lactate.
To determine the reason for the conditional ethanolamine autotrophy in the psd1⌬ strain, we first analyzed its ability to synthesize the aminoglycerophospholipids via the de novo pathway using [ 3 H]serine incorporation. Under all conditions the incorporation of radiolabel into PtdEtn was decreased to the same extent. Therefore, decreased Psd2p activity does not explain the conditional ethanolamine auxotrophy observed in psd1⌬ cells.
We measured the size of the phospholipid pools to determine whether alterations in lipid composition explain the conditional ethanolamine auxotrophy. As observed with the [ 3 H]serine labeling, under all conditions examined the lipid profile of the psd1⌬ strain was altered compared with the wild-type strain. The content of the PtdEtn pools of the psd1⌬ and psd1⌬ dpl1⌬ strains were decreased compared with the parental strain by 70 -80% when measured in the absence of ethanolamine and by 10 -70% in the presence of ethanolamine. It appears that the Kennedy pathway-derived PtdEtn cannot fully compensate for the lack of Psd1p activity in these cells. Surprisingly, in the psd1⌬ and psd1⌬ dpl1⌬ strains, the size of the PtdCho pools were equal to or larger than wild-type cells in all cases. This indicates that a lack of PtdCho is not the cause of the ethanolamine auxotrophy.
The overall lipid profiles of the psd1⌬ strain grown under the various conditions do not provide a clear explanation for the inviability of cells either at 37°C, or in lactate, when grown in the absence of ethanolamine. However, an examination of the absolute levels of PtdEtn shows that under permissive conditions the PtdEtn content of the psd1⌬ strain is at least 8% of the total phospholipid, but under the nonpermissive conditions the PtdEtn content was around 4% of the total phospholipid. This indicates there may be an essential requirement for PtdEtn in yeast, and that in the psd1⌬ strain the Psd2p and Dpl1p enzymes cannot supply enough PtdEtn to fulfill both PtdCho synthesis and this essential requirement when cultured at 37°C, or when mitochondrial function is required.
Comparison of the growth characteristics and lipid compositions of the psd1⌬ and psd1⌬ dpl1⌬ strains indicates that Dpl1p does not dramatically alter the steady-state PtdEtn levels in the cell. However, the presence of Dpl1p does provide enough additional PtdEtn to sustain long term growth of cells at 30°C under ethanolamine-free conditions.
To test the hypothesis that PtdEtn is an essential lipid, we analyzed the growth phenotype of the psd1⌬ psd2⌬ dpl1⌬ strain. This strain cannot make PtdEtn via the de novo pathway (3,29) or from sphingolipid degradation (6, 7) and can only synthesize PtdEtn when supplemented with ethanolamine. The psd1⌬ psd2⌬ dpl1⌬ strain is an ethanolamine auxotroph that cannot be rescued by choline. This clearly demonstrates that either PtdEtn or one of its metabolites is essential. The psd1⌬ psd2⌬ strain, which can make PtdEtn from sphingolipid degradation via Dpl1p in the absence of ethanolamine, can grow in the presence of ethanolamine, choline, or the ethanolamine analogues propanolamine or monomethylethanolamine. psd1⌬ psd2⌬ cells grown in the presence of propanolamine accumulate PtdPrn at the expense of PtdCho and PtdEtn. In 2 mM propanolamine, the PtdEtn levels within the cell drop to 1% of the total phospholipid pool, and this may be the minimum level of this lipid necessary for growth. PtdPrn must have the ability to compensate for the low levels of PtdEtn and PtdCho, but there is at least one function for PtdEtn or one of its metabolites that requires the exact structure of the ethanolamine head group.
The various metabolic roles for PtdEtn in the cell are shown in Fig. 9. In addition to serving as a precursor for PtdCho, PtdEtn is involved in the glycosylphosphatidylinositol (GPI) modification of proteins. The GPI anchor contains three phosphoethanolamine moieties that are thought to be transferred from PtdEtn (30 -32) by the action of Mcd4p, Gpi13p, and Gpi7p (33)(34)(35). The GPI anchor is linked to proteins through the amine group of one of the phosphoethanolamine moieties, and a second phosphoethanolamine group appears to be essen- FIG. 8. PtdPrn is a hexagonal phase lipid. Dioleoyl-PtdEtn, dioleoyl-PtdCho, and dioleoyl-PtdPrn (40 -50 mg) were hydrated at neutral pH and their 31 P NMR spectra were analyzed as described under "Experimental Procedures." The temperature of the samples were controlled by a VT unit, and the sample was equilibrated at each temperature for 30 min prior to the data acquisition. tial for anchor synthesis in yeast (34). The GPI anchored proteins are essential for cell wall maintenance, because Gas1p, an abundant GPI-linked protein, is a glucanosyltransferase responsible for covalently linking 1,3-␤-linked glucan chains to each other and to other cell wall components (36). Our screen for mutants involved in PtdSer metabolism and transport has previously isolated a mcd4 mutation (9). To determine whether GPI anchor synthesis is the critical role for PtdEtn, we unsuccessfully tried to rescue the psd1⌬ psd2⌬ dpl1⌬ strain with choline or propanolamine plus osmotic support (1 M sorbitol), 2 which stabilizes weak cell walls and rescues many strains with cell wall defects (37). This indicates that either there is another critical function for PtdEtn, or that the defect in the cell wall in the psd1⌬ psd2⌬ dpl1⌬ strain cannot be rescued by osmotic support.
PtdEtn is also used to directly modify proteins to allow association with membranes. Recently, it was shown that the yeast protein Apg8p is reversibly modified with PtdEtn through a covalent bond between the C-terminal glycine of the protein and the amine of PtdEtn (38 -40). Apg8p is critically involved in autophagy, which is responsible for the bulk import of cytosol into the vacuole for protein degradation, and PtdEtn modification is essential to its function (40). Direct protein lipidation by PtdEtn may also be the critical role for this lipid. However, it is still unknown whether PtdEtn modification is required for the function of any essential gene products.
In our studies we designed a specific test to determine whether the cell requirement for PtdEtn could be replaced by PtdPrn, a structurally related lipid capable of forming the hexagonal phase. The NMR analysis of PtdPrn revealed that this lipid was capable of forming hexagonal phase structures over a wider temperature range than PtdEtn. Although propanolamine can substitute for the ethanolamine requirement of psd1⌬ and psd1⌬ psd2⌬ strains, it cannot do the same for the psd1⌬ psd2⌬ dpl1⌬ strains. It is noteworthy that Prn does reduce the ethanolamine requirement for psd1⌬ psd2⌬ dpl1⌬ strains from 0.5 mM to 50 M. These data provide compelling evidence that PtdEtn plays an essential function in cells that is independent of the physical properties of this lipid.