Identification of a Novel Lysophospholipid Acyltransferase in Saccharomyces cerevisiae*

The incorporation of unsaturated acyl chains into phospholipids during de novo synthesis is primarily mediated by the 1-acyl-sn-glycerol-3-phosphate acyltransferase reaction. In Saccharomyces cerevisiae, Slc1 has been shown to mediate this reaction, but distinct activity remains after its removal from the genome. To identify the enzyme that mediates the remaining activity, we performed synthetic genetic array analysis using a slc1Δ strain. One of the genes identified by the screen, LPT1, was found to encode for an acyltransferase that uses a variety of lysophospholipid species, including 1-acyl-sn-glycerol-3-phosphate. Deletion of LPT1 had a minimal effect on 1-acyl-sn-glycerol-3-phosphate acyltransferase activity, but overexpression increased activity 7-fold. Deletion of LPT1 abrogated the esterification of other lysophospholipids, and overexpression increased lysophosphatidylcholine acyltransferase activity 7-fold. The majority of this activity co-purified with microsomes. To test the putative role for this enzyme in selectively incorporating unsaturated acyl chains into phospholipids in vitro, substrate concentration series experiments were performed with the four acyl-CoA species commonly found in yeast. Although the saturated palmitoyl-CoA and stearoyl-CoA showed a lower apparent Km, the monounsaturated palmitoleoyl-CoA and oleoyl-CoA showed a higher apparent Vmax. Arachidonyl-CoA, although not abundant in yeast, also had a high apparent Vmax. Pulse-labeling of lpt1Δ strains showed a 30% reduction in [3H]oleate incorporation into phosphatidylcholine only. Therefore, Lpt1p, a member of the membrane-bound o-acyltransferase gene family, seems to work in conjunction with Slc1 to mediate the incorporation of unsaturated acyl chains into the sn-2 position of phospholipids.

It has long been known that phospholipids commonly contain an asymmetrical distribution of acyl chains (1,2). Saturated acyl chains are usually found at the sn-1 position and unsaturated acyl chains at sn-2. The reason for this asymmetry may be that the liquid crystal transition temperature (T c ) for phospho-lipids with two saturated acyl chains, such as distearoyl phosphatidylcholine (58°C), is much higher than that for 1-stearoyl-2-oleoyl phosphatidylcholine (3°C) (3). The incorporation of unsaturated acyl chains prevents membranes from assuming the gel phase at common physiological temperatures.
During the de novo synthesis of phospholipids (Fig. 1A), acyl chains are incorporated into the sn-2 position by 1-acyl-snglycerol-3-phosphate acyltransferase (AGPAT), 2 also known as lysophosphatidic acid acyltransferase. The acyl-CoA substrate specificity for the reaction in rat liver microsomes (4) and by overexpressed human AGPAT1 (5,6) and AGPAT2 (6,7) shows only a mild, if any, preference for unsaturated species. Thus, whereas the reaction produces a necessary intermediate, phosphatidic acid, in the synthesis of phospholipids and triglycerides, the substrate specificity does not match the acyl chain composition of most phospholipids. However, the reaction clearly has physiological importance because mutations in AGPAT2 result in the dramatic lack of many, but not all, adipose depots in congenital general lipodystrophy (8). Given this necessity of AGPAT2 function, it is surprising that there are nine AGPAT paralogs in humans. AGPAT6 has been implicated in triglyceride synthesis in mammary glands (9). AGPAT8 encodes for an endoplasmic reticulum glycerol-3phosphate acyltransferase (10), and AGPAT9 is highly homologous to lysophosphatidylcholine acyltransferase 1 (LPCAT1) (11). Whether the members of this gene family are responsible for all AGPAT activity in mammals cannot be practically determined. The only AGPAT gene family member in the Saccharomyces cerevisiae is SLC1. SLC1 deletion mutants contain no AGPAT activity in lipid particles and ϳ50% activity in microsomes (12). Clearly, a unique AGPAT exists in the yeast genome.
The product of AGPAT, PA, can be used to synthesize phospholipids by two pathways (Fig. 1A) (reviewed in Ref. 13). Additionally, after de novo synthesis, phospholipids can undergo remodeling (Fig. 1B). This involves deacylation at the sn-2 position by phospholipase A 2 and reacylation by acyl-CoA-dependent lysophospholipid acyltransferase (Lands' Cycle) (14). There are many possible physiological functions of this remodeling that may be organism-and cell type-specific. In addition to incorporating unsaturated acyl chains for membrane fluidity, specific incorporation of the polyunsaturated arachidonyl into the sn-2 position may prime phospholipase A 2 -inducible signal cascades that involve eicosanoid synthesis (15). Alternatively, preferential incorporation of saturated acyl chains occurs during surfactant production in lungs (16). The remodeling enzymes may also re-esterify exogenous lysophospholipids generated by extracellular lipases. For instance, yeast can absorb and esterify lysophosphatidylethanolamine (lysoPE) (17). In mammals, this process may prevent the inflammation that lysophospholipids can induce (18). To date, the only cloned lysophospholipid acyltransferases are the mouse LPCAT1 and lyso-platelet-activating factor acyltransferase/ LPCAT2 (19 -21). These are primarily expressed in lungs, have a mild preference for saturated acyl-CoA, are specific for lysophosphatidylcholine (lysoPC), and have conserved motifs in common with the AGPAT gene family. In yeast, acyl-CoA-independent LPCAT (i.e. transacylase) activity is mediated by Taffazin (22), and acyl-CoA dependent LPCAT activity has been reported (2, 23), but the enzyme had not been identified.
In this study, the application of synthetic genetic array (SGA) analysis lead to the identification of the enzyme, Lpt1p. SGA is an automated approach to create double gene deletion mutants by crossing a gene deletion strain of interest and the ϳ5,000 viable yeast gene deletion strains. Inviability of the double deletion mutants (i.e. synthetic lethality) defines a relationship whereby mutation in two different genes, neither by itself lethal, causes cell death when combined in a haploid genome (24). Such genetic interactions are of particular interest because they can identify genes whose products perform complementary biological functions.

EXPERIMENTAL PROCEDURES
Materials-Synthetic complete medium and yeast nitrogen base were obtained from Q-biogene (MP Biomedicals). Myristoyl lysophosphatidylethanolamine and oleoyl lysophosphatidylserine were from Avanti Polar Lipids (Alabaster, AL). [1-14 C]Palmitoyl-lysoPC (55 mCi/mmol) was from Perkin-Elmer Life Sciences. Nourseothricin was obtained from Werner BioAgents. All other chemicals were obtained from either Sigma or Fisher.
Pulse Labeling-Pulse labeling of yeast was performed for 30 min with log phase cultures cultured in YPD with [ 3 H]oleate as described previously (30). Phospholipids were resolved by thin layer chromatography in chloroform:methanol:acetic acid:water (50:25:8:4), and neutral lipids were resolved as described previously (30). Each lane was cut according to lipid standards and counted by liquid scintillation. Assays were performed on a minimum of two independent strains of each genotype on two different days. Statistical analysis was performed using t tests.
Preparation of Cell Lysates, Mitochondria, and Microsomes-Yeast cells were grown in YP containing either 2% glucose or 3% glycerol at 30°C into early stationary phase (A 660 ϭ 1.0). SC-His medium containing 2% galactose was used for yeast harboring a plasmid. The cells were harvested and washed, and cell lysates were prepared as described previously (30). The mitochondria were prepared using a previously published protocol (31). Briefly, the cells were grown in YP medium containing 3% glycerol, harvested, and lysed, and crude mitochondria were precipitated by centrifugation at 12,000 ϫ g. Resuspension of the pellet, followed by sucrose step gradients allowed purified mitochondria to be collected. The microsomes were purified from the 12,000 ϫ g supernatant by precipitation by centrifugation at 100,000 ϫ g for 1 h (30). Lowry assays were used to measure protein concentration.
LPCAT Assay, Radioactive Substrate-LPCAT activity was measured by the incorporation of [1-14 C]palmitoyl lysoPC into PC. The reaction contained 100 mM Tris-HCl, pH 7.4, 50 M [1-14 C]palmitoyl lysoPC (50,000 dpm/nmol), 1-110 M of the respective acyl-CoA, and 3 g of cell lysate protein in a final volume of 100 l. Fixed time assays were performed for 5 min at 28°C. The reactions were stopped by adding chloroform:methanol (2:1), and lipids were extracted and resolved as described elsewhere (20). EZ-Fit software was used for nonlinear regression, curve fit analysis to calculate K m and V max . To calculate V max /K m , V max was first changed to nM/min/mg to remove volume from the units of the ratio.

RESULTS
SGA Screen with slc1⌬ Identified LPT1-To identify the second AGPAT in S. cerevisiae, a SGA screen (24) was carried out using an slc1⌬ query strain. The primary screen identified 52 nonessential genes that resulted in a synthetic lethal/sick phenotype when disrupted in combination with deletion of SLC1 (data not shown). Although the primary screen often contains a number of false positive interactions, some functionally relevant genes were identified including the fatty acid elongase gene, ELO2. The screen also identified two genes, PMT5 and YOR175C, that were previously shown to exhibit genetic interactions with SLC1 (33). PMT5 encodes for a component of the dolichyl-phosphate-mannose-protein mannosyltransferase complex, whereas a function for YOR175c (from here on referred to as LPT1) 3 was uncharacterized previously. To confirm the synthetic-lethal interaction, independent slc1⌬ and lpt1⌬ haploid strains were generated and mated to form compound heterozygous diploids. Following meiosis, 80 haploid spores were isolated by tetrad dissection. Sixty spores were viable, and all were of the wild-type, slc1⌬ or lpt1⌬ genotype. In line with Hardy Weinberg equilibrium, 1 ⁄ 4 of the progeny were therefore inviable and slc1⌬lpt1⌬, confirming the synthetic lethal interaction.
Lpt1p Primary Structure-The LPT1 gene encodes for a predicted protein of 619 amino acids and 72.2 kDa. Computer algorithm analysis predicts seven transmembrane domains (33). The C-terminal four amino acids, KKEE, agree with the conserved ER retention signal KKXX (34). This is consistent with studies that have shown Lpt1p to co-purify with ribosomes (35) and a Lpt1pgreen fluorescent protein chimera to localize to the endoplasmic reticulum (36). In terms of homology, Lpt1p belongs to the membrane-bound o-acyltransferase (MBOAT) family of proteins (37). Within the MBOAT domain (Pfam PFO3062), Lpt1p shares ϳ25% identity with three characterized members of the MBOAT family in humans (Fig. 2). The presence of this domain in Lpt1p is consistent with it being an acyltransferase.
Determining Whether Altering LPT1 Expression Alters AGPAT Activity-Initial experiments characterizing the acyltransferase activity in yeast cell lysates were performed by measuring the release of coenzyme A from acyl-CoA substrates. Assays with and without lysoPA were performed in parallel so that lysoPA-independent CoA generation could be measured and subtracted. This "background" activity was not different between wild-type and lpt1⌬ samples. Two acyl-CoA species, palmitoyl-CoA and oleoyl-CoA, were provided to represent saturated and monounsaturated substrates. To establish the amount of Slc1-independent AGPAT activity in yeast, slc1⌬ cell lysates were assayed and showed a 60% reduction, with either acyl-CoA, compared with wild-type yeast (Fig. 3A). In contrast, lpt1⌬ cell lysates showed no statistically significant changes in AGPAT activity. However, overexpression of LPT1 resulted in a 7-fold increase in AGPAT activity (Fig. 3B). This suggests that Lpt1p can mediate the reaction, but in the early stationary phase of growth, from which these cells were harvested, Slc1 is the primary AGPAT.
Lpt1p Esterifies Other Lysophospholipids-Because yeast have been shown to mediate LPCAT activity (2, 23), we tested whether Lpt1p is also responsible for that activity. Similar assays to the above were performed by replacing lysoPA with lysoPC. Activity was seen with wild-type yeast cell lysates supplied with palmitoyl-CoA, oleoyl-CoA, and arachidonyl-CoA representing saturated, monounsaturated, and polyunsaturated substrates (Fig. 4A). Unsaturated acyl-CoAs showed 4 -7fold higher activity than the saturated palmitoyl-CoA. No activity was observed in lpt1⌬ cell lysates with any of the acyl-CoAs, suggesting that Lpt1p is the main, if not only, LPCAT in yeast. Similar assays using [ 14 C]lysoPC and oleoyl-CoA showed essentially no activity with lpt1⌬ cell lysates (0.2 Ϯ 0.2 nmol/ min/mg) and robust activity in wild-type lysates (55.6 Ϯ 4.7 nmol/min/mg). To address the possibility that Lpt1p is an accessory protein, necessary for the expression and/or activity of the true LPCAT, Lpt1p was overexpressed in wild-type yeast. LPCAT activity was increased 7-fold compared with wild-type yeast harboring an empty vector (Fig. 4B), consistent with Lpt1p being the catalyst of the observed activity. LysoPC-independent activity was modestly increased (11.5 versus 6.9 nmol/ min/mg) with LPT1 overexpression.
Because Lpt1p may be the only LPCAT in yeast, we hypothesized that it will have broad substrate specificity regarding lysophospholipids. Replacing lysoPC with lysoPE, lysophos- phatidylinositol (lysoPI), or lysophosphatidylserine (lysoPS) showed distinct acyltransferase activity (Fig. 5, A and B). Comparing the enzyme activities of each lysophospholipid with oleoyl-CoA suggests that Lpt1p has a substrate preference of lysoPC ϭ lysoPI ϭ lysoPS Ͼ lysoPE (Figs. 4 and 5). Because Lpt1p can use multiple lysophospholipids as substrates, it should be referred to as a LPLAT.
Acyl-CoA Substrate Specificity for Lpt1p-The ability to esterify lysophospholipids suggests that Lpt1p is involved in phospholipid remodeling. This hypothesis predicts that Lpt1p will preferentially utilize unsaturated acyl-CoA substrates. To test this, we performed substrate concentration series experiments with [ 14 C]lysoPC and the four acyl-CoA species commonly found in yeast: palmitoyl-CoA (16:0), palmitoleoyl-CoA (16:1), stearoyl-CoA (18:0), and oleoyl-CoA (18:1). Arachidonyl-CoA (20:4) was also included to represent polyunsaturated acyl-CoAs. Curve fitting of the Michaelis-Menten plots (Fig. 6) determined the apparent K m and V max (Table 1). Palmitoyl-CoA and stearoyl-CoA showed relatively low K m and low V max values, whereas the monounsaturated palmitoleoyl-CoA and oleoyl-CoA showed relatively high K m and high V max values and arachidonyl-CoA showed intermediate values. Because all of the reactions have the same, although undetermined, enzyme concentration, V max /K m should be proportional to the catalytic efficiency (k cat /K m ). This value tended to be higher for the unsaturated acyl-CoAs than the saturated acyl-CoAs. However, because of the very low K m for palmitoyl-CoA, this trend was not absolute.
Subcellular Location of LPLAT Activity-Previous studies have identified acyl-CoA-dependent LPCAT activity in both yeast microsomes (23) and mitochondria (22). To determine the subcellular location of LPLAT activity, wild-type yeast organelles were separated by differential and density gradient centrifugation. Using lysoPC as a representative lysophospholipid in LPLAT assays, microsomes contained 10-fold enriched activity compared with whole cell lysates, whereas mitochondria showed no enrichment (Table 2). Because microsomes comprised a much larger fraction of total cell protein, they contained 65% of total cellular LPLAT activity. Mitochondria con-    OCTOBER 19, 2007 • VOLUME 282 • NUMBER 42

JOURNAL OF BIOLOGICAL CHEMISTRY 30565
tained 2% of total activity with the remainder either lost during extraction or in another cellular compartment. No activity was detected in lpt1⌬ mitochondria, suggesting that there is not a second, mitochondrial-specific LPCAT.
Role of Lpt1p in Oleate Incorporation into Glycerolipids-We next addressed the physiological importance of Lpt1p mediated LPLAT activity. Wild-type and lpt1⌬ strains were pulse-labeled for 30 min with [ 3 H]oleate. Incorporation of the radiolabel into PC in lpt1⌬ strains was 30% less than in wild-type yeast (Fig. 7). Incorporation into PE, PI, and PS and neutral lipids was not affected. Because the product of AGPAT activity, PA, is required for the de novo synthesis of phospholipids and triglycerides, the selective effect on PC synthesis may be due to attenuated remodeling. If this is the case, then of the exogenous oleate incorporated into PC in wild-type yeast, 70% is via de novo synthesis and 30% is via remodeling. Why the phenotype was limited to PC is unclear. It may be that the phospholipase A 2 component of the remodeling cycle preferentially generates lysoPC.

DISCUSSION
AGPAT mediates the second step in de novo triglyceride and phospholipid synthesis. This pathway seems to be well conserved through evolution as evidenced by all of the enzymes of the pathway in S. cerevisiae identified to date having a human homolog (38). These studies were initiated to identify a novel AGPAT in yeast that would allow a more complete analysis of phospholipid metabolism in this model organism and potentially identify novel human homologs. Because AGPAT activity is required for the de novo synthesis of phospholipids, a major component of cellular membranes, if there is only one additional AGPAT in yeast besides SLC1, a synthetic lethality screen should identify this gene. Systematically crossing slc1⌬ haploids with each of the ϳ5000 viable single-gene deletion haploids identified a total of 52 synthetic-lethal interactions. One of the identified genes was ELO2 (FEN1), which encodes for an acyl-CoA elongase required for ceramide synthesis (39). Interestingly, a gain of function mutation, SLC1-1, complements yeast that have deficient sphingolipid synthesis. Viability is conferred by the synthesis of atypical PI species that contain very long chain acyl groups, presumably with similar biophysical properties to sphingolipids (40). Perhaps the nonmutated SLC1 produces a small amount of such PI species that are required when acyl-CoA elongation is limited.
LPT1 was also identified in the screen and chosen for further characterization because of its containing a motif found in a family of MBOATs (37). In S. cerevisiae, the family members are the acyl-CoA sterol acyltransferases, Are1, Are2 (41), and the

TABLE 1 Kinetic parameters of Lpt1p
Kinetic data from Fig. 6 were used to calculate apparent K m and V max .

TABLE 2 Subcellular localization of acyl-CoA-dependent LPCAT activity
Wild-type yeast were grown in YP with 3% glycerol to A 660 ϭ 1. Cell lysates, mitochondria, and microsomes were isolated as described under "Experimental Procedures." LPCAT activity was measured using the spectrophotometric method as described in the legend to Fig. 4. The data represent the means Ϯ S.E. (n ϭ 3). glycosylphosphatidylinositol anchor remodeling protein Gup1 (42), and its paralog, Gup2. In humans, the family includes ACAT1, ACAT2, and DGAT1. Lpt1p does not share sequence similarity with Slc1 or any other AGPAT. It also does not share sequence similarity to the lysophosphatidylcholine acyltransferases LPCAT1 (20) and LPCAT2 (21) recently identified in mouse lung nor the human lysophosphatidylglycerol acyltransferase (43). Lpt1p does, however, share ϳ25% overall amino acid sequence identity to three uncharacterized human gene translation products. Whether these are functional homologs is under current investigation. The function of Lpt1p was initially investigated by comparing in vitro AGPAT activity in wild-type and lpt1⌬ cell lysates. Coenzyme A release was used to monitor AGPAT activity as has been described for measuring AGPAT, LPCAT, and DGAT activity in rat liver homogenates (32). In a validation of the assay, deletion of SLC1 caused a 60% reduction in AGPAT activity, which agreed with a previous study that found slc1⌬ microsomes incorporate 50% less [ 14 C]glycerol 3-phosphate into phosphatidic acid than wildtype cells (12). No statistically significant decrease in activity resulted by deleting LPT1. However, overexpression of LPT1 in a wild-type strain conferred a 7-fold increase in AGPAT activity. These data, along with the inviability of scl1⌬lpt1⌬, suggest that Lpt1p is the second yeast AGPAT. The largely unchanged AGPAT activity in lpt1⌬ may be due to SLC1 up-regulation.
Subsequent experiments found that Lpt1p has a more pronounced role in esterifying lysophospholipids besides lysoPA. Deletion of LPT1 completely abrogated LPCAT activity and overexpression of LPT1 in a wild-type yeast strain increased this activity 7-fold. Cell fractionation studies detected the majority (65%) of whole cell LPCAT activity in microsomes, consistent with Lpt1p residing in the ER (35,36). Although others detected acyl-CoA dependent LPCAT activity in yeast mitochondria (22), we recovered little acyl-CoA dependent activity (2%) in wild-type mitochondria and none in lpt1⌬ mitochondria. Supplying wild-type yeast lysates with lysoPE, lysoPI, and lysoPS also showed acyl-CoAdependent acyltransferase activity. No activity was observed in lpt1⌬ lysates even with providing saturated and unsaturated acyl-CoA species. The substrate preference for lysophospholipids seems to be lysoPC ϭ lysoPI ϭ lysoPS Ͼ lysoPE. However, these differences may reflect a difference in solubility in the assay buffer and/or ability to intercalate into membranes during the assay and not a difference in enzyme affinity. The ability to esterify lysoPE agrees with the recent finding that yeast can transport exogenous lysoPE and esterify it (17). The use of lysoPI as an acyl acceptor has been rarely described in the literature: in rat brain (44), pancreas (45), and cow brain (46). This broad use of acyl acceptors by Lpt1p prompted using the term LPLAT to describe its activity. Although no evidence was found for Lpt1p-independent LPLAT activity in yeast, other such enzymes may exist that are not expressed under the growth conditions used here, require assay components that were not supplied, or use other substrates such as short or very long acyl-CoAs.
As for the four long chain acyl-CoA species commonly found in yeast (47), their utilization was studied in depth with substrate concentration series experiments. Because these substrates likely undergo kinetically distinct steps, such as membrane insertion, prior to interacting with Lpt1p, all of the kinetic parameters are considered to be apparent. With lysoPC as the acyl acceptor, the saturated acyl-CoAs showed low K m and low V max values, and the monounsaturated acyl-CoAs showed relatively high K m and high V max values. For oleoyl-CoA, the K m is lower and the V max higher compared with a previous study of yeast LPCAT (K m ϭ 152 M, V max ϭ 11 nmol/min/mg) (23). The values also differed from the rat liver LPCAT activity with unsaturated acyl-CoA (K m ϭ 1-10 M; V max ϭ 8 -13 nmol/min/mg) (48). Because V max is proportional to the turnover rate of the enzyme and K m is proportional to the enzyme affinity for substrate, V max /K m yields a single value for comparing substrate utilization. This value tended to be higher for unsaturated acyl-CoA species than saturated species. The presence, and not number, of double bonds seems to be critical for substrate utilization because arachidonyl-CoA has four and was actively used.
Comparing the kinetic parameters to the Michaelis-Menten plots indicates some discordance. The reason may be that at high acyl-CoA concentration, activity is limited by a detergent effect. Perhaps a better way to analyze the kinetic data is to compare the activity at physiological cellular concentrations of each acyl-CoA. Because yeast have ϳ0.1 nmol of each acyl-CoA in 10 9 cells (47) and the average volume of a yeast haploid is 70 m 3 , the cellular concentration of each acyl-CoA is ϳ2 M. If this gross estimate is correct, then Lpt1p likely mediates similar activity with saturated and unsaturated acyl-CoA inside cells. If the physiological acyl-CoA concentrations are 10 M or higher, then the use of unsaturated acyl-CoAs will predominate.
The physiological function of Lpt1p-mediated LPLAT activity was also investigated by pulse labeling wild-type and lpt1⌬ strains with [ 3 H]oleate. There was a selective reduction in incorporation into PC by 30%. It seems unlikely that this difference is due to decreased AGPAT activity because phosphatidic acid provides a precursor for all the phospholipid species, synthesized via diacylglycerol or CDP-diacylglycerol, and triglyceride. More likely, the difference is due to the deficient LPLAT activity in lpt1⌬. The lack of an effect on incorporation into PE or PI may reflect a low cellular abundance of lyso-species of these phospholipids. Whether S. cerevisiae has a phospholipase A 2 with preference for PC has not been determined. However, yeast actively use PC as an acyl donor for triglyceride synthesis and thus actively produce lysoPC via a phospholipid diacylglycerol acyltransferase reaction mediated by Lro1p (49). This may be the main source of Lpt1p substrates. In addition to constitutive remodeling, Lpt1p may also function to limit the abundance of lysophospholipids and avoid inappropriate stimulation of adenylate cyclase in yeast (50). LPLAT activity also allows exogenous lysophospholipids to be salvaged after uptake and directly utilized for cell membranes (17).
In S. cerevisiae, the preferential incorporation of unsaturated acyl chains into the sn-2 position of phospholipids seems to be achieved by a combination of substrate selectivity during de novo synthesis, primarily by Slc1p, in the Kennedy Pathway and remodeling by Lpt1p in the Lands cycle. This is supported by a deletion of both genes being lethal and the preference of both reactions for unsaturated acyl-CoA substrates. Within the family of membrane-bound o-acyltransferases, LPT1 and its orthologs may comprise a novel branch with important roles regarding phospholipid composition.