The glycerophosphocholine acyltransferase Gpc1 is part of a phosphatidylcholine (PC)-remodeling pathway that alters PC species in yeast

Phospholipase B–mediated hydrolysis of phosphatidylcholine (PC) results in the formation of free fatty acids and glycerophosphocholine (GPC) in the yeast Saccharomyces cerevisiae. GPC can be reacylated by the glycerophosphocholine acyltransferase Gpc1, which produces lysophosphatidylcholine (LPC), and LPC can be converted to PC by the lysophospholipid acyltransferase Ale1. Here, we further characterized the regulation and function of this distinct PC deacylation/reacylation pathway in yeast. Through in vitro and in vivo experiments, we show that Gpc1 and Ale1 are the major cellular GPC and LPC acyltransferases, respectively. Importantly, we report that Gpc1 activity affects the PC species profile. Loss of Gpc1 decreased the levels of monounsaturated PC species and increased those of diunsaturated PC species, whereas Gpc1 overexpression had the opposite effects. Of note, Gpc1 loss did not significantly affect phosphatidylethanolamine, phosphatidylinositol, and phosphatidylserine profiles. Our results indicate that Gpc1 is involved in postsynthetic PC remodeling that produces more saturated PC species. qRT-PCR analyses revealed that GPC1 mRNA abundance is regulated coordinately with PC biosynthetic pathways. Inositol availability, which regulates several phospholipid biosynthetic genes, down-regulated GPC1 expression at the mRNA and protein levels and, as expected, decreased levels of monounsaturated PC species. Finally, loss of GPC1 decreased stationary phase viability in inositol-free medium. These results indicate that Gpc1 is part of a postsynthetic PC deacylation/reacylation remodeling pathway (PC-DRP) that alters the PC species profile, is regulated in coordination with other major lipid biosynthetic pathways, and affects yeast growth.

The steady-state PC molecular species profile is the result of the molecular specificity of the PC biosynthetic routes combined with post synthetic acyl chain exchange (21). The PE methylation pathway produces predominantly diunsaturated PC species, whereas the CDP-choline pathway produces a more mixed profile (21,24). Furthermore, a number of studies have provided evidence for the postsynthetic remodeling of PC species (20,25). Indeed, PC remodeling was clearly demonstrated by pulse-chase studies using deuterium-labeled (methyl-D 3 )methionine followed by electrospray ionization-MS/MS analysis. Those studies revealed a postsynthetic increase in monounsaturated PC species (32:1 and 34:1) at an expense of diunsaturated PC species (32:2 and 34:2) (24). Prior to this work, an acyltransferase responsible for remodeling PC to more saturated species in S. cerevisiae had not been identified.
Here we undertook a more thorough examination of PC remodeling via PC-DRP and the role of the GPC acyltransferase, Gpc1. Importantly, we report that Gpc1 is a key player in postsynthetic PC remodeling events that result in more saturated PC species. In addition, we find that Gpc1 is regulated with respect to other aspects of lipid metabolism and that loss of Gpc1 impacts stationary phase viability.

GPC can be hydrolyzed by Gde1 or acylated by Gpc1
Cellular GPC has two potential metabolic fates. The glycerophosphodiesterase encoded by GDE1 can hydrolyze GPC to produce glycerol-3-phosphate and choline (14,19). Additionally, GPC can be acylated (likely at the sn-1 position) to LPC by Gpc1, a recently described acyltransferase (17). LPC can be converted to PC by Ale1, consistent with its role as a lysophospholipid acyltransferase (18,26). To gain further insight into the relative flux of GPC toward hydrolysis versus acylation, in vivo metabolic labeling was employed in strains lacking GPC1 and/or GDE1 (Fig. 2). Cells were grown in the presence of [ 14 C]GPC, which enters through the Git1 transporter (27)(28)(29), and label incorporation in the lipid fraction was determined. The [ 14 Fig. 2A; therefore only the label in PC is quantified. As shown in Fig. 2A, a gpc1⌬ mutant incorporates roughly 35% less label into PC as compared with WT, consistent with its role as a GPC acyltransferase. A gde1⌬ mutant incorporates 50% less label into PC, indicating that a large portion of the label incorporated into PC in a WT strain under low phosphate conditions originates from free [ 14 C]choline released through the action of Gde1. A gde1⌬ gpc1⌬ double mutant incorporates even less label, confirming the importance of both metabolic routes in determining the ultimate fate of GPC. The label remaining in the gde1⌬ gpc1⌬ double mutant is likely the result of an uncharacterized acyltransferase and/or glycerophosphodiesterase that can act on GPC.

Gpc1 and Ale1 are central players in the stepwise acylation of GPC to LPC and then to PC
To demonstrate that Gpc1 produces LPC in vivo and thereby confirm its cellular role, we designed an alternative labeling scheme and employed an ale1⌬ mutant to block the second acylation step (Fig. 2B). In the experiment represented in Fig.  2A, cells were grown under low-phosphate conditions to induce the expression of GIT1 (28), a condition that also induces the expression of GDE1 (14,30). Here, a plasmid harboring GIT1 under the control of the constitutive ADH1 promoter (31) was employed to allow cell growth under normal phosphate level, a condition that represses the expression of GDE1 and therefore reduces the hydrolysis of GPC once it enters the cell (14,30). In addition, the labeling time in the presence of [ 14 C]GPC was reduced to roughly two generations. Under these conditions, a 3-fold increase in labeled LPC was detected in an ale1⌬ mutant as compared with WT, and that increase was counteracted by simultaneous deletion of GPC1 (see the ale1⌬ gpc1⌬ double mutant). At the same time, labeled PC levels decreased in an ale1⌬ mutant and decreased further in an ale1⌬ gpc1⌬ double mutant. The fact that PC levels were reduced by only 50% in the ale1⌬ strain suggests that other enzymes may exist that are capable of converting LPC to PC in the cell. However, an additional possibility is that the increased LPC being produced is hydrolyzed, ultimately releasing free choline that could be incorporated into PC via the Kennedy pathway. These results confirm the formation of the immediate product of Gpc1 activity, LPC, in vivo, and the role of Ale1 in converting LPC into PC.

Gpc1 impacts the PC molecular species profile
A potential function for the PC-DRP is to modify the acyl chain content. To determine whether Gpc1 activity affects the PC molecular species profile, LC-MS/MS was performed (parent ion scanning for m/z 184). The cells were not supplemented with exogenous GPC for these experiments, so the substrate for Gpc1 activity presumably arose through endogenous phospholipase B activity (19). Fig. 3A includes the range of PC species produced by The PC-DRP is indicated by red arrows. The PC-DRP includes deacylation of PC to GPC by PLB1 or NTE1 followed by stepwise reacylation of GPC to LPC by GPC1 and LPC to PC by ALE1. The de novo PC synthesis routes are indicated by dotted arrows. Gene names are in italics. PLA, phospholipase A; PLB, phospholipase B.

PC remodeling in S. cerevisiae
yeast. Focusing on the major species, the data indicate that the loss of Gpc1 results in a decrease in monounsaturated PC species (32:1, 34:1) and an increase in diunsaturated PC species (32:2, 34:2) as proportions of the total PC pool. Overexpression, as predicted, has the opposite effect (Fig. 3B). When GPC1 was placed under the control of the GAL1 promoter and cells were grown on galactose, there was an increase in monounsaturated PC species (32:1 and 34:1) and a decrease in diunsaturated PC species (32:2, 34:2). In total, these results indicate that Gpc1 impacts the PC species profile and suggest that Gpc1 activity favors the utilization of saturated acyl-CoA species when acylating GPC.

Loss of Pct1 and Ale1 has minor effects on the PC species profile
Because Ale1 converts LPC to PC as part of the PC-DRP, we analyzed its impact on the PC molecular species profile. If Ale1 The cells were harvested, phospholipids were extracted and separated, and PC was quantified as described under "Experimental procedures." B and C, indicated strains contained plasmid in which GIT1 was constitutively expressed under the control of the ADH1 promoter. The cells were grown in the presence of [ 14 C]choline-GPC for 4 h and harvested, and phospholipids were extracted and analyzed as described under "Experimental procedures." The data are normalized to the WT containing ADH1-GIT1. Roughly 100-fold more PC was detected as compared with LPC in the WT strain. The data represent the averages of three independent cultures Ϯ S.D. A t test was performed to determine significance as indicated. *, p Յ 0.05; **, p Յ 0.005; ***, p Յ 0.0005.

PC remodeling in S. cerevisiae
were the only LPC acyltransferase in the cell, and assuming that Ale1 does not affect the species profiles of the lipid precursors of PC (PE and diacylglycerol), the profile of the ale1⌬ mutant should resemble that of the gpc1⌬ mutant, because the reacylation pathway would stall at LPC, and the LPC produced would likely be rapidly hydrolyzed by phospholipases. In fact, the ale1⌬ mutant did not resemble the gpc1⌬ strain in terms of PC species, because a slight decrease in PC32:2 and an increase in PC34:2 were its only significant changes from WT. The ale1⌬ gpc1⌬ mutant, like gpc1⌬, exhibited an increase in diunsaturated species and a decrease in monounsaturated species as compared with WT, although the magnitude of the changes was not identical between the two strains (Fig. 4). These results suggest that there may be more acyltransferases capable of acylating LPC in the cell (32,33).
The steady-state PC molecular species profile in a cell represents a combination of PC biosynthesis, PC degradation, and PC acyl chain remodeling. In the experiments presented in Fig.  3, cells were grown in the absence of free choline, so the PE methylation pathway, not the Kennedy pathway, was primarily responsible for PC biosynthesis. However, flux through the Kennedy pathway can also occur as a result of free choline released through phospholipase-mediated turnover. To examine the extent to which PC turnover resulting in free choline release followed by resynthesis contributes to the observed PC species profile, we utilized a pct1⌬ mutant in which flux through the Kennedy pathway is blocked. As shown in Fig. 4, the PC species profile in pct1⌬ was largely unchanged from WT, in agreement with Ref. 22, but did result in a small but significant increase in 32:1 PC. That increase may be the result of the observed uptick in GPC1 transcript upon PCT1 deletion (see Fig. 6), as evidenced by the fact that 32:1 again decreases in a pct1⌬ gpc1⌬ double mutant. Overall, these results indicate that the PC species profiles generated under our experimental conditions are primarily the result of PC synthesis via the PE methylation pathway followed by PC remodeling via PC-DRP.

Loss of Gpc1 does not affect PE, PI, and PS species profiles and slightly decreases total 16:0 FA content
We reported previously that Gpc1 had limited ability to utilize acyl-CoAs to acylate glycerophosphoethanolamine in an in vitro system (17). To probe the possibility that Gpc1 acylates glycerophosphoethanolamine to affect PE in vivo, we analyzed the PE molecular species. As shown in Fig. 5A, we did not detect any significant difference in PE molecular species profile when comparing a WT strain to a gpc1⌬ strain. Hence, Gpc1 does not play a major role, if at all, in PE remodeling in vivo under the conditions employed here.
The effect of GPC1 deletion on the phosphatidylinositol (PI) and phosphatidylserine (PS) species profiles was also examined. As shown in Fig. 5 (B and C), no significant differences between WT and gpc1⌬ were found. In addition, the overall fatty acid A and B, the indicated strains were grown to late log phase, the cultures were harvested, and the lipids were extracted. PC species were separated and analyzed using LC-MS/ MS, as described under "Experimental procedures." The data represent averages of three independent cultures Ϯ S.D. A t test was performed to determine significance as indicated. *, p Յ 0.05. The indicated strains were grown to late log phase, the cultures were harvested, and the lipids were extracted. PC species were separated and analyzed as described under "Experimental procedures." The data represent averages of three independent cultures Ϯ S.D. A two-way analysis of variance was performed to determine significance for each strain compared with the WT. *, p Յ 0.05; **, p Յ 0.004; ***, p Յ 0.0001.

PC remodeling in S. cerevisiae
composition of all cellular lipids was determined by GC. As shown in Fig. 5D, WT and gpc1⌬ fatty acid compositions are largely the same, with the exception of a small but significant decrease in C16:0 in gpc1⌬. This change is consistent with the fact that the gpc1⌬ mutant displays a decrease in monounsaturated PC species and may reflect the loss of a small C16:0 sink resulting from the inability of gpc1⌬ to add C16:0 to GPC during the PC remodeling process. The results indicate that loss of GPC1 does not induce large-scale changes in fatty acid composition but rather highly specific effects.

GPC1 expression is up-regulated by attenuation of PC biosynthesis and by inositol limitation
To gain further insight into the physiological role of Gpc1, we examined its transcriptional regulation in response to alterations in major lipid biosynthetic pathways. To examine PC biosynthesis, we employed a cho2⌬ mutant (attenuated PE methylation pathway), a pct1⌬ mutant (blocked Kennedy pathway), and a cho2⌬ pct1⌬ mutant. The PE methylation pathway for PC biosynthesis is not completely blocked in the cho2⌬ pct1⌬ double mutant, because of a partially redundant function supplied by the methyltransferase, Opi3 (34). As shown in Fig. 6A, GPC1 message levels, as measured by qRT-PCR, are increased roughly 3-fold in each single mutant and in the double mutant as compared with WT. Thus, inhibition of either pathway for PC biosynthesis causes an increase in GPC1 message, but the effect is not additive.
A number of phospholipid biosynthetic genes are regulated by the presence of inositol in the medium via a complex mechanism involving the modulation of phosphatidic acid levels at the ER, which in turn impacts the transit of the Opi1 repressor protein into the nucleus (35,36). Thus, inositol availability has a general impact on phospholipid composition and flux through the various pathways (10,35). As shown in Fig. 6B, there is a roughly 5-fold decrease in GPC1 message upon inositol supplementation.

Inositol supplementation represses GPC1 expression and increases diunsaturated PC species
Following the qRT-PCR findings (Fig. 6), the effect of inositol availability was examined more thoroughly, because this perturbation produced the largest transcriptional response and Figure 5. Loss of GPC1 has no significant effects on PE, PS, and PI species profiles and slightly decreases the total C16:0 acyl chain content. The indicated strains were grown to log phase and harvested, and the lipids were extracted. A-C, PE (A), PS (B), and PI (C) species were analyzed by MS as detailed under "Experimental procedures." D, total acyl chain composition was determined using GC. The data represent averages of four independent cultures Ϯ S.D. A t test was performed to determine significance as indicated. *, p ϭ 0.0008.

PC remodeling in S. cerevisiae
would not confound the interpretation of the PC species profile by impacting PC biosynthesis (as would the cho2⌬ and cho2⌬ pct1⌬ mutants). As seen in Fig. 7A, the decrease in GPC1 transcript upon inositol supplementation was reflected in a decrease in Gpc1 as measured by Western blotting analysis. It should be noted that the PC species profiles presented in Fig. 3 were performed on cells grown in the absence of inositol (IϪ) to maximize GPC1 expression. When grown in the presence of inositol (Iϩ), there is a clear increase in diunsaturated PC species and a clear decrease in monounsaturated PC species (Fig.  7B) in a WT strain, similar to the effect of GPC1 deletion (Figs. 3A and 7B) in IϪ medium. A similar pattern of changes in PC molecular species profile upon inositol limitation was reported previously (37). Loss of Gpc1 and inositol supplementation together result in an even greater increase in diunsaturated species, indicating that factors beyond Gpc1 are at play in impacting the PC species profile under these conditions.

Loss of Gpc1 impacts growth and stationary phase viability in inositol-free medium
Growth assays were performed to assess the phenotypic consequences of diminished PC remodeling as a function of inositol availability. On plates, gpc1⌬ displayed a slight defect in growth in IϪ conditions, but not Iϩ conditions (Fig. 7C). This difference was more clearly evident in liquid culture. As shown in Fig. 8A, gpc1⌬ mutants displayed an increase in lag time, resulting in slower growth overall, although the rate of doubling during logarithmic phase was comparable with WT. To determine whether decreased ability of cells to regrow from stationary phase was the reason behind the increased lag time, a recultivation experiment was performed. The cells were grown to stationary phase (as in Fig. 8A), but the cultures were incubated longer, for a total of 4 days. On days 3 and 4, the cells were taken from the original cultures and inoculated in fresh YNB IϪ medium. By day 3, gpc1⌬ lagged well behind WT in its ability to reinitiate growth as measured at three time points over 32 h ( Fig. 8B; Day 3). In 4-day-old cultures, no growth was detected in the gpc1⌬ cultures as measured over 32 h (Fig. 8C; day 4). These results indicate that loss of Gpc1 results in a loss in stationary phase viability in inositol-free medium. Western blotting analysis was performed to examine Gpc1 protein abundance as a function of growth phase. Although Gpc1 was found to be expressed in logarithmic phase (Fig. 9), little or no expression was detected in stationary phase (24 and 48 h of growth). This finding suggests that no further remodeling of PC via this pathway occurs once the cells reach stationary phase.

Discussion
PC metabolism consists of interconnected pathways of synthesis, degradation, recycling of catabolites, and remodeling ( Fig. 2A). Deacylation of PC via phospholipases of the B type produces GPC. Our previous studies have shown that GPC is found both intracellularly and extracellularly and that it can be transported into the cell via the Git1 permease (14,29). A well-

PC remodeling in S. cerevisiae
characterized pathway for GPC degradation is its hydrolysis to choline and glycerol-3-phosphate by the glycerophosphodiesterase, Gde1 (14,15). Prior to the identification of GPC1, GPC acyltransferase activity had been detected in Xanthomonas campestris, S. cerevisiae, and plants (16,38,39). Our recent identification of the gene encoding Gpc1 provided a tool for examining the relative flux of GPC between degradation versus acylation. When considering the label incorporated into PC from exogenous GPC, roughly 50% occurs via Gde1 activity producing free choline that is subsequently utilized by the Kennedy pathway. Roughly 30% is acylated by Gpc1, and another 20% is presumably metabolized by unknown enzymes. Thus, reacylation is not a trivial pathway for GPC conversion to PC. Through the use of mutants and radiolabeling (Fig. 2B), the primary roles of Gpc1 and Ale1 in the stepwise conversion of GPC to PC as part of the PC-DRP were confirmed. . Western blotting analysis was performed using anti-HA mouse IRdye 680 and goat anti-mouse IRdye 800. Glucose-6-phosphate dehydrogenase was used as a loading control. Protein bands were quantified using Image Studio TM software. The data represent two biological replicates. B, the indicated strains were grown to late log phase before harvesting. Phospholipids were extracted, and PC species were analyzed using LC-MS/MS. The data represent averages of three independent cultures Ϯ S.D. A t test was performed to establish significance. ****, p Յ 0.0005. C, strains were cultured in synthetic medium, harvested, and resuspended in sterile water. 10-fold serial dilutions (5 l) were spotted onto plates containing (ϩInositol) or lacking (ϪInositol) inositol.

PC remodeling in S. cerevisiae
Although the acylation of GPC to PC clearly occurs in the cell, the synthesis of PC at a level that supports growth requires either a functional PE methylation pathway or a functional Kennedy pathway (25,40). Hence, we hypothesized that the primary purpose of GPC reacylation may not be bulk synthesis but rather the synthesis of specific PC species as part of a remodeling pathway. Evidence for postsynthetic PC remodeling in yeast has been described by multiple research groups (20,25,(41)(42)(43). Importantly, de Kroon and co-workers (22) have used stable-isotope pulse-chase analysis coupled with MS to prove the postsynthetic formation of more saturated PC species as a function of time and demonstrated the involvement of Plb1. Prior to the identification of Gpc1 activity (16,17), PC remodeling was believed to involve the removal of a single acyl chain at the sn-2 position, followed by reacylation by a LPC acyltransferase, a cycle first described by Lands (44). Once LPC is formed, it can be acylated to PC by Ale1 (18,26,45), with some publications indicating that Ale1 prefers incorporation of an unsaturated acyl-CoA species at the sn-2 position of 1-acyl LPC (18,(45)(46)(47)(48). A recent publication has provided evidence that Ale1 may also have limited ability to acylate the sn-1 position of the synthetic LPC analog, 2-alkyl-sn-glycero-3-phosphocholine (43). Importantly, we posit that the previous step in the remodeling pathway, the acylation of GPC, is carried out by the glycerophosphocholine acyltransferase activity of Gpc1, adding another dimension to the process of PC remodeling, namely the potential exchange of both acyl chains. Of course, this finding does not preclude the possibility of PC remodeling

PC remodeling in S. cerevisiae
occurring via other acyltransferase or transacylase activities (32,33). PC molecular species profiling of cells lacking GPC1 (Fig. 3A) or overexpressing GPC1 (Fig. 3B) clearly indicates that Gpc1 impacts the PC species profile, producing a higher content of more saturated PC species (32:1 and 34:1) when present. The impact of Gpc1 on phospholipid molecular species is limited to PC, because PI, PE, and PS species profiles are unchanged by loss of Gpc1.
The obvious general question arising from these results is under what circumstances are Gpc1 activity and PC remodeling beneficial? The PC molecular species profile is highly dynamic (49,50). Given that changes in PC saturation status may well affect a membrane property such as fluidity, we reasoned that Gpc1 must be co-regulated with regards to other lipid metabolic processes to maintain optimal bilayer function. As a first approach, we utilized qRT-PCR to assess changes in transcript levels by attenuating the methylation pathway (cho2⌬) or blocking the Kennedy pathway (pct1⌬) either singly or in combination results in an uptick in GPC1 message levels. Granted, this experiment does not have a straightforward interpretation, because the inhibition of the major PC biosynthetic pathways may cause an increase in GPC reacylation as an end toward preserving PC, not remodeling PC. By similar reasoning, we did not test the effects of choline supplementation in this study (but will be examined in future studies), because that perturbation would increase biosynthesis through the CDPcholine pathway, potentially obscuring the interpretation of the findings. Nonetheless, our results clearly show that the GPC1 message increases upon inhibition of PC biosynthesis.
Inositol availability has far-reaching transcriptional and physiological effects in yeast. Inositol is required for PI biosynthesis. In turn, PI is a precursor for phosphoinositides, polyphosphates, inositol sphingolipids, and glycosylphosphatidylinositol anchors (10,35,37). Inositol can be synthesized by the cell and/or imported into the cell from the medium (51). In the absence of exogenous inositol, a number of genes involved in phospholipid metabolism are transcriptionally induced (10,35). Inositol supplementation results in a 5-fold down-regulation of GPC1 message and a 3-fold decrease in protein level (Figs. 6B and 7A). Down-regulation of GPC1 transcript and protein levels by inositol supplementation resulted in changes in the PC species profile, providing an explanation for a similar result reported by Henry and colleagues (37) prior to the identification of Gpc1.
Most genes responsive to inositol limitation contain one or more UAS INO elements (consensus sequence 5Ј-CATGT-GAAAT-3Ј) in their promoters and are regulated by Ino2, Ino4, and Opi1 via the Henry Regulatory Circuit (35,52,53). The transcription factors, Ino2 and Ino4, typically bind to UAS INO as a heterodimer to activate transcription when inositol is limiting, and the Opi1 repressor binds to Ino2 to block transcriptional activation when inositol is available. The Gpc1 promoter contains a sequence (CTTGTGAATA) that is 3 bp off from the consensus Ino2-Ino4 binding sequence (54) at Ϫ193 to Ϫ202. In addition, it contains a site (CATTTG) for Ino4-dependent/ Ino2-independent binding (55) at Ϫ182 to Ϫ187 bp. Further evidence for control by the Henry Regulatory Circuit is provided by microarray studies performed prior to the characterization of Gpc1 in which Opi1 (56) and Ino4 (57) were shown to affect transcription of the GPC1 ORF, YGR194w. Further studies would be required to unequivocally prove binding of Ino2 and/or Ino4 to the GPC1 promoter.
In media lacking inositol, where GPC1 is up-regulated, gpc1⌬ mutants display decreased stationary phase viability (Fig. 8). However, we did not detect Gpc1 expression in the stationary phase, suggesting that the alteration in PC species that occurs via PC-DRP during log phase may prepare the membrane for events that occur in later stages of growth. Gpc1 is predicted to be an integral membrane protein with eight transmembrane domains (17). A definitive localization study has not been performed for Gpc1 but will be the focus of future studies and may shed light on its role in cell viability. Of note, the changes in lipid metabolism resulting from IϪ medium are accompanied by the induction of multiple stress-response pathways, including the unfolded protein response, the protein kinase C-mitogen-activated protein kinase pathway, and others (35). Thus, the loss of cell viability observed under IϪ conditions upon loss of GPC1 is likely the result of a complex series of events. In summary, our data indicate that Gpc1 is part of PC-DRP, a remodeling pathway that provides the opportunity for postsynthetic alteration of both acyl chains species and, thereby, to impact the biophysical properties of the membrane.

Strains, plasmid, media, and growth conditions
The S. cerevisiae strains used in this study are presented in Table 1. Strains were maintained aerobically at 30°C with shaking or on a roller drum. Growth was monitored by measuring

PC remodeling in S. cerevisiae
optical densities at 600 nm (A 600 nm ) using a BioMate3 Thermo Spectronic spectrophotometer. The media used in this study were yeast peptone dextrose or yeast nitrogen base (YNB) with 2% glucose and amino acid composition as described in Ref. 58. The YNB medium was made according to DIFCO manual but lacking inositol. For studies requiring inositol, 75 M of inositol was provided. Transformations of autonomously replicating plasmids were performed using the lazy bones plasmid transformation protocol as described (59). For integrating DNA into the genome, the high-efficiency transformation procedure was used (60). The empty vector, pRS426, and vector overexpressing GIT1 gene ADH-GIT1 were from Ref. 31.

Construction of deletion strains
The WT (BY4742) S. cerevisiae strain was purchased from Open Biosystems (Thermo Scientific, Huntsville, AL). The deletion strains used in this study were made using PCR-based homologous recombination technique as described in Refs. 61 and 62). Nutritional markers, LEU2 and URA3, were amplified from pRS415 and pRS416, respectively, using primers listed in Table 2.

Construction of chromosomal C-terminal 3؋HA tag for GPC1
The 3ϫHA-URA3 cassette was amplified from plasmid pPMY-3ϫHA (63) using the primer set shown in Table 2. The PCR product was transformed into WT strain, and transformants were selected on YNB plates lacking uracil. Control PCR was performed on genomic DNA extracted from transformants to verify the integration of the entire cassette into the genome. To counterselect against the URA3 gene, the verified colonies were then reselected on a YNB plate containing 1 mg/ml of 5-fluoroorotic acid. Genomic DNA was extracted from colonies that grew on 5-fluoroorotic acid plates and were used as template to check for the insertion of 3ϫHA at the C-terminal end. Colonies producing the correct amplicon are JPV846 (WTϩGPC1-3ϫHA) used in the Western blotting analyses.

In vivo labeling and lipid isolation
For the long-term labeling experiment represented in Fig.  2A, strains were grown in YNB inositol-free medium containing 5 M of [ 14 C]GPC (Х200,000 cpm/ml) (American Radiolabeled Chemicals 3880) and a low concentration of KH 2 PO 4 (200 M) to induce expression of the Git1 transporter for GPC uptake. Cultures were allowed to grow to logarithmic phase and harvested, and the cell pellets were treated with 5% TCA for 20 min on ice. Following centrifugation, the supernatant was discarded, and cell pellets were incubated at 60°C for 60 min with 1 ml of ESOAK (95% ethanol, diethylether, H 2 O, pyridine, NH 4 OH (28 -30%); 15:5:15:1:0.036 v/v/v/v/v). The tubes were centrifuged to pellet the debris, and 1 ml of lipid-containing supernatant was transferred to fresh tubes containing 2.5 ml of chloroform/methanol (2:1) and 0.25 ml of 0.1 M HCl. Following vortexing and low-speed centrifugation, the bottom layers containing glycerophospholipids were dried under N 2 . Lipids were then suspended in chloroform/methanol (2:1) and spotted onto silica gel TLC plates (Whatman, Millipore Sigma 105626), and plates were developed in chloroform:methanol:acetic acid:water; (85:15:10:3.5 v/v/v/v) (39). The TLC plates were imaged using a Typhoon 8200 phosphorimager. For Fig. 2B, ImageQuant software was used to quantitate the PC and LPC spots. For Fig. 2A, no LPC was detected, allowing the radioactivity (cpm) in the lipid fraction to be converted to pmol PC/ODU based on the specific activity of the exogenous label.
For the experiments aimed at detecting LPC (Fig. 2B), strains containing either empty vector or ADH-GIT1 (31) were grown in YNB inositol-free medium containing 5 M of [ 14 C]GPC (Х200,000 cpm/ml) and a standard level of KH 2 PO 4 . The cultures were inoculated at an A 600 nm of ϳ0.2 from startup cultures. Cultures were allowed to grow for 4 h before harvesting ϳ3.2 ODUs of cells for each strain. Lipid extraction, separation, and quantitation were as described in previous paragraph. LPC and PC standards (Avanti Polar lipids) were used to verify lipid migration on the TLC plates. Radiolabeled GPC was purchased from American Radiolabeled Chemicals, and unlabeled GPC was from Sigma Scientific.

Analysis of PC and PE molecular species profiles by MS
The indicated strains were grown to logarithmic phase, and 30 ODUs of each were harvested. The cell pellets were suspended in 2 ml of 5% TCA, followed by a 15-min incubation on ice. After centrifugation, the cell pellets were washed with 20 ml of Nano-pure H 2 O. Phospholipids were extracted from the pellet as described above and were finally resuspended in 500 l of CHCl 3 :MeOH (1:1, v/v). A 5-l lipid sample was injected into an Agilent 1200 HPLC system coupled to an Agilent 6460 triple quadrupole LC/MS system. Lipid separations were performed on a Varian PLRP-S (150 ϫ 2.1 mm, 5 m) column at a flow rate of 0.35 ml/min, and the column was maintained at 50°C. Samples were eluted with a linear gradient using of 75% methanol in water (A) and 100% methanol (B). Both solvents contained 10 mM ammonium acetate and 0.1% formic acid. Electrospray ionization-MS was performed with scanning mode set to multiple reaction monitoring set at m/z 184.1 product ion to detect PC molecules. PE species were detected by neutral loss scanning at 141 atomic mass units in positive ion mode. The MS parameters were optimized at: collision energy, 25 eV for PC and 15 eV for PE; capillary voltage, 3500 V; fragmentor voltage, 100 V; dwell time, 200 ms; drying gas temperature, 320°C; sheath gas temperature, 350°C; sheath gas flow, 11 liters/min Reverse GTAAAGCGCCTGTAAATAAAAGCTCTCAAAGTTAACAGATAAATGAAGTGAACTATCTTACTGTAGGGCGAATTGGG PC remodeling in S. cerevisiae and drying gas flow, 10 liters/min. The data were acquired and analyzed using Agilent Mass Hunter Work Station software. Reported values for various PC and PE species were corrected for isotope effect (64).

Analysis of PI and PS molecular species profiles by MS
Total lipid extracts were dissolved in CHCl 3 :MeOH 1:1 and injected in a Ultra-High-Performance Liquid Chromatography (UHPLC) system equipped with a Kinetex HILIC column (50 ϫ 4.60 mm, 2.6 m; Phenomenex, Utrecht, The Netherlands) with a SecurityGuard ULTRA HILIC precolumn. Lipids were eluted using a binary gradient of acetonitrile:acetone 9:1 (v/v) (with 0.1% formic acid; eluent A) and 50 mM ammonium formate in acetonitrile:H 2 O 7:3 (v/v) (with 0.1% formic acid; eluent B). The linear gradient was from 100% A to 50% A in 1 min, followed by isocratic elution at the latter composition for an additional 2 min before a 1-min cleaning of the column with 100% B. The flow rate throughout the analysis was 1 ml/min. MS analysis was performed on an LTQ-XL mass spectrometer (Thermo Scientific) using electrospray ionization and detection in the negative mode (source voltage Ϫ4.2 kV). Full-scan mass spectra were recorded over a range of 450 -950 m/z. The data were converted to mzML format and analyzed using XCMS version 1.52.0 (65) running under R version 3.4.3, including isotope correction.

Analysis of acyl chain composition
Total cellular acyl chain composition was analyzed as described previously (66). Briefly, total lipid extracts corresponding to 100 nmol phospholipid phosphorus were dried, and transesterified in methanolic H 2 SO 4 (40:1 v/v) at 70°C. Fatty acid methyl esters (FAMEs) were extracted with hexane and separated on a Trace GC (Interscience, Breda, NL) equipped with a biscyanopropyl-polysiloxane column (Restek, Bellefonte, PA). FAMEs were identified, and signal intensity was calibrated using a commercially available standard (Nu-Chek, Elysian, MN). Acyl chain compositions are presented as mol % of total FAMEs recovered.

RNA extraction and qRT-PCR
RNA was extracted from logarithmic phase cells using the hot phenolic RNA extraction protocol (67). The extracted RNA sample was quantified using Thermo Scientific NanoDrop UVvisible spectrophotometry. DNase treatment was performed on 7 g of RNA using the TURBO DNA-free TM kit (Applied Biosystems) as per the manufacturer's protocol. The Verso TM SYBR Green one-step QRT-PCR ROX kit (Thermo Scientific) was used to determine the message levels on an Applied Biosystems StepOne-Plus TM real-time PCR thermocycler. The primer nucleotide sequence sets used in these studies are shown in Table 3. The optimal primer concentration (70 -100 nM) and tem-plate concentration (100 ng/25 l reaction) were determined by performing a standard curve on each primer set. Reverse transcription was performed at 50°C for 15 min followed by 95°C for 15 min for RT inactivation and polymerase activation. The thermal profile for amplification were 40 cycles at 95°C for 15 s, 54.5°C for 30 s, and 72°C for 40 s (62). Primer set specificity was determined by melting curve analysis. A no-template control and a no-RT control were used to validate the absence of contaminants in RNA samples. Experiments were performed using three independent biological replicates, and each replicate was analyzed in experimental triplicates. The data were normalized to the endogenous control gene, SNR17, and analyzed by the ⌬⌬C t method (68,69).

Protein extraction and Western blotting analysis
For Fig. 7A, the cells were grown in YNB medium containing or lacking 75 M of inositol. Cultures corresponding to 25 ODUs of cells were harvested at log phase. The cell pellet was resuspended in ice-cold extraction buffer (120 mM NaCl, 50 mM Tris-HCl, pH 7.5, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1% Nonidet P-40, 0.1% SDS, and 1% Triton X-100) containing 1ϫ Yeast/Fungal Protease Arrest TM mixture (G-Biosciences catalog no. 786-333). Lysis was performed using glass bead for eight cycles of the following: 30 s of vortexing and 30 s of incubation on ice. The cell debris was pelleted by centrifugation at 2000 ϫ g for 5 min. For Fig. 9, the cells were grown in YNB medium lacking inositol. Equivalent ODUs of cells were harvested at log and stationary (24 h) and late stationary phases (48 h). The cell were lysed for 10 min on ice with 50 l of extraction buffer (0.2 M NaOH, 0.2% mercaptoethanol) containing 1ϫ Yeast/Fungal Protease Arrest TM mixture. Proteins were precipitated with 50 l of 50% TCA and pelleted at 12,000 ϫ g for 5 min. The pellets were suspended in 120 l of dissolving buffer (4% SDS, 0.1 M Tris-HCl, pH 6.8, 4 mM EDTA, 20% glycerol, 2% 2-mercaptoethanol, 0.02% bromphenol blue) (70). The samples were heated at 37°C for 15 min. Protein in the supernatant was quantified using the Bradford assay kit (Thermo Fisher). An equivalent amount of protein was added to each lane. Western blotting analysis was performed as described in Ref. 17. Mouse monoclonal anti-HA (Sigma-Aldrich, catalog no. H3663) was used as primary antibody, and goat anti-mouse antibody (LI-COR, catalog no. 925-32210) was used for detection. The G6PD was used as the loading control (17).

Recultivation from a stationary phase culture
Indicated strains were grown in YNB medium lacking inositol. Cultures were restarted at an A 600 nm ϳ0.005, and growth curves were determined by measuring optical density at 600 nm (A 600 nm ) as described previously for 4 days (all data points not shown). The ability to grow following recultivate was examined by restarting the cultures at A 600 nm ϳ0.005 from day 3 and day 4 of the cultures used in the growth curve. The stationary phase was confirmed by microscopy, where most cells were in the unbudded G 0 phase. Growth of the recultivated cultures was monitored at three different time points.