Role of the Yeast Phosphatidylinositol/Phosphatidylcholine Transfer Protein (Sec14p) in Phosphatidylcholine Turnover and INO1 Regulation*

In yeast, mutations in the CDP-choline pathway for phosphatidylcholine biosynthesis permit the cell to grow even when the SEC14 gene is completely deleted (Cleves, A., T., E., Champion, K., Ait-ken, J., Dowhan, M., Cell 64, 789–800). We report that strains carrying mutations in the CDP-choline pathway, such as cki1 , exhibit a choline excretion phenotype due to production of choline during normal turnover of phosphatidylcholine. Cells carrying cki1 in combination with sec14 ts , a temperature-sensitive allele in the gene encoding the phosphatidylinositol/phosphatidylcholine trans-porter, have a dramatically increased choline excretion phenotype when grown at the sec14 ts -restrictive temperature. We show that the increased choline excretion in sec14 ts cki1 cells is due to increased turnover of phosphatidylcholine via a mechanism consistent with phospholipase D-mediated turnover. We propose that the elevated rate of phosphatidylcholine turnover in sec14 ts cki1

Phospholipid transfer proteins capable of exchanging phospholipids between membrane bilayers in vitro have been extensively characterized, but the role of these proteins in vivo has been more difficult to establish (1)(2)(3)(4)(5). A breakthrough in this analysis occurred with the discovery that the product of the Saccharomyces cerevisiae SEC14 1 gene, which is required for the secretory process, encodes a PI/PC 2 transfer protein (5). Deletion of the yeast SEC14 gene in an otherwise wild type cell is lethal and inactivation of the SEC14 gene product in a temperature-sensitive (sec14 ts ) mutant leads to arrest of the secretory pathway at the late Golgi stage (6). Analysis of suppressors that permit the sec14 ts mutant to grow at the restrictive temperature led to the discovery that mutations in the CDP-choline pathway for PC biosynthesis (Fig. 1) suppress the sec14 mutant phenotype, allowing wild type growth even in strains carrying a total deletion of the SEC14 gene (1).
The CDP-choline pathway, first described by Kennedy and Weiss (7), is one of two routes for synthesis of PC in eukaryotic cells, including yeast. The second route for synthesis of PC, originally described by Bremer and Greenberg (8), involves methylation of PE. Yeast cells can utilize either the CDPcholine or the PE methylation pathway, or a combination of the two, for net PC synthesis (9) (Fig. 1). Deletion of the genes in the CDP-choline pathway is not lethal in yeast and, indeed, appears to have little effect on growth (9,10). It has been widely assumed that the CDP-choline pathway in yeast functions largely for the utilization of exogenous choline. However, recent studies have suggested that the CDP-choline pathway contributes substantially to PC biosynthesis even in the absence of exogenous choline (10 -12). In the absence of exogenous choline, the yeast cell synthesizes PC predominantly via methylation of PE.
Conversely, yeast cells can survive the complete and simultaneous deletion of the genes encoding the two phospholipid methyltransferases that carry out the three-step conversion of PE to PC, provided choline is supplied in the growth medium (13).
However, deletions of genes encoding enzymes in either of these two pathways result in subtly different phenotypes. For example, the deletion of either of the phospholipid methyltransferases does not suppress the sec14 growth phenotype (1,14). Such mutants are, however, unable to repress the INO1 gene (encoding inositol-1-phosphate synthase; see Fig. 1) in response to exogenous inositol, unless PC biosynthesis is restored via the CDP-choline pathway (13,15). And while the deletion of genes encoding enzymes of the CDP-choline pathway does not affect INO1 regulation in response to inositol (9), such mutants suppress the sec14 phenotype (1,14).
The INO1 gene is the most highly regulated of a set of genes encoding enzymes of phospholipid biosynthesis that are subject to complex coordinate control. All these genes contain a conserved promoter element, UAS INO , that includes within it the canonical binding site, CANNTG, for transcription factors of the basic helix-loop-helix class (16,17). In the present report, we show that inactivation of the sec14 ts gene product leads to acceleration of PC turnover, an event that is not in itself lethal, as long as the liberated choline cannot be reused in PC biosynthesis. At the same time, we document a relationship between PC turnover and regulation of the INO1 gene and, by extension, other co-regulated genes of phospholipid metabolism.
Yeast Transformation-Yeast transformation was performed by the lithium acetate method (18) with minor modifications.
Assay for Opi Ϫ (Overproduction of Inositol) and Opc Ϫ (Overproduction of Choline) Phenotypes-To test for the Opi Ϫ phenotype (see Refs. 19 and 20 for a complete method description), strains were patched onto synthetic I Ϫ media and allowed to grow at the indicated temperatures for 2 days. The plates were then sprayed with a suspension of a diploid tester strain (AID) homozygous for ino1 and ade1 (Table I) and incu-bated for another 2 days.
For the Opc Ϫ phenotype, strains were patched onto plates containing synthetic I ϩ or I Ϫ medium lacking choline (C Ϫ ), and were allowed to grow at the indicated temperature for 2 days. They were then sprayed with a tester strain, cho2 opi3 (Table I), auxotrophic for choline and were incubated for another 2 days at 30°C or 37°C or another 3 days at 25°C.
␤-Galactosidase Assay-Each strain was transformed to uracil prototrophy with plasmid pJH359 (Ϫ359 to Ϫ119 INO1-CYC1-lacIЈZ), as described by Lopes et al. (21). The transformed strains were grown to mid-logarithmic growth phase in I ϩ or I Ϫ medium. The ␤-galactosidase assays were performed as described by Lopes et al. (21), except that aliquots were removed from the reaction mix at 5, 10, and 15 min. For temperature shift experiments, the strains were initially grown in repressing conditions (I ϩ medium) at 25°C, the cultures were filtered, and the cells used to inoculate separate cultures to be grown at 25°C or 37°C in either repressing (I ϩ ) or derepressing conditions (I Ϫ ). Samples were harvested at intervals and assayed for ␤-galactosidase activity.
Northern Analysis-RNA probes were synthesized from plasmids previously described (22). Plasmids were linearized with a restriction enzyme and transcribed, in the presence of [ 32 P]cytidine 5Ј-triphosphate, using an RNA polymerase as follows (plasmid/restriction enzyme/RNA polymerase): pAB309⌬/EcoRI/SP6 (TCM1), and pJH310/ HindIII/T7 (INO1). Transcription was performed according to manufacturer's protocol for the SP6/T7 Riboprobe Combination System (Promega). The experimental cultures were grown at the indicated temperatures to early logarithmic phase of growth (A 595 between 0.2 and 0.25), and RNA was isolated using glass bead disruption and hot phenol extraction (23). The TCM1 ribosomal protein gene, which is expressed constitutively with respect to inositol and choline availability, was used as a standard for RNA loading (24). Northern hybridization was performed, visualized by autoradiography, and quantitated with an AMBIS 4000 phosphoimager (AMBIS, Inc.), as described (9).
Phospholipid Analysis-To determine steady state phospholipid

FIG. 1. Diagram of 32 P-H 3 PO 4 and [ 14 C]choline labeling routes in S. cerevisiae.
Labeled phosphorus ( 32 P) is represented by *P, and labeled choline ( 14 C) is represented by †C. Water-soluble molecules that can be taken up from the media are circled (I, inositol; P, phosphorus; C, choline). Intracellular water-soluble metabolites are shown in ovals (I-1-*P, inositol 1-phosphate; Glu-6-*P, glucose 6-phosphate; †C-*P, choline phosphate; CD*P- †C, cytidine-diphosphate choline). Glycerophosphoinositol (Gro*PIns) is an extracellular metabolite, which can be taken up from the media. Lipids are shown in rectangles (DAG, diacylglycerol; *PA, phosphatidic acid; CD*P-DG, cytidine diphosphate-diacylglycerol; *PG*P, phosphatidylglycerol phosphate; *PG, phosphatidylglycerol; *PS, phosphatidylserine; *PE, phosphatidylethanolamine; *PC, phosphatidylcholine; *PI, phosphatidylinositol; *PI*P, phosphatidylinositol phosphate; *PI**P2, phosphatidylinositol bisphosphate. Solid lines represent routes of metabolic conversion. Dashed lines indicate potential flux across the plasma membrane. composition, strains were grown in I ϩ or I Ϫ synthetic media containing 10 Ci of [ 32 P]orthophosphate/ml. The cultures were harvested in midlogarithmic phase (A 595 ϭ 0.4 -0.6) after five to six generations of growth at the indicated temperature. Labeled lipids were extracted (25), individual phospholipid species were resolved by two-dimensional paper chromatography (26) and quantified by liquid scintillation counting. For phospholipid turnover experiments, cells were labeled as described above, harvested, washed twice in fresh media lacking labeled orthophosphate, and suspended at A 595 ϭ 0.1 under the indicated culture conditions. Aliquots of these cultures were removed at indicated times, and phospholipids were analyzed. The rate of phospholipid synthesis in vivo was determined by pulse labeling the cells for 30 min with 50 Ci of [ 32 P]orthophosphate/ml as described by Kelley et al. (27), followed by lipid extraction and separation by paper chromatography, as described above. The amount of [ 32 P]phosphate incorporated into sphingolipids was determined by deacylating the extracted lipids and separating the sphingolipids by one-dimensional chromatography (28).
Metabolic Labeling with [ 14 C]Choline-Strains were grown overnight at the indicated temperatures in I Ϫ media containing 1 Ci/ml 1,2-[ 14 C]choline at a concentration of 10 M. Cultures were harvested during mid-logarithmic phase, washed twice in fresh non-radioactive media, and suspended at A 595 ϭ 0.1 in I Ϫ C Ϫ media. At the indicated time points, aliquots of the cultures were removed, and the cells were pelleted by centrifugation. The supernatant was saved as the "media" fraction. The cell pellet was processed as described (25) for the extraction of lipids with one addition; following treatment of the cell pellet with trichloroacetic acid to permeabilize the cell membrane, the supernatant and subsequent pellet washes were combined and saved as the "intracellular water-soluble fraction of the cell." This fraction contained the vast majority of the intracellular water-soluble counts, as evidenced by the fact that Ͼ90% of the counts found in the only other cellular fraction (the lipid fraction) were shown by chromatography to be PC.
Separation of Choline-containing Water-soluble Metabolites-Cation exchange chromatography on Bio-Rex 70 resin (50 -100-mesh) was used to separate quantitatively the 1,2-[ 14 C]choline in the aqueous media washes or the trichloroacetic acid extracts from other water-soluble 1,2-[ 14 C]choline-containing metabolites (29). The aqueous samples (3-5 ml) were neutralized with 1.0 M Tris buffer, pH 8.0, as necessary, and applied to a 1-ml Bio-Rex 70 column. The column was then washed with 5 ml of H 2 O, followed by 10 ml of 50 mM glycine, 500 mM NaCl, pH 3.0, to elute the choline retained by the resin. The fractions were counted to determine the percentages of various metabolites present.

Conditional Inositol and Choline Excretion Phenotype in
Strains Carrying the Sec14 ts Allele in Combination with CDPcholine Pathway Mutations-The overproduction of inositol (Opi Ϫ ) phenotype ( Fig. 2) is associated with misregulation of the INO1 gene (9,16,19). We report here a related choline excretion phenotype (Fig. 3), which we are designating as Opc Ϫ (overproduction of choline). The Opi Ϫ and Opc Ϫ phenotypes of the strains used in this study are shown in Figs. 2 and 3 and are summarized in Table II. As reported previously (9,30), strains carrying only mutations in the CDP-choline pathway exhibited no Opi Ϫ phenotype. However, all of the CDP-choline pathway mutants exhibited the Opc Ϫ phenotype to varying degrees. No Opi Ϫ phenotype was observed for the strain carrying the sec14 ts mutation (but no mutation in the CDP-choline pathway) at its permissive temperature, 25°C, or at 30°C (Table II). (This strain will be referred to from here on as sec14 ts CKI1 ( Table I) because most of the subsequent studies were done with the set of strains carrying combinations of the sec14 ts and cki1 mutation.) A slight Opc Ϫ phenotype was observed for the sec14 ts CKI1 strain, but only in I Ϫ media at 30°C.
At 30°C, a temperature that is still permissive for sec14 ts , the double mutants containing sec14 ts in combination with each of the CDP-choline pathway lesions (i.e. sec14 ts cki1; sec14 ts cct1; sec14 ts cpt1) had an Opc Ϫ phenotype somewhat stronger than the phenotype exhibited by strains carrying only the CDP-choline pathway mutation in question (Table II). At 25°C, none of these double mutant strains exhibited an Opi Ϫ

FIG. 2.
Overproduction of inositol (Opi ؊ ) phenotype. Wild type, sec14 ts cki1, SEC14 cki1, sec14 ts CKI1, and opi1 strains were tested as described under "Experimental Procedures." Overexpression and excretion of inositol by strains result in growth of the tester strain, as observed by a halo around the strain being tested. The opi1 (19) strain is included as a positive control. phenotype. However, at 37°C, the temperature at which the sec14 ts gene product is inactivated (5), all three double mutants exhibited Opi Ϫ phenotypes, as well as dramatically stronger Opc Ϫ phenotypes than those seen in these same strains at 30°C or in strains carrying only the respective CDP-choline pathway lesions at any temperature ( Fig. 2; Table II).
Choline Metabolism-In the absence of exogenous choline supplementation, yeast cells make PC primarily via methylation of PE (Fig. 1). The choline excretion phenotype of the CDP-choline pathway mutants (Fig. 3) suggests that choline is being produced in the de novo synthesis of PC via methylation of PE, followed by PC turnover (see metabolic pathway, Fig. 1). Equivalent turnover presumably occurs in the normal course of lipid metabolism in wild type strains, but the choline is rapidly reutilized via the CDP-choline pathway without escaping from the cell. The increase in choline excretion at the restrictive temperature in the double mutants, carrying the sec14 ts allele in combination with one of the CDP-choline pathway mutations, therefore, suggests that accelerated turnover of PC is occurring.
To test this idea, we labeled the wild type, SEC14 cki1, sec14 ts CKI1, and sec14 ts cki1 strains with [ 14 C]choline in I Ϫ C Ϫ medium. [ 14 C]Choline is incorporated into PC via the CDPcholine pathway (Fig. 1), but strains carrying the cki1 mutation have very reduced capacity to incorporate [ 14 C]choline into PC.
We found that incorporation of [ 14 C]choline into PC in SEC14 cki1 and sec14 ts cki1 strains was approximately 13% of the wild type level, comparable to previous reports for cki1 strains (9,31). Analysis of the chloroform-soluble label extract from all four strains labeled at 25°C (i.e. wild type, sec14 ts CKI1, SEC14 cki1, and sec14 ts cki1) demonstrated that Ͼ90% of the label was associated with PC.
To study PC turnover, [ 14 C]choline-labeled cells were shifted to unlabeled medium and the fate of the label was charted. The pattern of label transfer between the lipid-soluble pool, the intracellular water-soluble pool, and the medium following the shift to unlabeled medium is shown for three strains (wild type, SEC14 cki1, and sec14 ts cki1) at 37°C (Figs. 4 and 5). All four strains were tested in this fashion at 25°C as well (data not shown); only two patterns of turnover were observed. The "wild type" pattern was exhibited by both wild type and sec14 ts CKI1 at 25°C and by wild type at 37°C (Fig. 4A). The "cki1" pattern was exhibited by both SEC14 cki1 and sec14 ts cki1 strains at 25°C and by the SEC14 cki1 strain at 37°C (Fig. 4B).
In strains exhibiting the "wild type" pattern of choline metabolism (i.e. the wild type strain at both 25°C and 37°C and sec14 ts CKI1 only at the permissive temperature of 25°C), the cells at the time of transfer to unlabeled medium had a significant amount (19 -29%) of their cellular 14 C label in a watersoluble intracellular pool (Fig. 4A). The remaining cellular label was lipid-associated. A small amount of label (2-7%) was associated with extracellular free choline at the start of the experiment. During the first 3 h after the shift to unlabeled medium, the free choline from the medium was taken up, the water-soluble intracellular pool declined to a few percent, and the label appeared in the lipid-associated pool, which was shown to be greater than 90% PC. Very little label (10% or less of the total label) appeared in the medium throughout the 6-h chase (Figs. 4A and 5). In the case of wild type cells, the extracellular label that did appear was not in the form of free choline (Fig. 5). Most likely, it is glycerophosphocholine, known to be excreted by wild type yeast cells (32).
In strains exhibiting the "cki1" pattern of choline metabolism (i.e. SEC14 cki1 at both 25°C and 37°C, and sec14 ts cki1 only at the permissive temperature of 25°C), a greater proportion of total cellular label was present in the lipid fraction at the start of the experiment (Fig. 4B). The water-soluble intracellular pool was 12% or less of total label. In the SEC14 cki1 strain at both temperatures, as well as the sec14 ts cki1 strain at 25°C, there was little or no transfer of label from the intracellular soluble pool to PC following the shift to unlabeled medium. However, steady loss of label from PC (about 7% in 3 h) was observed, with a corresponding appearance of label associated with free choline in the growth medium (shown for the SEC14 cki1 strain at 37°C in Figs. 4B and 5).
Finally, a third and very distinctive pattern of PC turnover was exhibited by the sec14 ts cki1 strain at 37°C (Fig. 4C). At 37°C, the sec14 ts cki1 strain had an initial label distribution at the time of shift to unlabeled medium similar to the SEC14 cki1 strain, but it lost label from PC much more rapidly (54% in the first 3 h). Again, the label appeared in the medium as free choline (Figs. 4C and 5). This finding is consistent with the strong Opc Ϫ phenotype associated with the sec14 ts cki1 strain at 37°C (Fig. 3). We also examined the loss of 14 C label from PC in the sec14 ts cki1 and SEC14 cki1 strains when the cells were labeled with [ 14 C]choline to steady-state at 25°C and were shifted to unlabeled medium at 37°C. The SEC14 cki1 strain, which had identical patterns of turnover at 25°C and 37°C (Fig. 4B), not surprisingly showed a similar labeling pattern when shifted from 25°C to 37°C. However, when shifted to 37°C after labeling at 25°C, the sec14 ts cki1 strain, which had different labeling patterns at the two temperatures, exhibited the pattern of 14 C loss from PC observed when the labeling and the turnover were both carried out at 37°C (Fig. 4C). Thus, we conclude that the accelerated pattern of turnover must have occurred immediately upon shifting the sec14 ts cki1 strain to 37°C.
Phospholipid Composition-At 25°C in I Ϫ medium, all four strains exhibited phospholipid compositions (Table III) similar to each other and to other published reports for these (12) and other strains (16) under similar growth conditions. The compositions of the sec14 ts cki1 and wild type strains grown in I Ϫ medium were also analyzed at 30°C and were found to be similar to the compositions obtained at 25°C. In I ϩ medium at 37°C, the three strains capable of growth (i.e. wild type, SEC14 cki1 and sec14 ts cki1) exhibited phospholipid compositions similar to those previously reported for this growth condition (12). In general, the phospholipid compositions of cells grown in I ϩ , compared with I Ϫ medium, contained a higher proportion of PI (Table III). At 37°C in I Ϫ medium, the sec14 ts cki1 strain, however, contained a proportion of PI (19% of total cellular phospholipids) approximately twice the proportion of PI observed in the wild type strain under these same growth conditions. With the exception of the sec14 ts cki1 strain at 37°C, the PI content of all of the strains grown at 25°C or 37°C in I Ϫ medium was between 9 -15% of the total phospholipid. In I ϩ medium at 37°C in all of the strains tested, the proportion of PI was 28 -32% of total phospholipid (Table III).
Phospholipid Turnover-Cells were labeled to steady-state with 32 P, as shown in Table III, and the retention and distribution of the 32 P into the various phospholipids was tracked after shifting the cells into unlabeled medium (Fig. 6). At 25°C, no distinctive differences among the four strains were observed (data not shown). At 37°C, the patterns of 32 P label retention in the wild type and SEC14 cki1 strains (Fig. 6, A and B) were similar to each other and to the results obtained in all four strains at 25°C (data not shown). In all four strains at 25°C, and in wild type and SEC14 cki1 cells at 37°C (Fig. 6, A and B), label was gradually lost from all lipids except PC, where it accumulated. In all cases except sec14 ts cki1 at 37°C (Fig. 6C), the ratio of label remaining in PI versus PC tended to decrease over time (Fig. 6D). In sec14 ts cki1 at 37°C, however, the PI/PC ratio started high and increased during the course of the 6-h turnover experiment. The label associated with the category "other," which includes sphingolipids, also rose in the sec14 ts cki1 strain during the course of the turnover experiment at 37°C. In the sec14 ts cki1 strain at 37°C, sphingolipids accu- FIG. 4. Phosphatidylcholine turnover in wild type, SEC14 cki1, and sec14 ts cki1 strains at 37°C. Strains were grown in I Ϫ medium containing 1 Ci/ml [ 14 C]choline to mid-logarithmic phase. At time zero, cells were harvested, washed, and recultured in non-radioactive I Ϫ medium. At the indicated times, aliquots of the culture were analyzed for radioactivity. Total radioactivity incorporated into the sec14 ts cki1 and SEC14 cki1 strains was only 13% of the incorporation in wild type, consistent with previous reports (see Footnote 3). Data expressed as a percentage of total label recovered in each fraction at each time point after the shift to unlabeled medium: culture medium (E), cellular water-soluble fraction (f), and cellular chloroform-soluble fraction (q). Data are the average of two independent experiments. mulated label during the course of the experiment (Fig. 6E). In the sec14 ts cki1 strain, the proportion of label associated with PA remained fairly constant over time (Fig. 6C), whereas it dropped about 2-fold in the wild type (Fig. 6A) and in the SEC14 cki1 (Fig. 6B) strains in the first 3 h after transfer to unlabeled medium.
Pulse Labeling of Phospholipids-Because of the nature of the product-precursor relationships shown in Fig. 1, 32 P introduced during a 30-min pulse labeling period will show a very different pattern of distribution than during steady-state labeling used to assess phospholipid composition (Table III). The three strains able to grow at 37°C were pulse-labeled with 32 P for 30 min, and in all three cases, the overall incorporation was greater per OD unit of culture in I ϩ medium compared with I Ϫ medium (Table IV). The ratio of label recovered in association with PI compared with PC (PI/PC ratio) was higher in all three strains grown in I ϩ as opposed to I Ϫ medium (Table IV). However, in I Ϫ medium in the sec14 ts cki1 strain, a higher proportion of label was associated with PI and there was a higher PI/PC ratio than in either the wild type strain or the SEC14 cki1 strain.
When examined at 25°C, the sec14 ts CKI1 strain showed a pattern of label incorporation during the 30-min pulse that was similar to the wild type strain at 37°C. The sec14 ts CKI1 and sec14 ts cki1 strains were also pulse-labeled with 32 P following a FIG. 6. Phospholipid turnover as a function of growth temperature. Strains were grown to mid-logarithmic phase in I Ϫ medium supplemented with 10 Ci [ 32 P]orthophosphate/ml, as described in Table III. The starting phospholipid compositions were identical to those shown in Table III for each respective strain and growth conditions at 37°C. The cells were harvested, washed, and inoculated into inositol-free non-radioactive media. At the indicated times, lipids were extracted and analyzed as described under "Experimental Procedures." The data shown in A, B, and C represent the relative percentages of the total label retained in each phospholipid species. "Other" lipids include CDP-DG, phosphatidylmonomethylethanolamine, phosphatidyldimethylethanolamine, cardiolipin, and lipids retained near the origin. The ratio of PI/PC (D) was calculated from the relative percentage of 32 P label associated with PI versus PC. Label associated with sphingolipid at 37°C is depicted in E.

TABLE III
Effect of inositol and temperature on phospholipid composition Yeast strains were grown for five to six generations at the indicated temperatures in I ϩ or I Ϫ medium containing 10 Ci/ml [ 32 P]orthophosphate. The cells were harvested, and phospholipids extracted and resolved as described under "Experimental Procedures." The term "Other" includes CDP-DG, PMME, PDME, CL, and polar lipids, including sphingolipids, remaining near the origin. Values represent the percentage of total lipid-associated 32  shift from 25°C to 37°C in I Ϫ or I ϩ media. The labeling pattern of the sec14 ts cki1 strain under these conditions (Table IV, part B) was quite similar to the pattern seen when the same strain was maintained continuously at 37°C in I ϩ or I Ϫ media. However, the sec14 ts CKI1 strain incorporated less 32 P label per OD unit of culture 4 h after the shift to 37°C as compared with the sec14 ts cki1 strain, and a much higher proportion of the label was incorporated into PC in the sec14 ts CKI1 strain at 37°C than in any other strain tested under any growth condition (Table IV,

part B).
Regulation of the INO1 Gene in the sec14 ts cki1 Strain-The Opi Ϫ phenotype, observed in the sec14 ts strain at 37°C, is associated with misregulation of INO1 (19). Therefore, we investigated INO1 regulation in four strains: wild type (SEC14 CKI1), sec14 ts CKI1, SEC14 cki1, and sec14 ts cki1. At 25°C, all four strains showed repression of the INO1 transcript in response to inositol when tested by Northern blot analysis ( Fig. 7; Table V). The sec14 ts CKI1 strain was also examined at 30°C and showed normal regulation. At 25°C or 30°C in each of these strains, the INO1 transcript is substantially repressed (i.e. 10-fold or more) in cells grown in I ϩ medium compared with cells grown in I Ϫ medium ( Fig. 7; Table V).
At 37°C, only three of the strains will grow: wild type, SEC14 cki1, and sec14 ts cki1; the wild type and SEC14 cki1 strains both exhibited INO1 normal regulation at 37°C (Table  V, Fig. 7). However, consistent with its Opi Ϫ phenotype, the sec14 ts cki1 strain showed an abnormal pattern of INO1 expression at 37°C. In the absence of inositol, the INO1 transcript was expressed at a level somewhat higher than the wild type derepressed level. Furthermore, when inositol was added to the growth medium, in the absence of choline, the level of INO1 transcript was repressed only about two-fold, compared with more than 10-fold repression seen in the wild type strains (Table V). When inositol and choline were added together to the growth medium, no repression of INO1, at all, was seen in the sec14 ts cki1 strain at 37°C (Table V).
Kinetics of INO1 Induction in the sec14 ts cki1 Strain Shifted to 37°C-To study the kinetics of INO1 induction, we used the INO1 lacZ reporter construct, described under "Experimental Procedures." This construct was transformed into the four strains used to study INO1 regulation, and the strains were assayed for ␤-galactosidase activity under various growth conditions. To establish that the reporter construct was regulated in a fashion resembling the native INO1 transcript (Fig. 7; Table V), we initially studied expression of ␤-galactosidase in cells grown to mid-logarithmic phase (Table VI). The sec14 ts CKI1 strain will not grow at 37°C and, therefore, only three of the four strains, namely wild type, SEC14 cki1, and sec14 ts cki1, were tested at 37°C. The overall pattern of expression of the INO1 lacZ construct (Table VI) was very similar to the expression of the native INO1 transcript measured by Northern blot analysis (Table V; Fig. 7). At 25°C and 30°C, all four strains showed ␤-galactosidase regulation in response to inositol comparable to wild type, except that the sec14 ts cki1 strain had a slightly higher level of ␤-galactosidase under repressing conditions (I ϩ medium). At 37°C, the wild type and SEC14 cki1 strains, showed the normal pattern of regulation, i.e. repression of INO1 lacZ in response to inositol, while the sec14 ts cki1 strain exhibited misregulation of INO1 lacZ at 37°C ( Table   TABLE IV Pulse labeling of phospholipids In A, strains grown to mid-log phase (A 595 ϭ 0.4 -0.6) in I ϩ or I Ϫ medium at the indicated temperatures were pulsed for 30 min with 50 Ci/ml [ 32 P]orthophosphate. The lipids were then extracted and analyzed as described under "Experimental Procedures." In B, yeast cells were grown at 25°C and then shifted to 37°C. After 4 h of growth, the cells were pulse-labeled for 30 min with 50 Ci/ml [ 32 P]orthophosphate. The lipids were extracted and analyzed. The amount of 32 P incorporated into lipid ( 32 P Incorp.) is presented as the counts/min per optical density unit at A 595 ϫ 10 3 (cpm/ODU ϫ 10 3 ). The relative percentage of 32 P label in the individual phospholipid species is presented as a percentage of the total 32 P incorporated into lipid. The PI/PC ratio was calculated from the percentages of PI versus PC. "Other" lipids include CDP-DG, PMME, PDME, CL, as well as polar lipids that remained near the origin.  7) at 37°C in I Ϫ medium, the sec14 ts cki1 strain expressed a level of ␤-galactosidase about 2-fold higher than its wild type and SEC14 cki1 counterparts. Also consistent with Northern blot analysis (Table V) in I ϩ medium at 37°C, only about a 2-fold repression of the INO1 lacZ reporter construct was observed in the sec14 ts cki1 strain (Table VI).
Using the reporter construct containing the lacZ gene under the control of an INO1 promoter fragment, we investigated the kinetics of INO1 derepression. At 25°C, all three strains tested (wild type, sec14 ts CKI1, and sec14 ts cki1) showed rapid derepression of the INO1 lacZ construct when shifted from I ϩ medium to I Ϫ medium (data shown only for sec14 ts cki1; Fig.  8B). The wild type strain showed a similar pattern of derepression when shifted from I ϩ medium at 25°C to I Ϫ at 37°C (Fig.  8H). In each of these cases, ␤-galactosidase activity plateaued as the cultures approached stationary phase, indicating that new expression of ␤-galactosidase had stopped, as expected, at about the time when INO1 expression is known to be repressed as cells enter stationary phase (33,34). 3 Upon shift from I ϩ medium at 25°C to I Ϫ medium at 37°C, the INO1 lacZ construct in the sec14 ts cki1 strain showed more rapid derepression than in other strains shifted to I Ϫ medium (Fig. 8D). Unlike wild type shifted to I Ϫ at 37°C (or any of the three strains shifted from I ϩ to I Ϫ at 25°C), derepression of the INO1 lacZ construct in the sec14 ts cki1 strain at 37°C continued well into stationary phase. The sec14 ts cki1 strain shifted from I ϩ medium at 25°C to I ϩ at 37°C also exhibited derepression of ␤-galactosidase. The initial derepression of the INO1 lacZ construct in the sec14 ts cki1 strain shifted from 25°C to 37°C in I ϩ medium was less rapid than when the cells were shifted to I Ϫ medium at 37°C. However, at 37°C, even in I ϩ medium, the derepression of the INO1 construct continued after the sec14 ts cki1 cells became stationary. The sec14 ts CKI1 strain stopped growing after 5-6 h upon shifting to 37°C (Fig.  8, E and F). However, under these conditions the INO1 lacZ construct derepressed in I Ϫ medium to a level exceeding the wild type strain grown under comparable conditions (Fig. 8,  compare F (sec14 ts CKI1) and H (wild type)). When shifted from 25°C to 37°C in I ϩ medium, no derepression of the INO1 lacZ construct was observed in the sec14 ts CKI1 strain (Fig. 8E). DISCUSSION Studies using wild type yeast cells have detected very little PC turnover (32). Consistent with these earlier studies, our analysis of PC turnover in wild type yeast using both 32 P label and [ 14 C]choline label, revealed very little evidence of extensive PC turnover (Figs. 4 -6). In cells carrying a mutation in the CDP-choline pathway, it is possible to examine phospholpid metabolism under conditions in which reutilization of the choline liberated by PC turnover is largely blocked. Under these circumstances, we detected substantial PC turnover and the choline liberated by this process appeared in the growth medium (Figs. 4 and 5), providing a phenotype (Opc Ϫ ) that is readily detected in a plate assay (Fig. 3). We assume that the extent of PC turnover is similar in wild type, but that choline liberated by turnover in wild type cells is immediately reincorporated into PC via the CDP-choline pathway. Thus, the cki1 (cct1 or cpt1) genetic backgrounds provide unique opportunities to explore the extent of PC turnover in yeast.
The choline excretion phenotype provides, for the first time in yeast, a means of estimating the extent of activity in the CDP-choline pathway in cells growing in the absence of choline. Recent studies have suggested that this pathway contributes substantially to PC biosynthesis even in the absence of exogenous choline (10 -12). Our analysis substantiates this hypothesis. Reutilization of choline liberated by turnover is, apparently, a major function of the CDP-choline pathway in yeast cells growing in the absence of choline. Based on the choline excreted by the SEC14 cki1 strain at 37°C (Figs. 4 and 5), we estimate that some 7% of cellular choline is recycled via PC turnover and re-synthesis in each generation period (about 3 h). This must be an underestimate because cki1 cells retain some capacity to reuse free choline, as evidenced by our ability to label the cells with [ 14 C]choline in the first place. Nevertheless, it is possible to use the choline excretion plate phenotype (Opc Ϫ ) to carry out a qualitative comparison of the extent of PC turnover in different strains, under different growth conditions. For example, in the SEC14 cki1 strain, we observed that the extent of the choline excretion ring was affected by temperature and by the presence of inositol (Fig. 3, Table II).
The Nature of PC Turnover in Yeast-We propose that the free choline found in the growth medium of the cki1 bearing strains most likely arises from phospholipase D (PLD)-mediated PC turnover which produces free choline and phosphatidic acid (PA). It is now recognized that PLD-mediated hydrolysis is a major route of PC breakdown in a variety of mammalian cell types (36). The bulk of the [ 14 C]choline excreted by the SEC14 cki1 strain detected in the growth medium was free choline (Fig. 5). This is not consistent with phospholipase B (PLB)mediated turnover, which produces glycerophosphocholine 3 V. Jiranek, J. A. Graves, and S. A. Henry, submitted for publication.   Fig. 4. Strains were grown in I ϩ or I Ϫ medium with (C ϩ ) or without (C Ϫ ) 1 mM choline. RNA analysis was performed as described under "Experimental Procedures." Hybridization with the TCM1 probe served as a loading control. Quantitation was performed with an AMBIS 4000 phosphorimager. Data are expressed as a percentage of the expression observed in the wild type (wt) strain at 37°C in I Ϫ C Ϫ conditions.

Strain
Temperature  8. Kinetics of INO1 derepression in sec14 and sec14 ts cki1 mutants. Yeast strains containing plasmid pJH359 were grown at 25°C under repressing conditions (I ϩ medium), to the mid-logarithmic phase of growth. The cells were harvested by filtration and transferred to fresh media (I ϩ or I Ϫ ) at the permissive (25°C) or non-permissive (37°C) temperature. At the indicated time points, the A 595 of the cultures (O) and ␤-galactosidase activity (⌬) was determined. Graphs A-D represent the sec14 ts cki1 strain shifted to: I ϩ medium at 25°C (A), I Ϫ medium at 25°C (B); I ϩ medium at 37°C (C); and I Ϫ medium at 37°C (D). Graphs E and F represent the sec14 ts CKI1 strain shifted to I ϩ medium at 37°C (E) and I Ϫ medium at 37°C (F). Graphs G and H represent the wild type (SEC14 CKI1) strain shifted to I ϩ medium at 37°C (G) and I Ϫ medium at 37°C (H).
Phosphatidylcholine Turnover and INO1 Regulation in Yeast 20881 (37), rather than free choline. Neither can phospholipase C (PLC)-mediated turnover, which produces choline phosphate and diacylglycerol (DAG), readily account for the free choline excreted by the SEC14 cki1 mutant. Moreover, we propose that PC turnover accelerates via a PLD mechanism when the sec14 ts cki1 mutant is raised to the restrictive temperature, resulting in increased choline excretion. Consistent with this hypothesis, we found that the 14 C label that appears in the medium of the sec14 ts cki1 strain at 37°C is again predominantly free choline and that the generation of this pool of choline correlates with the loss of label from PC (Figs. 4 and 5). At 37°C, the sec14 ts cki1 strain lost more than 50% of its PC-associated 14 C choline label in 3 h (Figs. 4 and 5), or approximately one generation time. The SEC14 cki1 strain, in contrast, lost approximately 7% of its PC-associated label in this same amount of time.
When PC turnover occurs via a PLD-mediated route, one molecule of PA is generated for every molecule of free choline produced and the associated 32 P label is not lost from the lipid fraction (Fig. 1). Lipid-associated 32 P label recycled via PLD re-enters into the PA pool and can be directly reused in the series of reactions leading to the reformation of PC via the PE methylation pathway using CDP-DG, PS, and PE as intermediates. Recycled, labeled PA can also serve as a precursor to PI, via the CDP-DG branchpoint in the pathway, as shown in Fig.  1. In the experiments shown in Fig. 6, the distribution of label from 32 P into the various lipids in wild type and SEC14 cki1 cells 3 and 6 h after removal from labeled medium reveals that total 32 P is only slowly lost from the chloroform-soluble pool. Even in sec14 ts cki1 cells, which lose 50% or more of the [ 14 C]choline label associated with PC per generation at 37°C, 32 P is lost very slowly from the total lipid pool. These results are consistent with major turnover occurring via a PLD-mediated mechanism, as discussed above. These observations also explain why substantial PC turnover has not been detected previously in labeling studies of wild type yeast cells (32). In contrast to PLD-mediated turnover, the PLC and PLB-mediated routes of phospholipid turnover are predicted to lead to loss of 32 P from the chloroform-soluble pool because the 32 P label is cycled into water-soluble products such as choline phosphate and glycerophosphocholine (Fig. 1). Our data, therefore, suggest that the bulk of PC turnover in yeast cells is carried out via a PLD mechanism. While 32 P was lost only slowly from total lipid in the sec14 ts cki1 mutant at 37°C, the pattern of 32 P distribution during the 6-h chase was strikingly different than that observed in the same strain at 25°C or in any of the other strains, including SEC14 cki1 at 37°C. In the sec14 ts cki1 strain at 37°C, the ratio of PI to PC was high at the start of the turnover experiment, and, unlike the other strains, this ratio increased as the experiment progressed. At 37°C in the sec14 ts cki1 strain, 32 P label tended to accumulate in PI and in the category "other," which includes sphingolipids (Fig. 6), a pattern consistent with elevated turnover via a PLD-mediated mechanism. Each time PC serves as a substrate in PLD-mediated turnover, PC-associated 32 P recycles as PA and has a certain probability of passing into PI and then into sphingolipids. Accumulation of lipid-associated 32 P into PI-and inositol-containing sphingolipids is, thus, an expected consequence of elevated PLD-mediated turnover of PC. This pattern of label accumulation will be further accentuated when high amounts of inositol are present, either from exogenous sources or from endogenous synthesis, due to derepression of the INO1 gene, as will be discussed.
Role of PC Turnover in Suppression of the sec14 ts Phenotype-We do not believe that accelerated turnover of PC is the cause of lethality in the sec14 ts CKI1 strain when this strain is elevated to the restrictive temperature. The sec14 ts cki1 strain grows quite well at 37°C in the absence of a functional SEC14 gene product and yet exhibits greatly increased PC turnover (Figs. 4 and 5). Indeed, in the face of massive ongoing PC turnover, this strain has a doubling time comparable to that of the wild type and SEC14 cki1 strains. We propose that the cki1 lesion suppresses the lethality caused by inactivation of Sec14p at the restrictive temperature precisely because it prevents the resynthesis of PC following accelerated turnover. It seems highly probable, as has been proposed previously (1,12,14), that Sec14p plays a role in regulating the phospholipid content of Golgi. Sec14p is apparently required for removal of PC synthesized via the CDP-choline pathway from its immediate site of synthesis, presumably in the Golgi (12). If PC is not removed by the action of Sec14p, its rate of turnover via PLD apparently increases, as seen in Fig. 4. Under such circumstances, in the sec14 ts CKI1 strain when the CDP-choline pathway is not blocked, PC could be immediately re-synthesized, setting up a futile cycle.
Why is excess PC detrimental to the secretory process, and how could increased PC turnover lead to a bypass of the need for Sec14p? Recently, Seaman (38) pointed out similarities between the yeast and mammalian secretory pathways and called attention to recent work by Ktistakis et al. (39), demonstrating a role for PLD and PA in the secretory pathway in mammalian cells. Ktistakis et al. (39) showed that PA produced by PLD-mediated turnover was sufficient for coatomer binding to Golgi membranes even in the absence of ADP-ribosylation factor. PLD-mediated turnover also hydrolyzes PC, high levels of which may interfere with vesiculation and/or membrane fusion events required for the ongoing events of secretory pathway. In the sec14 ts cki1 strain, elevation to the restrictive temperature leads to both increased turnover of PC and to increased net PI synthesis. The combination of these effects leads to the restoration of the PI/PC balance, which McGee et al. (12) proposed is important in the ongoing secretory pathway. We propose that the increased synthesis of PI is due both to the greater availability of the PA precursor and also to the regulatory link to INO1 regulation, which we discuss below. Increased PI biosynthesis could lead, in turn, to increased synthesis of PIP and PIP 2 , all of which are negatively charged phospholipids, expected to facilitate the ongoing secretory process. PIP and PIP 2 are also known to stimulate PLD activity in yeast as in mammals (40). This argument is consistent with the fact that Sec14p dysfunction results in increased PC content in Golgi membranes (12) and that Golgi enriched membranes from mammalian cell types have high PLD activity (39).
Role of PC Turnover in Regulation of the INO1 Gene-We propose that the derepression of the INO1 gene, which occurs rapidly following the shift of the sec14 ts cki1 strain to 37°C (Fig. 8) even in the presence of inositol, is directly correlated to and caused by the increase in the rate of PC turnover that occurs when Sec14p is inactivated. It has long been known (16,41) that there is an association between PC and inositol metabolism. Structural gene mutants with defects at every step ( Fig. 1) in the PA to PC pathway (i.e. PA 3 CDP-DG 3 PS 3 PE 3 3 3 PC), exhibit Opi Ϫ phenotypes and misregulate INO1 and other co-regulated genes (for reviews, see Refs. 16, 42, and 43). The structural gene mutants in this series include the cdg1(cds1) mutant (44), which is defective in CDP-DG synthase (45), which catalyzes the first step in this reaction series (i.e. PA 3 CDP-DG) and the opi3 mutant, which is defective in PL methyltransferase, which catalyzes the last two reactions (i.e. phosphatidylmonomethylethanolamine 3 phosphatidyldimethylethanolamine 3 PC) (15,46,47). Similar Opi Ϫ phenotypes are present in structural gene mutants defective in enzymes for every other metabolic step in the sequence between PA and PC (Fig. 1), including cho1, (48) cho2, (9, 13), and psd1. 4 In the current study, we demonstrate that elevation of the sec14 ts cki1 strain to the restrictive temperature results in increased turnover of PC and also generates a signal for derepression of INO1. The derepression of INO1 that we observed upon shift to the restrictive temperature exhibits remarkably rapid kinetics (Fig. 8) and correlates to the increase in PC turnover. In a previous study (9), we employed strains containing combinations of mutations in the PE methylation and CDPcholine routes for PC biosynthesis. In that study, we demonstrated that no single metabolite in either pathway for PC biosynthesis correlated to the production of the signal for repression/derepression of INO1. The metabolic conditions previously tested for their roles in INO1 repression/derepression included relative PC content and free choline availability. Neither of these factors was correlated to INO1 derepression (9). Rather, the metabolic signal controlling INO1 regulation appeared to be linked to the overall ability of increased PC biosynthesis to stimulate cell growth (9). The data presented in the current study suggests that INO1 derepression is correlated to PC turnover. We propose that INO1 derepression occurs in response to a metabolic signal generated in the course of the overall alteration in phospholipid metabolism produced by increased PC turnover. We are currently testing these ideas in cells containing mutations in PLD1, the major phospholipase D of the yeast cell (49 -51), first isolated as SPO14 (35,40).