Glycerophosphocholine-dependent growth requires Gde1p (YPL110c) and Git1p in Saccharomyces cerevisiae.

Glycerophosphocholine is formed via the deacylation of the phospholipid phosphatidylcholine. The protein encoded by Saccharomyces cerevisiae open reading frame YPL110c effects glycerophosphocholine metabolism in vivo, most likely by acting as a glycerophosphocholine phosphodiesterase. Deletion of YPL110c causes an accumulation of glycerophosphocholine in cells prelabeled with [14C]choline. Correspondingly, overexpression of YPL110c results in reduced intracellular glycerophosphocholine in cells prelabeled with [14C]choline. Glycerophospho[3H]choline supplied in the growth medium accumulates to a much greater extent in the intracellular fraction of a YPL110Delta strain than in a wild type strain. Furthermore, glycerophospho[3H]choline accumulation requires the transporter encoded by GIT1, a known glycerophosphoinositol transporter. Growth on glycerophosphocholine as the sole phosphate source requires YPL110c and the Git1p permease. In contrast to glycerophosphocholine, glycerophosphoinositol metabolism is unaffected by deletion of YPL110c. The open reading frame YPL110c has been termed GDE1.

The yeast Saccharomyces cerevisiae synthesizes and degrades the major glycerophospholipids via pathways that are very similar to those employed by higher eukaryotes (1,2). One pathway of phospholipid degradation in yeast as well as higher eukaryotes is deacylation, which results in the formation of water-soluble glycerophosphodiesters. S. cerevisiae cells growing in medium containing nonlimiting amounts of P i and inositol deacylate phosphatidylinositol via the action of phospholipases of the B type, Plb3p and Plb1p, to produce glycerophosphoinositol (GroPIns) 2 (3)(4)(5). Much of the GroPIns produced is excreted into the medium, and external GroPIns can be transported into the yeast cell in times of nutritional stress (phosphate or inositol limitation) via the permease encoded by GIT1 (see Fig. 1) (6,7). Git1p, a member of the major facilitator superfamily of transport proteins (8), was originally isolated based upon its ability to confer growth to an inositol auxotroph supplied with GroPIns as its inositol source (6). Subsequent studies have shown that although inositol limitation up-regulates Git1p transport activity, phosphate limitation does so to a much greater extent and GroPIns can be used as the sole source of phosphate for the cell (7). The enzyme(s) required for GroPIns catabolism remain uncharacterized.
Phosphatidylcholine (PC) deacylation also occurs in S. cerevisiae, resulting in the formation of intracellular and extracellular glycerophosphocholine (GroPCho) (Fig. 1). In general, S. cerevisiae produces more internal than external GroPCho (9). However, external GroPCho production increases as the pH of the medium is raised above 5 (11). Similarly, internal GroPCho production increases upon increased flux through the CDP-choline pathway for PC synthesis as a consequence of temperature elevation or choline supplementation (9). S. cerevisiae Nte1p, a phospholipase B and homolog of human neuropathy target esterase, is responsible for the production of intracellular GroPCho via PC deacylation (10). Plb1p (11) is thought to be primarily responsible for the formation of extracellular GroPCho (Fig. 1). As is true for S. cerevisiae, various mammalian cells respond to an increase in PC synthesis by increasing PC deacylation (12)(13)(14), suggesting a conserved role for this degradative pathway in maintaining PC homeostasis. The build-up of GroPCho in the cell has been associated with a number of disease processes, including cancer (15) and Alzheimer disease (16). In addition, GroPCho has been implicated in diverse cellular functions such as maintenance of renal osmolarity (17), inhibition of lysophospholipase activity (18), and inhibition of phosphatidylinositol transfer protein alpha (19). Clearly, cell physiology impacts, and is impacted by, GroPCho levels. In turn, the level of GroPCho in the cell is a function of both its formation via PC deacylation and its degradation via glycerophosphodiesterases. Glycerophosphodiesterase encoding genes and glycerophosphodiesterase activities acting upon GroPCho have been reported for several cell types, including Escherichia coli (20), Haemophilus influenzae (21), carrot cell wall (22), kidney (23), and brain (24). GroPIns-specific glycerophosphodiesterase activities have been observed in various rat tissues (25,26), and a gene encoding a glycerophosphoinositol glycerophosphodiesterase (GDE1/MIR16) has been cloned from rat (27).
The use of S. cerevisiae as a powerful model for studying phospholipid metabolism is well established (1,2). An understudied aspect of this metabolism is that of the glycerophosphodiesters, such as GroPCho, produced through phospholipid deacylation. Thus, these studies were undertaken to further our knowledge of glycerophosphodiester metabolism in this important model organism. We demonstrate that the protein encoded by YPL110c (here named GDE1) affects glycerophosphocholine levels in the cell, most probably by acting as a glycerophosphodiester phosphodiesterase. Furthermore, we report that GroPCho is transported intact into the cell in a manner dependent upon the GroPIns permease, Git1p.

MATERIALS AND METHODS
Strains and Media-Strains were maintained on YEPD medium (1% yeast extract, 2% Bactopeptone, 2% glucose). The base medium for experiments was chemically defined synthetic media (4) lacking inositol (IϪ). Where indicated, IϪ medium was supplemented with 75 M inositol to make Iϩ medium. For experiments involving the overexpression of GDE1 from the GAL1 promoter (see Fig. 4), the strains were grown in IϪ medium lacking dextrose but containing 2% raffinose and 2% galactose. For experiments involving the addition of exogenous GroPIns or GroPCho (Sigma; catalog number G4007), the base medium was altered by substituting 1g of KCl for 1g of KH 2 PO 4 /liter and adding KH 2 PO 4 to a final concentration of 0.2 mM (low P i ) or 10 mM (high P i ) (7). Strains obtained from Research Genetics (JPV125 and JPV126) were checked by PCR to confirm the expected gene deletion. Strain JPV131 was isolated following tetrad dissection of diploid JPV126. To make the gde1⌬ ypl206c⌬ (JPV431) double mutant, the KanMX marker of strain JPV125 was exchanged with HIS3 using the marker swap plasmid M4754 (28). Plasmid M4754 was digested with NotI, and the released fragment containing HIS3 flanked by regions of the KanMX gene was used to transform JPV125. Hisϩ colonies were checked by PCR to verify integration at the correct location, and the resulting strain was named JPV433. Strain JPV431 was isolated following tetrad dissection of the diploid formed by crossing JPV433 with JPV131 (TABLE ONE).
Construction of pGAL-GST-GDE1 Allele-Plasmid pFA6a-kanMX6-PGAL-GST (29) was used as template to amplify a module for insertion into the genome at the 5Ј end of GDE1. The 5Ј ends of the forward and reverse primers bore 40 nucleotides homologous to the target gene sequences, followed by 20 nucleotides homologous to the plasmid. The underlined sequences are homologous to the target genes: forward primer, 5Ј-TAA TTG CGA CTT CCA CGT GTT CGC AGG TGG AGC AAT GTT ACA GAA TTC GAG CTC GTT TAA AC-3Ј, and reverse primer, 5Ј-CTC TGG AAT GCG ATG ATT GGC AAA GGT TTT TCC GAA CTT CAT ACG CGG AAC CAG ATC CGA TT-3Ј. The PCR product was transformed (30) into JPV203, and G418-resistant colonies were selected (31). To verify integration at the correct location, genomic DNA was isolated and used as template in PCR using forward primer, 5Ј-TTC TTT CTT TTT ATG CAT CTT-3Ј, and reverse primer, 5Ј-CAA CGA CTT GTA GCC AAC ATA-3Ј. Integration of the PCR module resulted in the following changes in the genomic DNA: (i) deletion of ϳ500 bp immediately upstream of the native GDE1 start codon and replacement with the GAL1 promoter and (ii) fusion of the GST gene to the 5Ј end of GDE1. The resulting strain was named pGAL-GST-GDE1 (JPV436).  (27), water-soluble fractions were processed by extracting three times with water-saturated ether to remove trichloroacetic acid, drying down, and resuspending in solvent A (27).

Construction of Plasmid
[ 3 H]Choline-GroPCho Labeling and Metabolite Analysis-Strains were grown in Iϩ low P i medium containing 10 M [ 3 H]choline-GroP-Cho to the late logarithmic/early stationary phase of growth (A 600 ϭ 1-1.5). The counts associated with the membranous and water-soluble fractions were determined through trichloroacetic acid extraction as described previously (9). Ion exchange chromatography was employed for the separation of [ 3 H]choline-containing metabolites (33).
Analysis of GroPIns Metabolism-Strains were grown for several generations in IϪ medium containing either low P i or no P i (7) and supplemented with 5-25 M [ 3 H]GroPIns. Counts associated with the membranous and water-soluble fractions were determined through trichloroacetic acid extraction as described previously (9). The watersoluble counts were separated by anion exchange chromatography (4).
Phospholipid Composition-Wild type and gde1⌬ mutant strains were grown to logarithmic phase in IϪ or Iϩ media containing 10 Ci/ml [ 32 P]orthophosphate. Labeled lipids were extracted (34), and individual phospholipid species were resolved (35) by two-dimensional chromatography.
Western Blot Analysis-Western analysis was performed as described previously (7) using a monoclonal antibody against GST (Covance Inc.; catalog number MMS-112P) as the primary antibody.
Cell Disruption and in Vitro Assays of Phosphodiesterase Activity-Several growth conditions, three methods of cell disruption, and multiple assay conditions were employed in attempts to detect in vitro glycerophosphodiester phosphodiesterase activity in wild type and GIT1 overexpressing strains of S. cerevisiae. The strains were grown in medium in which both inositol and phosphate concentrations were varied (Iϩ low P i , IϪ low P i , Iϩ high P i , and IϪ high P i ). Disruptions were performed using either glass bead breakage (36), detergent lysis using celLytic Y reagent (Sigma), or enzymatic lysis using a celLytic Y plus reagent kit (Sigma) in the presence of 1 mM phenylmethylsulfonyl fluoride or yeast protease inhibitor mixture (Sigma; catalog number P8215). Following disruption, the lysates were centrifuged for either 90,000 ϫ g or 20,000 ϫ g for 1 h. In each experiment, lysates, resuspended pellets, and supernatants were assayed for activity. The basic assay mix contained 55 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 0.5 mM [ 3 H]GroPCho, and 40 -100 g of protein in a total volume of 100 l, and the reactions were allowed to proceed for 1 h at 30°C. Variations on this assay included substitution of ZnCl or CaCl for MgCl, the addition of 5 mM dithiothreitol, the addition of various concentrations of detergents (Triton X-100, Nonidet P-40, and SDS), changing the pH of the buffer to 8.5 or 6.0 (50 mM MES), and allowing the reaction to proceed for several hours. The reactions were stopped by the addition of 300 l of cold methanol, followed by the addition of 300 l of chloroform to produce two phases. The upper layer was removed, dried down, and suspended in solvent A for subsequent HPLC analysis. In some cases, the reaction products were analyzed by ion exchange chromatography (33).

Gde1p (YPL110cp) and YPL206c Contain Glycerophosphodiester
Phosphodiesterase (GPDE) Motifs-A thorough understanding of the importance of phospholipid deacylation to cellular physiology requires knowledge about all aspects of the metabolism, including the fate of the resulting glycerophosphodiesters. Proteins containing GPDE motifs (Pfam accession number PF03009) as described by Pfam (38) are widespread in nature, with over 500 proteins containing the motif being reported by InterPro (www.ebi.ac.uk/interpro/) (39). Two S. cerevisiae gene products, Gde1p and YPL206c, contain GPDE motifs that could potentially be involved in the hydrolysis of GroPCho and/or GroPIns (Fig. 2). The focus of this work, Gde1p (Fig. 2), also contains ankyrin repeats (cd00204) and an SPX domain (pfam03105), as predicted by the NCBI CDART tool (db.yeastgenome.org/cgi-bin/protein/getDomain) (40,41). Ankyrin repeats mediate protein-protein interactions in a variety of proteins (42). SPX domains (named after yeast Syg1P and Pho81P and human XPR1) have been found in yeast proteins associated with G-protein signaling (Syg1p) (43) and phosphate sensing and transport (Pho81p, Pho90p, and Pho91p) (44). GDE1 encodes a protein of 1223 amino acids with a predicted molecular mass of 138 kDa. The UniProt accession number for GDE1, located on the left arm of chromosome 16, is Q02979.
A gde1⌬ Mutant Accumulates [ 14 C]Choline-GroPCho upon Labeling with [ 14 C]Choline-We began our studies by analyzing the distribution of choline-containing metabolites in strains bearing deletions in YPL110c and YPL206c. The cells were radiolabeled to steady state with [ 14 C]choline and the distribution of radiolabel upon chasing with nonradiolabeled choline for four h was analyzed. In wild type yeast, exogenous [ 14 C]choline becomes incorporated into phosphatidylcholine. Label that appears later in the trichloroacetic acid-soluble fraction of the cell is mostly in the form of GroPCho (9). Deletion of GDE1 resulted in an increase in water-soluble associated [ 14 C]choline-labeled compounds at the expense of membrane-associated counts (Fig. 3A). The distribution of [ 14 C]choline counts in the YPL206c⌬ strain, in contrast, was very similar to that of the wild type strain (Fig. 3A). All of the membrane-associated [ 14 C]choline counts were found to reside in PC (data not shown). HPLC analysis of the trichloroacetic acid-soluble metabolites indicated that in both strains, the majority of trichloroacetic acid-soluble counts (Fig. 3B) (Fig. 4B). Western analysis employing antibodies against the GST tag confirmed that the fusion protein was expressed under inducing conditions (Fig. 4C). Similar results were obtained when GDE1 was overexpressed from a multicopy plasmid, B94, under the control of its own promoter. The wild type strain containing GDE1 on a multicopy plasmid contained fewer trichloroacetic acid-soluble [ 14 C]choline-associated counts as compared with a wild type strain containing an empty vector (Fig. 4A), which translated into roughly 2-fold less internal GroPCho (Fig. 4B).
Because our [ 14 C]choline labeling studies with GDE1 delete and GDE1 overexpressing strains were consistent with the assignment of GDE1 as a GroPC phosphodiesterase encoding gene, we attempted to detect in vitro enzymatic activity. Despite growing the cells under a variety of conditions, employing three cell disruption techniques, and making multiple alterations in the assay components (see "Materials and Methods"), we were unable to detect any GPC phosphodiesterase activity in cell lysates or fractions derived from wild type or GDE1 overexpressing strains. Two types of overexpressing strains were employed: a strain bearing a GDE1-GST allele under the control of the GAL1 promoter and a strain bearing multicopy GDE1 under the control of its own promoter. Furthermore, we could detect no activity in a Gde1-GST fusion protein partially purified over a glutathione-agarose column. As a positive control, we were able to detect GroPC hydrolysis using a GroPC phosphodiesterase purified from mold (Sigma; catalog number G1642). Potential reasons for the lack of detectable in vitro activity are enumerated under "Discussion." Exogenous [   This study to increase in low phosphate conditions, we reasoned that growing the cells in low P i medium might maximize GroPCho uptake and utilization. When a wild type strain was grown in low P i medium supplemented with 10 M [ 3 H]choline-GroPCho, a small amount GroPCho was found intact in the trichloroacetic acid-soluble fraction of the cell (Fig. 5A). In addition, label was found in the membrane fraction, suggesting that choline was liberated from GroPCho via a glycerophosphodiesterase and subsequently incorporated into PC. In support of that reasoning, we found [ 3 H]choline in the medium fraction of the wild type strain (Fig. 5B). In contrast, growth of the gde1⌬ mutant in the presence of [ 3 H]choline-GroPCho resulted in a much different labeling pattern. As compared with wild type, the gde1⌬ mutant exhibited an   increased accumulation of [ 3 H]choline-GroPCho in the trichloroacetic acid extract (Fig. 5C), less counts associated with the membrane fraction (Fig. 5A), and little or no [ 3 H]choline in the medium or trichloroacetic acid extract (Fig. 5, B and C). Importantly, little or no [ 3 H]choline-GroPCho incorporation (Fig. 5A) occurred in a strain bearing a deletion in the gene encoding the glycerophosphoinositol (GroPIns) transport protein, GIT1 (Fig. 1), suggesting that GroPCho, in addition to GroPIns, is a substrate for Git1p. GDE1 and GIT1 Are Required for Utilization of GroPCho as Phosphate Source-S. cerevisiae was shown previously to be able to utilize GroPIns as its sole source of phosphate (7). We found that wild type S. cerevisiae could also use GroPCho as sole phosphate source but that deletion of GDE1 abrogated that ability (Fig. 6). As expected from the labeling results with [ 3 H]choline-GroPCho (Fig. 5), the GIT1 gene was also required for growth on GroPCho (Fig. 6). In contrast, the sole choline transporter of S. cerevisiae, encoded by the HNM1 gene (46), was not required.
Deletion of GDE1 and/or YPL206c Does Not Alter Catabolism of Exogenous GroPIns-The role of the two potential glycerophosphodiesterase encoding genes in GroPIns metabolism was also investigated. The majority of the GroPIns produced by S. cerevisiae is excreted into the medium (3). Furthermore, S. cerevisiae transports GroPIns into the cell where it is catabolized, presumably by a glycerophosphodiesterase, and its inositol portion is used in the synthesis of phosphatidylinositol (4). Thus, we analyzed the ability of the mutants to incorporate [ 3 H]inositol derived from exogenously supplied [ 3 H]GroPIns into phosphatidylinositol. No differences in labeling patterns were observed between wild type, gde1⌬, YPL206c⌬, and gde1⌬ YPL206c⌬ strains (Fig.  7A). Finally, strains bearing deletions in GDE1 and/or YPL206c grew similarly to wild type when GroPIns was supplied as sole phosphate source (Fig. 7B).
Steady State Phospholipid Composition-The phospholipid composition of wild type and gde1⌬ mutant strains was compared for cultures grown in IϪ and Iϩ synthetic media. No significant difference between the mutant and wild type strain was observed.

DISCUSSION
The studies reported here represent a substantial advancement in our understanding of the GroPCho metabolism in S. cerevisiae. We report the identification of a novel gene product, Gde1p, involved in GroPCho turnover. In addition, we report that GroPCho is taken up by yeast cells, that its transport across the plasma membrane is dependent upon the Git1p permease, and that GroPCho, like GroPIns (7), can be used as the sole source of phosphate for the cell. Also implicit from the described studies is that GroPCho acts as a source of choline for use in the synthesis of PC.
Several lines of evidence indicate that GDE1 encodes a glycerophosphodiesterase responsible for the hydrolysis of GroPCho in the cell. To begin with, upon uniform labeling of cells with [ 14 C]choline, the trichloroacetic acid-soluble portion of a gde1⌬ mutant contains more GroP-Cho as compared with wild type (Fig. 3), whereas strains overexpressing GDE1 contain less GroPCho (Fig. 4). Because exogenous [ 14 C]choline must first be incorporated into PC before a deacylation event liberates GroPCho into the trichloroacetic acid extract, it could be argued that Gde1p acts to negatively regulate PC deacylation, instead of acting to hydrolyze GroPCho. Arguing against that interpretation of the data is that a similar build-up of internal GroPCho is seen when a gde1⌬ mutant is exogenously labeled with [ 3 H]choline-GroPCho (Fig. 5). Furthermore, the product of glycerophosphodiesterase activity, [ 3 H]choline, was detected in the medium of a wild type strain labeled with [ 3 H]choline-GroPCho, whereas little or no [ 3 H]choline was detected in the medium or trichloroacetic acid extract of a gde1⌬ mutant (Fig. 5). Thus, the enzymatic reaction liberating choline from GroPCho that occurs in a wild type strain is greatly decreased in a gde1⌬ mutant. The final piece of evidence indicating a role for Gde1p in GroPCho hydrolysis is the finding that a wild type strain, but not a gde1⌬ mutant, can utilize GroPCho as the sole phosphate source (Fig. 6).
The absence of any label derived from [ 3 H]choline-GroPCho within the cells of a git1⌬ mutant indicates two important facts: (i) Git1p is required for GroPCho transport and (ii) label enters the cells as an intact GroPCho molecule. If GroPCho were degraded extracellularly, [ 3 H]choline would require its transporter, Hnm1p, to enter the cells, not Git1p. Indeed, the increased intracellular GroPCho observed in a gde1⌬ mutant strongly argues for the role of Gde1p as an intracellular GroP-Cho glycerophosphodiesterase. The ability of yeast to transport GroPCho through Git1p is an exciting, if unexpected, finding. Our previous work has shown Git1p to be a GroPIns transporter with little affinity for GroPCho (6). Indeed, short term transport assays indicate that the rate of GroPCho transport is roughly ten times less than that of GroPIns and that cells grow less robustly when 100 M GroPCho is supplied in the medium as compared with 100 M GroPIns (data not shown). Our previous work has demonstrated that GIT1 expression is up-regulated by phosphate limitation and that Git1p is required for the utilization of GroPIns as the phosphate source for the cell. We now report that Git1p is required for GroPCho uptake and for utilization of GroPCho as a phosphate source. These findings broaden the substrate range for Git1p and strengthen the suggestion that one role for Git1p is to provide a mechanism for surviving low phosphate stress by scavenging glycerophosphodiesters. Because S. cerevisiae secretes phospholipases B (5), it is easy to envision a situation in which phospholipases are dispatched to hydrolyze phospholipids, releasing glycerophosphodiesters, which can then be taken up into the cell.
In addition to being used as a phosphate source, our studies also indicate that GroPCho acts as a precursor for PC synthesis. Upon labeling with [ 14 C]choline, the build-up of GroPCho seen in a GDE1 mutant occurs at the expense of incorporation of label into PC (Fig. 3), suggesting that the pool of choline found in GroPCho is normally recycled back into PC biosynthesis. Similarly, overexpression of GDE1 (Fig. 4) results not only in less GroPCho accumulation but greater incorporation of choline label into PC, suggesting accelerated recycling of the GroPChoderived choline pool into PC. Finally, upon labeling with [ 3 H]choline-GroPCho, choline label is clearly found associated with the membrane fraction of the cell.
Strong homologs of Git1p are found in many fungal and plant species, including Candida albicans, Neurospora crassa, and Arabidopsis thaliana (www.ncbi.nlm.nih.gov/BLAST/). Although clear sequence  homologs have not been noted in mammalian cells, a definitive determination of the existence of Git1p-like proteins in mammals awaits a better understanding of the elements responsible for conferring glycerophosphodiester substrate specificity to the transport protein. Interestingly, evidence for GroPIns-4-P transport activity in Swiss 3T3 cells has been reported (47).
Our labeling data suggest that GroPCho is hydrolyzed by Gde1p to produce choline and glycerolphosphate (Fig. 1). The subsequent metabolism employed by S. cerevisiae to grow on glycerolphosphate in the absence of inorganic phosphate (Fig. 6) is not known but will be the subject of future research. The finding that the gde1⌬ mutant incorporates some label from [ 3 H]choline-GroPCho into the membrane fraction (Fig. 5) suggests that another gene product may exist that is capable of releasing free [ 3 H]choline for subsequent synthesis of PC. However, that hypothetical gene product is not robust enough to allow for detection of [ 3 H]choline in the medium or trichloroacetic acid extract or to support growth of a gde1⌬ mutant on GroPCho. The in vivo hydrolyzing ability of Gde1p appears to be specific for GroPCho, because we could detect no alterations in GroPIns metabolism in a gde1⌬ mutant (Fig.  7A), and gde1⌬ could utilize GroPIns as sole phosphate source as well as wild type (Fig. 7B). Deletion of the other S. cerevisiae open reading frame containing a GPDE motif, YPL206c, has no effect upon GroPIns or GroPCho metabolism based upon the labeling and growth experiments described here. The nature of the metabolite(s) hydrolyzed by YPL206cp is under study.
We can envision four potential reasons for our inability to detect in vitro glycerophosphodiesterase activity: (i) The GDE1-GST fusion protein may not be active. Arguing against this interpretation of the data is that a strain (JPV436) bearing the fusion construct clearly contains less intracellular GroPCho upon choline labeling than the corresponding wild type (Fig. 4). Also, we were similarly unable to detect in vitro activity in a wild type strain bearing multicopy GDE1 (JPV203 ϩ plasmid B94). (ii) The protein may be rapidly degraded upon cell disruption. Arguing against protein degradation being the major factor is that the vast majority of immunoreactivity resides in a single band upon Western analysis of fractions derived from JPV436. (iii) We may not have employed the proper assay conditions. This possibility can never be completely ruled out. We varied a number of standard variables (pH, metal ions, protease inhibitors, reducing agents, and detergent), to no avail. (iv) We may have disrupted necessary protein-protein or proteinlipid interactions in the course of cell disruption. This would seem to be the most likely possibility. The facts that massive overexpression of the GDE1-GST fusion allele results in roughly a 4-fold decrease in GroPCho and that multicopy GDE1 confers only a 2-fold decrease in GroPCho suggest that there may be some other limiting factor required for glycerophosphodiesterase activity. That limiting factor may be an interacting protein or perhaps a membrane surface. Similar to our results, glycerophosphodiesterase activity catalyzed by the product of the E. coli upgQ gene could not be detected in cell extracts (48). The authors concluded that the glycerophosphodiesterase was physically dependent upon another protein, possibly the transport protein responsible for glycerophosphodiester uptake (48). Gde1p activity is not dependent upon the transporter required for GroPCho uptake, as evidenced by the finding that a git1⌬ mutant labeled with [ 14 C]choline has wild type GroPCho levels, not the increased levels seen in a gde1⌬ mutant (data not shown). However, the predicted domain structure of Gde1p suggests potential protein-protein interactions. Ankyrin repeats (38) are located in the middle of Gde1p, roughly localized between residues 300 and 600. Ankyrin repeats have been found in organisms ranging from viruses to humans, and their role in mediating protein-protein interac-tions is well documented (42). The SPX (38) domain located at the N terminus of Gde1p also has the potential to mediate a protein-protein interaction. The SPX domain of yeast Syg1p binds to the G-protein ␤-subunit and inhibits transduction of the mating pheromone signal (43). Interestingly, a number of proteins involved in phosphate sensing and metabolism also contain N-terminal SPX domains (38). Our finding that Gde1p is required for the utilization of GroPCho as phosphate source, together with microarray studies reporting increased GDE1 (YPL110c) expression in low phosphate conditions (45), would seem to be in keeping with the observation that Gde1p has an SPX domain. Gde1p was predicted by a global analysis of green fluorescent proteintagged proteins (49) to be cytoplasmic. We found the GST-tagged version of the protein in the membrane fraction upon cell disruption by detergents and in both the membrane and soluble fractions upon glass bead disruption (data not shown). Future studies on Gde1p will include analyses of its cellular localization, its potential binding partners, and its role in phosphate metabolism.