Transporter-Driven Glycerophosphocholine (GPC) Toxicity Is Conserved from Fission Yeast to Budding Yeast: Roles for Inositol Pyrophosphates and Gde1 Regulation in Fission Yeast

TL;DR

This study shows that GPC toxicity in yeast is driven by transporters and regulated by inositol pyrophosphates and Gde1, a process conserved across yeast species.

Contribution

The study reveals a conserved mechanism of GPC toxicity involving transporters, inositol pyrophosphates, and Gde1 regulation in fission and budding yeast.

Findings

Tgp1 in fission yeast specifically transports GPC, causing growth impairment but not reduced viability.

Gde1 hydrolyzes GPC and its deletion or mutations in key domains suppress GPC toxicity.

GPC toxicity in overexpressing Tgp1 or Git3 is independent of Gde1 and IP8 in both fission and budding yeast.

Abstract

Glycerophosphocholine (GPC) and glycerophosphoinositol (GPI) are phospholipid metabolites generated by phospholipase-mediated deacylation. In budding yeast, they enter cells via the Git1 permease; in fission yeast, the homolog is Tgp1. This study investigates why GPC is toxic to asp1-STF mutants, where Tgp1 is upregulated due to loss of Asp1 pyrophosphatase, resulting in elevated inositol pyrophosphate 1,5-IP8. We show that S. pombe Tgp1 specifically transports GPC, explaining why GPC, but not GPI, impairs growth. Increased GPC uptake slows doubling time but does not reduce viability. Toxicity is relieved by deletion of Gde1, a phosphodiesterase that hydrolyzes GPC to choline and glycerol-3-phosphate. Mutations in either the Gde1 active site or SPX domain also suppress toxicity, and radiolabeling confirms both domains are required for enzymatic activity. GPC is toxic in cells vastly overexpressing Tgp1 even without elevated IP8, but Gde1 loss does not suppress this effect. Similarly, in S. cerevisiae overexpressing the Candida albicans Git3 transporter, GPC provision causes toxicity independent of Gde1. Loss of Gpc1, the acyltransferase converting GPC to lysophosphatidylcholine, does not alter toxicity in either yeast. These findings highlight a conserved process by which GPC regulates growth and reveal a role for IP8 in modulating this process.

Figures

Full text

1. Introduction

Glycerophosphodiesters are metabolites produced through phospholipase-mediated deacylation of phospholipids. In Saccharomyces cerevisiae (budding yeast), they are produced internally and can be released extracellularly [1,2,3,4]. The glycerophosphodiesters glycerophosphocholine (GPC) and glycerophosphoinositol (GPI) can be imported into the cell through the transporter Git1 in S. cerevisiae. Git1 has the highest affinity for GPI and less affinity for GPC [5,6]. Git1 was the first glycerophosphodiester transporter to be characterized in a eukaryotic cell [6]. Git1 homologs exist in multiple fungal species, and four Git1 paralogs have been characterized in Candida albicans [7,8,9,10]. The Candida paralogs display specificity for either GPI (CaGit1) or GPC (CaGit3 and CaGit4). In the fission yeast Schizosaccharomyces pombe, Tgp1 (a 528-aa polypeptide) is the sole Git1 homolog [11,12,13,14,15] whose substrate specificity has not been determined. Tgp1 is more closely related in its primary structure to CaGit3 (204 positions of side chain identity) and CaGit4 (197 identities) that transport GPC than it is to S. cerevisiae Git1 (149 identities) or CaGit1 (157 identities) that are selective for GPI transport. Expression of S. cerevisiae Git1 and S. pombe Tgp1 is repressed during growth under phosphate-replete conditions and is transcriptionally upregulated during phosphate limitation to aid in phosphate acquisition by importing extracellular glycerophosphodiesters that can be metabolized by intracellular enzymes. Gde1 is an intracellular glycerophosphodiesterase that hydrolyzes GPC to choline and glycerol-3-phosphate (G3P); expression of gde1 is upregulated during acute phosphate starvation of budding and fission yeast [4,12].

tgp1 is part of a three-gene fission yeast PHO regulon that includes phosphate acquisition genes pho1 (cell surface acid phosphatase) and pho84 (inorganic phosphate transporter) [16]. The PHO genes are repressed under phosphate-replete conditions by upstream lncRNA-mediated transcriptional interference and are induced during phosphate starvation when synthesis of the interfering lncRNAs is turned off [12,15]. PHO lncRNA 3′-processing and termination is a sensitive control point in PHO mRNA repression, insofar as transcriptional interference can be tuned by increasing or decreasing the frequency with which RNA polymerase II terminates lncRNA transcription prior to encountering the mRNA promoter [14,15]. Genetic maneuvers that enhance precocious termination of lncRNA transcription result in derepression of PHO mRNA expression in phosphate-replete cells, and those that reduce the probability of lncRNA termination prior to the mRNA promoter result in hyper-repression of the flanking PHO mRNAs relative to their basal levels [15].

Inositol pyrophosphates, especially 1,5-IP_8_, are signaling molecules that impact fission yeast PHO gene expression by acting as agonists of precocious RNA 3′-processing/termination of upstream lncRNA synthesis, thereby relieving the lncRNA-mediated interference [14,15]. Inositol pyrophosphates are also involved in the transcriptional control of phosphate-responsive genes in budding yeast and plants, albeit via mechanisms that do not involve transcriptional interference [17,18,19,20].

1,5-IP_8_ is synthesized by the sequential action of position-specific kinases Kcs1, which converts IP_6_ to 5-IP_7_, and Asp1, which converts 5-IP_7_ to 1,5-IP_8_. Asp1, the key agent in IP_8_ dynamics, is a bifunctional enzyme composed of an N-terminal kinase domain that synthesizes 1,5-IP_8_ and a C-terminal pyrophosphatase domain that hydrolyzes 1,5-IP_8_ back to 5-IP_7_. An inactivating mutation of the Asp1 pyrophosphatase active site (H397A) increases intracellular 1,5-IP_8_ [21,22] and results in upregulation of the tgp1, pho1, and pho84 mRNAs by 21-fold, 6-fold, and 4-fold, respectively [14]. Asp1 pyrophosphatase-inactivating nonsense alleles—named asp1-STF6 and asp1-STF9 that truncate the 920-aa Asp1 protein at Trp386 and Trp493, respectively—result in a severe growth defect on YES (Yeast Extract with Supplements) medium. A screen for extragenic suppressors led to the highly instructive findings that: (i) truncation or deletion of tgp1 completely restored growth of asp1-STF6/9 cells on YES medium; (ii) asp1-STF6/9 cells grew normally on an enriched synthetic medium ePMGT; (iii) titrating yeast extract into ePMGT reprised the growth defect, signifying that yeast extract likely contains a toxic substance imported by Tgp1; (iv) titrating GPC into ePMGT recapitulated the asp1-STF6/9 growth defect; and (v) RNA-seq showed that tgp1 mRNA increased by 43-fold and 47-fold, respectively, in asp1-STF6 and asp1-STF9 cells grown in ePMGT medium [11]. The greater upregulation of tgp1 (and ensuing toxicity) in asp1-STF6/9 cells vis-à-vis asp1-H397A cells correlated with a greater increase in 1,5-IP_8_ levels [21].

These results engendered a working model whereby the increased IP_8_ in asp1-STF6/9 cells leads to overexpression of tgp1 and thus increased import of GPC. Consistent with this idea, we found that induced overexpression of tgp1 in asp1^+^ wild-type cells also elicited toxicity dependent on GPC in the medium [11]. It is conceivable that elevated levels of intracellular GPC are toxic per se, or that increased GPC perturbs phospholipid metabolism, for example, by driving excessive synthesis of lysophosphatidylcholine (LPC) via GPC acyltransferase Gpc1 or overproduction of phosphatidylcholine (PC) via the sequential actions of Gpc1 and the lysophospholipid acyltransferase Ale1. Alternatively, elevated GPC leads to its conversion into elevated levels of non-lipid metabolites that are toxic. Consistent with the latter scenario, we found that ablating Gde1 suppressed the growth defect of asp1-STF6/9 cells on YES medium at 30 °C [11,13,23]. The 1134-aa Gde1 protein consists of an N-terminal SPX IP_8_-binding/sensor domain, a central ankyrin repeat domain, and a C-terminal glycerophosphodiesterase catalytic domain. Mutagenesis tests suggested that IP_8_ binding and phosphodiesterase catalytic activity are both required for Gde1’s function in mediating the growth defect of asp1-STF6/9 cells on YES medium [11,13,23].

The goal of this study was to examine the molecular underpinnings of the growth inhibition caused by overzealous import of GPC. Using metabolic labeling, we investigated the substrate specificity of Tgp1 and related that to the relative toxicity of GPC versus GPI. We find that GPC prolongs the doubling time of asp1-STF cells but does not elicit cell death. We examined the role of Gde1, which revealed that the inositol pyrophosphate binding site in the SPX domain is essential for Gde1 hydrolysis of GPC in vivo and for GPC toxicity in asp1-STF fission yeast. Interdiction of GPC conversion to LPC (by ablating the acyltransferase Gpc1) or G3P conversion to phosphatidylglycerol (by eliminating the phosphatase Gep4) did not impact asp1-STF6/9 toxicosis. The toxicity studies were extended to budding yeast, enabled by the heterologous expression of the Candida albicans Git3 transporter gene that has a high affinity for GPC [7,8,9,10]. Overall, this study defines the parameters of GPC toxicity in terms of transporter specificity and downstream GPC metabolism.

2. Materials and Methods

2.1. Media and Strain Maintenance

All S. pombe and S. cerevisiae strains (Table 1) were grown aerobically at 30 °C in liquid culture or on agar plates. To maintain S. cerevisiae, we used YPD (yeast peptone dextrose) or YNB (yeast nitrogen base) media that were prepared as described [24]. YNB ura-dropout medium was used to select for S. cerevisiae cells transformed with pRS416 and p416-CaGit3 plasmids. All S. pombe strains were maintained in enhanced Pombe Minimal Glutamate plus Thiamine (ePMGT) medium that was prepared as described [11,12]. ePMGT was used for all the experiments except where indicated, where we also tested growth on YES (Yeast Extract with Supplements) agar medium. A Thermo Fisher Scientific BioMate160 spectrophotometer, Waltham, MA, USA, was used to assess cell growth by measuring absorbance at 600 nm (A600) as indicated in the figure legends.

2.2. Sources of 3H-Inositol-GPI and 14C-Choline-GPC

Radiolabeled GPI was produced as described [5,28]. In brief, ^3^H-inositol-phosphatidylinositol (20 µCi, American Radiolabeled Chemicals, Inc., St. Louis, MO, USA) was dried under nitrogen gas. Chemical deacylation was performed by adding 100 µL methylamine reagent (27% of 40% methylamine, 46% methanol, 11% n-butanol, 16% H_2_O) and incubating at 50 °C for 1 h. The sample was dried via SpeedVac centrifugation, resuspended in H_2_O (250 µL), and sonicated. The sample was mixed with 250 µL of N-butanol:petroleum ether:ethylformate (20:4:1), and the aqueous (bottom) layer was transferred to a new tube. Following a second extraction with 250 µL of N-butanol:petroleum ether:ethylformate, the aqueous fraction was dried via SpeedVac centrifugation and resuspended in 200 µL MeOH:H_2_O (1:1). Recovery of radiolabel (in counts per minute [CPM]) was determined using a PerkinElmer Tri-Carb 4910 TR scintillation counter, Hopkinton, MA, USA. To confirm its identity, the newly synthesized ^3^H-inositol-GPI was analyzed by thin layer chromatography (TLC) alongside ^3^H-inositol-PI and ^3^H-inositol-GPI standards using the solvent system chloroform:methanol:ammonium hydroxide (6:10:4.4) [28]. The TLC plate was exposed to a Fujifilm Imaging Plate BAS-TR2025, Valhalla, NY, USA and visualized via a Typhoon 8200 phosphorimager, Uppsala, Sweden.

^14^C-choline-GPC was purchased from American Radiolabeled Chemicals, Inc. (Item# ARC 3880).

^3^H-inositol-GPI and ^14^C-choline-GPC were used in transport assays as described below in Section 2.3.

2.3. 14C-Choline-GPC and 3H-Inositol-GPI Transport Assays

S. pombe cultures (2 mL) were inoculated thickly from fresh plates. Cultures were allowed to grow for 6–8 h, at which point additional ePMGT medium (3 mL, for a total volume of 5 mL) was added. The following day, cultures were restarted in ePMGT media at an A600 of 0.8. Cultures were allowed to grow for two hours to an A600 of approximately 1.0. At this point, the radiolabeled compound of interest (^14^C-choline-GPC at 50,000 CPM/mL; ^3^H-inositol-GPI at 5000 or 10,000 CPM/mL) was added as a mixture with non-labeled GPC or GPI (300 nM final concentration). Transport parameters were optimized by testing several time points and concentrations of GPC to establish conditions within the linear range to be used for the experiments presented here. Time zero samples were taken, as were samples following 30 min of growth. Per time point, 1 mL of cells from each culture was added to an aliquot of carrier cell preparation (100 µL; see Section 2.4 below) and centrifuged for 1 min at 6000 rpm on a benchtop microcentrifuge. Samples were separated into supernatant (~1000 µL) and pellet (resuspension volume 100 µL water) fractions. Half the volume of each fraction was added to 5 mL of scintillation fluid (MP Biomedicals EcoLume), and CPMs were determined on a PerkinElmer Tri-Carb 4910 TR scintillation counter. Readings were multiplied to calculate total counts per fraction and ultimately represented as percentages in pellet vs. supernatant or as pmol/min/ODU. Experiments were performed using three independent cultures (biological triplicates). Note that this method is similar to that employed for C. albicans [9] but altered from the method typically used for S. cerevisiae [6].

2.4. Carrier Cell Preparation

To facilitate pelleting of cells in transport assays and in vivo metabolic labeling, we prepared “dead” carrier cells as follows: S. cerevisiae cells were grown in liquid culture, harvested by centrifugation, resuspended in 100 µL of 5% trichloroacetic acid (TCA) per ODU (1 A600 unit), and incubated for 10 min. After centrifugation, the pelleted material was resuspended in 200 µL of water per ODU. An aliquot (100 µL, 0.5 ODU) of the carrier cell preparation was used for each 1 mL sample analyzed in transport and metabolic labeling experiments.

2.5. In Vivo Metabolic Radiolabeling with 14C-Choline-GPC

S. pombe cultures (2 mL) were inoculated thickly from fresh plates. Cultures were allowed to grow for 6–8 h, at which point 3 mL of additional ePMGT medium was added for a total volume of 5 mL. Experiments were performed using three independent cultures. The following day, cultures were restarted in ePMGT medium at an A600 of 0.8. Cultures were allowed to grow for two hours to an A600 of approximately 1.0 before a mixture of radiolabeled GPC (^14^C-choline-GPC, 50,000 CPM/mL) and non-labeled GPC (300 nM final concentration) was added. Samples were taken at time 0 and after 15, 30, and 45 min of incubation. At each time point, aliquots (1 mL) of the cultures were added to 100 µL of carrier cell preparation and centrifuged. Samples were separated into supernatant (~1000 µL) and pellet (resuspension volume 100 µL H_2_O) fractions. An aliquot of each supernatant fraction was subjected to liquid scintillation counting (LSC) to represent extracellular GPC. Pellets were treated with 5% TCA for 20 min, centrifuged, and the supernatant was transferred to a new tube. Pellets were then resuspended in 1M Tris-HCl (pH 8), vortexed, and centrifuged again. Together, the supernatants from TCA and Tris-HCl resuspensions were pooled, and a portion was counted to represent the intracellular fraction (1000 µL volume). Finally, the remaining pellet (membrane fraction) was resuspended in 500 µL Tris-HCl (pH 8), and a portion was subjected to LSC. Readings were multiplied to calculate total counts per fraction and ultimately represented as a percentage in each fraction.

Intracellular fractions were processed via column chromatography as described in detail [29] to separate labeled choline (resultant of GPC metabolism) from GPC. 50W x8 H+ Dowex resin was prepared and pipetted to make 250 µL columns. Columns (disposable Kimble Kontes Disposaflex^®^ Polypropylene Funnel, Vineland, NJ, USA, item #420160) were conditioned by adding 1 mL of a 1:1 5% TCA: 1 M Tris-HCl (pH 8) mixture. Following conditioning, a portion of each intracellular fraction was diluted 1:6 with water and applied to a column. [^14^C]GPC was eluted with a 1-mL and 2-mL H_2_O wash. [^14^C]choline, produced through GPC catabolism, was eluted with 5 mL of HCl. Standards were used to verify the separation procedure, and the radiolabel incorporated into each metabolite was quantified by liquid scintillation counting.

2.6. Growth Analyses

To assess fission yeast growth on agar plates, cultures were grown in liquid ePMGT medium until A600 reached 0.2–0.6 and then adjusted to an A600 of 0.1. Aliquots (3 µL) of serial five-fold dilutions were spotted on ePMGT agar or YES agar as indicated in the figure legends.

For both S. pombe and S. cerevisiae, liquid growth curves were generated using a Molecular Devices SpectraMax i3 microplate reader, San Jose, CA, USA. The 96-well plates were incubated at 30 °C with hourly shaking, followed by absorbance readings at 600 nm (A600). Data points plotted represent measurements taken at 4-h intervals. For experiments involving S. pombe, 2 mL cultures were initially prepared in ePMGT medium at the start of day zero. Approximately eight hours later, medium was added to bring the total volume to 5 mL per culture, and the cultures were incubated overnight. The following day, a 96-well plate was inoculated at an initial density of A600 = 0.005. GPC was added as indicated. For growth experiments involving S. cerevisiae overnight cultures were used to inoculate a 96-well plate at A600 = 0.01 in YNB ura^−^ medium. GPC or GPI was added as indicated.

2.7. S. pombe Strain Constructions

To generate gep4∆ and gpc1∆ deletion strains, we used PCR amplification and standard cloning methods to first construct bacterial plasmids in which the entire gep4 or gpc1 coding sequences were replaced by a natMX antibiotic-resistance cassette. The disruption cassettes, in which the resistance marker is flanked by ~500–600 bp fragments of upstream and downstream gene-specific chromosomal DNA, were excised and transformed into S. pombe diploid cells. Nourseothricin-resistant transformants were selected and analyzed by diagnostic PCR to confirm correct integration at one of the gpc1 or gep4 loci. Upon sporulation of the heterozygous diploids and random spore analysis [30], nourseothricin-resistant gep4∆ and gpc1∆ haploids were isolated. Standard fission yeast methods were employed to generate strains with mutations in two differently marked genes [30,31].

2.8. Survival Assays

All cultures of fission yeast cells (3 replicates per strain) were maintained at an A600 of ≤0.8 by dilution into fresh medium as necessary. The wild-type, asp1-STF6, and asp1-STF9 strains were grown in ePMGT medium prior to the addition of GPC (500 µM). Wild-type cells harboring plasmids (p[nmt1-tgp1] or the empty vector control) were first grown in ePMGT-leu medium, wherein selection for the LEU2 plasmid is maintained, but the nmt1 promoter is repressed by the presence of thiamine [27]. Cells were harvested by centrifugation, washed with water, and resuspended in ePMG-leu medium to induce tgp1 expression from the nmt1 promoter. Cultures were grown for 18 h prior to the addition of GPC (500 µM). Immediately prior to the addition of GPC (time 0) and at various times afterwards, the number of cells in each culture was determined using a hemocytometer, and aliquots were plated to ePMGT or ePMGT-leu agar medium. Percent survival was determined as described in the figure legends.

2.9. Reverse Transcriptase Quantitative PCR Analysis

Total RNA was isolated from: (i) S. pombe wild-type (asp1^+^), asp1-STF6, and asp1-STF9 cells grown in liquid ePMGT medium at 30 °C to an A600 of 0.5 to 0.8; and (ii) asp1^+^ p[nmt1-41x] (empty LEU2 vector) and asp1^+^ p[nmt1-41x–tgp1] (tgp1 overexpressing) cells grown for 18 h in liquid ePMG–leu medium at 30 °C (three independent cultures for each strain). Cells were harvested by centrifugation, and total RNA was extracted via the hot phenol method. The RNAs were treated with DNase I, extracted serially with phenol:chloroform and chloroform, and then precipitated with ethanol. The RNAs were resuspended in 10 mM Tris HCl (pH 6.8) and 1 mM EDTA and adjusted to a concentration of 500 ng/μL. Reverse transcription was performed with 2 μg of this RNA template plus oligo(dT)18 and random hexamer primers by using the Maxima First Strand cDNA synthesis kit (Thermo Scientific, Waltham, MA, USA). After cDNA synthesis for 30 min at 55 °C, the reverse transcription reaction mixtures were diluted 10-fold with water. Aliquots (2 μL) were used as templates for tgp1^+^ and act1^+^ gene-specific quantitative PCR (qPCR) reactions directed by sense and antisense primers [32]. The qPCR reactions were constituted with the Maxima SYBR Green/ROX master mix (Thermo Scientific) and monitored with an Applied Biosystems (Waltham, MA, USA) QuantStudio 6 Flex Real-Time PCR system. The qPCR reactions were performed in triplicate for each cDNA population. The level of tgp1 cDNA was calculated relative to that of act1 cDNA by the comparative Ct method [33]. The actin-normalized levels of the tgp1 transcript in wild-type asp1^+^ cells were assigned a value of 1.0, and the tgp1 mRNA levels in the other strains were then normalized to the wild-type control value.

3. Results

S. pombe strains with enhanced tgp1 expression are sensitive to exogenous GPC, not GPI, in liquid medium. Previous studies showed that fission yeast asp1-STF strains overexpressing tgp1 display a growth defect on defined agar medium (ePMGT) supplemented with increasing concentrations of GPC [11]. Here we confirm that the toxicity also occurs in liquid ePMGT medium in a Tgp1-dependent manner (Figure 1A–C, Supplemental Figure S2A). Growth of asp1-STF9, a strain in which tgp1 mRNA is upregulated by 47-fold as gauged by RNA-seq [11], was inhibited by the addition of 250 or 1000 µM GPC (Figure 1B), whereas growth of wild-type cells was insensitive to 250 µM GPC and 1000 µM GPC (Figure 1A). asp1-STF9 tgp1∆ cells were also insensitive to 250 and 1000 µM GPC (Figure 1C), thereby affirming that Tgp1 was responsible for GPC toxicity. By contrast, growth of wild-type, asp1-STF9, and asp1-STF9 tgp1∆ cells was unaffected by 250 or 1000 µM GPI (Figure 1D–F).

Tgp1 imports GPC, but not GPI, in asp1-STF cells. To determine Tgp1 substrate specificity, transport assays were performed by providing ^14^C-choline-GPC or ^3^H-inositol-GPI to fission yeast cultures grown in ePMGT medium containing a total of 300 nM GPC or GPI. An initial experiment was performed in which asp1-STF9 cells in ePMGT medium were incubated with ^14^C-choline-GPC and, either immediately (time 0) or after 15, 30, or 45 min, separated by centrifugation into supernatant and pellet fractions, with the supernatant containing the labeled GPC not entering the cell, and the pellet comprising the labeled GPC transported into the cell. Supplemental Figure S1A shows that there was no GPC uptake at time 0 and that the extent of total GPC import increased linearly with incubation time up to 45 min, concomitant with a decrease in extracellular GPC. Thus, a 30-min incubation was employed for the experiments shown in Figure 2B,C, in which we measured GPC uptake by various S. pombe strains. We found that GPC was imported into asp1-STF6 and asp1-STF9 cells to mean extents of 41% and 76% of the input labeled GPC, respectively (Figure 2B). This corresponds to import rates of 4 pmol/min and 7 pmol/min per A600 unit of asp1-STF6 and asp1-STF9 cells, respectively (Figure 2C). By contrast, wild-type cells did not detectably import GPC during the 30 min incubation (Figure 2B) because tgp1 expression is actively repressed in wild-type cells by lncRNA-mediated transcriptional interference. Deletion of tgp1 effaced GPC uptake in the asp1-STF6 background (Figure 2B) and reduced GPC import in asp1-STF9 tgp1∆ cells to near-background levels seen for the asp1-STF9 strain at time 0 (Figure 2A,B).

There was no detectable import of labeled GPI by either wild-type or asp1-STF cells (Figure 2E), signifying that Tgp1 is specifically dedicated to the transport of GPC. This result is consistent with the finding that exogenous GPC caused growth inhibition of asp1-STF6/9 cells, but GPI provision did not (Figure 1), and the fact that Tgp1 has higher (72%) identity with CaGit3, a GPC transporter, than with ScGit1, which primarily transports GPI [34,35,36,37,38,39,40,41]. It is conceivable that S. pombe can import GPI via an as-yet unidentified transporter that is not recognizably homologous to Tgp1; however, this hypothetical GPI transporter is not expressed under the standard growth conditions employed here.

Gde1 abets GPC toxicity in asp1-STF cells. Since GPC must enter the cell to be toxic, we sought to determine if GPC itself or metabolic products of GPC are responsible for the growth inhibition of the asp1-STF strains. We focused first on the glycerophosphodiesterase Gde1, in light of the previous identification of loss-of-function gde1 mutations as extragenic suppressors of the growth defect of asp1-STF cells on YES medium [23]. Here we found that gde1∆ suppressed the growth defect of asp1-STF6/9 cells on ePMGT medium supplemented with increasing concentrations of GPC (Figure 3A). Gde1 consists of an N-terminal SPX inositol pyrophosphate-binding/sensor domain, a central ankyrin repeat domain, and a C-terminal glycerophosphodiesterase catalytic domain. Because the asp1-STF alleles underlying Tgp1-mediated GPC toxicity elicit increased intracellular 1,5-IP_8_ and 1-IP_7_, and given that Gde1 contains an inositol pyrophosphate-sensing SPX domain that could potentially affect Gde1 catalytic function, it was of interest to see whether inositol pyrophosphate binding and/or glycerophosphodiesterase activity were required for Gde1 to abet GPC toxicity. To generate a catalytically dead Gde1, we replaced the chromosomal gde1 gene with an active site double-alanine mutant H810A-R811A [23,42] (Figure 3B, right panel). The gde1-H810A-R811A allele phenocopied gde1∆ with respect to suppression of GPC toxicity to asp1-STF6/9 cells at 30 °C (Figure 3A).

To query the effect of perturbing the inositol pyrophosphate binding site of Gde1, we introduced a triple SPX domain mutation, Y21F-K25A-N139A (Figure 3B, left panel), into the chromosomal gde1 gene [23]. The gde1-Y21F-K25A-N139A allele phenocopied gde1∆ in alleviating GPC toxicity to asp1-STF6/9 cells at 30 °C (Figure 3A). Liquid growth assays in ePMGT medium containing 0, 250, or 1000 µM GPC affirmed that asp1-STF9 strains lacking Gde1 or bearing mutations in the SPX domain or active site of Gde1 are resistant to exogenous GPC (Supplemental Figure S2 panels E,F,G).

The SPX domain of Gde1 regulates its glycerophosphodiesterase activity. Because growth inhibition via GPC requires both the catalytic and SPX domains of Gde1, the role of the SPX domain of Gde1 for in vivo enzymatic activity was examined. Metabolic labeling studies were performed by adding ^14^C-choline-GPC to log-phase cultures. Following labeling, harvested cells were fractionated into membrane and water-soluble components. The water-soluble metabolites ^14^C-choline and ^14^C-choline-GPC were subsequently separated. Gde1 activity is indicated by the release of ^14^C-choline from ^14^C-choline-GPC (Figure 4A). The amounts of labeled choline versus GPC in the water-soluble fraction (expressed as % of total ^14^C label in the soluble and membrane fractions; average of three biological replicates) are plotted in Figure 4B. The label distribution in asp1-STF9 cells (17% choline versus 41% GPC) indicated that 29% of the soluble GPC was hydrolyzed to choline. The distributions of asp1-STF9 cells with mutant gde1 alleles were as follows: gde1∆ (3% choline versus 63% GPC), gde1-H810A-R811A (6% choline versus 51% GPC), and gde1-Y21F-K25A-N139A (5% choline versus 44% GPC) (Figure 4B). These results affirm that Gde1 is principally responsible for the hydrolysis of cytosolic GPC via its catalytic domain and that inositol pyrophosphate engagement by the SPX domain is required for Gde1 activity in vivo.

Incorporation of label from ^14^C-choline-GPC into the membrane fraction (Figure 4C) may arise via two pathways: (i) the direct acylation of GPC by the acyltransferase Gpc1 followed by lysophospholipid acyltransferase Ale1 (the GPCAT pathway) or (ii) the Kennedy pathway for phosphatidylcholine (PC) biosynthesis, which begins with free choline [33] (Figure 4A). asp1-STF9 gde1∆, and, to a lesser extent, asp1-STF9 gde1-H810A-R811A and asp1-STF9 gde1-Y21F-K25A-N139A cells displayed fewer membrane-associated counts than asp1-STF9, consistent with there being less free labeled choline available (Figure 4B) for incorporation via the Kennedy pathway.

Interdicting anabolic pathways of GPC and G3P utilization does not suppress asp1-STF6/9 toxicity. We considered the prospect that other metabolites of GPC or G3P might contribute to the growth defect of asp1-STF6/9 cells. As shown in Figure 5A, GPC is deployed as a precursor for the synthesis of PC via a two-step pathway in which GPC is converted to lysophosphatidylcholine (LPC) by the acyltransferase Gpc1 and then to PC by the acyltransferase Ale1. To query whether excess GPC might drive toxicity via increased LPC or PC formation, we deleted the gpc1^+^ gene, per se, and in combination with asp1-STF6 and asp1-STF9. (Deletion of gpc1 in the asp1-STF9 strain background resulted in fewer membrane-associated counts versus asp1-STF9 (Figure 4C), consistent with interdiction of the GPCAT pathway.) Spot tests for growth on agar medium showed that all single and double mutants grew as well as wild-type on ePMGT medium that lacks GPC. Deleting gpc1 had no effect on cell growth on YES medium. The instructive findings were that gpc1∆ did not relieve the severe growth defect of asp1-STF6 and asp1-STF9 cells on YES medium that does contain GPC (Figure 5B). Liquid growth experiments affirmed that asp1-STF9 gpc1∆ cells phenocopied asp1-STF9 cells with respect to GPC sensitivity (Supplemental Figure S3).

Given the genetic evidence implicating Gde1 in GPC toxicity of asp1-STF6/9, presumably via increased generation of G3P in response to elevated inositol-1-pyrophosphates, we queried whether this reflects increased flux of G3P into phosphatidylglycerol (PG), via a two-step pathway in which the enzyme Pgs1 catalyzes the reaction of G3P with CDP-diacylglycerol to form phosphatidylglycerol phosphate (PGP), which is then dephosphorylated by the Gep4 phosphatase to generate PG (Figure 5A). Whereas Pgs1 is essential in fission yeast, Gep4 is inessential. Thus, we deleted the gep4^+^ gene, singly and in tandem with asp1-STF6. gep4∆ did not suppress the asp1-STF6 growth defect on YES medium (Figure 5B).

Exogenous G3P does not elicit a growth defect in fission yeast. We had reported previously that adding choline (up to 4 mM) or glycerol-3-phosphate (up to 4 mM) in the medium has no effect on the growth of wild-type, asp1-STF6, or asp1-STF9 cells [11]. There is, however, a caveat to the G3P result, insofar as extracellular G3P will be hydrolyzed by the cell surface-associated and secreted acid phosphatase enzyme Pho1, which is strongly induced in asp1-STF6/9 cells (Figure 6A). To eliminate this potentially confounding issue, we queried whether deleting pho1 and pho4 (the two known cell surface acid phosphatases in S. pombe) would confer toxicity to extracellular G3P. We found that the pho1∆ pho4∆ asp1-STF6 and pho1∆ pho4∆ asp1-STF9 strains were impervious to added G3P up to 4 mM while remaining sensitive to inhibition by 125–500 µM GPC (Figure 6B). These findings suggest that: (i) induced Tgp1 per se is unable to transport G3P and (ii) fission yeast lack a dedicated transmembrane transporter for extracellular G3P.

Elevated GPC uptake does not impact asp1-STF cell viability as measured by colony-forming units. The preceding experiment did not distinguish whether the GPC toxicity to asp1-STF6/9 cells reflects slowed growth versus cell death. Thus, we asked whether chronic exposure to growth inhibitory concentrations of GPC affected fission yeast viability. S. pombe wild-type (WT), asp1-STF6, and asp1-STF9 cells were grown in liquid ePMGT medium containing 500 µM GPC, and aliquots were withdrawn at serial intervals and examined microscopically to determine cell counts. Nonlinear regression fitting of the data to an exponential growth equation revealed a doubling time of 2.3 h for wild-type cells versus 9.5 h for asp1-STF9 cells and 11.3 h for asp1-STF6 cells (Figure 7A). We gauged cell viability by plating equal numbers of cells taken from exponentially growing cultures sampled prior to (time 0) and after 12, 24, and 48 h exposure to GPC. The viable cell counts for each strain were normalized to the respective time 0 values (defined as 100% survival) and are plotted in Figure 7B. There was no significant difference in survival between wild-type and asp1-STF cells after 24 h of GPC treatment (74% to 84% survival rates) or after 48 h (WT 79% versus STF9 85%, p value 0.54; WT versus STF6 53%, p value 0.086). We conclude that GPC toxicity to the fission yeast asp1-STF mutants reflects a reduced rate of growth rather than an increased rate of cell death.

gde1 is not required for GPC toxicity in wild-type cells that overexpress tgp1 at high gene dosage. We reported previously that wild-type asp1^+^ cells and asp1∆ cells (lacking 1,5-IP_8_) are rendered sensitive to GPC toxicity when tgp1 is overexpressed from a high-copy plasmid under the control of the thiamine-repressible nmt1-41x promoter [11]. It was suggested that the high-copy overexpression leads to an even higher accumulation of intracellular GPC that elicits toxicity independent of inositol pyrophosphates. Consistent with this idea, we find that tgp1 mRNA levels (measured by RT-qPCR and normalized to act1 mRNA) are 230-fold higher in asp1^+^ wild-type cells bearing the nmt1-tgp1 plasmid vis-à-vis the empty vector control during growth in liquid ePMG medium lacking thiamine (Figure 8A). By contrast, tgp1 mRNA levels in asp1-STF6 and asp1-STF9 cells are only 41-fold and 35-fold higher than in wild-type cells, respectively (Figure 8A). These values are in excellent agreement with the 43 to 47-fold increases in tgp1 mRNA in asp1-STF6 and asp1-STF9 cells determined via poly(A)^+^ RNA-seq [11].

We assayed the survival of wild-type asp1^+^ fission yeast bearing the empty vector or nmt1-tgp1 plasmids prior to (time 0) and after incubation for 24 h in medium containing 500 µM GPC by plating equal numbers of cells (counted manually) on ePMGT agar medium lacking GPC and quantifying the colony counts. The 5% survival after high-copy nmt1-driven tgp1 expression was significantly lower than the 87% survival of control cells bearing the empty vector (Figure 8B). Thus, whereas growth is merely slowed in asp1-STF6/9 cells exposed to exogenous GPC, the much higher level of Tgp1 expression in nmt1-tgp1 cells elicits lethality.

We proceeded to ask whether this high-copy tgp1-driven GPC toxicity was dependent on Gde1. We found that in the presence of thiamine, when the nmt-41x promoter is repressed (tgp1 OFF state), wild-type gde1^+^ cells and gde1∆ cells were unaffected by carriage of the tgp1 expression plasmid when the medium was supplemented with 125 µM or 250 µM GPC (Figure 9). By contrast, in the absence of thiamine, when the nmt-41x promoter is activated (tgp1 ON state), inclusion of GPC in the medium severely inhibited the growth of gde1^+^ cells and gde1∆ cells carrying the tgp1 expression plasmid, without affecting the growth of cells bearing the empty plasmid vector (Figure 9). These results prompt the inference that G3P is not the sole “culprit” in GPC toxicity, i.e., the higher levels of Tgp1 expression attained at high gene dosage under nmt1 promoter control result in higher levels of GPC, vis-à-vis those in asp1-STF6/9 cells, that trigger toxicity in a pathway unrelated to Gde1 and G3P. Alternatively, it is conceivable that said higher levels of GPC activate a hypothetical Gde1-independent pathway of GPC conversion to G3P.

Exogenous GPC is toxic to S. cerevisiae strains with exacerbated GPC import. Our findings in S. pombe indicate that hyperactive import of extracellular GPC by Tgp1 results in a growth defect. To examine if unrestrained GPC transport is toxic to S. cerevisiae, wild-type budding yeast cells were transformed with a CEN URA3 plasmid expressing the Candida albicans high-affinity GPC transporter CaGit3 [7,8,9,10] under the control of the strong, constitutive TEF1 promoter. Growth of CaGit3-expressing budding yeast was inhibited progressively by inclusion of 10 and 25 µM GPC in the medium (Figure 10D), whereas the same concentrations of GPI had no effect (Supplemental Figure S4). GPC had no effects on wild-type cells bearing the empty plasmid vector (Figure 10A). The GPC toxicity to S. cerevisiae is apparent at lower doses (10 µM, 25 µM) than those required for toxicity in S. pombe, which may reflect differences in either the affinity or the relative expression levels of the transporters between the two organisms. Deletion of the S. cerevisiae GPC1 gene did not impact the GPC inhibition of the growth of Git3-expressing yeast (Figure 10E). However, deletion of the S. cerevisiae GDE1 gene rendered Git3-expressing cells more sensitive to inhibition by 10 µM GPC (Figure 10F). This increased sensitivity of gde1∆ compared to wild-type cells at 10 µM might be due to higher levels of intracellular GPC in the absence of the GPC catabolizing enzyme Gde1 [43,44]. At 25 µM GPC, however, both gpc1∆ and gde1∆ mutants are inhibited.

4. Discussion

Extracellular metabolites are both a source of key nutrients during periods of scarcity and can be potentially deleterious if allowed to accumulate to excess in the cell. The present study exemplifies this principle in fission yeast, where regulated expression of Tgp1, which we show here to be a specific transmembrane transporter of GPC, determines whether extracellular GPC is helpful or harmful. Tgp1 is strongly repressed under phosphate-replete growth conditions via lncRNA-mediated transcriptional interference with the tgp1 promoter. As shown here, this is reflected in the inability of phosphate-replete wild-type S. pombe cells to import radiolabeled GPC from the culture medium. Tgp1 expression is derepressed during acute phosphate starvation when synthesis of the interfering lncRNA is turned off. Tgp1 derepression in this context, coupled with depression of co-regulated genes encoding cell surface acid phosphatase Pho1, phosphate transporter Pho84, and extracellular 5′-nucleotidases Efn1/2 [12,16,26] allows for hydrolysis of extracellular phosphomonoesters and import of the resulting inorganic phosphate or, in the case of Tgp1, import of GPC that can serve as a source of phosphate after catabolism by intracellular enzymes. Any potential growth inhibition that might ensue from too much GPC uptake will be irrelevant in the context of phosphate starvation, insofar as phosphate-starved fission yeast stop growing after two cell divisions and enter a period of prolonged G0 quiescence [12].

Interdicting inositol pyrophosphate catabolism, as in the asp1-STF mutants, leads to elevated levels of inositol-1-pyrophosphates that trigger precocious transcription termination of the lncRNAs that repress tgp1 and pho1, resulting in their derepression under phosphate-replete conditions. As shown here, the increase in tgp1 mRNA in asp1-STF cells correlates with vigorous uptake of extracellular radiolabeled GPC and inhibition of growth, both of which are eliminated by tgp1 deletion. The yeast extract used to prepare standard YES growth medium apparently contains GPC in a quantity sufficient to cause inositol pyrophosphate toxicosis via Tgp1, which can be recapitulated in asp1-STF cells by adding yeast extract or pure GPC to a rich synthetic medium, ePMGT.

Here we reveal that Gde1, a glycerophosphodiesterase enzyme imputed to catabolize GPC to choline and G3P, abets GPC toxicosis in asp1-STF cells in a manner that depends on Gde1’s inositol pyrophosphate-binding SPX domain as well as its phosphodiesterase active site. Indeed, by tracking the fate of exogenous ^14^C-choline-GPC after cellular uptake, we demonstrate that its hydrolysis to choline depends on Gde1 and the predicted IP_8_ docking site on the SPX domain. To our knowledge, this is the first evidence that IP_8_ is an activator of this enzyme.

We find that GPC toxicity to fission yeast is a tunable phenomenon that correlates with the magnitude of Tgp1 induction and is not obligately connected to inositol pyrophosphate status. To wit, forced overexpression of Tgp1 at high gene dosage under the control of a strong inducible nmt1 promoter triggers lethal GPC toxicity in phosphate-replete wild-type cells, and this growth inhibition is not ameliorated by ablation of Gde1. Thus, G3P produced by Gde1 may be but one of several potential mediators of GPC toxicity that vary in impact according to how much GPC is imported. The findings that the growth defect of asp1-STF cells in YES medium was not impacted by eliminating phospholipid-synthesizing enzymes Gpc1 or Gep4 weigh against a model in which excess conversion of GPC to LPC or G3P to phosphatidylglycerol is causative of the growth defect.

Despite gaps in our understanding of the mechanism underlying GPC-induced growth inhibition in fission yeast, this study demonstrates that GPC toxicosis extends to the budding yeast S. cerevisiae. The need to regulate intracellular levels of potentially toxic metabolites such as GPC is addressed through diverse mechanisms. Focusing on transporter specificity and expression and metabolite conversion as key modulators of GPC toxicity, we compare strategies employed by three fungal species (Figure 11). In S. cerevisiae, the Git1 transporter exhibits low affinity for GPC, favoring GPI instead [6], rendering exogenous GPC non-toxic unless the high-affinity C. albicans Git3 transporter is heterologously expressed. In S. pombe, the Tgp1 transporter is GPC-specific, but toxicity only arises under conditions of elevated Tgp1 expression, such as in asp1-STF cells where growth is slowed, and in asp1^+^ nmt1-tgp1 cells where 230-fold overexpression elicits cell death. In contrast, GPC is not toxic to C. albicans, despite it having an elaborate system for glycerophosphodiester transport involving four Git1 homologs. Whereas CaGit1 is GPI-specific and CaGit2’s specificity remains uncharacterized, both CaGit3 and CaGit4 transport GPC, with CaGit3 representing the major activity [7,8]. It is noteworthy that GPC uptake in C. albicans is approximately 500-fold higher than in S. cerevisiae, but the organism does not display growth inhibition upon exogenous GPC addition. Our previous studies suggest this is due to rapid GPC catabolism that efficiently releases inorganic phosphate. Indeed, C. albicans cells grow equally well when provided with either GPC or inorganic phosphate as the sole phosphate source [7,8,9].

We investigated whether metabolic conversion might impact GPC toxicity. In S. cerevisiae and S. pombe strains with plasmid-driven excessive GPC transport in otherwise wild-type backgrounds, disruption of the proximal conversion steps mediated by Gde1 or Gpc1 did not affect toxicity. This suggests that GPC itself is the toxic species. GPC may act by directly or indirectly modulating a protein involved in cell growth. Indirect effects could include allosteric regulation of an enzyme producing a downstream metabolite or interaction with a non-enzymatic protein that influences proliferation. The growth-inhibitory effects of elevated intracellular GPC suggest it may play an unrecognized role in modulating cell growth under physiological conditions. Numerous lipid and lipid-derived metabolites are known to influence cellular processes, including proliferation—well-characterized examples include ceramides, sphingolipids, and fatty acids [45,46,47]. The inositol polyphosphates and pyrophosphates, which arise from polyphosphoinositide metabolism, represent another class of bioactive molecules. Notably, we demonstrate here that IP_8_ modulates S. pombe Gde1 activity in vivo through binding to its SPX domain.

Although studies on the regulatory roles of glycerophosphodiesters are limited, some evidence suggests they may possess modulatory properties. For instance, GPC and related compounds have been shown to inhibit lysosomal deacylating phospholipases A and B in Batten disease models [48]. Additionally, GPI has been reported to inhibit growth in S. cerevisiae when Git1 is constitutively expressed and retained at the cell surface in strains lacking the α-arrestins Aly1 and Aly2, which normally mediate Git1 trafficking to the vacuole [49]. Whether GPI and GPC share a common mechanism of toxicity in S. cerevisiae remains to be determined.

5. Conclusions and Future Directions

Cells experiencing high levels of GPC transport either through mutation or plasmid-driven transporter expression are growth inhibited. This suggests that GPC, a common intracellular metabolite found in all eukaryotic cells, may have growth modulatory properties under physiological conditions. Here we establish the foundational parameters by which GPC influences cell growth through transporter specificity and downstream metabolic pathways. Future work will build on these insights to address GPC’s regulatory roles. For example, lipidomic and metabolomic profiling under GPC stress could uncover altered metabolic states that clarify underlying mechanisms. Additionally, identifying suppressor mutations that mitigate GPC toxicity may reveal direct molecular targets of GPC.