Phosphatidylethanolamine Synthesis in the Parasite Mitochondrion Is Required for Efficient Growth but Dispensable for Survival of Toxoplasma gondii*

Background: Toxoplasma gondii is a common intracellular parasite of diverse host cells. Results: The parasite can produce phosphatidylethanolamine in its mitochondrion, endoplasmic reticulum, and parasitophorous vacuole, which together allow versatile lipid biogenesis. Conclusion: Multiple routes of lipid synthesis ensure parasite survival in discrete nutrient milieus. Significance: T. gondii offers an instructive model to study membrane biology of parasitic protists. Toxoplasma gondii is a highly prevalent obligate intracellular parasite of the phylum Apicomplexa, which also includes other parasites of clinical and/or veterinary importance, such as Plasmodium, Cryptosporidium, and Eimeria. Acute infection by Toxoplasma is hallmarked by rapid proliferation in its host cells and requires a significant synthesis of parasite membranes. Phosphatidylethanolamine (PtdEtn) is the second major phospholipid class in T. gondii. Here, we reveal that PtdEtn is produced in the parasite mitochondrion and parasitophorous vacuole by decarboxylation of phosphatidylserine (PtdSer) and in the endoplasmic reticulum by fusion of CDP-ethanolamine and diacylglycerol. PtdEtn in the mitochondrion is synthesized by a phosphatidylserine decarboxylase (TgPSD1mt) of the type I class. TgPSD1mt harbors a targeting peptide at its N terminus that is required for the mitochondrial localization but not for the catalytic activity. Ablation of TgPSD1mt expression caused up to 45% growth impairment in the parasite mutant. The PtdEtn content of the mutant was unaffected, however, suggesting the presence of compensatory mechanisms. Indeed, metabolic labeling revealed an increased usage of ethanolamine for PtdEtn synthesis by the mutant. Likewise, depletion of nutrients exacerbated the growth defect (∼56%), which was partially restored by ethanolamine. Besides, the survival and residual growth of the TgPSD1mt mutant in the nutrient-depleted medium also indicated additional routes of PtdEtn biogenesis, such as acquisition of host-derived lipid. Collectively, the work demonstrates a metabolic cooperativity between the parasite organelles, which ensures a sustained lipid synthesis, survival and growth of T. gondii in varying nutritional milieus.


Toxoplasma gondii is a highly prevalent obligate intracellular parasite of the phylum Apicomplexa, which also includes other parasites of clinical and/or veterinary importance, such as Plasmodium, Cryptosporidium, and Eimeria. Acute infection by Toxoplasma is hallmarked by rapid proliferation in its host cells and requires a significant synthesis of parasite membranes. Phosphatidylethanolamine (PtdEtn) is the second major phospholipid class in T. gondii.
Here, we reveal that PtdEtn is produced in the parasite mitochondrion and parasitophorous vacuole by decarboxylation of phosphatidylserine (PtdSer) and in the endoplasmic reticulum by fusion of CDP-ethanolamine and diacylglycerol. PtdEtn in the mitochondrion is synthesized by a phosphatidylserine decarboxylase (TgPSD1mt) of the type I class. TgPSD1mt harbors a targeting peptide at its N terminus that is required for the mitochondrial localization but not for the catalytic activity. Ablation of TgPSD1mt expression caused up to 45% growth impairment in the parasite mutant. The PtdEtn content of the mutant was unaffected, however, suggesting the presence of compensatory mechanisms. Indeed, metabolic labeling revealed an increased usage of ethanolamine for PtdEtn synthesis by the mutant. Likewise, depletion of nutrients exacerbated the growth defect (ϳ56%), which was partially restored by ethanolamine. Besides, the survival and residual growth of the TgPSD1mt mutant in the nutrient-depleted medium also indicated additional routes of PtdEtn biogenesis, such as acquisition of host-derived lipid. Collectively, the work demonstrates a metabolic cooperativity between the parasite organelles, which ensures a sustained lipid synthesis, survival and growth of T. gondii in varying nutritional milieus.
Toxoplasma gondii is an obligate intracellular parasite that causes cerebral and ocular toxoplasmosis in immunosuppressed individuals as well as congenital disease and abortions in animals and humans. It can infect and replicate in nearly all nucleated vertebrate cells and thus poses a major threat to global health. Besides that, the parasite serves as a model organism to understand the biology of parasitic and free living protozoa. Successful infection and transmission of T. gondii depends on the interconversion between its fast replicating tachyzoite and dormant bradyzoite stages. The lytic cycle begins with the infection of host cells by tachyzoites, which cause tissue necrosis and acute infection (1). Upon immune and physicochemical stress (pH, temperature, etc.), tachyzoites convert into bradyzoite-containing cysts (chronic infection) (2). Such cysts are commonly formed in the neural and muscular tissues and persist for the entire life of the infected host (2).
The promiscuous parasitism of T. gondii in assorted niches necessitates a discretionary access to host resources and fine tuning of the parasite's metabolism. The parasite resides in a parasitophorous vacuole, which provides a safe environment and interface for acquisition of nutrients. The rapid replication of intracellular tachyzoites and concomitant expansion of the parasitophorous vacuole are accompanied by significant membrane synthesis and nutritional import from the host cell (3). Similar to other eukaryotes, phospholipids account for a major fraction of parasite membranes, which include phosphatidylcholine (PtdCho), 2 PtdEtn, phosphatidylinositol, and PtdSer (4). T. gondii has the potential to synthesize its major phospholipids (4 -6) and harbors a relatively complete set of enzymes for de novo synthesis (4,5). Unlike PtdCho and PtdSer, for which the parasite harbors only one route, PtdEtn is made by decarboxylation of PtdSer and by the CDP-ethanolamine pathway (4). The physiological relevance of de novo phospholipid syntheses in T. gondii remains to be understood, however.
PtdEtn is one of the most abundant phospholipids in prokaryotes and eukaryotes, which contributes to the membrane integrity, membrane fusion/fission, protein stabilization, and autophagy events (7). It is enriched in the mitochondrial membrane and can be made by PtdSer decarboxylation and/or the CDP-ethanolamine pathway (8). PtdSer decarboxylase (PSD) enzymes are a unique subgroup of decarboxylases, which carry a pyruvoyl group at their active site (9). Bacteria and mammalian cells express a single PSD protein, whereas two to three distinct PSDs have been identified in yeast and plants, respectively (10 -12). Quite notably, T. gondii tachyzoites display a much higher (10-fold) PSD activity when compared with yeast and mammalian cells (4). We have shown that the PSD activity in T. gondii comprises membrane-bound and soluble fractions (13). The PSD attributed to the soluble activity is secreted/excreted into the parasitophorous vacuole (TgPSD1pv) via dense granules (13), whereas the other PSD has, so far, remained elusive. This study aimed to (a) characterize the membrane-associated PSD for its activity and subcellular location, (b) determine its relative importance for PtdEtn synthesis, and (c) examine the cooperativity of distinct routes of PtdEtn synthesis for membrane biogenesis, survival, and growth of T. gondii, if any.
Parasite Culture and Expression Constructs-Human foreskin fibroblasts (HFF; Cell Line Service) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), 2 mM glutamine, minimum essential medium (MEM) non-essential amino acids, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C and 5% CO 2 in a humidified incubator. Tachyzoites (⌬ku80-TaTi and ⌬ku80hxgprt Ϫ ) were maintained by serial passage in HFF monolayers at a multiplicity of infection (MOI) of 3. Parasite RNA was isolated using TRIzol and reverse-transcribed into first-strand cDNA. The open reading frames (ORFs) of TgPSD1mt, TgPSD1pv, and TgEPT1/2 were amplified from cDNA using PfuUltra II Fusion polymerase (Agilent Technologies). Primers used for PCRs are listed in Table 1.
Full-length TgPSD1mt or its truncated forms (residues 91-427 and 113-427) with a C-terminal HA-tag were ligated into the pTETO7SAG1-UPKO plasmid at NcoI/PacI restriction sites using T4 ligase and transformed into Escherichia coli XL-1b for molecular cloning and vector preparation. The use of the TETO7SAG1 promoter in the TaTi strain of T. gondii permits a tetracycline-regulatable expression of the genes (19). The ORFs of TgEPT1/2-HA were cloned into the pTgGRA2-UPKO plasmid using NsiI and PacI sites. The TgPSD1pv-HA construct in the pTgPSD1pv-UPKO vector was made in three sequential steps by cloning TgPSD1pv-5Ј-UTR, TgPSD1pv-HA ORF, and TgNTP3-3Ј-UTR. The 5Ј-UTR of TgPSD1pv (2 kb) along with the first 52 bp of the ORF was cloned using EcoRV and MscI in the UPKO vector. In the second step, the partial ORF (53-2904 bp) containing a C-terminal HA tag was cloned at the MscI and PacI restriction sites. Finally, the 3Ј-UTR of the TgNTP3 gene was inserted using PacI and NotI (primers in Table 1). Bacterial clones were confirmed by PCR and sequencing.
For 3Ј-insertional tagging (3ЈIT) of the TgPSD1mt gene with a C-terminal HA tag, its 1.3-kb crossover sequence fused to the epitope was amplified from tachyzoite gDNA and digested with XcmI and HpaI. It was cloned into XcmI/EcoRI-digested pTKO-HXGPRT vector, finally eliminating the EcoRI cloning site. The pTKO-HXGPRT-TgPSD1mt-3ЈIT construct was linearized by EcoRI present in the first half of the crossover sequence, transfected into the ⌬ku80-hxgprt Ϫ strain (17), and then selected for HXGPRT expression, as described above. The resulting strain expressed TgPSD1mt-HA under the control of its endogenous promoter and the TgGRA2-3ЈUTR.
Tachyzoite Growth and Replication Assays-The growth fitness of tachyzoites was examined by plaque assays, recapitulating successive rounds of lytic cycles (i.e. invasion, replication, and egression). Confluent HFF cells in 6-well plates were infected with 200 -400 parasites/well. The parasitized host cells were incubated in standard culture medium with 10% FCS for 7 days without perturbation (37°C, 5% CO 2 ). Cells were washed twice with PBS, fixed with ice-cold methanol (2 min), and stained with crystal violet (10 min). To set up nutrientdepleted assays, normal FCS was replaced by dialyzed (PAA) or by lipid-depleted FCS (Biowest), as indicated, at the time of infection. The plaques were imaged and quantified using the ImageJ suite (National Institutes of Health, Bethesda, MD). To determine parasite replication, confluent HFF cells cultured on coverslips in 24-well plates were infected with tachyzoites (MOI ϭ 1), incubated for 12-60 h, fixed, and then immunostained using anti-TgGap45 antibody (1:3000; rabbit). The mean numbers of parasites per vacuole were scored to compare the replication rates of the strains.
Yeast Complementation Assay-Yeast cells were grown in 2% yeast extract, 1% peptone, and 2% glucose or in synthetic uracil-free minimal medium (0.67% yeast nitrogen base; Difco) supplemented with appropriate amino acids and 2% glucose. The S. cerevisiae psd1⌬psd2⌬ mutants (BY23480 and PTY44) were transformed with the constructs expressing TgPSD1mt, TgPSD1mt⌬mtp, or ScPSD1. The TgPSD1mt and TgPSD1mt⌬mtp TgPSD1mt (91-427) and TgPSD1mt (113-427) cDNAs were cloned from parasite mRNA, and ScPSD1 ORF was cloned from yeast gDNA (primers in Table 1). The ORFs were ligated into pYES2.1 (Invitrogen) or pESC-Ura (Agilent Technologies) vectors expressing the URA3 gene for selecting yeast transformants in uracil-deficient medium. Both plasmids allow galactose-inducible expression of a protein of interest under the control of the GAL promoter of S. cerevisiae.
Transformation was performed using standard yeast protocols. Transgenic cells were grown in synthetic uracil-dropout minimal medium supplemented with 2 mM ethanolamine and 2% glucose at 30°C, and positive clones were identified by PCR and sequencing. For complementation assays, yeast cells were grown in liquid medium to an A 600 of ϳ0.1, and serial dilutions (1:5) were stamped or spotted on synthetic uracil-dropout plates containing either glucose, galactose, or lactate (2%) with or without ethanolamine (2 mM). Plates were incubated for 2-3 days at 30°C. For plasmid loss, yeast cells were serially propagated for 5 days under nonselective conditions (2% yeast extract, 1% peptone, and 2% glucose) and evaluated by replica plating on selective (uracil-free) and nonselective (uracil-replete) plates in synthetic minimal medium. They were also examined for their ability to grow with or without ethanolamine.
Subcellular Fractionation of S. cerevisiae-Yeast cells were grown to A 600 ϭ 0.4 -0.8 in synthetic uracil-dropout medium supplemented with 2 mM ethanolamine and 2% galactose. All subsequent steps were performed at 4°C. Cells were washed once with double-distilled H 2 O and suspended in 50 mM potassium phosphate buffer (pH 6.8) containing EDTA (3 mM), phenylmethylsulfonyl fluoride (PMSF) (0.5 mM), and sucrose (0.25 M). Cell extract was prepared by disrupting yeast cells with glass beads (0.5-mm diameter) using a bead beater (Biospec Products) and removing the beads and cell debris by centrifugation (2000 ϫ g, 5 min). Crude mitochondrial preparation was obtained by centrifuging cell extract (10,000 ϫ g, 10 min) and two washes with buffer. The resulting supernatant was used to obtain the microsomes and cytosolic fractions (100,000 ϫ g, 1 h). Yeast cells expressing ScPSD1 (positive control) (10) were included to ascertain the efficiency of fractionation.
PtdSer Decarboxylase Assay-The reaction samples were prepared in 50 mM potassium phosphate (pH 6.8) buffer containing 10 mM ␤-mercaptoethanol, 0.25 M sucrose, 0.5 mM PMSF, 1 mM EDTA, and standard protease inhibitors. PSD activity was determined by trapping 14 CO 2 released from Ptd[1-14 C]Ser on filter paper saturated with 2 M KOH (23). Ptd[1-14 C]Ser was made from L-[1-14 C]serine and dioleoyl CDP-diacylglycerol using PtdSer synthase. PtdSer synthase was purified from E. coli strain JA-200 harboring the plasmid pPS3155 as described previously (24). The assay was performed at 30°C in 16 ϫ 100-mm borosilicate tubes sealed with an airtight rubber septum holding the KOH-saturated filter paper. The 0.8-ml assay mixture contained 60 mM potassium phosphate (pH 6.8), 0.17 M sucrose, 0.35 mM PMSF, 2 mM EDTA, 0.5 mM 2-mercaptoethanol, 0.5 mM dioleoyl Ptd[U-14 C]Ser (0.1 Ci/mol), and 0.1% (w/v) Triton X-100. The reactions were started with 0.2 ml of yeast fractions and terminated after 1 h by the addition of 0.5 ml of 0.25 M H 2 SO 4 , introduced through the rubber septum using a hypodermic needle. The emitted 14 CO 2 was trapped for 1 h prior to recovering the filter paper for liquid scintillation spectrometry.
Metabolic Labeling of Extracellular Parasites-The purification of fresh host-free tachyzoites and metabolic labeling were performed as described before (4). Briefly, HFF cells were infected with tachyzoites (MOI ϭ 3), and parasites were collected 42 h postinfection. Samples were maintained on ice throughout the parasite isolation procedure. Infected cells were passed twice through 27-gauge needles to release the parasites, followed by removal of host cell debris by centrifugation ( The parasitized cells were incubated (48 h, 37°C, 5% CO 2 ), and the parasites were released using 27-gauge needles. Tachyzoites were washed twice with PBS, counted, and subjected to lipid extraction. Total radioactivity accumulated in parasite lipids was quantified to deduce the import of host phospholipids by the replicating tachyzoites.
Lipid Extraction and Measurements-Lipid extraction from T. gondii was performed using the Bligh and Dyer method (25). The chloroform phase containing lipids was either used for liquid scintillation spectrometry or dried under nitrogen gas stream and suspended in chloroform and methanol (9:1). Lipids were resolved by one-dimensional TLC on silica gel 60 plates in chloroform, methanol, and acetate (65:25:10, v/v/ v). To achieve a higher resolution, they were separated twice in chloroform, ethanol, water, and triethylamine (30:35: (26). Lipids were visualized by iodine or 8-anilino-1-naphthalenesulfonic acid staining and/or by autoradiography. All lipids were identified based on their co-migration with standards and quantified by liquid scintillation counting or by phosphorus assay (27).
Bioinformatics and Data Analyses-Initial identification of the parasite gene sequences was performed using the Toxoplasma genome database (ToxoDB) (28). TgPSD1mt ORF was experimentally annotated by PCR and sequencing, and submitted to the NCBI database (accession number DQ450198). Prediction of the mitochondrial targeting peptide was done using MitoprotII (29), and the CLC sequence viewer was used for the protein alignment. For phylogenetic analysis, protein sequences were aligned based on the PFAM hidden Markov models (PF02666) using hmmer3 (30), and sequences were trimmed using trimAl (31). The phylogenetic tree was built using PhyML (32) and the LG model for amino acid substitution. Data plotting and statistical analyses were performed using the GraphPad Prism suite. Assays were performed at least three times unless specified otherwise. The error bars show the S.E. The statistical analyses were done using Student's t test, analysis of variance, and Bonferroni's test (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001).

RESULTS
TgPSD1mt Is a Type I PtdSer Decarboxylase Localized in the Parasite Mitochondrion-Our earlier studies characterized a partially soluble PSD enzyme in T. gondii, which is secreted into the parasitophorous vacuole (TgPSD1pv) via dense granules (13). This work and subsequent database mining also indicated the presence of a second PSD gene in the parasite (ToxoDB; TGGT1_225550). We cloned its ORF (427 residues), which contains a PtdSer decarboxylase domain (184 -408 amino acids, Pfam domain PF02666) and a conserved LGST motif, required for the maturation and activity of PSD enzymes (Fig.  1A, dotted red box). In silico analysis of the protein sequence using MitoProtII predicted a high (98.5%) probability of import into the mitochondrion and a cleavage site between Ala-96 and Ser-97 (Fig. 1A, green box), which was confirmed by localization assays shown below. Hence, we named the protein TgPSD1mt (T. gondii PtdSer decarboxylase 1 mitochondrial).
Alignment of TgPSD1mt with P. falciparum and human orthologs revealed a fairly high conservation, particularly in the PSD domain (Fig. 1A). TgPSD1mt is 40% identical (62% similar) to PfPSD and 35% identical (51% similar) to HsPSD. The PSD proenzymes in prokaryotes as well as in eukaryotes are proteolytically processed between the Gly and Ser residues into a membrane-anchored ␤-subunit and a smaller ␣-subunit harboring the pyruvoyl moiety at its N terminus (33,34). The pyruvoyl prosthetic group is part of the active site and is needed for the catalytic function of PSD enzymes. A similar processing of TgPSD1mt is predicted to yield a mature protein with an ϳ32-kDa ␤-subunit and ϳ6-kDa ␣-subunit (see below; Fig. 5E). Phy-Phosphatidylethanolamine Biogenesis in T. gondii MARCH 7, 2014 • VOLUME 289 • NUMBER 10 logenetic clustering showed that TgPSD1mt belongs to the type I class PSD enzymes and is distinct from the previously described TgPSD1pv protein (Fig. 1B). Our database searches also identified only one PSD in most protozoan parasites (not shown) except for a small subgroup of coccidians (Eimeria and Neospora), which express two discrete PSD enzymes related to TgPSD1mt and TgPSD1pv (Fig. 1B).
To determine the subcellular location of TgPSD1mt in T. gondii, we generated stable transgenic parasites by 3Ј-HA tagging of the gene, which subsequently expressed TgPSD1mt-HA under the control of its native promoter ( Fig.  2A). Immunostaining of the intracellular parasites showed a mitochondrial localization of TgPSD1mt-HA (green), which co-localized with a known organelle marker (red; TgF1B) (15). Consistently, the enzyme was localized in the parasite mitochondrion when it was ectopically expressed under the control of a foreign promoter (see below; Fig. 5D). In contrast, TgPSD1pv, regulated by its endogenous promoter, was localized in the parasitophorous vacuole (Fig. 2B) (13). These data reveal two phylogenetically distinct PSD enzymes in T. gondii, one of which is secreted into the parasitophorous vacuole (TgPSD1pv), and a second one that is expressed in the parasite mitochondrion (TgPSD1mt).
TgPSD1mt Can Functionally Complement a psd1⌬psd2⌬ Yeast Mutant-To examine whether TgPSD1mt is a functional PSD enzyme, we expressed the parasite protein in an S. cerevisiae psd1⌬psd2⌬ mutant (Fig. 3). Ablation of the two native PSD proteins makes the yeast strains auxotrophic for ethanolamine, which is used to synthesize PtdEtn via the three-step CDP-ethanolamine pathway, of which the final reaction is located in the endoplasmic reticulum ( Fig. 3A) (35). We expressed TgPSD1mt cDNA under the regulation of the ScGAL10 promoter in the psd1⌬psd2⌬ yeast mutants. Empty vector and ScPSD1 expression plasmid served as negative and positive controls, respectively (Fig. 3B). As expected, all transgenic yeast strains were able to grow under ethanolamine-replete conditions. The empty vector (negative) control did not show any detectable growth in ethanolamine-free medium, whereas TgPSD1mt and ScPSD1 rescued growth of the mutant in a galactose-inducible manner. To validate the specificity of complementation, we allowed yeast cells to lose the plasmid in nonselective medium. Cells that lost the PSD expression construct also lost their ability to grow on uracil-dropout selective plates as well as in ethanolamine-deficient nonselective medium (not shown).
Next, we produced subcellular fractions of the transgenic yeast strains to determine the distribution of TgPSD1mt activity in the mitochondrial, microsomal, and cytosolic fractions. A psd2⌬ yeast strain (JSY9750), expressing only the native ScPSD1 in the mitochondria, was included to ascertain a suc-  MARCH 7, 2014 • VOLUME 289 • NUMBER 10 JOURNAL OF BIOLOGICAL CHEMISTRY 6815 cessful fractionation (Fig. 3C (10, 12)). Each fraction was used to perform PSD assays by trapping 14 CO 2 released by decarboxylation of radioactive PtdSer. As anticipated, the psd1⌬psd2⌬ mutant harboring the empty vector did not show any measurable activity, and overexpression of TgPSD1mt via the ScGAL10 promoter was quite evident when compared with the positive control. Similar to ScPSD1 (10, 12), a majority of the recovered TgPSD1mt activity (Ͼ80%) was expressed in the mitochondriaenriched fraction.

Phosphatidylethanolamine Biogenesis in T. gondii
The Mitochondrial Targeting Peptide in TgPSD1mt Is Required for Localization but Not for Activity-We next tested the importance of the predicted mitochondrial targeting peptide for localization and enzymatic activity. We expressed two HAtagged deletion variants of TgPSD1mt lacking the N-terminal peptide (TgPSD1mt (91-427) -HA and TgPSD1mt (113-427) -HA). None of the two truncated forms were localized in the parasite mitochondrion, and they displayed a diffuse fluorescence throughout the parasite instead (Fig. 4A). We also expressed TgPSD1mt  in the yeast mutant to test the role of the N-terminal sequence for the catalytic activity and subcellular distribution. Notably, the truncated PSD was catalytically active and functionally complemented yeast growth in ethanolamine-free cultures (Fig. 4B), although only ϳ17% of the TgPSD1mt (113-427) activity was present in the mitochondriaenriched fraction (Fig. 4C). These assays demonstrate that the N-terminal peptide is required for mitochondrial targeting but not for the catalytic function of TgPSD1mt.
TgPSD1mt Is Dispensable for the Parasite Survival-Our further work focused on examining the physiological significance of TgPSD1mt for T. gondii. Hence, we generated a tetracyclineregulatable conditional mutant in the ⌬ku80-TaTi tachyzoites (Fig. 5A). To this end, we first ectopically expressed a tetracycline-repressible TgPSD1mt-HA ORF (TgPSD1mt-HA r ) at the TgUPRT locus. Then the merodiploid strain expressing TgPSD1mt-HA r was subjected to the deletion of the TgPSD1mt locus by double homologous recombination. The mutant strain lacking the TgPSD1mt gene showed PCR bands specific for 5Јand 3Ј-crossovers (Fig. 5B). Subsequent DNA sequencing confirmed the successful recombination events. Finally, the ⌬tgpsd1mt/TgPSD1mt-HA r mutant showed a regulation of TgPSD1mt mRNA by anhydrotetracycline (Fig. 5C), verifying the conditional nature of the strain.
In agreement with the above data, TgPSD1mt-HA was localized in the mutant's mitochondrion and displayed a significant expression in the on state (no anhydrotetracycline) cultures (Fig. 5D). Its expression was markedly repressed by the drug within 1 day and disappeared after 3 days (off state). Immunoblot data confirmed a complete knockdown of TgPSD1mt-HA in the conditional strain treated with the drug for one (2 days) or two (4 days) passages (Fig. 5E). The presence of three distinct drug-repressible bands, probably corresponding to preproenzyme, proenzyme, and ␣-subunit, indicated proteolytic maturation of TgPSD1mt. The mutant could be cultured in the off state for a prolonged duration, which implied a nonessential nature of PSD. To further support the notion, we generated a knockout of T. gondii lacking the TgPSD1mt gene (Fig. 6A). The ⌬tgpsd1mt mutant could be maintained in culture for multiple generations, which confirmed the expendable nature of TgPSD1mt for the lytic cycle.
TgPSD1mt Is Required for an Optimal Parasite Growth and Replication-Following a successful genetic ablation of TgPSD1mt, we examined the overall growth of the two mutants in human fibroblasts by performing plaque assays (36). The parasites were allowed to grow in host-cell monolayers for multiple rounds of lytic cycles (Fig. 6B). The plaques were evaluated for their numbers and area to assess the relative growth fitness of the knock-out and conditional mutants (Figs. 6C and 7A and 7B). The ⌬tgpsd1mt strain exhibited an about 35% decrease in plaque size when compared with its parental strain (Fig. 6C). The ⌬tgpsd1mt/TgPSD1mt-HA r mutant displayed ϳ20% reduced plaque growth even in the absence of anhydrotetracycline (on state), which is probably due to a relatively weaker strength of the conditional promoter. A 7-day exposure to the drug accentuated the growth defect to 35% in the conditional mutant, whereas it had no apparent effect on the parental strain. Prolonged drug exposure for 10 days exerted a nearly 45% growth impairment in the conditional mutant (Fig. 7B). The plaque numbers for both mutants were similar to their respective parental strains (not shown). The average numbers of the parasites per vacuole during the course of infection confirmed that the observed growth defect is caused by an impaired replication of the mutant tachyzoites (Fig. 7C). These results reveal that TgPSD1mt is needed for an optimal growth and replication of T. gondii, although it is dispensable for the parasite survival.
The TgPSD1mt Mutant Up-regulates the CDP-ethanolamine Pathway-For downstream assays, we used the ⌬tgpsd1mt/ TgPSD1mt-HA r strain, which permitted a conditional testing (on/off states) and thus ensured the specificity of the stated observations. We first determined the phospholipid composition of parasites (Fig. 7D). Consistent with former results (4), phosphatidylcholine was the major lipid in the parental and mutant strains, followed by PtdEtn. Unexpectedly, no significant reduction in PtdEtn was apparent in the conditional mutant. The relative amounts of other major lipids were also unperturbed (Fig. 7D). These results prompted us to study the alternative routes of PtdEtn synthesis in the mutant. To discern a role of TgPSD1pv, we performed metabolic labeling of parasite lipids with radioactive serine (Fig. 8A), which is mainly incorporated into PtdSer, and into PtdEtn following decarboxylation of the nascent PtdSer (4). Compared with the parental strain, only a modest (but statistically not significant) decline in lipid decarboxylation was detected despite a shutdown of TgPSD1mt protein synthesis in the off state mutant (Fig. 5E). The data indicated a continual PtdEtn synthesis by decarboxylation of PtdSer in the mutant, which is probably due to TgPSD1pv present in the dense granules (13).
The parasite is also known to produce PtdEtn from ethanolamine via the three-step CDP-ethanolamine pathway (4); its subcellular location in T. gondii is not defined so far. We found two putative EPT proteins in the parasite (ToxoDB; TGGT1_261760 and TGGT1_257510), which could catalyze the third and final step of PtdEtn synthesis (the transfer of phosphoethanolamine to diacylglycerol). Both TgEPT1-HA and TgEPT2-HA proteins co-localized with a known ER marker, TgDer1-GFP (Fig. 8B) (18). To test whether the ERderived lipid can alleviate depletion of the mitochondrial PtdEtn, we executed labeling of parasite lipids with radioactive ethanolamine (Fig. 8, C and D). As expected, ethanolamine was primarily incorporated into PtdEtn and into phosphoethanolamine-ceramide, as deduced by their TLC migration with the lipid standards. Ethanolamine metabolism was not affected by anhydrotetracycline in the parental strain, whereas the mutant in its on state showed a modest increase in the nascent PtdEtn (Fig. 8C and D). The off state mutant, on the other hand, utilized about 36% more substrate into PtdEtn than its own on state and showed 60 -77% higher lipid labeling when compared with the parental strain (Fig. 8D). Collectively, these results demonstrate a sustained PtdSer decarboxylation as well as up-regulation of the CDPethanolamine route upon knockdown of mitochondrial PtdEtn synthesis.
The TgPSD1mt Mutant Can Tolerate a Depletion of Ethanolamine-To dissect the observed metabolic plasticity in T. gondii, we assessed whether the conditional mutant can survive simultaneous knockdown of TgPSD1mt and depletion of ethanolamine in culture (Fig. 9A). To this end, we set up plaque assays using dialyzed FCS. As anticipated, the TgPSD1mt mutant in the on state behaved similar to the parental strain, whereas in the off state, it exhibited an accentuated 56% growth defect that was partially restored by the addition of ethanolamine (Fig. 9B). A further inhibition of the mutant's growth, albeit modest, was observed when lipid-depleted serum was used in lieu of dialyzed FCS (Fig. 9C). The survival and resilient growth of the off state mutant in both serum types implied yet another route to drive PtdEtn biogenesis, potentially involving import of host-derived lipids. To test this, we first radioactively labeled the lipid pool in human fibroblasts and then allowed the conditional mutant to replicate and perform lipid biogenesis in prelabeled host cells (Fig. 9D). We measured a nearly 2-fold higher accumulation of radiolabeled lipid in the off state mutant. Although these data do not rule out a contribution of the CDP-ethanolamine pathway within the parasite, they are indicative of PtdEtn scavenging from the host cells when de novo synthesis by TgPSD1mt is shut down. Taken together, our results also reveal a surprising level of metabolic plasticity in the parasite with respect to its growth and lipid biogenesis under varying nutrient conditions.

DISCUSSION
Here we show that T. gondii produces PtdEtn by decarboxylation of PtdSer, which is known to be synthesized via a base exchange type PtdSer synthase (4). The parasite harbors two phylogenetically distinct PSD proteins to synthesize PtdEtn, FIGURE 5. Conditional mutagenesis allows a tetracycline-regulated knockdown of TgPSD1mt in T. gondii. A, scheme for generating the ⌬tgpsd1mt/ TgPSD1mt-HA r strain. A tetracycline-regulatable copy of TgPSD1mt-HA was inserted at the UPRT locus, and the TgPSD1mt gene was deleted using a knock-out plasmid (pTKO-TgPSD1mt-5ЈUTR-DHFR-TS-TgPSD1mt-3ЈUTR). The primer pairs to screen for 5Ј-and 3Ј-recombination are depicted in blue. B, PCR testing of the ⌬tgpsd1mt/TgPSD1mt-HA r mutant. Pyrimethamine-resistant parasite clones were screened by genomic PCR using the 5Ј-and 3Ј-crossover-specific primers (5ЈScr-F1/R1 and 3ЈScr-F1/R1). The parental gDNA was included as a negative control. C, PCR analysis to verify the regulation of TgPSD1mt mRNA by ATc. Total RNA isolated from the mutant or parental strain was used to generate cDNA and amplify TgPSD1mt using ORF-specific primers. D, immunostaining of the mutant showing ATc-regulated expression of TgPSD1mt-HA r . The untreated control and drug-treated (1 and 3 days) parasites were stained using anti-HA and anti-TgF1B antibodies (24-h infection). E, immunoblot image confirming the proteolytic processing and regulation of TgPSD1mt-HA r protein in the conditional mutant. ATc treatment was performed for 2 or 4 days in cultures, and fresh host-free parasites were subjected to protein isolation and immunoblot analyses. Note that pre-proenzyme, proenzyme, and ␣-subunit of TgPSD1mt-HA exhibit an aberrant migration in SDS-PAGE. A nonspecific band, which was not regulatable, was also observed in both strains. TgHsp90 (and nonspecific band) served as the loading control.
TgPSD1mt and TgPSD1pv, compartmentalized in the mitochondrion and parasitophorous vacuole, respectively. Most type I PSD enzymes localize in mitochondria except for PfPSD1, which is reported to be expressed in the parasite ER (37). Unlike Plasmodium but consistent with mammalian, plant, and yeast cells (38 -40), we demonstrate a typically conserved mitochondrial expression of TgPSD1mt. Plasmodium and Toxoplasma also harbor the ethanolamine-dependent route to make PtdEtn in the parasite ER (4,41). It therefore appears as if the synthesis of PtdEtn is confined to the endoplas-FIGURE 6. TgPSD1mt is nonessential for survival but is required for efficient parasite growth. A, scheme for generating the ⌬tgpsd1mt strain. The TgPSD1mt gene was deleted using a knock-out construct (pTKO-TgPSD1mt-5ЈUTR-HXGPRT-TgPSD1mt-3ЈUTR). The ⌬tgpsd1mt strain was confirmed by 5Ј and 3Ј PCR screening using the indicated primers (5ЈScr-F2/R2 and 3ЈScr-F2/R2). The parental gDNA was included as a negative control. B, schematized lytic cycle of T. gondii tachyzoites depicting the events of invasion, replication, and egression. Successive rounds of parasite cycles in confluent human fibroblasts over a 1-week period led to plaque formation. C, representative plaques formed by the ⌬tgpsd1mt mutant and parental (⌬ku80-hxgprt Ϫ ) strains (left) and quantification of their sizes (right). In total, 200 plaques of each strain from three assays were measured for their area using the ImageJ software (***, p Ͻ 0.001). Error bars, S.E. MARCH 7, 2014 • VOLUME 289 • NUMBER 10 mic reticulum in Plasmodium, whereas this lipid is produced at multiple sites in T. gondii (i.e. mitochondrion, ER, parasitophorous vacuole, and dense granules (this work) (13)). Such an autonomous and resourceful PtdEtn synthesis bestows a considerable metabolic plasticity to T. gondii in differing nutritional niches.

Phosphatidylethanolamine Biogenesis in T. gondii
The relative physiological significance of PtdSer decarboxylation and/or CDP-ethanolamine pathways varies among different organisms (8). For example, in the mouse, a deletion of mitochondrial PSD is embryonic lethal; however, ablation of ethanolamine-dependent lipid synthesis can be remunerated by PSD (42). The CDP-ethanolamine route is reported to be essential in protozoan parasites, Plasmodium and Trypanosome species (41,43,44). Conversely, Arabidopsis and Saccharomyces can tolerate a collective ablation of their PSD proteins or the CDP-ethanolamine pathway (11,40). Interestingly, both of these routes are not mutually exclusive in T. gondii, and a specific lipid transfer appears to exist between the parasite mitochondrion and ER. Such an interorganelle lipid transport is reported to occur between the ER and mitochondria and/or the Golgi network in yeast (Fig. 3A), plants, and mammalian cells (45). The existence of such a process remains to be established in T. gondii.
The fact that T. gondii survives a concurrent knockdown of the TgPSD1mt function and withdrawal of ethanolamine and lipids in culture implies additional routes for PtdEtn biogenesis, which involve TgPSD1pv activity and lipid trafficking from the host organelles. This notion is supported by our data showing a continual synthesis of PtdEtn from PtdSer despite a knockdown of mitochondrial PSD and accrual of ethanolamine-derived FIGURE 7. The ⌬tgpsd1mt/TgPSD1mt-HA r mutant shows an impaired growth and replication despite a normal phospholipid composition. A, representative plaques of the conditional mutant in the absence or presence of ATc (0.5 M). For treatment exceeding 7 days, parasites were precultured with ATc. B, relative plaque areas of the strains from A. 200 plaques of each strain from three assays were quantified (***, p Ͻ 0.001). C, replication curves of the parental and mutant strains. For ATc treatment, parasites were cultured for three passages in drug-containing medium and used to infect human fibroblasts. Parasitized cells were stained with anti-TgGap45 antibody at the indicated time points. The average parasite numbers per vacuole were estimated by scoring 150 vacuoles from three assays. D, phospholipid composition of the parental and mutant strains. Lipids were resolved by one-dimensional TLC (chloroform, ethanol, water, triethylamine (30:35:7:35, v/v/v/v)) and then quantified by phosphorus assay. Error bars, S.E. PtdCho, phosphatidylcholine; PEtn-Cer, phosphoethanolamine-ceramide.
host lipid(s) by intracellular parasites. It is known that the parasitophorous vacuole becomes associated with host mitochondria and ER (46), which are major sites of lipid synthesis in the mammalian cells (8,45). This contiguous association of host organelles with the vacuole corresponds to the aforementioned zones of privileged lipid exchange, where PtdEtn and/or PtdSer may traffic to the parasite. Host-derived PtdSer could also be used to produce PtdEtn by TgPSD1pv in the parasitophorous vacuole and dense granules. Akin to yeast and mammalian cells transporting PtdEtn/PtdSer across their membranes (47), the parasite genome harbors putative P4-ATPase permeases (ToxoDB), which may transport phospholipids across the vacuolar and plasma membranes in T. gondii. Finally, it was shown in Leishmania that degradation of sphingolipids by sphingosine 1-phosphate lyase can also contribute to PtdEtn synthesis (48). Our in silico analysis, however, did not reveal an ortholog in the Toxoplasma database.
It is an intriguing observation that TgPSD1mt is not essential for the parasite survival but is required for optimal growth. In mammalian cells, ablation of PSD is often associated with fragmented/deformed and dysfunctional mitochondria (42,49,50). A knockdown of TgPSD1mt, on the other hand, did not affect FIGURE 8. PtdEtn synthesis in the endoplasmic reticulum is up-regulated upon knockdown of mitochondrial PSD in T. gondii. A, labeling of parasite lipids with radioactive serine. Fresh extracellular parasites (5 ϫ 10 7 ) of the parental or mutant strain (untreated or treated with ATc for two passages) were incubated with [ 3 H]serine (1 Ci/nmol, 2 h, 37°C). Lipids were resolved by one-dimensional TLC (chloroform, methanol, acetate (65:25:10, v/v/v)) and subjected to liquid scintillation counting (n ϭ 3 parental; n ϭ 5 mutant). The differences between the conditions/strains are non-significant. B, immunolocalization of putative EPT enzymes (TGGT1_261760 and TGGT1_257510). The parasites were co-transfected with the vectors expressing TgEPT1-HA or TgEPT2-HA along with a TgDer1-GFP expression plasmid and stained with anti-HA antibody after 24 h of infection. C, labeling of parasite lipids with radioactive ethanolamine. The parental or mutant parasites (5 ϫ 10 7 ) were incubated with [ 14 C]ethanolamine (20 -40 nCi/nmol, 2 h, 37°C) followed by lipid isolation, TLC separation, and autoradiography. D, individual phospholipid bands from C were visualized by iodine vapor staining and scraped for scintillation counting (n ϭ 4 assays; **, p Ͻ 0.01; ***, p Ͻ 0.001). PEtn-Cer, phosphoethanolamine-ceramide. Error bars, S.E. the typical lariat shape morphology of the organelle in T. gondii (Fig. 5D). Likewise, the import of a nucleus-encoded F1B-ATPase protein to the organelle was also unaltered in the ⌬tgpsd1mt/TgPSD1mt-HA r strain (Fig. 5D). Finally, the mitochondrial membrane potential showed no difference between the on and off states, as judged by quantifications of DiOC 6 fluorescence and by mitotracker staining of the parasite organelle (not shown). It is nonetheless plausible that the parasite makes exclusive PtdEtn species in the mitochondrion not pro-duced otherwise, which may be needed for efficient parasite growth.
In conclusion, our work reveals that PtdEtn synthesis occurs at multiple distinct locations in T. gondii. The results suggest that synthesis of PtdEtn in this parasite parallels to the free living organisms, such as yeast and plant cells, and that PtdSer decarboxylation and the CDP-ethanolamine routes cooperate with each other. An interorganelle exchange of lipids ensures the parasite survival and growth in varying nutritional milieus. A, representative plaques formed by the parental or ⌬tgpsd1mt/TgPSD1mt-HA r strains in the media containing dialyzed serum with or without ethanolamine supplementation. The off state parasites were pretreated for two passages with ATc, and treatment was maintained during plaque cultures. B, relative plaque areas of the parental and mutant strains from A. In total, 250 plaques of each strain or condition from three assays were quantified. C, plaques formed by the conditional mutant in medium supplemented with either dialyzed or lipid-depleted serum. D, accumulation of radioactive lipids in the conditional mutant following its intracellular replication in human fibroblasts labeled for PtdEtn prior to infection. Isotope labeling of host cells was performed with [ 14 C]ethanolamine (40 nCi/nmol) in medium containing lipid-depleted serum (24 h, 37°C, 5% CO 2 ). Host cells were then infected with tachyzoites (MOI ϭ 2; 48 h), as described under "Materials and Methods." Lipids isolated from fresh parasites were analyzed by liquid scintillation counting to estimate the import of host-derived lipids (n ϭ 4 assays; *, p Ͻ 0.05; ***, p Ͻ 0.001). Error bars, S.E.
Last but not least, T. gondii with its established genetic manipulation tools offers an instructive model to explore the lipid biology of intracellular pathogens.