Apicoplast-Localized Lysophosphatidic Acid Precursor Assembly Is Required for Bulk Phospholipid Synthesis in Toxoplasma gondii and Relies on an Algal/Plant-Like Glycerol 3-Phosphate Acyltransferase

Most apicomplexan parasites possess a non-photosynthetic plastid (the apicoplast), which harbors enzymes for a number of metabolic pathways, including a prokaryotic type II fatty acid synthesis (FASII) pathway. In Toxoplasma gondii, the causative agent of toxoplasmosis, the FASII pathway is essential for parasite growth and infectivity. However, little is known about the fate of fatty acids synthesized by FASII. In this study, we have investigated the function of a plant-like glycerol 3-phosphate acyltransferase (TgATS1) that localizes to the T. gondii apicoplast. Knock-down of TgATS1 resulted in significantly reduced incorporation of FASII-synthesized fatty acids into phosphatidic acid and downstream phospholipids and a severe defect in intracellular parasite replication and survival. Lipidomic analysis demonstrated that lipid precursors are made in, and exported from, the apicoplast for de novo biosynthesis of bulk phospholipids. This study reveals that the apicoplast-located FASII and ATS1, which are primarily used to generate plastid galactolipids in plants and algae, instead generate bulk phospholipids for membrane biogenesis in T. gondii.


Introduction
Apicomplexan parasites include the etiological agents of a number of devastating human diseases, including malaria (Plasmodium spp.), toxoplasmosis (Toxoplasma gondii) and cryptosporidiosis (Cryptosporidium spp.). Most Apicomplexa harbor a plastid of prokaryotic origin, termed the apicoplast [1][2][3]. While the apicoplast lacks the enzymes involved in photosynthesis, this organelle contains many of the other metabolic pathways found in plant and algal plastids, including a prokaryotic type II fatty acid synthesis pathway (FASII) [4]. Since the apicoplast is essential for parasite survival, some of these pathways are considered attractive drug targets [5][6][7][8]. The discovery of the FASII pathway suggested that these parasites synthesise fatty acids (FA) de novo, rather than relying solely on supply from their host as was initially thought [4,9,10]. Subsequent genetic studies showed that components of the T. gondii FASII pathway (e.g. acyl carrier protein) are essential for the intracellular growth of the rapidly dividing tachyzoite stages [11,12]. FASII is also essential to parasite development in mosquito stages of P. falciparum [13] and liver stages of P. berghei and P. yoelii, the rodent malaria models [14,15].
Upon invasion of host cells, most apicomplexan parasites become surrounded by a membrane, termed the parasitophorous vacuolar membrane (PVM), which expands as the parasite develops and replicates. Both PVM expansion and parasite plasma/organelle membrane production during replication are reliant on recruitment of phospholipids (PLs) for membrane biogenesis. Indeed, PLs are the major lipid class found in T. gondii parasite membranes, accounting for up to 80-90% of the total lipid content [16,17]. PL synthesis is essential for parasite replication and enzymes involved in their assembly are promising drug targets [18][19][20]. In T. gondii, PL assembly is supported by both scavenged and de novo-synthesised FA, while mature PLs such as phosphatidylcholine (PC) may also be scavenged [21]. PC is the most abundant PL species in T. gondii and P. falciparum membranes [16,17]. PC synthesis is critical for T. gondii tachyzoite replication and Plasmodium blood and liver stages [22][23][24][25], while specific inhibitors of PC biosynthesis are currently in clinical trials as anti-malarial drugs [26,27]. Other important PLs include phosphatidylethanolamine (PE), accounting for up to 20% PL in these parasites [16,17], and phosphatidylinositol (PI), which is essential for apicoplast biogenesis, glycolipids, glycosylphosphatidylinositol anchors (GPI), membrane dynamics and integrity, and parasite survival [17,[28][29][30][31].

Disruption of TgATS1 causes defects in organellar and daughter cell development
To investigate the role of TgATS1 in parasite growth, we disrupted the endogenous copy of TgATS1 (eTgATS1) by insertion of a selectable marker (chloramphenicol acetyl transferase, CAT) into the TgATS1 ORF using a double recombination approach with a recombineering cosmid (S2B Fig, [52]). Four independent TgATS1-HA-iKO strains were generated: two bearing a TATi-inducible TgATS1-HA copy (TgATS1-HA-iKO), and two equivalent mutants in the TgATS1-HA was detected by Western blot as two distinct bands with apparent molecular masses of 72 kDa and 55 kDa (Fig 2A). These likely correspond to the pre-processed (pATS1, including the complete or partial N-terminal bipartite sequence) and mature (mATS1, apicoplast-resident) forms of the protein, respectively. Addition of ATc to the culture medium down-regulated TgATS1-HA-iKO expression, with pATS1 and mATS1 no longer detectable by Western blot after 3 and 4 days, respectively (Fig 2A). Plaque assays were performed to determine the effect of TgATS1 down-regulation on parasite growth. ATc treatment of parental lines produced no detectable growth defect (Fig 2B, top panels). ATc treatment of TgAT-S1-HA-iKO parasites resulted in almost complete ablation of host cell lysis and plaque formation ( Fig 2B, bottom right panel), although growth assays after 8 days of TgATS1 repression in a TgATS1-HA-iKO strain expressing cytosolic tdTomato indicated that some parasites may continue to grow, albeit slowly (S3 Fig).
To discern the cause of this growth arrest, we examined mutant parasites by IFA. Strikingly, we observed partial loss of apicoplast-localized FNR-RFP after 4 days of ATc treatment, with  [68]) and C. moschata (CmATS1, [47]). Residues strictly conserved between all species are highlighted in green, residues conserved in at least three species in cyan, residues conserved in apicomplexan sequences in yellow, and residues conserved between TgATS1 and plant ATS1 (AtATS1 and CmATS1) in blue. Brown triangles and grey ovals represent residues putatively involved in G3P or FA binding, respectively. Secondary structures [47,67] are represented above the sequence alignment by blue cylinders for α-helices and orange arrows for β-strands. Domain 1 of the protein (4-helix bundle) is symbolized with dashed lines and Domain 2 (α-β Domain) in solid lines. Residues putatively involved in binding the G3P substrate in CmATS1 (brown triangles) are strictly conserved in TgATS1, while those putatively involved in binding the acyl-ACP substrate are highly conserved (grey ovals). (B) Overlay of the CmATS1 crystal structure ( [47]) and the predicted TgATS1 3D structure. The overall structure and surface accessibility of CmATS1 (grey) and TgATS1 (green and magenta) is conserved and highly similar as observed in the ribbon representation. (C) Residues putatively involved in binding substrate (G3P) and those involved in the catalytic motif NHX 4 D of CmATS1 form a catalytic pocket with His-167 and Asp-172. (D) Both the motif and topology of the pocket are strictly conserved in TgATS1 (His-574 and Asp-579). (E, F) IFA shows that TgATS1 is a stromal-resident protein of the apicoplast, as confirmed by co-localization with fluorescence of (E) the chimeric apicoplast stromal FNR-RFP reporter protein co-expressed in the TgATS1-iKO parasite line and (F) anti-CPN60, a known marker of the apicoplast stroma [51]. Scale bars represent 2 μm. some parasites displaying no signal at all (Fig 2C). IFA against the apicoplast stromal protein, CPN60, revealed that loss of TgATS1 resulted in redistribution of CPN60 from a typically punctate apicoplast staining to a diffuse cytoplasmic staining (Fig 2C). This was also observed using antibodies against the apicoplast outer membrane marker ATRx1 (Fig 2D), suggesting a TgATS1 was detected by Western blot analysis using anti-HA antibody as two bands: the pre-mature form (pATS1,~75KDa) and the mature form (mATS1,~55KDa). Protein expression was down-regulated to undetectable levels by 3 or 4 days of ATc treatment (1 mg/mL ATc, numbers indicate days of culture). GRA1 (lower panel) served as a loading control. (B) Plaque assays performed in the absence (-) or presence (+) of ATc and fixed after 10 days show an impaired lytic cycle of TgATS1-HA-iKO parasites in the presence of ATc. (C) IFA of TgATS1-HA-iKO parasites using anti-HA antibody and apicoplast stromal markers FNR-RFP (two upper panels) and CPN60 antibody (lower panel) indicates loss of HA and apicoplast signals in the presence of ATc for 4 days (white arrows indicate normal apicoplast while red arrows indicate loss of apicoplast signal), as well as the cytosolic mis-localisation of apicoplast CPN60 (lower panel, zoomed area). (D) IFA of TgATS1-HA-iKO parasites using antibodies against the apicoplast stromal marker CPN60 and the apicoplast outer membrane marker ATRx1 confirms mis-localisation of CPN60 and ATRx1, indicating loss of apicoplast structure (lower panels, zoomed areas). White arrows indicate normal apicoplast and red arrows indicate a normal apicoplast. (E) Quantification of the number of intact apicoplasts relative to parasites and vacuoles in TgATS1-HA-iKO parasites following ATc treatment. A significant loss of apicoplasts was observed in the presence of ATc (upper graph) of up to 60% at days 5 and 6, compared to the wild type (parental) strain that contained 100% apicoplasts in all vacuoles regardless of the number of parasites per vacuole (lower panel). (n = 100 vacuoles). (F) IFA of TgATS1-HA-iKO parasites using anti-IMC1 antibody grown in the presence and absence of ATc indicates IMC structure defect (zoomed areas). Scale bars: 2.5 μm.
doi:10.1371/journal.ppat.1005765.g002 ATS1 Is Essential for T. gondii Lipid Synthesis loss of apicoplast structure. To investigate this further we quantified the number of parasites containing an intact apicoplast over the course of 6 days of ATc treatment ( Fig 2E). Although some parasites lost their apicoplast within the first day, the greatest loss appeared after 5-6 days of ATc treatment. Inner membrane complex (IMC) formation (as observed with anti-IMC antibodies) was also greatly affected and, together with apicoplast abrogation, was associated with changes in parasite morphology (Fig 2F). These observations suggested that TgATS1 is required for organelle formation and parasite division.
Transmission electron microscopy of TgATS1-iKO parasites revealed further defects in intracellular parasite morphology (Fig 3). Untreated parasites morphology was unperturbed (Fig 3A), containing apicoplasts with the canonical four membranes (Fig 3B). However, after 3 days of ATc treatment, a number of ultrastructural changes were observed including (i) the accumulation of large electron-lucent vesicles and abnormal intracellular compartments ( Fig  3C); (ii) the appearance of apicoplasts with grossly malformed membranes that had detached from each other ( Fig 3D); (iii) the complete loss of apicoplasts ( Fig 3E); (iv) and the accumulation of multiple vacuoles harbouring filamentous material (Fig 3F). Taken together, these studies show that loss of TgATS1 expression leads to marked changes in apicoplast morphology and defects in the biogenesis of this organelle and parasite development. In the presence of ATc, intracellular development of TgATS-iKO parasites was drastically affected, resulting in parasites bearing aberrant organelles shown by red arrows and large electron lucent regions shown by blue arrows (C). Apicoplast biogenesis was also affected in the presence of ATc, with only a few parasites bearing an apicoplast and, of those present, morphological aberrations were observed, including disorganized membranes and atypical stroma (D). Intracellular division also seemed affected upon TgATS1 disruption (E). Parasites often displayed large electron-lucent vesicles containing an unusual ribbon-like material shown by blue arrows (E, F). Scale bars are indicated in each figure.
TgATS1 disruption reduces FASII C14:0 generation and FA elongation To determine the role of TgATS1 in lipid biosynthesis, ATc-treated and untreated TgATS1-iKO parasites were labelled with U-13 C-glucose (i.e. glucose where all carbons are 13 C) and the de novo synthesis of fatty acids assessed by mass isotopomer distribution (MID) analysis of total parasite fatty acids by gas chromatography-mass spectrometry (GC-MS) analysis. MID involves quantification of the molecular ions of individual unlabelled fatty acid methyl esters and corresponding isotopomers generated by incorporation of 13 C-skeletons derived from glucose, after subtraction of natural abundance isotopomers. For example, the MID of the palmitoyl methyl ester (m/z 270) containing a backbone of 16 carbons involves quantification of all ions from m/z 270 (indicated as M0) to m/z 286 (indicated as M16). We have previously used this approach to distinguish the rate of FA synthesis via the plastid FASII and ER-based elongation pathways through sequential condensation of two-carbon units from 13 C-malonyl-CoA onto the growing FA chain (Fig 4, [8,12]).
Intracellular tachyzoites were labelled with U-13 C-glucose for 1-4 days in the presence or absence of ATc, and total parasite FA was analysed by GC-MS. As expected, a wide range of FA were labelled in the absence of ATc (Fig 5A and 5B, S4A Fig) [12,17,53,54]. Interestingly, incorporation was highest after 2 days and decreased by day 4, suggesting that parasites may switch from de novo synthesis to increased salvage of host fatty acids at later time points ( Fig  5B). Inspection of MIDs revealed uniform labelling of some of the saturated FA, C14:0, C16:0, and unsaturated FA C16:1, confirming that labelling was due to (apicoplast located) de novo synthesis rather than (ER-located) elongation reactions (Fig 5C and 5D, S4B Fig and S5 Fig).
Labelling of parental cell lines with U-13 C-glucose (or U-13 C-acetate, see below) was unaffected by addition of ATc, while host cell lipids were not labelled in either condition [12]. Addition of ATc to the TgATS1-iKO did not significantly affect 13 C-incorporation into C14:0, C16:0, C16:1, and C18:1 after 1 day (Day 1, S4A Fig), but resulted in a significant decrease of 13 C-incorporation into C14:0 by day 2 (Day 2, Fig 5A), and C14:0, C16:0, C16:1, and C18:1 by day 4 (Fig 5B). Importantly, the fraction of C14:0 and C16:0 molecules that were labelled after 4 days of TgATS1 repression incorporated 13 C up to the full complement of 14 or 16 atoms, indicating that FASII was still active (Fig 5D, S5 Fig). Interestingly, while cellular abundance of all fatty acids was slightly (yet significantly for five of the 11 fatty acids) reduced after 2 days of TgATS1 repression (Fig 5E), a different pattern was observed after 4 days ( Fig 5F). Here, cellular abundance of C14:0 was greatly and significantly reduced, with smaller reductions observed for C16:0, C16:1, C18:1, and C20:1 (significantly for the latter two). Conversely, longer chain saturated fatty acids increased in abundance (significantly for C20:0) ( Fig 5F). Taken together, this indicates that TgATS1 disruption results in reduced flux through the FASII pathway, which might be partly compensated by increased fatty acid recycling/remodelling after 4 days of TgATS1 repression.
To investigate the effect of TgATS1 disruption on FA elongation, intracellular T. gondii tachyzoites were labelled with U-13 C-acetate and harvested and analysed as above (see Fig 4 for labelling strategy schematic). U-13 C-acetate, through conversion to 13 C-acetyl-CoA and then 13 C-malonyl-CoA, is efficiently used by the ER-localized elongase machinery, but not apicoplast FASII, providing a specific measure of elongation [12,53,54]. In the absence of ATc, high levels of 13 C incorporation were observed in the longer chain (>C16) FA, with incorporation increasing with chain length (Fig 5G), as observed previously [53]. MID analysis suggested that 13 C-acetate was incorporated onto unlabelled C14:0 or C16:0 to produce longer chain fatty acids (as observed by isotopologues sequentially increasing by 2 mass units for all FA 16 carbons and longer; Fig 5H and 5I, S6 Fig). After 4 days of TgATS1 repression by ATc, 13 C incorporation was significantly reduced for some FA (C16:1, C18:0, C18:1, C20:0, C20:1, and C22:0), although not all ( Fig 5G). Interestingly, despite reduced abundance (Fig 5F), 13 C incorporation significantly increased in C14:0 and C16:0. Taken together, these analyses suggest that loss of TgATS1 results in a selective defect in the synthesis of saturated long chain fatty acids in the apicoplast, but that this can be partially compensated for by increased elongation of fatty acids scavenged from the host cell.  13 C-carbon source labelling strategies to assess the role of FASII and TgATS1 in T. gondii membrane biogenesis. TgATS1-iKO parasites were grown in the presence of 12 C-glucose or U-13 C-glucose in the presence or absence of ATc. Glucose is metabolized via the parasite glycolytic pathway into dihydroxyacetone-phosphate (DHAP) and phosphoenolpyruvate (PEP). These precursors are imported into the apicoplast via the apicoplast phosphate transporter (APT). PEP can then be transformed into acetyl-CoA via the action of the apicoplast pyruvate kinase (PK) and the pyruvate dehydrogenase complex (PDH). Acetyl-CoA serves as a substrate for the FASII pathway to generate growing acyl chains on an acyl carrier protein (ACP) scaffold. Glycerol 3-phosphate (G3P) can be generated from imported DHAP by the apicoplast glycerol 3-phosphate dehydrogenase (Gpda), or potentially by an as yet unidentified cytosolic G3P dehydrogenase (G3PDH). Both labelled and/or unlabelled G3P and FA can be used as a substrate by the apicoplast TgATS1 (glycerol 3-phosphate acyltransferase) to form lysophosphatidic acid (LPA). These LPA species can potentially be exported from the apicoplast towards the ER to be assembled into phospholipids (PL) and/or have its FA elongated via the three ER elongases (ELO). In order to assess ER elongation, TgATS1-iKO parasites were grown in the presence or absence of ATc together with 12 C-acetate or U-13 C-acetate, which is metabolized to cytosolic acetyl-CoA by acetyl-CoA synthetase (ACS). Collectively, this strategy enabled the determination of (i) apicoplast FASIIgenerated FA products, (ii) products assembled via TgATS1, and (iii) PL species generated via the apicoplast. Total lipid extracts from TgATS1-iKO parasites grown in the described conditions were analyzed by two mass spectrometry techniques. GC-MS was used for the measurement of total FA profiles and LC-MS/MS was used to determine the relative amounts of 13 C incorporation into individual PL molecular species. In the cartoon, unlabelled and labelled moieties are shown in black and red, respectively. The apicoplast is responsible for biosynthesis of bulk phospholipid classes, most of which are generated via TgATS1 We hypothesised that TgATS1 may be required for incorporation of plastid-synthesised FA into LPA, an obligate intermediate of PA de novo synthesis (Fig 4). PA is the essential precursor for all de novo-synthesised PLs, which are bulk components of all cellular membranes, including those of the apicoplast. To assess the role of TgATS1 on PL biosynthesis, lipids from TgATS1-iKO parasites grown in the presence or absence of ATc were extracted and analyzed by liquid chromatography-mass spectrometry (LC-MS/MS) for the three major PL classes: PC, PI and PE. Precursor ion scanning was used for PC and PI (precursors of m/z 184 in positive mode and m/z 241 in negative mode, respectively) and neutral loss scanning was used for PE (neutral loss of 141 u in positive mode).
In the absence of ATc, LC-MS/MS revealed a PL repertoire in untreated (and parental) parasites ( Fig 6A, 6B and 6C) similar to previous reports [16]. Our analyses detected 27 PC molecular species (Fig 6A), eight PI molecular species (Fig 6B), and 16 PE molecular species (Fig 6C), each class containing FA moieties of various lengths and degrees of saturation.
U-13 C-glucose labelling can be used to assess de novo synthesis of LPA and PA by observation of label incorporation into FA (as explained above) and the glycerol 3-phosphate (G3P) backbone to which the FA moieties are attached (Fig 4 shows possible combinations of label incorporation into LPA and PA). U-13 C-glucose incorporation into the glycerol and acyl moieties of each PL class was measured using PL-specific scanning, as above. Since choline and ethanolamine are not synthesised from glucose, and T. gondii lacks the Ino1 enzyme required for de novo myo-inositol synthesis, no PLs were observed with labelled head groups [31]. Substantial incorporation of 13 Catoms into G3P and FA was observed in PC, PI and PE (Fig 6D, 6E and 6F, respectively). The mass spectra of individual PL species contained +3, +15, or 17 isotopomers (Fig 6D, 6E and 6F), corresponding to labelling of (i) only the G3P backbone (i.e. where +3 corresponds to incorporation of 3 × 13 C atoms into the glycerol of PL) or (ii) the G3P backbone and an acyl chain (i.e. where +15 corresponds to incorporation of 3 × 13 C atoms into glycerol and 12 × 13 C atoms into an acyl chain, while +17 corresponds to incorporation of 3 × 13 C atoms into glycerol and 14 × Due to the complexity of the spectra obtained in these analyses, it was difficult to confirm the full identities of all 13 C-labelled lipids. We used a novel high-resolution chromatographic Analysis of FASII fatty acid biosynthesis and elongation in TgATS1-iKO parasites by metabolic labelling using stable isotope precursors. Tachyzoites of TgATS1-iKO parasites were labelled with U-13 Cglucose for 2 or 4 days or U-13 C-acetate for 4 days in the presence or absence of ATc. Lipids were extracted, derivatized to form fatty acid methyl esters (FAMEs), and analysed by GC-MS to determine 13 C incorporation. (A, B) 13 C incorporation into fatty acids from U-13 C-glucose in the absence (dark colour) or presence (light colour) of ATc for 2 (A) and 4 days (B). (C, D) Mass isotopologue distributions (MID) of FA (C14:0) from U-13 C-glucose in the absence or presence of ATc for 2 (C) and 4 days (D) (colour scheme as above). The xaxis indicates the number of 13 C atoms in each FAME, where 'm0' indicates the monoisotopic mass containing no 13 C atoms, while 'mX' represents that mass with 'X' 13 C atoms incorporated). MIDs for the all detected FAMEs using U- 13  This improved chromatography ensured that mass envelopes for each isotopologue series did not overlap, confirming that major PL species contained one or two labelled fatty acid moieties (an example of an MS/MS analysis for PC(30:0) is shown in S9  Fig). In all cases, the LPA moiety of the PL was labelled, providing direct evidence that apicoplast-synthesized FA can be assembled into LPA, thereby contributing to bulk PL composition.
Upon ATc repression of TgATS1 expression for four days, the abundance of individual ( Fig  6G) and total ( Fig 6H) PC and individual ( Fig 6I) and total ( Fig 6J) PI was reduced by up to 60%. Inspection of mass spectra revealed that TgATS1 repression led to a reduction in 13 C label incorporation in both PC ( Fig 6D) and PI (Fig 6E), as observed by the reduced levels of +15, +17 13 C-atoms in most PL isotopologues (compare red (-ATc) and green (+ATc) spectra), indicative of reduced LPA assembly and subsequent PL synthesis. These analyses suggest that TgATS1 assembles apicoplast-generated FA into the LPA precursor for PC and PI synthesis. Interestingly, while the abundance of individual ( Fig 6K) and total ( Fig 6L) PE was also reduced (by up to 40%) upon TgATS1 repression, the effect on 13 C incorporation was minimal (Fig 6F).
Neutral loss scanning for PE revealed the presence of an ion of m/z 661.53, which did not correspond to any predicted PE mass (Fig 6C), but was identified as PE-ceramide (d18:1/16:0) by MS/MS analysis (S10A Fig). The C16:0 moiety of this PE-ceramide contained up to 14 or even 16 13 C-atoms (S10B Fig), suggesting that ceramide is also synthesized using de novo-synthesized fatty acids from the apicoplast. The abundance of this lipid was not affected after 4 days of TgATS1 repression (S10C Fig) and, although label incorporation was affected under these conditions, fully labelled FA was still present (S10B Fig), suggesting both that FASII was still active when TgATS1 is disrupted and that TgATS1 was not significantly responsible for PE-ceramide precursor production.
To further investigate this apparent specificity in TgATS1 contribution to PL assembly, we quantified the relative amount of labelled (i.e. apicoplast-synthesised) and unlabelled (scavenged/remodelled) FA in PC, PI, and PE. Since the G3P backbone of all PLs may be generated by either a plastid or cytosolic G3P dehydrogenase (TGGT1_210260 or TGGT1_307570, respectively), we assumed that all species containing only 3 13 C-atoms were derived from a U-13 C-G3P esterified to unlabelled (i.e. not apicoplast origin) FA, while all species that contained more than three 13 C-atoms indicated a U-13 C-G3P esterified to at least one labelled (i.e. apicoplast origin) FA. In untreated (and parental) parasites, the total amount of apicoplastgenerated (i.e. labelled) FA was~70%, 74% and 42% for PC, PI and PE, respectively (Fig 7A,  7B and 7C, respectively), where PLs with shorter FA moieties contained proportionally more apicoplast-generated FA than PLs with longer FA (Fig 7D (PC), 7E (PI) and 7F (PE)), correlating with our quantitative labelling analysis (Fig 6G, 6I and 6K, respectively). Specifically, PC species 38 and PI species 36 carbons were assembled from >50% apicoplast-derived FA (Fig 7D and 7E, respectively), while this was true for PE species of only 34 carbons in length ( Fig 7F).
Precursor ion scans of m/z 241 for PI species also led to the identification of molecular species that were eventually identified as potential minor PE species. Fragmentation of these PE precursor ions indeed led to generation of m/z 241 in negative mode (S11 Fig). The analysis was also extended to the other PL species, namelyphosphatidylglycine (PG), phosphatidylserine (PS), and the newly discovered phosphatidylthreonine (PT) [55]. PG and PS were beneath the limit of quantification. The identification of two ions of m/z corresponding to published PT species (m/z 850.56, PT(40:5); m/z 878.59 PT(42:5)) was attempted. However, MS/MS fragmentation could not confirm the identity of these ions as being PT species TgATS1 is responsible for the synthesis of LPA (C14:0) Since the putative product of TgATS1 is predicted to be LPA, the direct precursor to PA, label incorporation into LPA and PA was investigated. Despite our improved chromatographic resolution, we did not detect free LPA and PA by LC-MS. Instead, we purified PL species by two dimensional high performance thin layer chromatography (2D-HPTLC) and analysed individual species by GC-MS. While LPA remained undetectable, PA was observed (albeit in low amounts) and contained C14:0, C16:0, C16:1, C18:0, and C18:1 FA moieties (Fig 8A). After 2 days of TgATS1 repression, 13 C incorporation into PA(C14:0) was reduced significantly (Fig  8A), while incorporation into longer FAs (C16:0, C18:0) was reduced by day 4, albeit insignificantly (Fig 8B). Analysis of the PC spot (Fig 8C, day 2; 8D, day 4), confirmed the LC-MS results that 13 C incorporation was significantly affected in PC species containing C14:0 FA. Importantly, the MID of PA(C14:0) in TgATS1-repressed parasites was very similar in pattern to, but of lower level than, the unrepressed control (Fig 8E, day 2; 8F, day 4), suggesting that (L)PA assembly is perturbed in these cells rather than FASII. The equivalent C14:0 MID in PC (Fig 8G, day 2; 8H, day 4), along with the relatively minor effects in label incorporation into PE and PE-ceramide (Fig 6F, S10B Fig), confirmed these results.
Although PA is downstream of the TgATS1 product LPA, we did not observe a significant reduction of total PA after 4 days of TgATS1 repression (Fig 8I). However, quantification of the fatty acid repertoire of PA revealed a significant reduction of LPA(C14:0) in the presence of ATc (Fig 8J and 8K), suggesting that this lipid may be a major product of TgATS1. To investigate whether LPA(C14:0) is important for parasite survival, we investigated whether exogenous LPA could rescue cell growth in plaque assays. Infected fibroblasts were treated with/without ATc, in the presence or absence of LPA(C14:0) or LPA(C16:0) in the culture medium ( Fig 8L). Strikingly, addition of LPA partially restored parasite growth, an effect that was only observed when using LPA(C14:0) (Fig 8L, lower left panel), and not LPA(C16:0) (Fig 8L, lower right  panel). These data confirm that TgATS1 has a specific role in generating LPA(C14:0) using apicoplast-derived fatty acids and that this species is critic.al for other PL biosynthetic pathways. U-13 C-glucose incorporation to fatty acids determines the apicoplast generates the FA moieties for production of most PC and PI molecular species. U-13 C-glucose was incorporated into ATc-untreated parasites grown in glucose-free medium. LC-MS/MS analysis of labelled molecular species that have incorporated 4 or more 13 C-atoms allowed quantification of apicoplast-generated species as shown in white, while all other species (mass shift +3) allowed quantification of non-apicoplast-generated species as shown in black. (A, B, C) Relative abundance for total PC (A), PI (B) and PE (C). (D, E, F) Relative abundance of individual molecular species for PC (D), PI (E), and PE (F). The data shows that the apicoplast generates the FA moieties for production of~70% PC,~72% PI, and~42% PE molecular species. Error bars indicate standard deviation, (n = 3 biological replicates). doi:10.1371/journal.ppat.1005765.g007

Discussion
Recent studies have shown that FA synthesized in the T. gondii apicoplast by the FASII enzyme complex can be trafficked to the parasite ER where they are further elongated and desaturated [12,53,54]. However, it remained unclear to what extent apicoplast-generated FA are incorporated into membrane phospholipids and why de novo FA biosynthesis cannot be bypassed by FA scavenging pathways.
Here we demonstrate that the apicoplast of T. gondii harbours a glycerol 3-phosphate acyltransferase, TgATS1, that is orthologous to the plant chloroplast ATS1 responsible for the initiation of the prokaryotic PA synthesis pathway. We show that TgATS1 is responsible for the recruitment of FASII-generated FA (most notably C14:0) into bulk synthesis of major phospholipid classes (PC, PE and PI) found throughout the parasite membranes and that this is critical for normal parasite growth and development, as also observed in P. yoelii liver stages [44]. A similar phenotype was observed in plants, although the loss of ATS1 function may be partially compensated by a second, ER-localised PA synthesis pathway of eukaryotic origin maintaining minimal plant development [56,57]. Despite having this second pathway, a similar compensation is not observed in T. gondii. This may be because plant (and algal) ATS1 assembles FASIIgenerated FA only into plastid-specific lipids, including PG and galactolipids [56,57]. In contrast, our data indicates that apicoplast lipid precursor synthesis has a much broader role in total lipid production in T. gondii, possibly as a result of having evolved to provide FA/lipids to the ER.
In algae and plants, FA and G3P can be assembled in the plastid via ATS1 to form LPA, which is the precursor for all other phospholipids. We show that TgATS1 is a typical, soluble algal/plant-like ATS based on sequence similarity and structural modelling. TgATS1 has a putative domain II containing all of the residues required for catalysis, including the catalytic pocket motif (NHX4D) arranged in a similar spatial arrangement to that of plant ATS1. Our results suggest that TgATS1 displays specificity regarding FA length (i.e. showing a preference for C14:0 over C16:0). Since FA specificity could have strong implications for membrane biogenesis during intracellular parasite development, FA selectivity should be addressed more closely in these parasites.

De novo synthesis of fatty acids in the apicoplast
Using U-13 C-glucose incorporation as a marker for FASII synthesis, we show here and in previous studies [54] that the apicoplast FASII machinery generates FA ranging from C12:0 to C16:0, and only a minor amount of C18:0. When TgATS1 is repressed (by addition of ATc) for 4 days, we observed reduced 13 C incorporation into most FA species (Fig 5B). Over the course of ATc treatment, we also observed severe disruption in apicoplast morphology and biogenesis, as well as changes in the IMC and most intracellular compartments (Figs 2C-2F and 3). We cannot distinguish between the possibilities that reduced 13 C incorporation into FA is directly linked to the loss of TgATS1, or rather the loss of the apicoplast (and hence concomitant loss of FASII). The possibility that FA biosynthesis per se is not disrupted following repression of TgATS1 is suggested by the finding that 13 C-incorporation into PE-ceramide was largely unaffected by TgATS1 repression, even after 4 days (S10B and S10C Fig). As ceramide synthesis is not dependent on TgATS1 (S10B and S10C Fig) [58], this finding suggests that FASII is maintained in TgATS1-repressed cells, at least until the apicoplast is irrevocably damaged. Maintenance of FASII activity was also confirmed by (i) the late loss of the apicoplast after 5-6 days of TgATS1 repression (Fig 2E) and (ii) the detection of fully labelled FA moieties in PL after 4 days of TgATS1 repression (Figs 5B, 6G-6L and 8B and 8D).
Using U-13 C-acetate, we also observed ER-localized elongation of FA. This is a key difference from the plant chloroplast, where C16:0 and C18:0 typically constitute the major FASII end products (for a review, see [38]). Interestingly, when TgATS1 is repressed, 13 C incorporation from 13 C-acetate is significantly increased in C14:0 (Fig 5G), most likely as an attempted compensatory mechanism responding to loss of C12:0 C14:0, C16:0 FA moieties. This suggests that the elongases could be used for de novo FA synthesis, or that 13 C-acetate/ 13 C-acetyl-CoA can enter the apicoplast and supply FASII when apicoplast integrity is compromised, although this is insufficient to support parasite growth. MID analysis of C14:0 (S6 Fig) suggests that an 8-carbon precursor (e.g. octanoate, C8:0) could be elongated under these conditions, perhaps reflecting increased scavenging of host lipids (such as lipoic acid) in an attempt to overcome reduced de novo PL assembly [46]. This increased scavenging may explain the low level of continued parasite growth observed after 8 days under TgATS1 repression (S3 Fig). Most observed lipids/phospholipids are affected by the loss of ATS1 in T. gondii, most significantly in the decrease of C14:0 (Figs 5B and 8K) and those PLs derived from LPA containing C12:0 and C14:0 (Fig 6). Moreover, when labelling with U-13 C-glucose, mature PL species with shorter FA (i.e. where the two acyl chains combined total 28 to 36 carbons) contained relatively more 13 C-labelled FA than PLs with longer FA moieties (>36 carbons). These PL species were much more affected by TgATS1 loss (Fig 6). These results suggest that the parasite preferentially uses shorter, apicoplast-synthesised FA (C12-C18) for PL production. Longer chain (>C18) FA can also be used for PL synthesis, but are likely of non-plastid origin (e.g. via scavenging from the host, and/or remodelling of existing membrane lipids). Since only LPA (C14:0) could successfully rescue the loss of TgATS1 (Fig 8L), it is possible that LPA(C14:0) may act not only as the substrate for subsequent PL assembly but also in a signalling role to control overall lipid biosynthesis, although this remains to be confirmed.

Function of the apicoplast in membrane biogenesis
Apicoplast FASII and ATS1 contribute to the synthesis of PL synthesis in the parasite, including the three PL classes, PC, PE and PI. This is a major departure from chloroplasts, the photosynthetic plastids of algae/plants, where the plastid-generated FA incorporated into LPA and PA via the PA synthesis pathway are solely used for the assembly of chloroplast lipids (i.e. galactolipids, sulfolipids and PG) [33]. Only under stressed conditions such as phosphate deprivation are chloroplast-derived glycerolipids transported to extra-plastidial membranes to compensate for loss of PL biosynthesis [42,59]. Instead of generating galactolipids, apicomplexan parasites may have evolved to export LPA (and/or PA) from the apicoplast to the ER to generate PL for global membrane biogenesis. The machinery for export of LPA/PA is yet to be identified and may involve the contact sites observed between apicoplast and ER membranes in T. gondii [60]. Interestingly, plant chloroplasts possess a multimeric ATP-binding cassette (ABC) transporter that allows PL and possibly PA import into chloroplasts [61]. It is plausible that similar machinery may have evolved in Apicomplexa to export LPA/PA from the plastid, in a manner reminiscent of the change in function of the apicomplexan ER-associated protein degradation (ERAD) machinery allowing import, rather than export, of protein into the apicoplast [62].
The mechanism by which TgATS1-generated LPA is converted to PA and then PL remains unclear. The genome of T. gondii (and Plasmodium spp.) contains a hypothetical complete plant-like PA synthesis pathway that is predicted to be in the apicoplast, including ATS1 and a lysophosphatidic acid acyltransferase (LPAAT, or ATS2). Interestingly, immunofluorescence studies in P. yoelii, suggest that ATS2 (PyLPAAT) partially localises to the ER, despite the presence of a predicted N-terminal bipartite apicoplast-targeting sequence [44]. It will be of interest to determine whether TgATS2 resides in the ER, and hence whether it could convert LPA exported from the apicoplast to PA in the ER. Alternatively, ATS2 could be dually targeted to, or transported between, both organelles.
Continued synthesis of PLs may be required for transport of membrane vesicles to and from the apicoplast and the maintenance of apicoplast integrity. Consistent with this notion, we show that repression of TgATS1 is associated with both the complete loss of the apicoplast organelles and/or the appearance of unusual vacuolar structures (Fig 3). In addition, internal compartment structure was highly affected upon the loss of TgATS1 (Figs 2F and 3), suggesting that TgATS1-generated PLs could be distributed throughout the cell to participate in global membrane biogenesis. Previous reports have hypothesised that apicoplast-derived FA may participate in apicoplast biogenesis in T. gondii [11] (and during liver stages of the rodent malaria parasites P. berghei and P. yoelii [14,15,44]). Our observations of defective apicoplast biogenesis in the TgATS1 mutants support this idea, although we could not conclude whether this phenotype was due to a lack of PL for maintenance of apicoplast integrity or a general 'loss of apicoplast/FASII' phenotype. It would be interesting to assess the origin of T. gondii apicoplast membrane lipids, as has been done for P. falciparum [17].
We observed that while 70-74% of PC and PI fatty acids were of apicoplast-origin (Fig 7A  and 7B, respectively), TgATS1 repression resulted in the reduction of~60% of PC and PI abundance (Fig 6H and 6J, respectively), suggesting that a considerable fraction (~10-15%) of PC and PI pools were also synthesised through this alternative, non-ATS1 route. The source of PA is not only from the ATS1-dependent apicoplast de novo pathway but also from PL recycling pathways, for example via DAG-kinase [63]. This likely explains why the loss of ATS1 did not significantly affect the total amount of PA (Fig 8J). Since PA is the central precursor for PL synthesis, the source(s) of PA should be proportionally reflected in all downstream PL. However, the loss of TgATS1 did not fully reflect this, since the incorporation of 13 C label from U-13 C-glucose into PC and PI was greatly affected by TgATS1 repression, while that of PE was less perturbed. This could be due to the difference in the substrate specificity of each synthetic pathway. For instance, elements/enzymes of the PC and PI biosynthetic pathways might have a preference for substrates with shorter acyl chains of apicoplast origin. Conversely, PE biosynthetic elements/enzymes might have a preference for longer acyl chains, likely derived from recycling pathways. Indeed, LC-MS/MS analyses revealed that PC and PI were present of shorter chain length (from PC(28:0) and PI(30:0)) than PE (from PE(32:1)) ( Fig 7D-7F).
An alternative hypothesis could be due to the pathways downstream of ATS1 and ATS2. De novo-synthesized PA is usually converted to diacylglycerol (DAG) or cytidyldiphosphate-DAG (CDP-DAG) for subsequent PL synthesis. The T. gondii genome encodes two putative CDP-DAG synthetases (CDSs): TGGT1 281980 (TgCDS1, a homolog to eukaryotic CDS), and TGGT1_263785 (TgCDS2). TgCDS1 (ER) and TgCDS2 (apicoplast) have recently been localized and their disruption affects the synthesis of PI and PG, respectively [64]. This also supports the hypothesis that each CDS may have its own substrate specificity and, moreover, that this specificity may be related to the localization of these enzymes. The two putative T. gondii phosphatidic acid phosphatases (PAPs), generating DAG [63], have been localized in the cytosol or in the vicinity of the IMC of T. gondii, suggesting that the conversion of PA to DAG may occur outside the apicoplast, but any site-dependent substrate specificity is not yet known. It is not yet clear whether the specificity to the substrate determines enzyme localization or vice-versa. In addition, PE synthesis has been shown to rely on (i) ER-localized ethanolamine phosphotransferases (EPTs), (ii) conversion of PS to PE by a mitochondrial PS decarboxylase (TgPDSmt), and (iii) a putative direct scavenging from the host cell [35]. These multiple sources and localizations of the PE biosynthetic pathways, especially those outwith apicoplast-ER interactions, may explain why PE is less affected by the loss of TgATS1 than PC and PI.

Conclusion
This study shows that apicoplast-generated FAs constitute an important source of precursors for bulk phospholipid biosynthesis during intracellular tachyzoite development. We show that apicoplast-synthesized FA are added to glycerol 3-phosphate by the enzyme TgATS1 to form LPA in the apicoplast. We propose that LPA is then converted to PA, and that LPA and/or PA are then trafficked to the ER, where they are ultimately converted to PC, PI, PE, and/or their precursors (DAG, CDP-DAG) (Fig 9). Mature PLs can then be trafficked to other organelles, including the apicoplast. Loss of TgATS1 is therefore associated with defects in PL synthesis and apicoplast biogenesis and function, including FA synthesis. T. gondii, and potentially other Fig 9. Proposed role of the apicoplast lipid assembly pathway. The glycolytic intermediates, dihydroxyacetone-phosphate (DHAP), and phosphoenolpyruvate (PEP) are imported into the apicoplast by the apicoplast phosphate transporter (APT) and converted to glycerol 3-phosphate (G3P) and acetyl-CoA, respectively. Acetyl-CoA is used by the FASII to generate, predominantly, C12 and C14 FA chains, which are transferred to G3P by TgATS1 to form lysophosphatidic acid (LPA). These LPA species are exported to the endoplasmic reticulum (ER) by an as yet unidentified transport system to generate bulk PC, PI and PE (shown as head groups with C, I and E, respectively). All PC, PI and PE species could then be elongated by elongase enzymes (ELO) before being exported to contribute to general parasite membrane biogenesis. Note: PI may also me assembled by PI synthase outside of the ER, most likely in the Golgi apparatus.
doi:10.1371/journal.ppat.1005765.g009 ATS1 Is Essential for T. gondii Lipid Synthesis apicomplexan parasites, therefore appear to have redirected FASII/ATS1 pathways of FA/LPA synthesis from galactolipid biosynthesis (as occurs in plants and algal plastids) to PL synthesis for global membrane biogenesis. Although the precise location(s) of conversion of LPA to PA remains to be determined, it is clear that proteins involved in LPA synthesis and trafficking, including TgATS1, may be potential drug targets.

Construct design
The TgATS1 open reading frame was PCR amplified using primers 5'-GATCTGATCAAAAA TGCTTTTCTTTCTGGCTCGTCTCC and 5'-GATCTCTAGATTCGAAATGCGGAGTGA TAAGTTTGTACAG, digested with BclI and XbaI, and ligated into the BglII and AvrII sites of vector pDt7s4H [51]. The resultant construct was transfected into TATi strain parasites and selected on pyrimethamine to produce the parental strain. This was subsequently cloned before characterisation. To disrupt the native TgATS1 locus, we amplified the CAT cassette from the vector piCG [70] using primers 5'-TCACGAACCGAAACGAAAAACGTCCAATCCCTCTT CACAGACGGCCCAGGATACGACTCACTATAGGGCGAATTGG and 5'-GTCGAACGGA TCCGGTTGGCTGAACACACCGTTGTCGTCTTGCCGGTCTCCCTCGACTACGGCTTC CATTGGCAAC. The resultant products were introduced into the cosmids TOXO218 and TOXOX36 using a recombineering approach, as described previously [52]. The resultant modified cosmids were transfected into the parental parasite strain expressing regulated TgATS1, selected on chloramphenicol, and cloned out. Clones were screened for successful disruption of the native TgATS1 gene using primers 5'-GCAGCAATAGTTCTTTTCAGG and 5'-AGGCGT CTTCGTGCGTATC, which will only give a band if the native TgATS1 gene is present. To further verify knockout of the native TgATS1 gene, Southern blotting was performed as described previously [71] using a probe amplified with the same primers as for the PCR screen.

Antibodies and immunofluorescence assays
Primary anti-CPN60 (rabbit) antibodies [51] were used at a dilution of 1:3000, anti-IMC1 at 1:1000, anti-HA (Rat, Roche) at 1:500, and anti-ATRx1 (Mouse) at 1:1000. Secondary AlexaFluor 488-and 546-conjugated anti-rat and anti-rabbit antibodies (Life Technologies) were used at 1/ 10000 dilutions, respectively. Parasites were fixed in PBS containing 4% paraformaldehyde for 30 min on ice. Samples were permeabilized with 0.1% Triton X-100 in PBS for 10 min on ice before blocking in PBS containing 3% BSA and incubation with primary antibodies then secondary antibodies diluted in the blocking solution. Labelled parasites were stained with Hoechst (1:10,000, Life Technologies) for 20 min and then washed three times in PBS then H 2 O. Coverslips were mounted onto slides prior to observation using a Zeiss epifluorescent microscope.

Electron microscopy
Intracellular tachyzoites were fixed in PBS containing 2.5% glutaraldehyde and 0.5% tannic acid (pH 7.2; Polysciences) for 15 min, followed by three washes in PBS. Samples were post-fixed in osmium tetroxide (OsO 4 ) in PBS for 2.5 min followed by three washes in PBS and water. Samples were stained overnight in 1% uranyl acetate, washed three times in water and dehydrated in a graded series of ethanol. Samples were embedded in LR white resin (London Resin) and ultrathin sections were observed on a Phillips Bio Twin (120 kV) electron microscope.

Phenotypic analyses
Plaque assays: HFF monolayers were infected with 500 parasites and allowed to develop under normal culture conditions for 10 days before staining with Crystal Violet (Sigma) and cell growth assessment by light microscopy for the presence of intact HFF.
Apicoplast quantification assay: TgATS1-iKO-FNR-RFP parasites were collected promptly after egress and inoculated onto new HFF monolayers and cultured for 0 to 6 days in the presence or absence of ATc. At each time point cultures were fixed with paraformaldehyde and stained with anti-IMC1 and anti-HA. The numbers of parasites per vacuole were counted for more than 100 vacuoles for each condition at each point time. Apicoplast loss was determined by observing the fluorescent FNR-RFP protein as an apicoplast marker.
Intracellular growth assays: We introduced cytosolic tdTomato into TgATS1-iKO parasites and performed fluorescence growth assays as described previously [71].
Lipid extraction and analysis of T. gondii tachyzoites Lipid extraction and analysis of tachyzoites was performed as previously described [12]. Intracellular tachyzoites (4 × 10 8 cell equivalents) were extracted in chloroform/methanol/water (1:3:1, v/v/v containing 50 nmol laurate (C12:0) as internal standard) for 1 h at 4°C, with periodic sonication. For MS analysis, polar and apolar metabolites were separated by phase partitioning. Total fatty acid analysis of the lipophilic fraction was analysed by GC-MS (fatty acids) and LC-MS (phospholipids). In all cases, lipids were extracted from the same cell number and experiments were repeated using each independent mutant line as a biological replicate.

Stable isotope labelling of T. gondii fatty acids and phospholipids
Stable isotope labelling using U-13 C-glucose or U-13 C-acetate (Cambridge Isotope Laboratories, USA), lipid extraction, and GC-MS analysis was performed as previously described [12,53]. Briefly, freshly infected HFF were incubated in the presence or absence of ATc (0.5 μM, Sigma-Aldrich) in either glucose-free medium supplemented with 8 mM U-13 C-glucose or low-glucose DMEM, supplemented with 8 mM U-13 C-acetate. In TgATS1 repression experiments, the 13 C-carbon source was added simultaneously with ATc. Other supplements (glutamine, sodium bicarbonate, and foetal bovine serum) were added according to normal culture conditions. Parasites were harvested at indicated time points and metabolites extracted and partitioned as above. An aliquot of the lipid extract was derivatised on-line using MethPrep II (Alltech) and the resulting fatty acid methyl esters were analysed by GC-MS as previously described [12,53]. All fatty acids were identified by comparison of retention time and mass spectra with authentic chemical standards and label incorporation was calculated as the percent of the metabolite pool containing one or more 13 C atoms after correction for natural abundance and the amount of 13 C-carbon source in the culture medium (as determined by GC-MS analysis). In these experiments, U-13 C-glucose or U-13 C-acetate were added at the same time as ATc, and so the label incorporation observed in +ATc samples may be from the initial period before TgATS1 was fully absent.

Liquid chromatography-mass spectrometry analysis
Total lipids were extracted and partitioned as above. These extracts were suspended in 100 μL 1-butanol/10 mM ammonium formate in methanol (1:1, v/v) and 0.5 μL aliquots were analysed using the following LC-MS method. An Agilent 1290 series liquid chromatography (LC) system (maximum pressure 1200 bar) comprising a vacuum degasser, binary pump, column oven and temperature controlled autosampler was interfaced with a Jetstream electrospray ionization triple Quadrupole (Agilent 6490 QQQ) or Quadrupole Time-of-Flight (Agilent 6550 QTOF) mass spectrometer. The LC parameters were as follows: column: 2.1 × 100 mm, 1.8 μm C18 Zorbax Elipse plus (Agilent); column temperature: 60°C; rate 0.6 mL/min. Gradient elution was from 45% mobile phase B to 100% B over 20 min, followed by 5 min at 100% B and a 3 min re-equilibration to 45% B. Mobile phase A: 10 mM ammonium formate in water; mobile phase B: water:acetonitrile:isopropanol, 5:20:75 (v/v/v) with 10 mM ammonium formate. The ESI source settings were: gas temperature: 250°C; gas flow rate: 20 L/min; nebulizer pressure: 45 psi; sheath gas temp: 350°C; sheath gas flow: 11 L/min; capillary voltage: 3000 V. In-spectrum calibration of the QTOF data was performed using reference ions of 121.0508 m/z and 922.0097 m/z which were supplied through the second ESI needle. A 10,000-count threshold was set for untargeted MS/MS experiments. All solvents were LC-MS grade (Burdick and Jackson) and 18.2 MΩ deionized water used.
The following scans were used for the three key lipid classes on the QQQ-MS: PC positive ionisation precursor scan of m/z 184, PE (and PE-Cer) positive ion neutral loss scan of 141 u, and PI negative ionisation precursor ion scan of m/z 241. These correlate to the relevant polar head groups. Data were analysed using Mass Hunter Qualitative (QTOF data) and Quantitative (QQQ data) software (Agilent).

PA quantification
Total lipid spiked with 25 nmol C13:0 fatty acid was extracted from U-13 C-glucose labelled parasites prepared as above using chloroform:methanol, 1:2 (v/v) and chloroform:methanol, 2:1 (v/v) in the presence of 0.1 M HCl. Pooled organic phase was subjected to biphasic separation by adding 0.1 M HCl. The organic phase was dried under N 2 gas and dissolved in 1-butanol. Total lipid was then separated by 2D-HPTLC with 1 μg PA(C17:0/C17:0) (Avanti Polar lipids) using chloroform/methanol/28% NH 4 OH, 60:35:8 (v/v) as the 1 st dimension solvent system and chloroform/acetone/methanol/acetic acid/water, 50:20:10:13:5 (v/v) as the 2 nd dimension solvent system [72]. The spot corresponding to PA was identified according to the migration of authentic PA standard, and subsequently extracted for GC-MS analysis (Agilent 5977A-7890B) after methanolysis using 0.5 M HCl in methanol incubated at 100°C for 1 h. Fatty acid methyl esters were identified by their mass spectrum and retention time compared to authentic standards. PA content was normalized according to the parasite cell number and internal standard.
Supporting Information S1 Fig. Related to Fig 1: A three-dimensional structure of TgATS1 was generated using CmATS1 as a model [47]. (A) The overall structure as observed in the ribbon representation of CmATS1α-carbons. Surface accessibility (B) and residues involved in substrate binding (G3P and acyl-ACP) and the catalytic motif NHX4D (C) of CmATS1. The structure (D), surface accessibility (E), and residues and motifs (F) are conserved and highly similar in TgATS1, forming similar grooves and pockets to those found in CmATS1 (G) Phylogeny of T. gondii The TgATS1 sequence was fused (i) to a HA-tag coding sequence at its 3'-terminus and (ii) to the tetracycline inducible promoter sequence (Pi) at its 5'-terminus (iTgATS1). The construct was transfected and randomly inserted into the TATi line genome, prior to endogenous gene (eTgATS1) replacement by a Chloramphenicol Acyltransferase (CAT) resistance cassette [73] via double homologous recombination using a specific CAT ATS1 KO cosmid [74]. Probes and restriction sites used for Southern blot are indicated by arrowheads and restriction enzymes (RE) names, respectively. Parasite growth rate was analysed over the course of 8 days by quantifying the fluorescence of tdTomato [71] expressed in the cytosol of TgATS1-iKO parasites in the absence (blue rectangles, control) or the presence of ATc. TgATS1-iKO were grown in the presence of ATc from day 0 to day 8 (red squares) or pre-treated with ATc for 3 days prior to the 8 day ATc treatment (green triangles). In the absence of ATc, TgATS1-iKO grew normally as observed by fluorescence levels but the presence of ATc substantially affected the amount of fluorescence, with this effects being strongest with 3 days of pre-treatment.  13 C-glucose metabolic labelling using stable isotope precursors. Tachyzoites of T. gondii conditional mutants for ATS1 were labelled with U-13 C-glucose for 1 day in the presence or absence of ATc. Lipids were extracted, derivatized, and the resulting FAMEs were analysed by GC-MS to determine 13 C incorporation. (A) The mean label incorporation from U-13 C-glucose into fatty acids is shown for parasites grown in the absence (dark red) and presence (light red) of ATc. (B) The MIDs for C14:0 labelled with U-13 C-glucose in the presence (red) and absence (pale red) of ATc. The x-axis indicates the number of 13 C atoms in each FAMEs ('m0' indicates the monoisotopic mass containing no 13 C atoms, while 'mX' represents that mass with 'X' 13 C atoms incorporated). Nomenclature Cx:y is shown where x is the number of carbons and y is the number of double bonds in the fatty acid chain. Error bars indicate standard deviation, where n = 2 biological replicates. Data shown has been background-subtracted for natural isotope abundance. Analysis of FASII fatty acid biosynthesis and elongation in TgATS1-iKO parasites by metabolic labelling using U-13 C-glucose. Tachyzoites of T. gondii conditional mutants for TgATS1 were labelled with U-13 C-glucose in the presence or absence of ATc. Lipids were extracted, derivatized, and the resulting fatty acid methyl esters (FAMEs) were analysed by GC-MS to determine isotope incorporation. MIDs for all detected FAMEs labelled with U-13 C-glucose in the absence and presence of ATc are shown in red and pale red, respectively. The x-axis indicates the number of 13 C atoms in each FAMEs ('m0' indicates the monoisotopic mass containing no 13 C atoms, while 'mX' represents that mass with 'X' 13 C atoms incorporated). Nomenclature Cx:y is shown where x is the number of carbons and y is the number of double bonds in the fatty acid chain. Error bars indicate standard deviation, where n = 4 biological replicates. Data shown has been background-subtracted for natural isotope abundance. iKO parasites by metabolic labelling using U-13 C-acetate. Tachyzoites of TgATS1-iKO parasites were labelled with 13 C-U-acetate in the presence or absence of ATc. Lipids were extracted, derivatized, and the resulting FAMEs were analysed by GC-MS to determine isotope incorporation. MIDs for all detected FAMEs labelled with U-13 C-acetate in the absence and presence of ATc are shown in purple and pale purple, respectively. MIDs suggested that saturated/ monounsaturated FAs incorporated 13 C-atoms onto C16:0 or C14:0 in a units of two (i.e. C18 incorporated 2 or 4, C20 incorporated 4 or 6, C22 incorporated 6 or 8, and C24 incorporated 8 or 10 13 C-atoms onto C16:0 or C14:0, respectively). The x-axis indicates the number of 13 C atoms in each FAME ('m0' indicates the monoisotopic mass containing no 13 C atoms, while 'mX' represents that mass with 'X' 13 C atoms incorporated). Nomenclature Cx:y is shown where x is the number of carbons and y is the number of double bonds in the fatty acid chain. Error bars indicate standard deviation, where n = 4 biological replicates. Data shown has been background-subtracted for natural isotope abundance. MIDs of PE-Cer(d18:1/16:0) extracted from TgATS1-iKO parasites grown in unlabelled conditions (black bars) or labelled with U-13 C-glucose for 4 days in the absence (white bars) or presence of ATc (grey bars). 'm0' indicates the monoisotopic mass containing no 13 C atoms, while 'mX' represents that mass with 'X' 13 C atoms incorporated. (C) Relative abundances of apicoplast-generated FA moieties (i.e. those containing 4 or more 13 C atoms) in PE-Cer(d18:1/16:0) extracted from TgATS1-iKO parasites grown in the presence and absence of ATc shows that the majority of FA is generated in the apicoplast and was not greatly affected by the disruption of TgATS1.