Ethanolaminephosphate cytidylyltransferase is essential for survival, lipid homeostasis and stress tolerance in Leishmania major

Glycerophospholipids including phosphatidylethanolamine (PE) and phosphatidylcholine (PC) are vital components of biological membranes. Trypanosomatid parasites of the genus Leishmania can acquire PE and PC via de novo synthesis and the uptake/remodeling of host lipids. In this study, we investigated the ethanolaminephosphate cytidylyltransferase (EPCT) in Leishmania major, which is the causative agent for cutaneous leishmaniasis. EPCT is a key enzyme in the ethanolamine branch of the Kennedy pathway which is responsible for the de novo synthesis of PE. Our results demonstrate that L. major EPCT is a cytosolic protein capable of catalyzing the formation of CDP-ethanolamine from ethanolamine-phosphate and cytidine triphosphate. Genetic manipulation experiments indicate that EPCT is essential in both the promastigote and amastigote stages of L. major as the chromosomal null mutants cannot survive without the episomal expression of EPCT. This differs from our previous findings on the choline branch of the Kennedy pathway (responsible for PC synthesis) which is required only in promastigotes but not amastigotes. While episomal EPCT expression does not affect promastigote proliferation under normal conditions, it leads to reduced production of ethanolamine plasmalogen or plasmenylethanolamine, the dominant PE subtype in Leishmania. In addition, parasites with episomal EPCT exhibit heightened sensitivity to acidic pH and starvation stress, and significant reduction in virulence. In summary, our investigation demonstrates that proper regulation of EPCT expression is crucial for PE synthesis, stress response, and survival of Leishmania parasites throughout their life cycle.


Introduction
Protozoan parasites of the genus Leishmania are transmitted through the bite of hematophagous sand flies. During their life cycle, Leishmania parasites alternate between flagellated, extracellular promastigotes in sand fly midgut and non-flagellated, intracellular amastigotes in mammalian macrophages. These parasites cause leishmaniasis which ranks among the top ten neglected tropical diseases with 10-12 million people infected and 350 million people at the risk of acquiring infection [1,2]. Drugs for leishmaniasis are plagued with strong toxicity, low efficacy, and high cost [3,4]. To develop better treatment, it is necessary to gain insights into how Leishmania acquire essential nutrients and proliferate in the harsh environment in the vector host and mammalian host.
To sustain growth, Leishmania parasites must generate abundant amounts of lipids including glycerophospholipids, sterols and sphingolipids. Phosphatidylethanolamine (PE) and phosphatidylcholine (PC) are two common classes of glycerophospholipids. Besides being major membrane constituents, PE and PC can function as precursors for several signaling molecules and metabolic intermediates including lyso-phospholipids, phosphatidic acid, diacylglycerol, and free fatty acids [5,6]. In addition, PE contributes to the synthesis of GPIanchored proteins in trypanosomatid parasites by providing the ethanolamine phosphate bridge that links proteins to glycan anchors [7]. PE is also involved in the formation of autophagosome during differentiation and starvation in Leishmania major [8,9] and the posttranslational modification of eukaryotic elongation factor 1A in T. brucei [10].
For many eukaryotic cells, the majority of PE and PC are synthesized de novo via the Kennedy pathway (Fig 1) [11]. In Leishmania, the key metabolite ethanolamine phosphate (EtN-P) is generated from of the sphingoid base metabolism or the phosphorylation of ethanolamine (EtN) by ethanolamine kinase [12] (Fig 1). EtN-P is then conjugated to cytidine triphosphate (CTP) to produce CDP-EtN and pyrophosphate by the enzyme ethanolaminephosphate cytidylyltransferase (EPCT) [13]. In the EtN branch of the Kennedy pathway, through the activity of ethanolamine phosphotransferase (EPT), CDP-EtN is combined with 1-alkyl-2-acyl-glycerol to eventually generate plasmenylethanolamine (PME) or ethanolamine plasmalogen, the dominant subtype of PE in Leishmania [14]. CDP-EtN can also be combined with 1,2-diacylglycerol to form 1,2-diacyl-PE (a minor subtype of PE in Leishmania) by choline ethanolamine phosphotransferase (CEPT) [15,16]. A similar branch of the Kennedy pathway is responsible for the de novo synthesis of PC (choline ! choline-phosphate ! CDP-choline ! PC), with CEPT catalyzing the last step of conjugating CDP-choline and diacylglycerol into PC as a dual activity enzyme [16][17][18]. Besides the Kennedy pathway, PE may be generated from the decarboxylation of phosphatidylserine (PS) in the mitochondria and PC may be generated from the N-methylation of PE [19][20][21].
In addition to biosynthesis, Leishmania parasites (especially the intracellular amastigotes) can take up and modify existing lipids to fulfill their needs [22][23][24]. For example, the enzyme CEPT which is directly responsible for producing PC and diacyl-PE (Fig 1) is essential for L. major promastigotes in culture but is not required for the survival or proliferation of amastigotes in mice [18]. While diacyl-PE is a minor lipid component in Leishmania, PC is highly abundant in both promastigotes and amastigotes [25,26]. These results argue that de novo PC synthesis is required to generate large amount of lipids to support rapid parasite replication during the promastigote stage (estimated doubling time: 6-8 hours in culture and 10-12 hours

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania in sand fly) [27,28]. In contrast, intracellular amastigotes can acquire enough PC through the uptake and remodeling of host lipids, which seems to fit their slow growing, metabolically quiescent state (estimated doubling time: 60 hours) [29,30]. Consistent with these findings, sphingolipid analyses of L. major amastigotes revealed high levels of sphingomyelin which could not be synthesized by Leishmania but was plentiful in mammalian cells [31]. Conversely, the biosynthesis of parasite-specific sphingolipid, inositol phosphorylceramide (IPC), is fully dispensable for L. major amastigotes [12,31,32]. Collectively, these data suggest that as Leishmania transition from fast-replicating promastigotes to slow-growing amastigotes, they undergo a metabolic switch from de novo synthesis to scavenge/remodeling to acquire their lipids [24].
Despite of these findings, it is important to explore whether intracellular amastigotes retain some capacity for de novo lipid synthesis. Our previous studies on the sterol biosynthetic mutant c14dm�suggest that is the case [33]. L. major c14dm�mutants lack the sterol-14αdemethylase which catalyzes the removal of C-14 methyl group from sterol intermediates. Promastigotes of c14dm�are devoid of ergostane-type sterols (which are abundant in wild type [WT] parasites) and accumulate 14-methyl sterol intermediates. However, lipid analyses of lesion-derived amastigotes demonstrate that both WT and c14dm�amastigotes contain cholesterol (which must be taken from the host since Leishmania only synthesize ergostane-type sterols) as their main sterol [33]. Parasite-specific sterols, i.e., ergostane-type sterols such as ergosterol and 5-dehydroepisterol in WT amastigotes and C-14-methylated sterols in c14dm� amastigotes are only detected at trace levels. Nonetheless, c14dm�amastigotes show significantly attenuated virulence and reduced growth in comparison to WT and C14DM add-back amastigotes, suggesting that the residual amounts of endogenous sterols (which cannot be scavenged from the host) play pivotal roles in amastigotes which may constitute the basis for sterol biosynthetic inhibitors as anti-leishmaniasis drugs [33,34].
In this study, we investigated the roles of EPCT (EC 2.7.7.14) which catalyzes the conversion of EtN-P and CTP into CDP-EtN and pyrophosphate in the Kennedy pathway (Fig 1). Our previous study on EPT in L. major revealed that EPT is mostly required for the synthesis of PME but not diacyl PE or PC [14]. Disruption of EPCT, on the other hand, may have a much more profound impact on the synthesis of PME, diacyl PE and PC (which can be generated from diacyl PE via N-methylation) (Fig 1). In mammalian cells, EPCT is considered as the rate limiting enzyme in the EtN branch of the Kennedy pathway and a key regulator of PE synthesis [35,36]. Disruption of EPCT in mice leads to developmental defects and embryonic lethality [37,38]. Additional studies report EPCT heterozygous mice have increased diacylglycerol and triacylglycerol in liver that resemble features of metabolic diseases, suggesting that EPCT is involved in maintaining the homeostasis of neutral lipids as well [38,39]. In Trypanosoma brucei (a related trypanosomatid protozoan), RNAi knockdown of EPCT reduces de novo PE synthesis leading to growth arrest and altered mitochondrial morphology in procyclics [16,40,41]. Based on these findings, it is of interest to determine the roles of EPCT in Leishmania especially during the disease-causing intracellular stage when parasites acquire most of their lipids via scavenging. Our results indicate that EPCT is essential for the survival of both promastigotes and amastigotes in L. major. In addition, the expression level of EPCT is crucial for stress tolerance, lipid homeostasis and the synthesis of GPI-anchored proteins. These findings may inform the development of new anti-Leishmania drugs.

Identification and functional verification of L. major EPCT
cruzi. The predicted L. major EPCT protein consists of 402 amino acids and bears 35-42% identity to EPCTs from Saccharomyces cerevisiae, Plasmodium falciparum, Arabidopsis thaliana, and Homo sapiens (S1 Fig). To confirm its function, the EPCT open reading frame (ORF) was cloned into a pXG plasmid (a high copy number protein expression vector) [42] and introduced into L. major wild type parasites (WT) as WT+EPCT. When promastigote lysate from WT+EPCT was incubated with [ 14 C]-EtN-P and CTP for 20 minutes at room temperature, we detected a radiolabeled product which exhibited similar mobility (retention factor) as pure CDP-EtN on thin layer chromatography (TLC), suggesting it was [ 14 C]-CDP-EtN (Figs 2A and S2A). Radiolabeled CDP-EtN was also formed when [ 14 C]-EtN-P and CTP were incubated with mouse liver homogenate, but not boiled lysates (Figs 2A and S2B). We did not detect significant amount of [ 14 C]-CDP-EtN using WT lysate, indicating that the basal level of EPCT activity in L. major promastigotes was below the level of detection with this approach (Fig 2A and 2B, activity from WT was similar to buffer alone and boiled lysate). To determine the cellular localization of EPCT, a GFP-tagged EPCT was cloned into pXG plasmid and introduced into WT promastigotes to generate WT+GFP-EPCT. These WT+GFP-EPCT cells displayed similar EPCT activity levels as WT+EPCT cells (Fig 2A and 2B; 11-12 times higher than WT or boiled lysate, p<0.001). By western blot, GFP-EPCT was detected at the predicted molecular weight of~73 kDa ( Fig 2C) and mainly found in the cytosolic fraction after sub-cellular fractionation ( Fig  2D). A previously reported cytoplasmic protein HSP83 [43] was detected in both the cytosolic and the large granule fractions (pellet 10K), whereas the plasma membrane-localized lipophosphoglycan (LPG) [44] was mainly found in the large granule fraction (Fig 2D). In addition, GFP-EPCT exhibited a cytoplasmic localization by fluorescence microscopy (Fig 2E). Together, these findings argue that L. major EPCT is a cytosolic enzyme capable of condensing EtN-P and CTP into CDP-EtN (Fig 1).

Chromosomal EPCT-null mutants cannot be generated without a complementing episome
To determine whether EPCT is required for survival in L. major, we first attempted to delete the two chromosomal EPCT alleles using the homologous recombination approach [45]. As demonstrated by Southern blot (S3A and S3B Fig), we successfully replaced one EPCT allele with the blasticidin resistance gene (BSD) using this approach and generated several heterozygous EPCT+/-clones. However, repeated attempts to delete the second EPCT allele failed to recover any true knockouts (results from one attempt were shown in S3C and S3D Fig), suggesting that a different method is needed to generate chromosomal EPCT-null mutants.

EPCT is essential for L. major promastigotes
To evaluate the essentiality of EPCT, EPCT+/-+pXNG4-EPCT and epct �+ pXNG4-EPCT promastigotes were cultivated in the presence or absence of ganciclovir (GCV). Because of the TK expression from pXNG4-EPCT, adding GCV would trigger premature termination of DNA synthesis [46]. Thus, in the presence of GCV, parasites would favor the elimination of pXNG4-EPCT (which can be tracked by monitoring GFP fluorescence level) during replication to avoid toxicity, if the episome is dispensable [46]. As shown in Fig 4A, after being cultivated in the presence of GCV for 14 consecutive passages, those chromosomal EPCT-null promastigotes still contained 60-74% of GFP-high cells, indicative of high pXNG4-EPCT retention levels after prolonged negative selection. In contrast, the pXNG4-EPCT plasmid was easily expelled from the EPCT+/-+pXNG4-EPCT parasites after 8 passages in GCV (their GFP level dropped to that of WT, Fig 4A and 4B), suggesting that the remaining chromosomal EPCT allele made the episome expendable. Without GCV, we detected a gradual loss of episome in the EPCT+/-+pXNG4-EPCT but not epct �+ pXNG4-EPCT parasites (Fig 4A), again arguing that the episome is indispensable in those chromosomal-null mutants. As controls, EPCT+/-+pXNG4-EPCT and epct �+ pXNG4-EPCT grown in the presence of nourseothricin (the positive selection) consistently retained 85-91% GFP-high cells ("+SAT" in Fig 4A and 4C).
Our analysis showed that 25-38% of epct �+ pXNG4-EPCT parasites became GFP-low after prolonged exposure to GCV (Fig 4A and 4D). To examine whether these GFP-low cells in epct � + pXNG4-EPCT could survive by themselves, they were separated from GFP-high cells by fluorescence-activated cell sorting (FACS) and individual clones were isolated after serial dilution. As illustrated in S5 Fig, when single clones were expanded from the sorted GFP-low population in the presence of GCV and absence of nourseothricin, they quickly regained GFP fluorescence. After two passages, two such clones showed similar percentages of GFP-high cells (70-74%) as the population prior to sorting (Figs 4D-4F and S5). These data suggest that GFP-low cells still contained a low level of episome and would increase the episome level during proliferation after they were separated from GFP-high cells. Additionally, quantitative PCR (qPCR) analyses were performed on these clones using primers targeting the GFP region (to determine total plasmid copy number) and the L. major 28S rDNA (to determine total parasite number). Results revealed~8 copies of pXNG4-EPCT plasmid per cell in the epct � + pXNG4-EPCT GCV-treated clones, whereas the EPCT+/-+pXNG4-EPCT clones isolated by the same process (GCV treatment ➔ GFP-low sorting ➔ serial dilution) contained <0.01 copy per cell (S6A Fig). We also sequenced the pXNG4-EPCT plasmid DNA from GCV-treated epct � + pXNG4-EPCT clones and did not find mutations in TK, GFP or EPCT. Finally, we found a significant growth delay in GCV-treated epct �+ pXNG4-EPCT but not EPCT+/-+-pXNG4-EPCT promastigotes in comparison to control cells (S6B Fig). This is consistent with the episome retention in epct �+ pXNG4-EPCT which leads to GCV-induced cytotoxicity. Together, these findings demonstrate the essential nature of EPCT in L. major promastigotes.

EPCT expression is crucial for L. major amastigotes
To investigate if EPCT is required for the survival of intracellular amastigotes, we infected BALB/c mice in the footpad with stationary phase promastigotes. After infection, half of the mice received daily GCV treatment, and the other half received equivalent amount of PBS cellular fractions of WT + GFP-EPCT cells were probed by anti-GFP (top), -HSP83 (middle), or -LPG (bottom) antibodies. (E) Log phase promastigotes of WT+GFP-EPCT were examined by fluorescence microscopy. DIC: differential interference contrast. Scale bar: 10 μm. DNA: staining with Hoechst 33342. Merge: overlay of GFP and DNA. https://doi.org/10.1371/journal.ppat.1011112.g002

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania (control group) for 14 consecutive days as previously described [48]. No significant changes in mouse body weight or movement were detected from GCV treatment (S7A Fig). Mice infected by WT and EPCT+/-parasites exhibited equally rapid development of footpad lesions that correlated with robust parasite replication, indicating that one chromosomal copy of EPCT is sufficient for the mammalian stage (Fig 5A and 5B). By comparison, mice infected by EPCT+/-+pXNG4-EPCT or epct �+ pXNG4-EPCT showed a 9-12-week delay in lesion development ( Fig  5A), and the delay was consistent with the significantly reduced parasite growth as determined by limiting dilution assay and qPCR (Figs 5B and S7B). These results suggest that elevated EPCT expression from pXNG4-EPCT, a high copy number plasmid (Figs 3C and 5C), may be responsible for the delayed lesion progression and amastigote growth from EPCT+/-+-pXNG4-EPCT and epct �+ pXNG4-EPCT promastigotes (Figs 5 and S7). To test this hypothesis, we introduced pXNG4-EPCT into WT parasites and the resulting WT +pXNG4-EPCT cells showed similar infectivity in mice as EPCT+/-+pXNG4-EPCT or epct �+ pXNG4-EPCT parasites (Fig 5), indicating that this plasmid can attenuate virulence in the WT background as well. In addition, the EPCT gene was cloned into a pGEM vector along with its 5'-and 3'-flanking sequences (~1 Kb each) and the resulting pGEM-EPCT was introduced into EPCT+/-parasites. Like cells with pXNG4-EPCT, these EPCT+/-+ pGEM-EPCT parasites displayed significantly reduced virulence and growth in BALB/c mice (S8 Fig), proving that these defects were caused by EPCT overexpression and not restricted to the pXNG4 plasmid.
While GCV treatment had little impact on infections caused by WT, EPCT+/-, or EPCT+/-+pXNG4-EPCT parasites, it caused an additional 3-4-week delay in lesion progression for epct � + pXNG4-EPCT (Fig 5A and 5B), suggesting that GCV-induced cytotoxicity is more significant in the chromosomal null mutants than others. To determine the pXNG4-EPCT copy number in amastigotes, we performed qPCR analyses on genomic DNA extracted from lesionderived amastigotes. The average plasmid copy number per cell was revealed by dividing the total plasmid copy number with total parasite number. As shown in Fig 5C, EPCT+/-+-pXNG4-EPCT amastigotes contained 8-13 copies per cell at week 4 post infection and that number gradually went down to 1.5-2.5 copies per cell by week 14-20; and GCV treatment caused reduction in plasmid copy number as expected. The reduction in plasmid copy number over time suggests that the episome is not required in EPCT+/-+pXNG4-EPCT amastigotes. Similarly, the episome copy number in WT +pXNG4-EPCT amastigotes decreased from 14-16 per cell at week 4 post infection to 2-4 per cell in week 14-20 and GCV treatment led to further reduction (Fig 5C). In comparison, epct �+ pXNG4-EPCT amastigotes maintained 12-24 plasmid copies per cell throughout the course of infection even with GCV treatment (Fig 5C). Thus, these chromosomal EPCT-null amastigotes must retain a high episome level to survive which is consistent with their increased sensitivity to GCV (Fig 5A and 5B), Together, these data argue that EPCT is critically important for the survival of intracellular amastigotes.
In addition to plasmid copy number, we also examined EPCT transcript levels by reverse transcription-qPCR using α-tubulin transcript as the internal standard (Fig 6). For WT parasites, EPCT mRNA level went down~60% from promastigotes to amastigotes, consistent with the relatively slow replication rate for amastigotes ( Fig 6A). As expected, the presence of pXNG4-EPCT resulted in a 10-18-fold increase in EPCT mRNA levels (compared to WT cells) during the promastigote stage ( Fig 6B). While the overexpression was less pronounced during the mammalian stage, we still observed a 2-8-fold increase in EPCT mRNA in epct � + pXNG4-EPCT amastigotes (week 14-20 post infection) in comparison to WT amastigotes ( Fig 6C). For EPCT+/-+pXNG4-EPCT and epct �+ pXNG4-EPCT amastigotes (week 14 post infection), EPCT transcript levels correlated with plasmid copy numbers (Figs 5C and 6C). The fact that GCV treatment increased EPCT levels in epct �+ pXNG4-EPCT may be due to a combination of factors including the level of GCV being below the minimum effective concentration and amastigotes starting rapid replication by week 14. For WT +pXNG4-EPCT amastigotes (week 14 post infection), GCV treatment had opposite effects on EPCT transcription (upregulation) and plasmid copy number (downregulation), which may reflect increased

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania transcription from the endogenous EPCT locus when the plasmid was depleted (Figs 5C and  6C).
Next, because PE and PI are involved in the biosynthesis of GP63 (a zinc-dependent, GPIanchored metalloprotease) and lipophosphoglycan (LPG), respectively, we examined whether EPCT under-or over-expression could influence the production of these surface glycoconjugates. Interestingly, immunofluorescence microscopy revealed a 4-5-fold reduction of GP63 in EPCT+/-, WT +pXNG4-EPCT, and EPCT+/-+pXNG4-EPCT parasites and a nearly 8-fold reduction in epct �+ pXNG4-EPCT (Fig 8A and 8C). On the other hand, the cellular levels of LPG (determined by western blot) were unaltered in EPCT mutant lines (Fig 8B and 8D). These findings argue that ECPT expression from its chromosomal locus is required for the proper synthesis of GP63.

EPCT overexpression affects stress response
To further explore how EPCT overexpression may compromise L. major virulence, we examined the response of mutants to starvation, acidic pH and heat stress as tolerance to these stress conditions are essential for parasite survival in the macrophage phagolysosome. Under normal conditions (complete M199 medium, pH7.4, 27˚C), EPCT+/-and overexpressors proliferated at similar rates as WT promastigotes in log phase and exhibited good viability in stationary phase (Figs 3E and 9A). However, if cells were transferred to phosphate-buffered saline (PBS, pH7.4) to test starvation tolerance, 50-60% of EPCT overexpressing cells (WT+pXNG4-EPCT, EPCT+/-+pXNG4-EPCT and epct �+ pXNG4-EPCT) died after 2-3 days in comparison to <20% death for WT parasites (Fig 9B). These overexpressors also showed a 24-48-hour growth delay if they were cultivated in a pH5.0 medium (Fig 9C). In addition, epct �+ pXNG4-EPCT parasites were slightly more sensitive to acidic stress (pH 5.0, Fig 9D) and heat (37˚C, Fig 9E and 9F) than WT in the stationary phase We did not detect any significant difference in mitochondrial superoxide level between WT and EPCT mutants (S13 Fig). Whether these defects are linked to the altered lipid contents in EPCT overexpressors remains to be determined. Regardless, sensitivity to stress conditions likely contributes to their lack of virulence.

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania

Discussion
Using a complementing episome assisted knockout approach coupled with negative selection, we demonstrate that EPCT is indispensable throughout the life cycle in L. major. The same method was applied to verify the essentiality of several genes in Leishmania [46][47][48]. The fact

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania that EPCT is essential in both promastigotes and amastigotes is in sharp contrast to cholinephosphate cytidylyltransferase (CPCT) which catalyzes the equivalent step in the choline branch of the Kennedy pathway by combining choline phosphate and CTP into CDP-choline (Fig 1). L. major CPCT-null mutants (cpct �) cannot incorporate choline into PC but contain similar levels of PC as WT parasites [49]. Loss of CPCT has no impact on promastigote growth

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania in culture or their virulence in mice [49]. A similar study on CEPT which is directly responsible for the generation of diacyl-PE and PC revealed that this enzyme is only required for the promastigote but not amastigote stage [18]. These findings indicate that intracellular amastigotes can survive and proliferate without de novo PC synthesis by relying on the uptake/ remodeling of host lipids.
Like with CEPT, chromosomal null mutants for EPCT cannot lose the complementing pXNG4-EPCT episome in culture, suggesting that de novo PE synthesis is required for the fastreplicating promastigotes (Fig 4). When the GFP-low cells were separated from GCV-treated epct �+ pXNG4-EPCT promastigotes by FACS, they quickly regained GFP expression, suggesting that these cells can only survive and proliferate in the presence of GFP-high cells (S5 Fig). Leishmania parasites are known to produce and exchange extracellular vesicles among individual cells [50] which may contribute to the survival of epct �+ pXNG4-EPCT parasites after GCV treatment.
The fact that EPCT is indispensable for amastigotes is intriguing, as amastigotes can acquire enough PC, which is far more abundant than PE [18]. One possibility is that the PE scavenging pathway, either direct uptake or remodeling of host lipids, is insufficient and amastigotes need certain level of de novo synthesis to meet the demand for PE. Alternatively, because of its central position in the Kennedy pathway (Fig 1), a loss of EPCT will not only abolish the de novo synthesis of PME and diacyl-PE, but also negatively affect the production of PC (from PE Nmethylation) and PS (from PE interconversion). In comparison to EPT, CPCT and CEPT, disruption of EPCT would result in a more pleiotropic effect on the synthesis of multiple classes of glycerophospholipids (Fig 1).
We observed a significant virulence attenuation from EPCT overexpression as parasites with pXNG4-ECPT or pGEM-EPCT showed delayed lesion progression and cell replication in mice (Figs 5 and S8). Such defects were not found with the episomal overexpression EPT, CPCT, CEPT, or FPPS (an essential enzyme required for sterol synthesis) [14,18,48,49]. In mammalian cells, endogenous EPCT is reported to have a bimodal distribution between the cytosol and smooth endoplasmic reticulum (ER) [51]. EPCT generates CDP-EtN for EPT and CEPT to synthesize PME and diacyl-PE, respectively (Figs 1 and 2). Both EPT and CEPT are localized in the ER [14,18]. Our results indicate that EPCT overexpression caused reduction in PME but had little effect on diacyl-PE or PC (Figs 7, S9 and S10). Perhaps EPCT overexpression (mainly in the cytosol) led to increased production of CDP-EtN in cytosol but not ER which reduced PE synthesis. Alternatively, the elevated level of CDP-EtN may cause substrate inhibition when substrate concentrations exceed the optimum level for certain enzymes which hinders product release [52]. We did not detect any obvious alteration in cell volume/shape or compensatory changes in the lipid composition of EPCT over-expressors based on our lipidomic analysis of PE, PC, PI and IPC. Effects of EPCT overexpression on neutral lipids (e.g., sterols, DAG and TAG) remain to be determined.
It is not clear whether the altered PME synthesis in EPCT overexpressing cells is responsible for their heightened sensitivity to starvation or acidic pH, although these defects likely contribute to the virulence attenuation in mice. Our previous report on L. major ept �mutants indicate that while these mutants are largely devoid of PME, they did not exhibit the same stress response defects as EPCT overexpressing cells [14]. It is possible that the increased cytosolic concentration of CDP-EtN, coupled with the depletion of EtN-P and CTP, causes cytotoxicity when parasite encounter starvation or acidic stress. Another possibility is that reduced PME synthesis compromised macroautophagy (a process required the conjugation of PE to ATG8 [9]) in EPCT overexpressing cells, resulting in hypersensitivity to starvation. We did not detect significant changes in mitochondrial superoxide production from EPCT under-or overexpression, suggesting that mitochondrial PE synthesis is not affected (S13 Fig). Consistent with the detrimental effect of EPCT overexpression on amastigote growth, we found a steady decline of pXNG4-EPCT copy number in WT +pXNG4-EPCT and EPCT+/-+-pXNG4-EPCT amastigotes (Fig 5C). GCV treatment of mice caused further reduction in plasmid level as expected (Fig 5C). Curiously, neither EPCT+/-+pXNG4-EPCT amastigotes nor WT +pXNG4-EPCT amastigotes could completely lose the episome like EPCT+/-+-pXNG4-EPCT promastigotes in culture after GCV treatment (Figs 4A and 5C). Our previous studies on farnesyl pyrophosphate synthase (FPPS) and CEPT have demonstrated that heterozygous mutants of those genes lost most of their respective episome (<0.2 copy per cell for FPPS+/− +pXNG4-FPPS and CEPT+/− +pXNG4-CEPT) within 6 weeks post infection [18,48]. It is possible that the slow replication rates of EPCT+/-+pXNG4-EPCT and WT +-pXNG4-EPCT amastigotes prevented complete plasmid loss.
For epct �+ pXNG4-EPCT amastigotes, they retained 12-24 plasmids per cell throughout the course of infection, despite the cytotoxic effects from GCV and EPCT overexpression (Fig 5C). We also detected elevated EPCT transcript from epct �+ pXNG4-EPCT amastigotes isolated from GCV-treated mice (Fig 6C). These findings suggest that the need for EPCT outweighs the negative impacts from GCV and EPCT overexpression for L. major during the mammalian stage. However, it is important to remember that GCV treatment of mice only lasted for two weeks, and the level of GCV might be below the minimum effective concentration in the later stage of infection. In addition, plasmid copy number and EPCT transcript level may not fully reflect the EPCT protein level.
While the alteration of EPCT expression had little impact on the cellular level of lipophosphoglycan (LPG), it did significantly reduce the expression of GP63, a zinc-dependent metalloprotease that plays pivotal role Leishmania infection through proteolytic cleavage of host complement proteins and subversion of macrophage signaling [53][54][55][56]. Synthesis of GP63 requires PE to serve as a donor to generate the EtN-P linkage between protein and GPI-anchor [7,57]. It is not known whether a specific subset of PE (PME or di-acyl PE) is preferentially used in GP63 synthesis. As illustrated in Fig 8A and 8C, EPCT+/-parasites contained only 10-15% of WT-level GP63, and the episomal expression of EPCT did not rescue this defect. These results suggest that EPCT must be expressed from its endogenous locus to fully support GP63 synthesis, raising the possibility that endogenously expressed EPCT is modified or compartmentalized differently from episomally expressed EPCT, as previously reported for other leishmanial proteins [58].
In summary, our study establishes EPCT as the most important enzyme in the Kennedy pathway. It is crucial for L. major survival during the promastigote and amastigote stages. Overexpression of EPCT alters lipid homeostasis and stress response, leading to severely attenuated virulence. Future studies will investigate how intracellular amastigotes balance de novo synthesis with scavenging to optimize their long-term survival and evaluate the potential of EPCT as a new anti-Leishmania target.

Ethics statement
All procedures involving live mice were performed as per approved protocol by the Animal Care and Use Committee at Texas Tech University (PHS Approved Animal Welfare Assurance No A3629-01).

Leishmania culture and genetic manipulations
Leishmania major strain FV1 (MHOM/IL/81/Friedlin) promastigotes were cultivated at 27˚C in a complete M199 medium (M199 with 10% heat inactivated fetal bovine serum and other supplements, pH 7.4) [59]. To monitor growth, culture densities were measured daily by hemocytometer counting. Log phase promastigotes refer to replicative parasites at densities <1.0 x 10 7 cells/ml, and stationary phase promastigotes refer to non-replicative parasites >2.0 x 10 7 cells/ml.
To delete chromosomal EPCT alleles, the upstream and downstream flanking sequences (~1 Kb each) of EPCT were amplified by PCR and cloned in the pUC18 vector. Genes conferring resistance to blasticidin (BSD) and puromycin (PAC) were then cloned between the upstream and downstream flanking sequences to generate pUC18-KO-EPCT:BSD and pUC18-KO-EPCT:PAC, respectively. To generate the EPCT+/-heterozygotes (ΔEPCT::BSD/ EPCT), wild type (WT) L. major promastigotes were transfected with linearized BSD knockout fragment (derived from pUC18-KO-EPCT:BSD) by electroporation and transfectants showing resistance to blasticidin were selected and later confirmed to be EPCT+/-by Southern blot as previously described [18]. To delete the second chromosomal allele of EPCT, we used an episome assisted approach as previously described [46]. First, the EPCT open reading frame (ORF) was cloned into the pXNG4 vector to generate pXNG4-EPCT and the resulting plasmid was introduced into EPCT+/-parasites. The resulting EPCT+/-+pXNG4-EPCT cell lines were then transfected with linearized PAC knockout fragment (derived from pUC18-KO-EPCT: PAC) and selected with 15 μg/ml of blasticidin, 15 μg/ml of puromycin and 150 μg/ml of nourseothricin. The resulting EPCT chromosomal null mutants with pXNG4-EPCT (ΔEPCT::BSD/ ΔEPCT::PAC + pXNG4-EPCT or epct �+ pXNG4-EPCT) were validated by Southern blot as described in Figs 3, S3 and S4. For EPCT activity assay and localization studies, the EPCT ORF was into the pXG vector or pXG-GFP' vector to generate pXG-EPCT or pXG-GFP-EPCT respectively, followed by their introduction into WT L. major promastigotes by electroporation. Finally, a pGEM-5'-Phleo-DST IR-EPCT-3' construct was generated by cloning the EPCT ORF along with its upstream-and downstream flanking sequences (~1 Kb each) into a pGEM vector [60]. The resulting plasmid was introduced into EPCT+/-to drive EPCT overexpression in the presence of its flanking sequences.

Sub-cellular fractionation
Promastigotes were fractionated following a protocol adapted from T. cruzi work [61]. Briefly, mid-log cells were harvested by centrifugation and washed in PBS. Cells were resuspended in a lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% v/v Tween-20, 2 x protease inhibitor) at 2 x 10 8 cells/ml. After 15 min incubation on ice (vortex every 2 min), cell lysates were cleared (1000g, 10 min, 4˚C) and subjected to sequential centrifugation at 10,000g (10 min, 4˚C) and then 100,000g (60 min, 4˚C) to isolate the cytosolic fraction. Pellets from 10,000g centrifugation (large granules) and 100,000g centrifugation (microsomal fraction) were resuspended in lysis buffer to the same volume as that of the cytosolic fraction. Western blot was performed as described below.

Fluorescence microscopy and western blot
An Olympus BX51 Upright Epifluorescence Microscope equipped with a digital camera was used to visualize the localization of GFP-EPCT and GP63 as previously described [14]. For GFP-EPCT, WT + pXG-GFP-EPCT promastigotes were attached to poly-L-lysine coated coverslips and fixed with 3.7% paraformaldehyde prior to visualization. GP63 staining of unpermeabilized parasites was performed using an anti-GP63 monoclonal antibody (#96/126, 1:500, Abcam Inc.) at ambient temperature for 30 min, followed by washing and incubation with a goat-anti-mouse-IgG-Texas Red antibody (1:500) for 30 min. DNA staining was performed using Hoechst 33342 (2.5 μg/ml) for 10 min. All images were acquired using the same exposure times (GFP: 60 ms, DNA: 40 ms, DIC: 20 ms) and processed with the same settings in Adobe Photoshop. Because the anti-GP63 monoclonal antibody (#96/126) does not recognized denatured GP63 on western blot, we determined the cellular level of GP63 by dividing the integrated density of GP63 fluorescence (minus background) with the integrated density of Hoechst (minus background). For each cell type, 200 cells were analyzed, and the average of GP63/Hoechst was determined and normalized to WT (as 1.0).

Promastigote essentiality assay
EPCT+/-+pXNG4-EPCT and epct �+ pXNG4-EPCT promastigotes were inoculated in complete M199 media at 1.0 x 10 5 cells/ml in the presence or absence of 50 μg/ml of GCV (the negative selection agent) or 150 μg/ml of nourseothricin (the positive selection agent). Every three days, cells were reinoculated into fresh media with the same negative or positive selection agents and percentages of GFP-high cells for each passage were determined by flow cytometry using an Attune NxT Acoustic Flow Cytometer. After 14 passages, GFP-low cells were isolated

PLOS PATHOGENS
De novo PE synthesis is essential in Leishmania from EPCT+/-+pXNG4-EPCT and epct �+ pXNG4-EPCT by fluorescence-activated cell sorting (FACS). Individual clones were then isolated via serial dilution in 96-well plates and expanded in the presence of GCV and absence of nourseothricin. The GFP levels and pXNG4-EPCT plasmid copy numbers of selected clones were determined by flow cytometry and qPCR, respectively.

Mouse footpad infection and in vivo GCV treatment
Female BALB/c mice (7-8 weeks old) were purchased from Charles River Laboratories International (Wilmington, MA). To determine whether EPCT is required during the intracellular amastigote stage, day 3 stationary phase promastigotes were injected into the left hind footpad of BALB/c mice (1.0 x 10 6 cells/mouse, 10 mice per group). For each group, starting from day one post infection, one half of the mice received GCV at 7.5 mg/kg/day for 14 consecutive days (0.5 ml each, intraperitoneal injection), while the other half (control group) received equivalent volume of sterile PBS. Footpad lesions were measured weekly using a Vernier caliper after anesthetization with isoflurane (air flow rate: 0.3-0.5 ml/hour). Mice were euthanized through controlled flow of CO 2 asphyxiation when lesions reached 2.5 mm (humane endpoint) or upon the onset of secondary infections. Parasite numbers in infected footpads were determined by limiting dilution assay [63] and qPCR as described below.

Quantitative PCR (qPCR) analysis
To determine parasite loads in infected footpads, genomic DNA was extracted from footpad homogenate and qPCR reactions were run in triplicates using primers targeting the 28S rRNA gene of L. major [18]. Cycle threshold (Ct) values were determined from melt curve analysis. A standard curve of Ct values was generated using serially diluted genomic DNA samples from L. major promastigotes (from 0.1 cell/reaction to 10 5 cells/reaction) and Ct values >30 were considered negative. Parasite numbers in footpad samples were calculated from their Ct values using the standard curve. Control reactions included sterile water and DNA extracted from uninfected mouse liver.
To determine pXNG4-EPCT plasmid levels in promastigotes and lesion-derived amastigotes, a similar standard curve was generated using serially diluting pXNG4-EPCT plasmid DNA (from 0.1 copy/reaction to 10 5 copies/reaction) and primers targeting the GFP region. qPCR was performed with the same set of primers on DNA samples and the average plasmid copy number per cell was determined by dividing total plasmid copy number with total parasite number based on Ct values.
To determine EPCT transcript levels, total RNA was extracted from promastigotes or lesion-derived amastigotes and converted into cDNA using a high-capacity reverse transcription kit (Bio-Rad), followed by qPCR using primers targeting EPCT or α-tubulin coding sequences. The relative expression level of EPCT was normalized to that of α-tubulin using the 2 −ΔΔ(Ct) method [64]. Control reactions were carried out without leishmanial RNA and without reverse transcriptase. Vantage TSQ instrument applying precursor ion scan of m/z 196 for PE, precursor ion scan of m/z 241 for PI and IPC in the negative-ion mode, and precursor ion scan of m/z 184 for PC in the positive-ion mode [66]. Individual lipid species and their structures were also confirmed by high resolution mass spectrometry performed on a Thermo LTQ Obitrap Velos with a resolution of 100,000 (at m/z 400). All lipidomic analyses were performed five times.

Leishmania stress response assays and determination of mitochondrial ROS level
For heat tolerance, L. major promastigotes grown in complete M199 media were incubated at either 27˚C or 37˚C. To test their sensitivity to acidic pH, promastigotes were transferred to a pH5.0 medium (same as the complete M199 medium except that the pH was adjusted to 5.0 using hydrochloric acid). For starvation response, promastigotes were transferred to PBS (pH 7.4). Cell viability was determined at the indicated times by flow cytometry after staining with 5 μg/ml of propidium iodide. Parasite growth was monitored by cell counting using a hemocytometer.
Mitochondrial superoxide accumulation was determined as described previously [34]. Briefly, log phase promastigotes were transferred to PBS (pH 7.4) and stained with 5 μM of MitoSox Red. After 25 min incubation at 27˚C, the mean fluorescence intensity (MFI) for each sample was measured by flow cytometry.

Statistical analysis
Unless otherwise specified, experiments were repeated three to five times. Symbols or bars represent averaged values and error bars represent standard deviations. Differences between groups were assessed by one-way ANOVA (for three or more groups) using the Sigmaplot 13.0 software (Systat Software Inc, San Jose, CA) or student t test (for two groups). P values were grouped in all figures (***: p < 0.001, **: p < 0.01, *: p < 0.05).  Table. List of oligonucleotides used in this study. Sequences in lowercase represent restriction enzyme sites. (PDF)