Phosphatidylethanolamine and phosphatidylcholine biosynthesis by the Kennedy pathway occurs at different sites in Trypanosoma brucei

Phosphatidylethanolamine (PE) and phosphatidylcholine (PC) are among the most abundant phospholipids in biological membranes. In many eukaryotes, the CDP-ethanolamine and CDP-choline branches of the Kennedy pathway represent major and often essential routes for the production of PE and PC, with ethanolamine and choline/ethanolamine phosphotransferases (EPT and CEPT, respectively) catalysing the last reactions in the respective pathways. Although the site of PE and PC synthesis is commonly known to be the endoplasmic reticulum (ER), detailed information on the localization of the different phosphotransferases is lacking. In the unicellular parasite, Trypanosoma brucei, both branches of the Kennedy pathway are essential for cell growth in culture. We have previously reported that T. brucei EPT (TbEPT) catalyses the production of ether-type PE molecular species while T. brucei CEPT (TbCEPT) synthesizes diacyl-type PE and PC molecular species. We now show that the two enzymes localize to different sub-compartments of the ER. By expressing a series of tagged forms of the two enzymes in T. brucei parasites, in combination with sub-cellular fractionation and enzyme activity measurements, TbEPT was found exclusively in the perinuclear ER, a distinct area located close to but distinct from the nuclear membrane. In contrast, TbCEPT was detected in the bulk ER.


Results and Discussion
According to their in vitro and in vivo substrate specificities, EPT/CEPT and CEPT/CPT enzymes catalyse the last step in the two branches of the Kennedy pathway, leading to the formation of the most abundant eukaryotic glycerophospholipid classes PE and PC, respectively. Although scattered data about the sub-cellular localization of CPT, CEPT and EPT in mammalian cells, yeast and Toxoplasma is available [23][24][25][26][27] , the results are conflicting and provide no clear information on the site(s) of de novo synthesis of PC and PE. In particular, a detailed study to localize these enzymes in a single cell is lacking. In T. brucei parasites, PE and PC synthesis by the Kennedy pathway is essential and the substrate specificities of TbEPT and TbCEPT have been characterized using RNAi cell lines to down-regulate the enzymes individually 20,38 . In the present study, we have analysed the localization of TbCEPT and TbEPT by immunofluorescence microscopy of differently tagged forms of the enzymes, and by sub-cellular fractionation and enzyme activity measurements.

Localization of tagged forms of TbEPT and TbCEPT. Based on its previously reported specificity
for the synthesis of ether-type PE molecular species in T. brucei procyclic forms 20 , we expected TbEPT to localize in glycosomes, which are the site of ether-type (alk-1-enyl-acylglycerol and alkyl-acylglycerol) precursor synthesis 40 . In contrast, we found that expression of a C-terminally 3xHA-tagged version of TbEPT (TbEPT-3xHA) showed a perinuclear staining, but no co-localization with the glycosomal marker aldolase ( Fig. 2A, top panels). A similar staining was also observed in parasites expressing an N-terminally 2xcmyc-tagged TbEPT or an in situ C-terminally 3xHA-tagged form of the enzyme ( Fig. 2A, middle and bottom panels, respectively). Remarkably, we observed little or no co-localization of TbEPT with the ER marker BiP. The exclusive perinuclear localization of TbEPT was independent of the position of the tag (N-or C-terminal), the nature of the tag (HA, cmyc or GFP (see below)) or the method used to express the tagged protein (inducible versus stable expression), and was seen in > 95% of cells, indicating that the results likely reflect the genuine localization of TbEPT. Together, these results indicate that TbEPT is not localized in the bulk ER, i.e. the site of PE formation in other eukaryotes, but in a sub-compartment in close proximity to the nuclear membrane.
In addition, we analysed the sub-cellular localization of TbCEPT in T. brucei procyclic forms by expressing an N-terminally 3xcmyc-tagged version of the enzyme. The results show an essentially complete co-localization of TbCEPT with the ER marker BiP (Fig. 2B).
Because of the close apposition of the nuclear membrane and the region of the ER closest to the nuclear membrane, the unique perinuclear staining pattern of TbEPT was further investigated. To distinguish between nuclear and ER membranes, procyclic forms expressing TbEPT and TbCEPT whose C-termini were fused to GFP were co-stained with an antibody (MAb414) directed against nuclear pore  Fig. 2A). However, although staining with MAb414 revealed a similar ring-like staining pattern, little co-localization with TbEPT-GFP was observed (Fig. 3A). In addition, in agreement with the staining of 3xcmyc-TbCEPT (Fig. 2B), TbCEPT-GFP showed staining of the bulk ER. Again, as for TbEPT-GFP, little co-localization was observed between TbCEPT-GFP and MAb414.
Localization of TbEPT and TbCEPT in a single cell. To localize both TbEPT and TbCEPT in a single parasite, TbEPT-GFP and 3xcmyc-TbCEPT were co-expressed and analysed by (immuno)fluorescence microscopy (Fig. 3B). The results show that the two enzymes co-localize in the ER region proximal to the nucleus, however, while TbEPT is confined to the perinuclear region only, TbCEPT is distributed throughout the ER network. Together, our results demonstrate that TbEPT is located exclusively in an ER sub-domain in close proximity but distinct from the nuclear membrane. In contrast, TbCEPT reveals a staining pattern typical for bulk ER, although a sub-fraction co-localizes with TbEPT.

Sub-cellular fractionation and enzyme activity measurements. The sub-cellular localization of
TbEPT and TbCEPT was also investigated using cell fractionation of different trypanosome cell lines. To compare the sub-cellular distribution of the tagged enzymes with endogenous TbCEPT activity, a mixed culture of T. brucei procyclic forms expressing either TbEPT-3xHA or 3xcmyc-TbCEPT was fractionated by differential centrifugation (Fig. 4A). Subsequently, in each sub-cellular fraction the presence of the tagged enzymes was analysed by SDS-PAGE and immunoblotting (Fig. 4B) and the enzymatic activity  of TbCEPT was determined (Fig. 4C). The results show the highest enrichment of 3xcmyc-TbCEPT in the microsomal fraction, while TbEPT-3xHA was enriched the most in the membranes and organelles fraction (Fig. 4B). These findings support the immunofluorescence results showing that the two tagged enzymes localize differently in T. brucei. The distribution of marker proteins for mitochondria (mtHSP70), ER (BiP) and glycosomes (aldolase) in the different fractions confirms previous results 42 and shows that the microsomal fraction is depleted of mitochondrial and nuclear (MAb414) membranes, while it contains the markers for ER and glycosomes (Fig. 4B). Determination of TbCEPT activity using [ 14 C] CDP-choline as substrate in the fractions after differential centrifugation showed highest [ 14 C]PC formation in the microsomal fraction (Fig. 4C). This result is in line with the enrichment of 3xcmyc-TbCEPT in the same fraction (Fig. 4B). Since TbCEPT activity was identical in parasites before and after induction of 3xcmyc-TbCEPT expression (results not shown), indicating that the tagged enzyme doesn't contribute significantly to endogenous TbCEPT activity or that 3xcmyc-TbCEPT while correctly localized is not active, the measured in vitro activity represents endogenous TbCEPT activity.
To determine the sub-cellular distribution of endogenous TbEPT activity, an RNAi cell line allowing down-regulation of TbCEPT expression 20 was used for the following reason. TbCEPT shows dual substrate activity, i.e. it not only catalyses the production of PC but also PE 20 and, thus, its activity contributes to [ 14 C]PE formation when using [ 14 C]CDP-ethanolamine as substrate. Therefore, to allow determination of TbEPT activity in the absence of TbCEPT, TbCEPT mRNA was silenced using tetracycline-inducible RNAi. Absence of TbCEPT activity was confirmed by the inability of TbCEPT-depleted parasites to form [ 3 H]PC using [ 3 H]choline as substrate (Fig. 4D, upper panels). In contrast, TbEPT-mediated formation of [ 3 H]PE using [ 3 H]ethanolamine as substrate was unchanged in TbCEPT-depleted parasites (Fig. 4D, lower panels). These results are in line with previous data 20 . Analysis of [ 14 C]PE formation in the fractions after differential centrifugation from TbCEPT-depleted parasites revealed highest TbEPT activity in the membranes and organelles and microsomal fractions (Fig. 4E). The result is in good agreement with the enrichment of TbEPT-3xHA in these fractions (Fig. 4B).
Further fractionation of the membranes and organelles fraction, containing 3xcmyc-TbCEPT and TbEPT-3xHA, using a linear 18-50% Nycodenz gradient (Fig. 5A) again revealed a different distribution pattern for the two enzymes (Fig. 5B). TbEPT-3xHA co-fractionated most closely with BiP and mtHSP70 at the top of the gradient, while 3xcmyc-TbCEPT showed a dual enrichment in the upper (fractions 1-3) and middle fractions (fractions [11][12][13]. While the distribution of 3xcmyc-CEPT partly overlapped with MAb14 and aldolase, the fractions showing the highest enrichment of the respective enzymes are clearly different. These results are consistent with the immunofluorescence pictures showing that TbEPT and TbCEPT are localized differently and are not enriched in glycosomal or nuclear membranes (Figs 2  and 3).

Conclusions
In T. brucei parasites, TbEPT and TbCEPT catalyse the last step in the CDP-ethanolamine and CDP-choline branches of the Kennedy pathway and, thus, determine the site(s) of synthesis of the two major glycerophospholipid classes PE and PC, respectively. Both pathways are essential for growth of parasites in culture 20,38 . Our work using immunofluorescence microscopy and immunoblotting to localize differently tagged forms of TbEPT and TbCEPT, in combination with sub-cellular fractionation and activity measurements to determine the distribution of the endogenous enzymes, showed that both TbEPT and TbCEPT are present in the ER. However, TbEPT is confined to the perinuclear region of the compartment while TbCEPT is distributed throughout the entire ER network. The reactions catalysed by TbEPT and TbCEPT, both of which are predicted to have multiple transmembrane domains and thus are tightly membrane-bound, combine water-soluble CDP-activated precursors with hydrophobic substrates, i.e. DAG and/or AAG. Although the site(s) of DAG production in trypanosomes has not been studied, it is generally believed that DAG is available in all membranes. In contrast, production of AAG in T. brucei has been localized in glycosomes 40 . Our observation that TbEPT is localized exclusively in the perinuclear ER implies that the hydrophobic substrate of the reaction, AAG, is available in this sub-compartment, or in very close proximity. At present, it is not known if a sub-population of glycosomes associates with the perinuclear ER or if AAG can be trafficked from glycosomes to other organelles. Mechanisms to transfer AAG from glycosomes to the perinuclear ER may involve protein-or vesicle-mediated transport, or direct membrane contact sites. EPT and CEPT/CPT enzymes are present in essentially all eukaryotic cells. It will be interesting to study if the two enzymes show a similar intracellular distribution in other unicellular organisms, or mammalian cells. Trypanosomes and culture conditions. T. brucei procyclic form strain 427 was cultured at 27 °C in SDM-79 43 containing 5% heat-inactivated fetal bovine serum (Invitrogen). T. brucei procyclic form strain 29-13 co-expressing a T7 polymerase and a tetracycline repressor 44 (obtained from Paul Englund, Baltimore, MD) was cultured at 27 °C in SDM-79 containing 10% heat-inactivated fetal bovine serum, 25 μ g/mL hygromycin, and 15 μ g/mL G418.

Stable and transient transfection of trypanosomes.
Trypanosomes were harvested at mid-log phase, washed once in phosphate-buffered saline (PBS; 137 mM sodium chloride, 2.7 mM potassium chloride, 10 mM disodium phosphate, 2 mM monopotassium phosphate, pH 7.4) and resuspended in 100 μ l TbBSF buffer (90 mM sodium phosphate, 5 mM potassium chloride, 0.15 mM calcium chloride, 50 mM HEPES, pH 7.3) 49 previously mixed with 10 μ g of plasmid or 4 μ g of purified PCR product (in situ tag). Electroporation was performed in 100 μ l nucleocuvettes using Lonza 4D Nucleofector System (pulse code FI-115, "Primary Cell P3" solution). GFP constructs were expressed transiently (circular plasmids were transfected into 427 strain and slides were prepared 18-24 hours after transfection) whereas HA and cmyc constructs were stably transfected into 29-13 strain (plasmids were linearized with NotI prior to transfection). Clones were obtained by limited dilution and selected in the presence of the corresponding antibiotic, 2 μ g/ml puromycin for HA and cmyc constructs, or 15 μ g/ml G418 for the in situ construct. Proper integration of the constructs was confirmed by PCR using primers binding upstream of the Scientific RepoRts | 5:16787 | DOI: 1 0. 10 38 /s re p1 67 87 recombination site and downstream of the inserted gene. Expression of HA-or cmyc-tagged proteins was induced by addition of 1 μ g/ml tetracycline to the culture medium.
Fluorescence microscopy. Trypanosomes were collected and washed once in cold PBS before being allowed to adhere onto microscopy slides (Thermo Scientific) for 10 min. Cells were fixed with 4% paraformaldehyde, washed once with cold PBS and permeabilized with 0.2% (w/v) Triton X-100 in PBS. Blocking was performed with 2% bovine serum albumin in PBS (blocking solution) for 30 min, followed by 45 min incubation of first antibody diluted in blocking solution. The following antibodies were used: mouse monoclonal α -cmyc 9E10 (Santa Cruz Biotechnology; diluted 1:200), mouse monoclonal α -HA.11 16B12 (Covance, diluted 1:250), rabbit α -BiP (a kind gift of James D. Bangs, University of Buffalo, Buffalo, NY; diluted 1:2500), rabbit α -aldolase (a kind gift of Paul Michels, University of Edinburgh, Edinburgh, Scotland; diluted 1:5000), and mouse α -nuclear pore complex proteins (MAb414, Covance; diluted 1:5000). After washing, the corresponding secondary antibodies goat anti-mouse and anti-rabbit AlexaFluor 594 and 488 (Invitrogen; diluted 1:800) in blocking solution were added for 45 min. For the preparation of GFP-expressing cells, slides were kept protected from light during the whole procedure. After washing, cells were mounted with Vectashield containing 4′ ,6-diamidino-2-phenylindole (DAPI, Vector Laboratories). Fluorescence microscopy was performed with a Leica DMI6000 B inverted microscope and pictures were acquired, processed and 3D-deconvolved with LAS X software (Leica Microsystems CMS GmbH).
Cell fractionation. Following a previously published procedure 50 , parasites were fractionated into a crude microsomes and a membranes and organelles fraction, the latter of which was subsequently fractionated using a Nycodenz gradient. Briefly, 2-4 × 10 10 cells were harvested by centrifugation and subjected to nitrogen cavitation at isotonic conditions. Membranes and organelles (containing most organelles) were separated from cytosol and microsomes (crude microsomes) by differential sedimentation. The membranes and organelles fraction was subjected to DNase treatment, resuspended in 80% Nycodenz, loaded underneath a 18-50% (w/v) linear Nycodenz gradient using a syringe and centrifuged for 45 min in a SW28 rotor at 27000 rpm at 4 °C. Fractions (20 × 2 ml) were collected from the top of the gradient. The crude microsomal fraction was diluted 1:4 in HKM buffer (50 mM HEPES, 25 mM KCl; 5 mM MgCl 2 ·6 H 2 O) and a clearing spin was carried out (30 min at 4 °C at 30000 × g in a SS34 rotor) to eliminate residual mitochondrial vesicles. Microsomes were collected from this cleared supernatant by ultracentrifugation (2 h at 4 °C at 100000 × g in a SW28 rotor). The microsomal fraction has previously been called crude ER fraction 42 .
CEPT assay. 80 μ g of protein from sub-cellular fractions were incubated for 50 min at 27 °C in 100 μ l assay buffer (50 mM Tris pH 8.0, 10 mM MgCl 2 ·6 H 2 O, 0.005% (w/v) Tween20), supplemented with 5 nmoles CDP-choline, 0.9 nmoles [ 14 C]CDP-choline, 2.5 nmoles diacylglycerol and 2.5 nmoles alkyl-acylglycerol. After incubation, lipids were extracted twice by addition of 250 μ l 1-butanol, centrifugation and isolation of the organic phase. Thin layer chromatography (TLC). Lipid classes were separated on silica gel 60 plates (Merck) by one-dimensional TLC using a solvent system composed of chloroform: methanol: acetic acid: water (25:15:4:2, by vol.). Lipid standards were run on each plate alongside the samples to be analysed. Radioactive lipids were detected by scanning the air-dried plate with a radioisotope detector (Berthold Technologies) and quantified using the Rita Star software provided by the manufacturer.