Emerging Microbes & Infections 2021, VOL. 10 https://doi.org/10.1080/22221751.2020.1870413 ATP2, The essential P4-ATPase of malaria parasites, catalyzes lipid-stimulated ATP hydrolysis in complex with a Cdc50 β-subunit Anaïs Lamya,b, Ewerton Macarini-Bruzaferroa,c, Thibaud Dieudonné a,d, Alex Perálvarez-Marín e , Guillaume Lenoir a, Cédric Montigny a, Marc le Mairea and José Luis Vázquez-Ibar a a Université Paris-Saclay, CEA, CNRS, Institute for Integrative Biology of the Cell (I2BC), Gif-sur-Yvette, France; bDepartment of Medical Biochemistry and Biophysics, Umeå University, Umeå, Sweden; cDepartment of Clinical Medicine (FMUSP), University of São Paulo, São Paulo, Brazil; dDANDRITE, Nordic EMBL Partnership for Molecular Medicine, Department of Molecular Biology and Genetics, Aarhus University, Aarhus, Denmark; eBiophysics Unit, Department of Biochemistry and Molecular Biology, School of Medicine, Universitat Autònoma de Barcelona, Cerdanyola del Vallés, Spain ABSTRACT Gene targeting approaches have demonstrated the essential role for the malaria parasite of membrane transport proteins involved in lipid transport and in the maintenance of membrane lipid asymmetry, representing emerging oportunites for therapeutical intervention. This is the case of ATP2, a Plasmodium-encoded 4 P-type ATPase (P4- ATPase or lipid flippase), whose activity is completely irreplaceable during the asexual stages of the parasite. Moreover, a recent chemogenomic study has situated ATP2 as the possible target of two antimalarial drug candidates. In eukaryotes, P4-ATPases assure the asymmetric phospholipid distribution in membranes by translocating phospholipids from the outer to the inner leaflet. In this work, we have used a recombinantly-produced P. chabaudi ATP2 (PcATP2), to gain insights into the function and structural organization of this essential transporter. Our work demonstrates that PcATP2 associates with two of the three Plasmodium-encoded Cdc50 proteins: PcCdc50B and PcCdc50A. Purified PcATP2/PcCdc50B complex displays ATPase activity in the presence of either phosphatidylserine or phosphatidylethanolamine. In addition, this activity is upregulated by phosphatidylinositol 4- phosphate. Overall, our work describes the first biochemical characterization of a Plasmodium lipid flippase, a first step towards the understanding of the essential physiological role of this transporter and towards its validation as a potential antimalarial drug target. ARTICLE HISTORY Received 19 October 2020; Revised 21 December 2020; Accepted 24 December 2020 KEYWORDS Malaria; P4-ATPases; lipid flippase; PfATP2; membrane transport proteins; heterologous expression Introduction new drugs to supplement the existing therapeutic Malaria is a global health problem affecting over 200 arsenal. In fact, the combination of drugs acting on millions of people worldwide, and responsible in different targets and/or stages is the most efficient 2018 of nearly 4,05,000 deaths around the globe, strategy to combat malaria as it minimizes the devel- mostly children[1]. Malaria is caused by parasites of opment of new drug-resistance mechanisms [3]. the genus Plasmodium, a pathogen with a complex Recent efforts combining genetic disruption tools life-cycle between the transmission vector, the mos- and phenotypic analysis in human and rodent malaria quito Anopheles, and a vertebrate host. Five different parasites have revealed that ∼78% of Plasmodium- Plasmodium species infect humans, P. falciparum encoded membrane transport proteins (MTPs) are being the most virulent one, accounting for ∼90% of either essential or necessary for normal grow and the mortality caused by this disease. Over the last 15 development [4–6], thus confirming their therapeutic years, artemisinin-combined therapies (ACT) have potential [7]. Among the annotated 144 MTPs, permitted a substantial reduction of the mortality of P. falciparum encodes 13 P-type ATPases [7], most this disease. However, P. falciparum strains resistant of them refractory to gene deletion [8]. P-type to current treatments, particularly to artemisinin, ATPases are a large family of primary transporters have already appeared in different regions of south that utilize the energy from ATP hydrolysis to pump Asia, representing a serious threat to eradicate malaria cations (subfamilies P1, P2 and P3)[9], lipids (subfam- as no effective vaccine is yet available [2]. The fight ily P4)[10], polyamines (subfamily P5B)[11], or even against malaria thus requires a continuous supply of to discolase single transmembrane domains (P5A CONTACT José Luis Vázquez-Ibar jose-luis.vazquez-ibar@i2bc.paris-saclay.fr Université Paris-Saclay, CEA, CNRS, Institute for Integrative Biology of the Cell (I2BC), Gif-sur-Yvette 91198, France Supplemental data for this article can be accessed https://doi.org/10.1080/22221751.2020.1870413 © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group, on behalf of Shanghai Shangyixun Cultural Communication Co., Ltd This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrest- ricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Emerging Microbes & Infections 133 subfamily)[12]. In Plasmodium, P-type ATPases blod stages [6]. A recent study in P. yoelii has shown maintain ion homeostasis during its intracellular that Cdc50A interacts and stabilizes the GCβ/Cdc50A life-cycle [13]. The P. falciparum Na+/H+ P-type complex, mandatory for the gliding motility and mid- ATPase (PfATP4) has been identified as the target of gut traversal of the ookinetes in the mosquito vector the antimalarial family of drugs spiroindolones [14]. and, consequently, for parasite’s transmission [29]. The P. falciparum homologue of the sarco/endoplas- Also, its related Cdc50A homolog in Toxoplasma gondii mic reticulum Ca2+ ATPase (SERCA) pump is mandatory for bringing its GCβ partner to the plasma (PfATP6) was initially proposed as an artemisinin tar- membrane prior cellular egression [30]. get [15], although studies using recombinant PfATP6 ATP2 and the rest of Plasmodium-encoded P4- did not support this [16]. Recently, gene-targeting ATPases still remain as putative transporters waiting approaches have also uncovered the important role for functional annotation. In addition, and despite for parasite’s progression of the P4 subfamily, also the importance of the P4-ATPase/Cdc50 association, known as lipid flippases [4–6]. In eukaryotes, P4- to date, the identity of the Cdc50-interacting partners ATPases maintain the asymmetric distribution of of ATP2 have not been revealed, neither in the parasite phospholipids in membranes by translocating phos- nor in vitro using recombinant proteins. Moreover, pholipids from the extracellular (or luminal) leaflet the biological role of such possible association is still to the cytoplasmic leaflet [10], key in important phys- unknown. Here, using recombinantly-produced pro- iological functions as vesicle budding and protein teins, we demonstrate that the Plasmodium chabaudi trafficking [17]. Moreover, new roles of these trans- ATP2 (PcATP2) forms heterodimers with two of the porters have emerged in recent years as membrane three Plasmodium-encoded Cdc50 proteins: PcCdc50B lipids are also implicated in cell-signalling events. In and PcCdc50A. Moreover, PcATP2/PcCdc50B dis- plants, the transport of lysophospholipids mediated played lipid-stimulated ATPase activity in the presence by a P4-ATPase is involved in root development and of two putative phospholipid substrates, and this size of stomatal apertures [18]. Also, the virulence of activity was slighly upregulated by the presence of some pathogens and drug disponibility have been cor- PI4P. We provide the first study of the function and related with the expression and activity of P4-ATPases structural organization of a Plasmodium lipid flippase, [19]. The genome of P. falciparum contains four puta- an emerging antimalarial target candidate. tive P4-ATPases: PfATP2, PfATP7, PfATP8 and PfATP11 [7], well conserved in Plasmodium species with the exception of PfATP11 [8]. In addition, Materials and methods P. falciparum encodes two P4-ATPase-like proteins Molecular cloning only present in the Apicomplexa phylum: GCα and GCβ, consisting of a N-terminal P4-ATPase-like domain Synthetic cDNAs encoding three putative ATP2 connected to a C-terminal guanylate cyclase domain. orthologs from P. falciparum (PF3D7_1219600), While there is some discrepancy with regard the vital P. berghei (PBANKA_1434800), and P. chabaudi role of ATP7 and ATP8 [4,5], all the studies agree that (PCAHS_1436800), and the three Cdc50 subunits the activites of PfATP2 and its ortholog in the mouse encoded per specie: P. falciparum (PF3D7_0719500, malaria parasite P. berguei are completely irreplaceable PF3D7_1133300, and PF3D7_1029400), P. berghei during the blood stages. Moreover, in a recent study (PBANKA_061700, PBANKA_091510, and [20], gene duplication of the gene encoding PfATP2 PBANKA_051340), and P. chabaudi (PCHAS_061870, was associated with drug-resistant phenotypes against PBANKA_091510, and PBANKA_051340) were two drugs antimalarial compounds. This eventually situ- ordered to Genscript (Piscataway, NJ). All the cDNAs ates PfATP2 either, as the target of these two antimalarial were codon-optimized by the provider for S. cerevisiae compounds or responsible of reducing the local concen- expression using the proprietary algorithm Optimum- tration of these drugs at the target site. GeneTM. In all cloning steps, we used the E. coli XL1- Most P4-ATPases form heterodimeric complexes Blue strain and the Luria–Bertani medium with the with members of the Cdc50/LEM3 protein family, appropriate antibiotic. and this association is essential for trafficking of the Single expression vectors: The cDNAs encoding the P4-ATPase/Cdc50 complex [21], and for activity [22]. three ATP2 orthologs were cloned in the pYeDP60 The recent Cryo-EM structures of four P4-ATPases in expression vector [31] between the EcoRI and NotI complex with their respective Cdc50 subunits have restriction sites, yielding the pYeDP60-ATP2-TEV- revealed the structural basis of P4-ATPase/Cdc50 BAD plasmid that contains in downstream position association [23–28] P. falciparum encodes three puta- the tobacco etch virus (TEV) protease cleavage-site tive Cdc50 proteins (annotated as Cdc50A, Cdc50B coding sequence followed by the biotin acceptor and Cdc50C) also well conserved in Plasmodium domain (BAD) coding sequence. To obtain the N- species. Gene-disruption of either of these putative terminal BAD tagged ATP2 constructs, pYeDP60- Cdc50 proteins affects parasite’s progression in the BAD-TEV-ATP2, we amplified by PCR the ATP2 134 A. Lamy et al. sequences incorporating the PmeI and SacI restric- (w/v) BactoTM Casamino Acids, 0.7% (w/v) Yeast tions sites at both, the 5′ and 3′ ends, and cloned nitrogen base (without amino acids and with into the pYeDP60 vector. The PcATP2-D596N ammonium sulphate), 2% (w/v) glucose, 20 µg/mL mutant was obtained using the QuikChange II XL adenine and 20 µg/mL uracil) for 24 h at 28°C. Cells site-directed mutagenesis kit (Agilent). The nine were mixed with ∼1 μg of plasmid DNA and 100 μg Cdc50 subunits sequences were cloned in the of salmon sperm DNA (denatured at 100°C for pYeDP60 vector between the EcoRI and BamHI 5 min and cooled down on ice). After vortexing, restriction sites. The resulting pYeDP60-Cdc50- 500 μL of 40% (w/v) PEG 4000, 100 mM lithium acet- TEV-10xHis vectors contain the TEV site followed ate, 10 mM Tris-HCl pH 7.5 and 1 mM EDTA and by a 10xHis tag at the C-terminal end of the proteins. 20 μL of 1 M DTT was added. The tube was vortexed Co-expression vectors: pYedBAD-BAD-Drs2p/ again and left overnight at room temperature. The Cdc50p-His was cloned previously [31]. To construct next morning the tube was centrifuged at 350 × g for the other co-expression vectors we used a previous strat- 2 min, the supernatant was removed, and the cells egy. The fragments containing the cassette Promoter- were re-suspended in 100 μL of S6A minimal media Cdc50-TEV-His10-Terminator from each pYeDP60- (SAU without uracil) and plated on S6A-agar plates. Cdc50-TEV-10xHis vector were amplified by PCR, The plates were left in the incubator at 28°C for 3–5 incorporating the SbfI restriction site at both the 5′ days until the transformed colonies appeared. and the 3′ ends. The PCR fragments were then cloned at the SbfI site of the corresponding species-specific Protein expression in S. cerevisiae single-expression vectors pYeDP60-ATP2-TEV-BAD and pYeDP60-BAD-TEV-ATP2. Transformed yeast colonies were pre-cultured in 5 mL of S6A media or 24 h at 28°C. For small-scale expression tests, the pre-culture was diluted to Introduction of the superfolder Green OD600 of 0.2 in 20 mL of YPGE2X rich media (2% fluorescent protein (w/v) BactoTM Peptone, 2% (w/v), yeast extract, 1% The cDNA encoding the superfolder Green Fluor- (w/v) glucose and 2.7% (w/v) ethanol) and cultured escent Protein (GFP) was PCR amplified, introducing for 30 h at 28°C. Then, the cell culture was cooled a NotI site followed by the human Rhinovirus 3C down to 18°C, and 20 g/L of galactose was added to protease sequence (3C-protease) at the 5′ , and the induce protein expression for 18 h at 18°C. XmaI site at the 3′ . The PCR product was then cloned For large-scale cultures used for membrane frac- into the NotI and XmaI sites of both the single- tionation, the first pre-culture was diluted to a expression pYeDP60-PcATP2-TEV-BAD and the OD600 of 0.1 in 50 mL of S6A media and incubated co-expression pYeDP60-PcATP2-TEV-BAD/PcCdc50B- at 28°C for another 24 h. Then, this second pre-culture 10xHis vectors to obtain the pYeDP60-PcATP2-3C_Pro- was diluted to an OD600 of 0.05 in 500 mL of YPGE2X tease-GFP-TEV-BAD and the pYeDP60-PcATP2- media. The cells were grown for 36 h at 28°C to con- 3C_Protease-GFP-TEV-BAD/PcCdc50B-10xHis vectors. sume the majority of glucose. The culture flasks were Using an identical cloning strategy, we also made the cooled down to 18°C, and 2% (w/v) of galactose same constructs but without the BAD domain by were added to induce protein expression. After 13 h adding a stop codon next to the XmaI site, leading at 18°C, a second addition of 2% (w/v) galactose was to the the pYeDP60-PcATP2-D596N-3C_Protease- done and the culture was left for five more hours. GFP/PcCdc50B-10xHis vector. To introduce the sGFP The cells were centrifuged at 4000 × g during 10 min at the C-terminal end of PcCdc50B, we amplified by at 4°C and washed two times with cold water. The PCR a DNA fragment encoding the 3C_protease fol- cell-pellet was weighted and re-suspended in 2 mL of lowed by the GFP with BamHI sites at both the 5′ and cold TEKS buffer (50 mM Tris-HCl pH 7.5, 1 mM the 3′ ens. The sequence was then cloned at the EDTA, 0.1 M KCl and 0.6 M sorbitol) per gram of BamH1 site of pYeDP60-PcCdc50B-TEV-10xHis vec- cells. After incubation at 4°C for 15 min, the cells tors, resulting in the pYeDP60-PcCdc50B-3C_pro- were centrifuged and the resulting pellet was flash- tease-GFP-10xHis vector. frozen in liquid nitrogen and stored at −80°C. Yeast transformation Membrane preparation The different proteins and protein complexes were To analyze the expression of the proteins induced in expressed in the Saccharomyces cerevisiae strain the small-scale cultures, a total membrane preparation W3031b Gal4-2 (α, leu2, his3, trp1::TRP1-GAL10- was done. A volume of each culture corresponding to GAL4, ura3, ade2-1, canr, cir+) [32]. For yeast trans- 10 optical density units at 600 nm was centrifuged at formation, 5 ml of non-transformed W3031b Gal4-2 800 × g for 10 min and 4°C. Cells were re-suspended strain was cultured in S6 AU minimal media (0.1% in 1 mL of ice-cold TEPI buffer (50 mM Tris-HCl Emerging Microbes & Infections 135 pH 6.8, 5 mM EDTA, 20 mM NaN3) supplemented in 5% (w/v) milk in PBS-Tween buffer, followed by a with SigmaFastTM EDTA-free Protease Inhibitor incubation with a second Goat Anti-Mouse IgG- Cocktail (PIC) tablets (Sigma-Aldrich) and 1 mM HRP conjugate antibody (Bio-Rad) diluted to 1: phenylmethylsulfonyl fluoride (PMSF, Sigma- 3000 in 5% (w/v) milk in PBS-Tween buffer. The Aldrich) and transferred into 1.5 mL Eppendorf ECL Western Blotting Detection kit (GE healthcare) tubes. After centrifugation at 1000 × g for 10 min was used for western blot revelation, and lumines- and 4°C, the pellet was re-suspended in 100 µL of cence was detected by a CDD camera. ice-cold TEPI buffer, and cells were broken after add- ing 100 µL of glass beads (0.5 mm of diameter) and Deglycosylation assays of PcCdc50 proteins vortexing for 25 min at 4°C. Then, TEPI buffer was added to a final volume of 1 mL and samples were cen- P3 membranes co-expressing PcATP2/PcCdc50A, trifuged for 5 min at 500 × g and 4°C. The supernatant PcATP2/PcCdc50B or Drs2p/Cdc50p were diluted to was then centrifuged at 100 000 × g for 90 min and 4° 4 mg/mL in HEPES-sucrose buffer. 20 µg of total C. The resulting membrane pellet was re-suspended protein were added to a glycoprotein denaturing directly in 200 µL of SDS-PAGE loading buffer con- buffer (0.5% (w/v) SDS, 40 mM DTT) and protein taining urea, heated at 30°C during 10 min, and sub- denaturation was performed at 100°C for 10 min in jected to SDS-PAGE and western blot analysis. a dry bath. Samples were cooled down on ice for Frozen cell-pellets from large-scale cultures were 5 min and spun down at 15,000 × g for 10 sec. 10 µL resuspended in 1 mL of TES buffer (50 mM Tris-HCl of denatured samples were enzymatically digested pH 7.5, 1 mM EDTA and 0.6 M sorbitol) per gram of with either, PGNase F or EndoH (New England cell-pellet, supplemented with PIC and 1 mM PMSF. Biolabs) using the buffers and protocols provided by The cells were broken with 0.5 mm glass beads using the supplier. The reaction was carried out for 1 h at the planetary mill Pulverisette 6 (Fritsch). The broken 37°C and results were analyzed by western blot. crude extract was recovered and the beads were washed with 1.5 mL of ice-cold TES buffer per gram of cell-pel- Detergent solubilization screenings let. The pH of the broken cells was adjusted to 7.5, and centrifuged at 1000 × g for 20 min at 4°C. The resulting P2 and P3 membrane fractions were solubilized with supernatant (S1) was centrifuged at 20 000 × g for eleven different detergents at 2 mg/mL of total protein 20 min at 4°C to obtain the second pellet (P2). The concentration in the membranes, and 10 mg/mL of resulting supernatant from the previous step (S2) was detergent concentration in solubilization buffer then centrifuged at 125,000 × g during 1 h at 4°C, (20 mM Tris-HCl pH 7.8, 10% (v/v) glycerol, obtaining the third pellet (P3). P3 pellet was re-sus- 150 mM NaCl, 1 mM PMSF and PIC). Solubilization pended in 0.2 mL of HEPES-sucrose buffer (20 mM was performed at 4°C or 20°C and at different incu- HEPES pH 7.6, 0,25 M sucrose) per gram of cell-pellet. bation times. 1 h and 20°C were the conditions that Aliquots of all membrane pellets were flash-frozen in showed the best results. After incubation, samples liquid nitrogen and stored at - 80°C. An appropriate were ultracentrifugated for 1 h at 125,000 × g and 4° amount of membranes containing the desired total C, the pellet was resuspended in the same volume as protein amount was mixed with loading buffer contain- the supernatant, and both, the pellet and the super- ing urea, heated at 30°C during 10 min, and subjected natant (detergent-solubilized sample) were subjected to SDS-PAGE and western blot analysis. to SDS-PAGE and western blot to analyze the solubilization efficiency of each experimental con- dition. Detergents used in the screen: n-decyl-β-D- Protein detection by western blotting maltopyranoside (DM), n-undecyl-β-D-maltopyrano- Total protein concentration in the different samples side (UDM) and octaethylene glycol monododecyl was measured by the Bicinchoninic Acid Assay Ether (C12E8) from Calbiochem; n-dodecyl-β-D- (BCA). Samples were subjected to SDS-PAGE and maltopyranoside (DDM), lauryl maltose neopentyl proteins were transferred to Polyvinylidene difluoride glycol (LMNG), 3-[(3-Cholamidopropyl)dimethylam- (PVDF) membranes (Immobilon®-P, Merck). The monio]-1-propanesulfonate hydrate (CHAPS), and ATP2 proteins fused to the BAD tag were detected n-dodecyl phosphocholine 12 (FosC12) from Ana- with the Avidin HRP-conjugate probe (Invitrogen), trace; n-Octyl-β-D-glucopyranoside (OG), n-Octyl-β- diluted to 1:20,000 in 5% (w/v) milk in PBS-Tween D-thioglucopyranoside (OTG) from Sigma-Aldrich; buffer (PBS + 0.02% (v/v) Tween®20). Cdc50 proteins and Laurydimethylaminoxide (LDAO), and 5-cyclo- fused to the 10xHis tag were detected using the His- hexyl-1-pentyl-β-D-maltoside (CYMAL-5) from Fluka. ProbeTM-HRP conjugate (ThermoFisher) diluted to Solubilization experiments were also performed in 1:2000 in 2% (w/v) BSA in PBS-Tween buffer. GFP- the presence of the cholesterol derivative, cholesteryl tagged proteins were detected with a primary mouse hemisuccinate (CHS) (Sigma-Aldrich) at a 5:1 (w/w) antibody IgG1 K Anti-GFP (Roche) diluted to 1:1000 detergent to CHS ratio. 136 A. Lamy et al. Co-immunoprecipitation assays media supplemented with ampicillin. When the OD600 reached ∼0.6, the temperature was dropped Co-immunoprecipitation assays were performed using to 20°C, and protein expression was induced with agarose beads coupled to nanobodies recognizing 1 mM isopropyl ß-D-1-thiogalactopyranoside (IPTG) the GFP (GFP-Trap®, ChromoTek), and following the (Sigma-Aldrich) for 20–24 h at 20°C. The cells were protocol suggested by the provider. 500 µL of P3 mem- spun down, flash-frozen in liquid nitrogen, and stored branes diluted to 2 mg/mL of total protein concen- at −80°C. The cell-pellet was re-suspended in 10 mL of tration were solubilized in solubilization buffer lysis buffer (PBS pH 8, 0.5 M NaCl, 5 mM imidazole, containing 10 mg/mL of DDM and 2 mg/mL CHS, 1 mM PMSF and 10 µg/µL of Lysozyme) and left for during 1 h at 20°C. After incubation, membranes 1 h at 4°C with gentle rotation. Cells were broken by were ultra-centrifuged at 125,000 × g for 90 min at 4° sonication, maintaining the cells always on ice. Cell C, and the supernatant was incubated for 1 h and 4°C lysate was centrifuged at 20,000 × g for 20 min at 4°C, with gentle rotation with 25 µL of GFP-Trap® beads, and the supernatant was mixed with 1 mL of previously equilibrated in washing buffer 1 (20 mM TALON® metal affinity beads (Clontech) previously Tris-HCl pH 7.8, 150 mM NaCl, 0.2 mg/mL DDM, equilibrated with equilibration buffer (PBS pH 8, 0.5 M 0.04 mg/mL CHS, 10% (v/v) glycerol). After incu- NaCl and 5 mM imidazole), and incubated for 1 h at bation, the flow-through was removed by centrifuging 4°C under gentle rotation. After incubation, the beads the sample at 100 × g for 30 s. The beads were then were sequentially washed with 20 column-volumes of washed by adding 500 µL of ice-cold washing buffer equilibration buffer, 10 column-volumes of equilibration 1 followed by centrifugation at 100 × g for 30 sec. A buffer with 20 mM imidazole, and 10 column-volumes second wash was done with 500 µL of washing buffer of equilibration buffer with 30 mM imidazole. The 2 (20 mM Tris-HCl pH 7.8, 500 mM NaCl, 0.2 mg/mL nanoGFP was eluted with two steps of 5 column- DDM and 10% (v/v) glycerol). Proteins bound to the volumes of equilibration buffer with 150 mM of imida- GFP-Trap® beads were eluted by adding 50 µL of zole, and 5 column-volumes of equilibration buffer 0.2 M glycine buffer pH 2.5 followed by centrifugation with 300 mM imidazole. The elution fractions were ana- at 100 × g for 30 sec. The low-pH of the elution was lyzed by SDS-PAGE, pooled and concentrated using a quickly neutralized by adding 5 µL of 1 M Tris-base Vivaspin® 20,5,000 Mw concentrator (Sartorius). The pH 10.4. This elution step was repeated one more time concentrated protein was dialyzed overnight at 4°C to and samples were analyzed by western blot. remove the imidazole, further concentrated and stored at 4°C until use. The purified nanoGFP was then cova- Fluorescence-detection size-exclusion lently bound to agarose beads. 1 ml of NHS-Activated chromatography Sepharose 4 Fast Flow beads (GE Healthcare Life Sciences) was washed with 10–15 column-volumes of P3 membranes were solubilized as described in the 1 mM HCl, and equilibrated with PBS pH 7.5. Then, previous section and 200 µL of the detergent-solubil- 1 mg of nanoGFP was added to the beads and left over- ized supernatant was injected in a Superose 6 10/300 night at 4°C under gentle rotation, keeping a nanoGFP GL gel-filtration column (GE healthcare) equilibrated to beads volume ratio of approximately 0.5–1. After with 20 mM Tris-HCl pH 7.8, 150 mM NaCl, 10% (v/ the overnight reaction, the remaining non-reacted sites v) Glycerol, 0.1 mg/mL DDM, 0.02 mg/mL CHS. Flu- were blocked by the addition of 10–15 column-volumes orescence-detection size-exclusion chromatography of 0.1 M Tris-HCl pH 8.5 followed by a 4 h incubation (FSEC) experiments were performed in an ÄKTATM with this buffer at 4°C. Finally, the beads were subjected purifier chromatography system (GE healthcare) to 3 cycles of two sequential washes of 3 column- with a fluorescence detector (FP-4025, JASCO) con- volumes of 0.1 M Tris-HCl, 0.5 M NaCl, pH 8.5 and nected “in line” with the column. The excitation and 3 column-volumes of 0.1 M Acetate buffer, pH 5.0, emission wavelengths of the fluorescence detector 0.5 M NaCl. Finally, the home-made GFP-Trap® beads were set, respectively, to 460 and 520 nm. FSEC chro- were equilibrated in PBS pH 8.0 and stored at 4°C matograms were plotted using Plot2 software and data until use. was normalized. Purification and immobilization of PcATP2/ Generation of home-made GFP-Trap® beads for PcCdc50B in home-made GFP-Trap® beads ATPase assays P3 membranes expressing either PcATP2-GFP-BAD/ We first expressed and purified a nanobody directed PcCdc50B-His or PcATP2-D596N-GFP were diluted against GFP (nanoGFP). The plasmid vector pOPINE up to 5 mg/mL of total protein concentration in solu- GFP nanobody was a gift from Brett Collins (Addgene bilization buffer (20 mM Tris-HCl pH 7.8, 150 mM plasmid # 49172) [33]. 500 mL of E. coli BL21 cells NaCl, 100 mM KCl, 5 mM MgCl2, 20% (w/v) glycerol, harbouring this plasmid were cultured at 37°C in LB 20 mg/mL DDM and 4 mg/mL CHS), supplemented Emerging Microbes & Infections 137 with PIC and 1 mM PMSF. After 1 h at 20°C under in 10 TMs. The large intracellular domains (ICD) gentle rotation, samples were ultracentrifuged at between TM2 and TM3 (ICD1) and TM4 and TM5 125,000 × g during 1 h at 4°C, and the supernatant (ICD2) were aligned using MAFFT [35] and modeled (detergent-solubilized proteins) were mixed with the independently in MODELLER [36]. ICD1 was mod- home-made GFP-Trap® beads previously equilibrated eled against templates 6KG7, 6ROH, and 1IWO. with 10 column-volumes of washing buffer (20 mM ICD2 was modeled against templates 6K7G, 6ROH, Tris-HCl pH 7.8, 150 mM NaCl, 100 mM KCl, 1IWO, 4HQJ, and 5MPM. All TMs were modeled by 5 mM MgCl2, 20% (w/v) glycerol, 0.5 mg/mL DDM, carbon alpha molecular replacement in UCSF Chi- 0.1 mg/mL CHS and 0.01 mg/mL POPS). After 2 h mera [37] using 6K7G and 6ROH as templates. The at 4°C of incubation with gentle rotation, the beads whole PcATP2 sequence was realigned using were washed with 20 column-volumes of washing MAFFT and manually refined against the ICD1, buffer. GFP-Trap® beads with bound PcATP2-GFP- ICD2, and the TMs. This alignment was used as tem- BAD/Cdc50B-10xHis were diluted twice in washing plate to build the overall PcATP2 model in MODEL- buffer and flash-frozen in liquid-N2. The concen- LER. PcATP2 model was embedded in an POPC tration of PcATP2-GFP-BAD in the beads was bilayer, energy minimized, and equilibrated for 10 ns quantified by measuring the GFP fluorescence in a using the CHARMM36F [38] force field. 96-well plate, and using purified GFP as reference. The protein content coupled to the GFP-Trap® beads was eluted with 0.2 M glycine buffer pH 2.5 prior Fluorescent confocal microscopy analysis by Coomassie-blue staining and western blot. 1 mL of small-scale cultures was centrifuged for 5 min at 2000 × g followed by two washing steps with 1 mL ATPase assays of PBS pH 7.4. The cells were re-suspended in 1 mL of PBS pH 7.4, and 2 µL were deposited on an agarose Between 5 and 10 µL of home-made GFP-Trap® beads pad (1% (w/v) agarose gel in milliQ water). The pic- coupled with PcATP2-GFP-BAD/Cdc50B-10xHis tures were taken in the Light Microscopy Facility of (wild-type and mutant), and corresponding to ImaGif (I2BC, Gif-sur-Yvette, France) using a Leica 100 ng of PcATP2-GFP-BAD, were diluted into confocal microscope SP8 controlled by the LAS-X 50 µL of the reaction buffer (50 mM MOPS-Tris pH software and using an oil immersion objective lens 7.0, 100 mM KCl, 5 mM MgCl2, 20% (w/v) glycerol, with 100 x magnification. Samples were excited with 1 mM NaN3, 0.5 mg/mL DDM, 0.1 mg/mL CHS), a 488 nm laser. supplemented or not (“no lipid” experiments) with 0.45 mg/ml of the phospholipid substrate and, when indicated, with 0.025 mg/ml of PI4P. Samples were Results incubated at room temperature for 10 min and ATP2 and Cdc50 proteins are highly conserved moved to ice for 15 min more. The reaction was in Plasmodium species and present in other initiated after the addition of 1 mM ATP-Mg2+ and apicomplexan parasites incubated at 37°C for 25 min. The reaction was stopped by dropping the test tubes into liquid N2. ATP2 is highly conserved in Plasmodium species. For To quantify the hydrolyzed ATP, we used the BIO- instance, P. falciparum ATP2 (PfATP2, VEuPathDB MOL® Green solution (Enzo Biochem Inc), a proprie- ID: PF3D7_1219600), P. vivax ATP2 (VEuPathDB tary mixing solution that uses a molybdate/malachite ID: PVX_123625), and the ATP2 orthologs from the green-based assay to quantify the released Pi. Thus, malaria mouse models P. berghei, (VEuPathDB ID: 100 µL of BIOMOL® Green solution was added to PBANKA_1434800) and P. chabaudi (VEuPathDB each sample, vortexed and incubated at room temp- ID: PCHAS_1436800, named in this work PcATP2) erature for 30 min. After incubation, samples were (Figure supplement 1) share ∼60% amino acid identity quickly spun down to pellet the beads, and 100 µL of [8]. In addition, homologs to ATP2 are also present in clear supernatant were deposited in a 96-well plate the genome of other disease-causing intracellular to measure the absorbance at 620 nm. Standard cali- parasites belonging, to the Apicomplexa phylum like bration curves at each experimental buffer condition Plasmodium (Figure supplement 1) [8], although it is were made using a phosphate standard solution still unknown if these transporters are also essential. provided by the BIOMOL® Green supplier. These sequences contain the typical membrane top- ology of 10 predicted transmembrane segments (TMs), the conserved P-type ATPase motif, DKTG, PcATP2 3D model (intracellular loop between TMs 4 and 5, where the To build the PcATP2 model, a divide and conquer aspartate is transiently autophosphorylated during strategy was taken. TOPCONS algorithm [34] was the transport cycle), and the specific P4-ATPase used to infer membrane protein topology, resulting motif DGET (intracellular loop between TMs 2 and 138 A. Lamy et al. 3), implicated in protein dephosphorylation [10]. Cdc50 proteins. S. cerevisie carries at least six different Moreover, the well-conserved PICL (or PISL) motif biotynilated proteins, and some of them are also present in TM4 of P4-ATPases is also conserved in detected in our western blots when using the Avidin these apicomplexan sequences (Figure supplement HRP-conjugate probe (see control experiment in 1). Interestingly, the glutamine in TM1 and the two lane 9, and figure legend of Figure 1 for descrip- asparagine residues in TMs 4 and 6, that coordinate tion)[40]. Nevertheless, the position of the bands of the substrate’s phosphatidylserine (PS) moiety in the different S. cerevisiae biotynitaled proteins did ATP8A1 [24] and in ATP11C [28], are also conserved not overlay the ones of the BAD-fused PcATP2. in these apicomplexan sequences (Figure supplement PcATP2 displayed a slightly different electrophoretic 1). In ATP8A1, the isoleucine and leucine residues mobility depending on the position of the BAD tag of the PICL motif in TM4 also stablish hydrophobic (Figure 1), previously observed in other P4-ATPases contacts with the PS moiety of the substrate [24]. tagged as well with the BAD at either of the two extre- The genome of P. falciparum encodes three Cdc50/ mities [31]. Also, BAD-PcATP2 displayed two electro- LEM3 proteins: PfCdc50A (VEuPathDB ID: phoretic bands, suggesting either a cleavage near the Pf3D7_0719500), PfCdc50B (VEuPathDB B ID: C-terminal end or two different electrophoretic mobi- Pf3D7_1133300) and PfCdc50C (VEuPathDB ID: lities of the same construct (Figure 1, lanes 2, 4, 6 and Pf3D7_1029400). Like ATP2, each Cdc50 protein 8). Western blot analysis also revealed that the three has close homologs in other Plasmodium species putative PcCdc50 proteins were also co-expressed (between 50–60% amino acid identity between Plas- with both BAD-PcATP2 and PcATP2-BAD (Figure modium species, depending on the Cdc50 protein, 1, bottom panels). We observed that the intensity of see as example the alignments of Cdc50B orthologs the electrophoretic mobilities of BAD-PcATP2 are in Figure supplement 2), as well as in other apicom- influenced by the presence of the PcCdc50 proteins plexan parasites. All these sequences contain the two since the lower (or fastest) band became practically predicted TMs connected by a large extracellular (or the only one when co-expressed with PcCdc50A luminal) domain, as well as the four well-conserved (Figure 1, lane 8). PcCdc50A displayed two electro- cysteines, known to participate in two disulphide phoretic mobilities, suggesting protein glycosylation bridges [23,24]. These disulphide bridges are impor- events typical of Cdc50 proteins (Figure 1, lanes 7 tant for the association of the Cdc50 protein with and 8). Overall, when either BAD-PcATP2 or the P4-ATPase and for the activity of the complex PcATP2-BAD were co-expressed with each of the [39]. PcCdc50 proteins neither the electrophoretic pattern nor the intensity of the PcATP2 bands changed sub- stantially. We discontinued working with PcCdc50C Screening Plasmodium-encoded ATP2 and because of its low detection yield compared with the Cdc50 proteins for S. cerevisiae heterologous other two Cdc50 subunits. production We also examined the expression yield in two To unravel the functional and structural features of different membrane fractions obtained from differen- PfATP2, we examined its heterologous expression tial centrifugation of BAD-PcATP2 co-expressed with together with its putative Cdc50 proteins in the yeast either PcCdc50A-His or PcCdc50B-His. We detected Saccharomyces cerevisiae. We screened for expression the presence of PcATP2 and the two PcCdc50 sub- of the ATP2 and Cdc50 sequences encoded by three units in both, the P2 and in the P3 membrane fraction different Plasmodium species: P. falciparum, (Figure 2A). Notably, PcATP2 changed its relative dis- P. berghei and P. chabaudi aiming at identifying tribution between P2 and P3 depending on the co- ATP2 and Cdc50 candidates with suitable expression expressed PcCdc50 subunit (Figure 2A and B). That yield for biochemical studies. For western blot detec- is, while the relative amount of PcATP2 in P2 or P3 tion, we fused a biotin acceptor domain (BAD) at was similar when co-expressed with PcCdc50B (Figure either the N-terminal (BAD-ATP2) or the C- terminal 2A), PcATP2 seems to be mostly accumulated in P2 (ATP2-BAD) ends of the ATP2 sequences, and a when co-expressed with PcCdc50A (Figure 2B). In 10xHis-tag at the C-terminal end of each Cdc50 addition, the relative distribution of PcATP2 between protein [31]. The P. chabaudi ortholog, PcATP2, was P2 and P3 followed the same relative distribution as its the only one that expressed in our system. The co-expressed PcCdc50 protein (Figure 2A and B, bot- sequence similarity of ATP2 and Cdc50 ortologs in tom panels). Plasmodium suggest similar functional and structural We screened a battery of eleven detergents to solu- features, thus making PcATP2 a fair paradigm of bilize BAD-PcATP2 and the co-expressed Cdc50 pro- PfATP2. teins from both, P2 and P3 membranes. PcATP2 and Figure 1 shows the expression profiles of the two Cdc50 proteins in P2 membranes could not be solubil- BAD-tagged versions of PcATP2, alone, or co- ized under any of the tested experimental conditions, expressed with each of the three P. chabaudi-encoded with the exception of the rather denaturing n- Emerging Microbes & Infections 139 Figure 1. Analysis of PcATP2 expression in S. cerevisiae membranes, alone or co-expressed with each of the three putative PcCdc50 subunits. 5 µg of total protein were loaded on each lane. Top panels, western blots revealed with the probe against the BAD tag. Bottom panels, western blots revealed with the HisProbeTM to detect the 10xHis tag. Gel bands corresponding to PcATP2 and the P. chabaudi Cdc50 subunits are indicated by arrows. Lane 1: single expression of PcATP2-BAD. Lane 2: single expression of BAD-PcATP2. Lane 3: co-expression of PcATP2-BAD and PcCdc50B-His. Lane 4: co-expression of BAD-PcATP2 and PcCdc50B-His. Lane 5: co-expression of PcATP2-BAD and PcCdc50C-His. Lane 6: co-expression of BAD-PcATP2 and PcCdc50C- His. Lane 7: co-expression of PcATP2-BAD and PcCdc50A-His. Lane 8: co-expression of BAD-PcATP2 and PcCdc50A-His. Lane 9: empty Vector, membranes expressing no Plasmodium proteins (negative control). The theoretical molecular weight mass of PcATP2-BAD or BAD-PcATP2 is 180 kDa. The theoretical molecular weight masses of PcCdc50B-His, PcCdc50C-His and PcCdc50A-His are, respectively, 45, 61 and 51 kDa. Lanes 1–6 were run in a 8% acrylamide gel. Lanes 7–9 were run in a gel contain- ing a gradient of acrylamide between 4–15%. Naturally occurring biotinylated S. cerevisiae proteins are indicated in lane 9: Acc1p, acetyl-CoA carboxylase; Dur 1/2p, urea carboxylase; Pyc 1/2p, pyruvate carboxylase isoforms 1 and 2, and Arc 1p, complex acyl- RNA[40]. dodecylphosphocholine (Fos-Choline-12). In contrast, biogenesis, explaining (at least, in part) the preferen- a fraction of PcATP2 and PcCdc50B from P3 mem- tial localization of PcCdc50A together with its co- branes was solubilized in N-dodecyl-β-D-maltopyra- expressed PcATP2 partner in P2 membranes (Figure noside (DDM) and, with less efficiency, in Lauryl 2, right panels). maltose neopentyl-glycol (LMNG) or Octaethylene glycol monodecyl ether (C12E8) (Figure supplement N-Glycosylation of PcCdc50A and PcCdc50B 3). The fact that only the PcATP2 and PcCdc50B frac- expressed in Sacharomyces cerevisiae tions present in P3 could be solubilized in DDM strongly suggests a folding defect of the P2-containing As seen in Figures 1 and 2, PcCdc50A-His presented proteins because DDM is know to be poorly efficient two electrophoretic bands, suggesting glycosylation fo solubilize unfolded or partially folded integral events typical of Cdc50 proteins. Glycosylation of membrane proteins[41]. Importantly, the solubil- Cdc50 proteins plays often important roles in the stab- ization efficiency of these detergents was greatly ility and the activity of the P4-ATPase/Cdc50 complex improved by the presence of the cholesterol derivative, [39,42]. Using the online server NetNGlyc (http:// cholesteryl hemisuccinate (CHS) (Figure supplement www.cbs.dtu.dk/services/NetNGlyc/), one potential 3). Due to the low amount of BAD-PcATP2 co- glycosylation site was predicted at position N260 expressed with PcCdc50A in P3 (Figure 2), the immu- within the FNDT motif situated in the extracellular nodetection of the DDM/CHS solubilized fraction of or luminal domain that connects TMs 1 and 2 of BAD-PcATP2 from P3 membranes was always chal- PcCdc50A (Figure supplement 4). This motif is con- lenging (Figure supplement 3, panel E). Interestingly, served in Cdc50 proteins from others organisms, we found that the lower band of PcCdc50A (identified including the S. cerevisiae Cdc50 protein, Cdc50p, in the next section as the non-glycosylated form of whose glycosylation state has already been observed PcCdc50A, Figure 3) was the only one co-solubilized when overexpressed in S. cerevisiae [23]. In contrast, with PcATP2 (Figure supplement 3, panel F or Figure no potential N-glycosylation sites were predicted in 4, right panel, lanes 1, 4 and 5). This suggests that gly- the extracellular or luminal sequence of PcCdc50B. cosylation of PcCdc50A (see next section) might affect To analyze the possible N-glycosylation of PcCdc50A the folding and/or membrane trafficking during and PcCdc50B when expressed in S. cerevisiae, we 140 A. Lamy et al. Figure 2. Analysis of membrane fractions of S. cerevisiae co-expressing PcATP2 with either PcCdc50B or PcCdc50A. Top panels, western blots revealed with the probe against the BAD. Bottom panels, western blots revealed with the HisProbeTM to detect the 10xHis tag. (A) membrane fraction co-expressing BAD-PcATP2 and PcCdc50B-His. (B) membrane fraction co-expressing BAD- PcATP2 and PcCdc50A-His. The bands corresponding to BAD-PcATP2, PcCdc50B-His and PcCdc50A-His are indicated. P2 and S2 are, respectively, the membrane pellet and the supernatant obtained after centrifugation at 20,000 × g. P3 and S3 are, respect- ively, the membrane pellet and the supernatant obtained after centrifugation at 125,000 × g. Each lane was loaded with the cor- responding membrane fraction or supernatant containing 1 µg of total protein. Biotinylated S. cerevisiae proteins are also indicated: Acc1p: acetyl-CoA carboxylase, Dur1/2p: urea carboxylase, and Pyc1/2p: pyruvate carboxylase isoforms 1 and 2[40]. performed enzymatic deglycosylation assays using PcCdc50B expressed in S. cerevisiae is not N- membranes co-expressing BAD-PcATP2/PcCdc50B- glycosylated. His or BAD-PcATP2/PcCdc50A-His, and using two deglycosydases: the amidase PNGase F and the N-gly- cosydase EndoH. As control, we used Cdc50p co- PcATP2 associates with PcCdc50B and expressed with its S. cerevisiae P4-ATPase partner, PcCdc50A in detergent micelles Drs2p (BAD-Drs2p/Cdc50p-His). Consistent with To assess the association of PcATP2 with PcCdc50B or previous studies [31], Cdc50p ran as two major PcCdc50A even after DDM/CHS solubilization, we bands in the western blot (Figure 3); the fully glycosy- performed immunoprecipitation studies. We fused lated fraction (∼50 kDa) and the non-glycosylated one (∼37 kDa). After incubation with either PNGase F or EndoH, the glycosylated band of Cdc50p disappeared while the intensity of the lowest non-glycosylated band increased. Similarly, the upper band of PcCdc50A disappeared after incubation with either PNGase or EndoH, while the lowest band became more intense with the same electrophoretic mobility (Figure 3), thus confirming the N-glysosylation of PcCdc50A. Conversely, the band corresponding to PcCdc50B remained unaltered after enzymatic treat- ment with regard both electrophoretic mobility and Figure 3. Deglycosylation assays of PcCdc50B and PcCdc50A intensity (Figure 3). Of note, just above the band of co-expressed with PcATP2. Membranes co-expressing PcCdc50B there is a faint band in the non-treated PcATP2/PcCdc50B, PcATP2/PcCdc50A or Drs2p/Cdc50p were protein sample that disappeared after enzymatic diges- enzymatically deglycosylated with PNGase F or EndoH, as indi- tion (Figure 3). This band could represent a non-specific cated. 10 µg total protein were loaded in the experiments with PcATP2/PcCdc50B and PcATP2/PcCdc50A, whereas 5 µg of protein since it is also present (although less intense) in total protein were loaded in the control experiment with the experiment with untreated PcCdc50A. Nevertheless, Drs2p/Cdc50p. The western blots were revealed with the His- we can conclude from these experiments that most of ProbeTM to detect the 10xHis tag. Emerging Microbes & Infections 141 Figure 4. Co-immunoprecipitation of PcATP2-GFP-BAD with PcCdc50B-His or PcCdc50A-His. Lanes 1–4 correspond to the different steps of the assay with membranes co-expressing PcATP2-GFP-BAD and PcCdc50B-His (A) or PcCdc50A-His (B): (1) input, deter- gent-solubilized membranes, (2) flow-through or non-bound protein fraction, (3) washings, and (4) elution. Lanes 5–8 are equiv- alent to lanes 1–4 but with membranes co-expressing PcATP2-BAD and either, PcCdc50B-His (A) or PcCdc50A-His (B). The blots on the top panel were revealed with an antibody against the GFP to detect PcATP2-GFP-BAD, and the bottom one with the HisPro- beTM to detect PcCdc50B-His and PcCdc50A-His (labelled with an asterisk). the superfolder green fluorescent protein (GFP) at the membranes co-expressing a non GFP-tagged C-terminal end of PcATP2 followed by the BAD PcATP2 with either PcCdc50B (PcATP2-BAD/ (PcATP2-GFP-BAD). We chose to tag PcATP2 at PcCdc50B-His) or PcCdc50A (PcATP2-BAD/ the C-terminal end rather than the N-terminal like PcCdc50A-His). In this case, none of the PcCdc50 the previous experiments because the GFP is a more proteins were detected in the elution fraction (Figure sensitive folding reporter when fused at the C-term- 4, lane 8 bottom panels), thus excluding a false positive inal end of the target protein [41]. The presence of result due to a non-specific binding of PcCdc50B-His both co-expressed proteins in P3 membranes was or PcCdc50A-His to the GFP-Trap® beads. confirmed by western blot (Figure supplement 5, We also analyzed PcATP2 and PcCdc50B associ- lanes 1 and 2). In addition, the relative distribution ation by fluorescence-detection size-exclusion chrom- of C-terminally tagged PcATP2 between P2 and P3 atography (FSEC). We used the GFP fluorescence to membranes, and the solubilization efficiency of monitor the elution profile of DDM/CHS-solubilized DDM/CHS of PcATP2-GFP-BAD/PcCdc50B-His or PcATP2 in a gel-filtration column, and analyzed the PcATP2-GFP-BAD/PcCdc50A-His from P3 mem- degree of monodispersity (or stability) in detergent branes were very similar to the N-terminally tagged micelles of PcATP2 in the presence or in the absence version of PcATP2 (not shown). Therefore, DDM/ of PcCdc50B. Detergent-solubilized PcATP2-GFP- CHS-solubilized PcATP2-GFP-BAD/PcCdc50B-His BAD eluted as a nearly symmetric wide peak centred or PcATP2-GFP-BAD/PcCdc50A-His were trapped at ∼11.5 ml (Figure 5, dotted line). The second peak in agarose beads coupled to an anti-GFP nanobody at ∼16 ml elution corresponded to the light-scattering (GFP-Trap®), and bound proteins were eluted with a of empty DDM/CHS micelles, visible in the chromato- low-pH buffer and analyzed by western blot. As gram because the instrumental settings used to detect expected, PcATP2-GFP-BAD was detected in both PcATP2-GFP-BAD. DDM/CHS-solubilized mem- elution fractions (Figure 4A and B, lanes 4 in top branes expressing no GFP-tagged protein (shadow panels). Moreover, PcCdc50B-His and PcCdc50A- chromatogram in Figure 5) and the FSEC profile of His also coeluted with PcATP2-GFP-BAD in their 1% DDM 0.2% CHS confirmed this. PcATP2’s elution respective experiments (Figure 4A and B, lanes 4 in profile did not change significatively when PcCdc50B- bottom panels), demonstrating that PcATP2 interacts His was co-expressed (Figure 5, solid line). However, with either PcCdc50B or PcCdc50A in the membrane the western blot analysis of the eluted fractions and after detergent solubilization. The western blot along the PcATP2-GFP-BAD elution peak confirmed signals of eluted PcATP2-GFP and PcCdc50A-His that PcCdc50B-His co-eluted with PcATP2-GFP- were much weaker than the ones of the PcATP2- BAD (Figure 5, upper panel) following a similar GFP/PcCdc50B-His experiment, in concordance elution profile than PcATP2-GFP-BAD. Since the with their relative amounts in the P3 membrane frac- theoretical molecular mass of PcATP2-GFP-BAD tion (Figure 2 and Figure supplement 5). As control, and PcCdc50B-His are, respectively, 205 and 47 kDa, we performed the same experiment using P3 the co-elution of both proteins supports their 142 A. Lamy et al. immunoprecipitation studies revealed the association of PcATP2 with either PcCdc50A or PcCdc50B. More- over, FSEC experiments also confirmed the associ- ation of PcATP2 with PcCdc50B. Detergent-solubilized PcATP2/PcCdc50B hydrolizes ATP in the presence of POPS and POPE, and senses PI4P In detergent micelles, P4-ATPases hydrolyze ATP only in the presence of their phospholipid substrates [43]. This property not only permits evaluating the activity of the transporter, but also it allows to assess substrate specificity. We assayed the functional properties of recombinant PcATP2-GFP-BAD/PcCdc50B-His com- plex by measuring the lipid-stimulated capacity of PcATP2 to hydrolize ATP. Detergent-solubilized PcATP2-GFP-BAD/PcCdc50B-His was immobilized in home-made GFP-Trap® beads, and the liberated Figure 5. Fluorescence-detection Size Exclusion Chromato- inorganic phosphate upon ATP hydrolysis was quan- graphy of detergent solubilized PcATP2 and PcCdc50B. Mem- tified spectrophotometrically after 25 min at 37°C in branes containing 5 mg/ml of total protein concentration the presence of two putative phospholipid substrates. were solubilized with 1% (w/v) DDM, 0.2% (w/v) CHS, and the supernatant after ultracentrifugation was loaded into a As negative control, we generated a predicted non- Superose 6 10/300 GL gel-filtration column equilibrated with functional mutant of PcATP2 after replacing the aspar- 20 mM Tris-HCl pH 7.8, 150 mM NaCl, 10% (v/v) Glycerol, tate residue at position 596 by an asparagine (Figure 0.1 mg/mL DDM, 0.02 mg/mL CHS, and connected to a fluor- supplement 1). This aspartate is part of the conserved escence detector. Normalized FSEC profiles of single- P-type ATPase motif DKTG, and undergoes phos- expression of PcATP2-GFP-BAD (dotted lines) and co- expression of PcATP2-GFP-BAD with PcCdc50B-His (solid phorylation and de-phosphorylation during each line). A profile of membranes expressing no protein is transport cycle [9]. Therefore, the resulting PcATP2- shown (in grey shadow). Normalized FSEC profile of single- D596N-GFP mutant is expected to be fully deficient expression PcCdc50B-GFP-His (dashed line). The presence of in ATP hydrolysis, as verified in all the tested P-type free eGFP is indicated. On top of the figure, analysis of ATPases [9]. The presence of both PcATP2 variants PcCdc50B-His co-elution with PcATP2-GFP-BAD (solid line and the co-expressed PcCdc50B in the GFP-Trap® chromatogram). Collected fractions of 200 µl between 10.4– 12 ml of the PcATP2-GFP-BAD elution were analyzed by wes- beads were confirmed by SDS-PAGE (Figure 6, upper tern blot using the HisProbeTM to detect the presence of panel). Although both PcATP2 versions were visible PcCdc50B-His. by Coomassie staining (Figure 6, upper panel, A), PcCdc50B was only visible after immunoblotting (Figure 6, upper panel, C). In addition, anti-GFP wes- association in detergent micelles. No significant shift tern blot analysis of the two PcATP2 variants eluted in the retention time of PcATP2-GFP-BAD was from the GFP-Trap® beads confirmed the absence of observed when PcCdc50B-His was present (Figure 5, protein degradation fragments containing the GFP solid line), likely due to the low-resolution of the col- (Figure 6, upper panel, B), thus validating the use of umn at the range of molecular weights where PcATP2- the GFP fluorescence to measure the concentration of GFP-BAD elutes. To discard the elution of free or PcATP2 in the beads. Immobilized PcATP2-GFP- non-associated PcCdc50B at the same retention time BAD/PcCdc50B-His (called “WT” hereafter) in GFP- than PcATP2, we fused the GFP at the C-terminal Trap® beads displayed ATP hydrolysis in the presence end of PcCdc50B (PcCdc50B-GFP-His) and per- of the phospholipids, POPS and POPE (grey bars in formed FSEC analysis of DDM/CHS-solubilized Figure 6), with an apparent ATPase activity between membranes expressing only PcCdc50B-GFP-His. ∼0.5 and 0.7 nmols Pi min−1 μg protein−1. In the The elution profile of PcCdc50B-GFP-His (Mw same conditions, the apparent activity obtained from 76 kDa) in the absence of PcATP2 showed a main the samples containing the functionally-impaired peak at ∼14.3 ml (dashed line in Figure 5), far from mutant PcATP2-D596N-GFP/PcCdc50B-His (called the PcATP2-GFP-BAD main elution (∼11.5 ml, “DN” hereafter) ranged from ∼0.2 to 0.4 nmols Pi dotted line Figure 5). Interestingly, the relatively min−1 μg protein−1 (white bars in Figure 6). Moreover, broad and asymmetric shape of the chromatogram in the absence of added lipids, the apparent activites of suggested a poor stability of PcCdc50B-GFP-His in both WT and DN (“no lipid” in Figure 6,) were in the DDM/CHS when PcATP2 is absent. Collectively, same range as the DN sample in the presence of Emerging Microbes & Infections 143 Figure 6. Lipid-stimulated ATPase activity of PcATP2/PcCdc50B complex. The apparent ATPase activity of the PcATP2/PcCdc50B complex was measured by spectrophotometric quantification of the released inorganic phosphate in the absence (no lipid) and in the presence of the indicated lipids. POPS, 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine, and POPE, 1-palmitoyl-2-oleoyl- sn-glycero-3-phosphoethanolamine. As indicated, ATPase assays were also performed in the presence (+) or in the absence (-) of phosphatidylinositol-4-phosphate (PI4P). Gray bars and white bars represent, respectively, the activity of the wild-type (WT) and the functionally-impaired D596N mutant (DN) at each experimental condition. The reaction was performed at 37°C during 25 min, and initiated by the addition of 1 mM ATP-Mg2+. The bars display the average value of n = 3 or n = 4 experimental data- points, and the indicated error is the standard deviation. The upper panel displays the SDS-PAGE of the acid-eluted content of the immobilized wild-type (WT) or D596N mutant (DN) agarose beads used for this experiments. (A) Coomassie-stained, (B) western blot using the antibody against the GFP to detect PcATP2-GFP, and (C ) western blot using the HisProbeTM to detect the 10xHis tag fused to PcCdc50B .* P < 0.05 (paired t-test). Note that the DN mutant of PcATP2 runs slightly faster because it lacks the BAD domain. POPS and POPE. The slightly higher apparent activity Nevertheless, we observed a larger increase of the of WT with respect DN in the “no lipid” conditions POPS or POPE-stimulated ATPase activity of WT in (“no lipid” in Figure 6) is probably due to the presence the presence of PI4P (Figure 6). The net increase of of lipids carried by the protein during purification. WT ATPase activity, that is the difference of the appar- Therefore, these results showed that the heterologously ent ATPase activities of WT in the presence and produced PcATP2 catalyzes ATP hydrolysis induced absence of PI4P after substracting the contribution of by the presence of the phospholipids POPS and the DN sample at each condition were 0.13 and 0.14 POPE. The C-terminal end of PcATP2 and its Plasmo- nmols Pi min−1 μg protein−1 in the experiments with, dium orthologs contain the equivalent residues respectively, POPS and POPE (Figure 6). In summary, Tyr1235 and His1236 of Drs2p, recently identified as the detergent-solubilized PcATP2/PcCdc50 complex a phosphatidylinositol-4-phosphate PI4P recognition produced in yeast is functionally competent as it cata- motif [23] (Figure supplement 6). In Drs2p, PI4P bind- lyzes lipid-stimulated ATP hydrolysis in the presence ing induces a strong increase of the substrate-stimu- of POPS and POPE, being also upregulated by PI4P. lated ATPase activity [44,45]. In our experimental set up, the presence of PI4P slightly increased the signal Discussion background in all samples, since we observed a small increase in the average apparent activity in all control In this paper we have demonstrated for the first time experiments (see the non-lipid assays and the DN that purified PcATP2/PcCdc50B complex catalyses assays in the presence of POPS and POPE), with ATP hydrolysis in the presence of phospholipids. increases of the apparent ATPase activity between The essential and irreplaceable putative lipid flippase 0.02 and 0.08 nmols Pi min−1 μg protein−1 (Figure 6). activity of ATP2 [4–6] confirms that lipid homeostasis 144 A. Lamy et al. is paramount for the parasite’s development, repre- probably being one of the reasons for what 73% of senting new opportunities for therapeutic interven- P. falciparum-encoded MTPs are still considered as tion. There are obvious future possibilities since, for putative [7]. However, the use of recombinant pro- instance, PfATP2 was associated with drug-resistant teins has been pivotal in malaria research, allowing phenotypes against two antimalarial drugs [20]. the identification of substrates and inhibitors [47,48], The challenge now is to understand why and providing protein sample for structural studies when the transport activity of ATP2 and the [49,50] or for confirming that PfATP6 is not an arte- other putative P4-ATPases are required within the misinin drug-target [16]. One possible reason to complex life-cycle of the malaria parasite. The intra- explain the difficulty of expressing P-type ATPases cellular development inside the parasitized cell encoded by apicomplexan parasites in heterologous depends on an initial heavy supply of lipids to hosts is that they contain non-conserved amino acid build the different membrane compartments, stretches within the two large cytoplasmic regions together with efficient mechanisms to ensure a suit- connecting TMs 2 and 3 and TMs 4 and 5 (Figure able lipid composition on each membrane, as ∼75% 7)[51], enclosing a high density of positively-charged of all lipids exhibit significant variations on its rela- and asparagine residues, a potential handicap for het- tive amount along parasite’s development, particu- erologous expression due to endoplasmic reticulum larly during gametogenesis [46]. The Plasmodium (ER) retention [52]. These extensions are absent in parasite has a limited capacity of synthesizing fatty many other P-type ATPases, and as judged by a 3D acids and phospholipids[46], and therefore, many model of PcATP2 based on the recent cryo-EM struc- lipids are scavenged from the host membrane or ture of Drs2p and ATP8A1 [23,24], they contain a from the serum where P4-ATPases might have a rel- large proportion of non-structured regions (Figure 7). evant role. Confocal microscopy images of S. cerevisie expressing To unravel the structural and functional features of PcATP2-GFP-BAD alone or co-expressed with ATP2 we screened three ATP2 orthologs for heter- PcCdc50B-His indicates that PcATP2-GFP-BAD ologous expression in S. cerevisiae. In the present is located in intracellular compartments (Figure sup- study we succeeded to express the P. chabaudi plement 7). This could indicate a possible retention of ATP2, PcATP2, a fair paradigm of Plasmodium PcATP2-GFP-BAD in the ER as observed before in ATP2 due to its high conservation in Plasmodium mammalian cell-lines for other P4-ATPase mutants species (Figure supplement 1). This expression system presenting some folding defect [53]. Logically, in the was found to be successful in our group for PfATP6 absence of organellar markers we are unable to make [16], or for a yeast P4-ATPase [31]. The production further conclusions about the exact localization of of recombinant malaria MTPs is still very challenging, PcATP2. Interestingly, a recent study showed that the Figure 7. Alignment of the cryo-EM structure of Drs2p and a 3D structural model of PcATP2. The 3D structural model of PcATP2 (cartoon representation) was aligned with the cryo-EM structure of the autoinhibited conformation of Drs2p (PDB ID 6roh, mesh representation [23]). Protein domains of PcATP2 in the cytoplasmic region between TMs 2 and 3 that are absent in Drs2p are depicted in blue. Protein domains of PcATP2 in the cytoplasmic region between TMs 4 and 5 that are absent in Drs2p are depicted in magenta. (A) Lateral perspective of the molecule. (B) Cytoplasmic view after 90° rotation of A through the x axis. The grey sha- dow square in A represents the putative position of the surrounding membrane. Emerging Microbes & Infections 145 P. falciparum K+ channel PfKch2 is also expressed in phospholipids. Two amino acid clusters determine, internal compartments of S. cerevisiae but it is active at least in part, the lipid head-group selectivity of after detergent solubilization and purification[54]. P4-ATPases [24]. In PcATP2, these clusters corre- Co-immunoprecipitation studies (Figure 4), spond to residues Gln65 and Leu66 in TM1, and together with the relative distribution of PcATP2 Ala455 and Asn456 in TM4 (Figure supplement 1). and PcCdc50A or PcCdc50B between P2 and P3 mem- PS-selective P4-ATPases normally contain polar resi- brane fractions (Figure 2) and detergent-solubilization dues at these two clusters (two Gln at the correspond- screenings (Figure supplement 3, panels E and F) ing positions 65 and 66 of PcATP2, and two Asn at demonstrated that the yeast-produced PcATP2 is the equivalent positions 455 and 456 of PcATP2) able to associate with both PcCdc50B and PcCdc50A, that establish H-bonds with PS. In contrast, PC-selec- two of the three Plasmodium-encoded Cdc50 iso- tive P4-ATPases usually contain non-polar residues forms. This promiscuous capacity to associate with at the same positions [24]. PcATP2 has one polar more than one Cdc50 β-subunit is known in other residue at each cluster (Figure supplement 1) that, P4-ATPases [10], although the physiological meaning in principle, could be sufficient to coordinate PS, of this dual interaction remains to be determined. We PE, and even PC. Although in the ATPase assays it also found that S. cerevisiae-expressed PcCdc50A was is possible the presence of a non-functional fraction N-glycosylated (Figure 3). In contrast, the majority of of PcATP2, it is also possible that PcATP2 is partially PcCdc50B expressed in S. cerevisiae seemed not to autoinhibited using a mechanism similar to its yeast contain N-glycosylation (Figure 3). N-glycosylation homolog Drs2p [23,45]. In fact, the GFAFS motif at is a typical signature of Cdc50 proteins [39], and its the C-terminus of Drs2p involved in enzyme’s auto- role seems to be organism-dependent. While in inhibition by interacting with the nucleotide binding Leishmainia infantum Cdc50 glycosylation affects the site [23], is also well conserved in PcATP2 (as well in catalytic activity of the P4-ATPase/Cdc50 complex the other Plasmodium ATP2 orthologs), although the [42], N-glycosylation of the Cdc50 β-subunit of the length of the C-terminal of the Plasmodium human ATP8A2 is required to form a stable complex sequences is ∼60 amino acid shorter (Supplementary with the P4-ATPases, with no apparent role on its cat- Figure 5). In addition, PcATP2 shares with Drs2p a alytic activity [55]. Interestingly, the P. yoeli Cdc50A few residues in a short loop region of the nucleotide expressed in both gametocytes and ookinetes does domain (residues 698–704 in Drs2p, see Supplemen- not seem to be glycosylated as judged by the western tary Figure 5) that directly interact with the GFAFS blots bands ([29], and Jing Yuan, personal communi- motif to stabilize the autoinhibitory conformation cation). Clearly, due to its potential functional and/or [23]. Moreover, and as suggested by the presence of stabilizing role, further studies to unravel the glycosy- a PI4P binding motif on its amino acid sequence lation state of Cdc50A and Cdc50B in the parasite are (Figure supplement 6)[23], PcATP2 is also upregu- necessary. lated by this lipid as we observed a statistically signifi- FSEC experiments confirmed the association of cant increase of POPS and POPE-stimulated ATPase PcATP2 with PcCdc50B, as well as the stability of activity in the presence of PI4P (Figure 6). PI4P is an this complex in DDM/CHS detergent micelles essential lipid for the malaria parasite since the inhi- (Figure 5). Our data also suggested that the cholesterol bition of PI4P synthesis in the Golgi apparatus com- analog CHS is important to stabilize this complex pletely blocks the late steps of the intraerythrocytic since its presence highly improved the yield of co- cycle of P. falciparum by interfering with membrane solubilization of PcATP2 and PcCdc50B with DDM biogenesis around the developing merozoites [56]. (Figure supplement 3). In the recent cryo-EM struc- PI4P is found at the plasma membrane and at the ture of the human ATP8A1 in complex with Golgi apparatus in all stages of the erythrocytic Cdc50A, a CHS molecule was found at the heterodi- cycle [57]. Interestingly, the P. berghei ATP2 was mer interface between TMs 7 and 10 of ATP8A1 also localized at the interface parasite/host [4]. It and TM2 of Cdc50A, suggesting a potential role of will be relevant to investigate if the lethal effects as CHS on preserving the heterodimer [24]. Although a result of depleting PI4P or suppressing ATP2 are Plasmodium parasites cannot synthesize de novo somehow connected and linked to the biogenesis of cholesterol, lipidomic studies have revealed the pres- the parasite’s membranes. ence of this lipid in intracellular membrane compart- In conclusion, our work provides the first insights ments of infected erythrocytes [46], most likely into the function and regulation of a Plasmodium obtained from the host cell-membrane. ATP2, as well as the identity of two Cdc50 β-sub- The PcATP2/PcCdc50B complex produced in units of this transporter. This information rep- S. cerevisiae showed ATPase activity in detergent resents a first step towards the understanding of micelles, in the presence of the aminophospholipids the essential role of this transporter during malaria POPS and POPE (Figure 6), thus demonstrating the infection, and towards its validation as antimalarial capacity of PcATP2 to recognize these two drug target. 146 A. Lamy et al. Acknowlegments function, and enigmas. Front Physiol. 2016;7. Article no. 275. We would like to thank Dr Christine Jaxel for all the support [11] van Veen S, Martin S, Van den Haute C, et al. and discussions during this project, and to Dr Eugene Dia- ATP13A2 deficiency disrupts lysosomal polyamine tloff and Dr Jing Yuan for critically reading the manuscript. export. Nature. 2020;578:419–424. We also thank Quentin Rochette for his early work on this [12] McKenna MJ, Sim SI, Ordureau A, et al. The endo- project. This work was supported by the ANR grants ANR- plasmic reticulum P5A-ATPase is a transmembrane 14-CE09-0022 to GL, and ANR-18-C811-0009-02 to JLVI, helix dislocase. Science. 2020;369:eabc5809. the French Infrastructure for Integrated Structural Biology [13] Kirk K. Ion regulation in the malaria parasite. Annu (FRISBI; ANR-10- INSB-05), and the Centre National de Rev Microbiol. 2015;69:341–359. la Recherche Scientifique (CNRS). In memory of Prof [14] Jiménez-Díaz MB, Ebert D, Salinas Y, et al. (+)-SJ733, Ronald H. Kaback. a clinical candidate for malaria that acts through ATP4 to induce rapid host-mediated clearance of plasmo- dium. Proc Natl Acad Sci U S A. 2014;111:E5455– Disclosure statement E5462. [15] Eckstein-Ludwig U, Webb RJ, van Goethem IDA, et al. No potential conflict of interest was reported by the author(s). Artemisinins target the SERCA of plasmodium falci- parum. Nature. 2003;424:957–961. [16] Cardi D, Pozza A, Arnou B, et al. Purified E255L Funding mutant SERCA1a and purified PfATP6 are sensitive This work was supported by Agence Nationale de la to SERCA-type inhibitors but insensitive to artemisi- Recherche [grant number ANR-14-CE09-0022,ANR-18- nins. J Biol Chem. 2010;285:26406–26416. C811-0009-02]; French Infrastructure for Integrated Struc- [17] Paulusma CC, Elferink RPJO. P4 ATPases–the physio- tural Biology [grant number ANR-10- INSB-05]. logical relevance of lipid flipping transporters. FEBS Lett. 2010;584:2708–2716. [18] Poulsen LR, López-Marqués RL, Pedas PR, et al. A phospholipid uptake system in the model plant arabi- ORCID dopsis thaliana. Nat Commun. 2015;6. Article no. 7649. Thibaud Dieudonné http://orcid.org/0000-0001-6988- [19] Huang W, Liao G, Baker GM, et al. Lipid flippase sub- 4121 unit Cdc50 mediates drug resistance and virulence in Alex Perálvarez-Marín http://orcid.org/0000-0002-3457- Cryptococcus neoformans. MBio. 2016;7:e00478. 0875 [20] Cowell AN, Istvan ES, Lukens AK, et al. Mapping Guillaume Lenoir http://orcid.org/0000-0002-8759-5179 the malaria parasite druggable genome by using in Cédric Montigny http://orcid.org/0000-0003-0905-9861 vitro evolution and chemogenomics. Science. 2018; José Luis Vázquez-Ibar http://orcid.org/0000-0002-8314- 359:191–199. 3378 [21] Paulusma CC, Folmer DE, Ho-Mok KS, et al. ATP8B1 requires an accessory protein for endoplasmic reticu- lum exit and plasma membrane lipid flippase activity. References Hepatology. 2008;47:268–278. [22] Bryde S, Hennrich H, Verhulst PM, et al. CDC50 [1] World malaria report. 2019th ed. Geneva, 2019. proteins are critical components of the human class- [2] Gosling R, von Seidlein L. The future of the RTS,S/ 1 P 4-ATPase transport machinery. J Biol Chem. AS01 malaria vaccine: an alternative development 2010;285:40562–40572. plan. PLoS Med. 2016;13:e1001994. [23] Timcenko M, Lyons JA, Januliene D, et al. Structure [3] Burrows JN, Duparc S, Gutteridge WE, et al. New and autoregulation of a P4-ATPase lipid flippase. developments in anti-malarial target candidate and Nature. 2019;571:366–370. product profiles. Malar J. 2017;16:26. [24] Hiraizumi M, Yamashita K, Nishizawa T, et al. Cryo- [4] Kenthirapalan S, Waters AP, Matuschewski K, et al. EM structures capture the transport cycle of the Functional profiles of orphan membrane transporters P4-ATPase flippase. Science. 2019;365:1149–1155. in the life cycle of the malaria parasite. Nat Commun. [25] Bai L, Kovach A, You Q, et al. Autoinhibition and acti- 2016;7:10519. vation mechanisms of the eukaryotic lipid flippase [5] Bushell E, Gomes AR, Sanderson T, et al. Functional Drs2p-Cdc50p. Nat Commun. 2019;10. doi:10.1038/ profiling of a plasmodium genome reveals an abun- s41467-019-12191-9. dance of essential Genes. Cell. 2017;170:260–272.e8. [26] He Y, Xu J, Wu X, et al. Structures of a P4-ATPase lipid [6] Zhang M, Wang C, Otto TD, et al. Uncovering the flippase in lipid bilayers. Protein Cell. 2020;11:458–463. essential genes of the human malaria parasite plasmo- [27] Nakanishi H, Irie K, Segawa K, et al. Crystal structure dium falciparum by saturation mutagenesis. Science. of a human plasma membrane phospholipid flippase. J 2018;360:eaap7847. Biol Chem. 2020;295:10180–10194. [7] Martin RE. The transportome of the malaria parasite. [28] Nakanishi H, Nishizawa T, Segawa K, et al. Transport Biol Rev. 2020;95:305–332. cycle of plasma membrane flippase ATP11C by cryo- [8] Weiner J, Kooij T. Phylogenetic profiles of all mem- EM. Cell Rep. 2020;32:108208. brane transport proteins of the malaria parasite high- [29] Gao H, Yang Z, Wang X, et al. ISP1-anchored polariz- light new drug targets. Microb Cell. 2016;3:511–521. ation of GCβ/CDC50A complex Initiates malaria [9] Palmgren MG, Nissen P. P-Type ATPases. Annu Rev Ookinete gliding motility. Curr Biol. 2018;28:2763– Biophys. 2011;40:243–266. 2776.e6. [10] Andersen JP, Vestergaard AL, Mikkelsen SA, et al. [30] Bisio H, Lunghi M, Brochet M, et al. Phosphatidic acid P4-ATPases as phospholipid flippases—structure, governs natural egress in Toxoplasma gondii via a Emerging Microbes & Infections 147 guanylate cyclase receptor platform. Nat. Microbiol. lipid translocase dephosphorylation is controlled by 2019;4:420–428. phosphatidylinositol-4-phosphate. J Biol Chem. [31] Azouaoui H, Montigny C, Ash M-R, et al. A high-yield 2012;287:13249–13261. co-expression system for the purification of an intact [45] Azouaoui H, Montigny C, Dieudonné T, et al. High Drs2p-Cdc50p lipid flippase complex, critically depen- phosphatidylinositol 4-phosphate (PI4P)-dependent dent on and stabilized by phosphatidylinositol-4- ATPase activity for the Drs2p-Cdc50p flippase after phosphate. PLoS One. 2014;9:e112176. removal of its N- and C-terminal extensions. J Biol [32] Lenoir G, Menguy T, Corre F, et al. Overproduction in Chem. 2017;292:7954–7970. yeast and rapid and efficient purification of the rabbit [46] Gulati S, Ekland EH, Ruggles KV, et al. Profiling the SERCA1a Ca2+-ATPase. Biochim Biophys Acta - essential Nature of lipid Metabolism in asexual Biomembr. 2002;1560:67–83. blood and Gametocyte stages of Plasmodium falci- [33] Kubala MH, Kovtun O, Alexandrov K, et al. Structural parum. Cell Host Microbe. 2015;18:371–381. and thermodynamic analysis of the GFP:GFP-nano- [47] Juge N, Moriyama S, Miyaji T, et al. Plasmodium fal- body complex. Protein Sci. 2010;19:2389–2401. ciparum chloroquine resistance transporter is a H+ [34] Tsirigos KD, Peters C, Shu N, et al. The TOPCONS -coupled polyspecific nutrient and drug exporter. web server for consensus prediction of membrane Proc Natl Acad Sci U S A. 2015;112:3356–3361. protein topology and signal peptides. Nucleic Acids [48] David-Bosne S, Florent I, Lund-Winther A-M, et al. Res. 2015;43:W401–W407. Antimalarial screening via large-scale purification of [35] Katoh K, Misawa K, Kuma K-I, et al. MAFFT: a novel Plasmodium falciparum Ca2+-ATPase 6 and in vitro method for rapid multiple sequence alignment based studies. FEBS J. 2013;280:5419–5429. on fast Fourier transform. Nucleic Acids Res. [49] Qureshi AA, Suades A, Matsuoka R, et al. The molecu- 2002;30:3059–3066. lar basis for sugar import in malaria parasites. Nature. [36] Fiser A, Šali A. MODELLER: generation and refine- 2020;578:321–325. ment of homology-based protein structure models. [50] Kim J, Tan YZ, Wicht KJ, et al. Structure and drug Methods Enzymol. 2003;374:461–491. resistance of the plasmodium falciparum transporter [37] Pettersen EF, Goddard TD, Huang CC, et al. UCSF PfCRT. Nature. 2019;576:315–320. Chimera – a visualization system for exploratory [51] Arnou B, Montigny C, Morth JP, et al. The plasmo- research and analysis. J Comput Chem. 2004;25: dium falciparum Ca(2+)-ATPase PfATP6: insensitive 1605–1612. to artemisinin, but a potential drug target. Biochem [38] Feller SE, MacKerell AD. An improved empirical Soc Trans. 2011;39:823–831. potential energy function for molecular simulations of [52] Fujita H, Kida Y, Hagiwara M, et al. Positive charges of phospholipids. J Phys Chem B. 2000;104:7510–7515. translocating polypeptide chain retrieve an upstream [39] Costa SR, Marek M, Axelsen KB, et al. Role of post- marginal hydrophobic segment from the endoplasmic translational modifications at the β-subunit ectodo- reticulum lumen to the translocon. Mol Biol Cell. main in complex association with a promiscuous 2010;21:2045–2056. plant P4-ATPase. Biochem J. 2016;473:1605–1615. [53] Tone T, Nakayama K, Takatsu H, et al. ATPase reac- [40] Jidenko M, Lenoir G, Fuentes JM, et al. Expression in tion cycle of P4-ATPases affects their transport from yeast and purification of a membrane protein, the endoplasmic reticulum. FEBS Lett. 2020;594:412– SERCA1a, using a biotinylated acceptor domain. 423. Protein Expr Purif. 2006;48:32–42. [54] Molbaek K, Tejada M, Ricke CH, et al. Purification [41] Rodríguez-Banqueri A, Errasti-Murugarren E, and initial characterization of Plasmodium falciparum Bartoccioni P, et al. Stabilization of a prokaryotic K + channels, PfKch1 and PfKch2 produced in LAT transporter by random mutagenesis. J Gen Saccharomyces cerevisiae. Microb Cell Fact. 2020;19. Physiol. 2016;147:353–368. doi:10.1186/s12934-020-01437-7. [42] García-Sánchez S, Sánchez-Cañete MP, Gamarro F, [55] Coleman JA, Molday RS. Critical role of the β- et al. Functional role of evolutionarily highly con- subunit CDC50A in the stable expression, Assembly, served residues, N-glycosylation level and domains Subcellular localization, and lipid transport activity of of the Leishmania miltefosine transporter-Cdc50 sub- the P 4 -ATPase ATP8A2. J Biol Chem. 2011;286: unit. Biochem J. 2014;459:83–94. 17205–17216. [43] Coleman JA, Kwok MCM, Molday RS. Localization, [56] McNamara CW, Lee MCS, Lim CS, et al. Targeting purification, and functional reconstitution of the P4- plasmodium PI(4)K to eliminate malaria. Nature. ATPase Atp8a2, a phosphatidylserine flippase in 2013;504:248–253. photoreceptor disc membranes. J Biol Chem. [57] Ebrahimzadeh Z, Mukherjee A, Richard D. A map of 2009;284:32670–32679. the subcellular distribution of phosphoinositides in [44] Jacquot A, Montigny C, Hennrich H, et al. the erythrocytic cycle of the malaria parasite Phosphatidylserine stimulation of Drs2p·Cdc50p Plasmodium falciparum. Int J Parasitol. 2018;48:13–25.