In this study, the following primary antibodies and antisera were used: Rabbit anti-HA antibody (Sigma-Aldrich); rat anti-HA antibody (Roche); mouse anti-GFP antibody (Sigma-Aldrich); rabbit polyclonal antisera against Pfs230 (Biogenes), Pfs25 (ATCC), Pfs16, AMA1 (Boes et al., 2015); mouse polyclonal antisera against falcilysin (Weißbach et al., 2017), Pf39 (Simon et al., 2009) and actin I (Ngwa et al., 2013). The generation of polyclonal mouse antisera against PfPNPLA2, 3 and 4 is described below.

Recombinant proteins of PfPNPLA2 (spanning aa 571-998), PfPNPLA3 (spanning aa 589-763) and PfPNLPA4 (spanning aa 2-187) were expressed for the generation of antisera. PfPNPLA2 was expressed with an N-terminal maltose binding protein (MBP)-tag using the expression vector pIH902. Cloning was mediated by the addition of the restriction sites EcoRI and PstI (for primer sequences see Supplementary Table 1 ). PfPNPLA3 was expressed with a C-terminal His-tag using pET28c (+) and cloning was mediated by the addition of NcoI/XhoI restriction sites. PfPNPLA4 was expressed as a fusion protein with an N-terminal MBP-tag using pMalc5x. Cloning was mediated by XmnI/PstI restriction sites. Recombinant fusion proteins were expressed using E. coli BL21 (DE3) RIL (Stratagene) and purified via affinity chromatography using amylose resin (New England Biolabs) for the MBP-tagged proteins and complete™ His-tag purification resin (Sigma-Aldrich) for the His-tagged proteins according to the manufacturers’ protocols. After PBS buffer exchange via filter centrifugation using Amicon Ultra 15 (Sigma-Aldrich), protein concentrations were determined by Bradford assay. 100 µg recombinant proteins emulsified in Freund’s incomplete adjuvant (Sigma-Aldrich) were injected into 6 weeks-old NMRI mice (Charles River Laboratories) via subcutaneous injection. After 4 weeks, a boost with 50 µg recombinant proteins was injected. Ten days after the boost, mice were anesthetized by intraperitoneal injection of a ketamine/xylazine mixture (Sigma-Aldrich) according to the manufacturer’s protocol and immune sera were collected via heart puncture. Sera were pooled from three mice immunized with the same antigen. Experiments for the generation of antisera in mice were approved by the animal welfare committees of the District Council of Cologne, Germany (ref. no. 84-02.05.30.12.097 TVA).

PfPNPLA2- and PfPNPLA4-KD parasite lines were generated via single cross-over homologous recombination using the pARL-HA-glmS vector (Flammersfeld et al., 2020). A 1015-bp gene fragment homologous to the 3’ end of the gene coding for PfPNPLA2 and a 840-bp gene fragment of the gene coding for PfPNPLA4 (excluding the stop codons) were amplified using corresponding primers (for primer sequences see Supplementary Table 1 ). Ligation of the inserts with the vector backbone was mediated by NotI/AvrII restriction sites. A synchronized wildtype (WT) NF54 culture containing 5% ring stages was loaded with 100 µg vector in transfection buffer via electroporation as described. Parasites carrying the vector were selected by the addition of WR99210 (Jacobus Pharmaceuticals) in a final concentration of 4 nM and successful integration of the vector was confirmed by diagnostic PCR using (1) 5’-integration primer, (2) 3’-integration primer, (3) pARL-HA-glmS forward primer and (4) pARL-HA-glmS reverse primer (for primer location see Supplementary Figure 2A ; for primer sequences see Supplementary Table 1 ).

Total RNA was isolated from synchronous ring, trophozoite and schizont cultures, as well as from Percoll-gradient enriched immature, mature and activated gametocytes as described before (Flammersfeld et al., 2020). RNA preparation and subsequent cDNA synthesis were performed as described (Flammersfeld et al., 2020). Transcripts for PfPNPLA2 (250 bp), PfPNPLA3 (250 bp) and PfPNPLA4 (250 bp) were amplified in 25 cycles using the corresponding primers (for primer sequences see Supplementary Table 1 ). To confirm purity of the asexual blood stage and gametocyte samples, transcript amplifications of ama1 (407 bp) and of ccp2 (286 bp) were performed using the corresponding primers. Amplification of the housekeeping gene aldolase (378 bp) was used as loading control.

Synchronized asexual cultures of a wildtype (3D7) strain and the PNPLA2-iKD line were maintained as previously described (Trager and Jensen, 1976; Trager et al., 1980). Parasite cultures (200 µl of compact infected RBCs) were collected by centrifugation (10 min at 1200 rpm at room temperature). Parasites were fixed with 4% paraformaldehyde and 0.0075% glutaraldehyde for 30 min at room temperature. Cells were centrifuged at 3000 rpm for 2 mins and the supernatant was discarded. The sample was then washed twice with PBS and the supernatant removed. The pellet was resuspended gently in 1 ml of 4% FBS (fetal bovine serum) in PBS and spun again as previously. Cells were permeabilized using 0.1% Triton X-100 in PBS for 10 min. Cells were pelleted and the sample was washed thrice with PBS. The pellet was saturated with 4% FBS in PBS and incubated at 4°C for 1 h. There after the primary antibodies of HA-tag (Roche) or anti-PNPLA2/PNPLA3/PNPLA4 (homemade) LBPA (Tebu bio) were added at dilutions of 1:500 or 1:50, respectively. The antibodies were then washed away thrice with PBS. The secondary antibodies (AlexaFluor488/546 anti-rat/mouse) were used at 1:1000 dilutions in FBS and incubated for 1 h. Hoechst stain was used at 1:25000 dilution for 20 mins and finally samples were washed with PBS thrice. Pellets were resuspended in 150 µl of PBS and then gently mixed with the same amount of Fluorogel. 15 µl of each sample was pipetted on a microscope slide and overlayed with a square coverslip. Samples were allowed to solidify in the dark at 37°C for 1 h. Specimens were examined with an Axio Imager; ZEISS, ×100 magnification. Images were treated using ImageJ software. Indirect immunofluorescence images were obtained for gametocytes as previously described (Flammersfeld et al., 2020).

Mixed asexual blood stages cultures of a wildtype (3D7) strain and the PNPLA2-iKD line were maintained as previously described (Trager and Jensen, 1976; Trager et al., 1980). Parasite cultures (200 µl of compact infected RBCs) were collected by centrifugation (10 min at 1200 rpm at room temperature). Cultures were resuspended in media with or without 2.5 mM GlcN and incubated for 48 h. These cultures were then collected by centrifugation (10 min at 1200 rpm at room temperature) and resuspended in prewarmed RPMI containing 100 µM Mitotracker™ and incubated for 45 min. After the incubation, specimens were examined with an Axio Imager; ZEISS, ×100 magnification. Images were treated using ImageJ software.

Western blotting was performed as previously described (Flammersfeld et al., 2020). Briefly, tightly sorbitol-synchronized schizont stage parasite samples underwent saponin lysis to remove red blood cell membrane (Hall et al., 1983). The sample was then resuspended in a lysis buffer containing an anti-protease cocktail (0.5% Triton X- 100, 4% w/v SDS, 0.5× PBS). SDS-PAGE buffer was added, and samples were denatured at 95°C for 10 mins. Equal sample volumes were then separated by SDS-PAGE and proteins were transferred onto a Hybond ECL nitrocellulose membrane (Amersham Biosciences) according to the manufacturer’s protocol. The blot was blocked using Tris-buffered saline containing 5% w/v skim milk, pH 7.5, followed by overnight incubation at 4°C with polyclonal mouse immune sera specific for Pf39 (dilution 1:5,000) or using polyclonal rabbit immune sera specific to anti-HA (1:1,000). After washing, membranes were incubated with the respective alkaline phosphatase-conjugated secondary antibody (Sigma-Aldrich) for 1 h at RT and developed in a solution of nitroblue tetrazolium chloride (NBT) and 5-brom-4-chlor-3-indoxylphosphate (BCIP; Sigma-Aldrich) for 5-30 min at RT. Blots were scanned and processed using the Adobe Photoshop CS software. Band intensities were measured using the ImageJ programme version 1.51f.

To analyze asexual blood stage replication and length of the intraerythrocytic replication cycle in the PNPLA2-HA-iKD parasite line, experiments were set-up as follows: Expression of PNPLA2 was knocked-down in transgenic PNPLA2-HA-iKD parasite line by treatment of the culture with GlcN (D-(+)-glucosamine hydrochloride; Sigma Aldrich) at a final concentration of 2.5 mM. GlcN treatment started one cycle before the experimental set-up. The GlcN-treated culture was compared to the untreated culture cultivated in normal cell culture medium. This assay was also done in minimal media only including RPMI, glucose, hypoxanthine (with 1 M NaOH), FA-free bovine serum albumin (60 µM) and FFA C16:0 (30 µM) and C18:1 (30 µM). To exclude an unspecific effect of GlcN, the parental WT line was used as control. For each assay, three experiments were performed, each in triplicate.

To induce knockdown of PfPNPLA2, a synchronized parasite culture of the Pf PNPLA2-HA-iKD line was treated with 2.5 mM GlcN for 72 h. Gametocytogenesis was subsequently induced with lysed blood cells while the GlcN treatment was continued. The gametocytemia was evaluated by Giemsa smears over a period of six days. Untreated Pf PNPLA2-HA-iKD cultures were used as controls.

Lipidomics analysis was performed on 3-4 independent cell harvests of the PNPLA2-HA-iKD line. Highly synchronous parasite cultures grown for at least 48 hours with or without glucosamine (ring/trophozoite or schizont stage) were harvested by transferring the cultures to pre-warmed 50-ml tubes and the total amount of parasites was determined by Giemsa staining and via counting on a hemocytometer. The cultures were metabolically quenched via rapid cooling to 0°C by suspending the tube over a dry ice/100% ethanol slurry mix, while continually stirring the solution. Following quenching, parasites were constantly kept at 4°C and liberated from the enveloping RBCs by mild saponin treatment with 0.15% w/v saponin/0.1% w/v BSA PBS for 5 min. After three washing steps with 1 ml ice-cold PBS followed by centrifugation, the parasite pellet was subjected to lipid extraction using chloroform and methanol (Vaid et al., 2008) in the presence of butylhydoxytluene, PC (C21:0/C21:0), C13:0 or C15:0 FA standards (Avanti). Lipid extraction, separation and analyses were all performed as previously described in Flammersfeld et al., 2020 (Flammersfeld et al., 2020). FA methyl esters were detected by GC-MS (Agilent 5977A- 7890B) according to the method described in Ramakrishnan et al. (2012) (Ramakrishnan et al., 2012) and quantified by Mass Hunter software (Agilent). Each FA was quantified according to a calibration curve generated with authentic fatty methyl ester standards.

Data are expressed as mean ± SD. Statistical differences were determined using unpaired two-tailed Student’s t test, as indicated. Values of p < 0.05 were considered statistically significant. Significances were calculated using GraphPad Prism 5 and are represented in the figures as follows: ns, not significant p > 0.05; *p < 0.05; **p < 0.01.

Gene PF3D7_1358000 encodes for a putative patatin-like phospholipase (Flammersfeld et al., 2018), termed PfPNPLA2. The protein is predicted to be a phospholipase A2 (PLA2) with a patatin-like phospholipase domain like that of PfPNPLA1 (Flammersfeld et al., 2020), and an estimated molecular weight of 238.2 kDa ( Figure 1A ). PfPNPLA2 comprises a transmembrane domain at the N-terminal region of the protein. Furthermore, the characteristic GXSXG motif is present in this protein between aa 1166 and 1171, and a conserved signature catalytic aspartate found in most eukaryotic PLA2 at residue 1201 ( Figure 1A ) (Malley et al., 2018). It has been shown that all plasmodial patatin-like phospholipase homologs share the above-mentioned motif and catalytic aspartate, supporting the classification of PNPLA2 (Flammersfeld et al., 2018; Flammersfeld et al., 2020). In addition to PfPNPLA1 and PfPNPLA2, two more members of the PfPNPLA family were previously identified, i.e. PfPNPLA3 (151 kDa, encoded by gene PF3D7_0924000) and PfPNPLA4 (283 kDa, encoded by gene PF3D7_0218600). PfPNPLA3 comprises, in addition to the central patatin-like phospholipase domain, two N-terminal transmembrane domains ( Supplementary Figure 1A ).

Diagnostic reverse transcriptase (RT)-PCR was performed to investigate expression of pfpnpla2, pfpnpla3 and pfpnpla4 in the P. falciparum blood stages. This showed the synthesis of transcripts, from all three genes, in the ring, trophozoite, and in schizonts stages. Similar transcript levels were detected in immature (stage II-IV) and mature (stage V) gametocytes as well as in gametocytes at 15’ post-activation (p.a.) ( Supplementary Figure 1B ). Transcript analysis of the housekeeping gene aldolase was used as a loading control, while purity of the asexual blood stage and gametocyte samples was demonstrated by amplification of transcripts for the asexual blood stage-specific gene pfama1 (apical membrane antigen 1) (Peterson et al., 1989) and for the gametocyte-specific gene pfccp2 (LCCL-domain protein 2) (Pradel et al., 2004). cDNA preparations lacking the reverse transcriptase were used to verify that samples were devoid of gDNA ( Supplementary Figure 1B ).

For functional characterization of PfPNPLA2, we generated a conditional PfPNPLA2-HA-iKD line (using the pARL-HA-glmS vector) by fusing the sequences coding for a hemagglutinin A (HA)-tag to the 3´-coding region of pfpnpla2 ( Supplementary Figure 2A ). In the PfPNPLA2-HA-iKD, the pfpnpla2 mRNA is expressed under the control of the glmS ribozyme, which catalyzes its own cleavage in the presence of GlcN, triggering transcript degradation. Vector integration was confirmed by diagnostic PCR ( Supplementary Figure 2B ). Immunoblotting of asexual blood stage lysate with an anti-HA antibody confirmed the synthesis of an HA-tagged PfPNPLA2 fusion protein in the untreated PfPNPLA2-HA-iKD line, running at a molecular weight of 238 kDa ( Figure 1B ). When the disruption of Pf PNPLA2-HA-iKD was induced by treating asexual blood stage parasites with 2.5 mM GlcN for 72 h, PNPLA2-HA-iKD levels were reduced compared to the untreated control ( Figure 1B ).

The untreated PfPNPLA2-HA-iKD line was used for in-depth protein expression analysis. Furthermore, an HA-tagged PfPNPLA4 parasite line was generated ( Figure S1B ), using the above mentioned pARL-HA-glmS vector. In addition, polyclonal mouse antisera were generated against recombinant PfPNPLA2, PfPNPLA3 and PfPNPLA4 peptides to be used for protein expression analysis. Immunofluorescence assays (IFA) revealed that PfPNPLA2 is expressed throughout all asexual and sexual blood stages ( Figure 1C ). PfPNPLA2 is predominantly expressed in the schizont stages and seems to localize to vesicular structures of individual merozoites formed within the parasite vacuole. PfPNPLA2 could further be detected at distinct loci within the cytoplasm of gametocytes ( Figure 1C ). Closer co-localization studies revealed that PfPNPLA2 does not colocalize with PfPNPLA3 (or PfPNPLA4), which suggests that each phospholipase has non-redundant functions ( Figures 1D, E ). Further co-localization IFA using Mitotracker™ revealed that, in the ring to schizont-stage parasites, this protein localizes to the mitochondrion ( Figure 1F ) (Mullin et al., 2006; Botté et al., 2013). The presence of PfPNPLA2 at the mitochondrion led us to investigate the effect of the loss of the protein on the mitochondrion’s morphology by IFA using Mitotracker. Loss of the protein leads the mitochondrion to take on abnormal morphologies during the trophozoite and schizont stages. Instead of the usual elongated and branched form, the parasites that have lost the protein have faded or misplaced mitochondrion, among other abnormalities, also confirmed statistically significant over parasite population ( Figure 1G ). Furthermore, previous data show the protein localizing close to the apicoplast (Burda et al., 2021) leading us to the idea that this protein would have a dual function.

To determine the importance of PfPNPLA2 for parasite survival, we performed growth assays upon presence or absence (+GlcN) of the protein. Agreeing with previous phenotypic scores predicted for the protein (Nolan et al., 2017; Singh et al., 2019), knocking PfPNPLA2 down in the PfPNPLA2-HA-iKD line, by treatment with 2.5 mM GlcN, had no significant effect on intraerythrocytic replication during a 72 h period in regular in vitro culture conditions, when compared to untreated PfPNPLA2-HA-iKD parasites and both treated or untreated WT NF54 parasites ( Figure 2A ). Furthermore, quantification of ring, trophozoite and schizont stages, after GlcN treatment, at different time points, did not show any major effect on parasite stage development of PfPNPLA2-HA-iKD and WT NF54 parasites ( Supplementary Figure 3 ), suggesting that the lack of PfPNPLA2 does not delay the blood stage cycle. However, we previously showed that the host nutritional and lipid content can influence the essentiality of a protein, changing it from dispensable in normal/rich lipid conditions, to become essential under low nutritional conditions (Amiar et al., 2020). To analyze this, parasites were cultivated in minimal media (Mi-Ichi et al., 2006) containing the minimal FA requirement for in vitro culture, i.e. C16:0 and C18:1 (in combination to FA-free BSA). Interestingly, in these lipid-limiting conditions, the PfPNPLA2-HA-iKD line, treated with GlcN, displayed a significant growth delay after 96 hours of intracellular development, linked to a decrease in ring stage parasites ( Figures 2C, D ). In sexual stage parasites, the depletion of this protein did not affect the gametocyte population ( Figure 2B ). These data may suggest that PNPLA2 is more important for asexual blood stage development than for gametocytogenesis, and the essentiality of this protein cannot be ruled out, as we were not able to attain a complete absence of the protein following knockdown. More importantly, this protein seems to be important under nutrient poor conditions and may strongly be dependent on the extracellular environment of the host.

To investigate the effect the loss of PfPNPLA2 has on overall parasite lipid metabolism, we conducted global lipidomic analyses using gas-chromatography mass spectrometry on all asexual parasite stages (synchronized ring, trophozoite and schizont stages). It should be noted that despite PfPNPLA2’s essentiality in a lipid poor environment, all lipidomic experiments were conducted in normal culture conditions (with Albumax) as otherwise, cultures did not yield high enough parasitemia, once the protein was knocked down.

We first sought to assess any potential defects in total lipid content and then quantified major lipid classes sustaining parasite survival, free FAs (building blocks of most lipids; FFA), diacylglycerols (major lipid precursors, DAG), PLs (major membrane lipids), and triacylglycerols (the main component of lipid storage bodies; TAG).

During ring stage development, the knockdown of PfPNPLA2 in the PfPNPLA2-HA-KD line had no significant effect on the global FA profile of the total parasite lipid content ( Figure 3A ). However, further analyses, through separation of neutral and total PLs groups via 1D thin-layer chromatography (TLC), revealed that upon PfPNPLA2 knockdown, there was significant increases in TAG, FFA and DAG, the lipid classes constituting the parasite’s neutral lipid content, but there was no impact on the total PL content of the parasite ( Figure 3B ). Further analyses to determine the FA composition of each lipid class showed that both FFA and DAG displayed significant increases in their C16:0 and C18:1 content ( Figures 3C, D ), which are predominantly scavenged from the extracellular media (Mi-Ichi et al., 2006; Mi-Ichi et al., 2007). This also correlates to our findings that growth defects following PfPNPLA2 knockdown occur under low lipid content only containing C16:0 and C18:1. However, no changes could be detected in the FA profiles of TAG or total PL ( Supplementary Figures 4A, B ) under PfPNPLA2 deficiency.

Knockdown of PfPNPLA2 in trophozoite stage parasites exhibited limited changes in the overall FA profile of the total lipid content of the parasite, such as a significant increase in C18:0 and a significant decrease in C18:1 ( Figure 3E ). Further analyses of parasite lipid classes showed no changes in total neutral lipid content (either FFA, DAG or TAG) or PL content ( Figure 3F ). This lack of phenotype might be directly linked to a lower activity of the enzyme during trophozoite stages, evidenced by its RNA expression being the lowest during trophozoite stages (PlasmoDB, 2022).

The most severe effects were observed during the schizont stage when the protein is most expressed. First, in the total lipid FA profile, the loss of PfPNPLA2 resulted in a significant decrease of the relative abundance of palmitic acid (C16:0) ( Figure 3G ). Similar to the ring stage, knockdown of PfPNPLA2 in schizonts resulted in a significant increase of the storage lipid, TAG and no change to PL. However, unlike in ring stage, in schizonts, the levels of both FFA and DAG were significantly reduced ( Figure 3H ), but there were no changes in their FA species profile ( Supplementary Figures 4C, D ). The reduction of PfPNPLA2 levels did not lead to any changes in the relative amount of total PL levels ( Figure 3H ). Analysis of the FA content of TAG accumulating in schizont stage in the absence of PfPNPLA2 revealed a significant increase of TAG molecular species containing C16:0 and C18:1 ( Figure 3I ), as for rings, suggesting increased host scavenging. It should also be noted that there were no changes in the total lipid abundance at any stage, despite seeing clear changes in individual lipid groups ( Figures 3A, E, G ).

These data collectively suggest that PfPNPLA2 is most important for FFA generation/acquisition as well as the synthesis of neutral lipids likely for storage and timely mobilization during the transition from the schizont to the ring stage parasites.

To further investigate the putative phospholipase activity of PfPNPLA2, notably determine its potential lipid substrates and products, we sought to analyze each PL class and their molecular content in schizonts by 1D or 2D HPTLC coupled to GCMS lipidomic analyses.

The disruption of PfPNPLA2 significantly impacted parasite PL homeostasis ( Figure 4A ). There were slight, yet significant reductions in the content of phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS) and phosphatidylinositol (PI) that seemed more marginal. PLA2 usually catalyze the hydrolysis of one of the FA moieties from PL, hence generating lysoPL and FA. The levels of major lysolipids, lysophophatidylcholine (LPC), and lysophosphatidic acid (LPA) were not significantly impacted in the absence of PfPNPLA2, suggesting they might not be the direct products of the enzyme. The most striking change, however, was an extremely significant increase of phosphatidylglycerol (PG). PG is usually a low represented class of membrane PL that is also the substrate for the essential de novo synthesis of the mitochondrial-signature lipid class, cardiolipin. However, no change was observed in the levels of cardiolipin (CL) in the absence of PfPNPLA2, again suggesting CL not being the product of the phospholipase. Interestingly, PG can also be the substrate of PLA2s to generate lysophosphoglycerol (LPG), or LBPA/bismonoacylglycerolphosphate (LBPA/BMP), of which both are important lipid precursor/intermediates also present in apicomplexa in low amounts (Hullin-Matsuda et al., 2007; Saric et al., 2016; Nolan et al., 2017; Amiar et al., 2020). LBPA/BMP can also possibly be made from cardiolipin but the usual substrate in mammalian cells is usually PG (Hullin-Matsuda et al., 2007). We thus sought to quantify the levels of LPG, LBPA, and PG, and determine their potential change in content upon the disruption of PfPNPLA2 ( Figure 4B ). Our approach thus reveals that upon the loss of the enzyme, there was a significant accumulation of PG together with a very significant decrease of LBPA, whilst LPG levels were not significantly impacted (though slightly reduced). This significant decrease in LBPA was further confirmed by immunofluorescence (using an anti-LBPA antibody), which showed a significant decrease in fluorescence within the parasite when the protein expression was inhibited by GlcN treatment ( Figures 4C, D ). Taken together, the data strongly suggest that PfPNPLA2 is responsible for the degradation of PG to form LBPA, which further induces PL homeostasis disruption (see hypothesis model in Figure 5 ).