CDC50 Orthologues in Plasmodium falciparum Have Distinct Roles in Merozoite Egress and Trophozoite Maturation

ABSTRACT In model organisms, type IV ATPases (P4-ATPases) require cell division control protein 50 (CDC50) chaperones for their phospholipid flipping activity. In the malaria parasite Plasmodium falciparum, guanylyl cyclase alpha (GCα) is an integral membrane protein that is essential for release (egress) of merozoites from their host erythrocytes. GCα is unusual in that it contains both a C-terminal cyclase domain and an N-terminal P4-ATPase domain of unknown function. We sought to investigate whether any of the three CDC50 orthologues (termed A, B, and C) encoded by P. falciparum are required for GCα function. Using gene tagging and conditional gene disruption, we demonstrate that CDC50B and CDC50C but not CDC50A are expressed in the clinically important asexual blood stages and that CDC50B is a binding partner of GCα whereas CDC50C is the binding partner of another putative P4-ATPase, phospholipid-transporting ATPase 2 (ATP2). Our findings indicate that CDC50B has no essential role for intraerythrocytic parasite maturation but modulates the rate of parasite egress by interacting with GCα for optimal cGMP synthesis. In contrast, CDC50C is essential for blood stage trophozoite maturation. Additionally, we find that the CDC50C-ATP2 complex may influence parasite endocytosis of host cell hemoglobin and consequently hemozoin formation.

pathology is caused exclusively by the asexual blood stage of the life cycle. Briefly, extracellular merozoites invade host erythrocytes (RBC) and transform rapidly into ring stages. Over the next ;24 h, these develop within a membrane-enclosed parasitophorous vacuole to form trophozoites which digest host cell hemoglobin and initiate DNA replication and endomitosis. The resulting schizonts undergo cytokinesis (segmentation) only upon maturation, forming daughter merozoites which are eventually released from the host cell ;48 h following invasion through a highly regulated egress process. The cycle is then reinitiated by the invasion of fresh erythrocytes by the newly released merozoites. A detailed molecular understanding of the biochemical pathways and proteins essential for blood stage development will inform discovery of novel targeted therapeutics that prevent malaria pathogenesis.
Merozoite egress and invasion are regulated by cyclic nucleotide signaling, conserved elements of which regulate multiple aspects of cell biology in model organisms and across the animal kingdom and can be effectively targeted pharmacologically (4). The second messengers, cyclic AMP (cAMP) and cyclic GMP (cGMP), are produced by cyclase enzymes and activate their respective cyclic nucleotide-dependent effector protein kinases PKA and PKG in a concentration-dependent manner (5). The activated kinases phosphorylate downstream targets that carry out effector functions, while the cyclic nucleotide signals are then broken down by phosphodiesterases (PDEs) (6,7). P. falciparum uses cyclic nucleotide signaling throughout its complex life cycle (7). Notably, cGMP signaling is required for egress of asexual blood stage merozoites (8,9) but also egress of gametes (10) and liver stage parasites (11,12) as well as for regulation of ookinete and sporozoite motility (12)(13)(14). In contrast, cAMP signaling has been shown to be required for sporozoite apical organelle secretion and invasion of hepatocytes (15), gametocyte deformability (16), and erythrocyte invasion (17).
P. falciparum has two guanylyl cyclases. While guanylyl cyclase beta (GCb) is dispensable in blood stages (9), guanylyl cyclase alpha (GCa) synthesizes cGMP in mature blood stage schizonts, where it plays an essential role in activating PKG to trigger egress (18). Both of the malaria parasite GCs are large integral membrane proteins with 22 predicted transmembrane domains (TMDs), the C-terminal segment of which constitutes the paired C1 and C2 guanylyl cyclase catalytic domains. Uniquely for cyclase enzymes, Plasmodium GCs (along with apicomplexan and ciliate orthologues) also contain an N-terminal type IV P-type ATPase (P4-ATPase)-like domain (18)(19)(20). In other organisms, P4-ATPases transport phospholipids from the outer to the inner leaflet of a lipid bilayer, maintaining the lipid asymmetry required for numerous functions, including membrane remodeling and vesicle formation (21). Recent studies in the apicomplexan parasite Toxoplasma indicate that this domain is critical to the role of its single guanylyl cyclase (T. gondii GC [TgGC]) in lytic growth, where it is essential for host cell attachment, invasion, and motility-dependent egress of tachyzoites (22)(23)(24)(25).
In model organisms, P4-ATPases require cell division control protein 50 (CDC50) chaperones for their phospholipid flipping activity (26,27). CDC50 proteins are integral membrane proteins with two TMDs that interact with a TMD of their partner P4-ATPases (28,29). The intervening loop between the CDC50 TMDs forms a beta-sheet-rich structure that contacts the luminal side of P4-ATPases (28,29). Studies in Saccharomyces cerevisiae have shown that CDC50 binding partners are required for the autophosphorylation of the catalytically active aspartic acid residue of the P4-ATPase, which is necessary for completion of the phospholipid flipping reaction cycle (30,31). P. falciparum encodes three putative CDC50 proteins, termed CDC50A (PF3D7_0719500), CDC50B (PF3D7_1133300), and CDC50C (PF3D7_1029400). Previous work in the mouse malaria model Plasmodium yoelii has shown that the CDC50A orthologue binds to GCb and is required for ookinete motility (20). Similarly, TgGC controls egress of tachyzoites (22,24,25) and binds to a Toxoplasma CDC50 partner that is required for its function (22). However, the functions of P. falciparum CDC50 orthologues have not been examined. Here, we show that both CDC50B and CDC50C, but not CDC50A, are expressed in the asexual blood stages and that CDC50B interacts with GCa whereas CDC50C is the binding partner of another putative P4-ATPase (ATPase2; PF3D7_1219600). We show that CDC50B modulates the efficiency of parasite egress by interacting with GCa for optimal cGMP synthesis. In contrast, CDC50C is essential for asexual blood stage trophozoite maturation due to a crucial role in endocytosis of host erythrocyte hemoglobin.

RESULTS
Generation of genetic tools to investigate the function of the P. falciparum CDC50s. To assess the biological functions of CDC50A, CDC50B, and CDC50C in P. falciparum blood stages, we generated three transgenic parasite lines designed to allow investigation of subcellular location and the effects of conditional disruption of each CDC50. The transgenics were generated in the genetic background of a 3D7 P. falciparum line that stably expresses dimerizable Cre (DiCre), the Cre recombinase activity of which is induced in the presence of rapamycin (RAP) (32,33). In each case, the target genes were "floxed" such that treatment with RAP would lead to excision of DNA sequences encoding a C-terminal region containing the second TMD of each protein ( Fig. 1A and B); this TMD has been shown in model organism CDC50-ATPase structures to interact with the most C-terminal helix of the ATPase binding partner (28,29). The constructs were designed so that following homologous recombination, the genes were also modified by fusion to sequences encoding a C-terminal triple hemagglutinin (HA) epitope tag.
Successful modification of the target genes was verified by PCR (Fig. 1C), and expression and RAP-induced truncation of tagged CDC50A-HA, CDC50B-HA, and CDC50C-HA in the respective transgenic parasites (termed CDC50A-HA:loxP, CDC50C-HA:loxP, and CDC50B-HA:loxP) was confirmed by Western blotting (Fig. 1D). Immunofluorescence analysis (IFA) of the transgenic lines (Fig. 1E) revealed a diffuse, partly peripheral signal in individual merozoites within mature segmented schizonts for both CDC50B-HA and CDC50C-HA. This was similar to the pattern observed upon costaining with the plasma membrane marker merozoite surface protein 1 (MSP1). While successful tagging and floxing of the CDC50A gene was also confirmed by PCR and Sanger sequencing, no protein expression could be detected in asexual blood stages. This suggested that CDC50A is not expressed in asexual stages, consistent with findings in P. yoelii, where CDC50A is expressed only in gametocyte and mosquito stages (20). Alternatively, since P. falciparum transcriptomic data indicate that the CDC50A gene is transcribed in schizont stages (34), the protein may be expressed but rapidly degraded as its GCb binding partner is not present in asexual blood stages (9,19).
Of the three isoforms, only CDC50C is essential for blood stage parasite growth. To investigate the essentiality of CDC50A, CDC50B, and CDC50C, highly synchronized ring stage cultures of each DiCre transgenic line were treated with RAP to induce excision of the sequence encoding the C-terminal TMD of each protein (Fig. 1C), and parasite replication was assessed using flow cytometry. RAP-treated CDC50A-HA:loxP and CDC50B-HA:loxP parasites displayed no significant growth inhibition over three cycles compared to matched control (dimethyl sulfoxide [DMSO]treated) parasites ( Fig. 2A). In contrast, CDC50C-HA:loxP parasites underwent complete growth arrest after cycle 1. Examination of the parasites by Giemsa staining showed that while new rings went on to form schizonts in DMSO-treated wild-type (WT) parasites, RAP-treated CDC50C-HA:loxP rings did not develop beyond the early trophozoite stage and displayed an accumulation of pycnotic parasites 30 h postinvasion with no expansion in parasitemia (Fig. 2B). It was concluded that CDC50C is essential for asexual blood stage survival.
CDC50B and CDC50C bind to distinct parasite flippase partners. In other organisms, CDC50 proteins interact with their cognate P4-ATPases and are required for their activity (28)(29)(30)(31). To determine whether CDC50B and CDC50C interact with P4-ATPases during P. falciparum blood stage development, we performed immunoprecipitation (IP) experiments from extracts of highly synchronized CDC50B-HA:loxP and CDC50C-HA:loxP schizonts. Western blot analysis confirmed the expected enrichment of the HA-tagged proteins from schizont lysates (Fig. 3A). The immunoprecipitated material FIG 1 (A) Representation of the three P. falciparum CDC50 proteins displayed from the N terminus to the C terminus. White boxes, transmembrane helices (TMDs) as predicted by TMHMM (59). Arrows indicate the point from which the protein products are truncated when the corresponding modified locus is excised in transgenically modified parasite lines. For CDC50A, this is from Phe341, for CDC50B from His235, and for CDC50C from Glu383. a.a, amino acids. (B) Schematic representation of the SLI strategy (58) used to produce the three CDC50 DiCre lines and resultant RAP-induced disruption of the modified genes. Double-headed arrows, regions amplified by PCR in panel C; red arrowheads, loxP sites; yellow lollipops, translational stop codons; white boxes, TMDs; light blue boxes, recodonized sequence (R.R.). (C) Diagnostic PCR analysis of gDNA from transgenic CDC50 parasite lines CDC50A-HA:loxP, CDC50B-HA:loxP, and CDC50C-HA:loxP verifying successful modification of target loci by SLI. Efficient excision of "floxed" sequences is observed in RAP treatment for all lines. Lane C represents amplification of a control locus (PKAc) to check gDNA integrity. PCRs 1 to 4 are represented in the schematic locus in panel B. PCR 1 screens for the WT locus, PCR 2 for 59 integration, PCR 3 for 39 integration, and PCR 4 for excision of the floxed sequence. See Table 1  Expression of GAPDH (PF3D7_1462800) is shown as a loading control. CDC50C-HA showed some residual expression after excision, but quantification shows an ;8-fold reduction in protein following RAP treatment (inset). No expression of CDC50A-HA was detected. Predicted molecular masses of (Continued on next page) was then analyzed by mass spectrometry in comparison with mock IP samples derived from 3D7DiCre parental parasites, to confirm this and to identify coprecipitating protein species. As shown in Fig. 3B, this confirmed high levels of enrichment of the HAtagged CDC50 bait proteins. In addition, in the case of the CDC50B experiments, we detected a .9 log 2 enrichment of peptides derived from GCa (Fig. 3B), while in the  CDC50C IPs, we detected .9 log 2 enrichment of peptides mapping to another putative P4-ATPase, a putative aminophospholipid flippase (PF3D7_1219600) (ATP2) (Fig. 3B). No other proteins were as significantly enriched in each pulldown. These results strongly suggest that CDC50B is a cofactor for GCa and that CDC50C is a cofactor for ATP2.
CDC50B is not required for GCa expression or trafficking but is crucial for optimal cGMP synthesis required for egress. GCa has a key role in egress as the source of cGMP required for PKG activation (18). Having determined that CDC50B interacts with GCa, we next investigated whether CDC50B also has a role in parasite egress.
To do this, we first compared the egress kinetics of mature DMSO-and RAP-treated CDC50B-HA:loxP schizonts by monitoring the appearance in culture supernatants over Volcano plot of P values versus the corresponding log 2 fold change in abundance compared to 3D7DiCre control samples (Fisher's exact test). A green line indicates a P of 22log 10 , and green dots represent peptides where the P is less than 22log 10 . Peptides for GCa and ATP2 were enriched to a P of less than 219log 10 . A list of enriched proteins is found in Table S1 in the supplemental material. time of proteolytically processed forms of the PV protein serine repeat antigen 5 (SERA5), which acts as a proxy for egress due to its release upon schizont rupture (35). As shown in Fig. 4A, these experiments revealed a marked reduction in the rate of egress over the sampling period in RAP-treated CDC50B-HA:loxP parasites compared to that of control DMSO-treated counterparts. Densitometric quantitation of data from three independent experiments indicated that CDC50B null schizonts undergo ;50% less egress than WT controls when sampled over a 2-h period (see Fig. S1 in the supplemental material). This was not due to a delay in schizont development, since microscopic examination of Giemsa-stained DMSO-and RAP-treated CDC50B-HA:loxP schizonts showed no detectable delay in parasite maturation, and analysis of DNA content by flow cytometry indicated no significant differences between formation of DMSO-and RAP-treated CDC50B-HA:loxP schizonts (Fig. 4A, lower panel). We further investigated the egress phenotype of CDC50B null parasites using a flow cytometry time course to assess the production of new rings by highly synchronized DMSO-and RAP-treated CDC50B-HA:loxP schizonts. We observed that while the number of new rings formed in the CDC50B null population between 45 and 53 h postinvasion was reduced, by 69 h postinvasion there was no statistically significant difference in ring parasitemia between the control and CDC50B null parasites (Fig. 4B). Microscopic examination of these newly generated ring stage parasites showed that those derived from the DMSO-treated parasites appeared more mature than those derived from the RAP-treated cultures (Fig. 4B, inset boxes), suggesting a delay to invasion in the latter. Together, the data indicate that CDC50B null parasites exhibit an extended erythrocytic life cycle, resulting in a delay to egress after schizont maturation but with no overall reduction in new ring stage formation in the following cycle.
In Toxoplasma gondii, the orthologue of CDC50B is required for correct subcellular trafficking of TgGC (22). To investigate whether this is also true in P. falciparum, we used a CRISPR-Cas9-based approach to fuse GCa to a C-terminal mCherry tag in the CDC50B-HA:loxP line, creating a parasite line called CDC50B-HA:loxP GCa-mCherry (Fig. S2). We failed to detect the tagged protein directly by fluorescence microscopy, possibly due to the previously reported very low abundance of GCa (18). However, Western blotting revealed an ;250-kDa signal in extracts of the CDC50B-HA:loxP GCa-mCherry schizonts, likely representing a proteolytic fragment of the tagged protein, since GCa is prone to proteolytic degradation in both P. falciparum and T. gondii (18,22) (Fig. 4C). Interestingly, Western blotting of extracts of RAP-treated CDC50B-HA:loxP GCa-mCherry schizonts indicated that there was no detectable reduction in the levels of GCa in the absence of CDC50B ( Fig. 4C and Fig. S3A). Exploiting the tagged GCa-mCherry line, we sought to confirm whether CDC50B was coprecipitated when GCa-mCherry was immunoprecipitated using red fluorescent protein (RFP)-trap beads which bind mCherry. Coprecipitation of CDC50B was observed, confirming CDC50B binding by GCa-mCherry (Fig. S4).
To examine the role of CDC50B in trafficking of GCa, we used an anti-mCherry antibody to localize GCa-mCherry by IFA in RAP-treated CDC50B-HA:loxP GCa-mCherry parasites. This revealed no obvious mislocalization of GCa in the absence of CDC50B, with a similar, diffuse signal detectable in both RAP-and DMSO-treated schizonts. In addition, in contrast to previous findings with T. gondii (22), no mislocalization of GCa in the endoplasmic reticulum (ER) or secretory pathway in the absence of CDC50B was detected, as judged by colocalization with the ER marker plasmepsin V (PMV) (Fig. 4D). Taken together, these results indicate that CDC50B binding is not important for the correct trafficking or stable expression of GCa. To seek more insight into the egress defect, we investigated whether ablation of CDC50B resulted in changes in cyclic nucleotide levels. To do this, we assayed extracts of DMSO-and RAP-treated CDC50B-HA: loxP schizonts by ELISA to quantitate cGMP and cAMP levels. This showed that CDC50B null parasites contained 53.67% (612.16%) less cGMP than DMSO-treated controls, while no significant difference in cAMP levels was observed (Fig. 4E). These reduced cGMP levels suggested that binding of CDC50B to GCa might be required for maximal GCa cyclase activity. To test this, we investigated whether the defect in egress of CDC50B null schizonts could be reversed by the addition of compounds that stimulate or mimic elevated cGMP levels. Egress was monitored in the presence and absence of the PDE inhibitor zaprinast or PET-cGMP, a membrane-permeable cGMP analogue known to activate parasite PKG (18,36). Treatment with either compound restored egress of CDC50B null schizonts to levels similar to those observed in control CDC50B-HA:loxP schizonts, confirming the requirement of CDC50B for optimal cGMP synthesis by GCa (Fig. 4F and Fig. S3B).
In view of the above results, as well as the previous observation that a lipid cofactor may stimulate egress in Toxoplasma and P. falciparum (22,37), we examined whether the potential phospholipid flippase activity of the P4-ATPase domain of GCa might be modulated by CDC50B binding. To do this, we investigated whether uptake of fluorescently labeled phosphatidylserine (PS), phosphatidylethanolamine (PE), and phosphatidylcholine (PC) were affected following disruption of CDC50B. DMSO-or RAP-treated CDC50B-HA:loxP late schizonts were incubated with fluorescent lipids and then analyzed by flow cytometry to determine their ability to accumulate lipids. We found no significant difference in the bulk uptake of measured lipids in schizonts in the presence or absence of CDC50B (Fig. S6).
CDC50C is not required for lipid uptake or protein export but plays an essential role in hemoglobin uptake from the host erythrocyte. As described above, RAP treatment of synchronous, newly invaded CDC50C-HA:loxP rings produced a CDC50C null parasite population that displayed a trophozoite arrest phenotype (Fig. 2B), suggesting an essential role for CDC50C in the trophozoite-to-schizont transition. Given this evidence that CDC50C plays a very different role from that of CDC50B, we decided to interrogate more precisely the functional role of CDC50C. Previous transcriptomic analysis has shown that CDC50C is transcribed throughout the asexual blood stage cycle, with relatively low levels of transcription in rings increasing to a peak in mature schizont stages (34). Consistent with this transcriptional profile, immunostaining of CDC50C-HA:loxP parasites detected expression in ring, trophozoite, and schizont stages (Fig. S7). Costaining of CDC50C in trophozoites with antibodies to ERD2 (a Golgi marker), PMV (an ER marker), or EXP2 (a PVM marker) showed that CDC50C displayed a diffuse cytosolic staining (Fig. 5A).
Phospholipid flippases have been shown to contribute directly to cellular lipid uptake (38)(39)(40). Initially, we speculated that the growth arrest of CDC50C null trophozoites may be due to a dysregulation of lipid uptake as a result of loss of function of the putative aminophospholipid flippase ATP2 partner. To test this notion, we labeled live RAP-and DMSOtreated CDC50C-HA:loxP trophozoites with the fluorescent aminophospholipid analogues nitrobenzoxadiazol (NBD)-PC, NBD-PE, and NBD-PS. Microscopy showed that lipid labeling was observed predominantly in parasites and not the erythrocyte membrane (Fig. 5B, right). No discernible difference in lipid uptake between CDC50C null and WT trophozoites was observed by flow cytometry (Fig. 5B), suggesting that CDC50C plays no essential role in the uptake of these phospholipids.
In model organisms, flippases also contribute to the production and maintenance of membrane asymmetry required for generation of trafficking vesicles, with specific flippases influencing exocytosis or endocytosis pathways (39,(41)(42)(43). Given that lipid uptake was unaffected in the absence of CDC50C, we considered it plausible that the ATP2-CDC50C complex may contribute to lipid homeostasis and trafficking in an analogous manner. P. falciparum trophozoites remodel their intracellular environment to create new permeation pathways that enable export of a wide variety of proteins into the host RBC via exocytosis, a process which is essential for trophozoite development (44). To examine whether trophozoite death in CDC50C null parasites could be attributed to changes in protein exocytosis, control or RAP-treated CDC50C-HA:loxP ring stage parasites were allowed to develop into trophozoites and were then analyzed by IFA to determine the localization of skeleton binding protein (SBP), a prominent exported protein in trophozoites. Puncta of SBP, characteristic of export, were evident within the RBC cytosol in both DMSO-and RAP-treated CDC50C-HA:loxP trophozoites (Fig. 6A), and quantification of these puncta indicated that there was no significant change in the levels of export of SBP in CDC50C null trophozoites (Fig. 6B). This suggested that loss of CDC50C has no impact on bulk protein export during trophozoite development.
As these results suggested that protein export and exocytosis were unaffected in CDC50C null trophozoites, we examined an essential endocytotic process. As intracellular asexual blood stage malaria parasites develop, they endocytose and digest host erythrocyte hemoglobin. A major by-product of this catabolic process is the sequestration of heme in the form of a characteristic crystalline product called hemozoin, which accumulates in the parasite digestive vacuole as large, refractile complexes that are easily visible by light microscopy. Microscopic examination of Giemsa-stained thin blood films indicated that while CDC50C null trophozoites displayed an apparently  normal morphology, the hemozoin crystals appeared smaller than those of DMSOtreated controls (Fig. 6C, right). To examine this in greater detail, we compared the ratio of the hemozoin crystal length to that of the parasite length in RAP-and DMSOtreated CDC50C null trophozoites. This confirmed a significantly decreased size of hemozoin in CDC50C null trophozoites, whereas parasite size was unaffected (Fig. 6C), suggesting a CDC50C-dependent defect in hemozoin formation. To further examine this, we purified and quantified hemozoin from parallel populations of RAP-and DMSO-treated CDC50C-HA:loxP trophozoites (45). As shown in Fig. 6D, this revealed ;30% less hemozoin in trophozoites lacking CDC50C. To investigate whether uptake of hemoglobin was affected in CDC50C null trophozoites, RAP-and DMSO-treated CDC50C-HA:loxP trophozoites were harvested at 36 h postinvasion and released from their host cells using saponin, and levels of intraparasite hemoglobin were quantified by Western blotting (Fig. 6E). The results indicated that the levels of hemoglobin within CDC50C null trophozoites were significantly reduced compared with those of control counterparts (Fig. S8). We further interrogated the role of CDC50C in the uptake of hemoglobin by using a food vacuole bloating assay by treatment of trophozoites with the cysteine protease inhibitor E64 (46). We found that while WT parasites displayed the characteristic bloating phenotype observed upon inhibition of hemoglobin digestion, RAP-treated parasites were largely unaffected (Fig. 6F), suggesting that these parasites are defective in hemoglobin transport to the food vacuole. Collectively, these data support a role for CDC50C in uptake and digestion of host erythrocyte hemoglobin that is essential for parasite development.

DISCUSSION
In this study, we have shown that CDC50B and CDCD50C proteins are expressed during asexual blood stage development and that they each bind to different putative P4-ATPase flippases-GCa and ATP2, respectively-which function at different developmental stages of the asexual blood stage cycle. CDC50B is dispensable for intraerythrocytic development, as is its orthologue in P. yoelii (20). Consistent with this, CDC50B null parasites display normal intracellular maturation and no significant replication defect over multiple erythrocytic cycles. However, we find that cGMP levels are reduced in CDC50B null parasites as well as rates of egress following schizont maturation. This effect is likely due to a delay in PKG activation, as this defect can be rescued by treatment of CDC50B null parasites with either the PDE inhibitor zaprinast or PET-cGMP, a membrane-permeable cGMP analogue known to be capable of activating apicomplexan PKGs (18). Collectively, our results strongly suggest that CDC50B acts to enhance cGMP synthesis by GCa. Recent studies in T. gondii have shown that the single GC (TgGC) also binds to a CDC50 designated CDC50.1 (22). We find that CDC50B is phylogenetically most closely related to CDC50.1 and CDC50.2 (see Fig. S9A in the supplemental material). In contrast to the present study, knockdown of CDC50.1 resulted in mislocalization of TgGC and a block in egress of T. gondii tachyzoites (22). The egress block could be rescued by adding a PDE inhibitor, implying that in the absence of CDC50.1, TgGC remains functional but produces cGMP with reduced efficiency (22). We observed no detectable mislocalization of P. falciparum GCa in the absence of CDC50B, with normal growth of parasites despite the reduced egress rates, indicating differences between the genera. We speculate that this may be due to differences in the threshold levels of cGMP required to activate PKG to trigger egress in each species.
Importantly, our work adds to the evidence supporting a role for CDC50s and the P4-ATPase domain of apicomplexan GCs acting functionally to stimulate maximal cGMP production required for egress. By analogy with other CDC50-flippase interactions, we speculate that this occurs through CDC50B binding to the P4-ATPase domain of GCa. Modulation of the activity of the C-terminal cyclase domain by the P4-ATPase may integrate a lipid-mediated trigger for egress, potentially by phosphatidic acid, as shown in T. gondii (22), or phosphatidylcholine (PC), as recently indicated in P. falciparum (37). However, we did not observe changes in the bulk uptake of fluorescent PC by schizonts lacking CDC50B. Recent structural examination of a human P4-ATPase: CDC50 complex has shown that CDC50 forms an intimate interaction with the TMDs of the P4-ATPase partner, with the loop domain between the two TMDs of the CDC50 forming an antiparallel beta-sheet structure that contacts the luminal side of the transmembrane loops of the P4-ATPase. Human CDC50 is glycosylated at several conserved asparagine residues, and the structure showed that interactions between CDC50 glycan moieties and P4-ATPase stabilize the functional complex (28). Intriguingly, an alignment of human CDC50a and the three P. falciparum CDC50s indicates that Asn180, at which glycosylation has been shown to interact structurally with its partner P4-ATPase (28), is absent from CDC50B but conserved in both CDC50A and CDC50C (Fig. S9B). N-glycosylation has been observed in P. falciparum (47), but this finding raises the possibility that CDC50B may be nonglycosylated. This observation may explain the finding in P. yoelii that GCb is degraded in the absence of its partner CDC50, suggesting that GCb is highly reliant on interactions with its CDC50 partner (CDC50A) for protein stabilization (20). In contrast, in our study, we observe that loss of CDC50B does not impact expression or stability of GCa, since GCa-mediated egress still occurs.
The revelation that CDC50C is essential for intraerythrocytic maturation of asexual blood stage P. falciparum trophozoites and that CDC50C binds to ATP2 suggests that CDC50C plays a role critical for ATP2 function. In contrast to our findings using native parasite-derived protein preparations, a recent in vitro study using recombinant protein indicated that ATP2 can bind CDC50B; however, CDC50C binding was not tested in that work, as the authors could not express it (48). Our study indicates that the essential function of CDC50C cannot be complemented by CDC50B.
Global transposon mutagenesis data suggest that the gene encoding ATP2 is essential in P. falciparum blood stages (49) and its orthologue is refractory to targeted deletion in Plasmodium berghei (50). While its cellular function is unknown, ATP2 has been implicated in resistance to two Medicines for Malaria Venture (MMV) "Malaria Box" compounds, mediated through a novel pathway involving gene copy number amplification. Functional characterization of the mechanism by which drug resistance is achieved remains lacking (51). Interestingly, Cowell et al. observed nonsynonymous mutations in genes encoding putative parasite Sec24 and Yip1 proteins (classically involved in vesicular trafficking) in drugresistant parasite lines containing ATP2 copy number variations (51). Here, we found that the ATP2-CDC50C complex influences endocytosis of hemoglobin during blood stage development possibly by influencing the phospholipid makeup of the cytostome, a structure that is crucial for hemoglobin uptake (52,53), and it remains possible that other endocytic pathways may also be affected by loss of ATP2 function, although these were not investigated. In yeast, different P4-ATPases contribute to distinct vesicular trafficking pathways (39,43). It is plausible that this could be similar in P. falciparum. We speculate that copy number modulation of ATP2 acquired during selection for drug resistance may modulate the endocytic pathway of the parasite so as to affect drug uptake, although further work is required to investigate this.
A recent study in P. yoelii has shown that the orthologue of CDC50C binds to a different P4-ATPase (ATP7) in ookinetes during parasite development within the mosquito (54). This indicates that CDC50C chaperones the activity of distinct P4-ATPases in different developmental stages of the parasite life cycle in both mammalian and insect hosts. Consistent with this, the transcriptomic profiles of ATP2 and ATP7 show that they are confined to asexual and insect stages, respectively. The same study demonstrated that the ATP7-CDC50C complex is required for PC uptake in ookinetes, and the authors suggested that this process may be required to allow mosquito midgut cell traversal, as CDC50C null or ATP7 null ookinetes could not achieve this. Intriguingly, alignment of ATP7 and ATP2 primary sequences alongside those of model P4-ATPases revealed that the "QQ motif" involved in defining substrate specificity is replaced by QL and QV, respectively (Fig. S9C). Given the similarity between these amino acid motifs, it is plausible that ATP2 also transports PC. Our finding that NBD-PC uptake is unaffected in CDC50C null trophozoites suggests that either ATP2 transports another phospholipid or that lipid uptake in trophozoites occurs via (multiple) redundant pathways. During the preparation of this article, a recent study found that the T. gondii orthologue of CDC50C, CDC50.4, binds ATP2B, an essential P4-ATPase that transports PS (55). This CDC50.4-ATP2B complex is required for efficient microneme secretion in tachyzoites with no defect observed during parasite intracellular development (55). It is plausible that the P. falciparum CDC50C-ATP2 complex may perform a similar role in egressed merozoites, but this could not be addressed in our study due to the block in intraerythrocytic development in CDC50C null parasites. Our work provides substantial new insights into the multifaceted, essential roles played by CDC50C proteins in malaria parasites and highlights potential species-specific differences between the roles of CDC50s in apicomplexan parasites.

MATERIALS AND METHODS
P. falciparum culture and synchronization. P. falciparum erythrocytic stages were cultured in human erythrocytes (National Blood Transfusion Service, United Kingdom) and RPMI 1640 medium (Life Technologies) supplemented with 0.5% Albumax type II (Gibco), 50 mM hypoxanthine, and 2 mM L-glutamine. Synchronous parasite cultures were obtained as described previously (56). Briefly, late segmented schizonts were enriched by centrifugation on a 60% Percoll (GE Healthcare) cushion, followed by the addition of fresh erythrocytes to allow invasion for 1 to 2 h under continuously shaking conditions. The remaining schizonts were then removed by sorbitol treatment to yield highly synchronous ring stage cultures. In all cases, induction of DiCre activity when required was by treatment for 2 to 4 h with 100 nM RAP (Sigma) as described previously (32,57). Control parasites were treated with vehicle only (1% [vol/vol] DMSO).
Oligonucleotide primers used in diagnostic PCR to detect integration and excision of transgenes are provided below in Table 1. CDC50B-HA:loxP GCa-mCherry was generated by transfection of CDC05B-HA:loxP. A linearized donor DNA which inserted mCherry in-frame with the C terminus of GCa, followed by a T2A peptide and BSD selection marker when integrated, and three pDC2-based (33) Cas9 guide RNA (gRNA) plasmids were cotransfected, each with different single guide RNA (sgRNA) targeting the C terminus of GCa. sgRNA sequences were as follows: sgRNA1, CTCTAAATTATTACAAAATA; sgRNA2, AGAAAAAACATTCAAGTATC; sgRNA3, ACGATGAAAAAAAGAAGAAG. Parasites were left to grow for 2 days posttransfection, followed by treatment with 5 mg/mL BSD to select for integrants. After the emergence of BSD-resistant parasites, gDNA was screened for correct integration. Donor sequences were constructed by amplifying a T2A BSD sequence from pDCIn (DiCre induction) (17) by PCR and cloning using a BsrGI site in frame with the C terminus of a donor DNA targeting GCa which had previously been constructed in the lab (18).
Parasite sample preparation and Western blotting. Parasite culture supernatant samples for egress and adhesin shedding assays were prepared from tightly synchronized cultures as previously described (17). Percoll-purified mature schizonts were resuspended in complete medium and allowed to further mature for 3 h until they were predominantly mature segmented schizonts. The experiment was then initiated by washing parasites with RPMI 1640 medium three times, followed by final resuspension at a 10% hematocrit in fresh warm RPMI 1640 medium. Culture supernatant aliquots (100 mL) were harvested at specified time points by centrifugation. The schizont pellet from time zero was retained as a pellet control sample.
Parasite extracts were prepared from Percoll-purified schizonts treated with 0.15% (wt/vol) saponin to remove erythrocyte material. To solubilize parasite proteins, phosphate-buffered saline (PBS)-washed saponin-treated parasite pellets were resuspended in 3 volumes of NP-40 extraction buffer (10 mM Tris, 150 mM NaCl, 0.5 mM EDTA, 1% NP-40, pH 7.5, with 1Â protease inhibitors [Roche]). Samples were gently vortexed and incubated on ice for 10 min, followed by centrifugation at 12,000 Â g for 10 min at 4°C. For Western blotting, SDS-solubilized proteins were electrophoresed on 4% to 15% Mini-Protean TGX Stain-Free protein gels (Bio-Rad) under reducing conditions and proteins were transferred to nitrocellulose membranes using a semidry Trans-Blot Turbo transfer system (Bio-Rad). Antibody reactions were carried out in 1% skimmed milk in PBS with 0.1% Tween 20 and washed in PBS with 0.1% Tween 20. Appropriate horseradish peroxidase-conjugated secondary antibodies were used, and antibodybound washed membranes were incubated with Clarity Western ECL substrate (Bio-Rad) and visualized using a ChemiDoc (Bio-Rad).
Immunofluorescence assays. Thin blood films were fixed with 4% formaldehyde in PBS and permeabilized with PBS containing 0.1% (vol/vol) Triton X-100. Blocking and antibody binding were performed in PBS-3% BSA (wt/vol) at room temperature. Slides were mounted with ProLong gold antifade mountant containing DAPI (49,6-diamidino-2-phenylindole) (Thermo Fisher Scientific). Images were acquired with a Nikon Eclipse Ti fluorescence microscope fitted with a Hamamatsu C11440 digital camera and overlaid in ICY bioimage analysis software or ImageJ. Superresolution images were acquired using a Zeiss LSM880 confocal microscope with an Airyscan detector in Airyscan SR mode. Antibodies used for Flow cytometry. For growth assays, synchronous ring stage parasites were adjusted to a 0.1% parasitemia-1% hematocrit suspension and dispensed in triplicate into six-well plates. Triplicate samples of 100 mL were harvested at days 0, 2, 4, and 6 for each well and fixed with 4% formaldehyde-0.2% glutaraldehyde in PBS. Fixed samples were stained with SYBR green and analyzed by flow cytometry.
For the measurement of egress and ring formation of highly synchronized DMSO-and RAP-treated CDC50B-HA:loxP parasites, a culture of CDC50B-HA:loxP with a 1-h invasion window was seeded in duplicate at 1% hematocrit and 2% parasitemia and treated with DMSO or RAP at 4 h postinvasion for 2 h. Triplicate samples of 100 mL were taken and fixed with 4% formaldehyde-0.2% glutaraldehyde in PBS at hourly intervals from 45 to 53 h postinvasion and at 69 h postinvasion the subsequent day. Fixed samples were stained with SYBR green and analyzed by flow cytometry. Schizont parasitemia was determine by gating high-signal SYBR-positive cells. Ring parasitemia was determined similarly but by gating lowsignal SYBR-positive cells.
Fluorescent lipid labeling. NBD-PC, NBD-PE, and NBD-PS (Avanti polar lipids) were dried and resuspended in RPMI 1640 medium to 1 mM stock solutions and stored at -20°C. Relevant parasite stages (trophozoites or late schizonts) from highly synchronous cultures were pelleted and washed twice with RPMI 1640 medium. Parasites were then resuspended in RPMI 1640 medium containing Hoechst stain with 1 mM NBD lipid or no lipid (negative control). Suspensions were incubated at 37°C for 30 min and subsequently pelleted by centrifugation. Pellets were then washed three times with prewarmed RPMI 1640 medium containing 5% bovine serum albumin (BSA), followed by resuspension in PBS. Suspensions were then diluted 1:10 and analyzed by flow cytometry on an Attune NxT. Samples were gated for Hoechst DNA positivity, and the resultant population was gated for NBD lipid fluorescence. This gate was set to exclude signal generated from staining uninfected erythrocytes. For trophozoite samples, a low Hoechst signal population was gated, and for schizont samples, a high Hoechst signal population was gated.
Immunoprecipitation. Tightly synchronized schizonts (;45 h old) of CDC50B-HA:loxP, CDC50B-HA: loxP GCa-mCherry, CDC50C-HA:loxP, and 3D7DiCre parental parasites were enriched on a 70% Percoll cushion. The schizonts were treated for 3 h with 1 mM C2 (to arrest egress), after which the cultures were treated with 0.15% saponin in PBS containing cOmplete Mini EDTA-free protease and PhosSTOP phosphatase inhibitor cocktails (both from Roche) for 10 min at 4°C to lyse the host erythrocytes. Samples were washed twice in PBS containing protease and phosphatase inhibitors and snap-frozen, and the pellets were stored at 280°C. Parasite pellets (70-to 100-mL packed volume) were resuspended in 3 volumes of NP-40 extraction buffer (10 mM Tris, 150 mM NaCl, 0.5 mM EDTA, 1% NP-40, pH 7.5, with 1Â protease inhibitors [Roche]). Samples were gently vortexed and incubated on ice for 10 min, followed by centrifugation at 12,000 Â g for 10 min at 4°C. Clarified lysates were then added to anti-HA antibody-conjugated magnetic beads (Thermo Scientific) or RFP-trap beads (ChromoTek) which had been equilibrated in NP-40 extraction buffer. Samples were incubated at room temperature for 2 h on a rotating wheel, after which beads were precipitated using a magnetic sample rack. The supernatant was removed, and beads were washed three times with NP-40 extraction buffer, followed by three washes with extraction buffer lacking detergent. Washed beads were then resuspended in trypsinization buffer [50 mM ammonium bicarbonate, 40 mM 2-chloroacetamide, and 10 mM Tris-(2-carboxyethyl) phosphine hydrochloride], and samples were reduced and alkylated by heating to 70°C for 5 min. Two hundred fifty nanograms of trypsin was added to the samples and heated at 37°C overnight with gentle agitation, followed by filtration using a 0.22-mm Costar Spin-X centrifuge tube filter (Sigma). Samples were then run on a LTQ Orbitrap Velos mass spectrometer (Thermo Scientific). The search engines Mascot (http://www .matrixscience.com/) and MaxQuant (https://www.maxquant.org/) were used for mass spectrometry data analysis. The PlasmoDB database was used for protein annotation. Peptides and proteins having a minimum threshold of 95% were used for further proteomic analyses, and peptide traces were analyzed using Scaffold5. Enrichment was determined by comparing results from tagged lines with that of immunoprecipitated material from 3D7DiCre parental parasites.
Measurement of hemozoin content. A culture of 5% parasitemia 1-h synchronized ring stage CDC50C parasites were treated at 1 h postinvasion with DMSO or RAP (100 nM) and then left to develop until the early trophozoite stage at 36 h postinvasion. Parasites were then harvested by saponin lysis and processed similarly to a reported method (45) to purify the hemozoin. Pellets were then depolymerized in 0.5 mL of 0.2 M NaOH solution, and the resultant heme content was measured by absorbance at 410 nm in a Spectramax iD5 plate reader.
Measurement of intracellular cyclic nucleotide levels. cAMP and cGMP in mature CDC50B schizonts were measured using enzyme-linked immunosorbent assay (ELISA)-based high-sensitivity direct cAMP and cGMP colorimetric assay kits (Enzo Life Sciences). Mature schizonts were Percoll purified from RAP-or DMSO-treated CDC50B-HA:loxP cultures, followed by resuspension and lysis in 0.1 M HCl solution. Samples were pelleted at 10,000 Â g, and the supernatant was collected and stored at 280°C until required. To perform the ELISA, samples and standards were acetylated to improve the detection sensitivity according to the manufacturer's instructions. Standards and samples were run in triplicate on the same plate, and absorbance at 410 nm was read with a SpectraMax iD5 plate reader. The standard was fitted to a sigmoidal curve and used to determine cyclic nucleotide concentrations in parasite samples. The remaining supernatant was assayed for protein concentration by a Bradford assay kit (Pierce). cGMP and cAMP reading were normalized by protein content from the Bradford assay.
Food vacuole bloating assay. We conducted experiments similarly to those previously reported (46). CDC50C-HA:loxP parasites with a 1-h invasion window were treated with DMSO or RAP at 1 h postinvasion for 2 h. At 24 h, 33 mM E64 was added and the parasites were left to develop for a further 8 h, after which they were stained with 4.5 mg/mL dihydroethidium for 20 min at room temperature to detect the food vacuole. Parasites were then washed twice with PBS and imaged on an EVOS fluorescence microscope. A minimum of 20 cells were counted per condition and scored for bloated or nonbloated food vacuoles.
Data availability. Data are publicly available via the PRIDE database under data set identifier PXD033834.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.