Plasmodium falciparum Guanylyl Cyclase-Alpha and the Activity of Its Appended P4-ATPase Domain Are Essential for cGMP Synthesis and Blood-Stage Egress

The clinical manifestations of malaria arise due to successive rounds of replication of Plasmodium parasites within red blood cells. Once mature, daughter merozoites are released from infected erythrocytes to invade new cells in a tightly regulated process termed egress.

invasion and subsequent parasite development (26). In another rodent malaria parasite, P. berghei, PKG is required for sporozoite motility and invasion of hepatocytes following injection of sporozoites into a host by the mosquito (27). Furthermore, conditional disruption of PKG reduces the release of merosomes containing merozoites into the bloodstream prior to erythrocyte invasion (28).
In contrast to GCb, transcriptomic data indicate that P. falciparum GCa is expressed in both gametocytes and the asexual blood stages of the life cycle (https://plasmodb .org/plasmo/). The GCa gene has so far proved refractory to disruption in asexual blood stages (23,25,29), consistent with an important role during this clinically relevant stage. Confirmation of a key role for cGMP signaling in blood stages was obtained using a chemical genetic approach which demonstrated that PKG is essential for schizont rupture (30). This observation was subsequently extended to show that just prior to merozoite egress, PKG regulates the discharge of the subtilisin-like protease SUB1 from organelles termed exonemes into the parasitophorous vacuole (PV) of mature P. falciparum schizonts, where it proteolytically processes a number of proteins required for merozoite egress and invasion (31,32). PKG activity is also required for the release of calcium ions (Ca 21 ) from internal stores that is a prerequisite for egress (22). In efforts to dissect the mechanistic basis of this PKG-dependent egress pathway, comparative phosphoproteomic analysis identified 69 proteins that are phosphorylated in a cGMP-dependent manner (33). Very recently, the essential role for PKG in egress was confirmed through conditional genetic approaches (34). Despite this clear evidence for a crucial role for PKG in the asexual blood-stage life cycle, the role of GCa and its appended P4-ATPase domain remains unexplored.
Here, we show that GCa is essential for P. falciparum blood-stage egress and that GCa-null parasites cannot synthesize cGMP or mobilize Ca 21 . Crucially, we show that activity of the P4-ATPase domain of GCa is essential and that it functions upstream of cGMP synthesis.

RESULTS
P. falciparum GCa is expressed during late asexual blood-stage development and localizes to cytoplasmic vesicular structures in newly formed merozoites. Upon invasion of a red blood cell by a malaria merozoite, the parasite transforms through ring and trophozoite stages and then undergoes DNA replication to form a multinucleated schizont that eventually segments to form a new generation of daughter merozoites. These are then released upon egress to invade fresh red cells and repeat the cycle. Invasion is driven by an actinomyosin-based contractile complex often referred to as the glideosome, which lies beneath the pellicular membrane of the merozoite. Data from several previous transcriptome studies show that GCa mRNA expression peaks at the schizont stage, while proteomic analysis indicates presence of the protein in schizonts and merozoites (http://plasmodb.org/). To define the timing of expression of GCa at the protein level and to determine its subcellular localization, we generated a P. falciparum line expressing GCa fused to a C-terminal triple hemagglutinin (3ÂHA) epitope tag. For this, we used the P. falciparum 1G5 clone, which constitutively expresses a dimerizable Cre recombinase (DiCre) that can be activated by treatment with rapamycin (RAP) (35). The tagging strategy used single crossover homologous recombination at the 39 end of the endogenous GCa gene, along with introduction of a human dihydrofolate reductase (hDHFR) selection cassette flanked by two loxP sites (Fig. 1A). Initial selection for transformants with the antifolate WR99210, followed by several rounds of drug cycling, enriched for parasites in which integration had occurred. The hDHFR selectable marker was then recycled by RAP-induced Cre recombinase-mediated excision of the floxed sequence, leaving behind a single loxP site immediately downstream of the 3ÂHA epitope tag and translational stop codon (Fig. 1A). Two parasite clones were obtained that showed sensitivity to WR99210, suggesting the expected genomic rearrangement with loss of the hDHFR selection marker (Fig. 1B). The absence of the hDHFR cassette was confirmed in GCa:HA clone 1 by PCR  (Fig. 1C), and this transgenic parasite clone (referred to as GCa:HA) was subsequently used in all further experiments.
Tightly synchronized cultures of the GCa:HA line were sampled at 4-h intervals from the early trophozoite stage (24 h postinvasion) to the mature schizont stage (48 h postinvasion) and analyzed by Western blotting using a-HA antibodies. This revealed a single specific signal migrating at ;175 kDa that was most intense in mature schizonts (Fig. 1D). This molecular mass is consistent with a C-terminal GCa-3ÂHA fragment predicted to comprise both the guanylyl cyclase catalytic domains (C1 and C2) plus the 12 associated transmembrane helices (Fig. 1E). No signal was detectable at the ;500-kDa mass expected for the full-length protein.
To confirm the timing of GCa expression at the single cell level and to reveal its subcellular localization, we performed dual staining immunofluorescence microscopy (IFA) with anti-HA in combination with known markers for different subcellular compartments. In mature GCa:HA schizonts GCa localized to intracellular foci, but not the plasma membrane as established by costaining with a merozoite surface protein 1 (MSP1) antibody (Fig. 1F, top panel). To further characterize the nature of the intracellular compartment occupied by GCa, we costained with antibodies that react with apical membrane antigen 1 (AMA1), a micronemal marker, or plasmepsin V, an endoplasmic reticulum-resident protein. The a-HA staining showed no significant overlap with either of these markers, nor with a nuclear stain (Fig. 1F, middle and bottom panels). We conclude that GCa localizes to non-apical, cytoplasmic vesicular structures and is maximally expressed in mature schizonts.
GCa is essential for asexual blood-stage growth and merozoite egress. To investigate the function and essentiality of GCa, we used the DiCre recombinase system (35) to inducibly disrupt the GCa gene. For this, we created a GCa conditional knockout (cKO) line by introducing a second loxP site into the GCa locus of the GCa:HA line. This additional loxP site was incorporated within an artificial SERA2 intron (loxPint) (36) inserted into the ATPase domain of GCa using marker-free CRISPR/Cas9-mediated gene editing (see Fig. S1A in the supplemental material). Integration of the loxPint into the GCa coding region was confirmed by PCR in the uncloned parasite population (see Fig. S1B) and, following limiting-dilution cloning, two clones were obtained with the desired modification (see Fig. S1C). One of these clones, called GCa:HA:cKO, was used in all further experiments.
Activation of Cre recombinase in the GCa:HA:cKO line by the addition of RAP was expected to excise DNA sequences encoding part of the ATPase domain and the entire cyclase domain of GCa, abrogating both enzymatic activities ( Fig. 2A). To examine the efficiency of excision, synchronous cultures of young ring-stage parasites were treated with either 50 nM RAP or the equivalent concentration of vehicle only (dimethyl sulfoxide [DMSO]) for 2 h and allowed to mature to the schizont stage. PCR analysis showed highly efficient RAP-induced excision (Fig. 2B), while Western blot analysis showed FIG 1 Legend (Continued) the hDHFR cassette. Promoters/59 untranslated regions (UTRs) are indicated by arrows, and 39 UTRs/terminators are indicated by lollipops. Triangles represent loxP sites; arrows with numbers represent the relative position of oligonucleotide primers used for diagnostic PCR. rGCa refers to recodonized GCa gene sequences. (B) Growth curve for two cloned GCa:HA lines sensitive to treatment with 2.5 nM WR99210, consistent with successful excision of the hDHFR cassette. The data presented are from counting parasites on Giemsa-stained blood smears. At least 100 parasites were counted per condition. Clone 1 was used for all further experiments. (C) Diagnostic PCR analysis confirming successful integration of the 3ÂHA tag and efficient excision of the floxed hDHFR cassette in GCa:HA clone 1. Lane 1, parental 1G5DC; lane 2, GCa:HA nonexcised; lane 3, GCa:HA clone 1 (excised). (D) Western blot showing a time course of GCa:HA expression in P. falciparum blood stages. Parasites were harvested from cultures synchronized to a 2-h invasion window at the times indicated, with representative microscopy images shown above each sample (hpi, hours postinvasion). Scale bar, 5 mm. Blots were probed with a monoclonal a-HA antibody to visualize the GCa:HA fusion protein and an anti-PKG antibody as a staging control. GCa:HA migrated as an ;175-kDa fragment, while full-length protein (predicted ;499 kDa) could not be detected. Note that the additional band at ;125 kDa (*) arises from a cross-reactivity of the a-HA antibody with an unrelated parasite protein, since it is also detected in extracts from the unmodified 1G5 parental P. To assess the impact of GCa disruption on parasite viability and growth, we used flow cytometry to assess the replication of DMSO-treated (control) and RAP-treated GCa:HA:cKO parasites over three erythrocytic growth cycles. This revealed a complete growth arrest resulting from disruption of the GCa locus (Fig. 2E). Examination of Giemsa-stained GCa:HA:cKO parasites in the cycle of gene excision (cycle 0) revealed that RAP-treated cultures developed normally and were able to form mature segmented schizonts with a DNA content indistinguishable from control schizonts (Fig. 2F). However, upon further incubation we observed an accumulation of schizonts and a complete absence of newly formed ring-stage parasites, indicating that GCa is required for egress (Fig. 2G). To further characterize and quantify the egress defect in the GCa-null parasites, highly synchronized mature cycle 0 schizonts from DMSO-and RAP-treated cultures were monitored by live time-lapse microscopy. This confirmed that GCa-null parasites were unable to egress (Fig. 2H). Close examination of individual frames of the time-lapse video series revealed no signs of swelling or PVM rupture in GCa-null schizonts (Fig. 2I). Consistent with the above results, the abundant PV-resident protein SERA5, which is released during egress, was not detectable in culture supernatant samples from RAP-treated cultures (Fig. 2J). Collectively, these results establish that GCa plays an essential role in the asexual blood-stage life cycle and is required for merozoite egress.
GCa is responsible for cGMP production and essential for calcium release in asexual blood-stage schizonts. The phenotypic similarity between the egress block observed in GCa-null schizonts and that produced by the PKG inhibitor compound 2 ( Fig. 2I and Fig. 3A) is consistent with GCa-null schizonts failing to activate PKG due to a lack of cGMP synthesis. To evaluate the impact of GCa disruption on cyclic nucleotide production, we compared intracellular cGMP concentrations in mature cycle 0 GCa:HA: cKO schizonts from DMSO-or RAP-treated cultures. For these experiments, in order to prevent egress of the DMSO-treated (control) schizonts, both sets of parasites under comparison were maintained in the presence of compound 2. This analysis showed that ablation of GCa led to a 94.5% reduction in cGMP levels compared to control parasites (Fig. 3B). Treatment with the PDE inhibitor zaprinast, which blocks cGMP hydroly- Error bars indicate the standard deviations. Note that from ;16 days after RAP treatment parasites emerged, but these were evidently not GCa-null since PCR showed that the GCa locus was intact (see Fig. S3A and B). (F) Comparison of the DNA content in wild-type and GCa-null schizonts. Schizonts obtained from synchronous DMSOand RAP-treated GCa:HA:cKO cultures were arrested using compound 2 to prevent egress and fixed at ;46 h postinvasion, and parasite DNA was stained with SYBR green. SYBR green fluorescence intensity was measured by FACS counting from technical triplicates. (G) Representative microscopy images of Giemsa-stained parasites from DMSO-and RAP-treated GCa:HA:cKO cultures at ;50 h postinvasion, showing the accumulation of schizonts in the RAP-treated culture while new ring-stage parasites formed in the DMSO control culture. Scale bar, 5 mm. (H) Combined DIC and fluorescence images from time-lapse video microscopy of DMSO-and RAP-treated GCa:HA:cKO schizonts taken at 5 min (T = 5) and 30 min (T = 30) after release from a compound 2 block applied to transiently prevent and subsequently synchronize egress. In each experiment, one subset of parasites was pretreated with Hoechst to stain the nuclei so that DMSO-and RAPtreated parasites could be viewed simultaneously in the same imaging chamber. In the top panel, DMSO-treated parasites are Hoechststained, while in the bottom panel, RAP-treated parasites are Hoechst-stained. Schizonts visible in the first frame that rupture over the course of the video are circled in white. Scale bar, 10 mm. The graph on the right shows a quantification of the percentage of DMSO-and RAPtreated schizonts that egressed in each 30-min video. Data were collected from six videos, with Hoechst-treated samples depicted in blue. Statistical significance was measured by unpaired t test, where "***" signifies P , 0.0001. (I) Time series of individual stills from GCa:HA:cKO time-lapse video microscopy in panel G. Images following representative schizonts for each condition (DMSO and RAP) from 8 to 12 min after compound 2 washout are shown. Note that the PVM around the RAP-treated schizont remains intact throughout the time series, whereas the DMSO control has completed egress by 12 min. Scale bar, 5 mm. (J) Western blot analysis monitoring the release of SERA5 (50 kDa) into the culture supernatant of DMSO-and RAP-treated GCa:HA:cKO schizonts, as a measurement of egress over time. Sampling times are indicated in minutes. Pellet samples were included as loading controls for full-length SERA5 (120 kDa). sis in P. falciparum schizonts (31) and thus artificially elevates cGMP levels, resulted in a significant increase in cGMP levels in the DMSO-but not the RAP-treated samples, confirming that GCa-null schizonts are unable to synthesize cGMP (Fig. 3B). Importantly, we detected no significant differences in cAMP levels between the control and GCanull parasites, nor did cAMP levels increase upon zaprinast treatment (Fig. 3C), consistent with our previous findings (31).
Next, we investigated whether Ca 21 release from internal stores into the cytosol, a PKG-dependent process that is essential for merozoite egress (22), was impaired in the GCa knockout parasites. Treatment with the PDE inhibitor zaprinast specifically induces PKG-mediated calcium release (22). We measured zaprinast-induced Ca 21 release in DMSO-and RAP-treated GCa:HA:cKO parasites by loading cells with the fluorescent calcium indicator Fluo-4 AM. As shown in Fig. 3D, while zaprinast treatment stimulated elevated cytosolic Ca 21 levels in control schizonts no such elevation of Ca 21 levels was detected in the GCa-null schizonts (Fig. 3D). Importantly, both control and GCa-null schizonts showed similar response levels to the calcium ionophore A23187, which allows Ca 21 ions to cross cell membranes, indicating that internal calcium stores were not affected in the absence of GCa (Fig. 3D). Together, these results clearly identify GCa as a functional GC and establish that it generates the cGMP signal in asexual blood-stage schizonts essential for PKG activation, PKG-dependent Ca 21 release, and ultimately merozoite egress.
Chemical complementation of GCa-null parasites with PET-cGMP rescues the egress defect. Since the egress phenotype observed in GCa-deficient parasites is most likely due to their inability to produce cGMP to activate PKG, we reasoned that bypassing the need for cGMP synthesis by directly activating PKG could potentially rescue the egress defect. We also anticipated that this might reveal a second phenotype resulting from loss of the ATPase domain. To test this notion, we supplemented cultures containing highly mature RAP-treated cycle 0 GCa:HA:cKO schizonts with either cGMP or two cGMP analogues, 1-NH 2 -cGMP and PET-cGMP, both of which have previously been shown to activate recombinant apicomplexan PKG (37,38). While cGMP and 1-NH 2 -cGMP show similar levels of membrane permeability, PET-cGMP is 50-fold more lipophilic (https://www.biolog.de/technical_info_lipophilicity-data) due to its possession of a b-phenyl-1,N 2 -ethenoguanosine extension to the purine ring of cGMP. After 1 h of incubation in the presence of various concentrations of cGMP, 1-NH 2 -cGMP, or PET-cGMP, microscopic examination of the cultures revealed that PET-cGMP was highly effective at rescuing the egress block, resulting in the release of merozoites at all concentrations tested ( Fig. 4A; see also Fig. S4A). While rings with normal morphology could be observed in the cultures containing the lower concentrations of PET-cGMP (15, 31.2, and 62.5 mM), the cultures supplemented with higher PET-cGMP concentrations contained numerous intracellular or extracellular pyknotic forms, indicating toxicity (Fig. 4B). Few rings were observed in the cGMP and 1-NH 2 -cGMP-treated cultures, indicating limited activity of these compounds (Fig. 4A). It was concluded that PET-cGMP efficiently reversed the egress defect displayed by the GCa-deficient parasites (Fig. 4A).
To determine the optimum concentrations of PET-cGMP to sustain replication of the GCa-null parasites, parasite proliferation was assessed over a 72-h period (1.5 erythrocytic cycles) in the continuous presence of various concentrations of PET-cGMP. Interestingly, this revealed that none of the concentrations tested could rescue replication of the GCa-null parasites (Fig. 4C), including those concentrations that efficiently rescued egress in the short-term assays described above. To try to understand this, we performed parallel assays assessing growth of control, DMSO-treated GCa:HA:cKO parasites in cultures supplemented with the same range of PET-cGMP concentrations. This revealed that PET-cGMP is toxic, with a half-maximal effective concentration (EC 50 ) of 12.49 6 0.06 mM (Fig. 4C). Given its capacity to rescue the egress defect in GCa-null parasites, we considered that the toxic effect of prolonged exposure to the compound is likely explained by premature activation of PKG as the parasites matured, perhaps resulting in premature egress and release of noninvasive merozoites, a phenomenon previously observed upon treatment of schizonts with zaprinast (31). Furthermore, since bioinformatic analysis has revealed that PKG is the only predicted cGMP effector in Plasmodium parasites, it is unlikely that the toxic effects observed following PET-cGMP treatment are due to the activation of off-target effectors. In support of this model, further work showed that mature DMSO-and RAP-treated GCa:HA:cKO schizonts tolerated short-term incubation with 30 mM PET-cGMP, followed by washing off the compound ;10 h later, after the majority of schizonts had ruptured and formed rings; under these conditions, the new rings successfully matured to form cycle 1 schizonts (see Fig. S4B). Based on these findings, we used flow cytometry to quantify the degree of rescue that could be achieved over a single egress/invasion cycle. Quantification of cycle 1 parasite levels the following day showed that the presence of PET-cGMP during egress and invasion produced a 3.6-fold increase in parasitemia in the GCa-null cultures, while parallel cultures of GCa-null parasites lacking PET-cGMP completely failed to expand (Fig. 4D). In contrast, the replication rate of control parasites was slightly reduced (from 6.5-to 5-fold) by similar short-term treatment with PET-cGMP, perhaps indicative of the low levels of toxicity of the compound (Fig. 4D).
These results confirmed the capacity of PET-cGMP to rescue the egress defect in the GCa-null parasites. Of particular significance, since the conditional strategy used to disrupt the GCa gene was designed to excise key segments of both the cyclase and the ATPase-encoding sequences, the results indicated either that the ATPase domain has no essential function or that its function is to stimulate or activate cGMP synthesis, allowing its role to be bypassed by the presence of the cGMP analogue.
A conserved catalytic Asp residue within the ATPase domain is required for parasite survival. The N-terminal portion of GCa encodes a putative P-type ATPase that shares closest homology with type IV ATPases (P4-ATPases) (5) which in other organisms flip phospholipids from the outer to the inner leaflet of a lipid bilayer (12,13). P-type ATPases possess 10 transmembrane helices that facilitate transport of  (39). This requires the following cytoplasmic components: a nucleotide binding domain (N-domain), which binds ATP; a phosphorylation domain (P-domain), which contains a highly conserved aspartate (Asp) residue that becomes autophosphorylated to form an aspartyl phosphate intermediate; and an actuator domain (A-domain), which dephosphorylates the phosphorylation domain (40,41) (Fig. 5A). ATP-dependent autophosphorylation of the conserved Asp, which lies within a DKTGT motif, leads to a rotational change in the actuator domain surrounding the phosphorylation site; dephosphorylation is then coupled to inward transport of phospholipids or ions in the respective P-type ATPase families. Alignment of the GCa N-terminal region with other P4-ATPases, including human ATP8A1, for which a crystal structure has recently been determined (12), shows that of the 56 residues identical in P4-ATPases and P2-ATPases (a Ca 21 transporting SERCA and a Na 1 -K1P-type ATPase) that are required for ATP binding and catalysis, 47 are also identical in GCa and 2 are similar, fully consistent with the GCa sequence representing a functional ATPase domain (see Fig. S5).
Previous functional investigations (reviewed in reference 9) and two recent structural studies on yeast Drs2p and human ATPase8A1 in complex with their respective CDC50 partners (12,13) have demonstrated the importance of the transmembrane domains of P4-ATPases, particularly TM1 to TM4 (but also TM6), for lipid binding and translocation. Key among these is the PISL motif in TM4 which is highly conserved in P4-ATPases and is also present in the GCa sequence. This motif constitutes an Malaria Parasite GC-Alpha Is Essential for Egress ® important difference between the phospholipid and cation transporting ATPases and is crucial for binding and translocation of their respective substrates. P4-ATPases have a PISL motif whereas the ion transporters have a PEGL motif. GCa has a PISI motif at this position which contains the crucial IS pair diagnostic of phospholipid binding/ translocation rather than cation binding/translocation (see Fig. S5). The QQ motif at the C terminus of TM1 contains noncharged polar residues (e.g., Q and N) which in human ATP8A1 determine selectivity for PS in P4-ATPases. Q and N can hydrogen bond to the negative charge of the polar head group of PS, whereas the presence of small nonpolar residues (A and G) in place of the QQ motif preclude hydrogen bonding to PS and are features of P4-ATPases that translocate PC (12). GCa has a QN pair corresponding to the QQ motif of human ATP8A1, which is compatible with flipping of PS.
To address whether the ATPase domain of GCa is required for blood-stage replication, we investigated whether parasites could replicate following substitution of the conserved Asp (D756) in a manner designed to block formation of the aspartyl phosphate intermediate and thus ablate any enzymatic activity of the ATPase. To do this, we employed marker-free CRISPR/Cas9-mediated gene editing to introduce an asparagine (Asn) substitution of GCa D756 (Fig. 5B) (Fig. 5C; see also Fig. S1), since this substitution has been shown to completely ablate the activity of the yeast Drs2p P-Type ATPase and is expected to cause minimal structural changes to the ATPase (42). In parallel control manipulations, we used a similar approach to introduce synonymous mutations that did not alter the amino acid sequence but that could be differentiated from the wild-type locus at the nucleotide level. While parasites with a modified locus appeared in two of two independent transfections in these control cultures by 2 weeks posttransfection (Fig. 5D), no parasites emerged from three independent D756N transfections performed in parallel (one was performed in parallel with the successful integration of the loxPint) even following extended culture (6 weeks). Our failure to obtain viable parasites harboring the D756N mutation after three attempts, while readily obtaining transgenic parasites possessing synonymous mutations of the D756 codon, strongly suggests that an active ATPase domain is required for parasite survival. These data, combined with our ability to rescue the GCa-null growth phenotype by chemical complementation with PET-cGMP, indicate that the ATPase domain acts upstream of and is required for GC activity and that the ultimate role of the ATPase domain is to facilitate cGMP synthesis.

DISCUSSION
Although the single GC in the related apicomplexan parasite T. gondii (14-17) and GCb in P. falciparum (23), P. berghei (24,25), and P. yoelii (8) have been previously analyzed by reverse genetics, there have as yet been no functional studies of GCa in the asexual blood stages of Plasmodium. Here, we have made important advances in understanding the essentiality and function of this key component of the cGMP signaling pathway.
Epitope tagging of the endogenous P. falciparum GCa locus indicated that the protein is located in puncta within the confines of individual merozoites within mature schizonts. We propose that these puncta represent cytoplasmic membranous structures, consistent with the architecture of the protein which has 22 predicted transmembrane helices. It is likely that the GCa topology is orientated such that the twin GC catalytic domains face the cytosol in order to allow synthesis of cGMP to activate PKG, the majority of which is cytosolic (43). Our observed localization for P. falciparum GCa is reminiscent of that recently described for GCa in P. yoelii gametocytes (44). In contrast, in Toxoplasma (14-17) the single GC has been localized to the plasma membrane, primarily at the apical pole, in extracellular parasites, a localization similar to that of GCb in P. yoelii ookinetes. Therefore, unlike PyGCb and TgGC, there is no evidence of subcellular partitioning of cGMP generated by PfGCa in asexual blood stages.
Epitope-tagged P. falciparum GCa migrated in Western blots as an ;175-kDa species, much smaller than the predicted full-length protein (;500 kDa). A similar phenomenon has been observed in P. yoelii, where GCa migrates at ;175 kDa instead of the predicted full-length ;450 kDa (44), as well as in T. gondii, where TgGC predominantly migrates at ;125 and ;75 kDa, with only small amounts of the ;460-kDa species corresponding to full-length protein detected in some studies (14,15,17,18). Given the evidence suggesting that the Toxoplasma GC needs to be full length to be functional (15), it seems that the 175-kDa species observed in our study likely represents an N-terminally truncated proteolytic fragment, most likely resulting from protein instability during detergent extraction and that full-length P. falciparum GCa protein is present but below detection limits in our experiments.
Conditional deletion of a large segment of GCa spanning both the GC catalytic domain and the ATPase-like domain prevented asexual blood-stage replication, with a selective block in egress. This finding was in contrast to results generated in a previous global transposon-based gene knockout study which suggested that GCa was dispensable for blood-stage replication (45). However, that study acknowledged that false negatives can be obtained using their approach depending upon the position of transposon insertion. In an in vivo global gene knockout study in P. berghei, parasites in which GCa was disrupted grew slowly and were close to the cutoff for those essential for blood-stage replication (46), which is more consistent with our findings in P. falciparum.
We have previously reported that GCa possesses all the conserved amino acid residues required for catalytic activity (6). However, the ability of GCa to synthesize cGMP had previously not been demonstrated. The lack of detectable cGMP generation in GCa-null parasites confirms that it is a functional GC and also indicates that GCb, which is expressed only during mosquito stage development, cannot compensate for the absence of GCa in blood stages. The lack of cGMP synthesis fully explains the egress phenotype of GCa-null parasites, because activation of PKG by cGMP is known to be essential for egress (30,31). PKG activity also induces calcium release from internal stores in P. falciparum schizonts, which is required for egress. This is concordant with our finding that GCa-null schizonts did not mobilize calcium in response to cGMP elevation by the PDE inhibitor zaprinast. Together, these results establish that GCa is the key regulator upstream of PKG activity in P. falciparum blood stages. Disruption of the Toxoplasma GC also abolishes the rises in cytosolic Ca 21 required for secretion of micronemal proteins (17) and motility (47,48). Recent work in Toxoplasma has revealed a feedback loop between cGMP and cAMP signaling (18). However, unlike in Toxoplasma, where cAMP signaling is a negative regulator of egress, P. falciparum merozoite egress is not affected following disruption of either ACb or PKA (49). Inhibition of soluble guanylyl cyclase activity following phosphorylation by PKG has been reported in other systems (50), and we have previously identified that PfGCa is phosphorylated at two independent sites in a PKG-dependent manner (33), pointing toward a possible feedback loop whereby activated PKG may regulate GCa. Future work will be required to determine the significance of these phosphosites.
It has been suggested that the chemical gradient of phospholipids generated by P4-ATPases is akin to the chemical gradient of ions created by P2-ATPases to mediate signal transduction (51). To investigate whether the ATPase-like domain in PfGCa is likely to be a catalytically active ATPase, we attempted to generate parasites possessing a substitution of a highly conserved aspartate (D756) that is autophosphorylated in functional ATPases. These parasites could not be selected, while those engineered to reconstitute the wild-type aspartate proliferated normally, providing evidence that the ATPase domain is catalytically active and serves an essential function. Our finding that the GCa-null phenotype could be rescued by addition of the membrane-permeable cGMP analogue PET-cGMP, strongly suggests that the function of the ATPase domain is upstream of and directly related to cGMP synthesis. Our findings are consistent with previous studies in Toxoplasma where the ATPase domain was also shown to be critical for GC function (14, 15, 17). Complementation of GC-null Toxoplasma with a panel of mutants demonstrated that the ATPase domain is required for trafficking of GC, activity and maximal GC function since stimulated micronemal secretion was only partially reduced in the ATPase D728A mutant, while natural microneme secretion was prevented (15). Although our data suggest that the activity of the ATPase domain of GCa is an essential upstream regulatory factor for cGMP synthesis, our study does not further address its biochemical activity or indeed how it facilitates cGMP synthesis. The single Toxoplasma GC is involved in sensing phosphatidic acid, as well as changes in pH and potassium levels, to mediate egress (14,17). A recent suggestion that P. falciparum GCa might flip phosphatidylcholine (52) to mediate cGMP-stimulated egress was not supported by direct evidence of GCa-mediated flipping of phosphatidylcholine. Our analysis has shown the presence of a QN pair in P. falciparum at the C terminus of TM1 in GCa, which corresponds to the QQ motif of human ATP8A1 (12), suggesting that the sequence is compatible with flipping of PS. Future work will be needed to determine whether regulation of cGMP synthesis by GCa requires flipping or sensing of phospholipids by the ATPase domain and whether generation of lipid asymmetry across membranes contributes to activation of cGMP synthesis. This could be achieved by complementation of our conditional knockout line with various versions of GCa harboring the mutations in the ATPase domain.
Synthesis of cGMP by GCa has been linked to the stimulation of gametogenesis by xanthurenic acid (XA) (20), but it is not clear whether XA stimulates GCa activity directly or indirectly through other protein mediators. However, a study in P. yoelii has recently identified a protein called GEP1 that interacts with GCa, showing that both are required for XA-stimulated gametogenesis (44). Although cGMP synthesis and PKG activation are required for merozoite egress, the nature of the upstream signal that activates GCa in Plasmodium blood stages is unknown. Just prior to natural egress in Toxoplasma, the parasitophorous vacuole is acidified which triggers micronemal secretion (17). A similar mechanism may operate in Plasmodium. Future work will be needed to determine the events that occur upstream of cGMP signaling in blood stages and to understand the role of the P4-type ATPase domain in mediating the egress signal.
Rat monoclonal anti-HA tag antibody (clone 3F10) was purchased from Roche LifeScience (Penzberg, Germany) and rabbit anti-human PKG antibody from Enzo Life Sciences (Farmingdale, NY). A rabbit polyclonal antibody against MSP1-30 (53) as well as a rabbit anti-AMA1 antibody raised against the ectodomain (54) and rabbit anti-SERA5 (55) were all described previously. A mouse monoclonal antibody to plasmepsin V was kindly provided by Daniel Goldberg (Washington University School of Medicine in St. Louis, MO). P. falciparum culture and synchronization. P. falciparum asexual blood stages were cultured in human erythrocytes (National Blood Transfusion Service, London, United Kingdom) and complete medium (CM) consisting of RPMI 1640 medium (Life Technologies, CA) supplemented with 0.5% AlbuMAX type II (Gibco), 50 mM hypoxanthine, and 2 mM L-glutamine. Parasite cultures were incubated at 37°C and gassed with 90% N 2 , 5% CO 2 , and 5% O 2 according to standard procedures (56). Parasitemias were routinely monitored by examination of thin blood films fixed with 100% methanol and stained with 10% Giemsa stain in phosphate buffer (8 mM KH 2 PO 4 , 6 mM Na 2 HPO 4 [pH 7.0]).
Tightly synchronous parasites were obtained by purifying segmented schizonts on a 70% isotonic Percoll (GE Healthcare, Arlington Heights, IL) cushion and allowing them to invade fresh erythrocytes for 1 to 2 h while shaking. Unruptured schizonts were lysed by treating with 5% D-sorbitol (Sigma) for 10 min (57) to obtain highly pure and synchronous ring-stage cultures.
Induction of DiCre activity was achieved by treating early ring-stage parasites (2 to 10 h postinvasion) with 50 nM RAP for 2 to 3 h. Control parasites were treated with an equivalent volume of the vehicle DMSO (0.5% [vol/vol]).
Transfection of P. falciparum schizonts. Highly synchronous late-stage schizonts were used for transfection as previously described (35)  Plasmid construction. Primers used throughout this study were ordered from Integrated DNA Technologies (IDT, Coralville, IA) and are listed in Table S1 in the supplemental material. To generate the GCa:HA parasite line, a 1.9-kb fragment corresponding to the 39 end of the GCa coding region was PCR amplified with primers 14 and 15 and cloned into pHH1_PreDiCre_A_deltaH_deltaE (35) via EcoRV and XhoI restriction sites, upstream of the sequence encoding a triple hemagglutinin (3ÂHA) tag. The resulting plasmid pHH1_PreDiCre_GCa-3ÂHA was transfected into the 3D7/1G5DiCre line constitutively expressing dimerizable Cre recombinase (35). Transfected cultures were selected with WR99210 and then subjected to drug cycling to enrich for parasites having integrated the plasmid via single crossover recombination. Cultures were finally treated with RAP to activate Cre recombinase to recycle the hdhfr resistance marker. A clonal GCa:HA line sensitive to WR99210 was selected and further modified by CRISPR/Cas9-mediated gene editing to introduce a loxPint sequence into the ATPase domain of GCa to generate the GCa:HA:cKO line. A pUC19-based repair template was generated by first amplifying a 545bp 59 homology region and a 627-bp 39 homology region from genomic DNA using the primer pairs 16/ 17 and 18/13, respectively. The two PCR products were fused by overlap extension PCR using primers 16/13 and InFusion cloned into the HindIII and EcoRI sites of pUC19. A synthetic recodonized region of the GCa gene from bp 2109 to bp 2340 containing a SERA2-derived loxPint was ordered as a gBlock (IDT) and InFusion cloned into the AflII and BamHI sites located between the 59 and 39 homology regions. The repair template was linearized using PvuI and transfected, along with three pooled pDC2 plasmids, each encoding the Cas9 protein, the hDHFR selection cassette (which confers resistance to the antifolate WR99210), and a unique sgRNA sequence targeting either TTTAATATGTGTTCTATAGC, TCTATAGCAGGAAAAACATA, or CATATTCATCATAATCATTT. To mutate the ATPase domain of GCa, the repair template used to introduce the loxPint was modified to introduce either a D756N mutation, which would block formation of the aspartyl phosphate intermediate, or a wild-type D756 synonymous mutation, which would serve as a control. The repair templates were generated by replacing the loxPint flanked by the BglII and KpnI sites with overlap extension PCR products from primer sets 19/20 and 21/22 to introduce the D756N and D756 alleles, respectively. Each repair template plasmid was linearized with PvuI, combined with the pool of three pDC2 Cas9 plasmids mentioned above, and transfected into wild-type 3D7 parasites.
Limiting dilution to generate clonal parasite lines. Clonal parasite lines were obtained by limiting dilution combined with a plaque formation readout as previously described (58). The hematocrit and parasitemia of parasite cultures were determined by using a hemocytometer and by counting Giemsastained thin blood smears. Briefly, parasite cultures were diluted to give 0.3 parasites in 200 ml of culture at 1% hematocrit per well in a 96-well flat-bottom plate. After 9 days, plaque formation was assessed by using an EVOS FL cell imaging system (Thermo Fisher Scientific). Wells containing single plaques were subsequently expanded and analyzed by PCR.
Diagnostic PCRs. Integration of the 3ÂHA-tagging construct into the GCa locus was confirmed using primers 1/2. RAP-induced excision of the hDHFR cassette to create the GCa:HA line was validated using primers 1/4. Integration of the artificial loxPint into the ATPase domain of the GCa locus to generate the GC:HA:cKO line was confirmed using primers 5/7, while primers 5/6 were used to detect the presence of wild-type locus. Cre-mediated excision was validated using the primer pairs 10/11 and 12/11 to detect the unexcised and excised loci, respectively. Primers 8/9, which amplify a segment in an unmodified distal locus, served as a DNA quality control.
SYBR green growth inhibition assays. To determine the effect of various test compounds on parasite growth, their EC 50 s were determined by using the SYBR green growth inhibition assay adapted from a previous study (59). Test compounds were added as a series of 2-fold serial dilutions in triplicate to 96well flat-bottom plates. Wells containing no drug or 10 nM chloroquine were also included in each plate and served as negative and positive controls, respectively. Synchronous ring-stage parasites were added to achieve a starting parasitemia of 2% at 1% hematocrit and incubated at 37°C in a sealed gassed box for 72 h. Parasites were then lysed in buffer containing 20 mM Tris, 5 mM EDTA, 0.008% saponin, 0.08% Triton X-100, and 1Â SYBR green I (Molecular Probes, Eugene, OR) at pH 7.5 and incubated for 1 h at room temperature. SYBR green fluorescence was measured using a SpectraMax M3 plate reader (Molecular Devices, San Jose, CA) with excitation and emission wavelengths of 485 and 535 nm, respectively. EC 50 values were determined by nonlinear regression analysis.
Fluorescence-activated cell sorting analysis to measure parasite growth and DNA content. Parasite cultures were seeded in triplicate wells per condition and samples were fixed in 4% formaldehye, 0.1% glutaraldehyde in PBS containing 1Â SYBR green I (Molecular Probes) and stored at 4°C overnight. The fixative was aspirated, and the cells were washed in PBS and then analyzed using a BD LSR II flow cytometer (BD Biosciences, Franklin Lakes, NJ), with 50,000 events collected for each sample. FlowJo 7 analysis software (FlowJo LLC, Ashland, OR) was used to analyze the data.
Immunofluorescence microscopy. Air-dried thin blood smears were fixed in 4% formaldehyde in phosphate-buffered saline (PBS) for 20 min at room temperature, followed by permeabilization with 0.1% Triton X-100 in PBS. Blocking and antibody binding steps were performed in PBS containing 3% bovine serum albumin. Dual staining experiments were performed sequentially, starting with rat anti-HA, to eliminate cross-reactivity of the anti-rat secondary antibody with mouse or rabbit IgG. The secondary antibodies used were anti-rat IgG antibody conjugated to Alexa Fluor 488, anti-rabbit IgG conjugated to Alexa Fluor 594, and anti-mouse IgG conjugated to Alexa Fluor 594, all highly cross-adsorbed. Slides were mounted in ProLong Gold Antifade Mountant containing 49,69-diamidino-2-phenylindole (DAPI; Thermo Fisher Scientific). Images were acquired at Â100 magnification using a Nikon Eclipse Ti fluorescence microscope fitted with a Hamamatsu C11440 digital camera and overlaid in ICY bioimage analysis software (icy.bioimageanalysis.org).
Time-lapse video microscopy. Parasite egress was monitored by differential interference contrast (DIC) coupled with fluorescence microscopy using a Nikon Eclipse Ti fluorescence microscope with a 60Â oil immersion objective and fitted with a Hamamatsu C11440 digital camera. Segmented schizonts treated with C2 (1.5 mM) overnight were Percoll enriched and resuspended in warm complete medium at 0.4% hematocrit, transferred to prewarmed Poly-L-Lysine m-Slide VI 0.4 (IBIDI, Planegg, Germany) imaging chambers and imaged on a temperature-controlled microscope stage held at 37°C. To visualize DMSO-and RAP-treated parasites simultaneously, one culture was stained with 1 mg/ml Hoechst 33342 prior to washing off C2 and pooling the cultures, as previously described (60). Images were taken every 5 s for a total of 20 to 30 min, and the resulting videos were processed and analyzed in ICY bioimage analysis software (icy.bioimageanalysis.org).
Microscopy of Giemsa-stained blood films. Thin blood films fixed with 100% methanol and stained with 10% Giemsa stain in phosphate buffer (8 mM KH 2 PO 4 , 6 mM Na 2 HPO 4 [pH 7.0]) were imaged using an Olympus BX51 microscope fitted with an Olympus SC30 digital color camera through a 100Â oil immersion objective. Images were processed in Graphic (Picta, Inc.).
Parasite protein extraction, SDS-PAGE, and immunoblotting. Saponin-released parasites were lysed in 4 pellet volumes of CoIP buffer (150 mM NaCl, 0.5 mM EDTA, 1% NP-40, 10 mM Tris [pH 7.5]) supplemented with cOmplete EDTA-free protease inhibitor (Roche, Basel, Switzerland). Samples were incubated on ice for 10 min and centrifuged at 12,000 Â g for 10 min at 4°C, and the supernatant was collected. Reducing SDS sample buffer was added, and proteins were resolved on 4 to 15% Mini-Protean TGX stain-free precast gels (Bio-Rad, Hercules, CA) or 3 to 8% NuPAGE Tris-acetate protein gels (Thermo Fisher Scientific) for high-molecular-weight proteins. Proteins were transferred onto nitrocellulose membranes using a semidry Trans-Blot Turbo transfer system (Bio-Rad) and blocked using 10% skimmed milk in PBS containing 0.1% Tween 20 (PBST). Antibody incubations were carried out in 1% skimmed milk in PBST and washed in PBST. After incubation with secondary antibodies conjugated to near-infrared dyes, washed membranes were dried between Whatman 3MM blotting papers and imaged using an Azure c600 imaging system (Azure Biosystems, Dublin, CA) or a ChemiDoc imaging system (Bio-Rad).
Egress assays. Highly synchronous mature segmented schizonts from DMSO-and RAP-treated GCa: HA:cKO cultures that were treated with C2 (1.5 mM) at the early trophozoite stage were enriched on a Percoll gradient and washed several times in prewarmed RPMI. Parasites were resuspended in RPMI at 3.25 Â 10 8 parasites/ml, and 65-ml aliquots were dispensed into Eppendorf tubes. To harvest samples at each time point, parasites were pelleted at 9,000 Â g, and culture supernatants were purified using 0.22mm Costar Spin-X centrifuge filters (Corning, Corning, NY). The parasite pellets from the first time point were retained as a parasite loading control. Samples were subjected to Western blot analysis and probed for SERA5 as a measure of merozoite egress.
Calcium release assays. Mature segmented schizonts from RAP-and DMSO-treated cultures were Percoll enriched, and ;1.25 Â 10 8 cells from each condition were incubated in phenol red-free RPMI containing 10 mM Fluo-4-AM (Invitrogen, Carlsbad, CA) in the dark at 37°C for 45 min. The parasites were washed twice in prewarmed phenol red-free RPMI and then incubated for 20 min to allow for de-esterification of the AM ester. The parasites were washed twice and resuspended in phenol red-free RPMI at 1.25 Â 10 8 parasites/ml. Next, 100 ml of resuspended parasites was added again to wells on the bottom half of a 96-well plate. Three wells containing phenol red-free RPMI were also included as a control. Baseline Fluo-4 fluorescence in each well was read at 22-s intervals for 3 min using a SpectraMax M3 plate reader (Molecular Devices) prewarmed to 37°C with excitation and emission wavelengths of 483 and 525 nm, respectively. The plate was removed from the reader onto a heat block prewarmed to 37°C, and the parasites were resuspended and transferred to wells containing test compounds to give the desired final concentrations of zaprinast (75 mM), ionophore A23187 (20 mM), and DMSO (1.5%). The plate was placed back in the reader and read for a further 5 min at 22-s intervals. All samples were run in triplicate. The relative fluorescence units from reads at each time point and condition were averaged, and baseline and DMSO control values were subtracted.
Measurement of intracellular cyclic nucleotide levels. Intracellular cyclic nucleotide levels in mature segmented schizonts were measured using enzyme-linked immunosorbent assay (ELISA)-based high-sensitivity direct cAMP and cGMP colorimetric assay kits (Enzo Life Sciences). Around 1.25 Â 10 8 Percoll-purified schizonts were obtained from RAP-and DMSO-treated cultures to which C2 (1.5 mM) had been added to prevent schizont rupture. The purified schizonts were incubated for 3 min in RPMI containing C2 only or C2 in the presence of the PDE inhibitor zaprinast (75 mM). Parasites were then pelleted at 9,000 Â g, resuspended in 100 ml of 0.1 M HCl, and incubated for 10 min at room temperature with intermittent vortexing to complete cell lysis. The samples were pelleted at 9,000 Â g, and the supernatant was collected and stored at -80°C until required. Once all biological replicates were collected, each sample was diluted by adding 400 ml of 0.1 M HCl. Samples and standards were acetylated in order to improve sensitivity according to the manufacturer's instructions.
The detection ranges were 0.078 to 20 pmol/ml and 0.08 to 50 pmol/ml for the cAMP and cGMP assays, respectively. All samples and standards were set up in duplicate. Absorbance was measured at 405 nm using a SpectraMax M3 plate reader (Molecular Devices).
Treatment of parasite cultures with cGMP analogues to rescue GCa KO phenotype. Highly synchronous mature segmented schizonts from RAP-treated cultures at ;48 h postinvasion were treated with cGMP, 1-NH 2 -cGMP, or PET-cGMP at concentrations ranging from 15.6 to 250 mM. Giemsa-stained thin blood smears were taken after 2 h, and parasites were scored for their viability based on morphology. Parasites in 10 microscopic fields per condition were assigned to schizont, merozoite/ pyknotic intracellular, or ring-stage categories. Counts were performed blindly by two researchers. To measure the effect of 30 mM PET cGMP on wild-type and GCa KO replication rates, the cGMP analogue was added to synchronous DMSO-and RAP-treated GCa:HA:cKO cultures when segmented schizonts appeared and washed off 10 h later; the parasites were then harvested for fluorescence-activated cell sorting (FACS) analysis at the ensuing early schizont stage.
Sequence alignments. Sequence alignments were performed using Clustal Omega, modified manually, and guided by the alignment presented in a recent P4-ATPase structural study (12).
Data analysis and statistical significance tests. GraphPad Prism 7 was used for all statistical analyses. The numbers of biological and technical replicates for each experiment are noted in the figure legends.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.