Plasticity and therapeutic potential of cAMP and cGMP-specific phosphodiesterases in Toxoplasma gondii

Graphical abstract


a b s t r a c t
Toxoplasma gondii is a common zoonotic protozoan pathogen adapted to intracellular parasitism in many host cells of diverse organisms. Our previous work has identified 18 cyclic nucleotide phosphodiesterase (PDE) proteins encoded by the parasite genome, of which 11 are expressed during the lytic cycle of its acutely-infectious tachyzoite stage in human cells. Here, we show that ten of these enzymes are promiscuous dual-specific phosphodiesterases, hydrolyzing cAMP and cGMP. TgPDE1 and TgPDE9, with a K m of 18 lM and 31 lM, respectively, are primed to hydrolyze cGMP, whereas TgPDE2 is highly specific to cAMP (K m , 14 lM). Immuno-electron microscopy revealed various subcellular distributions of TgPDE1, 2, and 9, including in the inner membrane complex, apical pole, plasma membrane, cytosol, dense granule, and rhoptry, indicating spatial control of signaling within tachyzoites. Notably, despite shared apical location and dual-catalysis, TgPDE8 and TgPDE9 are fully dispensable for the lytic cycle and show no functional redundancy. In contrast, TgPDE1 and TgPDE2 are individually required for optimal growth, and their collective loss is lethal to the parasite. In vitro phenotyping of these mutants revealed the roles of TgPDE1 and TgPDE2 in proliferation, gliding motility, invasion and egress of tachyzoites. Moreover, our enzyme inhibition assays in conjunction with chemogenetic phenotyping underpin TgPDE1 as a target of commonly-used PDE inhibitors, BIPPO and zaprinast. Finally, we identified a retinue of TgPDE1 and TgPDE2-interacting kinases and phosphatases, possibly regulating the enzymatic activity. In conclusion, our datasets on the catalytic function, physiological relevance, subcellular localization and drug inhibition of key phosphodiesterases highlight the previously-unanticipated plasticity and therapeutic potential of cyclic nucleotide signaling in T. gondii.

Introduction
Cyclic nucleotide signaling in apicomplexan parasites has been an active area of research in the last decade. Its phylogenetic divergence, modus operandi, and functional repurposing to enable the specialized lifecycle events in this class of clinically-relevant pathogens have attracted the most attention. Toxoplasma and Plasmodium are the two standard parasite models deployed to study apicomplexan biology, including cAMP and cGMP signaling. T. gondii -the only known species of Toxoplasma, is well known for its prominent ability to infect and reproduce in several warmblood organisms without geographic constraints [1]. The parasite undergoes asexual and sexual growth switching between multiple infec-tious stages, and exhibits exceptional promiscuity and metabolic plasticity, which underlie its widespread infection, inter-host transmission, reproduction, persistence, and pathogenesis. It is, therefore, imperative to examine the molecular mechanisms and determinants of infection and develop efficient anti-parasitic treatment strategies.
This work focuses on the tachyzoite stage of T. gondii responsible for the acute infection (tissue necrosis by recurrent lytic cycles). Tachyzoites can infect a broad range of nucleated host cells in humans and animals. The lytic cycle comprises several steps, such as gliding motility, invasion, proliferation, and egress [2]. Besides other known factors, protein kinase-dependent on cAMP (PKA) and protein kinase-dependent on cGMP (PKG) serve as the prime regulators of the lytic cycle events. Cyclic GMP signaling, for example, governs the calcium-dependent micronemal exocytosis needed for the motility-driven invasion and egress by the parasite. It is initiated by an exclusive guanylate cyclase fused to a P4-type ATPase (ATPase P -GC), and mediated by PKG [3][4][5][6][7]. On the other hand, cAMP signaling, facilitated by adenylate cyclase and PKA proteins, has been suggested as a negative regulator of PKG and associated Ca 2+ homeostasis/signaling [8,9]. In addition, cAMP is known to regulate the acute-chronic stage differentiation in T. gondii [10][11][12].
While the actuation of cAMP and cGMP signaling is relatively well studied, their counter-regulation remains poorly understood in T. gondii. Cyclic nucleotide phosphodiesterase (PDE) enzymes are crucial for the spatiotemporal repression of signaling cascades. T. gondii harbors a remarkably expanded repertoire of phosphodiesterases, and many of them are phylogenetically divergent from their human counterparts [13,14]. Our earlier work has revealed that of the 18 PDEs present in T. gondii, a panel of 11 proteins is expressed at different subcellular locations in tachyzoites [14], which was endorsed in a recent study [15]. We also demonstrated TgPDE8 and TgPDE9 as dual-specific (hydrolyzing cAMP and cGMP) enzymes and determined that the latter is not essential for the lytic cycle. TgPDE1 and TgPDE2, on the other hand, are presumed to be crucial for the lytic cycle based on conditional mutagenesis in tachyzoites [15]. Nonetheless, the substrate specificity, physiological relevance, functional redundancy and therapeutic potential of these and other PDEs remain severely underexplored in T. gondii, which inspired us to perform this study.

TgPDE2 is cAMP-specific, while other PDEs in tachyzoites can hydrolyze cAMP and cGMP
We performed the colorimetric enzyme assays using enriched preparation of the native PDEs. Transgenic tachyzoites encoding smHA-tagged proteins under the control of respective promoters were deployed to isolate the 11 PDEs expressed during the lytic cycle. As shown (Fig. 1A), these proteins were immunoprecipitated from the parasite extract, and PDE-enriched samples were tested by immunoblot and enzyme assays (Fig. 1B-E). Similar to the earlier work [14], we observed a protein band of the predicted size for most PDEs besides some proteolytic products (Fig. 1B-D). Except for TgPDE2, all other enzymes hydrolyzed both substrates, i.e., cAMP and cGMP (Fig. 1E). TgPDE2 degraded only cAMP with no evident activity for cGMP even at a much higher amount of the protein (20 lg) and substrate (200 lM). TgPDE2, with a catalytic rate of > 0.5 nmol/lg protein, was also among the most active enzymes. TgPDE1, TgPDE7 and TgPDE9 were similarly efficient in hydrolyzing cGMP and cAMP (Fig. 1E).
The second cohort of enzymes with relatively moderate activity (0.1-0.2 nmol/lg protein) included TgPDE6, TgPDE10, TgPDE13, and TgPDE18 (Fig. 1E). Except for TgPDE13, which was twice as functional with cAMP than cGMP, the other three PDEs displayed similar rates of catalysis with both substrates. The third and last set of phosphodiesterases comprising TgPDE5, TgPDE8, and TgPDE12 was much less active (<0.1 nmol/lg protein), correlating with their weak expression in tachyzoites (Fig. 1D). TgPDE5 appeared at least twice more active with cAMP than cGMP. Conversely, TgPDE12 hydrolyzed cGMP at a 5x higher rate than cAMP. Collectively, these assays revealed the catalytic specificity of PDEs expressed in tachyzoites and suggested a significant functional redundancy in the counter-regulation of cyclic nucleotide signaling.
2.2. Catalytic kinetics of TgPDE1, TgPDE2, TgPDE7 and TgPDE9 with cAMP and/or cGMP Given the high enzymatic activity, distinct subcellular locations and the yield of enriched proteins, we focused on the substrate kinetics of TgPDE1, TgPDE2, TgPDE7 and TgPDE9 (Fig. 2). At first, the enzyme and time dependence of each PDE were assessed under saturating amount of cAMP or cGMP (Fig. S1). Knowing the linearity of individual reactions, we tested the catalytic activity with 1-500 lM of substrates to calculate the K m and V max values. All four PDEs displayed the typical Michaelis-Menten kinetics showing a dependence of their catalysis on the substrate concentration ( Fig. 2A-E). TgPDE1, with a K m of 73 lM for cAMP and 18 lM for cGMP, had a 4-fold higher affinity for the latter ( Fig. 2A). TgPDE2, exhibiting the lowest K m (14 lM) among all, was functional only with cAMP (Fig. 2B). TgPDE7 with the K m values of 60 lM and 52 lM for cAMP and cGMP, respectively, displayed a similar affinity for both cyclic nucleotides (Fig. 2C). On the other hand, the hydrolytic activity of TgPDE9 for cAMP (K m , 118 lM) and cGMP (K m , 31 lM) was analogous to TgPDE1 (Fig. 2D). As elaborated below, the kinetic parameters of indicated PDEs enabled the interpretation of our mutagenesis and phenotyping datasets besides elucidating their pharmacological relevance in the context of known inhibitors.
A remarkably strong effect of BIPPO and zaprinast on TgPDE1 prompted us to perform the inhibition kinetics using different concentrations of inhibitors. The IC 50 of BIPPO for the cAMP and cGMP hydrolysis by TgPDE1 was 0.31 lM (Fig. 3E) and 0.51 lM (Fig. 3F), respectively. In contrast, zaprinast exhibited an approximately 5x (E) Phosphodiesterase activity of PDE-smHA proteins with cAMP and cGMP. The colorimetric enzyme assays were set up using 6 lg of PDE samples (TgPDE5, 10 lg) and 200 lM substrate (1 h, 37°C). The control reactions run alongside lacked the substrate or enzyme. The substrate-free enzyme-only reaction was subtracted from samples to quantify the PDE activity (normalized to the protein amount). The negative controls indicate the precipitated protein of the parental strain (N.D., not detectable). The data show the mean ± SE (n = 3-4 assays).
higher IC 50 value (1.49 lM) for cAMP (Fig. 3G). These data advocate that commonly-used inducers of parasite egress can inhibit cAMP as well as cGMP hydrolysis and indicate TgPDE1 as the primary target of these drugs. Our preceding work has shown that tachyzoite can survive the genetic deletion of TgPDE9 with no apparent defect in the lytic cycle [14]. Herein, we tested the physiological importance and functional redundancy of other designated PDEs by CRISPR/Cas9aided mutagenesis (Figs. S2-S4A). The gene-specific knockout constructs with 5 0 and 3 0 homology arms flanking a DHFR-TS selection cassette were transfected into respective progenitor parasite strains expressing smHA-tagged PDEs. Transgenic tachyzoites were selected by pyrimethamine, and mutants were isolated by limiting dilution. The genomic screening of clonal mutants by PCR revealed the occurrence of 5 0 and 3 0 -crossovers, confirming a successful replacement of the TgPDE7, TgPDE8, and TgPDE9 loci by the selection marker (Figs. S2-S4B). The loss of PDE expression in transgenic parasites was tested by immunofluorescence and immunoblot methods (Figs. S2-S4C). Unlike the matching progenitor strains, no HA signal was observed in the mutants. Plaque assays (Figs. S2-S4D), representing the periodic lytic cycles and thus the overall fitness of tachyzoites, disclosed a normal growth in the parental and progenitor strains, as expected. Surprisingly, however, none of the three mutants (DTgPDE7, DTgPDE8, or DTgPDE9) exhibited a growth defect, as deduced by the number and size of plaques.
We subsequently generated a double mutant to investigate the possible redundancy between the two apically-located dualspecific PDEs, i.e., TgPDE8 and TgPDE9 (Fig. 4). First, a conditional TgPDE9 mutant was made by 3 0 -genomic tagging with a mini auxin-inducible degron (mAID) using the HXGPRT selection marker (Fig. 4A). The mAID system enables maintaining viable parasites in the absence of indole-3-acetic acid (IAA, a type of auxin) if the gene is essential [4]. A plasmid expressing Cas9 nuclease and gene-specific sgRNA targeting the TgPDE9-3 0 UTR was transfected with a donor amplicon (5 0 and 3 0 -homology arms flanking mAID-3HA and HXGPRT selection cassette) into tachyzoites. Immunostaining confirmed the apical localization and protein integrity (Fig. 4B). The HA signal was not detectable within 1 h of auxin treatment, ratifying a fast and efficient conditional knockdown of TgPDE9 in tachyzoites (see immunoblot). In the second step, we deleted TgPDE8 in the TgPDE9-mAID-3HA strain by double homologous crossover and pyrimethamine selection (Fig. 4A). The eventual mutant was screened by genomic PCR (Fig. 4C). Compared to the parental parasites, neither the TgPDE9-mAID-3HA nor DTgPDE8/TgPDE9-mAID-3HA strain was affected in plaque assays irrespective of auxin supplementation ( Fig. 4D-F), suggesting no apparent functional overlap between the two enzymes during the lytic cycle of T. gondii.

TgPDE1 and TgPDE2 are partly redundant but individually needed for the parasite growth
To examine the relevance of TgPDE1 and TgPDE2, we made their conditional mutants by mAID-3HA tagging (Fig. 5A, S5A). The parasite strains were verified for localization and regulation by fluorescence imaging and western blotting. Both proteins were not detectable after incubation with IAA ( Fig. 5B, S5B). Depleting TgPDE1 and TgPDE2 compromised the growth of mutants by $56 % and $88 % compared to the respective controls ( Fig. S5-C-D). A moderate but significant decline in plaque numbers was also observed (Fig. S5E), confirming a requirement of each enzyme for the optimal reproduction of tachyzoites. Nonetheless, the residual growth of the conditional strains encouraged us to make a double mutant of TgPDE1 and TgPDE2, as described above for TgPDE8 and TgPDE9 (Fig. 4), and test their physiological cooperativity (Fig. 5A). Our attempts to ablate the TgPDE2 locus in the TgPDE1-mAID-3HA strain were futile; however, we could delete the TgPDE1 gene in the TgPDE2-mAID-3HA mutant (Fig. 5C). As anticipated, in plaque assays ( Fig. 5D-F), the parental strain was not affected. In contrast, the TgPDE2-mAID-3HA mutant showed strongly reduced growth ($80 %) upon IAA exposure. Equally, deletion of TgPDE1 in the TgPDE2-mAID-3HA strain impaired the DTgPDE1/TgPDE2-mAID-3HA mutant (ÀIAA sample, Fig. 5D-F). Importantly, a collective loss of both proteins aborted the parasite growth, as judged by the absence of plaques in the auxin-exposed double mutant (+IAA, Fig. 5D-F).
We next performed detailed phenotyping of the DTgPDE1/Tg-PDE2-mAID-3HA strain to evaluate the lytic cycle events, such as gliding motility, invasion, replication, and egress ( Fig. 6). In contrast to the parental and TgPDE2-depleted strains, the double mutant's motile fraction, trail length, and invasion efficiency were strongly reduced after auxin treatment ( Fig. 6A-C). Notably, the deletion of TgPDE1 (ÀIAA sample) exerted no evident defect across these features. In addition, knockdown of TgPDE2 alone only moderately attenuated the parasite replication; however, the double mutant treated with auxin showed an extreme decline in the number of proliferating parasites, as scored by the size of parasitophorous vacuoles (Fig. 6D). Intriguingly, the depletion of TgPDE2 in the DTgPDE1/TgPDE2-mAID-3HA strain exerted opposing effects on the parasite egress in early and late cultures (Fig. 6E), which was increased 36 h post-infection but declined after 60 h. No egress defect was observed in the TgPDE2-mAID-3HA strain at any tested time points. These phenotypic assays underpin the singular and collective significance of TgPDE1 and TgPDE2 enzymes and highlight their mutual interplay during the lytic cycle.
2.6. The DTgPDE1/TgPDE2-mAID-3HA strain is refractory to BIPPOinduced egress To further understand the role of signaling during egress, we deployed a chemogenetic approach utilizing a calcium ionophore (A23187) and inhibitors of ''cGMP-specific PDE" (BIPPO) as well as PKG (C2 or Compound 2), all of which are widely used to understand the biology of T. gondii. We tested their impact on egress of the parental, TgPDE2-mAID-3HA and DTgPDE1/TgPDE2-mAID-3HA strains cultured without or with auxin ( Fig. 6F-G). As envisaged, A23187, an ionophore activating calcium signaling downstream of cyclic nucleotides [21][22][23][24], triggered a complete egress of the three strains irrespective of the IAA treatment (Fig. 6F). Equally, BIPPO induced almost total lysis of all samples but the auxin-treated double mutant, which responded by only 20-25 % egress (Fig. 6G). In light of enzyme inhibition assays (Fig. 3A), these data entail TgPDE1 as a primary target of BIPPO. The resistance of auxin-treated double mutant to this drug appeared to be a com-bined outcome of TgPDE1 deletion (i.e., the absence of drug target) and rise in cAMP after knockdown of TgPDE2, leading to suboptimal activation of PKG and hyper-activation of PKA. We, therefore, examined the effect of PKG inhibitor, Compound 2 [25] on BIPPO-induced egress (Fig. 6G). The residual egress of the auxintreated double mutant was indeed completely blocked by Compound 2, suggesting that the process is mediated by PKG.

Ultrastructural imaging of intracellular tachyzoites
We performed the transmission electron microscopy of the DTgPDE1/TgPDE2-mAID-3HA strain to gain ultrastructural insight into the phenotype (Fig. 7). The double mutant cultured without auxin had a normal morphology with intact organelles, excluding a detrimental effect of TgPDE1 deletion (Fig. 7A). In contrast, a knockdown of TgPDE2 resulted in an aberrant/distorted shape of tachyzoites (Fig. 7B). Besides a much lower number of parasites per vacuole, we noted enlarged vacuolar space and impaired budding of progeny. The endodyogeny was arrested in auxin-treated cultures (Fig. 7C). We also observed a population of abnormalshaped tachyzoites with constricted terminal regions (see red arrow in Fig. 7D). In further assays, we utilized immunogold labeling of TgPDE1 and TgPDE2 to decipher their spatial distribution. Given the distinct apical presence of TgPDE9, we included it as a control for potential sample processing artifacts that may cause mislocalization of PDEs ( Fig. S6A-C). The quantification of images disclosed that a majority of parasites (>60 %) expressed TgPDE9-smHA in the conoid region at the apical pole near the plasmalemma (Fig. S6A). TgPDE1-smHA was detected mainly at the cytosolic periphery (38 %) and inner membrane complex (27 %), whereas TgPDE2-smHA was expressed in the cytosol (45 %), dense granules (24 %) and rhoptries (15 %).

TgPDE1 and TgPDE2 may be regulated by a specific kinase and phosphatase network
Our final assays explored the interaction network of TgPDE1 and TgPDE2 in T. gondii (Fig. S7). We precipitated PDEs and their protein binding partners using a-HA agarose beads and subjected them to liquid chromatography-mass spectrometric analysis. The parental strain was included as a negative control. The principal component analysis endorsed the proteomic dataset's technical and biological reproducibility, as illustrated by the grouping of samples in each cohort (Fig. S7A). The heatmap also displayed evident clusters of proteins binding to TgPDE1 and TgPDE2 but absent in the control sample (Fig. S7B). We detected 143 and 22 unique interactors of TgPDE1 and TgPDE2, respectively (p 0.01, at least twofold enriched compared to the parental control). In total, 38 proteins were bound to both phosphodiesterases (Fig. S7C). Importantly, no other PDE except the bait was present in immunoprecipitated samples (Table S1), validating the quality of enzyme assays involving TgPDE1 and TgPDE2. Based on our current understanding of PDEs in other model organisms and their contextual relevance to signaling in apicomplexan parasites, we shortlisted interacting proteins, including a group of kinases and serine/threonine phosphatases (Fig. S7D). We suspect that some of the proteins identified herein as potential interaction partners may be involved in regulating TgPDE1 and TgPDE2 catalysis (for a complete list, see Table S1).

Discussion
Toxoplasma gondii has evolved an expanded panel of highly divergent phosphodiesterases to counter-regulate the cyclic nucleotide signaling. All 18 PDEs encoded by its genome belong to the class I phosphodiesterases [13,14]. Here, we characterized the substrate specificity, physiological relevance, functional redundancy and spatial distribution of PDEs during the lytic cycle of T. gondii. This study reports the catalytic activity of 11 enzymes expressed in the tachyzoite stage. Except for TgPDE2 hydrolyzing only cAMP, others are promiscuous dual-specific proteins degrad-ing cAMP and cGMP. Therefore, unlike its mammalian host [26,27] and related parasite P. falciparum [28,29], T. gondii harbors a much larger set of dual-specific enzymes. We also found that TgPDE1 exhibits a 4x higher affinity for cGMP than cAMP. The mutagenesis, phenotyping and localization datasets reveal func- tional cooperation of TgPDE1 and TgPDE2 during the lytic cycle. Some of the findings above echo a recent study [15] overlapping with this and our earlier work [14].
The K m values of TgPDEs range from 14 to 118 lM, comparable to most class I phosphodiesterases. For example, of the 11 human PDE families, hPDE3 to hPDE11 display K m of 0.04-9 lM, whereas hPDE1 and hPDE2 are within 10 to 100 lM [30]. Our work also sheds light on the modus operandi of PDE inhibitors frequently used in parasitology research. We observed that BIPPO could potently inhibit cGMP hydrolysis by TgPDE1 and TgPDE7; however, it is the former enzyme that underlies the drug's effect during the lytic cycle. Moreover, another physiologically-critical enzyme, TgPDE2, is refractory to inhibition by BIPPO and zaprinast. Both drugs can inhibit cAMP catalysis by TgPDE9, although the dispensability of this protein excludes it as a drug target in tachyzoites. Potent inhibition of cAMP hydrolysis compared to cGMP by dual-specific TgPDE1 and TgPDE9 in BIPPO and zaprinast-treated samples can be explained by their differential affinity (K m ) for these substrates. Consistent with homology modeling of the Toxoplasma PDEs [13,14] and phenotypic studies performed in Plasmodium falciparum [29], our enzyme assays suggest that the two alleged cGMP-specific PDE inhibitors also perturb the cAMP pathway besides cGMP signaling. Thus, the enzyme kinetics and chemogenetic phenotyping presented herein offer a renewed prospect for developing novel PDE inhibitors and parasite-specific therapeutics.
This work uncovers remarkable plasticity in the counterregulation of cyclic nucleotide signaling, as exemplified by catalysis and mutagenesis of TgPDE1-2 and TgPDE7-9 proteins. Tachyzoites can survive the absence of TgPDE7, TgPDE8 and TgPDE9 enzymes, whereas they depend on the cooperation of TgPDE1 and TgPDE2 for specific events during their asexual growth. Notably, neither the knockout of TgPDE1 nor the knockdown of TgPDE2 affects the motility and invasion, but a loss of both compromises these features. Surprisingly, the DTgPDE1/TgPDE2-mAID-3HA mutant showed a higher (premature) egress at 36 h, which reversed into a moderate impairment at 60 h, implying programmed crosstalk of cAMP and cGMP signaling as the parasite nears the end of its lytic cycle. Regarding parasite replication, mutagenesis of TgPDE1 and TgPDE2 exerted a negligible effect, though the simultaneous loss of both phosphodiesterases delivered a potent phenotype. Unlike TgPDE1, we could not knockout TgPDE2, suggesting that any other orthologs cannot fully compensate for the physiological role of the latter enzyme. A dominant expression of TgPDE1 and TgPDE2 in the parasite cytosol may account for their partial functional redundancy. Varied subcellular localization of PDEs in tachyzoites also reflects a compartmentalized control of cyclic nucleotide signaling, warranting further studies. Last but not least, we demonstrate TgPDE1 and TgPDE2 as potential drug targets to control the acute infection of T. gondii.

Biological reagents
The RHDku80Dhxgprt [31] and RHDku80Dhxgprt-TIR1 Other antibodies against the HA epitope and TgSag1 were purchased from Takara-Bio (Japan) and Sigma-Aldrich (Germany). The secondary antibodies (Alexa488, Alexa594; IRDye 680RD, 800CW) and oligonucleotides (Table S2) were obtained from Thermo Fisher Scientific (Germany). The anti-HA mAb-conjugated agarose beads (clone HA-7) were procured from Sigma-Aldrich (Germany). The cell culture reagents were purchased from PAN Biotech (Germany), and other standard chemicals were supplied by Sigma-Aldrich and Carl Roth (Germany). The kits for isolation, cloning and purification of nucleic acids were acquired from Analytik Jena and Life Technologies (Germany). The PDE assay kits (colorimetric) were purchased from Abcam (UK) and Enzo Life Science (USA).

Host cell and parasite cultures
The human foreskin fibroblasts (HFFs; Cell Lines Service, Eppelheim, Germany) were grown to confluence and harvested for further passaging by trypsin-EDTA treatment. Cells were cultured in Dulbecco's modified Eagle medium containing glucose (4.5 g/L), 10 % heat-inactivated fetal bovine serum (FBS; PAN Biotech), 2 mM glutamine, 1 mM sodium pyruvate, 1x minimum Eagle's medium nonessential amino acids, penicillin (100 U/mL), and streptomycin (100 lg/mL) in a humidified incubator (37°C, 5 % CO 2 ). The tachyzoite stage of T. gondii was maintained by serial culture in confluent HFF monolayers using a multiplicity of infection (MoI) of 3. Parasites for all experiments were prepared by squirting infected cultures through a 27G syringe (2x) unless stated otherwise.

Lytic cycle assay
The impact of genetic manipulation on the lytic cycle of tachyzoites was determined by standard phenotyping methods, as described in our previous works [37,38]. For plaque assay, the confluent HFF monolayers in 6-well plates were infected with 200 parasites/well and incubated for 7 to 8 days without perturbation.
Samples were fixed with ice-cold methanol for 10 min and then stained with crystal violet solution for 15 min. Plaques were imaged and scored for size and number using the ImageJ software (NIH, Bethesda). To quantify the invasion efficiency, the HFF monolayers on coverslips placed in 24-well plates were infected with tachyzoites (MoI: 10) for 30 min at 37°C, followed by fixation with 4 % paraformaldehyde/PBS and neutralization with 0.1 % glycine/ PBS. Before permeabilization, samples were stained with the mouse a-TgSag1 antibody (1:1000) to visualize the non-invaded/ extracellular parasites. Cells were washed 3x with PBS, permeabilized with 0.2 % Triton X-100/PBS for 20 min, and stained with the rabbit a-TgGap45 antibody (1:10000) to score the invaded parasites. The fractions of invaded/intracellular parasites were determined to compare the invasion rates across the parasite strains.
To gauge the intracellular replication of tachyzoites, HFF cells grown on coverslips were infected (MoI:1, 40 h). Samples were subjected to permeabilization, neutralization, blocking, and staining with the rabbit a-TgGap45 antibody. The tachyzoite proliferation was assessed by enumerating parasitophorous vacuoles harboring a variable number of progeny. For the egress assay, the host cells were infected (MoI:1) for 36 h and 60 h, followed by immunostaining, as done for the invasion assay. The disrupted vacuoles with egressing parasites were quantified by a-TgSag1/Alex-a488 staining (green), and the fraction of intact vacuoles was scored based on a-TgGap45/Alexa594 labeling (red). To evaluate the gliding motility, parasites (4 Â 10 5 ) suspended in Hank's balanced salt solution were incubated to let them settle and glide (30 min, 37°C) on glass coverslips pre-coated with 0.01 % BSA (2 h). As mentioned elsewhere, samples were stained with a-TgSag1 and Alexa488 antibodies to visualize the gliding trails and parasites. The motile fraction was counted on the microscope, and trail lengths were quantified by the ImageJ program.

Immunoprecipitation of PDE proteins
The cell-free extract was prepared as described above. To precipitate the native PDEs, 50 lL of anti-HA agarose beads were added to the tachyzoite extract (2 mg protein). The volume was adjusted to 1 mL by a lysis buffer containing protease inhibitors. The pull-down reaction was set with constant rotation (4°C, 4 h). Afterward, protein-conjugated beads were pelleted (200 g, 30 s), washed 2x with ice-cold lysis buffer with protease inhibitors, and then once with the PDE dilution buffer (150 mM NaCl and 10 mM Tris-HCl, pH 7.4) [29]. Samples were given a final wash with 10 mM Tris-HCl buffer (pH 7.4) before using them for the enzyme assays.

PDE enzyme assay
The experiment was performed using colorimetric kits (Enzo Life Science, Netherlands; Abcam, UK) based on the enzymatic cleavage of 3 0 5 0 cAMP/3 0 5 0 cGMP to 5 0 AMP/5 0 GMP, which are further hydrolyzed by 5 0 -nucleotidase to their nucleoside and phosphate moieties. The phosphate group is quantified to determine the PDE activity. To set up the assay, immunoprecipitated proteins (1-10 lg) were suspended in the reaction buffer (50 lL), followed by the addition of cAMP or cGMP (200 lM). Samples were incubated at 37°C for 1 h and mixed with 100 lL of the green reagent (30 min, room temperature) to terminate the reaction. Subsequently, the OD 620 was measured to quantify the phosphate group. We also included a cAMP-specific PDE from the bovine brain as a positive control and several negative controls (no protein, no substrate) for validation purposes. The standards with varying phosphate amounts (0.25-4 nmol) provided by the kit were included in all experiments to quantify the enzymatic hydrolysis of cAMP and cGMP.

Proteolytic digestion for mass spectrometry
As described elsewhere [39,40], samples were processed by a single-pot solid-phase-enhanced preparation method. In brief, anti-HA agarose beads were incubated for 15 min at 60°C in an SDS-containing buffer (1 % w/v SDS, 50 mM HEPES, pH 8.0) to release proteins, which were afterward reduced and alkylated by dithiothreitol and iodoacetamide, respectively. They were supplemented with 2 lL of carboxylate-modified paramagnetic beads (Sera-Mag SpeedBeads, GE Healthcare, 0.5 lg solids/lL water), followed by adding acetonitrile to a final concentration of 70 % (v/v). Beads were allowed to settle for 20 min at room temperature. Subsequently, samples were washed twice with 70 % (v/v) ethanol in water and once with acetonitrile. Beads were suspended in 50 mM NH 4 HCO 3 supplemented with trypsin (Mass Spectrometry Grade, Promega) at an enzyme-to-protein ratio of 1:25 (w/w) and incubated overnight at 37°C. Acetonitrile was added to the samples to reach a final concentration of 95 % (v/v), followed by incubation at room temperature for 20 min. To maximize the yield, supernatants derived from this initial peptide-binding step were subjected to the peptide purification procedure [40]. Each sample was washed with acetonitrile. Paramagnetic beads from the original reaction and corresponding supernatants were pooled in 2 % (v/ v) dimethyl sulfoxide in water and sonicated for 1 min. After centrifugation (12500 rpm, 4°C), supernatants containing tryptic peptides were transferred into a glass vial for mass spectrometry analysis and acidified with 0.1 % (v/v) formic acid.

Liquid chromatography-mass spectrometry analysis
Tryptic peptides were separated using an Ultimate 3000 RSLCnano LC system (Thermo Fisher Scientific) equipped with a PEPMAP100, C18, 5 lm, 0.3 Â 5 mm trap (Thermo Fisher Scientific) and an HSS-T3 C18, 1.8 lm, 75 lm Â 250 mm analytical reversedphase column (Waters Corporation). Mobile phase A was water containing 0.1 % (v/v) formic acid and 3 % (v/v) DMSO. Peptides were separated using a gradient of 2-35 % mobile phase B (0.1 % v/v formic acid, 3 % v/v DMSO in acetonitrile) over 40 min at a flow rate of 300 nL/min. The total analysis time was 60 min including the wash and column re-equilibration (temperature, 55°C). Mass spectrometric analysis of eluting peptides was conducted on an Orbitrap Exploris 480 instrument platform (Thermo Fisher Scientific). The spray voltage was set to 1.8 kV, the funnel RF level to 40, and the capillary temperature was at 275°C. Data were acquired in data-dependent acquisition mode targeting the 10 most abundant peptides for fragmentation (Top10). Full MS resolution was set to 120,000 at m/z 200, and full MS automated gain control (AGC) target to 300 % with a maximum injection time of 50 ms. The mass range was adjusted to m/z 350-1500. For MS2 scans, the collection of isolated peptide precursors was limited by an ion target of 1x10 5 (AGC target value of 100 %) and maximum injection times of 25 ms. The fragment ion spectra were acquired at a resolution of 15,000 at m/z 200, and the intensity threshold was kept at 1E4. The isolation window width of the quadrupole was set to 1.6 m/z, and the normalized collision energy was fixed at 30 %. All data were acquired in profile mode using positive polarity.

Data analysis and label-free quantification
The raw data acquired with the Exploris 480 were processed by MaxQuant (v2.0.1) suite [41,42] using standard settings and labelfree quantification (LFQ) enabled for each parameter group, i.e., control and affinity-purified samples (LFQ min ratio count 2, stabilize large LFQ ratios disabled, match-between-runs). Data were searched against T. gondii proteome (UniprotKB/TrEMB, 8450 entries, UP000005641) and common contaminants. For peptide identification, trypsin was set as a protease, allowing for two missed cleavages. Carbamidomethylation was programmed as fixed, and methionine oxidation and acetylation of protein Ntermini were set as variable modifications. Only peptides with a length of 7 amino acids or more were considered. Peptide and protein false discovery rates (FDR) were 1 %. In addition, proteins were identified by the presence of at least two peptides. Statistical analysis was conducted using the student's t-test, which was corrected by the Benjamini-Hochberg method for multiple hypothesis testing (FDR, 0.01). Proteins with a minimum twofold enrichment in the affinity-enriched samples were considered.

High-pressure freezing and freeze substitution
HFFs cultured on sapphire disks (3 mm, coated with 0.01 % poly-L-lysine) were infected (MoI:4). For high-pressure freezing (HPF), sapphire disks were dipped into 1-hexadecene and placed onto a flat aluminum planchette (3 mm diameter) with cells facing upwards, which was then covered with another aluminum planchette (3 mm diameter, cavity 40 lm). The planchette sandwich was placed in an HPF holder and frozen using a Wohlwend HPF Compact 03 high-pressure freezer (Engineering Office M. Wohlwend GmbH, Switzerland). The frozen samples were stored in liquid nitrogen until freeze substitution (FS). For FS, the aluminum planchettes were opened in liquid nitrogen and separated from the sapphire disks, which were then immersed in a substitution solution containing 1 % osmium tetroxide, 0.1 % uranyl acetate and 3 % H 2 O in anhydrous acetone pre-cooled to À90°C. The FS was performed in a Leica EM AFS2 (Germany) following the protocol of 30 h (-90°C), 12 h (-60°C), 12 h (-30°C) and 1 h (0°C). Samples were washed 5x with anhydrous acetone, stepwise embedded in EPON 812 mixed with acetone (30 %, 60 %, 100 %) and finally polymerized for 48 h at 60°C. Ultrathin sections of 70 nm were prepared using a Leica UC7 ultramicrotome (Germany) and a 35°U ltra diamond knife (DiATOME, Switzerland). Sections were collected on formvar-coated grids and stained for 30 min with 2 % uranyl acetate and 20 min with 3 % lead citrate (Roth, Germany). Images were collected using the JEM 2100Plus system (200 kV, JEOL, Japan), equipped with a XAROSA CMOS 20MP camera (Emsis, Germany).

Data analysis, availability, and presentation
All assays were executed at least three independent times unless specified otherwise. The mass spectrometry data were processed using proprietary programs associated with each instrument. The datasets have been deposited to the ProteomeXchange Consortium (PXD032173) via the jPOST partner repository (JPST001521) (http://proteomecentral.proteomexchange.org, https://doi.org/10.1093/nar/gkw1080). Other results presented herein were analyzed and plotted using the GraphPad Prism v8 software. The error bars in graphs signify means with SE. The pvalues were computed by Student's t-test (*p 0.05; **p 0.01; ***p 0.001; ****p 0.0001). Images of transgenic strains and phenotyping assays (plaque, immunofluorescence, immunoblot, PCR etc.) show only a representative of the three or more biological replicates.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article. scholarship by the Vietnamese Government and SPP2225 EXIT. Support for the electron microscopy facility was granted by DFG CRC944, Z-project, DFG iBiOs PI405/14-1, DFG Priority Program SPP2225 EXIT (HE1964/24-1 and PS72/2-1). The funding agencies had no role in the design, execution, analyses, interpretation of the data or decision to publish this work.

Author contributions
NG conceived, designed and supervised the project; KCV standardized and performed the wet-lab assays; LR assisted KCV in cell culture; OEP supervised KCV and RF to conduct the electron microscopy; UD and ST carried out the mass spectrometry analysis; NG, PH and MH contributed reagents and resources; KCV and NG analyzed the data and drafted the original manuscript. All authors read, edited and approved the work.