We can emulate the endogenous signaling pathways that mediate the kinetic phase of the lytic cycle by treating isolated parasites with zaprinast, which stimulates PKG activation by inhibiting phosphodiesterases that degrade cGMP (Brown et al., 2016; Lourido et al., 2012). Zaprinast-stimulated motility occurs rapidly and in defined sequence in apicomplexans: initially, an increase in cGMP activates parasite protein kinase G (PKG), which phosphorylates substrates and stimulates Ca²⁺ release from internal stores (Brown et al., 2020; Lourido and Moreno, 2015). PKG likely performs functions that extend beyond regulating Ca²⁺ stores, for example by mobilizing diacylglycerol and phosphatidic acid (Lourido et al., 2012; Brown et al., 2017; Bullen et al., 2016; Bisio et al., 2019); however, the use or phosphodiesterase inhibitors like zaprinast allows us to stimulate endogenous Ca²⁺ release without flooding the cell with Ca²⁺, as is the case with ionophores. Ca²⁺-dependent protein kinases, such as TgCDPK1 and TgCDPK3, synergize with PKG to effectuate microneme secretion and parasite motility (Brown et al., 2016; Lourido et al., 2012; Lourido et al., 2010; McCoy et al., 2012). This signaling cascade is active within seconds of cGMP elevation; however, existing T. gondii phosphoproteomes compare changes at a single time point following Ca²⁺-ionophore stimulation (Treeck et al., 2014), providing only a snapshot of diverging signaling states. Here, we add kinetic resolution to these signaling pathways. We quantified dynamic changes in the phosphoproteome within a minute of stimulation with zaprinast and thus activation of the cGMP/Ca²⁺ pathway.

We collected five timepoints in the 60 s following stimulation (0, 5, 10, 30, and 60 s), as well as three DMSO-treated matched timepoints (0, 10, and 30 s), in biological duplicate (Figure 1A). Using TMTpro labeling methods (Li et al., 2020), we multiplexed 16 samples, allowing us to analyze a complete time course, with replicates and controls, in a single MS experiment. Phosphopeptides were enriched from the rest of the sample using sequential metal-oxide affinity chromatography, which maximizes phosphopeptide capture (Tsai et al., 2014). Our experiments quantified 4,055 parasite proteins, none of which exhibited more than a twofold change in abundance in the 60 s following stimulation (Figure 1—figure supplement 1).

Given the paucity of known phosphoreglatory interactions in apicomplexans compared to other organisms (Weiss et al., 2020), we employed several analysis approaches to maximize the identification of changing phosphosites. We first calculated phosphoregulation scores by summing peptide abundances of vehicle (DMSO) and zaprinast-treated samples, taking their ratios, and standardizing the values with a modified Z score (Figure 1B). From a phosphoproteome of 11,755 unique peptides with quantification values (belonging to 2,690 phosphoproteins), 839 phosphopeptides increased in abundance three modified Z scores above the median, whereas 154 decreased 1.8 Z scores below the median. Principal component analysis on the significant peptides distinguished the agonist treatment and time-course kinetics in the two principal components accounting for the greatest variability in the data (Figure 1C and Figure 1—figure supplement 1).

We leveraged the kinetic resolution of our comprehensive phosphoproteomics datasets to identify subregulatory networks. A Gaussian mixture-model clustering algorithm (Invergo et al., 2017) heuristically resolved four clusters for phosphopeptides arising from zaprinast treatment: three clusters increasing with different kinetics, and one decreasing (Figure 1D). On average, the 173 phosphopeptides belonging to cluster 1 increased sharply in abundance within 5 s of treatment and continued to increase for the remainder of the time course (Figure 1D), suggesting that they belonged to the first wave of phosphoregulation. This cluster was enriched for phosphoproteins associated with phosphodiesterase activity, phospholipid binding, and Ca²⁺ binding (Figure 1E), including PDE1, PDE2, PI4,5K, PI3,4K, PI-PLC, a phosphatidylinositol-3,4,5-triphosphate 5-phosphatase, a putative Sec14, TgCDPK2A and TgCDPK7, and PPM2B (Table 1 and Figure 1F).

Peptides in clusters 2 and 3 (173 and 527, respectively) increased more gradually and exhibited lower fold-changes than cluster 1 sites (Figure 1D). Cluster 2 was notably enriched in proteins functioning in transport of monovalent ions and lipids, cyclase activity, and Ca²⁺ binding (Figure 1E). This set includes two putative Ca²⁺-activated K⁺ channels and the sodium-hydrogen exchangers NHE1 and NHE3 (Table 1). ATPase2 and the guanylyl cyclase in this cluster both have putative phospholipid-transporting ATPase domains. PI4,5K, PI3,4K, and TgCDPK2A, noted in cluster 1, have additional phosphoregulatory sites belonging to cluster 2. Clusters 2 and 3 are enriched for phosphoproteins with metal transporter activity, including ATPases such as TgA1, the putative copper transporter CuTP, a putative P5B-ATPase, a putative E1-E2 ATPase, and TgATP4. As observed in other T. gondii Ca²⁺-stimulated phosphoproteomes, at later time points small-molecule transporters are phosphorylated, including MFS transporters, ApiAT5-3, and TgFNT2. The guanylyl and adenylyl cyclases are also extensively modified, as are the cyclase organizers UGO and AIP.

Cluster 3, which represents a latter wave of phosphoregulation, is uniquely enriched in proteins functioning in subcellular remodeling, vesicle trafficking, and glideosome activity (Figure 1E). This set includes two divergent kinesins (KinesinB and TGGT1_201230), a dynein light chain, and the myosin motors MyoC, MyoF, and MyoJ (Table 1). Several uncharacterized GTPase regulators are phosphorylated later in the zaprinast response, including putative ARF1 activators (TGGT1_225310 and TGGT1_266830) as well as five uncharacterized TBC domain proteins. The glideosome-associated membrane proteins GAPM2B and GAPM3, which link the IMC, alveolin network, and microtubules, are also dynamically phosphorylated.

Cluster 4, the only cluster characterized by decreasing phosphorylation, was functionally enriched in phosphoproteins involved in transporter and MAP kinase activity. ERK7 was dephosphorylated within 30 s of zaprinast treatment, whereas TgTLK2 was dephosphorylated only at the final 60 s time point. ERK7 regulates conoid and cytoskeletal stability during cell division, with secondary functions in parasite egress, motility, and invasion (Back et al., 2020; Dos Santos Pacheco et al., 2021; O’Shaughnessy et al., 2020). TLK2 regulates parasite replication (Smith et al., 2022; Varberg et al., 2018). Several phosphoproteins functioning in ion transport belonged to cluster 4, including a putative K⁺ voltage-gated channel complex subunit, a putative Vacuolar (H+)-ATPase G subunit, TgATP4, and the guanylyl cyclase (Table 1), which has a P-type ATPase domain with unknown ion specificity. Small-molecule transporters included a putative nucleoside transporter, ApiAT5-3, and a MFS family transporter. Cluster 4 was the only class of peptides that was not functionally enriched in Ca²⁺ binding proteins (Figure 1E).

We identified phosphoproteins with peptides belonging to several clusters (Figure 1G). Such proteins may have multiple phosphosites regulated with different kinetics by the same enzyme; or by different enzymes that alight upon the target at varying spatiotemporal scales. For example, SCE1 and TGGT1_309910 have phosphopeptides belonging to all three increasing clusters, likely resulting from phosphorylation by TgCDPK3 and PKG, respectively. Indeed, SCE1 was implied to be a TgCDPK3 target through a genetic suppressor screen (McCoy et al., 2017). TGGT1_309910 is the ortholog of Plasmodium falciparum Pf ICM1, a PKG substrate identified through proteomic interaction studies (Balestra et al., 2021), although it remains uncharacterized in T. gondii. By contrast, several proteins belong to both increasing and decreasing clusters, likely targets of both kinases and phosphatases. We consider these proteins candidate signaling platforms. Several such proteins localize to discrete domains of the apical complex, including the guanylyl cyclase; apical annuli proteins AAP2 and AAP5; apical polar ring proteins TGGT1_227000 and RNG2; and the apical cap protein AC10 and a recently identified conoid gliding protein CGP (Table 1). Along with the guanylyl cyclase, the cAMP-specific phosphodiesterase PDE2 and a Ca²⁺ influx channel with EF hands exhibit peptides belonging to both increasing and decreasing clusters. Our phosphoproteome thus identifies candidate mediators of the feedback between cyclic-nucleotide and Ca²⁺ signaling.

The waves of phosphoregulation largely corroborate the sequence of physiological events observed following zaprinast treatment. The first cluster likely includes the most proximal targets of PKG. Previous PKG-dependent phosphoproteomes from the related apicomplexan P. falciparum similarly implicate PI-PLC as a substrate of the kinase (Alam et al., 2015; Balestra et al., 2021; Brochet et al., 2014). The product of PI-PLC activity, IP₃, stimulates Ca²⁺ store release in parasites (Garcia et al., 2017). In turn, phosphoproteins in clusters 2, 3 and 4 likely contain the targets of Ca²⁺-regulated kinases and phosphatases.

The phosphoproteome identified numerous dynamic changes in response to Ca²⁺ release. However, this information alone is not sufficient to infer the enzymes responsible for phosphorylation, or which events may be functionally relevant. Thermal proteome profiling (TPP) has been used to identify small molecule–target interactions in living cells and cell extracts (Dai et al., 2019; Mateus et al., 2020; Savitski et al., 2014). TPP operates on the premise that ligand binding induces a thermal stability shift, stabilizing, or destabilizing proteins that change conformationally in response to the ligand, and such changes in stability can be quantified by MS (Dziekan et al., 2020; Reinhard et al., 2015). Cells are incubated with different concentrations of ligand and heated, causing thermal denaturation of proteins. The soluble protein is extracted and quantified with multiplexed, quantitative methods, giving rise to thousands of thermal denaturation profiles. Proteins engaging the ligand are identified by their concentration-dependent thermal shift. We previously used this method to identify the target of the antiparasitic compound ENH1 (Herneisen et al., 2020; Herneisen and Lourido, 2021) and to measure changes to the T. gondii proteome when depleting the mitochondrial protein DegP2 (Harding et al., 2020). In a conceptual leap, we reasoned that TPP could also detect Ca²⁺-responsive proteins if parasite extracts were exposed to different concentrations of Ca²⁺, allowing us to systematically identify the protein components of signaling pathways on the basis of biochemical interactions with Ca²⁺ and its effectors.

Intracellular free Ca²⁺ levels span three orders of magnitude, from low nanomolar in the cytoplasm to high micromolar in organelles like the ER (Lourido and Moreno, 2015). To measure protein thermal stability at precisely defined [Ca²⁺]_free, we combined crude parasite extracts with calibrated Ca²⁺ buffers in a solution mimicking the ionic composition of the cytoplasm. As a proof of principle, we measured the thermal stability of the Ca²⁺-dependent protein kinase TgCDPK1, as the conformational changes of this enzyme have been structurally characterized (Ingram et al., 2015; Wernimont et al., 2010). Parasite lysates were adjusted to 10 different concentrations of free Ca²⁺ and heated to 58 °C, based on prior estimates of the melting temperature of TgCDPK1 (Herneisen et al., 2020). As measured by immunoblot band intensity, TgCDPK1 was strongly stabilized by Ca²⁺ (Figure 2A), suggesting that our experimental system is sensitive to Ca²⁺-dependent stability changes. The calculated EC₅₀ using this approach was in the low µM range, consistent with studies using recombinant enzymes (Ingram et al., 2015; Wernimont et al., 2010).

The effect of Ca²⁺ on the global thermostability of the proteome has not been assessed. Therefore, we first generated thermal profiles of the T. gondii proteome without or with 10 µM Ca²⁺, which is representative of the resting and stimulated Ca²⁺ concentrations of cell cytoplasm (Lourido and Moreno, 2015). A thermal challenge between 37 and 67°C induced denatured aggregates, which were separated from stable proteins by ultracentrifugation. The soluble fraction was digested and labeled with isobaric mass tags, pooled, fractionated, and analyzed with an orbitrap mass spectrometer (Figure 2B). We detected 3,754 proteins, of which 2,381 yielded thermal profiles for both conditions in both replicates (Figure 2C and Figure 2—figure supplement 1); the remaining proteins were not detected in all experiments. The median melting temperatures—the temperatures at which proteins are 50% denatured—were 54.4 and 54.7°C in the lysates without and with Ca²⁺, respectively. The distribution of melting temperatures was largely overlapping in the two conditions (Figure 2D), suggesting that Ca²⁺-dependent changes in protein stability were restricted to a subset of proteins. We additionally calculated an area under the curve (AUC) metric by numerical integration using the trapezoidal rule (Herneisen and Lourido, 2021) to compare the stabilities of proteins with atypical melting behavior (Figure 2—figure supplement 1), such as components of the tubulin cytoskeleton or parasite conoid.

To discover proteins with Ca²⁺-dependent stability shifts in the initial temperature range experiment, we rank-ordered proteins by Euclidean distance (ED) scores (Dziekan et al., 2020) quantifying the shift in thermal profiles with and without Ca²⁺ (Figure 2B and E). The majority of proteins, such as DNA polymerase β (TGGT1_233820), exhibited similar melting behavior in both conditions (Figure 2F). Our analysis identified as Ca²⁺-responsive parasite-specific proteins with EF hands, including TgCDPK7 and the calmodulin-like proteins CAM1 and CAM2 (Table 1, Figure 2F and Figure 2—figure supplement 1). The ED metric identified the stability changes of both CAM proteins despite a lack of typical melting behavior, supporting the use of this statistic. Membrane proteins, including potential Ca²⁺-leak channels localizing to the ER (Barylyuk et al., 2020; TGGT1_255900 and 206320), also exhibited thermal shifts (Figure 2F). Several proteins specific to the apicomplexan parasite phylum exhibited Ca²⁺-regulation (Barylyuk et al., 2020; Sidik et al., 2016a), including an apical lysine methyltransferase (AKMT) that relocalizes during Ca²⁺-stimulated egress Heaslip et al., 2011; DrpC, which regulates the stability of parasite organelles; a hypothetical protein (TGGT1_214340) with no annotated domains (Figure 2—figure supplement 1); and enzyme subunits involved in cellular metabolism.

The data also inform hypotheses about Ca²⁺ homeostasis and energetics in the parasite mitochondrion and apicoplast. A divergent subunit of the mitochondrial complex III, TgQCR9, as well as components of the ATP synthase complex, were destabilized by Ca²⁺ (Figure 2—figure supplement 1). These include the ATP synthase subunits α, γ, 8/ASAP-15, and f/ICAP11/ASAP-10; and the ATP synthase-associated proteins ASAP-18/ATPTG14 and ATPTG1 (Table 1). TGGT1_209950, a conserved alveolate thioredoxin-like protein suggested to localize to the apicoplast by spatial proteomics (Barylyuk et al., 2020), was similarly destabilized by Ca²⁺ (Figure 2—figure supplement 1). The apicomplexan mitochondrion has been reported to sense cytosolic Ca²⁺ fluctuations (Gazarini and Garcia, 2004), and the apicoplast may take up Ca²⁺ through direct contact sites with the ER (Li et al., 2021). Furthermore, redox signals were recently reported to induce parasite Ca²⁺ signaling and motility (Alves et al., 2021; Stommel et al., 1997). Our resource identifies several proteins that may couple ion homeostasis and cellular metabolism. The recent structure of the apicomplexan ATP synthase suggests that all of the Ca²⁺-responsive elements identified here, including subunits restricted to apicomplexan parasites, protrude into the mitochondrial matrix (Mühleip et al., 2021), which has elevated Ca²⁺ (Giorgi et al., 2018). In the case of TGGT1_209950, plants and algae have Ca²⁺-sensing thioredoxins (calredoxins) in chloroplasts (Hochmal et al., 2016) with Ca²⁺-regulated activity mediated by EF hands (Charoenwattanasatien et al., 2020). TGGT1_209950 lacks EF-hands; however, structural modeling based on primary sequence and predicted secondary features (Kelley et al., 2015; Meier and Söding, 2015) suggested similarity to calsequestrin, which lacks a structured Ca²⁺-binding motif. Our approach has thus identified candidate Ca²⁺-responsive proteins in metabolic organelles that would have been missed by bioinformatic searches; although, additional experiments are needed to determine the interactions each of these candidates has with Ca²⁺.

The temperature range experiment has the advantage of generating complete thermal stability profiles, but does not inform the relative affinities of the Ca²⁺-dependent stability change. To address this gap, we examined protein stability across 10 Ca²⁺ concentrations (EGTA and 10 nM to 1 mM). Based on the temperature-range experiments, we selected thermal challenge temperatures of 50, 54, and 58 °C to target protein with a range of thermal stabilities in these dose-response experiments (Figure 2—figure supplement 1). We hypothesized that Ca²⁺-responsive proteins would exhibit sigmoidal dose-response trends in thermal stability, similarly to TgCDPK1 (Figure 2A). We performed four independent concentration-range experiments on two mass spectrometers using different separation methods (ultracentrifugation or filtration) to capture different types of aggregates.

The concentration range thermal-profiling experiments provide insight into the magnitude of the Ca²⁺-dependent change (AUC parameter) and its dose-dependency (EC₅₀) (Figure 2B). Clear responses were observed for several known Ca²⁺-binding enzymes with EF hands, including calmodulin, calcineurin, and several parasite Ca²⁺-dependent protein kinases (CDPKs; Figure 3A). These parameters can be used to classify proteins as stabilized (AUC >1; e.g., TgCDPK1, TgCDPK2A, and TgCDPK3) or destabilized (AUC <1; e.g., calmodulin and calcineurin B) by Ca²⁺ (Figure 3A). Furthermore, EC₅₀ measurements may inform specific hypotheses about a protein’s involvement in signaling or homeostasis, based on the Ca²⁺ concentration at which the protein is predicted to change.

Our results generated a de novo catalog of proteins with Ca²⁺-dependent stability. Of the 4,403 proteins identified with sufficient quantification values for curve fitting, 228 were classified as Ca²⁺-responsive based on exhibiting a dose-response trend (R²) greater than 0.8 with a stability change (AUC) of two modified Z scores from the median (Figure 3B, Figure 3—figure supplement 1, and Supplementary file 2). Functional enrichment showed that such Ca²⁺-responsive proteins were significantly enriched in Ca²⁺-binding functions, protein kinase activity, and metal affinity (Figure 3C). We examined the 40 EF hand domain–containing proteins detected in our mass spectrometry experiments (Figure 3D), 25 of which exhibited dose-responsive behavior. Signaling and trafficking proteins were tuned to respond at lower Ca²⁺ concentrations: TgCDPK2A and CaM responded robustly with nM EC₅₀; TgCDPK1, TgCDPK3, CnB, and Eps15 exhibited low µM EC₅₀. In the cases of TgCDPK1, CaM, and CnB, these values are consistent with the Ca²⁺ activation constants determined from studies with recombinant proteins (Feng and Stemmer, 2001; Feng and Stemmer, 1999; Wernimont et al., 2010). By contrast, myosin-associated subunits responded at higher levels: Myosin light chains (ELC1, MLC1, MLC5, and MLC7) and the calmodulin-like proteins CAM1 and CAM2 responded above 10 µM Ca²⁺ (Table 1). Two additional proteins (TGGT1_255660 and TGGT1_259710) consistently responded with µM EC₅₀ values but have not been functionally characterized (Figure 3—figure supplement 1). The remaining EF hand domain–containing proteins may bind Ca²⁺ outside of the experimental concentration range; or may not undergo detectable stability changes upon Ca²⁺ binding (Mateus et al., 2020).

Several signaling enzymes lacking canonical Ca²⁺-binding domains exhibited a dose-response trend (R² >0.8) in our thermal profiling experiments (Figure 3B and E). The PKA regulatory and catalytic subunits (TGGT1_242070 and TGGT1_226030) responded robustly to Ca²⁺ with 10–100 µM EC₅₀ values in several experiments. A putative RIO1 kinase (TGGT1_210830) also responded consistently to Ca²⁺, although the magnitude of the change was small (Figure 3—figure supplement 1). The atypical kinase ERK7 was destabilized at high Ca²⁺ concentrations (Figure 3—figure supplement 1), similarly to other Ca²⁺-responsive proteins detected at the parasite apex. Rhoptry and dense granule kinases had EC₅₀ values in between 100 µM and 1 mM, consistent with the high concentration of Ca²⁺ in the secretory pathway (Figure 3E). We also searched for Ca²⁺-responsive phosphatases. The known Ca²⁺-regulated phosphatase subunit, CnB, was destabilized (Figure 3F). Protein Phosphatase 1 (PP1) responded to Ca²⁺ consistently (Figure 3F and 5A), although the catalytic subunit is not an intrinsic Ca²⁺ sensor. Our resource thus places proteins without previously characterized Ca²⁺ responses within the broader Ca²⁺ signaling network.

Our search and annotation of Ca²⁺-responsive proteins also identified candidates that link Ca²⁺ and ion homeostasis. This list includes several transporters and channels (Table 1): TgCAX, a protein with structural homology to LDL receptors (TGGT1_245610), two proteins with structural homology to the mitochondrial Ca²⁺ uniporter (TGGT1_257040 and TGGT1_211710), and an apicoplast two-pore channel. Several Ca²⁺-responsive hydrolases were also found, including SUB1, a metacaspase with a C2 domain, and a Ca²⁺-activated apyrase. These proteins may function during the intracellular, replicative phase of the lytic cycle, for which the function of Ca²⁺ is relatively unexplored.

Several divalent cations regulate protein structure and activity in addition to Ca²⁺, including Mg²⁺ and Zn²⁺. Such metal-binding proteins might appear Ca²⁺-responsive in our thermal profiling approach. We buffered our parasite extracts with excess Mg²⁺ (1 mM) to mitigate non-specific changes caused by the concentration of divalent cations. Nevertheless, to determine whether our approach identified additional metal-binding proteins, we compared the EC₅₀ values of candidates that bind different divalent cations, based on the presence of Interpro domains and through manual annotation (Figure 3—figure supplement 1). The EC₅₀ values of putative Mg²⁺-binding proteins were mostly in the mM range, which may result from nonspecific displacement of Mg²⁺ by Ca²⁺. However, a subset of Mg²⁺-binding proteins were stabilized by micromolar Ca²⁺–for instance, glucosephosphate-mutase (GPM1) family proteins TGGT1_285980 A and B. The GPM1 ortholog in Paramecium, parafusin (PFUS), is involved in Ca²⁺-regulated exocytosis (Subramanian and Satir, 1992) A PFUS homolog in T. gondii has been observed to co-localize with micronemes in the apical third of the cell (Matthiesen et al., 2001). Therefore, the thermal stabilization of GPM1 could arise from Ca²⁺-dependent modifications to the enzyme (Subramanian et al., 1994; Subramanian and Satir, 1992).

We selected five proteins exhibiting consistent Ca²⁺-responsive behavior by MS for validation: an EF hand domain–containing protein (Eps15), two kinases (RON13 and PKA-C1), and two uncharacterized putative metal-binding proteins (TGGT1_286710 and TGGT1_309290). The first three candidates were selected for potential involvement in dynamic Ca²⁺-regulated processes. Eps15 (TGGT1_227800) was recently shown to mediate endocytosis in P. falciparum (Birnbaum et al., 2020) and localized to puncta bridging the inner membrane complex (IMC) and cytoskeleton in T. gondii (Chern et al., 2021). PKA-C1 is thought to antagonize Ca²⁺ signaling in T. gondii (Jia et al., 2017; Uboldi et al., 2018). RON13 is a secretory-pathway kinase that phosphorylates substrates in apicomplexan invasion-associated organelles called rhoptries (Lentini et al., 2021). The function of the remaining two proteins is unknown, although TGGT1_286710 contains a zinc finger domain and TGGT1_309290, annotated as a hypothetical protein, contains an HD/PDEase domain—both contribute to parasite fitness in cell culture (Sidik et al., 2016a).

We appended epitope tags to the endogenous locus of each candidate. These Ca²⁺-responsive proteins localized to diverse structures (Figure 4B). PKA-C1 resided at the parasite pellicle, as previously reported (Jia et al., 2017; Uboldi et al., 2018). Eps15 concentrated at puncta in the apical third of the parasite. TGGT1_286710 and TGGT1_309290 localized to the parasite nucleus, with TGGT1_286710 exhibiting additional staining in the residual body. Finally, RON13 appeared to stain the rhoptries (Lentini et al., 2021). To validate the Ca²⁺-dependent stability of individual candidates, we prepared parasite extracts as described earlier, but relied on an immunoblot readout instead of MS. In these five cases, stabilization of the candidates was confirmed in multiple biological replicates. Several controls (TUB1, MIC2, GAP45, and SAG1) were stable across all Ca²⁺ concentrations tested (Figure 4 and Figure 4—figure supplement 1). In the case of PKA-C1, Eps15, and both putative metal-binding proteins, the immunoblot experiments revealed an even higher EC₅₀ than was measured in the MS experiments. We conclude that the stability changes detected by our global proteomics methods are robust, although the precise features of Ca²⁺-dependent stability may differ based on the method used to assess them.

Our orthogonal proteomics approaches map Ca²⁺-responsive phosphorylation and thermal stability. Unexpectedly, the catalytic subunit of Protein Phosphatase 1 (PP1, TGGT1_310700) exhibited consistent stabilization by Ca²⁺ in our mass spectrometry experiments (Figure 3F and Figure 5A). The contribution of phosphatases to Ca²⁺ signaling in apicomplexans is poorly understood (Yang and Arrizabalaga, 2017) and the Ca²⁺-responsive behavior of PP1 has not been reported in other eukaryotes. The function of the phosphatase has not been directly studied in T. gondii, although PP1 inhibitors have been shown to block invasion of host cells (Delorme et al., 2002), and recent experiments in P. falciparum suggest that PP1 is required for the merozoite egress-to-invasion transition (Paul et al., 2020). Intriguingly, PP1 was recently observed to relocalize to the apical complex in highly motile Plasmodium berghei ookinetes (Zeeshan et al., 2021), suggesting that the enzyme may serve apicomplexan-specific, Ca²⁺-responsive functions in remodeling the parasite phosphoproteome.

To track localization during the T. gondii lytic cycle, we tagged the endogenous C terminus of PP1 with an mNeonGreen (mNG) fluorophore and Ty epitope. We confirmed the Ca²⁺-dependent stability of PP1 by immunoblotting (Figure 5B). Live microscopy revealed a diverse array of PP1 localizations in parasites (Figure 5C). In accordance with imaging in Plasmodium (Zeeshan et al., 2021), PP1-mNG was distributed diffusely in the cytoplasm, as well as in foci resembling the nucleus, centrosome, and in some parasites, the periphery. These diverse localizations may arise from the association of PP1 with distinct regulatory subunits forming different functional holoenzymes, as characterized in metazoans (Brautigan and Shenolikar, 2018). To determine whether PP1 exhibits Ca²⁺-dependent relocalization, we treated parasites with zaprinast and A23187. The PP1 signal intensity increased at the apical cap and pellicle following zaprinast stimulation (Figure 5C and Video 1). However, Ca²⁺ ionophore treatment failed to induce the same relocalization patterns (Figure 5C and Video 2). The dynamics of PP1 activity thus appear enhanced by cGMP signaling within the parasites, which is upstream of Ca²⁺ (Brown et al., 2016; Sidik et al., 2016b) and was similarly observed in Plasmodium (Paul et al., 2020).

The parasite apex and pellicle are hotspots for the signaling that potentiates motility and invasion, so the relocalization of PP1 suggests it may function in these processes. We created strains expressing the PP1 catalytic subunit with an endogenous C-terminal auxin-inducible degron (AID) for rapid conditional knockdown (Figure 5D). Conditional degradation of PP1, following treatment with indole-3-acetic acid (IAA) was confirmed by immunofluorescence and immunoblotting (Figure 5E and F). Parasites depleted of PP1 failed to form plaques (Figure 5G), implicating the phosphatase in one or more essential functions during the lytic cycle (Sidik et al., 2016a). Even in the absence of IAA, the PP1-AID strain formed small plaques, indicating substantial hypomorphism. PP1-AID parasites exhibited slower replication than untagged parasites, and this effect was exacerbated with IAA treatment (Figure 5H). Parasites depleted of PP1 had a reduced invasion efficiency (Figure 5I), although this effect was modest and subject to technical variation, likely arising from the hypomorphism of the C-terminally tagged strain. Parasites treated with IAA egressed more slowly than untreated parasites when stimulated with zaprinast; however, egress kinetics were indistinguishable with a Ca²⁺ ionophore agonist (Figure 5J). Together, these results suggest that PP1 holoenzymes function at multiple steps in the lytic cycle. At least one holoenzyme relocalizes when parasite cGMP/Ca²⁺ pathways are stimulated. Because parasites lacking PP1 exhibit delays specifically in zaprinast-induced egress, we hypothesize that the peripheral holoenzyme enhances Ca²⁺ signaling.

PP1 is one of the major serine/threonine phosphatases in eukaryotic cells (Brautigan and Shenolikar, 2018). As the catalytic subunit relocalizes during cGMP/Ca²⁺-stimulated transitions in apicomplexans, we hypothesized that PP1 dephosphorylates crucial targets during egress and motility. To identify putative PP1 holoenzyme targets, we first treated intracellular PP1-AID parasites with IAA or vehicle for 3 hr prior to mechanically releasing parasites for analysis. We performed sub-minute phosphoproteomics by resuspending the extracellular parasites in a zaprinast solution followed by denaturation in SDS to stop enzymatic activity after 0, 10, 30, and 60 s. Multiplexing with TMTpro reagents followed by phosphopeptide enrichment allowed us to compare the zaprinast time courses with or without PP1-depletion in biological duplicate (Figure 6A). Analysis of an unenriched fraction of the proteome revealed significant depletion of PP1, which we confirmed in parallel by immunoblot (Figure 6—figure supplement 1). The hypomorphism of the PP1-AID strain and reduced parasite yield resulted in a phosphoproteome with lower coverage: 6,916 phosphopeptides with quantification values. To identify peptides exhibiting PP1-dependent regulation, we selected those exhibiting a twofold difference in abundance between vehicle- and IAA-treated samples in both replicates for at least one time point. In total, 757 peptides passed this threshold.

To focus on zaprinast-dependent changes, we clustered these peptides on the basis of their abundance relative to the earliest time point of stimulation. The peptides fit into five clusters with respect to PP1-dependence and kinetics (Figure 6B). Clusters 1, 2, and 3 contained peptides increasing in abundance upon stimulation in PP1-depleted parasites. The peptides belonging to cluster 1 generally increase in abundance with zaprinast treatment, which occurs more rapidly with PP1 depletion. By contrast, the abundances of the 614 peptides belonging to clusters 2 and 3 were elevated in the absence of PP1, suggesting that PP1 antagonizes these phosphorylation events. Under normal conditions, peptides in cluster 4 decreased sharply in abundance between 10 and 30 s of zaprinast treatment and recovered by 60 s; however, these peptides did not change in abundance when PP1 was depleted. Peptides in cluster 5 increased gradually between 10 and 30 s and decreased to basal levels by 60 s; when PP1 is depleted, these peptides exhibit a delay.

The effect of PP1 disruption on the phosphoproteome was pervasive, which may reflect disruption of numerous PP1 holoenzymes. Many of the examined phosphopeptides exhibited substantial basal hyperphosphorylation in the absence of PP1 (Figure 6C). Given the likely pleiotropy of catalytic subunit depletion, we focused our analysis on perturbed pathways rather than individual targets (Figure 6D). Cluster 1 did not possess enough peptides for enrichment analysis. Cluster 2 was enriched in phosphoproteins functioning in transmembrane transport, including P-type ATPases and ABC transporters. Cluster 3 contained both the guanylyl and adenylyl cyclases and was further enriched in putative sodium-hydrogen exchangers and tubulin-tyrosine ligases. Cluster 4 was solely enriched for proteins involved in ubiquitin transfer, including TGGT1_280660, an uncharacterized HECT domain-containing protein (Figure 6C). Cluster 5 phosphoproteins were notably involved in cytoskeletal motor activity, actin binding, and RNA binding. Numerous apical proteins exhibited PP1-dependent phosphorylation (Table 1), including AC7, CIP1, and two hypothetical proteins—TGGT1_258090 and TGGT1_320030, localized to the conoid base and the second apical polar ring, respectively (Koreny et al., 2021). The protein abundances did not vary between vehicle and IAA treatment, indicating that the phosphopeptide abundance changes arose from dynamic covalent modifications. The PP1-dependent phosphoproteome therefore supports the existence of apical and peripheral PP1 holoenzymes, as observed by live microscopy.

The phosphoproteomics data pointed to ion homeostasis dysregulation in the absence of PP1 (Figure 6D). Numerous transporters are hyperphosphorylated when PP1 is depleted, including Ca²⁺ ATPases, proton transporters, and MFS and ABC-family transporters. We hypothesized that aberrant Ca²⁺ mobilization may underlie the phenotypes observed in the motile stages of the lytic cycle upon PP1 depletion (Figure 5), as Ca²⁺ release and signaling are required for egress and invasion. To monitor Ca²⁺ mobilization in parasites, we introduced the genetically encoded Ca²⁺ indicator GCaMP6f into a defined chromosomal locus (Chen et al., 2013; Herneisen et al., 2020). By tracking the vacuole fluorescence of PP1-depleted parasites, we observed delayed Ca²⁺ mobilization and egress following zaprinast treatment (Figure 7A–C, Figure 7—figure supplement 1, and Video 3). Despite the delay, IAA-treated parasites attained the same increase in Ca²⁺ levels prior to egress as vehicle-treated parasites (Figure 7—figure supplement 1). By contrast, Ca²⁺ mobilization upon A23187 treatment was unperturbed in PP1-depleted parasites (Figure 7D–F, Figure 7—figure supplement 1 and Video 4). Motivated by recent experiments suggesting that T. gondii exhibits Ca²⁺-activated Ca²⁺ entry (Vella et al., 2021), we loaded PP1-AID parasites with the ratiometric Ca²⁺ indicator Fura2-AM and observed both reduced Ca²⁺ entry and lowered resting cytoplasmic [Ca²⁺] (Figure 7G–I and Figure 7—figure supplement 1). The integration of our global datasets therefore identified PP1 as a Ca²⁺-responsive enzyme and precipitated discovery of its role in Ca²⁺ entry.

T. gondii parasites of the type I RH strain ∆ku80/∆hxgprt genetic background (ATCC PRA-319) were grown in human foreskin fibroblasts (HFFs, ATCC SCRC-1041) maintained in DMEM (GIBCO) supplemented with 3% inactivated fetal calf serum (IFS) and 10 μg/mL gentamicin (Thermo Fisher Scientific). When noted, DMEM was supplemented with 10% IFS and 10 μg/mL gentamicin.