Plasma Membrane Association by N-Acylation Governs PKG Function in Toxoplasma gondii

ABSTRACT Cyclic GMP (cGMP)-dependent protein kinase (protein kinase G [PKG]) is essential for microneme secretion, motility, invasion, and egress in apicomplexan parasites, However, the separate roles of two isoforms of the kinase that are expressed by some apicomplexans remain uncertain. Despite having identical regulatory and catalytic domains, PKGI is plasma membrane associated whereas PKGII is cytosolic in Toxoplasma gondii. To determine whether these isoforms are functionally distinct or redundant, we developed an auxin-inducible degron (AID) tagging system for conditional protein depletion in T. gondii. By combining AID regulation with genome editing strategies, we determined that PKGI is necessary and fully sufficient for PKG-dependent cellular processes. Conversely, PKGII is functionally insufficient and dispensable in the presence of PKGI. The difference in functionality mapped to the first 15 residues of PKGI, containing a myristoylated Gly residue at position 2 that is critical for membrane association and PKG function. Collectively, we have identified a novel requirement for cGMP signaling at the plasma membrane and developed a new system for examining essential proteins in T. gondii.

in T. gondii is paramount to successful invasion and egress; therefore, microneme secretion must be tightly regulated.
Microneme secretion is controlled by two key signaling pathways, calcium (Ca 2ϩ ) and cyclic GMP (cGMP), which direct calcium-dependent protein kinases (CDPKs) (6) and protein kinase G (PKG) (7,8), respectively, to transduce their respective signals through substrate phosphorylation. Phosphoproteomic studies have identified substrates for CDPK1 in T. gondii (9) and PKG in Plasmodium spp. (10,11), yet it is still unclear which phosphorylation events are required for the control of microneme secretion. Interestingly, Ca 2ϩ and cGMP may cooperate but also regulate specific steps in microneme secretion. In T. gondii, serum albumin stimulates PKG-dependent microneme secretion by elevating cGMP but not Ca 2ϩ (12). However, compounds that elevate Ca 2ϩ also stimulate microneme secretion (13)(14)(15) but only when they receive a second signal such as serum albumin (12). Moreover, elevated calcium cannot overcome PKG inhibition (7), suggesting that PKG may control a final step in this process and act as the master regulator.
Apicomplexan parasites encode a single PKG gene (16) that is refractory to deletion in T. gondii (17). Selective inhibition of apicomplexan PKG kinase activity is lethal (16)(17)(18)(19), supporting an essential role. Despite having a single gene, several genera of tissue cyst-forming coccidian parasites, including Toxoplasma, Hammondia, Neospora, and Eimeria species, express two isoforms of PKG that localize to the plasma membrane (PKG I ) and cytosol (PKG II ), respectively (16,17,19). In T. gondii, PKG I (residues 1 to 994) harbors an N-terminal dual-acylation motif that promotes stable association with the plasma membrane, whereas PKG II (residues 103 to 994), which is initiated from a second downstream methionine, lacks this motif and remains cytosolic (19). PKG can only be deleted in T. gondii in the presence of a second copy of the gene, including those that encode nonacylated mutant proteins (17), although this result might be due to overexpression. Thus, whether PKG I and PKG II are functionally distinct or redundant is currently unknown.
T. gondii is equipped with excellent genetic tools, including clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 genome editing (20,21). A recent genome-wide CRISPR study performed with T. gondii demonstrated that 40% of the 8,158 genes surveyed have a fitness defect in vitro, with perhaps only 10% of these being essential (PKG included), and identified~200 hypothetical fitness-conferring genes conserved only in apicomplexans (22). Since the vast majority of these genes have yet to be studied and are likely to be refractory to deletion, conditional genetic technologies are imperative for functional studies. To facilitate the study of essential genes, we adapted the auxin-inducible degron (AID) system that has been shown to degrade proteins of interest in other eukaryotes (23). We utilized a mini-AID (mAID) tagging system for conditional protein depletion that enabled us to resolve functional differences between PKG isoforms in T. gondii. This study also highlights a broadly applicable tool for functional analysis of essential proteins in T. gondii.

RESULTS
Generation of an AID system for conditional protein depletion in T. gondii. To rapidly deplete PKG isoforms in T. gondii, we developed an auxin-based system for conditional and specific protein depletion (Fig. 1A). We stably expressed a codonoptimized TIR1-3FLAG construct in a T. gondii RH line that lacks Ku80 (ku80 KO ) (Fig. 1B). To validate this system, we introduced a yellow fluorescent protein (YFP)-mAID-3HA reporter into TIR1-3FLAG parasites to selectively target this protein for proteasomal degradation ( Fig. 1B and C). Addition of auxin promoted rapid depletion of YFP-mAID-3HA, but not the control protein SAG1, within 15 min of treatment (Fig. 1D, top). Pretreatment of parasites with a proteasome inhibitor promoted YFP-mAID-3HA stabilization, confirming the role of the proteasome in YFP-mAID-3HA knockdown (Fig. 1D, middle). Efficient knockdown of YFP-mAID-3HA was independently confirmed and quantified by immunofluorescence (IF) microscopy ( Fig. 1E and F).
To adapt the degradation system for biologically important targets, we developed an efficient tagging strategy to incorporate the auxin degron into essential genes. CDPK1 was selected because it was previously determined to be essential with a transcriptional knockdown system (6). Using a recently described system for efficient CRIPSR/Cas9-mediated gene editing (20), we generated a tagged CDPK1-mAID-3HA gene in TIR1-3FLAG parasites (see Fig. S1A and B in the supplemental material). CDPK1-mAID-3HA was fully responsive to auxin treatment on the basis of IF microscopy (see Fig. S1C). Auxin-induced depletion of CDPK1-mAID-3HA, but not the control protein aldolase, occurred within minutes of treatment, as shown by Western blotting (see Fig. S1D and E). Degradation of CDPK1-mAID-3HA blocked plaque formation, a measure of the complete lytic cycle, on host cell monolayers (see Fig. S1F), validating the mAID system for the study of essential proteins in T. gondii.
Conditional depletion of the PKG I and PKG II isoforms confirms their essentiality in T. gondii. To study the function of the PKG isoforms in T. gondii, we utilized the mAID system to tag the endogenous locus. Addition of this fusion to the C terminus resulted in the production of both isoforms fused to mAID-3HA ( Fig. 2A and B). Previous work has established that PKG I possesses an N-terminal dual-acylation motif that targets it to the plasma membrane, whereas PKG II , lacking this motif, remains cytosolic (19). Consistent with this finding, the combined PKG isoform localizations were evident at both the plasma membrane and the cytoplasm (Fig. 2C). Additionally, both mAID- Rapid Conditional Protein Depletion in T. gondii ® 3HA-tagged PKG isoforms were detectable in parasite lysates by Western blotting and both were depleted by auxin treatment for 4 h (Fig. 2D). Depletion of PKG I,II -mAID-3HA completely blocked plaque formation on host monolayers ( Fig. 2E and F) but did not affect parasite replication (Fig. S2), suggesting a role in motility. . Importantly, both wild-type PKG I,II and PKG I -specific constructs fully rescued microneme secretion following PKG I,II -mAID-3HA depletion (Fig. 4B, lanes 3 to 6). Interestingly, PKG II -specific constructs only partially supported microneme secretion following PKG I,II -mAID-3HA  10), suggesting that microneme-dependent processes would also be affected. Similarly, depletion of PKG I,II -mAID-3HA caused an~75% reduction in invasion efficiency (Fig. 4C). Complementation with wild-type PKG I,II and PKG I -specific constructs fully rescued invasion, whereas only partial rescue was obtained with PKG II -specific constructs (Fig. 4C). Likewise, wild-type PKG I,II and PKG I -specific constructs, but not PKG II -specific constructs, fully supported calcium ionophore-induced egress from host cell monolayers following PKG-mAID-3HA depletion (Fig. 4D). Given the partial sufficiency of PKG II to function in microneme secretion and related processes, we asked whether PKG II alone could support parasite growth and fitness over the course of several days. Although PKG II -complemented lines could partially support microneme secretion, invasion, and egress, they were insufficient at supporting plaque formation (Fig. 4E). Conversely, parasites complemented with wildtype PKG I,II or PKG I -specific constructs were fully capable of forming plaques on human foreskin fibroblast (HFF) monolayers upon PKG I,II -mAID-3HA depletion (Fig. 4E).
To confirm that PKG II is dispensable, we developed a markerless genome editing strategy to introduce an M103A mutation into the endogenous PKG-mAID-3HA gene, thereby preventing PKG II translation (Fig. S4A). We cotransfected parasites with a Cas9-green fluorescent protein (GFP)/guide RNA plasmid that cuts 6 bp upstream from the M103 codon and a Cas9-shielded homology donor amplicon corresponding to exon 3 (252 bp) containing the M103A mutation. Following transfection, parasites transiently expressing Cas9-GFP were sorted with a fluorescence-activated cell sorter (FACS) (Fig. S4B) and expanded on host cell monolayers (Fig. S4C). We hypothesized that parasites that expressed Cas9 would only survive a double-stranded break in PKG if it were repaired with the donor M103A template. Single clones were isolated from the FACS-sorted population by limiting dilution, and a 1-kb fragment surrounding the PKG M103 allele was amplified by PCR and sequenced by the Sanger method. The M103A and shielding mutations were evident in a transfected clone but not the parental line, confirming successful markerless editing (Fig. S4D). We confirmed loss of PKG II at the protein level by Western blotting, demonstrating that this line only expresses PKG I (Fig. 5A). To rule out compensatory mutations during genome editing, we depleted PKG I (Fig. 5B) and found that parasites were incapable of growth ( Fig. 5C and D). Taken together, these data reveal that PKG II is dispensable, while PKG I is essential, for growth on host cell monolayers.
Plasma membrane association is critical for PKG function in T. gondii. The observation that PKG II was unable to fully complement microneme secretion and Rapid Conditional Protein Depletion in T. gondii ® sustain parasite growth was intriguing, given that the isoform-specific complementation constructs showed similar levels of protein expression (Fig. S3C). Other than expression, the only known difference between the PKG I and PKG II isoforms is the presence of an N-terminal extension that contains a dual myristoylation and palmitoylation motif that is required for the plasma membrane association of PKG I (19). We hypothesized that in T. gondii, PKG activity at the plasma membrane is critically important for PKG function and parasite fitness. To test this hypothesis, we first needed to define the PKG I motif that could direct a soluble protein to the plasma membrane. We generated an mNeon-6Ty fusion with the first 15 amino acids (aa) of PKG I that was able to direct this fluorescent reporter to the plasma membrane ( Fig. S5A to C). Plasma membrane association is likely to require glycine myristoylation at position 2 of PKG I since a G2A mutant version of this peptide fused to mNeon-6Ty remained cytosolic ( Fig. S5A to C). Likewise, we found that the first 15 aa of PKG I were sufficient to redirect PKG II from the cytosol to the plasma membrane ( Fig. 6A and B, top and middle). To rule out any unforeseen PKG I -specific functions that may be present in the 15-aa peptide, we also directed PKG II to the plasma membrane with a similar N-terminal acylated peptide from CDPK3 (24) (Fig. 6A and B, bottom). The expression levels and localization of the ectopic PKG II constructs were independent of the endogenous copy of PKG I,II (Fig. 6C). Using these tools, we asked whether PKG II directed to the plasma membrane by N-terminal PKG I aa 1 to 15 or CDPK3 aa 1 to 15 fusions was sufficient to support parasite growth and fitness following PKG I,II -mAID-3HA depletion in a plaque formation assay. We observed that parasites complemented with a cytosolic PKG II construct could not form plaques on host cell monolayers following auxin-induced PKG I,II -mAID-3HA depletion ( Fig. 6D and E, left column). Surprisingly, parasites complemented with plasma membrane-associated PKG II constructs were fully capable of forming plaques when PKG I,II -mAID-3HA was degraded by growth in auxin ( Fig. 6D and E, middle and right columns). Therefore, we conclude that in T. gondii, PKG localization to the plasma membrane, but not the cytosol, is essential for proper PKG function and parasite viability.

DISCUSSION
Chemical genetic studies have demonstrated that PKG activity is required for the lytic life cycles of T. gondii and other apicomplexan parasites (16)(17)(18)(19)25). However, technical limitations of these methods have precluded the assignment of specific functions to membrane and soluble forms of PKG in parasites that express two isoforms. To define the functional contributions of each PKG isoform, we developed an AID tagging system that allowed rapid and robust conditional protein depletion in T. gondii. Simultaneous knockdown of both PKG isoforms was lethal, confirming their essentiality. Using a highly efficient CRISPR/Cas9 system for genome editing, combined with the AID degradation system, we found that plasma membrane-associated PKG I was necessary and sufficient, whereas cytosolic PKG II was largely insufficient and dispensable. Importantly, we were able to impart sufficiency to cytosolic PKG II by artificially directing it to the plasma membrane with N-terminal PKG I or CDPK3 peptide fusions. The combination of CRISPR/Cas9-mediated editing with mAID conditional protein regulation provides a powerful system for interrogation of the roles of essential proteins, including closely related isoforms of PKG.
Much of what is known about PKG function in apicomplexans was inferred from the effects of the PKG inhibitors, including the trisubstituted pyrrole inhibitor (compound 1) (16) and the imidazopyridine inhibitor (compound 2) (26). PKG was validated as a target by using gatekeeper mutants that were resistant to compound 1 and 2 inhibition (27). However, both compounds 1 and 2 have activity on other kinases, including CDPK1, which controls functions similar to those of PKG (26). Additional PKGindependent effects of compound 1 include induction of tissue cyst development in T. gondii (28,29). Furthermore, gatekeeper sensitization alone may lead to artificial results due to functional compensation, as was recently described for a gatekeeper mutant form of CDPK1 in P. falciparum (30). Thus, although PKG appears to be an essential gene in T. gondii, systems that allow careful dissection of the roles of two different isoforms have been lacking.
Conditional gene regulation is a powerful approach for understanding gene product function, and several such systems have been implemented in T. gondii (reviewed in reference 31). Transcriptional systems are relatively slow since they require natural turnover of existing mRNA and/or protein. Gradual depletion of proteins can lead to artificial phenotypes or mask genuine phenotypes, as has been demonstrated for the yeast helicase gene MCM4 (23,32). In contrast, posttranslational systems allow the rapid depletion of gene products (proteins) of interest. For instance, ddFKBP is an intrinsically unstable regulatable protein degron that is stabilized by the compound shield-1 (33). When used previously in T. gondii, addition of shield-1 fully stabilized a ddFKBP-YFP reporter within 90 min, whereas removal of shield-1 depleted ddFKBP-YFP to the background level after 320 min (34), representing a substantial improvement in speed over other systems. Unfortunately, the ddFKBP knockdown system requires permanent treatment of shield-1, an expensive compound known to have some toxicity for apicomplexan parasites (35), prompting us to develop an alternative approach. Unlike the ddFKBP system, the AID system uses an inexpensive and innocuous plant hormone (auxin/indole-3-acetic acid [IAA]) that is applied only when knockdown of an AID fusion is desired (23). To regulate proteins of interest with the AID system, two novel genetic components are needed. First, the cell must express an auxin receptor called transport inhibitor response 1 (TIR1) that combines to form a functional SCF ubiquitin ligase complex (SCF TIR1 ) in the presence of auxin (23). Here we expressed a codon-optimized version of TIR1 in T. gondii, as earlier attempts with the OsTIR1 gene from rice failed. Second, proteins targeted for depletion must be linked to an AID, which may be as small as 68 aa (36). Addition of auxin promotes polyubiquitination of AID-tagged proteins by the SCF TIR1 complex, targeting them for proteasomal degradation. The AID system has previously been used in Plasmodium spp. to examine the roles of essential proteins (37,38), prompting us to import the system into T. gondii. In our experience, the AID system operates an order of magnitude faster than the ddFKBP system in T. gondii (30 min for AID, as shown here, versus 320 min [34]). Furthermore, truncated versions of AID function efficiently with only 43 to 68 aa (36,39), limiting potential interference caused by bulky fusions. On the basis of these properties, the mAID system provides an efficient ligand-off protein expression tool in T. gondii that features unmatched specific protein knockdown speeds with an inexpensive, nontoxic ligand (auxin/IAA).
We applied the mAID system to PKG to confirm its essentiality and resolve functional differences between PKG isoforms in T. gondii. We found that plasma membraneassociated PKG I was necessary and sufficient for wild-type level function, whereas cytosolic PKG II was largely insufficient and dispensable. PKG II mutants that were artificially targeted to the plasma membrane were also sufficient, revealing a novel plasma membrane association requirement for PKG function in this parasite. These results were unexpected for two reasons. First, most of the PKGs of apicomplexan parasites with sequenced genomes and gene annotations lack the dual-acylation motif required for plasma membrane association and are likely cytosolic (40). However, such cytosolic isoforms of PKG should still have access to the plasma membrane by diffusion and may transiently fulfill a functional requirement for PKG at this interface. In fact, PKG-dependent phosphoproteomic studies performed with Plasmodium spp. identified several potential substrates at the plasma membrane (10,11). Second, a previous study of T. gondii showed that endogenous PKG could be deleted in parasites expressing Eimeria tenella FLAG-PKG, where the FLAG epitope was thought to disrupt the dualacylation sequence and promote cytosolic localization, implying that plasma membrane association is not required for PKG function (17). However, since E. tenella FLAG-PKG was expressed by the T. gondii tubulin promoter (17), it is possible that its overexpression was responsible for compensating for loss of PKG I at the plasma membrane. In our studies with the endogenous promoter, we found that PKG II alleles were not sufficient to rescue function, unless directed to the plasma membrane by an N-terminal fusion carrying an acylation signal.
In addition to T. gondii, several other tissue cyst-forming coccidian parasites, including Hammondia, Neospora, and Eimeria species, also carry an additional isoform of PKG to function at the plasma membrane. The reason why this location is essential in T. gondii while a soluble form is sufficient in Plasmodium and other genera is uncertain. However, it is possible that by directing the enzyme to the membrane, the overall expression level can be reduced, making substrate engagement more specific and preventing nonproductive phosphorylation of undesirable targets. Consistent with this model, evidence that PKG can compensate for other plasma membrane-directed kinases such as CDPK3 during egress by T. gondii (41) suggests a necessity for tight control or spatial restriction of PKG activity. It is also unclear why species that retain both plasma membrane and cytosolic forms retain the latter form, as our functional studies performed here suggest that the cytosolic form is dispensable. Our findings may reflect a vestigial function for PKG II or alternatively suggest a role in vivo that has not been tested in the present study. Collectively, our studies demonstrate the power of combining CRISPR/Cas9 genome engineering with conditional protein regulation to determine gene function and essentiality. Our studies identify the plasma membrane as the central platform for cGMP effector signaling in T. gondii, which should greatly facilitate the identification of currently unknown substrates of PKG that are required for microneme secretion.

MATERIALS AND METHODS
depending on the experiment. Plaque formation was assessed by counting the zones of clearance on EtOH-fixed, crystal violet-stained HFF monolayers.
Parasite growth. Freshly harvested parasites were allowed to invade HFF monolayers grown on glass coverslips for 1 h to allow invasion, and then cultures were treated with 500 M IAA to deplete mAID-3HA fusion proteins or the vehicle (EtOH) and grown for an additional 23 h. Infected monolayers were fixed at 24 h, and the number of parasites per vacuole was determined by IF microscopy. In each experiment, 20 image fields containing approximately 10 to 20 vacuoles were analyzed per treatment condition.
Microneme secretion. Parasites grown in HFF monolayers were pretreated with 500 M IAA to deplete mAID-3HA fusion proteins or the vehicle (EtOH) for 4 h. Parasites were then syringe released, 3 m filtered, washed, and resuspended in extracellular (EC) buffer (5 mM KCl, 142 mM NaCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 5.6 mM D-glucose, 25 mM HEPES, pH 7.4), and 2 ϫ 10 7 parasites were stimulated with 1% (wt/vol) BSA--1% (vol/vol) EtOH (final concentrations) in EC buffer or in EC buffer alone for 10 min at 37°C in the presence or absence of 500 M IAA. Following stimulation, parasites were chilled on ice and pelleted at 400 ϫ g for 10 min. Excreted/secreted antigen (ESA) fractions were collected and centrifuged once more at 800 ϫ g for 10 min. The twice cleared cell-free ESA fractions were subjected to Western blotting to assess microneme secretion. In each experiment, one replicate per treatment per strain was analyzed.
Parasite invasion. Parasites grown in HFF monolayers were pretreated with 500 M IAA to deplete mAID-3HA fusion proteins or the vehicle (EtOH) for 4 h. Parasites were then harvested, resuspended in D3, and used to infect HFF monolayers grown in optically clear 96-well plates (Greiner) for 20 min at 37°C in the presence of 500 M IAA or the vehicle (EtOH). Invasion was stopped by formaldehyde fixation, extracellular parasites were exclusively labeled with mouse anti-SAG1--Alexa Fluor 594 conjugate, and unbound antibody was removed by washing with PBS. The monolayers were then permeabilized with 0.1% Triton X-100, and all parasites were labeled with mouse anti-SAG1--Alexa Fluor 488 conjugate. Host nuclei were stained with Hoechst 33342 dye. A Cytation 3 multimode plate imager running Gen5 software (BioTek Instruments) was used to image parasites and quantitate the number that invaded each host cell. In each experiment, five replicates per sample and 16 image fields per replicate were analyzed.
Parasite egress. Freshly harvested parasites were resuspended in D3 and counted, and 5 ϫ 10 4 parasites were allowed to invade HFF monolayers grown on glass coverslips for 20 h. Parasites were then pretreated with 500 M IAA for 4 h to deplete mAID-3HA fusion proteins or the vehicle (EtOH). To stimulate egress, parasites were treated with 4 M (final concentration) calcium ionophore A23187 or the vehicle for 5 min at 37°C. Egress was stopped by formaldehyde fixation and evaluated by IF microscopy following permeabilization, blocking, and immunolabeling with antibodies against the parasite (SAG1) and parasitophorous vacuole (GRA7). In each experiment, 10 fields per treatment per strain were analyzed.