The Toxoplasma gondii Rhoptry Kinome Is Essential for Chronic Infection

ABSTRACT    Ingestion of the obligate intracellular protozoan parasite Toxoplasma gondii causes an acute infection that leads to chronic infection of the host. To facilitate the acute phase of the infection, T. gondii manipulates the host response by secreting rhoptry organelle proteins (ROPs) into host cells during its invasion. A few key ROP proteins with signatures of kinases or pseudokinases (ROPKs) act as virulence factors that enhance parasite survival against host gamma interferon-stimulated innate immunity. However, the roles of these and other ROPK proteins in establishing chronic infection have not been tested. Here, we deleted 26 ROPK gene loci encoding 31 unique ROPK proteins of type II T. gondii and show that numerous ROPK proteins influence the development of chronic infection. Cyst burdens were increased in the Δrop16 knockout strain or moderately reduced in 11 ROPK knockout strains. In contrast, deletion of ROP5, ROP17, ROP18, ROP35, or ROP38/29/19 (ROP38, ROP29, and ROP19) severely reduced cyst burdens. Δrop5 and Δrop18 knockout strains were less resistant to host immunity-related GTPases (IRGs) and exhibited >100-fold-reduced virulence. ROP18 kinase activity and association with the parasitophorous vacuole membrane were necessary for resistance to host IRGs. The Δrop17 strain exhibited a >12-fold defect in virulence; however, virulence was not affected in the Δrop35 or Δrop38/29/19 strain. Resistance to host IRGs was not affected in the Δrop17, Δrop35, or Δrop38/29/19 strain. Collectively, these findings provide the first definitive evidence that the type II T. gondii ROPK proteome functions as virulence factors and facilitates additional mechanisms of host manipulation that are essential for chronic infection and transmission of T. gondii.

parasite survival by manipulation of host cells and innate immunity during acute infection (8,9), few secreted parasite molecules have been identified as playing crucial roles in successful chronic infection in vivo (10)(11)(12).
ROPs are secreted from T. gondii rhoptry organelles directly into host cells during host cell invasion. After their secretion into host cells, ROP5 and ROP18 tether to the cytosolic face of parasitophorous vacuole (PV) membrane (PVM) through N-terminal arginine-rich amphipathic helix (RAH) domains (19,(22)(23)(24). PVM-associated ROP18 phosphorylates host immunity-related GTPases (IRGs) to prevent their accumulation on the PVM and thereby preserve the integrity of the PV from destruction (19,25). Association of the ROP5 pseudokinase with ROP18 increases ROP18 kinase activity (26), and most likely the availability of host IRG substrates for their subsequent inactivation by the ROP18 kinase (27)(28)(29). ROP5 also associates in macromolecular complexes with ROP17, another PVM-associated kinase that phosphorylates and inactivates IRGs (22).
In laboratory strains of mice, South American and type I strains are virulent, type II strains exhibit reduced virulence, and type III strains are avirulent (30). While type I PVs avoid IRG accumulation (19), type II and type III PVs accumulate IRGs and are efficiently destroyed (31,32). To explain these virulence differences in archetypal clonal lineages, virulent ROP18 gene alleles were proposed to inhibit accumulation of IRG proteins on the PVs of strain types that also express a virulent ROP5 locus (27,28). This hypothesis explains increased virulence in type I strains that possess virulent ROP18 and ROP5 gene alleles. This hypothesis also explains avirulent type III strains that possess a virulent ROP5 locus and an avirulent and nonexpressed ROP18 gene allele. However, it is currently unclear how less virulent type II strains that carry a virulent ROP18 gene allele and an avirulent ROP5 gene locus resist host IFN-␥ to establish chronic infection.
In addition to ROP5, ROP17, and ROP18, many ROP proteins are active kinases or pseudokinases (ROPKs), and the ROPKs collectively are characterized as the rhoptry kinome. The T. gondii rhoptry kinome was originally identified to consist of a set of 34 unique ROPK proteins and 10 additional degenerate ROPK genes that contain large DNA insertions or deletions within their kinase domains (33). Subsequently, four additional T. gondii ROPK genes were identified as ROP47, ROP48, ROP49, and ROP50 (34). Targeted deletion of ROP47 or ROP48 in a type II strain does not affect cyst burdens or chronic infection in vivo (35). However, the roles of most ROPKs in virulence, cyst development, and chronic infection in less virulent type II strains have not been directly tested. Here, we developed gene deletions of 26 ROPK gene loci encoding 31 unique type II ROPK proteins and examined the ability of these mutants to establish chronic infection in mice. Our results show that while several type II ROPK molecules moderately affect cyst burdens, ROP5, ROP17, ROP18, ROP35, and ROP38/29/19 (ROP38, ROP29, and ROP19) are essential for chronic infection.
ROPK knockout strains with the most severe defects in chronic infection stage differentiate in vitro. One hypothesis is that ROPK knockout strains severely deficient in their ability to establish chronic infection may be defective in differentiation from the tachyzoite stage to the bradyzoite stage (4). To examine this, we used high pH to induce differentiation in vitro. Parental ⌬ku80 strain tachyzoites switched normally to the bradyzoite cyst stage based on expression of cytosolic green fluorescent protein (GFP) under control of the bradyzoite stage-specific LDH2 gene promoter (11) and on the development of the cyst wall structure that surrounds the bradyzoites as assessed by Dolichos biflorus agglutinin (DBA) lectin staining of the major cyst wall protein CST1 (12) (Fig. 2A). The ⌬rop5, ⌬rop17, ⌬rop18, ⌬rop35, and ⌬rop38/ 29/19 knockout strains exhibited no defect in their ability to differentiate to cyst stages in vitro ( Fig. 2A) or in their frequency of differentiation (data not shown). We next examined whether mice infected with high doses (Ͼ10 5 tachyzoites) of ⌬rop5, ⌬rop17, or ⌬rop18 tachyzoites could rescue normal cyst burdens. Cyst burdens remained extremely low even after highdose infection of mice with ⌬rop5, ⌬rop17, or ⌬rop18 tachyzoites (Fig. 2B). Therefore, it was not a lack of differentiation that hindered cyst development in these knockout strains.
The type II ME49 strain ROP5 gene locus is multiallelic and possesses~10 ROP5 gene alleles (16,17). Complementation of ⌬rop5 through expression of the ROP5A or ROP5C gene allele (see Fig. S2C in the supplemental material) failed to rescue cyst burdens (Fig. 3), even after high-dose infection (data not shown). To examine the ROP5 locus in the type II Pru strain, we PCR amplified the full coding region of ROP5A and ROP5C alleles from the ⌬ku80 parent strain and sequenced randomly selected clones. Eight ROP5C alleles were observed for each ROP5A allele (Fig. S3), suggesting that the Pru strain ROP5 locus is highly similar to the ME49 strain ROP5 locus (16,17).
The ⌬rop17 knockout strain exhibits mild egress and growth defects in vitro. ROPK knockout strains deficient in chronic infection may be compromised in their ability to replicate. To glob-ally assess ROPK knockout strains for growth defects, we first examined plaques representing zones of infection where repeated cycles of invasion, intracellular replication, and egress occur. While all other ROPK knockout strains and their complemented strains developed plaques that were similar in size to those of the ⌬ku80 parent (data not shown), the ⌬rop17 knockout strain developed smaller plaques (Fig. 4A). This small-plaque phenotype was rescued by expression of the ROP17 FLHA gene (Fig. 4A). However, the ⌬rop17 small-plaque phenotype was not associated with any significant defect in the replication rate (see Fig. S4A in the supplemental material) or in the number of parasites per PV at 21 h (data not shown) or 45 h postinfection (Fig. S4B). In addition, the ⌬rop17, ⌬rop5, and ⌬rop18 knockout strains infected host cells as efficiently as the ⌬ku80 parent (Fig. 4B). Consequently, we measured whether the ⌬rop17 strain exhibited any defect in the egress of replicated parasites from the PV. Tachyzoites egressed from a higher percentage of ⌬rop17 PVs after 45, 68, and 72 h of infection, and this early egress phenotype was rescued by complementation with ROP17 FLHA (Fig. 4C). However, the average numbers of ⌬rop17 parasites at secondary sites of infection after release from the PVs were not different from those of the ⌬ku80 parent or the ROP17 FLHA complemented strain (Fig. S4C), suggesting that this early egress phenotype did not immediately alter invasion or continued replication of ⌬rop17 parasites.

DISCUSSION
Toxoplasma gondii establishes a chronic infection of the host as an obligatory step for its biological transmission to new hosts (1). Herein, using the ⌬ku80 genetic background that enables efficient and reliable genetic dissection of parasite gene families (39,40), we deleted 26 ROPK gene loci encoding 31 known ROPK proteins and show for the first time that many type II ROPK proteins influence the ability of T. gondii to establish chronic infection. Moreover, several ROPKs, including ROP5, ROP17, ROP18, ROP35, and ROP38/29/19 were found to provide crucial influences on functions that were necessary to establish chronic infection.
The roles of ROP5, ROP17, or ROP18 molecules in establishing chronic infection of the host have not been assessed previously. Type II strains carry a virulent ROP18 gene allele and were previously proposed to behave in an avirulent fashion because type II strains also express avirulent ROP5 gene alleles (16,17,(26)(27)(28). We found that deletion of either type II ROP18 or ROP5 essentially abrogated chronic infection and markedly reduced virulence lethality by Ͼ100-fold. These results correlate with the significant defects found in the resistance of ⌬rop18 and ⌬rop5 PVs to IRG coating and killing, suggesting that ROP5 and ROP18 co- operate as type II virulence factors despite the absence of a highly virulent ROP5 gene allele. Most likely, it is the absence of ROP5 gene alleles that limits type II virulence in comparison to the more highly virulent type I and South American strain types (15)(16)(17). This interpretation is consistent with previous evidence showing that coexpression of the virulent type I ROP5 gene locus in the type II background increased virulence by~3 to 4 log units (28). Our study is the first to link the importance of type II ROP5 and ROP18 virulence functions to the development of cyst burdens and chronic infection.
Remarkably, nearly complete ablation of chronic infection in the ⌬rop5 and ⌬rop18 knockout strains was associated with only an~10% reduction in parasite survival in IFN-␥-stimulated MEFs in vitro. Therefore, our results do not exclude the possibility that in addition to IRG resistance mechanisms, non-IRG functions of ROP5 and ROP18 complexes could also influence chronic infection. The N-terminal ROP18 RAH2 domain establishes PVM association, which is required for both IRG resistance and virulence but not for intrinsic ROP18 kinase activity (23,24). The RAH2 domain also mediates ROP18 association with host ATF6␤, and this interaction triggers the proteasome-dependent degradation of host ATF6␤, and correspondingly, compromises adaptive immune responses mediated by CD8 ϩ T cells (38). However, this specific ROP18-ATF6␤ interaction has only been previously demonstrated using recombinant ROP18 molecules that were not PVM associated (38). The multifunctional ROP18 molecule also associates with the p65 subunit of NF-B to mediate its degradation (41). A second ROP18 ligand-binding pocket that influences virulence is also present in the ROP18 protein structure (42). Moreover, additional parasite proteins were recently identified in ROP5 and ROP18 complexes, suggesting there is unexpected biological complexity to their regulation and PVM-associated functions (22). Thus, future studies are necessary to decipher whether ROP5 and ROP18 IRG-independent mechanisms influence chronic infection.
Our findings establish that ROP17 is a key molecule that determines cyst development in type II strains. However, IRG resistance in the ⌬rop17 knockout strain was unaffected in IFN-␥-stimulated MEFs, and the ⌬rop17 strain exhibited only a low-virulence defect in comparison to deletion of ROP5 or ROP18. Previously, the deletion of ROP17 in the virulent type I RH strain was shown to reduce virulence, to increase Irgb6 coating of PVs, and to increase clearance of PVs in IFN-␥-stimulated macrophages (22). Together, these findings are consistent with previously reported evidence showing that the PVM-associated ROP17 kinase is polymorphic and under positive selective pressure in the virulent type I strain, which possesses 24-amino-acid differences compared with type II or type III ROP17 (22). Collectively, the data suggest that type II ROP17 is likely to play another role important for establishing chronic infection that is independent of the mild ROP17 virulence defect. The early egress phenotype of the ⌬rop17 knockout strain was not linked to an immediate defect in parasite invasion or replication, yet 13-day-old ⌬rop17 plaques were significantly reduced in their size. It is possible that ROP17 may function as a signaling molecule that influences the receptiveness of host cells to parasitism, the ability of the parasite to gain access to the central nervous system, or the development or maintenance of cyst burdens in the central nervous system. Recent elegant work based on rhoptry organelle secretion of Cre recombinase into host cells in contact with or invaded by T. gondii (43,   44) has demonstrated that ROP proteins are primarily injected into neurons within the brain (5). Another possibility is that the early egress phenotype of ⌬rop17 parasites may reduce the frequency of the overall success rate for tachyzoite differentiation to encysted bradyzoites in vivo. Alternatively, loss of ROP17 or other PVM-associated ROPK proteins may increase the susceptibility of the cyst wall to degradation by host chitinase, which is produced by alternatively activated macrophages in chronically infected mice (45). Future studies are necessary to more clearly identify the mechanisms that determine the key role of ROP17 in establishing chronic infection.
Severe reductions in cyst burdens were observed in the ⌬rop35 and ⌬rop38/29/19 knockout strains in the absence of any virulence defect in vivo or any defect in resistance to IRGs in vitro. These findings are intriguing in view of previous microarray evidence that defines ROP35 and ROP38 as the only ROPK genes that are significantly upregulated in less-virulent strain types (33). In particular, ROP38 transcripts are found to be more abundant in bradyzoite stages (increased Ͼ16-fold) compared to tachyzoite stages (33). Consequently, bradyzoite stage-specific expression of ROP38 may be important for the transition to or the maintenance of chronic stages in vivo. Because type II ⌬rop35 and ⌬rop39/29/19 knockout strains show severe impairments in cyst formation without changes in virulence, future studies may elucidate novel mechanisms that influence chronic infection.
ROP35 is expressed in Toxoplasma, Neospora, Eimeria, and Sarcocystis and maps to a basal ROPK cluster designated ROPKL (34). ROP38-related genes are expanded in both Toxoplasma and cyst-forming Neospora coccidian parasites (33). T. gondii ROP38 strongly influences the expression of a large number of host genes involved in mitogen-activated protein kinase (MAPK) signaling cascades, apoptosis, and host cell proliferation (33). ROP16 is another ROPK that strongly influences the expression of a large number of host genes (8,46,47), and inhibits the responsiveness of its host cell to IFN-␥ signaling (46). Knockdown of the host STAT1 transcriptional coactivator tailless (TLX) was recently shown to repress a subset of IFN-␥-activated genes in the brain, and mice lacking expression of TLX in the brain were found to be impaired in their ability to control chronic infection (48). These observations suggest that ROPK molecules, such as ROP38, ROP35, and ROP16, could influence chronic infection through regulation of host response genes. Our findings provide the first definitive evidence showing that the type II T. gondii rhoptry kinome provides anti-IRG virulence functions in acute infection as well as additional mechanisms of host manipulation necessary to establish chronic infection.

MATERIALS AND METHODS
Ethics statement. All mouse work was conducted in accordance with the recommendations in the Guide to the Care and Use of Laboratory Animals (49) and Association for the Assessment and Accreditation of Laboratory Animal Care guidelines. The animal protocol was approved by the Dartmouth College Committee on the Use and Care of Animals (Animal Welfare Assurance A3259-01, protocol bzik.dj2). All efforts were made to minimize suffering and pain.
Mice and generation of bone marrow-derived macrophages. Female 7-to 9-week-old C57BL/6 mice were purchased from Jackson Laboratories (Bar Harbor, ME) and were maintained at the Center for Comparative Medicine and Research at the Geisel School of Medicine at Dartmouth. Bone marrow-derived macrophages were isolated from bone marrow from C57BL/6 mice by differentiation in Dulbecco's modified essential medium (DMEM) supplemented with 10% FBS, 1ϫ minimal essential medium nonessential amino acids, 1 mM sodium pyruvate (Life Technologies), antibiotics, and 30% L929 culture supernatant as previously described (52). Bone marrow-derived macrophages were harvested after 5 days of differentiation.
Chronic infection. Freshly lysed high-viability type II tachyzoites were obtained as previously described (11,37). Tachyzoites were centrifuged at 900 ϫ g for 7 min, washed, and counted in Dulbecco's modified phosphate-buffered saline (DPBS). Mice were infected by intraperitoneal (i.p.) injection, and parasite viability was determined in a plaque assay at the time of mouse infection. To induce chronic infection, mice were infected i.p. with 2 ϫ 10 2 tachyzoites of the tested type II strain. Higher parasite doses (2 ϫ 10 5 or 2 ϫ 10 6 ) were used in certain experiments as indicated.
Tissue cyst burden assay. The brains from mice infected with type II strains were harvested at 3 weeks postinfection (11). The brains were homogenized in 2-ml volumes of sterile DPBS using a Dounce homogenizer. Cyst counts were performed on a minimum of 10% of each brain. Because Pru background cysts vary greatly in size and the brains of chronically infected mice contain many small, medium, and large cysts (11,53), Type II parasites deficient in ROP5 and ROP18 are less virulent than parasites deficient in ROP17, ROP35, or ROP38/29/19. The ⌬rop5, ⌬rop17, ⌬rop18, ⌬rop35, and ⌬rop38/29/19 mutants and complemented strains were evaluated for virulence lethality in mice. (A) C57BL/6 mice were infected i.p. with 2 ϫ 10 5 tachyzoites of the ⌬rop5, ⌬rop17, or ⌬rop18 knockout strain or the ⌬ku80 parent strain. Data shown are combined from two independent experiments, each with four mice per group. The P values were calculated by a log rank Mantel-Cox test, and a P of Ͻ0.05 was considered significant. (B) C57BL/6 IFN-␥ Ϫ/Ϫ knockout mice were infected i.p. with 2 ϫ 10 2 tachyzoites of the ⌬rop5, ⌬rop17, or ⌬rop18 strain or the ⌬ku80 parental strain. Data shown are from a single experiment with four mice per group. The P values were calculated by a log rank Mantel-Cox test, and a P of Ͻ0.05 was considered significant. ns, not significant. (C) C57BL/6 mice were infected i.p. with 2 ϫ 10 6 tachyzoites of the ⌬rop5, ⌬rop17, or ⌬rop18 strain or the ⌬ku80 parental strain or with 2 ϫ 10 6 tachyzoites of the ⌬rop5::ROP5A or ⌬rop5::ROP5C [shown together as ⌬rop5::ROP5A/C], ⌬rop17::ROP17 FLHA , ⌬rop18::ROP18, ⌬rop18::ROP18 KD , or ⌬rop18::ROP18 RAH2(ATF) complemented strain. Data shown are combined from two independent experiments, each with four mice per group. The P values were calculated by a log rank Mantel-Cox test, and a P of Ͻ0.05 was considered significant. (D) C57BL/6 mice were infected i.p. with 2 ϫ 10 7 tachyzoites of the ⌬rop5, ⌬rop17, or ⌬rop18 strain or the ⌬ku80 parental strain. Data shown are combined from two independent experiments, each with four mice per group. The P values were calculated by a log rank Mantel-Cox test, and a P of Ͻ0.05 was considered significant. (E) C57BL/6 mice were infected i.p. with 2 ϫ 10 5 tachyzoites of the ⌬rop35 or ⌬rop38/29/19 strain or the ⌬ku80 parent strain. Data shown are combined from two independent experiments, each with four mice per group. cysts were scored using dark-field microscopy with an inverted fluorescence phase-contrast microscope (Olympus CKX41) to reliably identify all cysts. The type II ⌬ku80 strain expresses green fluorescent protein (GFP) under control of the bradyzoite stage-specific LDH2 promoter (11). GFP-positive (GFP ϩ ) cysts were scored at a total magnification of ϫ150 that provided the highest sensitivity for the detection of GFP ϩ bradyzoites within cysts and to then also verify the presence of a translucent thick cyst wall in bright-field microscopy (11,37).
Generation of ROPK knockout strains. Deletion of ROP kinome gene loci was performed using the KU80 knockout strain of the type II Pru⌬hxgprt strain as previously described (11,37). Briefly, gene locus knockout targeting plasmids were assembled in yeast shuttle vectors pRS416 or pRS426 using yeast recombinational cloning to fuse three distinct PCR products with 31-to 34-bp crossovers in order; a 5= GOI (gene of interest) target gene flank, the HXGPRT selectable marker, and a 3= GOI target flank (see Fig. S1A in the supplemental material) (36). Knockout plasmids were engineered to delete at least 200 nucleotides of the 5= untranslated region (5= UTR) and the complete coding region of the ROPK gene locus as defined in the ToxoDB.org database (54). All primers used to construct knockout targeting plasmids and nucleotide definition of ROPK gene loci deletions are listed in Table S2 in the supplemental material. After plasmid validation by DNA sequencing, targeting plasmids were linearized at restriction sites inserted at either the 5= end of the 5=-targeting flank or at the 3= end of the 3=-targeting flank. Linearized targeting plasmids were transfected by electroporation into tachyzoites of the Pru⌬ku80⌬hxgprt strain and ROPK knockouts were selected in 50 g/ml mycophenolic acid and 50 g/ml xanthine, and parasites were cloned by limiting dilution 30 days after transfection. ROPK knockouts were validated by genotype analysis using PCR to measure the following (shown in Fig. S1A): (i) targeted deletion of the coding region of the targeted gene (DF and DR primers) in PCR 1, (ii) correct targeted 5= integration (CXF and 5=DHFRCXR primers) in PCR 2, and (iii) correct targeted 3= integration (3=DHFRCXF and CXR primers) in PCR 3 using knockout validation primers shown in Table S3.
Complementation of ROPK knockouts. Complementation plasmids were designed to complement ROPK knockout strains Pru⌬ku80⌬ropGOI through chromosomal integration and expression of wild-type or mutant gene alleles at the uracil phosphoribosyltransferase (UPRT) chromosomal locus (TGME49_312480) as previously described (11) (see Fig. S1B in the supplemental material). Complementation plasmids were developed in the pRS416 or pRS426 yeast shuttle vector using yeast recombination to fuse, in order, a 5= UPRT target flank, the complementing gene of interest with native 5= UTR, and the 3= UPRT target flank (Fig. S1B). Oligonucleotide DNA primers (Table S4) were used to generate the complementing genes, synthesized on one or two PCR products. ROP18 mutant gene alleles possessing a kinase-dead (KD) domain or deleted RAH2 domain [RAH2(ATF)] were engineered as previously described (19,38). Following plasmid assembly by yeast recombinational cloning, targeting plasmids were validated by DNA sequencing. Prior to transfection, plasmids were linearized via the unique restriction site PmeI. The parasites were cultured for 2 days in normal infection media, and the cultures were then switched to selection medium containing 2 M 5-fluorodeoxyuridine (FUDR) and cloned 30 days after transfection by limiting dilution. Targeting of complementing genes to the UPRT locus was validated by genotype analysis using PCR assays (strategy shown in Fig. S1B) to measure the following: (i) deletion of UPRT coding region in PCR 4, (ii) correct targeted 5= integration in PCR 5, and (iii) correct targeted 3= integration of the complementing transgene at the UPRT locus in PCR 6 using oligonucleotide DNA validation primers (Table S5).
Characterization of ROP5A and ROP5C gene alleles. Genomic DNA isolated from the ⌬ku80 parent strain was used as the template to PCR amplify the complete coding region of ROP5A and ROP5C gene alleles using primers ATGGCGACGAAGCTCGCTAGAC (forward) and GAGC CGTTTTCTCAAAGCGACTGAG (reverse). PCR products were ran-domly cloned into the pCR4 Topo vector (Invitrogen), and individual clones were sequenced. All sequences were managed with MacVector and aligned using ClustalW.
Intracellular replication rate assay. Parasite growth rate was determined using methods described previously employing a direct parasite per vacuole scoring approach (11). Briefly, triplicate monolayers of HFF cells were infected at a multiplicity of infection (MOI) of~0.2. and parasites were allowed to invade cells for 1 h. Monolayers were then washed three times in DPBS to remove extracellular parasites. At 21 h and 45 h postinfection, tachyzoites per vacuole were scored in at least 50 randomly encountered vacuoles.
Plaque assay. Confluent monolayers of HFF cells were infected with 200 tachyzoites of a type II strain in infection medium, and cultures were left undisturbed for 13 days to allow localized development of PFU. The medium in the cultures was removed, and the cultures were fixed in 50% methanol and 10% acetic acid and stained with 0.25% Coomassie brilliant blue for 2 days. Cultures were rinsed in water, air dried, and photographed (55).
Infection rate assay. Parasite infection rate was determined in PFU assays. Briefly, multiple monolayers of HFF cells in 25-cm 2 flasks were infected with 500 tachyzoites. At 30 and 60 min postinfection, the monolayers were washed three times in DPBS to remove extracellular parasites, and either infection medium was replaced, or parasites were not removed from continuously incubated control flask. The cultures were incubated continuously for 11 additional days to establish PFU. Percent infection rate was determined as the number of PFU present at each time divided by the number of PFU in the continuously incubated cultures.
Early egress assay. Tachyzoites in primary type II Pru⌬ku80 PVs egress at~72 h postinfection (37). HFF cells were infected at an MOI of 0.01, and 1 h after infection, the monolayers were washed three times in DPBS to remove extracellular parasites, and infection medium was replaced. At least 500 infection sites were scored at 21, 45, 68, and 72 h postinfection to measure the percentage of intact primary PVs or secondary sites of infection sites (after parasite egress from the PV). The number of tachyzoites present in secondary infection sites was also measured at 45 h and 72 h postinfection by scoring at least 100 secondary sites of infection.
PV killing assay. C57BL/6 MEFs (ATCC) were seeded into 24-well trays and were cultured in DMEM in 15% FBS. MEF monolayers were incubated without IFN-␥, or MEF monolayers were stimulated with 200 U/ml IFN-␥ (Peprotech) 24 h prior to parasite infection to activate host immunity-related GTPases and cell autonomous killing mechanisms. Triplicate wells of MEFs were infected with 200 or 1,000 tachyzoites, and plaques were allowed to develop for 5 to 7 days. The total number of PFU per well was counted microscopically, and the percentage of PFU survival was calculated as the number of PFU in IFN-␥-stimulated MEFs divided by the number of PFU scored in nonstimulated MEFs (no IFN-␥ treatment).
Immunofluorescence assay. HFF cells were cultured on circular microcover glasses and infected with parasites for~16 to 30 h. To visualize ROP5, the cultures were fixed in 4% paraformaldehyde and permeabilized with 0.01% Triton X-100. To visualize ROP17 and ROP18, the cultures were fixed with Histochoice (Amresco) and permeabilized in 0.1% saponin (Sigma) for 10 min. All samples were blocked with 10% FBS and incubated with primary antibodies for 1 h at room temperature (RT). The primary antibodies were rabbit anti-ROP5 (MO556; 1:3,000 dilution) (26), rabbit anti-ROP18 (ROP18-His WA525; 1:500 dilution; Sibley Laboratory), and anti-HA to visualize ROP17 FLHA (rabbit monoclonal anti-HA tag antibodies; 1:500 dilution; Cell Signaling). Preparations were washed three times with PBS and incubated 1 h at RT with a 1:1,000 dilution of secondary goat anti-rabbit IgG antibodies conjugated to Alexa Fluor 488. Samples were mounted in SlowFade gold with DAPI (Life Technologies).
In vitro cyst differentiation assay. Tachyzoites were differentiated in vitro into bradyzoites within cysts essentially as previously and elegantly described by Tobin and colleagues (58). Differentiation media contained Roswell Park Memorial Institute medium (RPMI) without bicarbonate supplemented with 2.05 mM L-glutamine (Hyclone), 20 mM HEPES-free acid (IBI Scientific), 1% XL-glutamine (a long-lasting stable form of glutamine; VWR), 1% FBS, and 1% penicillin-streptomycin. The pH of differentiation medium was adjusted to 8.1 with sodium hydroxide and filter sterilized. HFF cells were cultured on circular microcover glass, and confluent monolayers were infected with type II parasites at an MOI of~0.5. Three hours after infection, the infected cells were washed once in DPBS supplemented with Ca 2ϩ and Mg 2ϩ , and cultures were incubated in differentiation media for 3 days at 37°C in ambient air. Infected cells were fixed in 4% paraformaldehyde, and the excess was quenched with 0.1 M glycine. All samples were permeabilized and blocked in 3% FBS plus 0.2% Triton X-100 for 30 min at room temperature, and then they were incubated with a 1:250 dilution of rhodamine-labeled Dolichos biflorus agglutinin (Vector Laboratories) for 1 h at RT. The preparations were washed three times with DPBS, mounted in SlowFade gold antifade with DAPI (Life Technologies), and imaged by confocal microscopy.

ACKNOWLEDGMENTS
We are grateful to developers of the ToxoDB.org Genome Resource. Tox-oDB and EuPathDB are part of the National Institutes of Health/National Institutes of Allergy and Infectious Diseases (NIH/NIAID)-funded Bioinformatics Resource Center. We thank David Sibley for providing antibodies specific to ROP5 and ROP18. This work was sponsored by grants AI041930, AI075931, AI084570, AI097018, and AI104514 from the National Institutes of Health (NIH). L.M.R. was a trainee on NIH training grants 5T32AI007363 and 2T32AI007519. R.B.G. was a trainee on NIH training grant 2T32AI007519.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.