Parasite stage-specific recognition of endogenous Toxoplasma gondii derived CD8⁺ T cell epitopes

Abstract

Background

BALB/c mice control infection with the obligate intracellular parasite Toxoplasma gondii and develop a latent chronic infection in the brain, as do immunocompetent humans. IFN-γ producing CD8⁺ T cells provide essential protection against T. gondii, but the epitopes recognized have so far remained elusive.

Methods

We employed caged MHC molecules to generate ~ 250 H-2L^d tetramers and distinguish T. gondii-specific CD8⁺ T cells in BALB/c mice.

Results

We identify two T. gondii specific H-2L^d-restricted T cell epitopes, one from dense granule protein GRA4 and the other from rhoptry protein ROP7. H-2L^d/GRA4 reactive T cells from multiple organ sources predominate 2 weeks after infection, while the reactivity of the H-2L^d/ROP7 T cells peaks 6–8 weeks after infection. BALB/c animals infected with T. gondii mutants defective in establishing a chronic infection show altered levels of antigen-specific T cells, depending on the T. gondii mutant used.

Conclusions

Our results shed light on the identity and the parasite stage-specificity of two CD8⁺ T cell epitopes recognized in the acute and chronic phase of infection with T. gondii.

Introduction

Toxoplasma gondii is an obligate intracellular parasite infecting all homeothermic vertebrate hosts, with human infection rates of 20–90%. As the causative agent of toxoplasmic encephalitis, the parasite poses a severe health threat for immunocompromised individuals, especially AIDS patients, and causes congenital defects in newborns [1]. No human vaccines or drugs that eradicate the infection are available.

In humans and rodents, T. gondii exists as the rapidly multiplying lytic tachyzoites, which later during infection convert into slower growing bradyzoites, harbored in cysts of neural and muscular tissue [2]. Sequential secretion through three organelles, micronemes, rhoptries, and dense granules from tachyzoites mediates invasion and survival inside the host cell [3]. In Europe and North-America the prevalent T. gondii strains are divided into 3 types, with Type II strains covering over 70% of isolates obtained from human toxoplasmosis [4].

Like immunocompetent humans [5], BALB/c mice limit Type II and III T. gondii with fewer cysts in their brains as compared to susceptible strains [6], and develop a latent chronic infection. This is ascribed to the generation of H-2L^d-restricted CTLs and dependent on Type II parasites [7]. When the immune response wanes, the parasite may recrudesce from the bradyzoite to the tachyzoite stage, which can invade virtually any nucleated cell. By replicating unchecked, T. gondii can cause fatal toxoplasmic encephalitis. The generation of IFN-γ by innate NK cells and by CD4⁺ and CD8⁺ T lymphocytes is central to host resistance [8–10], however, no T. gondii derived CD8⁺ T cell epitopes have been reported to date. Published reports of T. gondii specific CD8⁺ T cell responses are based on studies where animals were vaccinated mostly with major parasite protein, the surface antigens (SAG1-3). The resulting CD8⁺ T cells responded to either parasite lysate or to peptide restimulation in vitro ([11, 12, 13, 14] and references therein). Whether these potential SAG epitopes are generated in the course of a natural infection is not known.

Problems inherent in the identification of parasite-derived T cell epitopes are the organisms’ large and complex genomes, their multi-faceted life cycles and their persistence in the host despite the presence of protective immunity. We used the caged MHC-tetramer technology [15–17] to generate an array of approximately 250 H-2L^d tetramers of defined specificity to screen for T. gondii-specific T cell epitopes in infected BALB/c mice. We identified CD8⁺ T cell epitopes derived from two distinct parasite proteins, GRA4 and ROP7. The GRA4-specific T cells are detected during the acute phase of the infection, while the T cells reactive to the ROP7 peptide persist during the chronic phase. Both types of T cells secrete protective IFN-γ. Mice infected with mutant parasites defective in establishing a chronic infection exhibit altered levels of the two epitope-reactive T cells throughout the course of infection, consistent with the ability of bradyzoites to sustain a ROP7 reactive CD8⁺ T cell response. The identification of endogenous T. gondii-derived epitopes, as distinct from the use of engineered T. gondii expressing model antigens, affords new opportunities for dissecting the immune response against the parasite.

Materials and Methods

Antibodies

J.F. Dubremetz the provided ROP7 antibody T₄3H₁ [18], D. Sibley the GRA4 antibody [19], J. Saeij the SAG2Y antibody [20] and J. Gaertig the 12G10 anti-tubulin antibody.

Parasite Strains

T. gondii Pru ΔHXGPRT and ME49 tachyzoites were propagated in human foreskin fibroblast monolayers grown in DMEM containing 10% FCS and penicillin/streptomycin. The 73F9 and 9xG4 mutants were previously isolated [21]. The mutant 76-E2 is described in the supplemental text.

Experimental animals and T. gondii infection

All animal protocols were approved by the MIT Committee on Animal Care. Swiss-Webster and BALB/c mice were from Taconic. 40 cysts of the T. gondii ME49 strain (a gift from G. Yap) were isolated from the brain homogenate of an infected Swiss-Webster animal and injected intraperitoneally (i.p.) into BALB/c animals for future analysis of the T cell response. Alternatively, BALB/c mice were infected with 5000 Prugniaud tachyzoites or 2×10⁶ γ-irradiated Prugniaud tachyzoites (15 kRad).

Composition of the Toxoplasma secretome gene list

For the generation of the list of 73 T. gondii target proteins see Supplemental Text.

H-2L^d Epitope Prediction

The 73 selected ORFs (Table S1) were analyzed by BIMAS [22], RANKPEP [23], SYFPEITHY [24] and predicted 9 amino acid residue epitopes that scored higher then 150, 92 and 21 respectively, were incorporated into the screen. Double hits were removed, to give 246 unique nonameric sequences.

Peptide synthesis

Conditional ligands p29-P6* to p29-P8* were constructed manually, using Fmoc-based solid-phase peptide synthesis. The MIT Center for Cancer Research (Cambridge, MA) biopolymers facility synthesized the peptides used for screening.

Protein Expression and Purification

Recombinant expression of murine β₂m, as well as the luminal portion of the H-2L^d heavy chain with a C-terminal BirA recognition sequence (plasmid gift of J. D. Altman), was accomplished following established protocols. Refolding of the MHC complex with conditional ligands p29-P6* to p29-P8* was followed by biotinylation, size-exclusion chromatography (S75) [25], and were stored at −80°C. Class I MHC tetramers [26] were produced by addition of Streptavidin-PE (Invitrogen) to monomer at a final molar ratio of 4:1, respectively, and peptide exchange was effected by irradiation at 365 nm (Stratalinker 2400 UV) as described [16].

Cell preparation and tetramer staining

The peritoneal cavity of BALB/c mice was lavaged and splenocytes and mononucleated cells from the brain were prepared. For screening, CD8⁺ T cells were isolated with a Milteny Biotech kit. To purify brain T cells, the mice were perfused intracardially with PBS, the brain homogenized and passed over a 35% Percoll solution followed by a 70%/35% Percoll gradient. Cell suspensions were treated with ethidium monoazide under exclusion of light, washed and irradiated with incandescent light. The cells were incubated for 45 min in 96-well plates (~1 × 10⁵ cells at 50 μL/well) with freshly prepared H-2L^d tetramer and FITC-conjugated anti-CD8 mAb (Becton Dickinson) followed by fixing with 4% formaldehyde and analyzed by flow cytometry.

Cyst purification

Brains from ME49-infected Swiss Webster mice were homogenized in 1% Tween 20 in PBS. The homogenate was passed over a 90%/30% Percoll gradient and centrifuged. The 90%/30% interface and the half of the 30% layer were collected and repeatedly washed with PBS.

In vitro intracellular IFN-γdetection

Splenocytes from Pru or ME49-infected BALB/c mice were seeded at 4×10⁶ cells per well and restimulated overnight with 10 μg/mL of peptide. Cells were treated for 4 hours with 10 μg/mL Brefeldin A, then labeled with tetramer and anti-CD8 mAb as described above and stained with anti-IFNγ mAb using the BD Cytofix/Cytoperm kit (Becton Dickinson).

Western Blot Analysis

ME49 tachyzoites, and bradyzoites harvested from mouse brain were lysed in PBS by freeze-thawing. 0.5, 2, 5 and 10 μg of the tachyzoite and bradyzoite lysates were analyzed by SDS-PAGE. Immunoblot analysis was performed with ROP7-, GRA4-, Sag2Y- or tubulin-antibodies at 1: 1000, 1:2000, 1:1000 dilution, respectively.

Results

Design and validation of conditional ligands for H-2L^d Class I MHC tetramers

MHC tetramers enable the direct visualization of antigen-specific T cells [26], and arrays of MHC tetramers can be produced rapidly using caged MHC complexes. This technology is based on transient occupation of Class I [15–17] or Class II [27] MHC molecules with a conditional ligand. Here, we apply this strategy to the identification of parasite-specific CD8⁺ T cell epitopes, by generating ~250 distinct H-2L^d MHC tetramers. We designed three photocleavable ligands based on the p29 peptide, which conforms to the H-2L^d consensus binding motif [28]. Following inspection of the crystal structure of H-2L^d in complex with p29 ([29] and Fig 1A) we surmised that the Ile, His and Asn amino acid residues at positions P6, P7 and P8, respectively, could be replaced with the photocleavable 3-amino-3-(2-nitro)phenyl-propanoic acid (Anp) residue. The resulting p29-P6*, p29-P7*, and p29-P8* ligands were synthesized and used for the production of caged H-2L^d tetramers.

Fig. 1. Design and application of conditional ligands for H-2L^d Class I MHC molecules.

(A) Rendering of the p29 epitope (orange) in complex with H-2L^d (omitted for clarity, PDB-ID 1LD9). Individual replacement of the P6 to P8 residues with the synthetic Anp-residue produced photocleavable derivatives p29-P6*, p29-P7* and p29-P8*, respectively. (B) Class I MHC tetramers composed of H-2K^b/SV9-P7* [16], or H-2L^d preloaded with either p29-P6*, p29-P7* or p29-P8* were UV-irradiated in the presence or absence of the peptides SIYRYYGL (SIY), SIINFEKL (SII), IPAAAGRFF (ROP7), QLSPFPFDL (QL9) as indicated, and used to stain OT-1 or 2C TCR transgenic cells. Functional staining reagents were obtained only in case of the correct pMHC-TCR combinations.

To validate the peptide exchange reaction, we used 2C TCR transgenic cells that recognize the H-2L^d/QL9 complex [30]. Only after UV irradiation in the presence of QL9 did we observe successful peptide exchange for the three H-2L^d tetramers preloaded with p29-based ligands as visualized by surface staining (Fig. 1B). Control staining, using an irrelevant H-2L^d ligand IPAAAGRFF (see below), established that staining of 2C T cells was strictly peptide-specific. Secondly, we applied these conditions to H-2K^b complexes carying photocleavable SV9-P7* [16]. Here, UV-induced peptide exchange yielded reagents that stain both OT-1 and 2C TCR transgenic cells when provided with the respective SIINFEKL and SIYRYYGL peptide, consistent with the ability of the 2C T cell clone to recognize both allogeneic (H-2L^d) and syngeneic (H-2K^b) peptide-MHC complexes [30]. The p29-P6*, p29-P7*, and p29-P8* photolabile peptides can thus be used as conditional ligands for H-2L^d to rapidly generate MHC tetramer arrays of defined specificity in a single step without loss of functional integrity. We arbitrarily chose to use H-2L^d/p29-P8* for screening purposes.

Screening for CD8⁺ T cell epitopes from Toxoplasma gondii

A Type II Pru line engineered to express the model H-2L^d antigen β-galactosidase, elicits a specific CD8⁺ T cell response that peaks 3 weeks post-infection [31]. We therefore mined the T. gondii database for tachyzoite-stage secreted proteins to compile a partial list of antigens that could be a source of T. gondii CD8⁺ T cell epitopes (Fig. 2A, B and Table S1). Three web-based predictive algorithms were used to analyze 73 selected proteins for the presence of candidate H-2L^d restricted nonameric epitopes. This gave 246 unique candidate sequences (Fig. 2B and Table S2) out of 33.525 possible nonameric peptides in this protein data set, which we then used to generate distinct H-2L^d tetramers from.

Fig. 2. Screening for endogenous Toxoplasma gondii-derived CD8⁺ T cell epitopes with caged MHC-tetramers.

(A) Schematic depiction of a T. gondii tachyzoite with the relevant organelles highlighted. (B) Origin and numbers of proteins, as well as the candidate 9-mer epitopes embedded therein, that were selected for MHC tetramer screening. Redundancy in the selected proteome generates a surplus of 263 epitopes from 246 unique peptide sequences (see also Table S2). (C) Staining of CD8⁺ T cell-enriched splenocytes with H-2L^d/SPMNGGYYM (GRA4) on day 14 p.i. or H-2L^d/IPAAAGRFF (ROP7) 42 days p.i. reveals the antigen-specific CD8⁺ T cell subpopulations after infection with T. gondii. These reagents were obtained separately after photocleavage and peptide-exchange on the caged H-2L^d/p29-P8* tetramer complex.

Splenic CD8⁺ T cells from BALB/c mice were collected 10 and 21 days after i.p. injection of 40 T. gondii ME49 cysts and stained with freshly generated H-2L^d tetramers. We found two different H-2L^d-restricted antigens in these independent screens; the peptide SPMNGGYYM from GRA4, and IPAAAGRFF derived from ROP7 (Fig. 2C).

Kinetics of the H-2L^d-tetramer positive T cells throughout the infection

We found different frequencies and distribution for the GRA4- and ROP7-specific CD8⁺ T cell response. Infected BALB/c mice exhibited GRA4-reactive T cells as early as 10 days p.i. (Fig. 2C and 3A) while the ROP7 CD8⁺ T cell response peaks in week 6 post-infection (Fig. 3A). For both epitopes the levels of tetramer-positive CD8⁺ T cells were 2 to 3 times higher in the peritoneal cavity (PEC) and brain than in the spleen. The parasite had converted to the bradyzoite stage since we could detect cysts in fixed brain sections from a BALB/c mouse 4 weeks p.i. (data not shown). Moreover, both epitopes were not only generated in BALB/c mice infected i.p., but also in animals orally infected with 5 ME49 cysts as evident from the presence of the corresponding tetramer positive CD8⁺ T cells (data not shown).

Fig. 3. The cellular immune responses to the GRA4- and ROP7-derived epitopes are kinetically distinct.

(A) Cell surface staining with the appropriate Class I MHC tetramer plotted as % tetramer-positive of the total CD8⁺ cells versus the time point post-infection shows maximal expansion of GRA4-specific CD8⁺ lymphocytes around 14 days. The CD8⁺ T cells recognizing the ROP7 epitope peak between 4–6 weeks post-infection. These phenomenon are generally comparable over the different tissues sampled; spleen, peritoneal cavity (PEC) and brain. Data shown are representative of ≥3 independent experiments. (B) To compare CD69 staining, cells were gated on the CD8⁺ population (infection) or the H-2L^d/GRA4 or H-2L^d/ROP7 specific population (+infection, 2 and 4 weeks p.i. respectively). The percentage of cells that are CD69-high, as specified per plot, is increased for the antigen-specific population. Pre-gating on CD8⁺ splenocytes, both populations of H-2L^d/GRA4 (2 weeks p.i.) and H-2L^d/ROP7 (4 weeks p.i.) specific T cells are high for the surface marker CD44. The percentage of antigen-specific cells is indicated in the lower quadrant. (C) In vitro stimulation of splenocytes from T.gondii-infected BALB/c mice with the addition of either no peptide (−), enterotoxin (ET), SPMNGGYYM (GRA4) or IPAAAGRFF (ROP7) epitope, or the known H-2L^d-binding peptide QLSPFPFDL (QL9), demonstrates that both H-2L^d/GRA4 and H-2L^d/ROP7 CD8⁺ T cells produce IFN-γ in an epitope-specific fashion. The gating for total CD8⁺ T cells excluded the tetramer-positive population.

The T. gondii-specific CD8⁺ T cells displayed upregulated early activation marker CD69 (Fig. 3B) as compared to uninfected CD8⁺ T cells populations, as well as tetramer-negative CD8⁺ T cells (data not shown). Surface expression of CD44 (Fig. 3B) was high, relative to the tetramer-negative CD8⁺ splenocytes, and persisted throughout the later stages of infection, consistent with their in vivo activation state. Upon restimulation in vitro, both the ROP7- and GRA4-specific CD8⁺ T cells produce IFN-γ in response to the corresponding peptide (Fig. 3C). Moreover, tetramer-reactive CD8⁺ T cells from chronically infected animals (week 8 post-infection) display renewed expression of CD69 on their surface after 12 hours of in vivo restimulation with 10⁵ T. gondii Pru tachyzoites (data not shown).

Parasites defective in establishing a chronic infection show altered levels of H-2L^d/ROP7 positive CD8⁺ T cells

To dissect the origin of the GRA4 and ROP7 antigens in the course of T. gondii infection, we used T.gondii Pru mutants defective in establishing a chronic infection ([21] and Supplemental Text). BALB/c mice infected with mutant versus wild-type parasites exhibited a CD8⁺ T cell response of altered magnitude at 4 and 8 weeks post-infection in the brain (Fig. 4A). Animals infected with the Pru mutant 76-E2 that produced fewer and smaller brain cysts (see Supplemental Text and Fig. S2A) exhibit the most H-2L^d/GRA4-specific CD8⁺ T cells at 4 weeks p.i. Interestingly, the ROP7-CD8⁺ T cell response was sharply reduced for this 76-E2 mutant compared to wild-type Pru at 8 weeks post-infection, showing the importance of establishing a chronic infection for development of a T cell response against ROP7. The ROP7-CD8⁺ T cell response peaked early at 4 weeks after infection with the Pru mutant 73F9 that is defective in nuclear trafficking [21]. Even though the 73F9 mutant has an approximately 200-fold reduction in the number of cysts in the brains of mice [21], microarray analysis shows that 73F9 is developing into bradyzoites faster than wild-type Pru in vitro (Paul Davis and David Roos, personal communication). Pru mutant 9xG4, which has a dramatically reduced lethality in IFNγ^−/−mice (see Supplemental Text), elicits a reduced CD8⁺ T cell response for both GRA4 and ROP7 at 8 weeks post-infection. Clearly, the kinetics of parasite stage conversion and the morphology of the bradyzoite stage influence CD8⁺ T cell specificity.

Fig. 4. Stage-specific delivery and expression of antigens affects the specific CD8⁺ T cell repertoire.

(A) T. gondii mutants defective in establishing a chronic infection elicit differential epitope-specific CD8⁺ T cell responses at 4 and 8 weeks post-infection. For details, see text (*, p < 0.05, Welch’s t test). (B) Western Blot of the expression of ROP7 versus GRA4 in T. gondii ME49 tachyzoites and in vivo-generated bradyzoites shows GRA4 to be strongly expressed in the tachyzoite stage only while ROP7 exhibits similar expression profiles in both parasitic stages. SAG2-Y is exclusively expressed in bradyzoites and alpha-tubulin was used as loading control.

The differences seen for the H-2L^d/ROP7-specific T cells prompted us to investigate if these CD8⁺ T cells could be detected in BALB/c animals infected with replication-deficient T. gondii 6 weeks post-infection. BALB/c mice infected with 2×10⁶ γ-irradiated T. gondii Pru tachyzoites failed to elicit an H-2L^d/ROP7 specific response in the brain or the spleen (data not shown). We confirmed that all animals were parasite-exposed by checking for serum IgG specific for T. gondii (Fig. S2B and C). The absence of H-2L^d/ROP7 CD8⁺ T cells in this model can be explained by a requirement of the parasite to present the ROP7 antigenic peptide from the bradyzoite stage to CD8⁺ T cells.

Rhoptries are organelles that secrete their contents during invasion of the tachyzoite stage [32] and their function during the bradyzoite stage has not been investigated. We therefore examined the expression levels of ROP7 and GRA4 in relation to each other in the tachyzoite and bradyzoite stage of T. gondii. ME49 bradyzoites were prepared from the brains of infected Swiss-Webster mice and the corresponding ME49 tachyzoites came from serial passage in vitro through fibroblasts (HFFs). GRA4 protein levels were higher in the tachyzoite than in the bradyzoite stage, while expression of ROP7 is comparable between the two (Fig. 4B).

Discussion

Even though IFN-γ producing CD8⁺ T cells protect mice against toxoplasmic encephalitis [33, 34], the epitopes that mediate this recognition are unknown. Conventional methods to discover CD8⁺ T cell epitopes are especially difficult and time-consuming for parasitic infections and account for this dearth of information. Therefore, we set up a Class I MHC tetramer-based screen to directly visualize parasite-specific CD8⁺ T cells derived from a primary T. gondii infection. We discovered two CD8⁺ T cell epitopes derived from the T. gondii proteins GRA4 and ROP7. They are unique to GRA4 and ROP7 and neither epitope is polymorphic for the Type I (GT1), Type II (ME49) and Type III (VEG) strains of T. gondii sequenced to date. Together, these common strains cover 90% of the worldwide isolates of T. gondii. Interestingly, the GRA4 protein induces the proliferation of T-lymphocytes from infected animals of an H-2^d and H-2^k background [35]. Improvements of protein annotation, secretory protein prediction and stage specific microarray transcription, as well as mass spectrometry proteome data, available through the Toxoplasma database, should facilitate further selection of proteins for epitope identification. Relaxation of the search criteria to include peptides of 8 and 10 amino acids may increase the yield of CD8⁺ T cell epitopes.

GRA4 is a protein secreted through the apicomplexan-specific dense granules (Fig. 2A) into the parasitophorous vacuole (PV). Here, GRA4 forms a multimeric complex with GRA6 and GRA2 and stably associates with the intravacuolar network [19]. Even though the protein composition of the dense granules within the PV is now known, the function of these proteins is still poorly understood [36]. We find the H-2L^d/GRA4 restricted peptide SPMNGGYYM in the annotation using the TgTwinScan model, but not with the other annotation algorithms (Fig. S1). Rhoptries are the second organelle released during invasion and some ROP proteins are injected directly into the host cell (Fig. 2A)[32]. ROP7 is a rhoptry bulb protein, a member of the ROP2 kinase family and is injected into the host cell cytoplasm relocalizing to the PV membrane, possibly interacting with the host cell cytoplasm [18, 37]. The epitope IPAAAGRFF in ROP7 maps to its hydrophobic N-terminal region [38]. Due to sequence similarity of proteins within the ROP2 family, a sequence related to the IPAAAGRFF epitope with a single amino acid difference is present in two other rhoptry proteins - IPAAALRFF for both ROP2 and ROP8. However, T cells from T. gondii-infected mice consistently failed to stain with H-2L^d/IPAAALRFF-tetramers (data not shown).

The frequency of the ROP7-specific CD8⁺ T cells appears somewhat lower than CD8⁺ T cells reactive to the secreted model antigen β-galactosidase [31]. Whether this implies the presence of other antigens important for the CD8⁺ T cell response to this stage of the life cycle of T. gondii or is simply the result of an over-expressed model antigen remains open. Of note, the model antigen-specific T cell population was detected only when β-galactosidase was expressed under a tachyzoite promoter. The identification of two endogenous T. gondii antigens that evoke CD8⁺ T cell responses with such different kinetics underscores the importance of knowing the true parasite-derived epitopes. The prominent GRA4-specific CD8⁺ immune response during the first two weeks following infection, during the T. gondii tachyzoite stage, correlates with the protein expression data (Fig. 4B). Delivery of the ROP7 antigen during the bradyzoite stage for presentation to CD8⁺ T cells is a definite possibility, as ROP7 is transcribed (Marianne Matrajt and David Roos, personal communication) and according to our data expressed in both the tachyzoite and bradyzoite stage (Fig. 4B). The failure to detect ROP7-specific CD8⁺ T cells during acute infection might indicate the presence of yet another population of CD8⁺ T cells that contributes to the early response. Moreover, the secreted protein pool in tachyzoites is large and complex, with other proteins presumably dominating the response. In bradyzoites, the mixture of secreted proteins is less diverse and the exchange with the host cell is dramatically reduced. ROP proteins are required to assure host cell survival [32]. Since bradyzoites have an extended life-span, the ROP proteins might have an as yet underappreciated role in bradyzoites, and if so, they are likely the most abundantly secreted factors.

The pathway for delivery of antigens into the MHC class I presentation pathway for proteins expressed into the parasitophorous vacuole remains elusive [39]. Both GRA4 and ROP7 are transcribed equally in Pru tachyzoites and ME49 cysts, as judged by microarrays (Amit Bahl and David Roos, personal communication). However, we observed altered protein levels for GRA4 in the two parasitic stages and it is tempting to speculate that escape from the PV for GRA4 is differentially regulated during the tachyzoite versus the bradyzoite stage. Indeed, GRA4 may be limited, or not secreted in the bradyzoite stage and is not part of the cyst wall [40, 41]. ROP7, on the other hand, is most likely associated with the PV membrane, possibly in the host cell cytoplasm [3, 18]. Regardless of level and timing of expression, ROP7 and GRA4 must access the class I processing machinery. Intracerebral CD8⁺ T cells infiltrate from the acute phase of T. gondii infection in response to a transgenically expressed tachyzoite-stage antigen, where they persist and finally are slowly eliminated by apoptosis [42]. Unknown expression levels and localization of the transgenic antigen chosen may account for the seemingly different behavior we see for the H-2L^d/ROP7-specific intracerebral T cells. Selective traffic of antigen-specific CD8⁺ T cells into the brain occurs in vivo and is dependent on expression of Class I MHC by cerebral endothelium and the presence of the cognate antigen [43]. Once in the brain, the CD8⁺ T cells can undergo additional proliferation [44]. With the identification of two stage-specific endogenous T. gondii CD8⁺ epitopes these questions now become tractable without the need for genetically engineered parasites [45].

Underperformance of the immune system causes recrudescence of T. gondii. Continuous surveillance by CD8⁺ T cells likely keeps T. gondii under control, and candidate CD8⁺ T cells capable of such surveillance include those that recognize ROP7, expressed even at the bradyzoite stage. Our results represent the first step toward a more complete characterization of the immune response against T. gondii. The possible identification of epitopes that afford protection against outgrowth of parasites in cysts may facilitate the development of strategies to vaccinate against, or otherwise control this widespread and clinically important pathogen.