Toxoplasma gondii strains were serially passaged in ‘Toxo medium’ [4.5 g/liter D-glucose in DMEM with GlutaMAX (Gibco, cat# 10566024), 1% heat-inactivated fetal bovine serum (FBS) (Omega Scientific, cat# FB-11, lot# 441164), 1% penicillin-streptomycin (Gibco, cat# 15140122)], in confluent flasks of monolayers of human foreskin fibroblasts (HFFs) and cultured at 37°C, 5% CO₂. HFFs were cultured in ‘HFF medium’ [4.5 g/liter D-glucose in DMEM with GlutaMAX (Gibco), 20% heat-inactivated FBS (Omega Scientific), 1% penicillin-streptomycin (Gibco), 0.2% Gentamicin (Gibco, cat# 15710072), 1X L-Glutamine (Gibco, cat# 21051024)]. Strains assayed are listed in Table S1 , some are generous gifts from Jeroen Saeij (University of California, Davis) and David Sibley (Washington University, St. Louis), and others were obtained from BEI Resources.

Transnuclear T57 (Kirak et al., 2010) and T-GREAT mice were bred in-house under specific pathogen free (SPF) conditions. T-GREAT mice (Kongsomboonvech et al., 2020) bear the same T cell receptor specificity as T57, but also express an Ifng : YFP reporter that allows measurement of Ifng transcript by detection of YFP fluorescence (Reinhardt et al., 2015). C57BL/6J (B6) (colony 000664) and P2x7r-/- (colony 005576) mice were purchased from Jackson Laboratories and kept in-house under SPF conditions. Hind bones from Ripk3-/- mice, B6.129-Ripk3^tm1Vmd (Newton et al., 2004), were provided by Laura Knoll (University of Wisconsin). Bone marrow cells from the hind bones of 6-8 weeks old mice were obtained and cultured in ‘BMDM medium’ [4.5 g/liter D-glucose in DMEM with GlutaMAX (Gibco), 20% heat-inactivated FBS (Omega Scientific), 1% penicillin-streptomycin (Gibco), 1X non-essential amino acids (Gibco, cat# 11140076), 1 mM sodium pyruvate (Gibco, cat# 11360070)] supplemented with 20% L929 conditioned medium, and then harvested after 6-7 days of differentiation. All animal protocols were approved by UC Merced’s Committee on Institutional Animal Care and Use Committee (AUP 20-0015). All mouse work was performed in accordance to the Guide to the Care and Use of Laboratory Animals of the National Institutes of Health and the Animal Welfare Act (assurance number A4561-1). Euthanasia of mice was performed by inhalation of CO₂ to effect of 1.8 liters per minute.

B6 BMDMs were plated at 2x10⁵ cells per well in a 96-well tissue culture-treated plate, in BMDM medium supplemented with 10% L929 conditioned medium, and incubated overnight. The next day, BMDMs were infected with T. gondii tachyzoites, in triplicates, in ‘T cell medium’ [RPMI 1640 with GlutaMAX (Gibco, cat# 61870127), 20% heat-inactivated FBS (Omega Scientific, cat# FB-11, lot# 441164), 1% penicillin-streptomycin (Gibco, cat#15140122), 1 mM sodium pyruvate (Gibco, cat# 11360070), 10 mM HEPES (Gibco), 1.75 µl of β-mercaptoethanol (Gibco, cat# 21985023) per 500 mL RPMI 1640 with GlutaMAX]. The infections were done at multiplicity of infection (MOI) of 0.6, 0.2, and 0.07. Approximately 2 hours post-infection, 5x10⁵ naïve lymph node cells and splenocytes obtained from a naïve T57 transnuclear mouse were added into all wells of infected BMDMs. The lymph nodes and spleens were processed and combined, and red blood cells were lysed with ammonium chloride-potassium (ACK) lysis buffer, prior to being added to the co-culture. The supernatants of the co-cultures were harvested 48 hours later for further analysis by ELISA according to the manufacturer’s instructions (Invitrogen eBioscience, cat# 88731477). The supernatants were analyzed at 1:2, 1:20, and 1:200 dilutions to obtain values within the linear range of the manufacture’s ELISA standards.

T-GREAT cells were co-cultured with infected BMDMs as described above for T57 CD8 T cells. After 14 to 18 hours of the co-culture, cells were harvested for flow cytometry. With preparations all done on ice, cells were washed with ‘FACS buffer’ [PBS pH 7.4 (Gibco, cat# 10010049), 2% heat-inactivated FBS (Omega Scientific)] and blocked with ‘blocking buffer’ [FACS buffer with 5% normal Syrian hamster serum (Jackson Immunoresearch, cat# 007-000-120), 5% normal mouse serum (Jackson Immunoresearch, cat# 015-000-120), and anti-mouse CD16/CD32 FcBlock (BD Biosciences, clone 2.4G2) at 1:100 dilution)]. Samples were stained at 1:120 dilution with fluorophore-conjugated anti-mouse monoclonal antibodies against CD8α PE (eBioscience) or BV510 (BioLegend) (clone 53–6.7), CD3ϵ APC-eFlour780 (eBioscience, clone 17A2), CD62L eFlour450 (eBioscience, clone MEL-14), and CD69 APC (BioLegend, clone H1.2F3). Samples were washed and then incubated with propidium iodide (PI) at 1:1000 dilution (Sigma, cat# P4170). Flow cytometry was performed on an LSRII (Becton Dickinson) or ZE5 (Bio-Rad) analyzer and processed with FlowJo software (version 10.8.1); PI+ cells were excluded from analysis.

HFFs were plated in 24-well tissue culture-treated plates in HFF medium. Confluent monolayer HFFs were infected with 100 and 300 parasites. Plaques were quantified 4-6 days after infection. Displayed results are from MOIs with similar viability, the equivalent of ~MOI 0.2 was chosen for most assays.

The QTL analysis for the CD8 T cell IFNγ response phenotype, including both 1D (scanone) and 2D scans (scantwo) as well as the effect plots, was performed using ‘R’ version 3.6.1 and the ‘R/qtl’ package (Broman et al., 2003). The concentration of CD8 T cell-secreted IFNγ that was obtained for each F1 IxII progeny was normalized to that of type II strain, and for the F1 IxIII progeny, the values were normalized to the type III strain. The average of normalized values obtained over multiple experiments was then Log10 transformed and used for the QTL analysis. For the F1 IIxIII progeny, the Log10 of the CD8 T cell IFNγ concentration from a single experiment was used. A thousand permutations were calculated to obtain significant threshold values (p ≤ 0.05 and p ≤ 0.1). Genetic markers and the allelic assignments for each F1 progeny were obtained from previous studies (Su et al., 2002; Saeij et al., 2006; Behnke et al., 2011) and publicly available databases (http://toxomap.wustl.edu/IxIII_Typing_Table.html). The genetic maps of chromosomes VIIb and VIII were stitched into a single chromosome and renamed as “VIIb-VIII”. Recent evidence suggests these are a single chromosome (Bunnik et al., 2019; Xia et al., 2021), and our revised genetic map reflects the correct genetic marker orientation in the stitched VIIb-VIII T. gondii chromosome.

To generate a double knockout strains, Cas9-expressing pSS013 plasmid (gift from Jeroen Saeij, University of California, Davis) containing single guide RNAs (sgRNAs) targeting exon 1 of the candidate ROCTR genes, TG_278878 and TG_278882, were co-transfected via electroporation with selectable markers hypoxanthine-guanine phosphoribosyl transferase (HXGPRT) or dihydrofolate reductase (DHFR-TS) into RH Δhxgprt Δku80 or ME49 Δhxgprt::FLUC (ME49 Δhpt), respectively, at a 5:1 ratio of sgRNAs/plasmid to PCR amplicon of the selectable marker. The HXGPRT amplicon was generated using the pTKO-att plasmid as the template DNA and the DHFR-TS amplicon was generated using the pLoxP-DHFR-mCherry plasmid as the template DNA. In the case of the HXGPRT amplicon, 20 bp of homology arms were introduced during the PCR to allow for homology directed repair and removal of the TG_278878 and TG_278882 in the RH Δhxgprt Δku80 genetic background. The transfection bulk populations were then selected with ‘MPA xanthine’ medium [4.5 g/liter D-glucose in DMEM with GlutaMAX (Gibco, cat# 10566024), 1% heat-inactivated fetal bovine serum (FBS) (Omega Scientific, cat# FB-11, lot# 441164), 1% penicillin-streptomycin (Gibco, cat# 15140122), 0.05 mg/ml mycophenolic acid (MPA) (Millipore Sigma, cat# 475913), 0.05 mg/ml xanthine (Alfa Aesar; cat# AAA11077-22)] or in Toxo medium containing 3 µM pyrimethamine (Millipore Sigma, cat# 46706) to select for the presence of HXGPRT or DHFR-TS insertion, respectively. The drug selected population was screened by diagnostic PCR for evidence of successful targeting and repair of the Cas9 cut site with the selectable marker and then cloned by limiting dilution in a 96-well plate in MPA xanthine or pyrimethamine selection medium. Wells with only one clone were isolated and candidate gene disruption was again confirmed by diagnostic PCR for successful marker integration and double gene deletion. Parasite genomic DNA was isolated using DNAzol (Invitrogen, cat# 10503027) and ethanol precipitation. Diagnostic PCR was performed with MangoMix, which contains MangoTaq DNA polymerase (Bioline, cat# BIO-25-33). PCR amplicons of the selectable markers were generated with Phusion polymerase (NEB, cat# M0530L). Sequencing of PCR products and plasmids was performed by UC Berkeley DNA Sequencing Facility and samples were prepared according to their protocol. Sequencing results were analyzed using Sequencher 5.4.6 and SnapGene Viewer 5.2.1. All oligos and plasmids used in this study can be found in Table S2 . The RH Δnsm, RH Δist and RH Δist Δnsm strains were made in an analogous manner as that previously described for ME49 (Rosenberg and Sibley, 2021).

HFFs were plated on coverslips with HFF medium in 24-well tissue culture-treated plates. Once confluent, the monolayer HFFs were infected with T. gondii and incubated at 37°C, 5% CO₂ for 16 hours. The samples were then fixed with 3% formaldehyde in phosphate buffered saline (PBS) for 20 minutes. After the fixation, these were blocked with blocking buffer (3% BSA, 5% normal goat serum, 0.2% Triton X-100, 0.1% sodium azide in PBS). To visualize GRA5, the infected cells were stained with mouse anti-GRA5 primary monoclonal antibody (BioVision, clone TG 17.113) at 1:500 dilution, followed by Alexa Fluor 488 goat anti-mouse IgG (Life Technologies, cat # A11029) secondary antibody at 1:3000 dilution. To visualize TGD057, the infected samples were stained with rabbit anti-TGD057 polyclonal antibody (gift from Nicolas Blanchard, INSERM) at 1:2000 dilution, followed by Alexa Fluor 594 goat anti-rabbit IgG (Life Technologies, cat# A11037) secondary antibody at 1:3000 dilution.

T. gondii lysates were prepared through syringe-lysis, resuspended in Laemmli buffer (125 mM Tris HCl, 30% glycerol, 2% SDS, 0.2% bromophenol blue), and denatured at 90-100°C for 5 minutes. The lysates were separated on 4-20% Mini-PROTEAN TGX pre-cast 540 gels (Bio-Rad, cat# 4561096) or 5-15% polyacrylamide gels and then transferred to nitrocellulose membranes. The membranes were blocked with 10% fortified bovine milk in ‘TBS-T 0.1%’ [Tris-Buffered Saline and 0.1% Tween] at room temperature for 1h, and then incubated with α-TGD057 antibody (gift from Nicolas Blanchard, INSERM) at 1:4000 dilution in TBS-T 0.1% overnight at 4°C. Membranes were washed with TBS-T 0.1% three times and later incubated with goat α-rabbit horseradish peroxidase (HRP)-conjugated antibodies (Southern Biotech, cat# 4030-05) at 1:4000 dilution for 1h at room temperature. Membranes were then washed again TBS-T 0.1% three times and developed with Immobilon^® Forte Western HRP Substrate (Millipore, cat# WBLUF0500). All blots were imaged via chemiluminescence on a ChemiDoc Touch (Bio-Rad, cat# 12003153).

Bar graphs represent the average value obtained for all experiments, with standard deviations indicated. For these, values from individual experiments are represented as dots. For each measured experiment, results between parasites strains or conditions were mainly represented as that relative to the T57 response elicited by wildtype BMDMs infected with type II strains (equal to 1). For data with normal distribution, one-way or two-way ANOVA with Bonferroni’s ad-hoc statistical tests were determined. For data with non-Gaussian distribution, Kruskal-Wallis and Dunn’s ad-hoc test were applied. P-values < 0.05 were considered significant. Statistical analyses were performed with GraphPad Prism version 8.3.0.

To identify T. gondii genetic loci responsible for the strain-differences in CD8 T cell responses to infections, and ultimately ROCTR, we assessed 29 strains of F1 progeny from a type I x type II cross (F1 IxII), 34 strains of F1 progeny from a type II x type III cross (F1 IIxIII), and 32 strains of F1 progeny from a type I x type III cross (F1 IxIII) in an antigen-specific CD8 T cell activation assay ( Figure 1A ) (Kongsomboonvech et al., 2020). In brief, C57BL/6J bone marrow-derived macrophages (BMDMs) were infected with T. gondii and co-cultured with splenocytes and lymph node cells from naïve transnuclear ‘T57’ mice. These mice were cloned from the nucleus of a single tetramer-positive T. gondii-specific CD8 T cell which have a single T cell receptor specific for the TGD057_96-103 epitope (T57 epitope) presented by H-2K^b MHC I (Kirak et al., 2010; Wilson et al., 2010). The supernatant was harvested at 48 hours of the co-culture and analyzed for IFNγ concentration by ELISA. Although the TGD057_96-103 epitope is conserved among all T. gondii strains, type I parasite strains (RH, GT1) induce low T57-specific CD8 T cell-mediated IFNγ responses, while type II (ME49, Pru) and type III strains (CEP) induce a much higher response ( Figure 1B ), as previously described (Kongsomboonvech et al., 2020). The CD8 T cell response differs greatly in response to infections of F1 IxII T. gondii strains ( Figure S1 ). For example, the CD8 T cell IFNγ response to SF46 is quite low, unlike the high IFNγ level elicited in response to SF20 infections ( Figure 1B ). These results suggest the potential to genetically map loci responsible for strain-differences in T57 IFNγ responses to an endogenous antigen, which localizes to the PV (Lopez et al., 2015; Kongsomboonvech et al., 2020) and actin cytoskeleton of the parasite (Wilson et al., 2010).

A genome-wide QTL scan was performed to detect genotype-phenotype correlations using F1 progeny derived from all three sexual crosses of the clonal lineages for the naïve T57 IFNγ response at 48 hours. QTL analysis of the F1 IxII cross revealed three suggestive peaks with an LOD value greater than 2 on T. gondii chromosomes X (chrX) and XII (chrXII) ( Figure 1C ). Following genome wide permutation testing (n=1000), the QTL on chromosome XII (“PEAK3”) with the logarithm of odds (LOD) score of 4.4, surpassed a threshold value of p = 0.10, but none of the other QTLs (“PEAK1”, “PEAK2”) returned significant values. Polymorphic candidate genes within the 1.5 LOD interval of PEAK3, a 106 Kb region spanning 13 genes (TGME49_chrXII:5754579-5860469), include most notably two tandem uncharacterized NTPases, TG_278878 and TG_278882 (ToxoDB.org) ( Table 1 ). The genetic marker corresponding to the maximal LOD score within PEAK3 (65.m01195_at9) is a SNP adjacent to TG_278878. On chromosome X, there are two noticeable peaks. The major peak towards the right end of chromosome X (PEAK2) is a large 1.1 Mb locus (TGME49_chrX:3782766-4930776) encoding 147 genes and includes the genetic marker 46.m03675_at7, which produced the highest LOD score of 3.6 within this chromosome ( Figure 1C ). The minor peak towards the left end of chromosome X (PEAK1) corresponds to a ~419 Kb region of 48 genes (TGME49_chrX:1383691-1802907) and encompasses the genetic marker 42.m03493_at7 with a LOD score of 2.3 ( Figure 1C ); this marker defines a SNP within the T. gondii dense granule GRA35. Notably, the ROP5 locus, which is a known virulence determinant encoded on chromosome XII and identified using this panel of F1 IxII progeny (Behnke et al., 2011), did not produce an identifiable QTL in our screen. Thus, while ROP5 can inhibit the CD8 T cell response (Rommereim et al., 2019; Kongsomboonvech et al., 2020), polymorphisms in ROP5 do not account for strain-specific differences observed for this phenotype, as previously suggested (Kongsomboonvech et al., 2020). The TGD057 antigen encoded by TG_215980 is similarly expressed and entirely conserved between clonal strains, and no phenotype-genotype correlation was observed at this locus ( Figure 1C ). Based on the effect plots, F1 IxII progeny that express type II alleles at the chromosome X genetic markers 42.m03493_at7 and 46.m03675_at7 induce higher CD8 T cell-mediated IFNγ responses compared to those that express type I alleles at these loci, respectively ( Figures 1D, E ). In contrast, F1 IxII progeny that express a type I allele at the chromosome XII genetic marker 65.m01195_at9 induce higher CD8 T cell responses compared to those that express the type II allele ( Figure 1F ), suggesting the genetic background of type I strains may mask the effect of the putative chrXII ROCTR. In search of loci that potentially modify the function of the ROCTR and lower the detection of significant QTLs, epistatic interactions and ‘interactive’ QTLs were calculated, but none were detected between PEAKS 1, 2, and 3 or other loci within the F1 IxII genetic cross ( Figure S2A , not shown).

Other sexual crosses examined included the F1 progeny derived from the IxIII and IIxIII crosses, but QTL mapping revealed no statistically significant peaks ( Figures S3 , S4 ). It can be concluded that the polymorphic rhoptries responsible for strain-differences in virulence previously identified through QTL analyses of these same panels of F1 progeny, such as ROP18 and ROP5 (from both F1 IIxIII and F1 IxIII crosses) encoded on chromosomes VIIa and XII, respectively (Saeij et al., 2006; Taylor et al., 2006; Reese et al., 2011), do not seem to account for the strain-differences in host CD8 T cell IFNγ responses to T. gondii infections. Moreover, the small effect QTLs suggest that the T57 CD8 IFNγ response may be controlled by multiple gene loci of T. gondii and subject to environmental input during the 48 hours co-culture of the CD8 T cells and infected macrophages. Top polymorphic ROCTR candidates for PEAKS 1, 2 and 3 are included in Table 1 .

Candidate genes on T. gondii chromosome XII that correspond to PEAK3 include TG_278878 and its adjacent gene TG_278882, which are nucleoside triphosphate hydrolases (NTPases) of the GDA1/CD39 family of ecto-ATPases with apyrase activity (ToxoDB.org). In general, NTPases are secreted following invasion and localize to the PV lumen and PVM (Bermudes et al., 1994; Sibley et al., 1994). The related NTPases, NTPase I and NTPase II (Bermudes et al., 1994), do not contribute to type I strain virulence in mice but deplete cellular ATP (Olias and Sibley, 2016) and are thought to be important for tachyzoite replication (Nakaar et al., 1999; Asai et al., 2002) and possibly egress (Stommel et al., 1997). Large quantities of parasite-derived NTPase can be detected in the serums of infected mice (Asai et al., 1987) and represents up to 2-3% of the entire protein in the tachyzoite (Asai et al., 1983). NTPases may impact the host response in another way. It is possible that the ATP hydrolysis activity of NTPases may dampen host inflammasome activation in response to T. gondii infections (Melo et al., 2011). Inflammasome activation, for both NLRP1 and NLRP3, has been shown to be important for T. gondii control (Cirelli et al., 2014; Ewald et al., 2014; Gorfu et al., 2014) and can be triggered through the binding of exogenous ATP to the purinergic receptor P2X7 (Jo et al., 2016; Amores-Iniesta et al., 2017). It is possible that T. gondii NTPases deplete the amount of cytosolic ATP, thus preventing inflammasome activation either by lowering available exogenous ATP required for P2X7 receptor (P2X7R) activation following egress, or by thwarting NLR-oligomerization which is an ATP-dependent process (Duncan et al., 2007). Previously, we described an NLRP3-dependent pathway that is required to induce CD8 T cell IFNγ responses to T. gondii infections (Kongsomboonvech et al., 2020). For these reasons, the TG_278878 and TG_278882 NTPases were pursued as candidates in our search of ROCTR.

Double mutant Δ278878 Δ278882 T. gondii parasite strains were generated in both the RH Δhxgprt Δku80 and ME49 Δhxgprt genetic backgrounds using CRISPR/Cas9. Parasites were given Cas9 and small guide RNAs (sgRNAs) targeting the first exon of each gene and deletion strains were selected for those that bore evidence of repair with a transfected selectable marker ( Figure 2 ). In the type II ME49 deletion strain, both NTPase genes were successfully disrupted and the DNA sequence between the two CRISPR/Cas9 cut-sites was replaced with a dihydrofolate reductase (DHFR) selectable marker (Δ278878::DHFR::Δ278882) as confirmed by PCR ( Figure 2A ) and sequencing of the edited locus (not shown). During attempts to generate a double deletion strain in the type I RH background, it became clear that the syntenic NTPase genes to TG_278878 and TG_278882, corresponding to TGRH88_065000 and TGRH88_064900 respectively, were significantly different from those of type I GT1 and type II ME49 strains. PCR performed with several primer pairs flanking, within and between TG_278878 and TG_278882 genes, consistently yielded PCR products indicative of a large-scale deletion between the two genes in the RH genetic background ( Figure S5 ). The recent release of the RH genome (GCA_013099955.1) confirms these results and indicates that a 4.5 Kb deletion occurred at this locus, in which the 3’ end of exon 4 for TG_278878 experienced a 338 bp truncation and was then fused to a TG_278882 gene missing 1.2 Kb of the 5’ end of exon 1. Whether TGRH88_065000 and TGRH88_064900 encode functional NTPases is unknown. Nonetheless, an RH Δ278878 Δ278882 strain was generated in which an HXGPRT selectable marker replaced the DNA sequence internal to the two CRISPR/Cas9 cut sites ( Figure 2B ). However, no difference in the T57 IFNγ response was observed comparing the NTPase deletion strains with their parental counterparts ( Figure 2C ). Finally, P2x7r-/- BMDMs were screened to address whether P2X7R contributed to the NLRP3-dependent T57 IFNγ response (Kongsomboonvech et al., 2020), however, the response to parasite-infected P2x7r-/- BMDMs is intact ( Figure 2D ), ruling against a major role for exogenous ATP triggered P2X7R signaling in this system. In summary, although there appears to be a slight reduction in the T57 IFNγ response to the ME49 NTPase deletion strain, this difference is not significant. Hence, if these NTPases are ROCTRs, they exert a marginal effect at best, which would be consistent with the small effect QTLs produced by the genetic mapping.

PEAK2 encompasses multiple genes, however one candidate was intriguingly close to the genetic marker 46.m03675_at7 that produced the maximal LOD score, the dense granule TgNSM. Recently, TgNSM was described to be exported to the host nucleus where it associates with the NCoR/SMRT co-repressor complex promoting the transcriptional repression of many IFN-stimulated genes (Rosenberg and Sibley, 2021). Importantly, TgNSM works together with another exported dense granule, TgIST, to inhibit the transcription of key regulators of necroptosis following IFNγ or IFNβ stimulation (Rosenberg and Sibley, 2021). Necroptosis is a programed cell death response that is mediated by the RIPK3/RIPK1 signaling complex (Pasparakis and Vandenabeele, 2015), and in certain contexts is initiated by IFN-STAT1 signaling (Thapa et al., 2013). Several studies have demonstrated that RIPK3-dependent necroptosis is potent at inducing CD8 T cell activation in vitro and in vivo (Yatim et al., 2015; Ren et al., 2017; Rana et al., 2020; Aaes and Vandenabeele, 2021). Hence the role of RIPK3 was explored. Importantly, its expression in BMDMs was found to be absolutely required for eliciting T57 IFNγ responses to T. gondii ( Figure 3A ). Moreover, TgIST is a potent repressor of STAT1 signaling via its recruitment of Mi-2/NuRD to phosphorylated STAT1 dimers (Gay et al., 2016; Olias et al., 2016) and we previously demonstrated STAT1 is a requirement for the T57 IFNγ response (Kongsomboonvech et al., 2020), hinting of a possible STAT1-RIPK3 axis that could be intersected by ROCTRs. Given these observations, single and double TgIST and TgNSM deletion strains were screened. However, no significant difference was observed between parental and deletion strains ( Figure 3B ), indicating TgNSM on chromosome X and TgIST are not ROCTRs. This supposition is supported by previous findings regarding the parasite’s export machinery, which is required for TgIST and TgNSM export from the PV to the host nucleus (Gay et al., 2016; Olias et al., 2016; Rosenberg and Sibley, 2021) but is dispensable for T57 IFNγ responses to T. gondii (Kongsomboonvech et al., 2020). Thus, while STAT1- and RIPK3-dependent processes are necessary for inducing CD8 T cell IFNγ responses to T. gondii, any modulation of these signaling pathways by specific T. gondii effectors can be overcome by host macrophages to support T cell activation and/or differentiation in this system.

The genetic marker 42.m03493_at7 that corresponds to PEAK1 is within the gene encoding the T. gondii dense granule GRA35. GRA35 was previously identified as an NLRP1 inflammasome activator in T. gondii infections of Lewis rat macrophages (Wang et al., 2019). GRA35 localizes to the PVM with the aid of GRA42 and GRA43 and remains in PV lumen in their absence (Wang et al., 2019). Therefore, all three dense granules were analyzed and the T57 IFNγ response to Δgra35, Δgra42, and Δgra43 deletion strains was measured as before. Compared to the parental RH strain, CD8 T cell IFNγ responses were slightly elevated to BMDMs infected with RH Δgra35 and RH Δgra42 strains, but significant differences were only observed in response to RH Δgra43 infections ( Figure 4 ). A similar trend was observed in the type II ME49 genetic background, but none of the differences between deletion and wildtype strains were of statistical significance ( Figure 4 ). Given the relatively small effect that GRA35 had on the phenotype, we did not pursue the generation of complementation strains to address whether GRA35 is a ROCTR accounting for some of the strain-differences in the T57 IFNγ response. Regardless, the upstream regulator of GRA35, GRA43, seems important for modulating the T57 IFNγ response to T. gondii.

Collectively, our data suggests that when removing members of the parasite’s PVM-targeting pathway, including the Golgi-resident protein aspartyl protease, ASP5 (Coffey et al., 2015; Hammoudi et al., 2015) and GRA43 (Wang et al., 2019), the host CD8 T cell IFNγ response increases (Kongsomboonvech et al., 2020) ( Figure 4 ). It has also been demonstrated for the PV-associated antigen, GRA6, that PVM targeting increases access into the host’s MHC I antigen processing pathway (Lopez et al., 2015; Jensen, 2016). Hence, we tested whether ASP5 or GRA43 affect where TGD057 localizes within the PV. Thus far, the Blanchard and Yap groups and our studies have shown through immunofluorescence assays (IFA) and/or differential centrifugation techniques that TGD057 is a PV resident protein within the PV lumen (Lopez et al., 2015; Kongsomboonvech et al., 2020) and within insoluble structural elements of the tachyzoite (Wilson et al., 2010). Here, the localization of TGD057 in Δasp5 and Δgra43 T. gondii strains was investigated. TGD057 was detected with a rabbit polyclonal α-TGD057 and the PVM and dense granules were marked by α-GRA5. RH Δgra43 expresses GFP hence GRA5 was not assessed in this strain. Regardless, TGD057 appeared as puncta and was distributed throughout the PV and tachyzoite in all strain types analyzed, irrespective of whether they are stimulatory (ME49, ME49 Δasp5, RH Δgra43) or not (RH, RH Δasp5) ( Figure 5A ). These results imply GRA43 and ASP5 regulate T57 IFNγ responses independently of where TGD057 localizes within the vacuole.

Protein export from the PV is dependent on the parasite’s Golgi protease ASP5 (Coffey et al., 2015; Hammoudi et al., 2015; Curt-Varesano et al., 2016) and the MYR1 PVM-associated translocation machinery (Franco et al., 2016). As noted above, while this machinery was dispensable for T57 IFNγ response induction, ME49 Δasp5 elicited a much greater response compared to its parental control (Kongsomboonvech et al., 2020). We previously noted that TGD057 possesses a TEXEL motif and argued from the literature that TGD057 bore no evidence for ASP5-mediated cleavage to assist antigen processing (Kongsomboonvech et al., 2020). To test this, we performed a simple western blot comparing wildtype and Δasp5 strains using the polyclonal α-TGD057 which recognizes both an N- and C-terminal peptide sequence flanking the putative ASP5 cleavage site of TGD057 (personal communication, Nicolas Blanchard, INSERM). Consistent with initial characterizations of this protein (Wan et al., 2004), TGD057 migrates at the expected size of 18 kDa. Furthermore, no peptide fragments of 5 and 13 kDa were observed, which would be generated if TGD057 were to be cleaved at its putative TEXEL sequence, nor did the signal intensity of TGD057 change in Δasp5 relative to parental strains ( Figure 5B , not shown). In summary, the role for ASP5 and GRA43 in this system is not to process nor localize TGD057 within the PV. Instead, their ability to target dense granules to the PVM may assist the localization of unidentified ROCTRs to the PVM, or regulate the PVM integrity, thereby controlling CD8 T cell activation phenotypes.

Following initial T cell receptor (TCR) stimulation by antigens (also known as ‘signal 1’), early activated T cells require additional signals including co-stimulation (‘signal 2’) and soluble factors (‘signal 3’) to initiate the production of cytokines like IFNγ. Since ROCTRs may potentially intersect each of these steps and to assist genetic mapping, CD8 T cell IFNγ differentiation was measured at an earlier stage and disentangled from phenotypes associated with T cell activation using T-GREAT CD8 T cells. T-GREAT mice encode the same T57 TCR and report IFNγ transcription by YFP fluorescence as measured by flow cytometry, which can be detected as early as 14-18 hours after activation by parasite-infected BMDMs (Kongsomboonvech et al., 2020). Importantly, Ifng transcription can be measured independently of CD69 expression, which is a proxy for early TCR signaling events mediated by MHC antigen presentation (Cebrián et al., 1988) and TGD057 release from the vacuole. Therefore, T-GREAT cells were used to distinguish between T. gondii ROCTRs that may regulate activation (i.e. CD69+) from those that regulate differentiation (Ifng : YFP+) of CD8 T cells. Naïve T-GREAT cells predominately express CD62L, which promotes lymphocyte homing to secondary lymphoid organs (Steeber et al., 1996), and do not express CD69, which retains activated cells within the secondary lymph organs by antagonizing S1PR1 (Shiow et al., 2006) ( Figure 6A ). In contrast, at 14 hours of co-culture with infected BMDMs, T-GREAT cells significantly upregulate CD69 and downregulate CD62L, which occur more readily in response to type II compared to type I strains ( Figures 6A , S6A ). Type II strains are also better at inducing the early Ifng transcriptional response which is mainly observed in the fully activated CD69+ CD62L- subset compared to the CD69+ CD62L+ or CD69- CD62L+ populations of T-GREAT cells ( Figures 6B , S6B , not shown). Hence, the frequency of CD69+ CD62L- T-GREAT cells ( Figure S6A ) and Ifng : YFP+ among activated CD69+ CD62L- T-GREAT cells ( Figure S6B ) were measured in response to F1 progeny of the IxII cross and genetic mapping was performed. Although QTLs with an LOD > 2 were not found for the CD69+ T-GREAT activation phenotype, two peaks were identified for the Ifng : YFP+ phenotype, including a small QTL with the same boundaries as PEAK1 identified for the T57 IFNγ response at 48 hours (TGME49_chrX:1383691-1802907), and a second QTL observed on chromosome VIIb-VIII (“PEAK4”, TGME49_chrVIIb-VIII: 796459-2341638) ( Figure 6C ). Importantly, covariate analysis revealed that PEAK4 is significant with an LOD = 4. 1 (additive-QTL, p < 0.05). No evidence is inferred for epistatic interactions between PEAK4 and PEAK1 ( Figure S2B ). Effect plots reveal that type I, compared to type II alleles, for both QTLs correlate with lower Ifng transcriptional responses of T-GREAT cells ( Figure 6D ).

PEAK4 encompasses a large 1.54 Mb region of 208 genes, with several putative dense granules and rhoptry proteins as ROCTR candidates, including polymorphic ROP16 ( Table 1 ). Type I and III alleles of ROP16 are known to induce the alternative activation (M2) program of infected BMDMs (Jensen et al., 2011) and can inhibit CD8 T cell expansion and IFNγ production in vivo (Tuladhar et al., 2019; Chen et al., 2020) and in vitro (Kongsomboonvech et al., 2020). Importantly, a single polymorphism in the ligand binding domain renders the type II allele unable to maintain STAT6 and STAT3 (Yamamoto et al., 2009). Therefore, we hypothesized that ROP16 is the ROCTR associated with the PEAK4 QTL. A type II Pru strain expressing the ROP16_I allele from the type I strain (Pru+ROP16_I ) was analyzed, as it induces M2 BMDMs and activates the aforementioned STATs to regulate a variety of immune-related genes (Saeij et al., 2007; Jensen et al., 2011; Jensen et al., 2013; Chen et al., 2020). In this system, ROP16 mediates an MOI-dependent effect on the early T-GREAT Ifng transcriptional response to T. gondii ( Figure 6E ). At higher MOIs the Ifng transcriptional response is decreased by Pru+ROP16_I , particularly at MOI 2.4, to levels that resemble those induced by the type I RH strain. In contrast, no discrepancy is observed between Pru+ROP16_I and its parental strain at lower MOIs of 0.6 and 0.2. Whether the MOI-dependent effect of ROP16 diminishes the phenotype-genotype correlation at PEAK4 is unclear. However, these results are consistent with our previous observations that Pru+ROP16_I reduces IFNγ secretion by T57 cells at 48 hours (Kongsomboonvech et al., 2020).