All procedures involving mice were reviewed and approved by the Institutional Animal Care and Use Committee of Dartmouth College (Animal Welfare Assurance Number #A3259-01) and were in accordance with the guidelines published in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

C57BL/6 female mice were purchased from Jackson Labs (Bar Harbor ME) and maintained at the Center for Comparative Medicine and Research at the Geisel School of Medicine at Dartmouth in specific-pathogen-free conditions.

All parasite cultures were maintained in vitro by serial passages in human foreskin fibroblast (HFFs) monolayers in Eagle's modified essential medium (EMEM) (Gibco) supplemented with 1% fetal bovine serum (FBS) (Life Technologies) (33).

All strains used or developed in this study are listed in Table S1. Knockout targeting constructs were developed and parasites were transfected following the previously described protocol (34). Primers used for construct development and knockout or complementation validation are in Table S2. OVA expressing strains were engineered by amplifying ptubP30-OVA by PCR from plasmid ptubP30-OVA/sagCAT (35) using primers ptub_R and DHFR.3′_F that contain overlaps with the UPRT 5′- and 3′- untranslated regions (UTRs) to generate plasmid pOVA, that targets OVA to the UPRT locus following FUDR negative selection (Figure S1A) (34). The Δrop5-OVA and Δgra2-OVA strains were complemented by targeting the WT gene, rop5C for Δrop5 and gra2 for Δgra2, to the endogenous gene locus and then using 6TX selection to isolate positive clones (Figures S2A, S2D). The Δrop18 strain was complemented with targeting plasmids containing WT ROP18 (ROP18), a kinase dead ROP18 (ROP18^KD) (36) or a ROP18 protein unable to interact with ATF6β (ROP18^ATF6β) (37). Each of the ROP18 complementation targeting plasmids contained the complementing ROP18 with a ROP18 promoter and ptubP30-OVA construct flanked by the UPRT 5′- and 3′- UTRs to target the plasmid to UPRT in the parental RHΔku80Δrop18::HXGPRT strain using 5-fluorodeoxyuridine (FUDR) selection to isolate positive clones (Figure S3A). Plasmid pROP18^KD was engineered to mutate the aspartic acid to an alanine at position 394 in ROP18 (38) by QuickChange site direct mutagenesis (Agilent). Plasmid pROP18^ATF6β was generated with primers that deleted amino acids 147 to 164 in ROP18 (37). Genotype verification of gene replacements and deletion events was assessed by PCR and sequencing as previously described (34).

HFF cells were cultured on coverslips and infected with parasites for 24–48 h. To visualize OVA expression the cultures were fixed with Histochoice (Amresco), and permeabilized with 0.001% Triton X-100 (Sigma). To visualize ROP18 the cultures were fixed with Histochoice and permeabilized in 0.1% saponin for 10 min. To visualize ROP5, cultures were fixed with 4% PFA (Electon Microscopy Sciences) and permeabilized with 0.01% Triton X-100. To visualize GRA2, cultures were fixed with 4% PFA and permeabilized with 0.1% saponin. All samples were blocked with 10% FBS, incubated with primary antibodies (1 h at RT): rabbit α-OVA (1:1,000, Bethyl Laboratories), rabbit α-ROP18 (ROP18-His WA525, 1:500, Sibley laboratory), rabbit α-ROP5 [MO556, 1:3,000, (23)], and mAb α-GRA2 (France-Delauw Laboratory, 1:500). Cultures were washed 3 times with PBS and incubated 1 h at RT with secondary antibodies conjugated to an Alexa Fluor (Invitrogen). Samples were mounted in ProLong Gold with DAPI (Invitrogen) and then imaged with a Nikon A1R SI confocal microscope (Nikon, Inc.). Confocal images were processed with FIJI (39).

Bone marrow derived macrophages (BMMΦ) were differentiated from bone marrow isolated from C57BL/6 WT or Irgm1/m3 knockout (18) mice in a 5-day culture with 30% L929-culture supernatant as previously described (40). Bone marrow derived dendritic cells (BMDC) were differentiated from bone marrow isolated from C57BL/6 or Irgm1/m3 knockout (18) mice in a 9-day culture with GM-CSF (Peprotech) as previously described (41). Purity was verified by FACS analysis (data not shown) and unless otherwise stated, all incubations with BMMΦs or BMDCs were performed at 37°C.

A previously described flow cytometry based secretion assay was used to measure the accumulation of OVA protein in the lumen of the PV to verify that different mutant strains of Toxoplasma expressed and secreted equivalent OVA into the PV lumen (42). BMMΦ and BMDCs were infected with Toxoplasma at a MOI of 2.5 and incubated for 24 h. Cells were then washed with PBS, fixed with PFA (4%) and permeabilized with saponin (0.05%) to selectively expose PV lumen proteins for analysis. The expression of OVA was compared to the expression of GRA5 in each PV. To identify the infected cells that were double positive for GRA5 and OVA, processed cells were stained for mouse α-GRA5 (1:1,000, Biotem, TG 17.113) and revealed using AF647 goat anti-mouse IgG (1:500), and stained with rabbit α-OVA (1:1,000, Bethyl) and revealed using AF488 goat anti-rabbit IgG (1:500).

In vitro antigen presentation by BMMΦ and BMDCs was measured using the B3Z CD8⁺ T cell hybridoma that expresses β-galactosidase upon recognition of the SIINFEKL peptide (OVA_257−264) presented on H-2K^b-molecules (Provided by N. Shastri, University of California, Berkley) (43). WT and Irgm1/m3 KO BMMΦ or BMDC (10⁵) were seeded on 96-well trays and incubated overnight (BMMΦ) or for 4–6 h (BMDC). To activate cells (primed) prior to infection they were cultured for 4–6 h with IFN-γ (100 U/ml, Peprotech) and TNF-α (10 U/ml, Peprotech). Parasites were added to the APCs at an MOI of 2.5 and incubated for 12 h. Infected cultures were then spun down and washed twice with RMPI medium without phenol red (Gibco) and B3Z cells (10⁵) were added to the culture and incubated for 12 h. Lysis buffer (44) containing CPRG (final concentration 100 μM) was added to the cultures and plates were incubated for 14 h. Absorbance was read at 562 nm, with 650 nm as the reference wavelength. Data is from three experiments, each conducted in triplicate then averaged and normalized with reference to uninfected controls and represented as the fold change compared to an uninfected control or as the percent increase of each knockout compared to RH-OVA for the same condition.

WT and Irgm1/m3^−/− BMMΦ and BMDCs (4 × 10⁶) were seeded on 6-well trays overnight or for 6 h, respectively. To quantitate the level of IFN-γ-dependent PV clearance, the Toxoplasma infected cells were either incubated for 4–6 h with normal culture media (non-activated cells) or media containing IFN-γ (final concentration of 100 U/ml, Peprotech) and TNF-α (final concentration of 10 U/ml, Peprotech) (primed cells), and then infected with 100 tachyzoites/well and incubated for 6 days at 37°C. The medium was removed from each well and the remaining monolayer was fixed and stained with coomassie brilliant blue to reveal plaques formed from the surviving Toxoplasma PVs. The number of plaque forming units (PFUs) per well were counted and the percent killing (PV clearance) was determined by dividing the difference of the number of PFUs in unprimed and primed cells by the total number of PFUs in the unprimed cells. The assay was performed in triplicate in two separate experiments.

WT and Irgm1/m3^−/− APCs were seeded on a coverslip and incubated overnight (BMMΦ) or for 4–6 h (BMDC), then primed with IFN-γ (final concentration of 100 U/ml, Peprotech) and TNF-α (final concentration of 10 U/ml, Peprotech) for 6 h. Coverslips were infected with parasites at an MOI of 4 for 45 min then washed and fixed with 4% PFA (Electron Microscopy Sciences). Cells were then permeabilized with 0.1% saponin (Sigma) and blocked in 10% FBS. For visualization, cultures were incubated with mouse α-GRA5 (1:2,000, Biotem, TG 17.113) and rabbit α-Irgb6 [1:1,000, (18)], then washed and incubated with secondary antibodies anti-mouse Alexa Fluor 568 and anti-rabbit Alexa Fluor 488 (Invitrogen). Coverslips were mounted in ProLong Gold with DAPI (Invitrogen) and imaged at 63x with a Nikon A1R SI confocal microscope (Nikon, Inc.). All images were processed with FIJI (39). A minimum of 500 vacuoles was scored for each strain for quantification of Irgb6 vacuole coating.

Groups of mice were infected intraperitoneally with 100,000 (10⁵) or 1,000 (10³) tachyzoites in 0.2 ml PBS. Parasite viability was determined by a plaque forming unit (PFU) assay (data not shown).

Statistical analysis was performed using PRISM software (Graphpad Software). Survival was analyzed by the Log-rank (Mantel-Cox) test performed using the Gehan-Breslow-Wilcoxon test; a P ≤ 0.05 was considered significant. All other statistical calculations were performed using the non-parametric Mann-Whitney test; a P ≤ 0.05 was considered significant.

The model T cell antigen ovalbumin (OVA) was previously engineered as a single copy gene to drive high-level expression and secretion of soluble OVA protein (amino acids 140–386) into the lumen of the parasitophorous vacuole (PV) (3, 13, 35). This OVA transgene was targeted to the UPRT locus in the genetically tractable type I strain RHΔku80 background (33, 45) and clones of RH-OVA expressing a single copy of OVA inserted into the UPRT locus were isolated after selection in 5-fluorodeoxyuridine (FUDR) (Figures S1A,B) (46). As expected, OVA accumulated specifically in the PV lumen inside of the PV membrane (PVM) and did not accumulate in parasites or in the host cell (Figures S1C,D). This PV localized OVA antigen contains the SIINFEKL epitope that can be presented by MHC-I at the host cell surface to prime H-2k^b restricted SIINFEKL specific CD8⁺ T cells if OVA antigen is released from the PV and is correctly processed by the host cell MHC-I antigen presentation machinery.

Irgm regulatory molecules such as Irgm1 and Irgm3 regulate the host cell IRG effector mechanisms that attack the PVM to clear PVs and parasites from infected host cells (16–18, 47). To investigate the role of IFN-γ priming, host cell Irgm1/Irgm3 molecules, and cell type in regulating the presentation of PV associated parasite antigens by MHC-I, wild type (WT) and Irgm1/m3 deficient (Irgm1/m3^−/−) bone marrow derived macrophages (BMMΦs) or bone marrow derived dendritic cells (BMDCs) were primed with IFN-γ, or left unprimed and then host cells were infected with RH or RH-OVA. Irgm1/m3^−/− BMMΦs or BMDCs lack functional IRG effector molecules. Presentation of the OVA SIINFEKL peptide by MHC-I on the host cell surface was measured using the sensitive and quantitative B3Z antigen presentation assay (43) in previously optimized assay conditions to determine the fold change (or fold-activation) in antigen presentation in host cells infected with RH-OVA compared to host cells infected with parental RH not expressing OVA antigen (3, 13). As expected, RH infected WT or Irgm1/m3^−/− BMMΦs or BMDCs induced only low background levels of B3Z CD8⁺ T cell activation (<1-fold change). In contrast, antigen presentation by RH-OVA infected WT BMMΦs primed with IFN-γ was markedly increased in comparison to unprimed WT BMMΦs, unprimed Irgm1/m3^−/− BMMΦs, as well as primed Irgm1/m3^−/− BMMΦs infected with RH-OVA parasites (Figure 1, top panel). However, in contrast to macrophages, antigen presentation was increased in both IFN-γ primed WT BMDCs as well as in IFN-γ primed Irgm1/m3^−/− BMDCs (Figure 1, bottom panel). These results show that presentation of PV localized OVA antigen by MHC-I is IFN-γ and Irgm1/m3 dependent in macrophages, and IFN-γ dependent but Irgm1/m3 independent in dendritic cells. Thus, IFN-γ was required for stimulating antigen presentation in parasite infected cells, whereas the role of Irgm1/m3 molecules in antigen presentation of parasite antigens was cell type dependent.

Since IFN-γ primed presentation of PV localized Toxoplasma antigens by MHC-I through both Irgm1/m3 dependent and Irgm1/m3 independent mechanisms (Figure 1), we hypothesized that the parasite may utilize a range of secreted effectors to limit antigen presentation by MHC-I. To address this hypothesis, we first identified candidate secreted Toxoplasma rhoptry (ROP) and dense granule (GRA) secreted proteins that have been previously localized to the PV, the PVM, the intravacuolar network (IVN) membranes within the PV lumen, or that have been linked to parasite resistance against host IFN-γ or IRG effectors. Knockouts of ROP5, ROP18, GRA2, GRA3, GRA4, GRA5, GRA6, GRA7, GRA8, and GRA12 were developed in the isogenic parental RH-OVA [Δku80] background. Several knockout strains were complemented with either the corresponding WT gene allele or with a gene allele possessing an engineered mutation (Figures S2, S3). All strains engineered in this study are listed in Table S1. OVA expression and secretion into the PV lumen was initially assessed by confocal microscopy, and as expected by virtue of sharing an identical single copy OVA gene, OVA accumulation in the PV appeared to be equivalent in each of these isogenic mutant strains (Figure S1D). In addition, we selectively examined the expression of PV localized OVA, compared to expression of PV localized GRA5, using a FACS based assay to verify that in a large population of Toxoplasma- infected macrophages (Figure S4A), or infected dendritic cells (Figure S4B), the expression of PV localized OVA was equivalent between the different isogenic mutant strains. Thus, under the antigen presentation assay conditions used here to measure the presentation of PV localized OVA antigen by MHC-I, the PVs between the different mutant strains expressed equivalent amounts of PV localized OVA antigen.

Given previous findings that Δrop5 PVs are easily physically disrupted by the action of Irgm1/m3 dependent host IRG effectors (20, 22, 23, 26, 48), we hypothesized that host cells infected with Δrop5 parasites may exhibit increased levels of antigen presentation by MHC-I molecules from the release of PV localized OVA antigen into the host cell cytosol following IRG-mediated destruction of the PV. To assess this hypothesis, we measured B3Z CD8⁺ T cell activation in unprimed or IFN-γ primed WT and Irgm1/m3^−/− BMMΦs and BMDCs infected with RH-OVA or Δrop5 parasites. Compared to RH-OVA, antigen presentation in cells infected with Δrop5 parasites was markedly increased in all conditions (Figure 2A). In contrast to antigen presentation by BMMΦs infected with parental RH-OVA that was dependent on both IFN-γ and Irgm1/m3, antigen presentation in BMMΦs infected with Δrop5 parasites was boosted by priming with IFN-γ, but was not strictly dependent on either IFN-γ or Irgm1/m3^−/− (Figure 2A). These results suggested that ROP5 controls two distinct pathways of antigen presentation by macrophages, one pathway is dependent on Irgm1/m3 molecules while the second pathway is independent of Irgm1/m3 molecules. In addition, IFN-γ priming of macrophages increases antigen presentation by both pathways. Remarkably, compared with BMMΦs, essentially identical patterns of Irgm1/m3 dependent and Irgm1/m3 independent antigen presentation were also observed in BMDCs infected with Δrop5 parasites (Figure 2B). Thus, the loss of ROP5 created identical patterns of antigen presentation in BMMΦs and BMDCs, whereas in cells infected with parental RH-OVA that expresses ROP5, antigen presentation by macrophages was dependent on Irgm1/m3 and IFN-γ, while antigen presentation by dendritic cells was dependent only on IFN-γ (Figures 1, 2).

The ROP5 locus in the virulent type I RH strain appears to express 2 copies of the ROP5A gene allele type, 2 copies of the ROP5B gene allele type, and 2 copies of the ROP5C gene allele type (22). All six of these ROP5 gene alleles were deleted in Δrop5 parasites. The ROP5C gene allele type was previously correlated with IRG evasion and virulence in mice (24). Therefore, we complemented Δrop5 parasites with a single copy of the ROP5C gene allele (Δrop5::ROP5C parasites) (Figure S2). Expression of the ROP5C gene allele partially rescued the Irgm1/m3 and IFN-γ dependent antigen presentation phenotype in BMMΦs infected with Δrop5::ROP5C parasites (Figure 2A). In contrast, antigen presentation levels were similar in WT and Irgm1/m3^−/− BMDCs infected with Δrop5::ROP5C parasites (Figure 2B), suggesting that the ROP5C allele substantially rescued the Irgm1/m3 and IFN-γ dependent antigen presentation phenotype in dendritic cells. These results also suggested that the ROP5C gene allele did not substantially rescue the Irgm1/m3 independent antigen presentation phenotype in macrophages or in dendritic cells.

To assess the role of Irgm1/m3 dependent IRG effectors and PV disruption in ROP5 mediated antigen presentation phenotypes, we measured IRG coating of the PVM and PV killing in IFN-γ primed BMMΦs and BMDCs infected with RH-OVA, Δrop5, or Δrop5::ROP5C parasites. Compared with parental RH-OVA, Irgb6 coating of the PVM was strikingly increased in IFN-γ primed WT BMMΦs (Figures 2C,D) and in IFN-γ primed WT BMDCs (Figures 2F,G) infected with Δrop5 parasites. BMDCs infected with Δrop5 parasites expressing a single ROP5C gene allele (Δrop5::ROP5C parasites) (Figure S2) resisted Irgb6 coating of the PVM as well as the parental RH-OVA strain (Figures 2F,G). In contrast, resistance to Irgb6 coating of the PVM was not rescued in BMMΦs infected with Δrop5::ROP5C parasites (Figures 2C,D). As expected, PV survival was not detected in WT BMMΦs infected with Δrop5 parasites (Figure 2E), though PV survival was partially rescued in WT BMMΦs infected with Δrop5::ROP5C parasites (Figure 2E). These results suggested that PV survival in IFN-γ primed macrophages required multiple ROP5 alleles to resist the IFN-γ and Irgm1/m3 dependent mechanisms of PV killing (Figure 2E), though a single ROP5C allele was sufficient for PV survival in dendritic cells (Figure 2H). In addition, these phenotypes correlated with reduced killing (~60%) of PVs in dendritic cells compared to killing (~100%) of PVs in macrophages infected with Δrop5 parasites. Compared to parental RH-OVA, killing of PVs was increased only ~3-fold in BMDCs infected with Δrop5 parasites, whereas killing of PVs was increased at least 20-fold in BMMΦs infected with Δrop5 parasites (Figures 2E,H). While PV killing was not detected in Irgm1/m3^−/− BMMΦs infected with Δrop5 parasites (Figure 2E), surprisingly, we found that ROP5 also regulated an Irgm1/m3 independent mechanism of PV killing in IFN-γ primed BMDCs (Figure 2H). Moreover, a single ROP5C gene allele rescued PV survival in IFN-γ primed Irgm1/m3^−/− BMDCs infected with Δrop5::ROP5C parasites (Figure 2H). Consistent with these ex vivo PV killing phenotypes, complementation of Δrop5 parasites with the ROP5C gene allele significantly rescued the acute virulence of Δrop5 parasites in mice (Figure S5). Collectively, these findings suggest that ROP5 limited antigen presentation in macrophages and dendritic cells by neutralizing Irgm1/m3 and IFN-γ dependent host IRG effectors to prevent PV disruption and the release of PV localized OVA antigen for presentation by host MHC-I molecules. ROP5 also blocked an Irgm1/m3 independent pathway for the presentation of PV localized antigens by MHC-I. This Irgm1/m3 independent pathway was boosted by priming host cells with IFN-γ, but was not significantly associated with PV clearance in macrophages or dendritic cells.

ROP5 is a central component of distinct ROP17 or ROP18 high molecular weight PVM associated protein complexes (25) where it regulates the ability of ROP18 to phosphorylate IRGs (23) to neutralize their PV killing effector functions (36). In contrast to Δrop5 parasites (Figure 2A, top panel), antigen presentation was not increased in unprimed WT or Irgm1/m3^−/− BMMΦs infected with Δrop18 parasites (Figure 3A, top panel). However, IFN-γ primed WT BMMΦs infected with Δrop18 parasites exhibited markedly increased antigen presentation compared to IFN-γ primed Irgm1/m3^−/− BMMΦs (Figure 3A, bottom panel). Together, these results suggested that ROP18 selectively suppressed an Irgm1/m3 and IFN-γ dependent pathway of antigen presentation in macrophages. Consistent with these results, Irgb6 coating of the PVM (Figures 3B,C) and PV killing (Figure 3D) was increased in IFN-γ primed WT BMMΦs infected with Δrop18 parasites. Complementation of Δrop18 parasites with the WT ROP18 gene allele (Δrop18::ROP18) (Figure S3) rescued these Irgm1/m3 and IFN-γ dependent antigen presentation, Irgb6 coating, and PV killing phenotypes (Figures 3A–D). In contrast, complementation of Δrop18 parasites with ROP18 mutants possessing a non-functional kinase-dead (KD) domain (ROP18KD) that fails to phosphorylate IRGs (36), or with a ROP18 mutant with a deletion of the N-terminal ATFβ6 domain (ROP18ATF6β) (Figure S3) that mediates ROP18 association to the host ATFβ6 sensor (37), as well as to the PVM (49, 50), failed to fully rescue these phenotypes (Figures 3A–D). Thus, the Irgm1/m3 and IFN-γ dependent ROP18 mediated suppression of antigen presentation in macrophages was dependent on the ROP18 kinase activity and the N-terminal ATFβ6 domain of ROP18. Collectively, these results suggested that ROP18 limited antigen presentation in IFN-γ primed macrophages primarily through its Irgm1/m3 dependent interaction with IRG effector proteins to prevent PV disruption and the release of PV localized OVA antigen.

In contrast to unprimed WT and Irgm1/m3^−/− macrophages infected with Δrop18 parasites, a significant increase in antigen presentation was observed in unprimed WT and Irgm1/m3^−/− BMDCs infected with Δrop18 parasites (Figure 3E, top panel). Moreover, in contrast to IFN-γ primed WT and Irgm1/m3^−/− macrophages (Figure 3A, bottom panel), the levels of antigen presentation were similar between IFN-γ primed WT and Irgm1/m3^−/− BMDCs infected with Δrop18 parasites (Figure 3E, bottom panel). This Irgm1/m3 independent pathway of increased antigen presentation in BMDCs infected with Δrop18 parasites was rescued by the WT allele of ROP18 (Δrop18::ROP18) (Figure 3E, bottom panel). Together, these results suggested that ROP18 suppressed antigen presentation in dendritic cells primarily through a mechanism that was not dependent on Irgm1/m3 molecules. Consistent with these results, the absence of ROP18 did not affect Irgb6 coating of the PVM in BMDCs (Figures 3F,G) and only weakly influenced PV killing (Figure 3H). In addition, ROP18's ability to suppress this Irgm1/m3 independent pathway of antigen presentation was significantly rescued in unprimed as well as in IFN-γ primed Irgm1/m3^−/− BMDCs infected with kinase-dead Δrop18::ROP18KD parasites, or infected with Δrop18::ROP18ATF6β parasites (Figure 3E). These results suggested that the kinase activity and the ATF6β domain of ROP18 were not strictly necessary for ROP18's ability to suppress the Irgm1/m3 independent mechanism of antigen presentation in dendritic cells.

To determine whether other PVM localized proteins also influenced antigen presentation, we evaluated the role GRA3, GRA5, and GRA8, three other secreted dense granule proteins that selectively localize to the PVM (51). GRA8 played no detectable role in regulating antigen presentation in macrophages or dendritic cells (Figures 4A,B). Antigen presentation was selectively, and slightly, increased in unprimed WT BMDC infected with Δgra5 parasites (Figure 4A). In contrast, antigen presentation was markedly increased in unprimed and in IFN-γ primed WT and Irgm1/m3^−/− BMMΦs and BMDCs infected with Δgra3 parasites (Figures 4A,B). Remarkably, the increases in antigen presentation in unprimed and in IFN-γ primed WT and Irgm1/m3^−/− BMDCs infected with Δgra3 parasites were similar in magnitude (Figure 4B). These results suggested that GRA3 regulated a mechanism of antigen presentation that was not specifically dependent on Irgm1/m3 molecules, IFN-γ, or cell type.

The PVM localized GRA7 molecule associates with ROP5/ROP18 high molecular weight protein complexes (25). Consistent with this localization, GRA7 was previously shown to coordinate its activities with the ROP18 kinase to enhance resistance to IRG mediated clearance in macrophages (52, 53). Consistent with this role for GRA7 and our hypothesis that PV disruption increases antigen presentation of PV localized antigens by MHC-I, antigen presentation was unaffected in unprimed BMMΦs and was markedly increased in IFN-γ primed WT BMMΦs, indicating that GRA7 suppressed a major IFN-γ and IRG dependent pathway of antigen presentation in macrophages. This GRA7 phenotype in BMMΦs was similar to, though slightly reduced in magnitude, in comparison to BMMΦs infected with Δrop18 parasites (Figure 3A). In addition, minor increases in antigen presentation were observed in primed Irgm1/m3^−/− BMMΦs (Figure 4A), and in unprimed Irgm1/m3^−/− BMDCs (Figure 4B) infected with Δgra7 parasites. As expected, Δgra7 PVs exhibited increased coating with Irgb6 in IFN-γ primed BMMΦs (Figures 4C,D), and consistent with this observation, increased Irgm1/m3 dependent killing of the Δgra7 PV (Figure 4E). Collectively, these results suggested that GRA7 limited antigen presentation primarily in IFN-γ primed macrophages through its Irgm1/m3 dependent interaction with IRG effectors to prevent PV disruption and the release of PV localized OVA antigen for presentation by MHC-I.

In addition to GRA7, the GRA12 molecule was also previously shown to be selectively associated with high molecular weight ROP5 and ROP18 PVM localized protein complexes (25). Moreover, GRA12 was previously localized to a nanotubular membrane system in the PV (54) that is called the nanotubular membranous network, or alternatively, the intravacuolar network (IVN). GRA2, GRA4, GRA6, and GRA12 localize to the membranes that comprise the intravacuolar network (IVN) (51). GRA6 is required for the maintenance of IVN membrane structures in the PV, while GRA2 is required for the assembly of the membranous nanotubules that comprise the IVN (55). GRA2 was previously reported to influence IRG association with the PVM association (24, 52), and GRA2 has also been reported to suppress the presentation of the endogenous membrane-bound GRA6 C-terminal epitope by MHC-I (42). To examine the role of IVN localized GRA proteins in regulating antigen presentation of the soluble OVA PV localized antigen, we engineered several isogenic OVA-expressing GRA knockout strains (Δgra2, Δgra2::GRA2, Δgra2Δgra4, Δgra2Δgra6, Δgra4, Δgra6, and Δgra12) (Figure S1). Antigen presentation was increased in unprimed and IFN-γ primed WT and Irgm1/m3^−/− BMMΦs and BMDCs infected with Δgra12 parasites. Antigen presentation levels in unprimed or IFN-γ primed WT and unprimed or IFN-γ primed Irgm1/m3^−/− BMMΦs and BMDCs infected with Δgra12 parasites were similar (Figures 5A,E). These results suggested that GRA12 regulated a mechanism of antigen presentation that was not dependent on Irgm1/m3 molecules, IFN-γ, or cell type.

Similar and significant increases in antigen presentation were observed in unprimed and IFN-γ primed WT or Irgm1/m3^−/− BMMΦs infected with Δgra2 or Δgra2Δgra4 parasites (Figure 5A). In addition, increased antigen presentation was observed in IFN-γ primed WT BMMΦs compared with IFN-γ primed Irgm1/m3^−/− BMMΦs infected with Δgra2 or Δgra2Δgra4 parasites (Figure 5A). In contrast, no increase in antigen presentation was observed in BMMΦs infected with Δgra4 parasites (Figure 5A), suggesting that it was the absence of GRA2 that was most important for the antigen presentation phenotype in BMMΦs infected with Δgra2Δgra4 parasites. Complementation of Δgra2 parasites with the WT GRA2 gene allele inserted into the UPRT locus significantly rescued these Δgra2 antigen presentation phenotypes in BMMΦs (Figure 5A). These results suggested that GRA2 regulated Irgm1/m3^−/− dependent and Irgm1/m3 independent mechanisms of antigen presentation in BMMΦs infected by Δgra2 or Δgra2Δgra4 parasites. However, in contrast to macrophages infected with Δrop5 (Figures 2C–E), Δrop18 (Figures 3B–D), or Δgra7 parasites (Figure 4E), increased Irgb6 coating of the PVM (Figures 5B,C) or increased PV killing were not observed in IFN-γ primed WT BMMΦs infected with Δgra2Δgra4 parasites (Figure 5D). These results suggested that deletion of GRA2 increased Irgm1/m3 and IFN-γ dependent antigen presentation in BMMΦs without any associated increase in IRG mediated PV clearance.

Antigen presentation levels were similar between IFN-γ primed WT BMDCs and IFN-γ primed Irgm1/m3^−/− BMDCs infected with Δgra2 or Δgra2Δgra4 parasites (Figure 5E). In contrast, only small increases in antigen presentation were observed in BMDCs infected with Δgra4 parasites (Figure 5E), suggesting that it was the absence of GRA2 that was necessary for the antigen presentation phenotype in BMDCs infected with Δgra2Δgra4 parasites. These results suggest that GRA2 regulated an Irgm1/m3 independent mechanism of antigen presentation in dendritic cells. Complementation of Δgra2 parasites with the WT GRA2 gene allele rescued the Irgm1/m3 independent antigen presentation phenotype in BMDCs (Figure 5E). No increase in Irgb6 coating of the PVM was observed in BMDCs infected with Δgra2 or Δgra2Δgra4 parasites (Figures 5F,G).

GRA2 was previously reported to limit CD8⁺ T cell recognition of the membrane-bound immunodominant HF10 epitope present at the C-terminus of the GRA6 antigen (42). This interpretation of GRA2-associated antigen presentation phenotypes was based on the assumption that the GRA6 molecule itself did not influence antigen presentation by MHC-I (42). Thus, to formally eliminate any potential role of the GRA6 molecule in affecting the antigen presentation phenotypes of mutants deleted for GRA2, we developed Δgra6 parasites as well as double knockout Δgra2Δgra6 parasites (Figure S1). No significant antigen presentation phenotype was observed in BMMΦs infected with Δgra6 parasites (Figure 5A), though antigen presentation was slightly increased in unprimed or IFN-γ primed WT or Irgm1/m3^−/− BMDCs infected with Δgra6 parasites (Figure 5E). However, the absence of both GRA6 and GRA2 molecules in Δgra2Δgra6 parasites markedly decreased or abolished the antigen presentation phenotypes observed in BMMΦs and BMDCs infected with Δgra2 or Δgra2Δgra4 parasites (Figures 5A,E). These results revealed that the expression of the GRA6 protein was required to observe the major increase in antigen presentation by MHC-I in host cells infected with Δgra2 parasite strains.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.