We have previously shown that ATG16L1 plays an important role in the IFN-γ-induced reduction of type II T. gondii (ME49) in mouse cells (mouse embryonic fibroblasts; MEFs) but not in human HAP1 cells (Figure 1A), suggesting an ATG16L1-independent IFN-γ-induced anti-T. gondii response in human cells (12). To elucidate the molecular mechanism, we next challenged MEFs and HAP1 cells with type I (RH) or type II (ME49) parasites. As shown previously (12), MEFs showed more efficient IFN-γ-induced reduction of type II parasites than of type I parasites. By contrast, IFN-γ stimulation could similarly and efficiently reduce the numbers of both type I and II parasites in HAP1 cells (Figure 1B). IFN-γ-induced degradation of arginine by iNOS, or of tryptophan by IDO, which consists of two members, IDO1 and IDO2, have been shown to be important for the anti-T. gondii response in mouse or human cells (18, 30, 31). To test whether iNOS or IDO (or both) are involved in the process in HAP1 cells, we examined iNOS, IDO1, or IDO2 mRNA expression in HAP1 cells (Figure 1C). Stimulation of IFN-γ led to strong induction of IDO1 and IDO2 mRNAs and weak induction of iNOS mRNA. Second, we tested the expression levels of these mRNAs in IFN-γ-stimulated cells followed by T. gondii infection (Figure 1C). T. gondii infection at 24 hours after IFN-γ stimulation did not interfere with the expression of iNOS, IDO1, and IDO2 mRNAs in HAP1 cells (Figure 1C). Furthermore, we treated HAP1 cells with a pharmacological inhibitor of iNOS known as aminoguanidine or an inhibitor of IDO known as 1-methyl-DL-tryptophan (1-DL-MT), and compared the parasite numbers. 1-DL-MT but not aminoguanidine treatment abolished the IFN-γ-induced reduction of T. gondii numbers in HAP1 cells (Figure 1D), strongly suggesting the anti-T. gondii function of IDO in the human HAP1 cell line.

Both IDO1 and IDO2 could be inhibited by 1-DL-MT (35, 36). Although both IDO1 and IDO2 mRNAs were highly induced by IFN-γ stimulation in HAP1 cells (Figure 1C), it remained to be seen which was more important for the IFN-γ-induced response in HAP1 cells. To clarify the contribution of IDOs in HAP1 cells, we generated IDO1 singly deficient (IDO1-KO), IDO2 singly deficient (IDO2-KO), and doubly deficient (IDO1/IDO2-DKO) HAP1 cells by CRISPR/Cas9 genome editing (Figures 2A,B, Figure S1A) and tested the IFN-γ-induced reduction of type II parasite numbers. Although IDO2-KO HAP1 cells functioned normally, IDO1-KO cells as well as IDO1/IDO2-DKO cells were completely defective in IFN-γ-induced parasite reduction (Figure 2C), suggesting that IDO1 but not IDO2 is essential for the IFN-γ-induced anti-T. gondii response in the human cell line. Kynurenine is a tryptophan metabolite of IDOs (50). Therefore, we measured kynurenine concentrations in HAP1 cells lacking IDOs (Figure S1B). Whereas, the kynurenine concentrations were increased upon IFN-γ treatment in wild-type HAP1 cells, such an increment was not observed in IDO1-KO or IDO1/IDO2-DKO HAP1 cells (Figure S1B). By contrast, IDO2-KO cells showed normal induction of kynurenine after IFN-γ stimulation (Figure S1B), correlating with the importance of IDO1 in the IFN-γ-induced reduction of parasite numbers and the degree of tryptophan degradation. In mouse cells, a T. gondii strain-dependent difference was observed in the IFN-γ-induced anti-T. gondii response (Figure 1B) (12, 13). By contrast, although wild-type cells exhibited greatly reduced numbers of type I parasites after IFN-γ stimulation, this IFN-γ-mediated reduction was not observed in IDO1-KO HAP1 cells (Figure S1C), suggesting the lack of strain dependence in this human cell line. Next, we analyzed whether IDO1 expression could rescue the defective anti-T. gondii response of IDO1-KO cells. To achieve this, IDO1-KO cells were transfected with a doxycycline-inducible IDO1 or the empty control vectors and the IFN-γ-induced anti-T. gondii response was tested (Figure 2D, Figure S1D). The doxycycline-inducible IDO1 expression led to a reduction of parasite numbers in the IFN-γ-stimulated IDO1-KO cells (Figure 2D). Next we tested whether IDO1 plays a role in the inhibition of T. gondii replication or in parasite elimination. The parasite numbers per vacuole in IFN-γ-stimulated IDO1-KO HAP1 cells were significantly higher than those of wild-type cells (Figures 2E,F). Whereas, the rates of T. gondii-infected wild-type or IDO1-KO cells were comparable 3 and 24 h post infection (Figures 2E,G), indicating that IDO1 inhibits T. gondii replication in IFN-γ-stimulated HAP1 cells. Taken together, these data suggest that IDO1 plays a critical role in the IFN-γ-induced anti-T. gondii response in HAP1 cells.

Next we assessed whether IDO1 is important for the IFN-γ-induced anti-T. gondii response in other human cells. IDO1 mRNAs were highly induced in foreskin fibroblasts (HFFs), a hepatocyte cell line (Huh7), and an epithelial cell line (HeLa) upon IFN-γ stimulation regardless of the subsequent T. gondii infection (Figure 3A). Then we generated IDO1-KO HFFs, Huh7, or HeLa cells by CRISPR/Cas9-genome editing (Figure 3B, Figure S2A) and analyzed the IFN-γ-induced reduction in T. gondii numbers. Among all of the cell types tested, IDO1-KO cells were defective in the IFN-γ-mediated anti-T. gondii response (Figure 3C). However, compared with IDO1-KO HAP1 cells or HFFs, both of which displayed complete loss of an anti-parasite response (Figures 2C, 3C), IDO1 deficiency in Huh7 or HeLa cells resulted in severely impaired or modest defects (Figure 3C), suggesting an IDO1-independent anti-T. gondii response in Huh7 and HeLa cells. In addition, Tryptophan 2,3-dioxygenase (TDO) was not involved to IFN-γ induced anti- T. gondii responses in Huh7 cells (Figure S2B).

A previous study demonstrated that ATG16L1 is involved in the IFN-γ-induced anti-T. gondii response in HeLa cells (23). Therefore, we hypothesized that ATG16L1 plays a role in the IDO1-independent response in Huh7 and HeLa cells. To examine this possibility, we generated ATG16L1-deficient (ATG16L1-KO) HFFs, Huh7, and HeLa cells by CRISPR/Cas9 genome editing (Figure 4A). Although ATG16L1-KO HFFs and Huh7 cells showed no increment of parasite numbers (Figure 4B) or parasite numbers per vacuole upon IFN-γ stimulation (Figures 4C,D), ATG16L1-KO HeLa cells exhibited modest defects (Figures 4C,D). Furthermore, 1-DL-MT treatment in ATG16L1-KO HeLa cells resulted in a less efficient IFN-γ-mediated response than in non-treated control cells (Figure S2C), suggesting that ATG16L1 is involved in the IFN-γ-induced anti-parasite response in a cell type-specific manner. Thus, IDO1 is involved in the IFN-γ-induced anti-T. gondii response in human cells of various origins (Figure 6A). However, given the dispensable role of IDO in HUVECs (51), the degree of importance of IDO1 in the anti-T. gondii response depends on cell type.

Since IDO1 plays a critical role in the IFN-γ-induced anti-T. gondii response in various human cells, we next explored the possible T. gondii virulence mechanisms targeting IDO1 in human cells. We selected TgIST, a T. gondii secreting effector molecule, as the candidate (47, 48), since the regulation of IDO1 expression by IFN-γ was shown to be regulated by STAT1 (52). To assess this possibility, we generated TgIST-KO type II parasites by CRISPR/Cas9 genome editing (Figures S3A,B) and asked whether TgIST deficiency affects T. gondii virulence in human cells (Figure 5A). As previously reported (47, 48), we also confirmed that TgIST-dependent suppression of the anti-T. gondii effect could be detected only when T. gondii-infected cells were subsequently stimulated with IFN-γ but not when IFN-γ-pre-treated cells were followed by T. gondii infection (Figure 5A, Figure S4A). By contrast, we observed an IFN-γ-dependent reduction in TgIST-KO parasite numbers (Figure 5A), suggesting that TgIST promotes parasite growth in the IFN-γ-post-stimulated cells. Phosphorylation of the Y701 residue of STAT1 (STAT1 Y701-p) was previously shown to be induced by T. gondii infection without IFN-γ stimulation and STAT1 Y701-p proteins were translocated to the nucleus (48). TgIST directly binds to STAT1 and recruits the chromatin-modifying Mi-2/NuRD complex to STAT1 transcriptional complexes. As a result, chromatin is remodeled and interferon-stimulated gene expression, including that of IRF-1, is decreased (47, 48). We tested whether T. gondii infection inhibits STAT1-dependent transcription in IFN-γ-post-treated HFF, Huh7, or HAP1 cells in a TgIST-dependent manner. Although phospho-STAT1 (STAT1 Y701-p) proteins were not detected in the nucleus in unstimulated cells, STAT1 Y701-p proteins were translocated to the nucleus upon IFN-γ-treatment. As previously reported (47, 48), wild-type T. gondii infection caused STAT1 Y701-p nuclear translocation even in unstimulated cells (Figure 5B). The subsequent IFN-γ-stimulation further induced translocation of STAT1 Y701-p to the nucleus (Figure 5B); however, STAT1 Y701-p was not active since STAT1-regulated gene products such as IRF1 and IDO1 mRNAs and proteins were not induced (Figure 5C, Figures S4B,C). By contrast, infection of TgIST-KO T. gondii did not result in STAT1 Y701-p nuclear translocation in unstimulated cells. Furthermore, the subsequent IFN-γ stimulation in TgIST-KO parasite-infected cells normally induced STAT1 Y701-p nuclear translocation in comparison with uninfected cells. Moreover, normal levels of IFN-γ-induced IRF1 and IDO1 expression suggested that the STAT1 activity in TgIST-KO T. gondii-infected cells was normal (Figure 5C, Figures S4B,C). Furthermore, IDO1-KO cells did not exhibit an IFN-γ-induced reduction in TgIST-KO parasites (Figure 5D). Conversely, growth of TgIST-intact wild-type T. gondii as well as TgIST-KO parasite growth was strongly inhibited by doxycycline-induced (thereby, STAT1-independent) IDO1 overexpression (Figure 5E), indicating that the pro-parasitic role of TgIST for T. gondii growth in IFN-γ-post-stimulated human cells is mainly due to inhibition of IDO1 but not other IFN-γ-inducible proteins (Figure 6B).

All T. gondii strains were maintained in Vero cells in RPMI (Nacalai Tesque) supplemented with 2% heat-inactivated FBS (JRH Bioscience), 100 U/ml penicillin, and 0.1 mg/ml streptomycin (Nacalai Tesque), as previously described (55). HAP1 cells were maintained in IMDM (Nacalai Tesque) containing 10% heat-inactivated FBS, and 100 U/ml penicillin, and 0.1 mg/ml streptomycin. HFFs, Huh7 cells were maintained in RPMI (Nacalai Tesque) containing 10% heat-inactivated FBS, and 100 U/ml penicillin, and 0.1 mg/ml streptomycin. HeLa cells and MEF cells were maintained in DMEM (Nacalai Tesque) containing 10% heat-inactivated FBS, 100 U/ml penicillin, and 0.1 mg/ml streptomycin.

Antibodies against IDO1 (13268-1-AP), HDAC1 (10197-1-AP), and IRF1 (11335-1-AP) was obtained from Proteintech. Antibodies against Phospho-Stat1 (Tyr701) (#9167) was obtained from Cell Signaling. Antibodies against ATG16L1 (PM040) was obtained from MBL. Anti-β-actin antibody (A1978) was obtained from Sigma. Antibodies against GAP45 was described previously (56). Recombinant human and mouse IFN-γ were obtained from Peprotech. 1-Methyl-DL-tryptophan (sc-224746) was obtained from Santa Cruz Biotechnology, Inc. Aminoguanidine hydrochloride (396494) was obtained from Sigma.

All genomic deficient cell lines were generated with the px330 plasmid CRISPR/Cas9 system. The insert fragment of IDO1, IDO2 ATG16L1 gRNA were generated by annealing primers. All the primers used in this study are listed in Table S1. These insert fragments were inserted into the BbsI site of the cloning vector containing U6 promoter to generate gRNA expressing plasmids pIDO1_gRNA1, pIDO1_gRNA2, pIDO2_gRNA1, pIDO2_gRNA2, pATG16L1_gRNA1, and pATG16L1_gRNA2, respectively. The insert fragment was cut out by XhoI and SalI from the pIDO1_gRNA2, pIDO2_gRNA2, and pATG16L1_gRNA2 vector, and ligated into the XhoI site of the pIDO1_gRNA1, pIDO2_gRNA1, and pATG16L1_gRNA1 vector to generate plasmids pIDO1_gRNA1/2, pIDO2_gRNA1/2, and pATG16L1_gRNA1/2. The insert fragment was cut out by KpnI and MluI from pIDO1_gRNA1/2, pIDO2_gRNA1/2, and pATG16L1_gRNA1/2 vector, respectively, and ligated into the KpnI and MluI site of the pEF6-hCas9-Puro vector.

Human cells were electroporated with the pEF6-hCas9-Puro vector containing target gRNA1/2 using NEPA21 (nepa gene). And then 24 h post-electroporation, 0.5–5 μg/ml puromycin was added for 5–10 days to select for cells with a stably integrated. Cells were plated in limiting dilution in 96-well plates to isolate single cell clones. To confirm complete target gene deficient, the IDO1, IDO2 and ATG16L1 protein expression were analyzed by Western blot. Atg16L1 KO MEF cells were described previously (12).

Plasmid pSAG1::Cas9-U6::sgUPRT that encoding Cas9 nuclease (GFP fusion) under control of the T. gondii SAG1 promotor was obtained from addgene (plasmid 54467). The primer sequences are listed in Table S1. The TgIST-targeting CRISPR/Cas9 plasmid (pSAG1::Cas9-U6::sgTgIST-1 or pSAG1::Cas9-U6::sgTgIST-2) was constructed in two steps. First, the overlap PCR method was used to generate gRNA expressing plasmids. The U6 promoter from ME49 driving expression of the TgIST specific sgRNA (pgTgIST-1 or pgTgIST-2) was amplified from pSAG1::Cas9-U6::sgUPRT. For first-step PCR, primer pairs TgU6_F and TgISTgRNA1-R, TgISTgRNA1-F and TgU6_R, TgU6_F and TgISTgRNA2-R, TgISTgRNA2-F and TgU6_R were used. For second-step PCR primer pairs TgU6_F and TgU6_R were used, and cloning it into the NotI and SacI sites of the pBluescript II SK(+) (plasmid 54467). Second, the Cas9-Ds-Red monomer cassette under the SAG1 promoter was cut out from pSAG1::Cas9-U6::sgUPRT (Ds-Red monomer fusion), and ligated into the KpnI and NotI site of the pgTgIST-1 or pgTgIST-2. To generate a construct for deleting the entire coding sequence of TgIST, flanking regions of 5′ outside the sgTgIST-1 and 3′ outside the sgTgIST-2 regions were used to surround the TgIST cassette. To generate a plasmid for inserting HXGPRT into the TgIST gene, upstream regions of sgTgIST-1 (894-bp) and downstream regions of sgTgIST-2 (670-bp) were amplified from ME49 genomic DNA by using primers TgIST targeting 5′_F and TgIST targeting 5′_R or TgIST targeting 3′_F and TgIST targeting 3′_R. These two fragments were ligated into the KpnI and XhoI or BamHI and NotI sites of pHXGPRT vector.

Prugniaud (Pru) T. gondii-expressing luciferase were filtered, washed and resuspended in Cytomix (10 mM KPO₄, 120 mM KCl, 0.15 mM CaCl₂, 5 mM MgCl₂, 25 mM HEPES, 2 mM EDTA). Parasites were mixed with 50 μg of sgTgIST-1 and sgTgIST-2 CRISPR plasmid along with 40 μg of the targeting vector linearized by Kpnl and SacI, and supplemented with 2 mM ATP, 5 mM GSH. Parasites were electroporated by GENE PULSER II (Bio-Rad Laboratories). Selection by growth for 14 days in 25 μg/ml mycophenolic acid (Sigma) and 25 μg/ml xanthine (Wako) were used to obtain stably resistant clone. And then parasites were plated in limiting dilution in 96-well plates to isolate single clones. To confirm the disruption of the gene encoding TgIST, we analyzed messenger RNA of TgIST from WT and TgIST-KO parasites by quantitative RT-PCR. In addition, we observed comparable in vitro growth and in vivo virulence to each other and to the parental line.

Total RNA was extracted, and cDNA was synthesized using Verso Reverse transcription (Thermo Fisher Scientic). Quantitative RT-PCR was performed with a CFX connect real-time PCR system (Bio-Rad Laboratories) using the Go-Taq Real-Time PCR system (Promega). The values were normalized to the amount of beta actin (β-actin) for human cells or tubulin for T. gondii in each sample. The primer sequences are listed in Table S1.

Cells were lysed in a lysis buffer (0.5% Nonidet P-40, 150 mM NaCl, and 20 mM Tris-HCl, pH 7.5) containing a protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail (Nacalai Tesque). The cell lysates were separated by SDS-PAGE and transferred to polyvinylidene uoride membranes and subjected to Western blot analysis using the indicated antibodies as described previously (57).

For the experiment using IFN-γ pre-stimulated cells, HFFs (6 × 10⁵), HAP1 (2 × 10⁶), HeLa (8 × 10⁵), or Huh7 (6 × 10⁵) cells were untreated or treated with 10 ng/ml IFN-γ for 24 h, and subsequently infected with the luciferase-expressing T. gondii (MOI = 0.5) for 24 h. For the experiment using IFN-γ post-stimulation cells, HFFs (6 × 10⁵), HAP1 (2 × 10⁶), HeLa (8 × 10⁵), or Huh7 (6 × 10⁵) cells were infected with the luciferase-expressing T. gondii (MOI = 0.5) for 8 h, and subsequently untreated or treated with 10 ng/ml IFN-γ for 24 h. To measure the number of T. gondii, all infected cells were collected and lysed by 100 μl of lysis buffer (Promega) and sonicated. After centrifugation at 20,000 × g at 4°C, the luciferase activity of the supernatant was measured using the Dual Luciferase Reporter Assay System (Promega) and GLOMAX 20/20 luminometer (Promega). The percentages of the activities in cytokines stimulated cells over those in unstimulated cells were shown as “Relative T. gondii numbers” in figures.

HFFs, Huh7, or HeLa cells were cultured on glass coverslips, and infected with T. gondii (MOI = 2) for the indicated time, and fixed in PBS containing 3.7% paraformaldehyde for 10 min at room temperature. Cells were permeabilized with PBS containing 0.002% Digitonin for 5 min and then blocked with 8% FBS in PBS for 1 h at room temperature. And then, cells were incubated with the indicated primary antibodies for 1 h at 37°C, followed by incubation with Alexa 488-, Alexa 594-, or Alexa 647-conjugated secondary antibodies (Molecular Probes) and DAPI for 1 h at 37°C in the dark. Finally, coverslips were mounted onto glass slides with PermaFluor (Thermo Scientific) and analyzed using confocal laser microscopy (FV1200 IX-83, Olympus).

The enzymatic IDO activity was evaluated by the calculation of the kynurenine concentration in the cell culture supernatant as previously described (58). Cells were cultured in 12-well plates and untreated or treated with 10 ng/ml IFN-γ for 24 h. The concentration of kynurenine in culture supernatant was measured using Ehrlich reagent method (59). Seventy microliters of culture supernatant was mixed with 35 μl of 30% trichloroacetic acid, and centrifuged at 8,000 x g for 5 min. Then 75 μl of the supernatant was added to an equal volume of Ehrlich reagent (0.8% p-dimethylaminobenzaldehyde in acetic acid) in a 96-well plate, and the absorbance was read at 490 nm. The values were compared with a standard curve with defined concentrations of kynurenine (Sigma Aldrich).

All statistical analyses were performed using Excel (Microsoft). All the experimental points and n values represent an average of each three biological replicates (three independent experiments). The statistical significance of differences in mean values was analyzed by using an unpaired two-tailed Student's t-test. P-values less than 0.05 were considered to be statistically significant.

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.