To study T cell intrinsic STAT1 signaling, we generated Stat1 ^T-KO mice which lack STAT1 selectively in CD4⁺ and CD8⁺ T cells starting at the double-positive stage of thymic development. Prior to infection, total CD4⁺ and CD8⁺ T cells counts were comparable to WT littermates, although, consistent with studies in Stat1^-/- mice (36), effector and memory cells were increased, as were FOXP3⁺ Treg ( Figures S1A–G ). Crucially, baseline frequencies of IFN-γ producing CD4⁺ and CD8⁺ T cells were also comparable to WT controls ( Figure S1I ).

Strikingly, upon challenge with T. gondii, Stat1 ^T-KO mice were able to control parasite replication but invariably succumbed 1-2 weeks post infection ( Figures 1A–C ). This was in stark contrast to WT controls, which typically survived to chronic disease (>1 month), and Stat1^-/- mice, which had rampant parasites ( Figure 1C ). We also noted histological lesions in spleen, liver and lungs reminiscent of Il10-/- and Il27ra-/- mice infected with T. gondii ( Figure S2A ) (3, 4). Total splenic CD4⁺ and CD8⁺ T cells were respectively lower and higher in Stat1 ^T-KO mice ( Figure 1D ), and both lineages were sharply reduced at the site of infection (i.e. peritoneum), likely reflecting efficient parasite clearance ( Figure 1D ). Consistent with this latter point, serum levels of IFN-γ and IL-12p40, key cytokines driving anti-Toxoplasma responses (1, 2, 37), were substantially higher in Stat1 ^T-KO mice that in WT controls ( Figure S2B ). Taken together, these data establish that T cell intrinsic STAT1 signaling is required for resistance to T. gondii and point to anti-inflammatory properties as most relevant.

To further assess T cell responses, we made single cell suspensions from spleens and peritoneal exudates (PEx) at seven days post-infection, then measured IFN-γ production in response to soluble Toxoplasma Antigen (sTAg) or agonist anti-T cell receptor antibodies (anti-CD3ϵ). IFN-γ is produced by both CD4⁺ Th1 and CD8⁺ Tc1 effector T cells, and strictly required for resistance to T. gondii (2). It is also a key STAT1 stimulus and, like STAT1, known to have anti-inflammatory properties (38). Frequencies of IFN-γ⁺ CD4⁺ and IFN-γ⁺ CD8+ T cells were comparable between WT and Stat1 ^T-KO mice at seven days post-infection with one notable exception: splenic CD4⁺ T cells re-stimulated with anti-CD3ϵ ( Figure 1E and Figure S3A ). However, it should also be noted that due to differences in overall cellularity ( Figure 1D ), there were always fewer total IFN-γ⁺ CD4⁺ and more total IFN-γ⁺ CD8⁺ present in Stat1 ^T-KO cultures, regardless of re-stimulation conditions ( Figures S3B, C ). Nevertheless, it is clear that despite a key role in vitro and other in vivo settings (22, 29), T cell intrinsic STAT1 signaling is not strictly required for IFN-γ production during acute toxoplasmosis.

T-bet is a key driver of T cell interferon production and Th1 differentiation (21). Given that T-bet is directly induced by STAT1 (22, 23) and required for resistance to T. gondii (27), we quantified T-bet⁺ IFN-γ⁺ T cells at day 7 post-infection. This analysis revealed a clear difference between the CD4⁺ and CD8⁺ compartments; CD4⁺ T-bet⁺ Th1 cells were significantly reduced in Stat1 ^T-KO mice but CD8⁺ T-bet⁺ Tc1 cells were not ( Figures 1F, G and Figure S3A ). Thus, we conclude that STAT1 is required for optimal Th1-type responses but dispensable for parallel Tc1-type responses. However, we also emphasize that T-bet⁺ IFN-γ⁺ CD4⁺ T cells were still plainly evident in Stat1 ^T-KO mice ( Figure 1H ), suggesting that STAT1-independent mechanisms are also in play, as previously suggested (28).

STAT1 both inhibits de novo Treg differentiation (39) and induces expression of T-bet in mature Treg, thereby enhancing their capacity to suppress Th1- and Tc1-type responses (9, 40, 41). Both properties were evident in Stat1 ^T-KO mice following infection with T. gondii. The overall frequency of CD4⁺ Foxp3⁺ Treg was increased relative to WT controls while, at the same time, the proportion of those expressing T-bet was sharply reduced ( Figures 1I, J ). Given that T-bet^-/- Treg are defective in the context of T. gondii infection (42), we interpret that a lack of T-bet⁺ Treg likely contributes to emergence of pathogenic T cell responses in Stat1 ^T-KO mice.

Considering that Stat1 ^T-KO mice generate robust Th1 and Tc1 responses when challenged with T. gondii, we reasoned that STAT1 deficiency may unleash pathogenic aspects IFN-γ-producing T cells. To engage this hypothesis, we sorted IFN-γ producing CD4⁺ T cells from infected WT and Stat1 ^T-KO mice and compared transcriptomes by RNA-seq ( Figure 2A and Figure S4A ). Downstream analysis uncovered >1000 Differentially Expressed Genes (DEG), 74% of which were higher expressed in STAT1-deficient cells ( Figure 2A and Table S1). Next, we tested positive and negative DEG for biological pathway enrichment against the KEGG database. Predictably, almost all enrichment was contained within the larger negative fraction (i.e. DEG that were higher expressed in Stat1 ^T-KO ), with T cell and Toxoplasma related pathways prominently represented ( Figures 2B , S4B, C and Table S2 ).

The surprise came when we inspected the ‘hits’ within these pathways and found several genes involved with Th2- and Th17-type responses ( Figure 2B ). This was remarkable because Th2- and Th17-type cells are typically scarce during acute toxoplasmosis. Two Th2-related hits, ll4ra and ll13, were especially notable because they were substantially expressed (TMP >10; Figure 2B ). The rest were far less abundant, typically < 5 TPM ( Figure 2B ). Flow cytometry confirmed that IFN-γ⁺ IL-13⁺ double-positive CD4⁺ T cells were readily detectable in Stat1 ^T-KO mice at day 7 post-infection ( Figures 2C, D , S3C ), while IFN-γ⁺ IL-4⁺ and IFN-γ⁺ IL-17A⁺ double-positive cells were not ( Figures S3C–G ). However, while far less abundant than IL-13⁺ single-positive cells ( Figure S3C ), IL-4⁺ and IL-17A⁺ single-positive cells were still increased relative to WT controls ( Figures S3D–G ), consistent with prior work showing that STAT1 can suppress Th2- and Th17-type responses (29, 43, 44). Flow cytometry also confirmed that IL-13⁺ CD4⁺ T cells express high levels GATA3, the lineage-defining transcription factor for Th2 cells ( Figure S3H ). Transcript levels for Il5, another emblematic Th2 cytokine, and Il17f, another emblematic Th17 cytokine, were as low as Il4, suggesting that the former are also not evident at protein level ( Figure 2B and Figure S3D ). Thus, T cells from infected Stat1 ^T-KO mice appear poised to express multiple Th2- and Th17-type cytokines but IL-13 is uniquely de-repressed and, given its pro-inflammatory nature, we propose that it contributes to the attendant immuno-pathology.

Aside from negative DEG, we also considered the possibility that positive DEG may contribute to the aberrant T cell responses seen in Stat1 ^T-KO mice. We were immediately drawn to IL-10, which is required to suppress T cell responses during infection with T. gondii, as well as numerous other pathogens (16, 45). Strikingly, we found that IL-10 was dramatically reduced at both transcript and protein levels relative to WT controls ( Figures 2B, E, F ). We also noted that double-positive IFN-γ⁺ IL-10⁺ cells, the main IL-10 producers in WT mice, were scarce in Stat1 ^T-KO mice ( Figure 2E ), and that IL-10-producing Treg cells were also diminished ( Figure S5 ), suggesting that STAT1-driven IL-10 production may be relevant across T cell subsets. Consistent with this latter point, serum IL-10 was also diminished, indicting systemic failure of IL-10 production ( Figure S2B ). Given the many known anti-inflammatory properties of IL-10, we interpret that diminished production of this cytokine likely contributes to immune-pathology in Stat1 ^T-KO mice challenged with T. gondii.

Along with anti-inflammatory features, our RNA-seq studies also illustrate pro-inflammatory features of STAT1 that are relevant during T. gondii infection. Ifng and Tbx21 transcripts were each comparable to WT controls, likely due to the pre-selection of IFN-γ⁺ cells, but both IL-12 receptors subunits were sharply reduced in STAT1-deficient cells ( Figure 2B ). This is in line with prior work showing that STAT1 induces expression of Il12rb1 and Il12rb2 and notable here because IL-12 can include both IFN-γ and IL-10 production in T cells, and is strictly required for resistance to T. gondii (37). Taken together, our cytometry and RNA-seq studies affirm that T cell intrinsic STAT1 signaling has both pro-and anti-inflammatory functions which are critical for resistance to T. gondii and further argue that only the latter are nonredundant.

STAT family transcription factors control gene expression in both direct and indirect ways (46). Direct regulation involves binding to DNA regulatory elements, most notably promoters and enhancers, and thereby instructing transcription of associated genes. Indirect regulation can take many forms but, here, we will focus on the ability of STATs to induce expression of other transcription factors which, in turn, mediate downstream effects. To address direct regulation, we cross-referenced DEG from our transcriptome studies with a STAT1 ChIP-seq dataset captured in CD4⁺ T cells cultured with IL-6 or IL-27 ( Table S3 ) (17). This analysis revealed a near even split between STAT1 bound and ‘unbound’ DEG, indicating that both direct and indirect forms of regulation are relevant ( Figure 3A ). We were also struck by the appearance of both IL-10 and IL-13 among the STAT1 bound fraction as they have reciprocal expression patterns (i.e. positively versus negatively regulated) and likely play opposing roles during acute toxoplasmosis (i.e. anti- versus pro-inflammatory) ( Figure 2B and Figure 3B ).

To further explore their relationship to STAT1, we curated a list of transcription factors that are known to regulate IL-10 or IL-13 and asked which are proximally bound downstream of IL-27 ( Table S4 ) (10, 45, 47–50). Surprisingly, we found that, while only a select few were called DEG in our transcriptome studies, nearly all were strongly bound by STAT1 ( Figure 3C and Figure S6 ). Given the ambiguity of genes bound by STAT1 but not called DEG, which cannot be readily included or excluded as direct targets, we next focused on genes that met both criteria. MAF immediately stood out because it is both critical for T cell IL-10 production and strongly diminished in STAT1-deficient cells ( Figures 3B, C ). By contrast, Irf4 was higher expressed in STAT1-deficient cells, as were AP-1 family transcription factors Jun and Junb ( Figures 3B, C and Figure S6 ). Also notable, STAT1 itself was the most strongly bound IL-10 regulator ( Figure 3C ), in line with the long-standing notion that STATs induce themselves (46). Taken together, these data argue that STAT1 controls T cell IL-10 and IL-13 production through both direct and indirect mechanisms.

We and others have shown that STAT3 signaling is enhanced in STAT1-deficient T cells, leading to enhanced STAT3-driven gene transcription (17, 51). Thus, we were intrigued to find that ‘STAT3 signaling’ was among the most enriched pathways in Th1 cells from infected Stat1 ^T-KO mice ( Figure 4A ). Building on this finding, we confirmed that phosphorylation of STAT3 at tyrosine 705, the instigating event for canonical STAT3 signaling, is enhanced in STAT1-deficient T cells. Importantly, increased STAT3 phosphorylation was evident whether cells were cultured with IL-27, a potent activator of both STAT1 and STAT3, or IL-21, a potent STAT3 activator but lesser STAT1 activator ( Figures 4B, C ).

To further probe the relationship between STAT1 and STAT3, we crossed Stat1 ^T-KO mice with Stat3 ^HIES mice, which bear a germline transgene encoding a dominant negative form of STAT3 that impairs DNA binding (52). Of note, this mutation is commonly found in humans afflicted with Hyper IgE Syndrome (HIES), a disease characterized by severe immunological abnormalities, including massively elevated serum immunoglobulins impaired vaccination responses, and impaired Th17-type responses leading to recurrent fungal and bacterial infections (53). The resulting Stat1 ^{T-KO HIES} mice therefore lack STAT1 selectively in T cells and have reduced STAT3 activity in all cells. They were born at expected mendelian rations and indistinguishable from WT littermates in terms of appearance and lifespan (not shown). This was in line with prior studies on the parent Stat3 ^HIES strain (52) but in stark contrast to humans afflicted with HIES, most of whom exhibit severe developmental and skeletal defects (53). Notwithstanding, the progeny Stat1 ^{T-KO HIES} strain mirrored the parent Stat3 ^HIES strain in that both bore key immunological features of HIES, including prominent serum IgE ( Figure S2C ) and impaired Th17 responses ( Figures S7A, B ). They also mirrored the parent Stat3 ^HIES strain in terms of increased CD8⁺ T cell IFN-γ production ( Figure S1I ), and, notably, far eclipsed them in terms of serum IgE levels ( Figure S2C ). Most of the other immunological features that were considered, including total and memory CD8⁺ T cells, more closely resembled the parent Stat1 ^T-KO strain ( Figures S1D–F ).

Stat1 ^{T-KO HIES} mice largely phenocopied Stat1 ^T-KO mice when challenged with T. gondii. They were able to control parasite replication but ultimately succumbed to immune-pathology characterized by aberrant, IL-13 producing Th1 cells ( Figures 5A–E ). STAT3 ^HIES mice also cleared the parasites and succumbed to immunopathology but T cells from these animals produced substantially more IFN-γ than WT controls and did not produce IL-13 ( Figures 5A–E ). It was also notable that Stat1 ^{T-KO HIES} mice varied from both parental strains in terms of total CD4⁺ and CD8⁺ T cell counts ( Figure 5F ). Nevertheless, we can conclude that STAT1 is uniquely required to limit outgrowth of IL-13-producing Th1 cells during acute toxoplasmosis.

STAT1 and STAT3 have each been implicated in the generation of IL-10-producing T cells during T. gondii infection (8). However, despite strong in vitro evidence, neither has been shown to be strictly required in vivo. Thus, we were intrigued to find that CD4⁺ IFN-γ⁺ IL-10⁺ T cells were nearly absent in Stat1 ^{T-KO HIES} mice but not STAT3 ^HIES mice, which were comparable to WT controls ( Figures 6A, B ). These data formally establish that STAT1 is dominant in vivo, so we next compared IL-10 production in vitro under conditions that mimic the inflammatory milieu of acute toxoplasmosis. Specifically, naive CD4⁺ T cells were cultured in the presence of IL-12 and IL-27, then cytokine production assayed by flow cytometry. Again, we found that the frequency of double-positive IFN-γ⁺ IL-10⁺ and single-positive IL-10⁺ events was far lower in Stat1 ^T-KO and Stat1 ^{T-KO HIES} cultures than in STAT3 ^HIES cultures, indicating that STAT1 indeed is the dominant inducer of IL-10 under strongly Th1 polarizing conditions ( Figures 6C, D ).

Given present and past evidence that STAT1 and STAT3 each promote T cell IL-10 production, we compared their distributions across relevant gene loci. To that end, we cross-referenced the STAT1 ChIP-seq dataset referenced above with a STAT3 ChIP-seq dataset that was also captured in CD4⁺ T cells treated with IL-27 (17), and a STAT4 ChIP-seq dataset captured in CD4⁺ T cells treated IL-12 (54). The latter was included because IL-12-STAT4 signaling is prevalent during acute toxoplasmosis and prior work has shown it to be important for T cell IL-10 production (55, 56). Notably, we found that STAT1, STAT3 and STAT4 co-localized at >3000 genomic regions ( Figure 6E and Table S5 ), including multiple sites along the proximal Il10 locus and sites associated with key IL-10 regulators, particularly MAF and IRF4 ( Figures 6E, F , S6 and Table S5 ). Thus, STAT1, STAT3 and STAT4 appear to compete for access to DNA regulatory elements that control expression of IL-10 and key upstream transcriptional regulators.

Stat1 flox/flox mice were generated as described (62), backcrossed to C57Bl/6 (>8 generations), and further crossed with C57Bl/6 Cd4 Cre mice (stain: 022071 from Jackson Labs, USA) to generate Stat1 ^T-KO mice. STAT3 ^HIES mice were generated as described (52) and crossed Stat1 ^T-KO to generate Stat1 ^{TKO HIES} mice. Stat1^{-/ -} mice were purchased from Taconic labs (USA) and backcrossed to C57Bl/6 (>8 generations). Wild type C57BL/6J mice were purchased from Jackson Labs (Strain: 000664). For antibody quantification, serum was collected from 12–16-week-old cohorts of uninfected mice and assayed using LEGENDplex assays (service by Biolegend, USA). For histology, lungs, livers and spleens were dissected at day 7 post-infection, then processed for H&E staining. Blinded scoring was performed by a veterinary pathologist. Both male and female mice were used, and all cohorts were sex and age matched. Animals were housed and handled in accordance with NIH and University of Miami guidelines and all experiments approved by the respective Animal Care and Use Committees.

Mice were injected intraperitoneally with approximately 20 cysts of the avirulent ME49 strain of T. gondii, as described (63). Parasite burden was determined by microscopy, specifically by counting infected cells within cytospin smears obtained from Peritoneal Exudate cells (PEx), or by PCR to detect the T. gondii B1 gene in DNA from whole lung, liver and spleen samples as described (64). PCR data are presented as 2^-ddCt comparing amplification of the B1 gene relative to the mammalian IL-2 locus such that negative values indicate that the parasite DNA is less abundant. All quantification was performed at day 7 post-infection.

Single cell suspensions were made from spleens and peritoneal exudates at day 7 post-infection as described (65). 0.5-1 x 10^6 cells/ml were then stimulated with either soluble Toxoplasma antigen (sTAg; 10 μg/ml) or plate bound anti-CD3ϵ (10 μg/ml; clone 17A2; BioXcell, USA). 18 hours later, cultures were pulsed with Brefeldin A for 2-3 hours (BFA; 10 μg/ml; Sigma, USA), then processed for intracellular flow cytometry. All cultures were in RPMI-1640 medium supplemented with 10% fetal calf serum, 1% sodium pyruvate, 1% nonessential amino acids, 10 mM HEPES, 0.1% β-Mercaptoethanol, 100 U/ml penicillin and 100 mg/ml streptomycin. For intracellular flow cytometry, cells were washed in phosphate-buffered saline supplemented with 0.5% bovine serum albumin and 0.1% sodium azide, then fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences) and stained with fluorochrome-labeled anti-cytokine antibodies, in conjunction with fluorochrome labelled anti-mouse CD4, CD8, CD25, CD44, FOXP3, GATA3 and/or T-bet. Dead cells were excluded using Live/Dead aqua (Thermo Fischer). All antibodies were purchased from Thermo Fisher, BD Biosciences or Biolegend.

Naive CD4⁺ T cells (CD4⁺ CD44^low CD25^-) were sorted from pooled lymph nodes and spleens of uninfected mice (>95% purity). 1 x 10^6 cells/ml were then stimulated with plate bound anti-CD3ϵ (10 μg/ml; clone: 17A2; BioXcell) and soluble anti-CD28 (1 μg/ml; clone;37.51 BioXcell) in the presence of IL-6, IL-12, IL-21 and/or IL-27. 72 hours later, cultures were pulsed with phorbol 12-myristate 13-acetate (PMA; 50 ng/ml; Sigma, USA) and Ionomycin (500 ng/ml; Sigma, USA) for 4 hours in concert with Brefeldin A for the final two hours (BFA; 10 μg/ml), then processed for intracellular flow cytometry. For detection of phospho-STAT proteins, cells were cultured with anti-CD3ϵ and anti-CD28 alone for 48 hours, pulsed with cytokine for 1 hour (without PMA, Ionomycin or BFA), then processed for intranuclear flow cytometry. Briefly, these were fixed with 2% formaldehyde, permeabilized with 100% methanol and stained with fluorochrome labelled anti-human/mouse pY701 STAT1 (Clone 4a; BD Biosciences) and anti-human/mouse pY703 STAT3 (Clone 4/P-STAT3; BD Biosciences) in conjunction with fluorochrome labelled anti-mouse CD4, CD25 and CD44. Mouse IL-6, IL-12, IL-21 and IL27 were purchased from R&D Systems (USA) and used at 10 ng/ml. All cultures contained anti-mouse IL-4 and IFN-γ (10 μg/ml each; clone 11B11 and XMG1.2; BioXcell).

T. gondii infections and ex vivo sTAg re-stimulations were performed as above (without BFA), then IFN-γ capture assay performed as described (15). 0.5-2 × 10⁵ PE-labeled, IFN-γ-secreting cells were sorted per group (<90% purity) and immediate lysed and stored in Trizol regent (Thermo)( Figure S4A ). Total RNA was later purified by phenol-chloroform extraction with GlycoBlue as co-precipitant (7-15 μg per sample; Life Technologies, USA). Poly(A)+ mRNA was then enriched by oligo-dT-based magnetic separation and single-end read libraries prepared with NEBNext Ultra RNA Library Prep Kit (New England Biolabs, USA). Sequencing was performed with a HiSeq 2500 (Illumina, USA) and 50 bp reads (20-50 × 10⁶ per sample) were aligned onto mouse genome build mm9 with tophat2, assembled with cufflinks and gene-level counts compiled with htseq-count. To minimize normalization artifacts, genes failing to reach an empirically defined count threshold were purged using htsfilter. 12-14 × 10³ genes were typically recovered post filtering, regardless of genotype or experimental condition. These were normalized and differentially expressed genes (DEG) called by quasi-likelihood F testing using edgeR (66). DEG call denotes >2 fold pairwise change and Benjamini-Hochberg (BH) adjusted p value < 0.05 ( Table S1 ). Transcripts per million (TPM) were compiled with edgeR. An offset value of 1 was added to all TPM and those failing to reach a value >2 TPM in any genotype/condition were purged. clusterProfiler (67) was used for hypergeometric testing (HGT) of positively or negative regulated DEG against the KEGG (68) and Molecular Signatures (MSigDB) databases ( Table S2 ) (69). KEGG pathway maps were rendered with pathview, and all other plots with ggplot2 or Datagraph (Visual Data Tools Inc., USA). All data are deposited to the NCBI Gene Expression Omnibus (GEO) under accession number :.

STAT1, STAT3 and STAT4 ChIP-seq datasets were downloaded from the NCBI Small Read Archive via GEO (GSE65621 and GSE22105). STAT1 and STAT3 datasets are from CD4⁺ T cells cultured in vitro with IL-6 or IL-27 (17). STAT4 dataset is from CD4⁺ T cells cultured in vitro with IL-12 (54). 50 bps reads were aligned using bowtie and non-redundant reads mapped to mouse genome mm9 with macs2 with default settings and input controls as reference for peak calling (70). homer was used to annotate peaks and test for DNA motif enrichment. bedtools intersect was used for peak overlap analysis using the -wa option. Gene proximal peaks were defined as occurring within introns, exons, UTRs or <20 kb of transcriptional start or end sites. Genome browser files were rendered with IGV.

Statistical variances and distributions were measured by paired t test or ANOVA, as indicated. Bonferroni correction was used to account for multiple testing in RNA-seq and ChIP-seq datasets. When present, error bars denote standard deviation across >2 biological replicates.