Type I interferon signaling attenuates regulatory T cell function in viral infection and in the tumor microenvironment

Regulatory T cells (Tregs) play a cardinal role in the immune system by suppressing detrimental autoimmune responses, but their role in acute, chronic infectious diseases and tumor microenvironment remains unclear. We recently demonstrated that IFN-α/β receptor (IFNAR) signaling promotes Treg function in autoimmunity. Here we dissected the functional role of IFNAR-signaling in Tregs using Treg-specific IFNAR deficient (IFNARfl/flxFoxp3YFP-Cre) mice in acute LCMV Armstrong, chronic Clone-13 viral infection, and in tumor models. In both viral infection and tumor models, IFNARfl/flxFoxp3YFP-Cre mice Tregs expressed enhanced Treg associated activation antigens. LCMV-specific CD8+ T cells and tumor infiltrating lymphocytes from IFNARfl/flxFoxp3YFP-Cre mice produced less antiviral and antitumor IFN-γ and TNF-α. In chronic viral model, the numbers of antiviral effector and memory CD8+ T cells were decreased in IFNARfl/flxFoxp3YFP-Cre mice and the effector CD4+ and CD8+ T cells exhibited a phenotype compatible with enhanced exhaustion. IFNARfl/flxFoxp3YFP-Cre mice cleared Armstrong infection normally, but had higher viral titers in sera, kidneys and lungs during chronic infection, and higher tumor burden than the WT controls. The enhanced activated phenotype was evident through transcriptome analysis of IFNARfl/flxFoxp3YFP-Cre mice Tregs during infection demonstrated differential expression of a unique gene signature characterized by elevated levels of genes involved in suppression and decreased levels of genes mediating apoptosis. Thus, IFN signaling in Tregs is beneficial to host resulting in a more effective antiviral response and augmented antitumor immunity.


Introduction
Regulatory T cells (Tregs) are a subset of CD4 + T cells, which express the transcription factor Foxp3, and are critical in forestalling both self-and non-self-reactive immune responses [1,2]. Tregs primarily mediate their suppressive function by targeting conventional effector T cell activation and differentiation, mainly by decreasing the functional activity of antigen presenting cells (APCs) [3]. The critical role of Tregs in autoimmunity is best observed in scurfy mice or patients with IPEX syndrome that are completely deficient in Tregs and succumb to systemic autoimmune disease at a young age. While Tregs must control the activation of T effector cells to prevent autoimmunity, it is also clear that enhanced activation of Tregs may result in the inhibition of host immunity directed against microbes (virus, bacteria, protozoa, fungi and helminth) or tumors leading to poor antimicrobial or antitumor immune response with the persistence of pathogens, and defective tumor immunity [4][5][6].
Many animal models of bacterial infection are characterized by the expansion of Foxp3 + Tregs including Listeria monocytogenes, Salmonella enterica, and Mycobacterium tuberculosis infections and the suppressive function of Tregs can result in increased bacterial load with systemic tissue invasion [7][8][9]. Similarly in viral infection, higher frequencies of Tregs are associated with enhanced titers of Hepatitis C virus RNA and Dengue virus [10,11]. Paradoxically, Tregs have been described to play an early protective role in local infection in animals models of Herpes simplex virus 2 and West Nile virus [12,13]. During early phases of human immunodeficiency virus infection, Tregs have been postulated to control virus replication in target CD4 + T cells [14]. On the other hand Tregs may play an important beneficial role in preventing exuberant inflammatory responses during infection with parasites such as Pneumocystis carinii [15] and Schistosoma mansoni [16]. Similarly, Tregs protect the host from parasitic infections such as Plasmodium sp., Toxoplasma gondii, as well as infection with the fungus, Candida albicans [17][18][19]. These complex roles played by Tregs during acute and chronic microbial infections necessitate a delicate balance between the Foxp3 + Tregs and effector T cells to mount effective immune responses against pathogens without the induction of destructive autoimmunity.
The immune response towards viruses and intracellular bacteria are mediated by type I interferons (IFNs) which control the replication of pathogens within host cells. IFNs are members of a multi-gene family of cytokines, which encode IFN-α and IFN-β. Both IFN-α and IFN-β signal through a shared common heterodimeric receptor IFN-α/β receptor (IFNAR) composed of IFNAR1 and IFNAR2 [20]. The interactions of IFNs with the IFNAR mediates activation of Janus family protein kinases to induce the phosphorylation of signal transducer and activator of transcription (STAT). The canonical pathway of Type I IFN signaling is initiated by phosphorylation of STATs (STAT1, STAT2), induction of IFN-regulatory factor-9, resulting in the formation of a tri-molecular complex, IFN-stimulated gene factor-3, which translocates into the nucleus to induce transcription of IFN-stimulated genes through binding of IFN-stimulated response elements [21]. Additionally IFNAR signaling can trigger noncanonical pathways such as activation of γ-activated sequences through homodimerization of STATs (STAT1, STAT3, STAT4, STAT5, STAT6), phosphoinositide-3-kinase/mammalian target of rapamycin pathway, and mitogen-activated protein kinase pathway [22].
IFNs may mediate an array of host protective functions including restricting viral replication [23], activation of NK cell cytotoxicity, maturation of APCs, clonal expansion and survival of antigen-specific CD4 and CD8 T cells during viral infection, promotion of B cell responses, and induction of apoptosis [24][25][26][27][28][29][30]. Type I IFNs have proven to be clinically useful in the treatment of chronic viral infections and certain types of leukemias [31]. Detrimental effects of type I IFNs have also been extensively documented during viral infections as well as during bacterial, fungal and parasitic infections [32]. One of the best examples of the complex regulation of antiviral immunity by type I IFNs is lymphocytic choriomeningitis virus infection (LCMV). Blockade of IFN signaling in acute infection with LCMV Armstrong infection results in abrogation of CD8 + T cell responses and defective control of infection [33]. In contrast, blockade of IFN signaling during persistent LCMV Clone (Cl)-13 infection diminished immunosuppressive signals and decreased levels of IL-10 and PD-L1 expressing immunoregulatory DCs. Virus titers in both serum and kidneys were also reduced. The cell type (s) mediating the immunosuppressive effects of IFN have not been defined [33,34].
Studies on the effects of type I IFNs on Treg function yielded conflicting results [35,36] and have not used experimental systems to examine the direct effects of IFNs on Treg cell homeostasis and functions. Recently, Srivastava et al (2014) demonstrated that mice infected with LCMV Armstrong manifested a decrease in the absolute numbers of splenic Tregs between days 4 and 7 post infection and that this reduction correlated with the expansion of both CD4 + and CD8 + T effector cells which peak on day 7 post infection. Furthermore, they also demonstrated a selective depletion of wild type (WT) Tregs on day 7 post infection of mixed bone marrow chimeras between WT mice and mice with a global deletion of IFNAR (IFNAR -/-). This latter result is difficult to interpret as our recent studies [37] have shown that IFNAR -/-Tregs in such chimeric (IFNAR -/x WT) mice are at a competitive disadvantage as are IFNAR -/-Tregs in heterozygous female IFNAR fl/fl x Foxp3 Cre/WT mice.
In this study, we used IFNAR fl/fl x Foxp3 YFP-Cre mice to determine the role of IFNAR signaling specifically in Tregs during acute and chronic LCMV infection as well as in models of colon adenocarcinoma and melanoma. We demonstrate that IFNAR signaling in Tregs during the course of both acute and chronic viral infection results in a decrease in their activation status and a decrease in their suppressive function in vivo. The hypersuppressive state of Tregs in the absence of IFNAR signaling results in decreased CD8 + effector T cells, enhanced T effector cell exhaustion, defective generation of antiviral memory CD8 + T cells, and enhanced LCMV persistence. Similarly, in the tumor models, enhanced tumor growth and failure to efficiently generate antitumor T effector cells were observed in the absence of IFNAR signaling in Tregs. The enhanced suppressor function in the absence of IFNAR signaling in Tregs was accompanied by the induction of a gene expression pattern which was similar in the acute and chronic infection models and may be responsible for the heightened suppressor function.

IFNAR signaling in Tregs modulates the generation of antiviral effector T cells in acute LCMV infection
Studies performed by Srivastava et al. (2014), provided preliminary evidence that IFNAR signaling inhibits the function of Tregs during an acute LCMV infection and that the absence of this inhibitory effect resulted in enhanced Treg function and impaired antiviral effector T cell function. Because this study did not definitively prove that the target of the suppressive function of IFNAR signaling was the Foxp3 + Treg, we generated Treg-lineage specific IFNAR deficient mice by crossing IFNAR fl/fl mice with Foxp3 YFP-Cre mice. We confirmed that the IFNAR was specifically deleted in CD4 + Foxp3 + Tregs and not in CD4 + Foxp3 -T cells, CD8 + T cells or B220 + B lymphocytes (S1A Fig). First, we compared the clearance of LCMV Armstrong from the sera of IFNAR fl/fl , IFNAR fl/fl x Foxp3 YFP-Cre , and IFNAR -/mice at early time points postinfection. We observed that IFNAR fl/fl x Foxp3 YFP-Cre mice showed significantly higher viral titers (D3, D7 and D10) than IFNAR fl/fl mice, and as expected IFNAR -/mice had the highest titers among the three groups [33,38] (Fig 1A). However, IFNAR fl/fl x Foxp3 YFP-Cre mice cleared LCMV Armstrong on day 14 post-infection.
This result is consistent with the studies of Srivastava et al. (2014) suggesting that the responses of the antiviral effector T cells early during LCMV Armstrong infection are compromised. Indeed, while both the frequencies and absolute numbers of CD8 + CD44 + T cells specific for GP33-41 and NP396-404 did not differ between IFNAR fl/fl and IFNAR fl/fl x Foxp3 YFP-Cre mice on day 14 post-infection (Fig 1B and 1C), the production of the effector cytokines IFN-γ and TNF-α was markedly diminished (Fig 1E and 1F). While the frequency and absolute number of CD8 + CD44 + T cells specific for GP276-286 were increased, the frequency of IFN-γ producing cells recognizing GP276-286 was still reduced (Fig 1D and  1G). We did observe a modest decrease in the frequency, but not absolute numbers of CD4 + Foxp3 -CD44 + T cells specific for LCMV GP66-76 ( Fig 1H) and this was accompanied by a marked decrease in the production of both IFN-γ and TNF-α by the GP66-76 specific cells ( Fig 1I).

Lack of IFNAR signaling results in higher numbers of activated Treg during LCMV Armstrong infection
Taken together, these studies are consistent with the possibility that the absence of signaling via the IFNAR in Tregs during LCMV Armstrong infection potentiated their suppressive activity resulting in a failure to fully activate LCMV-specific T effector cells [39]. We observed that Treg cell numbers from infected WT and IFNAR fl/fl x Foxp3 YFP-Cre mice on day 4 and day 5 are similar and that Treg cells from both the WT and IFNAR fl/fl x Foxp3 YFP-Cre mice decrease similarly on day 7 post-infection (S1B Fig). While Srivastava et al (2014) reported a marked decrease in WT Foxp3 + Tregs on day 7 post infection in mixed bone marrow chimeras, we did not observe a decrease in Tregs on day 5 and the reduction in Treg frequencies and absolute numbers on day 7 was seen in both IFNAR fl/fl and IFNAR fl/fl x Foxp3 YFP-Cre mice (S1B Fig). In contrast, Foxp3 + Tregs frequencies and numbers were increased on day 14 post LCMV Armstrong infection in IFNAR fl/fl x Foxp3 YFP-Cre mice. Most notably, the percentages and absolute numbers of activated Tregs were increased in IFNAR fl/fl x Foxp3 YFP-Cre mice compared to IFNAR fl/fl mice on day 5, 7 and 14 post Armstrong infection (S1C Fig). In addition to elevated levels of CD44, Tregs in IFNAR flf/fl x Foxp3 YFP-Cre mice also expressed higher percentages of other activation markers, including Ki-67 + , ICOS + and TIGIT + (S1D and S1E Fig), consistent with an activated phenotype and greater degree of proliferation at day 5 post Armstrong infection. We did not see any differences in the frequencies of activated CD4 + Foxp3and CD8 + T cells (S1F Fig

Absence of IFNAR signaling potentiates viral persistence in chronic LCMV infection
The role of Tregs in the maintenance of chronic viral infection and effector T cell exhaustion has been difficult to define as it has been technically challenging to specifically deplete Tregs without the induction of autoimmune disease [40][41][42]. To evaluate the role of IFNAR signaling in Tregs during persistent chronic viral infection, mice were infected with LCMV Cl-13. We initially examined viral titers by plaque assay in serum at different time points during infection. On days 8, 25, 35 and 43-post infection, IFNAR fl/fl x Foxp3 YFP-Cre mice had significantly Frequencies and absolute numbers of IFN-γ + and TNF-α + cytokine producing cells within CD8 + CD44 + GP33 Tet + , CD8 + CD44 + NP396 Tet + and CD8 + CD44 + GP276 Tet + T cells from day 14 acute LCMV infected mice are shown. (H and I) Spleen cells were analyzed (as in B-D) for GP66 Tet + T cells, and IFN-γ + and TNF-α + producing cells were assessed among gated CD4 + Foxp3 -CD44 + GP66 Tet + T cells from day 14 acute LCMV infected mice. Ã P < 0.05, ÃÃ P < 0.01, ÃÃÃ P < 0.001, and ÃÃÃÃ P < 0.0001 (unpaired two-tailed Student's t-test). Data are shown from two to three experiments involving three to nine mice per group (A), and representative of two independent experiments (B-I) with four to five mice per group (Mean±SEM).
The persistence of Cl-13 infection among IFNAR fl/fl x Foxp3 YFP-Cre mice led us to examine the kinetics and activation of Tregs during chronic LCMV infection. The frequencies and absolute numbers of Tregs were higher in IFNAR fl/fl x Foxp3 YFP-Cre mice on day 25-post infection, but not on day 46-post infection ( Fig 2C). However, the activation state of the Tregs as measured by CD44 expression was higher on both days 25 and 46 in IFNAR fl/fl x Foxp3 YFP-Cre mice compared to IFNAR fl/fl mice ( Fig 2D). In contrast, no significant differences were observed in the frequencies or absolute numbers of CD4 + Foxp3or CD8 +

Absence of IFNAR signaling in Tregs results in decreased virus-specific cytokine production and enhanced virus-specific T effector cell exhaustion
To determine if the enhanced Treg activation and diminished viral clearance in Cl-13 infected IFNAR fl/fl x Foxp3 YFP-Cre mice results in decreased antiviral T effector cell responses, we examined the LCMV-specific responses of effector T cells. On day 25-post infection, both the absolute numbers of GP33 and NP396 tetramer positive T cells were similar (Fig 3A and 3B), while the frequencies of IFN-γ and TNF-α producing cells were lower (Fig 3C and 3D). However, on day 46-post infection, the absolute numbers of both antigen-specific CD8 + T cells were significantly decreased (Fig 3A and 3B) and this was accompanied by a marked decrease in the frequencies and absolute numbers of IFN-γ and TNF-α producing cells (Fig 3C and 3D). A moderate increase in the absolute number of CD4 + Foxp3 -CD44 + GP66 Tet +  High levels of PD-1 expression are one of the hallmarks of T cell exhaustion. On both days 25-and 46-post infection, the frequency of PD1 expressing CD8 + and CD4 + Foxp3 -T cells were higher in IFNAR fl/fl x Foxp3 YFP-Cre mice (Fig 4A and 4B). Two other markers of T cell exhaustion, the transcription factor eomesodermin (EOMES), and CD39 can be co-expressed with PD-1 on exhausted T cells [43,44]. Higher percentages and absolute numbers of EOMES + PD-1 + cells within the CD8 + and CD4 + Foxp3populations were present in the IFNAR fl/fl x Foxp3 YFP-Cre mice (Fig 4C and 4D). Correspondingly, PD1 + CD39 + frequencies and total numbers were higher in gated CD8 + T cells from IFNAR fl/fl x Foxp3 YFP-Cre mice ( Fig 4E). While the frequencies of gated CD4 + Foxp3 -PD1 + CD39 + T cells did not differ between IFNAR fl/fl and IFNAR fll/fl x Foxp3 YFP-Cre mice, the absolute numbers of CD4 + Foxp3 -PD1 + CD39 + T cells were higher in Treg-specific IFNAR deficient mice ( Fig 4F). Similarly, the frequencies of PD-1 + T cells were greater among gated CD8 + CD44 + GP33 and CD8 + CD44 + GP276 Tet + T cells in IFNAR fl/fl x Foxp3 YFP-Cre mice (Fig 4G and 4H). CD8 + CD44 + NP396 Tet + and CD4 + Foxp3 -CD44 + GP66 Tet + populations from day 46 Cl-13 infected IFNAR fl/fl x Foxp3 YFP-Cre mice also had higher proportions of PD1 expressing cells (S4A and S4B Fig). In addition, it has been demonstrated that Tregs with higher levels of PD1 expression can mediate enhanced suppression in LCMV infection [45], we also found that Tregs from Cl-13 infected IFNAR fl/fl x Fox-p3 YFP-Cre mice had significantly higher expression of PD1 than controls on days 25   absence of IFNAR signaling during Cl-13 infection results in markedly reduced cytokine production by virus-specific CD8 + T cells as well as a phenotype consistent with exhaustion.

Absence of IFNAR signaling on Tregs results in defective generation of memory T cells in LCMV infection
Exhausted CD8 + T cells have reduced memory cell potential which is secondary to higher LCMV antigen persistence in infected mice [46]. To determine whether the enhanced T cell exhaustion phenotype observed in IFNAR fl/fl x Foxp3 YFP-Cre mice is associated with a reduction in the formation of virus-specific memory T cells, we examined the levels of expression of three memory cell makers (CD62L, CD127, CXCR3) on gated CD8 + CD44 + GP276 Tet + T cells ( Fig 5A). The frequencies and the absolute numbers of all three memory populations were reduced in IFNAR fl/fl x Foxp3 YFP-Cre mice on day 46 post infection compared to IFNAR fl/fl control mice (Fig 5B-5D). Similar results were seen within the CD8 + CD44 + GP33 Tet + population (S5A- S5D Fig). While the number of memory CD8 + T cells is usually inversely correlated with terminally differentiated T cells as measured by KLRG-1 expression [47], the frequencies and absolute numbers of CD8 + CD44 + GP276/GP33 Tet + KLRG-1 + T cells were also lower in IFNAR fl/fl x Foxp3 YFP-Cre mice ( Fig 5E and S5E Fig). We further examined the protective capacity of memory CD8 + T cells by re-infecting the day 30 Armstrong infected mice with Cl-13 virus. On day 5 post Cl-13 infection, IFNAR fl/fl x Foxp3 YFP-Cre mice had reduced frequencies and numbers of CD8 + CD44 + NP396 Tet + cells, and in addition GP33-and GP276-stimulated CD8 + CD44 + T cells from IFNAR fl/fl x Foxp3 YFP-Cre infected mice produced significantly less Granzyme B (GrB) positive and GrB/IFN-γ double positive cells compared to CD8 + T cells from control mice (Fig 5F and 5G), however IFN-γ positive cells are tended to be more in infected IFNAR fl/fl x Foxp3 YFP-Cre mice but they are not significant compared to control mice. Collectively, these results demonstrate that the higher load of virus is associated with a defect in the generation of virus-specific memory T cells in the absence of IFNAR signaling in Tregs.

Transcriptional profiling of Tregs during Armstrong and Cl-13 infection
In order to better understand the molecular basis for the enhanced activation and suppressive function of Tregs from Treg-specific IFNAR-deficient mice during acute and chronic LCMV infection, we performed high-throughput RNA sequencing on sorted CD4 + YFP + Tregs isolated from day 5 Armstrong-infected Foxp3 YFP-Cre and IFNAR fl/fl x Foxp3 YFP-Cre mice. Principal component analysis (PCA) showed distinct clustering of Tregs from Foxp3 YFP-Cre mice relative to Tregs from IFNAR fl/fl x Foxp3 YFP-Cre mice ( Fig 6A). A total of 586 genes were significantly differentially expressed (249 genes were down, and 337 genes were up) in IFNAR fl/fl x Foxp3 YFP-Cre mice (fold change 1.5 and above, adjusted P < 0.05) (Fig 6B). Among the 586 genes, 174 genes were identified in the interferome database [48] (interferome.its.monash.edu. au) as IFN-signaling related (fold change 1.5 and above, adjusted P < 0.05) (S6A Fig), and were excluded from further analysis. We elected to exclude IFNAR regulated genes in order to perform an unbiased downstream analysis, as IFNAR signaling regulates the transcription of up to 2000 genes. The remaining 412 differentially expressed genes were exclusively non-IFN related. Of these, 156 genes were upregulated in infected Foxp3 YFP-Cre mice, and 256 genes were upregulated in infected IFNAR fl/fl x Foxp3 YFP-Cre mice. Gene set enrichment analysis (GSEA) of the 412 non-IFN related genes revealed that the natural Treg vs conventional T cell gene set was enriched to a greater extent in IFNAR fl/fl x Foxp3 YFP-Cre mice [36 genes out of 42 were in core enrichment, Enrichment score (ES): 0.566, P < 0.01, FDR:0.0] (Fig 6C). We observed that 32 out of 412 non-IFN related genes were differentially expressed (fold change 1.5 and above, adjusted P < 0.05, genes normalized by z-score; 24 genes in IFNAR fl/fl x Foxp3 YFP-Cre mice and 8 genes in Foxp3 YFP-Cre mice were upregulated) and could be classified as Treg-signature genes as previously reported [49,50] (Fig 6D). Representative upregulated genes in IFNAR fl/fl x Foxp3 YFP-Cre mice include Areg, Arhpag20, Bub1b, Ccl12, Ccr5, Il1r1, Mki67 (Ki67), Ncf1, Nrp2, Tnfrsf9 (CD137), Tcf19, Uhrf1, and Wnt3; representative downregulated genes in IFNAR fl/fl x Foxp3 YFP-Cre include Cybb, Dapl1, Fam160a1, Il1r2, and Tnfsf8 (CD153). Some of the above upregulated genes from Tregs include Areg, Ccl12, Mki67, Ncf1, Tnfrsf9, Uhrf1 are well characterized to modulate the enhanced Treg suppressive and proliferative function [49,50]. Further, we also performed ingenuity pathway analysis (IPA) for non-  A). Splenocytes harvested on day 5 Cl-13 recall challenge after day 30 LCMV Armstrong infection (memory phase), and CD8 + CD44 + T cells were analyzed for frequencies and absolute numbers of NP396 Tet + T cells (F). Spleen cells from Cl-13 recall challenge were stimulated with GP33, GP276, and Golgi stop for 5 hours at 37 o C, frequencies and absolute numbers of CD8 + CD44 + T cells producing Granzyme B, and IFN-γ were analyzed (G). Ã P < 0.05, ÃÃ P < 0.01, ÃÃÃ P < 0.001, and ÃÃÃÃ P < 0.0001 (unpaired two-tailed Student's ttest). Data derived from one to two experiments involving three to eight mice per group (Mean±SEM).  Tregs type I interferon signaling in viral infection and cancer these associated genes in the pathways are shown (S1 and S2 Tables). Some of the associated genes in the networks include, transcription factors: Depdc1, E2f1, E2f8, Foxm1, Mybl2, and Uhrf1 are downregulated, and pydc4/Ifi16 (interferon gamma inducible protein 16) is upregulated in Tregs from Foxp3 YFP-Cre mice; cell cycle kinases: Ccnb2, Aurkb, and Chek1 are also downregulated in Foxp3 YFP-Cre mice Tregs; immune cell genes such Tnfrsf9 (CD137), Ccl2, Ccr5, and Il1r1 are downregulated, in contrast C5ar1, CD19, Il1r2, Itga2b (CD41), Ly6c1, and Tlr7 are upregulated in Foxp3 YFP-Cre mice Tregs. We also tested whether the reduced cell cycle gene signature in Tregs from Foxp3 YFP-Cre mice contributed to a greater degree of apoptosis, but Active caspase-3 staining showed no significant increase in the staining of Tregs from day 5 Armstrong infected Foxp3 YFP-Cre mice compared to Tregs from IFNAR fl/fl x Foxp3 YFP-Cre mice (S6D Fig). In parallel, we also performed RNA sequencing on Tregs from day 25 post LCMV Cl-13 infection. PCA showed less distinct clustering (Fig 6E), and surprisingly, only 36 genes were significantly differentially expressed (23 genes were downregulated, and 13 genes were upregulated in IFNAR fl/fl x Foxp3 YFP-Cre mice, fold change 1.5 and above, adjusted P < 0.05) in Tregs from Foxp3 YFP-Cre and IFNAR fl/fl x Foxp3 YFP-Cre mice (Fig 6F). Among those differentially expressed genes, 14 genes were identified in the interferome database [48] as IFN-signaling related (fold change 1. Few of the associated genes in the network include, kinases: Camk2b, and Hunk and cytoplasmic enzymes: Cpta1, and Scpep1 are down regulated, while proapoptotic factor Erdr1 is upregulated in Foxp3 YFP-Cre mice Tregs.

cells (as in
We identified fourteen genes (fold change 1.5 and above, adjusted P < 0.05, normalized by z-score) differentially expressed in Tregs from LCMV Armstrong infected mice which were similarly differentially expressed in Tregs from LCMV Cl-13 infected mice including 7 that were up-and 7 down-regulated (Fig 6G and 6H). We further validated the differential expression of some of these non-IFN related genes during chronic infection which are common to both infection model or unique to chronic infection alone by quantitative real-time PCR (S7D Fig). Several of the upregulated genes include a-kinase anchoring protein 1 (Akap1), calcium/ calmodulin-dependent protein kinase II b (Camk2b), Hormonally upregulated Neu-associated kinase (Hunk), Rab4a and Rasgrf2. Akap1 is associated with cAMP signaling, and it can act as a gap junction protein in facilitating the transfer of cAMP from Treg to effector T cells leading to inhibition of T cell receptor (TCR) signaling [51,52]. Furthermore, CamK2b, Hunk, Rab4a and Rasgrf2 have also been implicated in enhancement of Treg function [53][54][55][56][57]. Taken together, all these upregulated genes participate in heightened Treg suppressive function observed in the Tregs from IFNAR fl/fl x Foxp3 YFP-Cre mice during both acute and chronic LCMV infections.

Treg-specific IFNAR deficiency leads to an impaired antitumor immune response
Enhanced Treg cell function is very well documented in various tumor models and it has been associated with a poor prognosis [6]. To determine whether the absence of IFNAR signaling was associated with enhanced Treg suppression in a non-infectious setting, we utilized the mouse colon adenocarcinoma MC38 and mouse B16.F10 melanoma models. IFNAR fl/f x Fox-p3 YFP-Cre mice showed higher tumor incidence (MC38: n = 11/11, 100%; B16.F10: n = 5/5, 100%) and significantly increased volume than IFNAR fl/fl mice (MC38: n = 8/10 mice, 80%, B16.F10: n = 4/5, 80%) (Fig 7A, left and right panels). Tregs and CD4 + Foxp3cells isolated from tumor infiltrating lymphocytes (TIL) showed comparable frequencies in both strains of Tregs type I interferon signaling in viral infection and cancer mice, however CD8 + T cell frequencies from TIL of IFNAR fl/fl x Foxp3 YFP-Cre mice were increased. Nevertheless, both CD4 + Foxp3and CD8 + T cells from IFNAR fl/fl x Foxp3 YFP-Cre mice tended to proliferate less as assayed by Ki-67 expression (S8A- S8C Fig). Importantly, Tregs from IFNAR fl/fl x Foxp3 YFP-Cre mice TIL expressed significantly higher levels of CD44, enhanced proliferation and expression of PD-1 (Fig 7B). Conversely, both CD4 + Foxp3and CD8 + TIL from IFNAR fl/fl x Foxp3 YFP-Cre mice expressed lower levels of CD44 and markedly reduced levels of IFN-γ and TNF-α production compared to TIL from IFNAR fl/fl mice ( Fig  7C: gated on CD4 + Foxp3 -TILs and Fig 7D: gated on CD8 + TILs). These data strongly suggest that Tregs in TIL from IFNAR fl/fl x Foxp3 YFP-Cre mice have enhanced suppressor activity in the tumor microenvironment and the phenotype of these Tregs within TIL closely resembles the activated hypersuppressive phenotype observed during acute and chronic LCMV infections.

Discussion
Tregs mediate a multifaceted role in modulating the immune response to acute and chronic infectious agents. While their beneficial effects in decreasing immune pathology during the resolution phase of many infections is clear, Tregs can also mediate immune suppression resulting in pathogen persistence. During viral infections, rapid activation of the innate immune system generates inflammatory signals that can initially control the infection and ultimately influence the quality and magnitude of the adaptive antiviral effector T cell response. The best characterized innate inflammatory signals are the type I IFNs. During LCMV infection type I IFNs are produced in large quantities immediately following viral infection by plasmacytoid DCs as well as virus infected cells and primarily exert their effect on CD8 + T cells by extending their survival. In this report, we have demonstrated that type I IFNs can also exert beneficial effects by acting on Tregs to down-modulate their suppressive functions both early during the course of acute LCMV Armstrong infection and also later during virus persistence in chronic Cl-13 infection. Srivastava et al. (2014), have previously examined the effects of type I IFN on Tregs during the course of acute LCMV infection. They concluded that type I IFNs down-modulated Treg function but postulated that the effects of type I IFNs were secondary to a selective decrease in the number of highly suppressive effector Tregs, and secondary to the pro-apoptotic and antiproliferative actions of type I IFNs early in the course of infection. The major difficulty in the interpretation of this study is that the experimental model they used did not allow them to selectively examine the effects of IFNs on Tregs in the absence of its effects on other cell types. Similarly, one previous study also showed that Tregs were reduced in infected wild type compared to control mice during first week post LCMV-Docile infection, but Tregs expanded to a greater extent from the second weeks onwards. Treg expansion was more pronounced in IL-21R deficient mice, suggesting IL-21 signaling restricts proliferation of Tregs during LCMV infection [58]. The availability of mice with a conditional deletion of the IFNAR in Tregs allowed us to dissect the mechanistic basis of IFNAR signaling in Tregs resulting in their reduced suppressive function and in more efficient antiviral and antitumor immune responses.
Similar to the previous study [39], we found that the generation of antigen-specific CD8 + and CD4 + T cells was comparable in Treg-specific IFNAR-deficient mice and WT controls during acute LCMV infection, but that both the virus-specific CD8 + and CD4 + T cells in IFNAR fl/fl x Foxp3 YFP-Cre mice produced markedly reduced amounts of IFN-γ and TNF-α accompanied by a slower rate of viral clearance than the controls. Most importantly, we did not observe a decrease in the percentages or absolute numbers of Tregs in the WT control mice that differed from IFNAR fl/fl x Foxp3 YFP-Cre mice on day 4 or 5 post-infection. We did detect a decrease in Tregs in the IFNAR fl/fl X Foxp3 YFP-Cre mice on day 7 post-infection, but we observed an identical decrease in Tregs in the control IFNAR fl/fl mice. In contrast to the loss of memory/effector Tregs observed in WT mice by Srivastava et al. (2014), we observed an enhanced percentage of memory/effector Treg as defined by CD44 expression on days 5, 7 and 14 post-infection in IFNAR fl/fl x Foxp3 YFP-Cre mice, as well as higher levels of expression of Ki-67, ICOS and TIGIT. Taken together, our results are most consistent with an enhanced suppressive phenotype of Tregs in the absence of IFNAR signaling in acute virus infection albeit the mice ultimately cleared the infection.
We observed a similar but more profound suppressive phenotype in IFNAR fl/fl x Fox-p3 YFP-Cre mice following Cl-13 infection, as the mice had higher serum titers of virus and also had higher viral titers in lungs and kidneys for as long as 46 days post infection. We observed decreased numbers of antiviral antigen-specific CD8 + T cells accompanied by a profound decrease in effector cytokine production. Increased viral persistence resulted in marked expression of markers associated with T cell exhaustion (PD-1, CD39, EOMES) [43,44,46] and decrease in generation of antigen-specific memory CD8 + T cells. The decrease in the formation of memory T cells in IFNAR fl/fl x Foxp3 YFP-Cre mice in Cl-13 infection should be contrasted to the effects of IL-10 producing Treg cells in augmenting memory T cell formation following Armstrong infection [47]. Slight increases in both the percentages, but not the absolute numbers of Tregs were seen on days 25 and 46 post infection. However, at both times points, we observed a significant increase in the percentages of activated/effector Tregs. To determine if the hyperactivated/hypersuppressive phenotype observed in the absence of IFNAR signaling in Tregs was unique to viral infections, we also examined the responses of IFNAR fl/fl x Foxp3 YFP-Cre mice to transplantable tumor models. Markedly enhanced growth of the tumor was observed in these mice accompanied by an enhanced percentage of activated PD-1 + tumor infiltrating Tregs. In addition, the activation and cytokine production by both CD4 + Foxp3and CD8 + tumor infiltrating T cells were markedly suppressed. Taken together, these studies demonstrate that IFNAR signaling in Tregs plays a critical role in down-modulating, but certainly not abolishing, their suppressive function and may in viral infections orchestrate the balance between immunopathology and eradication of the virus.
To begin to elucidate the mechanistic basis for the suppressive function of IFNAR-deficient Tregs during both acute and Cl-13 infection, we performed high-throughput RNA sequencing of Foxp3 + Tregs from both controls and IFNAR fl/fl x Foxp3 YFP-Cre mice on day 5 Armstrong and day 25 Cl-13 infection. A group of Treg-signature genes (Areg, Arhpag20, Bub1b, Ccl12, Ccr5, Il1r1, Mki67 (Ki67) Ncf1, Nrp2, Tnfrsf9, Tcf19, Uhrf1, and Wnt3) were expressed at higher levels in IFNAR fl/fl x Foxp3 YFP-Cre mice than WT controls on day 5 Armstrong infection. These genes have previously been identified as Treg-Up signature genes [49,50] and their enhanced expression is consistent with the hyperactivated phenotype of the Tregs at that time point. We did not observe upregulation of this group of genes on day 25 of Cl-13 infection. Interestingly, both in Armstrong and Cl-13 infection, we observed the differential expression of a number of genes in IFNAR fl/fl x Foxp3 YFP-Cre Tregs which might also play a role in their enhanced suppressive function including Akap1, Camk2b, Rasgrf2 and Hunk. Akap1, which serves as a gap junction protein, is involved in transferring pools of cAMP from Treg to effector T cells resulting in inhibition of TCR signaling [51,52]. Camk2b participates in the activation of nuclear factor kappa-B, which in turn plays a role in Treg development by stabilizing Foxp3 [53,54]. Both Camk2b and Hunk have been shown to be have higher levels of expression in human Tregs than conventional effector T cells [55]. Lastly, Rasgrf2 is involved in stimulation of TCR signaling through activation of nuclear factor for activated T cells (NF-AT) [57]. Conversely, we also observed group of genes (Erdr1, Rell1, Tlr7) whose expression is downregulated in Tregs from both IFNAR fl/fl x Foxp3 YFP-Cre Armstrong and Cl-13 infected mice. These genes have been described as potentially playing a role in apoptosis [59][60][61] and thus may be involved in the reduced suppressive function of Tregs after IFNAR signaling. We did not observe any differences between IFNAR fl/fl x Foxp3 YFP-Cre and WT mouse Tregs in expression of Active caspase-3 on day 5 post Armstrong infection again consistent with our data that cell death is not playing a role in reduced Treg suppression in WT mice, although, we cannot exclude the involvement of other cell death pathways. Future studies involving over expression and/or deletion of these genes in Tregs will be needed to specifically implicate one or several of these genes in Treg-mediated suppression during viral infections. We have not yet performed similar gene expression studies in IFNAR sufficient and deficient Tregs derived from the tumor microenvironment and compared such data with Tregs from LCMV-infected mice.
Our study demonstrates that one of the multiple cellular targets of Type I IFNs during viral infection are Treg cells and that the functional result of this interaction is a downregulation of Treg suppressor function. A similar process may take place in the tumor microenvironment and may be responsible for some of the antitumor effects of this group of cytokines [62,63]. Thus, type I IFNs should be added to the long list of cytokines (IL-1β, IL-4, IL-6, IL-15, IL-21) [64][65][66][67][68][69][70][71][72] and members of the tumor necrosis factor superfamily (TNFSF) (GITR-L, 4-1BB-L, OX40-L and TNF-α) [73][74][75][76] that are purported to decrease Treg suppressive function in autoimmune and infectious disease models. However, the abrogation of suppression is more frequently mediated by the action of the cytokine or TNFSF member on the responder T cells resulting in resistance to suppression [73,77], whereas we have definitively demonstrated that Tregs are the targets cells in this model. While type I IFNs may attenuate Treg suppression, it remains clear that the major and undefined cellular target for the immunosuppressive effects of type I IFNs in chronic LCMV infection is not the Tregs, as IFNAR fl/fl x Foxp3 YFP-Cre mice have elevated viral titers, while LCMV Cl-13 infected mice treated with a neutralizing antibody against IFNAR have decreased viral titers [33,34]. Thus, an approach to target IFNAR in normal hosts for inhibition of Treg suppressive function in chronic infection or in cancer would be difficult.

Ethics statement
This study was carried out in strict accordance with the recommendations for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by National Institute of Allergy and Infectious Diseases Animal Care and Use Committee (Protocol No: LI-5E).

Mice
Foxp3 YFP-Cre (expressing Cre recombinase regulated by Foxp3 promoter) mice were purchased from Jackson Laboratories (Bar Harbor, ME). IFNAR -/mice were obtained by National Institute of Allergy and Infectious Diseases (NIAID), and maintained in Taconic Farms (Germantown, NY). IFNAR fl/fl mice were generously provided by Ulrich Kalinke (Paul-Ehrlich Institut, Langen, Germany), and crossed with Foxp3 YFP-Cre mice to generate Treg-lineage specific IFNAR-deficient mice. All strains of mice used in this study were age 8-12 weeks of age), gender matched, and bred in-house.

LCMV infection and plaque assay
LCMV Armstrong and Cl-13 viruses (Shevach Laboratory) were propagated in baby hamster kidney-21 fibroblast cells [American Type Culture Collection (ATCC), Manassas, VA]. Viral titers were determined by plaque assay using Vero African-green-monkey kidney cells (ATCC). Viral stocks were frozen at -80 o C until use. Mice were infected with the diluted virus in 1x sterile phosphate buffer saline (PBS) (Armstrong virus, 2x10 5 plaque forming unit (pfu)/ mouse, i.p., or Cl-13 virus, 2x10 6 pfu/mouse, i.v.). LCMV titers in sera and organs were determined by plaque assay using Vero cells as described [78].

Lymphocytes isolation
Spleens were harvested from naive (uninfected) and infected mice on indicated days and homogenized the tissues using cell strainer (

Cell surface, intracellular and tetramer staining
Cell surface staining was performed as described [79]. Briefly after harvest, spleen cells (

Gene expression profiling and bioinformatics analysis
Armstrong (day 5) and chronic LCMV infected (day 25) Foxp3 YFP-Cre and IFNAR fl/fl xFox-p3 YFP-Cre mice CD4 + YFP + sorted cells were lysed in RLT buffer (Qiagen, Valencia, CA). Total RNA was extracted using Qiagen AllPrep 96 DNA/RNA kit as described by the manufacturer (Qiagen, Valencia, CA), with one exception prior to the extraction, the RLT lysate was homogenized using Qiagen QIAShredder columns (Qiagen) to shear any contaminating gDNA. Samples were then subjected to on-column Dnase I treatment. All steps were performed using PCR amplicon-free laboratory equipment to further minimize background signal during RNA sequencing and library generation. A 150 ng aliquot from each sample was individually adjusted to 50 μl using nuclease-free water. Each sample was processed using Truseq Stranded mRNA Sample Preparation, Rev. E (Illumina Inc., San Diego, CA) using the included barcodes with the following modification: post-amplification libraries were purified with Ampure XP beads twice. The resulting DNA libraries were fragment-sized using a DNA1000 Bioanalyzer Chip (Agilent Technologies, Santa Clara, CA) and quantitated using KAPA Library Quant Kit with universal qPCR Mix (Kapa Biosystems, Wilmington, MA) on a CFX96 Real-Time System (BioRad, Hercules, CA). All eight-ten samples were diluted to a 2 nM working stock and pooled using equal volume amounts. An 11 pM titration point was used to cluster a paired end, RAPID 2-lane flowcell on a Hiseq 2500 DNA sequencer (Illumina). Libraries were run as 2 x 100 bp paired end reads on 2 lanes of an Illumina Hiseq 2500 sequencer, which produced 28.7 million reads per sample. Reads were trimmed for adapter sequence and filtered for low quality sequence using the FASTX-Toolkit. Remaining reads were mapped to the mouse genome assembly mm10 using Hisat2 [80]. Reads mapping to genes were counted using htseq-count [81]. Differential expression analysis was performed using the Bioconductor package DESeq2 [82]. Further analysis was performed using Partek Genomic Suite (Partek Incorporated) and Ingenuity pathway analysis (IPA) is used for obtaining top canonical pathways, networks based on co-expression, transcription factor binding sites and protein-protein interactions.

Gene set enrichment analysis
GSEA were performed on the set of 412 genes (Armstrong infection) using GSEA v2.2.3 from The Broad Institute [83]. GSEA was run using molecular signature database v.5.2 gene sets, except the C1: positional gene sets, with 1000 permutations and all default parameters except minimum size of 5.

Real time quantitative PCR
Total RNA samples from Foxp3 YFP-Cre and IFNAR fl/fl x Foxp3 YFP-Cre mice were extracted as described in the NGS gene expression profiling method section. cDNAs were prepared from Superscript IV first-strand cDNA synthesis kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. Presynthesized Taqman gene expression assays (Thermo Fisher Scientific) were used to amplify Akap1, Car2, Cpe, Eomes, Erdr1, Gpat2, Rab4a, Rell1, Rasgrf2, Sdc3, Tlr7, and Actb was used as internal control. Real time qPCR was performed using QuantStudio7 Flex Real time PCR system (Thermo Fisher Scientific) using Taqman universal master mix II with UNG (Thermo Fisher Scientific). Target gene expressions were calculated by 2 -dct and expressed as relative to Actb.

Quantification and statistical analysis
Flow cytometry data were analyzed using FlowJo software version 10.2 and or 10.3 (FlowJo LLC, Ashland, OR). Graphs were prepared by GraphPad Prism software version 7.0 (Graph-Pad Software, Inc. La Jolla, CA). Statistical analysis was done through unpaired two-tailed Student's t-test. All data in the graphs presented as Mean±SEM values, and error bars represent SEM. Data were considered statistically significant when P < 0.05, and represented as Ã P < 0.05, ÃÃ P < 0.01, ÃÃÃ P < 0.001, and ÃÃÃÃ P < 0.0001.

Data and software availability
RNA sequence information reported in this study is deposited in NCBI GEO under the accession number: GSE104517.