STAT1 Represses Cytokine-Producing Group 2 and Group 3 Innate Lymphoid Cells during Viral Infection

A carefully orchestrated immune response is vital for effective host defense and tissue homeostasis. Both innate and adaptive immune responses are precisely regulated to respond productively to a variety of different environmental insults while simultaneously restricting detrimental host immunopathology. Lymphocytes, both innate and adaptive, exhibit the expected functional diversity to accomplish this arduous task.

Innate lymphoid cells (ILC) are an expansive class of cells that derive from the common lymphoid progenitor and respond rapidly to cytokine stimuli in an Ag-independent manner (1). ILC have been classified into three major subsets, ILC1, ILC2, and ILC3, which largely mirror adaptive CD4⁺ Th1, Th2, and Th17 cells, respectively, in both transcriptional regulation and cytokine production. ILC1 include classical NK (cNK) cells as well as non-NK cell helper-like ILC1 and produce predominantly IFN-γ. ILC2 express high levels of IL-5 and IL-13. ILC3 include NKp46⁻ ILC3, NKp46⁺ ILC3, and lymphoid tissue inducer cells, with marked production of IL-17 and/or IL-22. ILC are activated by the local cytokine milieu including IL-12, IL-15, and IL-18 (ILC1); IL-33, TSLP, and IL-25 (ILC2); and IL-23 and IL-1β (ILC3). Moreover, several endogenous signals have been identified as inhibitors of ILC including prostacyclin (PGI₂), 1,25-dihydroxyvitamin D3, and the natural Ras-ERK inhibitor Spred1 (2–4). In particular, the activation of ILC2 induces detrimental immunopathology in the context of respiratory syncytial virus (RSV), influenza, and rhinovirus infection (5–9). Similarly, the activation of CD4⁺ Th2 and Th17 cells has previously been associated with maladaptive immunopathology and enhanced disease severity in the context of viral infection (10–14). Conversely, ILC1 have been shown to produce antiviral mediators in the context of viral infection and/or enhance viral clearance (15–17).

Coordinated responses to viral infection, namely the enhancement of antiviral Th1 cells and the restriction of Th2 and Th17 cells, are accomplished by several mechanisms. Multiple, unique signals are integrated to either promote or inhibit Th1, Th2, or Th17 cells. Notably, type I, II, and/or III IFNs are produced at high levels during viral infection and have convergent signaling through STAT1 (18). The presence of IFNs in the polarizing cytokine milieu strongly promotes the generation of Th1 cells and IFN-γ, whereas restricting Th2 and Th17 polarization and their associated cytokine production (19–25). These data implicate IFNs and STAT1 signaling as a principle coordinator of CD4⁺ T cell subset balance during viral infection.

Although many discrete pathways for activation and inhibition of ILC have been defined, it remains unclear how ILC responses are orchestrated collectively to provide an effective antiviral response. Given the central role of STAT1 signaling during viral infection and in the regulation of CD4⁺ T cell subset balance, we hypothesized that STAT1 signaling promotes ILC1 responses and restrains ILC2 and ILC3 responses to viral infection. To test this hypothesis, we evaluated cytokine-producing ILC responses in Stat1^−/− mice infected with RSV, a major pathogen of infants and the elderly with robust clinical morbidity in both developed and undeveloped countries and a particularly high mortality rate worldwide. ILC were significantly dysregulated in Stat1^−/− mice, with diminished IFN-γ⁺ ILC1 and enhanced IL-5⁺ ILC2, IL-13⁺ ILC2 and IL-17A⁺ ILC3 populations. This dysregulation was associated with increased production of ILC-associated cytokines in the lungs, pathophysiologic changes in the airways, and poor viral clearance. Concurrently, Stat1^−/− mice had significantly increased expression of IL-33 and IL-23, and abrogating IL-33 or IL-23 attenuated the cytokine-producing ILC2 and ILC3 responses in these mice, respectively. Together, these data demonstrate the critical role of STAT1 signaling in the proper regulation of ILC responses to viral infection.

Female 8- to 12-wk-old mice were used for all experiments. BALB/cJ and CByJ.SJL(B6)-Ptprc^a/J CD45.1⁺ congenic mice were obtained from The Jackson Laboratory. Stat1^−/− and Il33^−/− mice on a BALB/c genetic background were generated as previously described (26–28). Stat1^−/− Il33^−/− double-deficient mice were bred from these strains. Mice were housed in microisolator cages under specific pathogen-free conditions. All animal experiments were approved by the Vanderbilt University Medical Center Institutional Animal Care and Use Committee and were conducted in compliance with the 1996 Guide for the Care and Use of Laboratory Animals prepared by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, National Research Council.

RSV strain 01/2-20 was isolated from a patient in the Vanderbilt Vaccine Clinic and propagated in HEp-2 cells as previously described (29, 30). Mice were anesthetized with a ketamine-xylazine solution and inoculated with 3.0 × 10⁶ PFU of RSV 01/2-20 by intranasal instillation with 100 μl of viral preparation or mock preparation (lysed uninfected HEp-2 cells). Mice were sacrificed and harvested for tissue on days 0–9 postinfection. Viral titer was assessed by immunodetection plaque assay (31). In brief, lungs were collected and homogenized in 1 ml of MEM media using a BeadBeater (29). Debris was removed by centrifugation, and serial dilutions of the homogenate were plated on 80% confluent HEp-2 cells for 1 h while shaking. The cells were then overlaid with MEM supplemented with 10% FBS, penicillin G, streptomycin, gentamicin, amphotericin B, and 0.75% methylcellulose and incubated at 37°C for 6 d. Cells were subsequently fixed with methanol at −80°C for 2 h, blocked with 5% nonfat milk/PBS buffer, incubated at 37°C for 1 h each with primary goat anti-RSV (catalog no. AB1128, 1:250 dilution; EMD Millipore) and secondary HRP-conjugated donkey anti-goat (catalog no. 705-035-147, 1:500 dilution; Jackson ImmunoResearch) Abs, and visualized with 4-chloronaphthol.

Flow cytometric analysis was performed as previously described (5). In brief, lungs were minced and enzymatically disrupted with 1 mg/ml collagenase IV and 50 U/ml DNase I in RPMI 1640 with 5% FBS for 1 h at 37°C. Enzymes were inactivated with EDTA, and the disrupted lungs were passed through 70-μm strainers to generate single-cell suspensions. RBCs were lysed and cells were restimulated with 10 ng/ml PMA and 1 μmol/l ionomycin in the presence of 0.07% monensin in IMDM media supplemented with 10% FBS, 0.01 mmol/l nonessential amino acids, penicillin-streptomycin, and 1 mmol/l sodium pyruvate for 5 h at 37°C. In some experiments, cells were treated during the restimulation 1 h before collection with 0.1% BSA in PBS (vehicle) or 10 ng/ml recombinant murine IFN-α (catalog no. 12100-1; R&D Systems), IFN-β (catalog no. 12400-1; R&D Systems), IFN-γ (catalog no. 50-813-664; PeproTech), IFN-λ2 (catalog no. 4635-ML-025; R&D Systems), or IL-27 (catalog no. 2799-ML-010; R&D Systems). Restimulated cells were stained with a fixable viability dye (catalog no. 13-0868; Tonbo) and combinations of cell surface markers. Cells were fixed and permeabilized (catalog no. TNB-0607-KIT; Tonbo) per manufacturer’s instructions and stained for intracellular markers. All Abs used for flow cytometric analysis are listed in Supplemental Table I. Cells were subsequently evaluated on a BD LSR II Flow Cytometer and analyzed using FlowJo Software Version 10. cNK cells were defined as CD3⁻ CD19⁻ NKp46⁺ CD49b⁺ T-bet⁺ EOMES⁺ cells that produced IFN-γ. Non-NK cells ILC1 were defined as Lin⁻ CD127⁺ T-bet⁺ EOMES⁻ NKp46⁺ RORγt⁻ cells that produced IFN-γ. ILC2 were defined as Lin⁻ CD45⁺ CD25⁺ CD127⁺ cells that expressed either IL-5 or IL-13. ILC3 were defined as Lin⁻ CD90⁺ CD127⁺ RORγt⁺ cells that expressed either IL-17A or IL-22. Lineage markers (Lin) include CD3, CD5, B220, CD11b, Gr-1, 7-4, and Ter-119. Gating strategies are shown in Supplemental Fig. 1 (cNK and ILC1) and Supplemental Fig. 2 (ILC2 and ILC3).

Lungs were isolated and homogenized in 1 ml of MEM media using a BeadBeater. Cell debris was removed by centrifugation, and supernatants were plated for ELISA. ELISAs for IL-5 (catalog no. M5000), IL-13 (catalog no. M1300CB), IL-17A (catalog no. M1700), IL-22 (catalog no. M2200), IL-12 p70 (catalog no. M1270), IL-15 (catalog no. DY447), IL-18 (catalog no. 7625), IL-33 (catalog no. DY3626), TSLP (catalog no. MTLP00), IL-25 (catalog no. DY1399), IL-1beta (catalog no. MLB00C), and IL-12p40 (catalog no. DY2398) were obtained from R&D Systems and performed as per manufacturer’s instructions.

Lungs were collected and homogenized in 1 ml of TRIzol reagent (catalog no. 15596018; Ambion) using a BeadBeater. RNA was isolated per TRIzol manufacturer’s instructions. Samples were DNase treated (catalog no. 18068-015; Invitrogen) and cDNA was generated using SuperScript III (catalog no. 18080-051; Invitrogen) per manufacturer’s instructions. cDNA abundance was assessed using FAM-MGB TaqMan primers for Il23p19 (catalog no. Mm01160011_g1), Il33 (catalog no. Mm00505403_m1), and Gapdh (catalog no. Mm99999915_g1) from Applied Biosystems. Amplifications were carried out on an Applied Biosystems QuantStudio 12k Flex Real-Time PCR machine, and data were analyzed using Applied Biosystems QuantStudio 12k Flex Software v1.2.2. Il23p19 expression was normalized to Gapdh and expressed as a fold change calculated by comparing Stat1^−/− with wild-type (WT) samples by the ΔΔCt method at each time point.

Endotracheal tubes were placed into euthanized mice. Saline (0.8 ml) was instilled into the lungs and recovered with a syringe attached to the endotracheal tube. A total of 0.1 ml of the recovered solution was spun onto slides, and these cells were stained (catalog no. 22-050-272; Richard-Allan Scientific) to visualize macrophages, lymphocytes, neutrophils, and eosinophils. Two hundred cells on each slide were counted to assess cell-type frequencies. Total numbers were determined by multiplying the cell frequencies by the total number of cells recovered from the bronchoalveolar lavage (BAL). In some experiments, BAL fluid was centrifuged to remove cells and the supernatants were plated for ELISA.

To evaluate mucus, we instilled lungs with 0.8 ml of 10% neutral-buffered formalin and immersed them in excess 10% neutral-buffered formalin for 24 h at room temperature. Lungs were subsequently paraffin embedded and sectioned (5 μm) for histologic analysis. Slides were stained with periodic acid–Schiff (PAS). Small- and medium-sized airways were quantified for airway mucus by a trained pathologist blinded to the experimental conditions using the following scoring matrix: 0, no PAS⁺ cells observed in cross sections of medium to small airways; 1, <10 PAS⁺ cells observed in cross sections of medium to small airways; 2, >10 PAS⁺ cells observed in cross sections of medium to small airways; 3, >10 PAS⁺ cells observed in cross sections of medium to small airways with mucous strands observed in air spaces; or 4, >10 PAS⁺ cells observed in cross sections of medium to small airways with mucous plugging of airways.

Female 6-wk-old BALB/cJ or Stat1^−/− mice were lethally irradiated with 10 Gy of [¹³⁷Cs]. A 1:1 mixture of whole bone marrow was prepared from 6-wk-old CByJ.SJL(B6)-Ptprc^a/J CD45.1⁺ congenic and Stat1^−/− CD45.2⁺ mice. This bone marrow mixture was transferred via retroorbital injection into irradiated mice and allowed to reconstitute for 6 wk before RSV infection and analysis. Mice were maintained on antibiotic water (0.025% trimethoprim/0.125% sulfamethoxazole) for 2 wk after transplant.

Stat1^−/− mice were treated on days 0 and 3 postinfection i.p. with 250 μg of an anti–IL-23p19 subunit neutralizing mAb (Amgen) or an isotype control (clone MOPC-21; BioXCell). Lungs were isolated at day 6 postinfection for flow cytometric analysis.

All data were collated and analyzed in GraphPad Prism Version 5. Whenever possible, data were pooled from multiple experiments for analysis. Statistical analyses including Student t test, one-way ANOVA with Bonferroni posttest, or two-way ANOVA with Bonferroni posttests were performed using GraphPad Prism Version 5 as appropriate and as documented throughout the text.

To test the role of STAT1 signaling on cytokine-producing ILC populations during viral infection, we infected Stat1^−/− mice intranasally with RSV or mock inoculum (uninfected HEp-2 cell lysate) and measured ILC1, ILC2, and ILC3 cell numbers throughout the course of infection. Gating for ILC1, ILC2, and ILC3 are described in the 2Materials and Methods and representative gating are shown in Supplemental Fig. 1 (cNK and ILC1) and Supplemental Fig. 2 (ILC2 and ILC3). We defined ILC subsets by their cytokine production, because our primary interest in these studies was in understanding the balance of type 1, 2, and 17 cytokine production by ILC. Relative to WT mice, RSV-infected Stat1^−/− mice exhibited a significant decrease in the total number of IFN-γ⁺ cNK cells at days 3 and 6 postinfection and a significant increase in the total number of IL-5⁺ ILC2, IL-13⁺ ILC2, and IL-17A⁺ ILC3 at days 6 and 9 postinfection (Fig. 1). There were no significant increases throughout the course of infection in IFN-γ⁺ non-NK ILC1 or IL-22⁺ ILC3 with RSV infection compared with mock infection in either WT or Stat1^−/− mice, except a very modest but statistically significant increase in IL-22⁺ ILC3 at day 9 postinfection in Stat1^−/− mice (Fig. 1A, 1C). The IL-17A⁺ ILC3 were exclusively NKp46⁻, consistent with previous reports of ILC3 in the lungs (data not shown) (32).

Local proliferation of ILC populations is thought to be an important driver of inflammatory responses, including during RSV infection (5, 33, 34). To assess whether ILC were replicating in situ, we measured the expression of the cell cycle–associated protein Ki67 in IL-5⁺ or IL-13⁺ ILC2 and IL-17A⁺ ILC3. RSV-infected Stat1^−/− mice had increased expression of Ki67 in the IL-13⁺ ILC2 and IL-17A⁺ ILC3 compartments, but not the IL-5⁺ ILC2 compartment, compared with RSV-infected WT mice (Fig. 2). These data suggest that local proliferation of IL-13⁺ ILC2 and IL-17A⁺ ILC3 contribute to the increased numbers of these cells in Stat1^−/− mice during RSV infection.

Concurrent with the increases observed in cytokine-producing ILC2 and ILC3, we identified significantly increased expression of the cytokines IL-5, IL-13, and IL-17A at day 6 postinfection in RSV-infected Stat1^−/− mice compared with WT mice (Fig. 3A). Although unlikely to be the exclusive source of these cytokines, the increased expression of IL-5, IL-13, and IL-17A correlates with an increase in the number of cytokine-expressing ILC. We also identified robust BAL eosinophilia and neutrophilia and airway mucous accumulation at day 6 postinfection in RSV-infected Stat1^−/− mice (Fig. 3B, 3D, 3E). Importantly, Stat1^−/− mice exhibited poor viral control, as evidenced by both increased viral titers and delayed viral clearance compared with WT mice (Fig. 3C). These cytokine, BAL, and viral load data are broadly consistent with prior reports of a dysregulated Th2/Th17-skewed immune response in Stat1^−/− mice after RSV infection (27, 35–41). Collectively, these data demonstrate a critical role for STAT1 signaling in promoting IFN-γ⁺ ILC1 and restraining IL-5⁺ ILC2, IL-13⁺ ILC2, and IL-17A⁺ ILC3 during RSV infection that largely parallels the ineffective, pathologic immune response occurring in the lungs.

To interrogate whether cell-extrinsic factors contribute to the skewed cytokine-producing ILC response in Stat1^−/− mice, we measured the number of WT and Stat1^−/− ILC after RSV infection in a mixed bone marrow chimera model. WT (BALB/cJ) or Stat1^−/− mice were lethally irradiated and reconstituted via retroorbital injection with a 1:1 mixture of WT CD45.1⁺ (CByJ.SJL[B6]-Ptprc^a/J) or Stat1^−/− CD45.2⁺ whole bone marrow. Six weeks after reconstitution, mice were inoculated with RSV and the total numbers of cytokine-producing WT and Stat1^−/− ILC were enumerated at day 6 postinfection (Fig. 4). Mock-infected WT recipients had significantly more WT IL-5⁺ and IL-13⁺ ILC2 than Stat1^−/− IL-5⁺ and IL-13⁺ ILC2, suggesting a potential basal advantage of WT ILC2 in the WT recipient background (Fig. 4A–C). Upon infection, however, Stat1^−/− IL-5⁺ and IL-13⁺ ILC2 were significantly increased compared with WT IL-5⁺ and IL-13⁺ ILC2, demonstrating a cell-intrinsic advantage of IL-5⁺ and IL-13⁺ Stat1^−/− ILC2 during viral infection (Fig. 4A–C).

In Stat1^−/− recipient mice, both WT and Stat1^−/− IL-5⁺ and IL-13⁺ ILC2 were found in similar numbers with mock infection (Fig. 4A–C). Similar to WT recipient mice, Stat1^−/− IL-5⁺ and IL-13⁺ ILC2 were disproportionately activated upon RSV infection compared with WT IL-5⁺ and IL-13⁺ ILC2, although notably to a lesser degree than in the WT recipient, suggesting additional factors that may be present in the WT mouse that further potentiate cytokine-producing ILC2 responses (Fig. 4A–C). Collectively, these data indicate that cell-intrinsic differences are at least partially responsible for the enhanced responsiveness of Stat1^−/− ILC2 during RSV infection.

We similarly evaluated the number of IL-17A⁺ ILC3 in our mixed bone marrow chimera model (Fig. 4A, 4D). In WT recipients, we found equal numbers of WT and Stat1^−/− IL-17A⁺ ILC3, suggesting comparable fitness of these cells in the absence of a viral insult. Upon RSV infection, we identified significantly more Stat1^−/− IL-17A⁺ ILC3 than WT IL-17A⁺ ILC3, although the number of Stat1^−/− IL-17A⁺ ILC3 was comparable between mock and RSV infection in the WT recipient. In Stat1^−/− recipients, Stat1^−/− IL-17A⁺ ILC3 were found in higher frequencies in both mock and RSV-infected mice, with particularly robust expansion after RSV infection. In contrast, WT IL-17A⁺ ILC3 did not expand after RSV infection in the Stat1^−/− recipients (Fig. 4A, 4D). These data suggest that Stat1^−/− IL-17A⁺ ILC3 have a cell-intrinsic advantage over WT cells, and that this at least partially contributes to the dysregulation of IL-17A⁺ ILC3 during RSV infection. Moreover, these data suggest that an additional cell-extrinsic factor(s) present in the Stat1^−/− recipient may be required for dysregulation of the ILC3 population.

Our bone marrow chimera experiments suggest that both cell-intrinsic and cell-extrinsic factors may play a role in the dysregulation of cytokine-producing ILC during viral infection. Prior reports suggest that ILC2 are intrinsically susceptible to inhibition to type I and II IFNs, providing a likely explanation for the cell-intrinsic effect seen in our bone marrow chimeric mice (34, 42–45). However, STAT1-dependent cell-extrinsic factors that regulate ILC frequencies have not been described. The cytokine milieu is critical for the activation of ILC. We hypothesized that pro-ILC stimulatory cytokines were differentially regulated in Stat1^−/− mice compared with WT mice. We infected WT and Stat1^−/− mice with RSV and measured the expression of IL-12, IL-15, IL-18, IL-33, TSLP, IL-25, IL-1β, IL-23p19, and IL-12/IL-23p40 in the whole lung homogenate daily for 5 d beginning 12 h postinfection. All cytokines were measured by ELISA except IL-23, which was measured by quantitative PCR because of lack of sensitive and specific protein-based reagents. Strikingly, there was a delayed but substantial increase in IL-33 protein and mRNA in the whole lung homogenates, beginning at day 4.5 postinfection and continuing through day 5.5 postinfection with a statistically significant nearly 10-fold increase in Stat1^−/− mice compared with WT mice (Fig. 5D, 5E). IL-33 is stored as preformed cytokine in the nucleus of cells. To assess whether this increased expression of IL-33 in the whole lung homogenate led to an increase in the extracellularly available IL-33, we measured IL-33 concentrations in the BAL. We identified a significantly increased concentration of IL-33 in the BAL of RSV-infected Stat1^−/− mice compared with RSV-infected WT mice at day 5.5 postinfection, when peak levels of IL-33 were observed in the whole lung homogenate, indicating that IL-33 was being released into the airways (Fig. 5F).

Similar to IL-33, we observed a statistically significant 10-fold increase in Il23p19 expression in Stat1^−/− mice compared with WT mice, peaking at day 4.5 postinfection (Fig. 5J). IL-23 is a heterodimer formed between the IL-23p19 subunit and the IL-12/IL-23p40 subunit. IL-12/IL-23p40 was found in significantly lower levels throughout infection in Stat1^−/− mice compared with WT mice, although levels postinfection in Stat1^−/− mice were universally higher than those at the time of infection, suggesting an increased availability of the IL-12/IL-23p40 subunit for complexing with IL-23p19 and forming an active cytokine (Fig. 5K). IL-12, TSLP, and IL-1β were marginally, but statistically significantly, higher in WT mice at days 3.5, 0.5, and 0.5, respectively (Fig. 5A, 5G, 5I). Expression of IL-15, IL-18, and IL-25 was absent or detected at only very low levels (Fig. 5B, 5C, 5H). Given the central role of IL-33 and IL-23 in the activation of ILC2 and ILC3, respectively, these data suggest an ILC cell-extrinsic mechanism for their dysregulation during viral infection in Stat1^−/− mice. Importantly, the expression of IL-33 and IL-23 coincided with the onset of ILC dysregulation in Stat1^−/− mice, which occurred between days 3 and 6 postinfection (Fig. 1A–C).

To determine the role of increased IL-33 expression during RSV infection in Stat1^−/− mice, we generated Il-33^−/− Stat1^−/− double-knockout mice. We infected WT, Stat1^−/−, and Il-33^−/− Stat1^−/− mice with RSV or vehicle and measured the number of ILC2. RSV-infected Stat1^−/− mice had significantly increased numbers of IL-5⁺ and IL-13⁺ ILC2 compared with RSV-infected WT mice (Fig. 6A). However, compared with RSV-infected Stat1^−/− mice, Il-33^−/− Stat1^−/− mice had significantly decreased total numbers of IL-5⁺ and IL-13⁺ ILC2 (Fig. 6A). These data suggest that increased production of IL-33 contributes to the enhanced activity of IL-5⁺ and IL-13⁺ ILC2 in the Stat1^−/− mice compared with WT mice during viral infection.

Similarly, we sought to determine the role of differential IL-23 expression during RSV infection in Stat1^−/− mice. To assess this, we measured the number of IL-17A⁺ ILC3 in RSV-infected Stat1^−/− mice treated with an anti–IL-23p19 neutralizing Ab or an isotype control. RSV-infected Stat1^−/− mice had significantly increased numbers of IL-17A⁺ ILC3 compared with RSV-infected WT mice (Fig. 6B). However, compared with isotype-treated Stat1^−/− mice, Stat1^−/− mice treated with an anti–IL-23p19 neutralizing Ab had significantly decreased total numbers of IL-17A⁺ ILC3 (Fig. 6B). RSV-infected anti-IL-23p19-treated Stat1^−/− mice still had a significantly increased number of IL-17A⁺ ILC3 compared with WT RSV-infected mice, consistent with our bone marrow chimera experiments that suggested both cell-intrinsic and cell-extrinsic factors play a role in the dysregulation of IL-17A⁺ ILC3 during RSV infection (Figs. 4A, 4D, 6B). Collectively, these data suggest that increased production of IL-23 contributes to the enhanced activity of IL-17A⁺ ILC3 in Stat1^−/− mice compared with WT mice during viral infection.

ILC2 express IFN receptors and respond directly to IFNs (34, 44). We sought to determine whether IL-17A⁺ ILC3 also express IFN receptors. Initially, we evaluated for the expression of type I, II, and III IFNs in the first 5.5 d of RSV infection (Fig. 7A–D). We found that WT mice produced significantly more IFN-α, IFN-β, and IFN-λ during the first several days of RSV infection compared with Stat1^−/− mice. However, by day 5.5 postinfection, IFN-λ was detected in significantly higher levels in the lungs of Stat1^−/− mice compared with WT mice. No significant differences in IFN-γ expression were observed throughout the first 5.5 d of infection between WT and Stat1^−/− mice. We next assayed whether IL-17A⁺ ILC3 expressed receptors for these IFN species. Compared with isotype control staining, IL-17A⁺ ILC3 had detectable expression of IFN-αR, IFN-γR1, and the IL-27R (Fig. 7E). IL-27, like IFNs, signals through STAT1. Finally, we determined whether IFN treatment of IL-17A⁺ ILC3 ex vivo could induce phosphorylation of STAT1 to an activated state. Compared with vehicle treatment, culturing bulk lung cells ex vivo with IFN-β for 1 h significantly induced phosphorylation of STAT1 in IL-17A⁺ ILC3 (Fig. 7F). Similar phosphorylation was not observed with IFN-α treatment, and data were equivocal with high degrees of experiment-to-experiment variability for IFN-γ and IFN-λ2 treatments (Fig. 7F and data not shown).

The appropriate orchestration of immune cells is vital for promoting productive immune responses and restricting detrimental pathophysiology. In this article, we show that STAT1 is a broad regulator of cytokine-producing ILC responses in the context of viral infection, promoting antiviral IFN-γ⁺ ILC1 and restricting IL-5⁺ and IL-13⁺ ILC2 and IL-17A⁺ ILC3. To this end, STAT1 represses cytokine-producing ILC2 and ILC3 via both cell-intrinsic and cell-extrinsic mechanisms. Extrinsically, STAT1 deficiency led to a significant increase in the expression of IL-33 and IL-23 that correlated strongly with increases in IL-5⁺ and IL-13⁺ ILC2 and IL-17A⁺ ILC3 in these mice. Furthermore, the genetic deletion of IL-33 or the neutralization of IL-23p19 in the context of STAT1 deficiency partially attenuated the dysregulated activation of cytokine-producing ILC2 and ILC3, respectively. Collectively, these data demonstrate that cytokine-producing ILC responses are coordinated by STAT1-dependent cell-intrinsic and cell-extrinsic mechanisms in the context of viral infection.

Previous studies have started to elucidate the direct role of IFN signaling on ILC2 responses. Specifically, these studies have shown that type I and type II IFNs directly repress ILC2 responses in vitro and in vivo in the context of viral infection or airway inflammation (34, 42–45). Data from our bone marrow chimeric mouse experiments suggest cell-intrinsic factors, such as the previously described direct inhibition of ILC2 by IFNs (34, 43, 44), play a role in regulating ILC frequencies during viral infection. Consistent with this literature, we identified that type I, II, and III IFNs are expressed in the lungs throughout RSV infection, and that IL-17A⁺ ILC3 express receptors for type I and II IFNs, as well as IL-27, another activator of STAT1. These data suggest that ILC3 may also be modulated directly by IFNs. Our chimeric experiments further suggest that cell-extrinsic factors regulate ILC frequencies during viral infection via a previously undefined STAT1-dependent mechanism. These data were particularly strong for IL-17A⁺ ILC3, and consistent with these data we observed an increase in Il23p19 during RSV infection in Stat1^−/− mice. Although our bone marrow chimera experiments did not specifically identify a cell-extrinsic effect for STAT1 deficiency in regulating ILC2, we observed significantly enhanced production of IL-33 in Stat1^−/− compared with WT mice during RSV infection similar to Il23p19. Genetic deletion of Il33 in STAT1-deficient mice significantly reduced the number of IL-5⁺ and IL-13⁺ ILC2 during RSV infection. It is not entirely clear why a cell-extrinsic effect for cytokine-producing ILC2 was not observed in our bone marrow chimeric mice despite a clear effect of IL-33 in our other experiments. It is possible that changes in the cell phenotype such as a downregulation of the IL-33R posttransplant altered their capacity to respond, or that factors present in the WT recipient background in combination with donor Stat1^−/− ILC2 selectively potentiated these cells by a non-IL-33-dependent mechanism, masking our ability to detect a cell-extrinsic difference by this method. Regardless, our results using genetic deletion of Il33 in the context of STAT1 deficiency strongly suggest an important role for STAT1 repression of IL-33 in regulating cytokine-producing ILC2 responses.

Some other unexpected results were observed in our bone marrow chimera experiments. Specifically, WT donor IL-5⁺ ILC2 were observed at a high level in mock-infected WT recipient mice. This may be because of an advantage compared with other ILC subsets in reseeding the tissue after transplant. Additionally, Stat1^−/− IL-5⁺ and IL-13⁺ ILC2 were more frequent in WT recipient mice compared with Stat1^−/− recipient mice during RSV infection, suggesting that additional environmental factors present in WT mice may add to STAT1 deficiency in promoting cytokine-producing ILC2. Finally, RSV infection did not expand the number of WT IL-13⁺ ILC2, as had been observed in our intact animals at day 6 postinfection shown in Fig. 1B. The comprehensive network of activators and inhibitors of ILC in these mixed bone marrow chimeric mice will need to be further defined to better understand all of these data. However, broadly, these data in conjunction with previously published studies extend our understanding of the role of STAT1 signaling on the ILC response during viral infection.

A role for STAT1 in regulating IL-33 had been suggested previously, but these experiments were performed using ectopic expression of IL-33 and IFN-γ in lung fibroblasts in the context of intact or disrupted STAT1 signaling (46). Using an in vivo model of RSV infection, we are the first to show that endogenous STAT1 signaling is sufficient to negatively repress IL-33 expression and regulate immune cell numbers. Moreover, we found that STAT1 negatively regulated IL-23, consistent with our previous work (40). In this article, we report for the first time, to our knowledge, that this repression of IL-23 by endogenous STAT1 attenuates the IL-17A⁺ ILC3 response to viral infection.

We identified an increase in the proliferation marker Ki67 in IL-13⁺ ILC2 and IL-17A⁺ ILC3, but not IL-5⁺ ILC2. It is possible that proliferative burst occurs earlier in the IL-5⁺ ILC2 compartment. Alternatively, cytokine-producing ILC2 and ILC3 may also arise from the activation of existing quiescent ILC in the tissue, differentiation of new ILC in situ, recruitment of ILC from other tissues, or the shifting of one ILC class to another. Additional investigative approaches will be necessary to better understand how these cytokine-producing ILC2 and ILC3 are accumulating in the lungs of Stat1^−/− mice during RSV infection.

ILC subsets have been shown to participate in a variety of physiologic and pathophysiologic processes. Accordingly, significant attention has been paid to the identification of activators and inhibitors of ILC subsets that could be exploited therapeutically. For ILC2, such targets include lipid mediators PGI₂ (inhibitory), PGD₂ (activating), and LTD₄ (activating), as well as cytokines including IL-33 and TSLP (3, 47–51). Retinoic acid has been shown to promote ILC3 in the gastrointestinal tract, and accordingly dietary modification of retinoic acid derivatives including vitamin A may present a viable therapeutic target (52). However, an effective and nonpathologic immune response often requires the coordinated activation and inhibition of multiple cell types. Targeting pathways that orchestrate this broader response may present a high-yield, efficacious approach to treating disease and promoting health. Our data suggest that STAT1 signaling represents such a pathway, promoting beneficial antiviral ILC1 and repressing pathologic ILC2 and ILC3 responses by multiple mechanisms in the context of viral infection.

Endogenous mutations in STAT1 signaling alter disease susceptibility in humans. Chronic mucocutaneous candidiasis is a clinical syndrome characterized by recurrent infections of mucous membranes and skin with Candida species. A majority of these patients have been shown to have gain-of-function mutations in STAT1, which promote type 1 immune responses and repress type 17 responses, predisposing individuals to infection with fungal diseases (53–57). In addition, loss-of-function mutations in STAT1 have also been identified that confer significant susceptibility to viral and mycobacterial infections (58). Analysis of CD4⁺ Th subsets in the context of either gain-of function STAT1 mutations show a strong concordance between the STAT1 signaling and CD4⁺ T helper activity, with increases in STAT1 signaling promoting Th1 cells and restricting Th17 cells (55, 56). Our data suggest that ILC subsets are similarly regulated by STAT1 in mice, and it is intriguing to consider the role of ILC in human patients with mutations at the STAT1 locus, especially considering the prominence of ILC at mucosal sites.

Throughout, we focused our efforts on understanding the role of STAT1 signaling on cytokine-producing ILC, defining ILC1, ILC2, and ILC3 subsets stringently based on their capacity to produce canonical cytokines. Although this allowed us to key in on the effects of STAT1 signaling on the major functional outputs of ILC, we were limited using this approach in our ability to understand the entirety of the ILC pool including both cytokine-producing and non-cytokine-producing ILC. We chose to focus on cytokine production, as we found that during viral infection the expression of hallmark transcription factors, specifically GATA3, was not always present in IL-5– and IL-13–producing ILC. We reason that these cells may be ILC2 that have downregulated GATA3 as a negative feedback mechanism or possibly ex-ILC1/ILC3 that have shifted their cytokine production to a type 2 program but do not yet express detectable levels of GATA3 (data not shown). Similarly, it is unclear whether the changes in cytokine-producing ILC are due to changing total numbers of ILC subsets, increases in the percentage of those subsets expressing cytokines, or other changes in the phenotypes of these cells. Such analyses are beyond the scope of this work, because our primary focus was on the production of type 1, 2, and 17 cytokines from ILC during viral infection.

Our data demonstrate that STAT1 is a chief regulator of cytokine-producing ILC1, ILC2, and ILC3 during viral infection. STAT1 promoted IFN-γ⁺ ILC1 and acted to repress IL-5⁺ and IL-13⁺ ILC2 and IL-17A⁺ ILC3 via multiple mechanisms, both cell-intrinsic and cell-extrinsic. These data delineate a critical role of STAT1 in broadly orchestrating a productive cytokine-producing ILC response to viral infection.

We greatly appreciate the support and assistance of the Vanderbilt VANTAGE Core and the VUMC Flow Cytometry Shared Resource, where flow cytometry experiments were performed. The Vanderbilt VANTAGE Core is supported in part by a Clinical and Translational Science Award (Grant 5UL1 RR024975-03), the Vanderbilt Ingram Cancer Center (P30 CA68485), the Vanderbilt Vision Center (P30 EY08126), and National Institutes of Health/National Center for Research Resources (G20 RR030956). The VUMC Flow Cytometry Shared Resource is supported by the Vanderbilt Ingram Cancer Center (P30 CA68485) and the Vanderbilt Digestive Disease Research Center (DK058404).

The authors have no financial conflicts of interest.