NK Cells Alleviate Lung Inflammation by Negatively Regulating Group 2 Innate Lymphoid Cells

Innate lymphoid cells (ILCs) are an emerging family of innate immune cells that function in lymphoid organogenesis, tissue remodeling, inflammation, infection, and cancer by playing critical roles in first-line immune defenses (1, 2). The ILC family (3) consists of a potent innate IFN-γ producer (i.e., classic cytotoxic NK cells) (4, 5) and noncytotoxic group 1 ILCs (ILC1s) (6), group 2 ILCs (ILC2s) (7), and group 3 ILCs (8), including LTi cells. Orchestrated by specific transcriptional factors (9) and regulated by various extrinsic soluble factors (10), members of the ILC family are functionally distinct, showing striking similarity with Th cell subsets in terms of cytokine expression and potential effector functions (1). However, despite an ever-growing list of reported factors that regulate each member of the ILC family, interactions among these subsets of ILCs have not been investigated.

Among the subsets of these ILCs are IFN-γ–producing NK cells, IFN-γ–producing ILC1s, and Th2 cytokine–producing ILC2s, all of which are reminiscent of the two counteracting subsets of T cells: Th1 and Th2. It has long been established that the antagonism between Th1 and Th2 during T cell differentiation is critical for the properly shaped T cell immune responses (11). Whether this classical paradigm of cellular interactions exists between corresponding family members of ILCs remains intriguing.

ILC2s and NK cells were reported to be involved in allergic airway diseases. Through production of Th2 cytokines, ILC2s serve an important role in orchestrating type II immune responses by participating in allergic (7, 12, 13) and fibrotic (14, 15) lung inflammation. In response to intranasal challenges, such as protease allergen papain (16) or the profibrogenic chemotherapeutic agent bleomycin (15), lung epithelial cells and/or macrophages produce stress-induced cytokines, including IL-25, IL-33, and thymic stromal lymphopoietin, leading to activation of ILC2s. Upon activation, ILC2-produced IL-5 mediates eosinophilia (17), whereas ILC2-derived IL-13 activates goblet cells and leads to mucus secretion, thus contributing to type II inflammation. In contrast, NK cells from asthma patients are activated and function to restrain inflammatory effector cells (18). In experimental mouse models, NK cells were shown to limit OVA-induced Th2 responses and to promote the resolution of allergic airway inflammation (19). Also, therapeutically augmenting NK activity by type I cytokine administration alleviated type II inflammation in the lung (20, 21). These reports shed light on the regulatory role of NK cells in controlling type II immune responses in the lung. However, the roles of NK cells in the regulation of type II immune responses and allergic diseases have not been fully revealed. We demonstrate that NK cells suppress ILC2 expansion and cytokine production and alleviate allergic and fibrotic lung inflammation in mice.

Six- to eight-week-old wild-type (WT) C57BL/6 mice were purchased from Beijing Vital River Laboratory Animal Technology (Beijing, China). C57BL/6 IFN-γ–knockout (GKO) mice were purchased from the Model Animal Research Center (Nanjing, China). All mice were maintained in a specific pathogen–free facility at the Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences. All procedures were approved by the Institutional Animal Care and Use Committee.

Polyinosinic-polycytidylic acid (poly I:C) (high m.w.; InvivoGen, San Diego, CA) was dissolved in pyrogen-free saline. Recombinant murine IL-2, IL-7, IL-12, and IFN-γ were purchased from PeproTech (Rocky Hill, NJ). Recombinant murine IL-33 was purchased from R&D Systems (Minneapolis, MN). Anti–IFN-γ was purchased from Bio X Cell (West Lebanon, NH).

For papain-induced allergic lung inflammation in mice, 80 μg of papain (Sigma Chemical, St. Louis, MO) was administered intranasally to anesthetized mice for three consecutive days. ILC2s and eosinophils were analyzed 2 d after the last administration. For bleomycin-induced fibrotic lung inflammation, anesthetized mice were administered 0.1 U bleomycin intranasally (Melone Pharmaceutical, Dalian, China). ILC2s and eosinophils were analyzed 7 d later. To elicit ILC2s in the liver, temporary overexpression of mcIL-33 (22) in the liver was achieved by hydrodynamic injection of 2 μg of pcDNA3.1-mcIL-33. Numbers and function of ILC2s, as well as eosinophil numbers in the liver, were assessed 3 d after injection.

Liver mononuclear cells were isolated essentially as previously described (23). Lung mononuclear cells were isolated as previously described (24).

For NK cell depletion, mice were injected with 200 μg of anti-NK1.1 mAb PK136 (Bio X Cell), or 50 μg anti-ASGM1 Ab (eBioscience), when indicated, 24–48 h before other experimental procedures.

For NK cell adoptive transfer, recipient mice were injected i.p. with a suboptimal dose (20 μg) of anti-NK1.1 mAb PK136. Two days later, recipient mice were hydrodynamically injected with 2 μg of pcDNA3.1-mcIL-33 for induction of ILC2s. Sixteen hours after plasmid injection, 5 × 10⁴ FACS-purified donor hepatic NK cells were injected (23) into the left lateral lobe of the liver of recipient mice. Donor hepatic NK cells were isolated and purified from mice that were hydrodynamically injected with 2 μg of pcDNA3.1-mcIL-33 14–18 h earlier. ILC2s and eosinophils in the left lateral lobe of the liver of recipient mice were analyzed 3 d post–NK cell transfer.

We purchased FITC–anti-CD3, FITC–anti-Gr-1, PerCP/Cy5.5 or allophycocyanin–anti-NK1.1, PE–anti mouse Lineage Cocktail, PerCP/Cy5.5 or allophycocyanin–anti-ST2, PE–anti-GATA3, PE–anti-CD11b, PE–anti-CD71, PE–anti-CD98, FITC–anti-KLRG1, PE or allophycocyanin–anti-IL-5, PE or allophycocyanin–anti-Ki-67, PE–anti-IFN-γ, allophycocyanin–anti-IL-17A, allophycocyanin–anti-Foxp3, PerCP/Cy5.5–anti-CD127, PerCP/Cy5.5–anti-CD4, PerCP/Cy5.5–anti-CD11c, allophycocyanin–anti-DX5, and BV510–anti-CD45 from BioLegend; eFluor 660–anti-Siglec-F and PE or allophycocyanin–anti-IL-13 from eBioscience; and FITC–anti-ARG1 from R&D Systems. Prior to staining with Abs, a LIVE/DEAD Fixable Violet Dead Cell Stain Kit (Life Technologies) was used to exclude dead cells. Then cells were incubated with rat Ig for 30 min to block Fc receptors. Along with Fc blocking, CountBright Absolute Counting Beads (eBioscience) were added, according to the manufacturer’s instructions, for determination of cell number. For intracellular cytokine staining of ILC2s, as well as of IFN-γ and IL-17A in T cells, total lymphocytes were stimulated in vitro for 3 h with Cell Stimulation Cocktail (plus protein transport inhibitors) (eBioscience). ILC2s were gated on CD45⁺ Lin⁻ ST2⁺ KLRG1⁺ CD90⁺ LIVE/DEAD Violet⁻ single cells (Supplemental Fig. 1). For intracellular cytokine staining of cells other than ILC2s, Th1, and Th17, total lymphocytes were not stimulated in vitro before Ab staining. Eosinophils were gated on SSC^hi CD45⁺ CD11b⁺ Gr1⁺ Siglec-F⁺ CD11c⁻ LIVE/DEAD Violet⁻ cells. Subsets within CD3⁻ NK1.1⁺ cells were gated within CD45⁺ CD3⁻ NK1.1⁺ LIVE/DEAD Violet⁻ single cells, as indicated. Cells were then fixed and permeabilized using Foxp3 Fix/Perm solution (BioLegend) before intracellular cytokine staining.

A FACSAria III cell sorter (BD Biosciences) was used to purify CD3⁻NK1.1⁺ NK cells or CD45⁺ Lineage-ST2⁺ CD90⁺ KLRG1⁺ ILC2s. The purity of sorted NK and ILC2 populations was >95%, as verified by postsort flow cytometric analysis.

NK cells and ILC2s were cultured in RPMI 1640 medium (Corning) containing 10% FCS and 1× Penicillin-Streptomycin solution (HyClone), plus various stimulants, if indicated. Sorted ILC2s from the liver of mice treated with pcDNA3.1-mcIL-33 were rested for 5–7 d with 10 ng/ml IL-2 alone before seeding at 10,000 cells per well in 96-well plates for experimental treatments. To activate and expand ILC2s, rested ILC2s were cultured with 10 ng/ml IL-2, 10 ng/ml IL-7, and 10 ng/ml IL-33 for 3 d before flow cytometric analysis. For poly I:C–activated NK cells, NK cells were purified from total splenocytes of mice injected i.p. with 150 μg of poly I:C in 200 μl saline 18 h earlier. When indicated, splenocytes, splenocytes from NK cell–depleted mice, or sorted splenic NK cells were stimulated with 10 ng/ml IL-12 for 6 h, washed three times with culture medium, and cocultured with ILC2s. Anti–IFN-γ Abs (10 μg/ml) were used to neutralize IFN-γ in ILC2 cultures when indicated.

Lungs were fixed with 4% paraformaldehyde and embedded in paraffin, and 4-μm sections were stained with periodic acid–Schiff. Image was acquired with a Pannoramic MIDI slide scanner and analyzed with Pannoramic Viewer software (both from 3DHISTECH).

Amounts of IL-5 and IL-13 in bronchoalveolar lavage fluid (BALF), serum, or culture supernatant were determined using ELISA kits from Dakewe Biotech (Shenzhen, China) and BosterBio (Wuhan, China), respectively.

Statistically significant differences were determined using the Student t test (analyses between two groups) or ANOVA (analyses of more than two groups). The p values < 0.05 were considered significant.

To study the role of NK cells in the regulation of ILC2s, we first monitored changes in NK cells when ILC2s (Supplemental Fig. 1) were activated during the early stage of lung inflammation (Fig. 1A, 1B). In papain-induced allergic inflammation and bleomycin-induced fibrotic inflammation, increased NK cell numbers were observed in the lung (Fig. 1C, 1D), whereas such changes were not observed for other lymphocyte populations in this early stage (Fig. 1C–F). Also, an upregulation in IFN-γ production was observed in NK cells after papain (Fig. 1G) or bleomycin (Fig. 1H) administration, indicating an activated phenotype for NK cells. These data demonstrate that NK cells are activated during the early stage of lung inflammation.

To reveal the effects of NK cell activation on ILC2 expansion and activation, we cocultured the sorted poly I:C–activated NK cells and ILC2s in vitro. In the presence of ILC2-activating IL-2, IL-7, and IL-33, significant reductions in ILC2 expansion and IL-5 and IL-13 production were observed in the presence of NK cells in a dose-dependent manner (Fig. 2A). NK cells are robust IFN-γ producers upon activation (4). The receptor for IFN-γ was reported to be expressed on ILC2s (25–27), which was further induced upon cytokine stimulation (26). We found that the reductions in ILC2 activation in the presence of NK cells were compromised by neutralizing IFN-γ (Fig. 2A), suggesting that the inhibitory effects of poly I:C–activated NK cells on ILC2s might be mediated, at least in part, by IFN-γ. To confirm this speculation, we assessed ILC2 expansion and cytokine production in the presence of culture supernatant from poly I:C–activated NK cells (Fig. 2B). We observed similar inhibitory effects on ILC2s by culture supernatant of poly I:C–activated NK cells, which was also abrogated by the presence of anti–IFN-γ Ab (Fig. 2B). It is known that poly I:C indirectly triggers activation and IFN-γ production of NK cells in mice (28). We did not observe any inhibitory effects of poly I:C directly on ILC2 expansion or cytokine production (Fig. 2C). In addition to in vivo–activated NK cells, we tested NK cells activated by IL-12 in vitro. The presence of splenocytes or splenocytes depleted of NK cells, either at rest or stimulated by IL-12, did not affect the expansion or cytokine production of ILC2s (Fig. 2D). However, expansion and cytokine production of ILC2s were significantly suppressed by the presence of NK cells (Fig. 2D, 2E), whose effects were abrogated by neutralization of IFN-γ and were augmented by IL-12 preactivation of NK cells (Fig. 2D). The percentage of Ki-67⁺ ILC2s was reduced significantly in the presence of IL-12–activated NK cells (Fig. 2E), which was rescued by neutralizing IFN-γ (Fig. 2E). GATA-3 is required for maintenance of ILC2 function (29). We did not observe any changes in GATA-3 expression in ILC2s after coculture with NK cells (Fig. 2F). However, in accordance with the above data that the optimal activation of ILC2s was compromised in the presence of NK cells, we found that NK cells significantly inhibited expression of the metabolic checkpoint, arginase-1 (30), in ILC2s in an IFN-γ–dependent manner (Fig. 2G). Also, decreased expression of the transferrin receptor CD71 on ILC2s was observed in the presence of NK cells, which was abrogated by anti–IFN-γ (Fig. 2G). In the meantime, amino acid transporter CD98 expression on ILC2s was not significantly changed by NK cells (Fig. 2G). Together, we showed that NK cells inhibited the proliferation and cytokine production of ILC2s via IFN-γ in vitro.

To explore whether NK cells also inhibited ILC2s in vivo, we used a model of expanding ILC2s in the liver by overexpression of IL-33 locally. IL-33 overexpression resulted in accumulation of ILC2s in the liver (Fig. 3A). In contrast, ST2, the receptor for IL-33, was reported to be expressed on NK cells from the liver (31), and ST2 expression on NK cells was reported to increase in lung inflammation caused by cigarette smoke (32). IL-33 could act directly on NK cells to promote their responses (31–33). We confirmed the expression of ST2 (Fig. 3B) on hepatic NK cells in the steady-state and observed elevated ST2 expression on hepatic NK cells upon IL-33 overexpression (Fig. 3B). Meanwhile, enhanced activation of NK cells was observed early upon IL-33 overexpression, as evidenced by increased accumulation (Fig. 3C) and elevated expression of CD69 (Fig. 3D) and IFN-γ (Fig. 3E) by NK cells. Although T cells also were activated after IL-33 overexpression (Fig. 3D, 3E), they represented a minor IFN-γ producer in this context (Fig. 3F). In contrast, CD3⁻ NK1.1⁺ cells in the liver were reported to include three populations with different phenotypes and functional properties (6, 34, 35). Although they are all potent IFN-γ producers, conventional NK cells were found to be the primary source of IFN-γ early after IL-33 overexpression in the liver (Fig. 3F). Meanwhile, tissue-resident NK cells, as well as mucosal ILC1s, represented minor IFN-γ producers in this context (Fig. 3F). These effects of IL-33 overexpression on NK cell activation were consistent with previous reports that IL-33 enhances NK responses (31–33) and agreed with our observations that NK cells were activated during the early stage of lung inflammation (Fig. 1), indicating that, by using this cytokine-driven model of ILC2 activation, we should be able to study the effects of NK activation on ILC2 function, as well as ILC2-related processes.

Using the in vivo model of cytokine-driven ILC2 expansion described above, we set out to study the effects of NK cell activation on ILC2 expansion and function. We found that NK cell depletion by PK136 increased cell number and enhanced IL-5 and IL-13 production of ILC2s, whereas ILC2 expansion and cytokine production were significantly impaired by NK cell–activating poly I:C (Fig. 4A, 4B). Consistent with this, serum IL-5 and IL-13 were reduced in response to poly I:C and were increased in mice depleted of NK cells using PK136 (Fig. 4C). IL-5–dependent eosinophilia was also significantly suppressed by poly I:C and was aggravated in mice pretreated with PK136 (Fig. 4D). Importantly, we found that the ILC2-suppressing effects of poly I:C were abrogated in mice depleted of NK cells using PK136 (Fig. 4A–D). Furthermore, pretreatment of mice with another NK cell–depleting reagent, anti-ASGM1 Ab, also led to increased hepatic ILC2s and eosinophils, as well as elevated serum levels of IL-5 and IL-13, in mice with IL-33 overexpression (Supplemental Fig. 2A, 2B). Collectively, these data indicate that NK cells inhibit ILC2 activation in vivo and that poly I:C inhibits ILC2 activation in vivo in an NK cell–dependent manner.

We showed above that NK cells inhibited ILC2s via IFN-γ in vitro (Fig. 2). In this study, we found that cytokine-induced ILC2s in the liver were enhanced in GKO mice (Fig. 4E). Elevated serum concentrations of IL-5 and IL-13 (Fig. 4F), as well as aggravated eosinophilia (Fig. 4G), were also prominent in these mice. These data confirmed the inhibitory effects of IFN-γ on ILC2 activation (26, 27, 36). More importantly, the ILC2-suppressing effects of poly I:C were not observed in GKO mice (Fig. 4E–G), indicating that NK cell–activating poly I:C inhibits ILC2 activation vivo in an IFN-γ–dependent manner.

To determine the role of NK cell–derived IFN-γ on IL-33–driven ILC2 activation in this context, we adoptively transferred WT or IFN-γ–deficient NK cells into the liver of WT or GKO mice after IL-33 overexpression. Significantly fewer ILC2s (Fig. 4H) and eosinophils (Fig. 4I) were detected in WT and GKO recipients that received WT donor NK cells than in those that received IFN-γ–deficient donor NK cells. In contrast, comparable levels of ILC2s (Fig. 4H) and eosinophils (Fig. 4I) were observed between WT and GKO recipients that received the same WT or the same IFN-γ–deficient donor NK cells. These data demonstrate that NK cells were a sufficient and essential source of IFN-γ in the suppression of ILC2 activation in the liver early upon IL-33 overexpression and were consistent with our findings that NK cells suppressed ILC2s in vitro via IFN-γ (Fig. 2). Taken together, our data demonstrate that NK cells suppress ILC2 activation and expansion in vitro and in vivo via IFN-γ.

We showed above that NK cells inhibited ILC2s in in vitro and in vivo cytokine-driven ILC2 activation models. Next, we set out to determine whether NK cells also regulate ILC2s in the early stage of lung inflammation when ILC2s are activated. In papain-induced allergic inflammation and bleomycin-induced fibrotic inflammation, depletion of NK cells significantly increased ILC2 numbers in the lung (Fig. 5A, 5B, Supplemental Fig. 2C, 2D). Meanwhile, T cell, but not B cell, responses in papain-treated mice were modestly enhanced (Fig. 5C), whereas T and B cells were not increased in bleomycin-treated mice (Fig. 5D). These data demonstrate that the presence of NK cells restrains ILC2 expansion in the early stage of lung inflammation.

In addition, we found that depletion of NK cells by PK136 led to enhanced production of IL-5 and IL-13 by ILC2s in papain-induced (Fig. 6A) and bleomycin-induced (Fig. 6B) lung inflammation. In the absence of NK cells, the amounts of IL-5 and IL-13 were also increased in BALF of papain-challenged (Fig. 6C, Supplemental Fig. 2E) and bleomycin-challenged (Fig. 6D, Supplemental Fig. 2F) mice. In accordance with such changes in ILC2s, we found that the absence of NK cells also resulted in significantly aggravated IL-5–dependent eosinophilia in the lung, as shown by increased numbers of eosinophils in lung tissue (Fig. 6E, Supplemental Fig. 2C, 2D) and BALF (Fig. 6F, Supplemental Fig. 2C, 2D) of papain- and bleomycin-challenged mice. Furthermore, IL-13–dependent hyperplasia of goblet cells was observed in papain-challenged mice pretreated with PK136 (Fig. 6G). These data demonstrate that the presence of NK cells normally restrains ILC2 responses and alleviates lung inflammation during the early stage.

In this article, we show for the first time, to our knowledge, interactions between families of ILCs. Our data demonstrated that activation of NK cells suppressed ILC2 expansion and cytokine production in vitro and in vivo. We found that such regulatory effects of NK cells are primarily mediated by IFN-γ. With regard to ILC2 activation in the lung in response to cytokine administration or infection, as well as for ILC2 activation in the visceral adipose tissue (VAT), the regulatory role of IFN-γ was studied (26, 27, 36). IFN-γ repressed ILC2 function induced by IL-33 alone (26), IL-2 plus IL-33 (36), IL-7 plus IL-33 (27), and IL-2 plus IL-25 (26) in vitro. STAT1 was found to be responsible for the IFN-γ–mediated suppression of ILC2 activation induced by IL-33 alone (26) and IL-7 plus IL-33 (27) in vitro. Exogenous IFN-γ repressed ILC2 activation in the lung in response to administration of IL-2 plus IL-25 (26) or IL-33 alone (26, 27, 36), as well as to infection of Nippostrongylus brasiliensis (26). Aging and a high-fat diet, which are associated with increased level of IFN-γ, and the constitutively excessive expression of IFN-γ in Yeti mice, resulted in repressed ILC2 activation in VAT (36). Yeti mice or mice coinfected with Listeria monocytogenes showed diminished ILC2 activation in the lung in response to helminth infection (36). Ifngr1 deficiency, which leads to a defect in the action of IFN-γ, resulted in elevated ILC2 activation and exacerbation of type 2 immune responses in the lung caused by N. brasiliensis infection or IL-33 administration (26). Despite these data on the role of IFN-γ, the source of IFN-γ has not been clearly revealed. Results from different lines of IFN-γ reporter mice showed that NK cells were potentially robust IFN-γ producers in the lung and VAT (26, 36). In this article, we confirmed that NK cells are robust IFN-γ producers by intracellular staining of IFN-γ protein, and we further found that, early after IL-33 overexpression in the liver, NK cells are the major IFN-γ producers in the liver. Importantly, by adoptively transferring WT or IFN-γ–deficient NK cells into NK cell–depleted WT or GKO recipients, we showed that NK cell–derived IFN-γ was essential for IFN-γ–mediated suppression of ILC2 activation in this context. We also showed that NK cells suppressed expression of the ILC2-intrinsic metabolic checkpoint, arginase-1 (30), as well as metabolism-related receptor CD71, but not its critical transcriptional factor, GATA-3 (29), in ILC2s in an IFN-γ–dependent manner, suggesting that downregulating the optimal metabolic status might contribute, at least in part, to the ILC2-suppressive effects by NK cells/IFN-γ. Overall, our observations that NK cells suppressed ILC2s via IFN-γ were reminiscent of the classic antagonism between Th1 and Th2 differentiation of T cells, in which IFN-γ, the major effector cytokine of Th1 cells, suppresses Th2 differentiation (11).

NK cells were reported to show activated phenotypes in peripheral blood of asthma patients (18); however, the roles of NK cells in allergic airway diseases remain controversial in experimental mouse models (18, 19, 37, 38), indicating a context-dependent, or even stage-dependent, role for NK cells in these diseases. Our study focused on the effects of NK cells on ILC2 activation during the early stage of allergic and fibrotic inflammation in the lung, when the activity of NK cells are triggered. We showed that an absence of NK cells enhanced ILC2 responses and aggravated early lung inflammation. The ILC2-regulating effects of NK cells also were observed in in vitro and in vivo models of cytokine-driven ILC2 activation. However, whether NK cells also regulate ILC2 responses in later stages of type II immune responses, as well as in those that happen outside the lung and the liver, needs further investigation.

Our study illustrates the novel cellular interaction between two families of ILCs: NK cells and ILC2s. We show that activation of NK cells during the early stage of allergic and fibrotic lung inflammation limits ILC2 expansion and cytokine production, protecting the host from aggravated type II immunopathologies in the lung. Our results support the use of type I cytokine administration for treating type II allergic and fibrotic diseases, in that such treatments might trigger NK activity to restrain ILC2 function, therefore alleviating type II immunopathologies.

The authors have no financial conflicts of interest.