Type 2 innate lymphoid cells: at the cross-roads in allergic asthma

IL-5 is essential for eosinophil development and activation [53], and therefore, ILC2-derived IL-5 is likely to affect eosinophils. Indeed, several studies suggest that activated ILC2s contribute to eosinophilic inflammation. For example, Doherty et al. showed that intranasal Alternaria or LTD₄ administration induced pulmonary eosinophilia in Rag2 ^−/− mice (lack T, B, and NKT cells but do have ILC2s), but not in Il7ra ^−/− mice, which lack all ILCs in addition to B and T cells [16, 17]. Likewise, Halim et al. showed that intranasal papain treatment increased the number of eosinophils in the lungs of WT and Rag1 ^−/− mice, but not in ILC-deficient Rag2 ^−/− Il2rg ^−/− mice. Furthermore, depletion of ILC2s in Rag1 ^−/− mice with neutralizing CD25 antibodies decreased papain-induced eosinophilia, whereas adoptive transfer of ILC2s into Rag2 ^−/− Il2rg ^−/− enhanced papain-induced eosinophilia [25]. These studies indicate that activated ILC2s induce pulmonary eosinophilia independent of the adaptive immune system.

In a subsequent study, Halim et al. investigated whether ILC2s affect eosinophilia in the presence of Th2 cells. Adoptive transfer of WT bone marrow cells into Rag2 ^−/− Il2rg ^−/− mice restored papain-induced and IL-25-induced pulmonary eosinophilia, whereas adoptive transfer of ILC2-deficient Rora ^sg/sg bone marrow cells did not. The presence of other pulmonary leukocytes, including neutrophils, macrophages, DCs, NK cells, NKT cells, T cells, and B cells, was not differentially affected in papain-treated mice that had received WT bone marrow cells versus mice that had received Rora ^sg/sg bone marrow cells [54]. Additionally, in an influenza A virus (IAV)-induced pulmonary inflammation model, in vivo depletion of ILC2s and T cells with a neutralizing anti-Thy1.2 antibody in WT mice reduced eosinophil numbers in the lung, which could be rescued by the adoptive transfer of total Thy1.1⁺ cells (which include ILC2 and T cells) but not CD3⁺ Thy1.1⁺ lung cells (which only include T cells) [22].

Together, these results indicate that ILC2s contribute to eosinophilia in mouse models of allergic asthma, both in the absence and presence of the adaptive immune system. It seems likely that this effect is mediated by ILC2-derived IL-5, yet no study has directly assessed this. The only indication that ILC2-derived IL-5 is sufficient to induce eosinophilia came from a recent study that used IL-5 reporter (Red5) mice to investigate pulmonary IL-5 production [19]. Red5 mice express a tdTomato(RFP)-IRES-Cre replacement allele at the translation start site of the endogenous Il5 gene, resulting in production of RFP instead of IL-5 upon commitment to Il5 translation [55]. Il5 ^red/red mice (which are IL-5 deficient) had strongly diminished numbers of eosinophils in the lungs after repeated chitin inhalation, whereas ILC2 numbers were not affected. Strikingly, RFP expression was found exclusively in ILC2s [19], indicating that ILC2-derived IL-5 is critically required for chitin-induced eosinophilia. In addition, ILC2-derived IL-13 may also contribute to allergen-induced pulmonary eosinophilia. Chitin-treated Il13 ^−/− Il4 ^−/− mice had reduced eosinophil numbers in the lung, whereas Il4 ^−/− mice had not. The source of IL-13 after challenge with chitin was confined to ILC2s as elucidated by IL-13 reporter mice (Il13 ^smart/smart) [19]. In support, Barlow et al. found that IL-25-induced eosinophilia in the BAL was decreased in Il13 ^−/− mice compared to WT mice, which could partly be rescued by the adoptive transfer of WT ILC2, but not by transfer of Il13 ^−/− deficient ILC2 [56]. Together, these reports suggest that ILC2-derived IL-5 and IL-13 contribute nonredundantly to eosinophilia. This corresponds with earlier findings on allergen-induced eosinophilia, which show that IL-5 promotes development and survival of eosinophils, whereas IL-13 mediates recruitment of eosinophils through secretion of eotaxins [57]. In line with this, Van Dyken et al. found that chitin challenge increased expression of eotaxin-1 in the lungs and that chitin-induced eosinophilia was decreased in mice lacking the eotaxin-1 receptor CCR3 [19].

IL-13 acts directly on airway epithelial cells and smooth muscle cells and thereby induces AHR and mucus hyperproduction in allergic asthma [6]. As ILC2s are capable of producing IL-13 in response to allergen exposure, it is proposed that ILC2s also contribute to the pathogenesis of these asthma symptoms.

This hypothesis was supported by an influenza (H3N1)-induced pulmonary inflammation study. H3N1 infection induced AHR in Rag2 ^−/− mice, which was abolished after ILC2 depletion with a neutralizing Thy1.2 antibody. In addition, H3N1 infection failed to induce AHR in Il13 ^−/− mice, which was restored by adoptive transfer of WT ILC2s [21]. ILC2-derived IL-13 also affected AHR in cytokine- and allergen-based mouse models of pulmonary inflammation. Kim et al. found that Il13 ^−/− mice showed reduced AHR after IL-33- or glycolipid (α-Galcer) inhalation compared to WT mice, which could be rescued by adoptive transfer of WT ILC2s [18]. Barlow et al. found similar results for IL-25-induced AHR, and importantly, they showed that adoptive transfer of WT ILC2s, but not Il13 ^−/− ILC2s, could rescue IL-25-induced AHR in Il13 ^−/− mice [56]. This indicates that ILC2-derived IL-13 is sufficient to induce AHR in response to IL-25 inhalation. However, the contribution of ILC2-derived IL-13 to allergen-induced AHR within the context of other IL-13-producing cells remains to be determined.

Besides the involvement in AHR, some studies suggest that ILC2s also contribute to mucus hyperproduction in allergic asthma models. For example, papain inhalation induced mucus hyperproduction in the lungs of WT and Rag1 ^−/− mice, but not in the lungs of Rag2 ^−/− Il2rg ^−/− mice [25]. Furthermore, adoptive transfer of WT bone marrow cells, but not Rora ^sg/sg bone marrow cells, restored papain-induced mucus hyperproduction in Rag2 ^−/− Il2rg ^−/− mice [54]. These studies are indicative for an effect of ILC2s on mucus hyperproduction, yet they do not prove that this effect is mediated via IL-13 secretion.

Although these interactions are less well established, it appears that ILC2s also interact with other innate immune cells that are involved in allergic asthma.

It was recently suggested that ILC2s may contribute to allergen-induced accumulation of alternatively activated macrophages (AAMs, also known as M2 macrophages) [19]. IL-4 and IL-13 stimulate differentiation into AAMs, which typically express arginase-1 and chitinases [58]. AAMs are generally considered to be anti-inflammatory but are also implied to play a role in pathogenic type 2 responses [59]. The study that investigated chitin-induced eosinophilia found that chitin-challenged ll13 ^−/− Il4 ^−/− mice had reduced AAM numbers in the lung, whereas Il4 ^−/− mice had not. IL-13 was exclusively expressed by ILC2s, which suggested that ILC2-derived IL-13 contributed to chitin-induced AAM accumulation [19].

The same authors crossed Il5 ^Red5/Red5 or Il13 ^{YetCre/YetCre} mice with mice that carry a Gt(Rosa)26 ^DTA allele, to create a mouse model in which cells get killed by diphtheria toxin α as soon as they commit to IL-5 or IL-13 production, respectively. Challenging these mice with chitin led to deletion of ILC2s and a reduction in AAM accumulation and eosinophilia, which supports earlier findings [19]. Surprisingly, ILC2 deletion also increased the levels of several inflammatory cytokines including TNFα, IL-1β, and IL-23 and increased the activation of IL-17A-producing γδ T cells, which caused an increase in airway neutrophilia. Intriguingly, chitin-treated IL-5- and IL-13-deficient mice had similar numbers of γδ T cells and neutrophils as WT controls, suggesting that the suppressive effects of ILC2s on γδ T cells and neutrophils were not mediated via IL-5 or IL-13 signaling [19].

Two recent reports showed colocalization of ILC2s and mast cells, suggesting an interaction between these cells. Barnig et al. found ILC2s in the close proximity of mast cells in the human lung [30]. Roediger et al. used intravital multiphoton microscopy to track the movement of ILC2s in the skin of healthy mice and found that ILC2s specifically interacted with mast cells. Interestingly, these ILC2s constitutively expressed IL-13, and in vitro experiments showed that IL-13 suppressed the release of TNFα and IL-6 by mast cells, suggesting an inhibitory effect of ILC2s on mast cells in this context [42]. Mast cells and ILC2s might also interact via mast cell-derived PGD₂ and LTD₄, which is released upon allergen challenge-induced FcεRI cross-linking and might recruit and activate ILC2s [16, 28‐30]. However, this remains to be confirmed in vivo and in the context of allergic asthma. Roy et al. reported that mast cell chymase has the ability to degrade IL-33, indicating that mast cells can have a regulatory function in the balance of IL-33, implicating a feedback mechanism on the activation of ILC2s [60].

One recent study suggested interactions between ILC2s and basophils, as clusters of ILC2s and basophils were found in inflamed human and murine skin. The presence of basophils preceded the recruitment of ILC2s to atopic dermatitis-like skin lesions in mice. Interestingly, basophil-derived IL-4 was required for the optimal recruitment of ILC2s to these skin lesions, and ILC2s expressed the IL-4 receptor, suggesting a novel type of ILC2 regulation [61]. However, it remains to be established whether this applies to ILC2s at other locations and whether ILC2s are also able to affect basophils. Furthermore, depletion of basophils in a papain-induced pulmonary inflammation model had no effect on pulmonary eosinophilia, which argues against a major role of basophils in pulmonary inflammation and ILC2 regulation [62].

ILC2s might affect the adaptive immune response by influencing Th2 cell differentiation (possibly via DCs), and/or by influencing the activity of primed Th2 cells or IgE-producing B cells. This question was assessed by Halim et al., who compared papain-induced Th2 cell generation in the mediastinal LN (mLN) of WT BMT mice versus Rora ^sg/sg BMT mice [64]. Via intracellular cytokine staining, they showed that Th2 cell differentiation occurred 6 days after papain treatment in the mLN of WT mice, and that this response was impaired in Rora ^sg/sg BMT mice compared to WT BMT mice. Similar results were found in HDM- or fungal protease-allergen-treated mice. Notably, the adoptive transfer of WT ILC2s into papain-treated Rora ^sg/sg BMT mice restored the number of Th2 cells [64]. Together, these results suggested that ILC2s are involved in driving Th2 cell differentiation in response to allergen encounter.

Although it has been shown that ILC2s accumulate in the mLN during airway inflammation, it remains the question whether they play a direct role in skewing naive T cells toward Th2 differentiation. As IL-4 plays a critical role in Th2 cell differentiation [65], it was therefore investigated whether ILC2s might promote Th2 cell differentiation via IL-4. Some studies report that pulmonary ILC2s have the ability to produce small amounts of IL-4 after in vivo activation with IL-25 or IL-33 [10], OVA[10], papain [14], and HDM [10, 66], as measured by ELISA or intracellular FACS staining. IL-4 production by ILC2s may depend on specific activating stimuli, as LTD₄, as shown in vitro. In contrast, IL-33 was not able to induce IL-4 production by ILC2s in vitro [16]. It remains to be determined whether ILC2s also produce IL-4 in the close proximity of naïve CD4⁺ T cells in vivo.

However, differentiation toward Th2 cells can occur in an IL-4-independent pathway, as shown by the induction of Th2 cells in IL-4-deficient mice after N. brasiliensis infection [67]. In concordance, Halim et al. found that papain-challenged Il4 ^−/− mice displayed a normal Th2 response. Interestingly, they found that papain-treated Il13 ^−/− mice had strongly decreased Th2 cell numbers in the mLN compared to WT mice. Intracellular cytokine staining showed that ILC2s were the predominant source of IL-13 after papain treatment, suggesting that ILC2-derived IL-13 promoted Th2 cell differentiation. This was confirmed in a transgenic TCR model, in which CFSE-labeled OVA-specific OT-II T cells were injected into WT BMT mice or Rora ^sg/sg BMT mice. OVA plus papain treatment induced normal T cell proliferation but failed to induce Th2 cell differentiation in Rora ^sg/sg BMT mice. This effect could be rescued by adoptive transfer of WT ILC2, but not by Il13 ^−/− ILC2, indicating that ILC2-derived IL-13 is required for papain-induced Th2 differentiation. Notably, the IL-13 receptor was present on DCs but not on CD4⁺ T cells, suggesting that this effect was mediated via DCs. Indeed, DCs with CD40⁺ expression (a surface marker that is implied to induce Th2 differentiation) were present at normal numbers in the lungs but strongly decreased in the mLN of Rora ^sg/sg BMT mice, which could be rescued by IL-13 injections. Subsequent in vitro migration assays showed that IL-13 increased the capacity of DCs to migrate toward a CCL21 gradient, which is highly expressed in the lymph node [64].

Together, these results indicate that ILC2s might facilitate Th2 differentiation via inducing efficient migration of activated DCs from the lung to the mLN [64]. The question remains whether ILC2s might skew DCs toward a pro-Th2 DC phenotype, e.g., expression of OX40L, ICOS-L, Notch ligand Jagged1, and secreting cytokine IL-6 but not IL-12.

In addition to an indirect, DC-mediated interaction, recent studies suggest that pulmonary ILC2s and CD4⁺ T cells might interact directly with each other [68, 69]. In vitro studies suggested that CD4⁺ T cells can activate ILC2s via IL-2 signaling, which corresponds with earlier findings of IL-2-mediated activation of ILC2s [32, 42]. Wilhelm et al. demonstrated that IL-2 treatment enhanced the autocrine IL-9 production by ILC2s, resulting in increased IL-5 and IL-13 [14]. In contrast, Mirchandani did not find an increased IL-9 production after IL-2 treatment [68], suggesting that IL-2 can have distinct stimulating effects on ILC2s in different settings.

On the other hand, the presence of ILC2s also enhanced anti-CD3/anti-CD28-induced CD4⁺ T cell proliferation [69] and type 2 cytokine production [68, 69], whereas IFN-γ production remained constant [69] or even decreased [68]. ILC2s and CD4⁺ T cells likely interacted in a cell contact-dependent manner, because separation of CD4⁺ T cells and ILC2s in a transwell system strongly decreased ILC2-dependent CD4⁺ T cells activation [68]. Drake et al. showed that increased type 2 cytokine production by cocultured ILC2s and CD4⁺ T cells was partly dependent on OX40L expression by ILC2s, likely through interaction with OX40 on CD4⁺ T cells [69]. OX40L expression by DCs has been shown before to drive Th2 cell differentiation [70]. In addition, Drake et al. found that ILC2-derived IL-4 might play a role in CD4⁺ T cell activation, since coculturing of Il4 ^−/− ILC2s with WT CD4⁺ T cells strongly diminished total IL-5 and IL-13 production as compared to coculturing WT ILC2s and CD4⁺ T cells [69]. However, it should be noted that the authors measured total cytokine production by cocultured ILC2s and CD4⁺ T cells and did not formally prove which cell type was the source of these cytokines [69]. In contrast, Mirchandani et al. found that ILC2-mediated activation of CD4⁺ T cells was independent of OX40- and IL-4 signaling [68]. Instead, their results suggested that ILC2s might interact with CD4⁺ T cells by presenting antigens on MHCII, as OVA-peptide-pulsed ILC2s could induce proliferation of OVA-specific transgenic DO11.10 T cells, which was blocked by adding a neutralizing MHCII antibody [68].

The potential capacity of ILC2s to present antigens was demonstrated by Oliphant et al. [13]. ILC2s that were isolated from murine mesenteric lymph nodes expressed MHCII, CD80, and CD86. Furthermore, they were able to endocytose and process antigen in vitro, albeit at a much lower efficiency than DCs and B cells, but could not induce proliferation of OVA-specific OT-II transgenic (OTII Tg) T cells after coculture unless OVA peptide was added [13]. The proliferation was abolished after using a neutralizing MHCII antibody or by using MhcII ^−/− ILC2s but also after combined treatment with neutralizing CD80 and CD86 antibodies [13]. ILC2s may provide T cells with costimulation on CD28 [13]. The capacity to present allergens in vivo awaits confirmation and in comparison to other APCs, like DCs. Additionally, the capacity to present antigens may be specific for certain ILC2 subsets. For example, mLN ILC2s had substantially higher MHCII expression than bronchoalveolar and lung ILC2s [13]. Coculturing of peptide-pulsed ILC2s also strongly enhanced ILC2 proliferation and type 2 cytokine production, as compared to controls lacking peptide. Interestingly, this effect was blocked by treatment with neutralizing antibodies against MHCII, CD80, CD86, or IL-2 [13]. These results implied that ILC2-expressed MHCII, CD80, and CD86 were not only required for ILC2-mediated activation of CD4⁺ T cells but also for activation of ILC2s themselves, possibly by stimulating CD4⁺ T cells to provide them with IL-2.

To investigate reciprocal activation between ILC2s and CD4⁺ T cells in vivo, cotransfer experiments were performed in murine models of allergic asthma [68, 69]. Il7ra ^−/− mice (which are deficient in ILCs and T cells) were reconstituted with ILC2 and/or CD4⁺ T cells from naive mice, and subsequently treated with a combination of OVA and a cysteine protease (bromelain)[69]. OVA plus bromelain inhalation induced a small increase in pulmonary eosinophilia and IL-13 levels in the BAL of mice that had been reconstituted with ILC2s or CD4⁺ T cells, whereas a large increase was found in mice that had been reconstituted with ILC2s and CD4⁺ T cells. Remarkably, the latter increase was larger than the additive effect of ILC2- and CD4⁺ T cell-transfer alone, which suggested a synergistic effect [69]. Similar enhancing effects of ILC2s on CD4⁺ T cell proliferation and type 2 cytokine production were found by Mirchandani et al., who cotransferred ILC2s and DO11.10 CD4⁺ T cells in OVA plus IL-33-treated ST2-deficient mice [68]. However, for both studies, it cannot be excluded that interactions between ILC2s and CD4⁺ T cells in vivo were mediated via another cell type, for example DCs. A potential direct interaction between CD4⁺ T cells and ILC2 in vivo was assessed by Oliphant et al., who tested the requirement for MHCII expression by ILC2 for expulsion of intestinal N. brasiliensis in Il13 ^−/− mice [13]. Il13 ^−/− mice showed delayed helminth expulsion, which could be rescued by adoptive transfer of WT ILC2s, but not MhcII ^−/− ILC2s. However, no difference in CD4⁺ T cell numbers could be observed in mice that had received WT ILC2 versus MhcII ^−/− ILC2 [13]. This implied that ILC2-expressed MHCII was required for efficient helminth expulsion, yet not for proliferation of CD4⁺ T cells in vivo. Possibly ILC2s were dependent on MHCII expression to receive activating signals from CD4⁺ T cells (e.g., IL-2).

Although ILC2s contribute to the initiation of Th2 cell differentiation, they do not seem to be required for the activation of memory Th2 cells. ILC2 deficient mice were capable of eliciting a full-blown airway inflammation, when initial Th2 cell differentiation was induced by an i.p. injection with OVA with the adjuvant alum [63].

Recently, it was described that regulatory T cells might also regulate activation of ILC2s besides their recognized suppressive effect on effector Th2 cells. In a self-limited OVA-induced allergic airway inflammation, it was demonstrated that the de novo generation of regulatory T cells coincided with a decrease in IL-13 production by ILC2s in a TGF-β-dependent manner [71]. Interestingly, the persistence of airway inflammation, airway hyperreactivity and remodeling was demonstrated to be dependent on ILC2s rather than on antigen specific T cells in a chronic asthma model [72].

One of the initial reports that characterized ILC2s described interactions between ILC2s and B cells. Coculturing of ILC2s from the FALC with splenic B cells enhanced IgA production. Furthermore, ILC2s enhanced proliferation of peritoneal B1 cells but not B2 cells, which was dependent on ILC2-derived IL-5. In addition, B1 cells that were adoptively transferred into Rag2 ^−/− mice proliferated more than when they were transferred into Rag2 ^−/− Il2rg ^−/− mice, and cotransfer of B1 cells with ILC2 but not CD4⁺ T cells into Rag2 ^−/− Il2rg ^−/− mice induced B1 cell proliferation [32]. Therefore, the authors suggested that ILC2s were able to activate B1 cells in vivo. However, these results have not been replicated and might not be relevant for the role of ILC2s in allergic asthma.

Recently, a novel Thy-1⁺ Sca-1⁺ ILC2-like cell type was described in the murine spleen that did not express c-Kit, IL-7R, or ST2, but instead expressed IL-18R. These ILC2-like cells proliferated and produced IL-5 and IL-13 after in vitro stimulation with IL-2 plus IL-18, or after coculture with B cells. Interestingly, coculturing these ILC2-like cells with anti-CD40 plus IL-4-treated B cells strongly enhanced IgE production. This effect was abolished when the cell populations were separated in a transwell system, suggesting a contact-dependent interaction [73].

An interaction between ILC2s and B cells might be mediated via the costimulatory molecule ICOS, that is highly expressed by ILC2s and may interact with ICOS-ligand-expressing B cells [31]. Interestingly, ILC2s also express ICOS-L and provide self-stimulation, important for survival and cytokine production. The ICOS-ICOS-L interaction was demonstrated to be required for murine and human ILC2 mediated induction of airway inflammation and hyperreactivity [74]. However, these interactions as well as their potential consequences for B cell function and type 2 responses warrant further studies.