NK cells are considered the innate counterpart of memory CD8⁺ T cells. They share similar functions such as cytotoxicity and interferon-γ (IFN-γ) production and both express the transcription factors Eomes and T-bet. ILC1 on the other hand closely resemble T_H1 cells. Both express and depend on T-bet but lack EOMES and produce IFN-γ (16–19). NK cells and ILC1 are involved in protecting the organism against pathogens, viruses and tumors (16, 20, 21). Intraepithelial ILC1 can be found in Crohn's disease patients and contribute as a proinflammatory IFN-γ-producing population in an anti-CD40-induced colitis model in mice (22). NK cells are suggested to be important in enhancing inflammatory responses in a hapten based contact hypersensitivity mouse model and human allergic contact dermatitis (23, 24). Taken together these cell types are mainly involved in mounting a type 1 immune response.

ILC2, like T_H2 cells, highly express the transcription factor GATA3 and produce type 2 cytokines including interleukin-5 (IL-5), IL-13 and the epidermal-growth-factor-like molecule amphiregulin (7). ILC2 mediate pathology in a mouse model of atopic dermatitis and promote wound healing in an IL-33-dependent manner (25, 26). ILC2 promote type 2 driven immune responses by promoting T_H2 differentiation of naïve CD4⁺ T cells through production of IL-13, and by expression of MHC class II on their cell surface induce T cell priming (27–29). In addition, the inducible T-cell costimulatory (ICOS) molecule is highly expressed on ILC2 regulating their activation status and proliferation (30, 31). Moreover, activated ILC2 can express the TNF receptor superfamily ligand OX40L, which promotes local T_H2 cell proliferation and adaptive type 2 inflammation (32). Increased ILC2 numbers are linked to human allergic airway and skin diseases like allergic asthma atopic dermatitis (25, 33–36). Thus, type 2 immune responses are profoundly shaped by ILC2.

ILC3 share RORγt expression with T_H17 cells and can produce IL-17 and IL-22 thereby helping the organism to fight against bacteria and fungi and viruses, such as Citrobacter rodentium, Salmonella enterica, Candida albicans, and rotavirus (2, 7, 37–41). There are ILC3 expressing the chemokine receptor CCR6 which comprise lymphoid-tissue-inducer (LTi) cells and can be CD4⁺ or CD4⁻. These cells are crucially important in the embryonic development of many lymphoid organs, whereas in adult mice they reside mainly in cryptopatches of the intestine with low proliferation (42–45). In mice, CCR6⁻ ILC3 can express natural killer cell receptor such as NKp46 (NCR⁺ ILC3), loose RORγt expression and upregulate T-bet, finally leading to IFN-γ production (46–50). These “ex-RORγt⁺ ILC3” closely resemble ILC1. A large population of ILC3 can be found in the intestine where they are essential for maintaining barrier integrity and immunologic tolerance to commensal bacteria of the gut (51–53). IL-17 producing ILC3 are proposed to be involved in plaque formation in a psoriasis mouse model based on the topical application of the Toll-like receptor 7 (TLR7) agonist imiquimod (54). Finally, elevated numbers of ILC3 are found in blood and affected skin of psoriasis patients (55–57). Given this data ILC3 are part of type 3 immune responses and intestinal immunopathology.

Significantly more ILC2 can be found in lesional skin biopsies from patients suffering from atopic dermatitis in relation to skin from healthy individuals (25, 36). These ILC2 produce high amounts of the type 2 cytokines IL-5 and IL-13 and express the membrane bound IL-33 receptor ST2 as well-receptors for IL-25 and thymic stromal lymphopoietin (TSLP) (25, 36). These changes are even more profound when ILC2 are isolated from skin of house dust mite (HDM) allergic individuals that have been challenged epicutaneously with HDM extract. IL-33 is able to strongly enhance the expression of IL-13 and IL-5 and to increase the migratory capacity of isolated skin-derived ILC2 in vitro (36). Interestingly, ILC2 from atopic patients also express higher amounts of the killer cell lectin-like receptor G1 (KLRG1), which is even further elevated after stimulation with IL-33 or TSLP (36).

Human ILC2 express the prostaglandin D₂ (PGD₂) receptor chemoattractant receptor-homologous molecule expressed on T_H2 cells (CRTH2) (64, 65). PGD₂ which is mainly produced by mast cells induces ILC2 migration, production of type 2 cytokines and upregulation of the expression of IL-33 and IL-25 receptor subunits (ST2 and IL-17RA) in vitro (66). The effects of PGD₂ on ILC2 can be mimicked by the supernatant from activated human mast cells (through IgE-mediated degranulation) and inhibited by a CRTH2 antagonist highlighting a cross-talk between mast cells and ILC2 (66).

ILC2 respond to further mast cell mediators like cysteinyl leukotrienes, particularly LTE₄ (67). Human ILC express the functional leukotriene receptors CysLT₁ and its expression is increased in patients with atopic dermatitis (67). LTE₄ not only induces migration, promotes cytokine productions and upregulation of IL-33/IL-25 receptors in human ILC2 in vitro, but also enhances the pro-inflammatory effect of the epithelial cytokines IL-25, IL-33, TSLP, and of PGD2 as seen by increased production of IL-5 and IL-13. This effect of LTE₄ can be partially inhibited by adding the leukotriene antagonist montelukast. Finally, addition of IL-2 to LTE₄ and epithelial cytokines significantly further amplifies the activation of ILC2 (67). These findings clearly suggest a pathogenic role of ILC2 in the pathogenesis if atopic dermatitis in humans (Figure 2).

Topical application of a synthetic form of active vitamin D3 (MC903) to the skin of mice can mimic atopic dermatitis-like inflammation with a type 2 signature (68). Using the MC903 AD mouse model Salimi et al. and Kim et al. investigated inflammatory responses in the presence and absence of ILC2. When ILC2 are depleted in Rag1^−/− mice by administering an anti-CD90.2 and/or anti-CD25 monoclonal antibody this leads to an dramatically decreased ear swelling response (25, 36). Furthermore, using Rorα^sg/sg (Rorα-knockout) bone marrow chimeric mice which lack ILC2, a markedly reduced inflammatory response in the skin can be seen, highlighting ILC2 as a main proinflammatory cell in this type 2 inflammatory model (36). An increase in IL-33 and IL-25 expression has been reported in lesional skin of patients with AD compared with healthy individuals underlining an important role for these cytokines as proinflammatory ILC2 activating cytokines in AD (36, 69, 70). Strikingly, when flow cytometry assisted cell sorting (FACS)-purified ILC2 from MC903-treated C57BL/6 wild-type mice are adoptively transferred by intradermal injection into naïve C57BL/6 wild-type recipient mice, the recipient mice develop AD-like skin reactions with a type 2 T cell response indicating that these innate cells alone are capable of eliciting an AD-like skin response (25).

Another possible mouse model to study eczema like skin reactions are the “flaky tail” mice. These mice bear a frameshift mutation in the murine filaggrin gene (flg) resulting in expression of a truncated profilaggrin (~215 kDa) instead of the normal high-molecular-weight profilaggrin (>500 kDa) (71). Topical application of allergen to mice homozygous for this mutation results in cutaneous inflammatory infiltrates and enhanced cutaneous allergen priming with increased development of allergen-specific antibody responses (71). Saunders et al. characterized changes of ILC2 numbers and their cytokine production in flg-mutant mice (72). These mice show spontaneous atopic dermatitis-like inflammation and develop compromised pulmonary function. In the skin and skin draining lymph nodes of these mice, there is a significant increase in the frequency of IL-5-producing ILC compared to wild type animals. However, no differences in cell numbers are seen for ILC1 and 3. Furthermore, flg-mutant mice show higher skin infiltrates of eosinophils, mast cells and basophils (72). Even more astonishing, when flg-mutant mice are crossed with Rag1^−/− mice (Flg^ft/ftRag1^−/−) skin lesions but not lung inflammation occur as shown by cutaneous expansion of IL-5-producing ILC2, indicating that skin inflammation can develop independently of the adaptive immune system in these mice (72). Regulation of ILC responses by adaptive immune cells is also reported in other tissues (73). Finally, increased frequency of ILC2 can be found in skin blisters taken from non-lesional skin of patients with filagrin mutations compared with the skin of filagrin wildtype subjects (72). Taken together, loss of filagrin function in humans and mice is clearly linked to increased ILC2 activation and disease progression in atopic dermatitis.

This latter model, however, has been challenged recently by the work of Schwartz et al. which provides evidence that atopic dermatitis like lesions can evolve independent of ILC2 and ILC2-derived cytokines in Filaggrin-mutant (Flg^ft/ft) mice bred on an ILC2-deficient background (74). Interestingly, inflammation in these mice following MC903 treatment requires IL-1β and IL-1R1-signaling but is independent of NOD-, LRR- and pyrin domain-containing protein 3 (NLRP3) inflammasome activation and results in elevated numbers of IL-1β-responsive connective tissue mast cells (74). Finally, Flg^ft/ft mice do not develop skin inflammation under germ-free compared to SPF conditions indicting a crucial role for the microbiome in promoting proinflammatory immune responses in this mouse model (74). This issue will be discussed in more detail in a later section.

Development of ILC2 depends on the transcription factor receptor-related orphan receptors alpha (RORα) and lack of RORα results in impaired lung inflammation in response to protease allergen in mice despite normal T_H2 cell responses (75). Dai et al. provide evidence that a synthetic RORα/γ inverse agonist (SR1001) is able to suppress inflammation in the MC903-induced atopic dermatitis mouse models. Topical treatment with SR1001 reduces epidermal and dermal inflammation, suppresses the production of type 2 cytokines and TSLP, and reverses impaired keratinocyte differentiation (76). Since SR1001 also inhibits RORγ signaling it is quite possible that RORγt⁺ ILC3 functions may also be impaired (42). If topical inverse agonists for RORα may have anti-inflammatory functions in humans remains to be elucidated.

A crucial role for the IL-33/ILC2 axis in the pathogenesis of AD has been proposed by Imai et al. The authors generated a transgenic mouse line which overexpresses IL-33 in keratinocytes. These mice spontaneously develop an itchy dermatitis closely resembling AD at age 6–8 weeks with thickened epidermis, skin infiltration of eosinophils and mast cells, and high histamine and IgE levels in the blood (77). Moreover, IL-5 and IL-13 expressing ILC2 numbers are significantly increased in lesional skin, peripheral blood, and regional lymph nodes. Administering a neutralizing monoclonal anti-IL-5 antibody results in a marked reduction of the inflammatory response as shown by a decreased peripheral blood eosinophil count, milder thickened epidermis and lower inflammatory infiltrates including eosinophils (77). Unfortunately, a randomized, placebo-controlled parallel group design study in patients with AD could not detect a clinical improvement by administering a monoclonal antibody to human interleukin-5 (mepolizumab) in two single doses of 750 mg, given 1 week apart, despite a significant decrease in peripheral blood eosinophils (78).

Group 1 ILCs consisting of NK cells and ILC1 are involved in inflammatory bowel and allergic skin diseases in mice (12, 24, 87, 88). Regarding ACD Carbone et al. were able to characterize CD56^highCD16⁻CD62L⁻ NK cells in an ex vivo human model which accumulate in affected skin of hapten allergic human individuals and these NK cells release type 1 cytokines and induce keratinocyte apoptosis in vitro (23). In mice NK cells can be further subdivided into two distinct subsets: CD49a⁺DX5⁻ liver-resident (Trail⁺) and CD49a⁻DX5⁺ conventional NK cells (cNK) (12). Furthermore, cNK cells seem to express much higher amounts of the transcription factor EOMES (87). Liver-resident NK cells can mediate long-lived, antigen-specific adaptive recall responses to haptens like DNFB and oxazolone independent of B cells and T cells (24). Preceding was the finding that a CHS response to several haptens can be elicited in Rag2^−/− mice lacking T- and B-cells but not in mice that either contain dysfunctional NK cells (SCID × beige mice) or completely lack NK cells (Rag2^−/− Il2rg^−/− mice). A proper CHS response can be transferred by FACS-purified antigen-specific Thy-1⁺ Ly49C-I⁺ liver-resident NK cells from sensitized Rag2^−/− mice when transferred into naive Rag2^−/− Il2rg^−/− recipients (24). The same NK cell type seems to mount antigen specific immunity against certain viral pathogens as well (88). Our own investigations using the hapten TNCB support the role of EOMES⁺ cNK cells as the dominant proinflammatory innate cell type in the early phase of contact hypersensitivity. NK cell numbers increase significantly 24 h in the ear skin of mice after allergen challenge and produce type 1 marker cytokines like IFN-γ and TNF (89). Taken together, NK cells seem to represent a major driving force of the innate immune system in CHS pathogenesis pathogenesis (Figures 3A–D).

Very little is known about the involvement of helper-like ILC in the pathogenesis of CHS, however there has been some indirect evidence for it in the past. ILC2 are known to be a major source of IL-13 production thus playing a crucial role in innate type 2 immune responses to worms and inhaled allergens (90, 91). IL-13-deficient mice (Il13^−/−) show impaired T_H2 responses induced by epicutaneous ovalbumin (OVA) exposure whereas i.p. sensitization is normal and results in responses equivalent to wild type mice (92). Interestingly, Il13^−/− mice display an even enhanced ear swelling responses to the hapten DNFB, which is also known to elicit a type 1 T-cell driven immune response (93), compared to wild type mice. At the time, this finding was interpreted as a lack of T_H2-mediated suppression but it's tempting to speculate that impaired ILC2 function in this mouse model may also have contributed to a disinhibited and thus exaggerated type 1 immune response. We recently characterized cell numbers and cytokine production of all ILC subgroups (ILC1-3 and NK cells) during the elicitation phase of a CHS mouse model based on the hapten TNCB using an ILC reporter system (89). Numbers of ILC are elevated in skin draining lymph nodes, show an activated phenotype and produce elevated amounts of their marker cytokines IL-13 and IL-5 at late time points (48 and 72 h), i.e., during the resolution phase of the inflammatory response in the skin. On the other hand, NK cell numbers and their production of IFN-γ and TNF are highest 24 h after allergen challenge paralleling the strongest skin inflammation period (89). The latter is expected since TNCB is known to elicit a type 1 driven immune response (93, 94). However, lack of ILC achieved by either antibody mediated depletion using an anti CD90.2 mAb in Rag1^−/− mice or by using mice that selectively lack ILC2 [Rorα^sg/floxIl7r^Cre/+mice (29)] results in a significantly enhanced and long lasting inflammatory response (89). The ear infiltrate of ILC depleted mice show a tendency toward a more type 1 biased immune response indicated by increased numbers of T-bet⁺ CD4⁺ T-cells (89). This data supports the concept of a counter regulatory role for ILC2 in CHS (Figures 3A–D).

Some allergens like Fluorescein isothiocyanate (FITC) and papain rather induce allergic type 2 immune responses with increased IL-4 producing T_H2 cell infiltrates in murine skin when reapplied topically or intradermally (28, 95, 96), suggesting that ILC2 might rather play a proinflammatory role in these models. Along this line we demonstrated in a papain skin challenge model that lack of IL-13-producing ILC2 leads to a marked reduction of inflammation with less skin infiltrating T_H2 cells [(28); (Figures 3F–J)]. A first therapeutic approach in type 2 dominated allergic skin responses has been proposed by Bao et al. They demonstrate that ILC2 numbers are increased in the skin of FITC-challenged mice. In addition, intraperitoneal injection of the cycloartane triterpene saponin Astragaloside IV during the sensitization phase leads to a reduction of the inflammatory response as seen by a decreased ear swelling response, less production of pro-allergic cytokines like IL-33 and TSLP, and significantly reduced numbers of ILC2 in the skin of these mice (97). Thus, ILC2 seem to have contrary roles in type 1 and type 2 dominated allergic skin reactions, respectively (Figure 3).

ILC2 and ILC3 express MHCII molecules on their surface and can act as antigen presenting cells for helper T cells (29, 51, 52). Our own analysis of MHCII expression on ILC2 revealed that in skin draining lymph nodes of mice ~50% of the ILC2 express MHCII, while in the skin only ~3% express MHCII. Antibody mediated depletion of ILC leads to a significant reduction of MHCII positive ILC2 both in skin and LN (89). Currently, we can only speculate that ILC2 might regulate effector T cells in a direct fashion via MHCII. In line with this, Oliphant et al. recently demonstrated that MHCII expression on ILC2 and subsequent antigen presentation to CD4⁺ T cells is crucial for successful helminth expulsion in mice (29). The crosstalk between ILC2 and CD4⁺ T cells seems to involve IL-2 since activated CD4⁺ T cell-derived IL-2 has been shown to synergize with IL-33 to stimulate ILC2 (29). Thus, lack of ILC2 may lead to a higher availability of IL-2 for proliferation of other effector cells like NK cells leading to an augmented response in CHS.

Another way how ILC2 might crosslink innate and adaptive immunity is by expressing the lipid-presenting molecule CD1a. Other than classical MHC proteins that present peptides, CD1 molecules present endogenous and exogenous lipid antigens to T lymphocytes (98). In a CHS model using the poison ivy-derived lipid contact allergen urushiol, CD1a expressing Langerhans cells are important to promote CD1a-restricted CD4⁺ T cells to produce IL-17 and IL-22. Furthermore, treatment with blocking antibodies against CD1a alleviates skin inflammation dramatically (99). More recently Hardman et al. demonstrated in a human skin challenge model that skin-derived ILC2 not only express CD1a but are also capable of helping CD1a-reactive T cells to sense S. aureus components in an cytosolic phospholipase A2 (PLA2G4A) and TLR-dependent–dependent manner, suggesting a new role for ILC2 in lipid surveillance of the skin (100). Currently, it is unclear whether this also applies for the adaptive immune response against urushiol. Taken together CD1a expression on ILC2 seems to be clearly involved in shaping the phenotype of adaptive T cell responses.

Mashiko et al. reported significantly elevated frequencies of basophils, ILC and T_H2 cells in the lesional skin of AD patients compared to patients suffering from psoriasis. Interestingly, basophils and ILC2 are positively correlated in skin, whereas skin basophils are inversely correlated with blood ILC2 suggesting that skin basophils may attract circulating ILC2 to skin of AD patients by IL-4 production (101). Kim et al. detected elevated numbers of basophils and ILC that form clusters in inflamed human AD skin compared to control skin. Using the MC903-based AD mouse model in IL-4/GFP reporter mice, they demonstrated that murine basophil responses preceded ILC2 responses and those basophils are the dominant IL-4-producing cell type in inflamed skin. In addition, ILC2 express the IL-4 receptor IL-4Rα and proliferate in an IL-4-dependent manner. Finally using Il4^−/− mice Kim et al. provide evidence that especially basophil-derived IL-4 is necessary for proinflammatory ILC2 responses in the skin (102).

Most notably, Egawa et al. have shown that basophil-derived IL-4 converts Ly6C⁺CCR2⁺ inflammatory monocytes into anti-inflammatory M2 macrophages in an IgE-mediated chronic allergic inflammation (IgE-CAI) mouse model, a model where basophils rather than mast cells and T cells play a critical role for the elicitation of allergic response (103, 104). In this model, skin infiltrating monocytes acquire an M2-like phenotype in an IL-4R- and basophil-dependent manner and adoptive transfer of Ly6C⁺CCR2⁺ inflammatory monocytes dampens the exacerbated IgE-CAI in CCR2^−/− mice which also requires IL-4R signaling (103). Thus, it is tempting to speculate, that basophil-derived IL-4 may promote pro-inflammatory responses via ILC2 and anti-inflammatory signals via M2 macrophages at the same time, leading to a counterbalanced immune response. However, the role of ILC2 in the IgE-CAI model is not known so far.

On the other side, ILC2 have been shown to promote polarization of the anti-inflammatory M2 macrophages by producing type-2 cytokines (IL-4, IL-5, and IL-13) in an renal ischemia-reperfusion injury model and experimental cerebral malaria (105, 106). Furthermore, in obese mice PD-1^high ILC2 are inhibited by PD-L1 expressing M1 macrophages which is promoted by TNF. PD-1 blockade improves ILC2 function, reinforces type 2 innate responses and promotes adipose tissue homeostasis (107, 108). Interestingly, in an serum-induced arthritis mouse model ILC2 were indispensable for dampening proinflammatory IL-1β secretion by bone marrow-derived macrophages (109). Finally, basophil-derived IL-4 seems to be essential for M2 macrophage mediated trapping of Nippostrongylus Brasiliensis larvae in the skin during second infection of mice thereby leading to reduced worm burden in the lung (110). However, basophils had no apparent contribution to worm expulsion from the intestine highlighting their crucial role in the skin (110).

Taken together, there seems to be an intense crosstalk between basophils, ILC2 and macrophages involving cytokines like IL-4, IL-13, and IL1β and resulting in differential polarization of macrophages dependent on the disease model. How these three cell type interact in AD and CHS remains to be elucidated.

Using the protease-allergen papain which induces type 2 allergic airway and skin inflammation we showed that ILC2 are necessary for mounting an appropriate antigen specific T_H2 memory response and that ILC2 activation clearly precedes T_H2 involvement in papain induced airway and skin inflammation (28). Furthermore, ILC2-derived IL-13 is needed for the activation and expansion of an allergen-induced subset of dendritic cells (CD11b⁺CD103⁻IRF4⁺) which produce the T_H2 cell chemoattractant CCL17. Using ILC2-deficient mice, we demonstrated that dermal ILC2 are crucial to mediate expansion of CCL17⁺ dendritic cells after skin challenge with papain finally leading to an effective T_H2 memory response. Thus, ILC2 licensing of dendritic cells is a critical component of the memory T_H2 cell response to certain allergens at barrier sites (28).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.