Group 2 Innate Lymphoid Cells Express Functional NKp30 Receptor Inducing Type 2 Cytokine Production Free

This article is featured in In This Issue, p.1

Innate lymphoid cells (ILC) are novel effectors of the immune system that contribute to diverse forms of inflammation and defense (1). They depend on the transcriptional repressor ID2 and cytokines that signal through the common γ-chain of IL-2. The lack of expression of common lineage markers and expression of IL-7Rα are shared features in this family. Based on their phenotype and function they have been divided into three main groups (2). Group 1 ILC include NK cells and group 1 ILC that produce IFN-γ and TNF-α (3). Group 2 ILC (ILC2) express GATA-3 (4) and retinoic acid–related orphan receptor (ROR) α transcription factors (5), and we and others have shown that human skin ILC2 produce type 2 cytokines IL-4, IL-5, and IL-13 in response to IL-33, IL-25, thymic stromal lymphopoietin (TSLP), and PGD₂ (6, 7). Group 3 ILC are RORγt-dependent and comprise lymphoid tissue inducers, as well as natural cytotoxicity receptor (NCR)⁺ and NCR⁻ group 3 ILC that express IL-22 alone or in combination with IL-17, respectively (8–10). Innate lymphoid cell family members do not possess a rearranged Ag-specific receptor and are thought to rely on a combination of activating and inhibitory signals for their effector functions, but the mechanisms are unclear (11). Several studies have evaluated the expression of the NCRs, NKp44 (NCR2 and CD336), NKp46 (NCR1 and CD335), and NKp30 (NCR3 and CD337) on group 1 and 3 ILC in humans, but the expression of these markers has not yet been demonstrated on ILC2 in humans (12). In this study, we sought to investigate the expression and function of NKp30 on human ILC2.

NKp30 is an activating type I Ig-like transmembrane receptor that is not present in mice, but is found mainly on human NK cells (13). It has one Ig-like extracellular domain that connects to the transmembrane region by a short 6-aa stem. It is encoded in the extremely polymorphic telomeric end of the class III region of the human MHC locus (3, 14). Alternative splicing of exon 4 gives rise to three main distinct isoforms, NKp30a, NKp30b, and NKp30c. Upon interaction with its ligands, isoforms a and b are thought to convey cytotoxicity responses, release of IFN-γ and TNF-α by NK cells, and trigger dendritic cell maturation (3, 15, 16), whereas splice variant NKp30c reduces IFN-γ production and cytotoxicity and increases the production of IL-10 by NK cells (3). A prevalent expression of isoform c on NK cells has been seen in patients with gastrointestinal stromal tumor and is associated with reduced survival due to defective IFN-γ, TNF-α, and IL-12 production, defective NK–dendritic cell dialog, and increased immunosuppression and IL-10 production (3). Several physiological, tumor, and viral markers have been identified as NKp30 ligands, including CMV tegument protein pp65 (17), Duffy binding–like-1α domain of Plasmodium falciparum erythrocyte membrane protein-1 (18), NF HLA-B–associated transcript 3 (BAT3) (19), and tumor-associated cell surface protein B7-H6 (20).

In this study, we investigate the expression of NKp30 on ILC2 ex vivo and on cultured ILC2. Using quantitative PCR we identify the splice variants of NKp30 and show that incubation of ILC2 with plate-bound B7-H6 or cell lines expressing this protein induced production of type 2 cytokines. This interaction can be inhibited by NKp30-blocking Abs and the soluble blocking ligand, galectin-3. We further established that activation of NKp30 induces the canonical pathway of NF-κB signaling. We identify a functionally important activatory cell contact receptor for ILC2, showing the involvement of NKp30 in ILC2-induced type 2 immune responses.

PBMCs were isolated from healthy adult donors under local ethics approval (National Research Ethics Service Committee South Central, Oxford C, 09/H0606/71). ILC2 were isolated and cultured as previously described (6). Briefly, the lineage (CD3, CD4, CD8, CD14, CD19, CD56, CD11c, CD11b, CD123, and FcεRI)–negative, CD45⁺, CD127⁺, CRTH2⁺ ILC2 population was sorted into 96-well plates at the density of 100 cells per well and resuspended in MLR of gamma-irradiated PBMCs from three healthy volunteers (2 × 10⁶ cells/ml) coupled with 100 IU/ml IL-2. After 4–6 wk, the growing wells were tested by flow cytometry staining and resorted until a pure population of lineage⁻CRTH2⁺IL-7Rα⁺ ILC2 was achieved (Supplemental Fig. 1A). The keratinocyte line (HaCaT) was cultured in tissue culture flasks (Corning, Corning, NY) in DMEM media supplemented with 10% FCS at 37°C with 5% CO₂ and split on reaching confluence (approximately every 3–4 d). K562, Jurkat, and THP-1 cell lines were cultured in RPMI 1640 supplemented with 10% FCS, amino acids (MEM nonessential amino acids solution 11140-050; Life Technologies) and HEPES (83264; Sigma-Aldrich). Cells were maintained at 0.2 × 10⁶/ml density. For HaCaT incubation with cytokines, IFN-γ was used at the concentration of 300 U/ml (21–24). All other cytokines were used at a concentration of 100 ng/ml (25).

For FACS surface staining the cells were labeled with the following anti-human Abs purchased from BioLegend unless stated otherwise: CD3 (SK7, BD Biosciences), CD19 (SJ25C1, BD Biosciences), CD123 (FAB301C, R&D Systems), CD11b (DCIS1/18), CD11c (BU15, Abcam), CD8 (RPA-T8), FcεRI (AER-37 [CRA-1]), CD14 (MφP9, BD Biosciences), CD4 (MEM-241), CD45 (H130), ICOS (C398.4A), CD56 (B159), CRTH2 (BM16, Miltenyi Biotec), IL-7Rα (A019D5), Live/Dead violet (L34955, Invitrogen), NKp30 (clone AF29-4D12), NKp30 blocking Ab (clone 210845, R&D Systems; AF29-4D12, Miltenyi Biotec), phospho-IκBα (Ser^32/36; Cell Signaling Technology, 9246), anti–B7-H6 Ab (ab121794), B7-H6 blocking Ab (17BL.3), CD68 (Y1/82A), Siglec-8 (7C9), and CD16 (3G8).

RNA extraction was performed using an RNeasy Plus mini kit (Qiagen, 74134) and TurboCapture 96 mRNA kit (Qiagen, 72251). cDNA was prepared using an Omniscript RT kit. The following gene expression assays were purchased from Applied Biosystems: GATA3 (Hs00231122_m1), IL-5 (Hs01548712_g1), IL-13 (Hs00174379_m1), GAPDH (Hs99999905_m1), IL-4 (Hs00174122_m1), RORα (HS00536545_m1), NKp30a (Hs01553310-g1), NKp30b (Hs01561746-g1), and NKp30c (Hs01553311-g).

Coat Corning Costar 9018 (Nunc MaxiSorp) were coated with indicated concentration of recombinant human (rh)B7-H6 Fc chimera protein (R&D Systems, 7144-B7-050) or control protein overnight at 4°C. ILC2 (5 × 10⁴) were cultured on B7-H6 or isotype control–coated plates. After 24 h the supernatants were collected for cytokine analysis using ELISA or cytokine bead array. Where indicated the cells were preincubated with 10 μg/ml Galectin-1 (CF 1152-GA-050/CF, Bio-Techne), Galectin-2 (1153-GA-050, Bio-Techne), and Galectin-3 (10289-HNAE-E-SIB, Stratech) for 1 h before culture with plate-bound rhB7-H6 or cytokine-treated HaCaT cells.

Human IL-13 ELISA Ready-SET-Go! (88-7439-86), human IL-13 ELISA DuoSet (DY213-05), and human IL-13 ELISpot^BASIC (3470-2A) kits were purchased from eBioscience, R&D Systems, and Mabtech, respectively, and carried out according to manufacturers’ instructions.

Anti–B7-H6 Ab (Abcam, ab121794), isotype control (Abcam, ab37416), mouse anti-rabbit HRP (344002, MaxDiscovery), and anti–Galectin-3 Ab (AF1154, Bio-Techne) were used to stain formalin-fixed, paraffin-embedded skin tissue sections from healthy donors and adult atopic dermatitis (AD) patients with moderate–severe disease. The diaminobenzidine signal was quantified using the “Fiji” version of ImageJ.

Skin biopsies from healthy donors were cut into wide strips and incubated overnight in 2 U/ml dispase at 4°C. Epidermal sheet was peeled from the dermis by forceps and incubated for 15 min in 0.5% trypsin plus 0.02% EDTA at 37°C. The mixture was then pipetted repeatedly and passed through a 40-μm strainer.

Suction blister cups were applied to the lesional and nonlesional skin of adult AD patients with moderate–severe disease, and not on systemic therapy, at a vacuum pressure of 250–450 mmHg. Blisters were formed within 30–90 min. Twenty-four hours later fluid was aspirated using a 30-gauge needle. Fluids were then centrifuged at 1500 rpm for 5 min at 4°C and the concentration of cytokines was measured by multiplex array (Luminex).

The t tests were performed using GraphPad Prism version 6.00 (GraphPad Software, San Diego, CA).

To investigate recognition strategies that ILC2 employ to sense the microenvironment, the expression of NKp30 on ILC2 was examined. Cultured ILC2 lines were stained for NKp30 (NCR3) and high levels of NKp30 expression were detected (94.6%, Fig. 1A). However, IL-2 is known to induce upregulation of NKp30 on NK cells, and its presence in the culture media could potentially induce upregulation of NKp30 by ILC2 (26, 27). Therefore to evaluate the expression of this receptor ex vivo, lineage⁻ (CD3, CD4, CD8, CD14, CD19, CD56, IL3R, FcεRI, CD11b, CD11c) cells that express CD45, CD127 (IL-7Rα), and CRTH2 from peripheral blood of healthy donors were analyzed and 52.4 ± 11.5% of human ILC2 express NKp30 on their surface ex vivo (n = 7, Fig. 1B, with gating in Supplemental Fig. 1A; the expression of NKp30 on NK cells is shown as a reference). Supplemental Fig. 1B shows the expression of NCRs on group 1, 2, and 3 ILC ex vivo.

Alternative splicing of exon 4 of the NCR3 gene gives rise to different isoforms. In NK cells, cross-linking of each isoform induces a distinct signaling pathway and a different pattern of cytokine production. NKp30a and NKp30b isoforms are classified as immunostimulatory whereas NKp30c is thought to be immunoregulatory (3). To determine the expression of NKp30 isoforms on ILC2, the mRNA levels of NKp30a, b, and c isoforms in cultured ILC2 were measured with quantitative PCR. The data were normalized to the ubiquitously expressed GAPDH housekeeping gene, and ILC2 were found to predominantly express NKp30c (Fig. 1C). Taken together, these data raised the possibility that NKp30 may play a role in ILC2 function.

B7-H6 is a self-ligand that is expressed on the surface of some tumor cell lines (14). It belongs to the B7 family of receptors that encode ligands for CD28 and CTLA-4 (B7-1 and B7-2), but unlike other members of the B7 family it is thought to be selective for NKp30 and does not bind CD28 or other NCRs. It is composed of 2 Ig extracellular domains (a distal V-like and a proximal C-like domain) with an adjacent phase 1 intron. Both the front and back β sheets of NKp30 bind to the V-like domain of B7-H6 via hydrophobic interactions (14, 28). To investigate whether NKp30 expressed on ILC2 mediates effector functions, ILC2 were treated with a rhB7-H6 Fc chimeric protein, a selective NKp30 ligand. Cross-linking NKp30 receptor on ILC2 by plate-bound B7-H6 protein induced production of increased amounts of IL-13 (Fig. 2A). Galectin-3 is a β-galactosidase–binding protein expressed on the surface of tumor cells and promotes proliferation and metastasis. Galectin-3 interacts with N-glycosylated sites (Asn⁴², Asn¹²¹) of NKp30 ectodomain and interferes with ligand binding (29). Incubation of ILC2 for 1 h with soluble NKp30-blocking ligand, Galectin-3, before performing the B7-H6 plate-bound assay reduced the expression of IL-13 (Fig. 2B) whereas it did not affect PGD₂ stimulation, which is mediated through CRTH2 (data not shown). The inhibitory effect of Galectin-3 was not observed with Galectin-1 and Galectin-2 (Fig. 2B). Next, we screened a panel of tumor cell lines for their ability to activate ILC2, including cells with differential expression of HLA class I molecules to investigate whether HLA may influence ILC2 function through interaction with nonpolymorphic HLA-binding receptors. The human B lymphoblastoid cell line .221 lacks expression of surface class I. Native .221 cells or .221 cells transfected with specific HLA variants did not induce production of IL-13 by ILC2 (Fig. 2C). In contrast, K562 cells (HLA^low myelogenous leukemia line), THP-1 cells (monocytic leukemia line), and Jurkat cells (T cell leukemia line) induced IL-13 expression by ILC2 (Fig. 2C). Further analysis showed that these cell lines all express B7-H6 on their cell surface, whereas the .221 cell line was negative. There was no expression of BAT3, another NKp30 ligand, on the surface of the tumor cell lines (Fig. 2D).

Similar to its effect on ILC2 interaction with plate-bound rhB7-H6, Galectin-3 had an inhibitory effect on NKp30-mediated interaction of ILC2 and K562 (Supplemental Fig 1C). Comparing K562-mediated activation of ILC2 with PGD₂ and IL-33 activation showed that K562 cells are as potent inducers of IL-13 expression by ILC2 as are PGD₂ and IL-33 (Fig. 2E). To investigate whether IL-33 or PGD₂ can modify NKp30, we stimulated ILC2 with increasing concentrations of IL-33 and PGD₂. No significant change was observed in the expression of NKp30 isoforms on ILC2 following stimulation with IL-33 and PGD₂ (Supplemental Fig. 2A). To examine the effect of NKp30-mediated activation of ILC2 on the expression of cytokine receptors and CRTH2, we incubated ILC2 with plate-bound B7-H6 and isolated mRNA at various time points. Three hours after incubation with B7-H6 the mRNA levels encoding the IL-33 receptor (ST-2) and PGD₂ receptor (CRTH2) were significantly reduced (Fig. 2F). IL-25 receptor (IL-17RB) expression was moderately reduced. Following initial downregulation, the ST2 mRNA level increased above the basal level. To test whether K562 also induced expression of other cytokines, we used multiplex cytokine analysis. Activation of ILC2 by K562 cells induced production of IL-2, IL-3, IL-4, IL-5, IL-13, IL-8, and GM-CSF (Fig. 3A) but did not increase production of IFN-γ and IL-10 (data not shown). Analysis of the level of these cytokines in the blister fluid of skin of patients with AD confirmed the biological significance of the levels of produced cytokines by activated ILC2 (Supplemental Fig. 2B).

To confirm that the ILC2 activation is mediated by NKp30, we incubated the cells with increasing concentrations of NKp30-specific Ab for 1 h before culture with the K562 cell line. Blocking NKp30 reduced the K562-induced cytokine production in a dose-dependent manner (Fig. 3A, Supplemental Fig. 2C). Incubating K562 cells with B7-H6–blocking Ab (30) for 1 h before culturing with ILC2 significantly inhibited production of IL-13 (Fig. 3B).

NKp30-mediated activation of ILC2 is contact-dependent, as supernatant from cultured tumor cell lines could not activate ILC2 (Fig. 3C) and indeed 24-plex and 7-plex analysis of K562 supernatant did not reveal any known ILC2-stimulating cytokines or chemokines (ILC2-stimulating cytokines are shown in Supplemental Fig. 2D). In fact, ILC2 produced a similar pattern of cytokines either by K562-mediated cross-linking of NKp30 receptor or plate-bound rhB7-H6 activation, but rhB7-H6 provided a stronger signal (Supplemental Fig. 2E).

To confirm the NKp30-mediated rapid effector functions of ILC2 ex vivo, we incubated freshly isolated ILC2 from the blood of healthy donors with the K562 tumor cell line and showed that K562 alone does not produce IL-13, but it can indeed activate ILC2 and induce their production of IL-13 (Fig. 3D).

As discussed above, NK cells express different isoforms of NKp30, whereas ILC2 mainly express the NKp30c isoform. To compare the cytokine production profile of activated NK cells and ILC2, we incubated both cell populations with rhB7-H6. Upon interaction with B7-H6, NK cells produced significant amounts of IFN-γ, TNF-α, and IL-10 whereas ILC2 did not. ILC2 mainly produced type 2 cytokines, IL-13 and IL-5. Although NK cells also expressed these cytokines, the ability of ILC2 in producing type 2 cytokines was significantly higher (Fig. 3E).

Therefore, NKp30 expressed on ILC2 is functionally active and mediates interaction between ILC2 and immortal laboratory tumor cell lines, suggesting ILC2 may have a role in the interaction with malignant or stressed host cells.

Several studies have reported the expression of B7-H6 on tumor cell lines. To investigate whether B7-H6 is also expressed in nonmalignant tissue, sections of healthy skin and lesional skin biopsies of patients with AD were evaluated. Interestingly, we found expression of B7-H6 protein in healthy skin tissue, but this was confined to the basal layer of epidermis. In contrast, lesional skin biopsies from adult patients with AD showed B7-H6 expression throughout the suprabasal layers of the epidermis and myeloid-derived populations in the dermis (Fig. 4A, Supplemental Fig. 3A). Objective quantification of immunohistochemistry staining of B7-H6 expression in skin biopsies of healthy controls and AD patients is shown in Fig. 4B.

It is therefore plausible to speculate that widespread expression of B7-H6 in the skin of patients with AD can lead to NKp30-mediated activation of ILC2. This finding is compatible with our earlier observation that ILC2 resident in the skin of patients with AD show an activated phenotype (6). The presence of ILC2 in the epidermis (Ref. 31) and Supplemental Fig. 3B) favors their migration into or retention in the epidermis and supports their possible interaction with keratinocytes.

To further understand whether the widespread expression of B7-H6 in patients’ skin contributes to the NKp30-mediated activation of ILC2, the effect of the immortalized human keratinocyte cell line (HaCaT) on ILC2 activation was tested. A relatively low expression of B7-H6 (Fig. 4C) was observed in HaCaT cells. However B7-H6 is known to be upregulated under inflammatory conditions (32). Therefore, we incubated the HaCaT cell line with IFN-γ, TNF-α, epithelial cytokines, mast cell products, and type 2 cytokines.

Consistent with earlier reports, the proinflammatory cytokines IFN-γ and TNF-α induced significant upregulation of B7-H6. Interestingly, type 2 cytokines IL-13 and IL-4 also significantly upregulated B7-H6, whereas epithelial cytokines (IL-25, IL-33, and TSLP) and the mast cell derivative PGD₂ did not significantly upregulate cell surface expression of B7-H6 on HaCaT cells (Fig. 4C). We next treated HaCaT cells with increasing concentrations of IFN-γ or IL-13 and then washed, trypsinized, counted, and subsequently cultured the cells with ILC2 for 24 h. The ability of ILC2 to produce IL-13 was correlated with the dose of IFN-γ or IL-13 used to treat HaCaT cells (Fig. 4D). Interestingly, preincubation with soluble Galectin-3 for 1 h before culturing with IL-13– and IFN-γ–treated HaCaT cells reduced the expression of IL-13 (Fig. 4E). Galectin-3 is expressed by myeloid-derived cells (Ref. 33 and Supplemental Fig. 3A) at similar levels in healthy skin and lesional biopsies of AD patients as shown by objective analysis of Galectin-3 staining of five donors (Supplemental Fig 3C).

All NKp30 isoform monomers have arginine-charged residues in their transmembrane regions that couple with aspartate amino acids in CD3ζ homodimers or CD3ζ/FcRγ heterodimers by a salt bridge (34). CD3ζ is an ITAM-bearing adaptor molecule essential for transmitting downstream signaling (14). Isoform b is constantly associated with CD3ζ whereas isoform c shows a weaker interaction. NKp30 cross-linking triggers the canonical pathway of NF-κB activation. NF-κB dimers are bound to the inhibitory protein IκB. The IκB kinase complex (IKK) consists of IKKα, IKKβ, and regulatory protein NEMO and phosphorylates IκB. Phosphorylated IκB undergoes proteasomal degradation that releases NF-κB to translocate to the nucleus (35). To study the signaling pathway involved in NKp30-mediated ILC2 activation, intracellular FACS analysis of these cells confirmed the increase in phosphorylation of IκB (Fig. 5A). To evaluate whether the increase in production of cytokines was due to the release of preformed cytokines or to an increase in gene expression, we incubated ILC2 with plate-bound B7-H6 protein for up to 24 h, which increased production of IL-13 protein (Fig. 5B) and increased mRNA expression of IL-13, IL-4, IL-5, and AREG and enhanced expression of GATA-3 (Fig. 5C) but did not affect RORα expression (data not shown).

In 1999, Moretta and colleagues (36) identified natural cytotoxicity receptors with activating properties that can trigger an immune response on recognition of cognate cellular and viral ligands and therefore play an important role in NK cell antitumor and antiviral cytotoxicity. Although NCRs have been detected on type 1 and type 3 innate lymphoid cells, their specific functions have not yet been fully elucidated (37, 38). In the present study, we identified and characterized NKp30 expression on group 2 innate lymphoid cells, showing an important role in induction of type 2 cytokines.

NKp30 protein is encoded by the NCR3 gene located in the highly polymorphic telomeric end of the class III region (39). NK cell subpopulations can express all three of the most common isoforms (a, b, and c), and the relative contribution of each receptor depends on the expression levels on the cell surface. Interestingly, ILC2 showed predominant expression of the putative immunomodulatory splice variant, NKp30c, and lower expression of NKp30a and b isoforms. We show that NKp30 engagement on ILC2 augments the production of type 2 cytokines IL-4, IL-5, IL-13, and GM-CSF as well as other inflammatory cytokines such as IL-2, IL-3, and IL-8. These cytokines have a crucial role in allergic and inflammatory conditions, and genes encoding many of these cytokines are located at the same chromosome locus 5q31–33 (40). Proliferation and differentiation of CD34⁺ progenitor cells into basophils and mast cells are highly dependent on IL-3. IL-3 also regulates dendritic cell differentiation from monocytes (41, 42). IL-8 is a potent neutrophil chemoattractant (43, 44) and contributes to allergy and severe asthma (45, 46). GM-CSF has been increasingly appreciated as a type 2 cytokine (47, 48) and a neutrophil and eosinophil survival factor (49, 50).

Another difficulty that limits our understanding of the role of NKp30 is poor characterization of its ligands and their distribution on normal tissues under homeostatic and inflammatory conditions. Two self-ligands have been identified to bind to NKp30: a novel member of B7 family receptors, B7-H6, and a largely intracellular protein, BAT3, which can be expressed on the cell surface under certain conditions (14, 20, 28, 51). Consistent with activation of NKp30-expressing cells through B7-H6, our data demonstrated that plate-bound rhB7-H6 can induce production of IL-13 by ILC2 in a dose-dependent manner. Interestingly, incubation of multiple tumor cell lines with ILC2 showed that cell lines that activated ILC2 and induced production of IL-13, such as THP1, K562, and Jurkat, lacked cell surface staining of BAT3 but maintained B7-H6 expression, suggesting that B7-H6 is the main NKp30 ligand being recognized on the laboratory tumor cell lines. Indeed, IL-13 production was diminished upon prior incubation of ILC2 with increasing concentrations of NKp30-blocking Ab or the soluble NKp30-blocking ligand, Galectin-3. We showed that the IL-13 production occurs rapidly, consistent with an early innate immune function to respond during the initial phases of an immune response. Although stimulation with IL-33 and PGD₂ did not alter the expression of NKp30 isoforms, NKp30 mediated activation of ILC2 downregulated expression of cytokine receptors IL-33, IL-17RB, and PGD₂, suggesting a negative feedback mechanism making ILC2 less responsive to further activation at early stages. The expression levels of cytokine receptors were increased after 8 h, and in the case of ST2 this increase was above basal levels. Stimulation with epidermal cytokines may follow the initial NKp30-mediated activation of ILC2.

Until recently, B7-H6 was thought to be absent on normal tissues and restricted to tumor and transformed cells (51). However, in 2013 Matta et al. (30) showed that B7-H6 could be induced on nontransformed cells in various conditions of cell stress, such as infections and inflammation. B7-H6 was selectively upregulated on CD14⁺CD16⁺ proinflammatory monocytes and neutrophils when stimulated with TNF-α and IL-1β in vitro and in septic conditions in vivo (30). These data have guided us to propose some molecular bases for the direct interaction and cross-talk between NKp30⁺ ILC2 and ligand-expressing tissues. Staining of formalin-fixed, paraffin-embedded sections of normal human skin showed low levels of B7-H6 expression in the basal layer of epidermis whereas B7-H6–expressing cells were detected throughout the epidermis in lesional tissue sections of patients with AD. Dermal cell populations expressing B7-H6 were also observed that mainly consisted of CD11b⁺ and CD11c⁻CD68⁺ cells (Supplemental Fig. 3A).

The unstimulated keratinocyte cell line, HaCaT, showed low levels of B7-H6 expression, but incubation with proinflammatory cytokines TNF-α and IFN-γ and type 2 cytokines IL-13 and IL-4 increased expression of B7-H6. It has been well established that type 2 cytokines are found in high concentrations in lesions of AD patients (52–55), but type 1 cytokines IFN-γ and TNF-α have also been found, particularly during the chronic stage of AD (56).

The expression of B7-H6 by keratinocytes provides a novel perspective on the link between inflammation and NKp30-mediated activation of ILC2. Keratinocytes in AD lesions may not only activate ILC2 by producing epithelial cytokines, but also direct interaction via B7-H6 can trigger production of type 2 cytokines. Therapeutic strategies to modulate ILC2 activity may provide novel treatment approaches for individuals with AD.

We are grateful to Simon Kollnberger and Paul Bowness for provision of .221 cells. We also express our gratitude to Prof. Eric Vivier for providing B7-H6–blocking Ab.

The authors have no financial conflicts of interest.