Atopic dermatitis (AD) is known as a chronic inflammatory skin disease. It is referred to as atopic eczema with typical symptoms such as itchy, red, swollen, and cracked skin lesions. Although AD is widespread and on the rise in developed countries, the exact pathological mechanism of AD is not fully understood yet (1–3). Currently, it is a typical type 2 helper T (T_H2) cell-mediated hypersensitive immune disorder in which various immune cells are known to participate in the development of skin inflammation (4). In AD, T_H2 cells secrete IL-4, IL-5, and IL-13 known to promote allergic responses (5, 6). These T_H2 cytokines can stimulate IgE production from B cells, activate mast cells, and lead to infiltration of eosinophils or other immune cells into inflamed tissues (7). Recent studies have also reported that type 2 innate lymphoid cells (ILC2) have the function of T_H2 cells in peripheral tissues. ILC2 is also well known to initiate and participate in T_H2 cell-mediated responses by secreting T_H2 signature cytokines such as IL-5 and IL-13 in peripheral tissues (8, 9).

Natural killer (NK) cells are well known as a type of anti-microbial lymphocytes in innate immunity (10, 11). NK cells can release proteolytic enzymes and interferon (IFN)-γ to remove virus-infected cells, intracellular bacteria, and tumor cells (12, 13). NK cells can be classified into several subsets, depending on the profile of cytokine secretion and their function (14). Typically, NK1 cells (also called conventional NK cells) secrete IFN-γ and NK2 cells secrete T_H2 cytokines (15–18). Subsets of IL-17 secretion (NK17) or IL-22 secretion (NK22) NK cells have also been reported (19–21). Compared to the past classification of the NK subset, various innate lymphoid cell types have been recently introduced, and in particular innate lymphoid cells (ILCs) have been proposed as representative helper innate immune cells (22). In addition, various T cell-lineage subsets have recently been defined in innate T cells such as NKT or γδ T cells that partially share receptors with conventional NK cells (23, 24). Therefore, it is necessary to more clearly distinguish the classification of NK. Accumulating evidences have demonstrated that some types of NK cells have a suppressive function like regulatory T (Treg) cells by secreting IL-10 or TGF-β in transplantation, pregnancy, and some infections (25–30). Although the role of NK cells in allergic diseases including AD is poorly understood, recent studies have reported that the number of circulating NK cells in blood samples of AD patients are generally decreased, but increased in inflammatory skins (31–34). However, it remains unclear which NK subset secretes suppressive cytokines.

In this study, we demonstrated that the population of TGF-β⁺ NK cells was decreased in human PBMC and lymphoid tissues from mice with AD than in healthy control. We further found that TGF-β⁺ NK cells were largely included in CD1d^hiPD-L1^hiCD27⁺ NK cell subset. Of interest, AD severity was relieved after an adoptive transfer of CD1d^hiPD-L1^hiCD27⁺ NK cell subset in mice by inhibiting T_H2 immunity.

NK cells are well recognized as innate immune cells that participate in various immune responses to viral infections or cancers by exhibiting anti-viral or anti-tumor effects (46–48). Although NK cells are lymphocytes, like other types of innate lymphoid cells (ILCs), they are generally involved in both innate immunity and adaptive immunity. NK cells are fundamental immune cells that play important roles in the initiation of immune responses and body homeostasis (49). They usually secrete signature cytokine such as IFN-γ. They are involved in the formation of T_H1 immunity in the body. They are known to play a crucial role in eliminating infected or tumor cells through secretion of intrinsic digestive enzymes (50, 51). Previous studies have shown that the frequency of NK cells in the body is generally decreased in various cancer diseases. This trend is considered to be due to the immune escape mechanism of cancer cells from host immune surveillance (52, 53). Recent studies have reported that NK cells are decreased in PBMCs from AD patients (34). This trend predicts that maintaining the balance of the number of NK cells and the immune circumstance will be important for the pathogenesis of allergy diseases such as AD. Our results also showed that the frequency of NK cells (CD56⁺CD3⁻) in the blood of AD patients was reduced ( Figures 1A, B ). Previous studies have found that NK cells are decreased in the blood of atopic dermatitis patients but increased in AD lesions, suggesting that NK cells can migrate to control peripheral T_H2 responses (34). However, the definite cause and mechanism of the alteration of NK cells in AD remain unclear. Therefore, we focused on another perspective to understand NK cells in atopic dermatitis. In addition to the well-known classical function, NK cells are known to have a subset of regulatory functions like other immune cells (14, 54).

In the past, researchers have paid attention to the etiological aspect caused by changes in the distribution and activity of a subset of specific immune cells due to breakdown of the balance of immune state in the body. In the 2000s, it was found that some sub-phenotypes of immune cells, including regulatory T cells (Tregs), were regulatory subsets that could restore the body’s immune status to normal through immunomodulation or induction of immune tolerance (55). Previous studies have shown that T cells are Tregs, B cells are regulatory B cells (Bregs), monocyte/neutrophils are myeloid-derived suppressor cells (MDSCs), macrophages are M2 or alternative macrophages, dendritic cells (DCs) are tolerogenic DCs, and ILCs are regulatory ILCs. Thus, a regulatory phenotype has been reported for each immune cell (56–61). These regulatory immune cells can differentiate to have their own characteristic functions by secreting anti-inflammatory cytokines such as IL-10 and TGF-β to control pro-inflammatory cells (62).

It is also known that NK cells have a regulatory type (called NKreg or NK3) (14). In addition to dividing effector functions into NK1 or NK2 cells according to inflammatory types, it has been proposed that secreting IL-10 or TGF-β can control excessive inflammatory responses (27, 28). It has been reported that NK cells can secrete anti-inflammatory cytokines such as IL-10 and TGF-β through secretion profiling, although there are only a small population of NK cells in the whole body. Especially, IL-10-producing NK cells are known to control the activation of T cells while Prf1^−/− mice-derived IL-10⁺ NK cells regulate CD8⁺ T cells and contribute to immunological maintenance in mouse cytomegalovirus (MCMV) infection (25, 26). In human studies, a regulatory NK subset that can secrete IL-10 and TGF-β has been reported (26, 27). It has been confirmed that IL-10 and TGF-β-producing NK cells exist in peripheral blood mononuclear cells and decidua in pregnant women. In particular, TGF-β-producing NK cells are significantly increased in decidua. It has been suggested that these regulatory NK cells contribute to pregnancy tolerance (26, 27). The present study showed that TGF-β-producing NK cells were present in human PBMCs. It also revealed that not only NK cells, but also TGF-β⁺ NK cells were decreased in allergic patients ( Figure 1C ). We employed MC903, a representative atopic dermatitis mouse model, to find changes in TGF-β⁺ NK cells in atopic dermatitis mice. Although there was a difference in the total number of lymphocytes, interestingly, TGF-β expression was higher in NK cells than in other lymphocytes of mouse spleen. Thus, we could predict that NK cells are an important source of TGF-β ( Figure 1D ). After MC903 treatment, TGF-β⁺ NK cells were significantly reduced in mouse spleen and cervical LN (cLN), a draining lymph node (LN), whereas these cells were increased in ear tissues, a peripheral target site of the disease ( Figures 1E–G ). As mentioned above, these patterns are similar to the migration pathway of classical IFN-γ⁺ NK cells in atopic dermatitis (34).

Previous studies have predicted that NK cells also have a regulatory type that can secrete anti-inflammatory cytokines such as IL-10 and TGF-β and that they are expected to be involved in immune tolerance or regulation (14). However, elucidation of the characteristic phenotype of NKreg is still insufficient. Through this study, we analyzed the expression of major receptors in mouse TGF-β⁺ NK cells ( Figure 2 ) and found high TGF-β expression (32.6 ± 2.0%) in CD1d^hiPD-L1^hiCD27⁺ NK subsets ( Figure 3 ). However, it is necessary to evaluate whether our proposed TGF-beta-producing NK subsets overlap with T cells, NKT cells, γδT cells or helper ILC1s. Therefore, as a result of the confirmation, it can be seen that the TGF-beta-producing NK subsets is CD3⁻CD49a⁻CD49b⁺Eomes⁺T-bet⁺ conventional NK cells ( Supplementary Figure 3 ).

We proposed that these highly TGF-β-expressing NK cells were splenic CD1d^hiPD-L1^hiCD27⁺ subsets. Several human NK cell studies have suggested an antigen-presenting role of NK cells. CD1d is a non-polymorphic, MHC class I-like molecule (63, 64). It is well known that CD1d usually presents antigens such as glycolipids, including α-galactosylceramide (α-GC), to CD1d-restricted NKT cells (65). CD1d is mostly expressed in innate immune cells such as DC, macrophages, B cells, and ILCs. In our study, it was confirmed that CD1d expression of TGF-β⁺ NK cells was high than TGF-β⁻ NK cells (66, 67). PD-L1 is well known to be expressed in NK cells as a representative immunosuppression marker (68). The murine CD27 expressing NK cell is exhibits potent cytokine production and high migratory capacity (39). Their correlation with the distribution pattern of TGF-β-producing NK subsets in AD was also determined. As a result, with AD development, CD1d^hiPD-L1^hiCD27⁺ NK subsets were decreased in lymphoid organs such as spleen and cLN, but markedly increased in the ear ( Figure 4 ). Based on these results, we can explain that the CD1d^hiPD-L1^hiCD27⁺NK1.1⁺CD3⁻ phenotype in mice is due to TGF-β-producing regulatory NK subsets. In addition, the distribution of CD1d^hiPD-L1^hiCD27⁺ NK subsets was decreased in spleen and draining LN from AD mouse model, suggesting that this distribution might be increased in skin lesions where type 2 inflammation appears.

This study not only suggested a phenotype of TGF-β⁺ NK cells, but also confirmed that the proposed TGF-β⁺CD1d^hiPD-L1^hiCD27⁺ NK subsets could regulate type 2 inflammation like AD. To demonstrate the immunomodulatory effect of TGF-β-producing CD1d^hiPD-L1^hiCD27⁺ NK subset in vivo, we separated CD1d^hiPD-L1^hiCD27⁺ NK subsets and CD1d^loPD-L1^loCD27⁻ NK subsets and adaptively transferred before MC903 induction of AD using a mouse model. Compared to vehicle mice (MC903+PBS i.v.) and CD1d^loPD-L1^loCD27⁻ NK subsets transferred mice, significant inhibitory effects were observed in CD1d^hiPD-L1^hiCD27⁺ NK subsets ( Figures 5A, B ).

Type 2 innate lymphoid cells (ILC2s) have recently been proposed as important effector cells in allergic responses. They are involved in peripheral allergic conditions through secretion of IL-4, IL-5, and IL-13 (42–44). We tested how CD1d^hiPD-L1^hiCD27⁺ NK subsets could affect the activity of ILC2 in the AD mouse model and found a decrease of systemic ILC2 ( Figures 5C–E ).

AD is well known as a typical chronic allergic disease. The classical immune system is initiated as allergen, resulting in a hypersensitivity reaction mediated by innate immune cells followed by a T_H2-cell mediated chronic inflammatory response (4). It is difficult to adequately overcome atopic dermatitis by controlling only the initial hypersensitivity reaction by innate immune cells. Thus, we tested how TGF-β-producing CD1d^hiPD-L1^hiCD27⁺ NK subsets could affect the activity of T_H2 effector cells in AD through in vivo adaptive transfer. Results confirmed that the addition of TGF-β-producing NK subsets induced a decrease in T_H2 cells without controlling the number of total T cells in lymphoid organs, thereby blocking the activity of T cells polarizing with T_H2 ( Figure 6 ).

It is well known that TGF-β can induce T cells into Foxp3⁺ regulatory T cells (iTregs) among several immunomodulatory functions (36, 69). In addition, Treg cells are the most representative immunomodulatory cells known to regulate various allergic disorders such as atopic dermatitis through several studies (70, 71). Therefore, we considered whether the control of T_H2-mediated inflammatory responses such as ILC2s and T_H2 cells by CD1d^hiPD-L1^hiCD27⁺ NK subsets could be correlated with Foxp3⁺ Treg cells. In our results, CD1d^hiPD-L1^hiCD27⁺ NK subsets did not directly control the number of Foxp3⁺ Treg cells in lymphoid tissues compared to its inhibitory effect on T_H2-mediated inflammatory cells. On the other hand, the number of Foxp3⁺ Treg cells was decreased in in the ear from the CD1d^hiPD-L1^hiCD27⁺ NK-treated group ( Figures 6C, D ). It was found that Foxp3⁺ Treg cells were maintained by inhibiting T_H2 cells activity without changing the distribution or the number of CD4⁺ T cells in the spleen. Therefore, administration of CD1d^hiPD-L1^hiCD27⁺ NK subsets in an AD-induced state can increase the ratio of Foxp3⁺ Treg cells compared to PBS control ( Figures 6E, F ). The decrease of Foxp3⁺ Treg cells in CD1d^hiPD-L1^hiCD27⁺ NK subsets administered ear appeared to result in reduced infiltration of CD4⁺ cells in ears. It can be seen that the disease improvement effect of CD1d^hiPD-L1^hiCD27⁺ NK subsets administration lies in the inhibition of infiltration of peripheral CD4⁺ T cells rather than the induction of an increase of peripheral Treg cells. According to this, CD1d^hiPD-L1^hiCD27⁺ NK subsets mainly controls T cell-mediated inflammatory responses in lymphoid tissues. Thus, treatment with CD1d^hiPD-L1^hiCD27⁺ NK subsets decreased T_H2 but increased the ratio of Treg cells in the spleen and draining lymph node.

In conclusion, we found that CD1d^hiPD-L1^hiCD27⁺ was a unique TGF-β-producing NK subset and that treating such regulatory subset in a mouse AD disease model inhibited T_H2-mediated effector cells and helped improve disease exacerbation. Although more diverse mechanism studies and mutual evaluation in human studies are needed, results of this study suggest that NK cell-derived regulatory subset can be used in various ways as a novel immune disease treatment strategy.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

The studies involving human subjects were reviewed and approved by the Institutional Review Board of Jeju Halla General Hospital (CHH-2016-L13-01). The patients/participants provided their written informed consent to participate in this study. The animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Konkuk University.

HK and WC designed the experiments, analyzed the data, and wrote the paper. KM and JK performed most of the experiments. GN analyzed the human sample data. DL and MJ collected and analyzed flow cytometry data. JL, MK, and SH performed in vivo experiments. All authors contributed to the article and approved the submitted version.

This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (NRF-2020R1C1C1003676 and NRF-2021R1A2B5B03002157).

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.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.