CD83+CCR7+ NK cells induced by interleukin 18 by dendritic cells promote experimental autoimmune uveitis

Abstract Natural killer (NK) cells have been reported to play a pathological role in autoimmune uveitis. However, the mechanisms regarding NK cells in uveitis and factors that affect NK‐cell activation in this condition remain unclear. Here, we report that the number of CD3‐NK1.1+CD83+CCR7+ cells is increased in the inflamed eyes within a mouse model of experimental autoimmune uveitis (EAU), and these cells express elevated levels of NKG2D, CD69 and IFN‐γ. Adoptively transferring CD83+CCR7+NK cells aggravates EAU symptoms and increases the number of CD4+IFN‐γ+T cells and dendritic cells (DCs) within the eye. These CD83+CCR7+NK cells then promote the maturation of DCs and IFN‐γ expression within T cells as demonstrated in vitro. Furthermore, IL‐18, as primarily secreted by DCs in the eyes, is detected to induce CD83+CCR7+NK cells. In EAU mice, anti‐IL‐18R antibody treatment also decreases retinal tissue damage, as well as the number of infiltrating CD83+CCR7+NK cells, T cells and DCs in the inflamed eyes and spleens of EAU mice. These results suggest that CD83+CCR7+NK cells, as induced by IL‐18 that primarily secreted by DCs, play a critical pathological role in EAU. Anti‐IL‐18R antibody might serve as a potential therapeutic agent for uveitis through its capacity to inhibit CD83+CCR7+NK cells infiltration.

killer (NK) cells infiltrating the eye may be a critical factor which drives this disorder of the immune system to result in tissue damage. [1][2][3][4][5][6] However, the underlying mechanisms of uveitis remain unclear.
Recently, NK cells have been reported to play a role in experimental autoimmune uveitis (EAU). 3,4,6-8 NK cells, which are a component of the innate immune system, represent the first line of defence against infections and play a regulatory role in autoimmune diseases by regulating the secretion of cytokines or by interacting with other cells. [9][10][11][12][13][14][15] As activated NK cells produce a significant increase in the expression of IFN-γ in Behcet's disease, a type of autoimmune uveitis, 4,6 any prevention in the secretion of IFN-γ from NK cells would have the potential to alleviate uveitis. An alteration in the status of NK cells may also contribute to the recovery of Behcet's disease through suppression of Th1 responses, 4,6 which represents a pathogenic factor for uveitis. 6,[16][17][18][19] Depleting of NK cells would alleviate the symptom of EAU. 8 Thus, the status of NK cells plays a critical role with regard to the development of autoimmune uveitis. Previous findings from our laboratory have revealed that EAU increased the number of CD3 -NK1.1 + NK cells during its inflammatory stages within the eyes and spleens. 3 Moreover, these cells expressed increased levels of CD83 and secreted elevated levels of IFN-γ. 3 However, the role of such CD83 + NK cells in EAU remains unknown and the specific factors affecting activation of NK cells in uveitis are unclear.
Interleukin-18 is a pro-inflammatory cytokine and is originally defined as an IFN-γ-inducing-factor. 20 It has been reported that IL-18 promotes differentiation of the NK cell specifically, CD56 + /CD83 + / CCR7 + cells, as well as the migration of NK cells. 21 However, whether IL-18 promotes NK cell activation in uveitis is unknown. IL-18 is constitutively produced by haematopoietic cells, such as DCs, macrophages and neutrophils, [22][23][24] as well as in non-hematopoietic cells, such as microglial and epithelial cells. 25 IL-18 involves binding with its receptor, consisting of IL-18R1 and IL-18R accessory protein in a heterodimeric receptor complex, to initiate signal transduction by the myeloid differentiation primary response protein 88 (MyD88). The nuclear factor (NF) kappa-light-chain-enhancer of activated B cells (κB) and mitogen-activated protein kinase pathways (MAPK pathways) are then activated, which promote IFN-γ transcription and stabilization of IFNG mRNA. [26][27][28] Recently, the role of IL-18 in uveitis has attracted considerable attention as it has been shown in both clinical and EAU models, that IL-18 shows enhanced expression during the inflammatory phase of the disease, [29][30][31][32][33][34] and polymorphism of the IL-18 gene is closely related to the occurrence of uveitis. 35,36 Thus, these reports indicate that IL-18 might participate in the inflammation of EAU. However, the mechanisms of IL-18 in uveitis remain unknown.
In this report, we investigated the status of NK cells and their role in uveitis, and tested the capacity for IL-18 to influence NK cell status in uveitis. And then, the possible cell subsets involved in secreting IL-18 in uveitis were measured. In addition, we examined whether blocking of IL-18 receptor would alleviate the development of uveitis and affect the status of CD83 + CCR7 + NK cells, DCs and T cells. The increases in IL-18 observed in EAU might represent an important factor which promotes CD83 + CCR7 + NK cell activation, thereby participating in the development of uveitis.

| Experimental autoimmune uveitis
Pathogen-free female C57BL/6 (B6) mice (6-8 weeks of age) were purchased from the Peking Vital River Laboratory Animal Ltd., Beijing, China and were maintained under pathogen-free conditions, and were performed in compliance with guidelines of the care and use of laboratory animals were approved by the ethics committee of Shandong were sorted from EAU mice with use of a flow sorting instrument (BD FACSAria TM III; BD Biosciences, CA, USA). The mice were administered CD83 + CCR7 + NK cells (1 × 10 6 ) at 4 days after immunization through an intravenous injection. Thereafter, at 4-day intervals, mice were killed and the eyes, lymph nodes and spleens were harvested for H&E staining and flow cytometric analysis.

| Anti-IL-18R antibody treatment
Anti-IL-18Rα antibody was purchased from R&D Systems Inc (Minneapolis, MN, USA). On day 8 after immunization, EAU mice were treated with anti-IL-18R antibody (10 µg/mouse, 100 µg/mouse, 450 µg/mouse) [41][42][43] through an intravenous injection every day. At 8 days after administration of anti-IL-18R antibody, EAU mice were killed for harvesting of the eyes lymph nodes and spleens. H&E staining was used to assess the severity of retinal tissue damage. IgG (10 µg/mouse, 100 µg/mouse, 450 µg/mouse) was administered daily through an intravenous injection and was used as the negative control.
For evaluating anti-IL-18R antibody effects in in vitro experiments, NK cells were isolated from the eyes of EAU mice (10 6 /mL) and were pretreated with anti-IL-18R (1 µg/mL) for 24 hours. These NK cells were then washed twice with PBS and added to the DCs, T cells or combination of DCs and T cells.

| Recombinant mouse interleukin-18 binding protein treatment
On day 8 after immunization, EAU mice were treated with IL-18 BP (Balb Co. Ltd., Beijing, China, 10 µg/mouse) 44 through an intravenous injection every day. At 8 days after administration of IL-18 BP, the eyes, lymph nodes and spleens of EAU mice were harvested. H&E staining was used to assess the severity of retinal tissue damage.
For evaluating IL-18 BP effects in in vitro experiments, IL-18 BP (1 µg/mL) was added to the co-culture of NK cells and DCs or to the co-cultures of NK cells, DCs and T cells.

| Co-culture of NK cells with DCs or DCs combined with T cells
Dendritic cells were isolated from spleens or ocular cells from EAU mice using a CD11c+isolation kit (Miltenyi Biotec, Germany). In brief, the spleen was minced and digested with collagenase D and DNase I at 37°C and filtered through a 70-µmol/L nylon mesh.  For anti-IL-27 neutralizing antibody treatment, anti-IL-27 neutralizing antibody (10 µg/mL) was added to the co-culture system of DC and NK for 24 hours.

| Flow cytometric analysis
The cells from the eyes, lymph nodes and spleens were collected from naïve and EAU mice as previously described. 3

| ELISA analysis
Aqueous humour and serum were obtained from the mice. Aqueous humour and serum concentrations of IL-18 and IFN-γ were analysed with use of an ELISA kit of mouse IL-18 or IFN-γ (R&D Systems Inc).
The assay was performed according to the instruction manual provided.

| Statistical analysis
Data analyses were conducted using GraphPad Prism 5 (GraphPad Software, San Diego, CA, USA). Each experiment was performed in FU ET AL.
| 1829 duplicate and replicated three times. Two-tailed Student's t tests or ANOVAs were applied to establish the presence of statistically significant differences between two groups or among the multiple sets of data respectively. For data failing to show homogeneity of variance, nonparametric Kruskal-Wallis test was used for multiple independent samples. Data were presented as mean ± SEM and P < 0.05 (*), 0.01 (**) and 0.001 (***) were required for results to be considered statistically significant.

| CD3 -NK1.1 + CD83 + CCR7 + NK cells are increased within the eyes in the EAU model
Induction of EAU was confirmed by the demonstration of its symptoms within the eyes, indicated by clinical and histopathological scores as previously described. 37,38 The disorganization of retinal tissue within the eyes of EAU mice and the dynamic clinical and histopathological scores of EAU mice are presented in Figure S1.
Infiltrating CD3 -NK1.1 + NK cells within the inflamed eyes and spleens were found to increase on days 8-12 (the initiation stage of EAU, Figure S1).
In particular, approximately 57.6% ± 5.7% of infiltrated NK cells in the inflamed eyes were CD3 -NK1.1 + CD83 + CCR7 + cells, and 61.2% ± 5.7% of NK cells in the lymph nodes of EAU were . All of them were higher than those in the eyes, lymph nodes or spleen of control mice ( Figure 1A,B). While activating signals, such as CD69 and NKG2D within CD3 -NK1.1 + CD83 + CCR7 + cells expressed higher than those signals in CD3 -NK1.1 + CD83 -CCR7cells, expression of the inhibitory receptor, NKG2A, was not significantly different between those two cells subsets ( Figure 1C). These increased amounts of CD3 -NK1.1 + CD83 + CCR7 + cells were shown to express high levels of IFN-γ, perforin and granzyme B as compared with that observed in CD3 -NK1.1 + CD83 -CCR7cells ( Figure 1D). These results indicated that most of infiltrated NK cells within the eyes were CD83 + CCR7 + NK cells which could express IFN-γ.

| Pathogenic role of CD83 + CCR7 + NK cells within the EAU model
To analyse the role of CD83 + CCR7 + NK cells in EAU, CD83 + CCR7 + NK or CD83 -CCR7 -NK cells were isolated from the inflamed spleen on days 12-16 after immunization. These cells were then adoptively transferred into EAU mice that had been immunized 4 days prior (Figure 2A). The severity of symptoms of these mice was observed every 4 days after CD83 + CCR7 + NK cells or CD83 -CCR7 -NK cell transfers. Retinal tissue damage and lymphocyte infiltration were present within the eyes of mice receiving this transfer of CD83 + CCR7 + NK cells on day 12 after immunization, and the retinal damage of these mice was more severe than that obtained in mice without cell transfer or with CD83 -CCR7 -NK cell transfer ( Figure 2B). Both clinical and histopathological scores of the eyes from mice receiving CD83 + CCR7 + NK cell transfer were higher than those of mice without cell transfer or those receiving CD83 -CCR7 -NK cell transfer ( Figure 2C). The amount of lymphocyte subsets including CD11c + MHC-II + DCs and CD3 -NK1.1 + cells within the eyes of mice receiving CD83 + CCR7 + NK cell transfer was greater than that in mice without a CD83 + CCR7 + NK cell transfer or those with a CD83 -CCR7 -NK cell transfer ( Figure 2D). Similar phenomenon was found in the lymph nodes of the mice with CD83 + CCR7 + NK cell transfer, but not in the spleen ( Figure 2D and Figure S2). These data indicated that CD83 + CCR7 + NK cells could affect the development of EAU and influence infiltrating T cells and DCs.

| CD83 + CCR7 + NK cells promote maturation of DCs
To further analyse whether CD83 + CCR7 + NK cells could influence the status of DCs and T cells in vitro, CD83 + CCR7 + NK cells from the eyes of EAU mice were isolated and co-cultured with isolated 33D1 + CD11c low MHC-II low DCs (immature DCs). DCs expressed high levels of CD80, CD86 and CD54 when co-cultured with CD83 + CCR7 + NK cells as compared with those lacking NK cells in the culture ( Figure 3A) as well as those co-cultured with CD83 -CCR7 -NK cells. Furthermore, to determine whether such mature DCs could affect the status of T cells, isolated CD4 + T cells were co-cultured with DCs that were pretreated with either CD83 + CCR7 + NK or CD83 -CCR7 -NK cells. These T cells, co-cultured with DCs-pretreated-CD83 + CCR7 + NK cells, expressed greater levels of IFN-γ than those co-cultured with non-treated DCs or DCs-pretreated-CD83 -CCR7 -NK cells ( Figure 3B). However, neither CD83 + CCR7 + NK cells nor CD83 -CCR7 -NK cells alone were capable of affecting IFN-γ secretion in T cells ( Figure 3B).
CD83 + CCR7 + NK cells might play a pathogenic role to promote maturation of DCs which then enable T cells to express IFN-γ.
Because CD83 + CCR7 + NK cells could secrete IFN-γ to influence the statues of DCs. 21 IFN-γR on DCs was blocked by anti-IFN-γR antibody, and then these DCs were co-cultured with CD83 + CCR7 + NK cells. Expression levels of CD80, CD86 and CD54 in above DCs were lower than those without anti-IFN-γR antibody treatment, while no significant differences were observed in isolated DCs or anti-IFN-γR antibody pretreated isolated DCs ( Figure S3A). Expression levels of IFN-γ in T cells co-cultured with CD83 + CCR7 + NK cells pretreated anti-IFN-γR antibody-blocked-DCs were also lower than that observed in CD83 + CCR7 + NK cells pretreated DCs ( Figure S3B).

| Increasing levels of IL-18 within the serum and aqueous humour of the EAU model may participate in CD83 + CCR7 + NK cell induction
As IL-18 has been reported to be an important factor involved in inducing subsets of CD83 + CCR7 + NK cells, 21  as compared with that of CD83 -CCR7 -NK cells ( Figure S7).

| DCs participated in the production of IL-18 in EAU
As IL-18 is reported to be produced primarily by macrophages, neutrophils and DCs, 19

| Anti-IL-18R antibody decreases the percent of CD83 + CCR7 + NK cells in vitro
To further evaluate whether the effect of DC in NK cells is through IL-18, we blocked the IL-18R in NK cells with anti-IL-18R antibody, and then co-cultured with matured DCs isolated from EAU mice.
The expressions of CD69, CD83, CCR7 and NKG2D within anti-IL-18R-treated-NK cells co-cultured with DCs were decreased as compared with non-treated or IgG-treated-NK cells ( Figure 5A). With anti-IL-18R antibody treatment, the percent of CD83 + CCR7 + NK cells ( Figure 5B) and expression ( Figure 5C) and secretion ( Figure 5D) of IFN-γ in NK cells were all decreased as compared with those nontreated or IgG-treated-NK cell. Blocking of IL-18R also reduced the effects of DC on CD83 + CCR7 + NK cells ( Figure S9A,B), but had no effect on CD83 -CCR7 -NK cells ( Figure S9C,D).

| Anti-IL-18R antibody treatment relieves EAU symptoms and decreases NK cell infiltration within the inflamed eyes
To assess whether anti-IL-18R antibody could affect the symptoms of EAU, anti-IL-18R antibody was administered through the tail vein of EAU mice that had been immunized for 8 days. At 8 days after anti-IL-18R antibody injection, ocular cells and splenic cells were harvested for analysis ( Figure 6A). While there was no obvious retinal tissue damage in response to the anti-IL-18R antibody treatment ( Figure 6B), the outer nuclear layer of retinal tissue in these mice with IgG treatment was deformed ( Figure 6B). Histopathological and clinical scores of the eyes in anti-IL-18R-treated-mice were decreased as compared with that of EAU and IgG-treated mice ( Figure 6C). Although the symptoms of EAU were alleviated with higher doses of anti-IL-18R antibody, no significant difference of the

| 1833
symptomatic remission was shown between higher and lower dose of anti-IL-18R treatment ( Figure S10).
Concentrations of IFN-γ and IL-18 within the aqueous humour and serum were decreased in anti-IL-18R-treated-mice as compared with that observed in untreated mice ( Figure 6E).
We also found that anti-IL-18R-pretreated isolated NK cells failed to increase the expressions of CD80, CD86 and CD54 on DCs, compared with those without anti-IL-18R treatment ( Figure S11A). In addition, DCs pretreated with anti-IL-18R treated NK cells failed to affect IFN-γ expression in T cells ( Figure S11B).
These data indicated that anti-IL-18R antibody could suppress the activation of NK cells and DCs along with IFN-γ expression in T cells.

| DISCUSSION
Uveitis is a serious inflammatory disease that can result in visual disability and blindness. 1 The primary basis for recurrent uveitis factors regulating DC-NK interaction in EAU may be present, which will be investigated in further studies within our laboratory. In addition, IFN-γ has been recognized to have pleiotropic effects in autoimmunity. [58][59][60] The function of IFN-γ in autoimmune disease may be related to its dosage and time. 61,62 The mechanism of IFN-γ in EAU needs further studying.  and Innovation Project of Shandong Academy of Medical Sciences.

CONFLI CT OF INTEREST
All the authors have declared there are no financial conflicts of interest with regard to this work.

AUTHORS' CONTRIBUTI ON
QF and WL designed the study, performed experiments, analysed the data, drafted and revised the manuscript; XM performed the animal experiments and flow cytometry; XW, NS performed the cell cultures, organized the samples and performed the experiments; YL F I G U R E 6 Anti-IL-18R antibody relieved symptoms of EAU and decreased inflammatory factors in EAU. (A) Diagram of anti-IL-18R antibody treatment in EAU. Anti-IL-18R antibody treatment was administered daily starting at 8 days after immunization, H&E staining and clinical test was performed every 4 days, and flow cytometry analysis on day 16. (B) Example of histopathology of a representative eye section from anti-IL-18R-or IgG-treated mouse (haematoxylin and eosin, original magnification, ×200). Vasculitis and photoreceptor folding within the retina were present within the inflamed eye (hollow arrows). (C) Histopathological and/or clinical scores of EAU, anti-IL-18R-treated and IgGtreated mice (three experiments with N = 3/group values represent the mean ± SEM Kruskal-Wallis test, ***P < 0.001). (D) Percent of CD4 + IFN-γ + T cells, CD4 + IL-17 + T cells, CD4 + GM-CSF + T cells, NK cells and DCs as determined within the eyes, lymph nodes (LN) or spleens of EAU, anti-IL-18R antibody-treated and IgG-treated mice. (E) Expressions of IFN-γ and IL-18 in the aqueous humour and serum of EAU, anti-IL-18R antibody-treated and IgG-treated mice (data of D, E were from three experiments with N = 3/group, values represent the mean ± SEM, ANOVA test: *P < 0.05, **P < 0.01, ***P < 0.001) and JX performed statistical analysis of the data; YS generated the figures and tables. All the authors approved the final version of this manuscript.