Activation of Invariant NKT Cells Ameliorates Experimental Ocular Autoimmunity by A Mechanism Involving Innate IFN-γ Production and Dampening of the Adaptive Th1 and Th17 Responses¹ Free

Invariant NKT cells (iNKT)⁷ are considered to represent an innate subset of T cells. iNKT cells have a semi-invariant Vα14-Jα18 TCR repertoire specific for lipid Ags that are presented on the MHC class I-like CD1d molecule. α-Galactosylceramide (α-GalCer), a synthetic glycolipid derived from marine sponges, is a well known iNKT TCR ligand (1). Recently, natural ligands of iNKT TCR have been described, such as α-glucuronosylceramide and glycosphingolipids from Sphingomonadaceae (2, 3, 4), that resemble cell wall constituents of some Gram-negative bacteria, supporting an innate role of iNKT cells in some infectious diseases. Other studies also revealed regulatory and protective properties in various models of autoimmunity such as experimental autoimmune encephalomyelitis (EAE) and type-1 diabetes (reviewed in Ref. 1). Most of these studies used synthetic ligands for iNKT cell activation and therapeutic regimen.

Upon ligation of their invariant TCR with α-GalCer, iNKT cells rapidly produce large amounts of cytokines such as IFN-γ and IL-4 (5). Analogues of α-GalCer prepared by total synthesis led to the identification of ligands that induce a cytokine pattern that is more biased toward either a Th1-type (IFN-γ) or a Th2-type (IL-4) response. An example of the former is α-C-GalCer, whereas the latter includes the ligand (2s,3s,4r)-1-O-(α-_D-galactopyranosyl)-N-tetracosanoyl-2-amino-1,3,4-nonanetriol) (OCH) (6, 7). Some studies demonstrated altered biological effects of these α-GalCer analogues on autoimmunity and cancer that could be ascribed to the cytokine profiles they elicit (6, 7).

Experimental autoimmune uveitis (EAU) induced in animals by immunization with retinal Ags in CFA is a model for human autoimmune uveitis, a disease that accounts for ∼10–15% of severe visual handicap in the US. EAU is induced by immunization with the same retinal Ags that are recognized by uveitis patients and is dependent on CD4⁺ Th1 and Th17 effector cells (8, 9). EAU in mice is induced with the interphotoreceptor retinoid-binding protein (IRBP) or with its pathogenic fragments emulsified in CFA (8).

The relationship between NKT cells and eye-related immune responses is not well understood. Although iNKT cells have been shown to play an important role in the eye-related regulatory phenomenon known as anterior chamber-associated immune deviation (ACAID) (10, 11), the possible role of iNKT cells in regulation of EAU has not been established. In the present study, we examine the role of iNKT cells and the effects of the iNKT cell ligands on EAU. Our data show that although natural strain-specific variations or genetically induced lack of iNKT cells do not seem to affect the threshold of susceptibility to EAU, iNKT function seemed to correlate with susceptibility. Importantly, a pharmacological enhancement of these cells using glycolipid iNKT cell ligands was able to inhibit induction of disease. This appeared to be due at least in part to iNKT-produced IFN-γ and a consequent dampening of the adaptive Th1 and Th17 pathogenic effector responses.

B10.RIII, B10.A, C57BL/6, DBA/2, AKR, and BALB/c mice (wild-type (WT) and CD1d-knockout (KO)) were purchased from The Jackson Laboratory. All experiments were approved by the National Eye Institute Animal Care and Use Committee. Animal care and use conformed to Institutional guidelines and to the Association for Research in Vision and Ophthalmology guidelines on the use of animals in ophthalmic and vision research.

Bovine IRBP was purified as described (12, 13). CFA was purchased from Difco and was supplemented with additional Mycobacterium tuberculosis H37RA to 2.5 mg/ml. Purified derivative of tuberculin (PPD) was purchased from the Statens Seruminstitut. α-GalCer (KRN7000) was provided by the Kirin Brewery (14). α-C-GalCer was synthesized as described previously (6). OCH was synthesized by Drs. Chi-Huey Wong and Douglas Wu of the Scripps Research Institute, La Jolla, CA. Neutralizing anti-IFN-γ Abs (clone R4-6A2) were obtained from the Biological Resources Branch, National Cancer Institute.

Livers were perfused in situ through the hepatic portal vein with PBS and minced into small pieces in PBS with 2% FCS and 0.02% sodium azide (PBS/FCS/Az). The tissue was then pressed though a 200-gauge mesh and cells were suspended in cold (4°C) PBS/FCS/Az. Cells were washed twice (500 g for 7 min). The pelleted cells were resuspended in 37.5% isotonic percoll at room temperature (25 ml per liver) and spun at 680g for 12 min. The cell pellet made up of lymphocytes and erythrocytes was collected, washed once in PBS/FCS/Az, and erythrocytes were lysed with 2 ml red cell lysis buffer (Sigma-Aldrich) for 4 min. Cells were then recovered by centrifugation through a layer of FBS, resuspended in PBS/FCS/Az, filtered through 100-μm mesh, counted, and prepared for immunophenotyping by flow cytometry.

C57BL/6 and BALB/c mice were immunized with 150 μg of bovine IRBP emulsified in CFA supplemented with Mycobacterium tuberculosis, strain H37RA from Difco to 2.5 mg/ml. Clinical disease was evaluated by fundus examination in a masked fashion and was scored on a scale from 0 (no inflammation) to 4 (complete destruction of the retina) in half-point increments, as described previously (8). Eyes were harvested for histopathology 21 days after immunization. Disease was scored by an ophthalmic pathologist (C.-C. Chan) in a masked fashion as described previously (8).

Unless otherwise noted, 5 μg of either α-GalCer, α-C-GalCer, or OCH were added to and emulsified with the uveitogenic Ag preparation (IRBP/CFA 1:1 v/v). The emulsion was injected s.c., divided into three doses (both thighs and base of tail). Systemic IFN-γ neutralization was achieved by treatment with monoclonal anti-IFN-γ Ab, 150 μg/mouse injected i.p. on days −2, 0, and 2).

Cytokine production to α-GalCer analogues in culture was examined on splenocytes obtained from naive mice. Cell suspensions of 2.5 × 10⁶ cells/ml were incubated with 100 ng/ml of the stimulant and supernatants were collected after 48 h. Cytokine production in vivo to α-GalCer analogues was measured in sera collected at the indicated time points after i.p. injection of 5 μg of the analog. For determination of Ag-specific cytokine production and proliferation, spleens and lymph nodes draining the site of immunization (inguinal and iliac) were collected on day 21 and were pooled within each group. Proliferation to the indicated doses of Ag was assayed by [³H]thymidine uptake during the last 16 h of a 72-h culture on triplicate cultures of 0.2 ml, as described (15). Cytokine responses were determined using the Pierce Chemical multiplex SearchLight Arrays technology (Ref. 16 and http://www.endogen.com/services).

iNKT cells were enumerated by flow cytometry after exclusion of dead cells by DNA staining with 7-amino-actinomycin D. Cells were reacted with PE-labeled CD1d/α-GalCer tetramers, allophycocyanin-labeled β-TCR, and FITC-labeled anti-CD4 Abs. β-TCR-positive cells were gated, and the percentage of α-GalCer-positive cells (either CD4⁺ or CD4⁻) was determined by counting of 1 × 10⁵ viable β-TCR-positive cells. Multiplication of percentages by absolute numbers of lymphocytes that were isolated from each organ (β counter) resulted in the total iNKT numbers shown in Fig. 1.

All experiments were performed at least twice and results were highly reproducible. Figures show data from representative or from pooled experiments, as specified. EAU severity is represented by fundoscopy scores determined on days 19–21. All fundoscopy scores were confirmed by histopathology. Where appropriate, statistical analysis of EAU severity was performed using the Snedecor and Cochran z test for linear trend in proportions (17). This is a nonparametric, frequency-based test that takes into account both disease severity and incidence. Probability values of <0.05 were considered to be significant. Values determined to be significantly different from controls are marked with an asterisk in the figures.

EAU is a disease model where susceptibility varies considerably among different mouse strains. The strain with the highest known susceptibility is B10.RIII. B10.A is another susceptible strain that can develop high disease scores, but, unlike the B10.RIII strain, it requires administration of pertussis toxin at the time of immunization to develop EAU, as do all other susceptible strains. C57BL/6 and DBA/2 mice have mild to moderate susceptibility, and AKR as well as BALB/c mice are resistant to disease. We asked the question whether susceptibility to disease in a series of EAU-characterized mouse strains correlated with their numbers of iNKT cells. Fig. 1 a shows a schematic representation of typical EAU scores for six mouse strains, summarizing previously reported findings (18, 19, 20). Representative pictures of severe, mild and no disease are shown in the inset.

Most of the iNKT cells in the body are concentrated in the thymus, liver and spleen. We enumerated invariant TCR-bearing NKT cells in these three organs in the different mouse strains using flow cytometry, by binding of PE-labeled CD1d/α-GalCer-tetramers. Although there were strain-specific differences in iNKT cell numbers and variations in their content in the different organs, there was no correlation with strain-specific differences in disease susceptibility (Fig. 1, b and c). In addition, a low iNKT number in one organ (e.g., thymus B10.A and spleen AKR) tended to be counterbalanced by a higher number in another organ (e.g., liver of B10.A and AKR) in some strains. Consequently, the total number of iNKT cells was often similar between strains with different EAU susceptibilities (Fig. 1 b).

We next examined the percent of CD4⁺ iNKT cells of the total iNKT cells, as this subset plays a role in the eye-specific regulatory phenomenon known as ACAID (11, 21). CD4⁺ iNKT cells were enumerated by double staining for CD4 and for the invariant TCR using CD1d/α-GalCer tetramers. The data revealed that the differences in CD4⁺ iNKT cells between the strains were even less pronounced than total iNKT numbers and did not correlate with disease susceptibility (Fig. 1 d).

Lastly, we selected three strains that were either resistant, moderately susceptible, or highly susceptible to EAU (BALB/c, C57BL/6, and B10RIII, respectively) and tested their iNKT cells in vitro using two different ligands, α-GalCer and OCH, each known to elicit differing cytokine profiles (4, 22). Single-cell suspensions were made from whole spleens and were stimulated with the indicated ligands for 48 h. Analysis of culture supernatants revealed that iNKT cells in resistant BALB/c mice produce significantly higher levels of IFN-γ, IL-4, and IL-2, irrespective of the ligand used for stimulation (Fig. 2). Interestingly, the opposite was the case with IL-17, with the highest levels detected in supernatant of B10RIII splenocytes. iNKT cells have recently been identified as a source of innate IL-17 (23), although the significance of this response for autoimmune disease is yet to be defined. No interstrain differences were found in other cytokines examined, such as IL-10, IL-13, and IL-5 (not shown).

If iNKT cells had a role in raising the threshold of susceptibility to EAU, we would expect that iNKT deficiency would result in more severe disease. However, mice deficient in CD1d (lacking CD1d-dependent NKT cells) (24) did not show enhanced EAU susceptibility compared with their WT counterparts (Fig. 3). The time of onset as well as the course of disease as determined by periodic fundus examinations were also not affected (data not shown). Mice deficient in CD1d on the resistant BALB/c background remained resistant (Fig. 3). These data are consistent with observations made by others in the EAE model (25, 26, 27).

We next examined whether functional triggering of iNKT cells using invariant TCR ligands can affect EAU. Five micrograms of α-GalCer incorporated into the IRBP/CFA emulsion ameliorated EAU severity and incidence in C57BL/6 mice (Fig. 4,a). To examine the effect of α-GalCer analogues, mice were similarly treated with α-C-GalCer and OCH. The data showed that OCH protected no better than α-GalCer, whereas α-C-GalCer was the most effective (Fig. 4 b).

This pattern of protection was unexpected because OCH deviates the iNKT response to TCR ligation toward IL-4 production (whereas α-C-GalCer skews toward IFN-γ) and has moreover been shown in experimental models of arthritis, diabetes in the NOD mouse, and encephalomyelitis (EAE) to be more protective than α-GalCer in its original form through an IL-4 dependent mechanism (reviewed in Refs. 7, 22). We therefore examined whether our α-GalCer, α-C-GalCer, and OCH preparations had the expected effect on iNKT cytokine production. Data obtained by measuring IL-4 and IFN-γ in serum of mice injected with the three α-GalCer analogues confirmed that indeed the three analogues elicited cytokine profiles that were in keeping with what has been reported by others (Fig. 5). These data suggested that an IFN-γ-dominated cytokine profile elicited by α-C-GalCer is more efficient in protecting from EAU than a deviation toward an IL-4 dominated profile in this model.

As we have previously observed that systemically produced IFN-γ can have a protective role in EAU (19, 28, 29), we decided to examine whether the enhanced protective effect of α-C-GalCer compared with the other analogues was due to its ability to induce enhanced production of IFN-γ by iNKT cells. We therefore injected mice immunized for EAU in the presence of α-C-GalCer with neutralizing Abs to IFN-γ at the time of immunization, when innate production of IFN-γ induced by α-C-GalCer would be occurring (according to the timeline established in Fig. 5). EAU and adaptive IFN-γ and IL-17, representing pathogenic effector Th1 and Th17 cell responses, were examined. Mice protected from disease with α-C-GalCer had strongly reduced adaptive IFN-γ and IL-17 responses (Fig. 6). Neutralization of IFN-γ at the time of immunization restored EAU scores in α-C-GalCer-treated mice, but had no effect on disease progression in the control group, demonstrating a specific requirement for IFN-γ in the protection. In addition, subsequent Ag-specific production of IFN-γ and IL-17 was restored (Fig. 6, b and c), supporting the notion that innate production of IFN-γ elicited by α-C-GalCer had a role in the protection and in inhibition of adaptive responses induced by this analog.

In the present study, we demonstrate that iNKT cells can have a role in EAU regulation. Their role appears to be not in setting the threshold of susceptibility to EAU, as do the natural CD4⁺CD25⁺ regulatory cells whose function in deterring development of ocular autoimmunity we have characterized in recent studies (30, 31). Rather, they can inhibit developing disease following a pharmacological enhancement of their activity at or around the time of priming. In chronic autoimmunity, priming of new effector T cells is believed to be occurring on a continuous basis. Since endogenous ligands for iNKT cells exist in the body and can trigger iNKT activity (4, 5), it is conceivable that iNKT cells can participate in modulating the course of ocular autoimmune disease. Thus, there appears to be a “division of labor” between the natural CD4⁺CD25⁺ regulatory T cells and iNKT cells, with the former setting the threshold of susceptibility and the latter possibly regulating the autoimmune response after that threshold has been passed.

The group of Stein-Streilein (10, 11, 21) demonstrated that iNKT cells have a central role in ACAID, a prototypic regulatory phenomenon elicited by injection of Ag into the anterior chamber of the eye and its transport by eye-derived APC to the spleen. iNKT cells are recruited into the spleen via a mechanism involving MIP-2 and participate in priming the adaptive T regulatory cells typically associated with ACAID. Although prior elicitation of ACAID to IRBP can inhibit a subsequent episode of EAU (32), it is unlikely that the protection from EAU by iNKT that we observe here bears a relationship to their role in ACAID. In ACAID, the eliciting Ag originates from the eye, which has to be perturbed (injected with Ag) in order for this phenomenon to be observed, and iNKT activation, if any, occurs without additional manipulation. In contrast, in our study, pharmacological activation of iNKT cells is needed and is applied when the eye is still intact. Thus, it is conceivable that iNKT cells may regulate ocular immune responses at more than one level.

Studies in the models of experimental arthritis, NOD diabetes, and EAE (reviewed in Ref. 22) had indicated that activation of iNKT by OCH is more effective than by α-GalCer, which was attributed to its induction of IL-4 and Th2 skewing. We were therefore surprised to find that OCH was not more effective than α-GalCer in protecting from EAU, and that the most efficient protection followed administration of α-C-GalCer, which induces an IFN-γ dominated iNKT cytokine response. Thus, effectiveness of protection paralleled the innate IFN-γ inducing ability of the invariant TCR ligand. The protection was accompanied by reduction in the IRBP-specific adaptive Th1 and Th17 pathogenic effector responses, as judged by production of their respective hallmark cytokines IFN-γ and IL-17 to in vitro recall with IRBP. The functional role of IFN-γ in the protective and regulatory effects of iNKT are strongly supported by direct evidence showing that neutralization of innate IFN-γ reversed the protective effect of α-C-GalCer and restored the subsequent proinflammatory cytokine production of the adaptive response. This is not to say that the mechanism of protection is the same for all the three analogues. Our data do not negate the possibility that protection from EAU by OCH and by α-GalCer could involve IL-4, as was previously demonstrated in several other autoimmune disease models (22).

Our data are in line with some previous reports, which revealed that protection from autoimmune disease by iNKT may not always involve IL-4 and Th2 skewing. Studies by Lehuen and her colleagues (33, 34) demonstrated that even in the absence of IL-4, iNKT cells can control EAE and experimental type 1 diabetes. This was associated with a decrease in Th1-associated pathogenic autoimmune responses without inducing Th2 responses and was due at least in part to induction of anergy in the autoreactive T cells (35). Such a mechanism could also be involved in the prevention of EAU observed here. Although these studies did not directly implicate IFN-γ in these effects, participation of IFN-γ (rather than IL-4) in protection from EAE was suggested by Furlan et al. (27).

It should be noted that high systemic levels of IFN-γ early or late in the disease can be protective, but likely by different mechanisms. Initial production of IFN-γ would be mostly from NKT and NK cells, whereas later in disease Ag-specific Th1 cells are a major source of IFN-γ. We previously showed that early up-regulation of IFN-γ by injections of IL-12 inhibits development of EAU and associated immunological responses by aborting priming, through a process that involves induction of NO and apoptosis (29). The innate IFN-γ produced by α-C-GalCer-triggered NKT cells may well work in a similar fashion. In contrast, protective effects of IFN-γ later in the disease appear to be due to its role in elimination of spent effector cells by activation-induced cell death (36, 37). Thus, neutralization of systemic IFN-γ at that stage also enhances disease (19), although at that point it is not possible to distinguish between effects of IFN-γ produced by Ag-specific T cells and iNKT cells.

In summary, we have demonstrated that iNKT cells can actively participate in regulating the autoimmune response to immunologically privileged retinal Ags. This apparently occurs at a different level than their role in induction of ACAID. The mechanism involves the induction of innate IFN-γ production through ligation of the invariant TCR and results in inhibited development of adaptive Th1 and Th17 responses that represent pathogenic effector mechanisms in uveitis.

We thank Dr. S. Yamano of the Kirin Brewery, Tokyo, Japan for providing α-GalCer (KRN7000) and Drs. Chi-Huey Wong and Douglas Wu of the Scripps Research Institute, La Jolla, CA for synthesizing the OCH used in this study.

The authors have no financial conflict of interest.