These authors contributed equally to this work.
Cellular immune response
Critical role for IFN-γ in natural killer cell-mediated protection from diabetes
Version of Record online: 12 DEC 2007
Copyright © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
European Journal of Immunology
Volume 38, Issue 1, pages 82–89, January 2008
How to Cite
Lee, I.-F., Qin, H., Priatel, John J. and Tan, R. (2008), Critical role for IFN-γ in natural killer cell-mediated protection from diabetes. Eur. J. Immunol., 38: 82–89. doi: 10.1002/eji.200737189
- Issue online: 21 DEC 2007
- Version of Record online: 12 DEC 2007
- Manuscript Accepted: 2 NOV 2007
- Manuscript Revised: 11 SEP 2007
- Manuscript Received: 17 FEB 2007
- Canadian Diabetes Association
- NK cells;
Autoimmune diabetes in nonobese diabetic (NOD) mice can be prevented by a single injection of complete Freund's adjuvant (CFA), but the mechanisms mediating protection remains unclear. We previously showed that NOD mice immunized with CFA have a markedly reduced incidence of diabetes that is associated with a significant decrease in the number of β-cell-specific, autoreactive cytotoxic T lymphocytes and, furthermore, that the effect of CFA is mediated by natural killer (NK) cells. In this study, we report one mechanism by which NK cells regulate the onset of diabetes. Administration of CFA produced a rapid increase in NK cell frequency and function, including cytotoxicity and IFN-γ secretion. By co-transferring NK cells from IFN-γ-deficient (or wild-type) NOD mice and spleen cells from diabetic NOD mice to NOD/SCID recipients, we show that IFN-γ secretion by NK cells significantly influences the effect of CFA protection. In contrast, NK cytotoxicity does not appear to participate in CFA-mediated protection from diabetes. Our findings demonstrate that NK cells mediate the protective effects of CFA through secretion of IFN-γ.
type 1 diabetes
In nonobese diabetic (NOD) mice, an animal model for type 1 diabetes (T1D), a single injection of complete Freund's adjuvant (CFA) between the ages of 4 and 10 weeks prevents the development of hyperglycemia 1–4. We have previously shown that CFA injection of NOD mice effectively prevents the accumulation of β-cell-specific cytotoxic T lymphocytes (CTL) that recognize an immunodominant β-cell epitope derived from islet-specific glucose 6 phosphatase-related protein (IGRP), and that protection from disease was mediated by a subset of natural killer (NK) cells 5. However, it remains unclear how NK cells exposed to CFA or the effects of CFA might serve to regulate autoreactive CTL.
NK cells are important for both anti-viral immunity and lymphocyte regulation, serving these functions through their capacity for direct cellular cytotoxicity and release of immunomodulatory cytokines such as interferon (IFN)-γ 6, 7. Abnormalities in activity of NK cells have been associated with autoimmune diseases besides T1D, including systemic lupus erythematosus and multiple sclerosis 8–10. Compared to other mouse strains such as BALB/c or C57BL/6, NK cells from NOD mice secrete less IFN-γ following in vitro stimulation and have significantly decreased cytotoxicity against NK-specific cell lines 11–13. In addition, it has been reported that IFN-γ can have a protective effect on diabetes in NOD mice and that CFA protection from disease is dependent upon the presence of IFN-γ 14. Given these observations, along with the important role that NK cells have in the release of IFN-γ and in mediating CFA protection, we addressed the hypotheses that CFA increases the frequency and function of NK cells and that IFN-γ secretion or cytotoxicity by NK cells is critical for protection from disease.
CFA rapidly increases the frequency of NK cells in peripheral blood and spleen
Numerical and functional abnormalities in NK cells have been reported in NOD mice and been implicated in the etiology of T1D 11–13. Furthermore, it has been shown that injection of CFA limits the development of diabetes in NOD mice 1–4 by preventing the accumulation of β-cell-specific CTL in an NK cell-dependent manner 5. In this study, we further investigated the mechanisms by which NK cells mediate the effects of CFA. First, we compared NK cell numbers in female NOD mice treated with either CFA or PBS. NOD mice received CFA or PBS for 16 h and spleen cells were harvested for staining. Immunostaining revealed an increased fraction of splenic NK cells among total splenocytes shortly after CFA injection (Fig. 1A and B). Moreover, this rise in NK cell proportion is attributable to increases in their absolute numbers in spleen after CFA injection (mean ± S.D., 0.648 ± 0.014 × 106 in PBS-injected group vs. 1.94 ± 0.058 × 106 in the CFA-injection group). Temporal measurement of NK cell frequencies in peripheral blood showed a similar increase in the CD3-DX5+ NK population peaking at 1 h post injection, and returning to pretreatment levels after 24 h (Fig. 1C); these kinetics were more prolonged in the spleen (Fig. 1D). These data indicate that NK cell expansion, accumulation or mobilization occurs rapidly following CFA treatment.
To determine whether NK cell proliferation is responsible for elevations in splenic cell numbers, NOD mice were injected with bromo-2′-deoxyuridine (BrdU) intraperitoneally to measure NK proliferation following treatment with PBS or CFA. Peripheral blood and splenocytes were harvested to analyze BrdU incorporation using anti-BrdU antibody. Our results showed that beginning 30 min after CFA treatment, BrdU incorporation in NK cells is detectable in blood, and a significant increase in BrdU was observed 2 h after CFA treatment, suggesting NK cells rapidly undergo proliferation after CFA injection (Fig. 1E). However, we did not observe significant BrdU incorporation in splenic NK cells in this experiment (Fig. 1F), suggesting that the increased NK population in spleen at 16 h after CFA injection is due to extrasplenic proliferation and mobilization to the spleen. Together, our data indicated that both recruitment and proliferation play a role in observed NK cell number increases following CFA injection.
CFA increases cytotoxicity and IFN-γ secretion by NK cells
To determine whether NK cell function is enhanced by CFA, we evaluated their cytotoxicity and IFN-γ production following CFA treatment. Splenic NK cells were isolated from NOD mice 16 h after injection of PBS or CFA and cultured in IL-2. The NK cells were subsequently used to determine cytotoxicity and IFN-γ secretion upon recognition and stimulation by YAC-1 target cells. Consistent with prior reports, only low-level cytotoxicity was observed against YAC-1 cells using PBS-treated NK effectors derived from NOD mice (Fig. 2A). However, NK cells from CFA-treated NOD mice had significantly increased lytic activity against target cells (Fig. 2A). In addition to cytotoxicity, the ability to secrete IFN-γ was assessed by culturing IL-2-expanded NK cells from CFA- or PBS-treated mice for 4 h at 37°C in the presence of the Golgi transport inhibitor GolgiStop. Intracellular staining of cells from CFA-treated mice demonstrated an augmented capacity to produce IFN-γ (Fig. 2B). These data indicate that the increased NK functions may play a role for the protective function of CFA in diabetes. It has been reported that the NK abnormality in NOD mice is a deficient activity of the receptor NKG2D 15. In these animals, upon IL-2 activation, NK cells from NOD mice but not from C57BL/6 mice expressed NKG2D ligands, which resulted in down-regulation of the receptor NKG2D and reduced NK function. To examine whether CFA-increased IFN-γ secretion in NK cells is through the restoration of NKG2D expression, IFN-γ secretion was measured after treatment. Following CFA treatment (but not PBS), staining for intracellular IFN-γ NK cells demonstrated a substantial increase in the frequency of IFN-γ-producing cells (Fig. 2C and D). Moreover, we also observed a modest increased NKG2D expression by NK cells from CFA-treated mice [mean fluorescent intensity (MFI) of NKG2D expression from two groups are (mean ± SEM) 136.0 ± 4.435 in PBS-injected group and 193.8 ± 3.326 in CFA-injected group] (Fig. 2E). To investigate whether CFA treatment also affected the expression of NKG2D ligands on NK cells, we analyzed surface levels of RAE-1γ by flow cytometry (Fig. 3A and B) and RNA expression of RAE-1α, RAE-1β and RAE-1γ (Fig. 3C). Immunostaining showed that the percentage of NK cells expressing RAE-1γ is significantly decreased in CFA-treated mice (Fig. 3A and B). Quantitative real-time PCR results revealed reduced levels of RAE-1α, RAE-1β and RAE-1γ messages in NK cells from CFA-treated mice (Fig. 3C). These results suggest that CFA may restore NKG2D-mediated function of NK cells through down-regulation of NKG2D ligand.
IFN-γ-deficient NK cells fail to rescue NOD mice from diabetes
Other investigators have shown that IFN-γ has a protective role in diabetes and is also necessary for the protective effects of CFA 14, 16. To test the hypothesis that IFN-γ from NK cells is a critical mediator of protection, we performed adoptive transfer experiments using NK cells from NOD and IFN-γ-deficient (IFN-γKO) NOD mice. Before adoptive transfer, we tested the ability of cells from these two mouse strains to produce IFN-γ (Fig. 4A and B). We compared IFN-γ secretion of both IL-2-expanded NK cells from NOD and NOD.IFN-γKO mice after 4-h incubation in vitro (Fig. 4A) and NK cells directly ex vivo (Fig. 4B). As shown in Fig. 4A, intracellular IFN-γ in CD3–/DX5+ cell subsets from IFN-γKO mice was negligible compared to NOD mice after PBS treatment. We also determined the effect of CFA on both mouse strains. As expected, IFN-γ was not present in CD3+/DX5+ or CD3–/DX5+ cells from IFN-γKO mice following CFA administration (Fig. 4A, B and data not shown); however, consistent with previous data, expression of IFN-γ was increased in NOD mice following after CFA treatment (Fig. 4A and B). To determine the contribution of IFN-γ to NK-mediated protection from disease, we adoptively transferred diabetogenic spleen cells (derived from diabetic NOD female mice) to NOD/SCID recipients with or without NK cells obtained from either NOD or NOD.IFN-γKO mice. NK cells in NOD/SCID recipients were depleted prior to adoptive transfer to eliminate the confounder of endogenous NK cells. As shown in Fig. 4C, NK-depleted recipients injected with CFA and adoptively transferred with diabetogenic cells all developed diabetes within 5 weeks. Restoration of NK cells from NOD mice delayed onset of disease in CFA-treated animals confirming our previous finding that the protective effect of CFA can be mediated by NK cells. However, when NK cells derived from NOD.IFN-γKO mice were co-adoptively transferred, a rapid onset of disease occurred similar to NK-depleted recipients. Together, these data indicate that NK cells require IFN-γ to mediate the protective effects of CFA.
NK cytotoxicity is not required for the protective effects of CFA
The experiments in Fig. 4 do not exclude a protective role for NK cytotoxicity in addition to IFN-γ secretion, as the possibility remains that NK cytotoxicity is impaired in NOD.IFN-γKO mice. To assess a possible role for NK cytotoxicity, we assayed the lytic capacity of purified NK cells derived from CFA injected NOD or NOD.IFN-γKO mice by chromium release assay. The results indicated that, while NK cells from IFN-γKO mice are unable to secrete IFN-γ with or without CFA treatment, their cytotoxic capacity was unimpaired and remained similar to that of wild-type NOD NK cells (data not shown). Because adoptive transfer IFN-γKO NK cells did not restore protection, these data suggest that CFA-induced NK cytotoxicity does not participate in the protective mechanism.
We have further investigated the role of NK cells in mediating CFA-induced protection from diabetes in NOD mice. Our results show that CFA treatment increases the frequency of NK cells and, perhaps more importantly, the function of NK cells in pre-diabetic NOD mice. Furthermore, our data show that the secretion of IFN-γ by NK cells is necessary for protection. Numerical and functional defects in NK cells have been reported in NOD mice 11–13, but few studies have addressed the impacts of this dysfunction. Our data indicate that the NK cells present in NOD mice, while deficient in number and function, may be rescued by CFA injection. One possibility is that CFA, directly or through its actions on dendritic cells, activates NK cells to increase expression of NKG2D and to release IFN-γ, which may subsequently regulate the autoimmune response.
Our adoptive transfer experiments clearly implicate IFN-γ as a critical mediator of CFA protection. The importance of IFN-γ is also supported by a report from Serreze's group showing that BCG or CFA treatment fail to inhibit diabetes development in IFN-γnull mice 14. In that report, the subset of IFN-γ-secreting cells important for protection was not identified. In our adoptive transfer experiments, we have shown that NK cell-secreted IFN-γ plays a key role for the protective effect of CFA because NK cells from IFN-γKO mice, transferred back to CFA-treated recipients, were unable to rescue mice from disease. However, because we did not perform experiments in which CD4+ T cells or NKT cells were depleted, we cannot exclude the possibility that secretion of IFN-γ from NKT or CD4+ T cells also participates in CFA protection from diabetes. Several studies provide supporting evidence for a role of IFN-γ in negative regulation of immune cells. First, in vitro experiments have shown that IFN-γ induces apoptosis of activated CD4 T cells in a model of mycobacterial infection 17, 18. Second, injections of recombinant IFN-γ decreased the incidence of diabetes in NOD mice by down-regulating anti-islet effector cells and diminishing insulitis 19. Third, Singh et al.20 have reported that BCG immunization causes in vivo apoptosis of diabetogenic T cells in NOD mice by a mechanism involving IFN-γ and TNF-α. Finally, we previously observed that an immunodominant population of IGRP-specific, diabetogenic CTL was significantly decreased in spleen and pancreatic islets of CFA-injected NOD mice 5, a finding that was reproduced in the current set of experiments (data not shown). Therefore, in pre-diabetic NOD mice, increased IFN-γ production by NK cells after CFA treatment serves to down-regulate a population of activated β-cell-specific CTL, possibly by inducing T cell apoptosis. Indeed, in our preliminary experiments, purified NK cells incubated with BCG in vitro secretes more IFN-γ compared with purified NK cells in medium alone. Moreover, when we incubated activated T cells with the supernatant from BCG-incubated NK cells or NK cells incubated with medium alone, we found that the level of apoptosis on T cells incubated with BCG-treated NK cells is significantly increased. Our result suggests that the protective effect of IFN-γ in diabetes might be through the induction of T cell death programming (data not shown). One research group has directly injected recombinant IFN-γ to block the onset of diabetes 19; however, contradictory studies have suggested that local production of IFN-γ from invading cells in pancreatic islets results in β-cell destruction 21–24, thus raising caution with regard to direct manipulation of disease with IFN-γ. In other words, it may be safer to stimulate NK cells to produce IFN-γ with agents less toxic than CFA, than to directly administer IFN-γ.
Recent work by Tian and his colleagues 25 have also demonstrated a protective role for NK cells. In their report, NOD mice received a series of polyI:C injections (a well-known NK cell activator) and the onset of diabetes was delayed or prevented. Similar to our studies, their adoptive transfer model indicated that without NK cells, the protective effect of polyI:C was abrogated. Although these data support our hypothesis, they suggest that unlike NK1- or NK2-like phenotypes 26–29, long-term polyI:C-treated NK cells regulate disease by exhibiting an NK3-like phenotype that is able to produce increased levels of TGF-β and IL-10. Tian and his colleagues 25 further hypothesized that these regulatory cytokines act to inhibit the Th1 response to islet autoantigens. Our experiments do not rule out this hypothesis.
How the regulatory function of NK cells is restored by CFA is still unknown, but Lanier and colleagues 15 have suggested that defective NKG2D expression underlies NK defects in NOD mice. They found that NKG2D expression was dramatically decreased in IL-2-activated NK cells from NOD mice as compared to other mouse strains, and that reduced expression of NKG2D was likely due to co-expression of an NKG2D ligand, RAE-1, at the cell surface, resulting in self modulation of NKG2D expression and functional impairment of NK cells. Similarly, Coudert et al.30 have reported a decreased function of NKG2D in NK cells after long period of exposure to NKG2D ligand, and Reyburn and his colleagues 31 have also shown that when human NK cells are recipients of NKG2D-ligand membrane transfer following synaptic encounters with NKG2D ligand-bearing cells, NKG2D ligand transfer and co-localization with NKG2D on the NK cell membrane resulted in abrogated NKG2D function and reduced cytotoxic function. These findings suggest that co-expression of NKG2D and its ligands on the NK cell surface negatively regulates NK activity. Our data show that CFA increases the frequency of NKG2D+ NK cells, and, notably, this cell population is able to up-regulate IFN-γ secretion, suggesting that promotion of NKG2D function is one mechanism of CFA protection. Moreover, our data demonstrate that expression of RAE-1 on NK cells is decreased following CFA treatment, further arguing that the effect of CFA on NKG2D ligand may mediate NK function.
As mentioned in the previous paragraph, Lanier and colleagues 15 have shown a decreased expression level of NKG2D on NK cells. A later study done by the same group 32 in NOD mice suggested that NKG2D is involved in the progression of diabetes. However, in their report, they mainly focused on the correlation of NKG2D on activated CD8+ T cells and NKG2D ligand on pancreatic islets. They showed an increased expression of RAE-1 on prediabetic (12–16 weeks of age) pancreatic islets and that infiltrating autoreactive CD8+ T cells in the pancreas from 16-week-old NOD mice express NKG2D. In addition, when NOD mice were treated with neutralizing anti-NKG2D antibody, diabetes onset is prevented. These results suggest that the interaction of NKG2D with its ligands might be one mechanism by which pancreatic islet cells were destroyed by autoreactive cells, possibly CD8+ T cells. Interestingly, we find that CFA injection increases NKG2D expression on NK cells and that these cells are protective against diabetes in the NOD mouse. It is possible that the administration of CFA increases NKG2D function in NK cells to produce IFN-γ, and that the produced IFN-γ induces apoptosis on autoreactive CD8 T cells. Moreover, because NK cells from CFA-treated mice are protective, it is unlikely that NKG2D-expressed NK cells would cause the destruction of islet cells.
In conclusion, we have shown that NK cell number and function are positively affected by a single injection of CFA, and that the regulatory action of CFA in diabetes is mediated by IFN-γ secreted from NK cells. In light of these results, the use of NK cell activators may be considered as a novel approach for the prevention of T1D.
Materials and methods
Female NOD mice, NOD/SCID mice and IFN-γ knockout NOD (NOD.IFN-γKO) mice were purchased from The Jackson Laboratories (Bar Harbor, ME, USA), and bred in a specific pathogen-free facility at the animal care unit of the Child and Family Research Institute (CFRI). The Animal Care Committee, Faculty of Medicine, University of British Columbia approved the care and use of all animals.
CFA immunizations and assessment of diabetes
Unless otherwise indicated, 5–6 week-old NOD mice were given a single injection of CFA emulsion (100 μL) subcutaneously in the base of tails. Blood glucose was monitored once weekly using test strips (Lifescan, Milpitas, CA). Mice with a blood glucose measurement of greater than 33 mM were considered diabetic and killed.
Antibodies and flow cytometry
The following mAb were purchased from BD Pharmingen (San Diego, CA): FITC-conjugated anti-CD3; FITC-conjugated anti-CD8; FITC-conjugated anti-IFN-γ; FITC-conjugated anti-BrdU; PE-conjugated anti-DX5; PerCP-conjugated anti-B220, PerCP-conjugated anti-CD3; APC-conjugated anti-IFN-γ and FITC-conjugated streptavidin. Allophycocyanin-conjugated anti-NKG2D and biotin-conjugated anti-RAE-1γ antibodies were purchased from eBiosciences (San Diego, CA). Cells were washed in PBS and incubated for 30 min with the indicated conjugated antibodies in a total volume of 40 μL PBS containing 0.3% BSA. Cells were washed three times and resuspended in PBS containing 1% fetal calf serum and 2.5% paraformaldehyde. Immunostained cells were analyzed on a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA) using CellQuest and FlowJo software for data analysis.
BrdU treatment and immunostaining with anti-BrdU antibody
To label proliferating NK cells, NOD mice were injected i.p. with 0.8 mg BrdU (Sigma Chemical Co., St. Louis, MO) 1 h prior to sacrificing mice. Peripheral blood and spleen were harvested for BrdU staining. The technique for staining BrdU-positive cells has been described previously 33, 34. Briefly, cells were stained with surface markers on ice for 30 min. After surface staining, cells were fixed with fixing buffer for 10 min and permeabilized at 37°C for 1 h using permeabilization buffer contains 0.5% BSA, 2 mM EDTA and 0.1% Triton X-100. After washing twice, cells were incubated in DNase solution at 37°C for 10 min and were subsequently stained with FITC-conjugated anti-BrdU antibody at room temperature in the dark. After staining, cells were washed twice and transferred into PBS containing 0.5% BSA and 2 mM EDTA. Immunostained cells were analyzed on a FACSCalibur flow cytometer.
Purification of NK cells
Spleen cells were pooled from NOD mice and stained with anti-CD3, anti-CD8 and anti-DX5 mAb. CD3–DX5+CD8– cells were sorted using a FACSAria cell sorter (Becton Dickinson, San Jose, CA) at the CFRI flow cytometry core.
Intracellular IFN-γ immunophenotyping
NOD mice were injected with a single dose of CFA or PBS. Mice were later killed at indicated time point and spleens were harvested. Splenocytes were subsequently incubated with the indicated antibodies on ice for 30 min to label surface markers. After washing three times, cells were incubated in FACS permeabilizing solution for 30 min. The permeable cells were stained with allophycocyanin-conjugated anti-mouse IFN-γ antibody. After staining, cells were washed twice and fixed in PBS containing 1% FCS and 2.5% paraformaldehyde. Immunostained cells were analyzed on a FACSCalibur flow cytometer. For analyzing IFN-γ secretion on IL-2 cultured NK cells, cells were incubated with GolgiStop (BD Pharmingen) for 4 h at 37°C before immunostaining.
For cytotoxicity assays, target cells (YAC-1) were labeled with 100 μCi 51Cr for 90 min at 37°C in RPMI 1640 medium containing 10% FCS, and washed three times with medium. 51Cr-labeled target cells (1 × 104) and effector (NK) cells were mixed in U-bottom wells of a 96-well microtiter plate at the indicated E:T ratios and plated in triplicate. After a 4-h incubation, cell-free supernatant was collected from each well, and radioactivity was measured in a gamma-counter. The percentage of specific 51Cr release was calculated according to the formula: % specific lysis = (experimental – spontaneous) release/(maximal – spontaneous) release × 100.
Quantitative (real-time) PCR was carried out using the ABI 7000 (Applied Biosystems) according to the manufacturer's instructions. PCR primers were as follows: RAE-1α, sense GTGAAACGATTGAAATTCTTGATACC, antisense GCTTTTCCTTGGGCTGCTTT; RAE-1β, sense AAGTAACATAAACAAGACCATGACTTCAG, antisense TCCTCAACTTCTGGCACAAATTT; RAE-1γ, sense GACGGCAAATGCCACTGAA, antisense CCACTTTGGTGTTAGACACCTTGTC; HPRT, sense TGGAAAGAATGTCTTGATTGTTGAA, antisense AGCTTGCAACCTTAACCATTTTG. Total RNA was treated with DNase I, and then first-strand cDNA was synthesized using oligo dT primer. The cycling condition for real-time PCR were: 95°C for 10 min, followed by 40 cycles of 95°C for 30 s, 55°C for 1 min and 72°C for 30 s. Data were analyzed using the Sequence Detector v1.7 Analysis Software (Applied Biosystems).
Adoptive transfers were performed as previously described 5. Briefly, recipient female NOD/SCID mice, 4–8 weeks of age, were injected on day 1 with 50 μL anti-asialo GM1 antibody (Wako Bioproducts, Richmond, VA) to selectively deplete NK cells. On day 2, mice were given a single 100-μL injection of CFA emulsion (Sigma) in the base of their tails. On day 3, recipient mice were injected i.v. with diabetic donor spleen cells (2 × 107 viable cells) and 5 × 105 purified NK cells derived from NOD or IFNγ KO NOD mice. Blood glucose was monitored once weekly, using test strips (Lifescan).
A Student's t-test was used to calculate statistical significance where indicated, and a single factor ANOVA was used for multigroup comparison. A log-rank test was applied to compare survival curves.
We thank Dr. Lisa Xu and the CFRI FACS core for cell sorting. This work was supported by grants from the Canadian Diabetes Association and the Canadian Institutes of Health Research.
Conflict of interest: The authors declare no financial or commercial conflict of interest.
- 28Production of IL-5 by human NK cells and regulation of IL-5 secretion by IL-4, IL-10 and IL-12. J. Immunol. 1995 154: 144.and
- 34Measurement of T and B cell turnover with bromodeoxyuridine. In: and (Eds.), Current Protocols in Immunology, John Wiley and Sons, New York 2007 pp. 4.7.1–4.7.6.and