Deficiency in Type I Interferon Signaling Prevents the Early Interferon–Induced Gene Signature in Pancreatic Islets but Not Type 1 Diabetes in NOD Mice

Type 1 diabetes is caused by the autoimmune destruction of insulin-producing β-cells. Human type 1 diabetes and the NOD mouse are characterized by infiltration of immune cells into pancreatic islets. In NOD mice, myeloid cells are observed from ∼3–4 weeks of age (1–5), followed by an influx of β-cell–specific T cells that cause β-cell death. Diabetes occurs after 15 weeks of age.

Type I interferon (IFN) has been associated with the development of autoimmune diabetes in humans and mice. Elevated IFN-α mRNA transcripts were observed in the pancreata and islets of deceased diabetic patients (6). β-Cells from type 1 diabetic individuals with hyperexpression of HLA class I contained immunoreactive IFN-α (7,8). Transgenic mice expressing IFN-α in β-cells developed insulitis and diabetes, but the severity was dependent on the genetic background, suggesting IFN-α may be required for the initiation of type 1 diabetes in genetically susceptible hosts (9).

Transcriptional profiling of NOD islets and pancreatic lymph nodes (PLN) before T-cell infiltration of islets identified an IFN-induced gene signature, and it is hypothesized that this contributes to the initiation of diabetes (10–12). To address the significance of this gene signature, we generated NOD mice deficient in IFN-α receptor-1 (IFNAR1). All type I IFNs signal through this receptor, so IFNARI deficiency eliminates all type I IFN responses. Unexpectedly, we observed no effect of IFNAR1 deficiency on type 1 diabetes development in NOD mice, despite a reduction in type I IFN-regulated gene expression in the islets of knockout mice.

Mice were maintained at St. Vincent’s Institute, and all experiments were approved by the institutional animal ethics committee. Mice (129.IFNAR1^−/−) (13) were backcrossed onto the NOD/Lt genetic background for 11 generations and intercrossed to produce NOD.IFNAR1^−/− mice. NOD8.3 mice have been described previously (14). NOD8.3 were bred with NOD.IFNAR1^−/− mice to generate NOD8.3/IFNAR1^−/− mice. NOD/Lt mice were purchased from the Walter and Eliza Hall Institute of Medical Research (Melbourne, VIC, Australia). NOD.Tlr2^−/− and NOD.Tlr9^−/− mice were generated at the Comparative Genomics Centre by intercrossing C57BL/6.Tlr2^−/− and C57BL/6.Tlr9^−/− mice (provided by Dr. Shizuo Akira, Osaka University, Osaka, Japan) with NOD/Lt mice and then backcrossing hemizygous mutant progeny to NOD/Lt for 10 generations before intercrossing.

DNA samples were genotyped as described (15). DNA was processed for Illumina mouse medium density linkage panel containing 1,449 single nucleotide polymorphism (SNP) loci by The Centre for Applied Genomics (Toronto, ON, Canada). The Jackson Laboratory Mouse Genome Informatics (MGI) and National Center for Biotechnology Information (NCBI) databases were used to identify strain differences. DNA from the 11th generation was genotyped using high resolution melt analysis. TaqMan probes (Applied Biosystems, Grand Island, NY) were obtained for differential SNPs across the IFNAR1 interval. Analysis was performed using LightCycler 480 High Resolution Melting Master mix (Roche Applied Science) on a Roche Lightcycler 480 system with NOD and 129 DNA as controls.

Female mice were monitored for diabetes for 300 days as described previously (15). Mice with two consecutive blood glucose measurements of ≥15 mmol/L were considered diabetic. For immunohistochemistry, pancreata were snap-frozen in O.C.T. (Sakura Finetek, Torrance, CA). Preparation of sections, staining, and scoring for insulitis were performed as described previously (16).

Islets of Langerhans were isolated using collagenase P (Roche, Basel, Switzerland) and Histopaque-1077 density gradients (Sigma-Aldrich) as previously described (14).

Total RNA was extracted using Nucleospin RNA XS kits (Macherey-Nagel). First-strand cDNA was generated using High Capacity cDNA Reverse Transcription kits (Applied Biosystem). Real-time PCR analysis was performed using a Roche LightCycler 480 and TaqMan gene expression assays (Applied Biosystem). Pan-Ifnα primers and TaqMan probes (Applied Biosystem) were described previously (10). To determine relative expression, Ct values from test samples were normalized to Ct values from control NOD islets at 1 week of age using 2^−ΔΔCt.

CD8⁺ T cells from NOD8.3 mice or NOD8.3/IFNAR1^−/− mice were labeled with carboxy fluorescein succinimidyl ester (CFSE), resuspended at 2.5 × 10⁷ cells/mL in PBS, and 200 μL injected intravenously into the tail vein of recipient mice, as described previously (17). The mice were killed after 5 days, and the inguinal lymph node (ILN), PLN, and islets were harvested and analyzed by flow cytometry, as described below.

Analysis was performed using the FACSCalibur flow cytometer (BD Biosciences, San Jose, CA) and FlowJo analysis software (Tree Star, Inc., Ashland, OR). Islets were used freshly isolated or after 48 h culture with 100 units/mL IFN-γ (BioLegend) or 1,000 units/mL IFN-α (PBL Interferon Source) in CMRL-1066 (Life Technologies), as described (14). Islets were dispersed to single cells and infiltrating lymphocytes stained as described previously (14,15). For analysis of β-cell major histocompatibility complex (MHC) class I expression, dispersed islets were stained with antibodies recognizing CD45 (to exclude infiltrating lymphocytes) and biotinylated anti-mouse H2K^d monoclonal antibody (BD Pharmingen), followed by streptavidin-allophycocyanin (Invitrogen). PLN, ILN, and spleens were prepared and stained as previously described (15). For analysis of adoptive transfer, single-cell preparations of islets, ILN, and PLN were stained with anti-CD8 as described previously (17).

Statistical analysis was performed using GraphPad Prism software (GraphPad Software, Inc., San Diego, CA). All data shown as bar graphs are represented as mean ± SEM. Data were analyzed by log-rank analysis or one-way ANOVA with posttests as indicated.

Backcrossed NOD.IFNAR1^−/− mice were of the NOD genotype across the whole genome except for the region on chromosome 16 encompassing the Ifnar1 gene. Fine mapping of the SNPs surrounding the Ifnar1 locus for 129 DNA identified the region between and including Chr16:89,057,197 (rs4216078; GRCm38/mm10 assembly) and Chr16:92,546,724 (rs4219643), which includes the Ifnar1 gene. Diabetes susceptibility/resistance alleles have not been reported within this interval (18).

We examined the islets of NOD mice from 1 to 14 weeks of age for the expression of type I IFN-induced genes, Ifit1, Isg15, Oas1a, and Mx1, and the chemokine receptor Cxcr4. Expression of these genes was detected in NOD islets at 1 week of age. Except for Cxcr4, expression peaked at 3–4 weeks of age, declined at 5–6 weeks of age, and increased again at 10–14 weeks of age (Fig. 1A and B). Ifnα mRNA expression was also elevated at 3–4 weeks and at 10–14 weeks of age (Fig. 1C). Expression of all IFN-induced genes was significantly reduced in NOD.IFNAR1^−/− islets up to 6 weeks of age, consistent with a lack of type I IFN signaling in the NOD.IFNAR1^−/− mice and indicating that type I IFN is responsible for the increased expression of these genes. At 10–14 weeks of age, expression of Cxcr4 and Isg15 was increased above background in the NOD.IFNAR1^−/− islets, suggesting another cytokine(s) produced in islets after 10 weeks of age is inducing gene expression independent of type I IFN signaling (Fig. 1A and B).

In NOD mice, cell debris released from dying β-cells is poorly cleared and forms protein complexes that may stimulate Toll-like receptor (TLR) signaling pathways and type I IFN production (11,19). TLR2 and TLR9 are linked to diabetes onset in the NOD mouse (20–22). To investigate whether TLR2 or TLR9 signaling promotes IFN-induced gene expression in islets, we examined gene expression in NOD.Tlr2^−/− and NOD.Tlr9^−/− mice. Expression of Ifit1, Isg15, Oas1a, or Cxcr4 was the same in NOD, Tlr2^−/−, or Tlr9^−/− islets at 2, 4, and 6 weeks of age (Fig. 2 and data not shown). Mx1 expression was significantly reduced in NOD.Tlr2^−/− compared with NOD islets at 2 weeks of age. This suggests that TLR2 or TLR9 are unlikely to play a major role in the generation of type I IFN, resulting in increased gene expression within islets of NOD mice.

The expression of type I IFN in NOD islets may result in MHC class I upregulation on β-cells, resulting in recognition and killing by cytotoxic T lymphocytes. As expected, MHC class I expression on NOD β-cells increased with age (Fig. 3A–C) (16). MHC class I expression on NOD.IFNAR1^−/− β-cells was similar to NOD at each age, including at 5 and 7 weeks of age, when we detected type I IFN–induced gene expression within the islets. These results indicate that type I IFN is not required for the upregulation of MHC class I expression on β-cells in vivo in NOD mice.

NOD and NOD.IFNAR1^−/− islets and CD45⁺ cells also responded equally to recombinant IFN-γ in vitro, as determined by upregulation of MHC class I (Fig. 3D and E). As expected, NOD.IFNAR1^−/− islets did not upregulate MHC class I in response to IFN-α. Islet CD45⁺ cells did not respond to IFN-α (data not shown). This demonstrates the lack of response to type I IFN has no effect on responses to IFN-γ in the NOD.IFNAR1^−/− mice.

We next compared the development of insulitis in NOD and NOD.IFNAR1^−/− mice. The percentage of CD45⁺ leukocytes infiltrating the islets increased as both strains of mice aged, from 5–15% at 35 days of age to 35–40% at >95 days of age (Fig. 4A). The frequency of CD11c⁺ dendritic cells (DCs), F480⁺ macrophages, or CD4⁺ and CD8⁺ T lymphocytes in the islets was similar between NOD and NOD.IFNAR1^−/− mice at each age (Fig. 4B and C). Histological analysis of pancreata at 50 and >95 days of age (Fig. 4D) showed no difference among NOD, NOD.IFNAR1^+/−, and NOD.IFNAR1^−/− mice. These data demonstrate that lack of type I IFN signaling does not prevent development of insulitis in NOD mice.

We observed no difference in the absolute number or the percentage of CD4⁺ and CD8⁺ lymphocytes and B lymphocytes within the PLN, ILN, or the spleens of NOD and NOD.IFNAR1^−/− mice at any age (Supplementary Fig. 1A and B), indicating that the lack of type I IFN signaling does not affect immune cell populations. The percentage of conventional DCs (CD11c^highSiglecH⁻), plasmacytoid DCs (pDCs) (CD11c^midSiglecH⁺) (Fig. 5A), or F480⁺ macrophages (data not shown) in the spleens and lymph nodes of NOD.IFNAR1^−/− mice was also no different compared with NOD mice (Fig. 5A).

To determine if IFNAR1 deficiency effects priming of CD8⁺ T cells in the draining PLN, we adoptively transferred CD8⁺ T cells from NOD8.3 mice that recognize the β-cell antigen glucose-6-phosphatase catalytic subunit–related protein (IGRP) into NOD or NOD.IFNAR1^−/− mice (14). NOD8.3 T cells proliferated specifically in the PLN and islets of NOD and NOD.IFNAR1^−/− mice (Fig. 5B and C). NOD8.3 CD8⁺ T cells deficient in the IFNAR1 receptor also specifically proliferated in the PLN and islets of NOD mice. NOD8.3 mice lacking IFNAR1 developed diabetes at a similar rate and incidence to NOD8.3 mice (12 of 13 NOD8.3/IFNAR1^−/− mice with mean ± SEM onset of diabetes at 73 ± 6 days of age compared with 12 of 12 NOD8.3 mice with mean ± SEM onset of diabetes at 82 ± 24 days of age). These data demonstrate that deficiency of IFNAR1 receptors on DCs or CD8⁺ T cells has little effect on the proliferation or migration capacity of antigen-specific T cells.

Finally, cohorts of female mice were monitored for diabetes for 300 days (Fig. 6). All three genotypes developed diabetes between 90 and 120 days of age. At 300 days of age, the diabetes incidence of the NOD, NOD.IFNAR1^+/−, and NOD.IFNAR1^−/− mice was not significantly different. These data clearly show that a lack of type I IFN signaling neither delays nor protects mice from diabetes.

To definitively study the requirement for type I IFNs in the development of diabetes, we generated IFNAR1-deficient NOD mice. We detected a significant type I IFN–induced gene signature in NOD islets that peaked at 3–4 weeks of age and was ablated in NOD.IFNAR1^−/− islets. However, deficiency in type I IFN signaling had no effect on diabetes development. Therefore, type I IFNs are not essential for diabetes in the NOD mouse.

Gene deletion is a robust and direct method for testing disease mechanisms. A shortcoming of this approach may be compensation for the missing gene by another pathway. Another shortcoming, in the NOD model, is the extensive backcrossing required. Even when most genes are NOD-derived, those closely linked to the deleted locus can affect diabetes development (23). Neither of these problems applies here. The type I IFN–induced gene signature was ablated in NOD.IFNAR1^−/− islets, demonstrating compensation did not develop in the knockout mice. However, we cannot rule out the possibility that type I IFNs are involved in diabetes and that their role has been replaced by another cytokine, operating via a different gene signature, in the IFNARI^−/− mice. This may be the case for MHC class I expression, which is not affected because type II IFNs can compensate for any role type I IFNs play (16). The deficiency in type I IFN signaling had no effect on other stages in diabetes development, suggesting hitchhiker genes affecting diabetes are not present. The results also do not exclude a role for type I IFNs in human diabetes. However, this result in NOD mice must be considered when clinical trials of type I IFN neutralization or blockade are planned. Our results make it less likely that a benefit will be observed.

Our results suggest a link between the IFN-induced gene signature and diabetes is unlikely. Our observations of the type I IFN–induced gene signature are consistent with other studies (10,11,24). However, the effect on diabetes differs markedly. Diabetes was reduced in mice treated with an IFNAR1 neutralizing antibody or an antibody to deplete pDCs producing type I IFN. The reason(s) for these differences is unclear, but it is possible that the short duration of antibody administration at a critical time may reveal benefits not observed in IFNAR1^−/− NOD mice. An increase in immature DCs was observed in mice treated with IFNAR1-neutralizing antibody, and this would affect the production not only of type I IFNs but also of other cytokines and antigen presentation for T-cell priming (11,12,25). The depletion of pDCs could also affect these pathways (11,12,25).

The mechanism of the type I IFN–induced gene signature is unclear but may be due to cellular processes associated with tissue development, including remodelling and cell death. Activation of TLR2, TLR7, and TLR9 by the accumulation of islet cell debris and the consequent production of type I IFN may contribute to diabetes in NOD mice (11,20). Treatment of NOD mice with the TLR7/9 agonist IRS₉₅₄ prevented the development of diabetes, an effect attributed to alterations in the TLR9 signaling pathway. Similarly NOD mice genetically deficient in TLR2 or TLR9 are protected from diabetes (11,20–22). Deficiency of these receptors has been shown to reduce the priming of diabetogenic T cells, suggesting that the protective effect of knocking out these TLR pathways may be due to alterations in antigen presentation and is less likely to depend on effects on type I IFN (20,21).

In conclusion, we have shown that a deficiency in type I IFN signaling prevents the early IFN–induced gene signature in pancreatic islets but not spontaneous diabetes in NOD mice. We believe our study is the most direct analysis of the effects of type I IFN signaling, and the results dissociate the type I IFN–induced gene signature in islets from diabetes development in NOD mice.

Acknowledgments. The authors thank Dr. G. Karupiah (Australian National University, Canberra, ACT, Australia) for IFNAR1^−/− mice; Dr. T. Mysore (James Cook University, Townsville, QLD, Australia) for assistance with Tlr2^−/− and Tlr9^−/− mice; C. Kos, R. Ayala-Perez, S. Thorburn, S. Ivory, and L. Mackin (all from St. Vincent’s Institute, Fitzroy, VIC, Australia) for excellent technical assistance; and D. Novembre-Cycon, R. Branch, H. Abidin, and A. Gomes (all from St. Vincent’s Hospital Melbourne, Fitzroy, VIC, Australia) for animal husbandry.

Funding. This study was supported by a National Health and Medical Research Council (NHMRC) of Australia Program grant, a Juvenile Diabetes Research Foundation research grant, and fellowships from the NHMRC (A.G.B. and H.E.T.). This study was supported in part by the Victorian Government’s Operational Infrastructure Support Program.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. H.S.Q. and K.L.G. researched data, designed experiments, and wrote the manuscript. S.M.-H., A.K., A.H., S.F., and L.E. researched data. T.C.B., A.G.B., and T.W.H.K. contributed to experimental design and discussion and reviewed and edited the manuscript. H.E.T. contributed to discussion; designed experiments; and reviewed, edited, and wrote the manuscript. H.E.T. and K.L.G. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. The abstract of this study was presented at the 13th International Congress of the Immunology of Diabetes Society, Mantra Lorne, Victoria, Australia, 7–11 December 2013.