RIG-I– and MDA5-Initiated Innate Immunity Linked With Adaptive Immunity Accelerates β-Cell Death in Fulminant Type 1 Diabetes

Clinical profiles of three autopsied patients with fulminant type 1 diabetes have been reported (4). Briefly, case 1 was a 14-year-old boy who died from diabetic ketoacidosis, following flu-like symptoms 5 days earlier. Case 2 was a 25-year-old man who died from diabetic ketoacidosis, following sudden symptoms of nausea and epigastralgia 2 days earlier. Case 3 was a 29-year-old man who died from diabetic ketoacidosis, following slight fever, nausea, and vomiting 2 days earlier.

Pancreatic tissues from 10 nondiabetic men (mean age ± SD, 62 ± 10 years) with gastric carcinoma who had undergone partial pancreatectomy and from five autopsied nondiabetic men (65 ± 11 years) were used as nondiabetic control subjects for immunohistochemical analysis. Pancreatic tissues from four autopsied type 1 diabetic patients (44 ± 9 years) who had histopathological insulitis and glutamic acid decarboxylase autoantibodies (titer: 3.0 ± 1.5 U/mL, cutoff <1.5) were examined as type 1 diabetic control subjects.

Methods for immunohistochemical analyses have been reported previously (4). Primary antibodies used in this study are listed in Supplementary Table 1. The definition of insulitis and frequencies of insulitis and mononuclear cell (MNC) phenotypes in islets of cases 1–3 have been documented previously (4). The number of pancreatic acinar cells surrounded by CD8+ T cells was counted in the randomly selected 60 photos of pancreatic section in each case. A confocal laser-scanning microscope, Fluoview FV1000 (Olympus, Tokyo, Japan), was also used.

We obtained sera from 18 patients with fulminant type 1 diabetes (age [range]: 32.3 ± 13.5 [17–58] years, sex [man/woman]: 12/6, duration: 31.0 ± 64.1 [0–240] days), 27 patients with typical type 1 diabetes (age: 31.4 ± 14.7 [6–55] years, sex: 12/15, duration: 12.5 ± 25.4 [0.1–108] months), and 30 nondiabetic control subjects (age: 33.3 ± 13.3 [20–60] years, sex: 17/13). Serum level of IFN-γ was measured by ELISA (PBL Biomedical Laboratories, R&D Systems, Piscataway, NJ).

The Ethics Committee of the University of Yamanashi approved all of the procedures performed in this study. All patients gave informed consent for measuring serum IFN-γ.

Differences in variables between groups were compared using Student t test and ANOVA. Fisher exact test was used to compare frequencies of islets. Values are expressed as means ± SD unless otherwise mentioned.

MDA5 was strongly expressed in β-cells, α-cells, and other types of islet cells of fulminant type 1 diabetic pancreata (Fig. 1A–D). In nondiabetic control and type 1 diabetic control subjects, weak MDA5 expression was observed in a few α-cells (Supplementary Fig. 1). Significant expression of RIG-I was observed preferentially in β-cells in all three patients with fulminant type 1 diabetes (Fig. 1E–H), yet it was not expressed in nondiabetic and type 1 diabetic control subjects (Supplementary Fig. 1). Enterovirus-capsid protein (VP1) was detected in β- and non–β-cells of fulminant type 1 diabetic pancreata confirming our previous report (Supplementary Fig. 2) (4) but not type 1 diabetic control and nondiabetic control subjects.

Both TLR3 and TLR4 were expressed in MNCs that had infiltrated islets of fulminant type 1 diabetic pancreata but not nondiabetic and type 1 diabetic control subjects (Table 1).

In all three pancreata of the patients with fulminant type 1 diabetes, IFN-α and -β₁ were strongly expressed (Fig. 2A–F). Some MNCs that had infiltrated around or in islets and pancreatic acinar and ductal cells also expressed theses cytokines in fulminant type 1 diabetes (Fig. 2A and B) but not in either control subject. The numbers of pancreatic acinar cells surrounded by CD8+ T cells were 11/mm², 24/mm², and 22/mm², respectively, in the pancreatic sections of cases 1, 2, and 3. Most IFN-α–expressing cells were β-cells, α-cells, and islet non–β- and non–α-cells (Fig. 2C–F). Interferon regulatory factor (IRF)-7 (7) was strongly expressed in β- and α-cells (Fig. 2G–I) and mostly stained around and in the nucleus of the islet cells (Fig. 2G). Major histocompatibility complex class I (MHC-I) was hyperexpressed in all islet cell subsets of fulminant type 1 diabetic pancreata (Fig. 2J). Nondiabetic control and type 1 diabetic control subjects did not show expression of IFN-α, IFN-β₁, or IRF-7 and hyperexpression of MHC-I in their islets.

Remarkable CD11c+ cells migration to the islets was observed in most islets of fulminant type 1 diabetic pancreata (Table 1). Intraislet CD11c+ cells expressed MHC class-II molecules (Fig. 2K–N). Confocal microscopy showed that some CD11c+ cells contained β-cell debris positive for insulin (Fig. 2O). Such findings were not observed in islets of nondiabetic or type 1 diabetic control subjects. Most CD11c+ cells were also positive for CD1a and some for CD68 (Fig. 2P), likely representing DCs and macrophage subsets. CD56+ or CD57+ NK cells and Tregs (CD4+ Foxp3+ cells) were not detected in or around islets of fulminant type 1 diabetic pancreata and either control. Tregs, CD4+ Foxp3+ cells, were not detected in or around the islets or in exocrine regions of the pancreas in fulminant type 1 diabetic, nondiabetic control, or type 1 diabetic control subjects (Table 1).

IL-18 was expressed in islet cells in all three fulminant type 1 diabetes (Fig. 3A and E). Most residual β-cells expressed both IFN-γ and IL-18 (Fig. 3G and H). IL-18, IFN-γ, and CXCL10 colocalized in most β- and islet non–β-cells (Fig. 3A–H). IL-12 was not expressed in any cells in affected pancreata. A few islets of type 1 diabetic control subjects (mean [range]: 2.8% [0–5.2]) expressed IL-18 and IFN-γ but not CXCL10. Nondiabetic control subjects did not express IL-18, IFN-γ, and CXCL10.

Serum levels of IFN-γ in patients with fulminant type 1 diabetes were approximately three times higher than those in nondiabetic and type 1 diabetic control subjects (Fig. 3I).

Elevated expression of Fas in islet cells coincided with marked MNC infiltration in fulminant type 1 diabetic pancreata (Fig. 3J). The subsets of islet cells with Fas expression were mostly β-cells. Fas-ligand (FasL)-bearing MNCs infiltrated most islets of fulminant type 1 diabetic pancreata (Fig. 3K–M) (Table 1). In islet cells of nondiabetic control and type 1 diabetic control subjects, Fas was not expressed (Supplementary Fig. 3). FasL-bearing MNC infiltration of islets was observed in type 1 diabetic but not nondiabetic control subjects (Table 1) (Supplementary Fig. 3).

Retinoic acid–receptor responder 3 (RARRES3) (8,9) was expressed in β-cells of fulminant type 1 diabetic pancreata (Supplementary Fig. 4). Cleaved caspase 8, a marker of the Fas-mediated extrinsic apoptotic pathway, cleaved caspase 9, a marker of the activated non–Fas-mediated apoptotic pathway, and activated caspases 3, a marker of the end stage of β-cell apoptosis, were expressed specifically in islet β-cells (Supplementary Fig. 4). In islets of autopsied nondiabetic control subjects, RARRES3, cleaved caspases 8, -9, and -3 were not expressed (Supplementary Fig. 5). In type 1 diabetic control subjects, RARRES3, cleaved caspases 8, -9, and -3 were expressed weakly in some islet β-cells (Supplementary Fig. 5).

Both RIG-I and MDA5 were strongly expressed in β-cells of fulminant type 1 diabetic pancreata. MDA5 was also hyperexpressed in α-cells and non−β-/non–α-cells in affected islets. Hyperexpression of RIG-I and MDA5 with expression of IFN-α and -β₁ in β-cells suggests a crucial role of RIG-I and MDA5 for sensing and responding to enterovirus infection in the pancreas of patients with fulminant type 1 diabetes. Mutations of MDA5 genes have been implicated in reducing the risk of type 1 diabetes (10). Reports also noted RIG-I mRNA expression in human islets infected with Coxsackievirus B3 and B5 (11,12). We showed that IRF-7, a master transcription factor of IFN-α and -β (7), translocated to the nucleus and that IFN-α and -β, essential factors that protect β-cells against viral infection (6), were strongly expressed in both β- and α-cells. These results indicate that all islet cells are in an activated state of innate immunity in response to enterovirus in patients with fulminant type 1 diabetes.

Increased TLR3+ MNCs that infiltrate affected islets should participate in sensing viral RNA and subsequently destroy β-cells with RIG-I– and MDA5-initiated proinflammatory signal axes in the innate immune response against Coxsackievirus B3 (13). Intra- and peri-islet DCs and macrophage subsets drastically increased in number and showed active phagocytosis of enterovirus-infected β-cells, whereas MHC-I was hyperexpressed in all islet subsets. Some DCs and macrophage subsets also expressed MHC-II molecules. Activated innate immune responses including virus sensing by RIG-I and MDA5 with subsequent IFN-α and -β production and DC and macrophage activation will not only protect for enteroviral infection by upregulating RIG-I and MDA5 (12,14) but will also enhance the adaptive immunity cascades for islet cell destruction (6,15,16). Indeed, patients with fulminant type 1 diabetes showed elevated serum levels of IFN-γ.

CD4+ Foxp3+ cells, which represent a pivotal subset of Tregs, were not observed in or around the islets of fulminant type 1 diabetic pancreata, suggesting that the extremely polarized local condition to predominance for Th1 in response to enteroviral infection suppresses Treg differentiation from naive T CD4+ precursors (17). In turn, the Treg-depleted islet condition enhances Th1 cytokine (i.e., IFN-γ) generation.

Notably, IL-18, an IFN-γ–inducing factor, was extensively expressed in islet cells of infected fulminant type 1 diabetic pancreata. In response to viral infection, IL-18 is promptly secreted from virus-activated macrophages, DCs, and T cells (18), stimulating production of IFN-γ synergistically with IFN-α and -β through a unique pathway that sometimes occurs independently of IL-12 or NK cells (19). Conversely, IL-18 can be induced by IFN-γ alone or in combination with other cytokines in islet β-cells (20). Thus, for fulminant type 1 diabetes, enterovirus itself or enterovirus-activated T cells and macrophages most likely infiltrate islets to induce IL-18 production in these cells. In addition, IFN-α and -β, produced in islet cells and islet-infiltrating MNCs, can enhance IL-18–mediated signaling (21). Subsequently, islet-secreted IL-18 may induce IFN-γ production via receptors on the islet cells or islet stromal cells in an autocrine/paracrine manner. Once this positive autocrine/paracrine circuit for production of IL-18, IFN-γ, and CXCL10 is established in islet cells, destructive mechanisms involving CXCR3+ T cells and macrophages might persist until complete destruction of the β-cells (4).

We found that Fas was highly expressed in affected islet β-cells and islet-infiltrating FasL+ cells. Taken together with the finding that MHC-I and IFN-α, -β, and -γ were strongly expressed in affected islet cells, effecter mechanisms for β-cell apoptosis in fulminant type 1 diabetes are likely mediated in part by MHC-I and by the Fas-FasL pathway (22). Inflammation-induced Fas-FasL expression in β-cells was reported to lead to rapid and massive β-cell destruction (23). Other apoptotic mechanisms through the IFN-γ–dependent JAK/STAT pathway (24) and innate immune pathway (25) will also exert β-cell destruction.

This study was partly supported by grants from the Ministry of Education, Science, Sports and Culture and Ministry of Health, Labor and Welfare (Japan).

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

K.A. and Y.N. conducted immunohistochemical staining, RT-PCR data analysis, and discussed, reviewed, and edited the article. S.T. contributed to planning and discussion and edited the article. T.M. and A.S. sampled autopsied pancreas and participated in discussion. T.A., M.S., and H.S. contributed to discussion and reviewed and edited the article. S.T., T.M., M.I., D.A., and T.A. recruited serum samples, measured serum levels of IFN-γ, and participated in discussion. F.F., A.K., and M.K. contributed to analysis of immunostained sample data. J.I., H.F., and T.E. contributed to analysis of nondiabetic pancreas and Treg+ lymph nodes. T.K. analyzed data and wrote and edited the article.

The authors thank K. Hosaka, C. Imai, and S. Takei for excellent secretarial work; Drs. T. Momotsu and E. Okazaki (Niigata City General Hospital) for generous assistance; and Professors A. Nakao (Department of Immunology) and R. Kato (Department of Pathology) at the University of Yamanashi for critical assessment and assistance of the article.