Introduction

Type 1A diabetes results from the cell-mediated autoimmune destruction of pancreatic beta cells [1, 2]. It is a multi-step process that can be divided into three stages. The first stage involves the presentation of beta cell autoantigens by professional antigen-presenting cells. CD4+ (helper) T cells subsequently recognise the autoantigens presented by antigen-presenting cells expressing MHC class II molecules via the T cell receptor. Finally, CD8+ (cytotoxic) T cells, which are activated by Th1 helper T cells, attack and destroy the pancreatic beta cells.

Several effecter molecules are thought to kill beta cells in the final phase of insulitis, including Fas–Fas ligand, perforin–granzyme and inflammatory cytokines such as TNF-α or IL-1β [3, 4]. Among them, TNF-α or IL-1β induces apoptotic death of cells from a mouse beta cell line [5] or isolated mouse beta cells [6], while injection of neonatal non-obese diabetic (NOD) mice with a neutralising anti-TNF-α antibody protected against diabetes [7]. Insulitis develops in NOD mice lacking the TNF receptor 1 (TNFR1 or TNFRp55) similarly to in normal NOD mice, but progression to diabetes does not occur in these mice [8]. In addition, treatment with specific neutralising antibodies for IL-1β prevents cyclophosphamide-induced diabetes in NOD mice [9], while IL-1 receptor deficiency slows progression to diabetes in this mouse [10]. Furthermore, prevention of insulitis and diabetes has been achieved in NOD mice by treatment with a recombinant IL-1 receptor antagonist [11]. Finally, we have detected the production of TNF-α and IL-1β in the pancreas of patients with recent-onset type 1 diabetes [12]. All of the above findings suggest an important role of these inflammatory cytokines in the destruction of beta cells in vitro and in vivo.

To clarify the significance of inflammatory cytokines in the development of human type 1 diabetes, we investigated the production of these cytokines, as well as the populations of islet-infiltrating T cells, dendritic cells and macrophages, during and after beta cell destruction in pancreatic biopsy specimens from patients with recent-onset type 1 diabetes.

Subjects and methods

Subjects

We examined six patients with recent-onset type 1 diabetes (four men, two women). All of the patients fulfilled the criteria of the American Diabetes Association and WHO for a diagnosis of type 1 diabetes [13]. The mean age of the patients was 23.7 ± 6.4 years (mean ± SD) and the mean duration of diabetes was 3.3 ± 2.6 months (mean ± SD). Islet cell antibodies, GAD antibodies or both types of antibodies were detected in all six patients. Islet cell antibodies and GAD antibodies were identified by indirect immunofluorescence or with a radioimmunoassay kit, respectively (Table 1). Pancreatic body tissues from three patients (two men, one woman) who had gastric cancer and underwent partial pancreatectomy were used as the normal control. The age of control subjects was 52.3 ± 7.0 years (mean ± SD), and the glucose tolerance of these subjects was in the normal range before the operation. We have reported part of the histological findings for the type 1 diabetic patients and control subjects previously [2, 14, 15].

Table 1 Patients’ profiles
Full size table

Pancreatic sections

The study protocol was approved by the Ethics Committee of Osaka University Medical School and written informed consent was obtained from all patients. Pancreatic biopsy was performed as described previously in the past 16 years and there were no major complications [14]. Briefly, a laparoscope was introduced into the abdominal cavity from an incision near the umbilicus under local anaesthesia. Under direct inspection, one to three pieces of tissue were obtained from the body of the pancreas with biopsy forceps. Blocks of these biopsy specimens (20–40 mg in each patient) were immediately frozen in dry ice-isopentane, and 5-μm thick sections were cut on a cryostat. About 300 consecutive sections were cut from each biopsy specimen and stored at −80°C until use. Haematoxylin and eosin staining was performed in every 11th cryostat section to detect islets. Control tissues were prepared in the same manner as the biopsy specimens.

Immunohistochemistry

We used a double-immunofluorescence method to detect insulitis, mononuclear cell infiltration into the islets and production of TNF-α or IL-1β. After drying in air, the frozen sections were fixed in ice-cold acetone for 15 min and washed three times for a total of 15 min in PBS, pH 7.4. Endogenous biotin was blocked with a biotin blocking system (Dako Japan, Kyoto, Japan). Then the sections were incubated overnight at 4°C with one of the following primary antibodies: anti-human CD3 (T3-4B5; Dako Japan), anti-human CD68 (EBM11; Dako Japan) or anti-human CD11c (BU15; Immunotech, Marseille, France). After washing in PBS, the sections were incubated for 60 min at room temperature with biotinylated horse anti-mouse immunoglobulins (Vector Laboratories, Burlingame, CA, USA), and then for an additional 40 min with fluorescein isothiocyanate (FITC)-conjugated avidin (Vector Laboratories). These procedures were followed by incubation with guinea-pig anti-insulin antibodies (Dako Japan) or rabbit anti-glucagon antibodies (kindly provided by S. Iwasa, Takeda Chemical, Osaka, Japan) and incubation with the corresponding secondary antibody, rhodamine-conjugated anti-guinea-pig immunoglobulins (Chemicon International, Temecula, CA, USA) or Texas Red-conjugated anti-rabbit immunoglobulins (Amersham Pharmacia Biotech, Bucks, UK). After each incubation, the sections were washed three times with PBS.

Production of TNF-α and/or IL-1β was visualised by double immunofluorescence staining. First, staining was performed with anti-recombinant human TNF-α antibody or anti-recombinant human IL-1β antibody (R&D Systems, Minneapolis, MN, USA), biotinylated rabbit anti-goat immunoglobulins (Vector Laboratories) and rhodamine-conjugated avidin (Vector Laboratories). The sections were incubated with biotinylated rabbit anti-goat immunoglobulins for 60 min at room temperature and then for an additional 40 min with rhodamine-conjugated avidin. Next, the sections were blocked with an avidin–biotin blocking system (Dako Japan), washed in PBS and incubated with primary antibodies targeting T cells (CD3), macrophages (CD68) or dendritic cells (CD11c). After washing in PBS, the sections were incubated for 60 min at room temperature with biotinylated horse anti-mouse immunoglobulins (Vector Laboratories) and then for an additional 40 min with FITC-conjugated avidin (Vector Laboratories). To achieve localisation of the islets, staining for insulin or glucagon was done with the adjacent serial section.

Each section was washed in PBS, and mounted in aqueous mounting medium (Perma Fluor; Immunon, Pittsburgh, PA, USA) before being observed under a fluorescence microscope with epi-illumination (Olympus, Tokyo, Japan).

We examined 26–54 sections to detect mononuclear cell infiltration of islets and the production of cytokines in each subject.

Data are presented as the percentage of cytokine-positive cells with each cell surface marker in double immunofluorescence-stained sections.

Statistical analysis

The significance of differences between two groups was evaluated by the Mann–Whitney U test and a p value of less than 0.05 was taken to indicate the presence of a statistically significant difference.

Results

Phenotypic analysis of mononuclear cells infiltrating the islets

Pancreatic islets were detected in all the type 1 diabetic patients. Light microscopic examination of haematoxylin and eosin staining sections revealed that most of these islets were small and atrophic as compared with islets in normal pancreas. When islets with two or more infiltrating mononuclear cells were defined as insulitis-positive, 10–60 insulitis-positive islets were identified in the six patients by the double immunofluorescence method (Fig. 1). Infiltration of T cells, macrophages and dendritic cells was detected in 36.7 (15.4–66.7)%, 49.0 (27.7–68.4)% and 34.5 (18.8–47.4)% of samples from each patient. Comparison between islets with beta cells and islets without beta cells showed infiltration of 46.0 (17.4–83.3)% vs 12.5 (0–33.3)% by T cells (p < 0.01), 48.6 (28.9–100)% vs 49.2 (0–60.0)% by macrophages (NS) and 33.3 (0–80.0)% vs 35.7 (0–50.0)% by dendritic cells (NS; all above values median [range]) (Table 2). In contrast, no insulitis-positive islets at all were observed in three control subjects.

Fig. 1
figure 1

Pancreatic sections from a patient with recent-onset type 1 diabetes were stained with the following: a anti-insulin antibody (red) and anti-CD3 antibody (green) (case 2); b anti-insulin antibody (red) and anti-CD68 antibody (green) (case 4); and c anti-glucagon antibody (red) and anti-CD11c antibody (green) (case 1). Mononuclear cells are infiltrating the islets

Full size image
Table 2 Insulitis for CD3, CD68 and CD11c by immunohistochemical analysis
Full size table

Production of TNF-α

TNF-α was detected in the islets of all six patients (Fig. 2). TNF-α-positive cells were detected in 3.5 to 54.5% of the islets with insulitis, and production of TNF-α was identified both inside and around the islets. Although TNF-α production was not seen in T cells, it was found in 25.0 (4.3–46.9)% of macrophages and 11.8 (0–40.0)% of dendritic cells in samples from each patient (values median [range]) (Table 3). The prevalence of TNF-α-positive macrophages or dendritic cells was 22.2% in islets with T cell infiltration and 24.5% in those without. On the other hand, TNF-α-positive cells were not detected in the 110 islets from three control subjects.

Fig. 2
figure 2

Pancreatic sections from a patient with recent-onset type 1 diabetes were stained with the following: a anti-TNF-α antibody (red) and anti-CD3 antibody (green) (case 3); b anti-TNF-α antibody (red) and anti-CD68 antibody (green) (case 6); and c anti-TNF-α antibody (red) and anti-CD11c antibody (green) (case 3). TNF-α was detected in CD68-positive cells and CD11c-positive cells. Arrowheads b, c, representative double positive cells

Full size image
Table 3 TNF-α levels for mononuclear cells infiltrating the islets
Full size table

Production of IL-1β

IL-1β was detected in the islets of all six patients (Fig. 3). IL-1β-positive cells were detected in 12.5 to 47.6% of the islets with insulitis, and production of IL-1β was identified both inside and around the islets. IL-1β was produced by 1.8 (0–11.3)% of T cells, but by 19.2 (0–35.3)% of macrophages and 10.7 (0–31.3)% of dendritic cells in samples from each patient (values median [range]) (Table 4). The prevalence of IL-1β-positive macrophages or dendritic cells was 19.1% in islets with T cell infiltration and 40.5% in those without. On the other hand, IL-1β-positive cells were not detected in the 116 islets from three control subjects.

Fig. 3
figure 3

Pancreatic sections from a patient with recent-onset type 1 diabetes were stained with the following: a anti-IL-1β antibody (red) and anti-CD3 antibody (green) (case 1); b anti-IL-1β antibody (red) and anti-CD68 antibody (green) (case 5); and c anti-IL-1β antibody (red) and anti-CD11c antibody (green) (case 1). IL-1β was detected in CD3-positive cells, CD68-positive cells and CD11c-positive cells. Arrowheads indicate representative double positive cells

Full size image
Table 4 IL-1β levels for mononuclear cells infiltrating the islets
Full size table

Discussion

This study reveals the source of inflammatory cytokines (TNF-α and IL-1β) and the populations of islet-infiltrating cells in patients with recent-onset type 1 diabetes and insulitis.

The main source of TNF-α and IL-1β in the islets was not T cells, but islet-infiltrating dendritic cells and macrophages. We have two hypotheses about the role of these cytokines in the beta cell destruction that occurs in human type 1 diabetes. First, these cytokines directly damage beta cells, as shown in studies in vitro, in which these two cytokines acted synergistically to damage beta cells and induce their apoptosis [5]. Second, these cytokines damage beta cells via beta cell–T cell interaction. It has been reported that IL-1β induces production of Fas on the beta cell surface [16], and islet-infiltrating T cells can destroy beta cells via the Fas–Fas ligand system [17]. In the present study, T cells were detected only in islets with beta cells, whereas TNF-α- or IL-1β-positive dendritic cells/macrophages were detected both in islets with and without beta cells, suggesting these cytokines could damage beta cells without the interaction of T cells and beta cells. The higher prevalence of TNF-α- or IL-1β-positive dendritic cells/macrophages in the islets without T cell infiltration than in those with infiltration also supports the first hypothesis.

T cells in this study were mainly observed in the islets with beta cells, while dendritic cells and macrophages were found in both islets with or without beta cells, suggesting that beta cells are mainly destroyed by T cells in the development of human type 1 diabetes. Only a few infiltrating T cells were detected in the islets of NOD mice after the onset of overt diabetes [18]. The main target of these cells is thought to be insulin [19], so such findings would be reasonable, because the T cell target has already been destroyed.

This study also made it clear that CD11c-positive dendritic cells infiltrate islets of type 1 diabetic patients with insulitis. CD11c-positive cells are reported to have strong antigen-presenting activity in mice and humans [20]. In the NOD mouse, an animal model of autoimmune type 1 diabetes, CD11c-positive cells are found to infiltrate pancreatic islets from a very early stage until the late stage of insulitis [21]. Of course, our biopsy specimens were taken after the onset of overt diabetes, which means that we detected the CD11c-positive dendritic cells in the late stage of insulitis in human type 1 diabetes. The reason why dendritic cells were still in the islets after beta cells had already been destroyed remains to be clarified.

In conclusion, the present study demonstrated that macrophages and dendritic cells infiltrate the pancreatic islets and produce inflammatory cytokines (TNF-α and IL-1β) during the development of type 1A diabetes.