Immune Cell Infiltration, Cytokine Expression, and β-Cell Apoptosis During the Development of Type 1 Diabetes in the Spontaneously Diabetic LEW.1AR1/Ztm-iddm Rat

LEW.1AR1/Ztm-iddm rats were housed in the Institute of Laboratory Animal Science of Hannover Medical School as described previously (4). The animals originated from a breeding colony (4) that was maintained through mating of diabetic female and diabetic male IDDM rats (F7; seventh generation of the IDDM rat; see http://www.mh-hannover.de/institute/clinbiochemistry/iddmrat.htm for details). Homozygous pairing with both parents diabetic resulted in a diabetes incidence of ∼60%. All rats were maintained according to the FELASA (Federation of European Laboratory Animal Science Associations) guidelines. The animals were investigated at days 45, 50, 55, 59, 60, or 65 after birth. On the day the animals were killed, tissue samples of pancreas and spleen were collected for morphologic examination. All analyses were performed on the same set of animals. Blood glucose concentrations were determined using capillary blood samples from the tail vein by the glucose oxidase method (Glucometer Elite; Bayer, Leverkusen, Germany). Serum insulin concentrations were determined by a rat insulin microplate enzyme-linked immunosorbent assay (Mercodia, Uppsala, Sweden).

Tissue specimens of pancreas and spleen were fixed in 4% paraformaldehyde, in 0.15 mol/l PBS, pH 7.3, or rapidly frozen in liquid nitrogen. Fixed tissue was embedded in paraffin, and frozen tissue was stored at −70°C for cryostat histology. Additionally, small pancreatic tissue samples were fixed in 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 mol/l cacodylate buffer, pH 7.3, postfixed in 1% OsO₄, and finally embedded in Epon for electron microscopical analysis (5).

Serial paraffin and cryostat sections were stained by the avidin-biotin-complex (ABC) method (6). The slides were incubated overnight with the specific antibodies, followed by a 30-min incubation with biotinylated goat anti-rabbit IgG (1:100) or biotinylated rabbit anti-mouse IgG (1:100) (Santa Cruz Biotechnology, Santa Cruz, CA). The slides were then incubated with a mixture of streptavidin (1:100) and biotin-peroxidase (1:1,000) (Jackson Immuno Research, West Grove, IL). The peroxidase reaction was visualized using 0.7 mmol/l diaminobenzidine and 0.002% hydrogen peroxidase in PBS, pH 7.3. The following primary antibodies were used: insulin (1:500) (polyclonal A565; DAKO, Hamburg, Germany), infiltrating macrophages (ED1), tissue macrophages (ED2), dendritic cells (Ox 62), B-cells (IgD), pan–T-cells (R 73), CD4⁺ T-cells, CD8⁺ T-cells (β-chain), NK cells (CD 161), major histocompatibility complex class II (Ox 6), interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ) (1:500) (all monoclonal; from Serotec, Oxford, U.K.), inducible nitric oxide synthase (iNOS) (1:50) (polyclonal; Santa Cruz Biotechnology), and activated caspase 3 (1:200) (polyclonal; Biosciences Pharmingen, Brussels, Belgium). All antibodies were certified for immunohistochemistry and showed specific immunostaining in tissue sections of the IDDM rat.

Pancreatic sections were fixed on 3-Chamber SuperFrost Plus slides on a heating block at 100°C for 2 min. Subsequently, the slides were treated with proteinase K (20 μg/ml) for 20 min at 37°C. Proteinase K was inactivated thereafter at 95°C for 2 min followed by an overnight incubation with a RNase-free DNase solution (1 unit/ml) at 37°C. The in situ RT-PCR procedure was a modification of different protocols (7,8). After rinsing with PCR-grade water and air drying, reverse transcription was performed in a buffer containing a mixture of all nucleotides (1 mmol/l dATP, dGTP, dCTP, and dTTP) (Genecraft, Lüdinghausen, Germany), oligo-dT primer (Invitrogen, Paisley, U.K.), M-MLV (Moloney murine leukemia virus) reverse transcriptase (0.5 units/μl) (Invitrogen, Paisley, U.K.), RNasin (Promega, Madison, Wisconcin), and ddH₂O. Slides were incubated at 37°C for 1 h in a moist chamber, and thereafter reverse transcriptase was inactivated at 92°C for 2 min. PCR amplification was performed in a buffer containing a nucleotide mix (10 μmol/l digoxigenin 11-dUTP, 190 μmol/l dTTP, and 200 μmol/l each of dATP, dCTP, and dGTP), the specific forward and reverse primer (1.25 μmol/l) and Taq biotherm polymerase (0.1 unit/μl) (Genecraft), self-seal reagent (MJ Research, Waltham, MA), and ddH₂O. Fifteen microliters of the PCR mix was applied to each chamber of the slide and sealed by cover slips. The slides were then placed in a thermal cycler suitable for in situ amplifications (MJ Research). PCR amplification was performed according to the following protocol: initial denaturation at 95°C for 3 min, followed by 35–40 cycles with a denaturation at 95°C for 45 s, annealing at 57°C for 45 s, and extension at 72°C for 45 s. A final extension was performed for 10 min at 72°C. After removal of the cover slips, the slides were incubated with a blocking reagent for 1 h. Thereafter, the incorporated digoxigenin-labeled nucleotides were detected by an anti-digoxigenin antibody labeled with alkaline phosphatase (1:500) or rhodamine (1:300) at room temperature for 1 h. Alkaline phosphatase activity was detected by the NBT/BCIP (nitro blue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate toluidine salt) color reaction (overnight incubation at room temperature). The primer sequences were rat IL-1β 5′-GCATCCAGCTTCAAATCTC-3′ (forward) and 5′-GGTGCTGATGTACCAGTTG-3′ (reverse), rat TNF-α 5′-CTACTGAACTTCGGGGTGATCGGTC-3′ (forward) and 5′-CTGGTATGAAGTGGCAAATCGGCT-3′ (reverse), rat IFN-γ 5′-CATTGAA-AGCCTAGAAAGTCTG-3′ (forward) and 5′-CGACTCCTTTTCCGCTTC-3′ (reverse), rat iNOS 5′-CTTCAACACCAGGTTGT-CTGCAT-3′ (forward) and 5′-ATGTCATGAGCAAAGGCACACAGAAC-3′ (reverse), and rat procaspase 3 5′-TGGTACCGATGTCGATGCAG-3′ (forward) and 5′-TACCGCA-GTCCAGCTCTGTACC-3′ (reverse).

The cellular composition of the different subsets of infiltrating immune cells was analyzed immunohistochemically using specific antibodies on consecutive cryostat sections. Ten to 15 pancreatic islets were analyzed in each pancreas at the different time points during diabetes development (45–65 days of age). The migration pattern of mononuclear immune cells within the different stages of islet infiltration and the changes in the ratio of ED1⁺ macrophages to CD8⁺ T-cells were analyzed by series of 100 consecutive paraffin sections. Pancreatic tissue was scored without knowledge of the glycemic status of the animals. Islet infiltration was graded on a score of 0–4 as follows: stage 0, no immune cell infiltration of the islets (control); stage 1, immune cell infiltration restricted to the periphery of the islets; stage 2, weak invasive immune cell infiltration of the islets; stage 3, large numbers of immune cells within the islets; and stage 4, complete β-cell loss without infiltration (termed “end-stage islets”). A mean insulitis score was calculated from 20 to 30 islets per individual pancreas. The stages of islet infiltration were ultrastructurally confirmed by electron microscopy. The pancreatic β-cell area of the islets was identified by immunostaining for insulin and morphometric quantification by a computer-assisted system as described previously (9). Apoptotic cells were identified by ultrastructural analysis and additionally quantified after TdT-mediated dUTP-biotin nick-end labeling (TUNEL) staining (10) using an in situ cell-death detection kit (Roche, Mannheim, Germany). The criteria for β-cell apoptosis were a positive staining for insulin and TUNEL-positive nuclei. The immunostained sections were viewed using bright field illumination with a Zeiss Photomicroscope II (Zeiss, Oberkochen, Germany).

Extensive morphologic examinations of pancreata from IDDM rats at all ages up to 45 days revealed no signs of infiltration (data not shown). Between days 50 and 60, diabetes incidence increased from 5 to 60% with a particularly steep increase from 21 to 60% between day 59 and 60 (Table 1). At day 50, 20% of the pancreases from normoglycemic animals already showed signs of a faint infiltration of the islets with an insulitis score of 1.1 (Tables 1 and 2). Normoglycemic animals with pancreas infiltration could be observed until day 59, when 29% of the pancreases were infiltrated with an average insulitis score of 2.4 (Tables 1 and 2). The insulitis score was always higher in the diabetic animals than in the normoglycemic counterparts, with pancreas infiltration at defined time points between day 50 and day 59 (Table 1). Normoglycemic IDDM rats with pancreas infiltration showed a gradual loss of β-cell area approaching 70% at day 59 (Table 2), whereas a diabetic state was usually observed after β-cell loss of >70% (Table 3). Thus, the critical loss of the β-cell mass for onset of overt diabetes in IDDM rats was in the range of 50–70%. In diabetic animals (blood glucose >7 mmol/l) blood glucose concentrations increased from 10.9 mmol/l (day 50) to 19.5 mmol/l (day 60), whereas serum insulin concentrations decreased from 0.8 ng/ml (day 50) to 0.1 ng/ml (day 60) (Table 1).

At day 50, upon immunohistochemical analysis, ED1⁺ macrophages, which stayed in the range of 60–80 cells/mm² until day 59 (Table 2), were the predominant immune cell subpopulation within infiltrated islets in normoglycemic rats. A comparable ED1 macrophage infiltration in the range of 70–90 cells/mm² could also be observed in diabetic animals (Table 3). CD4⁺ T-cell islet infiltration ranged between 15 and 20 cells/mm² at day 50 and doubled until day 59 in normoglycemic rats with pancreas infiltration (Table 2) as well as in diabetic rats (Table 3). NK cells and B-cells could be detected in islets of normoglycemic rats with pancreas infiltration as well as in diabetic rats between days 50 and 59 at numbers of <20 and <10 cells/mm², respectively (Tables 2 and 3). CD8⁺ T-cell islet infiltration dramatically increased by a factor of five from days 50 to 59 in normoglycemic rats with pancreas infiltration, approaching a value of nearly 150 cells/mm² (Table 2) in parallel to the increase of the insulitis score (Table 1) and the decrease of β-cell area (Table 2). Diabetic animals manifesting diabetes between ages 50 and 60 days showed constantly high numbers (160–200 CD8⁺ T-cells per mm²) in the infiltrated islets (Table 3). The decrease of the β-cell area in normoglycemic rats with pancreas infiltration (Table 2) as well as in diabetic animals (Table 3) showed an excellent correlation (r² = 0.998) with the insulitis score (Table 1). Also, the CD8⁺ T-cell infiltration showed a high correlation (r² = 0.996) with the different stages, indicating that this immune cell subpopulation contributes preeminently to the infiltration in the late stages of β-cell destruction in the type 1 diabetes rat model (Table 4). Due to the sharp increase in the number of CD8⁺ T-cells, the relative percentage of ED1⁺ macrophages in the total number of immune cells decreased to 30% in stage 3 (Table 4). CD4⁺ T-cell infiltration was doubled when the insulitis score increased from stage 1 to stage 3, but the relative percentage always remained in the range of 10% of the total immune cell population (Table 4). At the early islet infiltration stage 1, the total number as well as the percentage of the CD4⁺ T-cells was identical to that of the CD8⁺ T-cells (Table 4). NK cells and B-cells represented only a minor fraction (<10%) of the infiltrating immune cells in islet stages 1–3 (Table 4).

At day 45 of life, all animals were still normoglycemic (Table 1). None of the pancreatic islets and the surrounding pancreas parenchyma of these diabetes-prone IDDM rats showed any signs of insulitis or peri-insulitis (Fig. 1A). Occasionally one to three immune cells, mostly ED1⁺ macrophages, were observed in the sections of the noninfiltrated islets corresponding to stage 0 in the classification of the islet infiltration stages (Fig. 2A and C and Table 4). There was also no accumulation of ED1⁺ infiltrating and ED2⁺ tissue macrophages as well as of major histocompatibility class II -and Ox 62–positive dendritic cells in the connective tissue around the vessels (data not shown). The pancreatic β-cells within the typical mantle islets were well granulated and showed no signs of apoptosis (Fig. 2B). None of the endocrine cell types of the islets as well as none of the different cell types of the exocrine parenchyma showed mRNA expression of the proinflammatory cytokines IL-1β (Fig. 3A) or TNF-α (Fig. 3D) in the cytoplasm. iNOS mRNA expression (Fig. 3G) was also not detectable in pancreatic β-cells, but a few islet cells showed a faint expression of procaspase 3 mRNA (Fig. 4A), confirmed by immunohistochemistry on the protein level (not shown).

At day 50, immune cells could be observed migrating from the vessels to the islets (Fig. 1B). The immune cells surrounded the duct structures in the exocrine parenchyma and the pancreatic islets, forming an incomplete rim, and had already started to migrate into the islet periphery, thus representing islet infiltration stage 1 (Fig. 1B). The center of the islets was not infiltrated by immune cells (Fig. 2D–F). At this early infiltration stage, representing an insulitis score of 1.2, the immune cells were composed mainly of ED1⁺ macrophages (Fig. 2D) and some T-cells, with an equal amount of CD4⁺ and CD8⁺ T-cells (Fig. 2F) in normoglycemic rats with pancreas infiltration and a higher amount of CD8⁺ T-cells in the diabetic animals (Tables 2 and 3). Some apoptotic β-cells could also be detected in the islets, with the typical signs of cell rounding and dense insulin staining (Fig. 2E). Quantitative analysis of TUNEL-positive β-cells revealed an increase of the apoptosis rate from 0.2% in healthy IDDM rats to 1.7% in normoglycemic IDDM rats, with pancreas infiltration at the age of 50 days (Figure 5). In parallel to the infiltration of the islet periphery, mRNA expression of IL-1β (Fig. 3B) and TNF-α (Fig. 3E) could be detected in the immune cells.

Notably, IFN-γ mRNA expression in the immune cell population could not be observed at this early stage of pancreas infiltration and also not in any of the later stages (not shown). ED1⁺ macrophages in the spleen of these animals, however, stained positively for IFN-γ transcripts (not shown). At variance from the IL-1β–and TNF-α–positive immune cells, in this early infiltration stage with a predominance of macrophages, none of the β-cells in the infiltrated islets expressed mRNA of one of the three cytokines (Fig. 3B and E). Interestingly, the immune cells of the infiltrate in the islet periphery as well as 38 ± 2% of the β-cells in the center of the islets showed a strong iNOS mRNA expression (Fig. 3H). None of the immune cells, but 12 ± 2% of the β-cells, both single cells and groups of 3–5, in the infiltrated islets revealed an increased expression of procaspase 3 mRNA (Fig. 4B), which correlated well with the appearance of activated caspase 3 protein in apoptotic TUNEL-positive β-cells (not shown).

At day 55, the migration of immune cells was progressed toward the center of the pancreatic islets in both normoglycemic animals with pancreas infiltration and diabetic rats, resulting in islets with infiltration stage 2 (Fig. 1C) or stage 3 (Fig. 1D). The numbers of infiltrating immune cells within the islets were higher than at day 50 (Fig. 2D, F, G, and I and Tables 2 and 3), leading to a predominance of CD8⁺ T-cells (Fig. 2I). Severely infiltrated islets (stage 3) showed a decrease of the pancreatic β-cell area with a loss of the architecture of a typical mantle islet, resulting in small residual cell clusters composed of different endocrine cell types (Fig. 2H). In parallel to the severe infiltration of the islets, a massive increase of β-cell apoptosis with a maximal rate of 6.7% was recorded (Fig. 5). Notably, at day 55, infiltration of the islets did not occur at the same time as the increasing percentage of infiltration in stage 1 (20%), stage 2 (40%), and stage 3 (40%) of the affected pancreas. In the transition from stage 1 to stage 3, the number of immune cells, all expressing IL-1β and TNF-α mRNA, increased in parallel to the islet infiltration stages (Fig. 3C and F) without expression in the β-cells, as described in the early infiltration stage (Fig. 3C and F). In contrast to the early islet infiltration stage (stage 1), iNOS (Fig. 3I) and procaspase 3 mRNA expression were no more detectable in pancreatic β-cells in severely infiltrated islets.

Immunohistochemically, IL-1β and TNF-α were localized in both macrophages and T-cells in the islet periphery and center. However, when compared with the mRNA expression results obtained by in situ RT-PCR, fewer immune cells expressed these proinflammatory cytokine proteins immunohistochemically in the cytoplasm, probably due to high functional activity with a quick release of the synthesized cytokines (data not shown).

Day 59 marks the transition from a 21% to a 60% diabetes incidence at day 60 (Table 1). Thus, this time point was of particular interest for the documentation of immune cell infiltration, cytokine expression, and pancreatic β-cell destruction during diabetes development. At day 59, the pancreata of IDDM rats showed a severe infiltration with an insulitis score of 2.4 in still normoglycemic animals with pancreas infiltration and a score of 3.3 in diabetic rats characterized by a high rate of β-cell apoptosis, leading to the ultimate diabetes incidence of 60% at day 60 (Table 1). Pancreatic islet infiltration was dominated by CD8⁺ T-cells (Fig. 2I and Tables 2 and 3). The rate of β-cell apoptosis was slightly lower than at day 55, which can be explained by the progressive decrease of the β-cell area due to cell death (Fig. 5). In the severely infiltrated islets, IL-1β and TNF-α mRNA transcripts in the increasing immune cell subpopulations showed the highest expression level (Fig. 3C and F). After the total loss of β-cells in severely infiltrated islets at stage 3, none of the other endocrine cell types expressed mRNA of any of the three proinflammatory cytokines (Fig. 3C and F) or iNOS (Fig. 3I). Only the immune cells of the islet infiltrate showed a persistent mRNA expression of iNOS (Fig. 3I).

A total of 70% of the islets from diabetic animals showed a severe infiltration (stage 3) with a high pancreatic β-cell apoptosis rate (Fig. 5), finally leading to a complete loss of the β-cell area (Table 3). The composition of severely infiltrated islets (stage 3) at day 60 was not different from that at day 59 with a predominance of CD8⁺ T-cells, i.e., 60% of the infiltrating immune cells (Table 3). There was no iNOS mRNA expression in other endocrine cell types of these severely infiltrated pancreatic islets (Fig. 3I). In fact, 30% of the islets could be classified as end-stage islets (stage 4) with a complete loss of β-cells (Table 3) and virtually no remaining immune cells (Table 3) increasing to 100% at day 65. However, ED1⁺ macrophages remained in the exocrine pancreas parenchyma, in particular in the connective tissue around the vessels, with a clear expression of TNF-α mRNA of the proinflammatory cytokine IL-1β, and IFN-γ was not expressed in the remaining ED1⁺ macrophages.

The IDDM rat is a model of human type 1 diabetes very well suited for the elucidation of the chain of events leading to pancreatic β-cell loss and therefore of crucial importance for the understanding of the etiopathology of the disease process and for a rational design of possible intervention trials. The pre-diabetic period lasts only 1 week in the IDDM rat in contrast to a longer pre-diabetic period of up to 1 month reported for BB rats (11,12), up to 3 months for NOD mice (13), and even years in humans (14,15).

Macrophages could be detected ultrastructurally and by immunostaining as the first immune cell subpopulation invading the islet periphery, thus acting as initiators of the autoimmune process, as also observed in the BB rat (16–19) and in the NOD mouse (20–22). However, like in the human situation, but in contrast to these animal models, the IDDM rat showed no signs of peri-insulitis. ED1⁺ infiltrating macrophages and dendritic cells are accumulated in the periductular and perivascular connective tissue adjacent to noninfiltrated islets in a variable time window in the diabetes-prone BB rat between 10 and 30 days (12,16) and in the NOD mouse 30–50 days before starting to migrate into the islet periphery and initiating the islet infiltration (20,21). The gradual shift to the predominance of CD8⁺ T-cells in the later infiltration stages was accompanied by a significant decrease of the β-cell area. The proportion of CD4⁺ T-cells was maintained at a stable level of ∼10% of all infiltrating immune cells during the progression of islet infiltration. CD4⁺ T-cell infiltration apparently plays a crucial regulatory role in promoting the recruitment of CD8⁺ T-cells. CD4⁺ T-cells are critically important apparently also for disease progression in the NOD mouse (23,24). B-cells in small proportion could already be detected in the initial islet infiltration in the IDDM rat and remained low (<5%) during the progression to severe insulitis. B-cells may participate in an initial priming event, as also supported by studies in NOD mice (25), but are apparently not a prerequisite for disease progression. NK cells were highest in the second stage of islet infiltration in the IDDM rat. This coincides with the highest β-cell apoptosis rate. Thus, NK cells are apparently involved in the process of β-cell destruction in this animal model rather than playing an immunoregulatory protective role. This conclusion is also supported by studies in BB rat and NOD mice (26–34). The predominance of CD8⁺ T-cells in severely infiltrated islets of IDDM rats is an intriguing finding of the present study. It is the absolute increase of CD8⁺ T-cells that correlates well with the apoptotic β-cell destruction and a reduction of the β-cell area in infiltrated islets. The immune cell composition of severely infiltrated pancreatic islets in the IDDM rat corresponds with the late infiltration stages observed in diabetic BB rats and NOD mice (16,17,35–37). A predominance of CD8⁺ T-cells has not been reported in normal NOD mice but in a T-cell–receptor transgenic NOD strain showing a selective recruitment of CD8⁺ T-cells to the islets and the subsequent rapid β-cell destruction leading to an accelerated diabetes onset (38). Furthermore, NOD mice showed an increase of a lymphoblast subpopulation developing into CD8⁺ T-cells in the spleen of acutely diabetic animals (39) and an increased priming of autoreactive CD8⁺ T-cells in lymph nodes after streptozotocin-induced β-cell death (40). CD8⁺ T-cells were also dominant in islet infiltrates of human pancreas biopsies from patients with recent-onset type 1 diabetes (1,2). Our data thus demonstrate the crucial relationship between macrophages in the initial and CD8⁺ T-cells in the late phase of infiltration and β-cell destruction in the type 1 diabetes rat, a model of spontaneous type 1 diabetes without inherited developmental defects of the cellular immune system in contrast to BB rat and NOD mouse (41). This model is therefore very well suited for studies on the role of cytokines in the disease process. For this purpose we used in the present study for the first time a systemic approach using a sensitive in situ PCR technique for the analysis of the development of the cytokine gene expression pattern.

Proinflammatory cytokines play an important role as immune mediators inducing complex signaling pathways finally resulting in β-cell death by apoptosis or necrosis (42–44). This prompted us to correlate the expression of the β-cell toxic cytokines IL-1β, TNF-α, and IFN-γ in infiltrated islets of the IDDM rat with the status of immune cell invasion and β-cell destruction. By in situ RT-PCR it could be shown for the first time that IL-1β and TNF-α were expressed in vivo in immune cells at an early infiltration stage. In this IDDM rat model, we could further demonstrate that the expression of these cytokines increased in parallel to the number of the immune cells with the progression of islet infiltration. Cytokine-positive cells were always immune cells, never β-cells or any other endocrine cell type in the islet. In situ RT-PCR is a method that can, with high sensitivity, detect the gene expression of cytokines in infiltrated islets in the in vivo situation. The mRNA detection correlates with protein expression as shown in this study but is independent of the release of the cytokine protein from the immune cell and is therefore a very reliable indicator of the activity status of the immune cell. In the BB rat, IFN-γ transcripts could be detected by RT-PCR in infiltrated islets (45,46). In islets from diabetic NOD mice, the expression of IFN-γ and other Th1 cytokines has also been reported (47–50). On the other hand, in a transgenic NOD mouse lacking the IFN-γ receptor on the β-cells, diabetes incidence was not different from that of wild-type mice, suggesting that pancreatic β-cell survival is not crucially affected by a lack of IFN-γ (51). However, the lack of IFN-γ expression in the IDDM rat, as reported here, provides clear evidence for the fact that β-cell destruction in autoimmune diabetes is not critically dependent upon an activation of IFN-γ. A possible explanation for this apparent discrepancy to BB rat and NOD mouse studies could be the fulminant development of β-cell destruction in the IDDM rat without any signs of peri-insulitis. In this IDDM rat model, the classical signs of endothelial cell activation as a prerequisite for immune cell adhesion were absent in the pancreatic parenchyma. Under in vivo conditions IFN-γ plays an important role in mediating adhesion mechanisms by upregulation of molecules such as intracellular adhesion molecule-1 on endothelial or immune cells (51,52). IL-1β and TNF-α, as the dominant proinflammatory cytokines, apparently act directly cytotoxic to pancreatic β-cells or indirectly sensitize T-cell–mediated cytotoxicity by changing the surface protein expression on the β-cell as the target cell in the autoimmune process (46).

Our observation of apoptotic β-cell death is in agreement with biopsy studies (53) in BB rats where infiltrated islets showed an initial increase of β-cell apoptosis at day 68 followed by an accelerated β-cell apoptosis rate at day 85, which coincides with the onset of diabetes, even though the absolute rates of β-cell apoptosis were ∼10 times lower due to the slower disease progression in the BB rat. In the NOD mouse model with its long pre-diabetic period, rates of apoptosis during the time course of diabetes manifestation could be monitored only in transgenic strains with a modulation of the T-cell receptor (54,55) with macrophages expressing IL-1β on the protein level (56). The IDDM rat showed an excellent correlation between islet infiltration, β-cell apoptosis, and onset of diabetes, demonstrating that islet infiltration by ED1⁺ macrophages and subsequently by CD8⁺ T-cells is the initial step of a process finally causing β-cell apoptosis. In particular, the parallel induction of iNOS and procaspase 3 in β-cells with the infiltration of immune cells secreting proinflammatory cytokines is strong evidence for a crucial role of proinflammatory cytokines in the process of apoptotic β-cell destruction in vivo, which is in agreement with in vitro data (57,58).

For the first time it was possible in this study to document under in vivo conditions a strong correlation between immune cell infiltration and the gene expression of proinflammatory cytokines, which leads to an upregulation of iNOS and procaspase 3 gene expression in the pancreatic β-cells. The particular, features of the IDDM rat make it possible to depict a chain of events that provides a mechanistic explanation for the development of the diabetic syndrome in this animal model of human type 1 diabetes.

This work has been supported by a grant from the Deutsche Forschungsgemeinschaft (Jo 395/1-1/2) to A.J. and by a National Institutes of Health grant (1R21AI55464-01) to S.L. and H.-J.H. Some of the results were obtained during thesis work by A.G.

The technical assistance of D. Lischke and U. Sommerfeld is gratefully acknowledged.