Islet-Specific Expression of IL-10 Promotes Diabetes in Nonobese Diabetic Mice Independent of Fas, Perforin, TNF Receptor-1, and TNF Receptor-2 Molecules1

Several death-signaling or death-inducing molecules have been implicated in β cell destruction, including Fas, perforin, and TNFR-1. In this study, we examined the role of each death-signaling molecule in the IL-10-accelerated diabetes of nonobese diabetic (NOD) mice. Groups of IL-10-NOD mice, each deficient in either Fas, perforin, or TNFR-1 molecules, readily developed insulitis, and subsequently succumbed to diabetes with an accelerated kinetics and incidence similar to that observed in their wild-type or heterozygous IL-10-NOD littermates. Similarly, a TNFR-2 deficiency did not block accelerated diabetes in IL-10-NOD mice and spontaneous diabetes in NOD mice. These results demonstrate that pancreatic IL-10 promotes diabetes independent of Fas, perforin, TNFR-1, and TNFR-2 molecules. Subsequently, when cyclophosphamide, a diabetes-inducing agent, was injected into insulitis-free NOD.lpr/lpr mice, none of these mice developed insulitis or diabetes. Our data suggest that cyclophosphamide- but not IL-10-induced diabetes is Fas dependent. Overall, these findings provide evidence that pancreatic expression of IL-10 promotes diabetes independent of the major death pathways and provide impetus for identification of novel death pathways precipitating autoimmune destruction of insulin-producing β cells.

I nsulin-dependent diabetes mellitus (IDDM) 4 is an autoimmune disease caused by the T cell-mediated destruction of insulin-producing ␤ cells in the pancreatic islets of Langerhans (1,2). The nonobese diabetic (NOD) mouse spontaneously develops IDDM and has been used as an animal model of human IDDM. In the NOD mouse, insulitis begins at 3-4 wk of age and diabetes by 14 wk of age (3).
Cytokines produced by the mononuclear cell infiltrate (T cells and APC) itself are clearly involved in the propagation of insulitis. Treatment of young NOD mice with anti-IL-10 mAb prevented the development of insulitis (4). Elsewhere, the expression of IL-10 in pancreatic ␤ cells correlated with the insulitis of NOD mice (5). BALB/c mice expressing the IL-10 transgene (tg) in their insulinproducing ␤ cells (IL-10-BALB/c mice) of the pancreas did not develop diabetes, but their offspring (IL-10-NOD mice) from backcrosses (N2-N3) to NOD mice became diabetic at an accel-erated rate (6). Similarly, NOD mice expressing the IL-10 transgene in glucagon-producing ␣ cells of the pancreas developed diabetes at an accelerated rate (5). These cumulative findings implied that IL-10 is an immunostimulatory factor in the IDDM of NOD mice. Our recent studies have demonstrated that the promotion of diabetes by IL-10 in NOD mice requires the participation of an autoreactive T cell repertoire (7). Depending upon the circumstances, pancreatic IL-10 promoted autoimmune diabetes via CD4 (8) or CD8 T cells (7).
Several studies have shown that a homozygous deficiency of the Fas, perforin, or TNFR-1 molecules dramatically affects the incidence of spontaneous insulitis and diabetes in NOD mice. Fasdeficient NOD.lpr/lpr mice were free from insulitis and diabetes (9,10), and a perforin deficiency reduced the incidence of diabetes by a dramatic 80 -85%, despite the mild insulitis in pancreata (11) of these mice, as confirmed by others (12). Additionally, TNFR-1-deficient NOD mice exhibited mild insulitis, but were completely resistant to spontaneous diabetes (13). Consequently, each of these pathways participates in either the initiation and/or effector phases of autoimmune diabetes.
In this study, we examined the roles of these death-signaling molecules (Fas, perforin, and TNFR-1), and of TNFR-2 as well, in the IL-10-accelerated autoimmune diabetes of NOD mice. The Fas-Fas ligand (FasL) pathway was also explored in cyclophosphamide (CYP)-induced diabetes of NOD mice. Determining how these molecules function, if at all, in the inception and acceleration of autoimmune diabetes is important to understanding how inflammatory stimuli such as cytokines in the target organ predispose mice to autoimmune diabetes and to devising appropriate therapeutic interventions.
Our results demonstrate that IL-10 promoted diabetes in NOD mice independent of the Fas, perforin, TNFR-1, and TNFR-2 molecules. A homozygous deficiency at perforin gene locus in IL-10-NOD mice slightly delayed the onset, but did not decrease the incidence of their diabetes. Although IL-10-NOD.lpr/lpr mice readily developed diabetes, CYP-injected NOD.lpr/lpr mice did not. This outcome suggests that IL-10-accelerated diabetes is Fas independent, whereas CYP-induced diabetes is Fas dependent. Finally, current findings may provide impetus for the delineation of additional death pathways, under the influence of cytokine-induced inflammation, precipitating autoimmune destruction of insulinproducing ␤ cells.

Materials and Methods
Mice NOD/shi mice were part of the rodent breeding colony at The Scripps Research Institute (La Jolla, CA). IL-10-BALB/c mice expressing the IL-10 transgene in their islets under control of the human insulin promoter were backcrossed to NOD/shi mice for 10 -11 generations to produce IL-10-NOD mice (7). The presence of the transgene was verified by PCR. Mice were housed under specific pathogen-free conditions.

Generation of Fas-deficient NOD.lpr/lpr mice
Initially, we backcrossed B6.MRL.lpr/lpr mice to NOD mice for two generations. The heterozygous offspring were intercrossed to get NOD.lpr/lpr mice. IL-10-NOD mice were then backcrossed to these mice to generate N3 to N4 mice of appropriate combinations.
Subsequently, to generate a later generation of IL-10-NOD.lpr/lpr mice, IL-10-NOD mice were backcrossed to NOD.wt/lpr or NOD.lpr/lpr mice of an N9 backcross generation. The Fas mutation (lpr) was verified in tail DNA by using two pairs of primer sets (10). The first pair was composed of NIL-1, 5Ј-CAG CAG GAA TCC TAT GAG GT-3Ј and NIL-2, 5Ј-CTC GCA ACG TGA ACG GTT CG-3Ј, yielding a band of 381 bp for the mutated allele. The second pair was composed of NIL-1, 5Ј-CAG CAG GAA TCC TAT GAG GT-3Ј and NIL-4, 5Ј-GCA GAG ATG CTA AGC AGC AG-3Ј, yielding a band of 265 bp for the wild-type allele and a band of 5.7 kb for the mutated allele.

Generation of perforin-deficient NOD mice
Perforin-deficient BALB/c mice were backcrossed onto NOD mice for two to three generations, and the resulting heterozygous mice of generation N2 or N3 were then intercrossed to generate perforin-deficient NOD mice and, subsequently, perforin-deficient IL-10-NOD mice. The perforin genotype was determined with PCR using three primers on DNA prepared from tail biopsies (perforin 12 primer, 5Ј-TGG CCT AGG GTT CAC ATC CAG-3Ј; perforin 17 primer, 5Ј-CGT GAG AGG TCA GCA TCC TTC-3Ј; perforin 26 primer, 5Ј-ATA TTG GCT GCA GGG TCG CTC-3Ј). The PCR yielded a 500-bp fragment for wild-type mice, a 350-bp fragment for KO mice, and 350-and 500-bp fragments for heterozygous mice.

MHC typing of mice
All the second backcross mice were tested by PCR for NOD, C57BL/6, and BALB/c MHC. The presence of I-A␤ d was determined on tail DNA by PCR using the following primer set: forward primer, 5Ј-GAT ACA TCT ACA ACC GGG AGG AG-3Ј, and reverse primer, 5Ј-CTG TTC CAG TAC TCG GCG TCT G-3Ј. PCR amplification yielded a103-bp product from BALB/c, but not NOD mice. The presence of I-E␣ d was tested in tail DNA by PCR using the following primer set: forward primer, 5Ј-ATG AGC TCC CAG AAG TCA TGG G-3Ј, and reverse primer, 5Ј-GGA GAG ACA GCA GCT CTC AGC-3Ј. PCR amplification yielded a 277-bp product from BALB/c, but not NOD mice. Mice were also tested for MHC class I molecules at K d with anti-K d mAb (clone SF1-1.1), K b with anti-K b mAb (clone AF6-88.5), D d with anti-D d mAb (clone 34-2-12), D b with anti-H-2D b mAb (clone 28-14-8), and class II I-A b molecule with anti-I-A b mAb (clone AF6-120.1) (PharMingen, La Jolla, CA) by flow cytometry.

Assessment of diabetes
Starting at 4 -5 wk of age, mice were tested for diabetes by weekly measurements of blood glucose (BG) levels using a one-step Bayer Glucometer Elite (Bayer Corporation, Elkhart, IN). Animals were considered diabetic when BG levels were Ͼ300 mg/dl. In most instances, the BG levels exceeded 500 mg/dl. Mice of both sexes were included in all the IL-10-NOD experiments, and female mice were employed for monitoring of diabetes in TNFR-1 KO and TNFR-2 KO mice.

Histological analysis
Lymphocytic infiltration of the islets was evaluated on hematoxylin and eosin (H&E)-stained paraffin sections of the pancreas. For insulin staining, paraffin-embedded sections of the pancreata were stained with an immunoperoxidase method using polyclonal Abs to porcine insulin, followed by a biotinylated secondary Ab and an avidin-biotin complex, as described earlier (16).
Analysis by H&E staining (Fig. 3D) revealed that the pancreatic islets from IL-10-NOD-scid/scid mice that were injected with NOD-perforin deficient (Ϫ/Ϫ) splenocytes were completely infiltrated with mononuclear cells, causing destruction of most of ␤ cells, leaving few in place. Conversely, pancreatic islets from NOD-scid/scid mice that were injected with NOD.perorin KO 5    (Ϫ/Ϫ) splenocytes exhibited only periinsulitis and contained intact insulin-producing ␤ cells. These findings further confirm the data of Fig. 3A that IL-10 promotes diabetes independent of perforin pathway.
Furthermore, the pancreatic islets from both groups of mice were extensively infiltrated with autoreactive lymphocytes (Fig.  5B), and their insulin-producing ␤ cells were destroyed (not shown). The islets from pancreata of 5-wk-old littermate NOD-TNFR-2-deficient (Ϫ/Ϫ) mice exhibited periinsulitis. These mice subsequently progressed to diabetes beginning at 18 wk of age and showed 60% incidence of diabetes by 24 wk of age (n ϭ 5). These findings imply that TNFR-2 signaling is irrelevant for spontaneous and accelerated diabetes of NOD and IL-10-NOD mice, respectively.

CYP fails to provoke autoimmune insulitis and diabetes in the absence of pancreatic inflammation
Finally, to test whether CYP would induce diabetes in a Fas-dependent manner, we injected CYP into Fas-deficient NOD.lpr/lpr FIGURE 6. A, CYP injection fails to induce diabetes in insulitis-free NOD.lpr/lpr mice. NOD.lpr/lpr (Ϫ/Ϫ) (n ϭ 10) and NOD.Fas/lpr (ϩ/Ϫ) (n ϭ 10) mice were injected i.p. with CYP (200 mg/kg body weight) on days 0 and 14, then monitored for diabetes at weekly intervals. Mice were considered diabetic when the BG values were Ͼ300 mg/dl. B, The insulitis index for CYP-treated NOD.lpr/lpr (n ϭ 10) and NOD.Fas/lpr (n ϭ 8) mice is shown. The values in the parentheses represent the number of islets scored for each category. C, Representative H&E-stained paraffin-embedded pancreatic sections from CYP-treated NOD.lpr/lpr and NOD.Fas/lpr mice (ϫ400).
Although approximately 80% of the islets from CYP-injected heterozygous (Fas/lpr) mice (n ϭ 10) showed severe insulitis, none of the islets from CYP-treated NOD.lpr/lpr mice (n ϭ 8) had any sign of lymphocytic infiltration in or near the islets (Fig. 6B). Occasionally, a perivascular infiltrate occupied the pancreatic tissue of CYP-treated NOD.lpr/lpr mice, as pictured in representative H&E-stained sections from NOD.Fas/lpr (ϩ/Ϫ) and NOD.lpr/lpr (Ϫ/Ϫ) (Fig. 6C). Since the pancreatic islets of CYP-treated NOD.lpr/lpr mice were free from lymphocytic infiltration and stained positively for insulin, yet those of CYP-treated heterozygous NOD.Fas/lpr mice were filled with mononuclear cells and stained only weakly for insulin-positive cells, Fas expression is a prerequisite for CYP-induced diabetes in NOD mice.

Discussion
The results from this study demonstrate that the expression of IL-10 in pancreatic islets bypasses the requirement for Fas, perforin, TNFR-1, and TNFR-2 molecules and that IL-10 can otherwise precipitate the diabetic process. In contrast to IL-10-accelerated diabetes in NOD.lpr/lpr mice, CYP fails to provoke diabetes in NOD.lpr/lpr mice. Therefore, our findings demonstrate that IL-10-accelerated diabetes is Fas independent and that CYP-induced diabetes is Fas dependent. This current study also demonstrates that for the first time, in contrast to the role of TNFR-1 in diabetes of the NOD mouse, deficiency of TNFR-2 failed to block spontaneous diabetes of NOD mice.
Destruction of ␤ cells in the spontaneous diabetes of NOD mice required Fas-FasL interaction. The Fas-FasL pathway appeared to be required for the initiation and/or effector phases of spontaneous autoimmune diabetes in former experiments with NOD mice (9,10,19), and with TCR tg NOD mice expressing islet-specific T cells (12,20). The current study with CYP-induced diabetes of NOD mice further highlights a role for Fas-FasL pathway in destruction of ␤ cells. However, islet transplantation experiments provided differing results. That is, the Fas-FasL pathway did not participate in the effector stages of diabetes, since NOD.lpr/lpr islets transplanted into recently diabetic NOD mice were completely destroyed by an autoimmune attack (21,22) or following CYP injection (22). Apparently different mechanisms participate in the destruction of ectopically transplanted islet grafts and of ␤ cells in situ. Our current findings demonstrated that expression of the IL-10 transgene in the pancreatic islets promoted accelerated diabetes of NOD mice in situ without a requirement for the Fas signaling. Additionally, we showed that splenocytes from diabetic IL-10-NOD.lpr/lpr mice transferred disease into NOD.scid/scid mice, reinforcing the implication that the Fas-FasL pathway is not required throughout this autoimmune process of IL-10-NOD mice.
If, as seems evident, the Fas-FasL pathway does not participate in this accelerated diabetes, presumably expression of the IL-10 transgene in the islets of NOD mice could awaken other death pathways such as those that use perforin or TNFR-1 or TNFR-2 molecules. However, our findings exclude that possibility. Therefore, the results described in this work contrast with earlier conclusions that perforin is required for spontaneous diabetes (11,12) and CYP-induced diabetes (11). TNFR-1 molecules were also considered a requirement for the spontaneous and CYP-induced diabetes of NOD mice (13). Since TNFR-2-deficient NOD mice developed spontaneous diabetes, we did not study the effect of TNFR-2 deficiency on CYP-induced diabetes. In addition, our current findings demonstrate that Fas is also required for CYP-induced diabetes of NOD mice. However, the actual cause, for the absence of disease, may have been the lack of intense inflammation in the pancreatic environment that is necessary for efficient APC activation, Ag presentation, and T cell activation, leading to the production of inflammatory mediators.
For example, we have shown that CD40-CD40L pathway is essential for the spontaneous autoimmune insulitis and diabetes of NOD mice, as demonstrated by Ab-blocking studies. This pathway appears to play a role in the initiation but not the effector phase of this disease process (16). The requirement for this pathway in spontaneous diabetes was confirmed by Green and coworkers (23) using CD40L-deficient NOD mice. However, this pathway was found dispensable for the accelerated diabetes of tg IL-10-NOD (7) and TNF-␣-NOD (23) mice. These findings and the results presented in the current study suggest that cytokine-induced inflammation in the pancreatic environment circumvented the requirement for the well-established costimulation pathways, thereby short-circuiting the onset of disease. This hypothesis is further supported by two additional observations: 1) Expression of the IL-10 transgene in diabetes-free BDC2.5 NOD mice leads to the development of diabetes. 2) CYP injection fails to provoke diabetes in insulitis-free NOD.lpr/lpr mice. We are of the opinion that the failure to observe diabetes in CYP-treated NOD.lpr/lpr mice is unrelated to the lpr-induced lymphoproliferative effect, because IL-10-NOD.lpr/lpr mice do develop diabetes, and their splenocytes can transfer diabetes into NOD.scid/scid mice with delayed kinetics.
It is well established that IL-10 promotes pathogenic cell-mediated and humoral autoimmunity. In fact, tg IL-10-C57BL/6 mice expressing IL-10 under control of the salivary amylase promoter developed a Sjögren's-like syndrome via a Fas-FasL pathway (24). Depending on the circumstances, then, IL-10 could exhibit its autoimmunostimulatory effect via Fas-dependent and Fas-independent pathways. IL-10 also seems to act as an immunostimulator in humoral autoimmunity through B cells (25). However, our previous findings demonstrate that, in T cell-mediated autoimmunity, IL-10 readily promotes autoimmune diabetes independent of B cells, because B cell-deficient IL-10-NOD mice also developed accelerated diabetes similar to that in their wild-type counterparts (7). Continual administration of IL-10 to NZB/W F 1 mice accelerated autoimmunity, whereas treatment with anti-IL-10 Ab delayed its onset (26). Apart from its pathogenic role in autoimmunity, IL-10 further exerted its immunostimulatory capacity with respect to tumor suppression, since IL-10 expressed under a class II promoter of C57BL/6 mice limited the growth of immunogenic tumors (27).
From the foregoing results, we conclude that accelerated diabetes in IL-10-NOD mice does not involve the classical death signaling molecules, Fas, perforin, and TNFR-1. Nor is the TNFR-2 molecule required for diabetes to arise in IL-10-NOD mice. Therefore, pancreatic IL-10 may promote diabetes via unique death pathway(s) involving TRAIL, TWEAK, and LIGHT molecules (28 -30), or its expression may promote diabetes via compensatory death pathways. To prevent such disease, these novel death pathways used by autoimmune responses to destroy insulin-producing ␤ cells must be uncovered along with molecules that interfere with their destruction.