Nonobese diabetic (NOD) mice transgenic for Fas ligand (FasL) on islet β cells (HIPFasL mice) exhibit an accelerated diabetes distinct from the normal autoimmune diabetes of NOD mice. This study was undertaken to define the mechanism underlying accelerated diabetes development in HIPFasL mice. It was found that diabetes in HIPFasL mice is dependent on the NOD genetic background, as HIPFasL does not cause diabetes when crossed into other mice strains and is lymphocyte dependent, as it does not develop in HIPFasLSCID mice. Diabetes development in NODSCID recipients of diabetic HIPFasL splenocytes is slower than when using splenocytes from diabetic NOD mice. β cells from HIPFasL mice are more susceptible to cytokine-induced apoptosis than wild-type NOD β cells, and this can be blocked with anti-FasL Ab. HIPFasL islets are more rapidly destroyed than wild-type islets when transplanted into nondiabetic NOD mice. This confirms that FasL+ islets do not obtain immune privilege, and instead NOD β cells constitutively expressing FasL are more susceptible to apoptosis induced by Fas-FasL interaction. These findings are consistent with the accelerated diabetes of young HIPFasL mice being a different disease process from the autoimmune diabetes of wild-type NOD mice. The data support a mechanism by which cytokines produced by the insulitis lesion mediate up-regulation of β cell Fas expression, resulting in suicide or fratricide of HIPFasL β cells that overexpress FasL.

Fas ligand (FasL),3 a type 2 membrane protein belonging to the TNF family, plays an important role in the induction of cell death. Ligation of Fas receptors by FasL results in apoptosis of Fas-expressing cells. Although β cells in type 1 diabetes die via the process of apoptosis (1), there remains considerable argument regarding the role of Fas in this β cell death. For β cell death to be mediated by the Fas pathway requires, firstly, that β cells are capable of expressing Fas and, secondly, that there is a source of FasL in the islets (either β cells express FasL or other islet cell populations express FasL, such as α-cells or islet-infiltrating lymphoid cells).

In answer to the first question, Fas has been shown by immunohistochemistry to be expressed by both murine and human β cells. Fas is expressed by β cells from old nonobese diabetic (NOD) mice and also by cytokine-stimulated β cells from young NOD mice (2, 3, 4). Fas is also expressed on β cells from human subjects with newly diagnosed type 1 diabetes (5, 6). Fas was shown to be up-regulated by normal human β cells in response to cytokine stimulation (7).

Pancreatic α-cells constitutively express FasL, and this could, therefore, be a source for FasL to ligate Fas on β cells (8). Human islets constitutively express FasL and undergo apoptosis in response to cytokine stimulation or incubation in high glucose concentrations (9), both of which up-regulate β cell Fas expression. Although there have been reported difficulties in showing FasL expression on β cells by flow cytometry (10), this may be due to methodological problems, such as destruction of FasL epitopes during β cell isolation and purification. Paradoxically, NOD mice, despite being prone to autoimmune diabetes, express lower, rather than higher, levels of FasL than nondiabetes-prone BALB/c mice (11). This is due to a mutation in the NOD FasL gene (11).

Consistent with a role for Fas in autoimmune diabetes, Fas-deficient NOD-lpr/lpr mice are resistant to diabetes development after transfer of diabetogenic T cells (12). It is also conceivable, however, that diabetes protection in this model is due to constitutive FasL expression by NOD-lpr/lpr T cells that are thereby rendered nonpathogenic (13). NOD-lpr/lprSCID mice have delayed onset and reduced incidence of diabetes after adoptive transfer of diabetogenic NOD spleen cells (14). NOD mice heterozygous for the FasL mutation gld, which have reduced functional FasL expression on T cells, but no lymphadenopathy, do not develop diabetes (14). Destruction of transplanted syngeneic islets is markedly delayed in NOD mice treated with anti-FasL Ab (15). In a transgenic model of autoimmune diabetes, CD8 T cells killed β cells in a Fas-dependent and perforin-independent manner (16). Similarly, transgenic CD4 T cells specifically killed cytokine-treated β cells expressing Fas and not untreated β cells or cytokine-treated β cells from Fas-deficient NOD-lpr/lpr mice (16). Hence there is considerable evidence that Fas plays an important role in β cell destruction. Evidence against a role for Fas in autoimmune β cell destruction is less impressive. Fas-deficient neonatal pancreas is still destroyed when transplanted into diabetic NOD mice (17). Although one group reported that anti-FasL neutralizing Ab did not prevent diabetes induction in the cyclophosphamide model of autoimmune diabetes (17), a recent report showed that anti-FasL Ab did, in fact, prevent diabetes induction in the cyclophosphamide model, but at the afferent, rather than efferent, level (18).

Chervonsky et al. (2) and ourselves (19) have independently shown that NOD mice that express a transgene for FasL on islet β cells (HIPFasL mice) develop an accelerated form of diabetes. These mice provide a unique insight into the possible role of the Fas pathway in immune β cell destruction. The following studies were undertaken to determine the mechanism for accelerated diabetes development in HIPFasL mice.

All mice, including the C57BL/6 and (B6×DBA/2)F1 mice used for backcrossing, were obtained from the Animal Breeding Establishment of the Australian National University (Canberra, Australia). All mice were housed and maintained under specific pathogen-free conditions in microisolator cages and given food and water ad libitum. Diabetes was defined as blood glucose levels >12 mmol/liter on at least three consecutive blood readings (Advantage II glucometer; Roche, Indianapolis, IN).

The cDNA for murine FasL (Immunex, Seattle, WA) was ligated to the human insulin promoter (Genentech, South San Francisco, CA) and a HepBsAg 3′ sequence to enable FasL expression on islet β cells. The resulting 4-kb construct was injected into NOD mice oocytes to produce HIPFasL founders.

All mice were screened at 30 days of age for the presence of the FasL transgene by PCR. Primers in the insulin promoter and FasL DNA produced a PCR fragment of 1.4 kb.

The presence of FasL expression was also confirmed using immunostaining. After isolation, single-cell suspensions of islets from HIPFasL mice and control mice were left in culture medium for 4 h to recover from trypsin treatment. Cells were then incubated with hamster anti-mouse FasL-biotin mAb (BD PharMingen, San Diego, CA) and matrix metalloproteinase inhibitor (KB8301; BD PharMingen) for 2 h before being washed and incubated with streptavidin-PE (BD PharMingen) for 1 h. They were then analyzed on a FACScan cytometer (Fig. 1 a).

FIGURE 1.

β cell FasL expression. a, Single-cell suspensions of islets from HIPFasLSCID and NODSCID mice stained with anti mouse FasL. b, FasL staining of HIPFasL pancreas. Small arrows highlight β cells staining positively for FasL, the medium arrow highlights the peri-islet mononuclear infiltrate, which is negative for FasL, and large arrows show the islet border.

FIGURE 1.

β cell FasL expression. a, Single-cell suspensions of islets from HIPFasLSCID and NODSCID mice stained with anti mouse FasL. b, FasL staining of HIPFasL pancreas. Small arrows highlight β cells staining positively for FasL, the medium arrow highlights the peri-islet mononuclear infiltrate, which is negative for FasL, and large arrows show the islet border.

Close modal

Immunostaining was also performed on frozen sections of pancreata of HIPFasL mice. Mice were sacrificed by cervical dislocation, the pancreas was collected, and fresh tissue was embedded in OCT compound (TissueTek; Miles, Elkhart, IN) and snap-frozen using liquid nitrogen. Before staining, 8-μm sections were cut and fixed in methanol. FasL was detected using goat anti-mouse FasL polyclonal Ab (Santa Cruz Biotechnologies, Santa Cruz, CA), donkey anti-goat IgG-biotin as secondary Ab and streptavidin-Texas Red as the signal fluorochrome (Fig. 1 b).

Single-cell suspensions were made from spleens of diabetic NOD mice. These cells were washed twice with HBBS and counted. A total of 5 × 107 splenocytes were resuspended in 300 μl of PBS and injected i.p. into NODSCID mice. Recipient mice were maintained under specific pathogen-free conditions and observed for diabetes development by weekly urine testing (Testape; Eli Lilly, Indianapolis, IN).

Islets were isolated as previously described (20). Animals were anesthetized using Avertin injection (12 μl/g i.p.; Sigma-Aldrich, St. Louis, MO). An abdominal incision was then made, and the common bile duct was identified and cannulated using a 30-gauge needle. A solution of Collagenase P (Roche; 2.5 mg/ml) in HBSS and 0.05% BSA was injected through the bile duct, and the pancreas was inflated. Pancreatic tissue was digested for 15 min in a 37°C water bath. Individual pancreatic islets were then handpicked and washed three times in HBSS containing FCS.

Isolated islets were transplanted under the kidney capsule using established techniques (21). Briefly, animals were anesthetized with Avertin (12 μl/g) given i.p. An incision was made in the left flank, and the kidney was exposed. Using a dissecting microscope, the kidney capsule was dissected to create a pocket where islet grafts were placed. The kidney was repositioned, and the abdominal incision was closed. In relevant experiments blood glucose levels were measured every second day following transplantation until the development of diabetes. At the time of graft collection, animals were sacrificed, and the grafted kidney was exposed and examined macroscopically in situ, then removed and fixed in 10% formalin. Histological analysis was performed using H&E staining and insulin and glucagon immunostaining (DAKO, Carpenteria, CA). The extent of graft destruction was expressed as the number of islets with destructive lesions divided by the total number of islets.

A total of 30 islets/well were incubated in 1 ml of islet culture medium (RPMI 1640, containing 2000 mg/liter glucose, l-glutamine, 10% FCS, and 50 μg/ml gentamicin) in 24-well plates at 37°C in 95% air and 5% CO2 for 24 h. In relevant experiments islets were incubated in the presence of 40 U/ml IL-1β (PeproTech, Rocky Hill, NJ) and 500 U/ml IFN-γ (PeproTech). FasL was neutralized using hamster anti-mouse FasL mAb (clone MFL3; BD PharMingen) at a final concentration of 5 μg/ml.

After 24-h incubation, single islet cell suspensions were prepared by resuspension of islets in a trypsin/HBBS solution (0.05% final concentration), incubation for 10 min in a 37°C water bath, and then drawing five times through a 19-gauge needle before being washed with HBBS. Single-cell suspensions were stained with propidium iodide (Sigma-Aldrich) at 2 μg/ml and analyzed on a cytometer (FACScan; BD Biosciences, San Jose, CA). The percent increase in β cells undergoing apoptosis was calculated using the following formula: (% apoptosis in stimulated β cells − % apoptosis in unstimulated β cells)/(% apoptosis in unstimulated β cells) × 100. This formula allows comparison between experiments with different basal levels of apoptosis.

Comparisons of means were made using the two-tailed Student’s t test. Comparisons of incidence were made using two-tailed Fisher’s exact test.

The transgenic line of HIPFasL mice demonstrated accelerated diabetes compared with nontransgenic littermates. This was particularly true for HIPFasL males, which became diabetic as early as 25 days of age. By 84 days of age, 44% of the HIPFasL males were diabetic compared with 0% of the NOD male littermate controls (p < 0.001; Fig. 2). At 160 days of age, the incidence of diabetes in HIPFasL males was 55%, and that in wild-type NOD male littermates was 10% (p < 0.001)

FIGURE 2.

Incidence of diabetes in male HIPFasL (•; n = 19) and NOD littermate controls (○; n = 31). HIPFasL male mice demonstrated a reduced age of onset and an increased incidence of diabetes (p < 0.001).

FIGURE 2.

Incidence of diabetes in male HIPFasL (•; n = 19) and NOD littermate controls (○; n = 31). HIPFasL male mice demonstrated a reduced age of onset and an increased incidence of diabetes (p < 0.001).

Close modal

HIPFasL females also became diabetic earlier and with a higher incidence than nontransgenic littermates. Because of the higher baseline rate of diabetes in NOD females, this difference was less marked than for HIPFasL males. HIPFasL females developed diabetes as early as 40 days of age, and by 84 days of age, 13.5% of the HIPFasL females were diabetic compared with 0% of the NOD female littermate controls (p < 0.05). At 160 days of age, the incidence of diabetes in HIPFasL females was 66%, and in wild-type NOD female littermates it was 51% (Fig. 3).

FIGURE 3.

Incidence of diabetes in female HIPFasL (•; n = 29) and NOD littermate controls (○; n = 33). HIPFasL female mice demonstrated a reduced age of onset of diabetes (p < 0.05).

FIGURE 3.

Incidence of diabetes in female HIPFasL (•; n = 29) and NOD littermate controls (○; n = 33). HIPFasL female mice demonstrated a reduced age of onset of diabetes (p < 0.05).

Close modal

A marked mononuclear infiltrate preceded the development of diabetes in HIPFasL and wild-type mice, with no significant evidence of a neutrophilic infiltrate (Fig. 4)

FIGURE 4.

Section of a HIPFasL 40 day-old male pancreas showing lymphocytic infiltration of the islet. H&E staining; magnification, ×400.

FIGURE 4.

Section of a HIPFasL 40 day-old male pancreas showing lymphocytic infiltration of the islet. H&E staining; magnification, ×400.

Close modal

To investigate whether the FasL transgene caused diabetes in other genetic backgrounds, HIPFasL mice were backcrossed with nondiabetes-prone strains. When HIPFasL mice were crossed with C57BL/6, the resultant (HIPFasL/NOD × C57BL/6) male F1 mice remained euglycemic when monitored for diabetes development up to 84 days of age (Fig. 5). When HIPFasL mice were crossed with (B6/DBA/2)F1 mice, only one of 15 (7%) HIPFasL/NOD × B6/DBA/2 male F1 mice developed diabetes by 84 days of age compared with eight of 18 (44%) HIPFasL male controls. Histological examination of the islets of nondiabetic F1 mice showed minimal and infrequent insulitis. These experiments indicate that the NOD genetic background is critical for the accelerated diabetes seen in mice expressing the HIPFasL transgene.

FIGURE 5.

Comparison of incidence of diabetes at 12 wk when the HIPFasL transgene was expressed in male mice of the following backgrounds: (C57BL/6 × DBA/2)F1×NOD)F1 (n = 15), (C57BL/6 × NOD)F1 (n = 25), and NOD (n = 18).

FIGURE 5.

Comparison of incidence of diabetes at 12 wk when the HIPFasL transgene was expressed in male mice of the following backgrounds: (C57BL/6 × DBA/2)F1×NOD)F1 (n = 15), (C57BL/6 × NOD)F1 (n = 25), and NOD (n = 18).

Close modal

To determine whether the accelerated diabetes in HIPFasL transgenic mice was T cell mediated, the HIPFasL transgene was crossed onto the NODSCID background. At 160 days of age zero of 10 (0%) of HIPFasLSCID mice had developed diabetes, and histological examination revealed no evidence of insulitis. This confirms that a functional immune system is necessary for FasL transgene-mediated accelerated diabetes.

To further examine the mechanism by which FasL expression accelerates diabetes, the ability of splenocytes from diabetic HIPFasL mice to transfer diabetes to NODSCID recipients was assessed. Splenocytes (5 × 107) from male diabetic HIPFasL mice were injected i.p. into male NODSCID recipients, and the mice were monitored for diabetes development. The time taken for diabetes to develop in NODSCID recipient mice was dependent on the age of diabetes onset of the donor HIPFasL mice (Fig. 6). When the donor’s age of diabetes onset was <90 days, it took 68 ± 12 days for diabetes to develop in the recipient, whereas when the donor’s age of onset was >90 days, the time to diabetes was 42 ± 17 days (n = 5; p < 0.01). This indicates that splenocytes of HIPFasL mice becoming diabetic at a young age have a reduced ability to cause β cell destruction compared with splenocytes from transgenic mice developing diabetes at a similar age as wild-type NOD mice.

FIGURE 6.

Time to onset of diabetes in NODSCID recipient mice after transfer of donor splenocytes from diabetic HIPFasL mice developing diabetes before or after 90 days of age. The time to diabetes was significantly longer when splenocytes from diabetic HIPFasL mice <90 days of age were used for transfers (68 ± 12 vs 42 ± 17 days; n = 5; p < 0.01).

FIGURE 6.

Time to onset of diabetes in NODSCID recipient mice after transfer of donor splenocytes from diabetic HIPFasL mice developing diabetes before or after 90 days of age. The time to diabetes was significantly longer when splenocytes from diabetic HIPFasL mice <90 days of age were used for transfers (68 ± 12 vs 42 ± 17 days; n = 5; p < 0.01).

Close modal

To determine whether HIPFasL β cells were susceptible to destruction by a nondestructive NOD islet infiltrate, we cotransplanted HIPFasL and NOD islets under the kidney capsule of nondiabetic NOD mice. Islets from HIPFasL or nondiabetic NOD male mice aged 100–120 days were transplanted under the kidney capsule of 70-day-old NOD male mice (n = 8). HIPFasL islets were grafted at the cephalic pole, and islets from NOD mice were grafted at the caudal pole of the kidney. The grafted kidney was harvested 8 wk after the original graft. At the time of harvesting no mice were diabetic. Grafts were inspected macroscopically, and lesions were classified as destructive or nondestructive as previously described (20). All islet grafts from NOD donors showed neovascularization with conserved structure of the graft and well-demarcated islet architecture, characteristic of nondestructive insulitis (Fig. 7,a). By comparison, islet grafts from HIPFasL donors showed swelling of the graft site with islets having a milky white appearance and being poorly defined, characteristic of destructive insulitis (Fig. 7,b). One of the HIPFasL grafts was no longer visible and had presumably already been completely destroyed. Histological examination of HIPFasL donor grafts sections revealed a dense mononuclear cell infiltrate around the grafted islets, with extensive islet destruction and with 36% of the remaining islets exhibiting destructive insulitis characterized by peri- and intraislet infiltrates, destruction of islet architecture (Fig. 7,d) and loss of insulin staining. By comparison, despite a similar degree of insulitis >99% of the NOD donor islet grafts exhibited conserved islet structure (Fig. 7 c) and strong insulin staining consistent with benign insulitis.

FIGURE 7.

HIPFasL and NOD islets were cotransplanted under the kidney capsule of 70-day-old NOD nondiabetic mice (n = 8) and harvested 8 wk later. a–d correspond to one representative animal. Macroscopic in situ examination of the NOD donor islet graft (a) shows neovascularization with preserved graft structure and well-demarcated islet architecture (arrows). H&E staining (c) shows nondestructive peri-islet infiltrate with intact islet structure (arrows). By contrast, the HIPFasL donor islet graft (b) demonstrates a cloudy white color and has lost islet definition (arrows). Histological examination of the HIPFasL graft (d) shows destructive insulitis with peri- and intraislet infiltrates and destruction of islet architecture (arrows).

FIGURE 7.

HIPFasL and NOD islets were cotransplanted under the kidney capsule of 70-day-old NOD nondiabetic mice (n = 8) and harvested 8 wk later. a–d correspond to one representative animal. Macroscopic in situ examination of the NOD donor islet graft (a) shows neovascularization with preserved graft structure and well-demarcated islet architecture (arrows). H&E staining (c) shows nondestructive peri-islet infiltrate with intact islet structure (arrows). By contrast, the HIPFasL donor islet graft (b) demonstrates a cloudy white color and has lost islet definition (arrows). Histological examination of the HIPFasL graft (d) shows destructive insulitis with peri- and intraislet infiltrates and destruction of islet architecture (arrows).

Close modal

The previous experiments suggested that there may be different mechanisms mediating diabetes development in young or old HIPFasL mice. To gain greater insights into these differences, NODSCID islet isografts were placed under the kidney capsule of diabetic HIPFasL mice according to their age of diabetes onset. Two groups of graft recipients were used; those in which the age of diabetes onset was <90 days (n = 5) and those in which it was >90 days (n = 5). Wild-type diabetic NOD mice normally destroy NODSCID islet isografts within 7 days. Twenty hours post-transplantation all grafted animals had returned to normoglycemia, and blood glucose was monitored every second day for 40 days. The return of persistent hyperglycemia was taken to reflect islet graft destruction, and this was confirmed by histological examination.

Fourteen days post-transplant, graft survival was significantly different between the two groups (p < 0.02). In graft recipients originally developing diabetes at >90 days of age, all islet grafts were destroyed within 7 days of transplantation (n = 5/5). By contrast, none of the grafts in those recipients originally developing diabetes at <90 days of age were destroyed at up to 14 days post-transplantation, and mean graft survival time was >90 days.

Some of the previous experiments could be explained if HIPFasL islets were more susceptible to apoptosis than wild-type NOD islets. To test this possibility, islets were isolated from HIPFasL mice (n = 11) and wild-type NOD controls (n = 9) and cultured at 30 islets/well for 24 h in the presence or the absence of IL-1β (40 U/ml) and IFN γ (500 U/ml). Single-cell suspensions were stained with propidium iodide at 2 μg/ml and then analyzed by flow cytometry.

The percentage of cells undergoing apoptosis in response to cytokine stimulation was significantly higher (p < 0.001) for HIPFasL (57.06 ± 38.32%) than wild-type NOD β cells (19.15 ± 13.26%; Fig. 8). This confirms that β cells expressing the FasL transgene are indeed more susceptible to cytokine-induced apoptosis than wild-type β cells.

FIGURE 8.

Islets incubated with IL-1β and IFN-γ for 24 h were assessed for apoptosis by staining with propidium iodide. Histograms show apoptosis of HIPFasL (a) and NOD (b) β cells after cytokine stimulation; data shown are from one representative experiment. c, Cytokine-induced apoptosis was lower for NOD (19 ± 13.3%; n = 11) than for HIPFasL (57 ± 38.3%; n = 14) β cells (p < 0.001).

FIGURE 8.

Islets incubated with IL-1β and IFN-γ for 24 h were assessed for apoptosis by staining with propidium iodide. Histograms show apoptosis of HIPFasL (a) and NOD (b) β cells after cytokine stimulation; data shown are from one representative experiment. c, Cytokine-induced apoptosis was lower for NOD (19 ± 13.3%; n = 11) than for HIPFasL (57 ± 38.3%; n = 14) β cells (p < 0.001).

Close modal

To confirm that the mechanism for the increased susceptibility of HIPFasL β cells to cytokine-induced apoptosis was mediated via Fas-FasL interaction, we repeated the above experiment in the presence of neutralizing anti-FasL Ab. Addition of anti-FasL resulted in a 70% reduction in HIPFasL β cell apoptosis from 42.77 ± 31.02% with cytokine stimulation down to 12.79 ± 22.63% with FasL neutralization (p < 0.001; Fig. 9).

FIGURE 9.

Effect of blocking FasL on cytokine-induced apoptosis of HIPFasL β cells. Islets were incubated in the presence of IL-1β and IFN-γ with and without FasL blocking Ab for 24 h. a, Histograms showing apoptosis of HIPFasL β cells after cytokine stimulation in the absence (a) or the presence (b) of anti-FasL neutralizing Ab. c, Apoptosis of HIPFasL β cells in the presence of anti-FasL neutralizing Ab was reduced from 43 ± 31.0% to 13 ± 22.6% (n = 7; p < 0.05).

FIGURE 9.

Effect of blocking FasL on cytokine-induced apoptosis of HIPFasL β cells. Islets were incubated in the presence of IL-1β and IFN-γ with and without FasL blocking Ab for 24 h. a, Histograms showing apoptosis of HIPFasL β cells after cytokine stimulation in the absence (a) or the presence (b) of anti-FasL neutralizing Ab. c, Apoptosis of HIPFasL β cells in the presence of anti-FasL neutralizing Ab was reduced from 43 ± 31.0% to 13 ± 22.6% (n = 7; p < 0.05).

Close modal

Transgenic FasL expression on NOD β cells results in an accelerated form of autoimmune diabetes. For reasons not currently explained, this effect is more pronounced in male than in female mice. A marked mononuclear infiltrate preceded the development of diabetes. This contrasts with reports of other FasL transgenic lines (22), where the infiltrate was predominantly comprised of neutrophils. Insulitis in HIPFasL mice is apparent as early as 20 days of age, and diabetes develops from 25 days of age. This accelerated diabetes is dependant on the NOD genetic background, as 39 of 40 mice expressing HIPFasL on a mixed background showed no histological evidence of insulitis, and only one of 40 developed diabetes. The reason why accelerated diabetes is dependent on the NOD genetic background is not known. One explanation is that for transgenic FasL to mediate β cell apoptosis, Fas must first be up-regulated. In the NOD mouse this stimulus is provided by cytokines produced by the insulitis lesion, such as IL-1 and IFN-γ. This stimulus for Fas expression would be absent, however, in F1 mice that did not have insulitis, thereby explaining their protection from FasL-induced apoptosis. This is supported by the fact that diabetes in the FasL model is T cell dependent, as expression of HIPFasL on the NODSCID background, which has no T or B cells, did not result in insulitis or diabetes. An alternative explanation is that particular genes in the NOD background may make NOD β cells particularly susceptible to Fas-mediated apoptosis. The previously suggested role of neutrophils is not supported by the current data, with diabetes instead being dependent on the presence of T cells.

The diabetes of HIPFasL mice could potentially be explained by the FasL transgene causing an acceleration of the normal autoimmune diabetes of NOD mice. Other features of autoimmune diabetes in the NOD include rapid disease recurrence in islet isografts (23) and the ability of splenocytes from diabetic donors to transfer disease to NODSCID recipients (24). HIPFasL mice do not, however, demonstrate the same features. Although splenocytes from HIPFasL mice transferred diabetes to NODSCID recipients, the time taken for diabetes to develop in the recipients was >50% longer than that for recipients of nontransgenic NOD splenocytes. In fact, the time to diabetes in NODSCID recipients of HIPFasL splenocytes was similar to that seen in recipients of age-matched nondiabetic NOD donor splenocytes. Even more uncharacteristically, when islets (NODSCID) were grafted into recently diabetic young or old HIPFasL mice, all grafts were destroyed within 7 days in HIPFasL mice that developed diabetes after 90 days of age, compared with long term survival of four of five grafts into HIPFasL that developed diabetes before 90 days of age. Diabetes eventually occurred in these graft recipients between 150 and 250 days of age.

The above findings when taken together suggest that despite the similarities there are some fundamental differences in the disease process of HIPFasL and wild-type NOD mice. It is possible that the disease processes are the same, but that the reduced ability of HIPFasL splenocytes to transfer disease reflects FasL on transgenic β cells causing the death of Fas-expressing T cells at the very same time that the β cells are themselves being destroyed. This could result in depletion of β cell-reactive clones capable of either destroying an isograft or transferring disease to NODSCID recipients. Another possibility is that diabetes in HIPFasL mice requires the presence of insulitis (thereby explaining the necessity for lymphocytes and an NOD genetic background), but that FasL is directly responsible for β cell death either through autodestruction (the local cytokine milieu up-regulates β cell Fas expression, leading to Fas-FasL interaction and cell suicide) or because FasL may activate lymphoid or other hemopoietic cells that then indirectly kill β cells through other effector pathways, e.g., TNF.

We have also shown that HIPFasL β cells are more susceptible than wild-type cells to cytokine-induced apoptosis. This increased susceptibility is due to the action of FasL, as neutralization of FasL with blocking Ab abrogates cytokine-induced apoptosis of transgenic β cells. This is consistent with a scenario in which a nondestructive insulitis that in nontransgenic NOD mice would not cause β cell destruction is sufficient to induce β cell apoptosis in HIPFasL mice. The fact that HIPFasL grafts are destroyed while wild-type NOD islets are preserved when cotransplanted into nondiabetic NOD mice clearly indicates that HIPFasL islets are highly susceptible to benign insulitis. This may occur through cytokine-mediated up-regulation of β cell Fas, subsequent FasL ligation, and cell death. Disease recurrence is not seen in isografts or transferable to NODSCID recipients because the destructive insulitis phase and its associated effector T cell populations have not yet been established. In normal NOD mice the absence of constitutive FasL expression by β cells prevents β cell destruction during the phase of benign insulitis. However, during the transition to destructive insulitis a change in the cytokine milieu may cause up-regulation of β cell FasL, thereby triggering β cell death. Whether this is a part of the mechanism of β cell destruction in the normal course of autoimmune diabetes needs to be established, but there is evidence that this could be the case (25, 26, 27).

Other investigators have shown that the Fas-FasL pathway plays a major role in β cell death in autoimmune diabetes (28, 29, 30). Our data suggest in addition that the level of islet FasL expression determines the level of cytokine-induced β cell apoptosis. The accelerated diabetes in FasL transgenic mice would appear, therefore, to be due to the interaction of FasL with Fas up-regulated on β cells in response to cytokines produced by infiltrating T cells. This would explain why diabetes does not occur in HIPFasLSCID mice.

We thank Dr. Shiva Reddy for assistance with islet immunohistochemistry; Stephanie Palmer, Darren Newington, and Megan Cramond for technical assistance; and Sharen Pringle for administrative help.

1

D.S. is the recipient of a scholarship from the Canberra Hospital Salaried Specialists Private Practice Fund.

3

Abbreviations used in this paper: FasL, Fas ligand; NOD, nonobese diabetic.

1
Kurrer, M. O., S. V. Pakala, H. L. Hanson, J. D. Katz.
1997
. β cell apoptosis in T cell-mediated autoimmune diabetes.
Proc. Natl. Acad. Sci. USA
94
:
213
.
2
Chervonsky, A. V., Y. Wang, F. S. Wong, I. Visintin, R. A. Flavell, C. A. Janeway, Jr., L. A. Matis.
1997
. The role of Fas in autoimmune diabetes.
Cell
89
:
17
.
3
Stassi, G., R. De Maria, G. Trucco, W. Rudert, R. Testi, A. Galluzzo, C. Giordano, M. Trucco.
1997
. Nitric oxide primes pancreatic β cells for Fas-mediated destruction in insulin-dependent diabetes mellitus.
J. Exp. Med.
186
:
1193
.
4
Suarez-Pinzon, W., O. Sorensen, R. C. Bleackley, J. F. Elliott, R. V. Rajotte, A. Rabinovitch.
1999
. β cell destruction in NOD mice correlates with Fas (CD95) expression on β cells and proinflammatory cytokine expression in islets.
Diabetes
48
:
21
.
5
Moriwaki, M., N. Itoh, J. Miyagawa, K. Yamamoto, A. Imagawa, K. Yamagata, H. Iwahashi, H. Nakajima, M. Namba, S. Nagata, et al
1999
. Fas and Fas ligand expression in inflamed islets in pancreas sections of patients with recent-onset type I diabetes mellitus.
Diabetologia
42
:
1332
.
6
Loweth, A. C., G. T. Williams, R. F. James, J. H. Scarpello, N. G. Morgan.
1998
. Human islets of Langerhans express Fas ligand and undergo apoptosis in response to interleukin-1β and Fas ligation.
Diabetes
47
:
727
.
7
Stassi, G., M. Todaro, P. Richiusa, M. Giordano, A. Mattina, M. S. Sbriglia, A. Lo Monte, G. Buscemi, A. Galluzzo, C. Giordano.
1995
. Expression of apoptosis-inducing CD95 (Fas/Apo-1) on human β cells sorted by flow-cytometry and cultured in vitro.
Transplant. Proc.
27
:
3271
.
8
Signore, A., A. Annovazzi, E. Procaccini, P. E. Beales, J. Spencer, R. Testi, G. Ruberti.
1997
. CD95 and CD95-ligand expression in endocrine pancreas of NOD, NOR and BALB/c mice.
Diabetologia
40
:
1476
.
9
Maedler, K., G. A. Spinas, R. Lehmann, P. Sergeev, M. Weber, A. Fontana, N. Kaiser, M. Y. Donath.
2001
. Glucose induces β cell apoptosis via upregulation of the Fas receptor in human islets.
Diabetes
50
:
1683
.
10
Thomas, H. E., R. Darwiche, J. A. Corbett, T. W. Kay.
1999
. Evidence that β cell death in the nonobese diabetic mouse is Fas independent.
J. Immunol.
163
:
1562
.
11
Kayagaki, N., N. Yamaguchi, F. Nagao, S. Matsuo, H. Maeda, K. Okumura, H. Yagita.
1997
. Polymorphism of murine Fas ligand that affects the biological activity.
Proc. Natl. Acad. Sci. USA
94
:
3914
.
12
Itoh, N., A. Imagawa, T. Hanafusa, M. Waguri, K. Yamamoto, H. Iwahashi, M. Moriwaki, H. Nakajima, J. Miyagawa, M. Namba, et al
1997
. Requirement of Fas for the development of autoimmune diabetes in nonobese diabetic mice.
J. Exp. Med.
186
:
613
.
13
Kim, S., K. A. Kim, D. Y. Hwang, T. H. Lee, N. Kayagaki, H. Yagita, M. S. Lee.
2000
. Inhibition of autoimmune diabetes by Fas ligand: the paradox is solved.
J. Immunol.
164
:
2931
.
14
Su, X., Q. Hu, J. M. Kristan, C. Costa, Y. Shen, D. Gero, L. A. Matis, Y. Wang.
2000
. Significant role for Fas in the pathogenesis of autoimmune diabetes.
J. Immunol.
164
:
2523
.
15
Suarez-Pinzon, W. L., R. F. Power, A. Rabinovitch.
2000
. Fas ligand-mediated mechanisms are involved in autoimmune destruction of islet β cells in non-obese diabetic mice.
Diabetologia
43
:
1149
.
16
Amrani, A., J. Verdaguer, B. Anderson, T. Utsugi, S. Bou, P. Santamaria.
1999
. Perforin-independent β cell destruction by diabetogenic CD8+ T lymphocytes in transgenic nonobese diabetic mice.
J. Clin. Invest.
103
:
1201
.
17
Kim, Y. H., S. Kim, K. A. Kim, H. Yagita, N. Kayagaki, K. W. Kim, M. S. Lee.
1999
. Apoptosis of pancreatic β cells detected in accelerated diabetes of NOD mice: no role of Fas-Fas ligand interaction in autoimmune diabetes.
Eur. J. Immunol.
29
:
455
.
18
Mahiou, J., U. Walter, F. Lepault, F. Godeau, J. F. Bach, L. Chatenoud.
2001
. In vivo blockade of the Fas-Fas ligand pathway inhibits cyclophosphamide-induced diabetes in NOD mice.
J. Autoimmun.
16
:
431
.
19
Petrovsky, N., D. Silva, L. Socha, R. Slattery, B. Charlton.
2002
. The role of Fas ligand in β cell destruction in autoimmune diabetes of NOD mice.
Ann. NY Acad. Sci.
958
:
204
.
20
Gazda, L. S., B. Charlton, K. J. Lafferty.
1997
. Diabetes results from a late change in the autoimmune response of NOD mice.
J. Autoimmun.
10
:
261
.
21
Bowen, K. M., L. Andrus, K. J. Lafferty.
1980
. Successful allotransplantation of mouse pancreatic islets to nonimmunosuppressed recipients.
Diabetes
29
:
98
.
22
Allison, J., H. M. Georgiou, A. Strasser, D. L. Vaux.
1997
. Transgenic expression of CD95 ligand on islet β cells induces a granulocytic infiltration but does not confer immune privilege upon islet allografts.
Proc. Natl. Acad. Sci. USA
94
:
3943
.
23
Wang, Y., O. Pontesilli, R. G. Gill, F. G. La Rosa, K. J. Lafferty.
1991
. The role of CD4+ and CD8+ T cells in the destruction of islet grafts by spontaneously diabetic mice.
Proc. Natl. Acad. Sci. USA
88
:
527
.
24
Rohane, P. W., A. Shimada, D. T. Kim, C. T. Edwards, B. Charlton, L. D. Shultz, C. G. Fathman.
1995
. Islet-infiltrating lymphocytes from prediabetic NOD mice rapidly transfer diabetes to NOD-scid/scid mice.
Diabetes
44
:
550
.
25
Amrani, A., J. Verdaguer, S. Thiessen, S. Bou, P. Santamaria.
2000
. IL-1α, IL-1β, and IFN-γ mark β cells for Fas-dependent destruction by diabetogenic CD4+ T lymphocytes.
J. Clin. Invest.
105
:
459
.
26
Darville, M. I., D. Liu, M. C. Chen, D. L. Eizirik.
2001
. Molecular regulation of Fas expression in β cells.
Diabetes
50
:
S83
.
27
Zumsteg, U., S. Frigerio, G. A. Hollander.
2000
. Nitric oxide production and Fas surface expression mediate two independent pathways of cytokine-induced murine β cell damage.
Diabetes
49
:
39
.
28
Giordano, C., P. Richiusa, M. S. Sbriglia, G. Pizzolanti.
1998
. The role of FAS/FASLig and apoptosis in β cell death in animal models: relevance for human IDDM.
Diabetes Metab. Rev.
14
:
194
.
29
Kreuwel, H. T., L. A. Sherman.
2001
. The role of Fas-FasL in CD8+ T-cell-mediated insulin-dependent diabetes mellitus (IDDM).
J. Clin. Immunol.
21
:
15
.
30
Wang, H., Y. Li, Q. Sun, G. Yang.
2000
. Oral administration of insulin to female nonobese diabetic mice inhibited diabetes and induced Fas ligand expression on islets of Langerhans.
Chin. Med. J.
113
:
433
.