It is widely proposed that reactive oxygen species (ROS) contribute to β-cell death in type 1 diabetes. We tested this in nonobese diabetic (NOD) mice using β-cell–specific overexpression of three antioxidant proteins: metallothionein (MT), catalase (Cat), or manganese superoxide dismutase (MnSOD). Unexpectedly, the cytoplasmic antioxidants, MT and catalase, greatly accelerated diabetes after cyclophosphamide and accelerated spontaneous diabetes in male NOD mice. This occurred despite the fact that they reduced cytokine-induced ROS production and MT reduced streptozotocin diabetes in NOD mice. Accelerated diabetes onset coincided with increased β-cell death but not with increased immune attack. Islets from MTNOD mice were more sensitive to cytokine injury. In vivo and in vitro studies indicated reduced activation of the Akt/pancreatic duodenal homeobox-1 survival pathway in MTNOD and CatNOD islets. Our study indicates that cytoplasmic ROS may have an important role for protecting the β-cell from autoimmune destruction.
It is generally believed that reactive oxygen species (ROS) contribute to autoimmune-mediated pancreatic β-cell destruction and that application of antioxidants would benefit β-cell survival in type 1 diabetes. To test this, a series of studies have been carried out using different models, including diabetic animals, transplanted islets, cultured primary islets, and insulin-producing cell lines. However, the data from those reports are inconsistent and ambiguous. For example, some in vitro studies (1,2) demonstrated that overexpression of antioxidant enzymes improved insulin-producing tumor cell resistance to ROS and cytokine toxicity. But other studies on primary islet cells, treated with antioxidant compounds (3,4) or islets that overexpressed antioxidant enzymes (5,6), found no protection against cytokine toxicity.
Several in vivo studies had positive results. Systemic treatment with antioxidant compounds delayed type 1 diabetes onset (7,8) and prolonged islet graft survival (9,10). However, some studies found that antioxidants were ineffective for prevention of type 1 diabetes (11). Moreover, a widely used antioxidant, N-acetyl-cysteine, accelerated the transfer of diabetes into NOD scid mice (12). The most successful approach, targeted β-cell overexpression of thioredoxin, produced significant protection against type 1 diabetes in NOD mice (13), but the role of antioxidant protection in this result can be debated because thioredoxin has additional potent antiapoptotic actions (14). Pancreatic β-cell–specific overexpression of the secreted form of superoxide dismutase (SOD) failed to modify the progression of type 1 diabetes in NOD mice (15). Thus, the published studies have not clarified the true roles of ROS and antioxidants in autoimmune type 1 diabetes at the level of the β-cell.
In addition to the pathophysiologically destructive actions of ROS, transient bursts of small amounts of cellular ROS have been recognized to play an important role in the action of a variety of growth factors, cytokines, and hormones, including insulin-regulated signal transduction pathways (16,17). Recently, ROS have been demonstrated to serve as a second messenger to facilitate insulin/IGF signal cascade, probably through the modulation of protein tyrosine phosphatase (PTP) (18,19). In β-cells, insulin/IGF signaling regulates growth, survival, and metabolism (20–22). Increased activity of insulin/IGF signal transduction by genetic manipulation or administration of exogenous IGF blocked cytokine-mediated inhibition of insulin secretion and islet cell apoptosis (23,24) and also provided protection from type 1 diabetes in NOD mice (25). However, it is still uncertain whether ROS modulation of insulin/IGF signaling affects the fate of β-cells during the progression of autoimmune diabetes.
In the present study, we explored the roles of ROS in type 1 diabetes using transgenic NOD mice with β-cell overexpression of metallothionein (MT), catalase, or manganese SOD (MnSOD). Our data demonstrated that none of these antioxidants provided any detectable benefit against autoimmune diabetes. On the contrary, the two cytoplasmic antioxidants, MT and catalase, markedly sensitized to autoimmune diabetes and islet cell death. Further analysis revealed that a β-cell survival signal pathway, the Akt/pancreatic duodenal homeobox-1 (PDX-1) pathway, was inhibited by MT and catalase transgenes during the progression of NOD diabetes or upon stimulation with cytokines. A PTP inhibitor, orthovanadate, counteracted the sensitizing effect of MT. These results imply that the normal cellular ROS activity may play an important role in activating protective responses of β-cells in autoimmune type 1 diabetes. High-level expression of cellular antioxidants may prevent activation of protective pathways, thereby sensitizing β-cells to apoptosis and type 1 diabetes.
RESEARCH DESIGN AND METHODS
Chemicals.
5-(6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate (CM-H2DCFDA) and PicoGreen were obtained from Molecular Probes (Eugene, OR). Alamar Blue was from Biosource International (Camarillo, CA). Cyclophosphamide (CYP), sodium orthovanadate, streptozotocin (STZ), recombinant mouse interleukin (IL)-1β, interferon-γ, tumor necrosis factor (TNF)-α, and anti-mouse β-actin antibody were from Sigma (St. Louis, MO). Rabbit antiserum to guinea pig insulin was from BioGenex (San Ramon, CA). Mouse anti-horse MT antibody was supplied by Dako (Carpinteria, CA). Sheep anti-human SOD and rabbit anti-human catalase were from Biodesign International (Saco, ME). Phospho-Akt (Ser473), Akt, phospho-Foxo-1 (Ser256), Foxo-1, cleaved caspase 3, and total caspase 3 antibodies were from Cell Signaling Technology (Beverly, MA). PDX-1 antibody was from Abcam (Cambridge, MA). Donkey anti-rabbit fluorescein isothiocyanate and Cy3-conjugated IgG were from Jackson ImmunoResearch Laboratories (West Grove, PA). Other reagents were from Sigma or Fisher.
Generation of transgenic NOD mice.
MT, catalase, and MnSOD transgenes, maintained on FVB background as previously described (5,6,26), were transferred onto the NOD background by repeated backcrossing with NOD mice (The Jackson Laboratory, Bar Harbor, ME) using the speed congenic strategy (27). In each generation, offspring were selected by PCR not only for the presence of transgene alleles but also for the lowest degree of heterozygosity for polymorphic microsatellite markers linked to 19 NOD diabetic susceptibility loci (Idd1 to Idd19). After five to seven backcrosses, founder NOD mice congenic for transgenes were established and identified as homozygous of all 19 NOD Idd alleles. Those founder mice were further backcrossed to NOD mice twice and then maintained by sib mating between transgenic and nontransgenic littermates. Mice used in the present study were littermates derived from backcross 7 for MT and MnSOD and backcross 9 for catalase. All mice were housed in ventilated cages at the University of Louisville Research Resources Center with free access to water and standard mouse diet. The animal procedures were approved by the Institutional Animal Care and Use Committee, which is certified by the American Association of Accreditation of Laboratory Animal Care.
Monitoring for CYP-accelerated, spontaneous, and STZ-induced diabetes.
To induce accelerated diabetes development, 6- to 7-week-old sex-matched transgenic and nontransgenic NOD littermates received two intraperitoneal injections of 200 mg/kg body wt of CYP, 2 weeks apart. Tail blood glucose level was measured every other day using a glucose meter (OneTouch Ultra; LifeScan, Milpitas, CA). In some experiments, mice were killed at the indicated days after the first injection of CYP for pancreas insulin measurement, pancreas sectioning, or pancreatic islet isolation. For spontaneous NOD diabetes, both female and male mice were tested every 2 weeks for tail blood glucose starting at the age of 8 weeks. For STZ-induced diabetes, 6- to 7-week-old sex-matched mice were treated with a single intraperitoneal injection of STZ in 0.1 mol/l sodium citrate (pH 4.5) at a dose of 180 mg/kg body wt, and tail blood glucose was monitored daily for 6 days. Mice were considered diabetic after two consecutive measurements of blood glucose >200 mg/dl. The onset of diabetes was dated from the first of these two sequential measurements.
Measurement of pancreas insulin, islet insulin, DNA, and glutathione peroxidase activity.
Islet isolation, culture, and measurement of these parameters were previously described (26). Briefly, pancreas or islet insulin was extracted with acid ethanol and measured with a mouse insulin enzyme-linked immunosorbent assay kit (ALPCO Diagnostics, Windham, NH) according to the manufacturer’s manual. Islet DNA was quantified with PicoGreen (Molecular Probes). Islet glutathione peroxidase activity was measured by a glutathione peroxidase assay kit (Cayman Chemical Company, Ann Arbor, MI) according to the manufacturer’s manual. The activity was determined from a standard curve obtained with bovine erythrocyte glutathione peroxidase (Sigma) and normalized by islet protein content.
Quantitative reverse transcription PCR.
The RNA expression of endoplasmic reticulum (ER) stress markers Bip and CHOP was measured by real-time quantitative PCR using the ABI 7300 sequence detection system (Applied Biosystems, Foster City, CA) with a modified protocol as previously described (6). Briefly, total RNA was purified from isolated islets and reverse transcribed to cDNA. Ten-nanogram RNA–derived cDNA was mixed with Taqman universal PCR master mix, appropriate primers and probe, and run at 50°C for 2 min, 95°C for 10 min, then 95°C for 15 s followed by 60°C for 1 min for 40 cycles. The assay identification numbers for the specific primers and probe of each gene were Mm00517691_m1 for Bip, Mm00492097_m1 for CHOP, and Mm00607939_s1 for β-actin (Applied Biosystems). The results for CHOP and Bip were normalized to β-actin mRNA level.
Histology and immunohistochemistry.
For pre-diabetic MT, catalase, MnSOD, and insulin staining, pancreas sections were taken from mice at the age of 4–5 weeks. For insulitis scoring and cleaved caspase 3 immunostaining, pancreas sections were obtained from mice not treated or treated with CYP for 8 and 6 days. The degree of insulitis was blindly scored by two observers on hematoxylin-eosin–stained sections using the following scale: 0, normal islet; 1, peri-insulitis; 2, invasive intrainsulitis; and 3, atrophic islet, defined as a shrunken islet with disarranged cell distribution but with no or few lymphocyte infiltration. In each mouse, three or four different slides (at least 50 μmol/l apart) in a total of 10–14 sections were examined. Results were expressed as mean islet percentages for each insulitis category from three or four mice.
The immunostaining procedure was previously described (6). Briefly, sections were deparaffinized, rehydrated, pretreated with target antigen retrieval solution (DakoCytomation, Carpinteria, CA), washed with PBS, blocked with serum, and incubated with primary antibody at 4°C overnight. The dilution for the antibodies was as follows: insulin 1:100; catalase, MT, and MnSOD 1:50; and cleaved caspase 3 1:500. At the end of the incubation, immune complexes were detected either by incubation with appropriate biotin-labeled IgG, followed by ABC reagent and diaminobenzidine chromogen (Vector Laboratory, Burlingame, CA), or by incubation with appropriate fluorescein isothiocyanate–or Cy3-conjugated IgG. Quantitative analysis of cleaved caspase 3 staining area was performed with ImagePro software (Media Cybemetrics, Silver Spring, MD). In each mouse, three to six different slides were examined. Results were expressed as mean percentage of cleaved caspase 3 staining area within an islet.
In vitro treatment with cytokine.
Islets isolated from transgenic and nontransgenic littermates were cultured in a medium containing cytokine mix at the indicated concentrations of IL-1β, interferon-γ, and TNF-α for 24 h. In some experiments, islets were cocultured with indicated concentrations of sodium orthovanadate (Sigma) in addition to cytokine mix. After cytokine treatment, either total islet protein was extracted for Western blot analysis or islet cell viability was assessed by Alamar Blue assay as previously described (6,28). The islet cell viability was calculated as the ratio of difference of Alamar Blue fluorescence before and after cytokine treatment over the fluorescence before treatment. Measurement of cytokine-induced ROS production in whole islets was performed as previously described (26). Briefly, islets were preloaded with 5 μmol/l CM-H2DCFDA for 1 h and treated with the indicated concentrations of cytokine mix for 16 h. After treatment, ROS production was visualized on a fluorescent microscope equipped with a digital camera.
In vitro treatment with CYP.
Assessment of CYP cytotoxicity to islets was carried out in a 96-well microplate with a modified procedure of Meyer et al. (29). Ten overnight-cultured islets from MT transgenic or nontransgenic mice on FVB background (26) were incubated for 5 h in 200 μl culture medium (pH 7.4) containing 4 mmol/l NADP+, 4 mmol/l glucose-6-phosphate, 2 mU/ml glucose-6-phosphate dehydrogenase, 75 μg protein/ml mouse liver microsome extract (30), and the indicated concentrations of CYP. After CYP treatment, the culture medium was replaced with 200 μl fresh culture medium and islets were incubated for another 16 h. After incubation, islet morphology was examined and islet DNA content was measured with PicoGreen (Molecular Probes).
Western blot analysis.
Fifty to 100 islets freshly isolated from mice without or with CYP treatment for 4 or 7 days or 200 cultured islets without or with cytokine mix treatment for 24 h were lysed in 40–50 μl cold lysis buffer (20 mmol/l Tris, pH 7.5, 150 mmol/l NaCl, 1 mmol/l EDTA, 1 mmol/l EGTA, 1% Triton X-100, 2.5 mmol/l sodium pyrophosphate, 1 mmol/l β-glycerophosphate, 1 mmol/l Na3VO4, 1 mmol/l dithiothreitol, 1 μg/ml leupeptin, and 1 mmol/l phenylmethylsulfonyl fluoride) by sonication and centrifuged at 11,000g for 30 min. Equal amounts of protein (10 μg for signaling proteins or 40 μg for cleaved caspase 3) were separated by SDS-PAGE (Bio-Rad) and transferred to polyvinylidine fluoride membranes. After blocking with 5% nonfat dry milk in 0.1% Tween-20 Tris-buffered saline solution, blots were incubated with antibodies against phosphorylated-Akt (1:1,000), Akt (1:1,000), phosphorylated-Foxo-1 (1:1,000), PDX-1 (1:7,000), cleaved caspase 3 (1:1,000), total caspase 3 (1:1,000), or β-actin (1:4,000) overnight at 4°C. Blots were further incubated with appropriate peroxidase-labeled secondary antibodies (1:2,000 dilution). The immune complexes were identified using the enhanced chemiluminescence detection system (Amersham).
Data analysis.
Data are presented as means ± SE. Statistical analysis was performed by ANOVA plus Tukey’s post hoc test for multiple comparisons. The cumulative incidence of diabetes was calculated for each group, and the significant difference was tested by Kaplan-Meier log-rank survival test.
RESULTS
Generation of speed congenic NOD mice with overexpression of antioxidants in pancreatic β-cells.
Using a speed congenic strategy (27), the MT (26), catalase (5), and MnSOD (6) transgenes were transferred from the FVB strain onto the NOD background by backcrossing for up to nine generations. Successive backcrosses were selected based on PCR-genotyping for all 19 NOD diabetic susceptibility loci (Idd1 to Idd19) until mice were homozygous for all NOD alleles. Mice used in this study were the littermates born from the intercross between transgenic and nontransgenic NOD mice.
Expression of transgenes on the NOD background was confirmed by immunohistochemistry on pancreas sections. As shown in Fig. 1, each of the transgenes was highly expressed in pancreatic islets. Western blot analysis of these proteins in various tissues, including intestine, liver, kidney, heart, brain, muscle, lung, spleen, thymus, bone marrow, and isolated islets, further demonstrated that the expression of transgenes was islet specific (data not shown). The subcellular localization of MT and catalase appeared to be cytoplasmic based on the diffuse, nonnuclear staining pattern. This is consistent with what we have previously observed for the same transgenes in β-cells of FVB mice (6) and observed by others in transgenic cardiac cells (31,32). Staining for the MnSOD transgene product was granular rather than diffuse, which is typical of mitochondrial localization. Mitochondrial targeting of transgenic MnSOD has also been shown by others (33), and this is the normal site of endogenous MnSOD.
Unexpected acceleration of NOD diabetes onset by cytoplasmic antioxidant proteins MT and catalase but not by mitochondrial MnSOD.
Antioxidants have been proposed as therapeutic agents to protect from autoimmune diabetes (10). We anticipated that one or more of the antioxidant transgenes would delay or prevent onset of NOD diabetes. The incidence of diabetes was first tested in transgenic and nontransgenic siblings following injection of CYP, which is an alkylating cytostatic drug widely used to accelerate and synchronize diabetes onset in NOD mice (34,35). However, the results were completely opposite to our initial expectation. Instead of delaying diabetes, the two cytoplasmic antioxidants transgenes remarkably accelerated NOD diabetes (Fig. 2A and B). At 20 days after the first injection of CYP, 93% of MT and 63% of catalase transgenic NOD mice were diabetic, but only ∼7% of control mice were diabetic. These results showed that MT had the most dramatic impact on accelerating diabetes and that catalase also significantly increased the rate of onset. The acceleration of β-cell damage as well as the stronger effect of the MT gene was confirmed by measurement of pancreatic insulin content. Eight days after CYP injection, insulin content in MTNOD and CatNOD mice were only at 22 and 36% of original pancreas insulin levels, respectively, whereas the control NOD mice still had 76% remaining (Fig. 2C). The MnSOD transgene, unlike MT and catalase, appeared to have no effect on diabetes onset in NOD mice; diabetes developed at a nearly identical rate in MnSOD and control NOD mice after CYP injection (Fig. 2D).
We considered the possibility that the increased sensitivity to diabetes in MTNOD and CatNOD mice may have been due to nonspecific, detrimental effects of transgene overexpression in β-cells. To rule this out, we compared parameters of islet function in transgenic and control mice. We performed the most complete analysis on MTNOD mice because they showed greater sensitization to CYP-induced diabetes than CatNOD mice. At 5–7 weeks of age, the health of transgenic islets was indicated by normal blood glucose, pancreas insulin content, islet DNA content, and islet insulin content in MTNOD and CatNOD mice (online appendix [available at http://diabetes.diabetesjournals.org]). In addition, the activity of another islet antioxidant protein, glutathione peroxidase (online appendix), was not altered by the overexpression of MT or catalase. Islet morphology (Fig. 1) and glucose tolerance tests (Fig. 3A and B) revealed no distinctions between transgenic and control animals. Our prior results (5,6,26) indicated that none of our three transgenes impaired isolated islet insulin secretion on their original FVB background, and preliminary results with MTNOD islets (data not shown) indicate normal insulin secretion. We also tested MTNOD mice for sensitization to STZ-induced diabetes (Fig. 3C). In contrast to the sensitization to CYP-induced diabetes, the MT transgene significantly increased resistance to STZ diabetes. These results demonstrated that the sensitization to type 1 diabetes was not due to general weakness of transgenic islets. We next considered the possibility that MT directly sensitizes β-cells to the toxicity of CYP. To test this possibility, we injected MT transgenic mice on the FVB background with the same CYP regimen used on NOD mice. These mice did not become diabetic (data not shown), indicating that the immune activation produced on the NOD background was necessary for CYP toxicity to MT islets. To test this more directly, isolated islets were exposed to CYP that had been activated by liver microsome extract. Activated CYP was found to be toxic to control islets (Fig. 3D and E), but MT overexpression almost eliminated this injury. These results demonstrated that MT was protective from the direct toxicity of CYP, and therefore MT sensitization to type 1 diabetes could not have been due to sensitization to the direct toxicity of CYP. We also considered the possibility that increased protein synthesis from overexpression of MT or catalase could lead to low-level ER stress, which could prime the β-cell for additional injury. To test this, we measured mRNA levels of CHOP and Bip, known markers and mediators of ER stress (36). As shown in Fig. 3F, neither MT nor catalase increased expression of these ER stress markers.
Lastly, we monitored the incidence of spontaneous diabetes in MT and catalase NOD mice. Since the major variable determining the time of onset of spontaneous diabetes is activation of immune cells (37) rather than the period of time required for β-cell death, we were unsure whether we could detect changes in diabetes onset that were due to increased β-cell sensitivity. However, for male NOD mice, both MT and catalase transgenes significantly accelerated the onset of spontaneous diabetes (Fig. 4A and B). At 200 days of age, male MTNOD and CatNOD mice had a diabetes incidence >85% compared with 35–40% in control male NOD mice. In female NOD mice, neither MT nor catalase produced a noticeable trend toward earlier onset (Fig. 4C and D).
Before the onset of NOD diabetes, many pancreatic islets are surrounded by lymphocytes that do not penetrate into the islet nor appear to target β-cells. This stage is referred to as peri-insulitis. Following some unknown trigger, more lymphocytes are targeted to the islet, and the lymphocytes enter the islet and attack the β-cells during the phase of intrainsulitis. We performed a semiquantitative, blind analysis of insulitis in MTNOD and NOD pancreas before and 8 days after CYP injection (Fig. 5). As shown in Fig. 5B, insulitis before CYP injection was similar in transgenic and control islets. Eight days after CYP, there was a trend in MTNOD pancreas (P = 0.07) for a smaller percentage of islets to exhibit intrainsulitis and a significantly higher percentage of MTNOD islets that were small and disrupted, which we called atrophic islets. These findings do not indicate that MTNOD islets were under more severe immune attack than NOD islets; rather, they suggest that MTNOD islets were more easily injured. To assess the extent of islet cell injury, cleaved caspase 3 immunostaining was examined (Fig. 5C and D). As shown in Fig. 5C, after CYP injection, MTNOD islets had a significant increased number of caspase 3–stained cells, generally located in the β-cell–rich center of the islets compared with control NOD islets. Image analysis revealed that the average percentages of cleaved caspase 3–stained areas within an islet after CYP injection were 11.7 and 5.5% in MTNOD and NOD mice, respectively (Fig. 5D, left). Similar analysis also confirmed increased caspase 3 staining in islets of CatNOD mice after CYP treatment (Fig. 5D, right).
To explore the mechanism behind our in vivo findings, we needed an in vitro system that reproduced the enhanced in vivo sensitivity of transgenic islets. For this purpose, isolated MTNOD and control islets were cultured and exposed to a cytokine mix, containing IL-1β, interferon-γ, and TNF-α, which are important mediators for β-cell destruction in autoimmune type 1 diabetes (38). Consistent with our in vivo findings, MTNOD islets were more sensitive to cytokine toxicity than NOD control islets, as revealed by more activated caspase 3 expression (Fig. 6A) and a greater reduction of islet metabolic activity (Fig. 6B). Preliminary results suggest that CatNOD islets are also more sensitive to cytokine treatment (data not shown). This culture system allowed us to examine cell signaling in MTNOD islets. The Akt/PDX-1 pathway has been shown to be important for controlling pancreatic β-cell survival (39), and this pathway is known to be enhanced by ROS (40). Therefore, we assessed the possibility that cytoplasmic antioxidants could inhibit this path and thereby increase apoptosis in MTNOD islets. As shown in Fig. 6C, MTNOD islets showed less activation of the Akt/PDX-1 pathway following cytokine treatment than NOD islets. They contained less phosphorylated Akt, less phosphorylated Foxo-1, and less PDX-1 than islets from identically treated NOD islets. We examined if reduced activation of the Akt/PDX-1 pathway was also true in vivo. As shown in Fig. 7, islets isolated 7 days after CYP treatment from MTNOD and CatNOD mice showed similar changes as we observed using the in vitro system; phosphorylated Akt and Foxo-1 were reduced and there was less expression of PDX-1 than islets from identically treated NOD mice. Since this change might be the consequence of β-cell death 7 days after CYP injection, we further examined the changes 4 days after CYP treatment. A smaller but still evident reduction of phosphorylation or expression of these proteins was seen in MTNOD and CatNOD transgenic islets (Fig. 7B and D).
Several studies have shown that ROS enhancement of signaling is through oxidation-induced inhibition of PTP activity (19,41). Cytokines also increase intracellular levels of ROS (42,43). In control NOD islets, cytokine treatment elevated ROS, as indicated by increased CM-H2DCFDA fluorescence (Fig. 8A). MT is a potent scavenger of ROS, as we have previously reported (28), and ROS production was markedly reduced in MTNOD islets (Fig. 8A). In addition, assays of CatNOD islets showed reduced CM-H2DCFDA fluorescence after exposure to cytokines or hydrogen peroxide (data not shown). Scavenging of ROS should protect PTP from inactivation. Elevated PTP activity might play a role in MT-induced sensitization to cytokines. If that is correct, then the inhibitor of PTP, sodium orthovandate, should reduce sensitization. As shown in Fig. 8B and C, sodium orthovanadate protected both morphology and metabolic activity in MTNOD islets but had no consistent effect in control islets. Vanadate addition essentially eliminated the sensitization of MTNOD islets to cytokines.
DISCUSSION
In this study, three transgenic lines of mice with β-cell–specific overexpression of antioxidant proteins MT, catalase, or MnSOD were backcrossed onto the NOD background by the speed-congenic method. None of the transgenes had any detrimental effect on pre-diabetic NOD β-cell function, and the mitochondrial antioxidant, MnSOD, had no effect on the occurrence of diabetes. The cytoplasmic antioxidant, MT, was shown to have several beneficial effects, including protection from STZ-induced diabetes, protection of cultured islets against CYP exposure, and reduced ROS production induced by cytokines. However, MT and another cytoplasmic antioxidant, catalase, greatly accelerated diabetes onset after CYP injection and significantly accelerated spontaneous diabetes in male NOD mice. The accelerated diabetes onset coincided with increased β-cell–cleaved caspase 3 staining but did not appear to be due to increased immune attack. Cultured islets from MTNOD mice were more sensitive to cytokine injury. In vivo and in vitro evidence was obtained, suggesting a role for reduced activation of the Akt survival pathway, possibly due to inhibition of protein tyrosine phosphatase.
A major goal of this project was to determine whether ROS damage to the β-cell is a major contributor to type 1 diabetes in the NOD mouse. Systemic treatment of NOD mice with a number of antioxidants has been shown to prevent or delay diabetes in this model (7–9). However, systemic treatment does not reveal whether protection occurs at the level of the β-cell or the immune system. To distinguish between these sites of protection, it is necessary to target β-cells or immune cells. The work of Hotta et al. (13) showed that targeting of the antioxidant thioredoxin to β-cells did prevent NOD diabetes. However, it is now recognized that thioredoxin has potent antiapoptotic actions (14) in addition to its antioxidant actions. One other antioxidant, extracellular SOD, expressed specifically in β-cells, failed to reduce NOD diabetes (15). Because of these inconclusive results, we decided to comprehensively test the benefit of β-cell antioxidant protection against type 1 diabetes using three different β-cell antioxidant transgenes, each of which has previously been shown to be effective against ROS and STZ diabetes (5,6,26). Our results were clear; none of these β-cell antioxidants reduced the incidence of NOD diabetes. This suggests that prior reports of protection with systemic antioxidants reflect primary beneficial actions on the immune system. In fact, several of these studies (8,44) have reported striking reductions in immune activation. The immune system may be an attractive therapeutic target for antioxidant action, but the β-cell does not appear to be.
Unexpectedly, two of the antioxidant transgenes strongly accelerated NOD diabetes. Following CYP treatment, diabetes developed in most MTNOD mice within 10 days of the first injection. In control NOD mice, only 2 mice out of a total of 42 mice became diabetic after the first CYP injection. CatNOD mice also had a rapid onset after CYP treatment, though somewhat slower than MTNOD mice. Spontaneous onset of diabetes was less dramatically affected than CYP-induced diabetes. Presumably, the lesser effect was due to the fact that the major factor determining the onset of spontaneous diabetes is how long it takes before immune cells are stimulated to attack the islet rather than how long it takes β-cells to die (37). Since the β-cell transgenes did not alter the immune system, the time for diabetes onset was not greatly altered. Sensitization to spontaneous diabetes by the MT and catalase transgenes was evident in males but not in females. This is in contrast with sensitization to CYP-induced diabetes, which showed no sex effect. It has been recognized for 20 years in normal NOD mice that there is a significant impact of sex on spontaneous diabetes (45) but not on CYP-induced diabetes (46,47). The basis for the effect of sex on diabetes remains uncertain in both our antioxidant transgenic NOD mice and in normal NOD mice (47). In our transgenic NOD mice, the basis for the sex difference presumably resides in the β-cell where the transgenes are expressed. This is different from normal NOD mice, where the sex difference is thought to be a function of the immune system (47).
Thorough analysis of pre-diabetic islet structure and function revealed no abnormalities in MTNOD or CatNOD islets that could provide an explanation for their accelerated diabetes onset. In fact, MT islets were more robust than control islets when directly challenged with STZ or CYP. This demonstrated that before immune attack the transgenic β-cells were normal. It is also possible that transgenic expression of nonmouse proteins may have promoted increased immune attack. However, we could find no difference in peri- or intrainsulitis before or after diabetes onset, indicating that the acceleration of diabetes was probably not due to a more aggressive attack by the immune system. These results led us to the hypothesis that MT and catalase may prevent activation of a β-cell mechanism that protects from autoimmune-induced cell death. While a multitude of reports (48–50) demonstrate protective effects of antioxidants in various cell types, there are also several reports that antioxidants can sensitize cells to damage and apoptosis. However, the mechanism of antioxidant-induced sensitization remains speculative. Also, we cannot completely rule out a role for FVB genes tightly linked to catalase or MT transgenes. These may not have been eliminated by our speed congenic backcross methodology and could have produced unexpected effects on NOD diabetes, as has been previously reported (51).
We chose to analyze the Akt/PDX-1 pathway that has been implicated as an important survival pathway in the β-cell. Akt kinase is activated upon phosphorylation, leading to phosphorylation of a number of targets including Foxo-1. Phosphorylation of Foxo-1 releases inhibition of PDX-1 transcription. Genetic manipulations of the β-cell that increase activity in this pathway, such as overexpression of Akt (52,53) or PDX-1 (54), promote survival. Correspondingly, modifications that inhibit this pathway, such as reducing expression of PDX-1 (55), increase β-cell apoptosis. Prior analysis of the role of the Akt/PDX-1 pathway has been limited to the context of type 2 diabetes. Since β-cell survival is also a component of type 1 diabetes, we assessed whether differential activity of the Akt/PDX-1 pathway might explain the greater β-cell death and earlier onset of type 1 diabetes in MTNOD and CatNOD mice. Our results with islets isolated 4 or 7 days after CYP injection showed that MT and catalase reduced phosphorylation of Akt and Foxo-1 while decreasing expression of PDX-1. This shows that activity of the Akt/PDX-1 survival pathway is reduced in MTNOD and CatNOD islets following CYP injection. Because these changes can occur in islet protein phosphorylation during isolation, an in vitro system was set up to test islets from MTNOD and NOD mice challenged by treatment with cytokines. Using this culture system we saw that MTNOD islets produced much lower levels of ROS than NOD islets following cytokine treatment. Despite this, MTNOD islets were significantly more sensitive to cytokine-induced β-cell death, as indicated by significantly greater loss of metabolic activity, much more activation of caspase 3, and greater structural degeneration. As we saw in vivo, phosphorylation of Akt and Foxo-1 and expression of PDX-1 were reduced after cytokine treatment of MTNOD islets. These results are consistent with the recent finding of Yang et al. (56) that Akt activation protects β-cells from cytokine injury. They show a clear association between decreased activity of the Akt/PDX-1 survival pathway and reduced survival of MTNOD and CatNOD islets.
Activation of β-cell Akt occurs following ligand binding of insulin and IGF-1 receptors via tyrosine-phosphorylated intermediates. These pathways are downregulated by PTP-1B. PTP-1B, like other tyrosine phosphatases, can be readily inhibited by ROS. Oxidation prevents phosphatase activity, resulting in temporary enhancement of the insulin/IGF-1 activation of Akt. The reduced phosphorylation of Akt in MTNOD and CatNOD islets could occur because antioxidants prevent ROS inhibition of PTP-1B. If true, then inhibition of PTP-1B activity with vanadate should be more protective in MTNOD islets than in NOD islets. This was found to be the case, as vanadate was not or only slightly protective to NOD islets, but it was very protective for MTNOD islets. Since PTP-1B is cytoplasmic, this would also explain why two cytoplasmic antioxidants sensitized β-cells to apoptosis while the mitochondrial antioxidant MnSOD did not. We recognize that our evidence for the role of PTP-1B is indirect and preliminary; islets provide insufficient material for direct assays of PTP-1B. Other actions of MT and catalase, such as alterations in nuclear factor κB (50) or Jun kinase (48), could produce sensitization to β-cell death.
To summarize our conclusions, we found no evidence for reduction of NOD diabetes by β-cell–targeted expression of antioxidants. In fact, two cytoplasmic antioxidants speeded up spontaneous diabetes in males, greatly accelerated CYP-induced diabetes in both males and females, and sensitized isolated islets to cytokine injury. This sensitization may result from lower activation of the Akt/PDX-1 survival system.
X.L. and H.C. contributed equally to this study.
X.L. is currently affiliated with the Department of Biology, Amgen, South San Francisco, California. H.C. is currently affiliated with the Department of Developmental Biology, Stanford University, Stanford, California.
Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org.
DOI: 10.2337/db05-1357
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Article Information
This work was supported by grants DK58100 and HL075080 and the Commonwealth of Kentucky Research Challenge Trust Fund and the University of Louisville Center for Genetics and Molecular Medicine.
We thank Patricia Kralik for help in setting up the cleaved caspase 3 studies.