Repair of cytokine-induced DNA damage in cultured rat islets of Langerhans

Treatment of cultured rat pancreatic islets of Langerhans with the combined cytokines interleukin-1beta (IL-1beta), interferon gamma (IFN gamma) and tumour necrosis factor alpha (TNF alpha) leads to DNA damage including strand breakage. We have investigated the nature of this damage and its repairability. When islets are further incubated for 4 h in fresh medium, the level of cytokine-induced strand breakage remains constant. If the nitric oxide synthase inhibitor N(G)-monomethyl-L-arginine (NMMA) is present during cytokine treatment, then strand breakage is prevented. If NMMA is added following, rather than during,the cytokine treatment and islets are incubated for 4 h, further nitric oxide synthesis is prevented and most cytokine-induced strand breaks are no longer seen. To investigate DNA repair following cytokine treatment, cells were transferred to fresh medium and incubated for 4 h in the presence of hydroxyurea (HU) and 1-beta-D-arabinosyl cytosine (AraC), as inhibitors of strand rejoining. In the presence of these inhibitors there was an accumulation of strand breaks that would otherwise have been repaired. However, when further nitric oxide synthesis was inhibited by NMMA, significantly less additional strand breakage was seen in the presence of HU and AraC. We interpret this, as indicating that excision repair of previously induced base damage did not contribute significantly to strand breakage. Levels of oxidised purines, as indicated by formamidopyrimidine glycosylase (Fpg) sensitive sites, were not increased in cytokine-treated islets. We conclude that in these primary insulin-secreting cells: (a) the DNA damage induced by an 18h cytokine treatment is prevented by an inhibitor of nitric oxide synthase, (b) much of the damage is in the form of apparent strand breaks rather than altered bases such as oxidised purines, (c) substantial repair is ongoing during the cytokine treatment and this repair is not inhibited in the presence of nitric oxide.     

Nitric Oxide Induces Ataxia Telangiectasia Mutated (ATM) Protein-dependent γH2AX Protein Formation in Pancreatic β Cells

In this study, the effects of cytokines on the activation of the DNA double strand break repair factors histone H2AX (H2AX) and ataxia telangiectasia mutated (ATM) were examined in pancreatic β cells. We show that cytokines stimulate H2AX phosphorylation (γH2AX formation) in rat islets and insulinoma cells in a nitric oxide- and ATM-dependent manner. In contrast to the well documented role of ATM in DNA repair, ATM does not appear to participate in the repair of nitric oxide-induced DNA damage. Instead, nitric oxide-induced γH2AX formation correlates temporally with the onset of irreversible DNA damage and the induction of apoptosis. Furthermore, inhibition of ATM attenuates cytokine-induced caspase activation. These findings show that the formation of DNA double strand breaks correlates with ATM activation, irreversible DNA damage, and ATM-dependent induction of apoptosis in cytokine-treated β cells.     

Insulin Secretion, DNA Damage, and Apoptosis in Human and Rat Islets of Langerhans Following Exposure to Nitric Oxide, Peroxynitrite, and Cytokines

Cytokine-induced damage may contribute to destruction of insulin-secreting beta-cells in islets of Langerhans during autoimmune diabetes. There is considerable controversy (i) whether human and rat islets respond differently to cytokines, (ii) the extent to which cytokine damage is mediated by induction of nitric oxide formation, and (iii) whether the effects of nitric oxide on islets can be distinguished from those of reactive oxygen species or peroxynitrite. We have analyzed rat and human islet responses in parallel, 48 h after exposure to the nitric oxide donor S-nitrosoglutathione, the mixed donor 3-morpholinosydnonimine, hypoxanthine/xanthine oxidase, peroxynitrite, and combined cytokines (interleukin-1beta, tumor necrosis factor-alpha and interferon-gamma). Insulin secretory response to glucose, insulin content, DNA strand breakage, and early-to-late stage apoptosis were recorded in each experiment. Rat islet insulin secretion was reduced by S-nitrosoglutathione or combined cytokines, but unexpectedly increased by peroxynitrite or hypoxanthine/xanthine oxidase. Effects on human islet insulin secretion were small; cytokines and S-nitrosoglutathione decreased insulin content. Both rat and human islets showed significant and similar levels of DNA damage following all treatments. Apoptosis in neonatal rat islets was increased by every treatment, but was at a low rate in adult rat or human islets and only achieved significance with cytokine treatment of human islets. All cytokine responses were blocked by an arginine analogue. We conclude: (i) Reactive oxygen species increased and nitric oxide decreased insulin secretory responsiveness in rat islets. (ii) Species differences lie mainly in responses to cytokines, applied at a lower dose and shorter time than in most studies of human islets. (iii) Cytokine effects were nitric oxide driven; neither reactive oxygen species nor peroxynitrite reproduced cytokine effects. (iv) Rat and human islets showed equal susceptibility to DNA damage. (v) Apoptosis was not the preferred death pathway in adult islets. (vi) We have found no evidence of human donor variation in the pattern of response to these treatments.     

Induction and repair of DNA strand breaks in cultured human fibroblasts exposed to various phenols and dihydrodiols of benzo[a]pyrene

Cultured human fibroblasts from healthy donors were incubated for 30 min with nine different benzo[a]pyrene (BP) derivatives in the presence or absence of liver microsomes from 3-methylcholanthrene treated rats. The induction and repair of DNA strand breaks were analysed by alkaline unwinding and separation of double and single stranded DNA (SS-DNA) by hydroxylapatite chromatography immediately after the incubation or at various times after the treatment. In the absence of microsomes DNA stand breaks were detected in fibroblasts exposed to 30 microM of each of the six BP phenols (1-, 2-, 3-, 7-, 9- or 11-OH-BP) and the three BP dihydrodiols (BP-4,5-, BP-7,8- or BP-9,10-dihydrodiol). After removal of the BP derivatives from the medium the DNA strand breaks disappeared within 24 h. alpha-Naphthoflavone (alpha-NF) caused a decrease in the induction of strand breaks by 1-, 3- and 9-OH-BP but did not affect the induction of strand breaks in cells exposed to BP-7,8-dihydrodiol. In the presence of microsomes DNA strand breaks were found after exposure to 30 microM of each of the six BP phenols (1-, 2-, 3-, 7-, 9- or 11-OH-BP), as well as BP-7,8- and 9,10-dihydrodiol. In contrast BP-4,5-dihydrodiol did not induce strand breaks under these conditions. The induction of strand breaks by BP-7,8-dihydrodiol was enhanced in the presence of cytosine-1-beta-D-arabinofuranoside (AraC). In all cases the DNA strand breaks had disappeared 24 h after removal of the BP derivatives and microsomes except after treatment with BP-7,8-dihydrodiol.     

Different effects of tert-butylhydroperoxide-induced peroxynitrite-dependent and -independent DNA single-strand breakage on PC12 cell poly(ADP-ribose) polymerase activity.

The short-chain lipid hydroperoxide analogue tert-butylhydroperoxide induces peroxynitrite-dependent and -independent DNA single strand breakage in PC12 cells. U937 cells that do not express constitutive nitric oxide synthase respond to tert-butylhydroperoxide treatment with peroxynitrite-independent DNA cleavage. Under experimental conditions leading to equivalent strand break frequencies, the analysis of poly(ADP-ribose) polymerase activity showed an increase in PC12 cells but not in U937 cells. The enhanced poly(ADP-ribose) polymerase activity observed in PC12 cells was paralleled by a significant decline in NAD+ content and both events were prevented by treatments suppressing formation of peroxynitrite. Although DNA breaks were rejoined at similar rates in the two cell lines, an inhibitor of poly(ADP-ribose) polymerase delayed DNA repair in PC12 cells but had hardly any effect in U937 cells. The results obtained using the latter cell type were confirmed with an additional cell line (Chinese hamster ovary cells) that does not express nitric oxide synthase. Collectively, our data suggest that tert-butylhydroperoxide-induced peroxynitrite-independent DNA strand scission is far less effective than the DNA cleavage generated by endogenous peroxynitrite in stimulating the activity of poly(ADP-ribose) polymerase.     

Singlet oxygen-mediated damage to cellular DNA determined by the comet assay associated with DNA repair enzymes

The damage profile produced by the reaction of singlet molecular oxygen with cellular DNA was determined using the comet assay associated with DNA repair enzymes. Singlet oxygen was produced intracellularly by thermal decomposition of a water-soluble endoperoxide of a naphthalene derivative which is able to penetrate through the membrane into mammalian cells. We found that the DNA modifications produced by singlet oxygen were almost exclusively oxidised purines recognised by the formamidopyrimidine DNA N-glycosylase. In contrast, significant amounts of direct strand breaks and alkali-labile sites or oxidised pyrimidines, detectable by the bacterial endonuclease III, were not produced.     

Antioxidant and pro-oxidant actions of flavonoids: effects on DNA damage induced by nitric oxide, peroxynitrite and nitroxyl anion

Antioxidant and pro-oxidant activities of flavonoids have been reported. We have studied the effects of 18 flavonoids and related phenolic compounds on DNA damage induced by nitric oxide (NO), peroxynitrite, and nitroxyl anion (NO⁻). Similarly to our previous findings with catecholamines and catechol-estrogens, DNA single-strand breakage was induced synergistically when pBR322 plasmid was incubated in the presence of an NO-releasing compound (diethylamine NONOate) and a flavonoid having an ortho-trihydroxyl group in either the B ring (e.g., epigallocatechin gallate) or the A ring (e.g., quercetagetin). Either NO or any of the above flavonoids alone did not induce strand breakage significantly. However, most of the tested flavonoids inhibited the peroxynitrite-mediated formation of 8-nitroguanine in calf-thymus DNA, measured by a new HPLC-electrochemical detection method, as well as the peroxynitrite-induced strand breakage. NO⁻ generated from Angeli’s salt caused DNA strand breakage, which was also inhibited by flavonoids but at only high concentrations. On the basis of these findings, we propose that NO⁻ and/or peroxynitrite could be responsible for DNA strand breakage induced by NO and a flavonoid having an ortho-trihydroxyl group. Our results indicate that flavonoids have antioxidant properties, but some act as pro-oxidants in the presence of NO.     

Oxidative stress mediates neuronal DNA damage and apoptosis in response to cytosine arabinoside

Cytosine arabinoside (AraC) is a nucleoside analog that produces significant neurotoxicity in cancer patients. The mechanism by which AraC causes neuronal death is a matter of some debate because the conventional understanding of AraC toxicity requires incorporation into newly synthesized DNA. Here we demonstrate that AraC-induced apoptosis of cultured cerebral cortical neurons is mediated by oxidative stress. AraC-induced cell death was reduced by treatment with several different free-radical scavengers (N-acetyl-L-cysteine, dipyridamole, uric acid, and vitamin E) and was increased following depletion of cellular glutathione stores. AraC induced the formation of reactive oxygen species in neurons as measured by an increase in the fluorescence of the dye 5-(6)-carboxy-2',7'-dichlorodihydrofluorescein diacetate. AraC produced DNA single-strand breaks as measured by single-cell gel electrophoresis and the level of DNA strand breakage was reduced by treatment with the free radical scavengers. These data support a model in which AraC induces neuronal apoptosis by provoking the generation of reactive oxygen species, causing oxidative DNA damage and initiating the p53-dependent apoptotic program. These observations suggest the use of antioxidant therapies to reduce neurotoxicity in AraC chemotherapeutic regimens.     

Effects of tannic acid and its related compounds on food mutagens or hydrogen peroxide-induced DNA strands breaks in human lymphocytes

The effect of tannic acid (TA), gallic acid (GA), propyl gallate (PA) and ellagic acid (EA) on DNA damage in human lymphocytes induced by food mutagens [3-amino-1-methyl-5H-pyrido (4,3-b) indole (Trp-P-2) and 2-amino-1-methyl-6-phenylimadazo (4,5-b) pyridine (PhIP) or H₂O₂ was evaluated by using single-cell electrophoresis (comet assay). The toxicity of these tested compounds (0.1–100 μg/ml) on lymphocytes was not found. These compounds did not cause DNA strand breaks at lower concentrations of 0.1–10 μg/ml. At a concentration of 100 μg/ml, TA and GA exhibited slight DNA damage, whereas PA and EA showed no DNA strand breaks. TA and its related compounds decreased the DNA strand breaks induced by Trp-P-2, PhIP or H₂O₂ at concentrations of 0.1–10 μg/ml. DNA repair enzymes endonuclease III (Endo III) and formamidopyrimidine-DNA glycoslase (FPG)] were used to examine the levels of oxidised pyrimidines and purines in human lymphocytes induced by H₂O₂. All the compounds at 10 μg/ml can reduce the level of FPG sensitive sites. However, only EA inhibited the formation of EndoIII sensitive sites. The results indicated that these compounds can enhance lymphocytes resistance towards DNA strand breaks induced by food mutagens or H₂O₂ in vitro.     

Induction and repair of DNA strand breaks in Bacteroides fragilis

Alkaline sucrose gradient sedimentation was used to establish whether strand breakage and repair take place in the DNA of UV-irradiated Bacteroides fragilis during the removal of pyrimidine dimers. A B. fragilis wild-type strain and two of its repair mutants, a mitomycin C sensitive mutant (MTC25) having wild-type levels of UV survival, and a UV-sensitive, mitomycin C sensitive mutant (UVS9), were investigated. Under anaerobic conditions, far-UV irradiation induced metabolically regulated strand breakage and resynthesis in the wild-type strain, but this was markedly reduced in both the MTC25 and UVS9 mutants. Approximately half of the strand breaks generated by the various strains were rejoined during further holding in buffer. Under replicating conditions, complete repair of strand breaks in the wild type was observed. Caffeine treatment under anaerobic conditions caused direct DNA strand breakage in B. fragilis cells but did not inhibit UV-induced breakage or repair.     

Effect of age on endogenous DNA single-strand breakage, strand break induction and repair in the adult housefly, Musca domestica

It has been suggested that genomic alterations involving DNA damage and the ability to repair such damage play an important role in cellular senescence. In this study, endogenous DNA single-strand breaks, the susceptibility of DNA to induced strand breakage and the capacity to repair these breaks were compared in postmitotic cells from young (3-day-old) and old (23-day-old) houseflies. DNA single-strand breaks did not accumulate during normal aging in the housefly. However, cells of the old flies exhibited a greater sensitivity to single-strand breakage induced by gamma-radiation and UV light. The capacity to repair these exogenously induced single-strand breaks declined with age. Results do not support the view that DNA single-strand breaks are a causal factor in aging in the housefly. An age-related increase in the susceptibility to undergo single-strand breakage suggests alterations in chromatin during the aging process.     

Measuring oxidative damage to DNA and its repair with the comet assay

Background

Single cell gel electrophoresis, or the comet assay, was devised as a sensitive method for detecting DNA strand breaks, at the level of individual cells. A simple modification, incorporating a digestion of DNA with a lesion-specific endonuclease, makes it possible to measure oxidised bases.

Scope of review

With the inclusion of formamidopyrimidine DNA glycosylase to recognise oxidised purines, or Nth (endonuclease III) to detect oxidised pyrimidines, the comet assay has been used extensively in human biomonitoring to monitor oxidative stress, usually in peripheral blood mononuclear cells.

Major conclusions

There is evidence to suggest that the enzymic approach is more accurate than chromatographic methods, when applied to low background levels of base oxidation. However, there are potential problems of over-estimation (because the enzymes are not completely specific) or under-estimation (failure to detect lesions that are close together). Attempts have been made to improve the inter-laboratory reproducibility of the comet assay.

General significance

In addition to measuring DNA damage, the assay can be used to monitor the cellular or in vitro repair of strand breaks or oxidised bases. It also has applications in assessing the antioxidant status of cells. In its various forms, the comet assay is now an invaluable tool in human biomonitoring and genotoxicity testing. This article is part of a Special Issue entitled Current methods to study reactive oxygen species - pros and cons and biophysics of membrane proteins. Guest Editor: Christine Winterbourn.     

Interaction between radiation and drug damage in mammalian cells. V. DNA damage and repair induced in LZ cells by adriamycin and/or radiation

A multi-drug-resistant cell line selected in increasing concentrations of Adriamycin and designated LZ (J. A. Belli, Radiat. Res. 119, 88-100, 1989) is shown to exhibit a survival response characterized by radiation sensitivity and Adriamycin resistance. To determine if this response is due to alterations in either the initial levels of damage induced or the repair of DNA damage, LZ cells and the parental V79 cells were exposed to either radiation or Adriamycin and the damage and repair were measured with alkaline or nondenaturing filter elution. After exposure to radiation, induction and repair of both single-strand and double-strand breaks were equivalent. LZ cells exposed to 100 micrograms/ml Adriamycin for 1 h contained no measurable damage while the same treatment induced breaks and crosslinks in V79 cells. Pretreatment of LZ cells for 1 h with Adriamycin before irradiation did not alter either the initial levels of induced damage or the repair of strand breakage. These results suggest that (1) mechanisms other than differential induction and repair of strand breaks are responsible for the increased radiation sensitivity in LZ, and (2) the lack of Adriamycin-induced DNA damage in LZ is at least partially responsible for the increased cell survival after treatment.     

Are glutathione S transferases involved in DNA damage signalling? Interactions with DNA damage and repair revealed from molecular epidemiology studies

Glutathione S-transferases (GSTs) are members of a multigene family of isoenzymes that are important in the control of oxidative stress and in phase II metabolism. Acting non-enzymically, GSTs can modulate signalling pathways of cell proliferation, cell differentiation and apoptosis. Using a molecular epidemiology approach, we have investigated a potential involvement of GSTs in DNA damage processing, specifically the modulation of DNA repair in a group of 388 healthy adult volunteers; 239 with at least 5 years of occupational exposure to asbestos, stone wool or glass fibre, and 149 reference subjects. We measured DNA damage in lymphocytes using the comet assay (alkaline single cell gel electrophoresis): strand breaks (SBs) and alkali-labile sites, oxidised pyrimidines with endonuclease III, and oxidised purines with formamidopyrimidine DNA glycosylase. We also measured GST activity in erythrocytes, and the capacity for base excision repair (BER) in a lymphocyte extract. Polymorphisms in genes encoding three GST isoenzymes were determined, namely deletion of GSTM1 and GSTT1 and single nucleotide polymorphism Ile105Val in GSTP1. Consumption of vegetables and wine correlated negatively with DNA damage and modulated BER. GST activity correlated with oxidised bases and with BER capacity, and differed depending on polymorphisms in GSTP1, GSTT1 and GSTM1. A significantly lower BER rate was associated with the homozygous GSTT1 deletion in all asbestos site subjects and in the corresponding reference group. Multifactorial analysis revealed effects of sex and exposure in GSTP1 Ile/Val heterozygotes but not in Ile/Ile homozygotes. These variants affected also SBs levels, mainly by interactions of GSTP1 genotype with exposure, with sex, and with smoking habit; and by an interaction between sex and smoking. Our results show that GST polymorphisms and GST activity can apparently influence DNA stability and repair of oxidised bases, suggesting a potential new role for these proteins in DNA damage processing via DNA damage signalling.     

Arsenite induces oxidative DNA adducts and DNA-protein cross-links in mammalian cells.

Arsenic is generally recognized as a nonmutagenic carcinogen because sodium arsenite induces DNA damage only at very high concentrations. In this study we demonstrate that arsenite concentrations above 0.25 microM induce DNA strand breaks in both human leukemia cells and Chinese hamster ovary cells. Therefore, DNA damage may be involved in arsenic-induced carcinogenesis. Formamidopyrimidine-DNA glycosylase and proteinase K greatly increased DNA strand breaks in arsenite-treated cells, providing evidence that a large portion of arsenite-induced DNA strand breaks come from excision of oxidative DNA adducts and DNA-protein cross-links. Because DNA strand breaks appear only temporarily during excision repair, the level of detectable DNA strand breaks will be low at any given time point. For this reason many previous studies have only detected low levels of DNA strand breaks. We also show that catalase, and inhibitors of calcium, nitric oxide synthase, superoxide dismutase, and myeloperoxidase, could modulate arsenite-induced DNA damage. We conclude that arsenite induces DNA adducts through calcium-mediated production of peroxynitrite, hypochlorous acid, and hydroxyl radicals.     

Genotoxicity of formaldehyde and an evaluation of its effects on the DNA repair process in human diploid fibroblasts

Formaldehyde treatment of human fibroblasts gave rise to DNA damage detected by a nick translation assay. This damage was not repaired by typical 'long-patch'-type excision repair as evidenced by the failure of DNA repair inhibitor post-treatment to elevate the amount of DNA strand breakage. In addition, the effects of formaldehyde on DNA repair were examined in light of a recent report suggesting that formaldehyde inhibited the repair of X-ray-induced strand breaks and UV- and benzo [a]pyrene diol epoxide-induced unscheduled DNA synthesis in human bronchial cells. We report that formaldehyde (1) was ineffective at inhibiting the sealing of X-ray- or bleomycin-induced DNA strand breaks, (2) did not inhibit the removal of pyrimidine dimers from cellular DNA at short treatment times, and (3) that the previously observed inhibition of unscheduled DNA synthesis was most likely due to the inhibition of uptake of labeled precursor into formaldehyde-treated cells. Thus, our findings are not consistent with the notion that formaldehyde inhibits the repair process in human fibroblasts. Finally, formaldehyde was shown to elevate the level of misincorporation of bases into synthetic polynucleotides catalyzed by E. coli DNA polymerase I, indicating that the mutagenicity of formaldehyde may be due to covalent alteration of DNA bases.     

Role of cyclooxygenase-2 in cytokine-induced beta-cell dysfunction and damage by isolated rat and human islets

Type I diabetes mellitus is an autoimmune disease characterized by the selective destruction of the insulin-secreting beta-cell found in pancreatic islets of Langerhans. Cytokines such as interleukin-1 (IL-1), interferon-gamma (IFN-gamma), and tumor necrosis factor-alpha (TNF-alpha) mediate beta-cell dysfunction and islet degeneration, in part, through the induction of the inducible isoform of nitric-oxide synthase and the production of nitric oxide by beta-cells. Cytokines also stimulate the expression of the inducible isoform of cyclooxygenase, COX-2, and the production of prostaglandin E(2) (PGE(2)) by rat and human islets; however, the role of increased COX-2 expression and PGE(2) production in mediating cytokine-induced inhibition of islet metabolic function and viability has been incompletely characterized. In this study, we have shown that treatment of rat islets with IL-1beta or human islets with a cytokine mixture containing IL-1beta + IFN-gamma +/- TNF-alpha stimulates COX-2 expression and PGE(2) formation in a time-dependent manner. Co-incubation of rat and human islets with selective COX-2 inhibitors SC-58236 and Celecoxib, respectively, attenuated cytokine-induced PGE(2) formation. However, these inhibitors failed to prevent cytokine-mediated inhibition of insulin secretion or islet degeneration. These findings indicate that selective inhibition of COX-2 activity does not protect rat and human islets from cytokine-induced beta-cell dysfunction and islet degeneration and, furthermore, that islet production of PGE(2) does not mediate these inhibitory and destructive effects.     

Synergistic induction of hydroxyl radical-induced DNA single-strand breaks by chromium(VI) compound and cigarette smoke solution

Chromium(VI) compounds and cigarette smoke are known human carcinogens. We found that K2Cr2O7 and cigarette smoke solution synergistically induced DNA single-strand breaks (0.23+/-0.04 breaks per DNA molecule) in pUC118 plasmid DNA. K2Cr2O7 alone or cigarette smoke solution alone induced much less strand breaks (0.03+/-0.01 or 0.07+/-0.02 breaks per DNA molecule, respectively). The synergistic effect was prevented by catalase and by hydroxyl radical scavengers such as deferoxamine, dimethylsulfoxide, d-mannitol, and Tris, but not by superoxide dismutase. Ascorbic acid enhanced the synergism. Glutathione inhibited strand breakage only at high concentrations. Electron spin resonance (ESR) studies using a hydroxyl radical trap demonstrated that hydroxyl radicals were generated when DNA was incubated with K2Cr2O7 and cigarette smoke solution. Hydroxyl radical adduct decreased dose-dependently when strand breakage was prevented by catalase, deferoxamine, dimethylsulfoxide, d-mannitol or Tris, but not significantly by superoxide dismutase. We also used ESR spectroscopy to study the effects of different concentration of ascorbic acid and glutathione. The results showed that hydroxyl radical, which is proposed as a main carcinogenic mechanism for both chromium(VI) compounds and cigarette smoke solution was mainly responsible for the DNA breaks they induced.     

Superoxide, nitric oxide, peroxynitrite and cytokine combinations all cause functional impairment and morphological changes in rat islets of Langerhans and insulin secreting cell lines, but dictate cell death by different mechanisms

We have shown that nitric oxide treatment for 30–90 min causes inhibition of insulin secretion, DNA damage and disturbs sub-cellular organization in rat and human islets of Langerhans and HIT-T15 cells. Here rat islets and beta-cell lines were treated with various free radical generating systems S-nitrosoglutathione (nitric oxide), xanthine oxidase plus hypoxanthine (reactive oxygen species), 3-morpholinosydnonimine (nitric oxide, super-oxide, peroxynitrite, hydrogen peroxide) and peroxynitrite and their effects over 4 h to 3 days compared with those of the cytokine combination interleukin-1, tumour necrosis factor- and interferon-. End points examined were de novo protein synthesis, cellular reducing capacity, morphological changes and apoptosis by acridine orange cytochemistry, DNA gel electrophoresis and electron microscopy. Treatment (24–72 h) with nitric oxide, superoxide, peroxynitrite or combined cytokines differentially decreased redox function and inhibited protein synthesis in rat islets of Langerhans and in insulin-containing cell lines; cytokine effects were arginine and nitric oxide dependent. Peroxynitrite gave rare apoptosis in HIT-T15 cells and superoxide gave none in any cell type, but caused the most beta cell-specific damage in islets. S-nitroso-glutathione was the most effective agent at causing DNA laddering or chromatin margination characteristic of apoptotic cell death in insulin-containing cells. Cytokine-induced apoptosis was observed specifically in islet beta cells, combined cytokine effects on islet function and death most resembled those of the mixed radical donor SIN-1.