Perspectives in Diabetes Does Nitric Oxide Mediate Autoimmune Destruction of P-Cells? Possible Therapeutic Interventions in l DDM JOHN A. CORBEll AND MICHAEL L. MCDANIEL

Cytokines have been implicated as immunological effector molecules that induce dysfunction and destruction of the pancreatic p-cell. The mechanisms of cytokine action on the p-ceii are unknown; however, nitric oxide, resulting from cytokine-induced expression of nitric oxide synthase, has been implicated as the ceiiular effector molecule mediating

mouse. This antidiabetogenic effect of cytokines appears to conflict with evidence suggesting that cytokines mediate &cell dysfunction. The role of nltric oxide in both cytokine-mediated p-cell dysfunction, and the antidiabetogenic effects of cytokines, as well as the potential therapeutic use of aminoguanidine, are evaluated in this study. Diabetes 41 :897-903, 1992

p-ceii dysfunction. Nitric oxide is a free radical that targets intraceiiular iron-containing enzymes, which resuits in the loss of their function ~hecytokine iL-1 p induces the formation of nltric oxide in isolated rat islets and the insuiinoma ceii line, Rin-m5F. NMMA and NAME, both inhibitors of nitric oxide synthase, compieteiy protect islets from the deleterious effects of IL-1 p. These inhibitors are competitive in nature and inhibit both the cytokine-inducible and constitutive isoforms of nitric oxide synthase wlth nearly identical kinetics. This may preclude their use as therapeutic agents because of increases in blood pressure which resuit from the inhibition of constitutive nltric oxide synthase activity. Aminoguanidine, an inhibitor of nonenzymatic giycosyiation of ceiiuiar and extraceiiuiar constituents associated wlth diabetic compiications, recently has been reported to inhibit nitric oxide synthase. Aminoguanidine is -40-fold more effective in inhibiting the inducible isoform of nitric oxide synthase, suggesting that aminoguanidine or analogues may serve as potential therapeutic agents to block diseases associated with nitric oxide production by the inducible isoform of nitric oxide synthase. in vivo administration of TNF iL-1 has been shown to induce antidlabetogenic effects in the NOD

From the Department of Pathology, Washington University School of Medi- cine, Saint Louis, Missouri.

Address correspondence and reprint requests to Michael L. McDaniel, Department of Pathology, Box 81 18, Washington University School of Medi- cine, 660 South Euclid Avenue, Saint Louis, MO 631 10.

Received for publication 11 March 1992 and accepted in revised form 1 April 1992.

IL-lp, interleukin-lp; NMMA, No-rnonornethyl-L-arginine; NAME, No-nitro L-arginine methyl ester; TNF, tumor necrosis factor; IL-1, interleukin-I; NOD, nonobese spontaneously diabetic; IDDM, insulin-dependent diabetes melli- tus; IFN-r, interferon; CFA, complete Freund's adjuvant; STZ, streptozocin.

ytokines have been implicated as immunologi- cal effector molecules that mediate B-cell dvs- function and destruction associated with IDDM. IDDM is characterized by lymphocytic infiltra-

tion of the pancreatic islet, followed by p-cell death (1). In animal models of immune-mediated diabetes, macro- phages have been demonstrated to be among the first inflammatory cells to infiltrate the islet (2,3). These mac- rophages express class II MHC antigens and secrete cytokines, including IL-1 and TNF. Using a crude cyto- kine preparation isolated from human peripheral blood mononuclear cells, Mandrup-Poulsen et al. (4) demon- strated that such a preparation affected rat islet function and viability in a concentration- and time-dependent fashion (5). IL-1 antibodies or removal of IL-1 from the crude cytokine preparation by affinity chromatography prevented these inhibitory effects, suggesting that IL-1 was required for cytokine-mediated inhibition of insulin secretion from rat islets (6). Pretreatment of isolated rat islets for 18-24 h with human recombinant IL-1 resulted in an inhibition of glucose-stimulated insulin secretion, protein biosynthesis, and islet-cell replication (7-11). Under these conditions, IL-1-induced impairment of p-cell function was followed by p-cell death in a concen- tration- and time-dependent fashion (10). These results led Mandrup-Poulsen et al. (1 0) to propose a model that suggests a triggering event occurs in a single islet and that it requires the presence of both macrophages and helper T-cells. This triggering event leads to the release, processing, and presentation of p-cell antigens, resulting

DIABETES, VOL. 41, AUGUST 1992 897

in the initiation of the immune response. The immune response results in an increase in the local concentration of cytokines IL-1, TNF, and IFN-y, which leads to the destruction of the p-cell via the inhibition of mitochondrial function. This is possibly caused by the production of oxygen free radicals, which function as secondary medi- ators. The process continues until the islet is devoid of p-cells. In this study, we examine the effects of cytokines on p-cell function and viability, and we propose a cellular effector mechanism by which cytokine-induced p-cell death requires the induction of the enzyme nitric oxide synthase, which catalyzes the formation of the free radi- cal nitric oxide. We also propose novel strategies for the prevention of or intervention in disease states (such as IDDM) associated with nitric oxide production.

CYTOKINES Cytokines are small antigen-nonspecific glycoproteins (10-30 kDa) that are synthesized and rapidly secreted by a variety of different cells in response to numerous stimuli. Cytokines produce many effects on their target cells, including the promotion of cell proliferation and gene induction. The specific cytokines required to induce p-cell dysfunction and destruction depend on the spe- cies being studied. IL-1 alone appears to be sufficient to induce p-cell dysfunction and destruction in the rat (10-12), whereas TNF, which alone has no apparent affect on islet function or viability, functions in a synergis- tic fashion with IL-1 (1 0-12). The synergistic effect of TNF and IL-1 appears to be a common theme in many cytokine-mediated effector systems. In the mouse, IFN-y and TNF appear to produce p-cell death (13); this same cytokine combination also appears to be cytotoxic to human islet cells (14,151). These results are striking because they illustrate the different cytokine require- ments among the mouse, rat, and human species in the mechanism of p-cell destruction.

Paradoxically, in vivo injections of TNF or IL-1 also produce antidiabetogenic effects in the NOD mouse (1 6). The mechanism by which these cytokines produce such variable effects is unknown. We propose that the antidi- abetogenic effects observed with in vivo administration of this cytokine may be the result of macrophage-mediated nitric oxide formation. L-arginine metabolism through the nitric oxide synthase pathway appears to mediate sup- pressor macrophage activity of mitogen-induced rat or mouse lymphocyte proliferation in vitro (17). In these studies, Mills (1 7) also has shown that inhibitors of nitric oxide synthase augment mitogen-induced lymphocyte proliferation. TNF induces nitric oxide formation by mac- rophages in vitro, suggesting that the antidiabetogenic effects of TNF in the NOD mouse may be caused by TNF-induced nitric oxide generation by macrophages, which results in the inhibition of lymphocyte proliferation. Furthermore, CFA also has been shown recently to prevent insulitis and diabetes in the NOD mouse (18). The prevention of diabetes was associated with an inhibition of lymphocyte proliferation. CFA contains en- dotoxins known to stimulate nitric oxide production by macrophages, further suggesting that nitric oxide pro-

L-ARGININE ,-b NITRIC OXIDE + L-CITRULLINE NITRIC OXIDE NO. SYNTHASE

4 L-NMMA

(COMPETITIVE INHIBITOR)

NO; + NO - 3

FIG. 1. L-arglnlne-dependent nlrlc axlde pathway.

duction may inhibit lymphocyte proliferation and thereby prevent diabetes. However, other defects in the NOD mouse (e.g, cytokine release) also might participate in the process.

BIOCHEMICAL MECHANISM OF IL-1 MEDIATED p-CELL DYSFUNCTION Oxygen radicals. Highly reactive oxygen free radicals have been suggested to function as secondary effector molecules that may mediate p-cell dysfunction and de- struction (1 0,19). It has been proposed that the oxygen free radicals, superoxide and hydroxyl radical, are liber- ated either by infiltrating macrophages or that cytokines secreted by infiltrating lymphocytic cells induce the for- mation of oxygen free radicals within the p-cell mitochon- dria (1 0). Islets appear to be sensitive to oxygen radical formation because of low concentrations of mitochondrial Mn superoxide dismutase (20) and glutathione peroxi- dase (21)-endogenous enzymatic scavengers of oxy- gen free radicals that convert the highly reactive superoxide anion to hydrogen peroxide, water, and mo- lecular oxygen. Desferrioxamine, an agent that binds iron, which prevents the formation of oxygen free radicals via the Fenton reaction, reduces the incidence of diabe- tes associated with multiple low-dose STZ injections (22). Free radical scavengers also have been used to prevent the inhibitory effects of cytokines on insulin secretion in vitro (23). Although controversial, nicotinamide, which at high concentrations functions as an oxygen free radical scavenger, has been reported to partially prevent or delay the onset of IDDM in children predisposed to diabetes based on the presence of islet-cell antibodies (24). In vitro free radical scavengers have been used to prevent the inhibitory effects of cytokines on p-cell func- tion; however, complete protection has never been achieved by any of these types of interventions (10,19). Nitric oxide. Nitric oxide is the product of the mixed functional oxidation of one of the guanidino nitrogens of L-arginine to L-citrulline by nitric oxide synthase (Fig. 1; 25,26). At least two different isoforms of nitric oxide synthase (constitutive and cytokine inducible) have been characterized. These isoforms are differentiated by gene expression and cofactor requirements. The constitutive isoform of nitric oxide synthase is a Ca2+ and calmodulin- dependent enzyme (27,28), while expression of the

DIABETES, VOL. 41, AUGUST 1992

other isoform is inducible by cytokines, and the activity of this isoform is Ca2+ and calmodulin independent (29,30). Both enzymes are flavoproteins containing bound FAD and FMN, are NADPH dependent, and require tetrahy- drobiopterin as an additional cofactor (28,29). The con- stitutive enzyme is present in endothelium, brain, platelets, and recently has been shown to exist in islets (28,31,31a). Endotoxin, TNF, IFN-.I, and IL-1 have been shown to induce the expression of the inducible isoform of nitric oxide synthase in macrophages, smooth muscle, microglial cells, mesangial cells, hepatocytes, and p-cells of pancreatic islets (28,30,32).

The biological functions of nitric oxide differ depending on the isoform of nitric oxide synthase. The constitutive enzyme releases low levels of nitric oxide in response to physical stimuli or receptor activation. Nitric oxide re- leased under these conditions functions as a signaling molecule that mediates several physiological responses, including endothelium-dependent relaxation and the in- hibition of platelet aggregation-it may also participate in glucose-stimulated insulin secretion (28,31,31a). These responses are believed to be mediated by the direct activation of guanylate cyclase by nitric oxide, resulting in the formation of cGMP. Nitric oxide is pro- duced in much larger quantities by the cytokine-induc- ible isoform of nitric oxide synthase and appears to function as an effector molecule that mediates cytostatic and cytotoxic effects (30). These effects apparently are caused by nitric oxide-mediated destruction of iron-sulfur centers of iron containing enzymes and result in an impairment of mitochondrial function and DNA synthesis (33,34).

Analogues of L-arginine, in which one of the guanidino nitrogens is either methylated or nitrosylated, have been shown to inhibit nitric oxide synthase. NMMA and NAME are two commonly used compounds that block nitric oxide formation. Both of these inhibitors appear to be competitive in nature and inhibit both the inducible and constitutive isoforms of nitric oxide synthase with nearly identical kinetics. Also, D-arginine does not serve as a substrate for nitric oxide synthase, which demonstrates that the stereochemistry of the anomeric carbon of L-argi- nine must be preserved for nitric oxide synthase activity (28,30). Formatlon of nltrlc oxide by pancreatlc islet p-cells. The formation of nitric oxide by islets pretreated with IL-1 was initially demonstrated by the ability of NAME and NMMA to block the inhibitory effects of IL-1 on glucose- stimulated insulin secretion (Fig. 2; 32,35). Importantly, NMMA completely prevents IL-1-induced inhibition of glucose-stimulated insulin secretion. NAME provided partial protection against the inhibitory effects of IL-1 on insulin secretion by islets (35) at a concentration of this competitive inhibitor (1 mM) that is identical to that of the substrate L-arginine (1 mM). In these studies, IL-1 also was shown to induce the formation of nitrite (an oxidative product of nitric oxide) and both NAME and NMMA block nitrite formation. The formation of nitric oxide by islets was confirmed by the detection of an IL-1-induced iron-nitrosyl complex using electron paramagnetic reso- nance spectroscopy by our laboratory (Fig. 3; 32). The

Glucose(mM) 3 X) 3 20 3 X) 3 20

Control 0.5 mM 5 unitstrnl IL-1P + NMMA IL-1D NMMA

FIG. 2. E M S of NMMA on IL-1 &Induced lnhlbltlon ot glucose-stlmulated lnsulln secretion by rat Islets. Islets, Isolated by collagenare digestion, were pretreated tor 8 h wlth 5 Ulml 11-1 8, 0.5 mM NMMA, or both. lnsulln secretion was measured as descrlbed prevlously (32).

generation of this iron-nitrosyl complex was prevented by pretreatment of islets with NMMA in addition to IL-1, thus demonstrating the requirement for nitric oxide synthase activity. Furthermore, the protein synthesis inhibitor cy- cloheximide is known to prevent IL-1-induced inhibition of glucose-stimulated insulin secretion (36). Cyclohexi- mide also blocked the formation of nitric oxide by islets (36a). These studies provided the first evidence that IL-1 induces the expression of nitric oxide synthase by islets and that nitric oxide may be the cellular effector molecule that mediates IL-1-induced impairment of islet function.

CELLULAR MECHANISM OF IL-1's ACTIONS Inhibition of mitochondrial function has been proposed as the mechanism by which IL-1 mediates an impairment in the insulin-secretory response of the p-cell. Pretreat- ment of islets with IL-1 results in an inhibition of glucose oxidation (9,37,37a), which appears to be mediated by nitric oxide because NMMA and cycloheximide com-

(Lron-Ninosyl Complex)

g = 2.04

ILI + NMMA 1.45

FIG. 3. IL-1 &Induced Iron-nltrosyl complex tormatlon by Islets. Isolated Islet8 (2500lcondltlon) were pretreated tor 18 h wlth 5 Ulml 11-18, 0.5 mM NMMA, or both. The Islets were Isolated and EPR spectroscopy was performed as descrlbed prevlously (32). The g = 2.04 teature Is chsractcwlstlc ot the generation ot an Iron-nltrosyl complex, whlch Is belleved to arlse trom the destruction ot Iron-sulfur centen ot nonheme iron wntalnlng enzymes by nltrlc oxlde. The delocallzed electron g = 2.0023 feature present In sll cells Is as Indicated. The tormatlon ot nltrlte, an oxldatlve produd of nblc oxlde, measured on the supernatants of the Islets used b r EPR spectroscopy also Is shown. Reproduced wlth permlsslon trom the Amerlcan Soclety for Blochemlstry and Molecular Blology.

DIABETES, VOL. 41, AUGUST 1992

+ mRNA t + Protein Synthesis t t I Activation + Guanylate Cyclase

Mitochondria *(

DNA Synthesis

I t

- 3 Oxidative Phosphoryiation - 71 lnoulln - /?I Cell Death

FIG. 4. Proposed mechanism of IL-1's actlon on the p-cell.

pletely prevent IL-1-induced inhibition of glucose oxida- tion (36a). The ability of IL-1 to induce the formation of intracellular iron-nitrosyl complexes by islets (Fig. 3) suggests that the cellular targets for nitric oxide are iron-sulfur centers of iron-containing enzymes (32,38). This is suggested by our recent observation that IL-1 induces an 80% inhibition of mitochondrial aconitase activity of dispersed islet cells, and that this inhibition is prevented by NMM A (unpublished observations). Aconi- tase is a mitochondrial enzyme that contains iron-sulfur clusters known to be sensitive to nitric oxide (33,34).

The formation of nitric oxide by the islet also may explain the ability of oxygen free radical scavengers to partially protect against the inhibitory effects of IL-1 on p-cell function. Nitric oxide appears capable of interact- ing with the superoxide anion to form hydroxyl radicals (39). Hydroxyl radicals may produce deleterious effects on mitochondrial function and DNA synthesis. Because the p-cell is unique in that it contains low levels of endogenous oxygen free radical scavenging enzymes (20,21), oxygen radicals generated by the interaction of superoxide with nitric oxide may further participate in p-cell destruction. Oxygen free radical scavengers under these conditions might be expected to partially protect against the inhibitory effects of IL-1 on p-cell function by removing the highly reactive hydroxyl radical.

We have proposed a schematic model by which IL-1 mediates p-cell dysfunction and destruction (Fig. 4). In this model, IL-1 is produced by activated macrophages present during islet infiltration. IL-1 interacts with its receptors on the p-cell triggering a series of signaling events that include the induction of mRNA transcription and protein translation. Receptors for IL-1 have been demonstrated on both the Rin-m5F and HIT insulinoma cell lines (40,41), and transcriptional and translational inhibitors have been shown to prevent IL-1-induced inhibition of glucose-stimulated insulin secretion by iso- lated islets (36). We propose that IL-1 induces the expression of nitric oxide synthase, which results in the generation of the free radical nitric oxide. Nitric oxide targets iron-sulfur containing enzymes (aconitase, ribo-

nucleotide reductase, and possibly electron transport chain enzymes; 30) resulting in an inhibition of mitochon- drial function and DNA synthesis. Nitric oxide mediated inhibition of these enzymes ultimately results in p-cell death. Macrophages infiltrating the islet also may pro- duce nitric oxide and further damage the islet p-cell. Syngeneic activated peritoneal macrophages recently have been demonstrated to kill islet cells by an L-arginine dependent pathway (42). Mechanism of p-cell destruction. The pancreatic islet is a complex microorgan that contains both endocrine cells (p-, a-, 6-, and polypeptide) and nonendocrine cells (macrophages, fibroblasts, endothelial, and dendritic). The p-cell is selectively destroyed in IDDM by a process believed to be mediated by nitric oxide. Although, IL-l- induced formation of nitric oxide by isolated islets has been demonstrated, the islet endocrine cell type respon- sible for nitric oxide formation is unknown (i.e., does IL-1 induce nitric oxide production by the p-cell?). This ques- tion is further complicated by the nonendocrine cells of the islet, which contain the cytokine-inducible isoform of nitric oxide synthase. Using purified populations of p-cells and a-cells isolated by fluorescence-activated cell sorting, we recently demonstrated that IL-1 induces the formation of nitric oxide by purified p-cells. In con- trast, IL-1 does not appear to induce the formation of nitric oxide by purified a-cells (unpublished observa- tions). The formation of nitric oxide by purified p-cells is believed to mediate IL-1-induced inhibition of insulin secretion because NMMA prevents the inhibitory effects of this cytokine. IL-1 also has been demonstrated to induce the formation of nitric oxide by the insulinoma cell line Rin-m5F as evidenced by IL-1-induced nitrite and iron-nitrosyl complex formation, both of which are pre- vented by NMMA (unpublished observations). Because the p-cell is a source of nitric oxide and because macrophages do not appear to produce nitric oxide in response to IL-1 (43), it appears that IL-1 directly induces the expression of nitric oxide synthase by the endocrine p-cell. We currently are examining whether IL-1-induced nitric oxide formation by the p-cell is sufficient to induce p-cell death, or if nitric oxide production by macro- phages is also required for p-cell death in vitro.

INTERVENTIONS An understanding of cellular mechanisms involved in the etiology of IDDM may lead to the development of thera- peutic agents to intervene in the disease process. Early efforts directed at the use of oxygen free radical scaven- gers have been disappointing. lmmunomodulation stud- ies, such as the previously described antidiabetogenic effects of TNF or IL-1 in the NOD mouse, have proven somewhat more successful. If our hypothesis that nitric oxide mediates p-cell death is correct, then inhibition of nitric oxide generation should prevent the onset of IDDM. This appears to be the case with the multiple low-dose STZ-induced model of autoimmune diabetes. Treatment of male CBA mice with multiple low doses of STZ for 5 days induces hyperglycemia and lymphocytic infiltration into the islet. However, NMMA treatment for 5 days

DIABETES. VOL. 41, AUGUST 1992

following STZ injections partially prevents hyperglycemia and reduces the extent of lymphocytic infiltration into the islet (44). These results provide the first indication that prevention of nitric oxide formation in vivo may block the onset of diabetes, however the cellular mechanism by which NMMA prevents STZ-induced diabetes is un- known. While NMMA is a potent inhibitor of nitric oxide synthase, it may not be suitable for clinical interventions because it is not selective for the inducible isoform of nitric oxide synthase and it also inhibits the constitutive isoenzyme. Inhibition of constitutive nitric oxide synthase by NMMA may prove undesirable because of impaired cell-cell communication and increases in blood pressure.

The current challenge is to identify selective and potent inhibitors for the two isoforms of nitric oxide synthase. Corticosteroids, including dexamethasone, have been shown to block the induction of the inducible isoform of nitric oxide synthase without affecting the activity of the constitutive enzyme (28). We recently demonstrated that dexamethasone also blocks IL-1-induced inhibition of glucose-stimulated insulin secretion and nitric-oxide for- mation by islets (unpublished observations). These re- sults suggest a potential promise for the use of corticosteroids to block the inducible isoform of nitric oxide synthase, although there are numerous disadvan- tages to these nonspecific approaches.

In collaboration with Drs. Joseph Williamson and Ron- ald Tilton at Washington University School of Medicine, we recently discovered an inhibitor (aminoguanidine) that is selective for the inducible isoform of nitric oxide synthase (45). Fig. 5 demonstrates that aminoguanidine blocks IL-1--induced nitrite formation from Rin-m5F cells (IC,, -10 pM) in a concentration-dependent manner similar to the effects of NMMA. In contrast, aminoguani- dine is -40-fold less effective than NMMA in increasing mean arterial blood pressure of rats. The selective prop- erties of aminoguanidine on the inducible isoform of nitric oxide synthase have been confirmed using purified con- stitutive and inducible nitric oxide synthase isoenzymes isolated from rat brain and macrophage, respectively (unpublished observations). Aminoguanid~ne is an ana- logue of L-arginine in that it contains two chemically equivalent guanidino nitrogen groups in addition to a hydrazine moiety (Fig. 5). This hydrazine moiety is be- lieved to confer selectivity for the inducible isoform of nitric oxide synthase because replacement with a methyl group (methylguanidine) results in the loss of selectivity and reduced potency (unpublished observations). These studies suggest that aminoguanidine may represent a selective and potent inhibitor of the inducible isoform of nitric oxide synthase and warrants examination as a potential therapeutic agent to prevent diseases associ- ated with the inducible isoform of nitric oxide synthase.

The remarkable ability of aminoguanidine to block nitric oxide formation also suggests that nitric oxide may mediate some of the complications associated with dia- betes. Brownlee et al. (46) have shown that aminoguani- dine is a potent inhibitor of nonenzymatic advanced glycation of cellular and extracellular constituents, a process believed to be associated with vascular and neural complications of diabetes. NMMA also recently

Concentration (pM)

B

20 - ~Gmonometh~~-argin~ne Amlnoguanldlne

P

9

M/W NHCHZCH2CH2CHCOOH

C"3 O T . . . . . . . . I . . . . . . . . I . . . . ...-

1 10 100 1000 pmoles

FIG. 5. Effect6 of amlnoguanldlne and NMMA on (A) IL-1p-Induced nltrb formation by Rln-mSF celb and (4 m a n arterial blood preeaum In rats. (A) Rln-m5F cells were Incubated In the presence of IL-1B, 11-18 and amlnoguanldlne (AG), or 11-1 $ and NMM A tor 18 h nt 37%. The production of n t L war maured on the supernatant of the cell cultures and is expressed as the percent InhlbMon of IL-1p-Induced n M b tormatlon. (4 lncnaalng pmolw of amlnoguanldlne or NMMA were Injected intravenoudy In separate rats after atablllzntlon of arterial pressure, and the peak prerwre I n c m w mr recorded 88 dncr ibd previously (45). Results are exprescrod as the percentage Increams In preesure above barellne. R e p r l M wlth parmlsalon from Corbat el al. Dlebao8,41:552-56, 1992. Copyright (c) 1982 by the American Dlabter ~ o c l e t l o n , Inc.

has been shown to block vascular dysfunction using the skin chamber granulation tissue model (45), suggesting that nitric oxide may participate in vascular dysfunction associated with diabetic complications.

FUTURE STUDIES Important questions that remain to be answered are whether human islets produce nitric oxide and whether nitric oxide mediates p-cell dysfunction and destruction in humans. These questions are complicated by the difficulty in demonstrating nitric oxide production by human tissue in vitro, although indirect evidence of nitric oxide generation by humans (based on urinary nitrite levels) has been obtained (47). We recently demon- strated that the cytokine combination of IL-1 and IFN-v induces nitric oxide production by human islets in vitro; and when presented alone, IL-1 , TNF, and IFN-v have no effect on nitric oxide generation by human islets. The production of nitric oxide by human islets was demon- strated by IL-1- and IFN-?induced nitrite formation and cGMP accumulation, both of which were blocked by NMMA (unpublished observations).

The recent identification of nitric oxide as a cellular effector molecule that mediates cytokine-induced p-cell dysfunction and destruction offers promise in the imple-

DIABETES, VOL. 41, AUGUST 1992

mentation of new strategies to unravel the cellular mech- anisms involved in IDDM. This experimental approach is further enhanced by the discovery of a selective and potent inhibitor of the inducible isoform of nitric oxide synthase (aminoguanidine). It will be of interest to deter- mine whether aminoguanidine prevents spontaneous di- abetes in both the NOD mouse and the BE rat, and if nitric oxide mediates the antidiabetogenic effects of TNF or IL-1 in vivo. These approaches are important as they may explain both the antidiabetogenic and cytotoxic effects of cytokines, and resolve the paradox involving the protective and cytotoxic role of cytokines in @-cell destruction associated with IDDM.

ACKNOWLEDGMENTS This work was supported by National Institutes of Health Grant DK-06181 (MLM) and T32 DKO-07296 (JAC).

We would like to thank Joan Fink, Connie Marshall, and Jin Lin Wang for assistance in the preparation of this manuscript; collaborators Jack Lancaster and Mike Sweetland (EPR spectroscopy); Mark Currie, Tom Misko, and Bill Moore (nitric oxide synthase enzymatic assays); and Joseph Williamson and Ronald Tilton (aminoguani- dine). We thank David Scharp, Washington University School of Medicine, for providing human islets. We also would like to thank John Turk, Sasanka Ramanadham, Paul Lacy, and Emil Unanue for their critical evaluation of this manuscript.

REFERENCES 1. Gepts W: Pathologic anatomy of the pancreas in juvenile diabetes

mellitus Diabetes 141619-33, 1965 2. Cooke A: An overview on possible mechanisms of destruction of the

insulin-producing p-cell. Curr Top Microbiol lmmunol164:125-42, 1990

3. Walker R, Bone AJ, Cooke A, Baird JD: Distinct macrophage subpopulations in pancreas of prediabetic BB/E rats: possible role for macrophages in pathogenesis of IDDM. Diabetes 37:1301-304. 1988

4. Mandrup-Poulsen T, Bendtzen K, Nielsen JH, Bendixen G, Nerup J: Cytokines cause functional and structural damage to isolated islets of Langerhans. Allergy 40:424-29, 1985

5. Mandrup-Poulsen T, Bendtzen K, Nerup J. Egeberg J, Nielsen JH. Mechanism of pancreatic islet cell destruction: dose-dependent cytotoxic effect of soluble blood mononuclear cell mediators on isolated islets of Langerhans. Allergy 41 :250-59, 1986

6. Bendtzen K, Mandrup-Poulsen T, Nerup J, Nielsen JH, Dinarello CA, Svenson M: Cytotoxicity of human pl 7 interleukin-1 for pancreatic islets of Langerhans. Science 232:1545-47, 1986

7. Comens PG, Wolf BA, Unanue ER, Lacy PE, McDaniel ML: Interleu- kin 1 is potent modulator of insulin secretion from isolated rat islets of Langerhans. Diabetes 36:963-70, 1987

8. Zawalich WS, Diaz VA: lnterleukin 1 inhibits insulin secretion from isolated perifused rat islets. Diabetes 35:1119-23, 1986

9. Sandler S, Andersson A, Hellerstrom C: Inhibitory effects of interleu- kin 1 on insulin secretion, insulin biosynthesis, and oxidative metab- olism of isolated rat pancreatic islets. Endocrinol 121 :1424-31, 1987

10. Mandrup-Poulsen T, Helqvist S, Wogensen LD, Malvig J, Pociot F, Johannesen J, Nerup J: Cytokines and free radicals as effector molecules in the destruction of pancreatic beta cells. Curr Top Microbiol lrnmunol 164: 169 -93. 1990

11. Elzirik DL: Interleukin-1 induced impairment in pancreatic islet oxidative metabolism of glucose is potentiated by tumor necrosis factor. Acta Endocrinol 1 1 9:321-25. 1988

12. Pukei C, Baquerizo H, Rabinovitch A: Destruction of rat islet cell monolayers by cytokines: Synergistic interactions of interferony, tumor necrosis factor, lymphotoxin, and interleukin 1. Diabetes 37:133-36, 1988

13. Campbell IL, lscaro A. Harrison LC: IFN-y and tumor necrosis factor-a. Cytotoxicity to murine islets of Langerhans. J lmmunol 141 :2325-29, 1988

14. Rabinovitch A, Sumoske W. Rajotte RV, Warnock GL, Cytotoxic effects of cytokines on human pancreatic islet cells in monolayer culture. J Clin Endocrinol Metab 71 : I 52-56. 1990

15. Soldevila G, Buscema M, Doshi M, James RFL, Bottazzo GF, Pujol-Borrell R: Cytotoxic effect of IFN-y plus TNF-a on human islet cells. J Autoimmun 4:291-306, 1991

16. Jacob CO, Aiso S, Michie SA, McDevitt HO. Acha-Orbea H: Pre- vention of diabetes in nonobese diabetic mice by tumor necrosis factor (TNF): Similarities between TNF-a and interleukin 1. Proc NaN Acad Sci USA 87:968-72,1990

17. Mills CD: Molecular basis of "suooressor" macroohaaes: Arainine metabolism via the nitric oxide' synthetase J lm~uno l 146:2719-23 1991 - . -. . - - - . . - - .

18. Mclnerney MF, Pek SB, Thomas DW: Prevention of insulitis and diabetes onset by treatment with complete Freund's adjuvant in NOD mice. Diabetes 40:715-25, 1991

19. Oberley LW: Free radicals and diabetes. Free Radical Biol Med 5:113-24, 1988

20. Asayama K, Kooy NW, Burr IM: Effect of vitamin E deficiency and selenium deficiency on insulin secretory reserve and free radical scavenging systems in islets: decrease of islet manganosuperoxide dismutase. J Lab Clin Med 107:459-64, 1986

21. Malaisse WJ, Malaisse-Lagae F, Sener A, Pipeleers DG: Determi- nants of the selective toxicity of alloxan to the pancreatic B cell. Proc Natl Acad Sci USA 79:927-30, 1982

22. Mendola J, Wright Jr. JR, Lacy PE: Oxygen free-radical scavengers and immune destruction of murine islets in allograft rejection and multiple low-dose streptozocin-induced insulitis. Diabetes 38:379- 85, 1989

23. Sumoski W, Baquerizo H, Rabinovitch A: Oxygen free radical scavengers protect rat islet cells from damage by cytokines. Dia- betologia 32:792-96, 1989

24. Elliott RE, Chase HP: Prevention or delay of type 1 (insulin-depen- dent) diabetes mellitus in children using nicotinamide. Diabetologia 34:362-65, 1991

25. Kwon NS, Nathan CF, Gilker C, Griffith OW, Matthews DE, Stuehr DJ: L-Citrulline production from L-arginine by macrophage nitric oxide synthase: the ureido oxygen derives from dioxygen. J Biol Chem 265:13442-45, 1990

26. Leone AM, Palmer RMJ, Knowles RG, Francis PL, Ashton DS, Moncada S: Constitutive and inducible nitric oxide svnthases incor- -, ~ ~- ~ ~

porate molecular oxygen into both nitric oxide and citrulline. J go1 Chem 266:23790-95, 1991

27. Bredt DS, Snyder SH: Isolation of nitric oxide synthetase, a calm- odulin-requiring enzyme. Proc Natl Acad Sci USA 87:682-85, 1990

28. Moncada S, Palmer RMJ, Higgs EA: Nitric oxide. physiology. pathophysiology, and pharmacology. Pharmacol Rev 43: 109-42, la01

29. ~ c h r DJ, Cho HJ, Kwon NS. Weise MF, Nathan CF: Purification and characterization of the cytokine-induced macrophage nitric oxide synthase: an FAD- and FMN-containing flavoprotein. Proc Natl Acad Sci USA 88:7773-77, 1991

30. Hibbs Jr. JB, Taintor RR. Vavrin Z, Granger DL, Drapier J-C, Amber IJ. Lancaster Jr. JR: Synthesis of nitric oxide from a terminal guanidino nitrogen atom of L-arginine: a molecular mechanism regulating cellular proliferation that targets intracellular iron. In Nitric Oxide From L-Arginine: A Bioregulatory System. Moncada S, Higgs EA, Eds. New York, Elsevier, 1990, p. 189-223

31. Schmidt HHHW. Warner TD, lshii K. Sheng H, Murad F: Insulin secretion from pancreatic p-cells caused by L-arginine-derived nitrogen oxides. Science 255721-23, 1992

3la.Laychock SG, Modica ME, Cavanaugh CT: L-arginine stimulates cyclic guanosine 33'-monophosphate formation in rat islets of! Langerhans and RINmSF insulinoma cells: evidence for L-arginine: nitric oxide synthase. Endocrinol129:3043-52, 1991

32. Corbett JA. Lancaster Jr. JR, Sweetland MA, McDaniel ML: Inter- leukin-1 &induced formation of EPR-detectable iron-nitrosyl com- plexes in islets of Langerhans. J Biol Chem 266:21351-54, 1991

33. Drapier J-C, Hibbs Jr. JB: Murine cytotoxic activated macrophages inhibit aconitase in tumor cells: lnhibition involves the iron-sulfur prosthetic group and is reversible. J Clin Invest 78:790-97, 1986

34. Stadler J, Billiar TR, Curran RD, Stuehr DJ, Ochoa JB, Simmons RL: Effect of exogenous and endogenous nitric oxide on mitochondria1 respiration of rat hepatocytes. Am J Physiol260:C910-16, 1991

35. Southern C, Schulster D, Green IC: Inhibition of insulin secretion by interleukin-1 p and tumour necrosis factor-a via an L-arginine-de- pendent nitric oxide generating mechanism. FEBS Lett 276:42-44 1990

36. ~ughes JH, Colca JR, Easom RA, Turk J, McDaniel ML: lnterleukin 1 inhibits insulin secretion from isolated rat pancreatic islets by a process that requires gene transcription and mRNA translation. J Clin lnvest 86:856-63, 1990

36a.Corbett JA, Wang JL. Hughes JH. Wolf BA, Sweetland MA, Lan-

DIABETES, VOL. 41, AUGUST 1992

caster Jr. JR, McDan~el, ML: Nitric oxide and cGMP formation induced by interleukin-1 p in islets of Langerhans: evidence for an effector role of nitric oxide in ~slet dysfunction. Biochem J (in press)

37. Sandler S. Bendtzen K, Borg LAH, Eizirik DL, Strandell E, Welsh N: Studies on the mechanisms causing inhibition of insulin secretion in rat pancreatic islets exposed to human interleukin-1 p indicate a perturbation in the mitochondria1 function. Endocrinol 124:1492- 1501, 1989

37a.Welsh N, Eizirik DL, Bendtzen K. Sandler S: Interleukin-1 p-induced nitric oxide production in isolated rat pancreatic islets requires gene transcription and may lead to inhibition of krebs cycle enzyme aconitase. Endocrinol 129:3167-73, 1 991

38. Lancaster Jr. JR, Hibbs Jr. JB: EPR demonstration of iron-nitrosyl complex formation by cytotoxic activated macrophages. Proc Natl Acad Sci USA 87:1223-27, 1990

39. Beckman JS, Beckman TW, Chen J, Marshall PA, Freeman BA: boarent hvdroxvl radical oroduction bv oeroxvnitrite: im~lications fdr'endothelia~ i;jury from'nitric oxide and superoxide. roc Natl Acad SCI USA 87: 1620-24. 1990

40. Eizirik DL, Tracey DE. Bendtzen K, Sandler S: An ~nterleukin-1 receptor antagonist protein protects ~nsulin-producing 8-cells against suppressive effects of interleukin-1 p. Diabetologia 34:445- 48. 1991

41. Hamrnonds P, Beggs M, Beresford G, Espinal J, Clarke J, Mertz RJ:

Insulin-secreting 8-cells possess specific receptors for Interleukin- 1 p. FEBS Lett 261 :97-100. 1990

42. Kroncke K-D, Kolb-Bachofen V, Berschick B, Burkart V, Kolb H: Activated macrophages kill pancreatic syngeneic islet cells via arginine-dependent nitric oxide generation. Biochem Biophys Res Commun 175:752-58, 1991

43. Drapier J-C, Wietzerbin J, Hibbs JB Jr: Interferon-v and tumor necrosis factor induce the L-arginine-dependent cytotoxic effector mechanism in murine macrophages. Eur J lmmunol 18:1587-92, 1988

44. Lukic ML. Stosic-Grujicic S, Ostojic N, Chan WL, Liew FY: Inhibition of nitric oxide generation affects the induction of diabetes by streptozocin in mice. Biochem Biophys Res Commun 178:913-20, 1991

45. Corbett JA, Tilton RG, Chang K, Hasan KS, Ido Y, Wang JL, Sweetland MA, Lancaster JR Jr, Williamson JR. McDaniel, ML: Arninoguanidine, a novel inhibitor of nitric oxlde formation, prevents diabetic vascular dysfunction. Diabetes 41 1552-56, 1992

46. Brownlee M, Vlassara H, Kooney A, Ulrich P, Cerami A: Aminoguani- dine prevents diabetes-induced arterial wall protein cross-linking. Science 232: 1629-32, 1986

47. McCall T, Vallance P: Nitric oxide takes centre-stage with newly defined roles. TIPS 13:l-6, 1992

DIABETES. VOL. 41, AUGUST 1992