The Histochemical Journal 33: 317–327, 2001.© 2001 Kluwer Academic Publishers. Printed in the Netherlands.

Immunoexpression of interleukin-1β in pancreatic islets of NOD mice duringcyclophosphamide-accelerated diabetes: co-localization in macrophagesand endocrine cells and its attenuation with oral nicotinamide

S. Reddy, M. Young & S. GinnDivision of Paediatrics and the Liggins Institute for Medical Research, University of Auckland School of Medicine,Private Bag 92019, Auckland, New Zealand

Received 30 January 2001 and in revised form 30 May 2001

Summary

During insulin-dependent diabetes mellitus, islet invading immune cells destroy beta cells over a prolonged asymptomatic pre-diabetic period. Cytokines synthesised and secreted by specific immune cells within the islet infiltrate may be crucial effectorsof beta cell destruction or protection during the disease. Interleukin-1β may be a key cytokine which may act in concertwith other cytokines in initiating and/or promoting beta cell destruction. We have examined this hypothesis in NOD mice byassessing the intra-islet expression and co-localization of interleukin-1β at different time-points following cyclophosphamideadministration. We have also tested the effects of long-term oral nicotinamide given to NOD mice in suppressing intra-isletexpression of the cytokine in this accelerated model.

Cyclophosphamide was administered to day 95 female NOD mice. Pancreatic tissues were examined by dual-label confocalimmunofluorescence microscopy for the expression and co-localization of interleukin-1β at days 0, 4, 7, 11 and at onsetof diabetes (day 14). Diabetes developed in 7/11 mice 14 days after administration of cyclophosphamide while nicotinamidecompletely prevented the disease. At day 0, interleukin-1β immunolabelling was observed in selective intra-islet macrophages,several somatostatin cells and in a few beta cells. However, at day 4, it was seen mostly in somatostatin and some beta cells.At day 7, an increasing number of interleukin-1β cells were observed within the islets and co-localized to several somatostatincells, beta cells and macrophages. The mean number of intra-islet interleukin-1β cells reached a peak at day 11 and wassignificantly higher than at day 7 (p = 0.05) and at day 14 (onset of diabetes; p = 0.03). At day 11, interleukin-1β

immunolabelling was also present in selective macrophages which co-expressed inducible nitric oxide synthase. At onset ofdiabetes, some macrophages, residual beta cells and somatostatin cells showed immunolabelling for the cytokine. Exposureof NOD mice to oral nicotinamide was associated with a considerably reduced expression of interleukin-1β cells withinthe islet at day 11 (p = 0.002). We conclude that cylophosphamide treatment enhances the expression of interleukin-1β

in selective macrophages, somatostatin and beta cells during the course of the disease. Its expression reaches a maximumimmediately prior to onset of diabetes. Interleukin-1β present in intra-islet macrophages, somatostatin and beta cells mayinfluence its expression by autocrine and paracrine means. Interleukin-1β expression within islet macrophages may alsoup-regulate inducible nitric oxide synthase within the same macrophage or adjacent macrophage populations. These intra-isletmolecular events may corroborate with other local cytotoxic processes leading to beta cell destruction. Oral nicotinamide mayattenuate intra-islet expression of interleukin-1β and thus inducible nitric oxide synthase during prevention of Type 1 diabetesin this animal model. The expression of interleukin-1β in specific islet endocrine cell-types shown in this study requires furtherinvestigation.

Introduction

Insulin-dependent diabetes mellitus is an autoimmune dis-ease characterised by selective destruction of beta cells overa prolonged asymptomatic pre-diabetic period (Eisenbarth1986, Bach 1994). At diagnosis of the disease, the islet infil-trate consists predominantly of CD8 cells and to a lesserextent CD4 cells, B-lymphocytes and macrophages (Bottazzoet al. 1985, Foulis et al. 1986, Itoh et al. 1993, Somoza et al.1994). Several studies have shown that prolonged exposureof rodent or human islets in culture to various cytokines such

as interleukin-1β (IL-1β), interferon-γ (IFN-γ ) and tumournecrosis factor-α (TNF-α) results in beta cell dysfunction andfinally death (Sandler et al. 1987, Mandrup-Poulsen et al.1985, 1989, Eizirik et al. 1994, Cetkovic-Cvrlje & Eizirik1994). More recent data indicate that other cytokines, suchas interleukin-4 (IL-4), can prevent diabetes in NOD mice(Rapport et al. 1993). It is likely that beta cell inhibitoryand protective cytokines are expressed and released locallywithin the inflamed islets. However, there is little informationon the cellular expression of various intra-islet cytokine pro-teins and their downstream signalling molecules during the

318 S. Reddy et al.

development of Type 1 diabetes, largely due to inaccessibilityof suitable pancreatic material. By immunohistochemistry,post-mortem pancreatic tissue from diabetic subjects ofrecent onset showed expression of IFN-α in beta cellsand IFN-γ in islet lymphocytes (Foulis et al. 1987, 1991,Yamagata et al. 1996). Using the reverse transcriptase poly-merase chain reaction, mRNAs for IFN-α, IFN-γ , TNF-α,IL-1, IL-4 and IL-6 have been demonstrated in similar patho-logical material (Huang et al. 1995). However, in these lat-ter studies, neither the cellular source/s of the cytokinesnor their distribution at the protein level was determined(Huang et al. 1995). More recently, expression of Fas ininsulin and glucagon cells and Fas ligand (FasL) in islet infil-trating immune cells have been demonstrated immunohisto-chemically in pancreatic biopsies from recent onset Type 1diabetic patients (Moriwaki et al. 1999). A similar find-ing was also reported previously in pancreatic tissues fromtwo newly-diagnosed subjects following death, except thatFas expression was reported exclusively in beta cells (Stassiet al. 1997). In both the latter studies, co-immunolabellingof beta cell inhibitory cytokines within the islet was notperformed.

Studies in the NOD mouse, a close model of human Type 1diabetes, suggest that the islet infiltrating immune cells maybe the actual cellular effectors and regulators of beta celldestruction. They may elicit their function via a Th1/Th2cytokine milieu and in concert with IL-1β (Mossman &Coffman 1989, Rabinovitch 1994, 1998, Mandrup-Poulsen1996). There are only limited data on the cellular expressionof cytokine proteins in situ and their correlation with diabetesin the NOD mouse also (Hancock et al. 1995, Pilstrom et al.1997). In the cyclophosphamide and adoptive transfer modelsof accelerated diabetes in the NOD mouse, we have recentlyshown a heightened expression of IFN-γ and interleukin-4(IL-4) in islet immune cells prior to onset of diabetes (Reddyet al. 2000a,b). During the early stages of cyclophosphamide-induced diabetes, IL-4 was also co-localized in selective betacells (Reddy et al. 2000b). Interleukin-1β, a proinflammatorycytokine has been proposed to be cytotoxic to beta cells dur-ing the early stages of pre-diabetes (Arnush et al. 1998a,b). Itpromotes gene expression of inducible nitric oxide synthase(iNOS) in pancreatic islets and in RIN cells in culture (Eiziriket al. 1992, Kwon et al. 1995). Surprisingly, systemic admin-istration of IL-1β to NOD mice protects from diabetes (Jacobet al. 1990). However, treatment of NOD mice with anti-IL-1β prevents cyclophosphamide-induced diabetes (Cailleauet al. 1997). In order to clarify the role of IL-1β in Type1 diabetes, we have examined its pancreatic cellular expres-sion, co-localization and intra-islet spatial distribution dur-ing cyclophosphamide-induced diabetes in the NOD mouse.We have also examined whether oral nicotinamide, whichis an agent known to prevent diabetes in this animal modeland is currently subject to evaluation in a large human clin-ical trial (the European Nicotinamide Diabetes InterventionTrial (ENDIT)), modulates intra-islet expression of IL-1β

(Yamada et al. 1982, Reddy et al. 1990, 2001, Elliott & Chase1991, Pozzilli 1998).

Materials and methods

NOD mice and assessment of diabetes

A colony of NOD mice has been established under con-ventional conditions at the Animal Resources Unit of theAuckland School of Medicine. The spontaneous diabetesrate is approximately 50% among females and less than 1%among males between the ages of 120 and 250 days. Diabetesis defined as the presence of heavy glycosuria on three consec-utive days and confirmed by a non-fasting blood glucose levelexceeding 12 mM. Non-fasting blood glucose levels rangedfrom 4 to 6 mM in non-diabetic NOD mice.

Study groups, administration of cyclophosphamideand tissue collection

Female NOD mice were obtained from several breeding pairs,weaned at day 21 and allocated at random into three groups(23 mice per group) as follows:

Group 1: Standard mouse chow (Diet 86; Reddy et al.2001) + drinking water (control group);Group 2: Standard mouse chow+drinking water (cyclophos-phamide group);Group 3: Standard mouse chow + 1% nicotinamide(New Zealand Tablet Manufacturers) as drinking water(nicotinamide + cyclophosphamide group).

Cyclophosphamide was administered to day 95 female NODmice to accelerate diabetes as described recently (Reddy et al.1999, 2000b,c). According to this regimen, approximately66% of mice develop diabetes 14 days after its administration.At day 95, all except 3 mice from Group 2 and 3 were injectedwith cyclophosphamide (Sigma, St Louis, USA) at a dose of300 mg/kg body weight by the intra-peritoneal route. Group 1mice received diluent (sterile water). Following injection ofcyclophosphamide or diluent, 3 mice from each group weresacrificed at days 0 (day 95 and without cyclophosphamideadministration), 4, 7, 11 and the pancreas harvested (seebelow). The remaining 11 mice from each group were fol-lowed until day 14 for the development of diabetes. At day 14,three mice from Groups 1–3 and all 7 diabetic mice fromGroup 2 were killed and the pancreas removed. The remain-ing non-diabetic mice from the day 14 group were monitoredfor diabetes for a further 30 days (Groups 2 and 3).

At the various time-points described above, animals werekilled by cervical dislocation and almost the entire pancreaswith a small portion of the adjacent spleen was removed,embedded in OCT and snap-frozen in isopentane, cooled inliquid nitrogen. Tissues were stored at –70 ◦C until histo-chemical and immunohistochemical analyses.

Primary antibodies, non-immune IgG,normal sera and IL-1β

Polyclonal antibodies against mouse IL-1β were purchasedfrom R & D Systems (Minneapolis, MN, USA). They were

Interleukin-1β in islets of NOD mice 319

produced in goats following immunisation with purifiedE. coli-derived recombinant mouse IL-1 (rmIL-1). IL-1β

specific immune IgG was purified by mouse IL-1β affin-ity chromatography (0.1 mg/ml immune IgG). This antibodyneutralises the biological activity of rmIL-1. Western blottingshows <25% cross-reactivity with recombinant human IL-1β (rhIL-1β). In direct ELISA, it shows no cross-reactivityagainst a range of cytokines including mouse IFN-γ , mouseIL-4 and mouse transforming growth factor-β (TGF-β). Ithas been recommended by the suppliers for immunohisto-chemical detection of mouse IL-1β in cultured cells or tissuesections.

Polyclonal antibodies to mouse macrophage iNOS (rabbitIgG fraction, 0.25 mg/ml) were purchased from Transduc-tion Laboratories (Lexington, Kentucky, USA) and havebeen shown in this laboratory to have similar immunohisto-chemical specificity to our previously employed anti-mouseiNOS from Upstate Biotechnology (New York, USA; Reddyet al. 1999, 2000a,c). Rat monoclonal antibodies to mousemacrophages (Mac-1, clone M/170) CD4 (clone GK 1.5) andCD8 (clone 53.6.72) in the form of culture supernatants werefrom Dr H. Georgiou of the Walter and Eliza Hall Institute,Melbourne, Australia. The same antibodies have been usedpreviously in immunohistochemical staining of sections ofpancreas from the NOD mouse (Reddy et al. 1995). Rabbitanti-glucagon and rabbit anti-somatostatin were from ICNPharmaceuticals, Inc, Aurora, Ohio, USA. Guinea pig anti-insulin serum was available in this laboratory.

In the immunohistochemical procedure, all primary anti-sera were titrated to give maximal immunohistochemicalreactivity. Normal sera from the goat, sheep, donkey, rabbit,guinea pig, rat and mouse were available in this laboratory.

Highly purified rmIL-1β was purchased from PeprotechIncorporated (New Jersey, USA).

Immunolocalization of IL-1β

An indirect immunofluorescence procedure was employedfor the immunohistochemical localization of IL-1β and wassimilar to a recently reported procedure for IL-4 and IFN-γimmunolabelling of pancreatic sections from the NOD mouse(Reddy et al. 2000a,b). During the immunohistochemicalprotocol (described below), sections were routinely washed inexcess phosphate-buffered saline (PBS), pH 7.5, containing0.3% saponin (Sigma, St Louis, MO, USA) at the end of theincubation period of each immunohistochemical reagent. Thesame buffer also acted as the diluent for all immunologicalreagents for cytokine staining. Prior to immunohistochemi-cal staining, sections were re-fixed in cold acetone, equili-brated in PBS-saponin and incubated with blocking serum(5% normal donkey serum) for 1 h at 37 ◦C. After washing,they were incubated with goat anti-IL-1β (1 : 50, 2 µg/ml)for 60 h at 4 ◦C followed by addition of donkey anti-goatIgG-biotin (1 : 200, Jackson ImmunoResearch Laboratories,Westgrove, Pennsylvania, USA) for 1 h at 37 ◦C. They werethen incubated with streptavidin-Texas Red (1 : 200, JacksonImmunoResearch Laboratories) for 1 h at 37 ◦C and examined

by light and/or confocal immunofluorescence microscopy(see below).

Dual immunolabelling of IL-1β with macrophages,iNOS, insulin, glucagon or somatostatin

Groups of three serial sections were first stained for IL-1β

as described above and then incubated with rat anti-Mac-1(1 : 1), rabbit anti-iNOS (1 : 200) or guinea pig anti-insulin(1 : 100) for 2.5 h at 37 ◦C. Prior to addition of anti-insulin (oranti-somatostatin and anti-glucagon, 1 : 50), the correspond-ing section was fixed for 10 min with cold 4% paraformalde-hyde in PBS to improve staining. Sections were then washedin PBS and incubated with either goat anti-rat IgG (H&L)-cyanin-2 (1 : 50; Rocklands, Gilbertsville, Pennsylvania,USA), goat anti-rabbit IgG-Alexa-488 (1 : 200; MolecularProbes, Eugene, Oregon, USA), donkey anti-guinea pigIgG-FITC or donkey anti-rabbit IgG-FITC (1 : 100; JacksonLaboratories) for 1 h at 37 ◦C. Sections were then washedand prepared for light and confocal microscopy. Sectionsfrom the diabetic tissue were also dual stained for IL-1β andsomatostain and IL-1β and glucagon as described for insulin,except that goat anti-rabbit IgG-Alexa-488 was employed assecondary antibody.

Triple and quadruple immunolabelling

Selected sections were sequentially immunolabelled forIL-1β (Texas red), iNOS (Alexa 488) and macrophages (don-key anti-rat IgG-cyanin 5) by the above protocol and exam-ined by confocal microscopy and islets imaged. In someinstances, following triple imaging of selected islets, sectionswere incubated with 5% normal rabbit serum for 1 h at 37 ◦C,washed in PBS and then incubated with anti-insulin. Rabbitanti-guinea pig IgG-peroxidase (1 : 100; Dako, Denmark)was applied for 1 h at 37 ◦C. Cellular sites of insulin were visu-alised with 3,3′-diaminobenzidine dihydrochloride (DAB;Sigma) in the presence of H2O2 and following counter-staining with haematoxylin. The same triple-labelled isletspreviously stained by immunofluorescence were re-imagedby light microscopy to reveal insulin immunoreactivity byimmunoperoxidase staining.

Microscopy and imaging

Immunostained sections and sections stained with haema-toxylin and eosin were examined with an Olympus UV-lightmicroscope. At least 10 islets from each pancreas and theentire exocrine region on each slide were analysed and rep-resentative islets photographed (Reddy et al. 1995).

Immunofluorescent sections were also examined by confo-cal laser scanning microscopy (Leica, Heidelberg, Germany)as described previously (Reddy et al. 1999, 2000a,b). A seriesof thin (approximately 1 µm) optical sections at differentplanes through the specimen was collected. For direct corre-lation studies following dual and triple immunolabelling, theappropriate excitation filters were chosen and the same optical

320 S. Reddy et al.

Figure 1. Confocal photomicrographs of representative islets dual-labelled for IL-1β (red) with macrophages (green) or insulin (green) from controlCD-1 mice and from NOD mice at 0, 4 and 7 days after administration of cyclophosphamide. (a)–(d): Control CD-1 mice. (a) and (b) islet dual-labelledfor IL-1β (a) and macrophages (b). Arrows point to IL-1β cells which are not macrophages. (c) and (d) show the same islet from an adjacent sectiondual-labelled for IL-1β (c) and insulin (d). Arrows point to a few beta cells, which co-express IL-1β. (e)–(h): Day 95 female NOD mice withoutcyclophosphamide injection (day 0). (e) and (f) islet dual-labelled for IL-1β (e) and macrophages (f). Arrows point to macrophages which express

Interleukin-1β in islets of NOD mice 321

sections imaged. Each digital set was collected on the confo-cal microscope and processed using the 3D software includedwith the Leica ScanWare operating system to construct az-series projection, a computer averaged assembly of all opti-cal sections in the data set. The FITC, Alexa 488 or cyanin-2distribution was assigned a fluorescent green colour and theTexas red distribution a bright red colour while cyanin-5 wasgiven a blue colour (Reddy et al. 1999). Separate as well asmerged images were saved in TIFF file format and assembledin Adobe Photoshop 5.5 (Adobe Systems, USA).

Cell enumeration

The intra-islet IL-1β immunoreactive cells were enumeratedeither microscopically or following photography. At least 10separate islets were analysed from each pancreas. Resultswere expressed as the mean number of IL-1β cells per islet±SEM.

Controls for immunohistochemistry

In the immunohistochemical procedure for IL-1β, iNOS,macrophages and insulin, the primary antibody was replacedwith buffer, normal sera from a goat, sheep, rabbit, mouse andguinea pig and normal rabbit IgG. In addition, primary (fortwo-step staining) and secondary steps (for three-step stain-ing) were omitted. IL-1β antibody was also preabsorbed for18 h and at 4 ◦C with highly purified rmIL-1β (0.1 µg anti-IL-1β IgG + 0.1 µg rmIL-1β per 50 µl PBS-saponin buffer) orsomatostatin (0.1 µg anti-IL-1β + 0.5 µg somatostatin per50 µl PBS-saponin buffer) before use in the immunohis-tochemical procedure. In the immunohistochemical proce-dure for insulin, the primary antiserum was absorbed withexcess highly purified bovine insulin (1 µg insulin per µl oforiginal immune serum) before use. In the double and triple-staining protocols, species-incompatible secondary antibod-ies (FITC, Alexa 488, cyanin-2 or biotin-linked) wereemployed.

Statistics

An analysis of variance was used to investigate the differ-ences in intra-islet IL-1β positive cells at the 5 time-points.A square root transformation was employed as the raw datawas not normally distributed. There was an overall differencein the number of intra-islet IL-1β cell numbers (p = 0.03).A multiple comparisons test using Tukeys method was usedto determine differences between the various time-points.

IL-1β while arrowhead points to an IL-1β cell which does not show immunolabelling for macrophages. Arrow with a longer tail points to an extra-isletcell weakly stained for macrophage and IL-1β. (g) and (h) same islet from an adjacent section dual-labelled for IL-1β and insulin. Arrows point to betacells which express IL-1β while arrowheads point to IL-1β cells which are insulin negative. (i)–(j): Islet 4 days after cyclophosphamide administrationdual-labelled for IL-β (i) and insulin (j). Arrows point to insulin cells with immunolabelling for IL-1β also. (k)–(l): Islet 7 days after cyclophosphamideadministration dual-labelled for IL-1β (k) and macrophages (I). Arrows point to macrophages which express IL-1β while arrowheads point to IL-1β

positive cells without immunolabelling for macrophages. Scale bars: 1 cm = 42 µm (a–d,k,l); 1 cm = 27 µm (e–j).

Results

Development of cyclophosphamide-induced diabetesand islet pathology

Diabetes developed in 7/11 mice by 14 days after admin-istration of cyclophosphamide (Group 2). Diabetes did notdevelop in Group 1 mice (without cyclophosphamide) andin mice given oral nicotinamide and cyclophosphamide(Group 3). In Group 2, the insulitis scores increased pro-gressively during cyclophosphamide-induced diabetes devel-opment except at day 4 when there was a marked reduction.The immune cells were comprised predominantly of CD4 andCD8 cells and macrophages as reported previously (Reddyet al. 1999, 2001).

Interleukin-1β expression, co-localizationand its intra-islet spatial relationship

In pancreatic sections from normal CD-1 mice a small num-ber of IL-1β immunoreactive cells were observed within theislets. Dual staining showed that IL-1β was co-localized insomatostatin cells and a few beta cells but not in macrophagesor glucagon cells (Figure 1a–d).

In the NOD mouse, IL-1β immunolabelling was observedin intra-islet cells during the study period. However, notabledifferences in cell numbers and co-localization were observedat different time-points following cyclophosphamide admin-istration. At day 0 of cyclophosphamide administration,(95-day-old NOD mice), the cytokine was co-localized inselective macrophages within the islets (Figure 1e,f). Afew beta cells and several somatostatin cells also con-tained IL-1β (Figure 1g,h). Immunolabelling for the cytokinewas not observed in glucagon cells. This cellular pat-tern of staining was also seen in NOD mice at vari-ous time-points following administration of diluent alone(Group 1).

Beta and somatostatin cell co-localization of the cytokinein cyclophosphamide-treated mice were also observed atday 4 (Figure 1i,j).

At day 7, an increasing number of beta cells with strongimmunolabelling for the cytokine were observed, either inclose proximity or distally to macrophages. This was accom-panied by slightly weaker immunolabelling for the cytokinein selective macrophages (Figure 1k,l; Figure 2a,b). At thisstage the intracytoplasmic distribution of the cytokine in betacells was often punctate and sometimes polar (Figure 2c).Occasionally some islets showed immunolabelling forthe cytokine almost exclusively in selective beta cells

322 S. Reddy et al.

Figure 2. Confocal photomicrographs of representative islets dual-, triple- and quadruple-labelled for IL-1β (red), macrophages (green or blue),iNOS (green) or insulin (green or brown) from NOD mice 7 and 11 days after cyclophosphamide administration. Islets from NOD mice 11 days aftercyclophosphamide treatment but exposed to oral nicotinamide from weaning are also included. (a)–(b): Same islet as in Figure 1(k,l) from an adjacentsection 7 days after cyclophosphamide treatment dual-labelled for IL-1β (a) and insulin (b). Arrows point to beta cells which co-express IL-1β.Arrowheads point to IL-1β cells which are macrophage-like (compare Figure 1k,l). (c): Higher confocal view of an IL-1β positive islet showingmore detailed immunocytoplasmic distribution of the cytokine. Arrows point to polar distribution while arrowhead points to punctate immunolabelling.

Interleukin-1β in islets of NOD mice 323

despite significant macrophage infiltration within the sameislet.

At day 11, intra-islet IL-1β cells were observed pre-dominantly in macrophages. At this stage, triple andquadruple immunohistochemistry showed an increasingnumber of macrophages which co-expressed IL-1β andiNOS (Figure 2d–g) while only a few beta cells closeto macrophages expressed the cytokine (Figure 2d,e,h).Many intra-islet macrophages also contained the cytokine inthe absence of iNOS immunolabelling. Some macrophagesdevoid of IL-1β also expressed the enzyme.

At onset of diabetes, weak to moderate expression of thecytokine was observed in macrophages, a few residual betacells and somatostatin cells (Figure 3a–f). At day 11, a smallproportion of macrophages residing in exocrine, perivascu-lar and sinusoidal areas expressed the cytokine (results notshown).

Female NOD mice exposed to oral nicotinamide fromweaning and given cyclosphosphamide at day 95 showedreduced expression of intra-islet IL-1β cells at day 11,despite significant macrophage infiltration (Figure 2i,j). Thecytokine was co-localized to a few beta cells and macrophages(Figure 2i–l).

The mean number ±SEM of IL-1β cells per islet whichexpressed the cytokine during the various stages follow-ing cyclophosphamide administration and in the presence ofnicotinamide (day 11 mice) are shown in Figure 4. Therewas a significant difference in the mean number of IL-1β

cells between day 11 and at onset of diabetes (p = 0.03;Tukeys method) and a weaker difference between days 7and 11 (p = 0.05; Tukeys method). In animals exposed tooral nicotinamide from weaning and treated with cyclophos-phamide at day 95, the mean number of intra-islet IL-1β

cells per islet at day 11 was significantly lower than in micenot given nicotinamide (p = 0.002; analysis of variance;Figure 4).

Controls

In the immunohistochemical procedure for IL-1β, immuno-staining was absent when the primary antiserum was replacedwith PBS-saponin or normal sera from a variety of species.Immunostaining was almost abolished when the primary anti-body was preabsorbed with excess rmIL-1β. However, pre-absorption of the primary antibody with excess somatostatinresulted in preservation of IL-1β immunolabelling. The useof species incompatible secondary antibodies also resultedin an absence of staining. In the dual staining procedure for

(d)–(h): Islet 11 days after cyclophosphamide treatment quadruple-labelled for IL-1β (e), iNOS (f), macrophages (g) and insulin (brown). (d) is amerged confocal view of (e)–(g). In (d)–(g) arrows point to macrophages which co-express IL-1β and iNOS while the arrows with a longer tail in (d, eand h) point to a beta cell with immunolabelling for IL-1β. Arrowheads in (h) point to the residual beta cell area in the inflamed islet. (i)–(l): Islet11 days after cyclophosphamide treatment from a NOD mouse exposed to oral nicotinamide. (i,j) islet dual-labelled for IL-1β (i) and macrophages(j) Note only a few strongly stained IL-1β cells (arrow). (k,l): same islet from an adjacent section dual-labelled for IL-1β (k) and insulin (l). Arrowspoint to β cells which express IL-1β in (k) and (l). Note that the majority of the cytokine positive cells within the islet show weak immunolabelling.Scale bars: 1 cm = 42 µm (a,b,d–h); 1 cm = 27 µm (i–l); 1.25 cm = 21 µm (c).

IL-1β with macrophages, insulin or iNOS, omission of oneof the primary antisera did not result in immunostaining ofthe antibody-devoid antigen.

Discussion

Interleukin-1β exerts multiple influences on immune,metabolic, endocrine and neuronal functions (Dinarello1991). Several in vitro studies strongly implicate IL-1as a key molecule which may act in concert with otherbeta cell cytotoxic cytokines during Type 1 autoimmunediabetes (Bendtzen et al. 1986, Sandler et al. 1987a,b,Mandrup-Poulsen et al. 1990, Southern et al. 1990,Rabinovitch et al. 1994, Mandrup-Poulsen 1996, Rabinovitch1998). Little information exists on the in situ expression of IL-1β protein within the islets and their cellular sources duringinsulin-dependent diabetes mellitus. Here, we have examinedthe cellular expression and co-localization of IL-1β withinthe pancreas of the NOD mouse following administration ofcyclophosphamide, including its ontogeny and spatial rela-tionship to beta cells and iNOS immunoreactive cells. Wehave also determined whether the anti-diabetogenic prop-erty of oral nicotinamide is associated with the attenuationof intra-islet IL-1β expression.

Our studies show that islet-infiltrating macrophages,somatostatin and beta cells are the major producers of IL-1β

during cyclophosphamide-induced diabetes in the NODmouse. The cytokine was absent in glucagon cells. However,the cellular sources and pattern of distribution of IL-1β variedsignificantly at different stages of cyclophosphamide-induceddiabetes. At day 0, the cytokine was seen in the three intra-islet cell-types while at day 4, when there is only minimummacrophage infiltration, it was observed predominantly insome beta cells and somatostatin cells. Whether IL-1β can beproduced constitutively by somatostatin and beta cells them-selves and in the absence of paracrine influence of IL-1β orother signalling factors released from adjacent macrophagesis unclear. Previous studies have shown that islets isolatedfrom non-diabetes prone mice and maintained in culture showlow constitutive expression of IL-1β mRNA which is ele-vated significantly following exposure to IL-1β (Welsh et al.1995). Since islets from non-diabetes prone mice containonly a low number of resident macrophages that may bedepleted during islet isolation and culture, it is likely that theremay be other intra-islet sources of the cytokine such as betacells, somatostatin cells, endothelial cells or fibroblasts. In thepresent study, a proportion of islet cells from non-diabetesprone CD-1 mice also expressed the cytokine either

324 S. Reddy et al.

Figure 3. Confocal photomicrographs of representative islets from a cyclophosphamide-treated diabetic NOD mice dual-labelled for IL-1β (red)and macrophages (green), insulin (green) or somatostatin (green). (a)–(b): Islet dual-labelled for IL-1β (a) and macrophages (b). Arrows point tomacrophages which express IL-1β. Note that a number of intra-islet macrophages either show weak or no immunolabelling for IL-1β. Arrowheadspoint to strongly labelled IL-1β cells which are not macrophages. (c)–(d): The same islet from an adjacent section dual-labelled for IL-1β (c) andinsulin (d). Note the presence of only a few weakly stained insulin cells (arrows in (d)) which are strongly immunolabelled for IL-1β (arrows in (c)).Arrowheads point to strongly labelled IL-1β cells which are devoid of insulin. (e)–(f): The same islet from an adjacent section dual-labelled for IL-1β

(e) and somatostatin (f). Arrows point to somatostatin cells which show strong immunolabelling for IL-β while arrowheads point to IL-1β cells whichare somatostatin negative. Scale bars: 1 cm = 27 µm.

Figure 4. The mean ± SEM number of intra-islet IL-1β cells at vari-ous time-points following administration of cyclophosphamide to day 95female NOD mice (solid bars) and in mice exposed to oral nicotinamidefrom weaning and treated with cyclophosphamide similarly (day 11;open bars). Cytokine cell numbers were analysed from three mice at eachtime-point. In each mouse at least 10 separate islets were enumerated forIL-1β.

in beta cells or somatostatin cells. In situ hybridizationapproaches may be necessary to prove unequivocally, the con-stitutive expression of IL-1β by endocrine cells. However, insupport of our present findings, previous immunohistochem-ical studies have demonstrated the expression of immuno-reactive IL-1β in hypothalamic neurons of the rat (Lechanet al. 1990).

Interleukin-1β has also been shown to release somatostatinand growth hormone releasing hormone from hypothalamicexplants (Honeger et al. 1991). In view of the common embry-ological origin of neuronal and islet endocrine cells, it ispossible that the latter cells may be capable of producingsome cytokines constitutively. In other studies, beta cells havebeen shown to express a number of cytokines. For example,interferon-α has been co-localized in beta cells during humaninsulin-dependent diabetes mellitus (Foulis et al. 1987). A ratinsulinoma cell line and mouse islet cells express IL-6 whenexposed to IFN-γ and TNF-α in vitro (Campbell et al. 1989).Transformed beta cells but not alpha cells or mouse islet cells,when exposed to IL-1β, produce mRNA for TNF-α (Yamadaet al. 1993). Recently, we colocalized IL-4 in CD4 cellsas well as in beta cells during cyclophosphamide-induced

Interleukin-1β in islets of NOD mice 325

diabetes (Reddy et al. 2000b). Whether the IL-1β immuno-labelling in beta cells observed in our study is due to receptormediated uptake and internalization of the cytokine by betacells is not clear. Previous studies have shown the presence ofIL-1 receptors on beta cells of non-diabetes prone and NODmice (Hammondds et al. 1990). In diabetic NOD mice, IL-1receptors show a sharp decrease, correlating with the inten-sity of islet destruction (Jafarian-Tehrani et al. 1995). It ispossible that some cytokines produced by islet cells may benecessary for normal islet function.

At day 11, macrophage expression of IL-1β became morepronounced. At this stage, there was considerable reduc-tion in the number of beta cells probably as a consequenceof IL-1β induced iNOS expression, production of NO andother free radicals and cytotoxic cytokines produced by intra-islet T cells. Cytokines such as TNF-α and IFN-γ mayalso regulate intra-islet IL-1β production. This time-pointimmediately before diabetes (day 11) also coincides with amarked reduction in pancreatic insulin and mRNA content(Kay et al. 1991). However, at onset of diabetes, apart frommacrophages, some of the remaining insulin cells and the isletre-distributed somatostatin cells also expressed the cytokine.These observations suggest that beta cells may also augmenttheir own destruction in concert with adjacent IL-1β express-ing cells and T cell-derived cytokines. The present results arealso in accord with some of the previous findings by otherson the expression of IL-1β mRNA in the NOD mouse dur-ing spontaneous diabetes (Welsh et al. 1995). In these stud-ies, freshly isolated islets from female NOD mouse at 5 and16 weeks of age showed significant levels of IL-1β mRNAwhich were much higher than in male NOD mice at the sameages. The presence of relatively high levels of IL-1β tran-script in 5-week-female NOD mice, when there is minimuminsulitis, suggests that cellular sources of the cytokine, otherthan macrophages, cannot be ruled out completely.

Several mechanisms have been proposed by which nicoti-namide may exert its anti-diabetogenic activity (Pociotet al. 1993). Here we have shown that oral nicotinamideprophylaxis completely prevents cyclophosphamide-induceddiabetes in NOD mice. Recently we have shown immuno-histochemically that NOD mice treated similarly showreduced expression of intra-islet iNOS, despite significantmacrophage infiltration (Reddy et al. 2001). Previous stud-ies have shown that exposure of rat islets or RIN cells toIL-1β leads to iNOS gene expression and NO production viaIL-1β-induced activation of the transcription factor, NFκB(Kwon et al. 1995). In other studies, nicotinamide dose- andtime-dependently inhibited and delayed IL-1β-induced NOproduction from rat islets in culture (Andersen et al. 1994).However, nicotinamide was not able to attenuate nitrite accu-mulation in the medium when rat islets were exposed tothe same cytokine for 4–6 days (Reddy et al. 1995). Ourpresent data suggest that nicotinamide may down-regulatethe production of intra-islet IL-1β which may lead to atten-uated iNOS expression and thus NO production. In a recentstudy, 10 mM nicotinamide inhibited NO synthesis in a mousemacrophage cell line by preventing NO synthase mRNA

induction but not affecting NO synthase activity (Hauschildtet al. 1991, 1992). Another study employing rat insuli-noma RINm5F cell line suggests that lower concentrations ofnicotinamide may inhibit iNOS enzymatic activity whereashigher concentrations may inhibit iNOS mRNA expression(Cetkovic-Cvrlje et al. 1993). In human studies, nicotinamidehas been shown to inhibit in vitro production of IL-12 andTNF-α in peripheral whole blood of subjects with Type 1diabetes or from those at high risk of developing the dis-ease (Kretowski et al. 2000). It inhibits TNF-α productionin monocytes from peripheral blood (Fukuzawa et al. 1997).Nicotinamide has also been shown to up-regulate manganesesuperoxide dismutase in RINm5F cells and may thereforeafford protection as a free radical scavenger (Andrade et al.1996). It also inhibits poly ADPribose polymerase (PARP)activity, stimulates beta cell regeneration during partial pan-createctomy and counteracts macrophage-induced islet celllysis (Yonemura et al. 1984, Kolb et al. 1990). Previous stud-ies have implicated PARP in the process of pro-inflammatorygene expression. In macrophages from PARP-deficient ani-mals, there is a reduction in the activation of NFκB, withsubsequent suppression of pro-inflammatory gene expression(Zingarelli et al. 1998). ICAM-1, TNF-α and iNOS expres-sion are suppressed in PARP-deficient cells (Szabo et al.1998, Oliver et al. 1999). Nicotinamide may modulate thesemolecular events through inhibition of PARP. Further insightson the role of nicotinamide in vivo may be gained by deter-mining the intra-islet levels of the vitamin, iNOS and IL-1β

in parallel during long-term treatment of NOD mice.In conclusion, we have shown that IL-1β is expressed by

intra-islet macrophages during cyclophosphamide-induceddiabetes. Immunolabelling for the cytokine is also seen insomatostatin and beta cells at specific stages of the disease.The heightened expression of the cytokine in intra-islet cellsat later stages of cyclophosphamide-induced diabetes sup-ports the findings of others who have shown that systemicadministration of either neutralising antibodies to IL-1β

or soluble IL-1 receptor results in a significant protectionfrom disease in this accelerated model (Cailleau et al. 1997,Nicoletti et al. 1994). Co-localization of IL-1β and iNOS inseveral intra-islet macrophages shown in this study suggeststhat this cytokine may induce iNOS expression by autocrinemechanisms and promote beta cell destruction. Recent stud-ies by others have shown that exposure of mouse islet cellsto IL-1β results in the up-regulation of beta cell Fas (Stassiet al. 1995, Yamada et al. 1996, Suarez-Pinzon et al. 1999,Loweth et al. 1998). Thus, production and release of IL-1β

locally within the islet may also promote Fas-mediated lysisof beta cells. Our studies add support to the view that atten-uation of intra-islet IL-1β or their key downstream signalsduring prediabetes may have therapeutic benefit.

Acknowledgements

This study was supported by a grant from the AucklandMedical Research Foundation. We thank Ms E. Robinson

326 S. Reddy et al.

for the statistical analyses and Ms Jacqueline Ross and MsCathy Olsen for assisting in image preparation.

References

Andersen HU, Jorgensen KH, Egeberg J, Mandrup-Poulsen T, Nerup J(1994) Nicotinamide prevents interleukin-1 effects on accumulatedinsulin release and nitric oxide production in rat islets of Langerhans.Diabetes 43: 770–777.

Andrade J, Conde M, Ramirez R et al. (1996) Protection from nicoti-namide inhibition of interleukin-1β-induced RIN cell nitric oxideformation is associated with induction of MnSOD enzyme activity.Endocrinology 137: 4806–4810.

Arnush M, Heitmeier MR, Scarim AL et al. (1998a) IL-1 produced andreleased endogenously within human islets inhibits β cell function.J Clin Invest 102: 516–526.

Arnush M, Scarim AL, Heitmeier MR et al. (1998b) Potential role ofresident islet macrophage activation in the initiation of autoimmunediabetes. J Immunol 160: 2684–2691.

Bach JF (1994) Insulin-dependent diabetes mellitus as an autoimmunedisease. Endocr Rev 15: 516–542.

Bendtzen K, Mandrup-Poulsen T, Nerup J, Nielsen JH, Dinarello CA(1986) Cytotoxicity of human pI interleukin-1 for pancreatic islets ofLangerhans. Science 232: 1545–1547.

Bottazzo GF, Dean BM, McNally JM, MacKay EH, Swift PGF,Gamble DR (1985) In situ characterization of autoimmune phenomenaand expression of HLA molecules in the pancreas in diabetic insulitis.N Engl J Med 313: 353–360.

Cailleau C, Diu-Hercend A, Ruuth E, Westwood R, Carnaud C(1997) Treatment with neutralising antibodies specific for IL-1β pre-vents cyclophosphamide-induced diabetes in non-obese diabetic mice.Diabetes 46: 937–940.

Campbell IL, Cutri A, Wilson A, Harrison LC (1989) Evidence for IL-6production by and effects on the pancreatic β-cell. J Immunol 143:1188–1191.

Cetkovic-Cvrlje M, Eizirik DL (1994) TNF-α and IFN-γ potentiatesthe deleterious effects of IL-1β on mouse pancreatic islets mainly viageneration of nitric oxide. Cytokine 6: 399–406.

Cetkovic-Cvrlje M, Sandler S, Eizirik DL (1993) Nicotinamide anddexamethasone inhibit interleukin-1-induced nitric oxide productionby RINm5F cells without decreasing messenger ribonucleic acidexpression for nitric oxide synthase. Endocrinology 133: 1739–1743.

Dinarello CA (1991) Interleukin-1 and interleukin-1 antagonism. Blood77: 1627–1652.

Eisenbarth GS (1986) Type 1 diabetes mellitus. A chronic autoimmunedisease. N Engl J Med 314: 1360–1368.

Eizirik DL, Cagliero E, Bjorklund A, Welsh N (1992) Interleukin-1β

induces the expression of an isoform of nitric oxide synthase ininsulin-producing cells, which is similar to that observed in activatedmacrophages. FEBS Lett 308: 249–252.

Eizirik DL, Sandler S, Welsh N, Cetkovic-Cvrlje M, Nieman A,Geller DA, Pipeleers DG, Berndtzen K, Hellerstrom C (1994)Cytokines suppress human islet function irrespective of their effectson nitric oxide generation. J Clin Invest 93: 1968–1974.

Elliott RB, Chase HP (1991) Prevention or delay of type 1(insulin-dependent) diabetes mellitus in children using nicotinamide.Diabetologia 34: 362–365.

Foulis AK, Farquharson MA, Meager A (1987) Immunoreactiveα-interferon in insulin-secreting β cells in type 1 diabetes mellitus.Lancet 2: 1423–1427.

Foulis AK, Liddle CN, Farquharson MA, Richmond JA, Weir RS (1986)The histopathology of the pancreas in type 1 (insulin-dependent) dia-betes mellitus: a 25-year review of deaths in patients under 20 yearsof age in the United Kingdom. Diabetologia 29: 267–274.

Foulis AK, McGill M, Farquharson MA (1991) Insulitis in type 1 (insulin-dependent) diabetes mellitus in man-macrophages, lymphocytes, andinterferon-γ containing cells. J Pathol 165: 97–103.

Fukuzawa M, Satoh J, Muto G, Nischimura S, Miyaguchi S, Qiang XL,Toyota T (1997) Inhibitory effect of nicotinamide on in vitro and in vivoproduction of tumour necrosis factor-α. Immunol Lett 59: 7–11.

Hammonds P, Beggs M, Beresford G, Espinal J, Clarke J, Mertz RJ (1990)Insulin-secreting β-cells possess specific receptors for interleukin-1β.FEBS Lett 261: 97–100.

Hancock WW, Polanski M, Zhang J, Blogg N, Weiner HL (1995)Suppression of insulitis in non-obese diabetic (NOD) mice byoral insulin administration is associated with selective expres-sion of interleukin-4 and -10, transforming growth factor-β, andprostaglandin-E. Am J Pathol 147: 1193–1199.

Hauschildt S, Scheipers P, Bessler WG (1991) Inhibitors of poly(ADP-ribose) polymerase suppress lipopolysaccharide-induced nitriteformation in macrophages. Biochem Biophys Res Commun 179:865–871.

Hauschildt S, Scheipers P, Bessler WG, Mulsch A (1992) Induction ofnitric oxide synthase in L929 cells by tumour necrosis factor-α is pre-vented by inhibitors of poly(ADP-ribose) polymerase. Biochem J 288:255–260.

Honegger J, Spagnoli A, D’urso R, Navarra P, Tsagarakis S, Besser GM,Grossman AB (1991) Interleukin-1β modulates the acute releaseof growth hormone-releasing hormone and somatostatin from rathypothalamus in vitro, whereas tumor necrosis factor and interleukin-6have no effect. Endocrinology 129: 1275–1282.

Huang X, Yuan J, Goddard A et al. (1995) Interferon expression in thepancreases of patients with type I diabetes. Diabetes 44: 658–664.

Itoh N, Hanafusa T, Miyazaki A et al. (1993) Mononuclear cell infiltrationand its relation to the expression of major histocompatibility complexantigens and adhesion molecules in pancreas biopsy specimens fromnewly diagnosed insulin-dependent diabetes patients. J Clin Invest 92:2313–2322.

Jacob CO, Aiso S, Mitchie SA, McDevitt HO, Acha-Orbea H (1990)Prevention of diabetes in non-obese diabetic mice by tumour necrosisfactor (TNF). Similarities between TNF-α and interleukin. Proc NatlAcad Sci USA 87: 968–972.

Jafarian-Tehrani M, Amrani A, Homo-delarche F, Marquette C,Dardenne M, Haour F (1995) Localization and characterization ofinterleukin-1 receptors in the islets of Langerhans from control andnon-obese diabetic mice. Endocrinology 136: 609–613.

Kay TWH, Campbell IL, Harrison LC (1991) Characterization of pan-creatic T lymphocytes associated with beta cell destruction in the non-obese diabetic (NOD) mouse. J Autoimmun 4: 263–276.

Kolb H, Burkart V, Appels B, Hanenberg H, Kantwerk-Funke G,Kiesel U, Funda J, Schraermeyer U, Kolb-Bachofen V (1990) Essentialcontribution of macrophages to islet cell destruction in vivo and in vitro.J Autoimmun 3: 117–120.

Kretowski A, Mysliwiec J, Szelachowska M, Kinalski M, Kinalska I(2000) Nicotinamide inhibits enhanced in vitro production ofinterleukin-12 and tumour necrosis factor-α in peripheral whole bloodof people at high risk of developing Type 1 diabetes and people withnewly diagnosed Type 1 diabetes. Diabetes Res Clin Prac 47: 81–86.

Kwon G, Corbett JA, Rodi CP, Sullivan P, McDaniel ML (1995)Interleukin-1β-induced nitric oxide synthase expression by rat pan-creatic β-cells: evidence for the involvement of nuclear factor κB inthe signalling mechanism. Endocrinology 136: 4790–4795.

Lechan, RM, Toni R, Clark BD, Cannon JG, Shaw AR, Dinarello CA,Reichlin S (1990) Immunoreactive interleukin-1β localization in therat forebrain. Brain Res 514: 135–140.

Loweth AC, Williams GT, James RFL, Scarpello JHB, Morgan NG(1998) Human islets of Langerhans express Fas ligand and undergoapotosis in response to interleukin-1β and Fas ligation. Diabetes 47:727–732.

Mandrup-Poulsen T (1996) The role of interleukin-1 in the pathogenesisof IDDM. Diabetologia 39: 1005–1029.

Mandrup-Poulsen T, Bendtzen K, Nielsen JH, Bendixen G, Nerup J(1985) Cytokines cause functional and structural damage to isolatedislets of Langerhans. Allergy 40: 424–429.

Interleukin-1β in islets of NOD mice 327

Mandrup-Poulsen T, Helqvist S, Molvig J, Wogensen LD, Nerup J (1989)Cytokines as immune effector molecules in autoimmune endocrinediseases with special reference to insulin-dependent diabetes mellitus.Autoimmunity 4: 191–218.

Mandrup-Poulsen T, Helqvist S, Wogensen LD, Molvig J, Pociot F,Johannsen J, Nerup J (1990) Cytokines and free radicals as effectormolecules in the destruction of pancreatic beta cells. Current TopicsMicrobiol Immunol 164: 169–173.

Moriwaki M, Itoh N, Miyagawa J, Yamamoto K et al. (1999) Fasand Fas ligand expression in inflamed islets in pancreas sections ofpatients with recent-onset Type I diabetes mellitus. Diabetologia 42:1332–1340.

Mossman TR, Coffman RL (1989) Th1 and Th2 cells: different patterns oflymphokine secretion lead to different functional properties. Ann RevImmunol 7: 145–173.

Nicoletti F, Dimarco R, Barcellini W et al. (1994) Protection from exper-imental autoimmune diabetes in the non-obese diabetic mouse withsoluble interleukin-1 receptor. Eur J Immunol 24: 1843–1847

Oliver FJ, Menissier-de Murcia J, Nacci C et al. (1999) Resistance toendotoxic shock as a consequence of defective NF-κB activation inpoly (ADP-ribose) polymerase-1 deficient mice. EMBO J 18: 4446–4454.

Pilstrom B, Bjork L, Bohme J (1997) Monokine-producing cells predom-inate in the recruitment phase of NOD insulitis while cells producingTh1-type cytokines characterize the effector phase. J Autoimmun 10:147–155.

Pociot F, Reimers JI, Andersen HU (1993) Nicotinamide: biologicalactions and therapeutic potential in diabetes prevention. Diabetologia36: 574–576.

Pozzilli P (1998) Prevention of insulin-dependent diabetes mellitus.Diabetes/Metab Rev 14: 69–84.

Rabinovitch A (1994) Immunoregulatory and cytokine imbalancesin the pathogenesis of IDDM: Therapeutic intervention byimmunostimulation? Diabetes 43: 613–621.

Rabinovitch A (1998) An update on cytokines in the pathogene-sis of insulin-dependent diabetes mellitus. Diabetes Metab Rev14: 129–151.

Rabinovitch A, Suarez-Pinson WL, Strydnaka K, Schultz R, Lakey,Warnock GL et al. (1994) Human pancreatic islet β destruction bycytokines is independent of nitric oxide production. J Clin EndocrinolMetab 79: 1058–1062.

Rapport MJ, Jaramillo A, Zipris D et al. (1993) IL-4 reverses T-cellproliferative unresponsiveness and prevents the onset of diabetes inNOD mice. J Exp Med 178: 87–99.

Reddy S, Bibby NJ, Elliott RB (1990) Early nicotinamide treatment inthe NOD mouse: effects on diabetes and insulitis suppression andautoantibody levels. Diabetes Res 15: 95–102.

Reddy S, Salari-Lak N, Sandler S (1995) Long-term effects of nicoti-namide induced inhibition of poly(adenosine diphosphate-ribose)polymerase activity in rat pancreatic islets exposed to interleukin-1β.Endocrinology 136: 1907–1912.

Reddy S, Wu D, Swinney C, Elliott RB (1995) Immunohistochemicalanalyses of pancreatic macrophages and CD4 and CD8 T cell subsetsprior to and following diabetes in the NOD mouse. Pancreas 11: 16–25.

Reddy S, Yip S, Karanam M, Poole CA, Ross JM (1999) An immuno-histochemical study of macrophage influx and the co-localizationof inducible nitric oxide synthase in the pancreas of non-obese(NOD) mice during disease acceleration with cyclophosphamide.Histochemical J 31: 303–314.

Reddy S, Karanam M, Krissansen G, Nitschke K, Neve J, Poole CA,Ross JM (2000a) Temporal relationship between immune cell influxand the expression of inducible nitric oxide synthase, interleukin-4

and interferon-γ in pancreatic islets of NOD mice following adoptivetransfer of diabetic spleen cells. Histochemical J 32: 195–206.

Reddy S, Karanam M, Poole CA, Ross JM (2000b) Dual-label immuno-histochemical study of interleukin-4- and interferon-γ -expressingcells within the pancreas of the NOD mouse during disease accel-eration with cyclophosphamide. Autoimmunity 32: 181–192.

Reddy S, Karanam M, Robinson E (2001) Prevention ofcyclophosphamide-induced accelerated diabetes in the NOD mouseby nicotinamide or a soy protein-based infant formula. Int J ExptlDiabetes Res 1: 299–313.

Sandler S, Andersson A, Hellerstrom C (1987) Inhibitory effectsof interleukin-1 on insulin secretion, insulin biosynthesis, andmetabolism of isolated rat pancreatic rat islets. Endocrinology 121:1424–1431.

Somoza N, Vargas F, Roura-Mir C et al. (1994) Pancreas in recentonset insulin-dependent diabetes mellitus. Changes in HLA, adhesionmolecules and autoantigens, restricted T cell receptor Vβ usage, andcytokine profile. J Immunol 153: 1360–1377.

Southern C, Schulster D, Green IC (1990) Inhibition of insulin secretionby interleukin-1β and tumour necrosis factor-α via an L-arginine-dependent nitric oxide generating mechanism. FEBS Lett 276: 42–44.

Stassi G, De Maria R, Trucco G, Rudert W, Testi R, Galluzzo A,Giordano C, Trucco M (1997) Nitric oxide primes pancreatic β cellsfor Fas-mediated destruction in insulin-dependent diabetes mellitus.J Exp Med 186: 1193–1200.

Stassi G, Todaro M, Richiusa P et al. (1995) Expression ofapoptosis-inducing CD95 (Fas/Apo-1) on human β-cells sorted byflow-cytometry and cultured in vitro. Transplant Proc 27: 3271–3275.

Suarez-Pinzon W, Sorensen O, Bleackley RC, Elliott JF, Rajotte RV,Rabinovitch A (1999) β-cell destruction in NOD mice correlateswith Fas(CD95) expression on β-cells and proinflammatory cytokineexpression in islets. Diabetes 48: 21–28.

Szabo C, Virag L, Cuzzocrea S et al. (1998) Protection againstperoxynitrite-induced fibroblast injury and arthritis development byinhibition of poly(ADP-ribose) synthatase. Proc Natl Acad Sci USA95: 3867–3872.

Welsh M, Welsh N, Bendtzen K, Mares J, Strandell E, Oberg C, Sandler S(1995) Comparison of mRNA contents of interleukin-1β and nitricoxide synthase in pancreatic islets isolated from female and malenonobese diabetic mice. Diabetologia 38: 153–160.

Yamada K, Nonaka K, Hanafusa T, Miyazaki A, Toyoshima H, Tarui S(1982) Preventive and therapeutic effects of large dose nicotinamideinjections on diabetes associated with insulitis: an observation innon-obese diabetic (NOD) mice. Diabetes 31: 749–753.

Yamada K, Takane N, Otabe S, Inada C, Inoue M, Nonaka K (1993)Pancreatic β-cell-selective production of tumour necrosis factor-αinduced by interleukin-1. Diabetes 42: 1026–1031.

Yamada K, Takane-Gyotoku N, Yuan X, Ichikawa F, Inada C, Nonaka K(1996) Mouse islet cell lysis mediated by interleukin-1-induced Fas.Diabetologia 39: 1306–1312.

Yamagata K, Nakajima H, Tomita K et al. (1996) Dominant TCR α-chainclonotypes and interferon-γ are expressed in the pancreas of patientswith recent-onset insulin-dependent diabetes mellitus. Diabetes ResClin Pract 34: 37–46.

Yonemura Y, Takashima T, Miwa K, Miyazaki T, Yamamoto H,Okamoto H (1984) Amelioration of diabetes mellitus in partially de-pancreatized rats by poly(ADP-ribose) synthetase inhibitors. Evidenceof islet B-cell regeneration. Diabetes 33: 401–404.

Zingarelli B, Salzman AL, Szabo C (1998) Genetic disruption of poly(ADP-ribose) synthetase inhibits the expression of P-selectin andintercellular adhesion molecule-1 in myocardial ischemia/reperfusioninjury. Circ Res 83: 85–94.