Dominant Negative MyD88 Proteins Inhibit Interleukin-1 b / Interferon- g -mediated Induction of Nuclear Factor k B-dependent Nitrite Production and Apoptosis in b Cells*

Insulin-dependent diabetes mellitus is an autoimmune disease in which pancreatic islet b cells are destroyed by a combination of immunological and inflammatory mechanisms. In particular, cytokine-induced production of nitric oxide has been shown to correlate with b cell apoptosis and/or inhibition of insulin secretion. In the present study, we investigated whether the interleukin (IL)-1 b intracellular signal transduction pathway could be blocked by overexpression of dominant negative forms of the IL-1 receptor interacting protein MyD88. We show that overexpression of the Toll domain or the lpr mutant of MyD88 in b Tc-Tet cells decreased nuclear factor k B (NF- k B) activation upon IL-1 b and IL-1 b /interferon (IFN)- g stimulation. Inducible nitric oxide synthase mRNA accumulation and nitrite production, which required the simultaneous presence of IL-1 b and IFN- g , were also suppressed by ; 70%, and these cells were more resistant to cytokine-induced apoptosis as compared with parental cells. The decrease in glucose-stimulated insulin secretion induced by IL-1 b and IFN- g was however not prevented. This was because these dysfunctions were induced by IFN- g alone, which decreased cellular insulin content and stimulated insulin exocytosis. These results

Insulin-dependent diabetes mellitus is an autoimmune disease in which pancreatic islet ␤ cells are destroyed by a combination of immunological and inflammatory mechanisms. In particular, cytokine-induced production of nitric oxide has been shown to correlate with ␤ cell apoptosis and/or inhibition of insulin secretion. In the present study, we investigated whether the interleukin (IL)-1␤ intracellular signal transduction pathway could be blocked by overexpression of dominant negative forms of the IL-1 receptor interacting protein MyD88. We show that overexpression of the Toll domain or the lpr mutant of MyD88 in ␤Tc-Tet cells decreased nuclear factor B (NF-B) activation upon IL-1␤ and IL-1␤/interferon (IFN)-␥ stimulation. Inducible nitric oxide synthase mRNA accumulation and nitrite production, which required the simultaneous presence of IL-1␤ and IFN-␥, were also suppressed by ϳ70%, and these cells were more resistant to cytokine-induced apoptosis as compared with parental cells. The decrease in glucose-stimulated insulin secretion induced by IL-1␤ and IFN-␥ was however not prevented. This was because these dysfunctions were induced by IFN-␥ alone, which decreased cellular insulin content and stimulated insulin exocytosis. These results demonstrate that IL-1␤ is involved in inducible nitric oxide synthase gene expression and induction of apoptosis in mouse ␤ cells but does not contribute to impaired glucose-stimulated insulin secretion. Furthermore, our data show that IL-1␤ cellular actions can be blocked by expression of MyD88 dominant negative proteins and, finally, that cytokine-induced ␤ cell secretory dysfunctions are due to the action of IFN-␥.
Type I or insulin-dependent diabetes mellitus is an autoimmune disease causing specific destruction of the insulin-producing ␤ cells of the islets of Langerhans (1). Cytokines, in particular IL-1␤, 1 IFN-␥, and TNF-␣, are thought to play an important role in this destruction mechanism, but their exact role in vivo is not completely understood. They are however involved in both direct effects through control of gene expression by ␤ cells and in indirect effects through activation of endothelial and inflammatory cells present within the islets. The particular role of IL-1␤ has been much studied. In islets tested in vitro, IL-1␤ has been proposed to directly suppress ␤ cell function through induction of the endogenous inducible nitric oxide synthase (iNOS) and the production of NO (2). However, most of this early work has focused on rat islets and insulinomas, which appear to be exquisitely sensitive to IL-1␤induced cell death and inhibition of insulin secretion. In contrast, there are conflicting recent results regarding the role of NO in human islet ␤ cell destruction, with many groups suggesting that NO may only partly or not at all be involved (3)(4)(5). Studies on isolated mouse ␤ cells or insulinomas have suggested the following intermediate situation: first, IL-1␤ alone does not appear to affect primary mouse ␤ cells (6), and NO generated by cytokine (IL-1␤/IFN-␥)-treated islets may only be one of the ways by which mouse ␤ cells are destroyed (7,8). In addition, a recent report suggested that NO, produced at lower levels, could even protect against necrosis induced by alloxan and streptozotocin in purified rat ␤ cells stimulated with IL-1␤ only (9). This study also showed that IL-1␤ stimulation of rat ␤ cellsledtoanNO-dependentincreaseinexpressionofmanganesedependent superoxide dismutase, heat-shock protein of 70 kDa, and heme-oxygenase 1, enzymes that are involved in the natural defense mechanisms against reactive oxygen species.
IL-1␤ action depends on its interaction with a specific plasma membrane receptor composed of the IL-1 receptor and its accessory protein (IL-1R AcP ). Activation of the receptor leads to association of MyD88 with the IL-1/IL-1R AcP complex and subsequent activation of c-Jun N-terminal kinase and NF-B via recruitment of a TNF receptor-associated factor 6/ IL-1 receptor-associated kinase-dependent pathway (10,11). The MyD88 protein is an adaptor molecule that also participates in the IL-18 and lipopolysaccharide receptor signaling pathways in mammals. It contains two protein-protein interaction domains, a Drosophila-like Toll domain and a Death domain. The Drosophila Toll receptor is an homologue of the mammalian IL-1 and lipopolysaccharide receptors. It leads to the activation of the Pelle-like kinase, which has a strong similarity to the IL-1 receptor-associated kinase that phosphorylates and targets to degradation the IkB-like Cactus protein, thereby allowing the NF-B homologue Dorsal to enter the nucleus and activate transcription. It is thus believed that MyD88 is involved in an ancient and very conserved innate immune response to microorganism mediated by the lipopolysaccharide and IL-1 receptor family (11)(12)(13)(14). MyD88 function depends on homodimerization, which requires intact Toll and Death domains and association of the protein with the IL-1 receptor through the Toll domain. Important for the present study, it has been recently reported that interference with the intracellular IL-1 signaling pathway can be achieved by overexpressing either the Toll domain of MyD88 or the adaptor containing a Death domain mutation. This mutation (MyD88F56N) is similar to the lpr mutation found in the Death domain of the FAS receptor, which blocks the Fas-Fadd interaction (11). These two dominant negative mutants of the MyD88 protein (MyD88Toll and MyD88lpr) could therefore be used to suppress the IL-1 intracellular signaling pathway at a site close to receptor activation.
To evaluate the potential of blocking this pathway by gene transfer in ␤ cells and the consequence of this inhibition on NF-B activation, NO production, and GSIS in response to cytokine treatment, we used the conditionally immortalized ␤Tc-Tet cells. These cells have been established in culture from islets of transgenic mice expressing the SV-40 T antigen under the control of the tetracycline operator/tetracycline transactivator system (15). They can be growth-arrested in the presence of tetracycline, and, in the non-proliferating state, their insulin content is similar to that of native ␤ cells, and they secrete insulin with the normal glucose dose dependence (16). Importantly, when transplanted under the kidney capsule of diabetic syngeneic mice, these cells can control blood glucose for several months (15,17). They are however rejected when transplanted in NOD mice. 2 These cells thus represent a unique model to evaluate the action of cytokines on ␤ cell function and survival when transplanted in a diabetic environment. We previously demonstrated that Bcl-2 overexpression in ␤Tc-Tet cells increased resistance against certain apoptotic stresses, including staurosporine, hypoxia, and cytokines (17).
Here we show that the presence of Bcl-2 did not change the ability of the cells to up-regulate the inos gene and to generate NO in response to IL-1␤ and IFN-␥. We show that IL-1␤ receptor signal transduction in mouse ␤ cell is MyD88-mediated and that the use of dominant negative forms of this protein could attenuate IL-1␤/IFN-␥-induced NO generation. Furthermore, we showed that the cells overexpressing the MyD88 dominant negative proteins showed increased resistance to cytokine-induced cell death and maintained their insulin secretory response to glucose. However, these cells treated with IFN-␥ plus IL-1␤ or IFN-␥ alone showed reduced GSIS, which could be explained by an IFN-␥-induced decrease of their intracellular insulin content and a decrease in the secretory activity stimulated by glucose.

MATERIALS AND METHODS
Cell Culture-␤Tc-Tet or CDM3D cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 25 mM glucose and supplemented with 15% horse serum (Amimed), 2.5% fetal bovine serum (Life Technologies, Inc.), 10 mM Hepes, 1 mM sodium pyruvate, 2 mM glutamine, at 37°C with 5% CO 2 . CDM3D cells are ␤Tc-Tet cells that have been modified to overexpress Bcl-2 (17). Growth arrest was induced by including 1 g/ml of tetracycline or 1 g/ml of deoxycycline (Fluka). For nitrite secretion measurements, medium was changed to RPMI 1640, which has a lower level of nitrite content.
Analysis of iNOS and MyD88 Expression by Northern Blot-Total RNA was isolated and analyzed by Northern (RNA) blot as described previously using specific probes prepared by random primer labeling (18).
Preparation of Lentiviral Vectors and Infection of ␤ Cells-The Toll and MyD88-lpr cDNA were kindly provided by Dr. J. Tschopp (Department of Biochemistry, University of Lausanne) (11). They both contain a FLAG epitope. They were subcloned into a modified SIN-18-phospho-glycerate kinase-woodchuck hepatitis virus vector (19,20), which contains a neomycin-resistance gene downstream of an internal ribosome entry site from encephalomyocardiatis virus (kindly provided by Dr. N. Deglon, Gene Therapy Center, University Hospital, Lausanne, Switzerland). High titer stocks of lentiviral vectors packaged by the multiply attenuated lentivirus CMV⌬R8.91 and pseudotyped with the vesicular stomatitis virus-G envelope protein (plasmid pMD-G) were prepared by transient transfection of 293T cells as described (17,21). Viral stocks of LacZ virus were titered on human 293T or rat 208F fibroblasts in six serial dilutions (5-to 3125-fold), and the viral titer was determined by counting the number of blue cell foci per well and dividing by the dilution factor. MyD88 virus titers were determined by p24 enzyme-linked immunosorbent assay and neomycin-resistant colony formation in infected 208F fibroblasts as described (21). Pure viral stocks were tested for the presence of replication competent retroviruses using the HeLa-p4 assay (22). In addition, target cells were co-cultivated with HeLa-p4 cells for 1 month, and thereafter the HeLa-p4 cells were tested for the presence of ␤-galactosidase enzymatic activity, or target cell supernatant was applied to virgin HeLa-p4 cells. None of these tests revealed the presence of tat-transducing activity above background level. CDM3D cells were transduced with an multiplicity of infection of 20, and G418 selection (400 g/ml) was applied 48 h after infection to select the pool of infected cells.
Apoptosis-24 h before induction of apoptosis, proliferating or growth-arrested ␤ cells were distributed over poly-lysine-coated 96-well microtiter plates (3 ϫ 10 3 cells/well). Medium was changed the next day, and the indicated amount of cytokines was added in the presence or absence of tetracycline. The recombinant mouse cytokines IL-1␤, TNF-␣, and IFN-␥ were purchased from Life Technologies, Inc. and were used at the indicated final concentration (unit/ml). The percentage of viable, apoptotic, and dead or necrotic cells were assessed as described previously (17,23,24). Medium was removed from the well and replaced with the same volume of medium containing 20 g/ml Hoechst 33342 (Fluka) and propidium iodide 10 g/ml (Sigma). After 5 min at room temperature, the cells were examined with an inverted fluorescence microscope with ultraviolet excitation at 340 -380 nm. In each experimental condition at least 500 cells were counted. A control plate was analyzed in parallel to determine spontaneous cell death, which was deduced from the in experimental values.
Insulin Secretion-Cells were plated in 24-well dishes at a density of 10 5 /well 48 h before incubation with cytokines. Following cytokine exposure for 48 h, cells were then incubated for 1 h in Hepes-buffered Krebs-Ringer bicarbonate buffer, pH 7.4 (KRBH), containing 0.5% bovine serum albumin with 2.8 mM glucose and 250 M isobutylmethylxanthine (Sigma). The medium was changed again with KRBH, 0.5% bovine serum albumin containing 2.8 or 16.7 mM glucose and isobutylmethylxanthine. Secreted insulin was quantitated by radioimmunoassay (Linco Research Inc.) as described (17,25). Intracellular insulin was determined in acid-ethanol cell lysates. Briefly, cells were lysed in 250 l of 75% ethanol, 1.5% concentrated hydrochloric acid. Aliquots of cell lysates were also analyzed for DNA content (26) to normalize the secretion data. Lysates in acid ethanol were neutralized with one-tenth volume of 1M Na 2 CO 3 , and DNA content was determined by fluorescence using a Fluoroskan-II microplate fluorometer (Labsystems, Helsinki, Finland) with the excitation filter set at 355 nm and the emission filter set at 460 nm.
Cytokine-induced Nitrite Accumulation-Nitrite accumulation in the conditioned culture medium was detected spectrophotometrically (at 540 nm) by the Griess reaction in the presence of 1 mM sulfanilamide and 0.1 M HCl (27). Concentrations (picomoles of NO 2 Ϫ /mg of protein) were calculated from the absorption before (A1) and after (A2) the addition of 70 mM naphthylethylenediamine and compared with a standard curve derived from NaNO 2 (0 to 20 M). Values shown are mean Ϯ standard deviation of at least three independent experiments performed in duplicate.
Western Blot-The MyD88-lpr or Toll proteins were detected in immunoblot using the anti-FLAG-M2 (Sigma) mouse monoclonal antibody as described (11). The iNOS protein level was determined in cell lysates prepared in 1% Triton X-100, 0.15M sodium chloride, and 10 mM Tris, pH 7.4 with 50 g/ml of phenylmethylsulfonyl fluoride and 2 g/ml of aprotinin at 4°C for 10 min. Western blot was performed as described previously (28) cytomegalovirus promoter), that was used to correct for transfection efficiency. 48 h after transfection, cells were stimulated with cytokines for 3 h, and relative activity of luciferase and ␤-galactosidase was determined as described (11).

Stimulation of iNOS Expression in CDM3D Cells by IL-1␤ and IFN-␥-
The effect of the pro-inflammatory cytokines on induction of iNOS was tested using the conditionally immortalized murine ␤Tc-Tet cells that had been previously modified to stably express Bcl-2. These cells, referred to as CDM3D, have been shown to be easier to cultivate in standard medium, with a much reduced basal level of apoptosis, to fully conserve the capability of the parental cell line to be growth-arrested by tetracycline, and to secrete insulin with a normal glucose dose dependence (17). IL-1␤, IFN-␥, and TNF-␣, alone or in various combinations, were tested for their ability to stimulate NO production, as assessed by measuring the amount of nitrite in the supernatant of exposed cells. IL-1␤ at 10 units/ml combined with IFN-␥ at 150 units/ml induced an NO production close to the maximal production achieved when the cells were exposed to a combination of all three cytokines (100 units/ml of IL-1␤, 100 units/ml of TNF-␣, and 150 units/ml of IFN-␥) (Fig. 1A). Adding IL-1␤ to higher doses in the presence of 150 units/ml of IFN-␥ did not further increase NO production (not shown). Production of NO was associated with an increase in iNOS mRNA levels that also required the combination of IL-1␤ and IFN-␥ (same dose as above) (Fig. 1B).
Lentivirus-mediated Transfer of the Dominant Negative forms of MyD88 in CDM3D Cells-CDM3D cells were infected with recombinant lentiviruses directing the expression of the Toll domain or lpr mutant of MyD88 and the neomycin gene. Pools of infected cells were selected and expanded in the presence of G418. The proteins were detected in transiently transfected 293T cells by Western blot analysis ( Fig. 2A). In the infected CDM3D cells, a lower level of expression did not permit detection with the FLAG antibody. Expression was however confirmed by Northern blot analysis (Fig. 2B). To test whether the MyD88 mutant-expressing cells were still functional in correcting diabetes in vivo, they were implanted under the kidney capsule of streptozotocin-diabetic C3H syngeneic mice. Correction of hyperglycemia was observed for one month, after which the animals were killed (data not shown).
Expression of MyD88-Toll or lpr Mutant Proteins Inhibit Induction of iNOS mRNA and NO Generation by IL-1␤/IFN-␥-CDM3D cells, MyD88-Toll CDM3D, and MyD88-lpr CDM3D cells were exposed to IL-1␤/IFN-␥ for 18 h, a time that gives the highest expression of iNOS mRNA. Fig. 3A shows that the mRNA for iNOS was strongly induced in the CDM3D cells but to a much lesser extent in the MyD88-Toll or lproverexpressing ␤ cells. Expression of iNOS protein in CDM3D cells was detected 24 h after initiation of cytokine treatment (Fig. 3B). Inducible NOS was undetectable in the MyD88-Toll CDM3D cells and reduced 4-fold in MyD88-lpr CDM3D cells as compared with parental cells. In agreement with the above results, accumulation of nitrite in the conditioned medium of cytokine-treated cells was decreased by approximately 70% in the two MyD88 mutant-expressing cell lines (Fig. 3C).

MyD88-Toll or lpr Mutants Inhibit inos Gene Transcription at a Level Upstream of NF-B Activation-NF-B
is the main factor controlling rodent inos gene transcription induced by IL-1␤ (30). We thus tested whether the lack of iNOS mRNA accumulation in the presence of the MyD88 dominant negative proteins was due to impaired activation of NF-B. To address this question, an NF-B-luciferase reporter gene and expression vector for either MyD88 dominant negative inhibitor were co-transfected in CDM3D cells. The cells were stimulated with cytokines for 3 h, and luciferase activity was measured. As shown in Fig. 4A, IL-1␤/IFN-␥ stimulation resulted in 2.9-fold Ϯ 0.3 (n ϭ 3) activation of the NF-B reporter construct. Interestingly, stimulation with IL-1␤ only also activated the reporter construct, albeit to a lower level (2.2-fold Ϯ 0.2 (n ϭ 3)). Co-transfection of MyD88-Toll with the NF-B reporter construct decreased both basal (0.7 Ϯ 0.1 (n ϭ 3, p Ͻ 0.05 versus untransfected cells)) and IL-␤/IFN-␥-stimulated (1.6 Ϯ 0.3-fold activation (n ϭ 3, p Յ 0.001)) of transcriptional activity. MyD88-lpr co-expression totally suppressed induction of the NF-B reporter by both IL-1␤ or IL-1␤ and IFN-␥ (p Ͻ 005, n ϭ 3), and the basal level was again significantly reduced as compared with cells tranfected with the NF-B reporter only (n ϭ 3, p Ͻ 0.05). When an iNOS reporter construct was co-transfected with the MyD88-Toll or lpr proteins (Fig. 4B) a near complete inhibition of iNOS reporter induction in response to IL-1␤/IFN-␥ stimulation was observed (Fig. 4B). As expected, IL-1␤ alone was unable to induce the iNOS reporter gene even at higher doses (data not shown). In addition the basal level of reporter activity was significantly reduced in both cells as compared with CDM3D cells (p Ͻ 0.001, n ϭ 3).
MyD88-Toll or lpr Mutant Proteins Increase Protection of CDM3D Cells from Cytokine-induced Apoptosis-We have shown previously that Bcl-2 expression in CDM3D cells conferred partial protection against cytokine-mediated induction of apoptosis (17). Fig. 5 shows that CDM3D, MyD88-Toll, and MyD88-lpr CDM3D cells were similarly resistant to a combination of 1000 units/ml of TNF-␣ and IFN-␥. Exposure of the cells to increasing concentrations of IL-1␤ (10 to 500 units/ml) in the presence of 1000 units/ml of TNF-␣ and IFN-␥ revealed an increased resistance of MyD88 mutant protein-expressing cells to induction of apoptosis.
Glucose-induced Insulin Secretion in Cytokine-treated Cells-To determine whether expression of MyD88-Toll or MyD88-lpr also protected CDM3D cells from inhibition of glucose-stimulated insulin secretion by cytokines, we performed secretion experiments on cytokine-treated cells. CDM3D cells, expressing or not expressing the MyD88 mutants, were exposed to 10 units/ml IL-1␤, 150 units/ml IFN-␥, or a combination of both for 48 h, and the cells were then exposed to 2.8 or 16.7 mM glucose for 1 h after a 1-h preincubation at 2.8 mM glucose. Fig. 6a shows that the rate of insulin secretion at 2.8 mM glucose after the various treatments was similar in the three cell lines. Secretion at 16.7 mM glucose was markedly stimulated in control conditions in the three lines tested, and pre-exposure of the cells to IL-1␤ did not affect GSIS. However, pretreatment with IFN-␥ markedly decreased GSIS in all three lines and had a similar effect as the combination of IL-1␤ and IFN-␥. To determine whether the decrease in GSIS was due to a defect in stimulated exocytosis or to a decrease in insulin content, total cellular insulin content was measured in each condition. Insulin content was slightly, but not significantly, decreased in IL-1␤-treated cells but markedly decreased in IFN-␥ or IL-1␤/IFN-␥-treated cells (Fig. 6b). This decrease in insulin content could not be explained by a decrease in cell viability, because at the concentrations of cytokines used no apoptosis was detected. Furthermore, measurement of cellular DNA content at the end of the secretion experiments showed similar content in all conditions. When the secretion was expressed as a percent of the total intracellular insulin content (Fig. 6c), the secretion rate of IFN-␥-and IL-1␤/IFN-␥-treated cells was also significantly decreased, whereas the secretion rate of IL-1␤-treated cells was not different from that of control cells.

DISCUSSION
Here we studied the effect of cytokines on the mouse CDM3D cells, which were derived from ␤Tc-Tet cells by overexpressing the Bcl-2 gene, and the possibility to further genetically modify these cells to make them resistant to the action of IL-1␤. Using CDM3D cells and CDM3D cells expressing MyD88 dominant negative proteins, we showed the following: 1) induction of iNOS gene expression and NO production requires the combination of IL-1␤ and IFN-␥; 2) the IL-1␤ intracellular signaling pathway leading to iNOS gene expression can be inhibited by MyD88 dominant negative proteins; 3) the presence of these dominant negative proteins improved resistance to cytokineinduced apoptosis, and 4) the reduction of GSIS was not prevented by the MyD88 mutant proteins, because it resulted mostly from the action of IFN-␥ on cellular insulin content and on the secretory response.
The induction of iNOS mRNA and NO production by the CDM3D cells requires the combined action of IL-1␤ and IFN-␥. This indicates that both cytokine-dependent intracellular signaling pathways need to be activated to increase iNOS gene expression. These data are similar to those obtained with human ␤ cells where iNOS gene expression also requires the combined exposure to both IL-1␤ and IFN-␥. This is however in contrast to previous findings by others studying rat islets or purified rat ␤ cells where iNOS gene induction is strongly activated by IL-1␤ alone (3,4). From this perspective, studying mouse islets may therefore be more relevant to the understanding of cytokine action on human ␤ cells in the pathogenesis of type I diabetes. However, whether the same intracellular pathways are activated by these cytokines in mouse and human ␤ cells is not yet established.
Expression of the Toll domain of MyD88 and of MyD88-lpr markedly reduced activation of NF-B transcriptional activity in response to IL-1␤ and IFN-␥ and induction of iNOS mRNA, protein expression, and NO production. As these dominant negative forms of MyD88 do not interact with the IFN-␥ signaling pathway, these data demonstrate that in the ␤ cell line studied, IL-1␤ intracellular signaling depends on MyD88 interaction with the IL-1 receptor. This is therefore similar to IL-1 signal transduction in TH-1 cells and macrophages where the absence of MyD88 led to a complete inhibition of NF-B activation by IL-1␤ (31)(32)(33)(34). The fact that NF-B activation is suppressed in the presence of the MyD88 dominant negative proteins also indicates a critical role for this transcription factor in the control of inos gene expression in mouse ␤ cells, as previously suggested in studies of rat islet cells (29,35).
There is considerable in vitro data suggesting that NO secreted by ␤ cells could be directly involved in their own demise (2,30,36). There is however very little evidence for a toxic effect of NO on ␤ cells in vivo. Circumstantial evidence for NO involvement in ␤ cell destruction comes from study of transgenic NOD mice overexpressing iNOS in their ␤ cells, which showed accelerated appearance of diabetes without insulitis (37). Another report however showed that islets from iNOS Ϫ/Ϫ mice transplanted in NOD-severe combined immunodeficient mice were not protected from diabetes transferred by cloned diabetogenic CD4 cells. In this particular experimental system, only TNF-receptor I null islets were able to survive after transplantation (38). These data therefore suggest that NO production by ␤ cells may not play a major role in islet destruction. Our present data support this previous conclusion. Indeed, NO generated by CDM3D cells in response to IL-1␤ and IFN-␥ seemed to have little role in induction of apoptosis, because at the dose of cytokines that induced an almost maximal production of NO, we could only detect a small rate of cell death. However, further increasing IL-1␤ concentrations (from 10 to 500 units/ml) increased markedly the rate of apoptosis without increasing significantly NO production. Therefore there is no direct relationship between NO production and induction of cell death. That the IL-1␤ signaling pathway is involved in inducing cell death is nevertheless indicated by the protection, albeit partial, conferred by expression of MyD88 dominant negative proteins. One way by which IL-1 could induce apoptosis independently of NF-B/iNOS activation could be by directly acti- Fig. 6. Effect of cytokines on glucose-stimulated insulin secretion. Cells were treated for 48 h with 10 units/ml of IL-1␤, 150 units/ml of IFN-␥, both cytokines together, or not treated. Insulin secretion was then evaluated following a 1-h incubation in the presence of 2.8 or 16.7 mM glucose. a, insulin secretion at low glucose concentration was not affected by cytokine treatment. At high glucose concentrations, GSIS was not modified by IL-1␤ alone but was reduced by IFN-␥ or IFN-␥/ IL-1␤. The -fold stimulation of GSIS by 16.7 mM glucose is indicated above the column of the non-treated cells. b, intracellular insulin content was not significantly modified by IL-1␤ alone but was markedly decreased (40 to 60%) by IFN-␥ and IFN-␥ and IL-1␤. c, secretion of insulin at 16.7 mM glucose, expressed relative to insulin content, is significantly reduced following IFN-␥ or IFN-␥/IL-1␤ treatment but not following treatment with IL-1␤ alone. Data are the mean Ϯ S.E. of three independent experiments, each performed in triplicate. *, p Ͻ 0.05 versus untreated cells.
vating the c-Jun pathway that is involved in stress-induced apoptosis and that can also be stimulated by IL-1␤ in ␤ cells (39 -42).
Glucose-stimulated insulin secretion is also altered by cytokine treatment. Here we show that exposing CDM3D cells to cytokines reduced GSIS only moderately in the presence of IL-1␤, and the reduction was more important in the presence of IFN-␥ and similar to that observed when both cytokines were present. This indicates that IFN-␥ plays a major role in reducing insulin secretion. The effect of IFN-␥ was 2-fold; there was a reduction in cellular insulin content, and when the secretion data are expressed as a percent of the insulin content, it appears that there was also a reduction of the secretory activity stimulated by glucose. IFN-␥ may thus decrease insulin gene expression at a pretranslational stage and impair the normal glucose signaling pathway to insulin granules exocytosis. Preliminary data indicated that the reduction in insulin content was indeed proportional to a decrease in cellular insulin mRNA content (not shown). The mechanism by which IFN-␥ mediates this inhibition is not yet known and will require further study.
Treatment of type I diabetes by cellular transplantation will expose the cells to an environment characterized by reduced oxygen tension and the presence of an inflammatory reaction and/or a reactivated autoimmune system. Cytokines produced in these conditions may impair function and limit survival of the transplanted cells (43). We have shown previously that transferring Bcl-2 into ␤Tc-Tet cells improved their resistance to hypoxia and other stresses (17) but does not completely protect against cytokine-induced apoptosis. Here we demonstrated that the stable lentivirus-mediated overexpression of MyD88 dominant negative proteins in CDM3D cells improved resistance to the pro-apoptotic action of cytokines. At the same time the unique characteristics of these cells to be growtharrested and to properly secrete insulin both in vitro and in vivo are preserved. Transplanting these genetically modified cells in NOD mice will allow us to evaluate whether inhibition of IL-1␤ signaling pathways may prolong their survival in the autoimmune environment of these mice. We are similarly engineering CDM3D cells to express genes blocking the TNF-␣/ Fas and/or IFN-␥ intracellular signaling pathways. Transplantation of these novel cell lines in NOD mice will allow us to gain some important insight in the role of these cytokines in ␤ cell destruction and to determine whether genetic engineering may confer sufficient protection to allow cell survival after transplantation in a diabetic environment.