Type 1 Diabetes: Cellular, Molecular & Clinical Immunology

Chapter 5 - Type 1 Diabetes Mellitus: An Inflammatory Disease Of The Islet
Regine Bergholdt*, Peter Heding*, Sif Groth Rønn*, Joachim Størling*, Morten Tonnesen*, Flemming Pociot, and Thomas Mandrup-Poulsen
Steno Diabetes Center, Gentofte, Denmark *Shared first authorship, listed in alphabetic order

Corresponding author: Thomas Mandrup-Poulsen, MD, PhD, Professor of Immunodiabetology, Chief Physician, Steno Diabetes Center, Niels Steensens Vej 2, DK-2820, Gentofte, Denmark, Fax: +45 44 43 82 32, tmpo@steno.dk

Updated 3/06, slides updated 3/02 Click to download Powerpoint slide set

Type 1 diabetes mellitus (T1D) is generally considered to be a classical autoimmune disease, in which immune tolerance is broken by environmental factors in genetically predisposed individuals, leading to recognition of specific β-cell antigens by helper T-cells that subsequently activate effector T-cells, and these effector T-cells then directly mediate β-cell killing. However, as reviewed in the previous chapters, the cellular and molecular pathogenesis of T1D is more complex.
Although there is broad agreement that autoimmunity is defined as termination of a natural unresponsive state to "self", exhaustive and unifying consensus on the definition and classification of autoimmune disease has been difficult to obtain. According to the classical papers by Witebsky (1;2) , definition of a disease as an autoimmune state depends on detectable circulating or cell-bound antibodies reactive with an autoantigen, identification of the autoantigen, presence of mononuclear cell inflammation in the target tissue and ability to transfer disease with lymphoid cells or with serum.
In T1D, there is still uncertainty regarding the nature of the autoantigen(s), passive transfer of the disease by the use of autoantibodies, e.g. by transplacental transfer from diabetic mother to non-diabetic fetus has not been demonstrated, and the transfer of disease by lymphoid cells from man to experimental animal has not been reproducibly demonstrated.
As discussed in the above chapters, and reviewed in (3) there is still uncertainty as to the importance of classical T-cell effector mechanisms, such as the Fas/FasL, perforin/granzyme and membrane-bound TNF/TNF-related apoptosis inducing ligand (TRAIL). Further, recent studies in the NOD mouse have clearly demonstrated that although β2-microglubulin-null and thereby MHC class I and CD8+ T-cell-deficient mice do not develop diabetes (4) , later studies have shown that class I restricted T-cells are needed for the initiation but not the later effector phase (5) .
Taken together, these observations indicate that β-cell destruction in T1D cannot yet be definitively classified as an autoimmune disease, but certainly as an immune-mediated disease. This is most clearly demonstrated by the ability of immunosuppressants to prevent destruction of residual β-cell function in recent onset T1D patients (6) .
A common feature shared by immune-mediated diseases is the build-up of an inflammatory infiltrate in the target organ. There is increasing evidence from other immune-mediated diseases such as inflammatory bowel disease, multiple sclerosis and rheumatoid arthritis that many components of the inflammatory response, including macrophages, CD4+ and CD8+ T-cells, inflammatory mediators in the form of cytokines and free oxygen and nitrite oxide radicals, prostaglandins, complement factors, etc. all contribute to tissue destruction. T1D does not seem to be exceptional in this regard. In the past 20 years, in vitro and animal studies have underlined the importance of inflammatory mediators (3;7,8,9) . Thus, combinations of the proinflammatory cytokines IL-1β, IFN- β , TNF-β and IL-6 are synergistically cytotoxic to β-cells, in rodent islets by inducing a mixture of β-cell necrosis and apoptosis, in human islets mainly by inducing β-cell apoptosis. These proinflammatory cytokines are present early in the inflammatory infiltrate in animals models with T1D, and antagonists of proinflammatory cytokines prevent diabetes development in such models (reviewed in (3) ). There is an extensive literature on the β-cell cytotoxic effects of free oxygen and nitric oxide radicals elaborated either by infiltrating immune cells or as a result of cytokine-induced β-cell specific expression of enzymes generating these radicals (inducible nitric oxide synthase).

Figure 1

Figure 5.1. The inflammatory model for the pathogenesis of Type 1 diabetes. For explanation see text.

Based on the above observation(s), we suggest an inflammatory model for the pathogenesis of T1D as depicted in Figure 5.1. The model suggests that neonatal β-cell mass remodeling by apoptosis causes liberation of modified β-cell antigen, or environmental factors, most likely in the form of common viruses, induce a MHC Class I restricted presentation of β-cell antigen. MHC Class I bound antigen is recognized by CD8+ T-cells that cause a limited MHC Class I restricted β-cell damage, either via cytotoxic cytokines as secreted IFN-β and/or secreted TNF-β/membrane-bound TNF-β or the perforin/granzyme system. Liberated β-cell components, such as insulin or GAD possibly in glycosylated immature forms or altered by apoptotic remodeling or cytokine induced reactive oxygen species into forms not previously "seen" by the immune systems (10) are taken up by dendritic cells in the islets and transported to regional pancreatic lymph nodes, where the antigens are processed and presented to CD4+ T-cells. After clonal expansion the CD4+ T-cells will home to the islets, tracing antigen gradients, chemokines and endothelial adhesion molecules induced by the early CD8+ T-cell mediated inflammatory response. The activated CD4+ T-cells will recruit and activate specific as well as non-specific inflammatory cells that then build up the inflammatory insulitis infiltrate. The effector phase of the β-cell destruction is mediated by cytokines via induction of proapoptotic signalling selectively in β-cells and/or by inducing β-cell expression of Fas, marking the β-cells for MHC Class II non-restricted CD4+ T-cell mediated killing via interaction between the Fas ligand on CD4+ T-cells and Fas on the β-cells.
There are many unanswered questions in this model: How and why is tolerance broken to the β-cell? What is the nature of the antigens liberated by neonatal remodeling or recognized by the CD8+ T-cell in the initiating phase? Are these antigens different or identical to those that cause activation of the CD4+ T-cell system, and how do these components become antigenic? What regulatory mechanisms normally suppress the CD4+ T-cell response and how is this suppression lost? Is the role of the CD4+ T-cell mainly to orchestrate the immunological response in the islet or does the Th1-cell have specific effector functions? What is the relative importance of cytokines and the Fas/FasL system as effector mechanisms? What are the specific cellular signalling mechanisms leading to necrosis/apoptosis of the β-cell? What is the relative role of nitric oxide dependent and nitric oxide independent intracellular toxic effectors? To which extent can this model be extrapolated to human T1D? These questions are important to clarify for our future possibilities to provide rational pharmacological intervention.
The inflammatory model of T1D has been extensively reviewed (3,8,9). The purpose of this chapter is therefore to focus on three areas of the inflammatory model depicted in Figure 5.1, where we believe particular advances have been produced in the past years:
1. studies of candidate genes of susceptibility that control individual components of the inflammatory process;
2. studies on how inflammatory mediators signal β-cell necrosis/apoptsis;
3. studies of β-cell response to mediators of this process using expression profiling by DNA chip microarray or proteomic analysis to identify deleterious and protective genes induced in the β-cell during the inflammatory process.

Genetic aspects of the inflammatory model for the pathogenesis of type 1 diabetes
The etiology of T1D is unknown, but it is recognized that both genetic and environmental determinants are important in defining disease risk. At least one locus that contributes strongly to T1D resides within the major histocompatibility complex (MHC) on chromosome 6p21, reviewed in Chapter 7. However, HLA genes (IDDM1) of the MHC region alone cannot explain the familial incidence of T1D and a number of non-HLA loci have been identified which have small yet significant effect on T1D risk. Non-MHC genes are likely to be important susceptibility genes of the initiation phase of T1D pathogenesis. The individual effect of each of these genes is expected to be small (λS ~ 1.1-1.2) and their identification by linkage analysis difficult. Novel analytical tools, combination of different approaches and larger sample sizes are crucial for finding non-MHC T1D susceptibility genes. In a recent T1D genome scan including 1435 T1D multiplex families, nine regions of the human genome were identified as conferring some degree of linkage to T1D (11) , the far most significant being the HLA region. For most of the other regions, except the INS (insulin) gene region ( IDDM2 ) on chromosome 11p15 and the CTLA4 gene region ( IDDM12 ) on chromosome 2q33, no obvious candidate gene has been identified. T1D susceptibility genes should therefore also be searched for by other means, one of them being studies of biological candidate genes, e.g. evaluating the genetic evidence supporting the inflammatory model for the pathogenesis of T1D. In the present chapter, we will focus exclusively on the contribution of candidate genes presumed to directly influence the pathogenetic process as shown in the inflammatory model (Figure 5.1). These candidate genes include cytokine genes, genes involved in T cell regulation and inflammation and genes involved in protective and deleterious mechanisms in the β-cell. It can be suspected that several of the events outlined in the inflammatory model for T1D is under genetic control. In Table 1, the various candidate gene variants are listed, as well as their chromosomal position and whether association to T1D has been demonstrated. For some of the polymorphisms a functional effect has been demonstrated, e.g. a correlation between specific alleles and the expression level of the transcript and/or protein or to promoter activity, suggesting a functional significance, see Table 5.1.

Gene Chromosome
Position In Gene Association Functional Significance/
HLA 6p21 Alleles of DQA1,
Yes Influences conformation of Ag-binding groove and thereby Ag-presentation. Reviewed in Chapter 7, (12).
IL1B 2q14 Pos.+3954 C/T Yes Allele-dosage effect
on LPS stimulation. T-allele: Increased production
    Pos.-511, C to T Yes Effect on IL-1 secretion  
IL1RN 2q14.2 Intron 2, 86bp repeat Yes/No 2-allele: Increased expression
1/1: reduced expression
IL1R1 2q12 5'UTR (PstI) Yes   (18,20)
    5'UTR (HinfI) Yes Allele dosage effect (21)
IFNG 12q14 Intron 1, CA-repeat Yes/No 2-allele: increased in vitro expression (22,23,24,25,26)
TNFA 6p21.3 Microsatellite Yes Specific alleles correspond to different expression-levels. (27,28,29)
    Pos.-308   A/G Yes/No Association, independent of HLA,   demonstrated. (30,31,32)
IL12B 5q31.1-q33.1 Pos.+1159 C/A (3'UTR) Yes/No Conflicting (33,34,35,36,37,38,39)
IL6 7p21 Pos.-174 G/C Yes/No Allele dosage effect (40,41)
IL10 1q31-q32 Microsatellite (Pos.-1.1 kb) Yes Association of same allele as in studies of Multiple Sclerosis and Rheumatoid Arthritis (41,42,43)
IL4R 16p12.1-p11.2 Pos.+375 A/C Yes   (44,45)
    Pos.+389 G/T Yes   (44,45)
    Pos.-3223 C/T Yes Opposite allele of that demonstrated associated to allergy and asthma. (46)
IL18 11q22.2-q22.3 Pos.-607 C/A Yes/No   (43,47,48,49,50)
    Pos.-137 G/C Yes/No   (43,47,48,49,50)

Genes affecting T-cells:



2q33 Exon 1, 3'UTR Yes Allele dosage effect (51,52)
CD4 12p12 -1188 (CTTTT-repeat) Yes Allele dosage effect (53,54)
FAS 10q24.1 15 mutations No   (55)
    3 mutations No   (56)
In ?-cells:          
INS 11p15.5 -596, 5'UTR VNTR Yes Different classes: different INS transcription in pancreas and thymus Reviewed in Chapter 7
NOS2 17q11.2 Several, Exon 16: Yes   (58)
NFKB 4q24 CA repeat in regulatory region Yes/No   (59,60)
SOCS1 16q13.13 8 promoter-variants No   (61)
SOCS3 17q25.3 3 Promoter-variants No   (62)
BCL2 18q21.3 Ala43Thr Yes/No Increased sensitivity to apoptosis (63,64)
SOD2 6q21 TaqI RFLP Yes   (65)

Table 5.1. Cytokine gene polymorphisms associated to T1D. For some of the polymorphisms a functional effect has been demonstrated, e.g. a correlation between specific alleles and the expression level of the transcript and/or protein or to promoter activity, suggesting a functional significance.

The HLA locus accounts for 40-50% of the genetic susceptibility to T1D (66;67) . The HLA-DQ8 and HLA-DQ2 alleles in the class II region are considered to be the major genetic risk factors and seem to affect class II mediated antigen presentation in regional pancreatic lymphnodes, as well as the role of HLA-DQ restriction in the development of central and peripheral tolerance (68;69) . The HLA region is not further discussed in this chapter, but reviewed in Chapter 7.
The cytokine genes are obvious candidate genes as they are effector molecules in the inflammatory model of the pathogenesis of T1D. It is well documented that cytokine production and action are under genetic influence and thus, it can be expected that cytokine gene polymorphisms may contribute to T1D risk.
Specific alleles of the interleukin-1β gene, IL1B , the interleukin-1 receptor antagonist gene, IL1RN and the interleukin-1 receptor type 1 gene, IL1RI , polymorphisms are significantly associated with T1D (13;16;18;19;21) , and it has been examined whether disease-associated alleles have quantitative consequences for the synthesis and secretion of the peptides. Assessment of composite haplotypes of IL1A , IL1B and IL1RI seems to give the best prediction (70,71,72,73) , an approach also used in studies of the IL-1 gene cluster in other diseases, e.g. peridontitis (74,75,76,77)  and gastric cancer (78) . For the IL1B +3954 C/T polymorphism a clear allele-dosage effect of the T1D associated T-allele has been demonstrated in LPS stimulated monocytes, with higher IL-1β secretion when this allele is present, and highest when present in duplicate (13) . Similarly, T1D patients homozygous for the disease-associated allele of the IL1RN gene had lower circulating levels of IL-1RA (16) , whereas increased IL-1RA levels have been found for the 2-allele (15;17) . In the IL1RI gene, a polymorphism in the 5´UTR showed linkage to T1D and significant differences in IL-1RI plasma-levels correlated to genotype (21) . These studies suggest genetic regulation of monokine antagonism in T1D.
Functional effects of some of the variants of the interferon-gamma gene, IFNG , and the tumor necrosis factor-alpha gene, TNFA , have been demonstrated, see Table 1. Regarding TNFA polymorphisms, although different expressions of different genotypes have been observed (27;30;79;80) , effects independent of HLA class II are difficult to demonstrate. However, data supports TNFA as important, genetically and functionally, in T1D. The promoter polymorphism at position -308 is the most extensively studied and established, in terms of functional significance and HLA-independent association, see Table 1.
The significance of another cytokine gene, the IL12B gene in T1D, is not yet clear. Association has been shown in some T1D populations (39) but not in several others (33,34,35,36,37,38)   suggesting genetic heterogeneity. Functional studies on IL12B variants have been conflicting as well (37,38,39;81) . The action of the interleukin-6 gene, IL6 , via its receptor complex, is also believed to be under genetic control, and it is likely that variants in the gene encoding IL-6 may regulate IL-6 action at different levels. Several studies have evaluated genetic variants in the human IL6 gene in T1D, Table 1, findings have, however, been quite conflicting. In T1D the C allele of the most studied promoter SNP, -174 G/C, seems convincingly associated to T1D, at least in females (40) . Smaller studies demonstrated association to T1D in general (41) , although others could not demonstrate association (82) .
Genetic variants in other pro-inflammatory cytokines as interleukin-18, IL18, and transforming growth factor β, TGFB1, genes have also been observed. IL18 polymorphisms have demonstrated association to T1D, however, not consistently reproduced (47,48,49) . TGFB1 did not show independent association to T1D (83), but have been associated with nephropathy in T1D (83) .
Also genetic variation in anti-inflammatory cytokine genes and their possible associations to T1D, is potentially interesting. Variants in the IL10 , IL4 , IL4R and IL13 have been evaluated. Some variants in the IL4 , IL4R and IL10 have shown independent association, as shown in Table 1, some however, only when evaluating haplotypes of several polymorphisms (e.g. haplotypes of SNPs in IL4 and IL13 (44) ), whereas others ( IL10 -819 C/T and -592 C/A) only show association with sub-phenotypes, i.e. adult-onset T1D in Japan (84) .
Thus, cytokine gene polymorphisms are beyond doubt of importance in T1D, by complex genetic control of the actions of cytokines in the pathogenesis of T1D. To fully understand the role of cytokines in the pathogenesis of diabetes and the genetic influence on this, it is relevant to further investigate the downstream signalling pathways. Genetic variation in transcription factors, signalling molecules etc, e.g. NFκB, SOCS-1, SOCS-3, IRS-2 genes are likely to influence the net effect of cytokines on different target tissues.
Polymorphisms in the NF-κB gene ( NFκB ) have been associated with T1D (59) , although not replicated in a large Danish family collection (60) . Furthermore polymorphisms in the SOCS1 and SOCS3 genes have been examined, but no association to T1D has been demonstrated (61;62) . Variants in IRS2 have to our knowledge not yet been evaluated in T1D.
A comprehensive analysis of genetic variation in cytokines, cytokine receptors and signalling molecules may be needed to provide the complete picture of the genetic aspects of inflammatory mediators in the pathogenesis of T1D.
The major role of CD4 and CTLA-4 lies in the activation of CD4+ T cells upon antigen stimulation and in the following clonal expansion of T lymphocytes. Polymorphisms in CD4 have shown T1D association, supported by an allele dosage effect (85;86) . The CTLA4 region on chromosome 2q33 has been linked with susceptibility to several autoimmune diseases; the encoded molecule is a co-stimulatory receptor, involved in, and conferring an inhibitory effect on T-cell activation. Several CTLA4 gene variants have been identified. These include polymorphisms in the 5' flanking and promoter region, one coding SNP, an A49G variant leading to a threonine to alanine replacement in the signal peptide, and polymorphisms in the 3´UTR. Many of these variations have been associated with autoimmune diseases, including T1D, and may be common susceptibility factors in autoimmunity in general (87) . The most comprehensive SNP and LD mapping analysis of this locus (51) identified a G6230A SNP in the 3'UTR as the predominant marker for T1D risk although the presence of causative SNP(s) in the 5' end of the gene was not ruled out. The G6230A SNP was reported to correlate with higher mRNA level of soluble CTLA-4 in unstimulated T-cells from individuals heterozygous for the T1D protective haplotype ( A49, A6230 ) compared to the predisposing haplotype ( G49, G6230 ) (51) . This observation is not easily compatible with the observation in other autoimmune diseases, where higher levels of soluble CTLA-4 were found in patients vs. controls, and the fact that blockade of the CD28/CTLA-4 pathway by anti-CTLA-4-immunoglobulin seems to be a promising treatment in autoimmune diseases (87) . Thus, further studies are needed to clarify the functional role of CTLA4 in T1D pathogenesis. Based on the functional data observed in (51) and other studies (87) no clear molecular model to explain the increased risk for autoimmunity has yet emerged and additional studies are warranted. One can, however, hypothesize that the combined effect of the diabetes associated polymorphisms in CD4 and CTLA4 might lead to a more pronounced T cell activation and clonal expansion.
FasL mediated apoptosis is important in maintaining peripheral self-tolerance. However, FAS and FASL polymorphisms and haplotypes were not associated with T1D (55;56) , but may still play a role in the pathogenesis of T1D, albeit this is not under genetic control.
The insulin gene, INS , which is expressed specifically in the β-cell and in the thymus, is an early detectable autoantigen in T1D. The class I alleles of the INS VNTR, which confers genetic risk to T1D, lead to lower insulin expression in the thymus as well as higher insulin expression in the β-cell compared to the dominant protective class III alleles. This may attenuate the development of central tolerance to insulin whilst providing high antigen expression in the β-cell (88) . Certain class III alleles, that silence thymic INS expression, also confer genetic predisposition to T1D (89) . There is evidence for interaction between the INS and HLA loci in conferring susceptibility to T1D (90) .
Free oxygen radicals may play a role in the immune mediated β-cell destruction. SOD2 and NOS2 , whose gene products are the O2- scavenger manganese superoxide dismutase (MnSOD) and inducible NO synthetase (iNOS) respectively, are important candidate genes in this regard. Polymorphisms in the MnSOD gene, SOD2 , are associated with susceptibility to T1D (65) . Cytokine induced β-cell destruction correlates with MnSOD activity (91) . Variants of MnSOD with reduced activity have been described (92) and these might enhance the sensitivity to cytokine mediated β-cell destruction and hereby the susceptibility to T1D. The reduced activity of MnSOD might be under genetic control - although this has not been shown. Furthermore, upregulation of MnSOD protects islet cells from IL-1 induced damage (93) . A Ser608Leu polymorphism has been identified in NOS2 (94) . The Leu608 -allele was preferentially transmitted to HLA DR3/DR4 positive diabetic offspring indicating an interaction between the HLA locus and NOS2 (94) . The polymorphism is located in close proximity to an autoinhibitory domain of the molecule and it is possible that it might affect NO production. Furthermore the A14 allele of a NOS2 promoter polymorphism has in two studies been shown to exert protection against severe diabetic retinopathy (95) and diabetic nephropathy (96) , respectively. Interestingly, the protective allele has in a promoter activity assay been shown to exhibit increased promoter activity (95) .
Key regulatory elements in apoptotic as well as signal transduction pathways might be under genetic control in the pathogenesis of T1D. Polymorphisms in genes encoding central signalling molecules in the apoptotic pathways might play a role both at the level of the T-cell as well as at the level of the β-cell, while gene polymorphisms in signal transduction pathways are mainly thought to exert their effects in the β-cell itself possibly affecting the magnitude and direction of the response of the β-cell to cytokines. Examples of such disease associated polymorphisms are reported in the antiapoptotic molecule Bcl2 (63,64) .
As can be seen from Table 1 and summarized above, several of the polymorphisms genetically associated with T1D either affect cytokine/monokine production, the regulation of the central as well as the peripheral immune tolerance and the character of the immune response, or β-cell defense mechanisms. The combined effect of many of these polymorphisms supports the inflammatory model for the pathogenesis of T1D in that they lead to high cytokine/monokine production, a T cell response skewed towards a Th1 response and poor β-cell defense mechanisms against damage by free radicals.
It has been speculated that a balance between protective and deleterious mechanisms in and around the β-cells exists (97) . There is evidence that at least some of the mechanisms involved in this balance are under genetic control. In a mathematical model of the onset of T1D (98) , it has been proposed that the process consists of separate elements, each with a quantitative effect and that they can each be rate-limiting in the process, suggesting that T1D can be viewed as a dynamical process, in which an unstable condition has arisen. A combination of multiple events in different cascades and signalling pathways, all pushing in the same direction leading to T1D, is therefore likely. The key regulatory elements of these events are not known in details and the genetic basis for them therefore also unknown. Many of the "key molecules" identified by transcriptome and proteome analyses, as discussed below, have yet to be analyzed in this context. The combination may differ between ethnic groups, between families or between individuals, hence specific candidate genes or regions are very difficult to identify. Even though association might be difficult to show, a functional significance of a gene in such a cascade is possible. Knowledge about the "functional genomics" of the genes, i.e. genes involved in β-cell destruction, is therefore still important.

Signalling and Molecular Effector Pathways
β-cell response to inflammatory mediators
Over the years, numerous in vitro and in vivo studies have supported that cytokines play a central role in the pathogenesis of T1D (3) . That IL-1 β may have a negative impact on β-cell function- in vivo, was demonstrated in a study where repetitive injections of IL-1 β induced transcient diabetes in rats (99) . Also, neutralization of circulating IL-1 β by use of IL-1 β antibodies and the soluble IL-1 receptor delayed diabetes onset in non-obese diabetic (NOD)-mice (100;101) . Likewise, it was recently demonstrated that IL-1R knock out NOD mice developed diabetes significantly slower than wild type mice (102). Similar, NOD mice either lacking a functional IFN-γ receptor or Bio Breeding (BB)-rats that have been subjected to an IFN-γ neutralizing antibody therapy did not develop diabetes (103,104,105) . Studies of TNF-α receptor knock out in NOD mice have demonstrated that inhibition of this cytokine signalling can prevent diabetes (106;107) . That each of these cytokines are capable of delaying and/or preventing diabetes, may indicate that they either represent individual and essential steps in a linear route towards the disease, and/or that a synergistic and cumulative effect of the cytokines is necessary for disease development, as proposed by the model.
Evidence has proposed that cytokine mediated β -cell dysfunction involves the toxicity exerted by intracellular generated reactive oxygen species (ROS) (3) , with the capacity to oxidize and thereby damage critical cellular components, such as proteins and DNA, that in turn may have a devastating impact on the affected tissue (108) . Unfortunately, β -cells appear to be particularly susceptible to oxidative stress, due to a relatively low "scavenger" enzyme expression potential (109;110) . In accordance with this hypothesis, anti-oxidant administration or expression of antioxidant enzymes reduce cytokine-induced β-cell death in both rodent and human β -cells (111,112,113) . Previously, these findings were believed to be entirely associated with a direct protection of the β -cell against the harmful effects of ROS. However, expression of the scavenger enzyme manganese superoxide dismutase (MnSOD) inhibits activation of the transcription factor NFvB, possibly by modulating the redox-environment essential for NFκB translocation and DNA- binding (114) . As discussed in detail above, NFκB plays a critical role in the pro-apoptotic signalling exerted by cytokines. Being a diverse group of molecules with variable oxidative potential, the ROS most studied in cytokine mediated β -cell toxicity is NO. This molecule is formed during an enzymatic conversion of L-arginine to L-citrulline by the enzyme inducible nitric oxide synthase (iNOS) (3) . Being the first NFκB dependent gene identified in the β -cell (115) , the iNOS gene and protein have been subjected to numerous studies indicating a critical involvement of this molecule (116) . In addition to NO generation, exposure of human islets to cytokine combinations induces production of other ROS such as the superoxide anion (O2-) (113) , which if reacting with NO may generate the highly reactive radical peroxynitrite (ONOO-) (116) .
Several mechanisms of ROS-mediated β -cell toxicity have been elucidated. Thus NO inactivates the mitochondrial Krebs cycle enzyme aconitase by nitrosylation of Fe-S groups thereby preventing glucose oxidation and ATP generation in rodent β -cells (117) . ROS mediate DNA strand breaks, leading to activation of DNA repair mechanisms including the enzyme poly(ADP ribose)polymerase (PARP), have been shown to mediate β-cell death through depletion of cellular nicotinamide adenine dinucleotide (NAD+) (118) . Finally, NO may induce β -cell death by causing endoplasmic reticulum (ER) stress caused by depletion of ER Ca2+(119) (120) and potentiate JNK activation (121) .
Interestingly, recent studies analysing islets from iNOS-/- transgenic mice have suggested that NO production is predominantly involved in promoting necrosis, as purified β-cells from these animals are rescued from cytokine-induced necrotic, but not apoptotic cell death (122) . Further, despite induction of iNOS and subsequent NO production, cytokine-mediated death of human islets can occur independently of iNOS-generated NO (123,124,125) .

Intracellular signalling elicited by cytokines
Cytokines induce β-cell destruction by activating a complex network of intracellular signalling cascades. In the following, the signal transduction mechanisms and pathways activated by cytokines will be introduced.
Traditionally, most experimental work in the field of cytokine-mediated β-cell death deals with effects caused by IL-1β, the main β-cell cytotoxic cytokine. There are three members of the IL-1 family, IL-1α, IL-1β, and IL-1 receptor antagonist (IL-1Ra) (126) . IL-1α and IL-1β bind and signal through the same IL-1 receptor and have therefore similar cellular effects. IL-1Ra also binds to the IL-1 receptor, but is not capable of inducing intracellular signalling. Thus, IL-1Ra functions as a natural receptor antagonist and constitutes an important physiological modulator of the IL-1 system. The properties of IL-1Ra as an inhibitor of IL-1 signalling have made it useful as an experimental tool to study the involvement of IL-1 in various disease models including models of diabetes. Two IL-1 receptors have been cloned and characterized, IL-1 receptor type I and II (IL-1RI/II). Both receptors are membrane spanning, but only IL-1RI has an intracellular domain allowing the initiation of intracellular signalling. Thus, while IL-1RI is the signalling receptor, IL-1RII appears to act as a competitive decoy receptor constituting yet another element for modulation of IL-1 activity (126;127) . Expression of IL-1RI has been detected in most cells including pancreatic β -cells (128;129) . IL-1β signal transduction is initiated by ligand binding to IL-1RI allowing docking of the IL-1R accessory protein (IL-1AcP) (Figure 5.2).

Figure 5.2. Basic IL-1 signal transduction. IL-1 binding to IL-1RI allows docking of the IL-1AcP. This leads to recruitment of IRAK4 via the adaptor protein MyD88. IRAK4 activates IRAK1 by phosphorylation, which then is capable of interacting with TRAF6. TRAF6, in turn, interacts with a preformed complex consisting of TAK1-TAB1-TAB2. Once activated, TAK1 is then able to activate the NFκB pathway via IKK activation and the MAPK pathways via MAPKK activation. The activation of NFκB and AP-1 transcription factors by MAPKs ultimately results in altered gene expression.

The function of IL-1RAcP is probably to increase the avidity of IL-1RI for IL-1. Following this, IL-1R-activated kinase 4 (IRAK4) is recruited to the receptor complex via the adaptor protein MyD88 (myeloid differentiation primary response gene 88). IRAK4 then interacts with and phosphorylates recruited IRAK1 leading to autophosphorylation of IRAK1 (127;130) . Phosphorylated, activated IRAK1 then interacts with TNF-receptor-associated factor-6 (TRAF6) thereby allowing TRAF6 to translocate and interact with a protein complex consisting of TGF β -activated kinase 1 (TAK1) and TAK1 binding proteins 1 and 2 (TAB1 and 2) at the plasma membrane. These protein interactions result in the activation of TAK1. Two main signalling pathways are activated by TAK1; the MAPK pathways and activation of the transcription factor NFκB. As outlined in more detail below, MAPKs lead to phosphorylation of a broad spectrum of cellular proteins including transcription factors of the activator protein (AP) -1 family such as c-jun and ATF2 (131,132,133) . The NFκB pathway is activated via TAK1-mediated phosphorylation and activation of IκB kinase (IKK) - a process that may potentially involve the NFκB inducing kinase (NIK) (134-136) . Activated IKK phosphorylates the cytosolic inhibitor of NFκB, inhibitory κB (IκB), which in the non-phosphorylated state is bound to NFκB. The phosphorylation of IκB targets it for poly-ubiquitination and degradation by the proteasome thereby freeing NFκB to translocate to the nucleus, bind to DNA promoter and enhancer sequences and regulate the transcription of target genes (134;137) . Both the MAPK and NFκB pathways are required for the apoptotic response in β-cells following exposure to cytokines (reviewed below).

The MAPK Pathways
MAPKs comprise a family of serine and threonine protein kinases. Three major groups of mammalian MAPKs have been identified; ERK, p38 and JNK (131,132,133) . Studies on insulin-secreting cells, intact rat islets and purified primary rat β-cells have demonstrated that IL-1β is a potent activator of all three MAPKs (138,139,140,141,142,143) . Several distinct MAPKs in each group have been identified, each of which are encoded by separate genes; the ERK group consists of ERK1 and ERK2; the p38 group contains four ( α, β, γ, δ) members; and the JNK subgroup contains JNK1, JNK2 and JNK3. In addition, multiple MAPK isoforms are generated by alternative splicing. For example, the JNK1/2/3 gene transcripts give rise to ten different kinase isoforms (131,132,133) . Whereas ERK is typically activated by growth and survival factors, JNK and p38 are generally referred to as stress-activated since they are activated by many cellular stresses including osmotic/heat shock, UV irradiation and pro-inflammatory cytokines. Therefore ERK, with some exceptions, is associated with proliferation and survival, whereas JNK and p38 normally convey stress and apoptosis signalling (131,132,144;145) .

The Roles of MAPKs in β-Cell Apoptosis
The functional roles of MAPKs in cytokine-mediated β-cell apoptosis have been evaluated. Using the pharmacological inhibitors PD098059 (a MEK1/2 inhibitor) and SB203580 (an inhibitor of p38), the roles of ERK and p38 in IL-1 β signalling in β-cells have been investigated. These studies revealed that blocking ERK or p38 partially (~25 and ~40%, respectively) decreased apoptosis induced by cytokines in primary rat β-cells, intact rat islets and SB203580 in human islets (142,143,146) . On the other hand, PD098059 in human islets almost fully protects against IL-1 β-induced β-cell apoptosis (147) , suggesting that ERK may be more important for apoptosis in human β-cells than in rodents. It is not clear how ERK and p38 is involved in mediating apoptosis, although ERK may increase NFκB transcriptional activity (148) and p38 may affect the expression of members of the Bcl-2 family of apoptosis-regulating proteins (146) .
The role of JNK in β-cell destruction was first elucidated in transfection experiments with the natural JNK scaffold protein JIP-1/IB1 or the JNK-binding domain (JBD) of JIP-1/IB1 and by the use of cell-permeable peptide inhibitors. These studies demonstrated that blocking the JNK pathway confers pronounced protection against apoptosis induced by IL-1 β in insulin-secreting rodent cells 138,149,150,151) . Further, taking advantage of adenovirus-mediated overexpression of a dominant negative (DN) (kinase dead) JNK1 form, a recent study reported that JNK is a critical component in oxidative stress-induced suppression of insulin gene transcription in primary islet cells (152) . The same study reported that transplantation of streptozotocin-induced diabetic nude mice with islets overexpressing DN-JNK1 preserved insulin gene expression in islet grafts and reduced hyperglycemia compared to control mice. Evidence for involvement of JNK in human islet cell death in response to cytokines has also been obtained. Thus, the small molecule JNK inhibitor SP600125 delayed cytokine-induced suppression of human islet viability (153) , and 17 β-estradiol protects against cytokine-induced human islet cell death - an effect correlating with reduced JNK activity (154) . Finally, silymarin, a polyphenolic flavonoid, inhibits IL-1 β activation of JNK and blocks the adverse effects of cytokines on viability and insulin secretion in insulin-secreting cells and human islets (155) . Taken together, ample evidence favours MAPKs, and in particular JNK, as crucial factors in mediating cytokine-induced β-cell failure and apoptosis.

Mechanisms of JNK-Mediated Apoptosis
How does JNK mediate cytokine-induced β-cell apoptosis? In general, the most explored function of JNK signalling is the regulation of transcription factors - mainly c-jun. However, despite some potential targets for JNK in apoptosis signalling have been identified, the events downstream of JNK leading to cell death are not well understood. JNK phosphorylates c-jun on Ser63 and Ser73 thereby increasing its transcriptional capability. However, JNK also phosphorylates other AP-1 transcription factors e.g. junB, junD and ATF2. The transcriptional regulation of critical target genes by AP-1 may therefore play an important role in JNK-mediated apoptosis (144) . Recent findings in β-cells have indicated that various stress stimuli including cytokines results in JNK-dependent induction of the ATF3 gene, a member of the ATF/CREB family of transcription factors (156) . Using isolated ATF3 knockout mouse islets, it was shown that impaired ATF3 induction decreases cytokine- and NO-induced cell death (156) . These findings suggest that JNK may induce transcription-dependent apoptosis in β-cells following cytokine exposure.
Another potential transcription factor that is a target of JNK in apoptotic signalling is the tumor suppressor p53. Phosphorylation of p53 by JNK prevents ubiquitin-mediated degradation of p53 by the proteasome, thus stabilizing p53 thereby increasing the half-life of the protein (157). p53 may therefore contribute to JNK-dependent apoptosis. However, as p53 is dispensable for JNK-induced apoptosis, the precise role of p53 in JNK-mediated apoptosis is unclear (158) . In insulin-secreting cells we found no evidence for p53 accumulation in response to JNK-activating stimuli, suggesting that p53 may not be a target of JNK in β-cells (159) .
An additional transcription factor regulated by JNK is c-Myc. JNK phosphorylates c-Myc, and overexpression of a mutant, non-phosphorylatable c-Myc protects against JNK-mediated apoptosis (160) . Interestingly, overexpression of c-Myc in β-cells leads to failure of β-cell function and apoptosis (161,162) , i.e. effects similar to those mediated by JNK. However, whether c-Myc is a downstream target of JNK in β-cells remains to be established.
An alternative mechanism by which JNK induces apoptosis is via the regulation of members of the Bcl-2 family of apoptosis-regulatory proteins. Thus, JNK has been shown to mediate the phosphorylation of the mitochondria-associated Bcl-2 and Bcl-X L proteins leading to the inhibition of their anti-apoptotic functions (163,164,165) . Further, the pro-apoptotic Bcl-2 protein BAD also undergoes JNK-mediated serine phosphorylation promoting the pro-apoptotic function of BAD (166;167) . Such mechanisms behind JNK-mediated apoptosis would not be dependent upon de novo protein synthesis and would comprise an attractive model for stress-induced, JNK-dependent apoptosis. The findings that overexpression of Bcl-2 in insulin-producing cells and mouse or human islets affords protection against destruction by cytokines in vitro (168,169,170,171) , indicate that regulation of the level and function of Bcl-2 family proteins might well be essential in the regulation of β-cell apoptosis. However, whether JNK plays a role for post-translational modification of Bcl-2 family proteins in β-cells awaits clarification.

The Role of NFκB in β-Cell Apoptosis
The involvement of NFκB in cytokine-induced β-cell death has been elucidated by adenoviral gene transfer of a non-phosphorylatable, and thus a non-degradable form of IκB - the so-called IκB super-repressor. Infection of primary purified rat β-cells with adenovirus containing the IκB super-repressor resulted in decreased apoptotic (and necrotic) cell death induced by a combination of IL-1 β and IFN-γ (172) . Similarly, experiments with human islets have shown that NFκB inhibition by the IκB super-repressor protects against IL-1 β-stimulated, Fas-triggered apoptosis (173) . Hence, in addition to MAPKs, NFκB also plays an important role in cytokine-induced β-cell apoptosis. The use of high-density oligonucleotide arrays has provided insight into which genes are regulated by NFκB. Of the ~150 genes whose expression is altered by a 24-hour exposure to IL-1 β plus IFN-γ, 66 genes were found to be regulated by NFκB in primary rat β-cells (174) . Thus, since NFκB regulates multiple genes in β-cells and it is reasonable to assume that NFκB-dependent β-cell apoptosis relies on changes in the expression of multiple genes.

IFN-γ and TNF- a potentiate the toxic effects of IL-1β in β-cells
In contrast to IL-1β, TNF-α and IFN-γ do not have any major cytotoxic effects alone on β-cells. However, TNF-α and especially IFN-γ are known to strongly potentiate the cytotoxic effect of IL-1 β(175;176) . Furthermore, while IL-1 β is cytotoxic to rodent islets, human islets seem more resistant and require combinations of cytokines (8) . Signalling elicited by IFN-γ is much more linear than that of IL-1. Thus, following binding of IFN-γ to surface receptors, transcription factors of the STAT (signal transducer and activator of transcription) family are activated via so-called Janus kinases at the intracellular part of the receptor complex (177). A dose-dependent effect of IL-1 β on MAPK (ERK, JNK and p38) activation in isolated rat islets has been found (178) , whereas TNF-α alone only induced very modest MAPK activation and INF-γ alone surprisingly gave rise to a reduction of MAPK activity. However, in combination, TNF-α and IFN-γ potentiated IL-1 β-induced MAPK activation (178). This suggests that the potentiating effect of TNF-α and IFN-γ on IL-1 β β-cell cytotoxicity may be assigned to a synergistic activation of MAPKs. The molecular mechanisms responsible for this cross-talk are not fully understood.

Calcium and β-Cell Apoptosis
In all mammalian cells, Ca2+ functions as a second messenger to convey extracellular signals (hormones, growth factors etc.) into a cellular response. In the β-cell, Ca2+ first and foremost plays a crucial role for the physiological function of the β-cell, i.e. the stimulus-secretion coupling. However, a number of studies have pointed to the fact that Ca2+ may also be a key player in the regulation of β-cell apoptosis. When insulin-secreting cells are incubated with serum from T1D patients or exposed to Apolipoprotein CIII (present in T1D sera), they undergo apoptosis in a manner dependent on Ca2+ influx via L-type channels (179;180) . Further, mouse islet β-cell apoptosis induced by high glucose or a K+ channel inhibitor can be prevented by blocking L-type Ca2+ channels (181) . IL-1 β-induced apoptosis also seems to involve Ca2+ as suggested by the finding that blockade of L-type channels abrogates IL-1 β-induced mouse islet β-cell apoptosis (182) . Further, apoptosis caused by a combination of cytokines (IL-1 β+ IFN-γ+ TNF-α) can be prevented by T-type channel blockade in β TC3 insulin-secreting cells (183) .   Also, IFN-γ+ TNF-α-mediated suppression of mouse islet and MIN6 cell viability was reversed by L-type channel blockade (184) . In further support of a role of Ca2+ in cytokine-induced apoptosis is the finding that calbindin-D 28k , a cytosolic Ca2+ -binding protein, protects insulin-secreting cells from cell death induced by a mixture of IL-1 β , IFN-γ and TNF-α (185) . Thus, a substantial amount of evidence points toward an important role of Ca2+ in the regulation of β-cell apoptosis. What is unclear, however, is how Ca2+ promotes β-cell apoptosis. Recent studies have pointed towards Ca2+ as an important mediator of cytokine-induced MAPK activation (147;186) , suggesting that one mechanism behind Ca2+ -induced apoptosis is via activation of pro-apoptotic MAPK signalling.
Cytokines have been shown to cause ROS-dependent down-regulation of sarco/endoplasmic reticulum Ca2+ ATPase (SERCA) at the mRNA level in β-cells (174;187;188)   potentially resulting in depletion of endoplasmic reticulum (ER) Ca2+ content and induction of so-called ER stress. The ER serves as the main intracellular Ca2+ store for releasable Ca2+ . However, the ER also serves several other important functions including post-translational protein modifications and folding, and assembly of newly-synthesized secretory proteins. Further, the ER is the site of synthesis of lipids and sterols. The ER is exquisitely sensitive to alterations in homeostasis and any perturbation of ER function causes ER stress which can lead to cell death by apoptosis. Thus, as in many other cells, it has been shown that inhibition of SERCA by the drug thapsigargin induces apoptosis in insulin-producing cells and islets (189) . In line with the observation that cytokines suppress SERCA2b expression potentially leading to ER stress, it has recently been shown that ER is depleted for Ca2+ in β-cells exposed to cytokines or ROS (120;190;191) . Together, these observations highlight that ER stress is likely to be involved in mediating β-cell dysfunction and destruction following exposure to cytokines.
Several signalling pathways are activated by ER stress (192;193) . The transcription factor CHOP is strongly induced at the level of transcription following ER stress. Transcription of the CHOP gene is induced via the ER transmembrane proteins IRE1α, PERK and ATF6. ATF6 activation by ER stress is achieved via proteolysis and release of ATF6, while activation of IRE1α and PERK result in activation of the transcription factors XBP-1 and ATF4, respectively. Mouse islets from CHOP deficient animals were shown to be much more resistant to NO-induced, ER stress-mediated β-cell apoptosis as compared to control mouse islets (190) . IRE1α is also involved in JNK activation by ER stress (194) . IRE1α leads to JNK activation via the adaptor protein TRAF2 and the kinase ASK1 (192) . Finally, a member of the caspase family (caspase-12) is specifically activated and released from the ER following ER stress, and is involved in ER stress-induced cell death (195) .

Negative regulation of cytokine signalling by SOCS
As can be seen from the above, cytokines are able to activate numerous signalling pathways in the cell, leading to changes in gene-expression and overall cell-homeostasis. Often, these changes are absolutely crucial to maintain a normal body function. However, regulation of the cytokine response is just as essential in order to avoid the disastrous effects seen in conditions with excessive cytokine signalling - for example in the pathogenic process of diabetes. There are three principal known groups of cytokine-signalling inhibitors (196) . One group is the tyrosine phosphatases SHP-1 (SH2-domain containing phosphatase-1) and SHP-2 which are involved in dephosphorylation of signalling components such as the JAK-proteins (197;198) . Another group known as PIAS (protein inhibitors of activated STATs) inhibits activated STAT-1, STAT-3 and STAT-4. The exact mechanism of action remains to be elucidated, but PIAS may inhibit STAT-DNA binding or they may act as transcriptional co-repressors of STATs (199,200,201) . The last group is the suppressors of cytokine signalling - or the SOCS-proteins. The SOCS-proteins constitute a family of proteins that acts in a classical feedback loop to suppress signalling initiated by multiple cytokines and interestingly they have proven to be very attractive targets with respect to diabetes.
Studies performed both in β-cell lines and in primary β-cells have shown that one member of the SOCS-family, i.e. SOCS-3 protects β-cells from the toxic effects of IL-1 β and IFN-γ (202;203) . SOCS-3 inhibits IL-1β and IFN-γ induced β-cell apoptosis through an inhibition of IL-1 β and IFN-γ activated signalling molecules including STAT-1, NFκB and the MAP kinases JNK, p38 and ERK. Moreover, diabetes is delayed in transgenic NOD-mice having constitutive expression of SOCS-1 in their β-cells (204) , further demonstrating the potential of SOCS-proteins as β-cell protectors. Endogenous expression of the SOCS-proteins is induced by cytokines, but apparently this expression is delayed or at least not sufficient in the β-cells, making these vulnerable targets of destructive cytokines such as IL-1β and IFN-γ. Utilization of SOCS-proteins in the clinic could be based on stem cell therapy or on transplantation strategies. For example it has recently been shown that allograft rejection is delayed in SOCS-1 expressing islets (205) .

Expression Profiling Studies of Cytokine-Induced β-cell Destruction
Cytokine-induced β-cell destruction is primarily mediated by apoptosis in human islets (3;206) and requires de novo protein synthesis (207,208,209,210). Although not completely clarified, the proximal mechanisms in cytokine signal transduction have been extensively studied. However, the exact distal intracellular molecular events responsible for β-cell death are still poorly understood. Thus, identification of cytokine-mediated changes in gene and protein expression profiles is likely to provide useful information for future prevention and/or treatment strategies of T1D. Over the last decade development and improvement of transcriptomic- and proteomic technologies have made it possible to study expression profiles of both mRNA transcripts and proteins. Transcriptomic techniques as microarrays or genechip arrays displaying several thousands of mRNA and Expressed Sequence Tags (EST) are techniques to measure the dynamics of a genetic network at the mRNA level. Proteomic experiments, typically involving protein or peptide separation steps coupled to the identification of thousands of peptides by mass spectrometry, are useful means of large-scale or global analysis of the protein complement of cells or tissues (211) . Individually, transcriptome and proteome analyses have become powerful tools to decipher the complex genetic networks altered in response to environmental insults or disease, and a fast growing number of studies using these technologies in the study of cytokine-mediated β-cell destruction in T1D have been published. Data from such studies provide novel insights into molecular patterns of the dying β-cell.
Since T1D is caused by a selective destruction of the insulin-producing β-cells, leaving neighbouring endocrine cells intact (212), β-cells seem to be more sensitive to the toxic effects of cytokines. This increased sensitivity is believed to be an acquired trait associated with maturation from stem cell to active insulin-producing β-cell ( 138) . Microarray and proteomics have been used to investigate changes in gene and protein expression profile during β-cell differentiation (213,214,215) . As expected, these studies reveal a detailed but complex picture of gene- and protein expression, suggesting that the cumulative pattern of changes favours a transition from dynamic stability to dynamic instability, and thereby increased cytokine sensitivity of the insulin-secreting β-cell.  
When β-cells and islets are exposed to cytokines multiple changes in the expression profiles of mRNA and proteins are seen. These changes are illustrated by up- and down-regulation as well as de novo synthesis of diverse groups of genes. Several of these genes are putative targets for the transcription factor NFκB (174) . Regulation of these genes contributes to the loss of differentiated β-cell functions and triggers both pro- and anti-apoptotic mechanisms in the β-cell. However, interpretation of these data is complicated because it is difficult to discriminate between early "primary" effects and later "secondary" effects of cytokine exposure. Microarray studies have in this way previously demonstrated that almost half of the late (8-24 hours after cytokine exposure) gene expression changes are mediated by NFκB-dependent NO production (188)), indicating the important role of this radical for the late effect of cytokines. Furthermore, NFκB also regulates expression of other transcription factors like c-Myc, pancreatic duodenum homeobox 1 (Pdx-1) and islet factor 1 (Isl-1) (174) . Thus, NFκB seems to be a "master switch" of cytokine-induced β-cell dysfunction and death through primary and secondary effector mechanisms (Figure 5.3).
Genes and proteins altered by cytokine exposure can be clustered according to biological function and/or based on their temporal profile of expression, allowing an integrated understanding of biological processes that may explain some of the cytokine-induced effects on the β-cell phenotype. Supporting previous findings of cytokine-mediated inhibition of mitochondrial energy generation (216;217) , cytokines decrease expression of several genes related to the mitochondrial respiratory chain (187;218;219) , resulting in reduced energy generation in response to cytokine exposure. Furthermore, cytokines decrease expression of several genes related to differentiated β-cell functions and preservation of β-cell mass, including β-cell specific transcription factors (Pdx-1 and Isl-1), key-components of insulin synthesis and secretion (GLUT2, prohormone convertase-1/2, glucokinase and insulin), and diverse receptors for integrins and growth factors (174;187;188;218) . These changes result in decreased insulin production (117;220) and reduced growth capacity of the cytokine-exposed β-cells or islets (221) .
Cytokines (in particular IL-1β and IFN-γ) induce production of free radicals like superoxide (O2-), hydrogen peroxide (H2O2) and NO (8;222;223) , which may have toxic effects on the β-cell. As a response, several "protective" proteins, such as heat shock proteins (HSPs), glutathione-S-transferase (GST), manganese superoxide (MnSOD) and metallothionin are upregulated after cytokine-exposure of β-cells (187;224) . These proteins are involved in the scavenging of free radicals, but this increase in "defence/repair" mechanisms are paralleled by decrease in expression of other "defence-genes" and seem to be insufficient to prevent cytokine-mediated β-cell destruction.
Furthermore, several genes and proteins involved in apoptosis have been identified in purified β-cells and islets exposed to cytokines by proteome- and microarray analyses. These include lamins A and B and transforming growth factor β (TGF β) receptor interacting protein (218) , caspase-1 (225) , death protein-5 and Fas (187;226)   which are all up-regulated following IL-1 β and/or IFN-γ exposure. These findings support that the apoptotic machinery is initiated in β-cells and islets after cytokine exposure (206;227) .

Figure 5.3. Important pathways changed in expression profiling studies after cytokine-exposure.

Chemokines are expressed in the islets of Langerhans during insulitis. Chemokines are small molecules involved in the migration and activation of leukocytes. The specificity of the chemokine system is derived from both the release of specific chemokines in inflammatory reactions and the regulated expression of their receptors (reviewed in (228,229,230)). It has been demonstrated that chemokines are expressed in pancreatic islets during early insulitis (231;232). These findings are supported by microarray studies of FACS-purified Chemokines are expressed in the islets of Langerhans during insulitis. Chemokines are small molecules involved in the migration and activation of leukocytes. The specificity of the chemokine system is derived from both the release of specific chemokines in inflammatory reactions and the regulated expression of their receptors (reviewed in (228,229,230)). It has been demonstrated that chemokines are expressed in pancreatic islets during early insulitis (231;232). These findings are supported by microarray studies of FACS-purified β-cells and β-cell lines where the cytokine-induced expression of several chemokines, cytokines and cell adhesion molecules suggest the β-cell as being active participant in leukocyte homing insulitis (187;188).   
Data from transcriptomic- and proteomic studies are complex but provide a global picture of the intracellular processes compared to a gene-by-gene approach. However, results obtained must be regarded as snapshots in narrow time windows of a highly dynamic process that is constantly influenced by the internal and external environment. Since not all mRNAs are translated into protein, and posttranslational protein-changes are important for the function, correlation between transcriptomic- and proteomic studies is often unsatisfactory (215;233;234). Further discrepancies might be explained by e.g. differences in detection sensitivity, sample preparation, experimental conditions, differences in mRNA- and protein turnover and alternative splicing of mRNAs.
Development of methodology and instrumentation in the field of transcriptomic- and proteomic techniques is progressing rapidly, and requires advanced software. Future studies using data mining- and cluster analysis software may identify relevant groups or clusters of genes suited for pharmacological intervention and/or transplantation strategies in T1D.

Conclusions and Perspectives
There have been remarkable advances in our understanding of the molecular pathogenesis of T1D over the last decade. The Human Genome Project and new screening technologies are generating an amount of data unprecedented in biology. Science now has the tools to characterize the pathophysiological nuances and inherited variations that interact over time and lead to common diseases as T1D.
The multiple approaches to understand cytokine-mediated β-cell cytotoxicity presented here may be useful for identification of relevant mechanisms for β-cell death after cytokine exposure. Data obtained by the outlined methodologies may unravel the hitherto most detailed and complex picture of the molecular processes leading to β-cell destruction in vitro . Although the picture is complicated and far from complete we are observing the ailing β-cell through a new window and the challenge is now to learn to fully understand what we see.
Genetic, signal transduction and expression studies have revealed that a race between protective and deleterious mechanisms takes place in β-cells when exposed to cytokines, and that the balance between these mechanisms is crucial. The overall picture in the cytokine-exposed β-cell is: Decrease in energy generation, insulin production and β-cell function, increase in NO production, activation of the MAPK and NF-kB cytokine-signalling pathways, increases in pro-apoptotic proteins and genes, and decreases in cellular defense proteins.
The proposed model (Figure 5.1) may prove valuable in identifying therapeutical intervention strategies in T1D and serve as a tool for testing such strategies. By using transgenic or knock-out mice of identified susceptibility genes, by detailed expression profiling e.g. of primary and secondary effects, and by modifying signal transduction new targets for intervention in the diseases process are likely to be identified.
Manipulation of the costimulatory signals, e.g. by CTLA-4 activating immunoglobulins, has been envisioned as a potential strategy for achieving therapeutically useful immunosuppression or tolerance. CTLA4-Ig has been used successfully in other immune-mediated diseases as psoriasis (85;86), treatment of graft-versus-host disease in allogenic bone marrow transplantation (235) , in mouse models of systemic lupus erythematosus (SLE) (236) , and the effect of CTLA4-Ig in experimental models of RA is also promising (237) . Clinical randomized trials on the use of CTLA4-Ig in T1D are however missing.
The demonstrated effectiveness of superoxide dismutase and other oxygen free radical scavengers in preventing the development of disease recurrence in transplanted pancreatic tissue (238,239,240)   and from in vitro studies (reviewed in (3) ) suggests that agents directed at blocking inflammatory mediators, such as interleukins and oxygen radicals, may be effective in preventing damage to the susceptible pancreatic β-cells. Interestingly, IFN-β for multiple sclerosis and TNF-α antagonists for rheumatoid arthritis and Crohn's disease are amongst the first new treatments for autoimmunity approved by the Food and Drug Administration in 20 years.
Further genetic, signal transduction and expression profiling analyses within T1D research may eventually complete the picture that the first analyses have just begun to make visible. The perspective of this is the development of new and specific intervention modalities in β-cell destruction in T1D. Making the β-cell more resistant to mediators of the immune system may prolong the survival of transplanted islets or engineered β-cells, and potentially prevent the ongoing β-cell destruction in predisposed individuals.

Reference List - links to PubMed available in Reference List.

Chapter 5 Powerpoint slide set - Updated 3/02

For comments, corrections or to contribute teaching slides, please contact Dr. Eisenbarth at: george.eisenbarth@ucdenver.edu

Back to Type 1 Diabetes Table of Contents