ROLE OF NITRIC OXIDE AND VIRAL PRODUCTS IN PANCREATIC β-CELL DYSFUNCTION AND DEATH

UNIVERSITE LIBRE DE BRUXELLES Faculté de Médecine

Laboratoire de Médecine Expérimentale Promoteur: Prof. D. L. Eizirik

ROLE OF NITRIC OXIDE AND VIRAL PRODUCTS IN PANCREATIC β-CELL

DYSFU N CTION AND DEATH

Dongbo LIU

This thesis is presented for obtaining the degree of ¨PhD in Biomedical Science¨

Academic Year 2003-2004

CONTENTS

List of reports constituting this thesis 3

List of abbreviations 4

Summary 6

Résumé 9

1. Introduction 13

1.1 The autoimmune process and type 1 diabetes mellitus 13 1.2 Insulitis and β-cell damage in T1DM 16

1.2.1 Insulitis 16

1.2.2 β-cell death: apoptosis or necrosis? 17

1.3 Initiators, contributors and effectors of β-cell death: cytokines, chemokines,

viral infections and nitric oxide 25

1.3.1 Cytokines 26

1.3.2 Chemokines 30

1.3.3 Fas and FasL 32

1.3.4 Viral infections as triggers of insulitis and β-cell death 34 1.3.5 Viral infections and double-stranded RNA (dsRNA) 37 1.3.6 Nitric oxide and β−cell damage 37 1.3.6.1 Structure and regulation of the iNOS promoter 39

1.3.6.2 Intracellular targets for NO 40

1.3.7 Genes and transcription factor networks regulating β-cell death 40

2. Aims of the Study 42

3. Results 44

4. Conclusions and Perspectives 48

5. References 63

Acknowledgements 92

List of reports constituting this thesis

I. Liu D, Pavlovic D, Chen MC, Flodström M, Sandler S, Eizirik DL 2000 Cytokines induce apoptosis in β-cells isolated from mice lacking the inducible isoform of nitric oxide synthase (iNOS

). Diabetes 49: 1116-1122

II. Liu D, Darville M, Eizirik DL 2001 Double-stranded ribonucleic acid (RNA) induces β-cell Fas messenger RNA expression and increases cytokine-induced β-cell apoptosis. Endocrinology 142: 2593-2599

III. Liu D, Cardozo AK, Darville M, Eizirik DL 2002 Double-stranded RNA cooperates with interferon-γ and IL-1β to induce both chemokine expression and nuclear factor-κB- dependent apoptosis in pancreatic β-cells: potential mechanisms for viral-induced insulitis and β-cell death in type I diabetes mellitus. Endocrinology 143: 1225-1234

IV. Rasschaert J, Liu D, Kutlu B, Cardozo AK, Kruhøffer M, Ørntoft TF, Eizirik DL

2003 Global profiling of double stranded RNA- and IFN-γ-induced genes in rat pancreatic β-

cells. Diabetologia 46: 1641-1657

LIST OF ABBREVIATIONS

ADAR RNA-specific adenosine deaminase

Apaf-1 apoptotic protease activating factor-1

APC antigen-presenting cell

AS Argininosuccinate synthetase

BB Bio-breeding

BM bone marrow

CAD caspase-activated deoxyribonuclease

CARD caspase recruitment domain

caspase cysteine aspartase

CEB/P CCAAT/enhancer-binding protein

CHOP/GADD153 CEB/P homologous protein

CL chemokine ligand

CTL cytotoxic T-cell

CSF colony-stimulating factor

Ctyc cytochrome c

DC dendritic cells

DED death effector domain

DRAF1 dsRNA-activated transcription factor1

dsRNA double-stranded RNA

eIF-2 eukaryotic initiation factor-2α

EMC Encephalomyocarditis

ER endoplasmic reticulum

ERK extracellular signal-regulated kinases

EST expressed sequence tag

FACS fluorescence activated cell sorter

FAD flavine adenine dinucleotide

FADD/MORT1 Fas-associating death domain protein/mediator

of receptor-induced toxicity

FAK focal adhesion kinase

FasL Fas ligand

FMN flavine mononucleotide

GAD

65 kDa isoform of glutamic acid decarboxylase

GAS gamma-activated sequence

Gld generalized lymphoproliferative disease,

complementary mutation of FasL

IA2 tyrosine phosphatase-like protein antibodies

IAA insulin autoantibodies

ICA islet cell cytoplasmic antibodies

IFN interferon

IKK inhibitor of κB (I-κB) kinase

IL interleukin

iNOS inducible nitric oxide synthase

IP-10 (CXCL10) interferon-γ - inducible protein 10

IRF interferon regulatory factor

ISGF-3 interferon-stimulated gene factor 3

ISRE IFN-stimulated response element

JAK Janus family of tyrosine kinases

JIK c-Jun N-terminal inhibitory kinase

JNK c-Jun NH

-terminal kinase

JNKK JNK kinase

KO knock out

LCMV lymphocytic choriomeningitis virus

Lpr lymphoproliferation, complementary mutation of Fas

MAP mitogen-activated protein

MAPK Mitogen-activated protein kinase

MCP-1 (CCL2) macrophage chemoattractant protein – 1

MEKK MAP/Erk kinase kinase

MIP macrophage inflammatory protein

NAD

nicotinamide adenine dinuleotide

NF-κB nuclear factor-κB

MHC major histocompatibility complex

NIK NF-κB-inducing kinase

NK natural killer

NO nitric oxide

NOD non-obese diabetic

OAS oligoadenylate synthetase

OMM outer mitochondrial membrane

PARP poly(ADP)ribose polymerase

Pdx-1 pancreatic duodenal homeobox

PIC polyinosinic-polycytidylic acid

PKA2 p21-activated kinase 2

PKR IFN-inducible RNA-dependent protein kinase

PT permeability transition

RANTES (CCL5) regulated upon activation normal T cell

expressed and secreted

RIP receptor-interacting protein

RNOS reactive NO species

ROS reactive oxygen species

RT-PCR reverse transcriptase-polymerase chain reaction SERCA sarcoplasmic-endoplasmic-reticulum Ca

ATPase

STAT signal transducer and activator of transcription

T1DM Type 1 diabetes mellitus

TCR T-cell receptors

TE thymic epithelium

Th T helper

TLR3 toll-like receptor 3

TNF tumor necrosis factor

TRADD TNF-R1-associated death domain protein

TRAF2 TNF-R-associated factor-2

SUMMARY

Type 1 diabetes mellitus (T1DM) is an autoimmune disease caused by progressive destruction of insulin-producing pancreatic

-cells. Both viral infections and the cytokines interleukin-1

(IL-1

) and interferon-

(IFN-

) have been suggested as potential mediators of

-cell death in early T1DM. Nitric oxide (NO) is a highly diffusible, short-lived free radical gas, which plays a significant role in several physiological processes in a diversity of tissues and organisms. Prolonged exposure of rodent or human pancreatic β-cells to combinations of cytokines induces the expression of the inducible form of nitric oxide synthase (iNOS) and Fas, NO production, and cell death. It also induces the expression of potential "defense"

genes, such as manganese superoxide dismutase (MnSOD) and heat shock protein (hsp) 70.

Recent studies have shown that NO, in addition to having cytotoxic actions, may also regulate gene transcription. It remains unclear whether NO mediates cytokine-induced gene expression and subsequent

-cell death. Previous studies using NO synthase blockers yielded conflicting results, which may be due to non-specific effects of these agents.

In the first part of our work, we examined the role of NO in β-cell dysfunction and

death by using an iNOS knockout mice (iNOS

, background C57BL/6x129SvEv). We

evaluated the effects of cytokines on gene expression, as determined by reverse transcriptase-

polymerase chain reaction (RT-PCR), and viability, as determined by nuclear dyes, of

pancreatic islet cells or fluorescence-activated cell sorter (FACS)-purified

-cells isolated

from iNOS knockout mice or their respective controls (C57BL/6x129SvEv). The combination

of cytokines used was interleukin-1

(50 U/ml) +

-interferon (1000 U/ml) + tumor necrosis

factor-

(1000 U/ml). The lack of cytokine-induced iNOS activity in the iNOS

islet cells

was confirmed by RT-PCR and nitrite determination. Cytokines induced a > 3-fold increase in

Fas and MnSOD mRNA expression in wild-type (wt) and iNOS

islets. On the other hand,

hsp 70 was induced in wt but not in iNOS

islets. Prolonged (6-9 days) exposure of wt islets to cytokines lead to an 80-90% decrease in islet cell viability, whereas viability decreased by only 10-30% in iNOS

islet cells. To determine the mode of cytokine-induced cell death, FACS-purified

-cells were exposed to the same cytokines. After 9 days, the apoptosis index was similarly increased in wt (39

3%) and iNOS

(33

4 %)

-cells. On the other hand, cytokines increased necrosis in wt (20

4 %) but not in iNOS

(7

3 %)

-cells. From these data, we conclude that: 1) NO is required for cytokine-induced hsp 70 mRNA expression, but not for Fas and MnSOD expression; 2) cytokines induce both apoptosis and necrosis in mouse

β

-cells; 3) cytokine-induced apoptosis is mostly NO-independent, whereas necrosis requires NO formation.

In the second part of our work, we examined the role of the viral product double-

stranded RNA (dsRNA) in β-cell dysfunction and death. DsRNA is produced by many viruses

during their replicative cycle. We investigated whether dsRNA (here utilized as synthetic poly

IC (PIC)) modifies the effects of IL-1

and IFN-

on gene expression and viability of rat

pancreatic

-cells and the role of NO in this process. FACS-purified rat

-cells were exposed

for 6-16 h (study of gene expression by RT-PCR) or 6-9 days (study of viability by nuclear

dyes) to PIC and/or IL-1

or IFN-

. PIC increased the expression of Fas and Mn superoxide

dismutase mRNAs by 5-10-fold. IL-1

and a combination of PIC + IFN-

but not PIC or IFN-

alone) induced expression of iNOS and consequent NO production. Induction of iNOS

expression by PIC + IFN-

requires NF-

B activation, as suggested by transfection

experiments with iNOS promoter-luciferase reporter constructs into primary

-cells. The PIC-

responsive region in the Fas promoter is located between nucleotides -223 and –54. Site-

directed mutations at the NF-

B and C/EBP binding sites prevented PIC-induced Fas

promoter activity. Increased Fas promoter activity was paralleled by enhanced susceptibility

of PIC + cytokine-treated

-cells to apoptosis induced by FasL. Combinations of IL-1

+

IFN-

, PIC + IFN-

or PIC + IL-1

induced a 2-3-fold increase in the number of apoptotic

- cells. Blocking of iNOS activity decreased PIC + IL-1

-, but not PIC + IFN-

-, induced apoptosis. β-cell infection with an adenovirus encoding the NF-

B inhibitor AdI

B

prevented both necrosis and apoptosis induced by PIC + IL-1

or PIC + IFN-

. mRNAs for several chemokines and one cytokine were induced by PIC, alone or in combination with IFN-

, in pancreatic

-cells. These included IP-10 (CXCL10), IL-15, MCP-1 (CCL2), fractalkine (CX3CL1) and MIP-3

(CCL20). There was not, however, induction of IL-1

expression. PIC has an additive effect on cytokine-induced

-cell death, by both NO- dependent (in the case of IL-1

) and NO-independent (in the case of IFN-

) mechanisms.

To further elucidate the molecular mechanisms involved in PIC + IFN-γ-effects, the global profile of genes modified by these agents was analysed by high-density oligonucleotide arrays representing 8000 probes in primary rat β-cells. Following a 6 or 24h treatment with IFN-γ, PIC or IFN-γ and PIC, we observed changes in the expression of 51 to 189 genes. IFN- γ modified the expression of MHC-related genes, and also of genes involved in β-cell metabolism, protein processing, cytokines and signal transduction. PIC affected preferentially the expression of genes related to cell adhesion, cytokines and dsRNA signal transduction, transcription factors and MHC. PIC and/or IFN-γ up-regulated the expression of several chemokines and cytokines that could contribute to mononuclear cell homing and activation during viral infection, while IFN-γ induced a positive feedback on its own signal transduction.

PIC + IFN-γ inhibited insulin and GLUT-2 expression without modifying pdx-1 mRNA

expression. Based on these findings, we propose an integrated model for the molecular

mechanisms involved in dsRNA + IFN-γ induced β-cell dysfunction and death.

RESUME

Le diabète mellitus de type 1 (DMT1) est une maladie auto-immune provoquée par la destruction progressive des cellules β pancréatiques productrices d'insuline. Des infections virales, ainsi que deux cytokines, l’interleukine-1β (IL-1β) et l’interféron-γ (IFN-γ), ont été proposées comme médiateurs potentiels de la mort des cellules β dans le DMT1 précoce. Le monoxyde d’azote (NO) est un gaz à radical libre, hautement diffusible, à demi-vie courte, qui joue un rôle significatif dans plusieurs processus physiologiques dans une diversité de tissus et d'organismes. L'exposition prolongée des cellules β pancréatiques humaines ou de rongeur à des combinaisons de cytokines induit l'expression de la forme inductible de la NO synthase (iNOS) et de Fas, la production de NO, et la mort cellulaire. L'expression de gènes potentiellement impliqués dans les mécanismes de défense cellulaire, tels que la superoxyde dismutase – dépendante du manganèse (MnSOD) et la protéine ¨heat shock¨ (hsp) 70, est également induite. Des études récentes ont montré que le NO, outre le fait d'exercer des actions cytotoxiques, pourrait aussi réguler la transcription de gènes. Il n’est cependant pas encore clairement établi si le NO est un médiateur de l’induction de l'expression de gènes et de la mort de la cellule β par les cytokines. Des études antérieures utilisant des bloqueurs de la NO synthase ont montré des résultats contradictoires, qui pourraient être dus à des effets non spécifiques de ces agents.

Dans la première partie de notre travail, nous avons examiné le rôle du NO dans la

dysfonction et la mort des cellules β en utilisant une souris knockout iNOS (iNOS

,

background C57BL/6x129SvEv). Nous avons évalué les effets des cytokines sur l'expression

de gènes, par la technique de "reverse transcriptase-polymerase chain reaction" (RT-PCR), et

sur la viabilité de cellules d'îlots pancréatiques ou de cellules β purifiées par la technique de

tri cellulaire basé sur la fluorescence (FACS), isolées au départ de souris knockout iNOS ou

de leurs contrôles respectifs (C57BL/6x129SvEv), par l'utilisation de colorants nucléaires. La combinaison des cytokines utilisée était la suivante : interleukine-1β (50 U/ml) + interféron-γ (1000 U/ml) + facteur de nécrose tumorale α (1000 U/ml). L’absence d’induction de l’activité de l’iNOS par les cytokines dans les îlots iNOS

a été confirmée par RT-PCR et mesure de la formation de nitrite. Les cytokines induisaient une augmentation de trois fois de l'expression de l'ARNm de Fas et de la MnSOD dans les îlots de type sauvage (wt) et iNOS

. D'autre part, hsp 70 était induite dans les îlots wt mais pas dans les îlots iNOS

. L'exposition prolongée (6-9 jours) d'îlots wt aux cytokines a conduit à une diminution de 80-90% de la viabilité des cellules insulaires, alors que la viabilité diminuait seulement de 10-30% dans les cellules insulaires iNOS

. Afin de déterminer le mode d'action des cytokines dans l'induction de la mort cellulaire, des cellules β purifiées par FACS ont été exposées à ces mêmes cytokines. Après 9 jours, l'index apoptotique augmentait de manière similaire dans les cellules β wt (39 ± 3%) et iNOS

(33 ± 4%). D’autre part, l’index nécrotique augmentait dans les cellules β wt (20% ± 4%) mais pas dans les cellules β iNOS

(7% ± 3%). De ces données, nous pouvons conclure que : 1) le NO est requis pour l'expression de l'ARNm de hsp 70 induite par des cytokines; 2) les cytokines induisent à la fois l'apoptose et la nécrose dans les cellules β de souris; 3) l'apoptose induite par les cytokines est principalement NO- indépendante, tandis que la nécrose requiert la formation de NO.

Dans la seconde partie de notre travail, nous avons examiné le rôle de l'ARN viral

double brin (dsRNA) dans la dysfonction et la mort des cellules β. Le dsRNA est produit par

de nombreux virus durant leur cycle de réplication. Nous avons recherché si le dsRNA (ici

utilisé sous la forme de poly IC synthétique (PIC)) modifie les effets de l’IL-1β et l’IFN-γ sur

l'expression de gènes et sur la viabilité des cellules β pancréatiques de rat; nous avons

également investigué le rôle du NO dans ce processus. Dans ce but, des cellules β de rat

purifiées par FACS ont été exposées pendant 6-16h (étude de l'expression de gènes par RT-

PCR) ou 6-9 jours (étude de la viabilité par colorants nucléaires) au PIC et/ou à l’IL1-β et l’IFN-γ. Le PIC augmentait de 5-10 fois l'expression des ARNm de Fas et de la MnSOD.

L’IL-1β et l'association de PIC et de l' IFN-γ (mais pas le PIC ou l’IFN-γ seul) induisait l'expression de l’iNOS et la production de NO conséquente. L'induction de l'expression de l’iNOS par le PIC + IFN-γ requiert l'activation de NF-κB, comme suggéré par des expériences de transfection avec des constructions du promoteur de l’iNOS couplé au gène rapporteur luciférase dans des cellules β primaires. Des expériences similaires out montré que la région de réponse au PIC dans le promoteur Fas est localisée entre les nucléotides –223 et –54. La mutation des sites de liaison de NF-κB et de C/EBP inhibait l'activité du promoteur Fas induite par le PIC. L’induction de l'activité du promoteur Fas était parallèle à une susceptibilité plus grande à l'apoptose induite par FasL des cellules β traitées aux cytokines + PIC. Les combinaisons IL-1β + IFN-γ, PIC + IFN-γ ou PIC + IL-1β induisaient une augmentation de 2-3 fois du nombre de cellules β apoptotiques. Le blocage de l'activité de l’iNOS diminuait de manière significative l'apoptose induite par PIC + IL-1β, mais pas par PIC + IFN-γ. L'infection des cellules β par un adénovirus exprimant IκB

, un inhibiteur de NFκB, inhibait à la fois la nécrose et l'apoptose induite par PIC + IL-1β ou PIC + IFN-γ. Les ARNm de plusieurs chémokines (IP-10 (CXCL10), MCP-1 (CCL2) , fractalkine (CX3CL1) et MIP-3α (CCL20)) et d’une cytokine (IL-15) étaient induits par le PIC, seul ou en combinaison avec l’IFN-γ, dans les cellules β pancréatiques. Cependant, il n'y avait pas d'induction de l'expression de l’IL-1β. Le PIC a un effet additif sur la mort des cellules β induite par les cytokines, par des mécanismes à la fois dépendant du NO (dans le cas de l’IL- 1β) et indépendant du NO (dans le cas de l’IFN-γ).

Nous proposons que le dsRNA, généré pendant une infection virale, pourrait

contribuer au dysfonctionnement de la cellule β à la fois en induisant l'expression de

chémokines et de l’IL-15, agents potentiels du développement de l’insulite, et en agissant en synergie avec les cytokines produites localement afin d'induire l'apoptose de la cellule β. La production de NO et l'activation du facteur de transcription NFκB semblent jouer un rôle central dans les effets délétères du dsRNA dans les cellules β pancréatiques.

Afin d'élucider les mécanismes moléculaires impliqués dans les effets du PIC et de l' IFNγ, nous avons analysé le profil d'expression génique des cellules β primaires de rat et les modifications d'expression induites par ces agents grâce à la technique des puces à ADN ("microarray"). Après une exposition de 6 ou 24h à l' IFNγ, au PIC ou à une combinaison de ces deux agents, nous avons observé des modifications de l'expression de 51 à 189 gènes. L' IFN-γ modifie l'expression de gènes MHC-connexes, ainsi que l'expression de gènes impliqués dans le métabolisme de la cellule β, dans les synthèse et sécrétion des protéines ainsi que dans les voies de signalisation des cytokines. PIC affecte préférentiellement l'expression de gènes impliqués dans les processus d'adhérence cellulaire, dans les voies de signalisation des cytokines et du dsRNA, ainsi que l'expression de gènes codant pour des facteurs de transcription et les gènes MHC-connexes. L'exposition des cellules pancréatiques au PIC et/ou à l' IFN-γ induit l'expression de plusieurs chemokines et cytokines qui pourraient contribuer à l'attraction et à l'activation des cellules mononucléaires pendant l'infection virale;

l' IFN-γ exerce un feedback positif sur sa voie de signalisation intracellulaire. Le traitement au

PIC + IFN-γ inhibe l'expression de l'ARNm de l'insuline et du GLUT-2, le transporteur du

glucose spécifique aux cellules pancréatiques β sans toutefois modifier l'expression de l'

ARNm codant pour le facteur de transcription pdx-1. En se basant sur les résultats obtenus,

nous proposons un modèle général pour les mécanismes moléculaires impliqués dans le

dysfonctionnement et la mort de la cellule β observés après exposition au dsRNA + IFN-γ.

1. Introduction

1.1 The autoimmune process and type 1 diabetes mellitus

Autoimmune diseases, such as Type 1 diabetes mellitus (T1DM), are caused by autoantibodies or autoreactive T cells that provoke inflammation, functional alterations, and anatomical lesions (1). Four criteria are used to define autoimmune diseases (2):

1. The disease can be transferred by the patients' antibodies or T cells.

2. Immunosuppressive therapies can slowdown or prevent the disease.

3. Manifestation of humoral or cell-mediated autoimmunity directed against the target organ is present in the disease.

4. Sensitization against an autoantigen present in the target organ can experimentally induce the disease.

T1DM is clinically defined as a condition of severe or absolute deficiency of insulin secretion (3), resulting from specific loss of insulin-secreting β cells (3-5). As an autoimmune disease, experimental diabetes can be transferred from diabetes-prone NOD mice and BB rats (two animal models of T1DM) into nondiabetic syngeneic animals by spleen cells from diabetic animals (6-8), or induced in athymic rats by transfer of normal spleen cells (9). After pancreas transplantation between human identical twins, diabetes re-appears most likely because of the infiltration of the transplanted pancreas by the recipient autoimmune cells (10).

β-cell damage can be delayed in human diabetes by immunosuppressive agents such as

cyclosporine (11;12), Steroids(13), Thymopoietin (14) OK3 (15) etc, and it can be prevented

in animal models of autoimmune diabetic NOD mice and BB rats by cyclosporine (16-19),

Deoxyspergualin (20), Rapamycin (21) and anti-IFN

monoclonal antibodies (22). Regarding

the manifestations of anti-β-cell autoreactivity, most pre-diabetic patients have islet-specific

autoantibodies including islet cell cytoplasmic antibodies (ICA), insulin autoantibodies

(IAA), antibodies against IA-2 (IA-2-Ab) and the 65 kDa isoform of glutamic acid decarboxylase (GAD

-Ab), which are immune markers for T1DM (23). Moreover, T cells from T1DM patients proliferate in response to β-cell autoantigens (24). Further indirect evidence support the autoimmune nature of human T1DM; including the presence of insulitis (infiltration of the islets of Langerhans by mononuclear cells) (25-27) and the association of T1DM with HLA genes (associated with most autoimmune diseases) (23). In animal models, thymic anomalies were observed in both NOD mice (28-30) and BB rats (31), decreased IL-4 production (32) and delayed T cell apoptosis (33-35) were observed in NOD mice, while lymphocytopenia (36) and increased NK cell activity were observed (37)in BB rats.

A major characteristic of the immune system is that B and T cells are physiologically tolerant to most self-antigens. During the process of precursor T-cell receptor gene recombination, only 10% of T-cells succeed in expressing both the α and β chains of the T- cell receptor (38), while the other T-cells are eliminated by apoptosis. Once a complete T-cell receptor is expressed on the cell surface, the T-cells in the thymus undergo negative or positive selection (38). During positive selection, the newly rearranged T-cell receptors (TCRs) expressed on developing CD4

CD8

thymocytes interact with MHC molecules on cortical thymic epithelium (TE)(39). During negative selection, developing T cells interact with TE and bone marrow (BM)-derived antigen (Ag)-presenting cells or thymic medullary dendritic cells (DCs), resulting in the deletion of self-reactive T cells(39). In this case the T- cell receptors bind with high affinity to self-antigens presented by thymic antigen-presenting cells and the cell is eliminated by apoptosis, thus decreasing the risk of autoimmunity.

Those autoreactive T-cells that resist the apoptotic process are delivered to the periphery. T-

cells expressing T-cell receptors that recognize a foreign peptide cross-reactive with a self-

peptide may undergo further maturation to finally express either CD4 or CD8 markers (40).

In T1DM, the self-tolerance to β-cell antigens is lost(1), perhaps due to a defective negative selection involved in central tolerance mechanism (39). This could happen either because the β-cell target autoantigens are not present in the thymus at sufficient concentration to induce negative selection or because these antigens are present in the thymus but diabetogenic epitopes are subdominant or cryptic and do not give rise to negative selection (41). The evidence that neonatal intrathymic islet grafts prevents the onset of T1DM in BB rats supports this hypothesis (42;43).

Viral infection may play an important role in the loss of self-tolerance. Viral infections modify HLA gene expression through IFN production, and might lead to immunogenic expression of a subdominant or cryptic autoantigen on the surface of the target cell (41;44- 49). It may also be involved in the expression of a neoantigen on the β-cell surface or bypass anergized T-cells by molecular mimicry (41;44-49). These, and other potential mechanisms of viral-induced diabetes, are discussed below (1.3.4).

The balance between Th1 and Th2 cells probably contributes to trigger T1DM (50- 54). Antigen-specific T cell activation leads to the differentiation of naive CD4

Th cells into Th1 and Th2 clones, which are defined by their pattern of cytokine production and functions.

Th1 cells produce IL-2, IFN-γ, and TNF-α and stimulate cell-mediated immunity and

delayed-type hypersensitivity reactions. Th2 cells produce IL-4, IL-5, IL-10, and IL-13 and

promote humoral responses(55;56). Th1 cytokines promote Th1 and inhibit Th2 activity,

whereas Th2 cytokines induce Th2 but block Th1 activity (50;51). Th1 cells and their

cytokines play a direct role in the onset and progression of T1DM (50;54). Thus, recent-onset

T1DM is associated with predominance of Th1 cytokines and decreased production of Th2

(IL-4) cytokines (57-60). Adoptively transferred diabetogenic Th1 cells provoke insulitis in

NOD/SCID mice (61), and there is dominance of Th1 cytokines in the female, but not male

NOD mice, which is consistent with the higher prevalence of diabetes in females (62). Direct

and indirect mechanisms are involved in Th1 cytokine-induced or –accelerated β-cell death.

Th1 cytokines directly activate macrophages and CD8

cells and increase infiltration by these cells in the islets, leading to release of cytotoxic mediators such as NO and oxygen radicals.

Th1 cytokines also induce the activation and expansion of bystander autoreactive T-cells and suppress the production of soluble cytokine antagonists, which indirectly contribute to β-cell destruction (50;51). Of note, pancreatic homing of Th2 cells and predominance of Th2 cytokines were found in insulitis associated with new-onset T1DM (63-66). Surprisingly, under some experimental conditions expression of Th2 cytokines accelerates β-cell destruction instead of overcoming the autoimmune assault (67-69). Moreover, anti-IL-10 antibodies prevented peri-insulitis and insulitis in NOD mice (65). The role of Th2 cytokines in T1DM is complex and depends on the relative contribution of IL-10 in the process. IL-10 may induce necrosis via closing the microvasculature and activating T and B cells (50;51). In addition, Th2 cytokines enhance the MHC class II expression and change the expression of endothelium-bound adhesion, thus stimulating the homing of macrophages, B cells, and eosinophils (50;51). Finally, by activating resident immune cells and promoting pancreatic infiltration, Th2 cytokines amplify the cascade of anti-β-cell immunity (50;51). Thus, it seems that predominantly Th-1-mediated infiltration comprises CD8

and CD4

T cells, leading to persistent and sustained immune attacks and β-cells apoptosis. (50;51). Th2-mediated infiltration includes eosinophils, macrophages, and fibroblasts, leading to β-cell death by necrosis(50;51). It remains to be determined which is the exact role for Th1/Th2 cells in human T1DM.

1.2 Insulitis and β-cell damage in T1DM 1.2.1 Insulitis

Insulitis, the infiltration of mononuclear cells in and around the islets of Langerhans,

is the consequence of an anti-β-cell-mediated immune response. T1DM in animal models, and

probably also in human disease, has two phases, which are, to some extent, superimposed: a.

periinsulitis and insulitis, the phase when leukocytes invade the islets; b. overt diabetes, when the invading mononuclear cells have killed so many β-cells that there is not enough insulin secretion to maintain normal glycaemia. (70). In rodent models, invasive and destructive insulitis is preceded by periinsulitis (mononuclear cell infiltrate around the islets) (41).

Destructive insulitis can be transferred to non-diabetes-prone mouse, rat, or human pancreas indicating that the anomaly is at least in part due to the immune system.(71-73). The defect of the immune system is mostly expressed in T-cells, because diabetes can be transferred to healthy recipient by purified T cell population (6;74) or T cell clones (75;76). Both CD8

T- cells and CD4

T-cells are involved in this chronic inflammatory infiltrate. CD8

T-cells directly recognize β-cells via the MHC class I protein, but CD4

T-cells cannot recognize β- cells directly because these cells do not express MHC class II proteins. CD4

T-cells recognize antigens presented by local antigen-presenting cells, like dendritic cells, macrophages and B cells. Thus, CD4

T-cell-depending β-cell death proceeds indirectly, without antigen-specific interaction between CD4

T-cells and β-cells. CD4

T cells also participate in the activation of CD8

T-cells by activating antigen presenting cells (77). By using chimeric mice, in which the bone marrow-derived antigen-presenting cells, but not the islet β-cells, are capable of presenting antigen to T-cell, it was demonstrated that insulitis and β-cell destruction can proceed in the absence of direct recognition of islet β-cell surface antigen by T-cells (78). Thus, the interactions between macrophages, CD4

T-cells and CD8

T-cells that establish a chronic inflammatory lesion are crucial for β-cell destruction in T1DM. Cytokines, chemokines, and Fas-FasL interaction are important factors in this process (see below, 1.3).

1.2.2 β-cell death: apoptosis or necrosis?

Cell death is a crucial event in the development, maintenance and function of multicellular organisms, but when excessive or inadjusted it can lead to tissue-specific diseases. Necrosis and apoptosis are two different forms of cell death. They present different morphological features, underlying different signalling and effector mechanisms (79).

Necrosis usually appears in cases of overwhelming cell injury, leading to loss of selective permeability of the cytoplasmic membrane. This can either be due to direct damage to the cytoplasmic membrane, or indirectly, as the result of severe energy depletion. As consequence, water and calcium ions enter the cell, leading to cellular swelling and subsequent rupture of plasma and organelle membranes and dissociation of organized structures (79).

Apoptosis is a form of programmed cell death that enables metazoans to control cell number in tissues and to eliminate damaged, abnormal, misplaced, non-functional or potentially dangerous individual cells (80). Unlike necrosis, which is secondary to acute, nonphysiological injury, apoptosis is a programmed ¨cell suicide¨. Thus while necrotic cells lyse and release their cytoplasmic and nuclear contents into the intercellular milieu, inducing inflammation, apoptotic cells shrink, lose intercellular contacts, show dense chromatin condensation, cytoplasmic blebbing and cellular fragmentation into small apoptotic bodies.

Apoptotic cells are phagosized and eliminated at an early stage by neighbouring cells or macrophages, without causing a major inflammatory response. (81).

Studies in animal models suggest that apoptosis is the main mode of β-cell death in

autoimmune-mediated diabetes mellitus (82-84). In NOD mice, β-cell apoptosis precedes islet

lymphocytic infiltration (85). β-cell apoptosis was also observed in the multiple-low-dose-

streptozotocin-treated mice (86), interferon-γ transgenic mice (under the control of rat

glucagon promoter) (87), and in infiltrated islets of BDC2.5/NOD.scid mice (an accelerated

model of diabetes, which has CD4

T-cells bearing a transgenic T-cell receptor but are devoid

of CD8

T-cells) (88). In diabetes prone BB rats, the time course of islet cell apoptosis correlates closely with the appearance of early insulitis and decreased pancreatic insulin staining (68 days of age) (89). A further increase of islet cell apoptosis is observed in diabetes prone BB rats at day 85, coinciding with the onset of diabetes (89). Of note, β-cell apoptosis was observed in two young patients who died in ketoacidosis after diagnosis of diabetes (90).

The major players in apoptosis are cysteine proteases cysteine aspartase (caspase), adaptor proteins, proteins from the tumor necrosis factor receptor family and from the Bcl-2 family (81) (Fig 1, Fig 2)). Apoptosis-related receptors contain a death domain (DD), which interacts with DD-containing adaptors; these adaptors then interact, again through DD domains, with death effector domain (DED)-containing effector proteins(81;91;92).

Caspases are synthesized as inactive proenzymes which are activated by cleavage at specific Asp residues to activate enzymes containing both large (p20) and small (p10) subunits to form heterotetramers (92;93). Caspases are not only responsible for degradation of cellular substrates during the end stage of apoptosis, but are also critical for the onset of cell death (93;94). At least 14 caspases have been identified. Among them, some are sufficient to promote self-processing (initiator caspases like caspase 2, 8, 9 and 10), while some are directly responsible for the proteolytic cleavages that lead to cell disassembly (effector caspases like caspase 3, 6, 7 and 12) (81;92;94;95). Caspases participate in apoptosis in a well-organised manner. First they inactivate proteins that protect living cells from apoptosis, such as Bcl-2 and I

/DEF45 (an inhibitor of CAD; caspase-activated deoxyribonuclease).

Second, they can directly disassemble cell structures. For example, caspases cleave a single

site of lamins causing lamina (the major structural proteins of the nuclear envelope) to

collapse thus contributing to chromatin condensation (93;94). Caspases also

DDDDDED DED DDDD DD DDDEDDED DDDDDED DED DDDDDED DED DDDDDD DD DDDEDDEDDD DDDDDEDDEDDEDDED

Caspase 8 Caspase 8

Effector caspases

Apoptosis Apoptosis

Death signals FasL

Death receptors Fas

Adaptors FADD FADD

Activator caspases

Fig 1 Apoptosis signaling by Fas (adapted from ref. 80).

DDDD DDDDDD DD

Death signals

Death receptors

Adaptors

Activator caspases

DDDED

Caspase 8

Effector caspases

Apoptosis

FADD

DEDDDDDDEDDEDDED

Caspase 8 Caspase 8

Effector caspases

Apoptosis Apoptosis

FADD

DEDDDDD

DDDD

TRADD TRADD

TNF

TNFR1

DDDD

RIP

TRAF2

AP1/2

MEKK1 JNKK

JNK c-Jun

NIK IKK

IκB/NF-κB NF-κB

Fig 2 Apoptosis signaling by TNFR1 (adapted from ref. 80).

indirectly reorganize cell structures by cleaving proteins involved in cytoskeleton regulation, like gelsolin, focal adhesion kinase (FAK) and p21-activated kinase 2 (PKA2). Finally, caspases dissociate the regulatory and effector domains of key proteins. For example, they inactivate or deregulate proteins involved in DNA repair (like DNA-PK), mRNA splicing (like U1-70K) and DNA replication (like replication factor C) (94).

Adaptor proteins connect caspases and up-stream regulators of apoptosis. Associations between adaptor proteins and caspases or TNF-R family members are mediated by homotypic interactions between death domain (DD), death effector domain (DED) (in case of caspase 8 and 10), and the caspase recruitment domain (CARD) (in case of procaspases 1, 2, 4, 5, 9, 11, and 13) (Fig 1, Fig 2) (81;92). The DD is a region in the cytoplasmic part of CD95 and related TNF-R family members, and also in adaptor molecules such as Fas-associating death domain protein/mediator of receptor-induced toxicity (FADD/MORT1), TNF-R1-associated death domain protein (TRADD), and receptor-interacting protein (RIP) (Fig 1, Fig 2)(81). After cross-linking with the DDs of a TNF-R family member, the adaptors induce the aggregation and activation of caspases through the death effector domain (DED) (81).

Depending on the cell type and the signals to the cell, members of the TNF-R family have pleiotropic actions, such as triggering proliferation, survival, differentiation, or death (81). These receptors are activated by members of the TNF ligand family. Most of these ligands are synthesized as membrane-anchored trimers, and extensive receptor cross-linking is required for signalling (Fig 1, Fig 2) (81).

Members of Bcl-2 family are proapoptotic or antiapoptotic (81;92). The

proapoptotic/antiapoptotic balance of Bcl-2 family members will effect apoptosis both by

cytochrome c release into the cytoplasm and by caspase activation (92). Bcl-2 itself, Bcl-x

,

and Mcl-1 are typical inhibitors of apoptosis. They contain four domains with similar

sequence (Bcl-2 homology regions BH1–BH4). Some viral homologs also promote cell

survival and have three or four regions with extensive amino acid sequence similarity with Bcl-2 (81;91). Other members in this family contain either one (‘BH3-only” proteins like Bid, Bad, Bim, Noxa, and Puma) or three (“BH1-3” proteins, including Bax, Bak, and Bok) homology regions and act as apoptosis enhancers (81;91).

Several key apoptosis signalling pathways have been described in different cell types, including the pancreatic β-cells. Among them, are: A. stress-induced and Bcl-2 family- regulated. Apaf-1 and caspase-9 are required in this process; B. Fas or TNF-RI-activated, requiring FADD and caspase-8. This pathway is usually not blocked by Bcl-2 or its homologs (81;81;91); C. The ER stress pathway, which is mediated by the transcription factor CHOP/GADD153 and caspase-12 (95-98).

Under stress, BH3-only proteins might induce mitochondrial permeabilization directly via Bax/Bak-lipid interaction, while Bax can also promote permeabilization transition (PT) pore formation by regulating ER Ca

stores (91). Both lead to the release of proapoptotic mitochondrial proteins including cytochrome c (91). In the cytoplasm, cytochrome c, together with Apaf-1, ATP, and caspase 9, forms the apoptosome, which activates caspase 9 and upstream initiator caspases, thus initiating the apoptosis cascade (92).

FasL is a homotrimeric molecule binding three Fas molecules (Fig 1). Fas ligation leads to clustering of the receptors' death domain; FADD then binds to the clustered receptor death domain through its own death domain. FADD also contains a ¨death effector domain¨

(DED) that binds to an analogous domain repeated in tandem within the zymogen form of

caspase-8 (Fig 1). The DED domain is a member of homophilic interaction domains termed

caspase recruitment domain (CARD), which is found in several caspases with prodomains,

including caspase –2, -8, -9 and -10 (80;81). Upon recruitment by FADD, caspase-8

oligomerization drives its activation through self-cleavage. Caspase-8 then activates

downstream effector caspases such as caspase-9 committing the cell to apoptosis (Fig 1) (80;81).

TNF trimerizes TNF-R1 upon binding, leading to expression of proinflammatory and immunomodulatory genes via activation of the transcription factors NF-κB and AP-1 (Fig 2) (80). There are pre-existing cellular factors that can suppress the apoptotic stimulus generated by TNF, as suggested by the fact that in most cell types TNF only triggers apoptosis when protein synthesis is blocked (80;81). After binding with TNF-RI, TNF induces clustering of the receptors' death domain (Fig 2). The clustered receptor death domains then bind to the adaptor TRADD (TNFR-associated death domain) through death domain. TRADD recruits TNF-R-associated factor-2 (TRAF2), leading to activation of NF-κB and JNK/AP-1, whereas FADD mediates activation of apoptosis (Fig 2) (80). TRAF2 and RIP activate NIK (NF-κB- inducing kinase), which then activates IKK (Inhibitor of κB (I-κB) Kinase Complex). IKK phosphorylates I-κB, which is degraded in the proteasome, releases NF-κB, which then translocates to the nucleus and activates transcription (80). The pathway from TRAF2 and RIP to JNK involves the cascade mitogen-activated protein (MAP) kinases MEKK1 (MAP/Erk kinase kinase –1)-- JNKK (JNK kinase)--JNK (80;81).

The endoplasmic reticulum (ER) regulates the folding, exporting and processing of newly synthesized proteins. Various genetic and environmental stresses, such as glucose deprivation, perturbation of calcium homeostasis, and exposure to free radicals cause accumulation of unfolded proteins in the ER and induce ER stress (99). Calcium is an important signalling molecule in the ER. Ca

is released from the lumen of the ER via Ca

channels and transported back to the lumen of ER via the sarcoplasmic-endoplasmic-

reticulum Ca

ATPase (SERCA) (100). The ER chaperones calreticulin and calnexin are

important in this signalling process (100). The effector caspase-12 is specifically activated

during ER stress (101;102), while the transcription factor CHOP/GADD153 is up-regulated in

parallel, also contributing to cell death (101). In response to ER stress, via binding to IRE1 (a proximal sensor), c-Jun N-terminal inhibitory kinase (JIK) modulates IRE1 -TRAF2 (tumor necrosis factor receptor-associated factor 2) complex formation. TRAF2 interacts with procaspase-12, leading to its clustering and cleavage (103). Pancreatic β-cells have a highly developed ER due to their specialised functions of insulin synthesis and secretion (98;104;105). When ER functions are severely impaired, CHOP/GADD153 is induced in β- cells, leading to activation of c-Jun NH2-terminal kinase and/or caspase-12, and apoptosis (104). It has been shown in FACS-purified β-cells that the expression of mRNAs for GADD153/CHOP and TRAF-2 ARE induced by cytokines, while expression of SERCA2 is decreased after cytokine treating (106).

Mitogen-activated protein kinases (MAPK) are central in signalling β-cell apoptosis (83). There are three groups of MAPK: the extracellular signal-regulated kinases (ERKs), the c-Jun NH

-terminal kinase (JNK), and the p38 protein kinases (83). ERK is mainly activated by mitogens and cellular stressors, while JNK and p38 are mainly activated by cellular stress including cytokines (83;107) and viral infection (108). The JNK pathway is a major signal for cytokine-induced β-cell apoptosis (83), phosphorylating serine and threonine residues in several substrates, including downstream protein kinases and transcription factors or precursors of transcription factors. This triggers a cascade of signal transduction that culminates in cell death (83;84).

1.3 Initiators, contributors and effectors of β-cell death: cytokines, chemokines, viral infections and nitric oxide

Two main models of β-cell death have been suggested. In the first, β-cells are killed

by cytotoxic T-cells using perforin or granzymes as effector molecules. Perforin, a pore-

forming protein, is a key component of cytotoxic T-cell (CTL) granules. It causes lysis of the

target cell by binding to cell membrane and aggregating to form a polyperforin-lined pore.

Granzymes are a family of at least 11 serine proteases, which cause DNA fragmentation.

Granzymes A and B are the main mediators of apoptosis (82;84). In the other model, islet- invading T-helper cells release cytokines and chemokines, inducing mononuclear cell infiltration and endothelial cell activation. Activated endothelial cells express adhesion molecules and liberate inflammatory mediators, while recruited macrophages are stimulated by interferon (IFN)-γ to produce interleukin (IL)-1β and tumor necrosis factor (TNF)-α, which in synergy with IFN-γ lead to β-cell toxicity via β-cell-specific induction of iNOS and other apoptosis-activating pathways. In addition, IL-1β induces Fas expression in β-cells, rendering them susceptible to lysis triggered by FasL expressed on the surface of Th1 and cytotoxic T-cells (82;84).

1.3.1 Cytokines

Cytokines are a class of small polypeptide proteins, which are secreted mostly by cells from the immune system. In the normal immune system, autoreactive T cells are prevented from expressing their autoreactive potential by other regulatory (suppressor) T cells (109).

The direction taken by the autoreactive and regulatory T cells, in terms of Th phenotype, is

largely regulated by their respective cytokine products (110). Th1 cells secrete mostly IL-2

and IFN-γ, whereas Th2 cells secrete IL-4 and IL-10. Some cytokines are produced by both

Th1 and Th2 cells, such as IL-3, TNF-α and chemokines (110;111). Th1 cells and their

cytokine products are important mediators of cell-mediated immunity. IFN-γ, TNF-α, and

TNF-β activate vascular endothelial cells to recruit circulating leukocytes into the tissues at

the site of antigen challenge, and they in turn activate macrophages to eliminate the antigen-

bearing cell. Furthermore, IL-2 and IFN-γ activate cytotoxic T cells and NK cells to destroy

target cells in respectively an MHC-dependent and/or MHC-independent fashion. In contrast,

Th2 cytokines are mostly stimulators of humoral immune responses (antibody production,

especially IgE) by B-lymphocytes. The Th1 cytokine IFN-γ inhibits the production of the Th2 cytokines IL-4 and IL-10; these, in turn, inhibit Th1 cytokine production (110;111).

In NOD mice and BB rats, β-cell destructive insulitis is associated with increased expression of proinflammatory cytokines (IL-1, TNF-α, and IFN-α) and type 1 cytokines, whereas non-destructive insulitis is associated with increased expression of type 2 cytokines and TGF-β (111). Destruction of islet grafts in NOD mice is also associated with increased local expression of IL-1β and IFN-γ (112;113). Type 1 cytokines activate: (a) macrophages to produce proinflammatory cytokines (IL-1 and TNF ), (b) cytotoxic T cells that interact specifically with β-cells and destroy them by upregulating MHC class I expression on the β- cells (an action of IFN-γ), or by inducing Fas (CD95) expression in β-cells (mainly an action of IL-1, (114;115)). Furthermore, the cytokines IL-1, TNF-α, and IFN-γ are directly cytotoxic to β-cells, by inducing expression of pro-apoptotic genes and the formation of oxygen free radicals, nitric oxide, and peroxynitrite (84;111). In addition to data from NOD mouse and BB rat models, there is evidence for cytotoxic actions of IL-1, TNF-α, and IFN-γ on human islets in vitro, and IFN-α and IFN-γ have been detected in islets of patients with recent-onset T1DM (84;110).

The cytokines IL-1

, TNF-

and IFN-

has been shown to induce rodent and human

-cell death mostly by apoptosis (84;88;116-118). As described above, these cytokines may act as mediators of pancreatic islet

-cell functional impairment and destruction, and in animal models of type 1 diabetes mellitus, IL-1β, IFN-

, and TNF-α are produced in and around the islets during insulitis (84;111;119-122).

Interferons are a group of cytokines with a crucial role in the host defense against

microbial and viral agents (123). IFNs are classified into two types: type I IFNs (“Viral IFNs”)

includes IFN-α, IFN-β, and IFN-ω, while type II IFN (“Immune IFN”) is IFN-γ. The viral IFNs are mostly induced by virus infection, whereas immune IFN is induced by mitogenic or antigenic stimuli, and, in some cases, also by viral infection (124). Most cell types infected by virus are able to synthesize IFN-α/β in cell culture (125) but not it is unclear whether β-cells produce these cytokines. IFN-

is produced mostly by T-lymphocytes and natural killer (NK) cells and bind to a species-specific cell surface receptor (125). IFN-γ production is stimulated by macrophage-derived cytokines, including TNF-α, IL-12, IL-18 and IFN-γ itself (126). It works through activation and regulation of both specific and nonspecific immune responses (127;128). IFN-mediated signalling and transcriptional activation of gene expression occurs via the JAK-STAT pathway (84;124). In response to IFN stimulation, the signal transducer and activator of transcription (STAT) family of proteins become tyrosine phosphorylated by the Janus family of tyrosine kinases (JAK). For IFN-α/β, the phosphorylated forms of Stat- 1α/β and Stat-2, together with an additional non-STAT protein, p48 (also known as IRF-9), translocate to the nucleus and constitute a complex known as ISGF-3. The ISGF-3 trimeric complex binds to a cis-acting DNA element – IFN-stimulated response element (ISRE). In the case of IFN-γ, the phosphorylated Stat-1α factor homodimerizes, translocates to the nucleus, and binds to a different cis-acting element – the gamma-activated sequence (GAS) (84;124).

As part of their antiviral activities, IFNs induce expression of several protein in virus-infected

cells. Among these proteins, IFN-inducible RNA-dependent protein kinase (PKR), the 2', 5’-

oligoadenylate synthetase (OAS), RNase L, the RNA-specific adenosine deaminase (ADAR),

and the Mx protein GTPases are important in the antiviral actions (124). PKR phosphorylates

protein synthesis factor eIF-2α, leading to an inhibition of translation. Both IFN treatment and

virus products, such as dsRNA, synergize to inhibit eIF-2 and change the translational

pattern of the host cell (124). OAS catalyzes the synthesis of oligoadenylates of the general

structure ppp(A2'p)

A (2-5A), which possess a 2',5'-phosphodiester bond linkage. RNase L, a

latent endoribonuclease, is activated by binding 2-5A oligonucleotides. Activated RNase L catalyzes the degradation of both viral and cellular RNAs, including cellular rRNA. The expression, regulation and function of the OAS and the 2-5A-dependent RNase L have been characterized extensively in IFN-treated and virus-infected cells (124). ADAR edits viral RNA transcripts and cellular pre-mRNAs by delaminating adenosine to yield inosine at posttranscriptional level, which then alters the functional activity of viral and cellular RNAs and affect the biological processes (124). Proteins MxA and Mx1 have well- characterized antiviral function and belong to the Mx family of proteins. Mx alone is sufficient to block the replication of virus in the absence of other IFN-inducible proteins. Mx proteins are inducible by IFN-α and IFN-β but not by IFN-γ, and are GTPases that belong to the superfamily of dynamin-like GTPase, which associate with viral protein complexes. The spectrum of antiviral activities of the Mx proteins, and the molecular mechanisms by which they act to inhibit viral replication, are dependent on the specific Mx protein, its subcellular site of localization, and the type of challenge virus examined (124).

IFN-γ may be involved in the pathogenesis of diabetes via: a. Up-regulation of

expression of MHC molecules; b. the development of a Th1 cell phenotype; c. macrophage

activation, d. direct induction of β-cell apoptosis; e. induction of synthesis of adhesion

molecules and chemokines (126;129) (112;130). IFN-

has been found in mononuclear cells

infiltrating islets of pre-diabetic NOD mice and BB rats (128;131), and during early onset of

diabetes in humans (132). Transgenic mice expressing the IFN-

gene under the control of the

insulin promoter have autoimmune β-cell destruction and diabetes (60), which can be

prevented by anti-IFN-

-antibodies (133). Treatment with IFN-

-antibodies also prevents

diabetes in NOD mice (134). On the other hand, transgenic expression of IFN-

under the

control of the glucagon promoter induces apoptosis in a proportion of

-cells without leading

to insulitis or diabetes (87). This indicates that IFN-

production alone is not sufficient to

induce overt diabetes mellitus. In vitro, IFN-

synergizes with IL-1β and TNF-α to induce direct

-cell damage (135-138). IFN-

is required for IL-1

-induced NO production by human islets (135), while in rat and mouse islets it potentiates IL-1

-induced iNOS mRNA expression and NO production (139;140).

IL-1

induces the expression of a variety of genes and proteins that contribute to both acute and chronic inflammatory responses. These include induction of secondary cytokines such as IL-6 and colony-stimulating factor (CSF), induction of fever and other components of the acute phase response, induction of adhesion molecules and chemokines, and induction of inducible nitric oxide synthase (iNOS) (141;142). IL-1β induces iNOS mRNA expression and NO production in rat islets (143), which is paralleled by decreased insulin synthesis and release (135;144). IL-1β is cytotoxic to whole rat islets (116), but it does not cause the death of FACS-purified single β-cells (145). However, a combination of IL-1β + IFN-

+ TNF-α or IL-1β + IFN-γ kill purified rat (146), mouse (147), and human β-cells (136). The induction of iNOS in human islets requires combination of IL-1β and IFN-γ (137;148;149). On the other hand, cytokine-induced β-cell dysfunction and death in human islets and apoptosis in mouse islets are apparently NO independent. (129;136;137). IL-1β also inhibits mouse islets insulin release, (pro)insulin biosynthesis and insulin mRNA expression (150).

1.3.2 Chemokines

Once immune tolerance is broken and the β-cells are recognized as targets by the immune system, the pancreatic islets are progressively invaded by mononuclear cells.

Mononuclear cell migration is regulated by chemokines, a large family of small proteins with

four conserved cysteines linked by disulfide bonds (151). Chemokines are subdivided into

four families according the relative position of their cysteine residues. The C family lacks the

first and third cysteine residues found in the other three families. There are two members in

this family: chemokine ligand (CL) 1(lymphotactin-α) and CL-2 (lymphotactin-β). Members of this family mostly recruit T lymphocytes. In the CC family of chemokines (β-chemokines), the first two-cysteine residues are adjacent to each other. They evoke migration of monocytes, lymphocyte and dendritic cells (DC). Members of this family include macrophage inflammatory protein (MIP) -3α (CCL20), macrophage chemoattractant protein (MCP) -1 (CCL2) and regulated upon activation normal T cell expressed and secreted (RANTES (CCL5)). In the CXC family of chemokines (α-chemokines), one amino acid separates the first two-cysteine residues. They are chemotactic for neutrophils (glutamic acid (E)-leucine (L)-arginine (R) amino acid motif positive) or lymphocytes (ELR negative). Members of this family include IL-8 (CXCL8), SDF-1 (CXCL12), GROα (CXCL1) and IP-10 (CXCL10).

Fractalkine (CX3CL1) is the only member of the last family (CX

C family). Fractalkine (CX3CL1) is a membrane-bound glycoprotein in which the first two-cysteine residues are separated by three amino acids. Fractalkine (CX3CL1) binds to the membrane via a mucin stalk, allowing it to function as both an adhesion molecule (possibly aiding DC-T-cell interactions) and as a chemoattractant (for IL-2 activated natural killer (NK) cells and CD8

T cells) (152;153). During leukocyte movement chemokines provide signals to convert the low- affinity, selectin-mediated interaction into high-affinity, integrin-mediated interactions between leukocytes and endothelial cells that leads to extravasation of leukocytes (152). In different animal models, the neutralization of specific chemokines result in inhibition of leukocyte infiltration and tissue pathology (154-158). The main stimuli for chemokine production are early pro-inflammatory cytokines (such as IL-1 and TNF-α), bacterial products (like lipopolysaccharide) and viral infection (152).

The CC chemokines macrophage chemoattractant protein – 1 (MCP-1 (CCL2)),

macrophage inflammatory protein (MIP-1β) (CCL4), regulated upon activation normal T cell

expressed and secreted (RANTES (CCL5)), and the CXC chemokine interferon-γ - inducible

protein (IP-10 (CXCL10)) are highly expressed in islets obtained from diabetes-prone NOD mice at the early stages of insulitis, and probably contribute to mononuclear cell homing and adhesion (112;130;159;160). Moreover, increased levels of IP-10 (CXCL10) and MCP-1 (CCL2) have been observed in the serum of T1DM patients at the disease onset (161;162). It is noteworthy that the β-cells themselves, upon exposure to IL-1β and/or IFN-γ, express several chemokines and cytokines, including IP-10 (CXCL10), MCP-1 (CCL2), fractalkine (CX3CL1), MIP-3α (CCL20), and IL-15 (112;130;158;163-166). Expression of these chemokines is regulated to a large extent via NF-κB activation (106;167).

1.3.3 Fas and FasL

The cell death receptor Fas (CD95) is expressed on a wide variety of cells. FasL (CD95L) is a membrane protein typically found on the surface of activated T cells and natural killer (NK) cells (126). Apoptosis induced by FasL has at least two functions in the normal homeostasis of mammals -- deletion of activated lymphocytes from the periphery after an immune response, and killing of target cells (e.g. virus-infected or tumor cells) by CTLs or NK cells (126). Fas expression can be upregulated by the cytokines IFN-γ, IL-1β and TNF-α in some cell types (126).

In NOD mice, increased Fas expression is well correlated with the time of a main

“diabetes checkpoint”, around 12 weeks of age (168). β-cell Fas expression has also been reported in inflamed islets during progression to overt diabetes (126), while FasL is expressed mostly in islet-infiltrating cells in patients with insulitis, with CD8

as the most prevalent FasL-positive phenotype. FasL expressed on macrophage and CD4

are also detected (169).

Other data comes from mice with lpr (Lymphoproliferation, Complementary Mutation of Fas)

and gld (Generalized Lymphoproliferative Disease, complementary mutation of FasL)

mutations (170) (171). NODlpr mice did not develop insulitis or diabetes either

spontaneously or after adoptive transfer of diabetic NOD splenocytes or the CD8

islet- specific T cell clone G9C8 (170;172). The incidence of diabetes is markedly reduced following transfer of diabetogenic T cells into irradiated NODlpr mice (170). Transgenic expression of Fas ligand (FasL) on β-cells in NOD mice may result in accelerated T1DM (168). NOD mice heterozygous for the gld (with reduced function of FasL expression but no lymphadenopathy) failed to develop T1DM. Even without large number of FasL-expressing T cells, cells that could impair the function of adoptively transferred cells, NOD-lpr/lpr- scid/scid mice still showed delayed onset and reduced incidence of T1DM as compared to their Fas-expressing NOD-scid/scid littermates, after adoptive transfer of diabetogenic NOD spleen cells (168;173;174). Recent data from β-cell specific expression of a dominant- negative point mutation in a death domain of Fas (lpr

or Fas

) (173) and neutralizing anti- FasL antibody (174) suggest that Fas/FasL interactions contribute to the early CD4

T-cell mediated β-cell death. Interference with this interaction in the induction phase of diabetes may retard or prevent disease progression in NOD mice. All these data suggest that induction of Fas expression on β-cells, and their consequent destruction by Fas-FasL interaction, is an important pathogenic mechanism in T1DM.

Fas expression can be up-regulated on primary β-cells and β-cell lines by IL-1

(114;126;175;176), and Fas-expressing β- cells are sensitive to apoptosis induced by soluble

FasL or agonist anti-Fas antibody in vitro(126). In vitro studies of Fas up-regulation by

cytokines and Fas-mediated apoptosis have been performed on human islets, suggesting that,

like in mouse and rat islet cells, human islet cells can also be killed in a Fas-dependent

manner(177).

The Fas promoter, originally described in liver cells (178), has binding sites for NF- κB and C/EBP (figure 3) Transcriptional regulation of the rat Fas gene by IL-1β has been studied in fluorescence-activated cell sorting (FACS)-purified primary β-cells and insulin-

producing RINm5F cells(114). In IL-1β–treated cells, the promoter activity of the constructs pFas-811 luc and pFas-223 luc was induced, whereas pFas-54 luc was not responsive to IL- 1β, indicating that the IL-1β-responsive region is located between nucleotides -223 and -54 (Fig 3)(114). By mutation analysis, it was demonstrated that both NF-κB and C/EBP binding sites are required for IL-1β induction of the Fas promoter (Fig 3)(114).

1.3.4 Viral infections as triggers of insulitis and β-cell death

T1DM is a multifactorial autoimmune disease, probably caused by the interaction of environmental factors with an inherited predisposition (23;179). Over 20 different regions of the human genome show some linkage with the disease. The strongest linkage (around 50%)is with the human leukocyte antigen (HLA) genes lying within the major histocompatibility complex (MHC) region on the short arm of chromosome 6 (IDDM1). T1DM is also linked to the region of the insulin gene on chromosome 11p (IDDM2) (23;179;180). Additional

+1

-811 -223

EcoRI -54

NciI +126 NF-κB

C/EBP Oct C/EBP

Ets Ets

Luc Luc Luc

Fig 3. Schematic representation of the rat Fas promoter

(reproduced from ref. 114)

putative candidate genes for diabetes predisposition are IDDM4 (chromosome 11q13), IDDM5 (chromosome 6q25), IDDM7 (chromosome 2q31), IDDM8 (chromosome 6q27), IDDM10 (chromosome 10p11–q11), IDDM12 (chromosome 2q33), IDDM13 (chromosome 2q35), and IDDM15 (chromosome 6q21)(180).

In addition to genetic predisposition, increasing evidence suggests that environmental factors play a key role in the pathogenesis T1DM (1;181;182). Thus, more than 60% of identical twins are discordant for the disease, and diabetes prevalence varies from country to country. After migration to the countries with a high frequency of T1DM, individuals from low T1DM frequency countries become more susceptible to the disease than their compatriots, excluding the possibility of ethnic genetic differences as the sole explanation for the prevalence of diabetes (183;184). Finally, diabetes incidence is increasing in most countries, pointing to an environmental influence. In NOD mice and BB rats some nonimmunological interventions, like changing diet (for instance by eliminating cow milk proteins) decrease diabetes prevalence (41) (49;185;186).

Viruses are one of the main candidates as environmental triggers of human T1DM (45). Viruses may act as initiating agents for autoimmunity to β-cells, and at least 6 human and 9 animal viruses have been suggested to cause clinical or experimental T1DM (45).

Encephalomyocarditis (EMC) virus, mengovirus, reovirus and retrovirus are associated with

the diabetes in mice; coxsackie B virus, particularly B4, in mice and humans; foot-and-mouth

virus in pigs and cattle; rubella virus in hamsters and rabbits; bovine viral diarrheamucosal

disease virus in cattle; and KRV in rats(45). There also are direct or indirect evidence

showing that cytomegalovirus, Epstein-Barr virus and varicella zoster virus are associated

with T1DM in humans (45). Human pancreatic

-cells can be infected in vitro by different

enteroviruses, leading to persistent infection, functional impairment and, in some cases, cell

death (187). Enterovirus antibody levels are increased in children with newly diagnosed